Biochimica et Biophysica Acta 1817 (2012) 1565–1575

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r .com/ locate /bbabio

Evolution and diversification of Group 1 [NiFe] hydrogenases. Is there a phylogeneticmarker for O2-tolerance?

Maria-Eirini Pandelia a,⁎, Wolfgang Lubitz a, Wolfgang Nitschke b

a Max-Planck-Institut für Bioanorganische Chemie, Stiftstrasse 34‐36, D45470, Mülheim a.d. Ruhr, Germanyb Laboratoire de Bioénergétique et Ingénierie des Protéines, CNRS, FR3479, Aix-Marseille University, France

Abbreviations: DMKn, demethylmenaquinone n, whthe prenyl units present in the side-chain; UQn, ubiqunumber of the prenyl units present in the side-chheterodisulfide reductase; Dsr, dissimilatory sulfate redchemical name 7-mercaptoheptanoylthreoninephosphaanion with the formula HSCH2CH2SO3

−

⁎ Corresponding author at: 104 Chemistry Building,PA 16803, USA.

E-mail address: mxp65@psu.edu (M-E. Pandelia).

0005-2728/$ – see front matter © 2012 Elsevier B.V. Alldoi:10.1016/j.bbabio.2012.04.012

a b s t r a c t

a r t i c l e i n f o

Article history:Received 11 March 2012Received in revised form 21 April 2012Accepted 24 April 2012Available online 1 May 2012

Keywords:[NiFe] hydrogenasesPhylogenetic treeO2-toleranceBioenergeticsIron–sulfur cluster

Group 1 hydrogenases are periplasmic enzymes and are thus strongly affected by the “outside world” the cellexperiences. This exposure has brought about an extensive heterogeneity in their cofactors and redox part-ners. Whereas in their majority they are very O2-sensitive, several enzymes of this group have been recentlyreported to be O2-tolerant. Structural and biochemical studies have shown that this O2-tolerance is conferredby the presence of an unusual iron–sulfur cofactor with supernumerary cysteine ligation (6 instead of 4 Cys,hence called ‘6C cluster’). This atypical cluster coordination affords redox plasticity (i.e. two-redox transi-tions), unprecedented for this type of cofactors and likely involved in resistance to O2. Genomic screeningand phylogenetic tree reconstruction revealed that 6C hydrogenases form a monophyletic clade and are un-expectedly widespread among bacteria. However, several other well-defined clades are observed, which in-dicate early diversification of the enzyme into different subfamilies. The various idiosyncrasies thereof areshown to comply with a very simple rule: phylogenetic grouping of hydrogenases directly correlates withtheir specific functions and hence biochemical characteristics. The observed variability results from gene du-plication, gene shuffling and subsequent adaptation of the diversified enzymes to specific environments. Animportant factor for this diversification seems to have been the emergence of molecular oxygen. Hydroge-nases appear to have dealt with oxidative stress in various ways, the most successful of which, however,was the innovation of the 6C-cluster conferring pronounced O2-tolerance to the parent enzymes.

© 2012 Elsevier B.V. All rights reserved.

1. Introduction

Chemiosmotic energy conversion, life's penultimate way toharvest energy, taps the ΔG available in redox disequilibria betweenreducing and oxidizing substrates [1,2]. Whereas the nature of the an-cestral electron sink exploited by nascent life is still a matter of debate[3], molecular hydrogen almost certainly was the first electron source[1,2]. [NiFe] hydrogenases, extant enzymes using the H+/H2 couplefor bioenergetic ends, are found in a wide variety of phyla in Archaeaand Bacteria. Molecular phylogeny suggests them to fall into four dis-tinct groups, the so-called Group 1 to 4 hydrogenases, a distinctionsupported by phylogenetic markers such as subunit compositionand operon structure [4,5].

ere n refers to the number ofinone n, where n refers to theain; MK, menaquinone; Hdr,uctase; CoB, coenzyme B (fullte); CoM, coenzyme M is the

Pennsylvania State University,

rights reserved.

Group 1 hydrogenases are generally designated as membrane-associated enzymes coupling the H+/H2 couple to quinones1 in met-abolic pathways with a variety of redox substrates. They are exportedto the periplasmic space via an N-terminal signal peptide located inthe small subunit that is known as the twin-arginine translocationmotif (i.e. RRXFXK) [7], whereas the large subunit lacking this signal-ing motif is co-translocated with the small subunit. They representthe both phylogenetically most widespread and best studied group.A subclass – the so-called O2-tolerant Group 1 hydrogenases – has re-cently gained extensive attention, due to their potential biotechno-logical importance. Only a few cases of O2-tolerant enzymes havebeen studied so far, e.g. the enzymes from Aquifex aeolicus (Hase I)[5,8], Ralstonia eutropha (MBH) [9], Hydrogenovibrio marinus [10]and Escherichia coli (Hyd-1) [11,12]. Whereas Group 1 hydrogenasesare in general typified via the extensively studied Desulfovibrio en-zymes [13] (Fig. 1), the O2-tolerant ones are often considered as theexotic outliers of the group.

Very recently, crystal structures of three O2-tolerant enzymes[10,12,14] showed the catalytic [NiFe] site and number of iron–sulfur

1 The archaeal hydrogenases from Methanosarcinales do not contain the usual qui-nones, but methanophenazines (MP), which are functional menaquinone analogs [6](see also below and SI).

Fig. 1. Localization of the metal cofactors in the H2-reduced state of the O2-sensitiveperiplasmic hydrogenase from Desulfovibrio vulgaris Miyazaki F (pdb id: 1H2R, top)and of the membrane-bound O2-tolerant hydrogenase from Hydrogenovibrio marinus(pdb id: 3AYX, bottom). The large subunit (dark grey ribbon) contains the hetero-bimetallic [NiFe] site and the small subunit (white ribbon) contains the three iron–sulfur centers in an almost linear arrangement. The glutamate-rich surface near thedistal [4Fe4S] (1His3Cys) cluster situated close to the surface is shown with redcolor. Note the difference in the structure of the cluster proximal to the [NiFe] site be-tween the two enzymes (6C cluster in the H. marinus hydrogenase).

1566 M-E. Pandelia et al. / Biochimica et Biophysica Acta 1817 (2012) 1565–1575

cofactors to be conserved, albeit with nuances. The most prominentmodification is that the [4Fe4S] cluster proximal to the [NiFe] site(Fig. 1) has been replaced by a novel [4Fe3S] center, coordinated by atotal of six cysteine residues (6C cluster), instead of the canonical four.The [4Fe3S] is able to perform two functionally meaningful single-electron transitions [15], unlike traditional [4Fe4S] clusters. Its uniqueredox plasticity is promoted by redox-dependent structural changesresulting from its surplus cysteine coordination. Site-directedmutagen-esis indeed showed that O2-tolerance of 6C-containing hydrogenasesis dominantly linked to the presence of the [4Fe3S] cluster [16,17].

R. eutropha is a β-proteobacterium belonging to the so-calledcrown-group of bacterial prokaryotes. A. aeolicus, by contrast, is amember of one of the earliest branching bacterial phyla, theAquificales. Branching-off of the Aquificales is considered to have oc-curred prior to the advent of oxygenic photosynthesis and theresulting rise in O2 levels in the biosphere. The simultaneous presenceof the oxygen-tolerant 6C-hydrogenases in these two species there-fore raises a number of evolutionary questions: Does it represent anevolutionary product that marks transition to a more oxygenic life-style? Has the trait of the 6C cluster been transferred laterally? Has

this strategy to cope with O2 appeared several times independentlyor may it be a relatively ancient trait?

To better understand the origin(s) and evolutionary pathways ofthe O2-tolerant 6C-hydrogenases, we have re-analyzed the topologyof the phylogenetic tree of the Group 1 family with specific attentionto structural marker characteristics of the complete hydrogenase en-zyme complexes. The obtained results suggest surprising evolution-ary pathways not only for the oxygen-tolerant subgroup but also forthe whole ensemble of Group 1 enzymes.

2. Materials and methods

BLAST (Basic Local Alignment Search Tool) searches were per-formed using web tools from the National Center for BiotechnologyInformation (NCBI) (http://blast.ncbi.nlm.nih.gov/Blast.cgi) [18]. Pro-tein BLAST analyses used the amino acid sequences of the small [NCBIReference Sequence: NP_213454.1, Locus tag: aq_660], large [NCBIReference Sequence: NP_213455.1, Locus tag: aq_662], cytb [NCBIReference Sequence: NP_213456.1, Locus tag: aq_665] subunits ofthe A. aeolicus Hydrogenase (Hase) I as queries. Multiple sequenceanalysis and alignments were performed with the programs ClustalX(version 2), MUSCLE (http://www.ebi.ac.uk/Tools/msa/muscle/) andfurther refined (taking advantage of structural information) usingSeaview (version 4). The trees were constructed with the ClustalX(version 2) and the MEGA 5 [19] software using the Bootstrap NJalgorithm without excluding protein gaps. An exhaustive and up-to-date inventory of homologs was carried out on current sequencedatabases; the alignments were checked to remove highly divergentsequences, gaps, and ambiguously aligned positions. For this purpose,also the GUIDANCE web-server (http://guidance.tau.ac.il/overview.html) was additionally used as a comparative tool for assigning a con-fidence score for each residue, column, and sequence in an alignmentand for projecting these scores onto the multiple sequence align-ments. For the purpose of rooting our phylogenetic trees, we havetested the robustness of our trees by modifying the precise outgroupenzymes and varying the number of ingroup representatives. Inthe present study, we constructed our trees using as a singleoutgroup the soluble hydrogenase from A. vinosum DSM 180(YP_003443825.1), since it gave essentially the same results. Thephylogenetic trees were constructed using the sequences of thesmall subunits (iron–sulfur cluster containing) and the large subunits([NiFe]-site containing) of the respective enzymes. Both provided es-sentially the same tree topology. Therefore, only trees constructedbased on the sequences of the large subunit module are presented,which, due to their higher number of informative sites, yieldedbranching order supported by higher bootstrap values. Trees werevisualized and edited further with the open source programTreeView (version 1.6.6). All structures were obtained from theRCSB PDB protein database (www.pdb.org) [20]. Redox potentials ofb-cytochromes in the main manuscript are given in Table 2. Theredox potentials for the NarI cytochrome in nitrate reductases havebeen measured in two different organisms in E. coli [21] and inP. denitrificans [22]. These potentials are significantly more positivecompared to those measured for the cytbI, occurring in the majorityof the Group I hydrogenases studied up to date as well as Formate De-hydrogenases [23–25]. Lists of all the organisms that have been usedfor the construction of the phylogenetic trees have been attached as aseparate Excel Spreadsheet.

3. Results and discussion

3.1. Group 1 heterogeneity; gene cluster organization, subunit composi-tion and nature of iron–sulfur clusters

Group 1 hydrogenases are generally considered to consist of thelarge and the small subunit (Fig. 1) and to be connected to the

Fig. 2. (Left) Gene cluster arrangements of Group 1 hydrogenases. The types of the three iron–sulfur clusters situated in the small subunit are distinguished for each case ([4Fe] ascubic and [3Fe] cubic with missing corner pictograms). The cluster proximal to the [NiFe] site shows extensive variations between different species in its primary coordinationmotif. The characteristic primary motifs are given in the beginning of each line (a) to (g), the special coordinating amino acids and the pictograms of the clusters are highlightedin the respective colors. (Right) Structure of the classical four-cysteine ligated cluster typically encountered in the proximal position in Group 1 enzymes (pdb id: 1H2R, H2-reducedD. vulgaris MF). The six-cysteine coordinated (6C) cluster in the O2-tolerant H. marinus hydrogenase (pdb id: 3AYX, H2-reduced state); the two extra cysteine residues are coloredin red.

1567M-E. Pandelia et al. / Biochimica et Biophysica Acta 1817 (2012) 1565–1575

quinone pool via a di-heme cytochrome b (cytb) that, together withthe hydrophobic C-terminus of the small subunit, anchors the en-zyme to the membrane. These prototypical Group 1 hydrogenasesare encoded by clustered genes arranged in the order small-large-cytbI-subunit (Fig. 2a). In the small subunit, a [3Fe4S] cluster is locat-ed in the medial position between two [4Fe4S] centers. The clusterproximal to the [NiFe] site is an all-cysteine (4Cys) ligated [4Fe4S]cluster and the one most distant from the [NiFe] site (distal) is coor-dinated by three cysteines and one histidine (3Cys1His) (Table 1,Fig. 1). However, both genome surveys [4] and biochemical studies[5,11,26] have shown numerous members of the group featuring sub-stantially different make-ups with respect to all these characteristics.Our comparative genome screening approach for Group 1 hydroge-nases came up with 7 basic arrangements differing in gene clusterarchitecture, presence of additional/alternative subunits and struc-ture of the iron sulfur clusters (Fig. 2).

Group 1 thus starts to look more like a mixed crowd than a homo-geneous group. Membership to Group 1 was initially based on se-quence markers [4] but later shown to correspond to phylogeneticclustering. In the following, we will show that the seven variantsshown in Fig. 2 actually correspond to a phylogenetic subclusteringand that the subclades most likely arose in response to specific ener-getic requirements and environmental constraints.

Table 1Binding motifs of the three iron–sulfur clusters occurring in the membrane bound hydrogenmiddle one and distal the most distant one (Fig. 1). The two extra cysteines of the proxima

Classes Proximal

6C Hases CXCC X93–94 CX4CX28–30CPrototypical/HybA CXGC X91–96 CX33–38CCxxN (bacteria) CXGN X94–96 CX39–40CCxxD (bacteria) CXGD X92–94 CX40Csulfate reducers (bacteria) CXGC X94 CX32–35CBacteria Isp CXGC X92–96 CX33CCrenarchaea Isp (CxxN) CXGN X114–138 CX80–82CCrenarchaea Isp (CxxD) CXGD X118–178 CX81–83CEuryarchaea/Actino 3x[4Fe] CXGC X92–107 CX37–49CEuryarchaea DxxC 3x[4Fe] DXGC X107 CX49C

3.2. Phylogeny orders the apparent Group 1 heterogeneity

Phylogenetic analysis has been carried out using 20 archaeal and93 bacterial sequences, based on sequence homology of the large cat-alytic and the small electron-transferring subunits. The number of se-quences retrieved in Blast searches was equilibrated to avoiddisproportionate trees resulting from species biases in genomic data-bases. An unrooted radial phylogram based on the aligned sequencesof the large subunits is shown in Fig. 3. For rooting of the tree, theGroup 3 soluble hydrogenase from Allochromatium vinosum wasused as outgroup (Fig. 4). We have tested the robustness of ourtrees by modifying the precise outgroup enzymes and varying thenumber of ingroup representatives. Two observations are notewor-thy: (a) The clades discernible were strictly robust, that is, nevermixed or coalesced. These clades are further supported by evolution-ary marker traits, i.e. specific idiosyncrasies with respect to gene clus-ter organization, nature of the membrane-anchor subunit and ligandsto redox centers. (b) Branching topologies in the archaeal domainwere relatively stable. This was also true for the lowest two phyla inthe bacterial part of the tree, i.e. Actinobacteria and δ-proteobacteria(Fig. 3). Both these two phyla also figure prominently at the base ofrecent 16S rRNA-based species trees [27]. Considering the rapid se-quence of radiations (Figs. 3, 4) and the low bootstrap values for the

ases of Group 1 (Figs. 1, 2). Proximal is the cluster closest to the [NiFe] site, medial thel 6C cluster occurring in known O2-tolerant hydrogenases are indicated in bold.

Medial Distal

CX11 PX6CX2C HX2C X24CX5CCX11 PX6CX2C HX2C X24CX5CCX11 PX6CX2C HX2C X24CX5CCX11 PX6CX2C HX2C X24CX5CCX11–13 PX6CX2C HX2C X24CX5CCX11 PX6CX2C HX2C X24CX5CCX13 PX6CX2C HX2C X21CX5CCX13 PX6CX2C HX2C X21CX5CCX11 CX6CX2C HX2C X24CX5CCX11 CX6CX2C HX2C X24CX5C

Fig. 3. Radial phylogram of the Group 1 hydrogenases based on the sequence homology and comparison of the large subunit (contains the catalytic [NiFe] site). The Group 3 solublehydrogenase from A. vinosum DSM 180 (YP_003443825.1) has been included as outgroup. The tree has been built by the neighbor-joining (NJ, distance-matrix) method consideringgap positions and accounting for multiple substitutions. The tree has been built using 1000 bootstrap samplings.

1568 M-E. Pandelia et al. / Biochimica et Biophysica Acta 1817 (2012) 1565–1575

corresponding nodes (Fig. 4), it is not surprising that the branchingorder of the higher clades varied substantially as a function of thewhole sequence ensemble.

3.3. Bioenergetics and molecular O2; an intricate interplay driving molec-ular evolution

3.3.1. Oxygen-sensitive hydrogenases, the likely ancestral typeIn the bacterial domain, the earliest branching part consists of

Actinobacteria, sulfate-reducing and ε-proteobacteria. These organ-isms are strict anaerobes and contain the simplest structures of[NiFe] gene operons (Fig. 2, lines c, d, g) as well as a four or three-cysteine ligated proximal tetranuclear (most likely) [FeS] cluster.The same types of hydrogenases (disregarding deviant proximal clus-ter ligation, see below) occur in many Archaea (Fig. 2 lines c, d). It istherefore likely that these come close to the ancestral Group 1 hy-drogenases having existed in the Last Universal Common Ancestor(LUCA). These hydrogenases are extremely oxygen-sensitive and typ-ical representatives are those from Desulfovibrio (periplasmic) andWolinella species (membrane-attached and reducing menaquinone(MK)). The [NiFe] enzymes from Desulfovibrio are the most thorough-ly studied and are classified as Group 1 despite their interaction witha soluble cytochrome c3 (cytc3) rather than with quinones (Fig. 5).

The properties of the higher radiations (Fig. 3) suggest rapid di-versification from this ancestral structure into new subfamilies,based on gene duplication events and recruitment of additional sub-units according to the Redox-protein-construction-kit principle [28].

Most of these higher clades predominantly contain members frommore recently branched phyla with only a few isolated cases of obvi-ous lateral gene transfer into specific members of Actinobacteria andδ-proteobacteria.

3.3.2. Hydrogenases with low oxygen-tolerance

3.3.2.1. HybA-hydrogenases. HybA-enzymes contain in between thesmall and large subunit genes two other additional genes, hybA andhybB (Fig.2, line f). The HybA protein contains 16 conserved cysteinesprobably ligating four [4Fe4S] clusters and belongs to the family oftetracubane electron-transfer proteins [25,28]. HybB codes for amembrane-integral protein, predicted to have 9–10 transmembrane he-lices and frequently annotated as “putative cytochrome subunit”. How-ever, there are no conserved canonical binding sites for hemes. Its highsimilarity with a class of membrane-associated proteins including PsrC(polysulfide reductase), DmsC (dimethyl sulfoxide reductase) and NrfD(assimilatory nitrite reductase) strongly suggests that HybB serves as aquinone reduction site [29] and, similar to PsrC, is devoid of b-hemes(Fig. 5). HybA-hydrogenases are clustered into a monophyletic clade(Figs. 3, 4) with mainly γ- and very few α- and β-proteobacterial mem-bers. The best characterized example is Hydrogenase-2 (Hyd-2) from E.coli [11,30,31] and this enzyme provides important clues for deducingthe metabolic function of HybA-hydrogenases. Hyd-2 is bidirectional invitro and tolerates limited concentrations of O2 under very reducing con-ditions [11]. It shows an increased tendency for hydrogen production.The observation that Hyd-2 can perform hydrogen catalysis at low

Fig. 4. Rooted phylogram of Group 1 hydrogenases constructed from the aligned sequences of the large subunit gene. The soluble Group 3 hydrogenase from A. vinosum has beenused as outgroup. A detailed description of the organisms used is contained in the SI. Selected organisms have been annotated: the crenarchaeal hydrogenases from Thermoproteustenax (CCC80712.1) and Thermoproteus uzoniensis 768–20 (YP_004338974.1), the MBH from R. eutropha (NP_942644.1), MBH from H. marinus (provided by Higuchi and co-workers)8, Hyd-1 from Escherichia coli (NP_415492.1), Hase I from A. aeolicus (NP_213455.1), MBH from A. vinosum DSM 180 (YP_003443990.1), HynL from Thiocapsa rose-opersicina (AAC38282.1), the membrane associated hydrogenase from Alteromonas macleodii str. ‘Deep ecotype’ (YP_004425482.1), the MBH from Paracoccus denitrificansPD1222 (YP_916875.1). Full and open squares denote the bootstrap values of high and small confidence, respectively. The tree was built using 1000 bootstraps repetitions. Differentcolors have been used to indicate the respective clades (see text).

1569M-E. Pandelia et al. / Biochimica et Biophysica Acta 1817 (2012) 1565–1575

oxygen concentrations (~1% O2) [11] may indicate that these enzymesencounter, and thus have to copewith, oxygen in vivo, albeit at relativelylow concentrations. These hydrogenases, therefore, likely emerged at apoint in the Earth's past when oxygen appeared in the environment.The present species distribution of HybA-hydrogenases suggests themto have emerged prior to the radiation of α-, β- and γ-proteobacteria.Precisely this region of the proteobacterial subtree has been suggestedpreviously, based on quinone content, to mark the emerging impact ofoxygen in the biosphere [32].

Hyd-2 from E. coli couples periplasmic hydrogen oxidation(−420 mV) to cytoplasmic fumarate reduction (+30 mV) [11,30,33].The redox potentials of the catalytic [NiFe] site and the iron–sulfurclusters have not been determined so far. By comparison of their elec-trochemical properties, the reduction potentials may be inferred asclose to those of oxygen-sensitive hydrogenases [15,34] (Fig. 6).

A striking property of the HybA-hydrogenases is the absence of b-hemes in their membrane-anchor subunit which likely couples H2 ox-idation to quinone reduction. For Hyd-2, absence of b-hemes finds anatural rationalization a) in the low ΔG available for electron transferfrom H2 to fumarate and b) localization of H2 and fumarate catalyticsites with respect to the cytoplasmic membrane. H2 oxidation re-leases protons into the periplasm and builds up a proton gradient.Shuttling the derived electrons across the membrane for quinonereduction and hence proton uptake from the cytoplasm further in-creases this gradient. However, transferring electrons against the150 mV electrostatic potential of an energized membrane consumescommensurable redox energy. The succinate/fumarate couple is sub-stantially less oxidizing than most common terminal acceptors (ni-trogen oxides, Fe(III), O2, etc.), making the ΔE of the donor/acceptorcouple insufficient for sending electrons against the membrane

Fig. 5. The four redox partner modules of Group 1 hydrogenases. (a) cytbI features the two-helix motif (the 4 His ligands of the 2 hemes come from two helices). The structuralmodel shown is that of the apo-cytbI from A. aeolicus constructed based on the γ subunit of E. coli Formate Dehydrogenase-N (FdnI, pdb id: 1KQF). Hemes were included after struc-tural alignment. (b) cytbII (or Isp1) has the three-helix motif (the 4 His ligands of the 2 hemes come from three helices). The structural model shown is that of the apo-cytbII fromRhodobacter sp. SW2 (ZP_05843854.1) constructed based on the γ subunit of E. coli nitrate reductase (NarI, pdb id: 1Q16). (c) The cytochrome-devoid HybB protein. It belongs tothe superfamily typified by the γ subunit of polysulfide reductase, for which the crystal structure with boundmenaquinone is shown (PsrC-MK7, pdb id: 2VPZ). (d) The soluble cytc3(4 hemes) from D. vulgaris MF (pdb id: 1J0O), is the redox partner of [NiFe] hydrogenases from sulfate-reducing bacteria.

1570 M-E. Pandelia et al. / Biochimica et Biophysica Acta 1817 (2012) 1565–1575

potential and, as a result, the b-hemes are disposed of as in the case ofpolysulfide reductase [29].

The conserved presence of hybB suggests that throughout thisclade, the parent electron transfer chains use weakly oxidizing termi-nal acceptors. This scenario might, at first sight, seem in conflict withthe occurrence of one α- and two β-proteobacterial representatives,i.e. species that usually are devoid of low potential menaquinones.The redox potential of ubiquinone, however, would be too high to op-erate such “short” chains toward weak oxidants. It is intriguing thatone out of these three cases belongs to a group known to producerhodoquinone [35], a low-potential quinone proposed to be derivedfrom ubiquinone [36]. It seems likely to us that low-potential qui-nones (likely rhodoquinone) also exist in the two remaining α- andβ-proteobacterial species.

3.3.2.2. Isp-hydrogenases. In Isp-hydrogenases two additional openreading frames, called isp1 and isp2, are positioned in between thesmall and large subunit genes (Fig. 2, line e). Homology modelingand hydrophobicity analyses predict that Isp1 has five transmem-brane helices and binds two hemes via the so-called two-helixmotif. By virtue of sequence similarity, Isp1 belongs to the b-heme su-perfamily (Fig. 5, cytbII cytochrome) including NarI [6,37–39], HdrEfrom methanogenic Archaea [40], DsrM from A. vinosum [38] andHmeC2 from A. fulgidus [37], whose role, when known, is to oxidizethe quinone pool [37,38]. All of these isp1-like genes are followed intheir gene clusters by the isp2 gene. Isp2 has high sequence similarityto the catalytic subunit HdrD of heterodisulfide reductase fromMethanosarcina barkeri [26,40]. Hdr proteins typically harbor twoconventional [4Fe4S] clusters and two tetranuclear clusters with aunique CX31–38CCX33–34CXXC motif (5C) [6,40,42]. The C-terminalcopy of these 5C clusters is thought to mediate the reductive cleavageof a disulfide bond [40] and Isp2 contains only this C-terminal copy.

Isp-hydrogenases form two separate subclades, one in Archaeaand one in Bacteria (Fig. 3), leaving us confronted with an evolution-ary conundrum, which can be rationalized by two distinct scenarios.Either the isp1–isp2-module, present in other enzyme systems[37,38,40] was inserted twice independently into an ancestralsmall-large subunit gene cluster or the tree resolution for very deepevolutionary times may be insufficient. The “actual tree” may in thatcase correspond to two separate subtrees for prototypical and Isp-

2 HmeC is more properly referred to as DsrM [41], though when it was isolated fromArchaeglobus fulgidus it was given the name HmeC.

hydrogenases, respectively, which would have both co-existed inLUCA. Weak sequence divergence may preclude resolving the twosubtrees and thus cause the apparent collapse into a single tree. Atthis point, both (or still further) scenarios are possible.

Gleaning the metabolic role of Isp-hydrogenases is more difficultthan for HybA-enzymes, since for no member an unambiguous func-tion has been elucidated so far. Typical representatives arethe membrane-attached enzymes from Thiocapsa roseopersicinaand A. vinosum. In their purified state they only contain the catalyticheterodimer (HynSL) and lack Isp1–Isp2. Mutational studies showedthat Isp1–Isp2 are critical for in-vivo H2 cycling; in their absence hy-drogen production is abolished, whereas hydrogen oxidation is ~70%decreased [26]. This strong link between hydrogenase function andpresence of the isp genes together with their central localization inthe operon (rendering co-regulation and co-expression very likely)demonstrate that the two Isp proteins are an integral part of the hy-drogenase complex in vivo.

The best characterized Isp-hydrogenase, A. vinosum HynSL [43,44],is oxygen-sensitive arguing for a role in an anaerobic bioenergeticchain. The redox potentials of its catalytic [NiFe] states are close tothe H2/H+ couple (−420 mV) [45], the [3Fe4S] cluster titrates at+10 mV [46] and the two [4Fe4S] clusters at ≤−300 mV [43], i.e.rather similar to those reported for the oxygen-sensitive hydroge-nases from the Desulfovibrio anaerobes [15,34] (Fig. 6). Unfortunately,the redox properties of the associated Isp1–Isp2 modules are notknown. Other homologous biochemically characterized cases are theheme/iron–sulfur complex (HmeCD)3 from A. fulgidus [37], theDsrMK complex from Desulfovibrio desulfuricans ATCC 27774 [41],proposed to be involved in disulfide reduction (linking quinones todissimilatory sulfite reduction) with so-far unknown thiol substrates.Similar to Isp2, HmeD and DsrK contain only the C-terminal 5Ccluster. Their midpoint potentials in the substrate-bound state is+90 mV [37] and+130 mV [41], respectively and thus more positivethan that of HdrE from Methanothermobacter marburgensis(−185 mV for CoM-HDR,−30 mV CoB-HDR) [40]. The reduction po-tentials of cytbII-type hemes (Isp1) will depend on the chemical na-ture of the quinone pool and the substrate of the Hdr-type protein(Isp2). With the exception of HdrE from M. thermophila, linked tomethanophenazine (E=−165 mV) [6,47], the redox potentials ofcytbII-type hemes are more positive than those of cytbI

3 HmeCD is more properly referred to as DsrMK [41].

Fig. 6. Redox scheme of the different Group 1 enzymes constructed from spectroscopic and biochemical data. 6C-hydrogenases: the [NiFe], iron–sulfur clusters' and cytbI redox po-tentials are those measured for A. aeolicus Hase I. Hydrogenases from sulfate reducers (SR): redox potentials given are for D. vulgaris MF and D. gigas. Isp-hydogenases: redox po-tentials are those measured for A. vinosum. The redox potentials of the Isp1–Isp2 complex were assumed similar to those of HmeCD/DsrMK (Table 2). HybA-hydrogenases: redoxpotentials were deduced from electrochemical data on E. coli Hyd-2. Redox potentials of the HybA iron–sulfur clusters were assumed to lie close to those of DmsB/NarH. MK-reactive hydrogenases: redox potentials were considered similar to those of SR and HybA-hydrogenases (based on the W. succinogenes cytbI redox potentials and phylogenetic po-sition of these enzymes). On the right the redox potentials of electron acceptors are given. Note that sulfate is a much weaker oxidant than O2; linking of the respective enzymes tothese two different electron acceptors thus dictates their different redox properties.

1571M-E. Pandelia et al. / Biochimica et Biophysica Acta 1817 (2012) 1565–1575

[22,23,25,48] (Fig. 6, Table 2). Methanophenazine in methanogens isa functional analog of menaquinone [47,49]. It is a small hydrophobicquinone-like redox cofactor with an isopropenoid side-chain (seeSupplementary information). Its role in the energy metabolismof methanogens is similar to that of quinones in mitochondria andbacteria. The final electron acceptor of the HynSL–Isp1–Isp2 super-complex is unknown. However, it seems to us that the presence ofcytbII and the heterodisulfide-related Isp2 protein strongly indicatesa link to a sulfur-dependent metabolic pathway [26]. Thishypothesis is supported by the observation that HynSL fromT. roseopersicina is expressed in the presence of thiosulfate [26].

Which kind of redox reaction between sulfur compounds and theH2/H+ couple might the Isp-hydrogenases catalyze? The gene clusterand observed targeting sequences suggest an enzyme consisting of aperiplasmic hydrogenase module (small-large subunit), a cytoplasmicredox protein converting sulfur compounds (most likely S\S bonds)and a b-cytochrome serving as transmembrane electron wire. Thisstructure allows for several bioenergetically sensible electron-transferscenarios. The simplest one might stipulate periplasmic H2 oxidation(liberating two protons), followed by transmembrane electron-transfer (against the membrane potential) and ultimately reducingunspecified disulfide bonds (consumption of two protons in the cyto-plasm). This ultra-short electron-transfer chain can generate a proton

Table 2Binding motifs of the two b-cytochrome types encountered in Group 1 hydrogenases. cytbI beligands to the hemes come from three helices. cytbII or more commonly Isp1 belongs to thehemes come from two helices. The last column contains the midpoint redox potentials fopotential.

Subgroup cytbI

CXCC (bacteria) HX40–46H HX13H

DSGC (Archaea) HX32H HX13HCXGC (bacteria) HX37–42H HX13HCXGD (bacteria) HX38–41H HX13HCXGN (bacteria) HX38–41H HX13HCXGC (Archaea/actinobacteria [4Fe]) HX30–33H HX13HCXGN (bI‐like Crenarchaea) HX58H HX14HFormate dehydrogenase (W. succinogenes pdb id: 1QLB) HX38H HX13HCXGC (Isp bacteria)CXGN (Isp Crenarchaea)CXGD (Isp Crenarchaea)Nitrate reductase (E. coli pdb id: 1Q16)Nitrate reductase (P. denitrificans)DsrM (A. vinosum)

a [19]. b [44]. c [17]. d [18]. e [34].

gradient and thus be energetically viable under certain growth condi-tions. More complicated electron-transfer models can be thought of,if involvement of quinones is envisaged. Electrons issued both from hy-drogen and sulfur oxidation to disulfides might merge to reduce qui-none at the cytoplasmic face of the membrane. The quinol wouldthen go downstream toward higher potential terminal acceptors (oreven pass through the Rieske/cytb complex). Finally, even the reversescenario is possible, despite its apparent counterintuitive aspects: theIsp-complex might act as a terminal electron acceptor from reducedquinol via electron bifurcation to hydrogen pulled by reduction of ahigher potential acceptor (the S\S bond).

Recently, the heterodimeric (SL) hydrogenase from theγ-proteobacterium Alteromonas macleodii was reported to toleratethe presence of 1–3% oxygen [50]. This enzyme shares a strong ho-mology (60–69%) with Isp-hydrogenases [50] but lacks isp1–isp2.Phylogenetic analysis places the A. macleodii enzyme in the midst ofIsp-hydrogenases (Figs. 3, 4), whereas absence of Isp1–Isp2 likely in-dicates coupling to altered electron acceptors.

Although the T. roseopersicina HynSL was initially reported to berelatively O2-resistant, recent studies showed the activity to be lostat ~1% O2 [50]. Similarly, the A. vinosum HynSL is considered to bevery oxygen-sensitive. In this context, the decreased O2-sensitivityof the A. macleodii enzyme is intriguing. This difference may be

longs to the FdnI superfamily. It adopts the so-called three-helix motif, i.e. the histidineNarI superfamily. It adopts the so-called two-helix motif, i.e. the histidine ligands to ther the different di-heme cytochromes quoted versus the standard hydrogen electrode

cytbII (Isp1) Redox potential, mV

−20, −110 (A. aeolicus)a

+130, −60 (A. aeolicus)a

−100, −240 (W. succinogenes)b

−20, −190b

HX12H HX17HHX12–13H HX17HHX13H HX17HHX9H HX17H −10, +125c

HX9H HX17H +95, +210d

HX9H HX17H +60, +110e

1572 M-E. Pandelia et al. / Biochimica et Biophysica Acta 1817 (2012) 1565–1575

explained considering that O2-tolerance of enzymes from the Isp-clade may crucially depend on their in vivo operating potential andconsequently their precise metabolic role.

3.3.3. Oxygen-tolerant 6C-hydrogenasesThe most intriguing clade is perhaps that containing the oxygen-

tolerant hydrogenases. Already more than 30 years ago, an atypicalEPR signature was observed in hydrogenases from R. eutropha andP. denitrificans [51,52], never seen before in those from sulfate reducers[34,53]. These signatures, suggestive of an additional paramagneticcenter, subsequently resurfaced in the spectra of other Group 1hydrogenases that are members of this clade (A. aeolicus [5,15],Azotobacter vinelandii [54], Hydrogenophaga sp.AH-24 [55], Cupriviadusmetallindurans [46] and E. coli [11]).

The identity of the unknown cofactor remained enigmatic for almosttwo decades andwas elucidated only recently [15]. EPR electrochemicaland Mössbauer studies identified this paramagnetic center to a second,higher potential redox transition of the proximal cluster [15]. Thediamagnetic “oxidized” proximal cluster can be reduced to a state thatresembles spectroscopically [4Fe4S]1+ ferredoxins (S=1/2). However,in the O2-tolerant hydrogenases, it can furthermore be super-oxidizedto a form reminiscent of the paramagnetic 3+ state of HiPIP centers[15]. The paramagnetism of the super-oxidized proximal cluster isresponsible for the atypical complex spectroscopic signatures ofthese hydrogenases [8,11,56,57], which result from magnetic inter-actions with the neighboring paramagnetic [NiFe] and [3Fe4S]1+

centers [15].The recent crystallographic structures from two members of this

family [10,12,14] revealed that the proximal cluster is actually a[4Fe3S] cluster with in-total six coordinating cysteine residues(Figs. 1, 2 line b). These six cysteines are fully conserved in this family,hence called 6C clade. The sulfur atom of the first extra cysteine (C) inthe primary binding motif CTCC replaces one of the inorganic sulfidesand thus becomes an intrinsic cluster ligand [10,14], whereas the sec-ond cysteine terminally coordinates one of the Fe atoms [10,14](Fig. 1, bottom). In line with the dual redox character of the 6C clus-ter, the structures of the super-oxidized and reduced forms showunique features. The CC tandem cysteine motif appears to be criticalfor its structural and redox plasticity, stabilizing the super-oxidizedform by offering an additional coordinating ligand via deprotonationof the CC peptide amide nitrogen [10]. These redox-dependent struc-tural changes promoted by the surplus cysteine coordination, allowthe 6C [4Fe3S] cluster to carry out two single-electron transitions. In-terestingly, a direct correlation of O2-tolerance and the presence ofthe 6C cluster could be demonstrated via cysteine deletion mutantsin R. eutrophaMBH [16]. These were found to show growth limitationfor oxygen concentrations higher than 2%, the O2-sensitivity of thepurified enzyme was significantly increased [16], whereas the atypi-cal EPR features due to the presence of the super-oxidized proximalcluster were absent.

Our phylogenetic trees demonstrate that the notion of O2-tolerant6C hydrogenases being an oddity among Group 1 enzymes is mis-guided (Figs. 3, 4). The 6C cluster is not restricted to enzymes from(micro)aerophilic organisms, but is also found in those from strict an-aerobes (e.g. Heliobacteria, Chlorobi). The presence of the completegene sequences in these operons and the wide distribution of 6C-enzymes in the Chlorobi/Firmicutes phyla, shows that they encodefunctional hydrogenases rather than representing pseudogenes. Inthe light of the proposed rescue function for this cofactor against ox-idative stress [16], this was unexpected.

The clade of 6C-hydrogenases is monophyletic and all membersshare the 6C characteristics (i.e. likely the structure of the [4Fe3S] prox-imal cluster). 6C-type enzymes are found in the crown-groupα-, β-, γ-proteobacteria but also in δ-proteobacteria, Bacteroidetes, Aquificales,Chlorobiaceae and Firmicutes. Since the 6C cluster confers a substantialfitness advantage in environments at least occasionally exposed to O2,

a wide species distribution driven by lateral gene transfer appears like-ly. However, the topology of the 6C-clade strongly argues for a differentscenario. The 6C tree in fact splits into two well-separated parts; (a)Bacteroidetes, Chlorobiaceae and Firmicutes (Heliobacteria andDesulfitobacteria) and (b) α-, β- and γ-proteobacteria. In the upperpart (a) of the 6C subtree, low bootstrap values hamper conclusionson vertical or lateral gene transfer. The cleavage, however, betweenProteobacteria and Chlorobiaceae/Firmicutes/Bacteroidetes, is stronglybootstrap-supported and reflects 16S rRNA species trees [27] (Fig. 4).

Aquificales appear as a subgroup branching at the base of the prot-eobacterial sequences. As shown previously, a high percentage of theirgenomes was imported laterally from ε-proteobacteria [58] andAquificales may even serve as repositories for ε-proteobacterial genesthat are difficult to recognize in extant ε-proteobacteria. It thereforeappears likely that the group of Aquificales actually represents se-quences arising from ε-proteobacteria, making this “side” of the 6Ctree closely resemble the proteobacterial species phylogeny (Figs. 3, 4).

After the branching-off of the Aquificales, the next higher node splitsinto two distinct clades, both of which contain α-, β- and γ‐prote-obacteria (Fig. 4). The upper clade is characterized by high bootstrapvalues and reproduces the cleavage into α-, β-, γ-proteobacteriacharacteristic for species trees [27]. The lower clade features verylow bootstrap values and an extreme mixture of α-, β-, γ- andδ-proteobacterial representatives. It cannot be decided whether the dis-ordered topology is due to pervasive lateral gene transfer or whether itreflects insufficient phylogenetic signal. Irrespective of this ambiguity, agene duplication event in the ancestor of α-, β-, γ-proteobacteria ap-pears to have given rise to these two distinct micro-families of6C-hydrogenases.

The tree topologies shown in Figs. 3 and 4 suggest a provocativeconclusion: 6C-hydrogenases appear to have emerged prior to the di-vergence of Chlorobia/Heliobacteria on one and α-, β- and γ-proteo-bacterial radiations on the other side. In the traditional thinking thatHeliobacteria and Firmicutes have diverged prior to the origin of ox-ygenic photosynthesis, the presence of 6C-enzymes in these specieswould indicate that their appearance pre-dates the emergence oflight-driven O2 production. Therefore, either the 6C cluster appearedwithout an apparent link to O2-tolerance or we need to rethink ourevolutionary history of oxygenic photosynthesis. The problem weare facing is not entirely new and has been raised previously basedon the presence of an additional cofactor also suspected to conferO2-tolerance in the Rieske/cytb complexes of Heliobacteria andFirmicutes [59]. Indeed, molecular phylogenies of several O2-depen-dent or O2-tolerant enzymes are in favor of a deeper than previouslythought ancestry of oxygenic photosynthesis (manuscript in prepara-tion). A more detailed discussion of this topic, however, is beyond thescope of the present article.

The expression of 6C-hydrogenases is now commonly thought tobe linked to hydrogen oxidation in the presence of O2 or even neces-sarily with O2 acting as terminal electron acceptor, i.e. to theKnallgas reaction. In addition to the special proximal cluster, how-ever, the 6C-enzymes studied so far stand out by high redox poten-tials of all three iron sulfur clusters [15,46] (Fig. 6), a property oftencorrelated to O2 serving as terminal acceptor. In several species, bycontrast, 6C-hydrogenases have been described to work under an-aerobic or microaerobic conditions using high-potential acceptorsother than O2, e.g. nitrogen oxides. The 6C Hyd-1 from E. coli isexpressed in anaerobic cultures, whereas there is evidence that itspresence is of importance when the cells experience microaerobicconditions [11,60]. E. coli deletion mutants containing only Hyd-1were active when grown on high-potential electron acceptors suchas nitrate and O2 [11,60]. The 6C-enzyme from T. roseopersicina(HupSL) is induced under anaerobic conditions and upon decreasingthe thiosulfate concentration during growth, which will lead to an in-crease in cellular redox potential [26] although the ultimate electron ac-ceptor is not known. In the obligate phototrophic Chlorobiaceae and

1573M-E. Pandelia et al. / Biochimica et Biophysica Acta 1817 (2012) 1565–1575

Heliobacteria, electrons produced from 6C-hydrogenases need to endup reducing the (relatively high-potential) photo-oxidized reactioncenter pigment.

From the presently available data it seems likely to us that en-zymes from the 6C clade are not only characterized by their specialproximal cluster, but also by a common redox up-shifted electron-transfer FeS chain and correspondingly high-potential redox partners(down the line to the terminal acceptor) (Fig. 6). Whereas the emer-gence of the 6C cluster was possibly driven by the appearance of O2,the redox upshift is certainly due to the use of high-potential terminalacceptors such as oxygen, nitrogen oxides or photosynthetic reactioncenters. Whether the common ancestor of all 6C-hydrogenases origi-nated in a “Knallgas”‐type, a denitrifying or a photosynthetic bioener-getic chain remains an open question.

Another open issue is how this ancestor was able to synthesize thesophisticated 6C [4Fe3S] cluster. For [NiFe] hydrogenases no specificmaturation genes for the assembly of the iron–sulfur clusters havebeen identified so far suggesting that biosynthesis of thesecofactors occurs via standard biosynthetic pathways (i.e. ISC, SUF)[61,62]. So, is the 6C cluster the result of a more specialized assemblyapparatus?

Recent studies on the “Knallgas” β-proteobacterial hydrogenasefrom Rhizobium leguminosarum employing deletion of accessorygenes suggested that hupGIJ (homologous to hoxORT in R. eutropha)[63] are involved in the maturation of the iron–sulfur cluster sub-unit [64]. More precisely, the hoxR gene of R. eutrophaMBH was pro-posed to be important for oxygen-tolerance and presumablyinvolved in the formation of the [4Fe3S] proximal cluster [63]. Intu-itively, this proposal is tempting but fails to explain the absence ofthe hoxR gene from all species having 6C-hydrogenases outside theα-, β-, γ-proteobacterial family. HoxR therefore cannot be essentialfor maturation of the 6C cluster. HoxR has an all-cysteine ligatedrubredoxin site, indicative of its possible role as a redox proteinagainst oxidative stress conditions (electron safety valve to protecteasily oxidizable centers). Since adaptation to higher oxygen levelsis probably a multi-step process, it seems more likely to us thatHoxR is rather involved in accrued protection of the 6C cluster (iflinked to the 6C cluster at all). The exclusive occurrence of hoxR inthe most recent branching orders of the proteobacteria corroboratesthis scenario.

3.4. The proximal cluster, a pivotal redox center in [NiFe]-hydrogenases

The previous section illustrates the outstanding role of the proximalcluster in modulating the overall properties of [NiFe]-hydrogenases.Due to its close spatial proximity to the [NiFe] center, the proximalcluster has in fact been discussed in the past as being to some degreepart of the catalytic core of the enzyme [65]. From this point of viewit is not astonishing that hydrogenases extensively experimentedwith its coordination pattern (Fig. 2, Table 1). The most frequent caseof the primary CxxC motif is the CTGC pattern. Two cases of archaealhydrogenases fromMethanosarcinaleshave replaced the leading coordi-nating cysteine with an aspartate (DSGC). A second, more commonmodification is the replacement of the second coordinating cysteineby an aspartate (i.e. CAGD, CTGD) or an asparagine (i.e. CEGN, CAGN).For these cases, there is no direct experimental evidence that a[4Fe4S] cluster is still formed. However, aspartate may act via its car-boxylic oxygen either as a mono- or a bi-dentate ligand and aspartateligation indeedoccurs in the [4Fe4S] ferredoxin from Pyrococcus furiosus[66] and the oxygen-sensing [4Fe4S] cluster of Bacillus subtilis [67].Asparagine ligation is rarer, although this residuemay also ligate throughits side-chain amide oxygen. The sole spectroscopically characterizedhydrogenase with a proximal cluster having an asparagine instead ofa cysteine, is the cyanobacterial-like enzyme from Acidithiobacillusferrooxidans [68]. The aerobically isolated A. ferrooxidans enzyme showssignals reminiscent of a proximal [4Fe4S] cluster magnetically

interacting with the [NiFe] site. The fact that the cluster is already para-magnetic (reduced) in the as-purified enzyme suggests a higher reduc-tion potential induced by the putative asparagine ligation. However,these signals were absent in spectra from whole cells, suggesting thepossibility of cluster damage during purification [68]. The fact that thishydrogenase shows optimal activity with relatively high-potential elec-tron acceptors indicates that the redox potentials of the iron–sulfur clus-ters are more positive than those of standard hydrogenases [15,34]. ACEGN motif for the proximal cluster may therefore be correlated with ahigher redox potential (similar to P. furiosus ferredoxin) [66].

There are, so far, no biochemical, biophysical or enzymatic datasuggesting specific roles for all these variants in Archaea and Bacteria.From the experiences with the 6C cluster, however, we would predictthat these deviant ligation patterns are functionally meaningful andbench-characterizations are thus much needed. Group 1 hydroge-nases are expected to still have a few more surprises in stock.

4. Conclusions

Group 1 hydrogenases are periplasmically oriented, in contrast tomembers of the other groups, and thus are exposed to the “outsideworld” with its variable oxidizing substrates and specific environmen-tal stress factors, the most obvious of which certainly being O2. Specificadaptations to these varying conditions rationalize Group 1's extensivediversification into subclades optimizing the enzyme for diversechemiosmotic chains. To modulate proton-coupled electron-transferprocesses and decrease O2-lability, Group 1 enzymes particularly ex-periment with the ligation of the proximal cluster, the cofactor closestto the catalytic site and at the interface of the two subunits. Their mostsuccessful evolutionary invention to this end is the 6C cluster, shownby mutation studies to be crucial for O2-tolerance. Its most notable pe-culiarity is a second redox transition promoted by redox-dependentstructural changes related to its surplus cysteine coordination. Themost logical role of the 6C cofactor's third redox state, rather than par-ticipating in catalysis, is that of an electron-safety valve helping to effi-ciently neutralize reactive oxidative species by releasing an additionalelectron. The 6C cluster correspondingly is widely distributed in hy-drogenases from many more recent phyla, demonstrating that this co-factor is not an odd phenomenon but a wide-spread means to copewith increasing oxidative conditions and most likely direct exposureto O2.

Acknowledgements

We would like to thank Pascale Infossi and Marie-Therese GiudiciOrticoni for helpful comments and Ines Pereira for stimulating discus-sions. This work has been supported by the Max-Planck Society andthe Solar-H2 program.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.bbabio.2012.04.012.

References

[1] N. Lane, J.F. Allen, W. Martin, How did LUCA make a living? Chemiosmosis in theorigin of life, Bioessays 32 (2010) 271–280.

[2] W. Nitschke, M.J. Russell, Hydrothermal focusing of chemical and chemiosmoticenergy, supported by delivery of catalytic Fe, Ni, Mo/W, Co, S and Se, forced lifeto emerge, J. Mol. Evol. 69 (2009) 481–496.

[3] A.L. Ducluzeau, R. van Lis, S. Duval, B. Schoepp-Cothenet, M.J. Russell, W. Nitschke,Was nitric oxide the first deep electron sink? Trends Biochem. Sci. 34 (2009)9–15.

[4] P.M. Vignais, B. Billoud, Occurrence, classification, and biological function of hy-drogenases: an overview, Chem. Rev. 107 (2007) 4206–4272.

[5] M. Brugna-Guiral, P. Tron, W. Nitschke, K.O. Stetter, B. Burlat, B. Guigliarelli, M.Bruschi, M.T. Giudici-Orticoni, [NiFe] hydrogenases from the hyperthermophilic

1574 M-E. Pandelia et al. / Biochimica et Biophysica Acta 1817 (2012) 1565–1575

bacterium Aquifex aeolicus: properties, function, and phylogenetics, Extremophiles 7(2003) 145–157.

[6] R.K. Thauer, A.K. Kaster, H. Seedorf, W. Buckel, R. Hedderich, Methanogenic archaea:ecologically relevant differences in energy conservation, Nat. Rev. Microbiol. 6(2008) 579–591.

[7] G. Voordouw, Evolution of hydrogenase genes, Adv. Inorg. Chem. 38 (1992) 397–422.[8] M.E. Pandelia, V. Fourmond, P. Tron-Infossi, E. Lojou, P. Bertrand, C. Leger, M.T.

Giudici-Orticoni, W. Lubitz, Membrane-Bound hydrogenase I from the hyperther-mophilic bacterium Aquifex aeolicus: enzyme activation, redox intermediates andoxygen tolerance, J. Am. Chem. Soc. 132 (2010) 6991–7004.

[9] G. Goldet, A.F. Wait, J.A. Cracknell, K.A. Vincent, M. Ludwig, O. Lenz, B. Friedrich,F.A. Armstrong, Hydrogen production under aerobic conditions by membrane-bound hydrogenases from Ralstonia species, J. Am. Chem. Soc. 130 (2008)11106–11113.

[10] Y. Shomura, K.S. Yoon, H. Nishihara, Y. Higuchi, Structural basis for a [4Fe–3S]cluster in the oxygen-tolerant membrane-bound [NiFe]-hydrogenase, Nature479 (2011) 253–256.

[11] M.J. Lukey, A. Parkin, M.M. Roessler, B.J. Murphy, J. Harmer, T. Palmer, F. Sargent,F.A. Armstrong, How Escherichia coli is equipped to oxidize hydrogen under dif-ferent redox conditions, J. Biol. Chem. 285 (2010) 3928–3938.

[12] A. Volbeda, P. Amara, C. Darnault, J.M. Mouesca, A. Parkin, M.M. Roessler, F.A.Armstrong, J.C. Fontecilla-Camps, X-ray crystallographic and computational stud-ies of the O2-tolerant [NiFe]-hydrogenase 1 from Escherichia coli, Proc. Natl. Acad.Sci. U. S. A. 109 (2012) 5305–5310.

[13] A. Volbeda, M.H. Charon, C. Piras, E.C. Hatchikian, M. Frey, J.C. FontecillaCamps,Crystal-structure of the nickel–iron hydrogenase from Desulfovibrio gigas, Nature373 (1995) 580–587.

[14] J. Fritsch, P. Scheerer, S. Frielingsdorf, S. Kroschinsky, B. Friedrich, O. Lenz, C.M.T.Spahn, The crystal structure of an oxygen-tolerant hydrogenase uncovers anovel iron–sulphur centre, Nature 479 (2011) 249–252.

[15] M.E. Pandelia, W. Nitschke, P. Infossi, M.T. Giudici-Orticoni, E. Bill, W. Lubitz,Characterization of a unique [FeS] cluster in the electron transfer chain of the ox-ygen tolerant [NiFe] hydrogenase from Aquifex aeolicus, Proc. Natl. Acad. Sci. U. S.A. 108 (2011) 6097–6102.

[16] T. Goris, A.F. Wait, M. Saggu, J. Fritsch, N. Heidary, M. Stein, I. Zebger, F. Lendzian,F.A. Armstrong, B. Friedrich, O. Lenz, A unique iron–sulfur cluster is crucial for ox-ygen tolerance of a [NiFe]-hydrogenase, Nat. Chem. Biol. 7 (2011) (p. 310-U87).

[17] M.J. Lukey, M.M. Roessler, A. Parkin, R.M. Evans, R.A. Davies, O. Lenz, B. Friedrich,F. Sargent, F.A. Armstrong, Oxygen-tolerant [NiFe]-hydrogenases: the individualand collective importance of supernumerary cysteines at the proximal Fe–S clus-ter, J. Am. Chem. Soc. 133 (2011) 16881–16892.

[18] S.F. Altschul, W. Gish, W. Miller, E.W. Myers, D.J. Lipman, Basic local alignmentsearch tool, J. Mol. Biol. 215 (1990) 403–410.

[19] K. Tamura, D. Peterson, N. Peterson, G. Stecher, M. Nei, S. Kumar, MEGA5: molec-ular evolutionary genetics analysis using maximum likelihood, evolutionary dis-tance, and maximum parsimony methods, Mol. Biol. Evol. 28 (2011) 2731–2739.

[20] H.M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T.N. Bhat, H. Weissig, I.N.Shindyalov, P.E. Bourne, The protein data bank, Nucleic Acids Res. 28 (2000)235–242.

[21] N.R. Hackett, P.D. Bragg, The association of 2 distinct b-cytochromes with the re-spiratory nitrate reductase of Escherichia coli, FEMS Microbiol. Lett. 13 (1982)213–217.

[22] A.L. Ballard, S.J. Ferguson, Respiratory nitrate reductase from Paracoccusdenitrificans — evidence for 2 b-type hemes in the gamma-subunit and propertiesof a water-soluble active enzyme containing alpha-subunit and beta-subunit, Eur.J. Biochem. 174 (1988) 207–212.

[23] P. Infossi, E. Lojou, J.P. Chauvin, G. Herbette, M. Brugna, M.T. Giudici-Orticoni,Aquifex aeolicus membrane hydrogenase for hydrogen biooxidation: role of lipidsand physiological partners in enzyme stability and activity, Int. J. Hydrogen Ener-gy 35 (2010) 10778–10789.

[24] R. Gross, J. Simon, F. Theis, A. Kroger, Two membrane anchors of Wolinellasuccinogenes hydrogenase and their function in fumarate and polysulfide respira-tion, Arch. Microbiol. 170 (1998) 50–58.

[25] M. Jormakka, S. Tornroth, B. Byrne, S. Iwata, Molecular basis of proton motiveforce generation: structure of formate dehydrogenase-N, Science 295 (2002)1863–1868.

[26] L.S. Palagyi-Meszaros, J. Maroti, D. Latinovics, T. Balogh, E. Klement, K.F.Medzihradszky, G. Rakhely, K.L. Kovacs, Electron-transfer subunits of the NiFe hy-drogenases in Thiocapsa roseopersicina BBS, FEBS J. 276 (2009) 164–174.

[27] F.D. Ciccarelli, T. Doerks, C. von Mering, C.J. Creevey, B. Snel, P. Bork, Toward au-tomatic reconstruction of a highly resolved tree of life, Science 311 (2006)1283–1287.

[28] F. Baymann, E. Lebrun, M. Brugna, B. Schoepp-Cothenet, M.T. Giudici-Orticoni, W.Nitschke, The redox protein construction kit: pre-last universal common ancestorevolution of energy-conserving enzymes, Philos. Trans. R. Soc. Lond. B Biol. Sci.358 (2003) 267–274.

[29] M. Jormakka, K. Yokoyama, T. Yano, M. Tamakoshi, S. Akimoto, T. Shimamura, P.Curmi, S. Iwata, Molecular mechanism of energy conservation in polysulfide res-piration, Nat. Struct. Mol. Biol. 15 (2008) 730–737.

[30] A. Dubini, R.L. Pye, R.L. Jack, T. Palmer, F. Sargent, How bacteria get energy fromhydrogen: a genetic analysis of periplasmic hydrogen oxidation in Escherichiacoli, Int. J. Hydrogen Energy 27 (2002) 1413–1420.

[31] N.K. Menon, C.Y. Chatelus, M. Dervartanian, J.C. Wendt, K.T. Shanmugam, H.D.Peck, A.E. Przybyla, Cloning, sequencing, and mutational analysis of the Hyb-operon encoding Escherichia coli hydrogenase-2, J. Bacteriol. 176 (1994)4416–4423.

[32] B. Schoepp-Cothenet, C. Lieutaud, F. Baymann, A. Vermeglio, T. Friedrich, D.M.Kramer, W. Nitschke, Menaquinone as pool quinone in a purple bacterium,Proc. Natl. Acad. Sci. U. S. A. 106 (2009) 8549–8554.

[33] T.V. Laurinavichene, A.A. Tsygankov, The involvement of hydrogenases 1 and 2 inthe hydrogen-dependent nitrate respiration of Escherichia coli, Microbiology 72(2003) 654–659.

[34] M. Teixeira, I. Moura, A.V. Xavier, J.J.G. Moura, J. LeGall, D.V. Der Vartanian, H.D.Peck, B.H. Huynh, Redox intermediates of Desulfovibrio gigas [NiFe] hydrogenasegenerated under hydrogen — Mössbauer and EPR characterization of the metalcenters, J. Biol. Chem. 264 (1989) 16435–16450.

[35] H. Miyadera, A. Hiraishi, H. Miyoshi, K. Sakamoto, R. Mineki, K. Murayama, K.V.P.Nagashima, K. Matsuura, S. Kojima, K. Kita, Complex II from phototrophic purplebacterium Rhodoferax fermentans displays rhodoquinol-fumarate reductase activ-ity, Eur. J. Biochem. 270 (2003) 1863–1874.

[36] B.C. Brajcich, A.L. Iarocci, L.A.G. Johnstone, R.K. Morgan, Z.T. Lonjers, M.J. Hotchko,J.D. Muhs, A. Kieffer, B.J. Reynolds, S.M. Mandel, B.N. Marbois, C.F. Clarke, J.N.Shepherd, Evidence that ubiquinone is a required intermediate for rhodoquinonebiosynthesis in Rhodospirillum rubrum, J. Bacteriol. 192 (2010) 436–445.

[37] G.J. Mander, E.C. Duin, D. Linder, K.O. Stetter, R. Hedderich, Purification and char-acterization of a membrane-bound enzyme complex from the sulfate-reducingarchaeon Archaeoglobus fulgidus related to heterodisulfide reductase frommethanogenic archaea, Eur. J. Biochem. 269 (2002) 1895–1904.

[38] F. Grein, I.A.C. Pereira, C. Dahl, Biochemical characterization of individualcomponents of the Allochromatium vinosum DsrMKJOP transmembrane complexaids understanding of complex function in vivo, J. Bacteriol. 192 (2010)6369–6377.

[39] M.G. Bertero, R.A. Rothery, M. Palak, C. Hou, D. Lim, F. Blasco, J.H. Weiner, N.C.J.Strynadka, Insights into the respiratory electron transfer pathway from the struc-ture of nitrate reductase A, Nat. Struct. Biol. 10 (2003) 681–687.

[40] R. Hedderich, N. Hamann, M. Bennati, Heterodisulfide reductase frommethanogenic archaea: a new catalytic role for an iron–sulfur cluster, Biol.Chem. 386 (2005) 961–970.

[41] R.H. Pires, S.S. Venceslau, F. Morais, M. Teixeira, A.V. Xavier, I.A.C. Pereira, Charac-terization of the Desulfovibrio desulfuricans ATCC 27774 DsrMKJOP complex — amembrane-bound redox complex involved in the sulfate respiratory pathway,Biochemistry 45 (2006) 249–262.

[42] N. Hamann, G.J. Mander, J.E. Shokes, R.A. Scott, M. Bennati, R. Hedderich, Cyste-ine-rich CCG domain contains a novel [4Fe–4S] cluster binding motif asdeduced from studies with subunit B of heterodisulfide reductase fromMethanothermobacter marburgensis, Biochemistry 46 (2007) 12875–12885.

[43] J.M.C.C. Coremans, C.J. Vangarderen, S.P.J. Albracht, On the redox equilibrium be-tween H-2 and hydrogenase, Biochim. Biophys. Acta 1119 (1992) 148–156.

[44] H. Ogata, P. Kellers, W. Lubitz, The crystal structure of the [NiFe] hydrogenasefrom the photosynthetic bacterium Allochromatium vinosum: characterization ofthe oxidized enzyme (Ni-A state), J. Mol. Biol. 402 (2010) 428–444.

[45] B. Bleijlevens, F.A. van Broekhuizen, A.L. De Lacey, W. Roseboom, V.M. Fernandez,S.P.J. Albracht, The activation of the [NiFe]-hydrogenase from Allochromatiumvinosum. An infrared spectro-electrochemical study, J. Biol. Inorg. Chem. 9(2004) 743–752.

[46] K. Knüttel, K. Schneider, A. Erkens, W. Plass, A. Muller, E. Bill, A.X. Trautwein,Redox properties of the metal centers in the membrane-bound hydrogenasefrom Alcaligenes eutrophus CH34, Bull. Pol. Acad. Sci., Chem. 42 (1994) 495–511.

[47] M. Tietze, A. Beuchle, I. Lamla, N. Orth, M. Dehler, G. Greiner, U. Beifuss, Redox po-tentials of methanophenazine and CoB-S-S-CoM, factors involved in electrontransport in methanogenic archaea, Chem. Biochem. 4 (2003) 333–335.

[48] R. Gross, R. Pisa, M. Sanger, C.R.D. Lancaster, J. Simon, Characterization of themenaquinone reduction site in the diheme cytochrome b membrane anchor ofWolinella succinogenes NiFe-hydrogenase, J. Biol. Chem. 279 (2004) 274–281.

[49] U. Beifuss, M. Tietze, S. Baumer, U. Deppenmeier, Methanophenazine: structure,total synthesis, and function of a new cofactor from methanogenic archaea,Angew. Chem. Int. Ed. 39 (2000) 2470.

[50] W.A. Vargas, P.D. Weyman, Y.K. Tong, H.O. Smith, Q. Xu, [NiFe] Hydrogenase fromAlteromonas macleodii with unusual stability in the presence of oxygen and hightemperature, Appl. Environ. Microbiol. 77 (2011) 1990–1998.

[51] K. Schneider, D.S. Patil, R. Cammack, Electron-spin-resonance properties ofmembrane-bound hydrogenases from aerobic hydrogen bacteria, Biochim.Biophys. Acta 748 (1983) 353–361.

[52] K. Knüttel, K. Schneider, H.G. Schlegel, A. Muller, The membrane-bound hydroge-nase from Paracoccus denitrificans — purification and molecular characterization,Eur. J. Biochem. 179 (1989) 101–108.

[53] R. Cammack, V.M. Fernandez, E.C. Hatchikian, Nickel–iron hydrogenase, Inorg.Microb. Sulfur Metabol. 243 (1994) 43–68.

[54] J.H. Sun, D.J. Arp, Aerobically purified hydrogenase from Azotobacter vinelandii —activity, activation, and spectral properties, Arch. Biochem. Biophys. 287 (1991)225–233.

[55] K.S. Yoon, N. Tsukada, Y. Sakai, M. Ishii, Y. Igarashi, H. Nishihara, Isolation andcharacterization of a new facultatively autotrophic hydrogen-oxidizingbetaproteobacterium, Hydrogenophaga sp AH-24, FEMS Microbiol. Lett. 278(2008) 94–100.

[56] K.S. Yoon, K. Fukuda, K. Fujisawa, H. Nishihara, Purification and characterizationof a highly thermostable, oxygen-resistant, respiratory [NiFe]-hydrogenase froma marine, aerobic hydrogen-oxidizing bacterium Hydrogenovibrio marinus, Int. J.Hydrogen Energy 36 (2011) 7081–7088.

[57] M. Saggu, I. Zebger, M. Ludwig, O. Lenz, B. Friedrich, P. Hildebrandt, F. Lendzian,Spectroscopic insights into the oxygen-tolerant membrane-associated [NiFe] hy-drogenase of Ralstonia eutropha H16, J. Biol. Chem. 284 (2009) 16264–16276.

1575M-E. Pandelia et al. / Biochimica et Biophysica Acta 1817 (2012) 1565–1575

[58] B. Boussau, L. Gueguen, M. Gouy, Accounting for horizontal gene transfers ex-plains conflicting hypotheses regarding the position of aquificales in the phylog-eny of bacteria, BMC Evol. Biol. 8 (2008) 272.

[59] W. Nitschke, R. van Lis, B. Schoepp-Cothenet, F. Baymann, The “green” phylo-genetic clade of Rieske/cytb complexes, Photosynth. Res. 104 (2010)347–355.

[60] T.V. Laurinavichene, A.A. Tsygankov, H-2 consumption by Escherichia coli coupledvia hydrogenase 1 or hydrogenase 2 to different terminal electron acceptors,FEMS Microbiol. Lett. 202 (2001) 121–124.

[61] E.M. Shepard, E.S. Boyd, J.B. Broderick, J.W. Peters, Biosynthesis of complex iron–sulfur enzymes, Curr. Opin. Chem. Biol. 15 (2011) 319–327.

[62] M. Fontecave, S. Ollagnier-de-Choudens, Iron–sulfur cluster biosynthesis in bac-teria: mechanisms of cluster assembly and transfer, Arch. Biochem. Biophys.474 (2008) 226–237.

[63] J. Fritsch, O. Lenz, B. Friedrich, The maturation factors HoxR and HoxT contributeto oxygen tolerance of membrane-bound [NiFe] hydrogenase in Ralstoniaeutropha H16, J. Bacteriol. 193 (2011) 2487–2497.

[64] B.M. Wolfe, S.M. Lui, H. Manyani, L. Rey, J.M. Palacios, J. Imperial, T. Ruiz-Argueso,Gene products of the hupGHIJ Operon are involved in maturation of the iron–sulfur subunit of the [NiFe] hydrogenase from Rhizobium leguminosarum bv.viciae, J. Bacteriol. 187 (2005) 7018–7026.

[65] L.A. Sazanov, P. Hinchliffe, Structure of the hydrophilic domain of respiratorycomplex I from Thermus thermophilus, Science 311 (2006) 1430–1436.

[66] S. Sham, L. Calzolai, P.L. Wang, K. Bren, H. Haarklau, P.S. Brereton, M.W.W. Adams,G.N. La Mar, A solution NMR molecular model for the aspartate-ligated, cubanecluster containing ferredoxin from the hyperthermophilic archeaon Pyrococcusfuriosus, Biochemistry 41 (2002) 12498–12508.

[67] I. Gruner, C. Fradrich, L.H. Bottger, A.X. Trautwein, D. Jahn, E. Hartig, Aspartate 141is the fourth ligand of the oxygen-sensing [4Fe–4S](2+) cluster of Bacillus subtilistranscriptional regulator Fnr, J. Biol. Chem. 286 (2011) 2017–2021.

[68] O. Schröder, B. Bleijlevens, T.E. de Jongh, Z.J. Chen, T.S. Li, J. Fischer, J. Forster, C.G.Friedrich, K.A. Bagley, S.P.J. Albracht, W. Lubitz, Characterization of a cyanobacterial-like uptake [NiFe] hydrogenase: EPR and FTIR spectroscopic studies of the enzymefrom Acidithiobacillus ferrooxidans, J. Biol. Inorg. Chem. 12 (2007) 212–233.