Crystal structure of 2-hydroxyl-6-oxo-6-phenylhexa-2,4-dienoic acid (HPDA) hydrolase (BphD enzyme) from the Rhodococcus sp. strain RHA1 of the PCB degradation pathway1

doi:10.1006/jmbi.2001.4737 available online at http://www.idealibrary.com on

J. Mol. Biol. (2001) 309, 1139±1151

Crystal Structure of 2-Hydroxyl-6-oxo-6-phenylhexa2,4-dienoic Acid (HPDA) Hydrolase (BphD enzyme) from the Rhodococcus sp. Strain RHA1 of the PCB Degradation Pathway Narayanasamy Nandhagopal1, Akihiro Yamada2, Takashi Hatta3 Eiji Masai2, Masao Fukuda2, Yukio Mitsui1{ and Toshiya Senda1* 1

Division of Protein Engineering, and 2

Division of Microbial Engineering, Department of BioEngineering, Nagaoka University of Technology Nagaoka, Niigata 9402188, Japan 3

Research Institute of Technology, Okayama University of Science Okayama 703, Japan

2-Hydroxyl-6-oxo-6-phenylhexa-2,4-dienoic acid (HPDA) hydrolase (the BphD enzyme) hydrolyzes a ring-cleavage product of an aromatic compound generated in a biphenyl/polychlorinated biphenyl (PCB) degradation pathway of bacteria. The crystal structure of the BphD enzyme Ê resolution by the multiple isomorphous has been determined at 2.4 A replacement method. The ®nal re®ned model of the BphD enzyme yields Ê resolution with reasonable geometry. The an R-factor of 17.5 % at 2.4 A BphD enzyme is an octameric enzyme with a 422 point-group symmetry. The subunit can be divided into core and lid domains. The active site of the enzyme is situated in the substrate-binding pocket, which is located between the two domains. The substrate-binding pocket can be divided into hydrophobic and hydrophilic regions. This feature of the pocket seems to be necessary for substrate binding, as the substrate is composed of hydrophilic and hydrophobic parts. The proposed orientation of the substrate seems to be consistent with the general catalytic mechanism of a/b-hydrolases. # 2001 Academic Press

*Corresponding author

Keywords: X-ray crystallography; a/b-hydrolase; substrate speci®city; PCB degradation; structure function relationship

Introduction The halo aromatics, found in paints, refrigerants, and pesticides are dif®cult to break down and are usually harmful to most living systems. PolychloriPresent addresses: N. Nandhagopal, Department of Biological Sciences, 1392 Lilly Hall of Life Sciences, Purdue University, West Lafayette, IN 47907-1392, USA; T. Senda, Biological Information Research Center, National Institute of Advanced Industrial Science and Technology, Aomi, Kohto-ku, Tokyo 135-0006, Japan. {Deceased Abbreviations used: HPDA, 2-hydroxyl-6-oxo-6phenylhexa-2,4-dienoic acid; PCB, polychlorinated biphenyl; MCP, meta-cleavage product; HOHD, 2hydroxy-6-oxohepta-2,4-dienoic acid; CE, combination extension of the optimal path; VAST, vector alignment search tool; FAH, fumarylacetoacetate hydrolase; MIR, multiple isomorphous replacement; r.m.s., root-meansquare. E-mail address of the corresponding author: [email protected] 0022-2836/01/051139±13 $35.00/0

nated biphenyl (PCB) is one of the well-known harmful halo aromatic pollutants. PCBs have been utilized in various ways because of their thermal and chemical stability. Concurrently, PCBs have accumulated in the environment due to their chemical stability. In 1973, Ahmed & Focht discovered that bacteria can degrade PCBs.1 Since then, the use of bacteria has been one of the most attractive means of removing PCBs from the environment, and a large number of bacteria that can degrade PCBs have been isolated and characterized.2 ± 8 It has been proposed that bacteria degrade PCBs through their biphenyl degradation pathway.9,10 The degradation of biphenyl begins with the hydroxylation of an aromatic ring by a biphenyl dioxygenase. The resultant diol compound is converted to a catechol derivative, namely 2,3-dihydroxybiphenyl, by a dihydrodiol dehydrogenase. The 2,3-dihydroxybiphenyl is subjected to an extradiol ring-opening reaction (meta-cleavage reaction) by a dihydroxybiphenyl dioxygenase, resulting in the formation of 2-hydroxyl-6-oxo-6-phenylhexa# 2001 Academic Press

1140 2,4-dienoic acid (HPDA). The HPDA is hydrolyzed by meta-cleavage product hydrolase (MCP-hydrolase) (Figure 1). The resultant benzoic acid is degraded by another degradation pathway similar to the biphenyl degradation pathway. The 2-hydroxypenta-2,4-dienoic acid is eventually converted into the intermediates of the central metabolism through the tricarboxylic acid cycle.11 Rhodococcus sp. strain RHA1 is one of the strongest PCB degraders ever isolated.12,13 In this bacterium, biphenyl or ethylbenzene induces the expression of the enzymes involved in biphenyl/ PCB degradation.14 The induction results in simultaneous expression of the multiple isozymes involved in biphenyl and ethylbenzene degradation. The multiplicity of enzymes expressed in this bacterium seems to support its strong PCB degradation ability. In the meta-cleavage step, two isozymes, BphC and EtbC, are induced,14 and in the subsequent hydrolysis step, three isozymes, BphD, EtbD1, and EtbD2, are induced.15 The EtbC enzyme has an extremely wide substrate speci®city and can transform both catechol (one aromatic ring) and 2,3-dihydroxybiphenyl (two aromatic rings). In contrast, all the MCP-hydrolases in Rhodococcus sp. strain RHA1, BphD, EtbD1, and EtbD2, have narrow substrate speci®city. The BphD enzyme preferentially hydrolyzes HPDA derived from biphenyl, but cannot hydrolyze 2hydroxy-6-oxohepta-2,4-dienoic acid (HOHD), which is the meta-cleavage product originating from ethylbenzene. In contrast, EtbD1 and EtbD2 can hydrolyze HOHD, but not HPDA.15,16 Thus, the lack of a bphD gene causes an accumulation of HPDA in the cell, resulting in a failure of the catabolism of biphenyl.17 The introduction of the bphD gene into the bphD mutant strain results in a recovery of the capacity for biphenyl/PCB catabolism.18 The substrate speci®city of this enzyme is one of the important points in understanding the mechanism of PCB degradation in this bacterium. In order to determine why the MCP-hydrolases have such narrow substrate speci®city, we began to elucidate

Crystal Structure of BphD

the three-dimensional structure of the BphD enzyme derived from the Rhodococcus sp. strain RHA1. The RHA1 BphD enzyme is an octameric enzyme with a total molecular mass of ca 250 kDa.16 Each subunit is composed of 285 amino acid residues.17 The catalytic triad residues of this enzyme, Ser110, Asp235, and His263, are completely conserved among the related meta-cleavage product hydrolases. Furthermore, the amino acid sequence of the BphD enzyme suggests that this enzyme belongs to the a/b-hydrolase family, similar to lipase, haloalkane dehalogenase, etc.16 Here, we report for the ®rst time the crystal structure of the BphD enzyme from RHA1. The overall structure of the BphD enzyme is typical of the a/b-hydrolases and can be divided into two domains, with the active site of the enzyme being located between the two domains. The substratebinding pocket of the BphD enzyme is tubular in shape and consists of hydrophobic and hydrophilic parts. This character seems to determine the orientation of the bound substrate, which is consistent with the general catalytic mechanism of a/bhydrolases. The substrate speci®city of the MCPhydrolases will be discussed based on the sequence alignment of the related enzymes. A short preliminary report on the present structure has already been published.19

Results Quality of the refined structure After 15 cycles of re®nement, the crystalloÊ resgraphic R-factor for the BphD enzyme at 2.4 A olution was reduced to 17.5 %, incorporating 32 water molecules. The root-mean-square deviations Ê in bond lengths and from ideality were 0.007 A 1.2  in bond angles (Table 1). The ®nal model is composed of all but two amino acid residues at the N terminus, for which no corresponding electron densities were found in the 2Fo ÿ Fc map. The re®ned structure of the BphD enzyme has a good geometry. The polypeptide chain forming one subunit of the BphD enzyme has 234 non-

Table 1. Statistics of the crystallographic re®nement Ê) Resolution ( A Completeness (%) Used reflection (F > 1s) R-factor (%) Rfree (10 % set) (%) No. of protein atoms No. of solvent atoms Ê 2) Average temperature factor ( A Main-chaina Side-chaina Root-mean-square deviation from ideality Ê) Bond lengths ( A Bond angles (deg.)

Figure 1. Reaction catalyzed by the BphD enzyme in the biphenyl degradation pathway.

20-2.4 90.0 15,274 17.5 22.7 2200 32 39.7 37.5 (2.2) 41.9 (3.6) 0.007 1.2

a Root-mean-square deviation of the temperature factors is given in parentheses.

1141

Crystal Structure of BphD

glycine and non-proline amino acid residues. Among these, 215 amino acid residues (91.9 %) have f, c angles within the most favored regions of the Ramachandran plot.20 Among the remaining 19 residues, 18 (7.7 % of the total) have dihedral angles in the additional allowed regions. Only one residue (Ser110, one of the catalytic triad residues) is in the disallowed regions of the f ÿ c plot (see below). Extra density around Ser110 It should be noted that the 2Fo ÿ Fc and Fo ÿ Fc maps show an extra density around the side-chain of Ser110 (Figure 2). Since the extra density around Ser110 is present in all the electron-density maps derived from various independent sets of native data, it does not appear to be an experimental artifact. The distance between the OG atom of Ser110 Ê, and the center of the extra density is about 2 A suggesting that the density is unlikely to represent a water molecule that forms hydrogen bond with the OG atom of Ser110. Crystallographic re®nement assuming a metal ion for the extra density resulted in an extremely higher temperature factor Ê 2 for Ca2‡ and Mg2‡, Ê 2 and 41.8 A of the ion (52.1 A Ê 2). respectively) than that of OG of Ser110 (30.5 A Furthermore, no proper ligands for the metal ion can be found except for Ser110, and the size of the density is rather large for the metal ion. Thus, it is unlikely that a metal ion coordinates OG of Ser110. The extra density, thereby, seems to represent a chemical modi®cation to Ser110. Based on the shape of the difference density map, the small group, which is composed of three or four light atoms (such as carbon, nitrogen and oxygen), seems to be covalently bound to Ser110. In order to interpret the electron density more precisely, crystal structure at much higher resolution and a detailed biochemical analysis will be necessary.

Overall structure of the octameric BphD enzyme The overall structure of the BphD enzymes is composed of eight monomers related by the 422 point-group symmetry (Figure 3), which is identical with the crystallographic symmetry. The octameric structure can be regarded as a stack of two planar rings, each of which consists of four subunits related by a 4-fold rotational symmetry. The Ê along overall dimensions of the molecule are 82 A Ê along the 2-fold axis the 4-fold axis and 105 A perpendicular to the 4-fold axis. Across the 2-fold axis, two b-sheets, each consisting of eight bstrands from the adjoining subunits, form a single large b-sheet consisting of 16 b-strands. The quaternary structure of the BphD enzyme is quite different from that of other a/b-hydrolases such as lipase,21,22 haloalkane dehalogenase,23 etc. This is the ®rst example of the octameric a/b-hydrolase whose crystal structure has been determined. Although multimer formation sometimes results in the enzyme having an allosteric character, there has been no report of an allosteric character for the BphD enzyme.24,25 Subunit structure The subunit structure of the BphD enzyme is classi®ed as being one of the a/b-hydrolase folds.26,27 The subunit of the BphD enzyme can be divided into two domains, the core domain (residues 1-140, and residues 210-285) and the lid domain (residues 141-209) (Figure 4(a)). The loop regions connecting the two domains (residues 135143, and residues 208-215) are located at one of the sidewalls of the subunit. The temperature factor of Ê 2) is much higher than this region (more than 60 A that of other parts of the molecule (Table 1). This ¯exible loop region is designated as the hinge region.

Figure 2. The 2Fo ÿ Fc electron-density map around the active site (stereo pair). Extra density can be observed around the side-chain of Ser110. Catalytic residues are labeled. This Figure was prepared using the program XtalView44 and Raster3D.45

1142

Crystal Structure of BphD

a rather unfolded four a-helix bundle. The disposition of helices a6 and a7 deviates signi®cantly from the canonical conformation of the four a-helix bundle, as the loop region connecting strand b3 and helix a1 of the core domain is plugged into the space between helices a6 and a7. The atoms in the Ê 2 on an lid domain exhibit high B-factors (49.4 A average), with those in helices a6 and a7 being sigÊ 2 on an average). Helices ni®cantly higher (58.7 A Ê 2 on an a4 and a5 have lower B-factors (35.7 A average) than helices a6 and a7, probably due to their contact with the adjacent subunit in the octamer. Comparison with other a /b b-hydrolases The crystal structure of the BphD enzyme was compared with those of other a/b-hydrolases, lipases (1EX9,22 1HLG21), haloalkane dehalogenase (1B6G23), and epoxide hydrolase (1CR630) (PDB IDs are given in parentheses) (Table 2). These comparisons show that the core domains are well superposed with one another, with a small rootÊ . In contrast, the mean-square deviation of ca 1.9 A lid domains exhibit signi®cant diversity in their conformations. The database search with CE and VAST has revealed that the lid domains of the haloalkane dehalogenase (1B6G) and soluble epoxide hydrolase (1EHY) exhibit signi®cant similarity to that of the BphD enzyme. Figure 3. Overall structure of the BphD enzyme viewed (a) along the 4-fold axis and (b) along the 2-fold axis, which is perpendicular to the 4-fold axis. Each subunit is shown in a different color. In order to indicate the location of the core and lid domains in the octameric structure, the core and lid domains of subunit 1 are colored light green and green, respectively. In (a), the subunit number is labeled with the same color as the corresponding subunit. The 1-2 and 1-5 interfaces are indicated by red arrows. This Figure was prepared using the programs MOLSCRIPT46 and Raster3D.45

In the core domain, there is an eight-stranded bsheet in which strand b2 is disposed in a fashion antiparallel with the rest (Figure 4(b)). The fold of the domain can be regarded as a canonical fold of a/b-hydrolase.27 The atoms in the b-sheet have Ê 2 on average) than other lower B-factors (29.5 A parts of the molecule, suggesting that this part forms a structural core of the subunit. The lid domain is located above the core domain (Figure 4(a)) and is primarily composed of four ahelices. The database search using CE (combination extension of the optimal path28) and VAST (vector alignment search tool29) retrieved several all-a-proteins like uteroglobin, cytochrome c, cytochrome b562, myohemerythrin etc. The fold of the lid domain has signi®cant similarity to that of uteroglobin (PDB ID: 2UTG), which can be regarded as

Interaction between the subunits In the octameric structure, there are two different types of interactions between subunits: (i) the interaction between subunits in the planar ring (e.g. between subunits 1 and 2, see Figure 3(a)), the 1-2 interaction; and (ii) the interaction between subunits along the 4-fold axis (e.g. between subunits 1 and 5, Figure 3(b)), the 1-5 interaction (Figure 3(b)). The residues around the hinge region of subunit 1 and the residues on helices a4 and a5 of subunit 2 are involved in the 1-2 interaction (Figure 3(a)). The buried surface area of the interface is relatively Ê 2). The interaction between subunits 1 small (897 A and 2 is primarily an electrostatic one, with ®ve

Table 2. Results of the least-squares ®ttings between the BphD enzyme and other related enzymes Protein (PDB ID) Lipase (1EX9) Lipase (1HLG) Epoxide hydrolase (1CR6) Haloalkane dehalogenase (1B6G) a

All residues r.m.s. Ê) deviationa (A

Core domain r.m.s. Ê) deviationa (A

2.0 (143/285)b 2.0 (188/379) 1.8 (181/319)

1.9 (129/218) 1.9 (173/253) 1.7 (172/243)

2.1 (205/310)

1.9 (169/228)

r.m.s. deviation: root-mean-square deviation. Number of structurally matched residues (Ca atoms) in the least-squares ®tting/number of amino acid residues of the target protein. b

Crystal Structure of BphD

1143

Figure 4. (a) Subunit structure of the BphD enzyme (stereo pair). The eight-stranded b-sheet and a-helices in the core-domain are colored yellow and blue, respectively. The lid domain is colored red. The catalytic residues are shown by a ball-and-stick model. The substrate-binding pocket is indicated by the red arrow. The N and C termini are labeled. Strands b1 to b8 and helices a1 to a10 (except for a8, which cannot be seen in this Figure) are also labeled in white and yellow, respectively. This Figure was prepared using the programs MOLSCRIPT46 and Raster3D.45 (b) Schematic drawing of the subunit of the BphD enzyme. The arrows and cylinders represent b-strands and a-helices, respectively. Cylinders colored gray and black represent a-helices that are located behind and in front of the eight-stranded b-sheet, respectively. The positions of the catalytic residues are indicated by red circles.

hydrogen bonds between the two subunits (Table 3). The interaction between the Arg149 of subunit 2 and residues on subunit 1 is intense. The arginine residue is plugged into the cavity of subunit 1, which is formed by residues Leu213 to Pro219, Met82, and Gln118, and forms hydrogen bonds with Gln118 and Ser212 (Table 3). The 1-5 interaction occurs between the subunits related by a crystallographic 2-fold axis (Figure 3(b)). The b8 strands from both of the subunits form an antiparallel b-sheet. The analysis of

the 1-5 interface has revealed that this interface is similar in nature to that of the hydrophobic core of the protein molecule. At the 1-5 interface, hydrophobic residues are buried inside the interface and are surrounded by hydrophilic residues, some of which form hydrogen bonds with residues on the counterpart subunit (Table 3). The residues taking part in the 1-5 interaction have lower temperature factors than those of residues in the 1-2 interaction, suggesting the rigid nature of the 1-5 interaction. These ®ndings suggest that the 1-5 interaction is

1144

Crystal Structure of BphD

Table 3. List of the hydrogen bonds between subunits 1-2 interaction Subunit 1

Subunit 2

Ê) Distance ( A

Arg149 NH1 Arg149 NE1 Arg149 NH2 Arg158 NH2 Arg158 NH1

2.7 2.9 3.1 2.7 3.1

1-5 interaction Subunit 1

Subunit 5

Ê) Distance ( A

Arg233 NH1 Arg233 NH1 Arg233 NH2 Arg269 NH2 Thr247 O Leu250 O Lys251 O Glu254 OE1 Leu255 O Val257 N (Leu255 N (Val257O

Thr247 O Leu250 O Lys251 O Glu254 OE1 Arg233 NH1 Arg233 NH1 Arg233 NH2 Arg269 NH2 Val257 N Leu255 O Val257 O Leu255 N

2.7 2.9 2.9 2.8 2.7 2.9 2.9 2.8 2.7 2.7 3.4)a 3.4)a

Gln118 OE1 Ser212 O Ser212 O Glu122 OE1 Glu122 OE2

a Although these two have a rather long distance between proton donor and acceptor, these are included in the Table to indicate the residues involved in the 1-5 interaction.

very tight, despite the relatively small buried-surÊ 2). face area (876 A Active site The active-site residues have been identi®ed as Ser110, Asp235, and His26315 and are situated inside the cavity between the core and the lid domains (Figure 4(a)). These residues are completely conserved among the related enzymes (Figure 5). Ser110 is present in the sharp turn between strand b5 and helix a3 (Figure 4(b)). The f, c angles of Ser110 occur in an unfavorable region of the Ramachandran plot ((f, c) ˆ (61.1  , ÿ114.8  )), resulting in its strained conformation. It is known that the strained turn structure, the socalled nucleophile elbow, is well conserved in all the a/b-hydrolases.27,31 The second member of the catalytic triad, Asp235, is on the loop connecting strand b7 and helix a9 (Figure 4(b)). The OD2 atom of this resiÊ ) with the ND1 due forms a hydrogen bond (2.76 A atom of His263, which is the catalytic base of the enzyme (Figure 6). The other carboxyl oxygen atom of Asp235, the OD1 atom, forms a hydrogen Ê ) with the N atom of Ile237 (Figure 6). bond (2.90 A His263 is located on the loop connecting strand b8 and helix a10 (Figure 4(b)). The OG atom of Ser110 cannot form a hydrogen bond with the NE2 atom of His263 due to the orientation of the Ser110 sidechain (Figure 6). A rotation of the w-angle of Ser110 is needed to form a hydrogen bond between OG of Ser110 and NE2 of His263. The modi®ed part of Ser110 represented by the extra density shown in Figure 2, however, could make interactions with Ê ), NE2 of the NH group of Ala37 (less than 3 A

Ê ) and a water molecule in the His263 (ca 3.2 A Ê ). active site (ca 3.2 A Substrate-binding pocket The tubular cavity harboring the catalytic residues should be the substrate-binding pocket (Figure 4(a)). The substrate-binding pocket is ca Ê deep with a diameter of ca 5 A Ê , which is 20 A large enough to accommodate the substrate, HPDA. The substrate-binding pocket is formed by helices of the lid domain and loop regions of the core domain. The core-domain residues form the bottom and walls of the pocket, and residues in the lid domain provide a cap for the pocket. The bottom of the pocket is composed of Gly36, Ala41, Asn46, residues Asn109-Met111, Val136, Val238, and Thr242 (Figure 7(b)). The loop regions, b3-a1, b6-a4, a7-a8, b8-a10, and b7-a9, form the wall. The entrance to the cavity is located opposite the hinge region of the subunit and is surrounded by the N-terminal parts of helices a1 and a10 (in the core domain) as well as the a6 helix (in the lid domain). In the octameric structure, the entrance is exposed to the solvent, allowing the substrate to access the pocket. Catalytic residues are present at the bottom Ê deep in of the pocket and are situated ca 12 A relation to the entrance. The pocket can be divided into two parts at the Ser110 position. The parts proximal and distal to the entrance are designated, respectively, as P and D-parts. Part of the surface of the P-part is lined by polar residues, Asn46, Asn109, and Gln266 (Figure 7(b)). The NE atom of Trp264 present on the surface of the P-part is a potential site for an electrostatic interaction. In addition, the mainchain C1O groups of residues Ala37 to Gly40 are also present on the surface of the P-part, providing additional sites for the electrostatic interaction with the substrate. In contrast, the surface of the D-part is lined exclusively by hydrophobic residues (Figure 7(b)). Thr242, which is the only polar residue on the surface of the D-part, presents the methyl group of its side-chain on the surface of the D-part, leaving its hydroxyl group facing outward from the surface. It should be noted that no NH or C1O group of the main-chain is present on the surface of the D-part. The D-part is lined only by hydrophobic side-chains. It is interesting to note that, while comparing the amino acid residues forming the P and D-parts among related MCP-hydrolases, the surface residues on the core-domain of the P-part are well conserved among the related MCP-hydrolases (Figures 5 and 7).

Discussion Active site structure and catalytic mechanism Although the BphD enzyme exhibits low amino acid sequence identity with other forms of a/b-

Crystal Structure of BphD

1145

Figure 5. Multiple sequence alignment between the BphD enzyme (Rhodococcus sp. strain RHA1) and other related MCP-hydrolases. Amino acid sequence alignments of the BphD enzyme and its related proteins were carried out using the program ClustalX.47 The obtained results were assessed and modi®ed manually. The amino acid sequences used for the sequence alignment are as follows (the origin and the primary accession number of databases are given in parentheses): (1) BphD (Rhodococcus sp. strain RHA1; Q9KWQ6); (2) BphD (Pseudomonas cepacia LB400; P47229); (3) BphD (Pseudomonas azelaica HBP1; O06648); (4) EtbD1 (Rhodococcus sp. strain RHA1; O82828); (5) EtbD2 (Rhodococcus sp. strain RHA1; O82829); and (6) TodF (Pseudomonas putida F1; P23133). All of these sequences were retrieved from the SWISS-PROT and TrEMBL databases.48,49 Completely conserved residues among the presented sequences are colored red, and residues almost entirely conserved (®ve sequences out of six) are shown in orange. The residues that are completely conserved in each group (group I or II) are colored blue. The HGXGPG motif is marked on the top of the table. The residues indicated by pink rectangles compose the lid domain. The catalytic residues are indicated by green rectangles. The secondary structure and its designation are shown below the aligned sequences; a and b represent the a-helix and b-strand, respectively.

1146

Crystal Structure of BphD

Figure 6. Ball-and-stick representation of the active site residues (stereo pair). The hydrogen bonds are shown as green dotted lines. Labels of catalytic residues are presented in red. This Figure was prepared using the programs MOLSCRIPT46 and Raster3D.45

Figure 7. Inner molecular surface of the substrate-binding pocket. The inner surface of the (a) lid-domain side and that of the (b) core-domain side are shown. The substrate that is modeled based on the least-squares ®ttings (see Materials and Methods) is also shown with a ball-and-stick model. The inner surface is colored based on the conservation ratio of the surface residues. Completely and mostly conserved residues in the sequence-alignment table (see Figure 5) are shown in red and orange, respectively. The residues that are completely conserved in each group (group I or II) are colored blue (see Figure 5). The catalytic residues of the BphD enzyme are colored green. This Figure was prepared using the programs SPOCK50 and Raster3D.45

Crystal Structure of BphD

hydrolase (less than 20 % sequence identity with lipases (1EX9 and 1HLG), haloalkane dehalogenase (1B6G), epoxide hydrolase (1CR6)), a least-squares ®tting between the BphD enzyme and the other a/b-hydrolases exhibits that the relative disposition of the catalytic triad residues is nearly the same as those of other a/b-hydrolases. Furthermore, the ®tting has revealed that His35 and Gly36 of the BphD enzyme, which are located on the loop connecting strand b3 and helix a1, are almost completely conserved in the a/b-hydrolases. Since the corresponding residues of His35 and Gly36 in the a/b-hydrolases are thought to form the oxyanion hole,27,32,33 His35 and Gly36 may be involved in the formation of the oxyanion hole in the BphD enzyme. It should be noted that the H35G36X37G38P39G40 (X ˆ small amino acid, subscripts show residue number in BphD) motif in the BphD enzyme is completely conserved among the MCP-hydrolases (Figures 5 and 6). The common features suggest that the catalytic mechanism of BphD seems to be similar to that of the other a/bhydrolases. Despite the common features described above, there is one signi®cant difference between BphD and other a/b-hydrolases: the modi®cation of Ser110. Since the enzyme used in the crystallization is active, the modi®ed Ser110 could play a functional role in the catalytic reaction. However, we cannot rule out the possibility that the modi®cation happens during the crystallization. It should be noted that the BphD enzyme hydrolyzes carboncarbon bonds, which are of signi®cantly lower reactivity than the bonds cleaved by most hydrolases. Fumarylacetoacetate hydrolase (FAH), another type of carbon-carbon bond hydrolase, is distinguished from other catalytic triad-containing hydrolases by the involvement of an enzymebound metal ion and an unusual oxyanion hole.34 Thus the modi®ed Ser110 may be required to hydrolyze carbon-carbon bonds. It is interesting to note that the modi®ed part of Ser110 interacts with His263 and the oxyanion hole. In order to elucidate the nature of the modi®cation of Ser110, more biochemical work should be done. Structural feature of the substratebinding pocket Although the BphD enzyme shows a low degree of sequence identity with related MCP-hydrolases (ca 30 %), residues forming the bottom and walls of the substrate-binding pocket are well conserved between the BphD enzyme and the other MCPhydrolases (Figures 4(b), 5 and 7(b)). This ®nding suggests that the structure of the bottom and walls of the substrate-binding pocket in the BphD enzyme is likely to be very similar to those of the other MCP-hydrolases, suggesting that this part seems to be involved in speci®c interactions with HPDA (see below). In contrast, the amino acid residues in the lid domain exhibit signi®cant diversity among the MCP-hydrolases (Figures 5 and 7(a)),

1147 suggesting that residues in the lid domain are not involved in the speci®c interaction with the substrate. It should be noted that the lid domain of the lipase adopts two conformations, open and closed conformations, to accommodate a large substrate.35,36 However, the lid domain of the BphD enzyme cannot adopt an open conformation, as the lid domain is involved in the interaction with the adjacent subunit. The size of the substrate for the BphD enzyme cannot be larger than that of its substrate-binding pocket. Orientation of the substrate in the substratebinding pocket The substrate-binding pocket of the BphD enzyme is composed of a hydrophilic P-part and a hydrophobic D-part. The substrate of the BphD enzyme, HPDA, is also divided into two parts, the hydrophobic aromatic ring part and the hydrophilic dienoic acid part (Figure 1). These characteristics regarding the hydrophobicity of the substrate-binding pocket and the substrate seem to determine the orientation of the substrate in the substrate-binding pocket; the aromatic part of HPDA binds to the hydrophobic D-part, and the dienoic acid part binds to the hydrophilic P-part (Figure 7). Although there is no positively charged residue on the surface of the P-part to interact with the negatively charged carboxyl group of the bound substrate, a conformational change (ca 130  rotation of the w2 angle) of Arg186 makes it possible to form a tight salt bridge between the sidechain of Arg186 and carboxyl group of the substrate. This interaction may be common among the related enzymes, as Arg186 is completely conserved among them (Figure 5). It should be noted that the proposed orientation of the substrate is also consistent with the catalytic mechanism of the a/b-hydrolases. According to the general catalytic mechanism of the hydrolases, 2-hydroxypenta-2,4-dienoate (Figure 1) should be released upon acyl-enzyme complex formation. A water molecule then attacks the acyl-enzyme complex, leading to the dissociation of benzoic acid from the enzyme. In order to correspond with the general catalytic mechanism, especially the sequence of the product release from the enzyme, the dienoate and aromatic ring parts should be located in the P and the D-parts, respectively. Substrate specificity of the MCP-hydrolases The MCP-hydrolases found in the aromatic compound degradation pathways are classi®ed into two groups; one is found in the biphenyl degradation pathway (group I, HPDA hydrolases) and the other in the degradation pathways of compounds with single aromatic rings (group II, HOHD hydrolases). Since both groups of enzymes have narrow substrate speci®city,15,16 the structure

1148

Crystal Structure of BphD

of the substrate-binding pocket of these enzymes should re¯ect its substrate speci®city. Analysis of the substrate-binding pocket has revealed that the amino acid residues lining the surface of the P-part are well conserved among all the enzymes in groups I and II (Figure 7). Since the dienoate part of the substrate is the common structural motif of HPDA and HOHD, it is reasonable to assume that the dienoate part interacts with the P-part, which is composed of the conserved amino acid residues between groups I and II. In contrast, the amino acid residues present on the surface of the D-part are not well conserved among the related enzymes. Thus, it is reasonable to assume that the D-part accommodates a structurally different part of the substrate between groups I and II (a phenyl group for group I, and an alkyl group for group II). The residue corresponding to Met111 of the BphD enzyme, which is located at the bottom of the D-part, is completely conserved among the group I enzymes (Figure 5). It should be noted that the corresponding residue in the group II enzymes is phenylalanine (Figure 5). This ®nding suggests that this residue may play a key role in determining the substrate speci®city of the MPC-hydrolases. The results of the modeling study suggest that the replacement of Met111 with phenylalanine make the volume of the D-part insuf®cient to accommodate the phenyl group of HPDA. The difference in the volume in the D-part may contribute to the substrate speci®cities of EtbD1 and EtbD2. However, the narrow substrate speci®city of the BphD enzyme cannot be explained only by the volume of the D-part in the substrate-binding pocket, as HOHD is easily accommodated into the substrate-binding pocket of the BphD enzyme. It should be noted that HOHD cannot ®ll the space in the D-part of the RHA1 BphD enzyme due to the lack of a phenyl group. The BphD-HOHD complex may be thermodynamically unstable because of the large hydrophobic cavity inside the mol-

ecule. More detailed physicochemical analysis should be carried out in order to clarify this point.

Materials and Methods Crystal preparation Puri®cation and crystallization of the BphD enzyme from Rhodococcus sp. strain RHA1 were carried out as described.37 The BphD enzyme used for the crystallization was active. The activity of the enzyme can be demonstrated visually by a change of HPDA solution from yellow to colorless upon adding the enzyme. The enzyme activity of BphD in the crystal, however, was not determined. Heavy-atom derivative crystals were obtained by soaking the crystals in the standard buffer (0.1 M cacodylate buffer (pH 6.5), 10 % (v/v) isopropanol, 25 % (w/v) PEG4000 ) containing the heavy-atom Ê, compounds. The cell dimensions are a ˆ b ˆ 110.8 A Ê in space group I422, with one subunit per c ˆ 136.4 A Ê 3/Da.). asymmetric unit (VM ˆ 3.26 A Data collection All the X-ray diffraction data were collected at 24  C using an R-AXIS IIc (Rigaku) area detector with CuKa radiation (generated by a Rigaku RU-200 rotating anode generator operating at 4.95 kW and focused by a SUPPER double-focusing mirror). The data were processed with the program PROCESS that was installed on the R-AXIS IIc system. Structure determination and crystallographic refinement The crystal structure was solved by the multiple isomorphous replacement (MIR) method using a crystal of the selenomethionyl BphD enzyme as one of the heavyatom derivatives.19 The selenomethionyl BphD enzyme was produced essentially following the procedure described by Hendrickson et al.38 For the crystallographic calculations, the CCP4 program suite was used.39 Using Ê resthe ®ve heavy-atom derivatives, the phases for 2.6 A olution were calculated with the program MLPHARE, resulting in a ®gure of merit of 0.43. The structure deterÊ resolution are shown in mination statistics at 2.6 A Table 4.

Table 4. Structure determination statistics of BphD No. of reflections observed/unique Ê ); (resolution (A completeness (%)) Native (RAXIS) Hg(SCN)2 (RAXIS) K2PtCl4 (RAXIS) KAu(CN)2 (RAXIS) Se-Met (RAXIS) Se-Met, Hg(SCN)2 (RAXIS) a

46,418/16,372 (2.4; 91) 68,763/10,608 (2.5; 70) 69,021/6355 (2.65; 51) 40,768/9668 (2.8; 86) 84,382/8259 (2.6; 91) 74,188/11,022 (2.64; 86)

Figure of merit (centric/acentric) ˆ 0.87/0.57.

Rmerge (%)

Riso (%)

Rcullis (%)

Phasing power

6.3

-

-

-

8.0

20.6

0.51

1.3

7.6

5.4

0.78

0.6

6.2

12.7

0.79

0.5

6.2

9.2

0.79

0.5

9.3

16.7

0.74

0.7

1149

Crystal Structure of BphD

The quality of the resultant MIR electron-density map was suf®cient for overall chain tracing. The electrondensity map was con®dently interpreted using the PROTEIN package of the program QUANTA (Molecular Simulation Inc.), with the aid of the methionine positions as appearing in the difference Fourier map that were calculated using the (selenomethionyl ÿ native) amplitudes as coef®cients. Crystallographic re®nement of the model was carried out using the program X-PLOR.40 One cycle of the crystallographic re®nement consisted of a simulated annealing procedure with the program X-PLOR and a modelcorrection procedure with the program QUANTA using 2Fo ÿ Fc and Fo ÿ Fc maps. In the course of the simulated annealing procedure, free R-factors41 were used to assess the progress of the re®nement. A 10 % fraction of the whole data was randomly selected and exclusively utilized for calculating the free R-factor. In correcting the obtained model, a series of f-c plots20 was used to ®nd the regions having bad geometries. Tools used for geometrical analyses and least-squares fittings Program PROCHECK42 was used to analyze conformational variations from the de®ned norm. The secondary structures of the BphD enzyme were analyzed using the program PROCHECK. Least-squares ®ttings were carried out using the program O.43 After manually superposing structures and the information regarding the corresponding Ca atoms of both the structures given, the routine for the leastsquares ®tting of the program O was used to complete the ®tting. Docking study The docking study of the substrate was carried out based on the crystal structures of epoxide hydrolase (1CR630) and lipase (1EX922), both of which have a substrate/inhibitor in their crystal structures. The two structures were ®tted to the crystal structures of the BphD enzyme using least-squares ®ttings, and the substrate of the BphD enzyme was placed manually to satisfy the substrate-binding features of the epoxide hydrolase and the lipase. An energy minimization and a conformational search of the bound substrate were not carried out in the present study. Therefore, the binding geometry of the modeled substrate (Figure 7) represents only an approximate orientation of the substrate. Accession number The coordinates for the BphD structure have been deposited with the RCSB Protein Data Bank with the accession code 1C4X.

Acknowledgments We thank Dr V. Nagarajan of the present group for critical reading of the manuscript. We also thank Dr T. Nonaka of the present group for crystallographic discussions. This study was partly supported by the Promotion of Basic Research Activities for Innovative Bioscience (PROBRAIN) in Japan and Grants-in-Aid for Scienti®c

Research from the Ministry of Education, Science and Culture of Japan, no. 10490017 given to Y.M. and no. 1169215 given to Y.M. and T.S.

References 1. Ahmed, M. & Focht, D. D. (1973). Degradation of polychlorinated biphenyls by two species of Achromobacter. Can Micobiol. 19, 47-52. 2. Furukawa, K. & Matsumura, F. (1976). Microbial metabolism of polychlorinated biphenyls. Studies on the relative degradability of polychlorinated biphenyl components by Alcaligenes sp. J. Agric. Food Chem. 42, 543-548. 3. Furukawa, K. (1982). Microbial degradation of polychlorinated biphenyls. In Biodegradation and Detoxi®cation of Environmental Pollutants (Chakrabarty, A. M., ed.), pp. 33-57, CRC Press Inc., Boca Raton, FL. 4. Furukawa, K. & Chakrabarty, A. M. (1982). Involvement of plasmids in total degradation of chlorinated biphenyls. Appl. Environ. Microbol. 44, 619-626. 5. Bedard, D. L., Unterman, R., Bopp, L. H., Brennan, M. J., Haberl, M. L. & Johnson, C. (1986). Rapid assay for screening and characterizing microorganisms for the ability to degrade polychlorinated biphenyls. Appl. Environ. Microbiol. 51, 761-768. 6. Bedard, D. L., Wagner, R. E., Brennen, M. J., Haberl, M. L. & Brown, J. F., Jr (1987). Extensive degradation of Aroclors and environmentally transformed polychlorinated biphenyls by Alcaligenes eutrophus H850. Appl. Environ. Microbiol. 53, 1094-1102. 7. Bedard, D. L., Haberl, M. L., May, R. J. & Brennan, M. J. (1987). Evidence for novel mechanisms of polychlorinated biphenyl metabolism in Alcaligenes eutrophus H850. Appl. Environ. Microbiol. 53, 1103-1112. 8. Kimbara, K., Hashimoto, T., Fukuda, M., Koana, T., Takagi, M., Oishi, M. & Yano, K. (1989). Cloning and sequencing of two tandem genes involved in degradation of 2, 3-dihydroxybiphenyl to benzoic acid in the polychlorinated biphenyl-degrading soil bacterium Pseudomonas sp. strain KKS102. J. Bacteriol. 171, 2740-2747. 9. Furukawa, K. & Miyazaki, T. (1986). Cloning of a gene cluster encoding biphenyl and chlorobiphenyl degradation in Pseudomonas pseudoalcaligenes. J Bacteriol. 166, 392-398. 10. Kohler, H. P., Kohler-Staub, D. & Focht, D. D. (1988). Cometabolism of polychlorinated biphenyls: enhanced transformation of Aroclor 1254 by growing bacterial cells. Appl Environ Microbiol. 54, 19401945. 11. Kikuchi, Y., Yasukochi, Y., Nagata, Y., Fukuda, M. & Takagi, M. (1994). Nucleotide sequence and functional analysis of the meta-cleavage pathway involved in biphenyl and polychlorinated biphenyl degradation in Pseudomonas sp. strain KKS102. J Bacteriol. 176, 4269-4276. 12. Seto, M., Kimbara, K., Shimura, M., Hatta, T., Fukuda, M. & Yano, K. (1995). A novel transformation of polychlorinated biphenyls by Rhodococcus sp. strain RHA1. Appl. Environ. Microbiol. 61, 33533358. 13. Seto, M., Masai, E., Ida, M., Hatta, T., Kimbara, K., Fukuda, M. & Yano, K. (1995). Multiple polychlorinated biphenyl transformation system in the grampositive bacterium Rhodococcus sp. strain RHA1. Appl. Environ. Microbiol. 61, 4510-4513.

1150 14. Hauschild, J. E., Masai, E., Sugiyama, K., Hatta, T., Kimbara, K., Fukuda, M. & Yano, K. (1996). Identi®cation of an alternative 2, 3-dihydroxybiphenyl 1, 2-dioxygenase in Rhodococcus sp. strain RHA1 and cloning of the gene. Appl. Environ. Microbiol. 62, 2940-2946. 15. Yamada, A., Kishi, H., Sugiyama, K., Hatta, T., Nakamura, K., Masai, E. & Fukuda, M. (1998). Two nearly identical aromatic compound hydrolase genes in a strong polychlorinated biphenyl degrader, Rhodococcus sp. strain RHA1. Appl. Environ. Microbiol. 64, 2006-2012. 16. Hatta, T., Shimada, T., Yoshihara, T., Yamada, A., Masai, E., Fukuda, M. & Kiyohara, H. (1998). Meta-®ssion product hydrolases from a strong PCB degrader Rhodococcus sp. RHA1. J. Ferment. Bioeng. 85, 174-179. 17. Masai, E., Sugiyama, K., Iwashita, N., Shimizu, S., Hauschild, J. E. & Hatta, T. et al. (1997). The bphDEF meta-cleavage pathway genes involved in biphenyl/ polychlorinated biphenyl degradation are located on a linear plasmid and separated from the initial bphACB genes in Rhodococcus sp. strain RHA1. Gene, 187, 141-149. 18. Fukuda, M., Simizu, S., Okita, N., Seto, M. & Masai, E. (1998). Structural alteration of linear plasmids encoding the genes for polychlorinated biphenyl degradation in Rhodococcus strain RHA1. Antonie von Leeuwenhoek, 74, 169-173. 19. Nandhagopal, N., Senda, T., Hatta, T., Yamada, A., Masai, E., Fukuda, M. & Mitsui, Y. (1997). Threedimensional structure of microbial 2-hydroxyl-6-oxo6-phenylhexa 2,4-dienoic acid (HPDA) hydrolase (BphD enzyme) from Rhodococcus sp. strain RHA1, in the PCB degradation pathway. Proc. Japan Acad. 73B, 154-157. 20. Ramachandran, G. N. & Sasisekharan, V. (1968). Conformations of polypeptides and proteins. Advan. Protein. Chem. 23, 283-437. 21. Roussel, A., Canaan, S., Egloff, M. P., Riviere, M., Dupuis, L., Verger, R. & Cambillau, C. (1999). Crystal structure of human gastric lipase and model of lysosomal acid lipase, two lipolytic enzymes of medical interest. J. Biol. Chem. 274, 16995-17002. 22. Nardini, M., Lang, D. A., Liebeton, K., Jaeger, K.-E. & Dijkstra, B. W. (2000). Crystal structure of Pseudomonas aeruginosa lipase in the open conformation. The prototype for family I. 1 of bacterial lipases. J. Biol. Chem. 275, 31219-31225. 23. Ridder, I. S., Rozeboom, H. J. & Dijkstra, B. W. (1999). Haloalkane dehalogenase from Xanthobacter Ê resolution. Acta autotrophicus GJ10 re®ned at 1.15 A Crystallog. sect. D, 55, 1273-1290. 24. Seah, S. Y. K., Terracina, G., Bolin, J. T., Riebel, P., Snieckus, V. & Eltis, L. D. (1998). Puri®cation and preliminary characterization of a serine hydrolase involved in the microbial degradation of polychlorinated biphenyls. J. Biol. Chem. 273, 22943-22949. 25. Seah, S. Y. K., Labbe, G., Kaschabek, S. R., Reifenrath, F., Reineke, W. & Eltis, L. D. (2001). Comparative speci®city of two evolutionarily divergent hydrolases involved in microbial degradation of polychlorinated biphenyls. J. Bacteriol. 183, 11511516. 26. Ollis, D. L., Cheah, E., Cygler, M., Dijkstra, B., Frolow, F. & Franken, S. M. et al. (1992). The a/b hydrolase fold. Protein Eng. 5, 197-211.

Crystal Structure of BphD 27. Heikinheimo, P., Goldman, A., Jeffries, C. & Ollis, D. L. (1999). Of barn owls and bankers: a lush variety of a/b hydrolases. Structure, 7, R141-R146. 28. Shindyalov, I. N. & Bourne, P. E. (1998). Protein structure alignment by incremental combinatorial extension (CE) of the optimal path. Protein Eng. 11, 739-747. 29. Gibrat, J. F., Madej, T. & Bryant, S. H. (1996). Surprising similarities in structure comparison. Curr. Opin. Struct. Biol. 6, 377-385. 30. Argiriadi, M. A., Morisseau, C., Hammock, B. D. & Christianson, D. W. (1999). Detoxi®cation of environmental mutagens and carcinogens: structurebased mechanism and evolution of liver epoxide hydrolase. Proc. Natl Acad. Sci. USA, 96, 1063710642. 31. Arand, M., Grant, D. F., Beetham, J. K., Friedberg, T., Oesch, F. & Hammock, B. D. (1994). Sequence similarity of mammalian epoxide hydrolases to the bacterial haloalkane dehalogenase and other related proteins. FEBS Letters, 338, 251-256. 32. Verschueren, K. H. G., Seljee, F., Rozeboom, H. J., Kalk, K. H. & Dijkstra, B. W. (1993). Crystallographic analysis of the catalytic mechanism of haloalkane dehalgenase. Nature, 363, 693-698. 33. Nardini, M., Ridder, I. S., Rozeboom, H. J., Kalk, K. H., Dick, R. R. & Dijkstra, B. W. (1999). The X-ray structure of epoxide hydrolase from Agrobacterium radiobacter AD1. J. Biol. Chem. 274, 1457914586. 34. Timm, D. E., Mueller, H. A., Bhanumoorthy, P., Harp, J. M. & Bunick, G. J. (1999). Crystal structure and mechanism of a carbon-carbon bond hydrolase. Structure, 7, 1023-1033. 35. Grochulski, P., Li, Y., Schrag, J. D., Bouthillier, F., Smith, P. & Harrison, D. et al. (1993). Insights into interfacial activation from an open structure of Candida rugosa lipase. J. Biol. Chem. 268, 12843-12847. 36. Grochulski, P., Li, Y., Schrag, J. D. & Cygler, M. (1994). Two conformational states of Candida rugosa lipase. Protein Sci. 3, 82-91. 37. Nandhagopal, N., Senda, T., Hatta, T., Yamada, A., Masai, E., Fukuda, M. & Mitsui, Y. (1997). Crystallization of a hydolase from Rhodococcus sp. strain RHA1, the BphD enzyme, in the PCB degradation pathway. Protein Pept. Letters, 4, 211-214. 38. Hendrickson, W. A., Horton, J. R. & LeMaster, D. M. (1990). Selenomethionyl proteins produced for analysis by multiwavelength anomalous diffraction (MAD): a vehicle for direct determination of threedimensional structure. EMBO J. 9, 1665-1672. 39. Collaborative Computational Project Number 4 (1994). CCP4 suite: programs for protein crystallography by collaborative computational project number 4. Acta Crystallog. sect. D, 50, 760-763. 40. BruÈnger, A. T. (1992). X-PLOR, Version 3.1 Manual, Yale University Press, New Haven, CT. 41. BruÈnger, A. T. (1992). Free R value: a novel statistical quantity for assessing the accuracy of crystal structures. Nature, 355, 472-475. 42. 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. 43. Jones, T. A., Zou, J.-Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallog. sect. A, 47, 110-119.

Crystal Structure of BphD 44. McRee, D. E. (1993). Practical Protein Crystallography, Academic Press, San Diego, CA, USA. 45. Merritt, E. A. & Murphy, E. P. (1994). Raster3D Version 2.0. A program for photorealistic molecular graphics. Acta Crystallog. sect. D, 50, 869-873. 46. Kraulis, P. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallog. 24, 946-950. 47. Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. & Higgins, D. G. (1997). The ClustalX windows interface; ¯exible strategies for

1151 multiple sequence alignment aided by quality analysis tools. Nucl. Acid Res. 25, 4876-4882. 48. Bairoch, A. & Apweiler, R. (1997). The SWISS-PROT protein sequence database: its relevance to human molecular medical research. J. Mol. Med. 75, 312-316. 49. Bairoch, A. & Apweiler, R. (2000). The SWISS-PROT protein sequence data bank and its supplement TrEMBL in 2000. Nucl. Acids Res. 28, 45-48. 50. Christopher, J. A. (1998). The SPOCK Program Manual. The Center of Macromolecular Design, Texas A&M University.

Edited by R. Huber (Received 2 February 2001; received in revised form 30 April 2001; accepted 30 April 2001)