Superoxide Dismutase from the Eukaryotic Thermophile ...

doi:10.1016/j.jmb.2008.11.031 J. Mol. Biol. (2009) 385, 1534–1555

Available online at www.sciencedirect.com

Superoxide Dismutase from the Eukaryotic ThermophileAlvinella pompejana: Structures, Stability, Mechanism,and Insights into Amyotrophic Lateral Sclerosis

David S. Shin1, Michael DiDonato1, David P. Barondeau1, Greg L. Hura2,Chiharu Hitomi1, J. Andrew Berglund3, Elizabeth D. Getzoff1,S. Craig Cary4,5⁎ and John A. Tainer1,2⁎
1Department of MolecularBiology, The Skaggs Institute forChemical Biology, The ScrippsResearch Institute, La Jolla,CA 92037, USA2Advanced Light Source,Lawrence Berkeley NationalLaboratory, Berkeley, CA 94720,USA3Institute of Molecular Biology,University of Oregon, Eugene,OR 97403, USA4College of Marine Studies,University of Delaware, Lewes,DE 19958, USA5Department of BiologicalSciences, University of Waikato,Hamilton 3240, New Zealand
Received 31 July 2008;received in revised form7 November 2008;accepted 11 November 2008Available online25 November 2008

*Corresponding authors. J. A. TaineChemical Biology, The Scripps [email protected] addresses: M. DiDonato, S

CA 92121, USA; D. P. Barondeau, DeAbbreviations used: ALS, amyotro

taurus superoxide dismutase; cDNAloop; GK1, Greek key loop 1; GK2,buffered saline; S–S, disulfide regiondismutase; SmSOD, Schistosoma manoleracea superoxide dismutase; VL, v

0022-2836/$ - see front matter © 2008 E

Prokaryotic thermophiles supply stable human protein homologs forstructural biology; yet, eukaryotic thermophiles would provide moresimilar macromolecules plus those missing in microbes. Alvinella pompejanais a deep-sea hydrothermal-vent worm that has been found in temperaturesaveraging as high as 68 °C, with spikes up to 84 °C. Here, we used Cu,Znsuperoxide dismutase (SOD) to test if this eukaryotic thermophile canprovide insights into macromolecular mechanisms and stability bysupplying better stable mammalian homologs for structural biology andother biophysical characterizations than those from prokaryotic thermo-philes. Identification, cloning, characterization, X-ray scattering (small-angle X-ray scattering, SAXS), and crystal structure determinations showthat A. pompejana SOD (ApSOD) is superstable, homologous, andinformative. SAXS solution analyses identify the human-like ApSODdimer. The crystal structure shows the active site at 0.99 Å resolution plusanchoring interaction motifs in loops and termini accounting for enhancedstability of ApSOD versus human SOD. Such stabilizing features mayreduce movements that promote inappropriate intermolecular interactions,such as amyloid-like filaments found in SOD mutants causing theneurodegenerative disease familial amyotrophic lateral sclerosis or LouGehrig's disease. ApSOD further provides the structure of a long-soughtSOD product complex at 1.35 Å resolution, suggesting a unified inner-sphere mechanism for catalysis involving metal ion movement. Notably,this proposed mechanism resolves apparent paradoxes regarding electrontransfer. These results extend knowledge of SOD stability and catalysis andsuggest that the eukaryote A. pompejana provides macromolecules highlysimilar to those from humans, but with enhanced stability more suitable forscientific and medical applications.

© 2008 Elsevier Ltd. All rights reserved.

Keywords: thermophile; thermostable proteins; superoxide dismutase;amyotrophic lateral sclerosis; amyloid filaments
Edited by M. Guss
r is to be contacted at Department of Molecular Biology, The Skaggs Institute forarch Institute, La Jolla, CA 92037, USA. E-mail addresses: [email protected];

tructural Biology, Genomics Institute of the Novartis Research Foundation, San Diego,partment of Chemistry, Texas A&M University, College Station, TX 77842, USA.phic lateral sclerosis; ApSOD, Alvinella pompejana superoxide dismutase; BtSOD, Bos, copy or complementary DNA; DSV, deep submergence vehicle; EL, electrostaticGreek key loop 2; HsSOD, Homo sapiens superoxide dismutase; PBS, phosphate-; SAXS, small-angle X-ray scattering; ScSOD, Saccharomyces cerevisiae superoxidesoni superoxide dismutase; SOD, Cu,Zn superoxide dismutase; SoSOD, Spinaciaariable loop; XlSOD,Xenopus laevis superoxide dismutase; ZnBR, zinc binding region.

lsevier Ltd. All rights reserved.

1535Alvinella SOD Structures, Stability and Mechanism

Introduction

Macromolecules from microbial extremophileshave many applications in biology, biotechnology,and industry.1 These microbes inhabit environmentswith extreme temperatures, pressures, pH, salinityand/or metal content. Bacteria and archaea living athigh temperature frequently provide highly stableproteins.1,2 These proteins are often characterizedprior to or in lieu of human homologs for threereasons: First, thermostable proteins are often better“behaved” such that individual conformations andordered flexible regions can be kinetically trapped atmesophilic temperatures.3 This characteristic mayhave contributed to the higher success rate of solvingthermostable protein structures in various labora-tories and by high-throughput methods.3–6 Second,purification is facilitated by heat denaturation ofrecombinant host proteins.7–9 Third, some archaealthermophilic proteins have closer resemblance tohuman proteins than those from lower-temperatureunicellular model systems, such as bacteria andyeast, or offer an orthologous system with minimalcomplexity.3,6,10,11 These factors all aid insights intoresidues, domains, and subunit interactions withrelevance to ubiquitous proteins in human macro-molecular systems.7–9,12–15 However, the lack ofcomplexity in microbial protein networks can limitextrapolation of multicellular eukaryotic pathways,as microbial pathways may lack orthologs ofproteins that interact with the more ubiquitouscentral enzymes. In other cases, microbial proteinswill differ enough in amino acid sequence, tertiaryfold, or assembly, making comparisons to humanproteins difficult. These limitations of analyzingthermophilic prokaryotic macromolecules wouldbe overcome if a suitable multicellular eukaryoticthermophile could be found as a resource to supplymore human-like thermostable macromolecules foranalyses.One of the most promising such candidate

thermophilic eukaryotes is the Pompeii worm,Alvinella pompejana. The Alvinellidae family ofpolychaetous annelids represent the most thermo-tolerant eukaryotes known, based on in situtemperature measurements of A. pompejana'simmediate environment16,17 and laboratory obser-vations of Paralvinella sulfincola thermotaxis.18 A.pompejana inhabits deep-sea hydrothermal ventsalong the East Pacific Rise, the spreading ridge axisbetween the Pacific and North American Plates.16,17

Here, volcanic activity causes seawater superheat-ing plus mixing with metal ion and sulfide-richvent fluids. The resultant metal sulfide precipitatescreate black smoke and help form large ventingchimneys. These emit the hottest fluids (∼350 °C)from orifices plus hot water (b150 °C) diffusivelythrough the sides, where adult Pompeii worms (2–6 cm in length) live within self-constructed tubes.This most thermotolerant, multicellular eukaryoteis also eurythermal (temperature-range tolerant): anover 60 °C gradient can exist along A. pompejana's 6-to 8 -cm tube length, suggesting A. pompejana has

protein isoforms with different stabilities.A. pompejanathrives at pH ∼5.5 amid toxic heavy metal ions,some at concentrations 1000 times higher than thatof ambient seawater,19 and is thus an outstandingcandidate organism for isolating stable macromole-cules resistant to high temperature, metal ions, andlow pH. Mean temperatures within the worm tubesare as high as 68 °C with spikes up to 84 °C,exceeding the 55 °C upper limit predicted foreukaryotic survival.16,17,19

To test whether A. pompejana has promise as aeukaryotic resource for human-like thermophilicmacromolecules, we collected A. pompejana wormswith the Alvin submarine and generated copy orcomplementary DNA (cDNA) for sequencing, clon-ing and protein expression. From the availablesequences encoding full-length proteins, we choseto test Cu,Zn superoxide dismutase (SOD) for threereasons. First, SOD is biologically and medicallyimportant with existing questions and missinginsights regarding its catalysis and stability thatmight be addressed by homologous thermophilicstructures. Second, SOD assembles into distincteukaryotic and microbial dimers, providing a testof the functional homology in solution. Third, themultiple existing and reasonably high-resolutionstructures from many different species,20–27 includ-ing human (Homos sapiens SOD, HsSOD), provide adetailed basis for key cross-species comparisons to afirst thermophilic eukaryote protein crystal struc-ture. A. pompejana Cu,Zn superoxide dismutase(ApSOD) thus provides an important test case forA. pompejana structure determinations by X-raycrystallography for detailed active-site comparisonsand solution small-angle X-ray scattering (SAXS) forexamination of the assembly state in solution. Thusfar, microbial SOD structures are monomeric orexhibit different dimer interfaces unlike those fromeukaryotes.28–31 Therefore, using samples from aeukaryotic, rather than a prokaryotic, thermophileshould yield a more relevant structure for gaininginsights into human SOD assembly and stability.SOD is a master eukaryotic regulator of oxygen

radicals, with relevance to brain pathology, cancer,aging, and cell biology.32 Together, superoxide andnitric oxide can initiate arachidonate and lipidperoxidation associated with cell signaling and cellkilling, where the biological levels of these reactiveoxygen species are precisely controlled by the SODand nitric oxide synthase enzymes.32–34 SODcatalyzes the disproportionation of toxic superoxideradicals (O2

·−) to oxygen (O2) and hydrogenperoxide (H2O2). SOD is also a relatively stablemetalloenzyme in mesophiles, where human SODhas a two-component set of melting temperaturesof 75 and 83 °C.35 Yet, in spite of this stability andmultiple structures from different species, includingthat of human SOD, being solved over the course ofN25 years, a structure revealing any substrate orproduct complex has remained elusive. Wehypothesized that if ApSOD was more stable, thismay allow us to trap a product complex to addresskey questions regarding the SOD electron transfer

1536 Alvinella SOD Structures, Stability and Mechanism

and mechanism. Furthermore, mutations associatedwith SOD structural defects can cause amyotrophiclateral sclerosis (ALS), or Lou Gehrig's disease, afatal neurodegenerative disorder.36–40 Thus far,more than 20% of inherited or familial ALS casesare correlated with SOD1 gene defects.41,42 Thedestabilizing SOD mutations that cause ALS resultin the disease's characteristic and currently untrea-table selective destruction of motor neurons leadingto progressive paralysis and death. Since theprotein sequence and structures of SOD proteinsare conserved, we hypothesized that changes inamino acid sequence and structure in ApSODshould point to stabilizing features and provideinsights into how single-site ALS mutations cancause the framework destabilization identified asthe hallmark of ALS mutations,38 provided ApSODhas enhanced thermostability.We report here the sequencing, identification,

and cloning of the cDNA encoding ApSOD alongwith the expression, purification, and combinedbiochemical and structural analyses of its product.We discovered ApSOD has remarkably high level ofsequence identity with HsSOD and other mam-malian SOD enzymes. Yet, we found ApSODsubstantially more stable than HsSOD. Moreover,crystals from initial conditions diffracted justbeyond 1 Å resolution, the highest yet reportedfor a wild-type SOD structure. Despite the highdegree of sequence similarity between the struc-turally characterized homologs, we identifiedresidue differences that evidently enhance stabi-lity. Furthermore, the stability and homogeneityof ApSOD protein samples were sufficient tosolve a crystal structure in the presence of thereaction product H2O2 to examine catalysis andfeatures promoting conformational stability andto generate an ab initio SAXS structure tocharacterize quaternary assembly in solution.These results extend knowledge of SOD stabilityand catalysis and also suggest that eukaryote A.pompejana offers a unique resource of macromo-lecules of enhanced stability for science andtechnology.

Fig. 1. Alvinella pompejana. (a) A. pompejana worm (redarrow) projecting out of its vent tube with gills fullyextended. The A. pompejana constructed vent tube is notedby the yellow arrow. (b) Ventral view of an∼4 -cmPompeiiworm after collection (scale bar represents 1 cm) withcollapsed gills on the anterior side (red structures onleft) and a bulge (middle) from gas expansion due to the∼250-atm change in pressure following collection.

Results

A. pompejana sample collection, cloning, andhomology to human SOD

Due to its 60–84 °C habitat, we chose to test theeukaryotic deep-sea hydrothermal-vent worm A.pompejana for its potential as a resource of thermo-stable proteins for structural studies. To initiatestructural analyses of proteins isolated from thiseukaryotic thermophile, we collected A. pompejanaworms roughly 2–6 cm in length from worm tubeson black smoker chimneys at hydrothermal ventsites along the Axial Summit Caldera of the EastPacific Rise at a depth of approximately 2500 musing the deep submergence vehicle (DSV) Alvin.

The vent fluid within worm tubes averaged 60 °C asmeasured by a time-lapse temperature probe,consistent with previous recordings.16,19,43 The gill-bearing anterior ends of the worms project towardvent tube openings (Fig. 1a), while the posteriorends of the worms directly experience hot vent fluid.Live worms were removed from the tubes andplaced in a container filled with an RNA-stabilizingsolution to protect RNA during transfer from theseafloor to the ocean surface (Fig. 1b).We extracted RNA from the posterior end of the

worm, which experiences the hottest vent-fluidenvironment, made cDNA libraries and sequencedthe libraries to search for and identify the cDNAcorresponding to the A. pompejana SOD1 processedmRNA transcript. The sequence encoding ApSODwas found and amplified by PCR and cloned into abacterial expression vector. Following expression,ApSOD protein purification steps included heatdenaturation at 65 °C for up to 1.5 h to removenonthermostable bacterial proteins. The isolation ofsoluble protein and detection of SOD activity(Supplementary Fig. S1) verified that ApSOD isthermostable and allows for efficient expression andpurification.Structure-based sequence alignment of ApSOD

with the known eukaryotic structures of bovine (Bostaurus or BtSOD),25,27 human (Homo sapiens or

Fig.

2(legend

onnext

page)



HsSOD),24,26 trematode (Schistosoma mansoni orSmSOD),20 frog (Xenopus laevis or XlSOD),21 spinach(Spinacia oleracea or SoSOD),22 and budding yeast(Saccharomyces cerevisiae or ScSOD)23 SODs revealedhigh conservation (Fig. 2a). At the amino acid level,ApSOD is strikingly conserved with mammalianSODs. Yet, phylogenetic analysis correctly places theApSOD sequence relative to the mammalian SODsequences (Fig. 2b). Among SODs with solvedstructures, BtSOD shares the greatest amino acidconservation with ApSOD (67% identity and 90%similarity over 151 residues), despite their geneticseparation (Fig. 2a). HsSOD is also very similar toApSOD, with 59% identity and 87% similarity over149 residues. Most ApSOD residues located atpositions corresponding with human ALS muta-tions either match the wild-type HsSOD residues ordiffer from both wild-type HsSOD and its ALSmutants. Only ApSOD Lys21 and Lys98 matchhuman ALS Lys mutations at HsSOD Glu21 andGlu100, respectively (Fig. 2a).

ApSOD crystal structure and active site

To test our hypotheses that proteins from athermophilic eukaryote will offer advantages forstructure determination and identification of stabi-lity traits, we crystallized the ApSOD protein andsolved its X-ray crystal structure. ApSOD proved asefficient for crystallization as it was for expressionand purification. Crystals in space group P6122grew in the first 24-well screen with ammoniumsulfate as the precipitant. The initial crystals,without additional refinement of experimentalconditions, diffracted X-rays beyond 1 Å resolution(Supplementary Fig. S2). Molecular replacementwith the human SOD probe 1PU039 successfullyprovided phasing information.The resulting electron densitymaps showed atomic

detail for residues 1–151, reflecting high order. Afterrefinement to 0.99 Å resolution, the structure had anR-factor of 11.2% and R-free of 13.8% (Table 1).Currently, this is the highest resolution and only b1 Åresolution wild-type SOD structure in the ProteinData Bank (PDB). Significantly, the resolution andquality of our ApSOD structure allowed us to refinethe active-site metal centers without geometricrestraints, determine experimentally defined stan-

Fig. 2. ApSOD sequence conservation, fold, and residueother eukaryotic Cu,Zn SOD proteins with solved structures: BS. oleracea; Sc, S. cerevisiae. Structural elements and secondary stbelow alignment mark ALS sites in HsSOD: green oval, ApSOhas a different residue; red star, ApSOD residue represents an Aligand; D, disulfide cysteine; B, bridging histidine; Z, zinc-bindicaps; S, stacking residue, S⁎, stacking residue hydrogen-bondinteractions noted in Fig. 8 and Table 2. Underlined numbersGlu68, which also makes side-chain interactions. (b) Phylogenedivergences. Labels at nodes represent bootstrap values and bStereo view of the ApSOD structure. Key structural elements inGK1 and GK2, Greek key loops 1 and 2, respectively; S–S, disloop); otherwise, β-strands are cyan and loops are gray. N anddimer interface on the opposite side of the subunit from the acwith bridging histidine His61 in yellow.

dard uncertainty errors for bond lengths and angles,and model anisotropic thermal displacement para-meters (B-factors).The ApSOD structure (Fig. 2c) resembles struc-

tures of other eukaryotic SOD proteins, but crystalscontain only one SOD subunit per asymmetric unit.The protein core consists of an eight-strandedantiparallel Greek key β-barrel,25 where loops β3/β4 and β6/β7 form +3 β-strand or Greek keyconnections25,44 (GK1 and GK2, respectively). Thestructure shares the short type II′ turn between β2and β3with BtSOD, as opposed to longer loops. Twoelongated loops extend from the β-barrel to form themetal-containing active-site channel and connec-tions critical for structural integrity and function(Fig. 2a and c). The β7/β8 or electrostatic loop (EL)acts in the guidance of the O2

·− substrate into theactive site,45 accounting for the enzyme's faster thandiffusion catalytic rate. In ApSOD, this loop isbounded by Cu ligand His118 and conservedArg141, which is implicated in hydrogen bondingwith O2

·−.46,47 The β4/β5 loop contributes to thedimer interface in other eukaryotic SODs and isstabilized via a disulfide bond between Cys55 andβ8 Cys144. This portion of β4/β5 is termed thedisulfide (S–S) subloop or region. Despite prolongedexposure to the reducing effects of the synchrotronX-ray beam, the ApSOD disulfide bond remainsoxidized, having a Sγ–Sγ distance of 2.248 ±0.008 Å,matching the native left-handed spiral confor-mation.25 The disulfide stability in the face ofreducing ionizing radiation is consistent with theframework stability of ApSOD. β4/β5 also encom-passes the Zn2+-binding region (ZnBR), as it containsall Zn2+-liganding residues. Zn2+ ligand His61divides the disulfide and zinc-binding subloopregions and also coordinates the catalytic Cu ion,so it is termed the bridging histidine. Since theβ4/β5loop connects the dimer interface, disulfide bondand the metal-binding sites together, their relativestabilities are interrelated.The ApSOD active-site geometry is well conserved

with HsSOD. Initial 2Fo−Fc electron density forApSOD revealed a predominantly reduced Cu(I) ion,based on its trigonal coordination. After severalrounds of crystallographic refinement, Fo−Fc differ-ence maps revealed partial occupancy for theoxidized Cu(II) ion, which was verified by R-factor

function. (a) Structure-based alignment of ApSOD witht, B. taurus; Hs, H. sapiens; Sm, S. mansoni; Xl, X. laevis; So,ructure of ApSOD are noted above the alignment. SymbolsD shares wild-type human residue; black square, ApSODLS mutation. Letters below alignment: C, copper-binding

ng ligand; H, H2O2-liganding residue; P, stabilizing prolineing partner. Numbers below alignment represent pairedrefer to main-chain atoms involved in interactions, excepttic tree for eukaryotic SODs shows estimated evolutionaryranch lengths are indicated (scale bar represents 0.1). (c)(a) are color coded with abbreviations (VL, variable loop;

ulfide region; ZnBR, zinc-binding region; EL, electrostaticC denote termini. The black bar (left) marks the potentialtive channel. Metal-liganding residues are shown as sticks

Table 1. Crystallographic data collection and analysis

Native H2O2 complex

Data collection and processingX-ray source BL12.3.1a BL12.3.1a

Wavelength (Å) 0.954 1.1002θ tilt (high-resolution

set) (°)20 (27) 22

Detector distance(high-resolution set) (mm)

350 (250) 350

Space group P6122 P6122Unit cell lengths: a, b, c (Å) 62.49, 62.49,

163.1962.56, 62.56,

163.75Unit cell angles: α, β, γ (°) 90, 90, 120 90, 90, 120Data range (last shell) (Å) 50–0.99

(1.03–0.99)50–1.35

(1.40–1.35)Observations (unique) 462,469

(98,085)179,249 (39,844)

Completeness (last shell) (%) 93.2 (54.0) 93.6 (58.0)Average redundancy

(last shell)4.7 (1.7) 4.5 (2.0)

Rsym (last shell)b 0.068 (0.29) 0.055 (0.32)I/σI (last shell) 15.45 (2.50) 22.62 (2.10)

RefinementResolution 45–0.99 45–1.35Reflections FN0

(cross-validation)93,084 (4,911) 37,795 (1994)

Data/parameter ratio 6.7 2.9Nonhydrogen protein

atoms (solvent)1137 (400) 1138 (321)

Rcryst (%)c 11.2 12.8Rfree (%)d 13.8 17.6

The rmsd between the structures was 0.09 Å for 151 Cα atoms.a Advanced Light Source, Berkeley CA.b Rsym is the unweighted R value on I between symmetry

mates.c Rcryst=∑hkl||Fobs(hkl)|−|Fcalc(hkl)||/∑hkl|Fobs(hkl)|.d Rfree= the cross validation R-factor for 5% of reflections

against which the model was not refined.


calculations and omit maps (Fig. 3a and b). Dualoccupancy of copper resulting from amixture of Cu(I)and Cu(II) oxidation states has been reported in otherhigh-resolution Cu,Zn SOD structures and can resultfrom reduction in the synchrotron X-ray beam.20,26,27

The electron density maps and refinement gaveaccurate Cu(I) and Cu(II) ion positions and Cu(I)–ligand bond lengths of 1.991±0.008, 1.930±0.007, and

Fig. 3. ApSOD active site. (a) Stereopair showing trigonaApSOD structure. Composite omit 2Fo−Fc density contouredwith experimentally defined standard uncertainty errors are nshown between respective bonds. (b) Rotation of (a) to show deare scaled to reflect occupancy. (c–f) Overlay of the 0.99 Å ressolvent atoms) and the 1.35 Å resolution ApSOD–H2O2 moSimulated annealing 2Fo−Fc omit electron density (light andApSOD structure and (d) for the 1.35 Å peroxide-containing selectron density at high levels, the center of mass for the constW0 and (d) H2O2 contoured at 5σ and 3σ, respectively. (c) Lowwhich suggests that while W0 lies near the hydrogen peroxidea site reserved for a larger molecule or that a minor compoposition. (d) Lower 2σ density for the peroxide-soaked structurealistic hydrogen-bonding distances to protein and solvent. HCu(II) is shown as dashes and bonds between copper and protand (d) to show details of other solvent atoms in the active site.and 3σ and (f) at 3.5 and 2.5σ. (c and e) When water is bound inrole in maintaining the active-site channel to attract superoxthrough interactions with the substrates and products.

1.970±0.008 Å for His44, His46, and His118, respec-tively (Fig. 3a), but alternative Cu(II) ligand positionswere not modeled. The Zn ion has full occupancy andis liganded by His61, His69, His78, and Asp81 in adistorted tetrahedral arrangement, consistent withother eukaryotic SOD active sites. The final ApSODmodel had refined occupancies of 80% for Cu(I) and20% for Cu(II), with the copper positions separated by1.06±0.01 Å.Ordered water molecules in the active-site chan-

nel provide insights into enzyme interactions withO2

·− substrate and H2O2 product (Fig. 3c and e).Final 2Fo−Fc electron density maps show that watermoleculeW0 resides near the expected position47 forthe Cu(II)-bound oxygen atom of the superoxideradical. The B-factor ofW0 is 20.7 Å2, as compared toaverage B-factors of ∼12.0 Å2 for neighboring watermolecules W1–W3, consistent with this site's abilityto hold a larger diatomic molecule. At high contourlevels, simulated annealing 2Fo−Fc omit electrondensity matches 2Fo−Fc electron density, identifyingthe center of mass for W0. However, at lowercontours, omit density for W0 extends away fromthe Cu ions, making a comma-shape. These resultssuggest that W0, with a refined occupancy of 96%,lies in a site destined for a larger molecule and mayhave some mobile character (see SupplementaryResults R1 and Supplementary Fig. S3).

H2O2 complex structure and active-site chemistry

SOD catalyzes the disproportionation of super-oxide to oxygen and hydrogen peroxide. In the firsthalf reaction: O2

·−+H++Cu(II)SOD→Cu(I)SOD+O2.In the second half reaction: O2

·− +H+ +Cu(I)SOD→ Cu(II)SOD+H2O2. Despite long-term effortsto characterize details of the SOD active-sitestructure and highly efficient reaction, the mechan-ism of H2O2 formation in the second half reactionremains controversial. In the originally proposedinner-sphere mechanism,47 the O2

·− substrate andH2O2 product interact directly with the Cu ion,His61, Arg141, and a water molecule. Later, thesecond half reaction was proposed to be outer

l coordination of Cu(I) by histidine ligands in the 0.99 Åat 3σ clearly defines individual atoms. Bonding distancesoted by arrows. Ligand–Cu–ligand angles with errors arensity for the two copper positions. Copper-colored spheresolution ApSOD model (light green carbon and dark greendel (salmon-colored carbon and red solvent atoms). (c)dark green) for the solvent in the 0.99 Å noncomplexedtructure (light and dark red). By contouring 2Fo−Fc omitituent in the predicted active site can be located: (c) waterer-level contours of 3σ show a residual curved tail for W0,O2 position, the water molecule is mobile, as it is bound innent of the occupancy may reside near the peroxide O1re is cylindrical and conforms to two oxygen atoms withydrogen bonds are shown as dots. The W0/H2O2 bond toein as continuous lines. (e and f) Rotation of the view in (c)For clarity of position andmovements (e) is contoured at 4the active site, W7 is bonded to Arg141, which may play a

ide. (d and f) Arg141 then likely plays a role in catalysis


sphere, based on the azide-binding geometry.48 Thisouter-sphere mechanism requires that the negativelycharged electron be transferred from positive Cu(I)to negative O2

·− over an intervening distance N3 Å,yet explains neither (1) the energetics driving therapid electron transfer nor (2) the coupling of proton

Fig. 3 (legend on

transfer with electron transfer. To resolve theseparadoxes and provide additional insights into theSOD mechanism, we used ApSOD to tackle thestructure of the SOD product complex.Although rapid catalysis by SOD previously

thwarted trapping of substrate or product complexes

previous page)


for structural studies, we were able to form the H2O2complex with ApSOD. The ApSOD crystal soakedaerobically in H2O2 was isomorphous with theunsoaked crystal, diffracted to 1.35 Å resolutionand allowed refinement to an R-factor of 12.8% andan R-free of 17.6% (Table 1). As expected, theApSOD–H2O2 complex electron density mapsrevealed a structure nearly identical in global foldand side-chain orientations to the ApSOD structure,but with a diatomic molecule in the positionpredicted for an O2

·−/H2O2 molecule within theactive site as well as subtle differences in thepositions and occupancies of nearby water mole-cules. These differences allowed detailed compar-isons of active-site stereochemistry relevant tounderstanding the enzyme mechanism (Fig. 3c–f).The initial 2Fo−Fc maps for the ApSOD–H2O2

complex readily revealed electron density for adiatomic oxygen species with an interatomic dis-tance of ∼1.5 Å in the predicted O2

·− binding site.We were able to refine two oxygen atoms withinthe site with full occupancy and the expected 1.49 ÅO–O bond length for H2O2. This ruled out thepossibility of simultaneous occupancy of two watermolecules in the site with combined occupanciestotaling 100% (see Supplementary Results R1 andSupplementary Figs. S3 and S4). Final 2Fo−Fc omitelectron density maps show the H2O2 moleculebound between Cu and Arg141 (Fig. 3d). In thestructure of the ApSOD–H2O2 complex, as in theApSOD structure, the Cu ion was found predomi-nantly in the reduced Cu(I) position, again suggest-ing X-ray-induced reduction. Following refinement,occupancies were 92% for the reduced Cu(I) positionand 8% for the oxidized Cu(II) position. In contrast,H2O2occupieda single binding site that is independentof the Cu ion redox state. In our ApSOD–H2O2 com-plex structure, the Cu(I) position is 3.47±0.07 Å awayfrom the proximal H2O2 oxygen atom O2, whichnearly superimposes with W0 in the noncomplexedApSOD structure (Fig. 3c and d and SupplementaryFig. S3), and water in other high-resolutionstructures26,27 (Supplementary Fig. S4) as well as theproximal atoms of bound anions in the Cu(II)containing ScSOD48 and BtSOD49 structures (Supple-mentary Fig. S5). Together, steric restrictions (Supple-mentary Fig. S4) and hydrogen bonding (Figs. 3c and

Fig. 4. Unified structure-based mechanism for SOD. (a andstructure and (b) the 1.35 Å H2O2 complex structure show alteofW0with H2O2, and serve as a guide to the mechanism schemmetals are colored matching the metal, where those to proteindenote movements and red dashed lines split the upper and loare blue. (c andd) First, anO2

·− radical displaces active-site Cu(IItermed O1, to Arg141 NH1, while the other, termed O2, bondN1 Å deeper into the active-site channel, while simultaneouslyrotating the side chain away from Cu(I), and O2 is released.bound within the same site, due to sterics imposed by pronegatively charged electron to the superoxide O2 atom, postoward O2

·− such that orbital overlap occurs for inner-sphere eleto or simultaneously with the electron transfer to form a coppeis converted to Cu(II) andH2O2 following addition of another pproduce H2O2. After the transfers, Cu(II) is coordinated to His6of H2O2 in the liquid state favoring reaction equilibrium towa

d and 4a and b) favor binding water, diatomic oxygenmolecules, and anions in this site.Consequently, we propose a unified general

mechanism for Cu,Zn SOD that takes into accountthis binding site, recent kinetic isotope data50 andsteric restrictions at the Cu(I) site. In this structure-based mechanistic proposal, both half reactionsproceed by an inner-sphere mechanism. In the firsthalf reaction (Fig. 4c–e), the O2

·− substrate binds Cu(II); Cu(II) is reduced to Cu(I), while O2

·− isoxidized to O2; and the bond between the Cuion and bridging ligand His61 Nɛ1 is broken,leaving Nɛ1 protonated. In the second half reaction(Fig. 4e–g), a proton from His61 Nɛ1 and anelectron from Cu(I) are donated to O2

·−; Cu(I) isoxidized to Cu(II), while O2

·− is reduced to H2O2(or HO2

−); and the bond re-forms between thecopper and His61. As we observed mobility for theCu ion, yet conservation of the substrate/productsite, we propose that the positively charged Cu(I)ion is attracted to and moves toward the nega-tively charged O2

·− substrate (Fig. 4e and f). To testthis, we manually moved the Cu(I) ion towardhydrogen peroxide O2 to see how closely we couldposition these atoms while maintaining bondingdistances between 1.9 and 2.1 Å to His44, His46,and His118 (for details, see Supplementary ResultsR2 and Supplementary Fig. S5). During this part ofthe reaction, the Cu ion could transition throughcoordination geometry similar to that predicted foranion binding45 and observed for azide binding,48

with His44, His61, His118, and the O2·−/H2O2

intermediate as equatorial ligands and W1 and/or His46 as axial. Significantly, we found that theCu ion could retain a distorted trigonal coordina-tion, while shortening the distance to the peroxideO2 to b2.8 Å, which is reasonable for inner-sphereorbital overlap. Facile distortion of trigonal Cu(I)geometry is promoted by its filled shell d10 electro-nic configuration, which is not subject to geometry-dependent ligand-field stabilization energy. Similardynamic elasticity and motion in a catalytic Zn site(also d10) has been proposed recently for themethionine synthases.51 For SOD, we propose thatelasticity in the Cu(I) geometry, coupledwith proteinand substrate mobility in solution, can permit directcontact with O2

·− for inner-sphere electron transfer;

b) Stereo views of ApSOD active sites from the (a) 0.99 Årnative solvent positions and bonding, due to replacemente. Hydrogen bonds are shown as purple balls and bonds toare solid. (c–g) Scheme for SOD mechanism. Green arrowswer copper positions. Atoms that form hydrogen peroxide) bindingwaterW0. ThenO2

·− hydrogen-bonds one oxygen,s Cu(II). (d and e) This reduces Cu(II) to Cu(I), shifting itbreaking its bond to His61. His61 Nɛ2 is then protonated,(e and f) During the second half reaction, a second O2

·− istein and bonds to His61, Arg141, and W4. To donate aitively charged Cu(I) is attracted to and moves upwardctron transfer. His61 Nɛ2 also donates a proton to O2

·− priorr–hydroperoxide intermediate. (f and g) This intermediateroton likely fromW4 in trans to avoid steric interference to1, and W2 andW3 likely reorient the H2O2 orbitals to thatrd the product.

Fig. 4 (legend on previous page)



no long-distance outer-sphere electron transfer isrequired.Our Cu-bound peroxide position and proposed

inner-sphere mechanism are consistent with kineticisotope results.50 These data support inner-sphereinteractions for the first half reaction and suggest arate-determining proton transfer for the second halfreaction at very high pH, where the bridging ligandHis61 would not be protonated. For an outer-spheremechanism, one might expect that the electrontransfer would be rate limiting with relativelyrapid proton transfer from His61 and/or adjacenthydrating water molecules. At physiological pH,electron transfer from the Cu(I) and proton transferfrom His61 could be coupled for the inner-spheremechanism, with superoxide being dismuted intoH2O2 at the position we define in our structure.The ApSOD structures show conserved, ordered

water molecules closely packed in the active-sitechannel (Supplementary Fig. S4). The hydrogen-bonding network formed by these conserved watermolecules aligns the substrate and promotescoupled proton and electron transfer (Fig. 4b andf). W4 (or the water molecule that occupies itsposition in other structures), which hydrogen-bondsto Arg141 Nɛ, rather than W2 (the proposed protondonor for the outer-sphere mechanism48), is theprobable second proton donor for H2O2 formation(Fig. 4b and f). The His61 Nɛ2–O2–O1–W4 stereo-chemistry is virtually planar with a trans dihedralangle of 174°, and with 137° Nɛ2–(O2–O1) and 121°(O2–O1)–W4 bond angles minimizing steric repul-sions during this reaction step. Water molecules W2and W3 are positioned to stabilize the H2O2 leavinggroup (Fig. 4b and g). The W2–(O1–O2) angle is 96°and the (O1–O2)–W3 angle is 76°, both of which arenear the 94.8° calculated angle for H–O–O hydrogen

Fig. 5. Thermal anisotropy probability ellipsoids for the AApSOD active-site structure. (b) The 1.35 Å ApSOD–H2Ocorresponding to the upper Cu ion position point between theion ellipsoid from the H2O2 complex structure defines the(Supplementary Fig. S5). The anisotropic B-factor ellipsoids forbond with Cu(II) in the upper position and to break the bond

peroxide bonds in the liquid or gaseous phase. TheW2–(O1–O2)–W3 dihedral angle of 142° is also inrange to stabilize the expected staggered anticlinicalconformation for H2O2 in solution. Thus, the con-served hydrogen-bonding network of ordered watermolecules in our ApSOD structures favors thebinding of substrate and product in the same locationand promotes coupled proton and electron transfer.The high resolution of these ApSOD structures

allowed us to calculate anisotropic B-factor para-meters, which provided support for Cu ion mobilitydescribed above (Fig. 5). In both ApSOD structures,most atoms were relatively isotropic; however,inspection of the active site revealed anisotropiesconsistent with our proposed mechanism: (1) thelong axes of the Cu(II) ion thermal ellipsoids pointtoward the peroxide O2 atom orW0, andW1; (2) theshape of the Cu(II) ellipsoid from the H2O2 complexstructure tracks the trajectory of the copper move-ments expected for transition from a trigonalcoordination arrangement to a square planar inter-mediate (Fig. 5 and Supplementary Fig. S3); and (3)the ellipsoids for the His61 side chain point towardH2O2 and, furthermore, match the pivot required tobind a moving copper ion. Thus, the high resolutionof these ApSOD structures allowed us to refine theproposed SOD catalytic mechanisms47,48 by visua-lizing how motions of the Cu(I) ion toward O2

·−

would allow inner-sphere electron transfer throughorbital overlap.

Subunit interactions in solution by SAXS

The oligomeric assemblies of homologous proteinscan differ between eukaryotes and prokaryotes,exemplifying another advantage of using a eukar-yotic thermophile for macromolecular structural

pSOD Cu-binding region. (a) The higher-resolution 0.99 Å2 complex active-site structure. The thermal ellipsoidsO2

·−/H2O2 binding site andW1. The shape of the upper Cuexpected path of Cu ion movement during catalysisthe His61 side chain match the motion expected to form ato Cu(I) in the lower position.


studies relevant to human disease. The SOD struc-tures from six eukaryotic species solved thus farexhibit a conserved dimeric assembly, whereasstructures of prokaryotic SODs are monomers ordissimilar dimers assembled with a distinctinterface.28–31 Surprisingly, only a single ApSODsubunit is found in the asymmetric unit of ourcrystals. The crystal packing dimer formed bycrystallographic 2-fold symmetry resembles eukar-yotic SOD dimers, yet only 11 out of 22 of the dimerinterface residues are fully conserved with those inthe other eukaryotic SODs with solved structures.So, we analyzed ApSOD's quaternary structure insolution by SAXS,52 additionally testing the suit-ability of A. pompejana to supply thermostableproteins that maintain human-like assemblies.Comparison of the experimental SAXS profile for

ApSOD with theoretical profiles calculated fromcrystal structures representing a single subunit anddimer of HsSOD39 and a bacterial SOD dimer fromActinobacillus pleuropneumoniae28 indicated aHsSOD-like dimer assembly (Fig. 6a). Furthercomparison of the ApSOD experimental scatteringintensity and pairwise interatomic distance function

Fig. 6. Determination of ApSOD dimer assembly and an abexperimental SAXS profile for ApSOD (black dots) to scatterinSOD dimer from A. pleuropneumoniae (magenta line) and of Hs(blue line), indicate that ApSOD assembles as a human-like SOfor ApSOD (black dots) to scattering profiles calculated fromand dimer (green line) further supports ApSOD dimer assemblfunctions from ApSOD SAXS data (black line) with P(r) funct(green line) crystal structure models confirms dimer assembly ithe interatomic pair distances (abscissa). (d and e) X-ray crApSOD dimer model (ribbon diagram) docked into an ab init

[P(r)] profiles with theoretical profiles calculatedfrom structures of the ApSOD single subunit andcrystallographic dimer also indicated a eukaryotic-type dimer assembly in solution (Fig. 6b and c). Theexperimental curve of the P(r) function and Dmaxvalue of 70 Å were in excellent agreement with thetheoretical P(r) function and calculated Dmax of 68 Åfor the dimer, but not the single subunit (calculatedDmax of 47 Å). Ab initio structure calculations basedon SAXS data converged on a well-defined mole-cular envelope in which the X-ray crystal structuredimer model was readily docked (Fig. 6d and e).Thus, ApSOD is homogeneous, stable enough forsolution structural analyses, and forms character-istic SOD dimers.

ApSOD stability and relations to ALS mutations

To evaluate if ApSOD is more stable than HsSOD,we characterized ApSOD biophysically and struc-turally. As SOD instability is implicated as animportant factor in ALS pathogenesis,32,37–40 resultson ApSOD stability may provide useful insights. Weused CD to monitor denaturation of ApSOD by

initio structure in solution by SAXS. (a) Comparison of theg profiles calculated from crystal structures of the bacterialSOD, both as a single subunit (orange line) and as a dimerD dimer. (b) Comparison of the experimental SAXS profilethe crystallographically defined ApSOD subunit (red line)y. (c) Comparison of the P(r) pairwise interatomic distanceions calculated for the single subunit (red line) and dimern solution. Scattering intensity (ordinate) is plotted againstystal structure of the crystallographic symmetry-relatedio SAXS structure solution (surface).

Fig. 7. ApSOD high stability. Pompeii worm SOD ismore stable than human SOD. Circular dichroismindicates that the unfolding midpoint for ApSOD dena-turation requires nearly 1 M higher guanidine than theunfolding midpoint for HsSOD.

Fig. 8. Additional loop and termini interactions may staeukaryotic SOD structures from the seven species listed inB-factors distinguishes more flexible, and therefore potentiallymore rigid regions (dark blue, low B-factor). The most flexdesignated top, functional, and terminal. (b–d) Residues in thstabilizing features in ApSOD that are underrepresented in othThe zinc-binding region is colored the same as the zinc ion in (variable loop (VL) is pink in (d).


guanidine titration. ApSOD proved exceedinglystable (Fig. 7). The unfolding midpoint of ∼5.1 Mguanidine is nearly 1 M higher in concentration thanthe ∼4.25 M unfolding midpoint of HsSOD.To identify structural traits that may confer added

stability,we compared theApSOD structurewith thehighest-resolution wild-type SOD structures fromhuman and five other eukaryotic species.20–23,26,27

As the overall number of residues (Fig. 2a) and fold(Fig. 8a) are well conserved among these SODs, theApSOD dimer interface surface area and hydrophobicpacking within the β-barrel were typical. We super-imposed the seven structures and color-coded thembycrystallographic B-factor to identify regions that aremore flexible and likely less stable (Fig. 8a). The outerturns and loops had the most flexibility, consistentwith results from molecular dynamic simulations53

andNMR.54 Conserved side-chain interactions among10 crystallographically independent subunits in theHsSOD crystal structure were proposed to stabilizesuch protruding and flexible structural elements.24

Comprehensive inspection and analysis of the super-imposed structures revealed that ApSOD has extraelectrostatic interactions compared to other SODs (Fig.8b–d and Table 2), including the most salt bridges

bilize ApSOD native fold and assembly. (a) Overlay ofFig. 2a. Coloring in a spectrum according to relative Cα

less stable, parts of the molecule (red, high B-factor) fromible parts are loops and turns localized in three areas,e three most flexible areas of the SOD structure that former structures (see Table 2). Subunits are colored separately.c) to distinguish it from the electrostatic loop (EL), and the


(Supplementary Table S1), consistentwith results froma large-scale comparison of thermophilic proteinstructures from T. maritima with their mesophilichomologs.5

In the “top” region of the β-barrel (Fig. 8a andb), the β1/β2 turn is secured and stabilizes thedimer interface by a charged hydrogen bondbetween Lys9 Nζ and Asn51 Oδ1 from the adjacentsubunit. β1/β2 Ser12 also hydrogen-bonds β8Leu142. The hydrogen bond between the β2Thr15 and β3 Thr34 side chains helps to anchorthe β1/β2 and GK1 loops and may also constrainGly35 to hydrogen-bond its N atom to the β5/β6turn. Within the “functional” region (Fig. 8a and c),which is composed of the Zn-binding and ELs,more interloop surface interactions join Glu68 andAsn76, Asp74 and Arg126, and Gly125 and Lys133.In the “terminal” region (Fig. 8a and d), Ile2 Nhydrogen-bonds to β2 Glu22 Oɛ2 across the openend of the N-terminal β-hairpin and also helpstether the adjacent β2/β3 variable loop, whereonly ApSOD and BtSOD share the short type II′turn. The C-termini are stabilized by symmetrichydrogen bonds between Thr150 Oγ1 to Asp50 Oδ1

across the dimer interface, next to the symmetricLys9 Nζ and Asn51 Oδ1 hydrogen bonds, describedabove. The majority of these anchoring interactionsfound within the β-barrel involve residues at the β-strand/turn interface, forming β–anchor–turn(βAt) and turn–anchor–β (tAβ) motifs. Altogether,14 ApSOD residues contribute to the abovehydrogen-bonding and electrostatic interactions.Of these only Asp50 and Asp74 are fully conservedamong the seven species. The unusually stablebovine SOD conserves half of these tetheringinteractions (Fig. 2a and Table 2).

Table 2. Stabilizing ApSOD features

Region ApSOD direct interaction Observed

Top1 Lys9 Nζb–Asn51 Oδ1 Bt2 Ser12 Oζ–Leu142 O —3 Thr15 Oζ1–Thr34 Oζ1 —Functional4 Glu68 N–Asn76 Oδ1 Bte, Xle

4 Glu68 Oɛ2–Asn76 Nδ2 Bte, Hse

5 Asp74 Oδ1–Arg126 NH1,2 Bt, Sm, Scf

6 Gly125 O–Lys133 Nζ SmTerminal7 Ile2 N–Glu22 Oɛ2 Sc,g So,g Xlg

8 Asp50 Oδ1–Thr150 Oγ1b Sc, So

ApSOD proline caps Conserved

Pro13 Bt, HsPro38 Sc, SoPro107 Sc, So, Xl

Interactions are numbered and match those in Fig. 2a; weaker water-a Refers to whether the first or second residue in the interaction lieb Spans dimer interface.c One subunit bonds adjacent to Asn14 N, this interaction is prevend Comparable; however, short ApSOD Thr–Thr side chains ensuree Comparable interaction is formed by alternate residues.f Comparable; however, interaction is limited to one hydrogen bong Comparable main chain–main chain interaction.h Refers to whether the proline lies within an ALS mutation site in

Each end of the β-barrel (Fig. 8b and d andTable 2) additionally contains “proline caps”(Pro13, Pro38, and Pro107) that may furtherpromote protein framework stability, defined asmain-chain and Cβ atom positions, by disfavoringflexibility and forming additional C–H…O bonds55between the prolines and intra- and interloopresidues. Of these turn–Pro–turn (tPt) anchoringproline cap motifs, Pro13 is conserved only withinthe mammalian proteins; whereas Pro38 andPro107 are conserved within spinach and yeast,and Pro107 also in frog. There is the potential totack down the β3/β4 GK1 loop, which containsPro38 in ApSOD, in the bovine, human, yeast, andfrog SODs by electrostatic interactions, exemplifiedby BtSOD's β3/β4 Glu38 and β5/β6 Lys89charged side-chain pair. BtSOD also has uniquePro100 and Pro121 caps within β6/β7 GK2 andthe β7/β8 EL.We examined the relationship of the ApSOD

residues discussed above with the human ALSmutation positions. Only 3 of 14 side chains involvedin hydrogen-bonding or electrostatic interactionsand 1 of 3 of the proline caps lie in ALS mutationsites; none represent actual ALS mutations. How-ever, ApSOD Lys21 and Lys98 represent the E21Kand E100K ALS mutations, respectively (Fig. 2a).These lysines lie at the bottom of the β-barrel, facingoutward, between the terminal region and the gapbetween the β-barrel and the zinc-binding subloop(Fig. 9a). In HsSOD, a charge complementarynetwork composed of Glu21-Lys30-Glu100 linksthe three adjacent β-strands. Therefore, the E21Kand E100K ALS mutations could destabilize HsSODby charge repulsion or create unwanted interactionswith negatively charged side chains on mutant

Potential Not possible ALS sitea

Hs Sc, Sm, So, Xl N, NSoc Bt, Hs, Sc, Sm, Xl Y, Y

Bt,d Hs,d Sm,d, So,d Xld Sc N, N

Hse Sc, Sm, So N, NSoe, Xle Sc, Sm N, N

Hsf, Sof, Xlf – Y, N– Bt, Hs, Sc, So, Xl N, N

— Bt, Hs, Sm N, YXlb Bt, Hs, Sm N, N

Not conserved ALS siteh

Sc, Sm, So, Xl NBt, Hs, Sm, Xl YBt, Hs, Sm N

mediated interactions are omitted.s within an ALS mutation site in HsSOD, respectively.

ted in ApSOD by sequestration of a N atom in Pro13 ring.rigidity.

d.

HsSOD.

Fig. 9. Implications for ALS. (a) Lys21 and Lys98 inApSOD represent charge reversal ALS mutations E21Kand E100K in humans, but are likely not destabilizing inApSOD due to charge neutralization at the structurallyadjacent and intervening position Thr28 (human Lys30).HsSOD is labeled and colored in orange. (b) A stackingarrangement bridges the C-terminus and Greek key loop 2across the ApSODdimer interface and continues across thesubunit to the zinc-binding region. In ApSOD, uncon-served His65 at the end of the stack distal to the dimerinterface makes a charged hydrogen or salt bridge bondwith unconserved Asp108. The two subunits are coloredyellow and teal, and loops within the yellow subunit arecolored according to Fig. 2a. A dot surface aids identifica-tion of the stacking elements. Dimer symmetry materesidues are labeled with “-B.” (c) The stacking arrange-ment in HsSOD. The two subunits are colored orange andlight brown. Side chains involved in ALS mutations areshown in green on one subunit and magenta on the other.The mutation sites lie at positions that bridge the dimerinterface and sandwich the stack together.


HsSODs within proximity. In ApSOD, a reversal ofcharge complementarity provides stability throughLys21-Thr28-Lys98.Closing the “bottom” edge of ApSOD is a striking,

extended hydrophobic stacking arrangement thatspans from the dimer interface to the oppositesubunit edge. Beginning where C-terminal Ile149stacks with adjacent subunit GK2 Arg113 andcontinuing through Phe62, Pro64, and His65 of theZnBR of β4/β5 (Fig. 9b), five side chains showfavorable ∼4 Å van der Waals stacking. At theoutside end, His65 forms an electrostatic interactionwith GK2 Asp108. Only parts of these strategicallyplaced stabilizing interactions occur in other SODs.For example, yeast SOD has the triple-ring portion,and frog SOD has a terminal electrostatic interac-tion. In HsSOD, Leu67 at the outside edge of thestack region (Fig. 9c) precludes an electrostaticinteraction, but loosely stacks between Pro66 andArg69 to support the Arg69 to ZnBR Glu77 saltbridge. Thus, HsSOD has weaker barrel edgeinteractions than those seen in ApSOD, suggestingthat this edge is less restrained by favorableinteractions. The most structurally important stabi-lizing interactions along this barrel edge representALS mutation sites: L67R, R115G, and I151T (Fig.2a). As C-terminal Ile151 and its contact partnerArg115 bridge the dimer interface and Leu67completes the end of the stack, the I151T andR115G mutations suggest that destabilization ofterminal interactions and the dimer interface mayplay a role in ALS by reducing native SOD fold andassembly stability and hence lowering the energeticbarrier to forming interactions leading to potentiallytoxic amyloid-like fibers and soluble aggregates.

Discussion

Here we test and extend promising studies of thethermostability56–58 and assembly59 of proteins fromA. pompejana, which inhabits one of the mostphysicochemically demanding environments onearth (Fig. 1). Before any macromolecule can becharacterized biochemically, biophysically, or struc-turally, it must be sufficiently stable to allowanalyses. Many planned studies on particularhuman proteins of medical relevance end at thestage of protein expression and purification andare replaced with characterization of homologousproteins to gain important insights into biologicalfunction. In other cases, where human protein isisolated, structural analyses may be limited byresolution, local disorder, or inability to trapcertain conformations due to instability of thesamples. Thus, understanding of human proteinshas been substantially advanced by characteriza-tions of their microbial homologs, and increasedmacromolecular stability plays a key role in thesesuccessful applications.3–10,12–15The discovery of ApSOD, its stability, dimeric

assembly, and close sequence and structural simila-rities to mammalian SODs support the great value of


A. pompejana as a thermophilic eukaryotic macro-molecule resource. We determined and presentherein the high-resolution structures (Fig. 2 andTable 1), dimer assembly in solution (Fig. 6), andhigh stability (Fig. 7) of ApSOD. These results fromthe A. pompejana enzyme are relevant to thechemistry, biophysics, and physiology of its humanSOD homolog and to the destabilization of HsSODby ALS mutations. Prior to this first structuralanalysis test case of a thermophilic eukaryoteprotein, more than 70 SOD structures have beensolved from a variety of different species. However,since we solved the first SOD structure over 25 yearsago and predicted its substrate/product site,25,47 astructure of SOD in complex with its substrate orproduct had been elusive. This highly stable eukar-yotic SOD, which preserved the human SOD dimerassembly, has provided active-site informationabsent from all previous SOD structures, suggestingA. pompejana may be a general resource for stableeukaryotic macromolecules and their assemblies.The biological importance of SOD catalysis and

function as a master regulator of oxygen radicals ineukaryotic cells, the causative role of HsSODmutants in ALS, and the many existing SODstructures, including a recent HsSOD structure at1.07 Å resolution,27 were key reasons to chooseApSOD for these comparative analyses. These newhigh-resolution ApSOD and ApSOD–H2O2 struc-tures establish conserved features, strategicallypositioned water molecules, and a framework forcomparative analysis of active-site stereochemistry.The ApSOD active-site structures (Fig. 3 andSupplementary Fig. S4) reveal three key points: (1)the O2

·−–H2O2 binding site is likely fixed, (2) theprotein movements during catalysis are minor, and(3) the most significant changes are associated withHis61 and the Cu ion. The high-resolution refine-ment of occupancies for the active-site constituentsand calculation of anisotropic thermal ellipsoidsdelineate these Cu ion and His61 ligand movements(Fig. 5). Together, our structures and analysessupport an updated and unified catalyticmechanism(Fig. 4). In contrast to Cu(I) outer-sphere electrontransfer to the O2

·− anion,48 our new structures andmechanism suggest that Cu(I) moves toward O2

·−

along a structurally facilitated trajectory for inner-sphere binding and orbital overlap (SupplementaryFig. S5). This inner-sphere mechanism is alsoconsistent with prior computational analyses bydensity functional theory and electrostatic modelingof the energetics of SOD throughout its redox cycle,including the energetics of the coupled proton/electron transfer.60 Our new testable model couplesCu(I)'s electron transfer to O2

·− with proton transferfrom His61 and formation of the His61–Cu bond.In fact, ApSOD active-site water molecules speci-

fically suggest that O2·− protonation in the second

half reaction is done in trans to avoid stericcrowding, distinct from the most recently publishedproposal (Fig. 4).48 Furthermore, water moleculesW2 and W3 are positioned to stabilize the newlyformed H2O2 product, analogously to hydrogen-

bond recognition of H2O2 by human catalase.61 Ashuman mitochondrial Mn SOD has a similar net-work of active-site water molecules, its specific rolescan now be tested in the light of our proposedmechanism for the Cu,Zn enzymes.62 Thus, theseApSOD structures open the door to comparativestudies relevant to all SODs, including the newlydiscovered Ni SOD.63

Of inherited ALS cases, N20% involve mutationsto the SOD1 gene, and SOD1 mutations have alsobeen detected in 12% of sporadic ALS cases.41,42

Destabilization of HsSOD protein to form amyloid-like aggregates with aberrant activity is hypothe-sized to cause the disease.36,38,39 Our comparativestructural analyses of eukaryotic SODs identifiedresidues that may contribute to the enhancedstability of ApSOD over HsSOD and also allowedus to test and extend the idea24 that electrostaticinteractions anchoring structural elements promoteprotein stability by maintaining the native fold andassembly. In ApSOD, specific hydrogen bonds andsalt bridges secure the ends of β-strands and loops,and strategically placed tPt “proline caps” rigidifythe otherwise flexible loops55 linking β-strands(Figs. 8 and 9). Significantly, the large functionalloops that contribute to substrate attraction andcatalysis also exhibit conspicuous stabilizing elec-trostatic and stacking interactions and hydrogenbonds in ApSOD. In the β-barrel core, ApSOD hassome shorter hydrophobic side chains, yet retainsclose packing, including the Phe43 anchor buriedadjacent to Cu ligand His44 to stabilize the activesite. Most residues identified as contributing to theenhanced ApSOD stability lie in sequence-variablepositions that do not disrupt the Greek key fold orcontribute to ALS in humans (Fig. 2a and Table 2).Moreover, the majority of the electrostatic anchoringinteractions involved in the Greek key fold are at theends of β-strands forming βAt and tAβ motifs.Destabilizing mutations occur in both Cu,Zn andMn human SODs; yet, those in Mn SOD are notassociated with amyloid-like aggregation or neuro-degenerative disease.64 Destabilizing ALS muta-tions in human Cu,Zn SOD appear to work byreducing β-barrel or dimer interface stability.32,38

Anchoring interactions at ends of α-helices wereidentified as important for their folding.65 OurApSOD results suggest analogous anchoring interac-tions at the ends ofβ-strands and loops stabilize theβ-structure. Based on comparisons of these anchoringinteraction motifs and sites with those mutated inSOD ALS patients, such anchoring interactions mayfurthermore avoid the formation of the inappropriatebeta-structural interactions and amyloid-like fila-ments associated with many neurodegenerative dis-eases including ALS. More generally, these ApSODanalyses suggest that functional protein stability,relevant to preservation of native fold and assemblyin vivo, can be substantially improved (or degraded)by substitutions or mutations that increase (ordecrease) the rigidity of the weakest or most flexibleparts of the structure, without altering core fold orassembly. Such “weakest links” generated through


mutations in loops and termini may induce localunfolding and framework destabilization that triggernucleation of amyloid-like filaments to form aggre-gates (Fig. 10). These structural analyses combinedwith the identification of other A. pompejana proteinsthat show thermostability traits56–58 suggest that thereis likely a structural basis, which includes extra salt-bridged interactions similar to those identified in T.maritima proteins,5 for enhanced thermostability ofother A. pompejana proteins, and that these proteinsmay aid insights into structure–function relationshipsin other systems.Overall, we showed here that ApSOD is more

stable than HsSOD biophysically (Fig. 7) andidentified interactions expected to enhance struc-tural stability (Figs. 8 and 9). Such stable SOD formshave been sought for biotechnology and medicalapplications for which ApSOD may be tested.66 Wefurthermore showed that the human-like ApSODdimer (Fig. 6) was able to trap a previouslyunobtainable H2O2 complex (Fig. 3), which extendsour detailed understanding of the SOD reaction andsuggests a unified general mechanism for catalysis(Fig. 4). Notably, this proposal resolves apparentparadoxes regarding the electron transfer despiterestrictive active-site geometry that would seemingly

Fig. 10. Implications for amyloid-like filament nucleationnucleation of amyloid-like filaments and formation of solubleover time both (a) ApSOD and (b) HsSOD are likely subjectedopen structures to expose regions normally sequestered by fowithin loop and termini regions may cause toxicity by shiftingto nucleation and growth on nonnative, amyloid-like fiber forthe increased stability of ApSOD, which has additional contacelectrostatic and hydrogen-bonding interactions; gray, N-terterminal and cross-dimer interaction; blue, proline cap; see Figgive ApSOD its higher stability to HsSOD. The interactions shmay enhance local perturbations by removing these and otherhindrance of normally favorable interactions and packing, res

block direct binding of O2·− to Cu(I). These initial

results support further analyses of the Pompeiiworm's value as a thermophilic eukaryote bylaboratories worldwide. For biochemists, biophysi-cists, and structural biologists, the Pompeii wormmay provide stable full-length eukaryotic macromo-lecules, and for X-ray crystallography, the additionalresolution to provide new insights into particularsystems of interest. Indeed, A. pompejana's ability tothrive in its harsh environment suggests that itsDNA, RNA, and proteins will be key resourcesthat may prove broadly useful in biotechnology,industry, and biomedical analyses and should betested accordingly.

Materials and Methods

Sample collection and cDNA sequencing

Using robotic arms housed on the DSV Alvin, wecollected A. pompejana worms from hydrothermal ventsites (9°N, 50/104°W17) at a depth of ∼2500 m (Fig. 1).Temperature measurements were made with a narrow∼0.3 -cm-diameter temperature probe. Worms wereplaced in an insulated device containing RNAlater (gift

and aggregate formation. (a and b) A unified model foraggregates. Due to flexibility in loop and termini regions,to reversible local structural perturbations that transientlyld and assembly interactions. ALS SOD mutants that lieequilibria towardmore open forms with decreased barriersming contacts. Support for this hypothesis comes from (a)ts in these regions, shown as spheres (purple, intrasubunitminal interaction; green, cross-dimer interaction; red, C-. 8 and Table 2) that tether these elements and thereby mayared in HsSOD are also shown as spheres. ALS mutationskey contacts, by introducing charge repulsion, or by stericulting in nonnative amyloid-like-promoting interactions.


from Ambion) for in situ stabilization of RNA. Wormswere then frozen and transported to the laboratory in dryice and then stored at −80 °C. Episymbiotic bacteria werelater removed from the dorsal side of the worms byscraping with forceps. A single worm was dissected, andIncyte Genomics Corporation generated cDNA librariesfrom the worm's posterior end, which experiences highertemperatures. The cDNAwere ligated into the pBluescriptKS+ vector. Approximately 5300 sequences were read(Genome Therapeutics). The A. pompejana SOD gene wasidentified by a tblastn search using the HsSOD proteinsequence as a query.

Sequence analysis

Using default blastp settings, we conducted a searchusing the predicted ApSOD protein query sequence on theNational Center for Biotechnology Information database,which yielded the Crassostrea gigas (pacific oyster), Bostaurus (cattle, predicted), Oryza sativa (rice, japonicacultivar group), Pinus pinaster (maritime pine), Ixodesscapularis (black-legged tick), and Bos grunniens (domesticyak) Cu,Zn SOD proteins as the top five matches. Thesehits indicated that the A. pompejana sequence wasconserved with other eukaryotic SODs, including thosefrom mammals. The sequences and tertiary folds of thehighest-resolution SOD proteins representing all eukar-yotic species with structures deposited at the ResearchCollaboratory for Structural Bioinformatics—B. taurus(cattle), H. sapiens (human), S. mansoni (trematode), X.laevis (frog), S. oleracea (spinach), and S. cerevisiae (buddingyeast)—were aligned and superimposed individuallywith ApSOD using SEQUOIA. A CLUSTALW sequencealignment was then manually edited by using thestructural overlays as guides. A consensus phylogenetictree (Fig. 2b) was generated with neighbor-joining and 100iterations of bootstrapping.

Cloning, expression, and purification

A cDNA library for cloning was generated on board theR/VAtlantis. Episymbiotic bacteriawere removedbybrieflyfreezing a single worm and then scraping with forceps. Theworm was then frozen and crushed with a mortar andpestle. PolyARNAwere isolated byusing thePolyAPurekit(Ambion). cDNAwere synthesized from PolyA RNA usingthe Marathon cDNA amplification kit (Clontech). PCRprimers 5′-GATAGG CCATCC AGG CCG TTT GCG TCCTGA AGG GA-3′ and 5′-GCG CGG TCG ACT TAC TCCTTT GTA ATA CCA ATG ACA CCA CA-3′ were designedfrom the ApSOD processed mRNA transcript sequenceidentified by cDNA sequencing and used to amplify thecoding region, with the exception of the most N-terminalresidues,whichwere substitutedwith an Sfi1 restriction site.The PCR primer for the C-terminal end contained a flankingSal1 restriction site. FollowingPCRamplification of the genefrom the cDNA library, thePCRproductwas ligated into thePCR2.1 TOPO vector (Invitrogen). The ApSOD gene wasremoved from the PCR2.1 TOPO vector and the HsSODgene was removed from the pPHSODC6AC111SlacIqR139expression vectorwith SfiI and SalI restriction enzymes. Theliberated ApSOD gene was then ligated into the linearizedexpression vector. The SfiI site was then mutagenized toyield the correct N-terminal sequence for ApSOD. Thevector contains a leader sequence that targets expressedprotein to the periplasm, where the leader is then removed.ApSOD protein was expressed and purified by follow-

ing previously established protocols for other SOD

proteins in our laboratory. Transformed SOD−/− Escher-ichia coli cells were cultured in Luria broth supplementedwith 100 μg/ml ampicillin at 37 °C. When cells reached anoptical density of 0.8 at 600 nm, 0.25 mM CuSO4 wasadded, and 0.4 mM IPTG was used to induce proteinsynthesis. Cells were cultured another 6–12 h beforeharvesting and resuspension in 100 ml of ice-cold Tris–Clbuffer (pH 7.5). Ice-cold 40% (w/v) sucrose (100 ml) and0.5M ethylenediaminetetraacetic acid (30 ml) were used toinduce osmotic shock. The suspension was rocked for30 min at 4 °C, then subjected to centrifugation. Thesupernatant was removed. The pellet was resuspended in200 ml H2O and centrifuged at 4 °C. The two supernatantswere combined and proteins were precipitated by theaddition of ammonium sulfate to 65% (w/v). Thissuspension was centrifuged and the pellet was resus-pended in 50 ml H2O. The protein solution was thendialyzed twice against 4 l of H2O and centrifuged. CuSO4was added to the supernatant at a final concentration of1 mM. The solution was heated to 65 °C for 30–90 min andcentrifuged. The supernatant was dialyzed against 4 l of100 mMMes (pH 5.5) overnight. Following another roundof centrifugation, the supernatant was applied to a POROSHS column and eluted with a NaCl gradient. ApSOD-containing fractions were pooled, dialyzed against 25 mMTris–Cl (pH 8.0), and applied to a POROS or HiTrap HQcolumn from which ApSOD protein was eluted with aNaCl gradient. Pooled fractions were concentrated to 2 mland applied to a Superdex 75 16/60 gel-filtration column.All column media were purchased from Pharmacia.ApSOD protein was metallated,27,39,67,68 following

established protocols,39,67 by dialyzing against equalmolar ZnSO4 in 10 mM sodium acetate (pH 5.5) buffer.Equalmolar CuSO4was then added to the dialysis solutionand allowed to equilibrate. Protein was then dialyzed inH2O and concentrated to 24–31 mg/ml. The color of theprotein solution was blue, indicating that the Cu ions wereincorporated and oxidized. Protein samples migrated as asingle band on nondenaturing polyacrylamide gels,indicative of a homogeneous protein species (Supplemen-tary Fig. S1). Purification of SODs directly from somesource organisms show that the N-terminal Met is oftenposttranslationally cleaved and the penultimate residueacetylated.44 Mass spectrometry analysis and N-terminalsequencing were used to verify ApSOD composition andshowed that the protein's amino acid sequence starts withthe second-position alanine, which was not acetylated.SOD activity was monitored with a polyacrylamide-gel-

based assay (Supplementary Fig. S1).39,40,69 Proteinsamples (10 μg) were migrated into a nondenaturing 4–20% Tris–glycine gel in native running buffer [50 mM Tris,380 mM glycine (pH 8.5)] by electrophoresis. DNAdouble-strand break repair protein Mre118 served as anegative control, and HsSOD (prepared as described70) asa positive control. The gel was soaked in Stain I buffer(0.2% nitroblue tetrazolium) and shaken for 20 min in thedark at room temperature. After being rinsed briefly withwater, the gel was soaked in Stain II buffer [0.0042% N,N,N′,N′-tetramethylethylenediamine (TEMED), 0.03 mMriboflavin, 360 mM NaKHPO4 (pH 7.8)] and shaken inthe dark at room temperature for 20 min. The gel wasdeveloped by illumination on a light box for 10 min andthen soaked in 10% (v/v) acetic acid for 5 min to stopdevelopment. Activity is indicated on the gel by bleaching.

X-ray crystal structure determination

ApSOD crystals were grown using vapor diffusion bymixing 1.5 μl of 24 -mg/ml protein solution with 1.5 μl


well solution from an ammonium sulfate grid screen. X-ray diffraction data were collected at Stanford Synchro-tron Radiation Laboratory (SSRL) beamline 11-1 fromcrystals that grew in 55% saturated ammonium sulfate,100 mM sodium citrate buffer (pH 5.5). These crystals,grown in initial unrefined conditions, were cryoprotectedin well solution supplemented with 20% (v/v) glycerol. Inan effort to collect all of the high-resolution data aftercollecting low-resolution data, the detector was translatedlaterally, placing the X-ray beam center in one corner.However, the combination of beamline physics with thecrystal's unit cell constants limited the resolution of thecombined data set to 1.03 Å resolution. Diffraction datawere processed with the use of the HKL2000 suite71 andTRUNCATE.72 Phases for this initial 1.03 Å data set werecalculated by molecular replacement using AMORE72

with HsSOD (PDB code 1PU039) as a search model. AnApSOD model was then generated and refined byiterative rounds of manual fitting with XFIT73 andrefinement initially with CNS74 and σ-a weighted terms,and in later rounds by SHELXL.75

Low- and high-resolution data sets with a combinedrange of 50–0.99 Å resolution were later collected from asingle crystal at the Advanced Light Source SIBYLSbeamline 12.3.1 using 2θ detector rotations (Supplemen-tary Fig. S2). The crystal was grown under the originalconditions, but cryoprotected by dehydration. Modifica-tion of the original crystal conditions or alternativecryoprotectants did not significantly alter the resolution.To dehydrate the crystal, additional saturated ammoniumsulfate was added to the crystallization wells to a final 90%(v/v) concentration and the drop was allowed toequilibrate for 30 min. The initial 1.03 Å model was thencombined with the 0.99 Å data set to generate new mapsfollowed by iterative rounds of model building andrefinement using SHELXL.For the H2O2 complex, a single crystal grown as

described above was transferred from its mother liquor toa 3 -μl drop solution containing 90% saturated ammoniumsulfate, 100 mM sodium citrate (pH 5.5), and 1 mM H2O2.Under the acidic conditions used, H2O2 with a pKa N11 isexpected to remain protonated, decreasing peroxidativereactions.76 The crystal was soaked over a 90% saturatedammonium sulfate well solution for 30min and cryocooled.Diffraction data were again collected by using a 2θ detectorrotation at the SIBYLS beamline 12.3.1 and were processedas described above, resulting in a 1.35 Å data set. The initial1.03 Å model was fit to data for the H2O2 complex by usingCNS rigid-body refinement, followed by cycles of modelbuilding and refinement, as described above.For both models, alternative main-chain and side-chain

conformations and lower-occupancy metal and solventsites found in Fo−Fc difference mapswere verified by CNS-generated simulated annealing and composite-omit maps.In later rounds of refinement, manual estimations and free-occupancy variables were used to refine the occupancies ofthese atoms. In the last round of refinement, theiroccupancies were fixed, allowing B-factors to converge.These occupancy estimates were checked using R-factorsand a variety of electron and difference density maps. R-free calculations77 were performed by removing 5% of thedata for the test set. The stereochemistry for each structurewas validated using MolProbity.78 Standard errors forbond lengths and angles for both individual structureswere calculated using SHELXL least-squares refinement.The number of salt-bridges in the ApSOD and othereukaryotic SOD structures was calculated with WHATIF.79

Bonds less than 4 Å5 and protein atoms involved in directbonding to metals were not used in calculating the totals.

For the 0.99 Å resolution ApSOD structure, the clearelectron density allowed modeling of residues 1–151 froma total of 152 residues, including nine alternative residueconformations, the Cu ion at two positions and one Zn ion.The model also includes two acetate, one sodium, and sixsulfate ions, and 354 ordered water molecules. For the1.35 Å ApSOD–H2O2 complex structure, residues 1–151were also modeled, with eight alternative residue con-formations, the Cu ion at two positions and one Zn ion.This model also includes a hydrogen peroxide molecule,one sodium and four sulfate ions, and 295 ordered watermolecules. As expected for the mild experimental condi-tions employed, we did not detect any proteinmodifications80,81 that occur when H2O2 is converted tomore reactive species.76,82,83

Small-angle X-ray scattering

SAXS data sets were collected on 15 -μl ApSOD proteinsamples at 2, 4, and 6 mg/ml concentrations in phosphate-buffered saline (PBS, pH 7.4) at 25 °C at the Advanced LightSource SIBYLS beamline 12.3.1 using a MAR CCD 165detector as detailed elsewhere.52 The X-ray beamwas tunedto 13 keV with an incident flux of approximately 1011

photons/s. The sample-to-detector distance was 1.46 m.The detector was centered normal to the primary X-raybeam, and data were collected with a scattering vector Q(Q=4π sin θ/λ, where 2θ is the scattering angle and λ is thewavelength) spacing range of 0.01 to 0.35 Å−1. Shortexposures of 10 s and long exposures of 100 s were taken foreach sample, flanked by corresponding blank buffersamples. Short and long data sets were merged withPRIMUS84 using the 10 -s exposures for the low-Q regionand the 100 -s exposures for the high-Q region. Both blankbuffer data for each experiment were subtracted from theprotein-scattering data and compared to ensure propernormalization due to flux variations. A concentrationgradient is required to ascertain whether scattering fromcomplex–complex interactions or small amounts of aggre-gation contribute to the overall scattering profile. Thesecontributions must be eliminated in order to furtheranalyze the scattering from the target molecule. There-fore, wider angle data out to Q=0.64 Å−1 were collectedonApSODproteinat a concentrationof 31mg/mlbyverticallymoving the detector 6 cm perpendicular to the beam.PRIMUS was used to calculate the radius of gyration

(RG) using the Gunier approximation and the low -resolution (QRGb1.3) data. Using the concentration beha-vior of the sample, we extrapolated the scattering profile toan effective 0 concentration by scaling the superimposabledata (Qb0.1 Å−1) and linearly fitting the intensity at eachvalue of Qb0.1 as a function of concentration.52 Themeasured RG from the 0 concentration scattering curvewas 20.3 Å (Supplementary Fig. S6). The high concentra-tions showed long-length interactions between indepen-dent dimers, observable in the low-angle region. As therepulsion solely affects the scattering at low Q values(Qb0.1), higher concentration data sets were merged withthe 0 concentration result. Thus, the data presented in Fig. 6are a composite of 2 -mg/ml data used in the low-angleregion (up toQ=0.1 Å−1) merged with the data collected atthe wider-angle setting (6mg/ml results were used for data0.15bQb0.25 and 31 mg/ml at 0.2bQb0.6).The P(r) functions and subsequent maximum interelec-

tron distances Dmax were calculated with GNOM.85

Calculated curves based on the human39 and A.pleuropneumoniae28 SODs and the 0.99 Å ApSOD X-raycrystal structure were generated with CRYSOL.86 HsSOD


chain A from 1PU0 was used to represent the singlesubunit of HsSOD. For the ApSOD structure, the lower-occupancy alternative atoms were removed and theoccupancies of the remaining atoms were set to 1.00prior to calculations. Ab initio 3D image reconstructionswere generated with GASBOR.87

Denaturation assays

Protein unfolding was monitored as previouslydescribed.39 ApSOD and HsSOD proteins were mixedwith increasing amounts of guanidine HCl (0 to 7 M in0.5 M steps) in PBS buffer (pH 7.4). The final concentrationfor each protein was 20 μM. Following incubation, theextents of protein unfolding were assessed by CDspectroscopy. Samples were loaded into a 0.1 -cm path-length quartz cell and the ellipticities were measured at218 nm with an AVIV model 2o2SF stopped-flow CDspectrometer in kinetics mode. Multiple measurementswere averaged to obtain final values. To convert raw datato fraction unfolded (Fu), straight lines were fit through theupper and lower plateau regions of the plots. Thefollowing formula was then used to normalize the data:

FU = YN � Yð Þ= YN � YDð ÞUN and UD represent the fits of the lines through the

upper and lower plateau regions, respectively, and U is theellipticity value at a given guanidine concentration. Datawere plotted as fraction unfolded versus guanidine concen-tration. A detailed analysis of the free-energy changes wereprecluded, as it was shown that wild-type SOD folding andunfolding are not fully reversible due to metal binding39

and illegitimate disulfide bond formation that producesaggregation.88 For accurate interspecies comparisons, weused wild-type protein rather than the less aggregationprone HsSOD Cys mutants used for in vitro studies.89

Accession numbers

The ApSOD mRNA sequence has been deposited inGenBank with accession number EU178106. Coordinatesand structure factors have been deposited in the PDB withaccession numbers 3F7L for the ApSOD X-ray crystalstructure and 3F7K for the ApSOD–H2O2 structure.

Acknowledgements

We thank B. R. Chapados, D. S. Daniels, J. A. Fee,K. Henscheid, K. Hitomi, and L. Noodleman forhelpful discussions and technical support. We thankthe staffs of SSRL beamline 11-1 and AdvancedLight Source beamline 12.3.1 for diffraction facilities.We thank the R/V Atlantis and DSV Alvin crews inaiding A. pompejana sample collection. This workwas supported by the 3rd Annual Incyte DiscoveryAward (D.S.S., S.C.C., J.A.T.), National Institutes ofHealth R01 GM037684 (E.D.G.), National SciencesFoundation LExEn NSF-9907666 (S.C.C.), Biocom-plexity OCE-0120648 (S.C.C.), and Department ofEnergy program Integrated Diffraction AnalysisTechnologies (J.A.T.). D.S.S. was supported in partby Skaggs Institute for Chemical Biology and RuthL. Kirschstein NSRA Fellowships.

Supplementary Data

Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.jmb.2008.11.031

References

1. Egorova, K. & Antranikian, G. (2005). Industrialrelevance of thermophilic Archaea. Curr. Opin. Micro-biol. 8, 649–655.

2. Yano, J. K. & Poulos, T. L. (2003). New understandingsof thermostable and peizostable enzymes. Curr. Opin.Biotechnol. 14, 360–365.

3. Fan, L., Williams, R. S., Shin, D. S., Chapados, B. R. &Tainer, J. A. (2008). Master keys to DNA replication,repair, and recombination from the structural biologyof enzymes from thermophiles. InThermophiles: Biologyand Technology at High Temperatures (Robb, F. T.,Antranikian, G., Grogan, D. & Driessen, A., eds), pp.239–263, CRC Press, Boca Raton, FL.

4. DiDonato, M., Deacon, A. M., Klock, H. E., McMullan,D. & Lesley, S. A. (2004). A scaleable and integratedcrystallizationpipelineapplied tomining theThermotogamaritima proteome. J. Struct. Funct. Genomics, 5,133–146.

5. Robinson-Rechavi, M., Alibes, A. & Godzik, A. (2006).Contribution of electrostatic interactions, compactnessand quaternary structure to protein thermostability:lessons from structural genomics of Thermotogamaritima. J. Mol. Biol. 356, 547–557.

6. Gregory, S. T. & Dahlberg, A. E. (2008). Structure andevolution of the Thermus thermophilus ribosome. InThermophiles: Biology and Technology at High Temperatures(Robb, F. T., Antranikian, G., Grogan, D. &Driessen, A.,eds), pp. 291–308, CRC Press, Boca Raton, FL.

7. Chapados, B. R., Hosfield, D. J., Han, S., Qiu, J., Yelent,B., Shen, B. & Tainer, J. A. (2004). Structural basis forFEN-1 substrate specificity and PCNA-mediatedactivation in DNA replication and repair. Cell, 116,39–50.

8. Hopfner, K. P., Karcher, A., Shin, D., Fairley, C., Tainer,J. A. & Carney, J. P. (2000). Mre11 and Rad50 fromPyrococcus furiosus: cloning and biochemical charac-terization reveal an evolutionarily conserved multi-protein machine. J. Bacteriol. 182, 6036–6041.

9. Shin, D. S., Pellegrini, L., Daniels, D. S., Yelent, B.,Craig, L., Bates, D. et al. (2003). Full-length archaealRad51 structure and mutants: mechanisms for RAD51assembly and control by BRCA2. EMBO J. 22,4566–4576.

10. Shin, D. S., Chahwan, C., Huffman, J. L. & Tainer, J. A.(2004). Structure and function of the double-strandbreak repair machinery. DNA Repair (Amsterdam), 3,863–873.

11. Shin, J. H., Kelman, L. M. & Kelman, Z. (2008).DNA replication in thermophiles. In Thermophiles:Biology and Technology at High Temperatures (Robb, F.T., Antranikian, G., Grogan, D. & Driessen, A., eds),pp. 265–277, CRC Press, Boca Raton, FL.

12. Fan, L., Arvai, A. S., Cooper, P. K., Iwai, S., Hanaoka,F. & Tainer, J. A. (2006). Conserved XPB core structureand motifs for DNA unwinding: implications forpathway selection of transcription or excision repair.Mol. Cell, 22, 27–37.

13. Fan, L., Fuss, J. O., Cheng, Q. J., Arvai, A. S., Hammel,M., Roberts, V. A. et al. (2008). XPD helicase structures


and activities: insights into the cancer and agingphenotypes from XPD mutations. Cell, 133, 789–800.

14. Hopfner, K. P., Karcher, A., Craig, L., Woo, T. T.,Carney, J. P. & Tainer, J. A. (2001). Structuralbiochemistry and interaction architecture of the DNAdouble-strand break repair Mre11 nuclease andRad50-ATPase. Cell, 105, 473–485.

15. Hopfner, K. P., Karcher, A., Shin, D. S., Craig, L.,Arthur, L. M., Carney, J. P. & Tainer, J. A. (2000).Structural biology of Rad50 ATPase: ATP-drivenconformational control in DNA double-strand breakrepair and the ABC-ATPase superfamily. Cell, 101,789–800.

16. Cary, S. C., Shank, T. M. & Stein, J. (1998). Worms baskin extreme temperatures. Nature, 391, 545–546.

17. Le Bris, N. & Gaill, F. (2007). How does the annelidAlvinella pompejana deal with an extreme hydro-thermal environment. Rev. Environ. Sci. Biotechnol. 6,197–221.

18. Girguis, P. R. & Lee, R. W. (2006). Thermal preferenceand tolerance of alvinellids. Science, 312, 231.

19. Di Meo-Savoie, C. A., Luther, G. W. & Cary, S. C.(2004). Physicochemical characterization of the micro-habitat of the epibionts associated with Alvinellapompejana, a hydrothermal vent annelid. Geochim.Cosmochim. Acta, 68, 2055–2066.

20. Cardoso, R. M., Silva, C. H., Ulian de Araujo, A. P.,Tanaka, T., Tanaka, M. & Garratt, R. C. (2004).Structure of the cytosolic Cu,Zn superoxide dismutasefrom Schistosoma mansoni. Acta Crystallogr., Sect. D:Biol. Crystallogr. 60, 1569–1578.

21. Carugo, K. D., Battistoni, A., Carri, M. T., Polticelli, F.,Desideri, A., Rotilio, G. et al. (1996). Three-dimen-sional structure of Xenopus laevis Cu,Zn superoxidedismutase b determined by X-ray crystallography at1.5 Å resolution. Acta Crystallogr., Sect. D: Biol.Crystallogr. 52, 176–188.

22. Kitagawa, Y., Tanaka, N., Hata, Y., Kusunoki, M., Lee,G. P., Katsube, Y. et al. (1991). Three-dimensionalstructure of Cu,Zn–superoxide dismutase from spinachat 2.0 Å resolution. J. Biochem. (Tokyo), 109, 477–485.

23. Ogihara, N. L., Parge, H. E., Hart, P. J., Weiss, M. S.,Goto, J. J., Crane, B. R. et al. (1996). Unusual trigonal–planar copper configuration revealed in the atomicstructure of yeast copper–zinc superoxide dismutase.Biochemistry, 35, 2316–2321.

24. Parge, H. E., Hallewell, R. A. & Tainer, J. A. (1992).Atomic structures of wild-type and thermostablemutant recombinant human Cu,Zn superoxide dis-mutase. Proc. Natl Acad. Sci. USA, 89, 6109–6113.

25. Tainer, J. A., Getzoff, E. D., Beem, K. M., Richardson, J.S. & Richardson, D. C. (1982). Determination andanalysis of the 2 Å-structure of copper, zinc super-oxide dismutase. J. Mol. Biol. 160, 181–217.

26. Hough, M. A. & Hasnain, S. S. (2003). Structure offully reduced bovine copper zinc superoxide dismu-tase at 1.15 Å. Structure, 11, 937–946.

27. Strange, R. W., Antonyuk, S. V., Hough, M. A.,Doucette, P. A., Valentine, J. S. & Hasnain, S. S.(2006). Variable metallation of human superoxidedismutase: atomic resolution crystal structures ofCu–Zn, Zn–Zn and As-isolated wild-type enzymes.J. Mol. Biol. 356, 1152–1162.

28. Forest, K. T., Langford, P. R., Kroll, J. S. & Getzoff, E.D. (2000). Cu,Zn superoxide dismutase structure froma microbial pathogen establishes a class with aconserved dimer interface. J. Mol. Biol. 296, 145–153.

29. Pesce, A., Capasso, C., Battistoni, A., Folcarelli, S.,Rotilio, G., Desideri, A. & Bolognesi, M. (1997).

Unique structural features of the monomeric Cu,Znsuperoxide dismutase from Escherichia coli, revealedby X-ray crystallography. J. Mol. Biol. 274, 408–420.

30. Spagnolo, L., Toro, I., D'Orazio, M., O'Neill, P.,Pedersen, J. Z., Carugo, O. et al. (2004). Uniquefeatures of the sodC-encoded superoxide dismutasefrom Mycobacterium tuberculosis, a fully functionalcopper-containing enzyme lacking zinc in the activesite. J. Biol. Chem. 279, 33447–33455.

31. Bourne, Y., Redford, S. M., Steinman, H. M., Lepock,J. R., Tainer, J. A. & Getzoff, E. D. (1996). Noveldimeric interface and electrostatic recognition inbacterial Cu,Zn superoxide dismutase. Proc. NatlAcad. Sci. USA, 93, 12774–12779.

32. Perry, J. J., Fan, L. & Tainer, J. A. (2007). Developingmaster keys to brain pathology, cancer and aging fromthe structural biology of proteins controlling reactiveoxygen species and DNA repair. Neuroscience, 145,1280–1299.

33. Crane, B. R., Arvai, A. S., Ghosh, D. K., Wu, C.,Getzoff, E. D., Stuehr, D. J. & Tainer, J. A. (1998).Structure of nitric oxide synthase oxygenase dimerwith pterin and substrate. Science, 279, 2121–2126.

34. Turner, S. R., Tainer, J. A. & Lynn, W. S. (1975).Biogenesis of chemotactic molecules by the arachido-nate lipoxygenase system of platelets. Nature, 257,680–681.

35. Lepock, J. R., Frey, H. E. & Hallewell, R. A. (1990).Contribution of conformational stability and reversi-bility of unfolding to the increased thermostability ofhuman and bovine superoxide dismutase mutated atfree cysteines. J. Biol. Chem. 265, 21612–21618.

36. Cardoso, R. M., Thayer, M. M., DiDonato, M., Lo,T. P., Bruns, C. K., Getzoff, E. D. & Tainer, J. A.(2002). Insights into Lou Gehrig's disease from thestructure and instability of the A4V mutant ofhuman Cu,Zn superoxide dismutase. J. Mol. Biol. 324,247–256.

37. Cleveland, D. W. & Rothstein, J. D. (2001). FromCharcot to Lou Gehrig: deciphering selective motorneuron death in ALS. Nat. Rev. Neurosci. 2, 806–819.

38. Deng, H. X., Hentati, A., Tainer, J. A., Iqbal, Z.,Cayabyab, A., Hung, W. Y. et al. (1993). Amyotrophiclateral sclerosis and structural defects in Cu,Znsuperoxide dismutase. Science, 261, 1047–1051.

39. DiDonato, M., Craig, L., Huff, M. E., Thayer, M. M.,Cardoso, R. M., Kassmann, C. J. et al. (2003). ALSmutants of human superoxide dismutase form fibrousaggregates via framework destabilization. J. Mol. Biol.332, 601–615.

40. Rakhit, R. & Chakrabartty, A. (2006). Structure,folding, and misfolding of Cu,Zn superoxide dismu-tase in amyotrophic lateral sclerosis. Biochim. Biophys.Acta, 1762, 1025–1037.

41. Majoor-Krakauer, D., Willems, P. J. & Hofman, A.(2003). Genetic epidemiology of amyotrophic lateralsclerosis. Clin. Genet. 63, 83–101.

42. Rosen,D. R., Siddique, T., Patterson,D., Figlewicz,D.A.,Sapp, P., Hentati, A. et al. (1993). Mutations in Cu/Znsuperoxide dismutase gene are associated with familialamyotrophic lateral sclerosis. Nature, 362, 59–62.

43. Luther, G. W., III, Rozan, T. F., Taillefert, M., Nuzzio,D. B., Di Meo, C., Shank, T. M. et al. (2001). Chemicalspeciation drives hydrothermal vent ecology. Nature,410, 813–816.

44. Getzoff, E. D., Tainer, J. A., Stempien, M. M., Bell, G. I.& Hallewell, R. A. (1989). Evolution of CuZn super-oxide dismutase and the Greek key beta-barrelstructural motif. Proteins, 5, 322–336.


45. Getzoff, E. D., Tainer, J. A., Weiner, P. K., Kollman,P. A., Richardson, J. S. & Richardson, D. C. (1983).Electrostatic recognition between superoxide andcopper, zinc superoxide dismutase. Nature, 306,287–290.

46. Fisher, C. L., Cabelli, D. E., Tainer, J. A., Hallewell,R. A. & Getzoff, E. D. (1994). The role of arginine 143 inthe electrostatics andmechanism of Cu,Zn superoxidedismutase: computational and experimental evalua-tion by mutational analysis. Proteins, 19, 24–34.

47. Tainer, J. A., Getzoff, E. D., Richardson, J. S. &Richardson, D. C. (1983). Structure and mechanismof copper, zinc superoxide dismutase. Nature, 306,284–287.

48. Hart, P. J., Balbirnie, M. M., Ogihara, N. L.,Nersissian, A. M., Weiss, M. S., Valentine, J. S. &Eisenberg, D. (1999). A structure-based mechanismfor copper–zinc superoxide dismutase. Biochemistry,38, 2167–2178.

49. Ferraroni, M., Rypniewski, W. R., Bruni, B., Orioli, P. &Mangani, S. (1998). Crystallographic determination ofreduced bovine superoxide dismutase at pH 5.0 andof anion binding to its active site. J. Biol. Inorg. Chem. 3,411–422.

50. Smirnov, V. V. & Roth, J. P. (2006). Mechanisms ofelectron transfer in catalysis by copper zinc super-oxide dismutase. J. Am. Chem. Soc. 128, 16424–16425.

51. Koutmos, M., Pejchal, R., Bomer, T. M., Matthews,R. G., Smith, J. L. & Ludwig, M. L. (2008). Metalactive site elasticity linked to activation of homo-cysteine in methionine synthases. Proc. Natl Acad.Sci. USA, 105, 3286–3291.

52. Putnam, C. D., Hammel, M., Hura, G. L. & Tainer, J. A.(2007). X-ray solution scattering (SAXS) combinedwith crystallography and computation: defining accu-rate macromolecular structures, conformations andassemblies in solution. Q. Rev. Biophys. 40, 191–285.

53. Khare, S. D. & Dokholyan, N. V. (2006). Commondynamical signatures of familial amyotrophic lateralsclerosis-associated structurally diverse Cu, Zn super-oxide dismutase mutants. Proc. Natl Acad. Sci. USA,103, 3147–3152.

54. Banci, L., Bertini, I., Cramaro, F., Del Conte, R. &Viezzoli, M. S. (2002). The solution structure ofreduced dimeric copper zinc superoxide dismutase.The structural effects of dimerization. Eur. J. Biochem.269, 1905–1915.

55. Bhattacharyya, R. & Chakrabarti, P. (2003). Stereo-specific interactions of proline residues in proteinstructures and complexes. J. Mol. Biol. 331, 925–940.

56. Burjanadze, T. V. (2000). New analysis of thephylogenetic change of collagen thermostability.Biopolymers, 53, 523–528.

57. Henscheid, K. L., Shin, D. S., Cary, S. C. &Berglund, J. A. (2005). The splicing factor U2AF65is functionally conserved in the thermotolerantdeep-sea worm Alvinella pompejana. Biochim. Biophys.Acta, 1727, 197–207.

58. Piccino, P., Viard, F., Sarradin, P. M., Le Bris, N., LeGuen, D. & Jollivet, D. (2004). Thermal selection ofPGM allozymes in newly founded populations of thethermotolerant vent polychaete Alvinella pompejana.Proc. R. Soc. London, Ser. B, 271, 2351–2359.

59. Jouan, L., Marco, S. & Taveau, J. C. (2003). Revisitingthe structure of Alvinella pompejana hemoglobin at 20Åresolution by cryoelectron microscopy. J. Struct. Biol.143, 33–44.

60. Konecny, R., Li, J., Fisher, C. L., Dillet, V., Bashford, D.& Noodleman, L. (1999). CuZn superoxide dismutase

geometry optimization, energetics, and redox poten-tial calculations by density functional and electrostaticmethods. Inorg. Chem. 38, 940–950.

61. Putnam, C. D., Arvai, A. S., Bourne, Y. & Tainer, J. A.(2000). Active and inhibited human catalase struc-tures: ligand and NADPH binding and catalyticmechanism. J. Mol. Biol. 296, 295–309.

62. Guan, Y., Hickey, M. J., Borgstahl, G. E., Hallewell,R. A., Lepock, J. R., O'Connor, D. et al. (1998). Crystalstructure of Y34F mutant human mitochondrialmanganese superoxide dismutase and the functionalrole of tyrosine 34. Biochemistry, 37, 4722–4730.

63. Barondeau, D. P., Kassmann, C. J., Bruns, C. K., Tainer,J. A. & Getzoff, E. D. (2004). Nickel superoxidedismutase structure and mechanism. Biochemistry,43, 8038–8047.

64. Borgstahl, G. E., Parge, H. E., Hickey, M. J., Johnson,M. J., Boissinot, M., Hallewell, R. A. et al. (1996).Human mitochondrial manganese superoxide dismu-tase polymorphic variant Ile58Thr reduces activity bydestabilizing the tetrameric interface. Biochemistry, 35,4287–4297.

65. Richardson, J. S. & Richardson, D. C. (1988). Aminoacid preferences for specific locations at the ends ofalpha helices. Science, 240, 1648–1652.

66. Hallewell, R. A., Laria, I., Tabrizi, A., Carlin, G.,Getzoff, E. D., Tainer, J. A. et al. (1989). Geneticallyengineered polymers of human CuZn superoxidedismutase. Biochemistry and serum half-lives. J. Biol.Chem. 264, 5260–5268.

67. Boissinot, M., Karnas, S., Lepock, J. R., Cabelli, D. E.,Tainer, J. A., Getzoff, E. D. & Hallewell, R. A. (1997).Function of the Greek key connection analysed usingcircular permutants of superoxide dismutase. EMBOJ. 16, 2171–2178.

68. Hayward, L. J., Rodriguez, J. A., Kim, J. W., Tiwari,A., Goto, J. J., Cabelli, D. E. et al. (2002). Decreasedmetallation and activity in subsets of mutantsuperoxide dismutases associated with familialamyotrophic lateral sclerosis. J. Biol. Chem. 277,15923–15931.

69. Crapo, J. D., McCord, J. M. & Fridovich, I. (1978).Preparation and assay of superoxide dismutases.Methods Enzymol. 53, 382–393.

70. Hallewell, R. A., Imlay, K. C., Lee, P., Fong, N. M.,Gallegos, C., Getzoff, E. D. et al. (1991). Thermo-stabilization of recombinant human and bovineCuZn superoxide dismutases by replacement offree cysteines. Biochem. Biophys. Res. Commun. 181,474–480.

71. Otwinowski, Z. & Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. InMethods in Enzymology (Carter, C. W., Jr & Sweets, R.M., eds), vol. 276, pp. 307–326. Academic Press,New York.

72. CCP4. (1994). The CCP4 suite: programs for proteincrystallography. Acta Crystallogr., Sect. D: Biol. Crystal-logr. 50, 760–763.

73. McRee, D. E. (1999). XtalView/Xfit—a versatileprogram for manipulating atomic coordinates andelectron density. J. Struct. Biol. 125, 156–165.

74. Brünger, A. T., Adams, P. D., Clore, G. M., DeLano,W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998).Crystallography & NMR system: a new software suitefor macromolecular structure determination. ActaCrystallogr., Sect. D: Biol. Crystallogr. 54(Pt 5), 905–921.

75. Sheldrick, G. M. & Schneider, T. R. (1997). SHELXL:high-resolution refinement. Methods Enzymol. 277,319–343.


76. Cabelli, D. E., Allen, D., Bielski, B. H. & Holcman, J.(1989). The interaction between Cu(I) superoxidedismutase and hydrogen peroxide. J. Biol. Chem. 264,9967–9971.

77. Brünger, A. T. (1993). Assessment of phase accuracyby cross validation: the free R value. Methods andapplication. Acta Crystallogr., Sect. D: Biol. Crystallogr.49, 24–36.

78. Richardson, J. S., Bryan, W. A., 3rd & Richardson, D. C.(2003). New tools and data for improving structures,using all-atom contacts.Methods Enzymol. 374, 385–412.

79. Rodriguez, R., Chinea, G., Lopez, N., Pons, T. &Vriend, G. (1998). Homology modeling, model andsoftware evaluation: three related resources. Bioinfor-matics, 14, 523–528.

80. Kurahashi, T., Miyazaki, A., Suwan, S. & Isobe, M.(2001). Extensive investigations on oxidized aminoacid residues in H2O2-treated Cu,Zn–SOD proteinwith LC–ESI–Q-TOF-MS, MS/MS for the determina-tion of the copper-binding site. J. Am. Chem. Soc. 123,9268–9278.

81. Uchida, K. & Kawakishi, S. (1994). Identification ofoxidized histidine generated at the active site of Cu,Zn–superoxide dismutase exposed to H2O2. Selectivegeneration of 2-oxo-histidine at the histidine 118. J.Biol. Chem. 269, 2405–2410.

82. Blech, D. M. & Borders, C. L., Jr (1983). Hydroper-oxide anion, HO-2, is an affinity reagent for theinactivation of yeast Cu,Zn superoxide dismutase:modification of one histidine per subunit. Arch.Biochem. Biophys. 224, 579–586.

83. Goldstone, A. B., Liochev, S. I. & Fridovich, I. (2006).Inactivation of copper, zinc superoxide dismutase byH2O2: mechanism of protection. Free Radic. Biol. Med.41, 1860–1863.

84. Konarev, P. V., Volkov, V. V., Sokolova, A. V., Koch,M. H. J. & Svergun, D. I. (2003). PRIMUS: aWindows PC-based system for small-angle scatter-ing data analysis. J. Appl. Crystallogr. 36, 1277–1282.

85. Svergun, D. (1992). Determination of the regular-ization parameter in indirect-transform methodsusing perceptual criteria. J. Appl. Crystallogr. 25,495–503.

86. Svergun, D., Barberato, C. & Koch, M. H. J. (1995).CRYSOL—a program to evaluate x-ray solutionscattering of biological macromolecules from atomiccoordinates. J. Appl. Crystallogr. 28, 768–773.

87. Svergun, D. I., Petoukhov, M. V. & Koch, M. H.(2001). Determination of domain structure of pro-teins from X-ray solution scattering. Biophys. J. 80,2946–2953.

88. McRee, D. E., Redford, S. M., Getzoff, E. D., Lepock,J. R., Hallewell, R. A. & Tainer, J. A. (1990). Changesin crystallographic structure and thermostability of aCu,Zn superoxide dismutase mutant resulting fromthe removal of a buried cysteine. J. Biol. Chem. 265,14234–14241.

89. Rumfeldt, J. A., Stathopulos, P. B., Chakrabarrty, A.,Lepock, J. R. &Meiering, E. M. (2006). Mechanism andthermodynamics of guanidinium chloride-induceddenaturation of ALS-associated mutant Cu,Zn super-oxide dismutases. J. Mol. Biol. 355, 106–123.

Superoxide Dismutase from the Eukaryotic Thermophile ...

Documents