Research Article 25

An archetypical extradiol-cleaving catecholic : the crystal structure of catechol 2,3-dioxygenase (metapyrocatechase) from Pseudomonas putida mt-2 Akiko Kita1, Shin-ichi Kita1,2, Ikuhide Fujisawa1, Koji Inaka2, Tetsuo Ishida3, Kihachiro Horiike3, Mitsuhiro Nozaki3 and Kunio Miki1*

Background: Catechol catalyze the ring cleavage of catechol and Addresses: 1Department of Chemistry, Graduate its derivatives in either an intradiol or extradiol manner. These have a key School of Science, Kyoto University, Sakyo-ku, 2 role in the degradation of aromatic molecules in the environment by soil bacteria. Kyoto 606-8502, Japan, Research Laboratory of Resources Utilization, Tokyo Institute of Technology, Catechol 2,3-dioxygenase catalyzes the incorporation of dioxygen into catechol Nagatsuta, Midori-ku, Yokohama 226-8503, Japan and the extradiol ring cleavage to form 2-hydroxymuconate semialdehyde. and 3Department of Biochemistry, Shiga University Catechol 2,3-dioxygenase (metapyrocatechase, MPC) from Pseudomonas putida of Medical Science, Seta, Ohtsu, Shiga 520-2192, mt-2 was the first extradiol dioxygenase to be obtained in a pure form and has Japan. been studied extensively. The lack of an MPC structure has hampered the *Corresponding author. understanding of the general mechanism of extradiol dioxygenases. E-mail: [email protected]

Results: The three-dimensional structure of MPC has been determined at 2.8 Å Key words: catechol 2,3-dioxygenase, extradiol dioxygenase, metapyrocatechase, non-heme resolution by the multiple isomorphous replacement method. The is a dioxygenase, X-ray crystallography homotetramer with each subunit folded into two similar domains. The structure of the MPC subunit resembles that of 2,3-dihydroxybiphenyl 1,2-dioxygenase, Received: 17 August 1998 although there is low amino acid sequence identity between these enzymes. Revisions requested: 24 September 1998 Revisions received: 28 October 1998 The active-site structure reveals a distorted tetrahedral Fe(II) site with three Accepted: 10 November 1998 endogenous ligands (His153, His214 and Glu265), and an additional molecule that is most probably acetone. Published: 24 December 1998

Structure 7 Conclusions: The present structure of MPC, combined with those of two January 1999, :25–34 http://biomednet.com/elecref/0969212600700025 2,3-dihydroxybiphenyl 1,2-dioxygenases, reveals a conserved core region of the comprising three Fe(II) ligands (His153, His214 and Glu265), one © Elsevier Science Ltd ISSN 0969-2126 tyrosine (Tyr255) and two histidine (His199 and His246) residues. The results suggest that extradiol dioxygenases employ a common mechanism to recognize the catechol ring moiety of various substrates and to activate dioxygen. One of the conserved histidine residues (His199) seems to have important roles in the catalytic cycle.

Introduction active site ferrous ion of this enzyme is easily oxidized to Dioxygenase containing non-heme iron as a sole the ferric form by various oxidizing agents, and the plays a key role in the oxidative biodegradation of a large enzyme is concomitantly inactivated [5,6]. The inclusion number of aromatic compounds by catalyzing the cleavage of a relatively large amount of acetone or ethanol in of aromatic rings. Catechol 2,3-dioxygenase (2,3-CTD, buffers protects the enzyme from such inactivation [6]. catechol:oxygen 2,3- [decyclizing]; EC Acetone is also a competitive inhibitor against catechol 1.13.11.2) catalyzes an extradiol cleavage of catechol to (inhibition constant 13 mM) [7]. form 2-hydroxymuconate semialdehyde with the insertion of two atoms of dioxygen (Figure 1) [1]. This production Crystal structures of dioxygenases have been determined of an intensely colored product from a colorless substrate for protocatechuate 3,4-dioxygenase (3,4-PCD) [8,9], has been used in a number of biotechnological assays. The soybean -1 [10], and 2,3-dihydroxybiphenyl catechol 2,3-dioxygenase from Pseudomonas putida mt-2 1,2-dioxygenase (DHBD from Pseudomonas cepacia LB400 (ATCC 23973), metapyrocatechase (MPC), has been and BphC from Pseudomonas sp. strain KKS102) [11,12]. extensively investigated to elucidate the reaction mecha- The lipoxygenase is a monomer, 3,4-PCD is a dodecamer, nism of the extradiol ring cleavage type enzymes. MPC is and both DHBD and BphC are octamers. The monomer a tetramer of identical subunits, each with a single non- subunits of DHBD and BphC have a molecular size heme ferrous ion of high spin state and 307 amino acids similar to that of MPC (~35 kDa). The structures of the α 2+ [2]; the enzyme is designated as ( -Fe )4 [3,4]. The 2,3-dihydroxybiphenyl 1,2-dioxygenases in the resting 26 Structure 1999, Vol 7 No 1

Figure 1

Proposed mechanism for the proximal extradiol cleavage of catechol and 3- or 4- 6 substituted catechol catalyzed by catechol 5 1 OH H153 H199 2,3-dioxygenase. The mechanism previously (Nε2) (Nε2)B R 4 2 OH H proposed by Que and Ho [21] is modified E265 2 3 based on the three-dimensional structures of H2O (Oε1) R1 O FeII II Fe two 2,3-dihydroxybiphenyl 1,2-dioxygenases H O H214 O 2 R2 O 2 – (Nε2) [11,12] and the present three-dimensional (HO ?) 2H O 2 R1 structure of catechol 2,3-dioxygenase (MPC).

OH The numbering for the catechol ring shown in

– B the figure is used in the text to discriminate O COO H R2 between the C2–C3 bond (proximal) and the R1 O C1–C6 bond (distal). In this numbering FeIII scheme, both MPC and 2,3-dihydroxybiphenyl R2 O O 1,2-dioxygenase catalyze the cleavage of the R1 – O bond between the 2- and 3-positions of their respective substrates, and protocatechuate BH 2H O 4,5-dioxygenase (a distal extradiol 2 B H O dioxygenase) catalyzes the cleavage of the BH FeII O bond between the 1- and 6-positions of R2 O protocatechuate. The oxygen atoms from O O II R1 – II O Fe O dioxygen are represented by a larger capital R Fe 2 R1 letter. Catechol, R = H, R = H; 2,3- O O 1 2 R O R2 O dihydroxybiphenyl, R =C H , R = H; 3- 1 O 1 6 5 2 Structure methylcatechol, R1 =CH3, R2 =H; – protocatechuate, R1 = H, R2 = COO . state revealed a square-pyramidal Fe(II) or Fe(III) metal conventional multiple isomorphous replacement (MIR) center coordinated by two histidines, one glutamic acid method using two heavy-atom derivatives. The structure and two solvent molecules. In the case of MPC, once the shows that the four subunit molecules are related by a active-site Fe(II) is oxidized to Fe(III) it is immediately noncrystallographic 222 symmetry. The final refined released from the holoprotein, whereas BphC can retain model includes all 307 amino acid residues with one Fe Fe(III) at the active site. This difference between MPC atom and an acetone molecule for each subunit and 112 and BphC is thought to be due to the difference in their solvent molecules. The 2Fo–Fc map is of good quality and Fe(II) binding environments. To elucidate both the shows continuous well defined electron density. Figure 2 general reaction mechanism of extradiol ring cleavage and shows the electron-density map based on the final model. the substrate-recognition system, it is necessary to deter- The crystallographic R factor for the reflections between mine the structure of MPC at atomic levels. 10 to 2.8 Å resolution was 0.200 with a free R factor of 0.280 based on a subset of 5% of the reflections. The Here we present the crystal structure of MPC determined model has good stereochemistry and 88% of all residues, at 2.8 Å resolution by X-ray crystallography. The structure except glycine, are in the ‘most favorable’ regions of the shows a tetrameric assembly and an acetone molecule Ramachandran diagram. Superpositions of the Cα coordi- directly bound to each of the active site iron atoms. This nates of each monomer in the asymmetric unit show root structure is not only the first of a dioxygenase showing a mean square (rms) deviations between monomers of less preference for small monocyclic substrates, such as cate- than 0.4 Å. The statistics for data collection, phasing and chol, but also the first of a tetrameric enzyme in an extra- refinement are summarized in Tables 1 and 2. diol ring cleavage type dioxygenase family. The structure of MPC provides important information relating to the Overview of the MPC monomer catalytic mechanism of a series of extradiol ring cleavage The MPC monomer is an elliptical molecule with approxi- type dioxygenases. mate dimensions of 55 × 50 × 55 Å. The structure of the monomer is shown in Figure 3. The monomer comprises Results and discussion N-terminal and C-terminal domains (residues 1–149 and Structure determination 150–307, respectively), which are structurally similar to It was difficult to obtain MPC crystals large enough for each other and related by a local twofold axis. The N-ter- X-ray diffraction work. After many attempts, by the direc- minal domain has eight β strands and two α helices in the tion-restricted stepwise seeding method, the crystals sequence 1S1, 1H1, 1S2, 1S3, 1S4, 2S1, 2H1, 2S2, 2S3, were successfully grown to a long prismatic form suitable 2S4; the C-terminal domain is composed of nine β strands for data collection [13]. The structure was solved by the and four α helices in the sequence 3S1, 3H1, 3S2, 3S3, Research Article Structure of catechol 2,3-dioxygenase Kita et al. 27

Figure 2 that each domain is composed of an eight-stranded half- opened β barrel (Figure 3), as found in 2,3-dihydroxy- biphenyl 1,2-dioxygenases [11,12]. The overall structure of the monomer closely resembles those of the 2,3-dihy- droxybiphenyl 1,2-dioxygenases, in spite of low sequence identity (below 20%; Figure 4). The superposition of Trp129 MPC and the DHBD or BphC monomers (with the exception of two loops, Gly130–Trp139 and Leu181– Trp139 Arg187, and two helical regions, Thr284–Lys291 and Arg296–Thr307) shows Cα rms deviations of 2.3 Å. The half-opened β barrel in each domain comprises two repeated βαβββ motifs. This feature has been proposed to be the result of gene duplication [14].

There are significant differences between the N- and C-ter- Leu131 minal domains at their ends and in several loop regions. In the C-terminal end region of the N-terminal domain (Gly130–Trp139) there is a long loop region which pro- trudes towards the adjacent subunit in the MPC tetramer (Figures 3 and 5). Three residues (Asp133, Val134 and Asn135) which are located at the most protruding point of the loop form β-sheet-like hydrogen bonds with the adja- cent subunit. In contrast, a region consisting of one strand and two helices (4S5, 4H2 and 4H3), which follows the last βαβββ motif, is observed only in the C-terminal end region of the C-terminal domain (Val280–Thr307). The last helix Val134 (4H3) is located so as to cover the open region of the β

Structure barrel in the C-terminal domain. The external long loop region of the N-terminal domain is not observed in the

Quality of the electron-density map. The 2.8 Å 2Fo–Fc map for the structures of DHBD and BphC. The strand-helix region of external loop region (residues 129–140). The figure was created with the C-terminal domain takes a different conformation from the program TURBO-FRODO [30]. the structures of DHBD and BphC [11,12].

The β barrel in the C-terminal domain is arranged as a 3S4, 4S1, 4H1, 4S2, 4S3, 4S4, 4S5, 4H2, 4H3. The rms funnel, the central part of which forms a cavity going Cα–Cα distances for 118 corresponding residues in each through the molecule. One end of the funnel is open to domain are calculated to be 1.9 Å, despite a sequence the molecular surface and the other end is open to an identity of only 14%. It is a remarkable structural feature intersubunit interface of the tetramer. The outer opening

Table 1

Data collection and phasing statistics.

† ‡ Crystal Resolution Total reflections Unique data Rsym* Completeness No. of sites Rcullis Phasing power (Å) (%)

Native 1 2.7 132,385 45,293 70.0 8.26 — — —

Native 2 2.8 95,686 44,949 82.1 5.05 — — —

Native (merged) 2.8 46,225 86.0 7.60 — — —

PBAC§ 3.5 93,309 25,737 56.0 7.89 4 0.620 2.23

PCMB# 3.5 85,127 25,158 53.6 10.16 4 0.842 1.00

*Rsym =|I–|/I, where I is the observed intensity and is the average intensity obtained from multiple observations of symmetry-related † ‡ reflections. Rcullis = root mean square (rms) lack of closure/rms isomorphous difference. Phasing power = < |FH| >/rms lack of closure, where |FH| § # is the heavy-atom structure-factor amplitude. PBAC, (CH3COO)2Pb. PCMB, ClHgC6H4COOH. 28 Structure 1999, Vol 7 No 1

Table 2 to the formation of mainchain hydrogen bonds. In DHBD and BphC, there is no protruding loop forming intermolec- Refinement statistics. ular hydrogen bonds [11,12], but the β-sheet-like hydro- Resolution (Å) 10–2.8 gen bonds are also formed between the strands at the No. of reflections 43,306 corresponding connecting loops and have an important No. of protein atoms 9912 role in the formation of the octamer. In MPC, the pres- No. of water molecules 112 No. of Fe sites 4 ence of the long protruding loop prevents the formation of No. of acetone molecules 4 an octameric structure. A tetrameric assembly of MPC is R factor* 0.200 constructed from two dimers through hydrophobic interac- R free† 0.280 tions between the 4H1 helices and through intermolecular 2 Mean B value for sidechains (Å ) 15.6 hydrogen bonds mediated by water molecules. Mean B value for mainchains (Å2) 12.8 Mean B value for water molecules (Å2) 14.2 Rms deviations from ideality The Fe environment bond lengths (Å) 0.022 The Fe- is located in the β barrel of the C-ter- bond angles (°) 1.94 minal domain. This site is located at the end of a large Σ Σ † *R factor = |Fobs—Fcalc|/ |Fobs|. The R free was calculated using 5% pocket, which is of a suitable size to accept substrates. The of the data. Fe atom is coordinated by three residues — His153 Nε2, His214 Nε2 and Glu265 Oε1 — which are included in the at the surface is surrounded by hydrophobic sidechains, β strands 3S1, 4S1 and 4S4, respectively. The three whereas the inner opening is buried in the center of the residues are conserved not only in the catechol 2,3-dioxy- tetrameric molecule. The active site is located in the genase family but also in 2,3-dihydroxybiphenyl 1,2-dioxy- cavity of the β barrel in the C-terminal domain, as men- genases. These ligand residues are hydrogen bonded to tioned later in detail. adjacent residues (Figure 6a); His153 Nδ1, His214 Nδ1, Glu265 Oε1 and Glu265 Oε2 form hydrogen bonds to Tetrameric assembly of MPC Asp152, His213, Ser216 and Thr253, respectively. The MPC tetramer is presented in Figure 5a. There is one MPC tetramer in the asymmetric unit with a noncrys- In the difference Fourier map, a large electron density was tallographic 222 point group symmetry. Each monomer is found near the Fe atom at the opposite side of the liganded in contact with the other three symmetry-related mol- residues (Figure 6b). The center of this electron density is ecules. The buried surface area of the monomer is calcu- about 3.9 Å away from the Fe atom. The density has an lated to be 2600 Å2, which corresponds to 21% of the total almost flat, oval shape and is somewhat too large to be accessible surface area. As shown in Figure 5b, two assigned as a water molecule. It is located between the imi- monomers form a tentative dimer mainly through a pair of dazole ring of His246 and the aromatic ring of Phe191; the intermolecular β-sheet-like hydrogen bonds formed imidazole plane is almost parallel to the flat plane of the between the protruding loop and the adjacent monomer; density. This density can be interpreted as several possible the three residues mentioned above (133–135) contribute molecules: tris(hydroxymethyl)aminomethane used in the

Figure 3

Structure of the MPC monomer. The colors are employed to clarify each motif: light blue, dark blue, light green and dark green correspond to the first, second, third and fourth βαβββ motif, respectively. The two loops and two helix regions that are specific to MPC are colored in brown. The Fe(II) atom is shown as a red sphere. (a) A top view of the MPC monomer. According to the order of the secondary structure elements, β strands and α helices in the nth βαβββ motif are represented as nS1, nH1, nS2, nS3 and nS4, respectively. The three residues liganded to the Fe(II) atom are shown as ball-and-stick models. (b) A side view of the MPC monomer rotated 90° from (a) along the horizontal axis. The figures were created using the program MOLSCRIPT [34] and rendered with RASTER3D [35]. Research Article Structure of catechol 2,3-dioxygenase Kita et al. 29

Figure 4

Structural comparison of extradiol ring (a) cleavage type dioxygenases. (a) Stereoview superposition of the Cα atoms of MPC and DHBD. The Cα chains of MPC and DHBD are drawn in green and brown, respectively. The Fe(II) atom of MPC is shown as a red sphere and the three residues liganded to the Fe(II) atom are shown in ball-and-stick representation. The figure was created with the program MOLSCRIPT [34] and rendered with RASTER3D [35]. (b) Sequence alignment of MPC and two 2,3- dihydroxybiphenyl 1,2-dioxygenases, DHBD and BphC. Common amino acid residues are shaded in gray. Amino acid residues that comprise the Fe environment (shown in Figure 6a) are marked with boxes and the (b) 1S11H11S21S3 three residues liganded to the Fe(II) atom are 1020304050 indicated in red. The thin blue and green lines MPC MNKGVMRPGHVQLRVLDMSKALEHYVEL LGL I EMDRDDQGRVY LKAWTEV 10203040 DHBD represent the N- and C-terminal domains of MS I RSLGYMGFAVSDVAAWRSFL TQKLGLMEAGT TDN GDL FR I DS BphC S I ERLGYLGFAVKDVPAWDHFL TKSVGLMAAGSAGD AALYRA DQ MPC, respectively. The secondary structure 102030 40 elements for MPC and DHBD are indicated above and below the corresponding 1S42S12H12S2 sequences: β strands are shown as arrows 60708090 MPC DKFSLVLREADEPGMDFMGFKVVDEDALRQLERDLMAYGCAVEQLPAGE

(dark blue, MPC; light blue, DHBD) and 5060708090 DHBDRAWRI AVQQGEVDDLAFAGYEVADAAGLAQMADKLKQAG I AVTTGDASLA

BphCRAWRIAVQPGELDDLAYAGLEVDDAAALERMADKLRQAGVAFTRGDEALM helices are shown as rectangles (dark pink, 50 60 708090 MPC; light pink, DHBD). The regions of disorder in DHBD and BphC are shown in 2S32S4 italicized letters. 100110120130140

MPC LNSCGRRVRFQAPSGHHFELYADKEYTGKWGLNDVNPEAWPRDLKGMAA 100110120 130 DHBDRRRGVTGL I TFADPFGLPLE I YYGASEVFEK PFLPGAAVS

BphCQQRKVMGL LCLQDPFGLPLE I YYGPAE I FHE PFLPSAPVS 100 110 120 130

3S13H13S23S3

150 160170180

MPC VR FDHALMYGDELPATYDL FTKVLGFYLAEQVLDE NGTRVA 140150160170180 DHBDGF L TGEQGLGHFVRCVPDSDKALAFYTDVLGFQLSDV I DMKMGPDVTVPA BphCGFV TGDQG I GHFVRCVPDTAKAMAFYTEVLGFVLSD I I D IQMGPET SVPA 140 150 160 170 180

3S44S14H1

190 200210220230 MPC QFLSLSTKAHDVAF IHHPEKGRLH HVS FHLE TWEDL LRAADL I SMTDTS I 190200210220230 DHBDYFLHCNERHHTLA IAAFPLPKR IHHFMLEVAS LDDVGFA FDRVDADG L I BphCHFLHCNGRHHTIALAAFPIPKRIH H FMLQANT I DDVGYA FDRLDAAG R I 190 200 210 220 230

4S24S34S44S5 4H2 240250260270280 MPC D I GP TRHGL THGK T I Y FFDPSG NRN EVFCGGDYNYPDHKPVTWTTDQLG 240250260270 DHBDTSTLGRHTNDHMVSFYASTPSG VEVEYGWSARTVDRSWVVVRHD BphC TSLLGRHTNDQTLSFYADTPSPM I EV EFGWGPRTVDS S WTVARHS 240 250 260 270

4H3 290300

MPC KA I FYHDR I LNERFMTVLT 280290 DHBD SPSMWGHKSVRD KAAARN KA Q BphC RTAMWGHKS VRRG 280 290

Structure

Tris-acetate buffer, citrate used as a precipitant, ethanol molecule fits well to this density with average B factors of used as an additive, sulfate used in the purification, bicar- 32.7 Å2 (maximum 40.9 Å2 and minimum 21.8 Å2). In the bonate from the air, and acetone added in all steps of the two species of 2,3-dihydroxybiphenyl 1,2-dioxygenases, preparation and crystallization to maintain enzymatic activ- similar electron density around the Fe atom is found at a ity. We have finally concluded that an acetone molecule is corresponding, but not exactly the same, position (~6 Å the most probable candidate, judging from the shape and away from the Fe atom in DHBD) [11,12]. For this size of the electron density and from the fact that acetone density, a t-butanol molecule was assigned in DHBD, is a competitive inhibitor [7]. The model of an acetone whereas no molecule was assigned in BphC. 30 Structure 1999, Vol 7 No 1

Figure 5

Schematic drawing of the assembly features of the MPC molecule. Protein monomers are drawn in purple or green with the protruding loops colored in yellow. Fe atoms are shown as red spheres and the three residues liganded to the Fe(II) atom are shown as ball-and-stick models. (a) A side view of the whole tetramer. (b) A tentative dimer of MPC. The figures were created using the program MOLSCRIPT [34] and rendered with RASTER3D [35].

Two residues, Tyr255 and Ser216, and a water molecule glutamate) and two water molecules are ligands, where hydrogen bonded to Thr249 are also found around the one histidine residue is an axial ligand. In the present MPC active site, but they are too far from the Fe atom MPC structure, such water molecules could not be found (more than 3.3 Å) to be ligand molecules. The average dis- near the Fe atom. Although not in an analogous position, a tances between the Fe atom and the O and N atoms of lig- water molecule was found in MPC oriented in a similar anded residues (His153 Nε2–Fe, 2.47 Å; His214 Nε2–Fe, manner to one of the two water molecules in DHBD, but 2.40 Å; Glu265 Oε1–Fe, 2.25 Å) are somewhat longer than it cannot be regarded as an Fe ligand because the Fe–N and Fe–O distances (1.9 Å–2.1 Å) determined by Fe–O(water) distance is more than 3.8 Å. X-ray absorption spectroscopic studies of MPC [15,16]. The average Fe–O(acetone) distance of 2.61 Å is much The active site Fe(II) atom is immediately released from longer than the other Fe–O distances, indicating the weak the protein molecule when it is oxidized to the ferric state. ligation between the Fe atom and the acetone molecule. As the electron density for the Fe atom is clearly observed, The Fe coordination geometries of MPC and DHBD are the Fe(II) atom is thought to be included at the metal- compared in Figure 7. The coordination geometry around binding site in the present structure. Acetone (10%) is the Fe atom in MPC could best be described as a highly usually included in buffer solutions used in the purification distorted tetrahedral. In the structures of the 2,3-dihydrox- and crystallization of the enzyme to prevent the release of ybiphenyl 1,2-dioxygenases [11,12], the Fe atom pos- the Fe atoms from the enzyme. The present MPC struc- sesses a square-pyramidal coordination geometry; three ture reveals that the putative acetone molecule is directly amino acid residues (two histidine residues and one ligated to the Fe atom blocking the Fe(II)-binding sites

Figure 6

The coordination geometry around Fe(II) in the MPC molecule. (a) Ball-and-stick representation of the Fe atom, surrounding amino acid residues and the acetone and water molecules. The carbon, nitrogen and oxygen atoms are represented as black, blue and red spheres, respectively. The thick and thin broken lines represent metal–ligand coordination and hydrogen bonds, respectively. The figure was created using the program MOLSCRIPT [34] and rendered with RASTER3D [35]. (b) The Fe-omitted Fo–Fc electron-density map. Fe(II) is represented by a red sphere. The acetone molecule is assigned to the electron density appearing above the Fe(II) atom. For clarity, only the residues His153, His214 and Glu265 are shown. The figure was created using the program TURBO-FRODO [30]. Research Article Structure of catechol 2,3-dioxygenase Kita et al. 31

Figure 7

Comparison of the arrangement of the Fe- binding site. Thick broken lines indicate metal–ligand coordination. The color scheme is the same as in Figure 6. (a) MPC and (b) 2,3-dihydroxybiphenyl 1,2-dioxygenase, DHBD.

opposite to the three protein ligands. In the case of pocket; such a crevice is not observed in the 2,3-dihydroxy- DHBD, the t-butanol molecule is thought to inhibit and biphenyl 1,2-dioxygenases. This crevice enables the Fe stabilize the Fe atom by blocking the entrance route from atom to be accessible from the molecular surface and could the opening of the funnel [11]. be a pathway through which 2-hydroxymuconate semialde- hyde, the product of the oxygenation reaction of MPC, is The substrate-binding pocket Catechol and dioxygen are thought to enter the active site through a channel in the opening region of the β barrel Figure 8 comprised mainly of hydrophobic amino acid residues (Val180, Val188, Ala189, Ile204, Leu287, Leu298, Phe302 and Met303). The sidechains of the residues on the 4H3 helix, mainly those of Phe302 and Met303, cover and narrow the open region of the β barrel of the C-terminal domain. Consequently, the substrate entrance of MPC is narrower than that of the 2,3-dihydroxybiphenyl 1,2-dioxygenases. This is consistent with the fact that cat- echol is much smaller than biphenyl. The inner channel wall of the active-site pocket is formed by the amino acid residues His153, Leu155, Ala189, Phe191, His199, His214, His246, Leu248, Thr249, Tyr255, Glu265 and Ile291 (Figure 8). Among these residues, in addition to the Fe(II) ligands, five other residues are also conserved in the structure of the 2,3-dihydroxybiphenyl 1,2-dioxyge- nases (Ala189, Phe191, His199, His246 and Tyr255). The acetone molecule is surrounded by residues Phe191, His199, His246, Tyr255 and Thr249 (Figure 8). If catechol occupies the same position as the acetone molecule in the enzyme–substrate complex, it will also interact with these residues. The structure of MPC with bound substrate or Molecular surface representation of the wall of the substrate-binding pocket in the active site of MPC. Two views from the entrance of the an inhibitor, such as o-nitrophenol, at the Fe(II) site will substrate-binding pocket are shown (left and right) with (a) the give more information about the reaction mechanism, numbering of several amino acid residues and (b) the position of the including its intermediate state. acetone molecule shown as a cyan sphere. The Fe atom is shown as a red sphere. The red arrow in the left-hand view of (a) indicates the crevice near the Fe atom at the edge of the active-site pocket. The As shown in Figure 8, a crevice near the Fe(II) atom (shown figure was generated using the program GRASP [36]. by a red arrow) is found at the edge of the active-site 32 Structure 1999, Vol 7 No 1

released from the active site. Figure 8 also shows how the catechols displace the two water ligands when they bind acetone molecule is accommodated in the substrate- to the MPC active site, and that one of the hydroxyl binding pocket. groups of the substrate is deprotonated. We call the site occupied by the deprotonated hydroxyl group in the MPC Mechanistic implications active site the A site (anionic site) and the site occupied The present MPC structure and the structures of DHBD by the non-deprotonated hydroxyl group the N site and BphC show that the strictly conserved residues in the (neutral site). There are two possible modes for substrate sequence of the extadiol dioxygenase family, His153, binding: the hydroxyl group at position 1 of the catechol Phe191, His199, His214, His246, Tyr255 and Glu265, ring occupies the site trans to His214 and that at position 2 form the central portion of the active site, and that these occupies the site trans to Glu265 (or vacant sixth site) and residues take approximately identical positions in the is deprotonated; or the hydroxyl group at position 1 occu- three-dimensional structure of the active site. Thus, the pies the site trans to Glu265 (or vacant sixth site) and that available structures of resting DHBD with Fe(II) at the at position 2 occupies the site trans to His214 and is depro- active site [11] and BphC complexed with 3-methylcate- tonated. In the second binding mode, the 3- or 4-substi- chol [12] can be used to estimate the active-site structures tuted group of a substrate undergoes steric conflicts with of MPC at rest and complexed with catecholic substrates, MPC, chiefly with the sidechain of Thr249 (Figures 6 and although the latter enzyme–substrate complex contains 8). Thus, the first mode may be favored and we tenta- Fe(III) and is inactive. MPC catalyzes the ring fission of tively propose the site trans to His214 as the N site. 4-substituted catechols, such as 4-chlorocatechol and 4-methylcatechol, more efficiently than that of catechol (as Que and Ho [21] recently proposed a general mechanism judged by kcat/KMc value, where kcat is the catalytic center for the extradiol cleavage of catechol. One of the key activity and KMc is the KM for the catechol derivatives) [7]. points of the proposed mechanism is the presence of a Catechols substituted at the 3-position, such as 3-methyl- bidentate monoanionic catecholate. In the MPC–substrate catechol, are also relatively good substrates for MPC [7]. complex, it is probable that the hydroxyl group at Importantly, the bond between the 2- and 3-positions is position 1 binds to the N site and the hydroxyl group at exclusively cleaved by MPC [17] (see Figure 1). position 2 is deprotonated and binds to the A site. The second key point of the mechanism is the presence of an On the basis of these facts, in the following, we consider active-site base in the neighborhood of the N site that can the reaction mechanism of 3- or 4-substituted catechols to attack the the hydroxyl group at position 1 of the bound explicitly discriminate between the hydroxyl groups at substrate. In the MPC structure, His199 is located near positions 1 and 2. The circular dichroism (CD), magnetic the N site and can provide the active-site base. The third circular dichroism (MCD) and X-ray absorption spec- key element of the reaction mechanism is that an O2 mol- troscopy (XAS) data on resting MPC support the presence ecule binds to the vacant sixth site of the Fe(II) centre in of a five-coordinated Fe(II) site in a square-pyramidal the enzyme–catechol complex. One of the atoms of the geometry [16,18]. The crystal structures of BphC and bound O2 then attacks the C3 atom from the side of the DHBD show that two water molecules bind to the active- catechol ring opposite to the active-site base; this is site Fe: one water molecule occupies the position trans to coupled with the deprotonation of the hydroxyl group at

His209 (His210 in the case of DHBD, which corresponds position 1 by the base. In MPC, O2 may bind to the sixth to His214 of MPC), and another water molecule occupies vacant site trans to His153. Because only the C2–C3 bond the position trans to Glu260 (Glu265 in the case of MPC). is cleaved in the reaction catalyzed by MPC, one of the

In the active site of MPC, only Glu265, Tyr255 and the atoms of the bound O2 molecule takes a position near C3 water molecules can take an anionic form. This suggests and far from C6. The conserved residues Tyr255 and that one of the water ligands is deprotonated to compen- His246 may have an important role in directing the O2 sate for the charge of the ferrous ion, producing an overall molecule approximately parallel to the C2–C3 bond. neutral site. Biological implications o-Nitrophenol, one of the competitive inhibitors of MPC, The microbial strategy for degrading aromatic molecules is anionic when bound to the MPC active site [19,20]. consists of two critical steps: the ring hydroxylation on Similarly, catechol derivatives complexed with MPC in adjacent carbon atoms and the ring cleavage of the anaerobic conditions exhibit absorption spectra resem- resulting catecholic metabolites. Catechol dioxygenases bling those of the corresponding substrates in monoan- catalyze the ring cleavage in either an intradiol or an ionic form (TI and KH, unpublished results). In the extradiol manner [17]. The extradiol dioxygenases structure of BphC in complex with 2,3-dihydroxybiphenyl utilize Fe(II), and rarely Mn(II), in the reaction. Most or 3-methylcatechol, the two hydroxyl groups of the sub- of these enzymes catalyze the cleavage of the proximal strates displace the two water ligands [12]. These results C2–C3 bond of 3- or 4-substituted catechol, but some suggest that the two hydroxyl groups of 3- or 4-substituted catalyze the cleavage of the distal C1–C6 bond. The Research Article Structure of catechol 2,3-dioxygenase Kita et al. 33

proximal extradiol dioxygenases are classified into two performed using the PROTEIN [27] program package. The calculated families on the basis of substrate preference: those pre- phases up to 3.5 Å resolution gave an overall figure of merit of 0.42 (PROTEIN statistics). The 3.5 Å map was improved by solvent flatten- ferring small monocyclic molecules, such as catechol ing, histogram matching and noncrystallographic averaging with the and 3- or 4-methyl catechol, and those preferring large program DM [28] in the CCP4 package [29]. bicyclic molecules, such as dihydroxybiphenyl [22]. Model building and crystallographic refinement In the present study, we report the first three-dimen- The quality of the improved 3.0 Å electron-density map was good enough to do chain tracing for most of the polypeptide chain, and a sional structure of catechol 2,3-dioxygenase (metapyro- polyalanine model was constructed using the graphics program catechase, MPC) from Pseudomonas putida mt-2 — an TURBO-FRODO [30]. Phases obtained by combining MIR phases with extradiol catechol dioxygenase that cleaves monocyclic the polyalanine model using the program SIGMAA [31] in the CCP4 substrates. Numerous kinetic and biophysical investiga- package were modified by solvent flattening and noncrystallographic averaging. This phase modification process was repeated several times. tions have been carried out on MPC, and models of the The model refinement was carried out by simulated annealing (heating reaction mechanism and the active site Fe(II) environ- up to 3000K) and conventional positional and B-factor refinement with ment have been built on the basis of these studies. The the program X-PLOR [32] using the Engh and Huber stereochemical structure of MPC therefore has important implications parameters. A random sample containing 5% of the data was excluded from the refinement and used for monitoring the course of the refine- for understanding the general mechanism of extradiol ment. The refinement was carried out with noncrystallographic ring cleavage. The MPC molecule is a homotetramer restraints in the earlier stages and the last refinements were performed and each subunit comprises two domains (the N- and without noncrystallographic restraints. When the crystallographic C-terminal domains). The active site is located in a R factor had dropped to 0.24, water molecules were added at positions σ with density higher than 3 in the Fo–Fc map calculated by X-PLOR. At cavity within the C-terminal domain. The cavity has every stage of the refinement, models were subjected to critical quality two openings: one is hydrophobic and open to bulk sol- analyses using the program PROCHECK [33]. The Fe ions were identi- vents and the other is narrow, hydrophilic and opens fied on the basis of their coordination and the height of their peaks. into an intersubunit interface. A putative acetone mol- ecule, which protects the enzyme from inactivation and The refined model The current model consists of 9916 atoms in total: protein atoms is a competitive inhibitor with respect to catecholic sub- (307 × 4 = 1228 amino acids in the tetramer and four Fe atoms), four strate, directly binds to the active-site iron, which is fixed acetone molecules and 112 water molecules. on the wall of the cavity by two histidines and one gluta- mate residue. Comparisons of the present structure with Accession numbers The coordinates for catechol 2,3-dioxygenase (metapyrocatechase) those of biphenyl-cleaving extradiol dioxygenases show have been deposited in the Brookhaven Protein Data Bank with acces- a strikingly similar active-site structure and reveal the sion code 1mpy. structural differences that determine substrate speci- ficity. The three active-site residues His199, His246 and Acknowledgements Tyr255, which are strictly conserved in all known extra- We would like to thank N Sakabe, A Nakagawa (currently at Hokkaido Uni- diol dioxygenases (more than 30), may have important versity), N Watanabe and M Suzuki of the Photon Factory for their kind help in intensity data collection and processing, which were performed under the roles in the catalytic cycle. approval of the Photon Factory Advisory Committee (proposal numbers 94G250 and 96G208). Thanks are also due to M Tamura (Hokkaido Uni- versity) for his help in starting this project and C Nakai, A Masui and S Kaida Materials and methods for their contribution at the initial stage of the work. KM is a member of the Crystallization, data collection and processing Sakabe Project of TARA (Tsukuba Advanced Research Alliance), University Initial thin needle-like crystals were obtained from sodium citrate solu- of Tsukuba. This work was partly supported by Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture, tions by the vapor-diffusion method. The best crystals were grown by Japan (number 09780605 to AK and number 08249217 to KH), by the the stepwise macroseeding method, adopting ethanol as the additive ‘Research for the Future’ Program (96L00502 and 97L00501 to KM) from [13]. The crystals belong to the space group P42212 with unit-cell the Japan Society for the Promotion of Science, and by the Hayashi Memor- dimensions of a = b = 264 Å and c = 59.8 Å [13]. The asymmetric unit ial Foundation for Female National Scientists to AK. contains one tetramer of MPC, the solvent content being 68%. X-ray diffraction data were collected at a wavelength of 1.0 Å using synchro- tron radiation with a screenless Weissenberg camera at the BL-6A References beam line of the Photon Factory, Institute of Material Structure 1. Kojima, Y., Itada, N. & Hayaishi, O. (1961). Metapyrocatechase: a new catechol-cleaving enzyme. J. Biol. Chem. 236, 2223-2228. Science, High Energy Accelerator Research Organization (formerly the 2. Tatsuno, Y., Saeki, Y., Nozaki, M., Otsuka, S. & Maeda, Y. (1980). National Laboratory for High Energy Physics) [23,24]. A native data set Mossbauer spectra of metapyrocatechase. FEBS Lett. 112, 83-85. used in the molecular refinement (native) was obtained by merging two 3. Nakai, C., Hori, K., Kagamiyama, H., Nakazawa, T. & Nozaki, M. native data sets, native 1 (c-axis rotation) and native 2 (a-axis rotation). (1983). Purification, subunit structure, and partial amino acid Data processing was performed using the programs WEIS [25], sequence of metapyrocatechase. J. Biol. Chem. 258, 2916-2922. DENZO and SCALEPACK [26]. 4. Nakai, C., et al., & Nakazawa, A. (1983). Complete nucleotide sequence of the metapyrocatechase gene on the TOL plasmid of Phasing Pseudomonas putida mt-2. J. Biol. Chem. 258, 2923-2928. 5. Nozaki, M., Kagamiyama, H. & Hayaishi, O. (1963). Two heavy-atom derivatives, lead acetate (PBAC) and 4-chloromer- Metapyrocatechase. I. Purification, crystallization and some properties. curibenzoic acid (PCMB) were used for the phase determination. Biochem. Z. 338, 582-590. Heavy-atom sites were determined by difference Patterson and Fourier 6. Nozaki, M., Ono, K., Nakazawa, T., Kotani, S. & Hayaishi, O. (1968). difference maps. There is a single heavy atom binding site per Metapyrocatechase. II. The role of iron and sulfhydryl groups. J. Biol. monomer in each derivative. Refinement of heavy-atom parameters was Chem. 243, 2682-2690. 34 Structure 1999, Vol 7 No 1

7. Kobayashi T., et al., & Nozaki, M. (1995). Overexpression of 31. Read, R.J. (1986). Improved Fourier coefficients for maps using Pseudomonas putida catechol 2,3-dioxygenase with high specific phases from partial structures with errors. Acta Cryst. A 42, 140-149. activity by genetically engineered Escherichia coli. J. Biochem. 32. Brünger, A.T. (1992). X-PLOR Version 3.1 Manual. Yale University 117, 614-622. Press, New Haven, CT. 8. Ohlendorf, D.H., Lipscomb, J.D. & Weber, P.C. (1988). Structure and 33. Laskowski, R.A., MacArthur, M.W., Moss, D.S. & Thornton, J.M. assembly of protocatechuate 3,4-dioxygenase. Nature 336, 403-405. (1993). PROCHECK: a program to check the stereochemical quality 9. Ohlendorf, D.H., Orville, A.M. & Lipscomb, J.D. (1994). Structure of of protein structures. (1993). J. Appl. Cryst. 26, 283-291. protocatechuate 3,4-dioxygenase from Pseudomonas aeruginosa at 34. Kraulis, P.J. (1991). MOLSCRIPT: a program to produce both detailed 2.15 Å resolution. J. Mol. Biol. 244, 586-608. and schematic plots of protein structures. J. Appl. Cryst. 24, 946-950. 10. Boyington, J.C., Gaffney, B.J. & Amzel, L.M. (1993). The three- 35. Merrit, E.A. & Murphy, M.E.P. (1994). Raster3D version 2.0 – a dimensional structure of an arachidonic acid 15-lipoxygenase. program for photorealistic molecular graphics. Acta Cryst. D Science 260, 1482-1486. 50, 869-873. 11. Han, S., Eltis, L.D., Timmis, K.N., Muchmore, S.W. & Bolin, J.T. (1995). 36. Nicholls, A., Sharp, K.A. & Honig, B. (1991). Protein folding and Crystal structure of the biphenyl-cleaving extradiol dioxygenase from a association: insights from the interfacial and thermodynamic PCB-degrading pseudomonad. Science 270, 976-980. properties of hydrocarbons. Proteins 11, 281-296. 12. Senda, T., et al., & Mitsui, Y. (1996). Three-dimensional structure of free form and two substrate complexes of an extradiol ring-cleavage type dioxygenase, the BphC enzyme from Pseudomonas sp. strain KKS102. J. Mol. Biol. 255, 735-752. 13. Kita, A., et al., & Miki, K. (1997). Crystallization and preliminary X-ray diffraction studies of expressed Pseudomonas putida catechol 2,3- dioxygenase. J. Biochem. 122, 201-204. 14. Bergdoll, M., Eltis, L.D., Cameron, A.D., Dumas, P. & Bolin, J.T. (1998). All in the family: structural and evolutionary relationships among three modular proteins with diverse functions and variable assembly. Protein Sci. 7, 1661-1670. 15. Bertini, I., Briganti, F., Mangani, S., Nolting, H.F. & Scozzafava, A. (1994). X-ray absorption studies on catechol 2,3-dioxygenase from Pseudomonas putida mt-2. Biochemistry 33, 10777-10784. 16. Shu, L., Chiou, Y.-M., Orville, A.M., Miller, M.A., Lipscomb, J.D. & Que, L.Jr. (1995). X-ray absorption spectroscopic studies of the Fe(II) active site of catechol 2,3-dioxygenase. Implications for the extradiol cleavage mechanism. Biochemistry 34, 6649-6659. 17. Nozaki, M., Kotani, S., Ono, K. & Senoh, S. (1970). Metapyrocatechase. III. Substrate specificity and mode of ring fission. Biochim. Biophys. Acta 220, 213-223. 18. Mabrouk, P.A., Orville, A.M., Lipscomb, J.D. & Solomon, E.I. (1991). Variable-temperature variable-field magnetic circular dichroism studies of the Fe(II) active site in metapyrocatechase: implication for the molecular mechanism of extradiol dioxygenases. J. Am. Chem. Soc. 113, 4053-4061. 19. Horiike, K., Kobayashi, T., Ishida, T. & Nozaki, M. (1991). O- Nitrophenol as an active site probe for catechol 2,3-dioxygenase. In International Symposium on and Oxygen Activation. (Yamamoto, S., Nozaki, M. & Ishimura, Y., eds), pp. 15-18, Yamada Science Foundation, Osaka, Japan. 20. Honjo, M., Ishida, T. & Horiike, K. (1997). Semi-micro-scale frontal gel chromatography of interacting systems of a protein and small molecules: binding of warfarin, tryptophan, or FMN to albumin, and o- nitrophenol to catechol 2,3-dioxygenase. J. Biochem. 122, 258-263. 21. Que, L.Jr. & Ho, R.Y.N. (1996). Dioxygen activation by enzymes with mononuclear non-heme iron active sites. Chem. Rev. 96, 2607-2624. 22. Eltis, L.D. & Bolin, J.T. (1996). Evolutionary relationships among extradiol dioxygenases. J. Bacteriol. 178, 5930-5937. 23. Sakabe, N. (1991). X-ray diffraction data collection system for modern protein crystallography with Weissenberg camera and imaging plate using synchrotron radiation. Nucl. Instrum. Methods A 303, 448-463. 24. Miyahara, J., Takahashi, K., Amemiya, Y., Kamiya, N. & Satow, Y. (1986). A new type of X-ray area detector utilizing laser stimulated luminescence. Nucl. Instrum. Methods A 246, 572-578. 25. Higashi, T. (1989). The processing of diffraction data taken on a screenless Weissenberg camera for macromolecular crystallography. J. Appl. Cryst. 22, 9-18. 26. Otwinowski, Z. & Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. In Methods in Enzymology Vol. 276. (Carter, C.W. Jr. & Sweet, R.M., eds), pp. 307-326, Academic Press, New York, NY. 27. Steigemann, W. (1992). PROTEIN version 3.1: a program system for the crystal structure analysis of proteins. Max-Planck Institut für Biochemie, Martinsried, Germany. 28. Cowtan, K. & Main, P. (1996). Phase combination and cross validation Because Structure with Folding & Design operates a in iterated density-modification calculations. Acta Cryst. D 52, 43-48. 29. Collaborative Computational Project No. 4. (1994). The CCP4 suite: ‘Continuous Publication System’ for Research Papers, this programs for protein crystallography. Acta Cryst. D 50, 760-763. paper has been published on the internet before being printed 30. Roussel, A. & Cambillau, C. (1989). TURBO-FRODO. In Silicon Graphics Geometry Partners Directory. pp. 77–78, Silicon Graphics, (accessed from http://biomednet.com/cbiology/str). For Mountain View, CA. further information, see the explanation on the contents page.