structure of an aromatic-ring-hydroxylating dioxygenase— … · 2017-02-14 · structure of an...

Structure of an aromatic-ring-hydroxylating dioxygenase —naphthalene 1,2-dioxygenaseBjörn Kauppi1, Kyoung Lee2† , Enrique Carredano1, Rebecca E Parales2, David T Gibson2, Hans Eklund1 and S Ramaswamy1*

Background: Pseudomonas sp. NCIB 9816-4 utilizes a multicomponentenzyme system to oxidize naphthalene to (+)-cis-(1R,2S)-dihydroxy-1,2-dihydronaphthalene. The enzyme component catalyzing this reaction,naphthalene 1,2-dioxygenase (NDO), belongs to a family of aromatic-ring-hydroxylating dioxygenases that oxidize aromatic hydrocarbons and relatedcompounds to cis-arene diols. These enzymes utilize a mononuclear non-hemeiron center to catalyze the addition of dioxygen to their respective substrates.The present study was conducted to provide essential structural informationnecessary for elucidating the mechanism of action of NDO.

Results: The three-dimensional structure of NDO has been determined at2.25 Å resolution. The molecule is an α3β3 hexamer. The α subunit has aβ-sheet domain that contains a Rieske [2Fe–2S] center and a catalytic domainthat has a novel fold dominated by an antiparallel nine-stranded β-pleated sheetagainst which helices pack. The active site contains a non-heme ferrous ioncoordinated by His208, His213, Asp362 (bidentate) and a water molecule.Asn201 is positioned further away, 3.75 Å, at the missing axial position of anoctahedron. In the Rieske [2Fe–2S] center, one iron is coordinated by Cys81and Cys101 and the other by His83 and His104.

Conclusions: The domain structure and iron coordination of the Rieske domainis very similar to that of the cytochrome bc1 domain. The active-site iron centerof one of the α subunits is directly connected by hydrogen bonds through asingle amino acid, Asp205, to the Rieske [2Fe–2S] center in a neighboringα subunit. This is likely to be the main route for electron transfer.

IntroductionInterest in the structure and mechanism of action of aro-matic-ring-hydroxylating dioxygenase systems [1] stemsfrom two sources. First, they play an important role in pro-viding the basic knowledge necessary for the develop-ment of bioremediation technology. Second, they oftenproduce single enantiomers of cis-arene diols which havebeen used as chiral synthons in the formation of inositols,prostaglandin E2α, conduritols, and numerous productswith biological activity [2–6]. Thus, they may provide anadditional, and potentially environmentally benign, alter-native to the use of chemical catalysts in asymmetrichydroxylation reactions [7]. One commercial application[8] is the oxidation of indole to indigo by the naphthalenedioxygenase system [9,10] in a reaction that does not gen-erate waste products.

A large family of more than 40 related multicomponentmononuclear non-heme iron oxygenase systems has beenidentified by protein purification or sequence analysis [1,11].The naphthalene dioxygenase system from Pseudomonas sp.NCIB 9816-4 is characterized by the extensive range of

substrates it can oxidize and the different types of reac-tions catalyzed with different substrates [6]. For example,naphthalene 1,2-dioxygenase (E.C. 1.14.12.12), the cat-alytic component, catalyzes the reaction shown below, andalso catalyzes many of the reactions reported for cyto-chrome P450 [12].

Naphthalene + NAD(P)H + H+ + O2 → (+)-cis-(1R,2S)-dihydroxy-1,2-dihydronaphthalene + NAD(P)+

The naphthalene dioxygenase system utilizes an iron–sulfur flavoprotein to transfer electrons from NAD(P)H toa Rieske [2Fe–2S] ferredoxin which in turn reduces theterminal oxygenase [13,14]. This catalytic portion of thenaphthalene dioxygenase system which we here call NDO(earlier called ISPNAP for iron–sulfur protein) is composedof non-identical subunits, α and β [15]. This change innomenclature is in accordance with the more recent classi-fication of iron–sulfur proteins by the Nomenclature Com-mittee of the International Union of Biochemistry andMolecular Biology [16]. The presence of a Rieske [2Fe–2S]center and mononuclear iron in the α subunit of NDO

Addresses: 1Department of Molecular Biology,Swedish University of Agricultural Sciences, Box590, BMCS-751 24 Uppsala, Sweden and2Department of Microbiology and Center forBiocatalysis and Bioprocessing, University of Iowa,Iowa City, Iowa 52242, USA.

†Present address: Department of Microbiology,Changwon National University, Changwon-Si,Kyongnam, South Korea.

*Corresponding author.E-mail: [email protected]

Key words: bioremediation, dioxygenase, electrontransfer, mononuclear iron, Rieske [2Fe–2S] center

Received: 26 November 1997Revisions requested: 8 January 1998Revisions received: 25 January 1998Accepted: 4 March 1998

Structure 15 May 1998, 6:571–586http://biomednet.com/elecref/0969212600600571

© Current Biology Ltd ISSN 0969-2126

Research Article 571

[15,17,18] and all other aromatic ring-hydroxylating dioxy-genases is based on biophysical studies with phthalate[19–21] and benzene [22,23] dioxygenases and deducedamino acid sequence data [11,24]. NDO was shown tocontain ferrous iron as indicated by the identification of aFe2+–nitric oxide complex by electron paramagnetic reso-nance (EPR) spectroscopy (MD Wolfe, JD Lipscomb, KLand DTG, unpublished data). The β subunit (present insome members of the family) has no cofactors and its rolein catalysis is unknown. The genes encoding the individ-ual components of NDO have been cloned and sequenced[25,26]. An overproducing recombinant Escherichia coli strainhas been used to purify large amounts of active NDO; thisfacilitated crystallization of the enzyme [27].

We now describe the three-dimensional structure of NDO.To the best of our knowledge, this is the first structure ofan asymmetric aromatic ring-hydroxylating dioxygenaseand can serve as the prototype structure for this uniquegroup of enzymes. We find that the quaternary organ-ization of the enzyme allows a close contact between thetwo redox centers. The Rieske [2Fe–2S] center of oneα subunit in the α3β3 molecule is positioned in close prox-imity to the mononuclear iron at the active site of a neigh-boring α subunit, suggesting a functional cooperativitybetween α subunits.

Results and discussionThe NDO structure was solved in a trigonal crystal formby a combination of the multiple isomorphous replace-ment (MIR) and multiwavelength anomalous dispersion(MAD) methods using selenomethionine (SeMet) labeledprotein. The structure has been refined at 2.5 Å resolu-tion. During refinement, a new orthorhombic crystal formwas found that diffracted to 2.25 Å resolution. This crystalform was used for the final refinement, giving an R factorof 0.19 and an R free of 0.23. The geometry of the three

noncrystallographic symmetry-related units was restrainedthroughout the refinement. An omit map around theactive site shows the quality of the electron density(Figure 1). The quality of the present model is good andmore than 90% of the residues are in the most favorableregion of the Ramachandran plot [28]. There are only tworesidues in the disallowed region (Cys309 in the α subunitand His188 in the β subunit), but inspection of bothresidues reveals that they are in regions of good electrondensity. Both of these buried residues are stabilized by anetwork of hydrogen bonds.

Quaternary structureNDO and most of the dioxygenases of this family ofenzymes have been characterized as α2β2 or α3β3 multi-mers [29]. A third group consists of oxygenases that onlycontain α subunits [30] (reviewed in [1]). On the basis ofgel-filtration analysis, NDO was originally reported tohave an α2β2 subunit composition [15]. From the crystalstructure, it is apparent that NDO is an α3β3 hexamer(Figures 2a and b). This is supported by recent analyticalultracentrifugation analysis, which gave a molecular massof 210 kDa for NDO (KL and DTG, unpublished data).The arrangement of the redox centers also strongly sup-ports an α3β3 quaternary structure (see below). Theresidues involved in the contacts between the α subunitsare to a large extent conserved within the sequences of theαβ family of ring-hydroxylating dioxygenases, suggestingthat hexamers may be the common organization.

The overall shape of the full hexameric complex resem-bles that of a mushroom (Figure 2a), in which the stemconsists of the β3 subunits and the cap the α3 subunits.The height of the mushroom is 75 Å and the diametersof the cap and stem are 102 Å and 50 Å, respectively.The trimer of heterodimers forms a very compact quater-nary structure, with three completely separated active

572 Structure 1998, Vol 6 No 5

Figure 1

Stereo omit Fo–Fc electron-density map at2.25 Å resolution of crystal form II (3 × rms).The map is for the region around the catalyticiron center with the ligands His208, His213and Asp362. The iron ion was not present inthe preceding refinement (close to the finalstages of refinement) and the ligands wererefined as alanine residues. FE FE

Asp362 Asp362

His208 His208

His213His213

Structure

sites but with the redox centers from different subunitsconnected across the subunit borders (Figure 2). Thethree α and β subunits form tight trimers which bury alarge part of their accessible surface area. Upon trimer-ization, 18,022 Å2 of solvent-accessible area (as deter-mined using the program SurfVol in WhatCheck) areburied. When the αβ heterodimer is formed, a total of4396 Å2 are buried. There are only a few mainchain–mainchain hydrogen bonds between the α and β sub-units. However, there are several sidechain-mediatedhydrogen bonds that are not conserved within the family.

The structure of the b subunitThe β subunit is dominated by a long twisted six-strandedmixed β sheet (Figure 3a,b). The strand order is 2-1-6-5-4-3.Three of the strands (β3, β4 and β5) are 14 residues inlength, which is unusually long for such a small protein.On the concave side of the pleated sheet, there are threehelices and on the convex side of the sheet is an additionalN-terminal helix. The central part of the contact area tothe other β subunits in the trimer is formed by the pleatedsheet, reminiscent of porin trimers [31]. Most of thehelices are on the outside of the trimer and are in contactwith other parts of the twisted sheet in other subunits.The N-terminal helix contacts the α2–α3 hairpin inanother β subunit.

Reports in the literature have suggested that the β sub-units of toluate, toluene and biphenyl dioxygenases maycontribute to substrate specificity [32,33]. However, otherstudies have indicated that the α subunit alone may controlsubstrate specificity in the benzene, 2-nitrotoluene and2,4-dinitrotoluene dioxygenase systems ([34]; JV Parales,REP, SM Resnick and DTG, unpublished data). Recentwork in our laboratory indicates that the β subunit of NDOdoes not contribute to the substrate range, regiospecificityor stereospecificity of NDO (REP and DTG, unpublisheddata). As the β subunit does not appear to be directlyinvolved in catalysis or substrate specificity, its main roleis probably structural. The atoms of the β subunit closestto the iron at the active site and the Rieske center areabout 10 Å away.

The β subunit shows a remarkable three-dimensional sim-ilarity to two other functionally unrelated proteins —scy-talone dehydratase (ScDH) [35] and nuclear transportfactor 2 (NTF2) [36] (Figure 3c). Superposition of thethree structures gives a root mean square (rms) differencebetween the β subunit of NDO and ScDH of 1.8 Å for104 aligned Cα positions and 1.9 Å for 106 Cα atoms incommon with the NTF2 structure (calculated using thelsq options in O [37]). It is remarkable that these threeproteins have such similar structures but completely dif-ferent functions. ScDH and the NDO β subunits are trimerswith the same quaternary structure arrangement, whereasNTF2 exists as a dimer. It has been proposed that ScDH

is derived from an ancestral NTF2 [38] despite the totallack of sequence identity (6–10%).

Research Article Naphthalene 1,2-dioxygenase structure Kauppi et al. 573

Figure 2

Structure of the NDO α3β3 hexamer. (a) The hexameric complex canbe regarded as a mushroom with the α subunits as the cap and the β subunits as the stem. The three αβ units are colored green, blue andred; the β subunits and the Rieske domains of the blue and redsubunits are in lighter colors than the catalytic domains of theirrespective α subunits. The Rieske domain of the green subunit iscolored in yellow to highlight its interactions with the dark red catalyticdomain. The Rieske center and the catalytic iron are shown in CPKmodel representation (iron is red and sulfur yellow). (b) View along themolecular threefold axis. (Figures were made using the programMOLSCRIPT [80] and rendered using Raster 3D [81,82].)

A sequence alignment based on structural superpositionshows that only three residues are the same in all threestructures. Interestingly, Asp46 in the NDO β subunit isequivalent to Asp31 in ScDH, which is in an active-siteAsp–His couple that is also conserved in NTF2. In NDO,the residue corresponding to this histidine is Arg118,which forms a hydrogen bond to the carbonyl oxygen ofresidue 193 close to the C terminus. The latter is boundinside the subunit with several hydrogen bonds. In con-trast to ScDH, the NDO β subunit does not have an enzy-matic function. Instead of forming an active-site cleft as inScDH, the cone-shaped NDO β subunit is completelyfilled by the C terminus, which is bent inwards, filling the

center of the cone. The negatively charged C terminusbinds to the internal Arg159 and Arg176 residues. Theother two residues that are present in all three structuresare Ile120, which is involved in hydrophobic interactions,and Lys164, which is on the surface. These residues are,however, not conserved in the β subunits of all αβ dioxy-genases that have been sequenced to date. Thus, theirconservation seems to be a coincidence.

Compared with ScDH and NTF2, there are two pro-nounced extensions in the NDO β subunit. The 24 residuesat the N terminus are involved in trimer interactionsbetween the β subunits. The long loop formed by residues


Figure 3

Structure of the β subunit. (a) The secondarystructure; α helices are shown as cylindersand β strands as arrows. (b) The tertiarystructure. The structure is dominated by along twisted six-stranded mixed β sheetwrapped around three α helices. (c) Stereocomparison of the NDO β subunit (green)with ScDH (red) and NTF2 (blue).

Structure

C

N

12 34512

34 6

(a)

(c)

(b)

68–85 is also involved in β subunit interactions and makesinteractions with the Rieske domain of the α subunit aswell. Residue 75 forms a mainchain hydrogen bond to themainchain of residue 92 of the Rieske domain.

The structure of the a subunitThe α subunit contains the active site mononuclear iron andthe Rieske [2Fe–2S] center. The α subunit has 25 β strandsand 13 α helices that have been numbered consecutively.The α subunit can be divided into two distinct domains: aRieske domain that contains the [2Fe–2S] center and thecatalytic domain that contains the non-heme ferrous ironion (Figure 4).

The Rieske domainThe Rieske domain (residues 38–158) consists of fourseparate β-sheet structures composed of β strands 3–15(Figure 4a). Two of the sheets are arranged in a β-sand-wich topology with β14-β15-β3 packing against the cat-alytic domain on one side and against the sheet formed byβ13-β6-β5-β4-β9 on the other side. Two β-hairpin struc-tures protrude roughly perpendicular to this sheet andform two fingers that hold the [2Fe–2S] center at theirtips. The first of these fingers is formed by β7 and β8,where β7 makes a 90° turn away from β6 in the previoussheet. The first two iron ligands in the Rieske [2Fe–2S]center are positioned in the loop between β7 and β8. Thesecond of the β-sheet fingers is composed of β10-β11-β12.The second pair of iron ligands to the Rieske center isfound in the loop between β10 and β11.

The Rieske domain in NDO has a very similar structure tothe Rieske domain of a water-soluble fragment of thecytochrome bc1 complex [39]. Superpositioning of theRieske domains of NDO and cytochrome bc1 results in anrms difference of 1.8 Å for 84 Cα atoms (using the lsqoptions in O with standard parameters [37]; Figure 5a).The first and second β sheets are very similar, as is theloop between the iron ligands. Residues 104–128 incytochrome bc1 are located on one side of the domain faraway from the Rieske center. Equivalent residues are notpresent in NDO. Another major difference between thetwo Rieske domains is the presence of a long loop of 15residues (118–132) in NDO. The analogous region incytochrome bc1 has only three residues (175–177), whichare at a similar position as the highly conserved Trp106 inNDO. The sidechain of this residue covers one side of theRieske center and stacks against the iron ligand His83.Cytochrome bc1 has a proline at the equivalent position.The longer loop in the NDO domain is to a large extentinvolved in subunit interactions with the catalytic domainin a neighboring subunit. Likewise, residues 89–97 inNDO, which differ significantly from the correspondingresidues in cytochrome bc1 (146–154), are also involved ininteractions with a β subunit. Thus, the two large differ-ences close to the Rieske center between the two domains

are due to subunit interactions within the NDO hexamer.The two iron-binding loops in the Rieske center of thecytochrome bc1 fragment are connected to each other by adisulfide bridge that covers one side of the iron center.NDO does not need a stabilizing disulfide bridge becausethe Rieske center in NDO is completely buried by sur-rounding α and β subunits.

From the structural similarities, it seems highly likelythat both Rieske domains stem from an ancestral Rieskeprotein. An alignment based on the structural superposi-tion shows a sequence identity of 16% for the approxi-mately 100 residues common to both domains. As shownin Figure 5b, the sequence similarities are most pro-nounced around the metal center. Four of six residuesare the same around the first set of ligands while three offive are the same around the second set.

The structure of the chloroplast Rieske protein has beenpublished recently [40]. In contrast to the cytochrome bc1domain, the chloroplast Rieske domain is divided into twodistinct subdomains. The subdomain binding the [2Fe–2S]center is virtually identical in the two cases. The rest ofthe domain shows striking differences despite a commonfolding topology.

The Rieske [2Fe–2S] centerIn the Rieske [2Fe–2S] center, Fe1 is coordinated byCys81 and Cys101 while Fe2 is coordinated by the NDatoms of His83 and His104 (Figure 5c). The two sulfideions bridge the two iron ions and form a flat rhombicarrangement. There are hydrogen bonds from the main-chain nitrogens of residues 84 and 86 to one of the sulfideions and from the mainchain nitrogens of residues 104 and106 to the other sulfide ion. The four iron-ligandingsidechains are hydrogen bonded in a second coordinationshell. Both histidine ligands are hydrogen bonded to car-boxylate residues in a neighboring subunit. Asp205 bindsto the NE of His104 and Glu410 binds to the NE ofHis83. The sulfur atoms of Cys81 and Cys101 are hydro-gen bonded to the mainchain nitrogens of residues 83 and103, respectively. The coordination is consistent withspectroscopic data obtained for enzymes of this dioxyge-nase family [23,41].

The ligands, iron atoms and sulfide ions in the Rieske[2Fe–2S] center of NDO superimpose closely on thecytochrome bc1 Rieske center structure, which was deter-mined at 1.5 Å resolution [39]. The deviations in coordi-nation geometry are within experimental error (Table 1).In spite of the structural similarities, the redox potentialsof the two proteins appear to be different. The mito-chondrial cytochrome bc1 Rieske center has a redoxpotential of +300 mV [42]. The redox potentials foraromatic ring-hydroxylating dioxygenases are very dif-ferent from the one in the cytochrome complex. For


example, the redox potential of benzene dioxygenase is–112 ± 15 mV [22]. This value was obtained from anaero-bic titration of the [2Fe–2S] center and the midpointpotential was not affected by interaction with iron orferredoxin. In view of their similar structures, as deduced

from the sequence similarities between benzene dioxy-genase and NDO, it is not unreasonable to predict thatthe redox potential for NDO would be in the samerange. The redox potential for phthalate dioxygenase is–150 mV [43].


Figure 4

Structure of the α subunit. (a) The secondarystructure; α helices are shown as cylindersand β strands as arrows. (b) Stereo diagramshowing the flow of the chain. Iron and sulfuratoms are shown as orange and yellowspheres, respectively. (c) The α subunit withthe Rieske domain in red and the catalyticdomain in light purple. The iron–sulfur centerand the catalytic iron with their respectiveligands are shown in a ball-and-stickrepresentation.

A50

A400

A150 A250A300

A100

B2

A350

A445

B100

A200

B194 B50

A1

B150A50

A400

A150

A100

B2

A300A250

A350

A445

A200

B100

B194 B50

A1

B150

C

228

238

1110

12

4 9

3

56

7

8

13

15

1916 21 2420 18 17 2522

23

12

N12

34

5 6 7

8

9

10-1 10-2

1112

Catalytic domainRieske domain

1314

(a)

(b)

(c)

Structure


Figure 5

Structure of the Rieske domain.(a) Superposition of the Rieske domains ofNDO (purple) and cyctochrome bc1 (yellow).The Rieske center is shown for NDO in ball-and-stick representation. (b) Sequencealignment of the Rieske domains of NDO (top)and cytochrome bc1 based on their three-dimensional structures. Identical residues areshown in bold type and residues used for thesuperposition are underlined. (c) Stereodrawing showing the coordination of theRieske [2Fe–2S] center by Cys81, Cys101,His83 and His104; sulfur atoms are shown inyellow, nitrogen in blue, and iron in orange.

83

104

103101

10681

83

104

103101

81106

Structure

40 50 60 70 NWL FLTHDSLIPA PGDYVTAKMG IDEVIVSRQN ---------- ---------- SKI EIKLSDIPE- -GKNMAFKWR GKPLFVRHRT KKEIDQEAAV EVSQLRDPQH 74 80 90 100 110 120 ------DGSIRAFLNV CRHRGKTLVS VEAGNAKGFV CSYHGWGFGS NGELQSVPFE DLERVKKPEWVILIGV CTHLGCVPIA -NAGDFGGYY CPCHGSHYDA SGRIRKGPAP 130 140 150 KDLYGESLNK KCLGLKEVAR VESFHGFIYG CF -------------LNLEVPSY EFTSDDMVIV G-

(c)

(b)

(a)

There is a difference in the hydrogen bonding for theNDO and cytochrome bc1 Rieske centers. In the latter,one of the cysteine sulfur atoms accepts two hydrogenbonds, whereas each cysteine sulfur in NDO accepts asingle hydrogen bond. The histidine ligands in cyto-chrome bc1 are exposed to the solvent and the bound iron

is close to the surface of the protein. In NDO they arecompletely buried. This difference has been suggestedto be partly responsible for the differences in redoxpotential [39]. Furthermore, both histidines in the NDORieske center are hydrogen bonded to carboxylates, whichmay influence the redox potential.


Table 1

Iron coordination.

A B C Average bc1 b6f

Rieske centerBond distances (Å)

Fe1–S1 2.2 2.2 2.3 2.2 2.2 2.3Fe1–S2 2.3 2.3 2.3 2.3 2.3 2.4Fe2–S1 2.3 2.3 2.3 2.3 2.2 2.3Fe2–S2 2.1 2.2 2.2 2.2 2.3 2.3Fe1–Fe2 2.8 2.7 2.7 2.7 2.7 2.7S1–S2 3.5 3.6 3.6 3.6 3.6 3.7Fe1–SG Cys81 2.2 2.3 2.3 2.2 2.3 2.3Fe1–SG Cys101 2.2 2.2 2.2 2.2 2.2 2.2Fe2–ND1 His83 2.2 2.2 2.0 2.1 2.2 2.2Fe2–ND1 His104 2.0 2.0 2.1 2.0 2.1 2.2S1–N 84 3.3 3.3 3.1 3.2 3.2 3.4(S1–N 86)* (4.0 4.1 4.0) 4.0 3.6 3.6S2–N 104 3.4 3.5 3.4 3.4 3.2 3.2S2–N 106 3.7 3.5 3.6 3.6 3.6 3.4SG Cys81–N 83 3.5 3.5 3.5 3.5 3.5 3.6SG–O* 3.1 3.1SG Cys101–N 103 3.5 3.6 3.6 3.6 3.8 3.6

Bond angles (°)Fe1–S1–Fe2 76 72 73 74 75 71Fe1–S2–Fe2 77 74 74 75 74 72S1–Fe1–S2 103 108 106 106 106 106S1–Fe2–S2 103 106 106 105 106 109Cys81 SG–Fe1–Cys101 SG 112 110 110 110 106 110His83 ND1–Fe2–His104 ND1 89 89 93 91 91 91

Active siteBond distances (Å)

Fe–NE2 His208 2.0 2.0 2.1 2.0Fe–NE2 His213 2.1 2.1 2.0 2.1Fe–O1 Asp362 2.3 2.2 2.2 2.2Fe–O2 Asp362 2.6 2.6 2.6 2.6Fe–O Water 2.3 2.2 2.3 2.3Fe–O Asn201 OD1† 3.6 3.6 3.8 3.7

Bond angles (°)His208 NE2–Fe–His213 NE2 112 108 111 110His208 NE2–Fe–O1 Asp362 142 141 141 142His208 NE2–Fe–O2 Asp362 93 93 92 93His208 NE2–Fe–O Water 99 96 96 97His213 NE2–Fe–O1 Asp362 92 93 95 93His213 NE2–Fe–O2 Asp362 99 100 102 100His213 NE2–Fe–O Water 121 120 121 121O1 Asp362–Fe–O2 Asp362 53 54 54 53O1 Asp362–Fe–O Water 92 98 94 95O2 Asp362–Fe–O Water 129 134 129 131His208 NE2–Fe–O Asn201† 72 74 70 72His213 NE2–Fe–O Asn201 173 174 172 173O1 Asp362–Fe–O Asn201 87 88 89 88O2 Asp362–Fe–O Asn201 86 86 86 86O Water–Fe–O Asn201 52 54 52 53

*Not present in NDO. †Asn201 is a potential iron ligand. The distancesand angles at the two iron centers are listed for the three α subunits inthe asymmetric unit (A, B and C) that were refined with restraints. Thevalues for the Rieske center are compared with the corresponding

parameters for the Rieske centers in the mitochondrial cytochrome bc1complex (PDB code 1rie) [39] and the photosynthetic cytochrome b6fcomplex [40] (PDB code 1rfs).

The catalytic domainThe catalytic domain is dominated by a nine-strandedantiparallel β sheet that extends over the whole α subunit(Figure 4c). The strand order is, from the top of the ‘cap’,24-25-17-18-19-20-21-22-16. The sheet has tight connec-tions between strands on the side that packs against theRieske domain whereas all insertions between the strandsare on the other side. Three of the insertions betweenthe strands contains the ligands to the catalytic iron andform the active-site cleft. One side of the active site isformed by the long helix, α10, which is broken into twoparts at residues 353–355 that loop out from the helix.The helix covers one part of the sheet in its full length.Asp362, a ligand to the catalytic iron, is positioned a half-turn after the break. The two other structural motifs thatform the active site extend like two fingers from differ-ent directions and pack against the long broken α10helix. Between strands β18 and β19 there are two helices(α8 and α9) covering one edge of the active site.Between strands β16 and β17, three helices (α5, α6 andα7) together with the loop (residues 228–234) form thethird part of the active site. This part contains two of thethree ligands to the active-site iron, His208 and His213,which are located in helix α6. Asp205, which is probablyimportant for electron transfer (see below), is found onthe tight turn between the α5 and α6 helices. The struc-ture of the active site is thus formed by the sheetforming the base with one helix packing against it andwith the two other structural elements sticking out fromthe sheet against this helix. The additional structural ele-ments of the domain form the top part of the mushroomon both sides of the main sheet. The additional small βsheet with three short antiparallel strands is also locatedin this part of the mushroom cap.

While the β subunit and the Rieske domain have similarstructural relatives, the structure of the catalytic domainseems to represent a new fold. A structural homologysearch of the Protein Data Bank using the DALI [44] andDEJAVU [45] programs found no protein structuressimilar to the catalytic domain of NDO.

Coordination of the iron at the active siteThe iron atom at the active site is coordinated by His208,His213, bidentately by Asp362 and by a water molecule.The geometry can be described as a distorted octahedralbipyramid with one ligand missing (Figure 6a; Table 1).Both histidines are hydrogen bonded to aspartic acid side-chains. His208 is hydrogen bonded to Asp205 and His213to Asp361. Both aspartic acids in the second coordinationshell, as well as the ligands, are completely conserved inthis family of enzymes. Asn201 is located close to the freeligand position of an octahedron but this distance is toogreat (3.75 Å) for this residue to be a ligand. If Asn201 isconsidered to be a ligand, a distorted octahedral geometrycan be formed (Figure 6b; Table 1). The square plane is

then formed by the bidentate coordinating Asp362, the NEatom of His208 and the water molecule. The axial ligandsare the NE atom of His213 and the sidechain oxygen ofAsn201 at the considerably longer distance.

Site-directed mutagenesis studies on toluene dioxygenase(TDO) led to the suggestion that the conserved aminoacids Glu200, Asp205, His208 and His213 may serve asligands to iron at the active site of this enzyme [11]. Theloss of TDO activity when Asp219 of TDO (analogous toAsp205 of NDO) is replaced by alanine can be explainedby its putative role in electron transfer (see below). Glu200,which was also identified as an essential residue on thebasis of site-directed mutational studies of TDO [11],forms contacts between α subunits in NDO and thus maybe important for α–α subunit interactions instead of ironcoordination as previously proposed.

Extensive biophysical studies of the active-site iron ofphthalate dioxygenase (PDO) have been carried out[46–48]. The results suggest that the iron exists in a dis-torted octahedral six-coordinate ferrous complex whichchanges to one [48] or two [47] five-coordinate specieswhen phthalate binds to the enzyme. The change incoordination state from six to five is attributed to aphthalate binding-dependent conformational change thatdisplaces a labile ligand and provides a site for oxygenbinding and activation [46–48]. A change in coordina-tion state can be envisaged for NDO with Asn201 as alabile ligand. However, if this is correct, the coordinationchange in PDO must involve a different residue fromthat in NDO. According to our sequence alignment, theresidue corresponding to Asn201 appears to be a glycinein PDO.

There is precedence for asparagine as an iron ligand inthe binuclear di-iron carboxylate enzyme purple acidphosphatase [49]. Asn201 is not completely conserved inthe dioxygenase family, but is replaced by a glutamine inmany of the enzymes. With slight differences in geome-try, a glutamine residue might form a similar coordina-tion with iron as the asparagine. This type of coordinationis seen in isopenicillin N synthase, where manganese sub-stituted for the active-site iron is coordinated by two his-tidines, one aspartate and a glutamine [50]. In the enzyme–substrate complex, the glutamine sidechain is replacedby the substrate [51].

The coordination of iron at the active site of NDO resem-bles to some extent that seen in other iron enzymes. Theextradiol-cleaving dioxygenases have ferrous iron coordi-nated by two histidines, one glutamate (monodentate) andtwo water molecules [52,53]. The same type of coordina-tion has also been found in tyrosine hydroxylase andrelated enzymes [54]. Lipoxygenases have iron coordi-nated by three histidines and the C terminus [55].


In many iron enzymes that utilize dioxygen, iron is coordi-nated by residues from distorted helices. Some ligand-containing helices in ribonucleotide reductase R2 [56,57],methane monooxygenase [58], cytochrome P450 [59], andlipoxygenase-1 [55] are to some extent distorted. Distor-tion is usually caused by the presence of extra residuesforming π-type helix turns. This type of distortion hasbeen considered to provide an oxygen pathway or toincrease conformational flexibility [56,60]. The situation isquite different in NDO. One of the ligands in NDO,Asp362, is located in a helix that is distorted to a muchlarger extent than in the enzymes mentioned above. Inthis helix, some of the residues in the center of the helix(Gly353, Pro354, Ala355 and Gly356) loop out, and thehelix then continues with a small tilt angle. The pro-truding residues in the NDO helix appear to anchor the αsubunit to the β subunit. Three regions of the catalyticdomain interact with the β subunit. These include the

long broken α10 helix, which contains one of the ligandsto the active-site iron, the β16 strand and the α7 helix.Interestingly, the residues that protrude out at the helixbreak are the main participants in the interactions with theβ subunit. It is primarily the C-terminal residues of the βsubunit, which are located inside the structure, that formmost of these contacts. These interactions do not seem tobe particularly important for keeping the enzyme in anactive conformation. Mutants lacking the last nine or24 C-terminal residues of the β subunit showed onlyslightly impaired activity, while a number of othermutants that seem to affect the packing of the β trimerhave a significant negative effect on activity (REP andDTG, unpublished data).

The active siteIron at the active site is located approximately 15 Å fromthe surface of the protein. From the surface, a narrow gorge


Figure 6

Coordination of iron at the active site of NDO.(a) Stereo drawing showing that thecoordination with His208, His213, Asp362(bidentate) and a water molecule results in aslightly distorted trigonal bipyramidalgeometry. The free position of an octahedronis directed towards Asn201, which is located3.75 Å from iron at the active site. Hydrogenbonds between the histidine and aspartateligands in the second coordination shell areshown as red dashed lines; atoms are instandard colors. (b) Stereo drawing of iron atthe active site with Asn201 as a hypotheticalligand shows that the coordination geometryis close to octahedral.

202

358

213

3.7Å

205201

362

361

208

202

358

3.7Å

213

201205

362

361

208

205

208

201

202

362213

361

205

208

201

202

362213

361

(a)

(b)

Structure

provides substrates with access to the iron ion (Figure 7). Atthe upper part of this gorge, two loops (residues 253–265and 223–240) shield part of the entrance and may act as lidscovering the substrate channel. The channel continues pastthe iron atom. The distance from the bottom of the cleft tothe surface of the protein is about 30 Å.

The narrowest part of the long channel is located near theactive-site iron where the channel is about 5 Å wide. Thispart of the channel contains the active-site iron and its twohistidine ligands, and also Asn201, Phe202 and Phe352.The region of the channel close to the active-site iron islined primarily by hydrophobic residues, consistent withNDO’s preference for hydrophobic substrates [6]. Thesehydrophobic residues probably position the substrate forthe reaction to take place. The pocket below the active-site iron is 7–8 Å wide, and is lined by the iron ligandAsp362, Ile191 (at the bottom of the channel), Pro198,Trp316, Val326, Asn363, Met366, Tyr103 (from a neigh-boring α subunit) and a conserved salt bridge betweenLys314 and Glu359. The residues in the gorge above theactive-site iron are Ala206, Val209, Leu217, Val260, Asn297,Leu307 and Trp358. The only polar sidechains in thecenter of the channel are the ligands to the catalytic ironand Asn201. Glu359 is approximately 7 Å from the active-site iron. The residues lining this plausible substrate cleftare not conserved in other dioxygenase family members.This is not unexpected, as the substrate specificities ofthe different members of the family vary.

Electron transfer between the Rieske center and the iron atthe active siteThe distance between the Rieske center and the iron atthe active site within a single α subunit is 43.5 Å. However,the distance is considerably shorter, 12 Å, between theRieske center of one α subunit and the active-site iron in aneighboring α subunit in the hexamer (see Figure 2). Thisis consistent with an estimated minimum separation ofapproximately 10 Å between the Rieske and mononucleariron centers for phthalate dioxygenase [46].

The two centers in the symmetry-related NDO α subunitsare connected through hydrogen bonds to Asp205. Thisresidue binds to His208, a ligand to the active-site iron, andto His104, a ligand to the Rieske center in a neighboringα subunit (Figure 8). This is in all probability the mainelectron-transfer pathway. The importance of Asp205 wasrealized earlier in a site-directed mutational study of TDOwhere the aspartate corresponding to Asp205 was changedto alanine and the enzyme lost all activity [11]. However,the structure of NDO suggests that this residue is not aniron ligand but may be involved in electron transfer.

There are also other hydrogen bonds between the two sub-units in the neighborhood of the iron centers. The side-chain of Trp106 and the mainchain carbonyl of residue 117

of the Rieske domain form hydrogen bonds to Tyr207close to the active site. Another connection between sub-units in this area is the charged hydrogen bond betweenthe sidechains of Arg84 of the Rieske domain and Glu200of the catalytic domain. All of these residues are conservedin all members of the αβ dioxygenase family. Tyr103 of theRieske domain and Asn201 form hydrogen bonds betweensubunits. These are not completely conserved residues.

Possible binding site for ferredoxin and the electron-transfer pathway to the Rieske centerThere is a pronounced depression in the Rieske domainclose to the [2Fe–2S] center that seems to be well suitedfor interactions with the ferredoxin component of the NDOsystem. The depression is formed by residues Lys97,Gly98, Val100, Gln115 and Ser116 from the Rieskedomain, Pro118 and Trp211 from the catalytic domain andresidues Ser75, Arg77, Arg 78, Pro105 and Trp108 fromthe β subunit. The arginine residues are located at theouter rim of the depression. The mainchain atoms of theresidues of the Rieske domain, and in particular the main-chain atoms of Cys101, are closest to the iron center. Theelectrons are probably transferred via these mainchainatoms to the Rieske center.

Substrate hydroxylationThe reaction mechanism of NDO is not known. Thedioxygenase reaction catalyzed by NDO is stereospecific,


Figure 7

Surface plot of NDO. A long narrow gorge leads from the surface ofthe catalytic domain to the iron at the active site (red). The probablesubstrate-binding site is the dark region at the bottom of the cleft.

which requires the substrate to be bound close to andfixed in a specific position relative to the iron-boundactivated dioxygen species during catalysis. The cis stereo-specificity implies that both oxygen atoms are added tothe ring from the same side. It is possible that the C1–C2of the substrate and the activated dioxygen line up parallelto each other. The iron in the active enzyme is believed tobe in the ferrous state but the oxidation state of the iron inthe crystals has not been determined. Dioxygen can bindto the ferrous ion and can be activated by accepting oneelectron from iron. The reactive species can then be asuperoxide radical like species or a peroxide. Cleavage ofthe dioxygen molecule may result in ferryl intermediatesas in other iron enzymes [51]. Whether proximity of thepositively charged iron can polarize the inert substrateremains to be shown.

The hydroxylation reaction requires a total input of twoelectrons. They can be transferred from the Rieskecenter to the active-site iron during the reaction, but atwhich step of the reaction this happens is not yet known.Investigations of the reduced form of the Rieske centerhave shown that the iron atom that is bound by two his-tidines is reduced to a ferrous ion [41]. An active siteferrous ion can be regenerated for the next activity cycleby electron transfer from ferredoxin via the Rieskecenter. The active enzyme may hence be considered asone where the catalytic iron is in the ferrous state and theRieske center is also reduced. There are no residues at

the active site that can facilitate proton donation, so it islikely that the protons come from water.

Biological implicationsNaphthalene 1,2-dioxygenase (NDO) belongs to a familyof bacterial enzymes that have an essential role in therecycling of carbon in nature. These enzymes are espe-cially important in the degradation of aromatic hydrocar-bons and related environmental pollutants. Knowledge ofthe NDO reaction mechanism is thus important in thedevelopment of bioremediation strategies for cleaning upenvironments contaminated with hazardous aromaticcompounds. An attractive alternative to bioremediationis the application of ‘green chemistry’, which refers tothe production of industrial chemicals by processes thatdo not generate hazardous waste. For example, a recom-binant strain of Escherichia coli expressing NDO, hasbeen used to synthesize indigo dye from glucose. cis-Arene diols produced by NDO and toluene dioxygenasehave been used in the synthesis of many products of bio-logical and economic importance.

The structure of NDO is the prototype for all members ofthe family of aromatic-ring-hydroxylating dioxygenases.The enzyme exists as a mushroom-shaped α3β3 hexamer.The α subunit contains a Rieske [2Fe–2S] center thattransfers electrons to mononuclear iron at the active siteof the enzyme. The structure of the domain containing theRieske center is strikingly similar to that of the soluble


Figure 8

Proposed route of electron transfer betweenadjacent α subunits. The Rieske center in αsubunit A is connected to the catalytic centerof the adjacent α subunit B. The connection isthrough Asp205, which is hydrogen bondedto His104 of the Rieske center and His208 atthe active site.

Fe

S Fe

S

C

O

S

Fe

α subunit B

NNH

NHNNH

N

Catalytic iron site

OH2

C

O

O

HN

N

O

S

Rieske [2Fe–2S]cluster

Asp205

α subunit A

Asp362

His83

Cys81

Cys101

His104

His208

His213

Structure

fragment of the Rieske [2Fe–2S] protein from the mito-chondrial cytochrome bc1 complex. This observation sug-gests that both Rieske [2Fe–2S] domains may be derivedfrom a common ancestral iron–sulfur protein. The struc-ture of the catalytic domain is characterized by a noveldomain fold that is dominated by a nine-stranded β sheet.The structure suggests cooperativity between adjacent αsubunits, where electrons from the Rieske [2Fe–2S]center in one α subunit are transferred to mononucleariron at the active site in the adjacent α subunit. The activesite is located at the bottom of a narrow channel wheremononuclear iron is coordinated by two histidines and oneaspartate residue. The smaller β subunit does not containany cofactors and may have a structural role in theholoenzyme.

Materials and methodsStructure determination and refinement of crystal form 1NDO was purified and crystallized as described [27]. The crystals belongto the rhombohedral space group R32 with cell dimensionsa = b = 178.9 Å, c = 323.6 Å (in the hexagonal setting). The crystalsgrew in 0.5 M sodium sulfate, 0.1 M Mes, pH 5.8, 5 mM nickel(II)sulfateand 22 mM 1,6-hexanediol in a cold room (6° C). Nickel was essential togenerate crystal form 1. The crystals grew in three days to a full size of0.5 × 0.5 × 0.1 mm as half or whole hexagonal plates. The crystals con-tained two αβ dimers per asymmetric unit. Their relationship is equivalentto a translation of approximately 1/2 in the z-direction as seen in a nativePatterson map (the peak is about 45% the size of the origin peak).

Data collectionThe data were collected on crystals frozen in 25–30% ethylene glycolunder cryogenic conditions (100K). DENZO and SCALEPACK [61]were used to index and merge each individual data set. Native datasets collected from different crystals proved to be non-isomorphous.The two best native data sets had a difference on F values of 23.1% inspite of small differences in cell dimensions.

Heavy-atom derivativesFor each derivative, the most similar native data sets were used. TheEMTS derivative gave interpretable Patterson maps and minor siteswere found from difference Fourier maps. The phases from this deriva-tive were used to locate sites in the other derivatives. Pt and Ta deriva-tives were investigated but hardly improved the phasing. Derivativedata were paired with the most isomorphous native data and refinedusing MLPHARE [62]. The phase probabilities for the different datasets were then combined using SIGMAA [63] which increased thequality of the electron density map. Refinement of heavy-atom sites inSHARP [64] together with solvent flipping in SOLOMON [65]improved the map further and β strands were seen in the bones repre-sentation. From this map, the relationship between the two non-crystal-lographically related molecules was confirmed.

MADThe structure of NDO was determined using a selenomethionine(SeMet)-substituted enzyme using the multiwavelength anomalous dif-fraction method (MAD). The SeMet-derivatized protein was generatedas described [27]. To produce large enough crystals, the pH wasraised to 6.0 and the temperature to 8°C. Otherwise, the same condi-tions were used as for the native crystals. A data set was collected ona rotating anode with a RAXIS-II imaging plate detector. Using theavailable heavy-atom phases, 26 of the 30 Se sites in the asymmetricunit were found after several consecutive rounds of phase refinement.

MAD data were collected on BM14 at ESRF in Grenoble in 15°batches at the peak wavelength, 0.97828 Å, at the inflection point,

0.97859 Å, and at the the remote wavelength, 0.90500 Å until a full60° data set had been collected.The EXAFS spectrum was used tochoose the wavelengths used in data collection [27]. The anomalousdifferences varied between the different batches of the MAD dataset. Therefore, each of the four batches were introduced individuallyin SHARP as incomplete data sets. It was absolutely crucial to dividethe batches. If all data were processed as one set, the anomaloussignal vanished and the resulting map was of bad quality.

In spite of the large content of the asymmetric unit (1286 residues), thecontribution by 20 Se atoms (only 20 of the 26 Se atoms refined wellusing only the MAD data) gave excellent phases and a beautiful 3.5 Åmap was calculated based on MAD experimental phases (combinedwith the Hendrickson–Lattmann coefficients input as external phaseinformation to SHARP from the EMTS derivatives and the home Se dataused as a derivative) and solvent flipping. Crystallographic details aregiven in Table 2. The electron density of the β strands seen in previousmaps was improved and the map contained many additional features.Connectivity between secondary structure elements began to appear.

At an early stage a huge peak (7σ) in the Fourier map was found whichwas interpreted as the [2Fe–2S] redox center. The coordinates for theiron ligands and the Rieske [2Fe–2S] center from cytochrome bc1 [39]could be modeled into the electron-density map and this was the start-ing point for tracing the sequence. The phases were slowly extended to3.0 Å using SOLOMON and most sidechains became visible. With thehelp of the SeMet positions the sequence could easily be traced. Atthis stage, all residues were built except a loop between residues225–235, four residues at the C terminus of the α subunit and the N-terminal methionine in the β subunit. All map and data manipulationswere done using the programs mapman and dataman [66].

Structure refinementThe structure was refined using the beta test version of the programCNS ([67] and AT Brünger, personal communication). The structure con-sists of two αβ dimers noncrystallographically (NCS) related by a transla-tion of about 0.5 in c. Electron density for one of the NCS units wasmuch better than density for the other. This has been observed in otherstructures [68]. This can be attributed to either static or dynamic disorderof one of the molecules in the asymmetric unit. The reason for this wasobvious once the structure was fully determined. The number of packinginteractions of the hexamer formed by this NCS unit was very low.Refinement started therefore by first model building the good NCS unitand then generating the other and this process was repeated at allstages of model building. We started refinement with positional NCSrestraints in CNS, and did several rounds of torsion angle dynamics[69–72]. A maximum likelihood target using amplitudes and phase prob-ability distribution was used for refinement [73,74]. The final refinementwas done using a 2.5 Å data set collected at BM14, at ESRF, Grenoble.The final R factor was 29.2% (for 49,712 reflections between 20 and2.5 Å and for reflections with Fobs > 3 × σ(Fobs)) and R free 33.3% (2654reflections). Bulk-solvent correction was applied as implemented in CNS.Only 16 residues of the surface regions were not in good density. At thispoint, there were 485 water molecules in every NCS unit. The geometryand the electron density of one of the NCS units was good.

Structure determination and refinement of crystal form 2NDO was purified as described [27]. It was crystallized by the hangingdrop method. The 4 µl drop was made of 2 µl of protein (40 mg/ml) and2 µl of the reservoir solution. The precipitant solution was 2.31 M(NH4)2SO4, 0.1 Mes pH 6.0 and 2% PEG. The crystals of form 2 wererectangular rods belonging to the orthorhombic space group I222 withcell dimensions a = 105.0 0Å, b = 174.0 Å, c = 282.5 Å. The cell volumesuggested that there are three αβ dimers in the asymmetric unit.

Data collectionData were collected on the crystals frozen in 30% glycerol undercryogenic conditions (100 K) at 0.95 Å wavelength. DENZO andSCALEPACK [61] were used to process and scale data. The crystals


diffracted to 2.1 Å resolution. However, we collected good data tobetter than 2.25 Å resolution. A total of 157 frames of data each of 1°oscillation were collected. Details are given in Table 1.

Structure solutionThe structure was solved by molecular replacement using the programAMORE [75]. The search model was the α subunit and the β subunitfrom the good NCS unit of the refined crystal form 1. The structure wassolved easily and the program found three α and three β subunits in theasymmetric unit. The starting correlation factor was 0.70 and the Rfactor 0.314, suggesting that the solution was correct. It was immedi-ately noticed that there was an α3β3 hexamer in the asymmetric unit.

Structure refinementThe structure was refined using the maximum likelihood programREFMAC [76], in combination with ARP [77], which was used to addwaters automatically. The iron centers were refined without anyrestraints. The van der Waals radii of these hetero atoms were set tovery small values. One round of refinement consisted of 10 cycles ofREFMAC (after each cycle of REFMAC, waters were added anddeleted automatically using ARP) with threefold NCS restraints, fol-lowed by model building using O and OOPS [37,45]. Averageddensity modified maps [78] revealed the missing loops and a few othermissing residues from form 1. These were added and further refine-ment continued as described above. All amino acid residues except theC-terminal residue of the α subunit and the N-terminal residue of theβ subunit were visible in the electron-density map and have beenmodeled. Each dimer contained 512 water molecules. The R factor andR free are 19.4% (for 110,479 reflections between 30 and 2.2 Å reso-lution) and 23.8% (5868 reflections) respectively. Bulk-solvent correc-tion was applied as implemented in the program REFMAC. Restrainedindividual isotropic B factors were refined.

Validation of the structure was done using the programs Procheck [79]and WhatIF (Vriend), using the validation server at EMBL, Heidelberg,Germany. Engh and Huber parameters were used for refinement. Therms deviations in bond lengths and bond angles are 0.02 Å and 2.1°,respectively. The overall coordinate ESU (estimated standard uncer-tainty) [76] based on R free is 0.2 (ESU based on maximum likelihoodis 0.13). These validations show that the structure is of high quality.None of the distances or angles changed significantly between crystal

forms 1 and 2, indicating that the structures are very similar. The twocrystal forms differ only in their packing. Data from crystal form 2 isreported as it is the higher-resolution structure.

Accession numbersThe coordinates and structure factors of naphthalene 1,2-dioxygenasehave been deposited in the Protein Data Bank with the accessioncodes 1ndo and 1ndosf, respectively.

AcknowledgementsWe thank Eric de la Fortelle for help with SHARP and Andy Thompson atESRF for help in data collection. This work was supported by grants fromthe the Swedish Natural Science Research Council (to HE and SR) and theSwedish Research Council for Forestry and Agriculture (to HE) and by USPublic Health Service Grant GM29909 from The National Institute ofGeneral Medical Sciences (to DTG).

References1. Butler, C.S. & Mason, J.R. (1997). Structure–function analysis of the

bacterial aromatic ring-hydroxylating dioxygenases. Adv. Microbial.Physiol. 38, 47-84.

2. Carless, H.A.J. (1992). The use of cyclohexa-3,5-diene-1,2-diols inenantiospecific synthesis. Tetrahedron Asymmetry 3, 795-826.

3. Sheldrake, G.N. (1992). Biologically derived arene cis-dihydrodiols assynthetic building blocks. In Chirality in Industry: The CommercialManufacture and Application of Optically Active Compounds.(Collins, A.N. & Crosby, J., eds.), pp. 127-166, John Wiley & SonsLtd., Chichester, UK.

4. Brown, S.M. & Hudlicky, T. (1993). The use of arene-cis diols insynthesis. In Organic Synthesis: Theory and Applications (Hudlicky,T., ed.) pp. 113-176, JAI Press, Greenwich, CT.

5. Hudlicky, T. & Reed, J.W. (1995). An evolutionary perspective ofmicrobial oxidations of aromatic compounds in enantioselectivesynthesis: history, current status, and perspectives. In Advances inAsymmetric Synthesis. (Hassner, A., ed.), pp. 271-312, JAI Press Inc.,Greenwich, CT.

6. Resnick, S.M., Lee, K. & Gibson, D.T. (1996). Diverse reactionscatalyzed by naphthalene dioxygenase from Pseudomonas sp. strainNCIB 9816. Indust. Microbiol. 17, 438-457.

7. Kolb, H.C., VanNieuwenhze, M.S. & Sharpless, K.B. (1994). Catalyticasymmetric dihydroxylation. Chem. Rev. 94, 2483-2547.

8. MacDonald, W. (1997). Bacteria to Dye For. Financial Times (NewYork), 15 May, p. 23.


Table 2

Crystallographic statistics.

Crystal Resolution Source, detector Rsym (%) Unique reflections Ι/σ(Å) All data Last shell All data Last shell Multiplicity Overall Last shell

completeness (%) completeness (%)

Native

NATX11 2.8 BW7B/DESY, MAR 11.8 32.8 47,184 (95) 2198 (90) 4.1 8.5 1.7NATEM3 2.5 BM14/ESRF, MAR 9.5 34.5 61,051 (96) 3671 (88) 3.3 7.3 1.6

Derivatives

EMTS2 3.5 Home, RAXIS-IIc 8.0 15.3 20,882 (81) 971 (76) 2.3 9.0 6.7SeMet 3.5 Home, RAXIS-IIc 10.4 20.4 25,121 (99) 1243 (90) 3.4 8.2 3.5SeMet peak 3.3 BM14/ESRF, MAR 10.2 21.9 28,835 (99) 1841 (99) 3.1 9.5 3.6SeMet edge 3.3 BM14/ESRF, MAR 9.7 21.2 28,838 (99) 1841 (99) 3.1 9.9 3.6SeMet remote 3.3 BM14/ESRF, MAR 10.0 22.0 28,240 (99) 1837 (99) 3.1 7.5 3.2I222 form 2.2 BM14/ESRF, MAR 9.1 38.6 117,067 (96) 5920 (97) 6.2 14.6 3.6

The table contains the final statistics of the MIR and MAD refinementas determined in the program SHARP. The number of Hg sites wasfour and the number of Se sites was 20. The total figures of merit for

acentric and centric reflections were 0.56 (0.49) and 0.47 (0.39),respectively. Values given in parentheses are for the last shell. After 20cycles of SOLOMON, the figure of merit was 0.79 (0.75).

9. Ensley, B.D., Ratzkin, B.J., Osslund, T.D., Simon, M.J., Wackett, L.P. &Gibson, D.T. (1983). Expression of naphthalene oxidation genes inEscherichia coli results in the biosynthesis of indigo. Science 222,167-169.

10. Murdock, D., Ensley, B.D., Serdar, C. & Thalen, M. (1993). Constructionof metabolic operons catalyzing the de novo biosynthesis of indigo inEscherichia coli. Biotechnology NY 11, 381-386.

11. Jiang, H., Parales, R.E., Lynch, N.A. & Gibson, D.T. (1996). Site-directed mutagenesis of conserved amino acids in the alpha subunitof toluene dioxygenase: potential mononuclear non-heme ironcoordination sites. J . Bacteriol. 178, 3133-3139.

12. Resnick, S.M. & Gibson, D.T. (1996). Regio- and stereospecificoxidation of fluorene, dibenzofuran, and dibenzothiophene bynaphthalene dioxygenase from Pseudomonas sp. strain NCIB 9816.Appl. Environ. Microbiol. 62, 4073-4080.

13. Haigler, B.E. & Gibson, D.T. (1990). Purification and properties offerredoxinNAP, a component of naphthalene dioxygenase fromPseudomonas sp. strain NCIB 9816. J. Bacteriol. 172, 465-468.

14. Haigler, B.E. & Gibson, D.T. (1990). Purification and properties ofNADH-ferredoxinNAP reductase, a component of naphthalenedioxygenase from Pseudomonas sp. strain NCIB 9816. J. Bacteriol.172, 457-464.

15. Ensley, B.D. & Gibson, D.T. (1983). Naphthalene dioxygenase:purification and properties of a terminal oxygenase component. J.Bacteriol. 155, 505-511.

16. Webb, E.C. (1992). Recommendations of the NomenclatureCommittee of the International Union of Biochemistry and MolecularBiology. In Enzyme Nomenclature. Academic Press, New York.

17. Suen, W.-C. & Gibson, D.T. (1994). Recombinant Escherichia colistrains synthesize active forms of naphthalene dioxygenase and itsindividual α and β subunits. Gene 143, 67-71.

18. Suen, W.-C. & Gibson, D.T. (1993). Isolation and preliminarycharacterization of the subunits of the terminal component ofnaphthalene dioxygenase from Pseudomonas putida NCIB 9816-4. J.Bacteriol. 175, 5877-5881.

19. Cline, J.F, Hoffman, B.M., Mims, W.B., LaHaie, E., Ballou, C.P. & Fee,J.A. (1985). Evidence for N coordination to iron in the [2Fe–2S]clusters of Thermus Rieske protein and phthalate dioxygenase fromPseudomonas. J. Biol. Chem. 260, 3251-3254.

20. Tsang, H.-T., Batie, C.J., Ballou, D.P. & Penner-Hahn, J.E. (1985). X-ray absorption spectroscopy of the [2Fe–2S] Rieske cluster inPseudomonas cepacia phthalate dioxygenase. Determination of coredimensions and iron ligation. Biochemistry 28, 7233-7240.

21. Tsang, H.-T., Batie, C.J., Ballou, D.P. & Penner-Hahn, J.E. (1996).Structural characterization of the mononuclear iron site inPseudomonas cepacia phthalate DB01 dioxygenase using X-rayabsorption spectroscopy. J. Biol. Inorg. Chem. 1, 24-33.

22. Geary, P.J., Saboowalla, F., Patil, D. & Cammack, R. (1984). Aninvestigation of the iron-sulphur proteins of benzene dioxygenase fromPseudomonas putida by electron-spin-resonance spectroscopy.Biochem. J. 217, 667-673.

23. Shergill, J.K., Joannou, C.L., Mason, J.R. & Cammack, R. (1995).Coordination of the Rieske-type [2Fe–2S] cluster of the terminaliron–sulfur protein of Pseudomonas putida benzene 1,2-dioxygenase,studied by one- and two-dimensional electron spin-echo envelopemodulation spectroscopy. Biochemistry 34, 16533-16542.

24. Neidle, E.L., Hartnett, C., Ornston, L.N., Bairock, A., Rekik, M. &Harayama, S. (1991). Nucleotide sequences of the Acinetobactercalcoaceticus benABC genes for benzoate 1,2-dioxygenase revealevolutionary relationships among multicomponent oxygenases. J.Bacteriol. 173, 5385-5395.

25. Simon, M.J., et al., & Zylstra, G. J. (1993). Sequences of genesencoding naphthalene dioxygenase in Pseudomonas putida strainsG7 and NCIB 9816-4. Gene 127, 31-37.

26. Parales, J.V., Kumar, A., Parales, R.E. & Gibson, D.T. (1996). Cloningand sequencing of the genes encoding 2-nitrotoluene dioxygenasefrom Pseudomonas sp. JS42. Gene 181, 57-61.

27. Lee, K., Kauppi, B., Parales, R.E, Gibson, D.T. & Ramaswamy, S.(1997). Purification and crystallization of the oxygenase component ofnaphthalene dioxygenase in native and selenomethionine-derivatizedforms. Biochem. Biophys. Res. Commun. 241, 553-557.

28. Ramakrishnan, C. & Ramachandran, G.N. (1965). Stereochemicalcriteria for polypeptide and protein chain conformation. Biophys. J. 5,909-933.

29. Mason, J.R. & Cammack, R. (1992). The electron-transport proteinsof hydroxylating bacterial dioxygenases. Annu. Rev. Microbiol. 46,277-305.

30. Nakatsu, C.H., Fulthorpe, R.R., Holland, B.A., Peel, M.C. & Wyndham,R.C. (1995). The phylogenetic distribution of a transposabledioxygenase from the Niagara River watershed. Mol. Ecol. 4, 593-603.

31. Weiss, M., S., Kreusch, A., Schiltz, E., Nestel, U., Welte,W.,Weckesser, J., Schulz, G.E. (1991). The structure of porin fromRhodobacter capsulatus at 1.8 Å resolution. FEBS Lett. 280, 379-382.

32. Harayama, S., Rekik, M. & Timmis, K.N. (1986). Genetic analysis of arelaxed substrate specificity aromatic ring dioxygenase, toluate 1,2-dioxygenase, encoded by TOL plasmid pWWO of Pseudomonasputida. Mol. Gen. Genet. 202, 226-234.

33. Hirose, J., Suyama, A., Hayashida, S. & Furukawa, K. (1994).Construction of hybrid biphenyl (bph) and toluene (tod) genes forfunctional analysis of aromatic ring dioxygenases. Gene 138, 27-33.

34. Tan, H.-M. & Cheong,C.-M. (1994). Substitution of the ISPa subunit ofbiphenyl dioxygenase from Pseudomonas results in a modification ofthe enzyme activity. Biochem. Biophys. Res. Commun. 204, 912-917.

35. Lundqvist, T., Rice, J., Hodge, N., Basarab, G., Pierce, J. & Lindqvist,Y. (1994). Crystal structure of scytalone dehydratase — a diseasedeterminant of the rice pathogen, Magnaporthe grisea. Structure 2,937-944.

36. Bullock, T.L., Clarkson, W.D., Kent, H.M. & Stewart, M. (1996). The1.6 Å resolution crystal structure of nuclear transport factor 2 (NTF2).J. Mol. Biol. 260, 422-431.

37. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. (1991). Improvedmethods for building protein models in electron density maps and thelocation of errors in these models. Acta Cryst. 47, 110-119.

38. Murzin, A.G. (1996). Structural classification of proteins: newsuperfamily. Curr. Opin. Struct. Biol. 6, 386-394.

39. Iwata, S., Saynovits, M., Link, T.A. & Michel, H. (1996). Structure of awater soluble fragment of the ‘Rieske’ iron–sulfur protein of bovineheart mitochondrial cytochrome bc1 complex determined by MADphasing at 1.5 Å resolution. Structure 4, 567-579.

40. Carrell, C.J., Zhang, H., Cramer, W.A. & Smith, J.L. (1997). Biologicalidentity and diversity in photosynthesis and respiration: structure ofthe lumen-side domain of the chloroplast Rieske protein. Structure 5,1613-1625.

41. Gurbiel, R.J., et al., & Hoffman, B.M. (1996). active-site structure ofRieske-type proteins: electron nuclear double resonance studies ofisotopically labeled phthalate dioxygenase from Pseudomonascepacia and Rieske protein from Rhodobacter capsulatus andmolecular modeling studies of a Rieske center. Biochemistry 35,7834-7845.

42. Trumpower, B.L. (1981). Function of the iron–sulfur protein of thecytochrome bc1 segment in electron-transfer and energy-conservingreactions of the mitochondrial respiratory chain. Biochim. Biophys.Acta. 639, 129-155.

43. Gassner, G.T., Ludwig, M.L., Gatti, D.L, Correl, C.C & Ballou, D.P.(1995). Structure and mechanism of the iron–sulfur flavoproteinphthalate dioxygenase reductase. FASEB J. 9, 1411-1418.

44. Holm, L. & Sander, C. (1995). Dali: a network for protein structurecomparison. Trends Biochem. Sci. 20, 478-480.

45. Kleywegt, G.J. & Jones, T.A. (1997). Detecting folding motifs andsimilarities in macromolecular crystallography. Methods Enzymol. 277,525-545.

46. Gassner, G.T., Ballou, D.P., Landrum, G.A. & Whittaker, J.W. (1993).Magnetic circular dichroism studies on the mononuclear ferrousactive-site of phthalate dioxygenase from Pseudomonas cepacia showa change of ligation state on substrate binding. Biochemistry 32,4820-4825.

47. Pavel, E.G., Martins, L.J., Ellis, W.R. Jr. & Solomon, E.I. (1994).Magnetic circular dichroism studies of exogenous ligand andsubstrate binding to the non-heme ferrous active-site in phthalatedioxygenase. Chem. Biol. 1, 173-183.

48. Tsang, H.-T., Batie, C.J., Ballou, D.P. & Penner-Hahn, J.E. (1996).Structural characterization of the mononuclear iron site inPseudomonas cepacia phthalate DB01 dioxygenase using X-rayabsorption spectroscopy. J. Biol. Inorg. Chem. 1, 24-33.

49. Klabunde, T., Sträter, N., Krebs, B. & Witzel, H. (1995). Structuralrelationship between the mammalian Fe(III)-Fe(II) and the Fe(III)-Fe(II)plant purple acid phosphatases. FEBS Lett. 367, 56-60.

50. Roach, P.L., et al., & Baldwin, J.E. (1995). Crystal structure ofisopenicillin N synthase is the first from a new structural family ofenzymes. Nature 375, 700-704.

51. Roach, P.L., et al. & Baldwin, J.E. (1997). Structure of isopenicillin Nsynthase complexed with substrate and the mechanism of penicillinformation. Nature 387, 827-830.


52. Han, S., Eltis, L.D., Timmis, K.N., Muchmore, S.W. & Bolin, J.T. (1995).Crystal structure of the biphenyl-cleaving extradiol dioxygenase from aPCB-degrading pseudomonad. Science 270, 976-980.

53. Senda, T., et al. & Mitsui, Y. (1996). Three-dimensional structures offree form and two substrate complexes of an extradiol ring-cleavagetype dioxygenase, the BphC enzyme from Pseudomonas sp. strainKKS102. J. Mol. Biol. 255, 735-752.

54. Goodwill, K.E., Sabatier, C., Marks, C., Raag, R., Fitzpatrick, P.F. &Stevens, R.C. (1997). Crystal structure of tyrosine hydroxylase at 2.3Å and its implications for inherited neurodegenerative diseases. Nat.Struct. Biol. 4, 578-575.

55. Boyington, J.C., Gaffney, B.J. & Amzel, L.M. (1993). The three-dimensional structure of an arachidonic acid 15-lipoxygenase.Science 260, 1482-1486.

56. Nordlund, P. & Eklund, H. (1993). Structure and function ofEscherichia coli ribonucleotide reductase protein R2. J. Mol. Biol.232, 123-164.

57. Kauppi, B., et al. & Eklund, H. (1996). The three-dimensional structureof mammalian ribonucleotide reductase protein R2 reveals a moreaccessible iron-radical site than Escherichia coli R2. J. Mol. Biol. 262,706-20.

58. Rosenzweig, A.C., Frederick, C.A., Lippard S.J. & Nordlund, P. (1993).Crystal structure of a bacterial non-haem iron hydroxylase thatcatalyses the biological oxidation of methane. Nature 366, 537-543.

59. Poulos, T.L., Finzel, B.C. & Howard, A.J. (1985). High-resolutioncrystal structure of cytochrome P450cam. J. Mol. Biol. 195, 687-700.

60. Poulos, T.L., Finzel, B.C., Gunsalus, I.C., Wagner, G.C. & Kraut, J.(1987). The 2.6-Å crystal structure of Pseudomonas putidacytochrome P-450. J. Biol. Chem. 260, 16122-16130.

61. Otwinowski, Z. (1993). Data collection and processing. InProceedings of the CCP4 Study Weekend. (Sawyer, L., Issacs, N. &Bailey, S., eds), pp. 56-62, Daresbury Laboratories, Warrington, UK.

62. Otwinowski, Z. (1991). Maximum likelihood refinement of heavy atomparameters. In Isomorphous Replacement and Anomalous Scattering.(Wolf, W., Evans, P.R. & Leslie, A.G.W., eds), pp. 80-88, SERCProceedings, Daresbury Laboratories, Warrington, UK.

63. Read, R.J. (1986). Improved Fourier coefficients for maps usingphases from partial structures with errors. Acta Cryst. A 42, 140-149.

64. de La Fortelle, E. & Bricogne, G. (1997). Maximum-likelihood heavy-atom parameter refinement for multiple isomorphous replacement andmultiwavelength anomalous diffraction methods. Methods Enzymol.276, 442-494.

65. Abrahams, J.P. & Leslie, A.G.W. (1996). Methods used in thestructure determination of bovine mitochondrial F1 ATPase. ActaCryst. D 52, 30-42.

66. Kleywegt, G.J. & Jones, T.A. (1996). xdlMAPMAN and xdlDATAMAN.Acta Cryst. D 52, 826-828.

67. Brünger, A.T. (1997). Patterson correlation searches and refinement.In Methods Enzymol. 276, 558-580.

68. Jones,T.A., Bergfors, T., Sedzik, J. & Unge T. (1988). The three-dimensional structure of P2 myelin protein. EMBO J. 7, 1597-1604.

69. Brünger, T.A., Kuriyan, J. & Karplus, M. (1987). Crystallographic Rfactor refinement by molecular dynamics. Science 235, 458-460.

70. Brünger, A.T., Krukowski, A. & Erickson, J. (1990). Crystallographic R-factor refinement by molecular dynamics annealing. Acta Cryst. A 46,585-593.

71. Brünger, A.T. (1992). The free R value: a novel statistical quantity forassessing the accuracy of crystal structures. Nature 355, 472-474.

72. Rice, L.M. & Brünger, A.T. (1994). Torsion angle dynamics: reducedvariable conformational sampling enhances crystallographic structurerefinement. Proteins 19, 277-290.

73. Pannu, N.S. & Read, R.J. (1996). Improved structure refinementthrough maximum likelihood. Acta Cryst. A 52, 659-668.

74. Adams, P.D., Pannu, N.S., Read, R.J. & Brünger, A.T. (1997). Cross-validated maximum likelihood enhances crystallographic simulatedannealing refinement. Proc. Natl. Acad. Sci. USA 94, 5018-5023.

75. Navaza, J. (1994). AMoRe: an automated package for molecularreplacement. Acta Cryst. A 50, 157-163.

76. Murshudov, G.N., Vagin, A.A. & Dodson, E.J. (1997). Refinement ofmacromolecular structures by the maximum-likelihood method. ActaCryst. D 53, 240-255.

77. Lamzin, V. & Wilson, K.S. (1993). Automated refinement of proteinmodels. Acta Cryst. D 49, 129-147.

78. Cowtan, K. (1994). DM. Joint CCP4 and ESF-EACBM Newsletter onProtein Crystallography. 31, 34-38.

79. Laskowski, R.A., MacArthur, M.W., Moss, D.S. & Thornton, J.M.(1993). PROCHECK: a program to check the stereochemistry qualityof protein structures. J. Appl. Cryst. 26, 283-291.

80. Kraulis, P. (1991). MOLSCRIPT: a program to produce both detailedand schematic plots of protein structures. J. Appl. Cryst. 24, 946-950.

81. Bacon, D.J. & Anderson, W.F. (1988). A fast algorithm for renderingspace-filling molecule pictures. J. Mol. Graph. 6, 219-220.

82. Merritt, E.A. & Murphy, M.E.P. (1994). Raster3D Version 2.0 - Aprogram for photorealistic molecular graphics. Acta Cryst. D 50,869-873.


structure of an aromatic-ring-hydroxylating dioxygenase— … · 2017-02-14 · structure of an...

Documents