- Email: [email protected]
Crystal Structure of Vestitone Reductase from Alfalfa (Medicago sativa L.)
doi:10.1016/j.jmb.2007.03.040
J. Mol. Biol. (2007) 369, 265–276
Crystal Structure of Vestitone Reductase from Alfalfa (Medicago sativa L.) Hui Shao, Richard A. Dixon and Xiaoqiang Wang⁎ Plant Biology Division, Samuel Roberts Noble Foundation, 2510 Sam Noble Parkway, Ardmore, OK 73401, USA
Isoflavonoids are commonly found in leguminous plants, where they play important roles in plant defense and have significant health benefits for animals and humans. Vestitone reductase catalyzes a stereospecific NADPH-dependent reduction of (3R)-vestitone in the biosynthesis of the antimicrobial isoflavonoid phytoalexin medicarpin. The crystal structure of alfalfa (Medicago sativa L.) vestitone reductase has been determined at 1.4 Å resolution. The structure contains a classic Rossmann fold domain in the N terminus and a small C-terminal domain. Sequence and structural analysis showed that vestitone reductase is a member of the short-chain dehydrogenase/reductase (SDR) superfamily despite the low levels of sequence identity, and the prominent structural differences from other SDR enzymes with known structures. The putative binding sites for the co-factor NADPH and the substrate (3R)-vestitone were defined and located in a large cleft formed between the N and C-terminal domains of enzyme. Potential key residues for enzyme activity were also identified, including the catalytic triad Ser129-Tyr164-Lys168. A molecular docking study showed that (3R)-vestitone, but not the (3S) isomer, forms favored interactions with the co-factor and catalytic triad, thus providing an explanation for the enzyme's strict substrate stereo-specificity. © 2007 Elsevier Ltd. All rights reserved.
*Corresponding author
Keywords: vestitone reductase; Medicago sativa; crystal structure; isoflavonoid; stereospecificity
Introduction Isoflavonoids are a diverse group of natural products that are found primarily in leguminous plants. The bioactivities of isoflavonoids impact plant, animal and human health. Simple isoflavones have been ascribed estrogenic, antiangiogenic, antioxidant, and anticancer activities,1–4 and the isoflavone genistein reduces susceptibility to mammary cancer in rats, inhibits prostate and stomach tumor growth in humans, and helps to prevent bone loss caused by estrogen deficiency in female mice.5–7
Abbreviations used: VR, vestitone reductase; rmsd, root-mean-square deviation; SDR, short-chain dehydrogenase/reductase; SOR, sorphorol reductase; IFR, isoflavone reductase; DFR, dihydroflavonol reductase; ANR, anthocyanidin reductase; 17β-HSD1, 17β-hydroxysteroid dehydrogenase type 1; MLCR, mouse lung carbonyl reductase. E-mail address of the corresponding author: [email protected]
Isoflavonoid phytoalexins are inducible antimicrobial compounds that have important roles in plant disease resistance.7,8 Pterocarpans are among the major chemical classes of isoflavonoid phytoalexins in legumes.9 Medicarpin from alfalfa (Medicago sativa),10 pisatin from garden pea (Pisa sativum),11 and maackiain from chickpea (Cicer arietinum) are among the most intensively studied pterocarpan phytoalexins.12 The stereochemistry of pterocarpans at the 6a and 11a positions (Figure 1) is important for their antifungal activity, since some fungal pathogens can break down the enantiomer of the pterocarpan present in their host plant, but not the opposite enantiomer.13 Alfalfa produces (−)-medicarpin, whereas pea and peanut (Arachis hypogaea) produce (+)-pisatin and (+)-medicarpin, respectively. Isoflavonoids originate from the amino acid L-phenyalanine, and their biosynthesis has been investigated extensively, with nearly all the enzymes now isolated and functionally characterized.14 Vestitone reductase (VR) is the central enzyme in a branch pathway for the conversion of non-enantiomeric isoflavones to enantiomeric pterocarpans, and
0022-2836/$ - see front matter © 2007 Elsevier Ltd. All rights reserved.
266
Vestitone Reductase Structure
Figure 1. The enzyme reaction catalyzed by M. sativa VR, and the structure of sophorol, the substrate for SOR from pea.
is involved specifically in the biosynthesis of medicarpin in alfalfa.15,16 Recombinant VR catalyzes the NADPH-dependent reduction of the 2-hydroxyisoflavanone (3R)-vestitone (the product of the isoflavone reductase reaction) to 7,2′-dihydroxy-4′methoxy-isoflavanol, which is further converted to medicarpin by 7,2′-dihydroxy-4′-methoxy-isoflavanol dehydratase (Figure 1). VR is highly stereochemically specific, and recognizes only the 3R conformation of vestitone, not (3S)-vestitone. It has been reported that VR and 7,2′-dihydroxy-4′methoxy-isoflavanol dehydratase form a weak association through protein–protein interactions.10 Recently, sorphorol reductase (SOR), a VR homolog, was identified for the biosynthesis of pisatin in pea.17 Similar to VR, SOR is a stereochemically specific enzyme; it recognizes only (−) sophorol, converting it to (−) 7,2′-dihydroxy-4′,5′-methylenedioxyisoflavanol, for the production of (+)-pisatin. There may be a large number of VR and SOR homologs for the biosynthesis of isoflavonoidderived pterocarpan phytoalexins in legumes, which are one of the largest families of crop plants with ∼ 20,000 species. However, nothing is known of the structural basis for catalysis by VR or SOR. In addition to VR, there are several other highly substrate-specific, NADPH-dependent reduction reactions in the formation of (iso)flavonoids in plants, including isoflavone reductase (IFR),18 and dihydroflavonol reductase (DFR) 19 and anthocyanidin reductase (ANR),20 two enzymes in the upstream branch leading to the biosynthesis of anthocyanins and condensed tannins. All are cytosolic enzymes with molecular mass of 35–37 kDa. VR is similar to DFR and ANR with an identity of ∼ 40%, but different from IFR, with no significant sequence identity (only ∼ 10%). Chalcone reductase, an enzyme that acts early in flavonoid metabolism,21 has no significant similarity to any of the other reductases in the pathway. So far, only structures of chalcone reductase and IFR have been reported,22,23 and the latter provides a structural basis for understanding the mechanism
for introduction of the stereochemistry at C3 of isoflavanones (Figure 1). However, since both chalcone reductase and IFR are unrelated to VR, it is necessary to solve the three-dimensional structure of VR, and related enzymes such as SOR that introduce the opposite stereochemistry, in order to understand how the stereochemistry at isoflavanone position C4 (position 11a of the pterocarpan) is introduced. These stereo-specific enzymes are potential targets for metabolic engineering to alter the stereochemistry of the host's pterocarpan phytoalexins in order to overcome their microbial degradation. Here, we present the crystal structure of VR from alfalfa at 1.4 Å resolution. The structure reveals distinct features different from other known NADPH-dependent enzyme structures. A comparative structural study further identifies the putative binding sites for the co-factor NADPH and the substrate (3R)-vestitone, and a catalytic triad (Ser129Tyr164-Lys168) with Tyr164 as a general base for enzymatic catalysis.
Results and Discussion Structure determination and overall structure The M. sativa VR was expressed in Escherichia coli and purified, and single crystals were obtained. The crystal structure of VR was determined using the multiwavelength anomalous dispersion method with a crystal of Se-Met-substituted enzyme, and refined at 1.4 Å resolution to an R-factor of 18.8% and Rfree of 20.3% with a native data set. Data collection, phasing and refinement statistics are presented in Table 1. The asymmetric unit contains one molecule of VR. This has overall dimensions of ∼ 60 Å × 47 Å × 51 Å , and a bi-domain structure consisting of two distinct domains, an N-terminal domain and a C-terminal domain (Figures 2 and 3). The large N-terminal domain is comprised of a Rossmann fold with an
267
Vestitone Reductase Structure Table 1. Data collection, phasing and refinement statistics
A. Data collection and phasing statistics Resolution(Å) Wavelength(Å ) Unique reflections Completeness(%) Rsym (%) I/σ(I) Figure of merit B. Refinement statistics R-factor (%) Rfree (%) Number of protein atoms Number of solvent atoms Average B-factors (Å2) rmsd from ideal values Bond lengths (Å) Bond angles (deg.)
Native
Se-peak
Se-inflection
Se-remote
1.4 0.9793 77,901 97.9(97.7) 5.1(39.2) 26.6(4.3)
2.0 0.9789 27,146 98.9(89.1) 11.8(42.6) 16.0(3.1) 0.47
2.0 0.9791 27,260 99.8(100) 12.9(38.4) 14.7(4.6)
2.0 0.9641 27,260 99.0(100) 10.7(36.6) 17.1(4.4)
18.8 20.3 2405 542 21.5 0.0044 1.28
Rsym = ∑hkl|I–|/∑I, where I is the observed intensity and is the average intensity from observations of symmetryrelated reflections. A subset of the data (10%) was excluded from the refinement and used to calculate the free R value (Rfree). R-factor = ∑||Fo|–|Fc||/∑|Fo|. Numbers in parentheses are for the highest resolution shell.
eight-stranded parallel β-sheet (β1–β7 and β10) flanked by nine α helices (α1–α7, α9, and α11) on both sides. The small C-terminal domain contains a
mixed parallel and antiparallel β-sheet (β8–β9 and β11–β13) and three α helices (α8, α10, and α12). These two domains form a cleft that would be the
Figure 2. Ribbon diagram of the VR structure. The α helices, β strands, and the N and C termini are labeled. Figures 2, 4, 6, and 7 were prepared with MOLSCRIPT41 and Raster3D.42
268
Vestitone Reductase Structure
Figure 3. Sequence alignment of VR from M. sativa, SOR from Pisum sativum, DFR and ANR from Medicago truncatula, E. coli UDP-galactose 4-epimerase, human 17β-hydroxysteroid dehydrogenase type 1 (17β-HSD1), and mouse lung carbonyl reductase. The secondary structure elements observed in the VR structure are shown above the alignment. Conserved residues are highlighted. This Figure was produced with ENDscript.43
binding sites for NADPH and (3R)-vestitone. Interactions observed between the N and C-terminal domains include hydrogen bonds formed between the OD2 atom of Asp232 and the OH atom of Tyr307, between the OD1 atom of Asp233 and the mainchain nitrogen atom of Lys306, and between the ND2 atom of Asn252 and the main-chain oxygen atom of
Asn294, and some weak interactions formed through water molecules. The two flexible regions between β2 and α2 (Pro40–Lys43) and between β4 and α4 (Asp90– Glu97) are disordered and invisible in the VR crystal structure. The first four residues of the N terminus are also disordered.
Vestitone Reductase Structure
Comparison with other NAD(P)H-dependent enzymes VR represents a large family of proteins, including VR homologs in legumes and other highly homologous reductases such as DFRs involved in flavonoid metabolism. A BLAST search against the Arabidopsis thaliana genome protein databases at The Institute for Genomic Research (TIGR) showed 20 hits that are ∼ 30–40% identical with the VR sequence. However, there was no highly homologous protein structure identified when searching against the RCSB Protein Data Bank (PDB). One of the similar proteins with known structures was E. coli UDP-galactose 4-epimerase (PDB 1XEL). Initial attempts to solve the structure of VR by molecular replacement using this epimerase structure as a search model failed, indicating that VR has some unique and special structural features. Amino acid sequence analysis, including a Pfam database search,24 showed that VR has features that place it in the epimerase/dehydratase family and the extended short-chain dehydrogenase/ reductase (SDR) superfamily (Figure 3).25,26 VR contains an N-terminal glycine-rich motif or fingerprint G12G13XG15XXG18, similar to the GXXXGXG se-
269 quence motifs involved in NAD(P)H binding of SDR enzymes, and a putative catalytic motif Y164XXXK168, a highly conserved active-site pattern for SDR enzymes. However, most members of the SDR superfamily contain about 250 amino acid residues. VR is larger, with 326 amino acid residues, similar to the epimerases, and may be classified as a member of the extended SDR superfamily. Structural comparison through the DALI search server 27 revealed similarities between VR and enzymes of the SDR or extended-SDR superfamily, such as E. coli UDP-galactose 4-epimerase,28 human 17β-hydroxysteroid dehydrogenase type 1 (17βHSD1), 29 and mouse lung carbonyl reductase (MLCR)30 (Figure 4(a) and (b)). These enzymes are similar in their N-terminal domains, but their C-terminal domains vary in both size and shape. Superimposing the structure of VR onto that of UDP-galactose 4-epimerase (PDB 1XEL) revealed structural similarity with root-mean-square deviation (rmsd) of 2.0 Å for 223 Cα atoms (using a 4.0 Å rmsd cutoff) and 16% sequence identity. Comparison to the regular SDR enzymes 17β-HSD1 (PDB 1FDT) and MLCR (PDB 1CYD), which are smaller and contain mainly one compact domain, also showed similarity, with rmsd values of 1.9 Å (157
Figure 4. A stereo diagram showing the superimposition of the structures of VR (cyan), epimerase (magenta in (a)); and MLCR (magenta in (b)). Glycine residues in the fingerprints are labeled, and catalytic triads are shown as ball- and -stick models. (a) Ligands NADH and UDP-glucose in the structure of epimerase, and (b) NADPH and 2-propanol in the structure of MLCR are shown in yellow as ball- and -stick models.
270 Cα atoms) and 2.0 Å (154 Cα atoms), respectively. Both 17β-HSD1 and MLCR superimpose mainly on the large N-terminal domain of VR. The structural conservation suggests that VR would interact with its co-factor NADPH and substrate vestitone in a mode similar to that observed for these and other SDR enzymes, although the sequence identity between VR and these SDR enzymes is very low (< 20%). Although the overall folding of VR is similar to that of other SDR enzymes, only about 68% of the Cα atoms of VR may be superimposed onto the epimerase structure, with a 4.0 Å rmsd cutoff, and about 48% onto 17β-HSD1 and MLCR, which have no distinct C-terminal domains. Compared to the VR C-terminal domain consisting of five β strands (β8–β9 and β11–β13) and three α helices (α8, α10, and α12) with ∼ 100 residues, there are only ∼ 30 residues forming mainly two α helices in MLCR structure (in the corresponding locations of α8 and β9 in the VR structure), and ∼45 residues forming three α helices and a long loop in the 17β-HSD1 structure (in the corresponding locations of α8, β9, and α10, and loop β8–α8 in the VR structure). The major differences between VR and epimerase are also in their C-terminal regions, which have quite different conformations and lengths around the regions of three loops β8–α8, α8–β9, and α10–β12,
Vestitone Reductase Structure
and the first two loops in VR are short due to some amino acid deletions (Figure 3). The major difference in their N-terminal domains is the region around helix α6 and loop α6–α7; VR has an extra helix α6 with an eight amino acid insertion. Most of these regions are around the cleft proposed as the putative binding site for co-factor and substrate. The fingerprint and putative binding pocket for the co-factor A large cleft was observed on the surface of the VR molecule, starting from the fingerprint G12G13XG 15 XXG18 region and extending to near the catalytic motif Y164XXXK168 region (Figure 5). The location of this highly conserved fingerprint region in the VR structure is between strand β1 and helix α1, similar to that of the fingerprint regions observed in the structures of epimerase bound with NADH and UDP-glucose, 17β-HSD1 with NADP+ and equilin, and MLCR with NADPH and 2propanol (Figure 4(a) and (b)). The fingerprint is a common NAD(P)H recognition motif for the SDR enzymes and other NADPHdependent enzymes. The portion of the large cleft near the fingerprint would therefore be the putative NADPH-binding pocket for VR. Comparison by superimposing structures showed that the putative
Figure 5. Molecular surface of VR. The electrostatic surface of the VR enzyme was calculated with PyMOL [http:// www.pymol.sourceforge.net/].
Vestitone Reductase Structure
binding pocket of VR is also similar to that of epimerase and MLCR (Figure 4(a) and (b)). NADPH would probably bind to VR in a conserved mode, as observed in the structures of epimerase, MLCR, and other members of the NAD(P)H-dependent SDR enzyme family. The NADPH pyrophosphate group may fit between loop β1–α1 (i.e. the fingerprint) and loop β4–α4 (Thr85–Pro88), and may form hydrogen bonds with some main chain atoms of these two loops. As revealed in the structures of 17β-HSD1 and MLCR, which use NADP+ or NADPH as a co-factor, a basic arginine residue, Arg37 in 17β-HSD1 and Arg39 in MLCR, forms hydrophobic contact with the adenine ring and/or electrostatic interaction with the 2′-ribose monophosphate group of NADPH. In the structure of VR, Arg37 is located in the equivalent position and may interact with the adenine and/or 2′-phosphate group of NADPH. There are several other basic residues, including Arg42 and Lys43 (in the disordered region), and Arg44 located close to this area. These basic residues are conserved in VR homologs. The co-factor-binding pocket of VR is more similar to those of 17β-HSD1 and MLCR, especially the segment interacting with the adenine base and the 2′-monophosphate group of NADPH, than that of epimerase, although the overall folding of VR is more similar to that of epimerase. This may be because VR, 17β-HSD1 and MLCR are NADPHpreferring enzymes, whereas epimerase uses NAD+ as a cofactor. Epimerase has an asparagine residue, Asn32, present in the position corresponding to Arg37 in VR, whereas in most of the NAD(H)preferring SDR enzymes, an aspartate residue in this position interacts with the adenine ribose of NAD(H). Catalytic triad and isoflavone substrate binding–implications for catalytic mechanism A Ser-Tyr-Lys triad was identified and proposed for the catalytic mechanisms of the SDR enzymes.26 For example, epimerase has the triad Ser124-Tyr149Lys153, 17β-HSD1 has Ser142-Tyr155-Lys159 , and MLCR has Ser136-Tyr149-Lys153.30 In the VR structure, the conserved Y164XXXK168 motif is located near one end of the large cleft (Figures 4(a) and (b), and 5). There are several serine residues present in the cleft near Tyr164, including Ser 87, Ser128, Ser129, Ser131, and Ser134. Structural comparison showed that Ser129 is located in the position corresponding to the serine of the catalytic triads of SDR enzymes (Figure 4(a) and (b)). This Ser129, together with Tyr164 and Lys 168, may act as the catalytic triad for VR. These catalytic triad residues are conserved in all VR homologs, and may be essential for enzyme catalysis and function. The pocket near the catalytic triad is presumably the substrate-binding site. The pocket is formed by loop β5–β6 (∼residue Ala132), loop α6–α7 (∼Phe160), loop β7–β8 (∼Pro192), helix α8 (∼Glu208), strand β9 (∼Arg225), and strand β12
271 (∼Leu290). The location of this pocket in the VR structure is similar to those of the substrate-binding pockets observed in the structures of UDP-galactose 4-epimerase and MLCR (Figure 4(a) and (b)). In the MLCR structure (PDB 1CYD),30 the substrate-binding pocket is very small and restricted by hydrophobic residues (i.e. Phe143, Leu146, Met137, Val138, Val181, Val190) and the catalytic triad, and its small substrates such as pyridine 3-aldehyde or acetone may be completely buried inside the enzyme molecule. In the epimerase structure (PDB 1XEL),28 the substrate-binding pocket is relatively long and extended to fit the long substrate UDPgalactose or UDP-glucose, and several charged residues, Asp295, Arg231, and Arg292, are located inside the pocket and interact with ribose hydroxyl and phosphate group of substrate. In contrast, the putative VR substrate-binding pocket is quite different in size, shape and overall features. The pocket in the VR structure is relatively hydrophobic, and is surrounded by the aromatic residues Phe193, Phe226, Phe135 and Tyr164, several other hydrophobic residues (e.g. Leu293, Val223, and Leu290), and a basic residue, His227, located in the bottom of the pocket. A molecular docking study of VR with co-factor and substrate using the program GOLD was carried out by defining the active site with the catalytic residue Tyr164 (Figure 6). In the model with the keto group at position 4 of vestitone docked close to the OH of Tyr164 (∼ 2.8 Å), the substrate (3R)-vestitone may stack its aromatic A and C rings with the nicotinamide ring of NADPH, its keto oxygen may also interact directly with the OG of the catalytic triad residue Ser129 (∼ 2.5 Å), and its B-ring may point to the molecular surface. The third catalytic triad residue, Lys168, has no interaction with substrate, and its NZ atom is ∼ 2.9 Å away from the 2′-hydroxyl group on the ribose moiety of NADPH. The keto group of (3S)-vestitone cannot be docked into a position for interacting with the catalytic residue Tyr164 when fitting it into the pocket. Its B-ring would be repulsed by His227 and other residues in the bottom of the pocket, and its A and C rings would not form a favored stacking interaction with the NADPH nicotinamide ring. This is consistent with previously reported enzyme assay results, that VR utilizes only (3R)-vestitone as substrate and shows no activity with (3S)-vestitone.15 The structural comparison and molecular docking study described above suggested that the interactions between VR enzyme, cofactor and substrate would be similar to those observed in other SDR enzyme structures, and VR may have an enzymatic mechanism similar to that proposed for SDR enzymes.26 The substrate (3R)-vestitone may interact with NADPH by stacking its aromatic ring with the nicotinamide ring of NADPH, which would attack and transfer its hydrogen atom onto the C4 atom of vestitone. The catalytic base residue Tyr164 would attack the oxygen atom at position 4 of vestitone from the opposite side, and Ser129 may form a hydrogen bond with this oxygen atom.
272
Vestitone Reductase Structure
Figure 6. A stereo diagram showing the putative ligand-binding pocket in the structure of VR. Some amino acid residues in the binding pockets are labeled and shown as bond models in cyan. The NADPH and (3R)-vestitone docked into the putative binding pocket of VR are shown as ball- and -stick models.
Lys168 may interact with the ribose of NADPH and form hydrogen bonds to its 2′ and/or 3′-hydroxyl groups, to stabilize the position of the nicotinamide moiety. The high substrate stereochemical specificity of VR is determined by the overall shape and features of the binding pocket. His227 located in the pocket might be a critical residue, and two other charged residues (i.e. Asp205 and Lys209) on the molecular surface and several hydrophobic residues (e.g. Phe193, Leu290, and Phe135) might have important roles for interacting substrate. The mobile region in VR and its relation to function Both of the two disordered regions in the VR structure are close to the putative NADPH binding site, and the disordered region between residues Asp90 and Glu97 is close to the substrate-binding site. The average temperature factors of the segments near the two disordered regions are about 40 Å2 (40.3 Å2 for residues Arg37–Asp45, and 38.2 Å2 for residues Ser87–Val100), much higher than the overall value of 21.5 Å2. The disorder and high-temperature factors of these regions indicate their flexibility, which is probably important for VR function. It has been reported that “open to closed” or “disordered to ordered” conformational changes occur upon binding of ligands to many SDR enzymes.26 Structural comparison between 17β-HSD1 apoenzyme and its complexes with cofactor and/or inhibitors showed open to closed conformational changes of the substrate-binding regions of the enzyme.29 Similarly, these regions in the VR structure may undergo a conformation change once cofactor or/and substrate bind to the enzyme. Although the co-factor NADPH or substrate vestitone were used for co-crystallization and soaking with native crystals, and several data sets were collected and analyzed, no electron density was observed for these ligands. As observed in the
crystal structure, the two disordered regions (Pro40– Lys43, and Asp90–Glu97) are close to the ligandbinding sites, indicating that ligand binding would cause the conformation changes and disrupt the current crystal lattice and packing formed in this specific crystallization condition, which is consistent with the observation of crystal damage during soaking experiments. To obtain crystals of the enzyme in complexes with ligands, it would be necessary to screen other crystallization conditions leading to different crystal forms. Another strategy is to mutate the catalytic residue Tyr164 to alanine or glycine for abolishing enzymatic activity, and using the inactive mutant for co-crystallization with both co-factor and substrate. These mutagenesis and crystallization trials are in progress. Comparison with other VR-homolog proteins-implications for substrate specificity SOR is a VR homolog identified in pea.17 SOR recognizes (–) sophorol, which has the same stereochemistry as (3R)-vestitone, but differs in the presence of the methylenedioxy ring between positions 4′ and 5′ of the B ring (Figure 1). The level of sequence identity between VR and SOR is 86%. A three-dimensional structural model for SOR was built using the program MODELLER with the VR structure as a template. The overall folding of SOR is similar to that of VR. Their putative substrate-binding pockets are very similar, with most of the residues conserved (Figure 7). There are only minor differences observed in the putative substrate-binding pockets. In the modeled structure of SOR, an alanine residue, Ala291, is present in the position corresponding to Pro291 in VR. The smaller residue in SOR may provide more space to accommodate the larger substrate. This suggests that SOR may possibly recognize a smaller molecule, such as (3R)-vestitone, as a substrate. A docking study showed that both (–) sophorol and
273
Vestitone Reductase Structure
(3R)-vestitone were docked into the active site of SOR structural model in the same orientation with similar fitness scores, since these two compounds are very similar in structure (Figure 1). However, the pterocarpans accumulating in pea contain the methylenedioxy substituent, presumably because the pathway architecture results in substitution occurring before reduction. Another minor difference is that a serine residue in SOR is present at the position corresponding to Ala165 of VR. The hydroxyl group of Ser165 in SOR may interact with the oxygen atom at position 5′ of sophorol directly or through a water molecule. A BLAST search showed that there is a protein in soybean (Glycine max) with 75% sequence identity to VR. It is annotated as a 2′-hydroxydihydrodaidzein reductase, but has not been functionally characterized. Sequence comparison showed that most of the residues present in the VR substrate-binding pocket are highly conserved in this protein, except for Ala130, Ser162 and Arg210, which correspond to Gly130, Gly161, and Lys209 in VR, respectively. This suggests that the soybean homolog may be involved in the biosynthesis of glyceollin, the peterocarpan phytoalexin in soybean. IFR precedes VR in the biosynthetic pathway of medicarpin in alfalfa and pisatin in pea. It is also an NADPH-dependent reductase, which reduces the 2,3 double bond of isoflavone to produce an asymmetric compound that is the substrate for VR or SOR. Although the level of sequence identity between VR and IFR is very low (∼ 10%), their threedimensional structures are very similar. Superimposition of the structures of VR and IFR gave an rmsd of 2.9 Å for 213 Cα atoms. IFR also contains a similar fingerprint region, but no YXXXK motif or catalytic triad. A single lysine residue, Lys144, was identified as the catalytic residue for IFR catalysis.23 The location of Lys144 in IFR is similar to that of Lys168 in VR. The lysine residue at this location in SDR enzymes was proposed to stabilize the nicotinamide moiety of NAD(P)H, and would not be involved directly in catalysis, including substrate binding or hydride transfer. However, in the cases of IFR and pinoresinol-lariciresinol reductase and phe-
nylcoumaran benzylic ether reductase, two IFR-like enzymes involved in lignan biosynthesis,31 there are no catalytic triads, and the lysine residue present in the substrate-binding pocket is the only residue that can act as the general base for catalysis. VR, IFR and other SDR enzymes likely share a common evolutionary origin from a progenitor enzyme with the same basic three-dimensional structure. Subsequent changes of catalytic residues led to new functions that helped plants adapt to their environment. DFR and ANR are two other NADPH-dependent reductases in the phenylpropanoid pathway, and have about 40% identity with VR. Sequence analysis showed that they are also members of the SDR enzyme family, containing similar fingerprint motifs and catalytic triads (Ser128-Tyr163-Lys167 in DFR, and Ser133-Tyr170-Lys174 in ANR, respectively) (Figure 3). This suggests that they may have catalytic mechanisms similar to that of VR rather than to IFR. DFR and ANR recognize flavonoids as substrates, and VR and IFR use isoflavonoids. The residues located in the substrate-binding pockets for substrate recognition and binding are very different in VR and DFR/ANR. The three charged residues in VR, Asp205, Lys209, and His227, are substituted by Pro204, Thr208, and Gln227 in DFR and by Ser211, Met215, and Gln234 in ANR. For the two positions that distinguish the catalytic pockets of VR and SOR (Ala165 and Pro291 in VR, Ser165 and Ala29 in SOR), Phe164 and Val290 are present in DFR, and Pro171 and Ile304 in ANR. There is an insertion of eight residues in the ANR sequence near the corresponding position of His227 in VR. All these differences may contribute to the different substrate specificities of these enzymes.
Materials and Methods Cloning, protein expression and purification The vestitone reductase cDNA from alfalfa (M. sativa L.) was cloned into the E. coli expression vector pHIS8, a derivative of pET28a kindly provided by Dr J. Noel. E. coli BL21(DE3) cells transformed with plasmid encoding the
Figure 7. A stereo diagram showing the superimposition of putative substrate-binding pockets in the structure of VR, and the modeled structure of SOR from pea. Some amino acid residues in the binding pockets of VR and SOR are labeled and shown as bond models in cyan and magenta, respectively.
274 VR construct were grown at 37 °C in LB medium containing 50 μg/ml of kanamycin until A600 nm = 0.6– 0.8. After induction with 0.5 mM isopropyl 1-thio-βgalactopyranoside (IPTG), the cultures were grown overnight at 16 °C. Cells were pelleted and resuspended in lysis buffer (50 mM Tris–HCl (pH 8.0), 500 mM NaCl, 10 mM imidazole, 20 mM β-mercaptoethanol). After lysis with a French press and centrifugation at 20,000g (SORVALL SA-600) at 4 °C for 20 min, Ni2+-NTA agarose was added to the supernatant containing the target protein. After incubation for 40–60 min, the mixture was transferred into a disposable column and washed extensively with lysis buffer (about 50 column volumes). Histagged proteins were washed from the column with elution buffer (50 mM Tris–HCl (pH 8.0), 500 mM NaCl, 250 mM imidazole, 20 mM β-mercaptoethanol). Incubation with biotinylated thrombin during dialysis overnight at 4 °C against 20 mM Tris–HCl (pH 7.5), 100 mM NaCl, 10 mM β-mercaptoethanol cleaved the N-terminal Histag. Dialyzed proteins were incubated with streptavidin agarose and Ni2+-NTA agarose to remove thrombin and the cleaved His tag, respectively. The proteins were further purified on Resource Q anion-exchange and Superdex-200 gel-filtration columns (GE Healthcare) and concentrated to 4–10 mg/ml in 10 mM Tris–HCl (pH 7.5), 10 mM NaCl, 5 mM β-mercaptoethanol. Selenomethionine-substituted VR was prepared by growing E. coli cells in M9 minimal medium supplemented with L-selenomethionine (Sigma) and purified by following the same protocol used for the native protein.
Vestitone Reductase Structure tively. A bulk solvent correction was applied. B factors were refined individually. Water molecules were added with Arp/wArp37 and checked manually for inclusion. The program PROCHECK 38 was used to check the model. All backbone ϕ-ψ torsion angles are within the allowed regions of the Ramachandran plot. Molecular modeling and docking The comparative modeling program MODELLER39 was used to generate a model for the sophorol reductase from pea. The structure of VR was used as a template in the structural modeling experiment. The three-dimensional model was obtained by optimally satisfying spatial restraints derived from the sequence alignment based on the CLUSTALX results and three-dimensional structure. The automated docking program GOLD40 was used to dock co-factor NADPH and substrate (3R)-vestitone into the VR enzyme active site. Default genetic algorithm parameters for controlling the operation of the docking process were used. The docking calculations were restricted to the putative binding pocket by defining the active site with residue Tyr164. GOLDscore was used to identify the lowest energy docking results. The hydrogen bonds and van der Waals contacts between ligands and enzyme were analyzed to identify the optimal binding mode. Minor manual adjustments of the GOLD solution were made using the program O. Protein Data Bank accession code
Crystallization and data collection Crystals of VR were grown using the hanging-drop, vapor-diffusion method. A sample(2 μl) of VR protein solution (∼ 4 mg/ml) was mixed with 2 μl of reservoir solution (0.1 M Tris–HCl (pH 8.0), 20% (w/v) polyethylene glycol 6000, 0.1 M MgCl2,). The mixture was equilibrated over the reservoir solution at 4 °C. Crystals grew over two weeks to dimensions of ∼ 0.3 mm × 0.2 mm × 0.1 mm. Before data collection, crystals were transferred to a solution of the same composition but supplemented with 20% (v/v) glycerol as a cryoprotectant, and flash-frozen in liquid nitrogen. Data from a crystal of VR were measured to 1.4 Å resolution with an ADSC Quantum 315 CCD detector at the SBC 19ID beamline at the Advanced Photon Source (APS), Argonne National Laboratory (Argonne, IL). The crystal belongs to space group P21212 (a = 69.0 Å, b = 92.3 Å, c = 63.3 Å). There is one molecule per crystallographic asymmetric unit with 50% (v/v) solvent content and a VM of 2.7 Å3/Da. Multiwavelength anomalous X-ray diffraction data at 2.0 Å resolution were collected from a Se-Met-labeled VR crystal, using three wavelengths, at the SBC 19BM beamline at APS. All data sets were processed using the program suite HKL2000.32
The atomic coordinates and structure factors for the structure of VR have been deposited with the RCSB Protein Data Bank (PDB) with the PDB ID 2P4H.
Acknowledgements We thank Drs X. He and S. Liu for initial cloning and protein purification, R. Guan for helpful advice in crystallography, M. Udvardi and L. Modolo for critical reading of the manuscript, and F. Rotella and Y. Kim at the Structural Biology Center beamlines 19BM and 19ID at the Advanced Photon Source, Argonne National Laboratory (Argonne, IL), for assistance with data collection. Argonne is operated by UChicago Argonne, LLC, for the U.S. Department of Energy, Office of Biological and Environmental Research under contract DE-AC02-06CH11357. This work was supported by the Samuel Roberts Noble Foundation.
Structure determination and refinement The structure of VR was determined using the multiwavelength anomalous dispersion method, and the data (40–2.0 Å) were analyzed with SOLVE.33 Three out of four selenium sites were located, yielding an overall figure of merit of 0.47. The program RESOLVE was used for electron density modification and automated modelbuilding.34 Interactive model building and crystallographic refinement were carried out with a native data set at 1.4 Å using the programs O35 and CNS,36 respec-
References 1. Dixon, R. A. (1999). Isoflavonoids: biochemistry, molecular biology abd biological functions. In Comprehensive Natural Products Chemistry (Sankawa, U., ed.), pp. 773–823, Elsevier, Oxford. 2. Dixon, R. A. & Sumner, L. W. (2003). Legume natural products: understanding and manipulating complex pathways for human and animal health. Plant Physiol. 131, 878–885.
275
Vestitone Reductase Structure 3. Dixon, R. A. & Ferreira, D. (2002). Molecules of interest: genistein. Phytochemistry, 60, 205–211. 4. Palevitz, B. A. (2000). Soybeans hit main street. Scientist, 14, 8–9. 5. Ishimi, Y., Miyaura, C., Ohmura, M., Onoe, Y., Sato, T., Uchiyama, Y. et al. (1999). Selective effects of genistein, a soybean isoflavone, on B-lymphopoiesis and bone loss caused by estrogen deficiency. Endocrinology, 140, 1893–1900. 6. Fritz, W. A., Coward, L., Wang, J. & Lamartiniere, C. A. (1998). Dietary genistein: perinatal mammary cancer prevention, bioavailability and toxicity testing in the rat. Carcinogenesis, 19, 2151–2158. 7. Dixon, R. A. & Steele, C. L. (1999). Flavonoids and isoflavonoids- a gold mine for metabolic engineering. Trends Plant Sci. 4, 394–400. 8. Dixon, R. A., Dey, P. M. & Lamb, C. J. (1983). Phytoalexins: enzymology and molecular biology. Advan. Enzymol. Relat. Areas Mol. Biol. 55, 1–136. 9. Dixon, R. A., Achnine, L., Kota, P., Liu, C.-J., Reddy, M. S. S. & Wang, L. (2002). The phenylpropanoid pathway and plant defence-a genomics perspective. Mole. Plant Pathol. 3, 371–390. 10. Guo, L., Dixon, R. A. & Paiva, N. L. (1994). The ‘pterocarpan synthase’ of alfalfa: association and coinduction of vestitone reductase and 7,2′-dihydroxy4′-methoxy-isoflavanol (DMI) dehydratase, the two final enzymes in medicarpin biosynthesis. FEBS Letters, 356, 221–225. 11. Cruickshank, I. A. M. & Perrin, D. R. (1960). Isolation of a phytoalexin from Pisum sativum L. Nature, 187, 799–800. 12. Daniel, S., Tiemann, K., Wittkampf, U., Bless, W., Hinderer, W. & Barz, W. (1990). Elicitor-induced metabolic changes in cell cultures of chickpea (Cicer arietinum L.) cultivar resistant and susceptible to Ascochyta rabiei. Planta, 182, 270–278. 13. VanEtten, H. D., Matthews, D. E. & Matthews, P. S. (1989). Phytoalexin detoxification: importance for pathogenicity and practical implications. Ann. Rev. Phytopathol. 27, 143–164. 14. Dixon, R. A., Harrison, M. J. & Paiva, N. L. (1995). The isoflavonoid phytoalexin pathway: from enzymes to genes to transcription factors. Physiol. Plant, 93, 385–392. 15. Guo, L., Dixon, R. & Paiva, N. L. (1994). Conversion of vestitone to medicarpin in alfalfa (Medicago sativa L.) is catalyzed by two indenpendent enzymes. J. Biol. Chem. 269, 22372–22378. 16. Guo, L. & Paiva, N. L. (1995). Molecular cloning and expression of Alfalfa (Medicago sativa L.) vestitone reductase, the penultimate enzyme in medicarpin biosynthesis. Arch. Biochem. Biophys. 320, 353–360. 17. DiCenzo, G. L. & VanEtten, H. D. (2006). Studies on the late steps of (+) pisatin biosynthesis: evidence for (−) enantiomeric intermediates. Phytochemistry, 67, 675–683. 18. Paiva, N. L., Edwards, R., Sun, Y., Hrazdina, G. & Dixon, R. A. (1991). Stress responses in alfalfa (Medicago sativa L.). 11. Molecular cloning and expression of alfalfa isoflavone reductase, a key enzyme of isoflavonoid phytoalexin biosynthesis. Plant Mol. Biol. 17, 653–667. 19. Beld, M., Martin, C., Huits, H., Stuitje, A. R. & Gerats, A. G. (1989). Flavonoid synthesis in Petunia hybrida: partial characterization of dihydroflavonol-4-reductase genes. Plant Mol. Biol. 13, 491–502. 20. Xie, D., Sharma, S. B., Paiva, N. L., Ferreira, D. & Dixon, R. A. (2003). BANYULS encodes anthocyani-
21.
22.
23.
24.
25.
26. 27. 28.
29.
30.
31.
32. 33. 34. 35.
36.
37.
din reductase active in plant flavonoid biosynthesis. Science, 299, 396–399. Welle, R. & Grisebach, H. (1989). Phytoalexin synthesis in soybean cells: elicitor induction of reductase involved in biosynthesis of 6′-deoxychalcone. Arch. Biochem. Biophys. 272, 97–105. Bomati, E. K., Austin, M. B., Bowman, M. E., Dixon, R. A. & Noel, J. P. (2005). Structural elucidation of chalcone reductase and implications for deoxychalcone biosynthesis. J. Biol. Chem. 280, 30496–30503. Wang, X., He, X., Lin, J., Shao, H., Chang, Z. & Dixon, R. A. (2006). Crystal structure of isoflavone reductase from alfalfa (Medicago sativa L.). J. Mol. Biol. 358, 1341–1352. Sonnhammer, E. L. L., Eddy, S. R. & Durbin, R. (1997). Pfam: a comprehensive database of protein domain families based on seed alignments. Proteins: Struct. Funct. Genet. 28, 405–420. Jörnvall, H., Persson, B., Krook, M., Atrian, S., Gonzalez-Duarte, R., Jeffery, J. & Ghosh, D. (1995). Short-chain dehydrogenases/reductases (SDR). Biochemistry, 34, 6003–6013. Tanaka, N., Nonaka, T., Nakamura, K. T. & Hara, A. (2001). SDR: structure, mechanism of action, and substrate recognition. Curr. Org. Chem. 5, 89–111. Holm, L. & Sander, C. (1993). Protein structure comparison by alignment of distance matrices. J. Mol. Biol. 233, 123–138. Thoden, J. B., Frey, P. A. & Holden, H. M. (1996). Molecular structure of the NADH/UDP-glucose abortive complex of UDP-galactose 4-epimerase from Escherichia coli: implications for the catalytic mechanism. Biochemistry, 35, 5137–5144. Sawicki, M. W., Erman, M., Puranen, T., Vihko, P. & Ghosh, D. (1999). Structure of the ternary complex of human 17beta-hydroxysteroid dehydrogenase type 1 with 3-hydroxyestra-1,3,5,7-tetraen-17-one (equilin) and NADP+. Proc. Natl Acad. Sci. USA, 96, 840–845. Tanaka, N., Nonaka, T., Nakanishi, M., Deyashiki, Y., Hara, A. & Mitsui, Y. (1996). Crystal structure of the ternary complex of mouse lung carbonyl reductase at 1.8 A resolution: the structural origin of coenzyme specificity in the short-chain dehydrogenase/reductase family. Structure, 4, 33–45. Min, T., Kasahara, H., Bedgar, D. L., Youn, B., Lawrence, P. K., Gang, D. R. et al. (2003). Crystal structures of pinoresinol-lariciresinol and phenylcoumaran benzylic ether reductases and their relationship to isoflavone reductases. J. Biol. Chem. 278, 50714–50723. Otwinowski, Z. & Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326. Terwilliger, T. C. & Berendzen, J. (1999). Automated MAD and MIR structure solution. Acta Crystallog. sect. D, 55, 849–861. Terwilliger, T. C. (2000). Maximum likelihood density modification. Acta Crystallog. sect. D, 56, 965–972. Jones, T. A., Zou, J.-Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for the building of protein models in electron density maps and the location of errors in these models. Acta Crystallog. sect. A, 47, 110–119. Brünger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). Crystallography and NMR system: a new software suite for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905–921. Lamzin, V. S., Perrakis, A. & Wilson, K. S. (2001). The
276
Vestitone Reductase Structure
ARP/WARP suite for automated construction and refinement of protein models. In International Tables for Crystallography–Crystallography of Biological Macromolecules (Rossmann, M. G. & Arnold, E., eds), vol. F, pp. 720–722. Kluwer Academic Publishers, Dordrecht. 38. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. (1993). PROCHECK—a program to check the stereochemical quality of protein structures. J. Appl. Crystallogr. 26, 283–291. 39. Marti-Renom, M. A., Stuart, A., Fiser, A., Sánchez, R., Melo, F. & Sali, A. (2000). Comparative protein structure modeling of genes and genomes. Annu. Rev. Biophys. Biomol. Struct. 29, 291–325.
40. Jones, G., Willett, P., Glen, R. C., Leach, A. R. & Taylor, R. (1997). Development and validation of a genetic algorithm for flexible docking. J. Mol. Biol. 267, 727–748. 41. Kraulis, P. J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallog. 24, 946–950. 42. Merritt, E. A. & Bacon, D. J. (1997). Raster3D: photorealistic molecular graphics. Methods Enzymol. 277, 505–524. 43. Gouet, P. & Courcelle, E. (2002). ENDscript : a workflow with web interface to display sequence and structure information. Bioinformatics, 18, 767–768.
Edited by M. Guss (Received 7 February 2007; received in revised form 12 March 2007; accepted 14 March 2007) Available online 21 March 2007
J. Mol. Biol. (2007) 369, 265–276
Crystal Structure of Vestitone Reductase from Alfalfa (Medicago sativa L.) Hui Shao, Richard A. Dixon and Xiaoqiang Wang⁎ Plant Biology Division, Samuel Roberts Noble Foundation, 2510 Sam Noble Parkway, Ardmore, OK 73401, USA
Isoflavonoids are commonly found in leguminous plants, where they play important roles in plant defense and have significant health benefits for animals and humans. Vestitone reductase catalyzes a stereospecific NADPH-dependent reduction of (3R)-vestitone in the biosynthesis of the antimicrobial isoflavonoid phytoalexin medicarpin. The crystal structure of alfalfa (Medicago sativa L.) vestitone reductase has been determined at 1.4 Å resolution. The structure contains a classic Rossmann fold domain in the N terminus and a small C-terminal domain. Sequence and structural analysis showed that vestitone reductase is a member of the short-chain dehydrogenase/reductase (SDR) superfamily despite the low levels of sequence identity, and the prominent structural differences from other SDR enzymes with known structures. The putative binding sites for the co-factor NADPH and the substrate (3R)-vestitone were defined and located in a large cleft formed between the N and C-terminal domains of enzyme. Potential key residues for enzyme activity were also identified, including the catalytic triad Ser129-Tyr164-Lys168. A molecular docking study showed that (3R)-vestitone, but not the (3S) isomer, forms favored interactions with the co-factor and catalytic triad, thus providing an explanation for the enzyme's strict substrate stereo-specificity. © 2007 Elsevier Ltd. All rights reserved.
*Corresponding author
Keywords: vestitone reductase; Medicago sativa; crystal structure; isoflavonoid; stereospecificity
Introduction Isoflavonoids are a diverse group of natural products that are found primarily in leguminous plants. The bioactivities of isoflavonoids impact plant, animal and human health. Simple isoflavones have been ascribed estrogenic, antiangiogenic, antioxidant, and anticancer activities,1–4 and the isoflavone genistein reduces susceptibility to mammary cancer in rats, inhibits prostate and stomach tumor growth in humans, and helps to prevent bone loss caused by estrogen deficiency in female mice.5–7
Abbreviations used: VR, vestitone reductase; rmsd, root-mean-square deviation; SDR, short-chain dehydrogenase/reductase; SOR, sorphorol reductase; IFR, isoflavone reductase; DFR, dihydroflavonol reductase; ANR, anthocyanidin reductase; 17β-HSD1, 17β-hydroxysteroid dehydrogenase type 1; MLCR, mouse lung carbonyl reductase. E-mail address of the corresponding author: [email protected]
Isoflavonoid phytoalexins are inducible antimicrobial compounds that have important roles in plant disease resistance.7,8 Pterocarpans are among the major chemical classes of isoflavonoid phytoalexins in legumes.9 Medicarpin from alfalfa (Medicago sativa),10 pisatin from garden pea (Pisa sativum),11 and maackiain from chickpea (Cicer arietinum) are among the most intensively studied pterocarpan phytoalexins.12 The stereochemistry of pterocarpans at the 6a and 11a positions (Figure 1) is important for their antifungal activity, since some fungal pathogens can break down the enantiomer of the pterocarpan present in their host plant, but not the opposite enantiomer.13 Alfalfa produces (−)-medicarpin, whereas pea and peanut (Arachis hypogaea) produce (+)-pisatin and (+)-medicarpin, respectively. Isoflavonoids originate from the amino acid L-phenyalanine, and their biosynthesis has been investigated extensively, with nearly all the enzymes now isolated and functionally characterized.14 Vestitone reductase (VR) is the central enzyme in a branch pathway for the conversion of non-enantiomeric isoflavones to enantiomeric pterocarpans, and
0022-2836/$ - see front matter © 2007 Elsevier Ltd. All rights reserved.
266
Vestitone Reductase Structure
Figure 1. The enzyme reaction catalyzed by M. sativa VR, and the structure of sophorol, the substrate for SOR from pea.
is involved specifically in the biosynthesis of medicarpin in alfalfa.15,16 Recombinant VR catalyzes the NADPH-dependent reduction of the 2-hydroxyisoflavanone (3R)-vestitone (the product of the isoflavone reductase reaction) to 7,2′-dihydroxy-4′methoxy-isoflavanol, which is further converted to medicarpin by 7,2′-dihydroxy-4′-methoxy-isoflavanol dehydratase (Figure 1). VR is highly stereochemically specific, and recognizes only the 3R conformation of vestitone, not (3S)-vestitone. It has been reported that VR and 7,2′-dihydroxy-4′methoxy-isoflavanol dehydratase form a weak association through protein–protein interactions.10 Recently, sorphorol reductase (SOR), a VR homolog, was identified for the biosynthesis of pisatin in pea.17 Similar to VR, SOR is a stereochemically specific enzyme; it recognizes only (−) sophorol, converting it to (−) 7,2′-dihydroxy-4′,5′-methylenedioxyisoflavanol, for the production of (+)-pisatin. There may be a large number of VR and SOR homologs for the biosynthesis of isoflavonoidderived pterocarpan phytoalexins in legumes, which are one of the largest families of crop plants with ∼ 20,000 species. However, nothing is known of the structural basis for catalysis by VR or SOR. In addition to VR, there are several other highly substrate-specific, NADPH-dependent reduction reactions in the formation of (iso)flavonoids in plants, including isoflavone reductase (IFR),18 and dihydroflavonol reductase (DFR) 19 and anthocyanidin reductase (ANR),20 two enzymes in the upstream branch leading to the biosynthesis of anthocyanins and condensed tannins. All are cytosolic enzymes with molecular mass of 35–37 kDa. VR is similar to DFR and ANR with an identity of ∼ 40%, but different from IFR, with no significant sequence identity (only ∼ 10%). Chalcone reductase, an enzyme that acts early in flavonoid metabolism,21 has no significant similarity to any of the other reductases in the pathway. So far, only structures of chalcone reductase and IFR have been reported,22,23 and the latter provides a structural basis for understanding the mechanism
for introduction of the stereochemistry at C3 of isoflavanones (Figure 1). However, since both chalcone reductase and IFR are unrelated to VR, it is necessary to solve the three-dimensional structure of VR, and related enzymes such as SOR that introduce the opposite stereochemistry, in order to understand how the stereochemistry at isoflavanone position C4 (position 11a of the pterocarpan) is introduced. These stereo-specific enzymes are potential targets for metabolic engineering to alter the stereochemistry of the host's pterocarpan phytoalexins in order to overcome their microbial degradation. Here, we present the crystal structure of VR from alfalfa at 1.4 Å resolution. The structure reveals distinct features different from other known NADPH-dependent enzyme structures. A comparative structural study further identifies the putative binding sites for the co-factor NADPH and the substrate (3R)-vestitone, and a catalytic triad (Ser129Tyr164-Lys168) with Tyr164 as a general base for enzymatic catalysis.
Results and Discussion Structure determination and overall structure The M. sativa VR was expressed in Escherichia coli and purified, and single crystals were obtained. The crystal structure of VR was determined using the multiwavelength anomalous dispersion method with a crystal of Se-Met-substituted enzyme, and refined at 1.4 Å resolution to an R-factor of 18.8% and Rfree of 20.3% with a native data set. Data collection, phasing and refinement statistics are presented in Table 1. The asymmetric unit contains one molecule of VR. This has overall dimensions of ∼ 60 Å × 47 Å × 51 Å , and a bi-domain structure consisting of two distinct domains, an N-terminal domain and a C-terminal domain (Figures 2 and 3). The large N-terminal domain is comprised of a Rossmann fold with an
267
Vestitone Reductase Structure Table 1. Data collection, phasing and refinement statistics
A. Data collection and phasing statistics Resolution(Å) Wavelength(Å ) Unique reflections Completeness(%) Rsym (%) I/σ(I) Figure of merit B. Refinement statistics R-factor (%) Rfree (%) Number of protein atoms Number of solvent atoms Average B-factors (Å2) rmsd from ideal values Bond lengths (Å) Bond angles (deg.)
Native
Se-peak
Se-inflection
Se-remote
1.4 0.9793 77,901 97.9(97.7) 5.1(39.2) 26.6(4.3)
2.0 0.9789 27,146 98.9(89.1) 11.8(42.6) 16.0(3.1) 0.47
2.0 0.9791 27,260 99.8(100) 12.9(38.4) 14.7(4.6)
2.0 0.9641 27,260 99.0(100) 10.7(36.6) 17.1(4.4)
18.8 20.3 2405 542 21.5 0.0044 1.28
Rsym = ∑hkl|I–|/∑I, where I is the observed intensity and is the average intensity from observations of symmetryrelated reflections. A subset of the data (10%) was excluded from the refinement and used to calculate the free R value (Rfree). R-factor = ∑||Fo|–|Fc||/∑|Fo|. Numbers in parentheses are for the highest resolution shell.
eight-stranded parallel β-sheet (β1–β7 and β10) flanked by nine α helices (α1–α7, α9, and α11) on both sides. The small C-terminal domain contains a
mixed parallel and antiparallel β-sheet (β8–β9 and β11–β13) and three α helices (α8, α10, and α12). These two domains form a cleft that would be the
Figure 2. Ribbon diagram of the VR structure. The α helices, β strands, and the N and C termini are labeled. Figures 2, 4, 6, and 7 were prepared with MOLSCRIPT41 and Raster3D.42
268
Vestitone Reductase Structure
Figure 3. Sequence alignment of VR from M. sativa, SOR from Pisum sativum, DFR and ANR from Medicago truncatula, E. coli UDP-galactose 4-epimerase, human 17β-hydroxysteroid dehydrogenase type 1 (17β-HSD1), and mouse lung carbonyl reductase. The secondary structure elements observed in the VR structure are shown above the alignment. Conserved residues are highlighted. This Figure was produced with ENDscript.43
binding sites for NADPH and (3R)-vestitone. Interactions observed between the N and C-terminal domains include hydrogen bonds formed between the OD2 atom of Asp232 and the OH atom of Tyr307, between the OD1 atom of Asp233 and the mainchain nitrogen atom of Lys306, and between the ND2 atom of Asn252 and the main-chain oxygen atom of
Asn294, and some weak interactions formed through water molecules. The two flexible regions between β2 and α2 (Pro40–Lys43) and between β4 and α4 (Asp90– Glu97) are disordered and invisible in the VR crystal structure. The first four residues of the N terminus are also disordered.
Vestitone Reductase Structure
Comparison with other NAD(P)H-dependent enzymes VR represents a large family of proteins, including VR homologs in legumes and other highly homologous reductases such as DFRs involved in flavonoid metabolism. A BLAST search against the Arabidopsis thaliana genome protein databases at The Institute for Genomic Research (TIGR) showed 20 hits that are ∼ 30–40% identical with the VR sequence. However, there was no highly homologous protein structure identified when searching against the RCSB Protein Data Bank (PDB). One of the similar proteins with known structures was E. coli UDP-galactose 4-epimerase (PDB 1XEL). Initial attempts to solve the structure of VR by molecular replacement using this epimerase structure as a search model failed, indicating that VR has some unique and special structural features. Amino acid sequence analysis, including a Pfam database search,24 showed that VR has features that place it in the epimerase/dehydratase family and the extended short-chain dehydrogenase/ reductase (SDR) superfamily (Figure 3).25,26 VR contains an N-terminal glycine-rich motif or fingerprint G12G13XG15XXG18, similar to the GXXXGXG se-
269 quence motifs involved in NAD(P)H binding of SDR enzymes, and a putative catalytic motif Y164XXXK168, a highly conserved active-site pattern for SDR enzymes. However, most members of the SDR superfamily contain about 250 amino acid residues. VR is larger, with 326 amino acid residues, similar to the epimerases, and may be classified as a member of the extended SDR superfamily. Structural comparison through the DALI search server 27 revealed similarities between VR and enzymes of the SDR or extended-SDR superfamily, such as E. coli UDP-galactose 4-epimerase,28 human 17β-hydroxysteroid dehydrogenase type 1 (17βHSD1), 29 and mouse lung carbonyl reductase (MLCR)30 (Figure 4(a) and (b)). These enzymes are similar in their N-terminal domains, but their C-terminal domains vary in both size and shape. Superimposing the structure of VR onto that of UDP-galactose 4-epimerase (PDB 1XEL) revealed structural similarity with root-mean-square deviation (rmsd) of 2.0 Å for 223 Cα atoms (using a 4.0 Å rmsd cutoff) and 16% sequence identity. Comparison to the regular SDR enzymes 17β-HSD1 (PDB 1FDT) and MLCR (PDB 1CYD), which are smaller and contain mainly one compact domain, also showed similarity, with rmsd values of 1.9 Å (157
Figure 4. A stereo diagram showing the superimposition of the structures of VR (cyan), epimerase (magenta in (a)); and MLCR (magenta in (b)). Glycine residues in the fingerprints are labeled, and catalytic triads are shown as ball- and -stick models. (a) Ligands NADH and UDP-glucose in the structure of epimerase, and (b) NADPH and 2-propanol in the structure of MLCR are shown in yellow as ball- and -stick models.
270 Cα atoms) and 2.0 Å (154 Cα atoms), respectively. Both 17β-HSD1 and MLCR superimpose mainly on the large N-terminal domain of VR. The structural conservation suggests that VR would interact with its co-factor NADPH and substrate vestitone in a mode similar to that observed for these and other SDR enzymes, although the sequence identity between VR and these SDR enzymes is very low (< 20%). Although the overall folding of VR is similar to that of other SDR enzymes, only about 68% of the Cα atoms of VR may be superimposed onto the epimerase structure, with a 4.0 Å rmsd cutoff, and about 48% onto 17β-HSD1 and MLCR, which have no distinct C-terminal domains. Compared to the VR C-terminal domain consisting of five β strands (β8–β9 and β11–β13) and three α helices (α8, α10, and α12) with ∼ 100 residues, there are only ∼ 30 residues forming mainly two α helices in MLCR structure (in the corresponding locations of α8 and β9 in the VR structure), and ∼45 residues forming three α helices and a long loop in the 17β-HSD1 structure (in the corresponding locations of α8, β9, and α10, and loop β8–α8 in the VR structure). The major differences between VR and epimerase are also in their C-terminal regions, which have quite different conformations and lengths around the regions of three loops β8–α8, α8–β9, and α10–β12,
Vestitone Reductase Structure
and the first two loops in VR are short due to some amino acid deletions (Figure 3). The major difference in their N-terminal domains is the region around helix α6 and loop α6–α7; VR has an extra helix α6 with an eight amino acid insertion. Most of these regions are around the cleft proposed as the putative binding site for co-factor and substrate. The fingerprint and putative binding pocket for the co-factor A large cleft was observed on the surface of the VR molecule, starting from the fingerprint G12G13XG 15 XXG18 region and extending to near the catalytic motif Y164XXXK168 region (Figure 5). The location of this highly conserved fingerprint region in the VR structure is between strand β1 and helix α1, similar to that of the fingerprint regions observed in the structures of epimerase bound with NADH and UDP-glucose, 17β-HSD1 with NADP+ and equilin, and MLCR with NADPH and 2propanol (Figure 4(a) and (b)). The fingerprint is a common NAD(P)H recognition motif for the SDR enzymes and other NADPHdependent enzymes. The portion of the large cleft near the fingerprint would therefore be the putative NADPH-binding pocket for VR. Comparison by superimposing structures showed that the putative
Figure 5. Molecular surface of VR. The electrostatic surface of the VR enzyme was calculated with PyMOL [http:// www.pymol.sourceforge.net/].
Vestitone Reductase Structure
binding pocket of VR is also similar to that of epimerase and MLCR (Figure 4(a) and (b)). NADPH would probably bind to VR in a conserved mode, as observed in the structures of epimerase, MLCR, and other members of the NAD(P)H-dependent SDR enzyme family. The NADPH pyrophosphate group may fit between loop β1–α1 (i.e. the fingerprint) and loop β4–α4 (Thr85–Pro88), and may form hydrogen bonds with some main chain atoms of these two loops. As revealed in the structures of 17β-HSD1 and MLCR, which use NADP+ or NADPH as a co-factor, a basic arginine residue, Arg37 in 17β-HSD1 and Arg39 in MLCR, forms hydrophobic contact with the adenine ring and/or electrostatic interaction with the 2′-ribose monophosphate group of NADPH. In the structure of VR, Arg37 is located in the equivalent position and may interact with the adenine and/or 2′-phosphate group of NADPH. There are several other basic residues, including Arg42 and Lys43 (in the disordered region), and Arg44 located close to this area. These basic residues are conserved in VR homologs. The co-factor-binding pocket of VR is more similar to those of 17β-HSD1 and MLCR, especially the segment interacting with the adenine base and the 2′-monophosphate group of NADPH, than that of epimerase, although the overall folding of VR is more similar to that of epimerase. This may be because VR, 17β-HSD1 and MLCR are NADPHpreferring enzymes, whereas epimerase uses NAD+ as a cofactor. Epimerase has an asparagine residue, Asn32, present in the position corresponding to Arg37 in VR, whereas in most of the NAD(H)preferring SDR enzymes, an aspartate residue in this position interacts with the adenine ribose of NAD(H). Catalytic triad and isoflavone substrate binding–implications for catalytic mechanism A Ser-Tyr-Lys triad was identified and proposed for the catalytic mechanisms of the SDR enzymes.26 For example, epimerase has the triad Ser124-Tyr149Lys153, 17β-HSD1 has Ser142-Tyr155-Lys159 , and MLCR has Ser136-Tyr149-Lys153.30 In the VR structure, the conserved Y164XXXK168 motif is located near one end of the large cleft (Figures 4(a) and (b), and 5). There are several serine residues present in the cleft near Tyr164, including Ser 87, Ser128, Ser129, Ser131, and Ser134. Structural comparison showed that Ser129 is located in the position corresponding to the serine of the catalytic triads of SDR enzymes (Figure 4(a) and (b)). This Ser129, together with Tyr164 and Lys 168, may act as the catalytic triad for VR. These catalytic triad residues are conserved in all VR homologs, and may be essential for enzyme catalysis and function. The pocket near the catalytic triad is presumably the substrate-binding site. The pocket is formed by loop β5–β6 (∼residue Ala132), loop α6–α7 (∼Phe160), loop β7–β8 (∼Pro192), helix α8 (∼Glu208), strand β9 (∼Arg225), and strand β12
271 (∼Leu290). The location of this pocket in the VR structure is similar to those of the substrate-binding pockets observed in the structures of UDP-galactose 4-epimerase and MLCR (Figure 4(a) and (b)). In the MLCR structure (PDB 1CYD),30 the substrate-binding pocket is very small and restricted by hydrophobic residues (i.e. Phe143, Leu146, Met137, Val138, Val181, Val190) and the catalytic triad, and its small substrates such as pyridine 3-aldehyde or acetone may be completely buried inside the enzyme molecule. In the epimerase structure (PDB 1XEL),28 the substrate-binding pocket is relatively long and extended to fit the long substrate UDPgalactose or UDP-glucose, and several charged residues, Asp295, Arg231, and Arg292, are located inside the pocket and interact with ribose hydroxyl and phosphate group of substrate. In contrast, the putative VR substrate-binding pocket is quite different in size, shape and overall features. The pocket in the VR structure is relatively hydrophobic, and is surrounded by the aromatic residues Phe193, Phe226, Phe135 and Tyr164, several other hydrophobic residues (e.g. Leu293, Val223, and Leu290), and a basic residue, His227, located in the bottom of the pocket. A molecular docking study of VR with co-factor and substrate using the program GOLD was carried out by defining the active site with the catalytic residue Tyr164 (Figure 6). In the model with the keto group at position 4 of vestitone docked close to the OH of Tyr164 (∼ 2.8 Å), the substrate (3R)-vestitone may stack its aromatic A and C rings with the nicotinamide ring of NADPH, its keto oxygen may also interact directly with the OG of the catalytic triad residue Ser129 (∼ 2.5 Å), and its B-ring may point to the molecular surface. The third catalytic triad residue, Lys168, has no interaction with substrate, and its NZ atom is ∼ 2.9 Å away from the 2′-hydroxyl group on the ribose moiety of NADPH. The keto group of (3S)-vestitone cannot be docked into a position for interacting with the catalytic residue Tyr164 when fitting it into the pocket. Its B-ring would be repulsed by His227 and other residues in the bottom of the pocket, and its A and C rings would not form a favored stacking interaction with the NADPH nicotinamide ring. This is consistent with previously reported enzyme assay results, that VR utilizes only (3R)-vestitone as substrate and shows no activity with (3S)-vestitone.15 The structural comparison and molecular docking study described above suggested that the interactions between VR enzyme, cofactor and substrate would be similar to those observed in other SDR enzyme structures, and VR may have an enzymatic mechanism similar to that proposed for SDR enzymes.26 The substrate (3R)-vestitone may interact with NADPH by stacking its aromatic ring with the nicotinamide ring of NADPH, which would attack and transfer its hydrogen atom onto the C4 atom of vestitone. The catalytic base residue Tyr164 would attack the oxygen atom at position 4 of vestitone from the opposite side, and Ser129 may form a hydrogen bond with this oxygen atom.
272
Vestitone Reductase Structure
Figure 6. A stereo diagram showing the putative ligand-binding pocket in the structure of VR. Some amino acid residues in the binding pockets are labeled and shown as bond models in cyan. The NADPH and (3R)-vestitone docked into the putative binding pocket of VR are shown as ball- and -stick models.
Lys168 may interact with the ribose of NADPH and form hydrogen bonds to its 2′ and/or 3′-hydroxyl groups, to stabilize the position of the nicotinamide moiety. The high substrate stereochemical specificity of VR is determined by the overall shape and features of the binding pocket. His227 located in the pocket might be a critical residue, and two other charged residues (i.e. Asp205 and Lys209) on the molecular surface and several hydrophobic residues (e.g. Phe193, Leu290, and Phe135) might have important roles for interacting substrate. The mobile region in VR and its relation to function Both of the two disordered regions in the VR structure are close to the putative NADPH binding site, and the disordered region between residues Asp90 and Glu97 is close to the substrate-binding site. The average temperature factors of the segments near the two disordered regions are about 40 Å2 (40.3 Å2 for residues Arg37–Asp45, and 38.2 Å2 for residues Ser87–Val100), much higher than the overall value of 21.5 Å2. The disorder and high-temperature factors of these regions indicate their flexibility, which is probably important for VR function. It has been reported that “open to closed” or “disordered to ordered” conformational changes occur upon binding of ligands to many SDR enzymes.26 Structural comparison between 17β-HSD1 apoenzyme and its complexes with cofactor and/or inhibitors showed open to closed conformational changes of the substrate-binding regions of the enzyme.29 Similarly, these regions in the VR structure may undergo a conformation change once cofactor or/and substrate bind to the enzyme. Although the co-factor NADPH or substrate vestitone were used for co-crystallization and soaking with native crystals, and several data sets were collected and analyzed, no electron density was observed for these ligands. As observed in the
crystal structure, the two disordered regions (Pro40– Lys43, and Asp90–Glu97) are close to the ligandbinding sites, indicating that ligand binding would cause the conformation changes and disrupt the current crystal lattice and packing formed in this specific crystallization condition, which is consistent with the observation of crystal damage during soaking experiments. To obtain crystals of the enzyme in complexes with ligands, it would be necessary to screen other crystallization conditions leading to different crystal forms. Another strategy is to mutate the catalytic residue Tyr164 to alanine or glycine for abolishing enzymatic activity, and using the inactive mutant for co-crystallization with both co-factor and substrate. These mutagenesis and crystallization trials are in progress. Comparison with other VR-homolog proteins-implications for substrate specificity SOR is a VR homolog identified in pea.17 SOR recognizes (–) sophorol, which has the same stereochemistry as (3R)-vestitone, but differs in the presence of the methylenedioxy ring between positions 4′ and 5′ of the B ring (Figure 1). The level of sequence identity between VR and SOR is 86%. A three-dimensional structural model for SOR was built using the program MODELLER with the VR structure as a template. The overall folding of SOR is similar to that of VR. Their putative substrate-binding pockets are very similar, with most of the residues conserved (Figure 7). There are only minor differences observed in the putative substrate-binding pockets. In the modeled structure of SOR, an alanine residue, Ala291, is present in the position corresponding to Pro291 in VR. The smaller residue in SOR may provide more space to accommodate the larger substrate. This suggests that SOR may possibly recognize a smaller molecule, such as (3R)-vestitone, as a substrate. A docking study showed that both (–) sophorol and
273
Vestitone Reductase Structure
(3R)-vestitone were docked into the active site of SOR structural model in the same orientation with similar fitness scores, since these two compounds are very similar in structure (Figure 1). However, the pterocarpans accumulating in pea contain the methylenedioxy substituent, presumably because the pathway architecture results in substitution occurring before reduction. Another minor difference is that a serine residue in SOR is present at the position corresponding to Ala165 of VR. The hydroxyl group of Ser165 in SOR may interact with the oxygen atom at position 5′ of sophorol directly or through a water molecule. A BLAST search showed that there is a protein in soybean (Glycine max) with 75% sequence identity to VR. It is annotated as a 2′-hydroxydihydrodaidzein reductase, but has not been functionally characterized. Sequence comparison showed that most of the residues present in the VR substrate-binding pocket are highly conserved in this protein, except for Ala130, Ser162 and Arg210, which correspond to Gly130, Gly161, and Lys209 in VR, respectively. This suggests that the soybean homolog may be involved in the biosynthesis of glyceollin, the peterocarpan phytoalexin in soybean. IFR precedes VR in the biosynthetic pathway of medicarpin in alfalfa and pisatin in pea. It is also an NADPH-dependent reductase, which reduces the 2,3 double bond of isoflavone to produce an asymmetric compound that is the substrate for VR or SOR. Although the level of sequence identity between VR and IFR is very low (∼ 10%), their threedimensional structures are very similar. Superimposition of the structures of VR and IFR gave an rmsd of 2.9 Å for 213 Cα atoms. IFR also contains a similar fingerprint region, but no YXXXK motif or catalytic triad. A single lysine residue, Lys144, was identified as the catalytic residue for IFR catalysis.23 The location of Lys144 in IFR is similar to that of Lys168 in VR. The lysine residue at this location in SDR enzymes was proposed to stabilize the nicotinamide moiety of NAD(P)H, and would not be involved directly in catalysis, including substrate binding or hydride transfer. However, in the cases of IFR and pinoresinol-lariciresinol reductase and phe-
nylcoumaran benzylic ether reductase, two IFR-like enzymes involved in lignan biosynthesis,31 there are no catalytic triads, and the lysine residue present in the substrate-binding pocket is the only residue that can act as the general base for catalysis. VR, IFR and other SDR enzymes likely share a common evolutionary origin from a progenitor enzyme with the same basic three-dimensional structure. Subsequent changes of catalytic residues led to new functions that helped plants adapt to their environment. DFR and ANR are two other NADPH-dependent reductases in the phenylpropanoid pathway, and have about 40% identity with VR. Sequence analysis showed that they are also members of the SDR enzyme family, containing similar fingerprint motifs and catalytic triads (Ser128-Tyr163-Lys167 in DFR, and Ser133-Tyr170-Lys174 in ANR, respectively) (Figure 3). This suggests that they may have catalytic mechanisms similar to that of VR rather than to IFR. DFR and ANR recognize flavonoids as substrates, and VR and IFR use isoflavonoids. The residues located in the substrate-binding pockets for substrate recognition and binding are very different in VR and DFR/ANR. The three charged residues in VR, Asp205, Lys209, and His227, are substituted by Pro204, Thr208, and Gln227 in DFR and by Ser211, Met215, and Gln234 in ANR. For the two positions that distinguish the catalytic pockets of VR and SOR (Ala165 and Pro291 in VR, Ser165 and Ala29 in SOR), Phe164 and Val290 are present in DFR, and Pro171 and Ile304 in ANR. There is an insertion of eight residues in the ANR sequence near the corresponding position of His227 in VR. All these differences may contribute to the different substrate specificities of these enzymes.
Materials and Methods Cloning, protein expression and purification The vestitone reductase cDNA from alfalfa (M. sativa L.) was cloned into the E. coli expression vector pHIS8, a derivative of pET28a kindly provided by Dr J. Noel. E. coli BL21(DE3) cells transformed with plasmid encoding the
Figure 7. A stereo diagram showing the superimposition of putative substrate-binding pockets in the structure of VR, and the modeled structure of SOR from pea. Some amino acid residues in the binding pockets of VR and SOR are labeled and shown as bond models in cyan and magenta, respectively.
274 VR construct were grown at 37 °C in LB medium containing 50 μg/ml of kanamycin until A600 nm = 0.6– 0.8. After induction with 0.5 mM isopropyl 1-thio-βgalactopyranoside (IPTG), the cultures were grown overnight at 16 °C. Cells were pelleted and resuspended in lysis buffer (50 mM Tris–HCl (pH 8.0), 500 mM NaCl, 10 mM imidazole, 20 mM β-mercaptoethanol). After lysis with a French press and centrifugation at 20,000g (SORVALL SA-600) at 4 °C for 20 min, Ni2+-NTA agarose was added to the supernatant containing the target protein. After incubation for 40–60 min, the mixture was transferred into a disposable column and washed extensively with lysis buffer (about 50 column volumes). Histagged proteins were washed from the column with elution buffer (50 mM Tris–HCl (pH 8.0), 500 mM NaCl, 250 mM imidazole, 20 mM β-mercaptoethanol). Incubation with biotinylated thrombin during dialysis overnight at 4 °C against 20 mM Tris–HCl (pH 7.5), 100 mM NaCl, 10 mM β-mercaptoethanol cleaved the N-terminal Histag. Dialyzed proteins were incubated with streptavidin agarose and Ni2+-NTA agarose to remove thrombin and the cleaved His tag, respectively. The proteins were further purified on Resource Q anion-exchange and Superdex-200 gel-filtration columns (GE Healthcare) and concentrated to 4–10 mg/ml in 10 mM Tris–HCl (pH 7.5), 10 mM NaCl, 5 mM β-mercaptoethanol. Selenomethionine-substituted VR was prepared by growing E. coli cells in M9 minimal medium supplemented with L-selenomethionine (Sigma) and purified by following the same protocol used for the native protein.
Vestitone Reductase Structure tively. A bulk solvent correction was applied. B factors were refined individually. Water molecules were added with Arp/wArp37 and checked manually for inclusion. The program PROCHECK 38 was used to check the model. All backbone ϕ-ψ torsion angles are within the allowed regions of the Ramachandran plot. Molecular modeling and docking The comparative modeling program MODELLER39 was used to generate a model for the sophorol reductase from pea. The structure of VR was used as a template in the structural modeling experiment. The three-dimensional model was obtained by optimally satisfying spatial restraints derived from the sequence alignment based on the CLUSTALX results and three-dimensional structure. The automated docking program GOLD40 was used to dock co-factor NADPH and substrate (3R)-vestitone into the VR enzyme active site. Default genetic algorithm parameters for controlling the operation of the docking process were used. The docking calculations were restricted to the putative binding pocket by defining the active site with residue Tyr164. GOLDscore was used to identify the lowest energy docking results. The hydrogen bonds and van der Waals contacts between ligands and enzyme were analyzed to identify the optimal binding mode. Minor manual adjustments of the GOLD solution were made using the program O. Protein Data Bank accession code
Crystallization and data collection Crystals of VR were grown using the hanging-drop, vapor-diffusion method. A sample(2 μl) of VR protein solution (∼ 4 mg/ml) was mixed with 2 μl of reservoir solution (0.1 M Tris–HCl (pH 8.0), 20% (w/v) polyethylene glycol 6000, 0.1 M MgCl2,). The mixture was equilibrated over the reservoir solution at 4 °C. Crystals grew over two weeks to dimensions of ∼ 0.3 mm × 0.2 mm × 0.1 mm. Before data collection, crystals were transferred to a solution of the same composition but supplemented with 20% (v/v) glycerol as a cryoprotectant, and flash-frozen in liquid nitrogen. Data from a crystal of VR were measured to 1.4 Å resolution with an ADSC Quantum 315 CCD detector at the SBC 19ID beamline at the Advanced Photon Source (APS), Argonne National Laboratory (Argonne, IL). The crystal belongs to space group P21212 (a = 69.0 Å, b = 92.3 Å, c = 63.3 Å). There is one molecule per crystallographic asymmetric unit with 50% (v/v) solvent content and a VM of 2.7 Å3/Da. Multiwavelength anomalous X-ray diffraction data at 2.0 Å resolution were collected from a Se-Met-labeled VR crystal, using three wavelengths, at the SBC 19BM beamline at APS. All data sets were processed using the program suite HKL2000.32
The atomic coordinates and structure factors for the structure of VR have been deposited with the RCSB Protein Data Bank (PDB) with the PDB ID 2P4H.
Acknowledgements We thank Drs X. He and S. Liu for initial cloning and protein purification, R. Guan for helpful advice in crystallography, M. Udvardi and L. Modolo for critical reading of the manuscript, and F. Rotella and Y. Kim at the Structural Biology Center beamlines 19BM and 19ID at the Advanced Photon Source, Argonne National Laboratory (Argonne, IL), for assistance with data collection. Argonne is operated by UChicago Argonne, LLC, for the U.S. Department of Energy, Office of Biological and Environmental Research under contract DE-AC02-06CH11357. This work was supported by the Samuel Roberts Noble Foundation.
Structure determination and refinement The structure of VR was determined using the multiwavelength anomalous dispersion method, and the data (40–2.0 Å) were analyzed with SOLVE.33 Three out of four selenium sites were located, yielding an overall figure of merit of 0.47. The program RESOLVE was used for electron density modification and automated modelbuilding.34 Interactive model building and crystallographic refinement were carried out with a native data set at 1.4 Å using the programs O35 and CNS,36 respec-
References 1. Dixon, R. A. (1999). Isoflavonoids: biochemistry, molecular biology abd biological functions. In Comprehensive Natural Products Chemistry (Sankawa, U., ed.), pp. 773–823, Elsevier, Oxford. 2. Dixon, R. A. & Sumner, L. W. (2003). Legume natural products: understanding and manipulating complex pathways for human and animal health. Plant Physiol. 131, 878–885.
275
Vestitone Reductase Structure 3. Dixon, R. A. & Ferreira, D. (2002). Molecules of interest: genistein. Phytochemistry, 60, 205–211. 4. Palevitz, B. A. (2000). Soybeans hit main street. Scientist, 14, 8–9. 5. Ishimi, Y., Miyaura, C., Ohmura, M., Onoe, Y., Sato, T., Uchiyama, Y. et al. (1999). Selective effects of genistein, a soybean isoflavone, on B-lymphopoiesis and bone loss caused by estrogen deficiency. Endocrinology, 140, 1893–1900. 6. Fritz, W. A., Coward, L., Wang, J. & Lamartiniere, C. A. (1998). Dietary genistein: perinatal mammary cancer prevention, bioavailability and toxicity testing in the rat. Carcinogenesis, 19, 2151–2158. 7. Dixon, R. A. & Steele, C. L. (1999). Flavonoids and isoflavonoids- a gold mine for metabolic engineering. Trends Plant Sci. 4, 394–400. 8. Dixon, R. A., Dey, P. M. & Lamb, C. J. (1983). Phytoalexins: enzymology and molecular biology. Advan. Enzymol. Relat. Areas Mol. Biol. 55, 1–136. 9. Dixon, R. A., Achnine, L., Kota, P., Liu, C.-J., Reddy, M. S. S. & Wang, L. (2002). The phenylpropanoid pathway and plant defence-a genomics perspective. Mole. Plant Pathol. 3, 371–390. 10. Guo, L., Dixon, R. A. & Paiva, N. L. (1994). The ‘pterocarpan synthase’ of alfalfa: association and coinduction of vestitone reductase and 7,2′-dihydroxy4′-methoxy-isoflavanol (DMI) dehydratase, the two final enzymes in medicarpin biosynthesis. FEBS Letters, 356, 221–225. 11. Cruickshank, I. A. M. & Perrin, D. R. (1960). Isolation of a phytoalexin from Pisum sativum L. Nature, 187, 799–800. 12. Daniel, S., Tiemann, K., Wittkampf, U., Bless, W., Hinderer, W. & Barz, W. (1990). Elicitor-induced metabolic changes in cell cultures of chickpea (Cicer arietinum L.) cultivar resistant and susceptible to Ascochyta rabiei. Planta, 182, 270–278. 13. VanEtten, H. D., Matthews, D. E. & Matthews, P. S. (1989). Phytoalexin detoxification: importance for pathogenicity and practical implications. Ann. Rev. Phytopathol. 27, 143–164. 14. Dixon, R. A., Harrison, M. J. & Paiva, N. L. (1995). The isoflavonoid phytoalexin pathway: from enzymes to genes to transcription factors. Physiol. Plant, 93, 385–392. 15. Guo, L., Dixon, R. & Paiva, N. L. (1994). Conversion of vestitone to medicarpin in alfalfa (Medicago sativa L.) is catalyzed by two indenpendent enzymes. J. Biol. Chem. 269, 22372–22378. 16. Guo, L. & Paiva, N. L. (1995). Molecular cloning and expression of Alfalfa (Medicago sativa L.) vestitone reductase, the penultimate enzyme in medicarpin biosynthesis. Arch. Biochem. Biophys. 320, 353–360. 17. DiCenzo, G. L. & VanEtten, H. D. (2006). Studies on the late steps of (+) pisatin biosynthesis: evidence for (−) enantiomeric intermediates. Phytochemistry, 67, 675–683. 18. Paiva, N. L., Edwards, R., Sun, Y., Hrazdina, G. & Dixon, R. A. (1991). Stress responses in alfalfa (Medicago sativa L.). 11. Molecular cloning and expression of alfalfa isoflavone reductase, a key enzyme of isoflavonoid phytoalexin biosynthesis. Plant Mol. Biol. 17, 653–667. 19. Beld, M., Martin, C., Huits, H., Stuitje, A. R. & Gerats, A. G. (1989). Flavonoid synthesis in Petunia hybrida: partial characterization of dihydroflavonol-4-reductase genes. Plant Mol. Biol. 13, 491–502. 20. Xie, D., Sharma, S. B., Paiva, N. L., Ferreira, D. & Dixon, R. A. (2003). BANYULS encodes anthocyani-
21.
22.
23.
24.
25.
26. 27. 28.
29.
30.
31.
32. 33. 34. 35.
36.
37.
din reductase active in plant flavonoid biosynthesis. Science, 299, 396–399. Welle, R. & Grisebach, H. (1989). Phytoalexin synthesis in soybean cells: elicitor induction of reductase involved in biosynthesis of 6′-deoxychalcone. Arch. Biochem. Biophys. 272, 97–105. Bomati, E. K., Austin, M. B., Bowman, M. E., Dixon, R. A. & Noel, J. P. (2005). Structural elucidation of chalcone reductase and implications for deoxychalcone biosynthesis. J. Biol. Chem. 280, 30496–30503. Wang, X., He, X., Lin, J., Shao, H., Chang, Z. & Dixon, R. A. (2006). Crystal structure of isoflavone reductase from alfalfa (Medicago sativa L.). J. Mol. Biol. 358, 1341–1352. Sonnhammer, E. L. L., Eddy, S. R. & Durbin, R. (1997). Pfam: a comprehensive database of protein domain families based on seed alignments. Proteins: Struct. Funct. Genet. 28, 405–420. Jörnvall, H., Persson, B., Krook, M., Atrian, S., Gonzalez-Duarte, R., Jeffery, J. & Ghosh, D. (1995). Short-chain dehydrogenases/reductases (SDR). Biochemistry, 34, 6003–6013. Tanaka, N., Nonaka, T., Nakamura, K. T. & Hara, A. (2001). SDR: structure, mechanism of action, and substrate recognition. Curr. Org. Chem. 5, 89–111. Holm, L. & Sander, C. (1993). Protein structure comparison by alignment of distance matrices. J. Mol. Biol. 233, 123–138. Thoden, J. B., Frey, P. A. & Holden, H. M. (1996). Molecular structure of the NADH/UDP-glucose abortive complex of UDP-galactose 4-epimerase from Escherichia coli: implications for the catalytic mechanism. Biochemistry, 35, 5137–5144. Sawicki, M. W., Erman, M., Puranen, T., Vihko, P. & Ghosh, D. (1999). Structure of the ternary complex of human 17beta-hydroxysteroid dehydrogenase type 1 with 3-hydroxyestra-1,3,5,7-tetraen-17-one (equilin) and NADP+. Proc. Natl Acad. Sci. USA, 96, 840–845. Tanaka, N., Nonaka, T., Nakanishi, M., Deyashiki, Y., Hara, A. & Mitsui, Y. (1996). Crystal structure of the ternary complex of mouse lung carbonyl reductase at 1.8 A resolution: the structural origin of coenzyme specificity in the short-chain dehydrogenase/reductase family. Structure, 4, 33–45. Min, T., Kasahara, H., Bedgar, D. L., Youn, B., Lawrence, P. K., Gang, D. R. et al. (2003). Crystal structures of pinoresinol-lariciresinol and phenylcoumaran benzylic ether reductases and their relationship to isoflavone reductases. J. Biol. Chem. 278, 50714–50723. Otwinowski, Z. & Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326. Terwilliger, T. C. & Berendzen, J. (1999). Automated MAD and MIR structure solution. Acta Crystallog. sect. D, 55, 849–861. Terwilliger, T. C. (2000). Maximum likelihood density modification. Acta Crystallog. sect. D, 56, 965–972. Jones, T. A., Zou, J.-Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for the building of protein models in electron density maps and the location of errors in these models. Acta Crystallog. sect. A, 47, 110–119. Brünger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). Crystallography and NMR system: a new software suite for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905–921. Lamzin, V. S., Perrakis, A. & Wilson, K. S. (2001). The
276
Vestitone Reductase Structure
ARP/WARP suite for automated construction and refinement of protein models. In International Tables for Crystallography–Crystallography of Biological Macromolecules (Rossmann, M. G. & Arnold, E., eds), vol. F, pp. 720–722. Kluwer Academic Publishers, Dordrecht. 38. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. (1993). PROCHECK—a program to check the stereochemical quality of protein structures. J. Appl. Crystallogr. 26, 283–291. 39. Marti-Renom, M. A., Stuart, A., Fiser, A., Sánchez, R., Melo, F. & Sali, A. (2000). Comparative protein structure modeling of genes and genomes. Annu. Rev. Biophys. Biomol. Struct. 29, 291–325.
40. Jones, G., Willett, P., Glen, R. C., Leach, A. R. & Taylor, R. (1997). Development and validation of a genetic algorithm for flexible docking. J. Mol. Biol. 267, 727–748. 41. Kraulis, P. J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallog. 24, 946–950. 42. Merritt, E. A. & Bacon, D. J. (1997). Raster3D: photorealistic molecular graphics. Methods Enzymol. 277, 505–524. 43. Gouet, P. & Courcelle, E. (2002). ENDscript : a workflow with web interface to display sequence and structure information. Bioinformatics, 18, 767–768.
Edited by M. Guss (Received 7 February 2007; received in revised form 12 March 2007; accepted 14 March 2007) Available online 21 March 2007