Structure of the Dissimilatory from the Hyperthermophilic Archaeon Archaeoglobus fulgidus

1 2 Alexander Schiffer ,2, Kristian Parey 1,2, Eberhard Warkentin , 3 3 Kay Diederichs 1, Harald Huber , Karl O. Stetter , Peter M. H. Kroneck 1 * and Ulrich Ermler2 *

1Fachbereich Biologig, Conservation of energy based on the reduction of sulfate is of fundamental Ma thema tisch­ importance for the biog~ochemical cycle. A key of this ancient Naturwissenschaftliche Sektion, anaerobic process is the dissimilatory sulfite reductase (dSir), which Universitiit Konstanz, 0-78457 catalyzes the six-electron reduction of sulfite to under Konstanz, Germany participation of a unique magnetically coupled -[4Fe-4S] center. 2Max-Planck-Institut fill' We determined the crystal structure of the enzyme from the sulfate-reducing archaeon Archaeoglobus fulgidus at 2-A resolution and compared it with that Biophysik, Max-von-Laue-Str. 3, of the phylogenetically related assimilatory Sir (aSir). dSir is organized as a 0 -60438 Frankfurt, Germany heterotetrameric (ar:'>h complex composed of two catalytically independent 3Institut flir Mikrobiologie und ar:'> heterodimers. In contrast, aSir is a monomeric protein built of two fused Archaeenzentrum, Universitiit modules that are structurally related to subunits a and r:'> except for a ferre­ Regensburg, Universitiitsstrafie doxin domain inserted only into the subunits of dSir. The [4Fe-4S] cluster of 31, 0 -93053 Regensburg, this domain is considered as the terminal site of the electron Germany transfer pathway to the siroheme-[4Fe-4S] center in dSir. While aSir binds one siroheme-[4Fe-4S] center, dSir harbors two of them within each ar:'> heterodimer. Surprisingly, only one siroheme-[4Fe-4S] center in each ar:'> heterodimer is catalytically active, whereas access to the second one is blocked by a tryptophan residue. The spatial proximity of ~he functional and structural siroheme-[4Fe-4S] centers suggests that the catalytic activity at one was optimized during evolution at the expense of the enzy­ matic competence of the other. The sulfite binding mode and presumably the mechanism of sulfite reduction appear to be largely conserved between dSir and aSir. In addition, a scenario for the evolution of Sirs is proposed.

Keywords: dissimilatory sulfate reduction; sulfite reductase; siroheme; Edited by M. Guss molecular evolution

Introduction to - II, predominantly between sulfate, elemental sulfur, and hydrogen sulfide. It evolved in an early The biogeochemical sulfur cycle includes reactions stage of prokaryotic life about 3.5 billion years ago in between sulfur compOLmds in oxidation states + VI hot and anoxic environments.' Billions of tons of sulfur compolmds per year are metabolized by var­ ious microbial species that use inorganic sulfur com­ 'Corresponding authors. E-mail addresses: pounds as terminal electron donors or acceptors for [email protected]; the purpose of energy conservation. In these dissi­ ulrich.ernuer®mpibp-frankfurt.mpg.de. milatory processes, specific microorganisms reduce Present address: A. Schiffer, Department of Chemical sulfate to hydrogen sulfide lmder anoxic conditions and Analytical Sciences/Structural Biology, Sanofi-Aventis and others oxidize sulfur compounds to sulfate 2 3 Deutschland GmbH, Industrial Park Hoechst, D-65926 Lmder oxic conditions. . The biochemical pathway 4 7 Frankfurt, Germany. of dissimilatory sulfate reduction - proceeds by Abbreviations used: Sir, sulfite reductase; dSir, activating sulfate (£0'= - 516 mV)8 to adenosine-5'­ dissimilatory sulfite reductase; aSir, assimilatory sulfite phosphosulfate (£°' =-60 mV)8 by ATP sulfurylase reductase; SirHP, aSir hemoprotein of . at the expense of ATP. Subsequently, adenosine-5'- 1064

phosphosulfate is hydrolyzed and reduced to sulfite and I) remain lmknown so far. The "I sublmit carries and AMP by adenosine-5'-phosphosulfate reduc­ a redox-active bond that might function tase, and the generated sulfite is finally reduced to in electron transfer to an external electron donor so 29 hydrogen sulfide by sulfite reductase (Sir) in a six­ far not identified. SubLmit I) is characterized by a 8 electron transfer process ; winged-helix motif, suggesting a role in DNA bind­ ing and regulation.3o While the (af)h core complex HS0 - 6e- 6H+;=HS- 3 H 0 3 + + + 2 of dSir has not been structurally characterized so P'(HS03- /HS- ) = - 116 mV far, a few crystal structures have been reported for aSirs and related , such as the Escherichia In sulfur-oxidizing organisms, this pathway pro­ coli hemoprotein in a tnmcated form31 and sub­ ceeds in the opposite direction starting from hydrogen sequently the Mycobacterium tuberculosis enzyme32 sulfide, elemental sulfur, or thiosuJfate.9 In contrast, and the spinach nitrite reductase.33 A characteristic assimilatory reduction of sulfate is achieved by a wide structural feature of the enzymes of the Sir family is variety of organisms of all three domains of life (but their trilobal architecture. Lobes 1 and 2 (solely fOlmd not of animals) and provides the cell with sulfur in in Sir) are composed of a mixed f,-sheet flanked oxidation state - II, which is vital for biosynthesis by a-helices, and lobe 3 is built by two attached of sulfur-containing amino acids and cofactors.IO, lI ferredoxin-like domains. Tn the central intersecting The latter process deviates in some organisms from points of the three lobes resides a siroheme center the dissimilatory pathway as adenosine-5'-phospho­ bridged to a [4Fe-4S] cluster via a cysteine thiolate, sulfate is first phosphorylated by adenosine-5'-phos­ forming a Lmique electronically coupled multimetal phosulfate kinase and reduced by phosphoadeno­ center. 31 sine-5'-phosphosulfate reductase, which is distantly In this report, we present the crystal structure of the related to adenosine-5'-phosphosulfate reductase. 12 dSir (af"> h core complex isolated from the hyper­ Sirs are considered as the key enzymes in both the thermophile A. julgidus at 2.0-A resolution. Although assimilatory and dissimilatory metabolism of sulfur. this organism has been shown to be of ancient They diverged according to sequence analysis stu­ origin,34 its Sir resulted from a lateral gene transfer dies from a common ancestor into four groups prior event from a deeply branched bacteriallineage,35,36 to the bacteriallarchaeal divergence13,14; the mono­ indicating nevertheless a slow rate of development meric assimilatory Sir (aSir),l5 the low-molecular­ and therefore an archaic natLUe of dSir. The struc­ weight aSir,1 6 the dissimilatory Sir (dSir; working tures of dSir and aSir are compared with regard to in both reductive and oxidative directions),9 and a their overall architecture, arrangement of metal co­ specia I dSir found in a few anaerobic only. 17 factors, electron transfer processes, and evolutionary Notably, assimilatory nitrite reductases, which development. convert nitrite to ammonia, also belong to the Sir family.1 8 dSirs have been isolated from several sulfate­ Results and Discussion reducing microorganisms, such as Des u!fovibrio vul­ garis,1 9 Desu!fovibrio desu!furicans, 20 Pyrobaculum The dSir of A. julgidus was purified to homo­ islandicum, 21 and Arcl1aeoglobus julgidus,22 and have geneity and crystallized lmder the strict exclusion of been characterized with respect to their molecular, dioxygen. The specific activity of 48.2 nmol sulfite spectroscopic, and kinetic properties. The dSirs stu­ min- 1 mg- 1 was in the same range as that re­ died so far are composed of a heterotetrameric (a[-'> h ported for dSir from various sulfate-reducing bacte­ core complex and, depending on the organism, two ria. 19,20,37 The diffraction data were phased with the additional small subunits, "I and 0, resulting in an MAD/STRAS method using intrinsic iron atoms and (a ~h"llll)lI multisubunit complex. 23 The subunits a, a mercury derivative (Table 1). Structural refinement P" "I, and I) have molecular masses of approximately converged to Rand Rfree values of 18.9% and 22.4%, 45, 43, 10, and 11 kDa, respectively, with sublmits a respectively, in the resolution range of 10.0-2.0 A. and ["> being related with respect to their primary se­ The asymmetric unit contained one (ar~h hetero­ quence. The (af">h core complex harbors the coupled tetra mer (Fig. 1). One af", dimer shows a well-shaped siroheme-[4Fe-4S] center as well as [4Fe-4S] clus­ electron density, and the structural analysis is mainly ters. The published stoichiometries range based on its data, whereas the denSity of the second from two to four sirohemes and three to six [4Fe-4S] a~ dimer is much less defined, in some loop regions clusters per (af)h heterotetramer, depending on the even disordered. The folds of the af) heterodimer organism and the purification procedure applied.13 of dSir are related to those of aSir and spinach nit­ Perhaps the most intriguing spectroscopic property rite reductase,31-33 but both sublmits additionally of Sir results from the presence of a set of complex contain an inserted ferredoxin domain (Figs. 1 and EPR signals assigned to high-spin S =5/2 and S =9 /2 2b). The structures of dSir and aSir reveal both a iron-sulfur cen ters. 24 The S = 5/2 resonances were high degree of relationship and interesting differ­ present in both aSirs and dSirs,24-2S whereas the ences that are discussed in the following. For com­ S = 9/2 signals were observed in several dSirs only, parative studies between aSir and dSir, we prefer­ including D. vulgaris24 and A. julgidus?,28 entially use the aSir hemoprotein of E. coli (called Sir activity can be clearly linked to the (af">h core SirHP hereafter) because of its highly resolved X-ray complex, while the ftmctions of the small subunits "I structure.31 1065

Table 1. Data statistics

Hg" Fe" Hg/Fe" Native Hg" peak inflection Fe" peak inflection remote Dntn collection Wavelength (A) 0.9393 1.0 1.009 1.733 1.742 0.95 Resolution range (A) 30.0- 2.0 40.0- 2.7 40.3-3.1 40.0- 2.9 40.0-3.1 40.0- 3.1 (2.16- 2.0) (2.9- 2.7) (3.2- 3.1) (3.1- 3.0) (3.2- 3.1) (3.2- 3.1) Space group P2, P2 1 P2, P2, P2, P2, Completeness (Ufo) 95.3 (79.4) 99.7 (99.9) 77.5 (79.5) 93.9 (99.7) 83.5 (84.5) 97.1 (98.7) Rsym (Ufo) 8.7 (39.1) 9.9 (43.8) 12.6 (60.9) 13.3 (49.7) 18.2 (67.9) 15.1 (57.2) 1/ a(J) 7.4 (2.3) 9.5 (2.9) 10.2 (2.3) 8.9 (2.4) 7.5 (1.7) 8.6 (2 .0) Red w1dancy 2.5 (2.4) 3.1 (3.1) 2.7 (2.9) 3.0 (2.9) 2.5 (2.3) 2.7 (2.5) Wilson B-value (N) 24.8 (25 .3)

Refinelllent statistics No. of residues, sirohemes, 1560,4, [4Fe-4Sj clusters, and 8,342 solvent molecules No. of molecules in a. u. 1 Resolution range (A) 20.0- 2.0 (2.05- 2.0) Reflections (F >Ocr) 108,315 Rworkingl Rfree (%) 18.9, 22.4 (27.1, 32.0) B-value (N ) a·" 11" a2, 1:1 2 28.4,21.4, 69.0,74.0 Bond-length deviation (A) 0.018 Bond-angle deviation (0) 1.86

a The MAD/SIRAS data set was collected with one crystal that was soaked for 1 h in the reservoir solution+0.05 mM thimerosal.

Oligomeric states single chain, by means of a 40-A-Iong linker segment that wraps arolmd the molecule.3 1 Both dSir-a and The characteristic trilobal architecture of Sir is dSir-f>' , and their equivalent moieties in aSir called realized in dSir by a heterodimer composed of tightly aSir-a (or SirHP-a) and aSir-b (or SirHP-b), respec­ associated sublmits a and [-'. (called dSir-a and dSir­ tively, are related by the same pseudo-2-fold sym­ f:',), and not by a monomer as in aSirs and assimilatory metry axis. This is documented by an rms value of nitrite reductases (Figs. 1 and 2). In aSir, equivalent 3.3 A (88% of the residues) between dSir and SirHP, parts of dSir-a and dSir-f:', have been fused into a compared with that of 2.4 A (89%) between dSir-a

Fig. 1. Structure of the dSir of A. fulgidus organized as an (a~'>h heterotetramer with a size of 125 Ax80 Ax 60 A. The heterotetramer is built of two attached arC) units (a subunits are shown in orange and red; [:) subunits, in light blue and dark blue). The C-terminal anns of both subunits are major constituents of their interface. The co fa ctors are shown as a stick model in green. 1066

(a)

(b) dSir-a

16

dSir-f3

245 299

320.~ C

Fig. 2. Architecture of Sir. (a) Stereoview of dSir and complete aSir from M. tllberwlosis superimposed. The trilobal structure is realized in dSir by an al3 heterodimer ana in aSirs (shown in gray) by a monomer. Lobe 1 (in red) is composed of residues a165-a416; .lobe 2 (in blue), l i 120- 1~3 66; and lobe 3 (in orange and light blue), the corresponding subdomains Cl.70-cd60 and 1330- r~ 115. For clarity, both inserted ferredoxin domains were omitted. (b) Schematic representation of the fold of subunits CI. and [1 using the same colors. The inserted ferredoxin domains are shown in yellow. The fold of aSir is marked as light gray margins. and SirI-IP-a and that of 4.0 A (89%) between dSir-~~ is largely preserved in all Sirs, the calculated rms and SirHP-b.38 Interestingly, both dSir-a and dSir-13 differences are clearly visible in the displacements of exhibit higher structural similarities to aSir-a than to the surrounding helices and loop segments. aSir-b. In line with this finding, aSir-a represents the Substantial structural differences between dSir-a major constituent of the siroheme binding and active and dSiH~ and between both and aSir are fmUld at site, which was consequently more conserved as the N- and C-terminal ends (Figs. 1 and 2b). Both aSir-b. Accordingly, dSir-a and dSir-f~ are structu­ ends are involved in intradimer and interdimer inter­ rally more related than SirHP-a and SirHP-b, as face formations (see below). Consequently, the buried reflected in sequence identities of 25% and 16% and surface area between dSir-a and dSir-~>' is larger than in rmsd values of 2.2 and 3.2 A, using 91 % and 83% of that between the aSir-a and aSir-b moieties of aSir, the C'" a toms, respectively. While the l3-sheet sca Hold which obviously compensates for the loss of free 1067

energy for separating the chains. The residual inter­ other (Fig. 1). Hence, a tetrameric architecture of face is mostly formed by equivalent segments in dSir dSir might rather be a relic from its development in and aSir, but the individual residues involved differ hot environments that was then conserved during completely. evolution. Oligomerization is a frequently used stra­ On the basis of gel filtration experiments, dSir of tegy for hyper thermophilic proteins to increase their A. fulgidus was previously described as an (al3h stability by reducing the surface area-to-volume 22 heterotetrameric enzyme complex (Fig. 1) that was ratio. 4·1 However, a role of the heterotetramer in the now confirmed by the X-ray structure. The buried transfer of six electrons to sulfite cannot be com­ surface area between two al3 heterodimers of pletely excluded as the molecular basis of this pro­ 4567 A2 (9.1 % of the entire surface)39 corresponds cess has not been established yet. to values fOlmd for other oligomeric proteins.40 The interface between the two al3 lmits is largely formed Siroheme-[4Fe-4S] center by the extended C-terminal segments of both subunits that are either wrapped around the COlmter The dissimilatory enzyme of A. fulgidus defini­ heterodimer or completely encapsulated by it. The tively harbors the same multimetal center composed C-terminal regions are not generally preserved in of a siroheme ring coupled to a [4Fe-4S] cluster via a dSir such that the heterotetrameric oligomeric state cysteine thiolate bridge, as observed in aSir. A novel cannot be predicted for all family members from feature of dSir compared with aSir results from the their primary structure. Moreover, the heterotetra­ presence of a second siroheme-[4Fe-4S] center in meric organization of dSir does not appear to be of each al3 heterodimer (Fig. 2a). The two centers are functional importance for the reduction of sulfite to termed 'ftmctional' and 'structural' in the following hydrogen sulfide. The reaction proceeds most likely as the latter is catalytically not productive (see in an independent manner at the ~ctive sites of the below). The presence of four siroheme-[4Fe-4S] two al?> units, which are about 50 A apart from each centers plus fOLlr additional [4Fe-4S] clusters in dSir

Fig. 3. Cofactors of the dSir of A. fulgidus. (a) Arrangement. of the siroheme-[4FHS] centers and the peripheral [4FHS] clusters in the a~ heterodimer. The functional simheme-[4FHS] center (at the right side) is the site of sulfite reduction and most likely receiv es the required electrons from the adjacent peripheral [4FHS] cluster. According to the vicinity of the sirohemes, the structural siroheme-[4Fe-4S] center could also serve as an electron donor for the functional center. The well-formed electron density at 2 A is shown at a contour level of 2u. (b) Surrotmding of the fourth iron of the [4FHS] cluster of the ftmctional siroheme-Fe/S cluster. Unexpectedly, the fourth iron is not ligated by a cysteine as fotmd for dSir-a and aSir but by a tlu-eorUne (Tlu·r~ 1 34 ). Alternatively, CysI:H77 approaches the iron to about 3 A, which is, however, too far away for coordination. 1068

of A. fulgidus, and presumably in all heterotetrameric structural differences exist. First, the macrocyclic dSirs, comprising altogether 36 Fe and 32 acid-labile siroheme ring system in dSir is less ruffled tha n that sulfur atoms, is compatible with conclusions from in SirHP, which concomitantly increases the dis tance sequence comparison studies but deviates from most between the siroheme and the adjacent [4Fe-4S] values determined by chemical analysis. 13 cluster. The closest contact between the CHB a tom of Interestingly, the two sirohemes are in close neigh­ the macrocyclic ring and th~ Sl sulfur of the [4F e-4S] borhood and are arranged in a manner that rings B cluster increases from 3.6 A in SirHP to 4.3 A in the and C project toward each other (Fig. 3). With a functional center of dSir. The distortions from minimal distance of 9.6 A between the acetate planarity of the functional and the structural carboxylates of ring C, they mutually participate in sirohemes of dSir are similar; however, the minimal forming their binding pockets and thereby rigidify distance to the [4Fe-4S] cluster increases to 4.6 A in and stabilize each other. Both sirohemes are multiply the latter. Second, the conformations of the siroheme linked via the polypeptide scaffold, which would acetate and propionate substituents are substan­ allow an electron transfer and perhaps weak elec­ tially altered as a consequence of the number, na­ tronic coupling between both. The most direct con­ ture, and conformation of the interacting positively nection is formed between the ring C carboxylate charged residues (Fig. 4). Only a few pOSitively group of both sirohemes media ted by Arga229 and a charged residues are conserved between the flmc­ solvent molecule via hydrogen bonds. tional and structural sirohemes .of dSir but none is In contrast, the site that would correspond to the conserved between dSir and SirHP. However, sev­ missing structural siroheme-[4Fe-4S] center in aSir eral basic side chains are topologically preserved has become substantially rearranged. Most impor­ both between the functional and structural siro­ tantly, the linker segment between aSir-a and aSir-b of dSir and between those of SirHP and dSir occupies the position of the siroheme (Fig. 2). Other (Fig. 4). It is conceivable that specific conformations structural features, such as the modified central of the basic residues will induce the pronOl.llced r~ -sheet of lobe 1 (i.e., devoid of the first strand) and saddle-shaped distortion of the siroheme ring in the shortened loops 35:41, 114:130, 166:170, and SirHP. A third difference is related to the binding 315:319 of SirHP-b compared with dSir-fJ" also mode of the [4Fe-4S] cluster bridged to the ftmctional prevent binding of the siroheme-[4Fe-4S] center siroheme moiety despite the high degree of conser­ (Fig. 2a). The of the coupled siroheme­ vation. Only three of the four iron atoms of dSir are [4Fe-4S] cluster in aSir is abolished by restructuring ligated to the invariant thiolate groups of Cys[3140, the polypeptide segment carrying the [4Fe-4S] clus­ Cys[3182, and Cysf3178 with distances between 2.2 ter and by replacing all four ligating cysteines of and 2.4 A The fourth iron atom forms a polar contact aSir-b, dSir-a, and dSir-[3. to the hydroxyl group of Thr[3134 (replaces the liga­ A comparison of the conformation and the ting cysteine in aSir) and to the thiol group of binding mode of the functional and structural Cys[3177 (Fig. 3b). Cys[3177 is replaced by a glycine i!l siroheme-[4Fe-4S] centers within dSir, and between dSir-a and in aSir. However, the distances of ca 3 A dSir and aSir, revealed that the positions of the between the iron and the side chain of coupled centers as well as the enveloping polypep­ Thrp,134 and the sulfur of Cysp,177, respectively, tide scaffold and its electrostatic properties are argue against a direct ligation. This is astonishing as mostly preserved despite any significant overall small conformational changes would place the thiol sequence identity. Nevertheless, a few significant sulfur in a ligating position.

I Gln121 1 ~ (a) I Ar~a80 I IHI.p139 I (b) NO I NE~ . ~E 2 [va@] dSir-u NH~. . aSir o iTn,4le fOG I· .... 0 ....-d H~" ./NHI ffiill!J & .::....NH2

~ ..•. NE2 ·IGlnl96 I 0 : .. ··NH 1Asn1161 ~NE2 O······OG 1ITii:illIl o NE~ ~DDI "" ' O o O· ······· .. NH I ~

(\.· •• NH IlliiillJ ~NH2 ·· · ·· . · 0 )Z.lli!lliJ 0. o o ... .. f "'NH'j" 'NH NH IA'Ra360 IIAI,a3ls l ~

Fig. 4. Interactions between the polypeptide and the siroheme- Fe/S centers in dSir-a (a) and SirHP (b). The sirohemes are embedded into a positively cl)arged pocket compensating for the negative charge of eight carboxylate groups. However, the conformation of the propionate and acetate groups and that of the contacting basic residues (highlighted in red) are not conserved between dSir and aSir. The functional siroheme of dSir interacts directly with 11 basic residues; the siroheme of SirHP, with 8. 1069

Catalytic reaction the two coupled siroheme-[4Fe-4S] centers, optimi­ zation at one active site automatically requires con­ Although each of the ap, heterodimers harbors two formational changes at the other. The functional complete sets of siroheme-[4Fe-4S] centers to cata­ siroheme-[4Fe-4S] center and the sulfite binding site lyze the six-electron reduction of sulfite to hydrogen might be adjusted by rigidifying the binding site of sulfide, a sulfite binding site is only formed in front the structural siroheme-[4Fe-4S] center, which is of the si-side of one (the functional) siroheme center supported by its lower temperature factor (S A2) (Fig. S). The substrate binding site is located at the compared with that of the ftmctional center. From end of an ~ lS-A-long fU1U1el formed by the inserted an economical point of view, the synthesis of a ferredoxin domain as well as lobes 1 and 3 of dSir-a siroheme-[4Fe-4S] center to stabilize the active site (Fig. Sa). Based on available structural informa­ of the enzyme appears fairly expensive. Therefore, tion,3 I-33 the siroheme center and the sulfite binding dSir has to be regarded as an intermediate sub­ site appear to be more deeply buried in dSir than in optimal solution of the evolutionary process that aSir. This finding might change when the natural was overcome during the evolution of aSirs. Usage electron donors bind to aSirs. Nevertheless, the of cofactors for structural reasons is rare in enzymes substrate binding site is more shielded in aSir mainly but not lmprecedented.42 Note also that a functional due to the covalently linked Tyr69-Cys161 pair and role of the structural siroheme-[4Fe-4S] center in the Arg64 in M. tuberculosis aSir and due to Lys91, Phe96, electron transfer process cannot be ruled out. and Met17S in spinach nitrite reductase. Note that The assignment of the substrate binding site at the these segments are partly tnmcated in SirHP. In functional siroheme is supported by the presence of contrast, the potential substrate binding site in front a bulky electron density at the position of the distal of the structural siroheme center is covered by a ligand although the nature of the ligand remains f3-bulge that is formed by prolonging the loop pre­ lmclear. The electron density shape certainly does ceding strand ~-',126:f3134 (Fig. Sb). The side chain of not match to a sulfite or a phosphate ion. COrnpared Trpf3119 is directed from the f>-bulge toward the with the structural siroheme center where the iron si-side of the siroheme and thus completely prevents atom is placed toward the re-face of the ring plane, sulfite binding. The tryptophan residue at position the iron atom of the functional siroheme moiety sits 119 is not strictly preserved but conservatively in the plane center and coordinates with the distal exchanged frequently by a phenylalanine. Addition­ ligand. Accordingly, the distance values between the ally, catalytically important amino acid residues bridging thiolate sulfur of Cysa223 and that of (present in front of the functional siroheme; see Cys~-', 182 and the iron atom of the structural and below) are essentially substituted and the negatively functional centers are 2.5 and 2.8 A, respectively. The charged acetate and propionate groups of rings A latter value corresponds to that fotmd in SirHP. and D point toward the si-side of the structural The yet unidentified distal ligand at the ftmctional siroheme and thus repel incoming anions (Fig. Sb). siroheme is embedded into a positively charged The prevention of catalysis at the structural pocket (Fig. Sa) mainly formed by side chains of siroheme-[4Fe-4S] center might be driven by the Arga98, Arga170, Lysa211, Lysa213, Thra133, and benefit of the enzyme to contain one optimized, four firmly bound solvent molecules arotmd them. highly productive active site as opposed to two less The four positively charged residues are strictly efficient active sites. Due to the close proximity of conserved in dSirs and aSirs, and their long side

(a)

Fig. 5. Potential sulfite binding site. (a) At the functional siroheme-[4Fe--4S] center, the four key residues Argcx98, Argcx170, Lyscx211, and Lyscx213 conserved in aSir and dSir constitute the pocket for binding sulfite and contribute significantly to catalysis. (b) At the structural siroheme- [4Fe--4S] center, the sulfite binding is blocked by a prolonged loop (in green) preceding strand rU26:rH34, from which the bulky side chain of Trpr~ 119 occupies the posi tion of the substra te in front of the siroheme iron. For comparison, the equiva lent loop in front of the functional siroheme-[4Fe--4S] center is also marked in green. 1070

chains adopt highly similar conformations and 12.1 A, respectively (Fig. 3), which are in the right positions emphasizing their catalytic importance. range to shuttle electrons at physiological rates.46 The Thra133, which is only conserved in dSirs, becomes most direct electron transfer pathway to the func­ replaced in aSirs by an arginine that points away tional siroheme- [4Fe- 4S] center extends from from the sulfite. Due to the nearly identical designs Cysr-', 178 to Cys0244 (ligands of the functional center of the sulfite binding pocket in aSir and dSir, we and the peripheral [4Fe-4S] cluster of dSir-l'» via a assume similar mechanisms for the destabilization tightly bound solvent molecule that interacts with of the S-O bond of sulfite and its protonation, which their thiol and main chain carbonyl groups, respec­ is described in detail for aSir. 31 tively. In comparison, the shortest link to the struc­ Our structural data of dSir from A. fulgidus clearly tural siroheme-[4Fe-4S] center is provided via document that the sequence IC- Xs-C-n-C-Xr C" Meta289, whose side chain interacts with the thiol represents an insufficient recognition motif for Sirs group of Cysa219 of the structural center and whose as it only recognizes the siroheme-[4Fe-4S] center. peptide amine nitrogen contacts a sulfur of the peri­ Actually, it solely identifies the [4Fe-4S] cluster, and pheral [4Fe-4S] cluster. Thus, a rapid electron trans­ not the sulfite binding site 'and the catalytically fer is geometrically feasible between both the flmc­ important amino acid residues. A conserved stretch tiona I and structural siroheme-[4Fe-4S] centers involved in sulfite binding helps define a second and their adjacent peripheral [4Fe-4S] cluster. The recognition sequence, "p-Y-K-a-K-s-K" (a, alipha­ latter [4Fe-4S] cluster might also channel electrons to tic; s, small), for dSirs containing the two invariant sulfite via the structural and functional siroheme­ lysines mentioned above. An analogous fingerprint [4Fe-4S] centers (Fig. 3). "P- R- K-a-K-a-s" can be defined for aSirs, whereby According to our current knowledge about the the "K-a-K" motif is shared by dSirs and aSirs. energy metabolism of A. fulgidus and other sulfate Notably, the recognition sequence contains at posi­ reducers, the electrons originate from a quinol pool tion 2 in dSir a tyrosine and in aSir an arginine, both in the cell membrane that is generated by oxidizing pointing to the ring D acetate. This might contribute organic substrates, such as lactate.47 The thereby to the different ring ruffling fmmd in SirHP and dSir. reduced coenzyme F420 is subsequently oxidized, The equivalent fingerprint "P-R-K-a-N-a-s" for and the quinone is reduced by a proton-translocat­ assimilatory nitrite reductase resembles that for aSir ing F420H2 menaquinone .48 Finally, except for the exchange of lysine by asparagine in the generated quinol is oxidized by the membrane position 5, as reported previously.31 protein complex Hme

Molecular evolution none of them assembles in a manner that a siro­ center and a [4Fe-4S] cluster juxtapose. Sirs have evolved over a period of about 3.5 billion Further catalytic optimization of Sir is accom­ years with different rates and under different envi­ plished by a gene duplication event, one of the major ronmental pressure and physiological constraints. sources of evolutionary development.53 This be­ Comparative studies reveal that the basic structural comes evident from the strong structural relation­ framework, including the unique electronically ship between dSir-a and dSir-f) and between aSir-a coupled siroheme-[4Fe-4S] center and most likely and aSir-b. The branching point between dSir and the mechanism of the six-electron reduction of sul­ aSir cannot be definitely derived. The high primary fite, has been remarkably conserved. On the other and tertiary structure relationship between the peri­ hand, the oligomeric state, the degree of internal pheral ferredoxin domains of dSir-a and dSir-r:'> (33%, structural relationships, the number of prosthetic nns = 1.9 A) suggests insertion prior to gene duplica­ groups, and the electron delivery system differ bet­ tion of dSir and consequently separated develop­ ween dSir and aSir. These observations prompted us ment between aSir and dSir before this event. The to propose a scenario of protein evolution as de­ absence of any overall sequence relationship bet­ picted in Fig. 6. ween aSir-a and dSir-a despite catalyzing the same It starts with the fusion of a larger air:'> domain reaction and originating from a common ancestor carrying a [4Fe-4S] cluster (lobe 1) and a smaller also argues for an earlier separation as that of dSir-a air:'> domain (the first moiety of lobe 3). This new and dSir-p, . The missing sequence identity between monomer has to be furnished (accidentally) with aSir-a and aSir-b in contrast to dSir-a and dSir-[?> surface properties that will allow weak interactions indicates that the molecular evolution of aSir has with a siroheme molecule and with a second mono­ been subjected either to a higher rate of amino acid mer in such a manner that the [4Fe-4S] cluster of one exchange or to an earlier gene duplication compared monomer faces the siroheme moiety of the other with dSir, with the latter possibility also arguing for (Fig. 2a). The juxtaposition of cofactors or metals branching prior to this event. protruding from different folding lmits is a common After gene duplication, the development of dSir theme in the evolution of cooperating cofactors, as focuses on the optimization of the active site at the observed, for example, in rubredoxin:NO/02 functional siroheme-[4Fe-4S] center. This is accom­ oxidoreductase5 1 and in methanol:cobalamrn me­ panied by modifications in the regions of the struc­ thyltransferase.52 In Sirs, the thiolate bridge between tural siroheme-[4Fe-4S] center that concomitantly the [4Fe-4S] cluster and the siroheme iron will eliminate its own enzymatic competence. In parallel, significantly contribute to stabilize the dimeric form. tetramerization of dSir occurs as the interface form­ A thereby generated weak activity for reducing ation is based on the differentiation of dSir-a and anions, such as nitrite and sulfite, will be enhanced dSir-r:'>, particularly of their C-terminal ends. After­ by spontaneous mutation events that increase the ward, divergence of the bacterial and archaealline­ monomer- monomer interactions and the affinity ages takes place as derived from sequence compa­ between the protein matrix and both the siroheme­ rison studies. The development of aSir might pass [4Fe-4S] centers and inorganic ions. Up to this through a stage described for dSir but goes sub­ point, aSir and dSir are presumably not distinguish­ sequently beyond that by fusing the two sublmits able. According to our currently limited knowledge and by removing the structural siroheme-[4Fe-4S] about the oligomeric composition of the low center. Both processes are coupled since the linker molecular-mass-aSir 16 and the unusual dSir, I7 segment between aSir-a and aSir-b complements these two enzymes do not fit into this concept as the abolished structural siroheme-[4Fe-4S] center.

4

half of --- lobe 3

lobe 1 ---4

Fig. 6. Evolution of Sir. The proposed route is based on th e assumption that branching between the dSirs and aSirs occurs prior to gene duplication. The following steps are illustrated: (1) dimerization; (2) optimization by spontaneous mutagenesis; (3) gene duplication; and (4) gene fu sion and tetrameri za tion. 1072

64 Together with gene duplication, gene fusion is the was brought to convergence using REFMAC5. •65 The 54 most important force driving protein evolution. ,55 problem was that only one al?> heterodimer has a clearly defined electron density, which hampers the applications of solvent flattening and molecular averaging phase improvement methods. Refined thermal displacement Materials and Methods B-factors are lowest for residues 56-59 and 84-90 of chain B within the rigid af'> dimer rather than ne ar the Purification center of gravity of the whole tetramer. There is a close correlation of the size of the B-values with the increasing distance from this 'cold' region. In addition, the electron The cultivation of A. fulgidus (DSM 4304T) was carried density of distant atoms shows a characteristic anisotropy out as previously described.34 Frozen cells were trans­ (extended perpendicular to the line to the cold region). ferred to an anaerobic chamber (95% N , 5% H ; Coy) and 2 2 This behavior is consistent with a rigid body Iibration of suspended in 1- 2 volumes of 20 mM potassium phosphate the whole tetramer that can be accounted for by the TLS buffer, pH 7.0, containing a few crystals of deoxyribonu­ (translation-Iibration-screw) model. Indeed, the refine­ clease I and 5 mM MgCl2 ·6H20. Cells were disrupted in a ment of the additional free 20 parameters (compared with French press, and the lysate was centrifuged at 100,000g. the ca 43,000 of the protein model) decreases the Rfree The soluble fraction was applied to a Q Sepharose Fast value by 3%. A view of the packing of the molecules in Flow column (1.6 cm x 10.0 cm; Amersham Pharmacia the crystal shows the 'cold' ends of the tetramers in con­ Biotech) equilibrated with 20 mM potassium phosphate tact to one another, related by the 21 axes. The same buffer, pH 7.0. dSir eluted in a linear gradient (0-1.0 M KCI) holds for the 'hot' ends. The Rfrcc value further decreases at about 0.54 M KCl. Fractions containing dSir were when applying the non-crystallographic symmetry as an combined and desalted by ultrafiltration (cutoff= 30 kDa; appropriate restraint. The refinement statistics are given Amicon) with subsequent dilution with 20 mM potassium in Table 1. The quality of the model was checked with phosphate buffer, pH 7.0, and 5% (v Iv) glycerol. The PROCHECK.66 There are four non-glycine residues in desalted protein was loaded onto a Resource Q15 column Ramachandran disallowed regions. Sequence discrepan­ (1.0 cm x 13 cm; Amersham Pharmacia Biotech) and eluted cies between the A. fulgidus genome67 and the previously by a linear gradient (0-1 M KCI) at about 0.27 M KCI. The determined gene sequence of dSir37 could be mostly cla­ combined fractions were concentrated by ultracentrifuga­ rified in favor of the latter. Figures 1, 2a, 3, 4 and 5 were tion and loaded onto a SuperdexTM 200 HiLoad™ 26/60 generated with PyMOL gel filtration column (2.6 cm x 60 cm; Amersham Pharma­ t. cia Biotech) equilibrated with 50 mM potassium phosphate buffer, pH 7.0, 150 mM NaCl, and 5% (v Iv) glycerol. Protein Data Bank accession code Protein was concentrated to 20 mg/ mI and stored at 100 K in 100 mM Tris-HCI, pH 7.0. The coordinates of dSir from A. fulgidus are deposited in the RCSB Protein Data Bankt with accession number 3c7b. Crystallization and data collection

Initial crystals were obtained with the hanging-drop method using the Hampton sparse matrix screening setup. Optimization led to a drop content of 11.11 of protein solu­ Acknowledgements tion and 1 f,l1 of reservoir solution composed of 20% PEG (polyethylene glycol) 4000, 0.1 M sodium citrate, 0.2 M This work was supported by ~he Max-Planck­ NaCl, and 5% 2-propanoJ. The crystals grew in space Gesellschaft and the Deutsche Forschungsgemein­ group P2 1 with unit cell parameters of 94.8 A, 69.4 A, schaft (ER 222/2-1,2; PK 451/32-3). We thank the 148.3 A, and 106.9°, with two a and two ~~ subunits in the 1 staff of the BW6 beamline (in particular, Gleb asymmetric unit (VM =2.6 A3 Da- , solvent content=53%). Bourenkov) at the DESY (Hamburg) as well as of They diffracted to around 2.0-A resolution. For freezing, crystals were incubated for 2- 5 min in a buffer containing the ID14.4 beamline in Grenoble for excellent tech­ 100 mM sodium citrate, pH 6.5, 20% PEG 4000, 0.1 M nical assistance and Hartmut Michel for continuous NaCl, 5% 2-propanol, and 15% glycerol. support. Native data were collected at beamline ID14.4 of the European Synchrotron Radiation Facility in Grenoble, whereas MAD data were collected at the BW6 beamline of References DESY in Hamburg. Data processing was performed with 57 the HKL56 and XDS program suites. Statistics of data 1. Canfield, D. E., Rosing, M. T. & Bjerrum, C. (2006). sets are summarized in Table 1. Early anaerobic metabolisms. Phi/os. Tran s. R. Soc. London, Ser. B, 361,1819- 1836. 2. Peck, H. D., Jr & Le Gall, J. (1994). Inorganic microbial Phase determination and refinement . Methods En zymol. 243, 3-682. 3. Postgate, J. R. (1984) . The Sulphate Reducing Bacteria, 2nd The [4Fe-4S1 clusters were found using SHELXDs8 and edit., Cambridge University Press, Cambridge, UK. further refined using SHARP59 The phases were calcu­ 4. Odom, J. M. & Peck, H. D. (1981). Localization of lated with SHARP and improved by solvent flatteningOO dehydrogenases, reductases, and electron transfer assuming a solvent content of 50%. Twofold molecular averaging within DM61 resulted in a poor electron density map that could, however, be finally interpreted using several iterative cycles of refinement and manual model t http://www.pymoJ.org building with the programs in CNSG2 and 0.63 Refinement t http://www.rcsb.org 1073

components in the sulfa te-reducing bacterium Des ltl­ Desulfovibrio vulgaris (Hildenborough). Eur. J. Biochelll. fovibrio gigas. J. Bacteriol. 147, 161-169. 233,873- 879. 5. Hansen, T A. (1994). Metabolism of sulfate-reducing 21. Molitor, M., Dahl, c., Molitor, 1., Schafer, U., Speich, prokaryotes. Antonie van Leeuwenhoek, 66, 165- 185. N., Huber, R & Deutzmann, R (1998) . A dissimilatory 6. Fri tz, G., Einsle, 0 ., Rudolf, M, Schiffer, A. & Kroneck, sirohaem-sulfite-reductase-type protein from the P. M. H. (2005). Key bacterial multi-centered metal hyperthermophilic archaeon Pyrobacu lu1I1 islandic!l1l1. enzymes involved in nitrate and sulfate respiration. Microbiology, 144, 529-541. J. Mol. Microbiol. Biotechnol. 10, 223-233. 22. Dahl, c., Kredich, N. M, Deutzmann, R & Triiper, 7. Fritz, G., Schiffer, A., Behrens, A., Bikhert, T, Ermler, H. G. (1993). Dissimilatory sulfite reductase from U. & Kroneck, P. M. H . (2007). Living on sulfate: Archaeoglobus fulgidus: physico-chemical properties of three-dimensional structure and spectroscopy of the enzyme and cloning, sequencing and analysis of the adenosine 5'-phosphosulfate reductase and dissim­ reductase genes. J. Gen. Microbiol. 139, 1817-1828. ilatory sulfite reductase. In MicrobiaL Sulfur Metabolism 23. Steuber, J. & Kroneck, P. M. H. (1998). Desulfoviridin, (Dahl, C. & Friedrich, c., eds), pp. 13-23, Springer, the dissimilatory sulfite reductase from DesuLJovibrio Berlin, Germany. desulfuricans (Essex)- new structural and functional 8. Thauer, R K, Jungermann, K & Decker, K. (1977). aspects of the membranous enzyme. Inorg. C/-zim. Acta, Energy conservation in chemotrophic anaerobic bac­ 276,52-57. teria. Bacterial. Rev. 41, 100-180. 24. Pierik, A. J. & Hagen, W R (1991). S =9/2 EPR signals 9. Hipp, W M ,. Pott, A S., Thurn-Schmitz, N., Faath, 1., are evidence against coupling between the s iroheme Dahl, C. & Triiper, H. G. (1997) . Towards the phylo­ and FelS cluster prosthetic groups in DesuLJovibrio geny of APS reductases and sirohaem sulfite reductases vulgaris (Hildenborough) dissimilatory sulfite reduc­ in sulfate-reducing and sulfur-oxidizing prokaryotes. tase. Eur. J. Biochem. 195, 505-516. Microbiology, 143, 2891- 2902. 25. Janick, P. & Siegel, L. M. (1982). Electron paramagnetic 10. Schwenn, J. D. (1997). Assimilatory reduction of inor­ resonance and optical spectroscopic evidence for ganic sulphate. In Sulfur Nutrition and Assimilation in interaction between siroheme and Fe4S4 prosthetic Higher (Cram, J. W, ed), pp. 3-23, Academic groups in Escherichia coli sulfite reductase hemo­ Publishing, The Hague, The Netherlands. protein subunit. Biochemistry, 21, 3538-3574. 11. Wirtz, M. & Droux, M. (2005). Synthesis of the sulfur 26. Hall, M. H., Prince, R H. & Cammack, R (1979). EPR amino acids: cysteine and methionine. Photosynth. Res. spectroscopy of the iron-sulphur cluster and sirohaem 86, 345-362. in the dissimilatory sulphite reductase from Desulfovi­ 12. Chartron, J., Carroll, K S., Shiau, c., Gao, H., Leary, brio gigas. Biochelll. Biophys. Res . COlllm.UI1. 581, 27-33. J. A., Bertozzi, C. R & Stout, C. D. (2006). Substrate 27. Wolfe, B. M., Lui, S. M. & Cowan, J. A. (1994). Desul­ recognition, protein dynamics, and iron-sulfur cluster foviridin: a multimeric-dissimilatory sulfite reductase in Pseudomonas aeruginosa adenosine 5'-phosphosul­ from DesuLJovibrio vulgaris (Hildenborough). Purifica­ fate reductase. J. Mol. Bioi. 364, 152-169. tion, characterization, kinetics and EPR studies. Eur. J. 13. Crane, B. R & Getzoff, E. D. (1996). The relationship Biochelll. 223, 79-89. between structure and function for the sulfite reduc­ 28. Schiffer, A. (2005). Structural and functional invest­ tases. Curl'. Opin. Stn/ct. BioI. 6, 744-756. igations on multi-site metalloenzymes of the biolo­ 14. Dhillon, A, Goswami, S., Riley, M., Teske, A. & Sogin, gical sulfur cycle. PhD thesis, UniversiUit Konstanz, M. (2005). Domain evolution and functional diversi­ Germany. fication of sulfite reductases. Astrobiology, 5, 18-29. 29. Mander, G. J. , Weiss, M. S., Hedderich, R, Kahnt, J., 15. Siegel, L. M , Murphy, M. J. & Kamin, H. (1973). Re­ Ermler, U. & Warkentin, E. (2005) . X-ray structure of duced nicotinam.ide adenine dinucleotide phosphate­ the gamma-subunit of the dissimilatory sulfite reduc­ sulfite reductase of enterobacteria: 1. The Escherichia coli tase: redox-active and structural disulfide bonds. hemoflavoprotein: molecular parameters and pros­ FEBS Lett. 579,4600-4604. thetic groups. J. Bioi. Chelll. 248,251- 264. 30. M.izuno, N., Voordouw, G., Miki, K , Sarai, A & 16. Moura, 1., LeGalJ, J., Lino, A. R , Peck, H. D., Fauque, G., Higuchi, Y. (2003). Crystal structure of dissirnilatory Xavier, A. V. et al. (1988). Characterization of tWo dissi­ sulfite reductase D (DsrD) protein- possible interac­ milatory sulfite reductases (desulforubidin and desul­ tion with B- and Z-DNA by its winged-helix motif. foviridin) from sulfate-reducing bacteria. Mossbauer Structure, 11, 1133- 1140. and EPR studies. J. Am. Chell/. Soc. 110, 1075- 1082. 31. Crane, B. R, Siegel, L. M & Getzoff, E. D. (1995). Sulfite 17. Carter, M. P., Tingey, J. , Bobik, T A & Roth, J. R reductase structure at 1.6 A: evolution and ca talysis for (2001). The alternate electron acceptor tetrathionate reduction of inorganic anions. Science, 270, 59-67. supports B12-dependent anaerobic growth of Salmo­ 32. Schnell , R, Sandalova, T., Hellman, U., Lindqvist, Y. & nella enterica serovar Typhimurium on ethanolamine Schneider, G. (2005). Siroheme- and [Fe4S4J-depen­ or 1,2-propanediol. J. Bacteriol. 183, 2463-2475. dent NirA from Mycobacteriul/'I tuberculosis is a sulfite 18. Knaff, D. B. (1996). Ferredoxin and ferredoxin-depen­ reductase with a covalent Cys-Tyr bond in the active dent enzymes. In O:tygenic Photosynthesis: The Light site. J. BioI. Chem . 280, 27319- 27328. Reactions (Ort, D. R & Yocum, C. F., eds), pp. 333-361, 33. Swamy, U., Wang, M., Tripathy, ]. M., Kim, S.-K., Kluwer Publishers, Dordrecht, The Netherlands. Hisasawa, M., Knaff, D. B. & All en, J. P. (2005). Struc­ 19. Marritt, S. J. & Hagen, W R (1996). Dissimilatory ture of spinach nitrite reductase: implications for sulfite reductase revisited: the desulfoviridin molecule multi-electron reactions by the iron-sulfur:siroheme does contain 20 iron atoms, ex tensively demeta lJ ated cofactor. Biochemistry, 44, 16054-16063. sirohaem, and an S =912 iron- sulfur cluster. EliI'. J. 34. Stetter, K. 0 ., Lauerer, G., Thomm, M. & Neuner, A. Biochem. 328, 724-727. (1987). Isolation of extreme thermophilic sulfate re­ 20. Steuber, J., Arendsen, A. F., Hagen, W R & Kroneck, ducers: evidence for a novel branch of archaebacteria. P. M. H. (1995). Molecul ar properties of the dissim­ Science, 236, 822- 824. ilatory sulfite reductase from Desulfovibrio desulfur­ 35. Klein, M. M., Friedrich, M., Roger, A. J., Hugenholtz, icans (Essex) and comparison with the enzyme from P., Fishbain, S., Abicht, H. et al. (2001). Multiple lateral 1074

gene transfers of dissimilatory sulfite reductase genes 52. Hagemeier, C. H., Krtter, M., Thauer, R. K, Warkentin, between major lineages of sulfate-reducing prokar­ E. & Ermler, U. (2006). Insight into the mechanism of yotes. J. Bacteriol. 183, 6028-6035. biological methanol activation based on the crystal 36. Stahl, D. A, Fishbain, S., Klein, M., Baker, B. J. & sh'ucture of the methanol:cobalamin methyltransfer­ Wagner, M. (2002). Origins and diversification of sulfate­ ase complex MtaBC. Proc. Nat! Acad. Sci. USA, 103, respiring microorganisms. An.ton.ie van Leeuwenhoek, 81, 18917- 18922. 189-195. 53. Lynch, M. & Conery, J. S. (2000). The evolutionary fate 37. Dahl, C. & Triiper, H. G. (2001) . Sulfite reductase and and consequences of duplicate genes. Science, 290, APS reductase from Archaeoglobus flilgidus. Methods 1151- 1155. EnzY1l10l. 331, 427-441. 54. Snel, B., Bork, P. & Huynen, M. (2000). Genome evo­ 38. Holm, L. & Sander, C. (1993). Protein structure com­ lution. Gene fusion versus gene fission. Trends Genet. parison by alignment of distance matrices. J. Mol. Bioi. 16,9- 11. 233, 123-138. 55. Vogel, c., Bashton, M., Kerrison, N. D., Chothia, C. & 39. Hubbard, S. J., Campbell, S. F. & Thornton, J. M. Teichmann, S. A. (2004) . Structure, function and evo­ (1991). Molecular recognition. Conformational analy­ lution of multidomain proteins. Curro Opin. Stntct. sis of limited proteolytic sites and serine proteinase BioI. 14, 208-216. protein inhibitors. J. Mol. Bioi. 220, 507-530. 56. Otwinowski, Z. & Minor, W. (1996). Processing of 40. Janin, J. (1997). Specific versus non-specific contacts in X-ray diffraction data co llected in oscillation mode. protein crystals. Nat. Struct. BioI. 4, 973-974. Methods E11Zy1l'lOI. 276, 307-326. 41. Jaenicke, R & Bohm, G. (1998). The stability of pro­ 57. Kabsch, W. (1993). Automatic processing of rotation teins in extreme environments. Curl'. Opin. Stmct. BioI. diffraction data from crystals of initially unknown 8,738- 748. symmetry and cell constants. ]. Appl. Crystallagr. 26, 42. Bornemann, S. (2002). Flavoenzymes that cata lyse 795-800. reactions with no net redox change. Nat. Prod. Rep. 19, 58. Schneider, T. R & Sheldrick, G. M. (2002). Substruc­ 761-772. ture solution with SHELXD. Acta Crystallogr., Sect. 0: 43. Siegel, L. M., Davis, P. S. & Kamin, H. (1974). Reduced Bioi. Crystal/ogr. 58, 1772-1779. nicotinamide adenine dinucleotide phosphate-sulfite 59. De la Fortelle, E. & Bricogne, G. (1997). Maximum­ reductase of enterobacteria: 3. The Escherichia coli likelihood heavy-atom parameter refinement for multi­ hemoflavoprotein: catalytic parameters and the se­ ple isomorphous replacement and multi wavelength quence of electron flow. J. Bioi. Chem. 249, 1572- 1586. anomalous diffraction methods. Methods Enzymal. 276, 44. Knaff, D. B. & Hirasawa, M. (1991). Ferredoxin­ 472-494. dependent chloroplast enzymes. Biochim. Biophys. Acta, 60. Abrahams, J. P. & Leslie, A G. W. (1996). Methods 1056,93-125. used in the structure determination of bovine mito­ 45. Fritz, G., Roth, A, Schiffer, A., BUchert, T., Bourenkov, chondrial F, ATPase. Acta Crystallogr. , Sect. 0: BioI. G., Bartunik, H. D. et al. (2002). Crystal structure of the Crystal/ogr. 52, 30-42. adenylylsulfate reductase from the hyperthermophili c 61 . Cowtan, K D. (1994). 'DM': an automated procedure archaeon Archaeogloblls fulgidus at 1.6 A resolution. for phase improvement by density modification. Joint Proc. Nat! Acad. Sci. USA, 99, 1836-1841. CCP4 and ESF-EACBM Newsletter on Protein Crystal­ 46. Page, C. c., Moser, C. C. & Dutton, P. L. (2003). lography, 31, 83- 91 . Mechanism for electron transfer within and between 62. Brunger, A, Adams, P. D., Clore, G. M., DeLano, proteins. Curl'. Opin. Chem. Bioi. 5,551-556. W. L., Gros, P., Grosse-Kunstleve, R W. et al. (1 998). 47. Moller-Zinhahn, D. & Thauer, R K (1990). Anaerobic Crystallography & NMR System: a new software lactate oxidation to 3 COz by Al'chaeoglobus fulgidus suite for macromolecular structure determinations. via the carbon monoxide dehydrogenase pathway: Acta Crystallogr. , Sect. 0: Bioi. Crystallogr. 54,905-921. demonstration of the acetyl-CoA carbon-carbon 63. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. cleavage reaction in cell extracts. Arch. Microbiol . 153, (1991). Improved methods for building protein mo­ 215-218. dels in electron density maps and the location of 48. Kunow, J. , Linder, D., Stetter, K O. & Thauer, R K errors in these models. Acta Crystallogr., Sect. A: Found. (1994). F4zoHz: quinone oxidoreductase from Archaeo­ Crystal/ogl'. 47, 110-119. globus fulgidus. Characterization of a membrane­ 64. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). bound multisubunit complex containing FAD and Refinement of macromolecular structures by the iron-sulfur clusters. Eur. ]. Biochem. 223, 503-511. maximum-likelihood method. Acta Crystallogr., Sect. 49. Mander, G. J., Duin, E. c., Linder, D., Stetter, K O. & 0: BioI. Crystal/ogr. 53, 240- 255. Hedderich, R (2002). Purification and character­ 65. Coll aborative Computational Project, Number 4. ization of a membrane-bound enzyme complex from (1994). The CCP4 Suite: programs for protein crystal­ the sulfate-reducing archaeon Archaeoglobus fulgidus lography. Acta Crystllllogr., Sect. 0: BioI. Crystallogr. 50, related to heterodisulfide reductase from methano­ 760-763. genic . EliI'. J. Biochelll. 269, 1895-1904. 66. Laskowski, R A, MacArthur, M. W., Moss, D. S. & 50. Hedderich, R, Berkessel, A & Thauer, R K. (1989). Thornton, J. M. (1993). PROCHECK: a program to Catalytic properties of the heterodisulfide reductase check the stereochemical quality of protein structures. involved in the final step of methanogenesis. FEBS ]. Appl. Crystllllogr. 26, 283-291. Lett. 255, 67-71. 67. Klenk, H.-P., Clayton, R A, Tomb, J. F., White, 0 ., 51. Frazao, c., Silva, G., Gomes, C. M., Matias, P., Coelho, Nelson, K E., Ketchum, K A et al. (1997). The R , Sieker, L. et al. (2000) . Structure of a dioxygen re­ complete genome sequence of the hyperthermophilic, duction enzyme from Oesllifovibrio gigas. Nat. Stmct. sulphate-reducing archaeon Archaeoglobus flllgidLis. BioI. 7,1041-1045. Nature, 390, 364-370.