Article pubs.acs.org/JACS
Reduction Potentials of [FeFe]-Hydrogenase Accessory Iron−Sulfur Clusters Provide Insights into the Energetics of Proton Reduction Catalysis Jacob H. Artz,† David W. Mulder,‡ Michael W. Ratzloff,‡ Carolyn E. Lubner,‡ Oleg A. Zadvornyy,† Axl X. LeVan,§ S. Garrett Williams,∥ Michael W. W. Adams,⊥ Anne K. Jones,∥ Paul W. King,‡ and John W. Peters*,† †
Institute of Biological Chemistry, Washington State University, 258 Clark Hall, Pullman, Washington 99163, United States Biosciences Center, National Renewable Energy Laboratory, 15013 Denver West Parkway, Golden, Colorado 80401, United States § Department of Chemistry and Biochemistry, Montana State University, 224 Chemistry and Biochemistry Building, Bozeman, Montana 59717, United States ∥ School of Molecular Sciences, Arizona State University, P.O. Box 871604, Tempe, Arizona 85287, United States ⊥ B216B Life Sciences Complex, Department of Biochemistry, The University of Georgia, Athens, Georgia 30602, United States ‡
S Supporting Information *
ABSTRACT: An [FeFe]-hydrogenase from Clostridium pasteurianum, CpI, is a model system for biological H2 activation. In addition to the catalytic H-cluster, CpI contains four accessory iron−sulfur [FeS] clusters in a branched series that transfer electrons to and from the active site. In this work, potentiometric titrations have been employed in combination with electron paramagnetic resonance (EPR) spectroscopy at defined electrochemical potentials to gain insights into the role of the accessory clusters in catalysis. EPR spectra collected over a range of potentials were deconvoluted into individual components attributable to the accessory [FeS] clusters and the active site H-cluster, and reduction potentials for each cluster were determined. The data suggest a large degree of magnetic coupling between the clusters. The distal [4Fe-4S] cluster is shown to have a lower reduction potential (∼ < −450 mV) than the other clusters, and molecular docking experiments indicate that the physiological electron donor, ferredoxin (Fd), most favorably interacts with this cluster. The low reduction potential of the distal [4Fe-4S] cluster thermodynamically restricts the Fdox/Fdred ratio at which CpI can operate, consistent with the role of CpI in recycling Fdred that accumulates during fermentation. Subsequent electron transfer through the additional accessory [FeS] clusters to the H-cluster is thermodynamically favorable.
■
INTRODUCTION Many of the most critical reactions in biology rely on intramolecular electron-transfer pathways, which couple external electron donors and acceptors through defined pathways of redox cofactors such as [FeS] clusters, flavins, cytochromes, and hemes. The fundamental mechanisms controlling the energetics and kinetics of intramolecular electron transfer in multiple redox center arrays have potential biotechnological applications including control of external reductant pools and biasing the directionality of catalysis.1,2 However, the study of multicofactor electron-transfer pathways is challenging due to overlap of spectroscopic properties from techniques that are typically used for the characterization. Hydrogenases are examples of enzymes that incorporate complex, multicofactor arrays for intramolecular electron transfer coupled to catalysis, in this case, reversible hydrogen activation.3 The [FeFe]-hydrogenases are enzymes that possess © 2017 American Chemical Society
a catalytic site H-cluster, consisting of a [4Fe-4S] cubane ([4Fe4S]H) with a cysteine thiolate linkage to a 2Fe subcluster ([2Fe]H), with an azadithiolate (ADT) bridging ligand and several carbon monoxide and cyanide ligands.4−6 The enzyme CpI has been investigated as a model for understanding the catalytic and structural properties of [FeFe]-hydrogenase.4,7−9 CpI features an assemblage of accessory clusters (F-clusters) organized as a stem of two [4Fe-4S] clusters, which branches to distal [4Fe-4S] and [2Fe-2S] clusters4 (Figure 1). The distal [4Fe-4S] cluster is coordinated by three cysteines and a histidine, similar to that of the [FeS] clusters in Complex I, and in contrast with the other two [4Fe-4S] clusters that have standard four cysteine coordination. [FeFe]-hydrogenases from different sources have a variety of different accessory cluster Received: March 1, 2017 Published: June 21, 2017 9544
DOI: 10.1021/jacs.7b02099 J. Am. Chem. Soc. 2017, 139, 9544−9550
Article
Journal of the American Chemical Society
site for the external electron donor Fd, and is shown to have a significantly negative reduction potential relative to the Fcluster chain. This implicates that, under physiological conditions, the arrangement and relative potentials of the accessory clusters presumably modulate the threshold ratio of oxidized and reduced electron carriers capable of driving proton reduction.
■
METHODS
Protein Expression and Purification. Heterologous protein expression was performed using the pET-Duet plasmid encoding CpI, as has been described previously.15−17 No maturase genes were present, yielding the apo form of the enzyme. The pET-DuetCpI plasmid was transformed into BL21 (DE3) Rosetta-2 competent cells (Novagen). Fresh transformants were inoculated into a 150 mL overnight expression culture. Four milliliters of the overnight culture was inoculated into 1 L of Terrific Broth media (EMD Millipore) supplemented with 5% glycerol, and the cultures were grown at 37 °C, 250 rpm, to an OD600 of 0.55−0.6. Two liters of culture was transferred into 2 L narrow-neck Erlenmeyer flasks (Kimble Kimax), and placed under continuous argon sparging at room temperature. The medium was then supplemented with sodium fumarate (25 mM final), ferric ammonium citrate (4 mM final), cysteine (2 mM final), and induced with IPTG (1.5 mM final). All further work was performed under anaerobic conditions. After 16−18 h of sparging, cultures were harvested by centrifugation for 6 min at 5400g, and the pellets were flash frozen in liquid nitrogen and stored at −80 °C until further use. Pellets were thawed and lysed in buffer of 50 mM Tris, pH 8, 5 mM NaCl, 5% glycerol, 20 mM sodium dithionite, cOmplete EDTA-free protease inhibitor cocktail (Roche Life Sciences), 140 μg mL−1 DNase and RNase, 120 μg mL−1 lysozyme, and 1% Triton X100. The lysate was stirred for 1.5 h at room temperature, and clarified for 1 h at 60 000g. Protein purification was performed on the benchtop anaerobically under continuous argon sparging. Clarified lysate was first purified over an anion exchange column (diethylaminoethanol sepharose, GE LifeSciences), and eluted with 50 mM Tris pH 8, 300 mM NaCl, 5% glycerol, and 20 mM NaDT. The eluate was collected and concentrated using Amicon Ultra-15 centrifugal concentrators with a 30 kDa cutoff (Millipore), loaded onto an affinity chromatography column (Strep-tactin resin, IBA), and eluted with 5 mM desthiobiotin. Purity of the protein was verified using SDS-PAGE and Western blot with antistrep antibody. The concentration of the protein was measured using the Bradford assay. Synthetic Reconstitution. Wild-type (WT) azadithiolate [2Fe]H and propanedithiolate (PDT) [2Fe]H reconstituted enzymes were prepared synthetically under anaerobic conditions according to previously published methods.5,18,19 The appropriate synthetic [2Fe]H was incubated with the apo-CpI in 100-fold excess for 1.5 h in an MBraun anaerobic box. Excess [2Fe]H was removed from the protein by exchange with a G-25 column (GE Healthcare). Cluster incorporation was verified with FTIR. Potentiometric Titrations. Steady-state potentiometric titrations were performed under strict anaerobic conditions in an Mbraun glovebox at room temperature. Protein was loaded on a G-25 column to remove sodium dithionite and eluted with 50 mM Tris pH 8, 200 mM NaCl, and 5% glycerol. The protein was then briefly treated with 1 mM thionin to oxidize it and buffer exchanged over the G-25 column again. The oxidized protein was added to a continuously stirred sample cell for potentiometric titrations along with 5 μM final concentration of each of the following redox-active dyes: indigo disulfonate (reduction potential = −125 mV), phenosafrinin (reduction potential = −255 mV), benzyl viologen (reduction potential = −361 mV), and methyl viologen (reduction potential = −440 mV), a final concentration of 10% glycerol, and a total cell volume of 3.2 mL. The potential was measured using a Ag/AgCl ORP electrode (Thermo Scientific 9678BNWP) and calibrated against a standard solution (Orion 967901, +220 mV vs SHE). All potentials reported herein are adjusted to the standard hydrogen electrode (SHE). The protein was
Figure 1. Depiction of CpI [FeFe]-hydrogenase (PDB 3C8Y), showing the branched conduit array, three cluster variations of the active site: (1) Apo, in which the protein is lacking the [2Fe]H, (2) the native H-cluster with an azadithiolate (ADT) [2Fe]H, and (3) the catalytically impeded H-cluster with a propanedithiolate (PDT) [2Fe]H.
arrangements, presumably adapted for their respective physiological functions.3,10,11 An electron paramagnetic resonance (EPR) study involving potentiometric titration of CpI has previously shown that the reduction potentials of the accessory cluster chain are approximately −420 ± 30 mV (pH 8), which is slightly more negative than the H-cluster at −400 ± 30 mV (pH 8.0).8,12 However, this study was unable to discriminate individual cluster EPR signals, nor the contribution to electronic properties and associated individual reduction potentials under turnover conditions.12 Previously, we were able to partially simulate the EPR spectra of dithionite-reduced CpI as a combination of three rhombic contributions, one of which arises from the well-characterized Hox signal of the H-cluster.13 The two remaining signals could not be assigned to specific accessory clusters.13 Furthermore, several other weaker, broad contributions proposed to arise from hyperfine couplings remained unaccounted for in the data simulations. Characterization of a construct containing only the N-terminal [2Fe-2S] domain14 revealed that the [2Fe-2S] accessory cluster exhibits a nearly axial signal with g-values of 2.047, 1.954, and 1.911 and a reduction potential of −390 ± 10 mV (pH 8.0). In this work, we have measured EPR spectra from CpI over a potential range that spans the turnover conditions, to identify individual cluster spectral contributions and assign reduction potentials to each [FeS] cluster. The results show evidence for potential-dependent, electronic interactions between the Hcluster and F-clusters in CpI, which may be further tuned by the branched arrangement of the conduit. Overall, these interactions are thought to strongly influence the energetics and kinetics of electron transfer and the reversible H2 oxidation activity. The unique 1His, 3Cys coordinated [4Fe-4S] cluster located at the enzyme surface is suggested to be the docking 9545
DOI: 10.1021/jacs.7b02099 J. Am. Chem. Soc. 2017, 139, 9544−9550
Article
Journal of the American Chemical Society titrated toward more negative potentials by addition of either 2 or 10 mM sodium dithionite in 2 μL increments. Following stabilization at a given potential, 200 μL samples were withdrawn, loaded into EPR tubes to a final concentration of 100 μM, and flash frozen in liquid nitrogen. For the apo-CpI titration, samples were withdrawn sequentially at −128, −259, −286, −336, −363, −400, −412, −420, −436, −440, and −457 mV. Temperature relaxation and power saturation profiles were collected for samples at −400 and −457 mV. For the CpIPDT titration, samples were withdrawn sequentially at −133, −260, −277, −297, −321, −351, −381, −402, −418, −444, −469, and −475 mV. The last sample (−475 mV) proved to be the most reduced the sample could achieve using dithionite as the reductant under this set of conditions. Temperature relaxation profiles were collected for samples at −381, −402, −418, −444, and −475 mV, and power saturation profiles at −381, −418, and −475 mV. Several samples were also prepared for analysis by Q-band EPR by a similar procedure, and 10 μL samples were withdrawn at −400, −420, and −485 mV followed by immediate freezing. EPR Spectroscopy. CW X-band EPR measurements were carried out on a Bruker Elexsys E-500 EPR spectrometer equipped with an SHQ resonator and in-cavity cryogen free VT system (ColdEdge Technologies) with a MercuryiTC temperature controller (Oxford). Spectra were collected at a frequency of 9.38 GHz, 1 mW microwave power, 15 K sample temperature (±0.2 K), a modulation frequency of 100 kHz, and a modulation amplitude of 10.0 GHz, although the power ranged from 0.1 to 80 mW for power series measurements, and the temperature was varied from 4 to 80 K for temperature series data. Pulse Q-band measurements (33.8 GHz) were carried out on a Bruker Elexsys E-580 CW/FT EPR spectrometer equipped with an EN5107D2 resonator and FlexLine cryostat (ColdEdge Technologies) with MercuryiTC temperature controller along with an additional cernox temperature sensor positioned on the probehead near the sample. Data were baseline corrected as necessary by either subtraction of a buffer reference or a manual user defined function in OriginLab. EPR simulations were carried out in MatLab using the EasySpin package20 and “esfit” fitting function. G-strain was used as the line broadening parameter, and data were fit using the simplex algorithm. The errors in the reported g-values are estimated at ±0.003. Spin counting was carried out by double integration of the spectra after manual baseline subtraction in OriginLab and referenced to coppertriethylamine samples with known concentrations (75−125 μM) measured at the same conditions. Nernst Fits. The reduction potentials of individual clusters were assigned by using signal amplitudes of the deconvoluted CpIPDT spectra as a function of potential, and then fit with the one-electron (n = 1) Nernst equation in OriginLab using the least-squares fitting method. Errors in the determined reduction potential are estimated to be ±8 mV. An EPR power saturation curve was used to verify that data for the Nernst fits were collected under nonsaturating conditions (Figure S1). Protein Docking. ClusPro 2.021 was used in the default balanced coefficients mode to dock CpI [FeFe]-hydrogenase PDB 3C8Y22 to the Fd from C. pasteurianum, PDB 1C1F.23 All top solutions were examined, and all showed docking of Fd near FS4C.
■
substituted by propanedithiolate (PDT). The resulting PDTsubstituted CpI (CpIPDT) exhibits very low proton reduction activity, but retains EPR spectra that closely resemble the native CpI24 (Figure S9), providing a robust platform to poise CpI samples at defined reduction potentials and establish a steady state near equilibrium. Eleven samples of apo-CpI were redox poised using mediators over −128 mV to −457 mV, and EPR spectra were recorded from each (Figure 2), as well as under different
Figure 2. EPR spectra from a potentiometric titration of apo-CpI (100 μM) from −128 to −457 mV. Microwave frequency, 9.38 GHz; microwave power, 1 mW; sample temperature, 15 K.
temperature and power regimes. Samples were frozen after stabilization of the solution potential.25 The resulting spectra consist primarily of one broad rhombic state with an additional rhombic feature and a g = 2.0 isotropic feature. The observed spectral features increase in intensity as the potential is lowered, but the spectral shape remains largely unchanged. The broadness of the signal, in conjunction with the fact that the main spectral feature is one rhombic system, suggests that the [4Fe-4S]H and the accessory [FeS] clusters (F-clusters) exhibit significant hyperfine and/or electron−electron coupling. The lack of [2Fe]H in apo-CpI and the relative difficulty in reducing the [4Fe-4S]H26,27 support an assignment of the complex EPR signal as arising from the reduced F-clusters. The nature of the interactions observed in the EPR line shape likely arises from electronic exchange interactions and averaging of the cluster spin systems, making it difficult to resolve individual signals.28,29 Interestingly, comparison of the apo-CpI signal to the CpIPDT with Hox signal subtracted shows distinct spectral features (Figure 3, red versus blue spectrum), despite the majority of the EPR signal arising from the same set of [FeS] clusters (F-clusters and ([4Fe-4S]H). This observation indicates that addition of [2Fe]H to make an intact H-cluster strongly influences the coupling interactions between F-clusters in CpI. In addition, it also indicates that apo-CpI is an inadequate model system for characterizing the electronic properties of how F-clusters in native CpI function as an electron-transfer chain. To avoid the inability to reach equilibrium that results from high catalytic turnover of reductant, we characterized a
RESULTS AND DISCUSSION
One of the major difficulties in generating and characterizing hydrogenases at defined potentials is that it is not possible to eliminate one of the substrates, protons, and thus titrations conducted at potentials below the H+/H2 reduction potential result in catalytic turnover and cannot reach equilibrium. This is a significant challenge for investigating CpI given that it catalyzes the reduction of protons at high rates at potentials ≤ −400 mV, and experimental conditions can be unstable and far from equilibrium. To mitigate this problem, we have employed samples lacking either the H-cluster [2Fe]H subcluster (apoCpI) or apo-CpI reconstituted with a synthetic analogue of [2Fe]H in which the native nonprotein ADT ligand is 9546
DOI: 10.1021/jacs.7b02099 J. Am. Chem. Soc. 2017, 139, 9544−9550
Article
Journal of the American Chemical Society
diamagnetic state. The lack of signal broadening at moderate (Figure 4, −402 mV) versus low (Figure 4, −475 mV) potentials in CpIPDT suggests that the unpaired spin mainly resides on the fully assembled H-cluster, diminishing the overall level of coupling. The reduction potential of the Hox in CpIPDT (−376 ± 8 mV, pH 8) is within the range of error reported12 for the CpIADT (−400 ± 30 mV, pH 8), which suggests that exchanging the bridging atom of the [2Fe]H from N-to-C does not dramatically alter the electronic properties of either the Hcluster or the F-clusters. This may not be true for all hydrogenases, however, as a previous study on the algal hydrogenase has shown that the enzyme reconstituted with the PDT-[2Fe]H has a reduction potential shifted to mildly more oxidizing potentials when compared to the native [2Fe]H.30 Several other signals grow in as the potential is decreased toward −402 mV. A very complex spectrum was observed at this potential, in agreement with those reported previously,13,31 which can be simulated as a series of six spin systems, including the well-characterized Hox and Hox-CO (g = 2.083, 2.020, 2.014) states, although the Hox-CO represents a relatively minor spin contribution to the signal (5% at −402 mV). At more negative potentials (−469 and −475 mV), the spectra broaden, presumably from weak spin−spin exchange among individual F-clusters, such as that reported in 2x[4Fe-4S] containing ferredoxins.32 We cannot exclude the possibility of additional spin contribution from EPR active, two electron reduced H-cluster states such as Hsred, which contains a [4Fe4S]H1+ cluster.26,27 It has been previously observed that the CrHydAPDT has residual levels of activity,24,33 which is consistent with our observation of drift in potential of the CpIPDT. Given the observed activity, we would anticipate that reduced states like Hsred are populated at some level. To further analyze the possibility of magnetic coupling at lower potentials, we performed a comparative EPR analysis at X and Q bands (Figure S10). The Q-band spectra were collected for samples prepared at −400, −420, and −485 mV. For the samples at −400 and −420 mV, only very minor shifts in gvalues are observed as compared to the X band. At −485 mV, the Q-band sample does not exhibit the broad features characteristic of the X-band at −475 mV. These observations suggest that the coupling interactions are strong only in the lowest potential samples. Deconvolution of the complex CpIPDT EPR spectra was accomplished with spectral simulations carried out with the “Pepper” function of EasySpin using previously described and simulated spin systems.13,14,34 The use of a combination of six spin systems successfully simulates the experimentally observed spectra (Figure 5). Individual g-values and weights for each component are presented in Table 1. At −402 mV, all clusters have a similar contribution to the overall spectrum, with the exception of a weak contribution from the Hox-CO signal. Interestingly, the Hox-CO signal has not been previously observed via FTIR for the PDT-reconstituted CrHydA1.30 This may reflect differences in the CpI enzyme as compared to CrHydA1 or differences in spectroscopic methodology. Alternatively, the signal assigned to Hox-CO may arise from degraded cluster. The four remaining unassigned signals in CpIPDT are proposed to result from each of the four F-clusters. The [2Fe-2S] FS2 cluster was attributed to the system with gvalues of 2.043, 1.959, and 1.911, based on similarity to values previously reported from a CpI N-terminal construct containing only FS2.14
Figure 3. EPR spectra of apo-CpI (100 μM) at −400 mV (red), CpIPDT at −402 mV (yellow), CpIPDT Hox (green), CpIPDT with Hox subtracted (gray), and the difference spectrum of reduced CpIPDT minus Apo CpI (blue). The incorporation of the [2Fe]H causes a large shift in the EPR spectra.
catalytically compromised CpI (CpIPDT) that otherwise retains a native structure and [FeS] cluster composition. EPR spectra from CpIPDT were measured by a redox dye mediated, potentiometric titration across a range from −133 to −475 mV (Figure 4). FTIR spectra were also collected later on
Figure 4. EPR spectra from a potentiometric titration of CpIPDT (100 μM) from −133 to −475 mV. Microwave frequency, 9.38 GHz; microwave power, 1 mW; sample temperature, 15 K.
thawed EPR samples to verify the integrity of the PDT-[2Fe]H H-cluster in chemically reconstituted samples (Figure S2). During the titration, there was a slow but noticeable drift toward more positive potentials, particularly at potentials below −418 mV. This is likely a result of a low level catalysis and an indication that the samples are at a steady state approaching equilibrium.24 Therefore, the samples were rapidly frozen in liquid nitrogen to trap the reduced states effectively. At more positive potentials, the EPR spectra consist of a typical, sharp Hox signal with g-values of 2.092, 2.039, and 2.00 with the entire F-cluster chain remaining in the oxidized 9547
DOI: 10.1021/jacs.7b02099 J. Am. Chem. Soc. 2017, 139, 9544−9550
Article
Journal of the American Chemical Society
value greater than 2 for the signal attributed to FS4A is unusual for a [4Fe-4S]+ cluster, suggesting the proximity of FS4A to the H-cluster may introduce magnetic coupling effects that distort or otherwise alter the spin distribution of the FS4A. Having identified the EPR spectrum for each of the individual [FeS] clusters, we independently fit each one to one-electron Nernst curves to determine the reduction potentials (Figure 6). On the whole, the reduction potentials
Figure 5. Spectral deconvolution of CpIPDT. The experimental spectrum can be simulated to close agreement with six individual signal components attributed to Hox, Hox-CO, FS4A, FS4B, FS4C, and FS2C clusters.
Notably, the FS4C system weight % increases to 30% of the observed signal at −444 mV, while the other systems are approximately the same or decreasing. This observation is consistent with FS4C as a lower potential cluster compared to the others. Also, for the overall simulation of the −444 mV spectrum (Figure S3), the lower contribution of Hox (4%) as compared to the other F-cluster signals is consistent with its initial conversion to Hred upon reduction. The signal attributed to FS4C, with g-values of 2.078, 1.952, and 1.88, was corroborated by EPR analysis of a CpI distal domain construct consisting of only 128 N-terminal residues and the corresponding ferredoxin-like domains binding the FS4C and FS2 clusters (Figures S4−S8). The absence of the Hcluster, FS4A, and FS4B clusters in the truncated protein makes it possible to specifically identify the FS4C and FS2 EPR signals, although, interestingly, the g-values of the FS2 signal are shifted slightly (∼0.004g) higher in the distal domain construct relative to the intact protein. A potentiometric titration of this distal domain construct (Figure S7) is in agreement with this cluster having a very negative reduction potential. The remaining two clusters, FS4A and FS4B, have assigned g-values of 2.077, 2.006, 1.982, and 2.120, 1.909, 1.856. While the two clusters cannot be unambiguously distinguished from this analysis, FS4B has tentatively been assigned to the g = 2.120 system on the basis of having a broader signal that may correspond with more coupling due to its closer proximity to three, rather than two, additional [FeS] clusters. The average g-
Figure 6. Proposed model of proton reduction in CpI (green). The His-coordinated FS4C has a more negative reduction potential than the other clusters, setting up a favorable electron-transfer landscape from Fdred (brown) through the conduit array toward the H-cluster (HC). Electronic interaction between FS2 and FS4B is thought to fine-tune the conduit.
of the CpI F-clusters are near that of the CpI H-cluster (−376 mV), with assignments of −366, −368, and −360 mV, for FS4A, FS4B, and FS2, respectively, making FS4A and FS4B nearly indistinguishable in reduction potential (Figures S11 and S14). Intriguingly, the Nernst curve for the FS4C signal did not reach a maximum intensity at the lowest measured potential (Figures 6 and S15). A reasonable estimate of the reduction potential value is ≤ −450 mV. This assignment of FS4C as the low-potential cluster is supported by reduction of the CpI distal domain construct with sodium dithionite prepared at more negative reduction potentials leading to an increase of the FS4C EPR signal (Figure S6).
Table 1. Components of EPR Spectra of CpI at −402 mV g-value
spin system 1 2 3 4 5 6 a
2.092 2.083 2.077 2.120 2.078 2.043
2.039 2.02 2.006 1.909 1.952 1.959
2.000 2.014 1.982 1.856 1.880 1.911
system weight %,a −402 mV
system weight %,a −444 mV
cluster assignment
17 5 17 20 20 21
4 4 15 23 30 24
Hox Hox-CO FS4A FS4B FS4C FS2
Percent contribution to global simulation. 9548
DOI: 10.1021/jacs.7b02099 J. Am. Chem. Soc. 2017, 139, 9544−9550
Article
Journal of the American Chemical Society
latter model and a hypothetical basis for a role of redox cluster arrays in catalytic bias. Collectively, the reduction potentials of the H- and F-clusters of CpI suggest that electron flow through the enzyme has been tuned to support a gradient in the reduction potential to efficiently direct electron flow. Under physiological conditions, the ability to evolve hydrogen is likely gated by the low potential FS4C. Only when a sufficient population of Fdred is present (e.g., during the disposal of excess reducing equivalents under fermentative metabolism) will electron donation to the FS4C of CpI be thermodynamically favorable. Electron transfer may then proceed rapidly through the FS4-pathway to the Hcluster. This model is similar to the proposed Complex I electron-transfer pathway model, where the initial [FeS] cluster of the electron-transfer pathway is thought to have the most negative potential, with the cluster at the opposite end of the electron-transfer pathway having the most positive potential.48 The CpI model is corroborated, in part, by current bioinformatics work that suggests that there has been coevolution between the protein environment surrounding both the H-cluster and the F-clusters.11,49,50 While it has previously been proposed that the most important factor in electron transfer between redox sites is a distance of
