© 2000 Nature America Inc. • http://structbio.nature.com
letters Structure of a dioxygen reduction enzyme from Desulfovibrio gigas Carlos Frazão1, Gabriela Silva1, Cláudio M. Gomes1, Pedro Matias1, Ricardo Coelho1, Larry Sieker1, Sofia Macedo1, Ming Y. Liu2, Solange Oliveira1,3, Miguel Teixeira1, António V. Xavier1, Claudina Rodrigues-Pousada1, Maria A. Carrondo1 and Jean Le Gall1,2 1
© 2000 Nature America Inc. • http://structbio.nature.com
Instituto de Tecnologia Química e Biológica, Universidade Nova de Lisboa, Av. da República, Apartado 127, 2781-901 Oeiras, Portugal. 2Department of Biochemistry and Molecular Biology, University of Georgia, Athens, Georgia 30602, USA. 3Universidade de Évora, 7001 Évora, Portugal.
Desulfovibrio gigas is a strict anaerobe that contains a wellcharacterized metabolic pathway that enables it to survive transient contacts with oxygen. The terminal enzyme in this pathway, rubredoxin:oxygen oxidoreductase (ROO) reduces oxygen to water in a direct and safe way. The 2.5 Å resolution crystal structure of ROO shows that each monomer of this homodimeric enzyme consists of a novel combination of two domains, a flavodoxin-like domain and a Zn-β-lactamase-like domain that contains a di-iron center for dioxygen reduction. This is the first structure of a member of a superfamily of enzymes widespread in strict and facultative anaerobes, indicating its broad physiological significance. With the increase of atmospheric oxygen levels, some organisms became capable of aerobic life by evolving specialized enzymes that are now quite well characterized. It is also well documented, especially in the case of sulfate reducing bacteria1, that
a
some strict anaerobes adapted to survive in transient oxygencontaining environments. However, the enzymes that enable them to survive in these adverse conditions are in general poorly understood. An exception to this are the enzymes in Desulfovibrio gigas. In addition to classic detoxifying enzymes such as catalase and superoxide dismutase2, this bacterium utilizes a pathway that enables energy conservation by substrate level phosphorylation in the presence of dioxygen, which is reduced to water by a rubredoxin:oxygen oxidoreductase (ROO)3. Under these conditions, a unique electron transfer chain composed of three proteins — NADH:rubredoxin oxidoreductase, rubredoxin, and ROO — reoxidizes the reduced pyridine nucleotides formed during glycolysis, at the expense of a polyglucose reserve stored during normal anaerobic growth3,4 (Fig. 1). This process revealed the first known physiological role for rubredoxin in anaerobes3,5. The components of this chain have been extensively characterized in biochemical3,6 and spectroscopic studies5. ROO is a homodimeric enzyme (each monomer of 43 kDa) containing one flavin mononucleotide (FMN) cofactor per monomer, which is directly reduced by rubredoxin. The heme content in ROO is substoichiometric and varies from purification to purification3, a situation reminiscent of that of bacterioferritins. Most importantly, the heme was shown not to be involved in the reaction with oxygen5. The first enzyme of this chain, NADH:rubredoxin oxidoreductase, is highly specific towards D. gigas rubredoxin5,6. Furthermore, sequence analysis showed that the gene encoding rubredoxin is adjacent to that of ROO in the D. gigas genome5. In contrast to another recently proposed mechanism for oxygen detoxification in anaerobes involving the intermediary formation of a superoxide anion and hydrogen peroxide7, ROO provides both oxygen scavenging and energy conservation mechanisms; the direct reduction of dioxygen to water avoids the production of the above mentioned reactive oxygen species.
b
c
Fig. 1 Oxygen scavenging by ROO. a, The D. gigas oxygen utilizing pathway, composed of NADH:rubredoxin oxidoreductase (NRO), rubredoxin (Rd), and ROO, allows survival under oxic conditions. When using their carbon reserves, the end products of fermentation are mainly glycerol and ethanol, but in the presence of oxygen, NAD + is regenerated and the products are further oxidized to acetate and CO2 (ref. 4). b, Distribution of ROO homologs among anaerobes and facultative aerobes: Mt, Methanobacterium thermoautotrophicum; Af, Archaeoglobus fulgidus; Tm, Thermotoga maritima; Ph, Pyrococcus horikoshii; Pa, Pyrococcus abyssi; Dg, Desulfovibrio gigas; Ec, Escherichia coli; Pg, Porphyromonas gengivalis; Rc, Rhodobacter capsulatus; Ct, Chlorobium tepidum; Syn, Synechocystis. Repeated abbreviations refer to several homologs in the same genome. The percentage of similarity to Dg ROO is given in parentheses. c, Amino acid sequence of D. gigas ROO (iron ligands in red, conserved or conservatively substituted residues in blue) and color coded secondary structure (blue, lactamase-like domain; red, flavodoxin-like domain; arrows, α-helices; boxes, β-sheets).
nature structural biology • volume 7 number 11 • november 2000
1041
© 2000 Nature America Inc. • http://structbio.nature.com
letters a
Fig. 2 ROO is a modular enzyme. a, The ROO dimer (monomers in blue and brown), showing the β-lactamase-like (light) and flavodoxin-like (dark) domains, iron (orange spheres) and FMN (stick model). A two-fold axis relates both monomers. b, The β-lactamase-like domain. Left, a ribbon diagram with termini labeled. Right, stereo view of the β-lactamase-like domain (blue) superimposed on β-lactamases from Stenotrophomonas maltophilia11 (green), Bacillus cereus12 (deep pink) and Bacteroides fragilis13 (gold). Additional ROO structural elements in the region corresponding to the substrate groove of β-lactamases are indicated in dark blue. c, The flavodoxin-like domain. Left, ribbon diagram. Right, stereo view of the flavodoxin-like domain (blue) superimposed on Desulfovibrio vulgaris32 (brown) and Clostridium beijerinckii17 (violet) flavodoxins.
© 2000 Nature America Inc. • http://structbio.nature.com
b
c
Each monomer is composed of two domains, a β-lactamaselike domain containing a di-iron center and a flavodoxin-like domain containing a FMN. The di-iron center and the FMN are located more than 35 Å apart. The homodimer shows a head-totail arrangement through a noncrystallographic two-fold axis10, which brings the di-iron site of one monomer close to the FMN of the other monomer (∼6 Å apart). The intradimer contacts Overall structure involve 12% of an otherwise solvent accessible area extending In the D. gigas genome, the gene encoding ROO is located in a through one helix of the flavodoxin-like domains plus the region dicistronic operon together with the gene encoding rubredoxin around the cofactors (Fig. 2a). — Northern analysis using either ROO or rubredoxin as the probe reveals a single mRNA of 1.6 kb. ROO functions as a β-lactamase-like domain homodimer3, with each monomer containing 402 residues The β-lactamase-like domain (residues 1–249) is organized in a (Fig. 1c), one FMN and a di-iron center. The 2.5 Å resolution αβ/βα sandwich, with the two inner β-sheets flanked by two sets structure was refined to an Rfree of 22.8% and an R-factor of of three solvent exposed α-helices (Fig. 2b). The N-terminal half 16.2% and includes the homodimer as two crystallographically of the β-sandwich is further decorated by an additional twoindependent monomers (a total of 802 residues as both N-ter- stranded β-sheet that covers the di-iron site. This site is located mini were not detected in the electron density), four cofactors within a shallow groove at the interface between the two sheets and 170 water molecules. PROCHECK8 indicated that seven and is surrounded by αβ-loops and the flavodoxin domain from residues are in the generously allowed regions of the the other monomer (Fig. 2a,b). The domain fold is essentially Ramachandran plot and shows the global stereochemical prop- that of Zn-β-lactamases (penicillin hydrolases), despite having erties to be better or similar to PROCHECK protein reference low ( 2σ(I)) R.m.s. deviations from ideal geometry (Å) Bond distances Angle distances Restraint planes 1 2
ESRF, BM14 1.7393 100 31.9–2.7 99.0 167,852 99,2731 96.0 18.9 6.4 2.76–2.70 84.8 3.9 31.9 2 6,570 170 7 26,292 33,720 17.94 (16.18) 24.77 (22.82) 0.005 0.019 0.024
Friedel mates were considered as independent measurements. Statistics of outer resolution shells.
-92 and -399 mV in Clostridium beijerinckii17), as is expected from the relative excess of basic versus acidic residues surrounding the isoalloxazine ring in ROO. This effect brings its redox potential closer to that of its electron donor, rubredoxin (∼6 mV)5. Dioxygen scavenging by ROO-like proteins The dimeric arrangement of ROO allows the di-iron site and FMN to be in close enough proximity that together they are able to provide the four electrons needed for full reduction of one oxygen molecule to two water molecules. Electrons donated by rubredoxin to the flavin5 can be transferred to the di-iron site via Glu 81, which is within van der Waals distance of the flavin C8 methyl group. By analogy to what is known for dioxygen activating di-iron enzymes15, a putative mechanism may be outlined. One should note, however, that the ROO di-iron site includes an acidic ligand showing an atypical orientation — that is, Glu 81, which has the basic cis His 146 as a neighbor. Because the protein was isolated and crystallized aerobically, the three-dimensional structure we describe most certainly corresponds to the diferric state. This is supported by the observed Fe–Fe distance, and by the observation that the dioxygen molecule is not bound to the di-iron center. According to the mechanisms proposed for the other di-iron enzymes, upon reduction to the diferrous state, the µ-O(H) bridge is removed, with the release of one water molecule. Subsequently, dioxygen binds transiently to Fe 2 (assuming a role reminiscent of that of copper B in the heme-copper oxidases) to form a peroxo intermediate, bridging the two iron atoms. Further electron transfer from the flavin moiety leads to the release of the second water molecule and the formation of the diferric µ-O(H) resting state. Protons, which are essential for the 1044
ESRF, BM14 1.74001 100 31.9–2.7 99.0 175,458 98,5301 95.7 22.4 4.4 2.76–2.70 82.9 4.4 26.9
ESRF, BM14 1.0331 100 31.9–2.7 99.0 180,991 92,2041 98.1 32.4 3.6 2.76–2.70 95.1 11.0 10.5
Methods Cloning and sequence analysis. The gene encoding ROO was cloned and sequenced as described5. An Escherichia coli-like consensus promoter, 136 base pairs upstream of the rubredoxin ATG, was identified using Promoter Prediction by Neural Network (NNPP)23.
nature structural biology • volume 7 number 11 • november 2000
© 2000 Nature America Inc. • http://structbio.nature.com
letters
© 2000 Nature America Inc. • http://structbio.nature.com
This shared high identity with its counterparts in sulfate reducing bacteria. A putative ρ-independent terminator downstream ROO coding unit was found using a terminator search program (GCG package24). Crystallization, diffraction data collection and structure determination. Aerobically purified native ROO3 was crystallized by vapor diffusion in a cold room (4–10 °C) to avoid pseudomerohedric crystal twins10. Sitting drops of 3 µl of precipitant solution and 3 µl of protein solution (10 mg ml-1 protein) were deposited on microbridges and equilibrated against the precipitant solution (10% (v/v) PEG 6K, 100 mM Tris-maleic acid, pH 6.0). The labile heme was lost upon crystallization, and the flavin content (one FMN per monomer) was determined by HPLC analysis. The chemical nature of the metal center was established from a crystal fluorescence spectrum and the crystallographic multiwavelength anomalous dispersion (MAD) determination at the Fe X-ray absorption edge. Two diffraction data sets were collected, a four-wavelength MAD data set to 2.7 Å resolution at ESRF beamline BM14 and a single wavelength set to 2.5 Å resolution at DESY, EMBL Hamburg outstation, beamline X11 (ref. 10). Diffraction images were processed with HKL25. The F°A values were extracted from the MAD data with MADSYS26, and used in SHELXS-97 (ref. 27) to locate three of the four Fe sites, which were further refined with SHARP28. The fourth Fe was located from a residual map and included in the refinement. The phases were improved with SOLOMON29 to an overall figure of merit (FOM) of 0.95 at 2.7 Å. Polypeptide tracing was performed followed by residue assignment. Electron density fit of the refined model was performed with XTALVIEW30. The two molecules in the asymmetric unit were refined to 2.5 Å using SHELXL31. Homologous 1–4 distances and atomic displacement parameters of homologous atoms were restrained to their common values. The atomic displacement parameters of water molecules with hydrogen bonds to equivalent atoms in each monomer were restrained to their common values. The di-iron sites and their ligands were restrained to a common geometry without target values. The residual lobe of electron density found near the di-iron site in difference Fourier maps was initially assigned to a water molecule, but its refinement led to a B-factor that was too low. Taking into account the crystallization conditions it was then tentatively assigned to a dioxygen molecule. Sequence and coordinates. The primary structure has been deposited in GenBank (accession number AF218053) and the coordinates have been deposited in the Protein Data Bank (accession number 1E5D).
nature structural biology • volume 7 number 11 • november 2000
Acknowledgments We thank the University of Georgia fermentation plant for growing the bacteria, G. Leonard and V. Stojanoff (ESRF, Grenoble) and A. Thompson (EMBL, Grenoble) for help with data collection, M. Regalla (ITQB) for HPLC flavin analysis. This work was supported by grants from the NIH and PRAXIS XXI.
Correspondence and requests for materials should be addressed to M.A.C. email:
[email protected] Received 8 May, 2000; accepted 30 August, 2000. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33.
Krekeler, H., Teske, A. & Cypionka, H. FEMS Bicrobiol. Ecol. 25, 89–96 (1998). dos Santos, W.G., et al. J. Bacteriol. 182, 796–804 (2000). Chen, L., et al. Biochem. Biophys. Res. Commun. 193, 100–105 (1993). Fareleira, P., Le Gall, J., Xavier, A.V. & Santos, H. J. Bacteriol. 179, 3972–3980 (1997). Gomes, C.M., et al. J. Biol. Chem. 272, 22502–22508 (1997). Chen, L., et al. Eur. J. Biochem. 216, 443–448 (1993). Jenney, F.E.J., Verhagen, M.F., Cui, X. & Adams, M.W. Science 286, 306–309 (1999). Laskowski, R.A., MacArthur, M.W., Moss, D.S. & Thornton, J.M. J. Appl. Crystallogr. 26, 283–291 (1993). Jabs, A., Weiss, M.S. & Hilgenfeld, R. J. Mol. Biol. 286, 291–304 (1999). Frazão, C., et al. Acta Crystallogr. D 55, 1465–1467 (1999). Ullah, J.H., et al. J. Mol. Biol. 284, 125–136 (1998). Carfi, A., et al. EMBO J. 14, 4914–4921 (1995). Concha, N.O., Rasmussen, B.A., Bush, K. & Herzberg, O. Structure 4, 823–836 (1996). Nordlund, P. & Eklund, H. Curr. Opp. Struct. Biol. 5, 758–766 (1995). Solomon, E.I., et al. Chem. Rev. 100, 235–350 (2000). Allen, F.H. & Kennard, O. Chem. Des. Autom. News, 8, 31–37 (1993). Ludwig, M.L. et al. Biochemistry 36, 1259–1280 (1997). Jones, D.T.J. Mol. Biol. 287, 797-815 (1999). Wasserfallen, A., Ragettli, S., Jouanneau, Y. & Leisinger, T. Eur. J. Biochem. 254, 325–332 (1998). Williams, R.J.P. & Frausto da Silva, J.J.R. The natural selection of the chemical elements: the environment and life’s chemistry 285–414 (Clarendon Press, Oxford; 1996). Valladares, M. et al. Biochemistry 36, 11534–11541 (1997). Osano, E. et al. Antimicrob. Agents Chemother. 38, 71–78 (1994). Reese,M.G.,Harris, N.L., & Eeckman, F.H. Biocomputing: Proceedings of the 1996 Pacific Symposium, (eds Hunter, L. & Klein, T.E.) 737–738 (World Scientific Publishing Co., Singapore; 1996). Wisconsin Package Version 10.0, Genetics Computer Group (GCG), Madison, Wisconsin. http://www.gcg.com Otwinowski, Z. & Minor, W. Methods Enzymol. 276, 307–326 (1997). Hendrickson, W.A. & Ogata, C.M. Methods Enzymol. 276, 494–523 (1997). Sheldrick, G.M. Acta Crystallogr. A 46, 467–473 (1990). De La Fortelle, E. & Bricogne, G. Methods Enzymol. 276, 472–494 (1997). Abrahams, J.P. & Leslie, A.G.W. Acta Crystallogr. D 52, 30–42 (1996). McRee, D.E. J. Mol. Graphics 10, 44–46 (1992). Sheldrick, G.M. & Schneider, T.R. Methods Enzymol. 277, 319–343 (1977). Watt, W., Tulinsky, A., Swenson, R.P. & Watenpaugh, K.D. J. Mol. Biol. 218, 195–208 (1991). Sali, A. & Blundell, T.L. J. Mol. Biol. 234, 779–815 (1993).
1045
