The Crystal Structure of Polyhydroxybutyrate Depolymerase from Penicillium funiculosum Provides Insights into the Recognition and Degradation of Biopolyesters

doi:10.1016/j.jmb.2005.12.028

J. Mol. Biol. (2006) 356, 993–1004

The Crystal Structure of Polyhydroxybutyrate Depolymerase from Penicillium funiculosum Provides Insights into the Recognition and Degradation of Biopolyesters Tamao Hisano1, Ken-ichi Kasuya2, Yoko Tezuka2, Nariaki Ishii2 Teruyuki Kobayashi3, Mari Shiraki3, Emin Oroudjev4, Helen Hansma4 Tadahisa Iwata5, Yoshiharu Doi5, Terumi Saito3 and Kunio Miki1,6* 1

RIKEN Harima Institute/ SPring-8, Koto 1-1-1, Mikazuki-cho, Sayo-gun Hyogo 679-5148, Japan 2

Department of Biological and Chemical Engineering, Faculty of Engineering, Gunma University, Tenjin-cho 1-5-1 Kiryu, Gunma 376-8515, Japan 3

Department of Biological Sciences, Faculty of Science Kanagawa University, Tsuchiya 2946, Hiratsuka, Kanagawa 259-1293, Japan 4

Department of Physics, University of California, Santa Barbara, CA 93106, USA 5 RIKEN Wako Institute, Hirosawa 2-1, Wako, Saitama 351-0198, Japan

Polyhydroxybutyrate is a microbial polyester that can be produced from renewable resources, and is degraded by the enzyme polyhydroxybutyrate depolymerase. The crystal structures of polyhydroxybutyrate depolymerase from Penicillium funiculosum and its S39A mutant complexed with the methyl ester of a trimer substrate of (R)-3-hydroxybutyrate have been ˚ and 1.66 A ˚ , respectively. The enzyme is determined at resolutions of 1.71 A comprised of a single domain, which represents a circularly permuted variant of the a/b hydrolase fold. The catalytic residues Ser39, Asp121, and His155 are located at topologically conserved positions. The main chain amide groups of Ser40 and Cys250 form an oxyanion hole. A crevice is formed on the surface of the enzyme, to which a single polymer chain can be bound by predominantly hydrophobic interactions with several hydrophobic residues. The structure of the S39A mutant–trimeric substrate complex reveals that Trp307 is responsible for the recognition of the ester group adjacent to the scissile group. It is also revealed that the substratebinding site includes at least three, and possibly four, subsites for binding monomer units of polyester substrates. Thirteen hydrophobic residues, which are exposed to solvent, are aligned around the mouth of the crevice, forming a putative adsorption site for the polymer surface. These residues may contribute to the sufficient binding affinity of the enzyme for PHB granules without a distinct substrate-binding domain. q 2005 Elsevier Ltd. All rights reserved.

6

Department of Chemistry Graduate School of Science Kyoto University, Sakyo-ku Kyoto 606-8502, Japan *Corresponding author

Keywords: polyhydroxyalkanoate; biodegradation; enzyme–substrate complex; subsite; polymer-adsorption site

Introduction Abbreviations used: PHA, polyhydroxyalkanoate; PHB, polyhydroxybutyrate; R3HB, (R)-3-hydroxybutyrate; H(R3HB)3M, methyl (3R)-3-[(3R)-3-{(3R)-3hydroxybutanoyloxy}butanoyloxy]butanoate; GlcNAc, N-acetylglucosamine; Man, mannose. E-mail address of the corresponding author: [email protected]

Polyhydroxyalkanoates (PHAs) are a class of aliphatic microbial polyesters produced by a variety of microorganisms as a carbon and energy-storage material.1 The most commonly occurring PHA is polyhydroxybutyrate (PHB), which is comprised of condensed (R)-3-hydroxybutyrate (R3HB) monomers. Secondary monomers can be incorporated into PHA, by varying the carbon source that is fed

0022-2836/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.

994 to the PHB-producing bacteria, by regulating genes involved in PHB biosynthesis, or by introducing a gene that affects a PHB biosynthetic pathway.2 PHAs have attracted commercial and academic interest, because they can be produced from renewable resources such as sugars and plant oils, and because the incorporation of secondary monomers affects the mechanical properties of the polymer, which can vary from thermoplastic to elastomer.3 Recent studies have focused on the elucidation of the mechanisms of PHA biodegradation and biosynthesis as well as production of new biodegradable polyesters.2,4,5 PHB is degraded by specific hydrolyzing enzymes, extracellular and intracellular PHB depolymerases (EC 3.1.1.75).6,7 Extracellular depolymerases are able to degrade partially crystallized (denatured) PHB, whereas intracellular depolymerases act on amorphous (native) PHB. Extracellular PHB depolymerases are typically comprised of three functional domains: a catalytic (320–400 aa), a linker (50–100 aa), and a substratebinding (40–60 aa) domain. The catalytic domain contains a lipase-like catalytic triad (serine, aspartic acid, and histidine residues), as well as a signature sequence of a pentapeptide [Gly-Xaa-Ser-Xaa-Gly] (where Xaa denotes any amino acid residue), known as the lipase box.6 Two types of catalytic domains have been identified, based on the position of the lipase box. In the type I domain, the lipase box is located in the middle of the primary structure, and in the type II domain, it is located at the amino terminus of the structure. The substrate-binding domain is responsible for the adsorption of the enzyme to the surface of the polymer, which permits the catalytic domain to interact with polymer chains, thus catalyzing their efficient hydrolysis.8,9 It has recently been proposed that the substrate-binding domain has an additional and more active function of disrupting the structure of the polymer.10,11,12 The role of the linker domain is not clearly understood, but it may function as a spacer that introduces a flexible region between the catalytic and substrate-binding domains to increase the hydrolytic efficiency of the catalytic domain.13 A recent study of a homologous protein to the linker domain, the fibronectin type III homology domain of cellobiohydrolase ChbA of Clostridium thermocellum, indicates that the linker domain may also exhibit a disruptive function against crystalline substrates, analogous to the substrate-binding domain. 14 The substrate-binding and linker domains are divided into two and three types, respectively, based on sequence similarity. Since these three functional domains are essential for the enzymatic degradation of water-insoluble polymers,13 it has been proposed that the degradation of the crystalline region of the polymer should proceed in three steps: adsorption of the enzyme to the polymer, non-hydrolytic disruption of the structure of the polymer, and hydrolysis.10,12,15–17 An extracellular PHB depolymerase from Penicillium funiculosum has been purified and

Crystal Structure of PHB Depolymerase

characterized.18,19 It has a molecular mass of about 33,000, and is the smallest of the known extracellular depolymerases, typically ranging from 40,000 to 57,000.6 It efficiently degrades PHB and a trimer of R3HB, although the degradation of PHB is less efficient by two orders of magnitude than a multidomain enzyme from Ralstonia pickettii.19 Furthermore, when the enzyme is inactivated by the inhibitor diisopropylfluorophosphate, it still retains the ability to bind reversibly to PHB.19 The issue of how depolymerases recognize a PHB chain with an intrinsically hydrophobic nature, as well as how such a small depolymerase effectively interacts with PHB, is of considerable interest. Here we report on the crystal structure of the PHB depolymerase from P. funiculosum at a reso˚ . The results show a single domain lution of 1.71 A structure with a circularly permuted variant of the a/b hydrolase fold. Furthermore, we determined the structure of the S39A mutant–R3HB trimer methyl ester (H(R3HB)3M) substrate complex at a ˚ , and reveal the binding mode resolution of 1.66 A for a trimer substrate. Solvent-exposed hydrophobic residues are aligned flat on the surface of the enzyme, serving as a polymer-adsorption site.

Results and Discussion Overall structure The crystal structure of the wild-type PHB depolymerase was determined at a resolution of ˚ with the crystallographic R and free-R factors 1.71 A of 19.0% and 22.1%, respectively. The shape of the molecule is globular with approximate dimensions ˚ !48 A ˚ !41 A ˚ , showing that the enzyme is of 52 A comprised of a single domain. The enzyme represents an a/b-type structure (Figure 1). It consists of an eight-stranded b-sheet composed of seven parallel and one antiparallel b-strands, nine a-helices surrounding the b-sheet, two concomitant two-stranded antiparallel b-sheets, and six 310-helices (Figures 2(a) and 3(a)). Four disulfide bridges, between Cys70–Cys79, Cys169–Cys180, Cys234–Cys241, and Cys250–Cys304 were identified. The topology of the secondary structure resembles that of the a/b hydrolase fold20 (Figure 2). A remarkable feature is that its polypeptide linkage is circularly permuted21 in comparison with that of the a/b hydrolase fold, that is, the positions of the N and C termini of the depolymerase are different from those of the canonical a/b hydrolase fold. A sequence analysis indicated that the enzyme can be classified as having a type II catalytic domain (K. K. et al., unpublished data). In the type II depolymerase structure, the location of the nucleophilic serine residue in the lipase box pentapeptide at the N-terminal region of the protein results in discontinuity of the polypeptide linkage in the middle of the central b-sheet, whereas in the canonical a/b hydrolase fold, the polypeptide linkage is continued from one edge to the other

Crystal Structure of PHB Depolymerase

995

Figure 1. Stereoview ribbon plot of a type II PHA depolymerase chain fold. a-Helices, b-strands, and 310-helices are colored in magenta, cyan, and orange, respectively. The mature form of the enzyme contains 319 amino acid residues with a calculated molecular mass of 33,569. The N and C termini of the chains are indicated. The catalytic triad residues Ser39, Asp121, and His155, four cystines Cys70–Cys79, Cys169–Cys180, Cys234–Cys241, and Cys250–Cys304, the glycosylated residue Asn144, and the attached oligosaccharide GlcNAc2Man3 are represented as ball-and-stick models. Figures were produced using the MOLSCRIPT43 and RASTER3D44 programs.

edge of the b-sheet. This is the first observation of a circularly permuted a/b hydrolase fold. This finding reveals an interesting relationship between the two types of catalytic domains, which are not apparently similar, with respect to the primary

structure. In fact, a comparison of the amino acid sequences of the type I enzymes with a modified sequence of P. funiculosum enzyme in which the circular permutation is cancelled so that Ser205 becomes an N-terminal residue show about a 28%

Figure 2. Topology diagram of (a) PHB depolymerase (type II) from P. funiculosum, and (b) the canonical a/b hydrolase fold. Boxes and arrows represent a-helices and bstrands, respectively. Positions of the catalytic residues are indicated as red circles.

996

Crystal Structure of PHB Depolymerase

Figure 3. Sequence alignments with (a) type II and (b) type I depolymerases. Catalytic triad, substrate-binding site, solvent-exposed hydrophobic, and disulfide bond-forming residues are colored in red, blue, magenta, and green, respectively. Note that Leu251, Leu298, and Pro301 are solvent-exposed residues contributing to the substrate-binding site. Secondary structural elements (magenta boxes, cyan arrows, and orange boxes indicate a-helices, b-strands, and 310helices, respectively) are aligned with the sequence of P. funiculosum depolymerase. Note that, although these enzymes except for that of P. funiculosum are composed of multiple domains, only the sequences of the catalytic domains are shown. (a) A signal sequence that is cleaved and not present in the mature protein is indicated in grey. Gly339, not included in the model due to disorder, is also indicated in grey. The sequence accession codes for enzymes of C. manganoxidans, Comamonas sp., C. testosteroni, and Schlegella sp. KB1a are BAA92354, AAA87070, BAA22882, and AAT09963, respectively. (b) The sequence of P. funiculosum enzyme is circularly permuted ([CP]) so that Ser205 is an Nterminal residue. A linkage site of the original N and C termini (G-339, 21-TAL) is indicated in orange. The sequence accession codes for Pausimonus lemoignei PhaZ3, PhaZ1, and PhaZ2 are AAB48166, CAA80310, and AAB17150, respectively.

homology (Figure 3(b)). Thus, both type I and II enzymes are related by an unexpected evolutionary relationship, that is, the type II depolymerase could have been derived from the type I enzyme, probably by gene duplication.22 Therefore, it is very likely that the structure of the type I catalytic domain resembles that of the P. funiculosum

enzyme, and exhibits an a/b hydrolase fold. A structural comparison using DALI23 resulted in a number of a/b hydrolase fold enzymes, including carboxylesterase from Pseudomonas fluorescens24 with the highest similarity of ZZ18.6. However, PHB depolymerase is largely different from these structural homologs in regions between b2 and a2,

997

Crystal Structure of PHB Depolymerase

b3 and a3, and b10 and a9, which contribute to the substrate-binding and the polymer-adsorption sites of the depolymerase (see below). The catalytic site It has been proposed that both type I and type II depolymerases contain lipase-like catalytic triads (serine, histidine, and aspartic acid residues).6 The catalytic residues of P. funiculosum depolymerase, Ser39, Asp121, and His155, are located after the first (b1), third (b3), and fourth (b4) b-strands, which are referred to as the nucleophilic elbow, the acidic turn, and the histidine loop for the canonical a/b hydrolase fold, respectively (Figures 1 and 2(a)). The main chain conformation of Ser39 with (f,j) values of (56.68, K128.38) exhibits characteristics typical for a nucleophilic residue of all of a/ b hydrolase fold enzymes. These residues are entirely conserved (Figure 3(a)), indicating that they are essential for catalysis, and the functional importance of Ser39 was confirmed in mutagenesis experiments. In spite of the circularly permuted fold, the spatial arrangement of the catalytic triad is very similar to that of lipases and other a/b hydrolase fold enzymes. The Nd1 atom of His155 and the carboxyl group of Asp121 are within ˚ ), whereas hydrogen bonding distance (about 2.8 A the Od atom of Ser39 and the N32 atom of His155 are not hydrogen bonded in the current model ˚ ). On the other hand, the Od atom of (about 3.9 A Ser39 is hydrogen bonded to the main chain amide group of Ser40, and a water molecule. This water molecule is also hydrogen bonded to the N32 atom of His155.

The substrate-binding site To understand the mechanism of substrate recognition by the enzyme as well as the catalytic mechanism on the basis of protein structure, we prepared a mutant protein S39A, and determined the structure of the S39A–H(R3HB)3M complex at a ˚ with crystallographic R and resolution of 1.66 A free-R values of 17.9% and 20.7%, respectively. The structures of the wild-type protein and the S39A mutant complexed with H(R3HB)3M are essentially ˚ for the same with an r.m.s. deviation of about 0.4 A a C atoms. No significant global structural change upon binding of H(R3HB)3M was observed. However, local conformational changes are observed for regions of the crystal contacts that are different between crystals of the wild-type and the mutant proteins due to crystallization in different space groups. In fact, the structure of the S39A protein complexed with H(R3HB)3M is almost identical to that in the uncomplexed form (data not shown). Therefore, these conformational differences are artifacts, produced during crystallization. Accordingly, although a slight difference of the conformation of Ser39 is observed, with a deviation of ˚ for the Ca position, this may be an about 0.54 A artifact due to different environments in terms of crystal contacts around the active site. Well-defined electron densities for H(R3HB)3M were obtained for the mutant–substrate complex data (Figure 4), which permitted us to build a reliable model of the bound substrate, and to investigate a substratebinding site and the binding mode of the substrate. We designate the monomer units of H(R3HB)3M as units 1–3 from its carboxyl end, and the corresponding sites (subsites) to which these monomer units

Figure 4. Simulated-annealing jmFoKDFcj electron density map for H(R3HB)3M bound to the S39A mutant, calculated from the model excluding H(R3HB)3M. Models of S39A and the R3HB trimer (H(R3HB)3M lacking the methyl group) are also shown as a line and a stick model, respectively. The map is contoured at 3.5s. Figures were produced using the PyMOL program (DeLano, W. L.; http://www.pymol.org/).

998 are bound as subsites 1–3, respectively. It should be noted that electron densities for the carboxyterminal methyl group of H(R3HB)3M were not visible, probably due to disorder. A crevice is formed on the surface of the enzyme, surrounded by structural elements between b2 and a2, b3 and a3, b5 and b6, and b11 and a9 (Figure 1). The catalytic residues Ser39, Asp121, and His155 are located in the crevice. The inside of the crevice serves as a substrate-binding site, which is of ˚ ) and width (w6 A ˚ ) to sufficient length (w15 A permit the incorporation of a single chain of the polymer (see Figure 7). A number of hydrophobic residues (Leu38, Tyr43, Tyr76, Thr123, Val124, Leu251, Tyr176, Leu298, Pro301, Trp307, Val308, and Trp310) as well as a cystine (Cys250–Cys304) are located inside the crevice, providing a hydrophobic environment favorable for the binding of polymer chains (Figure 5). These residues except for Leu251 are highly conserved in both type II and type I depolymerases (Figures 3), indicating that they are essential for interaction with PHB chains, and that they define the substrate specificity of the enzyme toward small side-chains of PHB. The structure of the S39A–H(R3HB)3M complex reveals three subsites for binding each monomer unit of the polymer substrate. Two methyl groups (side-chains) of units 1 and 2 of H(R3HB)3M interact with hydrophobic pockets of the substrate-binding site formed by Tyr43, Val124, and Trp307 (subsite 1), and Trp307, Tyr76, Val308, and Trp310 (subsite 2) (Figure 5). The size and shape of the two

Crystal Structure of PHB Depolymerase

hydrophobic pockets are such that they may be unable to accommodate side-chains larger than an ethyl group, consistent with the observed substrate specificity of this enzyme toward poly(3-hydroxyvalerate).19 The carbonyl oxygen of unit 2 is hydrogen bonded to the N31 of Trp307, which is the only hydrophilic interaction between H(R3HB)3M and the protein. On the other hand, the carbonyl oxygen of unit 3 is hydrogen bonded to a water molecule, which is further hydrogen bonded to Od1 of Asn300 and the main chain amide group of Asn302. These interactions define the substrate specificity of the enzyme toward polyesters with monomer units of three carbon atoms as a main chain. The binding site for unit 3 of H(R3HB)3M is situated at the surface of the protein, and is not clearly shaped compared to those for units 1 and 2. Unit 3 is surrounded by hydrophobic residues Tyr76, Leu298, Pro301, and Trp310, in addition to hydrophilic residues Ser299 and Asn302. However, the methyl group of unit 3 is exposed to bulk solvent, and is no longer embedded in the protein, unlike those of units 1 and 2. Thus it appears that the affinity of this subsite for a monomer unit is weaker than in the case of the other subsites. It should be noted that unit 3 is further surrounded by Asn314, Gly319, and Val321 of a neighboring protein molecule in the crystal lattice. As a result, the affinity of this site for unit 3 may be increased, which may be the reason for why electron densities for unit 3 are clearly observed in our X-ray data.

Figure 5. A close-up stereoview of the active site (the S39A–H(R3HB)3M complex). Side-chains of catalytic residues and residues forming the substrate-binding site in the crevice are represented as ball-and-stick models. The amide nitrogen atoms of Ser40, Cys250, and Asn302 are also shown. A model of the R3HB trimer (designated as (R3HB)3) is also shown as a black ball-and-stick model. A hydrogen bond between Nd1 atom of Trp307 and the carbonyl oxygen of an ester linkage adjacent to a scissile linkage, and one between His155 and Asp121 are indicated as green lines. The sidechain of Ser39 of the wild-type protein (shown in red) is superimposed on Ala39 of the mutant model, and a hydrogen bond between the Og atom of Ser39 and the amide group of Ser40 is tentatively indicated with a green broken line.

Crystal Structure of PHB Depolymerase

999

On the other hand, there should be a subsite (K1) to which a monomer unit linked with unit 1 via the scissile bond should be bound. Unfortunately, the disordered methyl group of H(R3HB)3M esterified with unit 1 does not provide sufficient information for subsite K1. Inspection of the structure indicates that hydrophobic residues Tyr176, Cys250, and Leu251 could serve as subsite K1. Implication for the catalytic mechanism The spatial arrangement of the catalytic residues indicates that the mechanism of the depolymerase reaction may be similar to that of a lipase/serine esterase.25 Thus it is possible that transient tetrahedral coordination of the carbonyl carbon atom could occur during the course of the catalytic reaction, and it could be stabilized by positively charged groups that make up an oxyanion hole. Inspection of the active site architecture indicates that the main chain amide group of Ser40 is part of the oxyanion hole. Another group that could potentially contribute to the oxyanion hole is the main chain amide group of Cys250. The catalytic reaction may thus proceed as follows (Figure 6): Ser39 plays a central role in the catalytic reaction, participating in a nucleophilic attack on the carbonyl carbon atom of a bound PHB chain. The His155–Asp121 hydrogen bonding system may enhance the nucleophilicity of the hydroxyl group of Ser39. Ser39 does not hydrogen bond to His155 in the ligand-free state, and it is likely that Ser39 changes its conformation upon binding of a substrate, forming a hydrogen bond with His155, and raising its nucleophilicity. A PHB chain is bound to the active site, with its four contiguous monomer units occupying the four subsites (K1 to 3) of the substrate-binding site. A carbonyl group next to the scissile ester linkage is hydrogen bonded to Trp307, which enables the carbonyl carbon of the scissile linkage to be located at a position that would permit it to be attacked by the hydroxyl group of Ser39. A tetrahedral coordination at the carbonyl carbon atom is transiently formed, which is stabilized by the oxyanion hole formed by the amide groups of Ser40 and Cys250. After a rearrangement of the bond formation, a covalent enzyme–substrate intermediate is subsequently formed, releasing a first product, with a hydroxyl group at the end of the chain. The intermediate is then hydrolyzed by the nucleophilic attack of a water molecule activated by the His–Asp system. The transiently formed tetrahedral coordination is again stabilized by the oxyanion hole. After bond rearrangement, the hydroxyl group of Ser39 is regenerated and a second product, with a carboxyl group at the end of the chain, is released, thus completing one cycle of the catalytic reaction. Note that His248, which is generally believed to form part of the oxyanion hole based on sequence conservation and the mechanistic analogy of the depolymerase to lipase,6 is located at the C-terminal end of strand b9, and neighbors the active site,

Figure 6. A proposed mechanism for the action of the PHB depolymerase from P. funiculosum. Upon binding of a PHB chain to the substrate-binding site, Ser39 may change its conformation to form a hydrogen bond with His155. Four contiguous monomer units (K1 to 3) are bound to the subsites, making predominantly hydrophobic interactions with the enzyme, with concomitant hydrophilic interactions. The carbonyl oxygen of unit 2 is hydrogen bonded to N31 of Trp307, and that of unit 3 is hydrogen bonded to Od1 of Asn300 and the main chain nitrogen of Asn302 via a water molecule. The methyl groups of units 1 and 2 are specifically bound to the hydrophobic pockets of the enzyme (see the text). The scissile bond is located between units K1 and 1. Nucleophilic attack on the carbonyl carbon involved in the scissile bond by Ser39 results in a formation of a covalent acyl-enzyme intermediate via a tetrahedral transition state, which is stabilized by two main chain amide groups of Ser40 and Cys250. The acyl-enzyme intermediate is subsequently hydrolyzed by a water molecule activated by His135 via formation of a similar tetrahedral transition state. After regeneration of Ser39, or liberation of the remaining product, a conformational change of Ser39 may occur, which disrupts a hydrogen bond with His155 and forms one with the amide group of Ser40.

1000 hydrogen bonding to the main chain carbonyl oxygen of Gly249 and the amide group of Ala283. Its side-chain does not point to the active site, and, therefore, it is unlikely that it would participate in catalysis. Solvent-exposed hydrophobic residues form a polymer-adsorption site To efficiently hydrolyze a water-insoluble substrate, it may be essential for the enzyme to reside on the surface for a period of time. In the case of a multi-domain depolymerase, the enzyme can reside on the surface of PHB via the substrate-binding domain.10,12,15–17 The single domain depolymerase from P. funiculosum lacks such a functionally distinct domain, but retains the ability to reversibly bind to polymers.19 Therefore it must have a region on its molecular surface for binding to the hydrophobic surface of a polymer. An inspection of the depolymerase structure revealed that a number of hydrophobic residues including threonine residues (Tyr75, Thr77, Tyr81, Tyr84, Thr122, Leu171, Thr173, Leu251, Tyr287, Ile289, Val297, Leu298, and Pro301) are aligned around the mouth of the crevice, and are exposed to the bulk solvent (it is assumed that threonine residues may play a dual role in hydrophobic and hydrophilic interactions with PHB) (Figure 7). It is likely that these residues contribute to adsorption to the surface of a PHB via hydrophobic interactions. Interestingly, these residues (except for Tyr84) are

Figure 7. A molecular surface representation of the P. funiculosum depolymerase centering on the mouth of the crevice. Positions of solvent-exposed hydrophobic residues (purple), as well as polar (green) and catalytic triad (cyan) residues, are indicated. Residues Leu251, Leu298, Ser299, Pro301, and Asn302 also contribute to the substrate-binding site. A model of the R3HB trimer bound in the crevice is shown as a yellow stick model.

Crystal Structure of PHB Depolymerase

aligned in a plane (Figure 7). A sequence alignment of type II depolymerases with a high homology (about 50%) reveals differences between single and multi-domain depolymerases with respect to these solvent-exposed residues (Figure 3(a)). Of these residues, Thr77, Tyr81, Thr122, Leu298, and Pro301 are highly conserved, or are replaced by other hydrophobic residues, whereas residues Tyr75, Leu171, Thr173, Leu251, Tyr287, and Ile289 are replaced by hydrophilic residues in several homologs with a multi-domain structure. It is possible that, in the case of the P. funiculosum enzyme, these residues collectively contribute to the increased ability of the catalytic domain to bind to the hydrophobic surface of the PHB, while for other homologs, the lower population of solvent-exposed hydrophobic residues results in an inability to adsorb to the surface of PHB.8,26 It is noteworthy that the hydrophilic residues that are located around the mouth of the crevice are Ser78, Asn82, Ser120, Ser168, Ser170, Ser174, and Asn302, and that they are highly conserved (Asn82, Ser174, and Asn302) or replaced by asparagine, threonine, glycine, or alanine residues, but not by hydrophobic residues in homologs. In the P. funiculosum enzyme, they may contribute to the weak affinity of this enzyme for PHB.19 The issue of whether the adsorption site contributes to the destabilization of ordered regions of PHB chains is an interesting topic. Studies of the degradation of PHB by multi-domain depolymerases10,12,15–17 indicate that the substrate-binding domain may play a critical role in introducing mechanical damage to ordered regions of the polymer upon binding to that region, as well as providing the catalytic domain a toehold to facilitate the hydrolysis. The introduction of mechanical damage may be the result of strong hydrophobic interactions between the substrate-binding domain and PHB chains, which destabilize the hydrophobic intra-chain and inter-chain interactions of the PHB chains. Atomic force microscopy observations of the degradation of lamellar crystals of PHB by the P. funiculosum enzyme showed that hollows occur at specific regions in a lamellar crystal of PHB, followed by fragmentation of the crystal (K. K. et al., unpublished data), which is a degradation process similar to that observed in the case of multi-domain enzymes. Thus it is likely that, although the effect may be smaller than the substrate-binding domain of multi-domain enzymes, as indicated by the weakness of the affinity for PHB, the adsorption site of the singledomain enzyme should introduce mechanical damage to ordered regions of PHB chains via hydrophobic interactions, thus elevating the mobility of PHB chains and creating a susceptible site for hydrolysis by the catalytic site of the enzyme, resulting in the efficient degradation of PHB. It appears that the cooperation of the nonhydrolytic action of the substrate-binding domain and the hydrolytic action of the catalytic domain is crucial for degradation of crystalline substrates.

1001

Crystal Structure of PHB Depolymerase

This mechanistic model was originally proposed by Reese et al.27 concerning degradation of cellulose, and revised by Din et al.28 It is interesting to note that there are similarities between PHB depolymerases and cellulolytic enzymes of the uncomplexed system29 in the modular structure and the process of degradation of insoluble polymer substrates with a high degree of crystallinity (typical ranges are 40–90% for cellulose,30 and 55 to 80% for PHB31). However, the single-domain PHB depolymerase is different from cellulolytic enzymes that lack cellulose-binding modules (e.g., endoglucanase III of Trichoderma reesei 32 and Humicola insolens33), in that these cellulolytic enzymes alone are not able to degrade crystalline cellulose completely.32,33 These single-domain cellulases may not need a cellulose-adsorption site on their molecular surface, probably because they usually work synergistically with other cellulolytic enzymes. On the other hand, the single-domain PHB depolymerase can degrade PHB completely, which can be attributed to its ability to interact with PHB via the PHB-adsorption site. In conclusion, the crystal structure of PHB depolymerase from P. funiculosum represents a novel member of the a/b hydrolase fold enzymes, with a circularly permuted polypeptide linkage. The data show that the enzyme is comprised of a single domain. Catalytic residues Ser39, Asp121, and His155 are located at topologically conserved positions. Amide groups of Ser40 and Cys250 may serve as an oxyanion hole. The substrate-binding site probably contains four subsites. Subsites 1 and 2 provide hydrophobic pockets for binding sidechains of substrates, and define specificity toward short-chain-length PHA. Trp307 serves as a recognition site for an ester group next to the scissile bond. The structure also reveals the PHB-adsorption site, formed by a number of solvent-exposed hydrophobic residues aligned around the cavity. The PHB-adsorption site contributes to the efficient binding affinity of the enzyme to PHB granules, which aids in the degradation of insoluble PHB granules without a substrate-binding domain. This is the first crystal structure of an enzyme that catalyzes the hydrolysis of microbial biopolyesters, and the information herein will facilitate further the development of biopolyester engineering based on enzyme–polymer interactions, toward the design of high-performance polymers with the simultaneous achievement of biodegradability and superior mechanical properties.

Materials and Methods Materials A methyl ester of a trimer of R3HB (methyl (3R)-3[(3R)-3-{(3R)-3-hydroxybutanoyloxy}butanoyloxy]butanoate; H(R3HB)3M) was synthesized as described.34

Preparation and crystallization of the wild-type enzyme The enzyme from P. funiculosum IFO 6345 was purified from the culture medium as described.19 Crystals were obtained by sitting-drop vapor-diffusion equilibrated against a reservoir containing 20% (w/v) polyethylene glycol 8000, 50 mM potassium dihydrogenphosphate, and 5% (v/v) glycerol, at 21 8C. The crystals belonged to the monoclinic space group P21 with cell dimensions aZ ˚ , and bZ107.98, and contained one 49.5, bZ68.9, cZ38.7 A molecule per asymmetric unit. They diffracted X-rays up ˚ resolution. Platinum-derivatized crystals were to a 1.7 A prepared by soaking native crystals in a solution containing 28% polyethylene glycol 8000, 0.1 M Mes (pH 6.0), 5% glycerol, and 10 mM K2PtCl4. Data collection Diffraction intensities were collected using a MarCCD detector at beamline BL44B2, SPring-8, Harima, Japan.35 A crystal was soaked for a few seconds in a cryoprotectant solution containing 28% polyethylene glycol 8000, 0.1 M Mes (pH 6.0), and 20% glycerol, and flash-frozen in a stream of nitrogen gas at 90 K. Diffraction intensities were recorded in a total of 180 frames. Data reduction was conducted using the HKL2000 program.36 Data collection statistics are summarized in Table 1. Phasing and refinement Phases were obtained by the single isomorphous replacement method with the incorporation of anomalous information from platinum atoms. Five platinum sites were identified and phases were calculated using the SOLVE program.37 Phases were improved and the initial model was built using the RESOLVE program.38,39 Manual rebuilding and refinement were conducted in several cycles using the XTALVIEW40 and CNS41 programs, respectively. The amino acid sequence used for model building was deduced from the nucleotide sequence (K. K., unpublished data). Crystallographic R Table 1. Data collection statistics Data set

Wild-type

Ptderivative

S39A– H(R3HB)3M

˚ Wavelength, A ˚ Resolution, A

1.0000 1.71 (1.78–1.71) P21

1.0720 2.17 (2.20–2.17) P21

1.0000 1.66 (1.72–1.66) C2

49.5 68.9 38.7 107.9 189,730

49.7 68.8 39.0 108.1 145,589

81.9 48.9 64.2 94.5 214,183

25,930

24,719

29,362

97.6 (75.6) 28.4 (3.1) 0.072 (0.240) 0.23

96.0 (75.4) 25.2 (6.0) 0.058 (0.162) 0.55

98.3 (89.9) 25.9 (5.0) 0.084 (0.262) 0.50

Space group Cell parameters ˚) a (A ˚) b (A ˚) c (A b (deg.) No. of measurements No. of unique reflections Completeness (%) I/s(I) Rmergea Mosaicity

Values in parentheses P P are for the outermost P Presolution shell. a Rmerge Z hkl i jIi ðhklÞKhIðhklÞij= hkl i Ii ðhklÞ, where Ii is the intensity of the ith observation and hIi is the mean intensity of the reflection.

1002 Table 2. Refinement statistics Data set

Wild-type S39A–H(R3HB)3M ˚ Resolution range (A) 47.1–1.71 40.8–1.66 No. of reflections 25,935 29,362 No. of atoms Protein 2,355 2,355 Sugar 61 14 Water 228 256 – 19 (R3HB)3 19.0/22.1 17.9/20.7 Rcrystallogr/Rfreea (%) r.m.s. deviations ˚) Bonds (A 0.005 0.005 Angles (8) 1.3 1.3 ˚ 2) Average B-factors (A Main chain 15.2 8.4 Side-chain 16.8 11.3 Sugar 35.1 13.8 Water 24.4 18.9 – 21.1 (R3HB)3 P P a Rcryst Z hkl jjFobs jðhklÞKjFcalc jðhklÞj= hkl jFobs jðhklÞ, and Rfree is a cross-validated R-factor.

and free-R (calculated with 10% of the total reflections) factors were converged to 19.0 and 22.1%, respectively, ˚ resolution. The current model against data up to a 1.71 A consists of 318 residues (Thr21–Gln338), a bifurcated oligosaccharide chain (GlcNAc2Man3) bound to Asn144, and 228 water molecules. The C-terminal residue Gly339 was not clearly defined in the electron density, probably due to disorder, and was excluded from the model. The refined model has reasonable stereochemistry, as verified with the PROCHECK program42 giving 0.4% (one residue) of the non-glycine residues in the generously allowed region of the Ramachandran plot. The outlier Ser39, one of the catalytic residues, is well defined in electron density, and its main chain conformation ((f,j)Z (56.68, K128.38)) is characteristics of this class of protein fold (the a/b hydrolase fold20). Refinement statistics are summarized in Table 2. Preparation of the S39A mutant protein Because attempts to purify recombinant proteins (either wild-type or mutant) from Escherichia coli cells have been unsuccessful due to the lack of efficiency of protein expression in that host cell, we prepared the mutant protein using a recombinant yeast as a host, according to manufacturers 0 protocols. A cDNA fragment of a gene encoding PHB depolymerase (phaZPfu) was amplified with a pair of primers 5 0 -GCAGGAATTCT AACGGCCCTACCTGCCTTC-3 0 (pPICZaNEco) and 5 0 TACTGAATTCCTAGCCTTGAAAGCCACTGACAAT-3 0 (pPICZaCEco), and recombinant plasmid pUC18 carrying phaZPfu at the EcoRI site (designated as pUCPFDP) as a template. Site-directed mutagenesis was performed using an LA PCRTM in vitro mutagenesis kit (TaKaRa Bio Inc, Kyoto, Japan). For mutation of the nucleophilic Ser39 to Ala, a synthetic oligonucleotide 5 0 -GCTAGCTAGAC CAGAGACAG-3 0 was used as a primer. The template for the mutagenesis was obtained by PCR using a pair of primers pPICZaNEco and pPICZaCEco, pUCPFDP as a template, and Deep Vent DNA polymerase (New England Biolabs, MA, USA). The PCR product was ligated to SmaI–digested pUC18, yielding a plasmid pUCPFDPTemp, which contained a complete phaZPfu gene. The mutated DNA fragment was inserted into pUC18, and

Crystal Structure of PHB Depolymerase

was transformed into E. coli DH5a. The mutated DNA fragment was amplified by the PCR method with a set of primers (pPICZaNEco and pPICZaCEco). EcoRIdigested PCR product was ligated into EcoRI-digested and dephosphorylated pPICZaB (Invitrogen Japan KK, Tokyo, Japan), yielding a plasmid, pPICZS39A. pPICZS39A was cloned in Pichia pastoris KM71H. P. pastoris carrying the phaZPfu gene was cultivated for three days at 30 8C in 1 l of BMGH medium under aerobic conditions on a reciprocal shaker (120 strokes/min) in ten 500-ml Sakaguchi flasks. Methanol was added at 24 h intervals, to maintain the concentration at 0.5%. The liquid culture of recombinant P. pastoris was centrifuged in a KUBOTA AG-5006 rotor at 10,000g for 10 min. The supernatant was filtrated by a mixed cellulose membrane filter with 0.45 mm pore size (Toyo Roshi Kaisha. Ltd., Tokyo, Japan) to remove contaminating mycelium. The mutant protein was purified by hydrophobic column chromatography using a Phenyl Toyopearl column (Tosoh, Tokyo, Japan) from the supernatant according to the same protocol as was used for the wild-type protein.19 Immunological cross-reactivity using an anti-PhaZPfu antibody confirmed the purified mutant depolymerase. The amino-terminal sequence, determined by Edman degradation, was AGILTALPAF, identical to that of the wild-type protein with an extra tetrapeptide (underlined), which was artificially added as a result of the mutagenesis procedure. The activity of the mutant protein was measured by the same method as that for the wild-type protein,19 resulting in no hydrolytic activity toward P(3HB) or p-nitrophenylbutyrate. Crystallization, data collection, and structure refinement of the S39A–substrate complex Crystals of the S39A protein were obtained by sittingdrop vapor-diffusion equilibrated against a reservoir containing 2.0 M sodium chloride, 100 mM acetate– NaOH buffer (pH 4.5), and 5% (v/v) 1,4-dioxane, at 21 8C. The crystals belonged to the monoclinic space ˚, group C2 with cell dimensions aZ82.0, bZ48.7, cZ64.8 A and bZ93.68, and contained one molecule per asymmetric unit. Crystals of the S39A protein–substrate complex were obtained by soaking S39A protein crystals in a solution containing 2.5 M sodium chloride, 100 mM acetate– NaOH buffer (pH 4.5), 5% 1,4-dioxane, 20% glycerol, and 10 mM H(R3HB)3M for about 5 min. They diffracted ˚ resolution. Data collection, reduction, X-rays up to 1.6 A and structure refinement were conducted using the same procedures as those for the native data. An initial model was obtained by the molecular replacement method using the coordinates of the wild-type enzyme as a search model. Crystallographic R and free-R (calculated with 10% of the total reflections) factors were converged to ˚ 17.9% and 20.7%, respectively, against data up to a 1.66 A resolution. The current model consists of 318 residues (Thr21–Gln338), one N-acetylglucosamine bound to Asn144, one trimer of R3HB (H(R3HB)3M lacking the carboxy-terminal methyl group), and 256 water molecules. N-terminal four residues, which are artificially added in the mutant protein due to the experimental design of the mutation, were not clearly visible in the electron density map. The refined model has reasonable stereochemistry, as verified with the PROCHECK program,42 giving 0.4% (one residue: Ala39, the mutation site) of the non-glycine residues in the generously allowed region of the Ramachandran plot. Data collection and refinement statistics are summarized in Tables 1 and 2, respectively.

Crystal Structure of PHB Depolymerase

Protein Data Bank accession numbers The atomic coordinates have been deposited in the RCSB Protein Data Bank† with accession codes 2D80 and 2D81 for the wild-type enzyme and the S39A–R3HB complex, respectively.

Acknowledgements We thank Yoko Yamagata for technical assistance, the staff of beamline BL44B2 at SPring-8 for help in data collection, and Jun Li for his contribution in the synthesis of the R3HB trimer methyl ester. We also thank Noritake Yasuoka for critical reading of the manuscript and invaluable discussions, and Christopher T. Nomura and Milton S. Feather for editing the manuscript. This work was supported, in part, by a grant from Ecomolecular Science Research II provided to the RIKEN Institute, by U.S. National Science Foundation MCB0236093 (E.O.), by a Grant-in-aid for the High-Tech Research Center Project (2002) from the Ministry of Education, Culture, Sports, Science, and Technology (MEXT) of Japan (to T.S.), and by a Grant-in-Aid for Young Scientists (A) from MEXT of Japan (Nos. 15685009 (to T.I.) and 17681009 (to T.H.)).

References 1. Anderson, A. J. & Dawes, E. A. (1991). Occurrence, metabolism, metabolic role, and industrial uses of bacterial polyhydroxyalkanoates. Microbiol. Rev. 54, 450–472. 2. Madison, L. L. & Huisman, G. W. (1999). Metabolic engineering of poly(3-hydroxyalkanoates): from DNA to plastic. Microbiol. Mol. Biol. Rev. 63, 21–53. 3. Doi, Y. (1990). Microbial Polyesters, VCH Publishers, New York. 4. Lenz, R. W. & Marchessault, R. H. (2005). Bacterial polyesters: biosynthesis, biodegradable plastics and biotechnology. Biomacromolecules, 6, 1–8. 5. Aldor, I. S. & Keasling, J. D. (2003). Process designed for microbial plastic factories: metabolic engineering of polyhydroxyalkanoates. Curr. Opin. Biotechnol. 14, 475–483. 6. Jendrossek, D. & Handrick, R. (2003). Microbial degradation of polyhydroxyalkanoates. Annu. Rev. Microbiol. 56, 403–432. 7. Kim, D. Y. & Rhee, Y. H. (2003). Biodegradation of microbial and synthetic polyesters by fungi. Appl. Microbiol. Biotechnol. 61, 300–308. 8. Fukui, T., Narikawa, T., Miwa, K., Shirakawa, Y., Saito, T. & Tomita, K. (1988). Effect of limited tryptic modification of a bacterial poly(3-hydroxybutyrate) depolymerase on its catalytic activity. Biochem. Biophys. Acta, 952, 164–171. 9. Kasuya, K., Ohura, T., Masuda, K. & Doi, Y. (1999). Substrate and binding specificities of bacterial polyhydroxybutyrate depolymerase. Int. J. Biol. Macromol. 24, 329–336. † http://www.rcsb.org/

1003 10. Murase, T., Suzuki, Y., Doi, Y. & Iwata, T. (2002). Nonhydrolytic fragmentation of a poly[(R)-3-hydroxybutyrate] single crystal revealed by use of a mutant of polyhydroxybutyrate depolymerase. Biomacromolecules, 3, 312–317. 11. Kikkawa, Y., Fujita, M., Hiraishi, T., Yoshimoto, M. & Doi, Y. (2004). Direct observation of poly(3-hydroxybutyrate) depolymerase adsorbed on polyester thin film by atomic force microscopy. Biomacromolecules, 5, 1642–1646. 12. Nobes, G. A. R., Marchessault, R. H., Chanzy, H., Briese, B. H. & Jendrossek, D. (1996). Splintering of poly(3-hydrxybutyrate) single crystals by PHB-depolymerase A from Pseudomonas lemoignei. Macromolecules, 29, 8330–8333. 13. Nojiri, M. & Saito, T. (1997). Structure and function of poly(3-hydroxybutyrate) depolymerase from Alcaligenes faeccalis T1. J. Bacteriol. 179, 6965–6970. 14. Kataeva, I. A., Seidel, R. D., III, Shah, A., West, L. T., Li, X.-L. & Ljungdahl, L. G. (2002). The fibronection type 3-like repeat from the Clostridium thermocellum cellobiohydrolase CbhA promotes hydrolysis of cellulose by modifying its surface. Appl. Environ. Microbiol. 68, 4292–4300. 15. Iwata, T., Doi, Y., Kasuya, K. & Inoue, Y. (1997). Visualization of enzymatic degradation of poly[(R)-3hydroxybutyrate] single crystals by an extracellular PHB depolymerase. Macromolecules, 30, 833–839. 16. Iwata, T., Doi, Y., Tanaka, T., Akehata, T., Shiromo, M. & Teramachi, S. (1997). Enzymatic degradation and adsorption on poly[(R)-3-hydroxybutyrate] single crystals with two types of extracellular PHB depolymerases from Comamonas acidovorans YM1609 and Alcaligenes faecalis T1. Macromolecules, 30, 5290–5296. 17. Murase, T., Iwata, T. & Doi, Y. (2001). Atomic force microscopy investigation of poly[(R)-3-hydroxybutyrate] lamellar single crystals: relationship between molecular weight and enzymatic degradation behavior. Macromol. Biosci. 1, 275–281. 18. Brucato, C. L. & Wong, S. S. (1991). Extracellular poly(3-hydroxybutyrate) depolymerase from Penicillium funiculosum: general characteristics and active site studies. Arch. Biochem. Biophys. 290, 497–502. 19. Miyazaki, S., Takahashi, K., Shiraki, M., Saito, T., Tezuka, Y. & Kasuya, K. (2002). Properties of a poly(3hydroxybbutyrate) depolymerase from Penicillium funiculosum. J. Polym. Environ. 8, 175–182. 20. Ollis, D. I., Cheah, E., Cygler, M., Dijkstra, B., Frolow, F., Franken, S. M. et al. (1992). The a/b hydrolase fold. Protein Eng. 5, 197–211. 21. Lindqvist, Y. & Schneider, G. (1997). Circular permutations of natural protein sequences: structural evidence. Curr. Opin. Struct. Biol. 7, 422–427. 22. Jeltsh, A. (1999). Circular permutation in the molecular evolution of DNA methyltransferases. J. Mol. Evol. 49, 161–164. 23. Holm, L. & Sander, C. (1996). Alignment of threedimensional protein structures. Methods Enzymol. 266, 653–662. 24. Kim, K. K., Song, H. K., Shin, D. H., Hwang, K. Y., Choe, S., Yoo, O. J. & Suh, S. W. (1997). Crystal structure of carboxylesterase from Pseudomonas fluorescence, an a/b hydrolase with broad substrate specificity. Structure, 5, 1571–1584. 25. Derewenda, Z. S. (1994). Structure and function of lipases. Advan. Protein Chem. 45, 1–52. 26. Kasuya, K., Takano, T., Tezuka, Y., Hsieh, W.-C., Mitomo, H. & Doi, Y. (2003). Cloning, expression and

1004

27.

28.

29.

30. 31.

32.

33. 34.

35.

Crystal Structure of PHB Depolymerase

characterization of a poly(3-hydroxybutyrate) depolymerase from Marinobacter sp. NK-1. Int. J. Biol. Macromol. 33, 221–226. Reese, E. T., Sui, R. G. H. & Levinson, H. S. (1950). The biological degradation of soluble cellulose derivatives and its relationship to the mechanism of cellulose hydrolysis. J. Bacteriol. 59, 485–497. Din, N., Damude, H. G., Gilkes, N. R., Miller, R. C., Jr, Warren, R. A. J. & Kilburn, D. G. (1994). C1–Cx revised: intramolecular synergism in a cellulose. Proc. Natl Acad. Sci. USA, 91, 11383–11387. Lynd, L. R., Weimer, P. J., van Zyl, W. H. & Pretorius, I. S. (2002). Microbial cellulose utilization: fundamentals and biotechnology. Microbiol. Mol. Biol. Rev. 66, 506–577. O’Sullivan, A. C. (1997). Cellulose: the structure slowly unravels. Cellulose, 4, 173–207. Holms, P. A. (1998). Biologically produced (R)-3hydroxyalkanoate polymers and copolymers. In Developments in Crystalline Polymers (Bassett, D. C., ed.), vol. 2, pp. 1–65, Elsevier, London. Sprey, B. & Bochem, H.-P. (1992). Effect of endoglucanase and cellobiohydrolase from Trichoderma reesei on cellulose microfibril structure. FEMS Microbiol. Letters, 97, 113–118. Schu¨lein, M. (1997). Enzymatic properties of cellulases from Humicola insolens. J. Biotechnol. 57, 71–81. Li, J., Uzawa, J. & Doi, Y. (1998). Conformational analysis of oligomers of (R)-3-hydroxybutanoic acid in solution by 1H NMR spectroscopy. Bull. Chem. Soc. Jpn. 71, 1683–1689. Adachi, S., Oguchi, T., Tanida, H., Park, S.-Y., Shimizu, H., Miyatake, H. et al. (2001). The RIKEN

36. 37. 38. 39. 40. 41.

42.

43. 44.

structural biology beamline II (BL44B2) at SPring-8. Nucl. Instrum. Methods Phys. Res. ser. A, 467–468, 711–714. 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. Terwilliger, T. C. (2002). Automated main-chain model-building by template-matching and iterative fragment extension. Acta Crystallog. sect. D, 59, 34–44. McRee, D. E. (1999). Practical Protein Crystallography 2nd edit., Academic Press, San Diego. Bru¨nger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). Crystallography and NMR system (CNS): a new software system for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905–921. 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. Crystallog. 26, 283–291. Kraulis, P. J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallog. 24, 946–950. Merrit, E. A. & Bacon, D. J. (1997). Raster3D: photorealistic molecular graphics. Methods Enzymol. 277, 505–524.

Edited by M. Guss (Received 27 September 2005; received in revised form 5 December 2005; accepted 7 December 2005) Available online 27 December 2005