Molecular dynamics and experimental investigation of H₂ and O₂ diffusion in [Fe]-hydrogenase

Abstract

The [Fe]-hydrogenase enzymes are highly efficient H₂ catalysts found in ecologically, and phylogenetically diverse microorganisms, including the photosynthetic green alga, Chlamydomonas reinhardtii. Although these enzymes can occur in several forms, H₂ catalysis takes place at a unique [FeS] prosthetic group, or H-cluster, located at the active site. Significant to the function of hydrogenases is how the surrounding protein structure facilitates substrate-product transfer, and protects the active site H-cluster from inactivation. To elucidate the role of protein structure in O₂ inactivation of [Fe]-hydrogenases, experimental and theoretical investigations have been performed. Molecular dynamics was used to comparatively investigate O₂ and H₂ diffusion in [Fe]-hydrogenase CpI. The results are compared to initial investigations of H₂ diffusion in [NiFe]-hydrogenase [1]. Our preliminary results suggest that H₂ diffuses more easily and freely than O₂, which is restricted to a small number of allowed pathways to and from the active site. These O₂ pathways are located in the conserved active site domain, shown experimentally to have an essential role in active site protection.

Introduction

Hydrogen production in the green alga, C. reinhardtii, is a strictly anaerobic process. Coupling to photosynthesis can be achieved under conditions of sulfur deprivation that results in a series of physiological events leading to reduced PSII activity, the respiratory consumption of residual O₂, and establishment of an anaerobic state [2,3]. Induction of hydrogenase gene expression and enzyme synthesis ultimately results in H₂-production. Although metabolic sequestration of O₂ overcomes the severe sensitivity of algal [Fe]-hydrogenase to O₂, the H₂-production efficiency is lower than theoretical maximum [2,3]. In order to develop an efficient, large-scale, photobiological H₂-production system, the O₂ sensitivity of algal [Fe]-hydrogenases needs to be addressed [2].

Inactivation of [Fe]-hydrogenase by O₂ involves diffusion of the diatomic gas from solvent to the active site, and subsequent chemical oxidation of the catalytic H- cluster [4]. The initial stages of diffusion involve molecular interactions between O₂ and protein, suggesting that the protein structure influences how O₂ approaches the active site. Previous studies of H₂ diffusion in [NiFe]-hydrogenase suggest that H₂ diffusion in the enzyme is not a random process, but rather tends to occur through a specific pathway, the H₂-channel [1]. To further investigate gas diffusion in hydrogenase, and to elucidate the initial events in the O₂-inactivation process, molecular mechanics methods were used to study O₂ and H₂ diffusion in [Fe]-hydrogenase CpI.

Methods and Results

Expression and O₂-sensitivity determination of [Fe]-hydrogenase

The expression of active the [Fe]-hydrogenases HydA1, HydA and HydAΔN were carried out in E. coli as previously reported [5]. Whole cell samples were solubilized, exposed to air, and aliquots assayed for residual H₂ production activity using the reduced methyl viologen assay [5].

Early biochemical studies of O₂-inactivation of [Fe]-hydrogenase suggested that differences exist in the O₂ sensitivities of enzymes isolated from different sources. The partially purified [Fe]-hydrogenase from C. reinhardtii exhibited an I₅₀ value of ~1 second when exposed to atmospheric levels of O₂ [6], whereas under similar conditions purified CpI exhibited an I₅₀ of 120–300 seconds [7] (Table 1). Our recombinantly expressed C. reinhardtii HydA1 [Fe]-hydrogenase was ~415-fold more sensitive to atmospheric O₂ (21%) than the C. acetobutylicum HydA (Table 1). To date the algal enzymes represent the simplest [Fe]-hydrogenases known, consisting solely of an active site domain. In contrast, the homologous bacterial enzymes, CpI and HydA, are more complex, possessing an additional electron-transfer domain [8]. This domain may also contribute to protection of the active site, and may explain the higher observed O₂- tolerance levels of these enzymes. Removal of the electron-transfer domain from HydA (HydAΔN), however, resulted in only a 3-fold decrease in O₂ tolerance (Table 1), and remained ~140-fold more tolerant to O₂ than HydA1. The differences in enzyme sensitivities conferred by the conserved active site domain strongly suggests that the amino acid composition of this domain is critical to protection of the H-cluster from O₂.

Table 1.

Comparison of algal and bacterial [Fe]-hydrogenase O₂ sensitivities.

[Fe]-hydrogenase	I50 value (seconds) in atmospheric O2
C. reinhardtii hydrogenase	<11
C. reinhardtii HydA1	<1
C. pasteurianum CpI	120 to 3002
C. acetobutylicum HydA	415 ± 115
C. acetobutylicum HydAΔN	145 ± 45

¹From [6].

²From [7].

Simulation setup and methods

Our CpI model is based on the X-ray crystal structure of [Fe]-hydrogenase CpI from Clostridium pasteurianum [8]. Missing H-cluster atoms from the CpI structure are modeled as a di(thiomethyl)amine as in [9]. The partial charges for the rest of H-cluster atoms were based on Ref. [10], with modifications of up to ±0.02e to preserve charge neutrality. The model was then embedded in a water box, resulting in a 57,000-atom system consisting of 9,000 hydrogenase atoms, 16,000 water molecules and 15 sodium ions. The full system was then equilibrated at constant temperature (310K) and pressure (1 atm) for 1ns.

Oxygen and hydrogen gas diffusion in CpI were investigated by all-atom molecular dynamics simulations of the outward diffusion of either O₂ or H₂ from the active site, in a similar manner to previous studies of gas diffusion across proteins [1, 11]. A charge-free model for O₂ and H₂ is used. In addition, for reasons of numerical stability, and to facilitate comparison between the two gases, H₂ with equivalent mass to O₂ (hH₂) was used in place of H₂. This way, O₂ and H₂ differ solely by their van der Waals (Lennard-Jones) parameters, molecular bond lengths, and spring constants.

In order to increase the sampling, the method of locally enhanced sampling (LES), otherwise known as the time-dependent Hartree approximation [12], was used to simulate 1,000 mutually invisible, simultaneous copies of the gas molecules (hH₂ or O₂) within a single protein-water system. Gas temperatures were regulated independently at 310K. Each of the 1,000 copies of either O₂ or hH₂ was initially placed at the same location as the H-cluster bound CO molecule in the CpI structure [4]. For hH₂, our simulations correspond to the realistic transport of H₂ out of the protein, but for O₂, we are observing a reverse of the normal O₂ diffusion from the bulk solvent to the active site.

LES simulations were performed with NAMD2 [13] using the CHARMM22 force-field [14]. A constant volume and temperature were used. The system was simulated with periodic boundary conditions, and Particle-Mesh Ewald method.

MD simulation of H₂ and O₂ diffusion in CpI

During the 1 ns equilibration, we observed a permanent and almost continuous tunnel- shaped cavity (using a 1 Å-radius probe) connecting the active site binding location to the solvent outside the protein at the location of the “H₂-channel” detailed for the Desulfovibrio desulfuricans [Fe]-hydrogenase reported by Nicolet et. al. [9,15].

Our preliminary simulations (6 or 4 independent 2 ns simulations, for either O₂ or H₂ diffusion respectively) suggest that the diffusion pathways of O₂ and hH₂ between the active site of CpI and the protein-solvent interface are different. More specifically, on one hand, O₂ is observed to transit across the protein barrier through a limited set of precisely defined channels, of which one is displayed in Fig 1a. We have observed O₂ motion across only two such channels, and in only one simulation did O₂ reach the solvent within the simulated time of 2 ns. On the other hand, in all four H₂ simulations, H₂ was observed exiting the protein. A sample of the allowed hH₂ pathways is shown in Fig 2a. In addition, hH₂ is occasionally observed (in 2–15% of all observed exits per simulation) to cut through the bulk protein by unique paths where no channel has been detected. A second important difference between H₂ and O₂ diffusion is that, during LES simulations with 1000 copies of the gas molecule sharing a unique protein trajectory, the different simultaneous hH₂ molecules are observed to spread out in the protein and take a multitude of exits (see Fig 2b–c). Oxygen molecules, in stark contrast, stay clumped together for most or all of the duration of the trajectories (see Fig 1b–c).

Figure 1.

Caption Sample trajectory of molecular diffusion of 1000 simultaneous and mutually invisible copies of di-oxygen (O₂) from the active site of CpI hydrogenase. Shown is (a) a superposition of all the O₂ positions over the 2.3 ns trajectory and snapshots of the O₂ trajectories after (b) 20 ps and (c) 2.3 ns. Figures were made using VMD [16].

Figure 2.

Caption Same as in Figure 1 for hH2. Shown is (a) a superposition of all the hH₂ positions of the 2.3 ns trajectory and snapshots of the hH₂ trajectories after (b) 300 ps and (c) 2.2 ns.

Conclusion

In summary we have shown that the diffusion of H₂ and O₂ gases inside of an [Fe]-hydrogenase is governed by the physical properties of both the gas itself and the protein structure. The dynamic process of gas migration dictates that static representations of structures can unsatisfactorily predict pathway selection. In this study, althoug a majority of H₂ was found to migrate through a series of conserved hydrophobic cavities, or a “H₂-channel” [15], H₂ is clearly able to diffuse during the same time period through a number of alternative routes. This is somewhat inconsistent with previous results shown for H₂ diffusion in [NiFe]-hydrogenase [1].

In contrast to H₂ diffusion which displays a degree of randomized behavior, O₂ diffusion is clearly limited to specific regions located within the conserved active site domain. The importance of this domain in O₂ diffusion agrees with the experimental results show here. The range of sensitivity levels found for algal and bacterial enzymes are also characterized by diversity among the amino acids that comprise the diffusion pathways. From the standpoint of engineering O₂ tolerance in [Fe]-hydrogenases, the sequences and structures of naturally occurring enzymes offer important clues on how to overcome the O₂-sensitivity of algal [Fe]-hydrogenases.

Abbreviations used

CpI

[Fe]-hydrogenase I from Clostridium pasteurianum

HydA

Clostridium acetobutylicum [Fe]-hydrogenase I

hH₂

heavy di-hydrogen

LES

locally enhanced sampling