Toward a mechanistic and physiological understanding of a ferredoxin:disulfide reductase from the domains Archaea and Bacteria

Disulfide reductases reduce other proteins and are critically important for cellular redox signaling and homeostasis. Methanosarcina acetivorans is a methane-producing microbe from the domain Archaea that produces a ferredoxin:disulfide reductase (FDR) for which the crystal structure has been reported, yet its biochemical mechanism and physiological substrates are unknown. FDR and the extensively characterized plant-type ferredoxin:thioredoxin reductase (FTR) belong to a distinct class of disulfide reductases that contain a unique active-site [4Fe-4S] cluster. The results reported here support a mechanism for FDR similar to that reported for FTR with notable exceptions. Unlike FTR, FDR contains a rubredoxin [1Fe-0S] center postulated to mediate electron transfer from ferredoxin to the active-site [4Fe-4S] cluster. UV-visible, EPR, and Mössbauer spectroscopic data indicated that two-electron reduction of the active-site disulfide in FDR involves a one-electron-reduced [4Fe-4S]1+ intermediate previously hypothesized for FTR. Our results support a role for an active-site tyrosine in FDR that occupies the equivalent position of an essential histidine in the active site of FTR. Of note, one of seven Trxs encoded in the genome (Trx5) and methanoredoxin, a glutaredoxin-like enzyme from M. acetivorans, were reduced by FDR, advancing the physiological understanding of FDR's role in the redox metabolism of methanoarchaea. Finally, bioinformatics analyses show that FDR homologs are widespread in diverse microbes from the domain Bacteria.

Disulfide reductases are universal in nature, where they interact with protein and small molecule substrates required for redox control and response to oxidative stress. Disulfide reductases include thioredoxin reductase (TrxR), 7 an NAD(P)H-dependent flavoenzyme containing an active-site disulfide that transfers reducing equivalents from the flavin to the disulfide of thioredoxin (Trx) by a disulfide-exchange mechanism (1,2). However, plant ferredoxin:thioredoxin reductase (FTR) is devoid of flavin and contains a novel active-site [Fe 4 S 4 ] cluster (3,4). Methanosarcina acetivorans, classified in the domain Archaea, produces a planttype ferredoxin:disulfide reductase (FDR) also devoid of flavin that contains an active-site [Fe 4 S 4 ] cluster revealed by the crystal structure (5). FDR is representative of a diverse family of disulfide reductases proposed to have evolved from an ancestral plant-type FTR catalytic subunit to meet a variety of specific ecological needs (6,37).
Preliminary characterization of FDR involved ferredoxin (Fdx)-dependent reduction of GSSG, a nonphysiological substrate absent in methane-producing species from the domain Archaea (methanoarchaea) (5). Although FDR homologs are widely distributed among diverse methanoarchaea, the physiological redox substrates are unknown (6,7). Trx is a potential physiological substrate for FDR from M. acetivorans, as it encodes seven Trx homologs (Trx1-Trx7) and a NADPH-dependent TrxR shown to reduce only Trx7 (8). However, NADP is not a primary electron carrier in methanoarchaea. Instead, the primary electron carriers are coenzyme F 420 (F 420 ) and Fdx, underscoring the physiological importance of the Fdx-dependent FDR.
Trxs are small redox proteins that reduce the disulfide bonds of proteins that are key to controlling a diverse array of essential processes in organisms from the domains Bacteria and Eukarya (1). Remarkably, little is known of the biochemistry and physi-ology of Trxs that are widespread in the domain Archaea, which includes diverse methanoarchaea (6 -10). The methanoarchaea play an essential role in the global carbon cycle. They are terminal organisms of microbial food chains, converting complex biomass to methane in a diverse set of anaerobic environments, annually producing nearly one billion tons of methane that contributes significantly to the greenhouse effect (11). Thus, it is of ecological importance to understand the physiology and biochemistry of redox control and response to oxidative stress in methanoarchaea.
A comparison of the crystal structures for FDR and FTR ( Fig.  1) reveals similarities and differences that necessitate a more comprehensive biochemical investigation of FDR. Both are devoid of flavin and contain a [4Fe-4S] cluster adjacent to the active-site disulfides in FDR (Cys 52 -S-S-Cys 84 ) and FTR (Cys 57 -S-S-Cys 87 ) (3,5). FTR is a ␣␤-heterodimer comprising a variable subunit and a catalytic subunit. The overall structure of FTR (Fig. 1A) is a concave disk measuring only 10 Å across the center, where the active-site [4Fe-4S] cluster and adjacent disulfide (Cys 57 -S-S-Cys 87 ) are located on opposite sides (5,12). The [4Fe-4S] cluster of FTR is proximal to the Fdx-binding site, and the disulfide is proximal to the substrate-binding site on the opposite side (Fig. 1). In contrast to FTR, FDR is a homodimer wherein the N-terminal domains contain the [4Fe-4S] cluster. Furthermore, the C-terminal domain has no significant sequence identity to the variable subunit of the heterodimeric FTR and contains a rubredoxin [1Fe-0S] center absent in FTR. The [1Fe-0S] center of one protomer is proximal to the [4Fe-4S] cluster of the other protomer in head-to-tail configuration of N-and C-terminal domains ( Fig. 1) (5). The [1Fe-0S] center is positioned on the side of the [4Fe-4S] cluster opposite the disulfide (Cys 52 -S-S-Cys 84 ), suggesting that the [1Fe-0S] center mediates electron transfer from Fdx to the [4Fe-4S] cluster (5). The variable subunit of FTR is apparently required for stability of the active-site region, as the catalytic subunit cannot be produced independently (4). Thus, it was proposed that the C-terminal domain of FDR has a similar stabilizing function in addition to a role for the [1Fe-0S] center in mediating electron transfer from Fdx to the active site (5). Similar to FTR, the crystal structure of FDR ( Fig. 1) indicates that the [4Fe-4S] cluster contains a unique iron ion coordinated by the sulfur atom of Cys 50 (5). The unique iron and the sulfur atom of Cys 50 are in van der Waals contact (2.9 and 3.1 Å, respectively) with the sulfur atom of Cys 84 that forms an active-site disulfide with Cys 52 . This arrangement results in a distorted tetrahedral iron site with a 3 -S-Fe-S(Cys 50 ) angle of 141°that opens toward the active-site disulfide analogous to FTR (Fig. 1). It is proposed that His 86 in FTR shuttles a proton from Cys 87 to substrate; however, catalytically active WT FDR contains Tyr 83 in the equivalent position of His 86 , identifying yet another anomaly between FTR and FDR (5,14).
Although preliminary characterization of FDR led to a proposed catalytic mechanism, it was largely based on activity with the surrogate substrate GSSG (5). Unlike FDR, FTR has been extensively characterized, leading to a proposed catalytic mechanism (Fig. 2) that is a guide for a more detailed investigation of FDR and comparison with FTR (3,14). In the resting state (structure A), a unique iron site of the [4Fe-4S] 2ϩ cluster  (12). Fdx, the catalytic subunit of FTR, the variable subunit of FTR, and Trx are colored blue, beige, brown, and green, respectively. The [4Fe-4S] cluster is shown in stick representation. Fdx and Trx interact exclusively with the catalytic subunit of FTR in opposite directions. B, active-site structure of FTR. Reproduced with permission (5). The active-site disulfide formed by Cys 57 and Cys 87 is contained in the red circle. C, crystal structure of the FDR dimer. Monomers are shown in red and blue. The iron and sulfur atoms of the [4Fe-4S] clusters and [1Fe-0S] centers are shown in a red and yellow ball representation, respectively. D, active-site structure of FDR dimer showing the adjacent C-and N-terminal domains containing the [1Fe-0S] centers and [4Fe4S] clusters, respectively. The active-site disulfide formed by Cys 52 and Cys 84 is contained in the red circle. Color codes for active-site structures of FTR (B) and FDR (D) are as follows: iron (green), sulfur (yellow), carbon (gray), nitrogen (blue), and oxygen (red). The Fe-S and S-S interactions are shown in black broken lines with distances in ångstroms. The angles bisecting the S-S interactions in B and D are 128.8 and 140.9°for FTR and FDR, respectively (see "Results").

Ferredoxin:disulfide reductase
in FTR weakly interacts with the sulfur of Cys 87 that forms the active-site disulfide with Cys 57 . The interaction is supported by the 3 -S-Fe-S(Cys 55 ) angle of 129°and increased Fe(II) character at the unique iron revealed by Mössbauer spectroscopy (15). In the mechanism proposed for FTR, one-electron reduction of structure A yields structure B, wherein the active-site disulfide is reduced by two electrons to the dithiol form with the thiolate of Cys 87 ligating the unique iron site, and the [4Fe-4S] 2ϩ cluster is oxidized to the EPR-active [4Fe-4S] 3ϩ form with S ϭ 1 ⁄ 2 ground state (16). Structure B has been observed as a transient intermediate in the native reaction, but it can be trapped by modification of FTR with N-ethylmaleimide (NEM) (14,16,17). The next steps in the reaction mechanism converting structure B to structure H involve transfer of the second elec-tron and attack by the thiol of Cys 57 on the substrate disulfide. The reaction is proposed to proceed by two possible mechanisms that depend on the order of these two steps involving structures F or G (12,(17)(18)(19). In structure B, the sulfur of Cys 87 ligates with the unique iron of the [4Fe-4S] 3ϩ cluster, thereby freeing the thiol of Cys 57 for nucleophilic attack on thioredoxin. In structure F, it is proposed that Cys 87 is protonated and anchored to the [4Fe-4S] 2ϩ cluster by a strong hydrogen bond to the sulfur atom of Cys 55 , promoting charge buildup on the unique iron consistent with the observed electron-rich [4Fe-4S] 2ϩ cluster and a partially valence-localized Fe(II)Fe(III) pair (17). It is proposed that His 86 plays an essential role in protonation of the cluster-interacting Cys 87 thiol based on nearly complete loss of activity for the H86Y variant and loss of site- Ferredoxin:disulfide reductase differentiated character of the unique iron site (14,18). Anchoring Cys 87 to the [4Fe-4S] 2ϩ cluster exposes Cys 57 free to attack substrate. In the final step for both mechanisms, disulfide/dithiol exchange leads to regeneration of the active-site disulfide (Cys 57 -S-S-Cys 87 ) on FTR (structure A) and release of the reduced substrate.
The mechanism of formation of the one-electron reduced intermediate (structure B) is unknown, although it is of general interest in understanding the site-specific chemistry of unique Fe-S clusters revealed in FDR and the recent crystal structure of the coenzyme M/coenzyme B (CoM-S-S-CoB) heterodisulfide reductase of methanoarchaea (20). Several mechanisms can be envisaged for the conversion of structure A to structure B, which have been hypothesized previously for FTR (3). First, the conversion could be concerted. Second, the conversion could involve initial reduction of the resting-state [4Fe-4S] 2ϩ cluster to the [4Fe-4S] ϩ form (structure C), followed by either a twoelectron attack on the sulfur of Cys 87 and cleavage of the activesite disulfide (Cys 57 -S-S-Cys 87 ), yielding structure B, or by oneelectron reductive cleavage of the disulfide bond by the reduced [4Fe-4S] 1ϩ , yielding the thiyl/thiol intermediate D, which could further undergo an electron transfer from the [4Fe-4S] 2ϩ cluster to the thiyl radical to yield structure B. The latter process may be similar to the reductive cleavage of S-adenosylmethionine to yield the 5Ј-deoxyadenos-5Ј-yl radical observed for radical S-adenosylmethionine enzymes (21). Alternatively, structure C could undergo disulfide exchange to yield structure E en route to the one-electron reduced intermediate B. Last, it is proposed that the initial electron is transferred to the disulfide moiety of A (Cys 57 -S-S-Cys 87 ), yielding the thiyl radicalcontaining structure D, followed by electron transfer from the [4Fe-4S] 2ϩ cluster to the thiyl radical, yielding structure B.
As discussed, FDR and FTR originate from different phylogenetic domains and differ in important structural features, predicting physiological and mechanistic variances not yet understood. Here, we present results of UV-visible, EPR, and Mössbauer spectroscopic approaches that further characterize FDR and address the role of [4Fe-4S] intermediates previously hypothesized for the mechanism of FTR. Results are also presented addressing roles for the active-site Tyr 83 , the C-terminal domain, and the [1Fe-0S] center of FDR that collectively sup-  Ferredoxin:disulfide reductase port a revised mechanism. Finally, we report substrates for FDR that expand an understanding of the physiological role for FDR in diverse methanoarchaea.

Protein purification
His 6 -tagged WT FDR and variants were produced in Escherichia coli strain ⌬iscR and anaerobically purified to electrophoretic homogeneity (Fig. S1). Gel filtration chromatography revealed that all proteins were dimeric as previously described for WT (5). The WT and Y83H variant contained 9.2 Ϯ 0.6 (n ϭ 3) and 9.7. Ϯ 0.6 (n ϭ 3) iron per dimer. The WT purified from E. coli cultured with 57 Fe contained 9.0 Ϯ 0.1 (n ϭ 3) iron per dimer. These results are consistent with one [4Fe-4S] cluster and one [1Fe-0S] center per monomer as revealed by the crystal structure ( Fig. 1) (5). As-purified WT FDR was exposed to air immediately before analyses to ensure a full complement of the oxidized [4Fe-4S] 2ϩ cluster (structure A, Fig. 2), hereafter referred to as the resting-state WT.

Spectroscopy of resting-state WT
As FDR and FTR derive from different phylogenetic domains and are structurally distinct, it is imperative that characterization of FDR track experimental approaches for FTR to determine similarities and differences. Thus, FDR was characterized spectroscopically beginning with the resting-state WT. The UV-visible spectrum showed a broad absorbance band similar to that described for the oxidized [4Fe-4S] 2ϩ cluster of FTR albeit with an A max of 404 nm (Fig. 3A) versus 410 nm for FTR (14). EPR spectroscopy (Fig. 4A) revealed effective g values of 9.63 ("M S ϭ 1 ⁄ 2" doublet of S ϭ 5 ⁄ 2), 4.81, and 4.34 ("M S ϭ 3 ⁄ 2" doublet of S ϭ 5 ⁄ 2), corresponding to a [1Fe-0S] 3ϩ center similar to rubredoxins (22,23). The EPR results, together with the previous crystal structure, establish the presence of a rubredoxin cluster in FDR.
FDR was investigated with Mössbauer spectroscopy to confirm and further characterize the [1Fe-0S] 3ϩ center and the apparent unique iron site of the [4Fe-4S] 2ϩ cluster as suggested by the crystal structure (Fig. 1). The 4.2-K spectrum (Fig. 5A), recorded in a weak magnetic field of 53 mT applied parallel to

Ferredoxin:disulfide reductase
the ␥-beam, confirmed the presence of an oxidized [4Fe-4S] 2ϩ cluster and a [1Fe-0S] 3ϩ center. The spectrum is dominated by quadrupole doublet features with a prominent shoulder on the high-energy side of the high-energy line, along with a small amount of a magnetically split component with parameters consistent with the [1Fe-0S] 3ϩ center observed by EPR (ϳ8% of total iron, black line). The dominant feature can be interpreted as the superposition of three quadrupole doublet components with an intensity ratio of 2:1:1 corresponding to three different iron sites, respectively (ϳ95% of total iron, yellow line). The simulation parameters (Table S1) are almost identical to those reported for the oxidized resting state of FTR (14). Site 1 exhibits parameters typical of [4Fe-4S] 2ϩ clusters having two valence-delocalized (Fe 2.5 ) 2 dimers (24,25). However, sites 2 and 3 have parameters indicative of a partially valence-localized Fe 2ϩ Fe 3ϩ pair (14). Site 3 with its larger isomer shift and quadrupole splitting is characteristic of the unique iron site of the [4Fe-4S] 2ϩ cluster of FTR, a result establishing the presence of a comparable unique iron ion in the [4Fe-4S] 2ϩ cluster of FDR (15,17). The crystal structure of FDR shows the unique iron that is in contact with the sulfur atoms of Cys 50 and Cys 84 (Fig. 1).

Spectroscopy of NEM-modified WT
The mechanism proposed for FTR includes a transient oneelectron-reduced intermediate (structure B, Fig. 2) with a [4Fe-4S] 3ϩ cluster stabilized by NEM modification of the reactive Cys 57 thiol (14,16,17). Thus, WT FDR was modified with NEM (NEM-WT) comparable with FTR to assess the role of structure B in the catalytic mechanism of FDR. The UV-visible spectrum of oxidized samples revealed an A max of 425 nm (Fig. 3B), and low-temperature EPR spectroscopy (Fig. 6B) showed nearly axial resonances (g ϭ 2.11, 1.996, 1.991), indicating an S ϭ 1 ⁄ 2 [4Fe-4S] 3ϩ cluster as reported for FTR (14). Dithionite reduction produced a UV-visible spectrum with the A max shifted to 404 nm (Fig. 3B), as reported for NEM-modified FTR, consistent with one-electron reduction of the [4Fe-4S] 3ϩ cluster to the two-electron-reduced state containing a [4Fe-4S] 2ϩ cluster (structure F) and the active-site disulfide (Cys 52 -S-S-Cys 84 ) reduced for exchange with substrate (14). The results support a mechanism for FDR in which the [4Fe-4S] 3ϩ cluster is a central intermediate (structure B, Fig. 2).
Further evidence for the [4Fe-4S] 3ϩ intermediate was obtained by Mössbauer spectroscopy of NEM-WT ( Fig. 5 (B and C) and Table S1), which revealed several spectral components. The majority of the iron (ϳ55%, yellow line) is in the site-differentiated [4Fe-4S] 2ϩ cluster with S ϭ 0 ground state observed in the nonmodified resting-state WT. The features of the ferric rubredoxin (ϳ8% of total iron, black line) are also present, consistent with the low-temperature EPR spectrum, along with a small amount (ϳ5%, green line) of adventitiously bound high-spin ferric iron, which is probably due to minor cluster degradation over the course of sample preparation. In addition, ϳ38% of the sample iron gives rise to magnetically split spectra in 53-mT and 8-T fields, suggesting that they emanate from an EPR-active cluster. Because EPR spectroscopy provides evidence for a [4Fe-4S] 3ϩ cluster, we have simulated this component with parameters similar to those previously reported for the [4Fe-4S] 3ϩ cluster in NEM-modified FTR (blue line) (14,15). The Mössbauer results are consistent with the presence of a [4Fe-4S] 3ϩ cluster (structure B, Fig. 2) and indicate that this species plays a role in the FDR mechanism.    (5). Dye-mediated redox titrations of NEM-WT at pH 7.0 were monitored by EPR spectroscopy (Fig.  S2). The data fitted with the Nernst equation were in good agreement with a one-electron (n ϭ1) reduction process. Midpoint potentials relative to the normal hydrogen electrode were determined to be Ϫ45.7 Ϯ 1.5 and Ϫ46.9 Ϯ 2.6 mV for the [4Fe-4S] 3ϩ/2ϩ and [1Fe-0S] 2ϩ/1ϩ redox couples, respectively. The result for the [4Fe-4S] 3ϩ/2ϩ couple contrasts sharply with the Ϫ145 Ϯ 10 mV (pH 7.0) midpoint potential for the FTR from Synechocystis (17).

Spectroscopy of reduced resting-state WT FDR
Reduction of the EPR-silent resting state (A) to the one-electron-reduced intermediate (B) was hypothesized to involve one-electron-reduced intermediate structure D, C, or E in the mechanism proposed for FTR (Fig. 2) (3). Thus, roles for these intermediates in the mechanism for FDR were investigated spectroscopically. The UV-visible spectrum of resting-state WT FDR was bleached by the addition of dithionite (Fig. 3A), indicating reduction of the [4Fe-4S] 2ϩ cluster. The low-temperature EPR spectrum at 16 K recorded 1 min after the addition of dithionite showed a set of resonances with g values of 2.03, 1.99, and 1.89 that were observed only up to 30 K and without significant broadening (Fig. 6A). This result is characteristic of a [4Fe-4S] 1ϩ cluster previously hypothesized for intermediate structures C and E in the mechanism proposed for FTR (Fig. 2) (Fig. 2) that was further investigated by incubating resting-state WT with excess dithionite and the samples frozen at time intervals for EPR spectroscopy (Fig. S3). At 1 min, the [4Fe-4S] 1ϩ signal accounted for ϳ60% of iron in the sample (Fig. 7). At 30 min, the [4Fe-4S] 1ϩ signal intensity decreased to ϳ5% of the starting signal without the appearance of new signals, including the S ϭ 1 ⁄ 2 [4Fe-4S] 3ϩ EPR signal of structure B. The results provide additional evidence for intermediate structure C or E in the mechanism for FDR. Although the S ϭ 1 ⁄ 2 [4Fe-4S] 3ϩ EPR signal of structure B was not detected, formation as a short-lived transient one-electron-reduced intermediate during reduction of the resting-state [4Fe-4S] 2ϩ cluster of structure A to the two-electron-reduced [4Fe-4S] 2ϩ cluster of structure F (Fig. 2) cannot be ruled out.
The mechanism was further investigated by Mössbauer spectroscopy of samples wherein the resting-state WT (Fig. 2, structure A) was treated with dithionite for 1 min and 30 min (Fig. 8). The spectra revealed the presence of three components in varying proportions. The first is a [4Fe-4S] 2ϩ cluster with even greater site differentiation (yellow lines in Fig. 8) than that observed for the resting state. This cluster exhibits sharp quadrupole doublet features in the 53-mT spectra. The 8-T spectra reveal that it has a S ϭ 0 ground state. Its parameters are virtually identical to those of two-electron-reduced FTR (14,17), and we therefore assign it to this state (Fig. 2, structure F). This component corresponds to ϳ57 and ϳ80% of total intensity in the spectra of the 1-and 30-min samples, respectively. The second component gives rise to broad, magnetically split subspectra in the 53-mT and 8-T spectra, suggesting that the features are associated with a complex exhibiting a half-integer spin ground state. Because EPR spectroscopy provides evidence for a [4Fe-4S] 1ϩ cluster, we have simulated these features with . F, reaction mixture minus MRX and Trx5. All solutions were made anaerobic, and the reactions were performed in an anaerobic glove bag containing a 95% N 2 and 4% H 2 atm. FDR was reduced with 2 mM dithionite, and the excess was removed by gel filtration with a PD-10 column. The reactions were monitored by production of thiols determined with Ellman's reagent (36). Reactions were performed in triplicate.

Table 1 Activities of WT FDR and variants with substrates MRX and Trx5
The standard reaction mixture

Ferredoxin:disulfide reductase
parameters typical of [4Fe-4S] 1ϩ clusters and assign it to structure C or E (Fig. 2) (27). This component amounts to ϳ38 and ϳ14% of total intensity in the spectra of the 1-and 30-min samples (blue lines). The Mössbauer data are consistent with the EPR data that support one-electron reduction of the resting-state FDR (structure A) [4Fe-4S] 2ϩ cluster yielding the [4Fe-4S] 1ϩ cluster (structure C or E) previously hypothesized for FTR (Fig. 2) (3). The spectra also reveal the presence of a reduced rubredoxin center (ϳ8%).

Role of the C-terminal domain
It has been proposed that the C-terminal domain of FDR has a stabilizing function analogous to that of the variable subunits of FTR (5). This role was investigated with a variant of FDR (⌬Rbx) deleted of the C-terminal domain (Fig. S4) that was purified with protocols for purification and air oxidation of WT. The expected monomer molecular mass of 17 kDa was confirmed by SDS-PAGE (Fig. S1). Size-exclusion chromatography (not shown) indicated a dimer as for WT (5). The variant contained 7.3 Ϯ 0.6 (n ϭ 3) iron per dimer, consistent with loss of the [1Fe-0S] center. CD spectroscopy (not shown) revealed the absence of random coils, which indicates that the variant folded properly.
UV-visible spectroscopy revealed a spectrum with an A max of 410 nm that was bleached by the addition of dithionite (Fig. S5), a result similar to resting-state WT with an oxidized [4Fe-4S] 2ϩ cluster. Identical results were obtained when the variant was reduced with the Fdx-regenerating system composed of NADPH, Fdx, and Fdx-NADP ϩ reductase (FNR) (not shown). The EPR spectrum, recorded 1.0 min after reduction with dithionite, revealed the presence of a S ϭ 1 ⁄ 2 [4Fe-4S] 1ϩ cluster with g values of 2.03, 1.99, and 1.89 (Fig. S5) as observed for WT. In contrast to WT, no EPR signal was observed in the low-field region for oxidized ⌬Rbx, a result consistent with absence of the [1Fe-0S] center. These results establish that the C-terminal domain of FDR is not required for integrity of the active-site [4Fe-4S] cluster.

Activities of WT and variants assayed with physiological substrates
Although FDR reduces GSH disulfide, it is a nonphysiological substrate absent in methanoarchaea (5). Thus, physiological substrates were identified to further characterize FDR and extend experimental approaches toward a physiologically relevant mechanistic understanding. M. acetivorans comprises a single TrxR and seven thioredoxin homologs (Trx1-Trx7) (28). The NADPH-dependent TrxR reduces Trx7, whereas electron donors to Trx1-Trx6 are unknown (8,10). Thus, Trx1-Trx6 are candidate substrates for FDR, as is the recently described methanoredoxin (MRX) with protein-disulfide reductase activity, and are encoded adjacent to the gene encoding FDR (29). Evaluation of Trx1-Trx6 and MRX showed that only Trx5 and MRX were reduced by FDR using Fdx as the reductant (Table  S2). Both MRX and Trx5 were also reduced by FDR that was prereduced with dithionite ( Fig. 9).
Activity of the ⌬Rbx variant with either Trx5 or MRX was only marginally less than for WT (Fig. S6 and Table 1), consistent with the C-terminal domain and [1Fe-0S] center dispens-able for integrity or reduction of the [4Fe-4S] active-site cluster of the variant with Fdx. Interestingly, Tyr 83 of FDR is in the equivalent position of His 86 in FTR, which is essential for catalysis (14,18). Thus, Tyr 83 was replaced with histidine and assayed with either Trx5 or MRX. Although the Y83H variant showed approximately twice the activity of WT (Table 1 and Fig. S6), the results indicate that histidine is a viable replacement for Tyr 83 .

Bioinformatic analysis
As the overall structure and active site geometry of FDR is novel, it was of interest to determine the extent to which homologs are present in nature. A search of the non-redundant National Center for Biotechnology Information-National Institutes of Health (NCBI-NIH) protein database with FDR as the query retrieved homologs from diverse species in the domains Bacteria and Archaea. The first 100 sequences had Ͼ98% coverage and Ͼ55% identity, for which representatives are shown in Fig. S7. The large majority of homologs in the domain Archaea derive from diverse methanogens, indicating that FDR plays a major role in redox maintenance of this group. All homologs contain the conserved active-site tyrosine and cysteine motifs ligating the active-site [4Fe-4S] cluster and [1Fe-0S] center, suggesting an identical catalytic mechanism. The results establish that FDR is the archetype of a novel disulfide reductase widespread in prokaryotes.

Discussion
The results presented advance an understanding of the catalytic mechanism of FDR and the physiological role it plays in methanoarchaea.

Mechanism
The mechanism proposed for FDR is similar to that proposed for FTR (Fig. 2), albeit with important exceptions (3,14,15,17). Similar to FTR, it is proposed that the resting-state WT (structure A) contains a unique iron in the oxidized [4Fe-4S] 2ϩ cluster that weakly interacts with the active-site disulfide (Cys 52 -S-S-Cys 84 ), as shown by Mössbauer data and predicted from the crystal structure (Fig. 1). Supported by spectroscopic and redox properties of NEM-FDR, a one-electron-reduced intermediate (structure B) is proposed in which the disulfide is cleaved and the cluster-interacting thiol of Cys 84 ligates to the unique iron site to yield an S ϭ 1 ⁄ 2 [4Fe-4S] 3ϩ cluster with five cysteinate ligands similar to that proposed for FTR. However, unlike in FTR, EPR and Mössbauer spectroscopy of reduced WT FDR identified a one-electron-reduced [4Fe-4S] 1ϩ cluster ascribed to structures C or E that is a proposed intermediate in reduction of the resting-state structure A en route to the one-electronreduced structure B of FDR. Intermediate E is disfavored, as it entails transient formation of a new disulfide bond. Thus, it is proposed that conversion of structure A to the one-electronreduced structure B proceeds via intermediate C (Fig. 2).
Conversion of intermediate C to B could occur by a twoelectron attack on the sulfur of Cys 84 and cleavage of the activesite disulfide or involve intermediate D. Regardless, it is doubtful that reduction of the resting-state active-site disulfide (Cys 52 -S-S-Cys 84 ) to the thiyl radical-containing structure D Ferredoxin:disulfide reductase proceeds directly without participation of intermediate structure C. This proposal is based on the reported structure of FTR showing that Fdx binds to the side of the [4Fe-4S] cluster opposite that of the solvent-exposed active-site disulfide. As the active-site architecture of FDR is similar to FTR, with the active-site disulfide exposed to solvent, it is proposed that intermediate C is involved in reduction of the resting state to yield structure D followed by electron transfer from the [4Fe-4S] 2ϩ cluster to the thiyl radical. This mechanism proposed for FDR is expected to apply to one-electron reduction of the plant-type FTR that awaits experimental evidence.
Mössbauer data of reduced resting-state FDR revealed a [4Fe-4S] 2ϩ cluster with more pronounced perturbation of the unique iron site, rationalized by a strong interaction with Cys 84 characteristic of structure F, as was observed previously for FTR (17). Thus, it is proposed that structure B is reduced to the two-electron-reduced structure F. The strong interaction of the unique iron site with Cys 84 is conceivably the consequence of a strong hydrogen bond between the thiol of Cys 84 and the sulfur atom of Cys 50 that leaves Cys 52 free to attack substrates. This interaction is the same proposed for FTR, wherein His 86 serves to protonate Cys 87 , promoting a hydrogen bond between the thiol of Cys 87 and the sulfur atom of Cys 55 (Fig. 2) (12, 17). However, FDR contains Tyr 83 ligated to Cys 84 of the active-site disulfide (Cys 52 -S-S-Cys 84 ) (Figs. 1 and 2). Conservation of Tyr 83 in all FDR homologs from diverse species in the domains Bacteria and Archaea (Fig. S7) indicates an essential function. Indeed, a past investigation showed that the Y83A and Y83F variants are inactive 8 when assayed with the surrogate substrate GSSG as described previously (5). Surprisingly, activity of the Y83H variant reported here increased nearly 2-fold relative to WT, in contrast to nearly complete loss of activity and sitedifferentiated character of the unique iron for the H86Y variant of FTR (18). Although the results presented here support a role for intermediate F, it is possible that Tyr 83 does not function in proton transfer assisting reduction of intermediate B to F, thereby favoring the pathway in which substrate binds to intermediate B forming intermediate G (Fig. 1). Further experimentation involving structural and spectroscopic analyses of WT and variants are necessary to determine the essential role for Tyr 83 . Regardless, the results establish that the active sites of FTR and FDR are not identical. Contrasting active sites could also explain the pronounced difference in redox potentials for the [4Fe-4S] 3ϩ/2ϩ couple of FTR and FDR.
The penultimate step in the proposed mechanism of FDR involves binding of substrates to the two-electron-reduced structure F, forming the heterodisulfide intermediate (structure H) analogous to that proposed for FTR. Viability of this proposal is supported by the finding that MRX and Trx5 are reduced with dithionite-reduced FDR containing the two-electron-reduced [4Fe-4S] 2ϩ cluster. However, binding of substrate to one-electron-reduced structure B and subsequent one-electron reduction to yield structure H, as proposed for FTR, cannot be ruled out at this juncture. In the final step, also analogous to FTR, the thiolate of Cys 84 is proposed to attack the heterodisulfide bridge of structure F, releasing reduced Trx5 or MRX and reforming the active-site disulfide (Cys 52 -S-S-Cys 84 ) of resting-state structure A.
It is expected that the mechanism of FDR applies to homologs in phylogenetically and physiologically diverse prokaryotes that have high sequence identity and conserved active-site motifs.
The results also address the role of the C-terminal domain of FDR and the [1Fe-0S] center in transfer of electrons from Fdx to the active site. Fdx-dependent activity was retained on deletion of the C-terminal domain, ruling out the previously proposed function of stabilizing the active site analogous to the variable subunits of FTR. However, this result does not exclude a role for the center in mediating electron transfer from Fdx to the activesite [4Fe-4S]. The C-terminal domain faces the side of the [4Fe-4S] cluster opposite that of the active-site disulfide (Cys 52 -S-S-Cys 84 ) analogous to the Fdx-binding site of FTR (Fig. 1). Thus, it is likely that the C-terminal domain blocks binding of Fdx adjacent to the [4Fe-4S] cluster, requiring the [1Fe-0S] center in WT as a point of entry for electrons from Fdx. A role for the center in intramolecular electron transfer is supported by redox potentials of the [4Fe-4S] 3ϩ/2ϩ and [1Fe-0S] 2ϩ/1ϩ couples that are nearly identical. Finally, the deduced sequences of all FDR homologs contain a C-terminal domain conserving the cysteine motif ligating the [1Fe-0S] center in FDR, further supporting an essential function for the center (Fig. S7). On the other hand, the C-terminal domain of FDR could be mobile and moved out of the way by interaction with Fdx (Fig. 1). Clearly, further investigation is necessary to resolve the function of the C-terminal domain and the [1Fe-0S] center.

Physiology
Our results advance an understanding of redox control in methanogens and the domain Archaea. Aerobic microbes classified in the domain Bacteria contain glutaredoxin, which is similar to Trx in structure and function, although it obtains reducing equivalents from GSH that is the product of an NADPH-dependent GSSG reductase (1). However, the glutaredoxin system is absent in strict anaerobes (1). Indeed, methanogens do not contain GSH, indicating the absence of a functional glutaredoxin system for redox maintenance (30 -32). Alternatively, nearly every sequenced methanogen genome has annotations for Trx homologues, indicating that Trx is a major thioldisulfide oxidoreductase involved in redox control (7,10). Notably, it was shown that Trx targets a host of diverse proteins in Methanocaldococcus jannaschii (7). Still, processes involved in providing reducing equivalents to Trxs of methanogens are largely unknown. Fdx and F 420 are the primary electron carriers in methanogens. M. jannaschii, a deeply rooted species, was shown to use a F 420 -dependent TrxR to reduce its primary Trx (7). M. acetivorans, a later-evolving species, lacks F 420 -dependent TrxR. Instead, M. acetivorans uses a NADPH-dependent TrxR that, of the seven Trxs, is specific only for Trx7. NAPDH is produced by the oxidation of Fdx and F 420 H 2 in M. acetivorans (8). The finding that FDR reduces Trx5 and MRX with Fdx, the primary electron carrier in acetate-grown M. acetivorans, reveals a pathway to reduce Trx directly from Fdx. Trx5 can participate in thiol-disulfide exchange but lacks disulfide reductase activity with insulin, the common surrogate substrate (8). Nonetheless, Trx5 is probably specific for the reduction of certain methanogen proteins. The results underscore the importance of determining roles for Trx5 and the remaining Trxs and the processes that supply reductant (33,34).
It was shown previously that MRX is reduced by HSCoM (HSCH 2 CH 2 SO 3 2Ϫ ), a coenzyme essential for the final step in all methanogenic pathways and also proposed to play a role in redox sensing (29). The finding that FDR also reduces MRX further advances a physiological understanding of both proteins and suggests that redox control in M. acetivorans is complex. The results support the previous proposal that FDR is representative of a diverse family of disulfide reductases proposed to have evolved from an ancestral plant-type FTR catalytic subunit to meet a variety of specific ecological needs (6).

Conclusions
A major advance toward the mechanistic and physiological understanding of FDR from M. acetivorans was accomplished with comprehensive experimental approaches that included EPR and Mössbauer spectroscopy, and site-specific variants. The study is the first for a unique family of disulfide reductases that evolved from an ancestral plant type and is distributed in diverse species from the domains Bacteria and Archaea. The results revealed similarities and differences between FDR and a plant-type FTR, leading to a proposed mechanism similar to that for FTR albeit with exceptions, reflecting differences in the active sites and subunit composition. Finally, the identification of substrates for FDR advances a physiological understanding of redox maintenance and response to oxidative stress in methanogens.

Experimental procedures
Materials NEM, Fdx, FNR from Spinacia oleracea, NADPH, and bovine pancreas insulin were obtained from Sigma-Aldrich. DNA oligonucleotides were from Integrated DNA Technologies, Inc. (Coralville, IA). The 57 Fe was purchased from ISO-FLEX (San Francisco, CA). DNA polymerase was purchased from Takara Bio Inc. (Mountain View, CA). All other chemicals used were of analytical grade.

Site-directed mutagenesis
The ⌬Rbx (minus the rubredoxin domain) and Y83H variants were constructed using the Q5 site-directed mutagenesis kit (New England Biolabs Inc., Ipswich, MA). In brief, mutations were introduced by directly amplifying the expression vector (pMA1659) containing the FDR gene using the following primers: ⌬Rbx, 5Ј-TGAATGCATCATCACCATCAC-3Ј (forward) and 5Ј-TCTCCATACCGGCTTTGAAAG-3Ј (reverse); Y83H, 5Ј-CGGGGCCTGTcatTGTGCCCTCT-3Ј (forward) and 5Ј-TAATCGTTCAGGTCAGGGTCTC-3Ј (reverse). The PCR products obtained after amplification were incubated with kinase-ligase-DpnI (KLD) enzyme mix (New England Biolabs) in buffer for 10 min at 21°C and transferred into 5-␣competent E. coli (New England Biolabs). The correct plasmids were validated by sequencing performed at the DNA Genomics Core Facility, Huck Institute of Life Sciences, Pennsylvania State University.

Protein expression and purification
To assure a full complement of iron, expression plasmids containing the genes encoding WT FDR (MA1659) and variants Y83H and ⌬Rbx were transformed into E. coli strain BL21(DE3) ⌬iscR (a gift from Prof. Dennis Dean, Virginia Tech) that was cultured anaerobically using a modified 3-liter spinner flask (Chemglass, Vineland, NJ) sparged with 1.0 atm of argon. Transformed cultures (100 ml) were grown overnight in lysogeny broth medium containing 100 mM MOPS (pH 7.4) and 1 mM ferric ammonium citrate that was the inoculum for the spinner flask cultures. The inoculum was grown aerobically at 37°C to an A 600 of 0.6 before adding inducer (250 M isopropyl 1-thio-␤-D-galactopyranoside) and 1 mM cysteine. The medium for 3-liter cultures was the same as for the inoculum except for amendment with glucose (0.5%, w/v) and sodium fumarate (25 mM) and sparging with 1.0 atm of argon to maintain anaerobic conditions. Cells were harvested at 21 h and stored at Ϫ70°C.
The recombinant WT FDR and variants Y83H and ⌬Rbx were purified anaerobically. Cells from 3-liter cultures were harvested by centrifugation. All subsequent steps were performed in an anaerobic glove bag (Coy Manufacturing, Ann Arbor, MI) using degassed buffers. The cell pellet was resuspended in 60 ml of 50 mM Tris-HCl buffer (pH 8) containing 300 mM NaCl (buffer A). Cells were lysed by sonication followed by centrifugation at 15,000 ϫ g for 30 min at 4°C to collect cell-free extract. The supernatant solution containing recombinant proteins was filtered (0.22 m) and loaded onto a nickel-Sepharose column connected to an Ä kta FPLC system (GE Healthcare). The proteins were eluted using a linear gradient of 0.0 -500 mM imidazole in buffer B (50 mM Tris-HCl (pH 8) containing 300 mM NaCl). Fractions containing the recombinant proteins were pooled and concentrated to 2 ml using a Vivacell 70 filter (Sartorius, Göttingen, Germany) fitted with a 10-kDa cutoff membrane and loaded onto a Superdex 200 column pre-equilibrated with 50 mM Tris-HCl (pH 8) containing 150 mM NaCl (buffer C). FDR eluted with buffer C as a single isolated peak. Fractions containing dark brown FDR were collected and concentrated to 2 ml using a 10-kDa cutoff membrane and stored at Ϫ80°C containing 10% (v/v) glycerol. Protein concentrations were determined by the Bradford assay (35). Purity was determined by SDS-PAGE and de novo sequencing at the Penn State Mass Spectrometer facility. The Trx proteins from M. acetivorans were heterologously produced and purified to homogeneity as described previously (8).

Determination of thiol content
The quantification of sulfhydryl groups was done using Ellman's reagent as described previously (36). Briefly, 100 l of reaction mixture was added to 900 l of a solution containing 0.2 mM 5,5Ј-dithiobis(nitrobenzoic acid) and 50 mM sodium acetate in 100 mM Tris buffer (pH 8.0). The samples were incubated at 21°C for 5 min, after which the optical absorbance was recorded at 412 nm.

Ferredoxin:disulfide reductase Activity assays
The concentrations of reactants are indicated under "Results" and in the supporting materials. Assays were conducted anaerobically with degassed buffers and in an N 2 atmosphere. The reduced Fdx electron donor was regenerated with NADPH and FNR. One unit of FNR reduces 1.0 mmol of cytochrome c per min at pH 7.5 and 25°C in the presence of spinach Fdx and NADPH. Reactions were initiated by the addition of FNR. Initial rates were determined by monitoring the oxidation of NADPH at 340 nm.

NEM modification
Alkylation was carried out by reducing FDR with an excess of dithionite and incubating for 30 min to ensure complete reduction of active-site cysteines. The reduced protein was placed on ice for 15 min and treated with a 5-fold molar excess of NEM for 10 min before quenching the reaction by exposure to air. Excess reagents were removed by using a PD-10 size-exclusion column.

EPR spectroscopy
CW EPR studies were carried out at X-band (9.39 GHz) using a Bruker E500 spectrometer and a cylindrical TE 011 mode resonator (Bruker BioSpin Corp., Billerica, MA). A ESR-910 liquid helium cryostat and ITC-4 controller (Oxford Instruments) maintained the temperature at 15 K. Determination of EPR signal intensities was done by measuring the individual EPR-active species under nonsaturating conditions. As the signals represent the first derivative of the absorption-type signal, the spectra were double-integrated, and the surface area of each signal was compared with that of a 10 mM copper-perchlorate standard (10 mM CuSO 4 , 2 M NaClO 4 , and 10 mM HCl). The values obtained this way were compared with the known enzyme concentration, which was set to 80% of the [4Fe-4S] cluster in dimeric FDR. Only the intensities of the [4Fe-4S] 1ϩ in WT and [4Fe-4S] 3ϩ in NEM-WT could be determined as described. The amount of EPR silent S ϭ 0 resting-state [4Fe-4S] 2ϩ was assigned as the difference between the concentration of FDR present minus the concentration of paramagnetic species.
Redox titration of NEM-WT was as described elsewhere (17). The titration was performed three times. Shown is a representative titration. The mean and S.D. are the results of the three determinations.

Mössbauer spectroscopy
Mössbauer spectra were recorded on constant-acceleration Mössbauer spectrometers from Seeco (Edina, MN) equipped either with a Janis SVT-400 variable-temperature cryostat (weak-field) or a Janis 8TMOSS-OM-12SVT variable-temperature cryostat (strong-field). Isomer shifts are reported with respect to the centroid of the spectrum of ␣-iron metal at room temperature. Simulations of Mössbauer spectra were carried out with the program WMOSS (Seeco, Edina, MN). Calculations invoking the spin Hamiltonian formalism were performed according to the equation below, in which the first term describes the electron Zeeman effect, the second and third terms describe the axial and rhombic zero-field splitting of the total electron spin ground state, the fourth term represents the interaction between the electric field gradient and the nuclear quadrupole moment, the fifth term describes the magnetic hyperfine interactions of the 57 Fe nucleus with respect to the total electron spin ground state, and the last term represents the nuclear Zeeman interactions of 57 Fe. All symbols have their usual meaning (13,25). Spectra were calculated in the slow relaxation limit.