The unmatched catalytic turnover rates of [FeFe]-hydrogenases require an exceptionally efficient proton-transfer (PT) pathway to shuttle protons as substrates or products between bulk water and catalytic center. For clostridial [FeFe]-hydrogenase CpI such a pathway has been proposed and analyzed, but mainly on a theoretical basis. Here, eleven enzyme variants of two different [FeFe]-hydrogenases (CpI and HydA1) with substitutions in the presumptive PT-pathway are examined kinetically, spectroscopically, and crystallographically to provide solid experimental proof for its role in hydrogen-turnover. Targeting key residues of the PT-pathway by site directed mutagenesis significantly alters the pH-activity profile of these variants and in presence of H2 their cofactor is trapped in an intermediate state indicative of precluded proton-transfer. Furthermore, crystal structures coherently explain the individual levels of residual activity, demonstrating e.g. how trapped H2O molecules rescue the interrupted PT-pathway. These features provide conclusive evidence that the targeted positions are indeed vital for catalytic proton-transfer.
[FeFe]-hydrogenases represent one of natures’ most effective classes of redox enzymes catalyzing the reversible reduction of protons to dihydrogen (H2) at turnover frequencies of up to 9000 s-11,2,3. Most [FeFe]-hydrogenases favor proton reduction while [NiFe]-hydrogenases are usually more biased toward H2 oxidation4. With their low catalytic over-potential5, [FeFe]-hydrogenases represent excellent models for a regenerative and likewise economically feasible H2 production. Their active center (“H-cluster”) can be structured into a standard [4Fe-4S]-cluster ([4Fe]H) and a diiron site ([2Fe]H). The latter is uniquely coordinated by three carbonmonoxide (CO) and two cyanide (CN−) ligands. They stabilize the cofactor in its protein environment6 and fine-tune its redox features7. An azadithiolate ligand (adt) further bridges the proximal (Fep) and the distal (Fed) iron center, which are differentiated according to their location relative to the [4Fe]H-cluster.
To achieve the extraordinary high turnover frequencies of [FeFe]-hydrogenases8, it can be implied that proton transfer (PT) is facilitated by distinct and optimized pathways. PT pathways span large distances through protein scaffolds, e.g. to enable proton-coupled electron transfer or proton translocation9,10,11,12. They usually comprise of a succession of protonatable or polar residues and protein-bound water molecules aligned at hydrogen-bonding distance13,14.
Based on the crystal structures of [FeFe]-hydrogenases CpΙ from Clostridium pasteurianum and DdH from Desulfovibrio desulfuricans, several putative PT pathways have been discussed15,16,17,18. Theoretical studies suggest that the most probable PT pathway comprises of strictly conserved residues E282, S319, E279, and C299 (from surface to H-cluster, numbering corresponds to CpI), including two protein-bound water molecules (Wat826, Wat1120 4XDC19, chain B) located between E279 and C29915,16,20. This pathway ends with C299 located in hydrogen-bonding distance to the amine head-group of the adt-ligand. Its identity and importance as a proton relay was verified in comparative studies on cofactor variants of HydA1 from Chlamydomonas reinhardtii21. Cornish and co-workers could show that amino-acid substitutions along the putative PT pathway dramatically decreased the catalytic activities. In particular, their study indicated a participation of surface-exposed residue E282 in catalytic PT15. In a second study they suggested a regulative function for positions R286 and S320 in the PT of CpI22. Furthermore, Morra and co-workers described that the pH optimum of variant C298D of [FeFe]-hydrogenase CaI from Clostridium acetobutylicum (corresponding to C299D in CpI) is shifted from pH 8 to pH 7, indicative for the involvement of C298 in PT23,24.
Although several studies were conducted, immediate experimental evidence for the relevance of residues in the putative PT pathway is missing, leaving an essential aspect of enzymatic performance opaque. In this study, site-directed mutagenesis (SDM) is used to investigate the PT pathway of two [FeFe]-hydrogenases, namely CpI and HydA1, which represent the largest (M3) and smallest (M1) type of monomeric [FeFe]-hydrogenases, respectively25. Most of the 22 SDM variants show strongly affected H2 release activities and pH optima. For 11 of these variants the crystal structure is solved, which facilitates a correlation of individual structural features (i.e. hydrogen-bonding distances) and catalytic performance. This provides insight into the minutiae of PT on the molecular level. Catalytically hampered SDM variants are analyzed by in situ attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy. When flushed with H2, these enzymes are found to adopt a key intermediate of hydrogen turnover, the recently described Hhyd state26,27,28,29,30. Hhyd accumulation under H2 clearly correlates with the diminished PT efficiency of the enzyme29. Herein we provide complementary kinetic, structural, and spectroscopic data, which allow to verify the PT pathway discussed above as the key route of catalytic PT in [FeFe]-hydrogenases.
H2 release assays and pH-dependent enzyme activities
For both CpI and HydA1, 11 SDM variants were generated to target residues along the putative PT pathway applying conservative and non-conservative exchanges (Fig. 1). Conservative exchanges (e.g. E → D) maintain the functional group of the targeted position, but due to other structural differences in the substitute residue, will affect the precise spatial placement and configuration of the functional group. In a highly ordered system such as the well-distanced H-bond chain of an evolutionarily optimized PT pathway, this should at least affect the efficiency of the functional aspect. Non-conservative exchanges (e.g. E → Q/A) delete the functional group entirely and therefore prohibit these substitute residues from rescuing the targeted function. For wild-type (wt) enzyme, H2 release activities of about 860 (HydA1) and 2600 (CpI) µmol H2 per mg per min were measured and defined as 100% activity29,31,32,33. Corresponding substitutions in CpI and HydA1 had similar effects on H2-release activity, except for variants E282QCpI (8.2%) and E144QHydA1 (0.4%). The strongest impacts were achieved when replacing E279CpI for A or Q, or C299CpI for A or S, resulting in activities <1%. This is well in line with the overall trend of an increasing impact of substitutions along the PT pathway in the direction from surface to H-cluster with C299D being the only exception (Fig. 1b). In general, conservative amino-acid exchanges had less dramatic effects than non-conservative substitutions. However, in some cases a non-conservative exchange to alanine retained a surprisingly large fraction of activity. Variants E282DCpI, C299DCpI, R286ACpI, and E282ACpI were only mildly affected and showed residual activities between 30 and 90%. For the same PT pathway position, different substitutions can have a dramatically different impact as exemplified by position E144HydA1/E282CpI. Here the exchange to glutamine diminished enzyme activity to only 0.4–8%, while the non-conservative exchange to alanine and the conservative exchange to aspartic acid retained between 46 and 81% activity. Further, there is a general trend for substitutions of the more surface-exposed PTP positions (R286, E282, and even S319) in CpI to have slightly lower impact on enzymatic activity as compared to the corresponding HydA1 variants. It might suggest for CpI a more open or flexible access to the PT pathway, which can tolerate variations slightly better than HydA1. This is especially obvious in the light of the 20-fold difference in the relative activity of E282QCpI, as compared to its HydA1 counterpart E144Q.
To link the role of individual residues in the putative PT pathway to substrate/product (H+) transfer, we probed the pH-activity profiles of variants with a high enough level of residual activity in terms of H2-release and H2-oxidation activity. As shown in Fig. 2a, native HydA1 is most active from pH 7 to pH 8 while wild-type CpI (Fig. 2b) clearly reaches its highest activity at about pH 8. Variant CpI-Y572A served as a negative control, with an amino-acid substitution outside of the putative PT pathway and features the pH-dependency profile of wild-type enzyme (see Supplementary Fig. 2a). In contrast to this behavior, the pH optima of all variants targeting the putative PT pathway were shifted to lower pH values, indicating that the limited PT efficiency can at least be partially rescued by an increased proton concentration.
Just as observed for catalytic activity, different variants of the same position in the putative PT pathway can cause significantly deviating shifts in pH optimum. In case of E282CpI, a substitution for aspartic acid shifted the optimum only slightly to pH 7.5 while an exchange to alanine or glutamine rendered the enzyme most active at pH 7 and 6, respectively (Fig. 2b). In most cases, the extent of the shift correlates to an overall loss in H2 release activity, however this has not been observed in all cases. The conservative variant E279D retains 10–30% of H2 release activity but exhibits a significantly stronger down-shift in pH optimum than the largely inactive variant E279Q (Fig. 2c). As shown in Fig. 2 and Supplementary Fig. 2, no significant differences between CpI and HydA1 and their respective SDM variants were observed, underlining the universal impact of the putative PT pathway. To probe the relevance of the putative PT pathway for catalytic proton release, we also investigated the pH-dependency of H2 oxidation. It is known from electrochemistry experiments with different [FeFe]-hydrogenases that the enzyme exhibits an increasing H2-oxidation rate with increasing buffer pH6. Accordingly, the pH range used in this assay was extended to pH 10. As shown in Supplementary Fig. 3, H2-uptake activity of native HydA1 generally enhanced with decreasing H3O+ concentrations, exhibiting a nearly fivefold rate-increase per pH unit between pH 6 and 8. After an intermediate drop between pH 8 and 9 the H2-oxidation activity further increased to nearly 20.000 µmol H2 per mg per min between pH 9 and 10. The absolute activities of SDM variants were significantly diminished compared to wild-type HydA1, reaching a pH optimum of at best 6% (R148A). Up to pH 9, variant R148AHydA1 showed a wild-type-like trend for the pH-activity profile of relative H2-oxidation activity, while instead of a second increase between pH 9 and 10, the activity strongly declined, suggesting that this second increase is connected with the titration of the guanidine base of R148. The pH-activity profile of E144AHydA1 was quite similar to wild-type HydA1, despite a flattening out of the local maximum at around pH 8. The latter suggests that this local activity maximum is depending on the presence of the surface-exposed Glu residue. S189A and C169D only showed single activity peaks at pH 8 or 9, respectively with very low H2-uptake rates of 1–2%, compared to wild type.
We recently showed that decreased proton release efficiency in the presence of H2 leads to an accumulation of the hydride state, Hhyd29. This H-cluster intermediate carries a terminal hydride and represents the first catalytic intermediate after heterolytic cleavage of H226,27,28,29. In previous studies we exploited this behavior to demonstrate that positions C169HydA1/C299CpI and E141HydA1/E279CpI contribute to catalytic PT29. Non-conservative substitutions to alanine that interrupted the PT pathway led to an enrichment of Hhyd instead of adopting a mix of reduced redox species (see below). Here this approach was applied to link PT activity of residues in the putative PT pathway to the steady-state equilibrium of redox species in the presence of H2. Employing ATR-FTIR spectroscopy, we probed CpI and HydA1 wild-type protein and the respective SDM variants at pH 8 under either N2 or H2 (Fig. 3). When flushed with N2, auto-oxidized HydA1 wild type and SDM variants uniformly exhibit peaks around 1964, 1939, and 1802 cm−1 characteristic for the Hox state34. HydA1 and CpI wild-type enzymes adopt a mixture of reduced states when exposed to H2, predominantly Hred/Hsred and Hred´34,35. In contrast, most variants populate the Hhyd state that can be enriched in wild-type enzyme only at pH 4 (see Supplementary Fig. 4). Similar results were achieved for the corresponding variants of CpI as shown in Fig. 3c, d. HydA1 variants C169D, S189A, E144D, and R148A, as well as corresponding CpI variants show wild-type-like spectra by adopting a mixture of reduced states instead of Hhyd (Fig. 3b). With the exception of S189AHydA1/S319ACpI, these variants retain a higher level of H2-release activity (Fig. 1b), which explains why in these cases Hhyd cannot be trapped at pH 8; just as wild-type enzyme, these variants populate Hhyd when titrated to significantly lower pH values (Supplementary Fig. 4). Interestingly, variant C169D and the corresponding variant of CpI (C299D) seem to be incapable of accumulating Hhyd under any of the conditions applied here.
For eight CpI variants (holoenzyme) and three HydA1 variants (apo-enzyme) protein crystals were obtained. Their structures were solved and refined to resolutions of 1.45–2.76 Å, allowing us to gain insight into the structural consequences of SDM. Corresponding to earlier crystal structure data, for all CpI variants a space group of P1 21 1 was observed with two copies in the asymmetric unit cell (chain A and B)19,25. Crystallization of HydA1 exclusively succeeded for apo-protein, which carries the [4Fe]H-cluster but lacks the [2Fe]H-site (Supplementary Fig. 5). However, comparisons of crystal structures of CpI apo- and holo-protein with apo-HydA1 demonstrated that a lack of cofactor does not affect the configuration of the putative PT pathway19,36. Variants of HydA1 apo-protein were crystallized in space group P32 2 1, with a single chain in the asymmetric unit. Overall, SDM did not induce unspecific structural changes as deduced from superposition with wild-type enzyme and corresponding root-mean square deviations of Cα atoms (Supplementary Table 2) and the the H-cluster was largely present for all CpI variants (Supplementary Table 5).
In the following, local structural changes in the putative PT pathway will be described. We start from the surface-exposed residues R286CpI and E282CpI, will continue addressing the median positions S319CpI and E279CpI, and end with C299CpI in the vicinity of the H-cluster.
According to their close inter-residue distance of 2.8 Å, R286CpI may function as a salt bridge partner of deprotonated E282CpI (Fig. 4, right and Supplementary Figs. 6, 7) and thus could be involved in the PT mechanism. R286CpI is further part of an extensive hydrogen-bonding network, which includes the carboxyl group of E282CpI, histidine residues H565CpI and H569CpI, and surface-bound H2O molecules. In case of variant R286A significantly fewer water molecules are observed in the same region (Supplementary Fig. 8). We therefore assume that the loss of the guanidine group in R286A destabilizes the network of proton-accepting/donating surface-bound H2O molecules near the entrance to the PT pathway. However, as the essential chain of PT pathway residues is not directly affected, the influence of this exchange on catalytic activity is comparatively weak (Fig. 1b).
In E282A, two water molecules occupy the space of the missing carboxylic acid group (chain B, Wat717 and Wat974) (Fig. 4, top middle and Supplementary Fig. 13a). The distances between these water molecules and the hydroxyl group of S319 are 2.5 and 3.1 Å (see Supplementary Fig. 7), respectively. They are close enough to rescue the PT activity between surface water and S319CpI further downstream the PT pathway. In the crystal structure of apo-E144AHydA1, one water molecule remains in a very similar position (Supplementary Fig. 5). It is therefore not surprising that both variants exhibit 50% wild-type activity (Fig. 1). In variant E282QCpI, the glutamine residue is potentially stabilized by two hydrogen bonds (H-bonds) unrelated to the putative PT pathway thus, blocking PT and rendering the enzyme largely inactive (Supplementary Fig. 7). Additionally, the two different conformations of Q282CpI in chain A and B (Supplementary Fig. 7) indicate structural flexibility at the entrance of the PT pathway, which may support the residual activity measured for this variant.
In variant S319ACpI, the carboxyl group of E282CpI is slightly shifted outward, probably due to the lack of the hydroxyl group at position 319CpI. In wild-type enzyme S319CpI acts as H-bond donor to E282CpI and drags its carboxy group further into the PT pathway (Fig. 4). In S319ACpI an unprecedented water molecule (Wat735) is located between E282 and A319, and the nearby water molecule Wat873 of wild-type CpI (chain B) is missing here, suggesting a translocation enabled by the unoccupied space of the missing hydroxyl group of position 319 and the slight outward shift of E282 (see Supplementary Fig. 12c). Although the distance to the carboxyl group of E279 amounts to 5.8 Å, the presence of Wat735 may explain the dramatically diminished yet still detectable H2 release activity of variant S319ACpI (6%).
The conformational differences between E279QCpI and wild-type CpI are insignificant, except that the carboxamide group of Q279CpI is slightly twisted relative to the original carboxyl group (Fig. 4). In the corresponding HydA1-variant E141Q, the glutamine residue precisely adopts the conformation of the glutamic acid residue in HydA1 wild type (Supplementary Fig. 5). For the corresponding alanine variant of both, HydA1 and CpI, shifts of the conserved water molecule in the putative PT pathway can be observed. While glutamine occupies the entire space of the native carboxyl group and thus clearly interrupts the putative PT pathway, an exchange to alanine may support additional H2O molecules to bridge the gap. However, in contrast to corresponding substitutions at the surface-exposed glutamate E282CpI, in neither E141AHydA1 (Supplementary Fig. 5) nor E279ACpI25 (Fig. 4) additional H2O molecules are observed. This explains why exchanges of the median glutamate to either alanine or glutamine lead to residual activities of below 1% (Fig. 1).
In case of C299DCpI, the carboxyl group of the side chain establishes H-bonds (2.7 Å) with the amine group of [2Fe]H and the “conserved” water molecule Wat708 (corresponding to Wat826 in chain B of 4XDC19) (Fig. 4). Unsurprisingly, protein samples of C169DHydA1 and C299DCpI both retain a rather high level of H2 release activity (30–80%, Fig. 1). This is not the case when substituting cysteine for alanine, which for both enzymes leads to a complete loss of catalytic activity. Interestingly, the crystal structure of C299ACpI reveals that the removal of the thiol group causes an additional H2O molecule to occupy the vacant space. While the additional H2O is expected to rescue PT activity, it obviously does not restore enzymatic activity. A summary of the characteristic experimental data gained for each enzyme variant is presented in Supplementary Table 1.
Fast proton transfer between bulk solvent and the H-cluster is a precondition for the high catalytic turnover rates of [FeFe]-hydrogenases. Mainly based on theoretical studies, four potential PT pathways have been discussed in literature15,16,17,18,20,37,38. The putative trajectory examined in this study has been favored as key catalytic PT pathway. However, unique experimental evidence for this assumption was lacking. A characterization of PT does not only help to understand the catalytic mechanism but will further contribute valuable parameters for the design of bio-inspired inorganic catalysts20. For artificial hydrogen catalysts, proton supply is often rate-limiting, e.g. due to the lack of a defined proton relay, thus making it necessary to add strong acids and adjust a very low solvent pH39. In this study, corresponding sets of SDM variants of CpI and HydA1 were generated to uncover the individual contribution of residues to PT in the putative PT pathway of [FeFe]-hydrogenases25,40. The complementary data gained from pH-activity profiles, crystal structure analysis, and ATR-FTIR spectroscopy unanimously show that the residues of this pathway are key for the supply and release of protons during catalytic turnover.
The crystal structures of wild-type CpI and its SDM variants reflect the Hox state19, which is commonly accepted to be the resting state of hydrogen turnover. To obtain a general survey on the putative hydrogen-bonding pattern in Hox, the distances between consecutive PT partners in wild-type CpI and the nine SDM variants were derived from the corresponding structures (Fig. 5 and Supplementary Fig. 7). On the basis of CpI crystal structure 4XDC19, PROPKA empirically predicted pKa values of 8.6 and 3.5 for E279 and E282, respectively (pH 8)41. The relatively high pKa of E279CpI reflects its hydrophobic environment compared to the surface-exposed E282CpI and suggests E279CpI to be largely protonated while E282CpI probably resides in the deprotonated state. Starting at the H-cluster, possible hydrogen-bonding interactions can be assumed for adt-NH/C299CpI and C299CpI/Wat826, both of which exhibit comparatively large distances of 3.5 and 3.2 Å42. Molecular dynamics simulations proposed a low H-bond occupancy between adt-NH and the thiol group of C299CpI and a stronger contact between E282CpI and R286CpI to be preconditions for H2 oxidation starting from Hox20. This proposal cannot be confirmed by our crystallographic data, which show the C299 thiol group in an intermediate orientation. However, we cannot rule out the possibility that such differences between theoretical data and structural information are influenced by the non-physiological crystallization conditions or low temperature crystal storage in liquid nitrogen, which might favor certain configurations in the PT pathway. The short distance of 2.5 Å between Wat826 and E279CpI suggests a strong H-bond, in contrast to the adjoined pair E279CpI/S319CpI, for which a distance of 3.6 Å again indicates a fairly weak interaction. The putative hydrogen bond between S319/E282 can be estimated to be of moderate strength (2.9 Å). E282CpI potentially interacts with at least one surface-bound H2O molecule (Wat990). Finally, arginine R286CpI (calculated pKa: 12.4) may serve as a putative salt bridge or hydrogen-bonding partner at close distance to E282CpI (2.8 Å). The combination of calculated pKa values and H-bond distances described in this study suggests for Hox the H-bond pattern depicted in Fig. 5.
Electron densities in the omit maps of crystal structures for wild-type CpI and most of the SDM variants examined here are unambiguously oriented (Fig. 6) and show low displacement factors (Supplementary Table 8). This is in contrast to quantum mechanics/molecular mechanics simulations by Long et al.38 who proposed a conformational bi-stability for the two glutamic acid residues E282 and E279 in the catalytic PT pathway. Our data rather favor a model in accordance with the Grotthuss mechanism, based on simultaneous deprotonation/protonation events according to a bucket line between strictly orientated residues and water molecules43. However, a bi-stability of the glutamate residues under turnover conditions might still be a valid interpretation, as it could also be regarded as switching between two different hydrogen-bonding patterns. In both cases, the direction of PT would be solely determined by the redox state of the [2Fe]H-cluster (see Supplementary Fig. 9 and Supplementary Discussion).
The FTIR data presented here were recorded under steady-state conditions applying changes in gas atmosphere and pH. As protons are reactants in hydrogen turnover (H2 ⇌ 2H+ + 2e−), it is plausible to assume that mutagenesis in the catalytic PT pathway influences the equilibrium of H-cluster species. This was exemplified above for those variants that accumulate Hhyd. A similar effect on the dynamic equilibrium of catalytic states has been observed for [NiFe]-hydrogenases44,45. The substitution of E17 for glutamine in the putative PT pathway of soluble hydrogenase I (SHI) of Pyrococcus furiosus disabled proton-coupled electron transfer (PCET) between two catalytic states (Nia-C and Nia-S)44, which was indicative by an accumulation of intermediates Nia-I1 and Nia-I2. For [FeFe]-hydrogenases, the hydride state has been demonstrated to accumulate under H2-oxidation conditions if proton efflux is severely restricted. This may be the result of an oversaturation of the native PT pathway due to the enhanced proton pressure, i.e. low bulk pH29,34. It can further be the result of [2Fe]H derivatization or eliminating protonatable side chains in the catalytic PT pathway29. The hydride state has been shown to exhibit an uncharged azadithiolate ligand (adt-NH)37,46 that implies a transient intermediate with a protonated ammonium cation, e.g. as proposed by Reijerse et al. (HhydH+)27. Accordingly, oxidation of Hhyd can be explained by PCET that relies on a functional release of protons. All PT pathway variants specifically react with H2 and it can be concluded that at least one proton is always injected into the hydrogen-bonding network, despite the compromised PT pathway. Single deprotonation appears possible as the next proton-binding site downstream C299CpI would be Wat826/Wat1120, which could form a Zundel-ion-like configuration upon protonation47 (H5O2+, Fig. 5 and Supplementary Fig. 12a). Neither of the two H2O molecules has been directly affected by mutagenesis. It may be concluded that PT is only blocked if a protonation of Wat826/Wat1120 yield a Zundel-ion-like configuration, stabilized due to increased pKa in comparison to H3O+.
The favorable position of the acidic asparagine side chain in the structure of C299DCpI effectively connects adt-NH of the [2Fe]H moiety with the conserved water molecule Wat708 in the otherwise intact PT pathway, rendering this variant significantly active. This comparatively high level of absolute activity and the fact that both corresponding variants, C169DHydA1 and C299DCpI, exhibit significant shifts in their pH-dependent activity optimum to the acidic range might explain why this type of variant refuses to accumulate Hhyd even at pH 4, while instead effectively continuing the turnover process as evident according to the comparatively dominant infrared bands for the reduced states Hred and Hsred (1890 and 1882 cm−1 for C169D) recorded at pH 4 (see Supplementary Fig. 4). In case of variant C299ACpI, the first deprotonation step (HhydH+ ⇌ Hhyd + H+) seems impossible but Fig. 3 shows that this variant very effectively accumulates Hhyd under H2. Thus, an alternative proton acceptor has to be assumed. The additional H2O molecule (Wat962), which is trapped in the space of the missing thiol group of variant C299A would be a plausible proton acceptor (Fig. 6 and Supplementary Fig. 13b). As it is well-positioned to bridge the adt-ligand (distance 3.4–3.7 Å) and Wat826 (distance 3.6–3.7 Å), it seems surprising that we are unable to measure any significant catalytic activity. This could be due to an unfavorable pKa difference among the Zundel-ion-like complex, the uncharged adt-ligand and glutamic acid E279Cp48. The pKa calculated for the adt-ligand in Hhyd is significantly larger than the one determined for Hox in mimics of the [2Fe]H-cluster49. However, no re-protonation of Hhyd is observed. We assume that the proton is trapped in a Zundel-ion-like configuration with a dangling H2O molecule Wat962 downstream of position A299 that prevents E279 from being re-protonated (Fig. 6a, C299A). Upon oxidation of H2, proton release via the H-cluster may restrict re-protonation of Hhyd and steady-state accumulation of HhydH+. Recently, an alanine was identified at the position homologous to C299CpI in the newly described sensory [FeFe]-hydrogenase HydS of Thermotoga maritima, which showed very low H2 release and oxidation activities (<5% of HydA1)50. However, HydS is clearly more active than the C → A variants of CpI and HydA1 suggesting a slightly different situation for HydS. Nevertheless, the very low activity level of this sensory-type [FeFe]-hydrogenase overall agrees very well with our results.
For variant E279ACpI the deprotonation of the [2Fe]H-cluster, which produces the hydride state would lead to the formation of the Zundel-ion-like configuration. However, due to the large distance between Wat875 and S319 (up to 4.5 Å) the proton cannot proceed any further. As a consequence, Hhyd with its deprotonated adt-NH ligand would be stabilized in presence of H2 (Fig. 6b, E279A). In case of the corresponding exchange at the surface-exposed glutamate in E282ACpI, two H2O molecules (chain B: Wat717 and Wat974) invading from the solvent are able to bridge the gap of the missing carboxyl group (Fig. 6c, E282A and Supplementary Fig. 13a) thereby rescuing a large fraction of H2-release activity (~50%). The fact that E282ACpI is rescued by an H2O moiety while in E279ACpI and C299ACpI the gap in the catalytic PT pathway cannot be closed, is probably connected to the superior accessibility at the protein surface and the specific functional requirements (such as e.g. pKa) of the position to be rescued within the PT pathway chain. This illustrates the significance of individual residues for the catalytic PT pathway, which appears to increase from surface to catalytic center. However, the lack of H2 release activity resulting from the non-conservative exchange in SDM variant E282QCpI (Fig. 6d, E282Q), which prohibits H2O access due to steric reasons demonstrates the necessity of a protonatable position at the entrance of the catalytic PT pathway (for SDM variants E279Q and S319A see Supplementary Figs. 10-11). Supplementary Table 9 provides a survey of the presumptive functions of all positions in the PT pathway, according to the conclusions drawn here.
The multilayered approach followed here, comprising enzyme kinetics, ATR-FTIR spectroscopy, and crystal-structure analysis, unambiguously verifies the herein proposed PT pathway to be the main trajectory for substrate/product transfer in [FeFe]-hydrogenases. From surface to active center the impact of substitutions along the PT pathway increases, with the inner part between E279CpI and C299CpI being the most vital positions for basic PT function. A model of the PT mechanism in line with the presented data further suggests a major role for the H2O-cluster and a Zundel-ion-like configuration comprising Wat826 and Wat1120 in regulating PT pathway function. External H2O molecules can rescue PT function (1) if provided sufficient space and accessibility, (2) if gaps to neighboring PT pathway positions do not exceed H-bond distance (≤4 Å), and (3) if the overall pKa-gradient of the PT chain is not severely imbalanced. How PT is precisely coupled and synchronized with electron transfer between [2Fe]H and [4Fe]H remains to be further elucidated, possibly by a time-resolved characterization of the catalytic mechanism under sub-turnover conditions. The variants described here are dramatically slowed down in PT efficiency and provide excellent models to resolve the individual steps. In conclusion, the herein presented data give valuable insights into the molecular parameters that enable and tune PT in redox active proteins, thus providing useful guidelines for de novo catalyst design.
Expression constructs were generated using the QuikChange® Site-Directed Mutagenesis Kit from Stratagene using either pET21b-HydA1Cr or pET21b-CpI as template and corresponding mismatch primers29, listed in Supplementary Table 6. The resulting constructs for the expression of SDM variants were verified via sequencing (3130xl Genetic Analyzer; Applied Biosystems) and used for the transformation of Escherichia coli expression host strain BL21 (DE3) ΔiscR51, which was kindly provided by Patrik R. Jones.
Protein expression and maturation
Handling of expression cultures and enzyme variants was entirely done under strictly anaerobic conditions in a glove box (Coy laboratory) atmosphere of 98.5% N2 and 1.5% H2. HydA1 and CpI proteins were expressed in absence of maturases HydE, -F, and -G52 as described previously and therefore lack the catalytically essential [2Fe]H sub-cluster in the original state after purification. Cell disruption was achieved by ultrasonication. Cell debris was separated from the supernatant by ultracentrifugation and filtration (pore size 0.2 µm). Protein purification was done employing strep-tactin/strep-tag II (IBA GmbH) affinity chromatography. HydA1 and CpI apo-proteins were maturated in vitro by adding the synthetic mimic of the native [2Fe]H-complex (Fe2[µ-(SCH2)2NH] (CN)2(CO)4[Et4N]2) ([2FeH]MIM) at a 10-fold molar excess31. [2FeH]MIM was synthesized following literature procedures and kept in 100 mM K2HPO4/KH2PO4 buffer (pH 6.8) at −80 °C53. After an incubation period of 1 h at 25 °C, size-exclusion chromatography was employed to remove redundant complex, using NAP-5 columns (GE Healthcare), which were equilibrated with 100 mM Tris–HCl (pH 8) containing 2 mM sodium dithiolate (NaDT) prior to use. Proteins were concentrated using 30 kDa Amicon Ultra centrifugal Filter units (Merck Millipore) and stored anaerobically at −80 °C. SDS-polyacrylamide gel electrophoresis and Bradford assays were employed to verify protein purity and to calculate the resulting protein concentration.
H2-evolution rates of enzyme variants (0.4–4 µg, depending on the level of residual activity) were determined employing a standard in vitro assay, comprising 100 mM NaDT as sacrificial electron donor and 10 mM methyl viologen as electron mediator in 0.1 M K2HPO4/KH2PO4, pH 6.8. The reaction volume of 2 ml was prepared in air-tight suba-seal vessels and purged with argon for 5 min prior to the H2-production period of 20 min in a shaking water bath, kept at 37 °C. The H2-production yield was determined by analyzing 400 µl of the sample headspace via gas-chromatography (GC-2010, Shimadzu). For the determination of pH-activity profiles, four different buffers were used and adjusted to individual pH values according to their respective buffer range, covering pH 5–10 with a resolution of 0.5 pH units: pH 5–6.5 (200 mM MES-NaOH, 2-(N-morpholino) ethanesulfonic acid); pH 7–7.5 (200 mM MOPS-NaOH, 3-(N-morpholino) propanesulfonic acid); pH 8–9 (200 mM Tris–HCl); and pH 10 (200 mM CAPS-NaOH, N-cyclohexyl-3-aminopropanesulfonic acid). pH-dependent H2-uptake activity was monitored performing a colorimetric assay at 25 °C with enzyme amounts between 0.5 ng and 1 µg in an atmosphere of 1 bar H2, using 10 mM benzyl viologen (Sigma-Aldrich) as electron acceptor (ε600: 10 mM−1 cm−1)6.
Crystallization and structural determination of variants
Vapor diffusion hanging drop and sitting drop methods were applied under anaerobic conditions at 277 K to crystallize the SDM variants using protein concentrations between 10 and 15 mg ml−1. Details of the crystallization conditions are summarized in Supplementary Table 3. Mounting was carried out after 5–10 days of crystal growth and crystals selected for data collection were flash-frozen and stored in liquid N2. Diffraction data were collected at 100 K in different beamlines as indicated in Supplementary Table 3 and processed using XDS54. Phenix55 and Coot56 were employed for molecular replacement (starting models for CpI and HydA1 were 4XDC19 and 3LX436 respectively) and structural refinement. The details of crystallographic data gained for each of the variants are summarized in Supplementary Table 4.
Wild-type and SDM variants of HydA1 and CpI were probed by in situ ATR-FTIR spectroscopy29,34,57 using a rapid-scan Tensor 27 spectrometer (Bruker Optik, Germany) equipped with a three-reflection ZnSe/silicon crystal ATR cell (Smith Detection, USA). The spectrometer was kept under anaerobic conditions in a vinyl glove box (Coy Laboratories, USA) under a water-free atmosphere of 99% N2 and 1% H2. All experiments were performed at room temperature, on hydrated films and in the dark. The oxidized state Hox was populated under 100% N2 ambient partial pressure while the reduced states (Hred/Hsred, Hred′, and Hhyd) were observed in the presence of H2 exclusively. All spectra shown in Fig. 3 were mathematically corrected for the broad combination band of liquid H2O at around 2120 cm−1 and normalized to unity.
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
The coordinates and structure factors for all structures are deposited in the PDB as 6GLY, 6GLZ, 6GM0, 6GM1, 6GM2, 6GM3, 6GM4, 6GM5, 6GM6, 6GM7, and 6GM8. Further data supporting findings of this study are available from the corresponding authors upon reasonable request. A Reporting Summary for this Article is available as a Supplementary Information file.
Frey, M. Hydrogenases: hydrogen-activating enzymes. ChemBioChem 3, 153–160 (2002).
Vignais, P. M. & Billoud, B. Occurrence, classification, and biological function of hydrogenases: an overview. Chem. Rev. 107, 4206–4272 (2007).
Vignais, P. M., Billoud, B. & Meyer, J. Classification and phylogeny of hydrogenases. FEMS Microbiol. Rev. 25, 455–501 (2001).
Lubitz, W., Ogata, H., Rüdiger, O. & Reijerse, E. Hydrogenases. Chem. Rev. 114, 4081–4148 (2014).
Krassen, H. et al. Immobilization of the [FeFe]-hydrogenase CrHydA1 on a gold electrode: design of a catalytic surface for the production of molecular hydrogen. J. Biotechnol. 142, 3–9 (2009).
Lampret, O. et al. Interplay between the CN- ligands and the secondary coordination sphere of the H-cluster in [FeFe]-hydrogenases. J. Am. Chem. Soc. 139, 18222–18230 (2017).
Gloaguen, F., Lawrence, J. D., Schmidt, M., Wilson, S. R. & Rauchfuss, T. B. Synthetic and structural studies on [Fe2(SR)2(CN)x(CO)6-x]x-as active site models for Fe-only hydrogenases. J. Am. Chem. Soc. 123, 12518–12527 (2001).
Hexter, S. V., Grey, F., Happe, T., Climent, V. & Armstrong, F. A. Electrocatalytic mechanism of reversible hydrogen cycling by enzymes and distinctions between the major classes of hydrogenases. Proc. Natl Acad. Sci. USA 109, 11516–11521 (2012).
Takahashi, E. & Wraight, C. A. Small weak acids reactivate proton transfer in reaction centers from Rhodobacter sphaeroides mutated at AspL210 and AspM17. J. Biol. Chem. 281, 4413–4422 (2006).
Chang, Y. H., Chuang, L. Y. & Hwang, C. C. Mechanism of proton transfer in the 3α-hydroxysteroid dehydrogenase/carbonyl reductase from Comamonas testosteroni. J. Biol. Chem. 282, 34306–34314 (2007).
Whitehead, S. J., Iwaki, M., Cotton, N. P., Rich, P. R. & Jackson, J. B. Inhibition of proton-transfer steps in transhydrogenase by transition metal ions. Biochim. Biophys. Acta 1787, 1276–1288 (2009).
Faxen, K., Salomonsson, L., Adelroth, P. & Brzezinski, P. Inhibition of proton pumping by zinc ions during specific reaction steps in cytochrome c oxidase. Biochim. Biophys. Acta 1757, 388–394 (2006).
Garczarek, F. & Gerwert, K. Functional waters in intraprotein proton transfer monitored by FTIR difference spectroscopy. Nature 439, 109–112 (2006).
Gerwert, K., Freier, E. & Wolf, S. The role of protein-bound water molecules in microbial rhodopsins. Biochim. Biophys. Acta 1837, 606–613 (2014).
Cornish, A. J., Gärtner, K., Yang, H., Peters, J. W. & Hegg, E. L. Mechanism of proton transfer in [FeFe]-hydrogenase from Clostridium pasteurianum. J. Biol. Chem. 286, 38341–38347 (2011).
Hong, G., Cornish, A., Hegg, E. & Pachter, R. On understanding proton transfer to the biocatalytic [Fe-Fe]H sub-cluster in [Fe-Fe]H2 ases: QM/MM MD simulations. Biochim. Biophys. Acta 1807, 510–517 (2011).
Nicolet, Y., Piras, C., Legrand, P., Hatchikian, C. E. & Fontecilla-Camps, J. C. Desulfovibrio desulfuricans iron hydrogenase: the structure shows unusual coordination to an active site Fe binuclear center. Structure 7, 13–23 (1999).
Sode, O. & Voth, G. A. Electron transfer activation of a second water channel for proton transport in [FeFe]-hydrogenase. J. Chem. Phys. 141, 22D572 (2014).
Esselborn, J. et al. A structural view of synthetic cofactor integration into [FeFe]-hydrogenases. Chem. Sci. 7, 959–968 (2016).
Ginovska-Pangovska, B. et al. Molecular dynamics study of the proposed proton transport pathways in [FeFe]-hydrogenase. Biochim. Biophys. Acta 1837, 131–138 (2014).
Berggren, G. et al. Biomimetic assembly and activation of [FeFe]-hydrogenases. Nature 499, 66–69 (2013).
Cornish, A. J. et al. Single-amino acid modifications reveal additional controls on the proton pathway of [FeFe]-hydrogenase. Biochemistry 55, 3165–3173 (2016).
Morra, S. et al. Site saturation mutagenesis demonstrates a central role for cysteine 298 as proton donor to the catalytic site in CaHydA [FeFe]-hydrogenase. PLoS ONE 7, e48400 (2012).
Morra, S. et al. The effect of a C298D mutation in CaHydA [FeFe]-hydrogenase: insights into the protein–metal cluster interaction by EPR and FTIR spectroscopic investigation. Biochim. Biophys. Acta 1857, 98–106 (2016).
Winkler, M., Esselborn, J. & Happe, T. Molecular basis of [FeFe]-hydrogenase function: An insight into the complex interplay between protein and catalytic cofactor. Biochim. Biophys. Acta 1827, 974–985 (2013).
Mulder, D. W., Guo, Y., Ratzloff, M. W. & King, P. W. Identification of a catalytic iron-hydride at the H-cluster of [FeFe]-hydrogenase. J. Am. Chem. Soc. 139, 83–86 (2017).
Reijerse, E. J. et al. Direct observation of an iron-bound terminal hydride in [FeFe]-hydrogenase by nuclear resonance vibrational spectroscopy. J. Am. Chem. Soc. 139, 4306–4309 (2017).
Rumpel, S., Sommer, C., Reijerse, E., Fares, C. & Lubitz, W. Direct detection of the terminal hydride intermediate in [FeFe] Hydrogenase by NMR spectroscopy. J. Am. Chem. Soc. 140, 3863–3866 (2018).
Winkler, M. et al. Accumulating the hydride state in the catalytic cycle of [FeFe]-hydrogenases. Nat. Commun. 8, 16115 (2017).
Cramer, S. P. et al. Terminal hydride species in [FeFe]-hydrogenases are vibrationally coupled to the active site environment. Angew. Chem. Int. Ed. 130, 10765–10769 (2018).
Esselborn, J. et al. Spontaneous activation of [FeFe]-hydrogenases by an inorganic [2Fe] active site mimic. Nat. Chem. Biol. 9, 607–609 (2013).
Noth, J. et al. [FeFe]-Hydrogenase with chalcogenide substitutions at the H-cluster maintains full H2 evolution activity. Angew. Chem. Int. Ed. 128, 8536–8540 (2016).
Noth, J. et al. Lyophilization protects [FeFe]-hydrogenases against O2-induced H-cluster degradation. Sci. Rep. 5, 13978 (2015).
Senger, M. et al. Protonation/reduction dynamics at the [4Fe–4S] cluster of the hydrogen-forming cofactor in [FeFe]-hydrogenases. Phys. Chem. Chem. Phys. 20, 3128–3140 (2018).
Mebs, S. et al. Bridging hydride at reduced H-cluster species in [FeFe]-Hydrogenases revealed by infrared spectroscopy, isotope editing, and quantum Chemistry. J. Am. Chem. Soc. 139, 12157–12160 (2017).
Mulder, D. W. et al. Stepwise [FeFe]-hydrogenase H-cluster assembly revealed in the structure of HydAΔEFG. Nature 465, 248–251 (2010).
Mulder, D. W. et al. Investigations on the role of proton-coupled electron transfer in hydrogen activation by [FeFe]-hydrogenase. J. Am. Chem. Soc. 136, 15394–15402 (2014).
Long, H., King, P. W. & Chang, C. H. Proton transport in Clostridium pasteurianum [FeFe] hydrogenase I: a computational study. J. Phys. Chem. B 118, 890–900 (2014).
Wang, N., Wang, M., Chen, L. & Sun, L. Reactions of [FeFe]-hydrogenase models involving the formation of hydrides related to proton reduction and hydrogen oxidation. Dalton. Trans. 42, 12059–12071 (2013).
Meyer, J. [FeFe] hydrogenases and their evolution: a genomic perspective. Cell. Mol. Life Sci. 64, 1063 (2007).
Olsson, M. H., Søndergaard, C. R., Rostkowski, M. & Jensen, J. H. PROPKA3: consistent treatment of internal and surface residues in empirical pKa predictions. J. Chem. Theory Comput. 7, 525–537 (2011).
Jeffrey, G. A. An Introduction to Hydrogen Bonding (Oxford Univ. Press, New York, 1997).
Cukierman, S. Et tu, Grotthuss! and other unfinished stories. Biochim. Biophys. Acta 1757, 876–885 (2006).
Greene, B. L., Vansuch, G. E., Wu, C.-H., Adams, M. W. & Dyer, R. B. Glutamate gated proton-coupled electron transfer activity of a [NiFe]-Hydrogenase. J. Am. Chem. Soc. 138, 13013–13021 (2016).
Evans, R. M. et al. Mechanistic exploitation of a self-repairing, blocked proton transfer pathway in an O2-tolerant [NiFe]-Hydrogenase. J. Am. Chem. Soc. 140, 10208–10220 (2018).
Pelmenschikov, V. et al. Reaction coordinate leading to H2 production in [FeFe]-Hydrogenase identified by nuclear resonance vibrational spectroscopy and density functional theory. J. Am. Chem. Soc. 139, 16894–16902 (2017).
Zundel, G. & Metzger, H. Energiebänder der tunnelnden überschuß-protonen in flüssigen säuren. Eine IR-spektroskopische untersuchung der natur der gruppierungen H5O2 +. Z. Phys. Chem. 58, 225–245 (1968).
Linke, K. & Ho, F. M. Water in Photosystem II: structural, functional and mechanistic considerations. Biochim. Biophys. Acta 1837, 14–32 (2014).
Huynh, M. T., Wang, W., Rauchfuss, T. B. & Hammes-Schiffer, S. Computational investigation of [FeFe]-hydrogenase models: characterization of singly and doubly protonated intermediates and mechanistic Insights. Inorg. Chem. 53, 10301–10311 (2014).
Chongdar, N. et al. Unique spectroscopic properties of the H-cluster in a putative sensory [FeFe] hydrogenase. J. Am. Chem. Soc. 140, 1057–1068 (2018).
Akhtar, M. K. & Jones, P. R. Deletion of iscR stimulates recombinant clostridial Fe-Fe hydrogenase activity and H2-accumulation in Escherichia coli BL21 (DE3). Appl. Microbiol. Biotechnol. 78, 853–862 (2008).
Kuchenreuther, J. M. et al. High-yield expression of heterologous [FeFe] hydrogenases in Escherichia coli. PLoS ONE 5, e15491 (2010).
Li, H. & Rauchfuss, T. B. Iron carbonyl sulfides, formaldehyde, and amines condense to give the proposed azadithiolate cofactor of the Fe-only hydrogenases. J. Am. Chem. Soc. 124, 726–727 (2002).
Kabsch, W. XDS. Acta Crystallogr. D. 66, 125–132 (2010).
Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D. 66, 213–221 (2010).
Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr. D. 60, 2126–2132 (2004).
Senger, M. et al. Stepwise isotope editing of [FeFe]-hydrogenases exposes cofactor dynamics. Proc. Natl Acad. Sci. USA 113, 8454–8459 (2016).
Sommer, C. et al. Proton coupled electronic rearrangement within the H-cluster as an essential step in the catalytic cycle of [FeFe] hydrogenases. J. Am. Chem. Soc. 139, 1440–1443 (2017).
Mebs, S. et al. Hydrogen and oxygen trapping at the H-cluster of [FeFe]-hydrogenase revealed by site-selective spectroscopy and QM/MM calculations. Biochim. Biophys. Acta 1859, 28–41 (2018).
Haumann, M. & Stripp, S. T. The molecular proceedings of biological hydrogen turnover. Acc. Chem. Res. 51, 1755-1763 (2018).
M.W. and T.H. gratefully acknowledge financial support from the VolkswagenStiftung (Design of [FeS]-cluster containing Metallo-DNAzymes (Az 93412)) Deutsche Forschungsgemeinschaft (DFG) (the Cluster of Excellence RESOLV EXC1069 and GRK 2341: Microbial Substrate Conversion). J.D. acknowledges support by the China Scholarship Council (CSC) and DFG-EXC1069. M.S. and S.T.S. thank the International Max Plank Research School (IMPRS) on Multiscale Biosystems and the Focus Area NanoScale (Freie Universität Berlin) for financial support. U.-P.A. and F.W. are grateful for financial support by the Deutsche Forschungsgemeinschaft (Emmy Noether grant to U.-P.A., AP242/2-1) and the Fraunhofer Internal Programs (Attract 097-602175). F.W. further thanks the Studienstiftung des deutschen Volkes for a PhD fellowship. We thank the staff of beamlines ID23-1 and BM30A at the ESRF and PXII at the SLS for technical support during X-ray data collection.
The authors declare no competing interests.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Duan, J., Senger, M., Esselborn, J. et al. Crystallographic and spectroscopic assignment of the proton transfer pathway in [FeFe]-hydrogenases. Nat Commun 9, 4726 (2018). https://doi.org/10.1038/s41467-018-07140-x
This article is cited by
JBIC Journal of Biological Inorganic Chemistry (2023)
Nature Communications (2021)