[Fe-Fe]-hydrogenase Reactivated by Residue Mutations as Bridging Carbonyl Rearranges: A QM/MM Study

Abstract

In the current work, we found aqueous enzyme phase reaction pathways for the reactivation of the exogenously inhibited [Fe-Fe]-hydrogenases by O₂, or OH^-, which metabolizes to H₂O^1,2. We used the hybrid quantum mechanics/molecular mechanics (QM/MM) method to study the reactivation pathways of the exogenously inhibited enzyme matrix. The ONIOM calculations performed on the enzyme agree with experimental results³, i.e., wild-type [Fe-Fe]-hydrogenase H-cluster is inhibited by oxygen metabolites. An enzyme spherical region with a radius of 8 Å (from the distal iron, Fe_d) has been screened for residues that prevent H₂O from leaving the catalytic site and reactivate the [Fe-Fe]-hydrogenase H-cluster. In the screening process, polar residues were removed, one at a time, and frequency calculations provided the change in the Gibbs’ energy for the dissociation of water (due to their deletion). When residue deletion resulted in significant Gibbs’ energy decrease, further residue substitutions have been carried out. Following each substitution, geometry optimization and frequency calculations have been performed to assess the change in the Gibbs’ energy for the elimination H₂O. Favorable thermodynamic results have been obtained for both single residue removal (ΔG_ΔGlu³⁷⁴ = -1.6 kcal/mol), single substitution (ΔG_Glu³⁷⁴_His = -3.1 kcal/mol), and combined residue substitutions (ΔG_Arg¹¹¹_Glu;Thr¹⁴⁵_Val;Glu³⁷⁴_His;Tyr³⁷⁵_Phe = -7.5 kcal/mol). Because the wild-type enzyme has only an endergonic step to overcome, i.e., for H₂O removal, by eliminating several residues, one at a time, the endergonic step was made to proceed spontaneously. Thus, the most promising residue deletions which enhance H₂O elimination are ΔArg¹¹¹, ΔThr¹⁴⁵, ΔSer¹⁷⁷, ΔGlu²⁴⁰, ΔGlu³⁷⁴, and ΔTyr³⁷⁵. The thermodynamics and electronic structure analyses show that the bridging carbonyl (CO_b) of the H-cluster plays a concomitant role in the enzyme inhibition/reactivation. In gas phase, CO_b shifts towards Fe_d to compensate for the electron density donated to oxygen upon the elimination of H₂O. However, this is not possible in the wild-type enzyme because the protein matrix hinders the displacement of CO_b towards Fe_d, which leads to enzyme inhibition. However, enzyme reactivation can be achieved by means of appropriate amino acid substitutions.

Introduction

[Fe-Fe]- and [Ni-Fe]-hydrogenases are two major classes of enzymes that reversibly catalyze the apparently simple reaction of protons and electrons to molecular hydrogen, 2H⁺ + 2e^- ⇄ H₂, which occurs in anaerobic media. In living systems, of the two metalloproteins, [Fe-Fe]-hydrogenases are mostly used for H₂ production, having a reactivity of about 2 orders of magnitude larger than [Ni-Fe]-hydrogenases. These enzymes are found in many bacteria, simple eukaryotes, and archaea where they provide H₂ for the metabolic processes of these life forms. By means of H₂ oxidation, ATP synthesis exploits H₂ as an energy source, whereas H₂ synthesis results from the metabolic disposal of excess electron (with available H⁺s), or from pyruvate fermentation. Proteins, such as ferredoxins, cytochrome C3, and cytochrome C6, act as physiological e^- donors or acceptors^4-6. The exploration of alternative energy sources has kindled great interest in hydrogenase research. The reason for studying biological H₂ production is to clarify the complex mechanism (for hydrogen synthesis), which may help researchers produce clean fuel⁷, using certain anaerobic organisms^8-11.

This theoretical study aims to find ways of making these enzymes function aerobically (to provide clean fuel, viz., H₂^α), because they become inactivated by exogenous ligands such as O₂, OH^-, and H₂O^1,3. Water is the metabolic product of the inactivated catalytic site, i.e., (Fe_d-)OO → (Fe_d-)H₂O¹, and it also binds to the hydrogenase active site in its resting state, viz., Fe^IIFe^{II12, 13}.

By performing Density Functional Theory (DFT) calculations on the H-cluster, with H₂O, OH^-, and O₂ bound to Fe_d, (redox states Fe^II-Fe^II), Liu and Hu³ have inferred, based on agreement between calculated and experimental vibrational frequencies of the three endogenous CO ligands, that OH^- is the oxygen species which inhibits hydrogenases. The X-ray structures of [Fe-Fe]-hydrogenases, from Clostridium pasteurianum (CPI)¹² and Desulfovibrio desulfuricans (DdH)¹⁴, can be used to theoretically investigate their functions via biochemical pathways^3,15. Since former DFT, and hybrid quantum mechanics/molecular mechanics (QM/MM) calculations^1-3,15-22 have shown success in clarifying certain aspects of the catalytic properties of the H-cluster, similar methodologies are also used in our investigation. In our study, as well as in other computational studies^1-3,15,19,20, CH₃-S^- has been substituted for cysteine, Cys³⁸², and H⁺ for the proximal cubane^β.

The active site of hydrogenases, viz., the H-cluster (Figure 1), is comprised of a dimetal complex, [Fe-Fe], with the metal centers being bridged by di(thiomethyl)amine (DTMA), and a cubane subunit, [Fe₄-S₄]²⁺. The iron atoms are coordinated by endogenous ligands, viz., two cyanides (CN^-), two terminal carbonyls (CO_t), and a bridging carbonyl (CO_b)^γ. An Fe atom, which is part of the proximal cubane subunit, [Fe₄-S₄]²⁺_p, is linked to the Fe_p of the di-iron subunit, [Fe_p-Fe_d], through a cysteinyl sulfur (or S_γ of Cys³⁸²).

Figure 1.

The H-cluster Structure.

In spite of the di-iron H-cluster subunit redox states, the proximal cubane remains in oxidation²³ state II, [Fe₄-S₄]²⁺_p. Computational and experimental [Fe-Fe]-hydrogenase H-cluster (and synthetic H-cluster analogues) research^{1-3,6,7,13-15,19,21,24-52} corroborates the potential redox states of the di-iron H-cluster subunit, Fe_p-Fe_d, where Fe_p^I-Fe_d^I, EPR silent, is the reduced di-iron H-cluster subunit, Fe_p^II-Fe_d^I, paramagnetic, is the partially oxidized, and catalytically active di-iron subunit, and Fe_p^II-Fe_d^II, EPR silent^14,53, is the fully oxidized, inactive biferrous subunit, and has an OH^- or H₂O molecule bound to the Fe_d^II.

By performing spectroscopic studies on [Fe-Fe]-hydrogenases, which have been purified from Clostridium pasteurianum and Desulfovibrio desulfuricans, their catalytic functions have been elucidated^12-14,54,55. An X-ray crystal structure of CPI hydrogenase shows an (inactivating) oxygen species that may be OH^-, or H₂O bound to the Fe_d of the H-cluster, while the other X-ray structure has an inactivating CO bound to Fe_d^12,13. For the current study, DdH has been selected because its crystal structure has a better resolution (viz., 1.6 Å), than CPI (viz, 1.8 Å)^12,14.

This investigation is subdivided into three parts, viz., thermodynamics, geometric, and electronic analysis, for both wild-type and mutated (residue substituted) DdH. These analyses were carried out in order to understand the thermodynamic results, their relationship to certain molecular spatial behavior, e.g., CO_b movement, and the electronic structural methods, such as frontier molecular orbitals (FMO), and natural bond orbital partial charges (NBO).

Methods

The ONIOM method⁵⁶ (DFT for the QM region, and the universal force field, UFF⁵⁷, for the MM region, implemented in Gaussian03⁵⁸) has been utilized to determine the reaction thermodynamics, viz., ΔG for individual reaction steps, of the [Fe-Fe]-hydrogenase H-cluster (the active group of DdH) reactivation.

Low spin states (singlet, and doublet), and low oxidation states (I, and II) have been used for the di-irons^3,19 in agreement with experimental and computational data.

The electronic structure of the hydrogenase H-cluster (without proximal cubane) has been determined using DFT method (B3LYP functional^59,60), and 6-31+G(d,p) basis set.

Hydrogen atoms were added to the X-ray crystal structure of DdH using the Gromacs program^61,62 [Brookhaven Protein Data Bank id.1HFE]. After the structure has been solvated [2043 H₂O molecules (in 1 nm layer peripheral to H-cluster)], six Na⁺ ions were randomly incorporated into the solvent (7 Na⁺ ions for clusters 1, 3, 4; 8 Na⁺ ions for cluster 2) to neutralize the negative charges²³ on the H-cluster, medial, and distal cubane/cysteines clusters^δ. At physiological pH (~7), Gromacs program considers negative charges on the acidic residues (28 Asp and 33 Glu), and positive charges on the most basic residues (44 Lys and 15 Arg).

Out of the 13 His found in DdH, only 2 are protonated, and these charges (in conjunction with those from Lys and Arg) are used to neutralize the negatively charged residues, viz., Asp and Glu. Thus, the overall apoprotein charge is zero, except the 12 negatively charged cysteines bound to the iron atoms of the 3 cubanes, where each has a charge of 2+.

Geometry optimizations have been performed in aqueous enzyme phase, where residues in the MM region (except the proximal cubane), and Fe_p and CO_t,p, in the QM region have been kept frozen^ε. The rational for freezing Fe_p, and CO_t,p arises from former optimizations² where they were found to spatially rearrange the least. Once geometry optimizations have been carried out for DdH, frequency calculations are performed in order to obtain thermodynamic data, viz., ΔG. Frequency calculations treat both the apoenzyme and the cubanes as partial charges, whereas [Fe_p-Fe_d] subunit is treated at DFT level.

The DdH apoenzyme and cubanes, viz., proximal, medial, and distal are included in the MM region. The QM region consists of the [Fe_p-Fe_d] subunit, (the moiety of H-cluster), and C_β, and S_γ (cysteinyl sulfur of Cys³⁸²). In order to avoid dangling bonds between the ONIOM layers, two linking hydrogen atoms were added between S_γ and Fe (of the proximal cubane), and between C_α and C_β (of Cys³⁸²).

The UFF charge equilibration method was utilized to describe the electrostatic interactions within the MM region of DdH, whereas the solvent charges (q_O = -0.706 a.u., and q_H = 0.353 a.u.) were obtained from literature³⁰.

Then, a DdH sphere with a radius of 8 Å from Fe_d was investigated regarding the potential inhibitory residues for H₂O removal. H-cluster hindering residues (for H₂O elimination), are identified by using QM/MM geometry optimizations, and then their influence on thermodynamics, and electronic properties of the catalytic site is assessed. Given that water is polar, candidate, potential inhibitory residues should also be polar. Then, potential, polar inhibitory residues are screened to identify the most probable residues that hinder H₂O from leaving the catalytic site. Screening is the process whereby polar residues are removed (from within a sphere of radius 8 Å), one at a time, which is followed by frequency calculations aiming to learn whether the binding energy of water has decreased. If successful, then further residue substitutions are performed, i.e., a neutral, polar residue is substituted for a neutral, nonpolar residue, and an acidic residue is substituted for a basic residue, and vice versa^ζ. Then, after each substitution, geometry optimization is performed, followed by frequency calculations to obtain the Gibbs’ energy of H₂O dissociation.

Lastly, Gibbs’ energies of H₂O and H₃O⁺ (viz., ΔG_H2O = -76.419750, and ΔG_H3O⁺ = -76.598767 Hartrees/molecule), which are needed for thermodynamic analysis, have been obtained by performing frequency calculations on these molecules which were positioned in the H-cluster cavity surrounded by the apoenzyme and the cubanes.

Results

Thermodynamic analysis

The reactivation mechanism of [Fe-Fe]-hydrogenase H-cluster essentially consists of three reaction steps, viz., protonation, reduction, and H₂O elimination. The reactivation pathways (Scheme I, II, and III) proceed with different combinations for the three steps.

Scheme I.

The Reactivation Pathway I of [Fe-Fe]-hydrogenase H-cluster.

Scheme II.

The Reactivation Pathway II of [Fe-Fe]-hydrogenase H-cluster.

Scheme III.

The Reactivation Pathway III of [Fe-Fe]-hydrogenase H-cluster.

For the hybrid calculations, the reductive step, 1 → 2 (Scheme I), of [Fe^II-Fe^II]-hydrogenase H-cluster proceeds rather endergonically, viz., ΔG_Enzyme = +42.6 kcal/mol (Table 1) relative to the gas phase (ΔG_{Gas Φ} = +8.2 kcal/mol, Table 1), which points out that DdH reduction is less spontaneous than in vacuum, emphasizing the stereoelectronic effects of the apoenzyme, the medial, and distal cubanes on the H-cluster. The protonation step, 2 → 3 (Scheme I), is highly exergonic (ΔG_{Enzyme Φ} = -317.9 kcal/mol, see Supplemental Table 1 for protonation free energies using an alternative protonation source). The Gibbs’ energy difference between the reactions of the two environments, (i.e., vacuum vs. enzyme Φ, Table 1), is +91.3 kcal/mol, and points to a more spontaneous reaction in gas phase. In step 3 → 4 (H₂O elimination), the hydrogenase H-cluster calculations show a rather small, endergonic result (ΔG_{Enzyme Φ} = +2.0 kcal/mol), as opposed to the exergonic gas phase outcome (ΔG_{Gas Φ} = -6.6 kcal/mol). The enzymatic H₂O removal is nonspontaneous due to the influence of the protein environment on the H-cluster electronic properties.

Table 1.

Native and residue removed DdH Gibbs’ energies (kcal/mol) for reaction steps of reactivation pathways I, II, and III.

Reaction steps	(+e-) 1 → 2	(+H+) 2 → 3	(-H2O) 3 → 4	(+H+) 1 → 2’	(-H2O) 2’ → 3’	(+e-) 3’ → 4	(+e-) 2’ → 3
Native DdH	+42.6	-317.9	+2.0	-236.8	+22.9	-59.4	-38.4
ΔSer62s*	+38.3	-313.6	+2.0	-232.6	+23.1	-63.9	-42.7
ΔArg111	+59.0	-332.4	-0.7	-251.4	+22.0	-44.7	-22.0
ΔTyr112	+47.5	-322.1	+1.6	-241.1	+22.8	-54.8	-33.6
ΔAsp144	+28.8	-305.0	+3.0	-223.4	+24.1	-73.8	-52.8
ΔThr145	+44.1	-318.8	-0.2	-237.9	+22.8	-59.9	-36.9
ΔGlu146	+32.5	-308.5	+2.5	-227.4	+23.6	-69.7	-48.6
ΔThr148	+41.8	-316.8	+1.8	-235.8	+22.9	-60.3	-39.2
ΔAsp150	+32.8	-308.1	+2.1	-227.0	+23.3	-69.5	-48.3
ΔThr152	+47.1	-322.6	+1.9	-241.6	+23.0	-55.1	-33.9
ΔGlu155	+36.5	-311.4	+1.9	-230.2	+23.0	-65.8	-44.7
ΔThr176	+43.8	-319.5	+2.1	-238.3	+23.2	-58.5	-37.4
ΔSer177	+38.3	-312.8	-0.5	-231.8	+22.9	-66.1	-42.7
ΔGln183	+46.3	-322.3	+2.3	-241.1	+23.3	-55.9	-34.9
ΔSer198	+42.6	-318.0	+2.1	-236.9	+23.2	-59.6	-38.5
ΔLys201	+48.3	-323.5	+1.7	-242.6	+22.7	-53.6	-32.6
ΔSer202	+46.3	-324.1	+2.4	-241.8	+22.5	-56.0	-35.9
ΔAsn207	+46.7	-321.6	+1.5	-240.6	+22.5	-55.2	-34.2
ΔSer230	+38.2	-314.0	+2.3	-232.9	+23.3	-64.0	-42.9
ΔLys237	+69.2	-345.2	+1.2	-264.4	+21.7	-32.1	-11.6
ΔLys238	+47.4	-322.1	+1.7	-241.5	+22.9	-54.4	-33.3
ΔGlu240	+25.6	-300.1	+0.8	-218.9	+23.8	-78.7	-55.6
ΔThr257	+42.8	-318.0	+1.9	-236.9	+23.1	-59.5	-38.3
ΔThr259	+45.9	-320.8	+1.8	-239.9	+23.0	-56.1	-35.0
ΔThr260	+42.1	-317.4	+2.0	-236.4	+23.1	-60.1	-39.0
ΔSer289	+44.0	-319.1	+1.9	-238.1	+22.9	-58.1	-37.0
ΔThr294	+40.4	-316.0	+2.2	-234.9	+23.2	-61.8	-40.8
ΔThr299	+40.9	-316.2	+1.9	-235.1	+23.0	-61.2	-40.1
ΔGlu374	+27.7	-301.3	-1.6	-220.2	+21.5	-76.5	-53.4
ΔTyr375	+41.5	-315.8	+1.1	-234.8	+22.3	-60.6	-39.5
ΔGln388	+44.0	-319.2	+2.0	-238.1	+23.1	-58.2	-37.1
Gas ϕ	+8.2	-409.2	-6.6	-328.3	-16.6	-62.7	-72.7

^*Residue removed DdH;

s = small chain

The DdH protonation, 1 → 2’ (Scheme II), provides an exergonic reaction, viz., ΔG_{Enzyme Φ}= -236.8 kcal/mol, proceeding less spontaneously than in gas phase, viz. ΔG_{Gas Φ} = -328.3 kcal/mol, correlating the trend in Scheme I. The enzyme Φ calculations, for the H₂O removal, 2’→3’, (Scheme II), confer an endergonic result, (ΔG_{Enzyme Φ} = +22.9 kcal/mol), as opposed to the exergonic gas Φ reaction step (ΔG_{Gas Φ} = -16.6 kcal/mol), which results in a difference of +39.5 kcal/mol. Again, the effect of the protein environment is manifest on the individual steps of the reaction mechanism. The final reductive step, 3’ → 4, of the aqueous enzyme Φ (Scheme II) proceeds exergonically, viz., ΔG_{Enzyme Φ} = -59.4 kcal/mol, close to the gas phase result, viz., ΔG_{Gas Φ} = -62.7 kcal/mol.

However, Scheme III, in contrast, definitely provides room for enhancements that could achieve H₂O removal because, in step 3 → 4, the hydrogenase H-cluster calculations provide a rather small, free energy, viz., ΔG_{Enzyme Φ} = +2.0 kcal/mol, which could be changed into an exergonic reaction via mutagenesis. For Scheme III, only the H₂O removal step is endergonic (viz., ΔG_{Enzyme Φ} = +2.0 kcal/mol, as in Scheme I), whereas the other remaining steps [protonation (viz., ΔG_{Enzyme Φ} = -236.8 kcal/mol), reduction (viz., ΔG_{Enzyme Φ} = -38.4 kcal/mol)] are exergonic.

The following investigation addresses possible reactivation mechanisms of DdH mutants, uses the QM/MM method for pathway I, II, and III, and aims for the removal of H₂O.

The wild-type hydrogenase has been residue manipulated in two ways. First of all, the considered residues (Table 1) have been removed one at a time in order to find which are responsible for keeping the water from being displaced. Subsequently, for the six culprit residues, that proved to hinder the removal of water, substitutions had been carried out (as describe in the Methods section), viz., Arg¹¹¹Glu, Thr¹⁴⁵Val, Ser¹⁷⁷Ala, Glu²⁴⁰His, Glu³⁷⁴His, and Tyr³⁷⁵Phe.

In Scheme I, the reduction step, 1 → 2, upon the removal of Glu²⁴⁰, becomes less endergonic by ΔG = +17.0 kcal/mol (relative to wild-type DdH) but not even close to make it spontaneous^η. For 2 → 3, with the removal of a basic residue, i.e., Lys²³⁷ (compared to the wild-type enzyme), the protonation step proceeds more exergonically, viz., by a Gibbs’ energy difference of +27.3 kcal/mol. The last step, 3 → 4, proceeds exergonically, viz., ΔG_{Enzyme Φ} = -1.6 kcal/mol, upon the removal of Glu³⁷⁴ from the apoenzyme, however Scheme I cannot proceed to completion due to the endergonic, reductive step, 1 → 2.

For Scheme II, an improvement has been obtained (as in Scheme I) upon the removal of exactly the same residue for each of the corresponding step, i.e., Lys²³⁷ for protonation (viz., ΔG_{Enzyme Φ} = -264.4 kcal/mol), Glu²⁴⁰ for reduction (viz., ΔG_{Enzyme Φ} = -78.7 kcal/mol), and Glu³⁷⁴ for H₂O elimination (viz., ΔG_Enzyme = +21.5kcal/mol). However, in spite of Gibbs’ energy improvements for all reaction steps, no matter what residues (Table 1) are removed, Scheme II is hindered from completion in the H₂O removal step.

In Scheme III, the only endergonic step is for the removal of H₂O (ΔG = +2.0 kcal/mol), which was, nevertheless, made to proceed exergonically by eliminating several residues, one at a time. The most promising residues are ΔArg¹¹¹, ΔThr¹⁴⁵, ΔSer¹⁷⁷, ΔGlu²⁴⁰, ΔGlu³⁷⁴, and ΔTyr³⁷⁵, and their respective deletion Gibbs’ energies are -0.7 kcal/mol, -0.2 kcal/mol, -0.5 kcal/mol, +0.8 kcal/mol, -1.6 kcal/mol, and +1.1 kcal/mol (Table 1). In 3 → 4, single residue substitutions have been carried out on the six residues that hinder the removal of water, viz., Arg¹¹¹Glu, Thr¹⁴⁵Val, Ser¹⁷⁷Ala, Glu²⁴⁰His, Glu³⁷⁴His, and Tyr³⁷⁵Phe, followed by (sequential) frequency calculations.

Then out of the most successful substitutions, two, three, and four residue combinations were examined by frequency calculations (Table 2). The 1^st two-residue combination is Glu³⁷⁴His, and Tyr³⁷⁵Phe giving ΔG = -5.1 kcal/mol; the 2^nd three-residue combination is Thr¹⁴⁵Val, Glu³⁷⁴His, and Tyr³⁷⁵Phe giving ΔG = -6.2 kcal/mol, and the 3^rd four-residue combination is Arg¹¹¹Glu, Thr¹⁴⁵Val, Glu³⁷⁴His, and Tyr³⁷⁵Phe giving ΔG = -7.5 kcal/mol.

Table 2.

DdH Gibbs’ energies (kcal/mol) of one, two, three, and four residue mutations for reaction step (3 → 4).

Residue substitutions for H2O removal step	3 → 4
Arg111Glu	-0.9
Thr145Val	-1.3
Ser177Ala	-0.1
Glu240His	+1.1
Glu374His	-3.1
Tyr375Phe	-2.1
Combinations of:
Glu374His and Tyr375Phe	-5.1
Thr145Val, Glu374His and Tyr375Phe	-6.2
Arg111Glu, Thr145Val, Glu374His and Tyr375Phe	-7.5

Geometric Considerations

In order to explain Gibbs’ energies between the aqueous enzyme Φ and the gas Φ calculations, DdH H-cluster and H-cluster geometries for the two phases are analyzed, and then the hydrogenase H-cluster distances to the six replaced, juxtaposed residues are presented for the aqueous enzyme Φ.

The wild-type DdH QM/MM calculations for H₂O removal, (3 → 4), reveal a contrasting picture regarding CO_b translation towards Fe_d relative to the gas phase H₂O elimination. That is, in Table 3^θ, 3 → 4, the iron-carbon distance, Fe_d-CO_b, essentially remains constant [(viz., 1.907 Å → 1.908 Å), whereas the gas Φ distance becomes smaller (viz., 1.945 Å → 1.850 Å), and the reaction (3 → 4) is exergonic]^2,15, which may explain why H₂O removal is endergonic for the enzyme Φ.

Table 3.

Interatomic distances (Å) for wild-type and mutated DdH, between Fe_p and Fe_d, Fe_p and CO_b, Fe_d and CO_b, and Fe_d and H₂O, before and after H₂O removal (3 → 4).

Before H2O removal	3 (Wild-type)	3 (Arg111 Glu mutant)	3 (Thr145 Val mutant)	3 (Glu374 His mutant)	3 (Tyr375 Phe mutant)
Fep-Fed	2.626	2.659	2.670	2.690	2.666
COb-Fep	1.959	1.963	1.968	1.975	1.991
COb-Fed	1.907	1.952	1.963	1.945	1.937
Fed-O(H2O)	2.127	2.205	2.181	2.223	2.184
After H2O removal	4 (Wild-type)	4 (Arg111Glu mutant)	4 (Thr145 Val mutant)	4 (Glu374 His mutant)	4 (Tyr375 Phe mutant)
Fep-Fed	2.587	2.584	2.616	2.589	2.582
COb-Fep	1.939	2.006	2.003	2.008	2.005
COb-Fed	1.908	1.928	1.935	1.928	1.906

Thus, it is ascertained that for the aqueous enzyme Φ, an endergonic ligand (H₂O) dissociation is manifested in the Fe_p-CO_b bond contraction (1.959 Å → 1.939 Å). The opposite trend is observed in gas phase, where the Fe_p-CO_b bond is elongated (2.013 Å → 2.232 Å) after H₂O elimination.

Out of the three reaction pathways presented here, only Scheme III is analyzed for geometrical considerations, for it is the only one having the potential for metabolic reactivation by means of residue substitutions. The latter scheme has only the water removal step to be overcome in order to proceed exergonically, while, on the other hand, the other schemes cannot be made exergonic by residue substitutions.

Next, an analysis is provided for the interatomic distances (Å), between Fe_p and Fe_d, Fe_p and CO_b, Fe_d and CO_b, and Fe_d and H₂O, before and after water removal, to compare the H₂O elimination Gibbs’ energies for the residue substituted DdH with the thermodynamics of the wild-type DdH.

For removal of water from the wild-type enzyme, the distance between Fe_d and CO_b is slightly increasing (Table 3), which corresponds to the endergonic step, viz., ΔG_{Enzyme Φ} = +2.0 kcal/mol. However, the H₂O removal from the mutated enzyme is exergonic (Table 2), correlating to the movement of the CO_b towards the Fe_d. As a result of mutating DdH, the following bond contractions, Fe_d-CO_b, have been obtained, viz., 0.024 Å for Arg¹¹¹Glu, 0.028 Å for Thr¹⁴⁵Val, 0.017 Å for Glu³⁷⁴His, and 0.031 Å for Tyr³⁷⁵Phe, corresponding to exergonic steps for water elimination (Table 2).

Note that a simultaneous bond elongation occurs [for all presented mutations (Table 3)] between the Fe_p and the bridging carbonyl, Fe_p-CO_b, when bond contraction for Fe_d-CO_b takes place. As a result of mutating DdH (vs. the wild-type enzyme), larger bond contractions between the iron atoms, Fe_d-Fe_p, have been obtained (ca. 0.1 Å, Table 3). It is also noticed that the bond length between the distal iron and water, Fe_d-H₂O, is longer, i.e., about 2.2 Å (vs. 2.1 Å) in the mutated DdH vs. the wild-type enzyme.

The following trend has been observed for some of the DdH mutants that the closer the substituted residue is to the H-cluster exogenous water, the more spontaneous the water removal step becomes. The exception is Tyr³⁷⁵Phe, in which the substituted amino acid is highly hydrophobic, although it gets closest to the exogenous oxygen atom of H₂O, it nevertheless has less effect on removing water compared to Glu³⁷⁴His mutant where the substituted amino acid (at a greater distance from H₂O) is not only polar but of opposite charge as well.

The negative partial charge on oxygen (-0.935) of H₂O repels the negatively charged carboxylate of Glu³⁷⁴ (q_COO^- = 0.547, -0.567, -0.563), thus making the water removal difficult. When Glu³⁷⁴ is replaced by a protonated histidine, the opposite effect is observed.

Frontier Molecular Orbital Analysis

Molecular orbital analysis is provided using frontier orbitals (HOMO, LUMO, and SOMO) in correlation with the formerly presented Gibbs’ energies. Essentially, reduction for all three pathways are carried out on closed-shell clusters, that is, an e^- is transferred into the lowest unoccupied molecular orbital, H-cluster LUMO.

In the case of open-shell H-cluster protonation (both gas and aqueous enzyme Φ), a σ-bond is formed between a H⁺ and (the exogenous ligand) OH^- by the interaction of electrons in the highest occupied molecular orbitals, viz., SOMO, and HOMO. Conversely, when a H⁺ is in close proximity to a closed-shell H-cluster, the resulting σ-bond is mainly due to the contribution of e^-s from HOMO and the H⁺.

The [Fe-Fe]-hydrogenase H-cluster 1 of the wild-type enzyme becomes reduced (Scheme I, 1 → 2), and according to the LUMO depiction (Figure 2), the transferred e^- appears to be localized in the vicinity of the di-iron atoms, Fe_p-Fe_d. However, according to the NBO partial charge results (Figure 3), the iron atoms (of hydrogenase H-cluster 1) do not have affinity for the approaching e^-, i.e., q_Fep = -1.291 a.u., and q_Fed = -1.109 a.u. The fact that the reduction is endergonic (ΔG_{Enzyme Φ} = +42.6 kcal/mol) corroborates the orbital analysis results, which indicate that e^- transfer to the cluster should be thermodynamically unfavorable due to existing negative charge on the di-iron atoms.

Figure 2.

The Frontier Molecular Orbitals from DFT Calculations (B3LYP/6-31+G(d,p)).

Figure 3.

The NBO Charges Obtained from DFT Calculations (B3LYP/6-31+G(d,p)). Charges are Given in a.u. for the Following H-Clusters: 1, 2, 2’, 3, 3’, and 4 Starting from the Top of the Columns.

Regarding [Fe-Fe]-hydrogenase H-cluster 2, (2 → 3), the open shells, SOMO_α, HOMO_α, and HOMO_β, have similar orbital distribution over OH^- (Figure 2), and could make a σ-bond with the incoming H⁺; in conjunction with the NBO partial charge of the hydroxyl oxygen, q_O = -0.991 a.u., the large H⁺ affinity for cluster 2, ΔG = -317.9 kcal/mol, (2 → 3) comes as no surprise.

In H-cluster 3, (3 → 4), the open shells, HOMO_α, HOMO_β, and SOMO_α, are diffused throughout the cluster except over the Fe_d-OH₂ bond, implying that the e^-s of the σ-bond, Fe_d-OH₂, reside in a lower energy state, which also explains the negative NBO partial charges of both Fe_d (q_Fed = -1.012 a.u.), and the O (q_O = -0.935 a.u.) of H₂O. This fact may explain, therefore, the affinity of Fe_d for H₂O, i.e., ΔG = +2.0 kcal/mol.

For [Fe-Fe]-hydrogenase H-cluster 1, (Scheme II, 1 → 2’), the HOMO is more diffused over Fe_d, DTMA bidentate ligand, and over the exogenous ligand, i.e., OH^-. In spite of greater e^- orbital diffusion over the DTMA ligand, the H⁺ becomes captured by OH^-, for it is a stronger base than the N of the DTMA bridge, as the NBO charges indicate (q_O. = -0.898; q_N = -0.682). Note that these analyses also agree with the calculated hydrogenase H-cluster proton affinity, i.e., ΔG = -236.8 kcal/mol.

For H-cluster 2’, (Scheme II, 2’ → 3’), HOMO is diffused over the center of Fe_p-Fe_d subcluster, CO_b, and over CN^- (Fe_d coordinated) which means that the e^-s of the σ-bond, Fe_d-OH₂, are situated in a lower energy state, explaining the negative NBO partial charges of both O (of H₂O, q_O = -0.878 a.u.), and Fe_d (q_Fed = -1.077 a.u.). This, then, accounts for the bond strength of Fe_d for H₂O, i.e., ΔG = +22.9 kcal/mol. In addition, from geometrical considerations, 2’ → 3’, it can be seen that CO_b-Fe_d bond becomes longer, viz., 1.932 → 1.942 Å [vs. 1.907 → 1.908 Å (in H-cluster 3, Scheme I, 3 → 4)], which partially accounts for the more endergonic H₂O elimination process, coinciding with an increase of one order of magnitude for the Gibbs’ energy difference (for water elimination in 2’ → 3’ vs. 3 → 4).

In H-cluster 3’, LUMO is mostly localized on the potential catalytic binding site. Upon the reduction of the H-cluster, (3’→ 4), the LUMO depiction shows that the e^- should become localized peripherally to the Fe_d. Then according to the NBO partial charges, i.e., q_Fep = -1.317 a.u., and q_Fed = -0.733 a.u., in conjunction with the above presented LUMO depiction, the cluster reduction occurs with a rather high spontaneity, i.e., ΔG_{Enzyme Φ} = -59.4 kcal/mol, although the di-iron atoms have rather high NBO charges^ι.

The LUMO on the wild-type hydrogenase H-cluster 2’, is delocalized on the two irons, Fe_p-Fe_d, DTMA sulfur atoms, and CO_b (Figure 2). Cluster reduction (Scheme III, 2’ → 3) occurs with a rather high spontaneity, i.e., ΔG_{Enzyme Φ} = -38.4 kcal/mol, although the di-iron atoms have high NBO negative charges, i.e., q_Fep = -1.305 a.u., and q_Fed = -1.077 a.u. [just as for H-cluster 3’, (3’ → 4)]. It ought to be noted that the e^- transfer occurs exergonically relative to the endergonic step (+42.6 kcal/mol) in Scheme I, 1 → 2, perhaps because the total charge on each cluster is different, i.e., cluster 2’ has charge 0 a.u., while 1 has a charge of -1 a.u.

Discussion

The comparison between gas phase and wild-type enzyme calculations, presented in the current study, unambiguously shows that CO_b migration is key to enzyme inhibition by O₂. When CO_b is close to the Fe_d, DdH becomes reactivated, while, on the other hand, when CO_b is further away from Fe_d, wild-type enzyme H-cluster inactivation is observed. The displacement of CO_b is controlled by the apoenzyme, and can be further modulated by amino acid substitutions. In the wild-type enzyme, the protein environment impinges CO_b away from the catalytic site, Fe_d, leading to an exogenously inhibited hydrogenase, and thus hindering H₂O elimination. On the contrary, suitable residue substitutions can reverse the enzyme inhibition by allowing CO_b to migrate towards Fe_d, concomitant with H₂O removal.

The potential reactivation pathway of Scheme I is rendered less promising by the protein environment due to step 1 → 2, which is rather endergonic.

Also, the reactivation pathway of [Fe-Fe]-hydrogenase H-cluster 4 (Scheme II) cannot be realized because of the high nonspontaneous step for H₂O removal (2’ → 3’, ΔG_{Enzyme Φ} = +22.9 kcal/mol). As a result, this scheme does not seem to provide room for mutagenic enhancements (residue substitutions) that could improve H₂O removal.

However, in electrochemical settings, a hydrogenase can be adsorbed onto an electrode surface where the endergonic reductive step may be reversed by intramolecular changes (viz., apoenzyme amino acid substitutions), or extramolecular modifications (viz., voltage adjustment, or solution tuning such as pH modification, salt concentration changes, etc.).

The last reaction pathway of Scheme III provides the best chance for reactivating DdH, because the only reaction step to overcome (H₂O removal) is barely endergonic (ΔG_{Enzyme Φ} = +2.0 kcal/mol), which can be accomplished by suitable amino acid substitutions.

DdH screening by residue deletions pointed out which residues may increase the removal of H₂O from the catalytic site. From residue deletion clues, single and multiple residue substitutions have led to exergonic H₂O removal (e.g., Arg¹¹¹Glu, Thr¹⁴⁵Val, Glu³⁷⁴His, and Tyr³⁷⁵Phe providing ΔG = -7.5 kcal/mol).

Conclusions

The DdH reactivation, according to pathway I, consists of an endergonic e^- transfer step (viz., ΔG_{Enzyme Φ} = +42.6 kcal/mol), followed by an exergonic H⁺ transfer step (viz., ΔG_{Enzyme Φ} = -317.9 kcal/mol), and then an endergonic H₂O removal step (viz., ΔG_{Enzyme Φ} = +2.0 kcal/mol).

For reactivation pathway II, the [Fe-Fe]-hydrogenase H-cluster H⁺ transfer occurs first (ΔG_{Enzyme Φ} = -236.8 kcal/mol), followed by H₂O removal (ΔG_{Enzyme Φ} = +22.9 kcal/mol), and then by e^- transfer (ΔG_{Enzyme Φ} = -59.4 kcal/mol), with all steps being exergonic, except for the removal of water.

Pathway III, however, proceeds by an exergonic protonation step (viz., ΔG_{Enzyme Φ} = -236.8 kcal/mol), an exergonic e^- transfer step (viz., ΔG_{Enzyme Φ} = -38.4 kcal/mol), followed by water removal step (viz., ΔG_{Enzyme Φ} = +2.0 kcal/mol), which is barely endergonic. Thus, the reason DdH intermediates of pathway III, rather than those of pathway I and II, were chosen to be residue mutated was to achieve a pathway that proceeds exergonically throughout. Results have been obtained for deleted, substituted, and combined (substitutions of) residues. Combinations of two, three, and four residues gave improved negative Gibbs’ energies for the removal of water, relative to single substituted residues.

In pathway III, the endergonic step, for H₂O removal, was made to proceed more spontaneously by removing inhibitory residues, one at a time. The promising residues are ΔArg¹¹¹, ΔThr¹⁴⁵, ΔSer¹⁷⁷, ΔGlu²⁴⁰, ΔGlu³⁷⁴, and ΔTyr³⁷⁵, and their respective Gibbs’ energies are -0.7 kcal/mol, -0.2 kcal/mol, -0.5 kcal/mol, +0.8 kcal/mol, -1.6 kcal/mol, and +1.1 kcal/mol.

Individual residues were substituted, viz., Arg¹¹¹Glu, Thr¹⁴⁵Val, Ser¹⁷⁷Ala, Glu²⁴⁰His, Glu³⁷⁴His, and Tyr³⁷⁵Phe. All substitutions resulted in improved spontaneity for H₂O removal, relative to residue deletions, except for ΔSer¹⁷⁷ → Ser¹⁷⁷Ala (ΔG = -0.5 kcal/mol → ΔG = -0.1 kcal/mol), and ΔGlu^{240 →} Glu²⁴⁰His (ΔG = +0.8 kcal/mol → ΔG = +1.1 kcal/mol).

From the successful, single residue substitutions, two, three, and four residue combinations were used to prepare DdH mutants for frequency calculations. The two residue combination, Glu³⁷⁴His, and Tyr³⁷⁵Phe, resulted in ΔG = -5.1 kcal/mol; the three residue combination, Thr¹⁴⁵Val, Glu³⁷⁴His, and Tyr³⁷⁵Phe, provided a ΔG = -6.2 kcal/mol, and the four residue combination, Arg¹¹¹Glu, Thr¹⁴⁵Val, Glu³⁷⁴His, and Tyr³⁷⁵Phe, gave a ΔG = -7.5 kcal/mol. Though the H₂O removal thermodynamic trend is not precisely cumulative, it seems to point in that direction.

The wild-type DdH H-cluster bond distance, i.e., Fe_d-CO_b, relative to that of the H-cluster (gas Φ), for H₂O elimination remains almost constant (viz., 1.907 Å → 1.908 Å), whereas the gas Φ² bond distance becomes smaller (viz., 1.945 Å → 1.850 Å), occurring with a concomitant exergonic H₂O removal; this, then, may partly explain why the removal of H₂O is exergonic for gas Φ, as opposed to the enzyme Φ.

Since pathway III has only the H₂O removal step to overcome to proceed exergonically, it was found to provide, to some extent, the sought after reactivation via residue substitutions. The step for H₂O removal from the wild-type hydrogenase (ΔG_{Enzyme Φ} = +2.0 kcal/mol) shows that the distance between the Fe_d and CO_b remains approximately the same. However, the H₂O removal from the mutated enzyme proceeds exergonically, correlating to the movement of the CO_b towards the Fe_d. The following bond contractions, Fe_d-CO_b, have been obtained, viz., 0.024 Å for Arg¹¹¹Glu, 0.028 Å for Thr¹⁴⁵Val, 0.017 Å for Glu³⁷⁴His, and 0.031 Å for Tyr³⁷⁵Phe, corresponding to exergonic steps for water elimination. Additionally, bond contractions have been obtained between the iron atoms (ca. 0.1 Å), concurrent with H₂O dissociation.

We conclude by postulating that even a single and proper residue (experimental) substitution, like ΔGlu³⁷⁴ → Glu³⁷⁴His (ΔG = -1.6 kcal/mol → ΔG = -3.1 kcal/mol), can reactivate the [Fe-Fe]-hydrogenase.

We presented here a complete mechanistic picture of the reactivation reaction of the H-cluster and the role of the bridging carbonyl in modulating this reactivation. We compared the gas phase results with enzyme calculations and show how careful residue mutations can have a significant effect on the electronic structure of H-cluster. We have identified for the first time amino acid residues surrounding the active site that can be used to modulate the reactivity of [Fe-Fe]-hydrogenase. Our calculations and the different analyses (i.e., thermodynamic, geometrical, and electronic) emphasize that all these structural data pertaining to different origins (electronic, geometric, thermodynamics) concur in the suggested picture for the reaction mechanism of the reactivation of wild-type [Fe-Fe]-hydrogenase.

Table 4.

Interatomic distances between the oxygen (of exogenous H₂O; Fe_d-OH₂, compound 3) and the juxtaposed atoms of the residue R-groups.

Mutated DdH	H2O oxygen	Å
Arg111Glu mutant
O(ε)	O(H2O)	17.507
O(ε)	O(H2O)	18.259
Thr145Val mutant
C(γ)	O(H2O)	9.616
C(γ)	O(H2O)	9.650
Ser177Ala mutant
C(β)	O(H2O)	6.077
Glu240His mutant
N(δ)	O(H2O)	12.395
N(ε)	O(H2O)	11.666
Glu374His mutant
N(δ)	O(H2O)	9.631
N(ε)	O(H2O)	9.038
Tyr375Phe mutant
C(ζ)	O(H2O)	7.724