1 Review

Trends in Biochemical Sciences Vol.00 No.X December 2008

Mechanics
of
gas
diffusion
within
 metalloenzymes
:
the
case
of
 hydrogenase
 Pierre
 Ceccaldi,
 Master
 Recherche
 Bioinformatique,
 Biochimie
 Structurale
 et
 Génomique,
 Université
 de
 la
 Méditerranée,
unité
de
Bioénergétique
et
d’Ingénierie
des
Protéines,
CNRS,
31
chemin
Joseph
Aiguier,
13009
 Marseille
 [Abstract] Hydrogenases, which catalyze the reversible H2 oxidation as a part of the bioenergetic metabolism in many microorganisms, are metalloenzymes for which hydrophobic gas channels have been identified. In this paper, I present the computational and experimental studies which were used for their characterization. Also recent experimental approaches for kinetics of gas diffusion in hydrogenase are reported. Hydrogenases are enzymes involved in H2 metabolism. They catalyze the reversible oxidation of hydrogen into electrons and protons. It is thought that they could be used in biotechnologies for either H2 production by photosynthetic organisms, or H2 oxidation in biofuel cells. However these enzymes are inactivated by oxygen, which appears to be the major obstacle for applications in industry. This has driven research on hydrogenases for many years. It has been shown that the buried active site is connected to the molecular surface by gas access pathways, determined as hydrophobic cavities. They may enable hydrogen to access the active site, also allowing inhibitors, such as O2 and CO, to bind it, inactivating the enzyme. In that case, understanding the mechanism of gas diffusion is crucial to engineer hydrogenases for biotechnological applications, but it is also of academic interest, since hydrophobic channels, or tunnels, exist in many other enzymes, including tryptophan synthase and Carbon Monoxide Dehydrogenase/Acetyl-CoA Synthase. Both experimental and computational methods have been used to characterize the channels and the mechanisms of gas diffusion in hydrogenase. Recent insights in kinetics of gas diffusion and inhibition in hydrogenase have confirmed that residues along the channel tune the rate of gas diffusion.

Structure/Function relationships : The most part of model hydrogenases are purifed from orgamisms of Desulfovibrio genus (Dg). They are classified in two families, NiFe and FeFe (also called Feonly), according to the metal composition of their acive site. The periplasmic NiFe hydrogenase is a heterodimer. The active site is buried in the large subunit (≈ 62 kDa), 30 Å away from the protein surface, whereas three iron-sulfur [FeS] clusters involved in the electrons transfer pathway are coordinated by the small subunit (30 kDa). The NiFe center has two CO and one CN-, both coordinated by the Fe atom. The enzyme is an element of the energyproduction process. Except that the FeFe active site binds one CO and one CN- on each Fe atom, the Fe-only enzyme presents the same organization of metallic centers : a buried binuclear active site and several [FeS] clusters connecting it to the surface. Its role in anaerobic metabolism is to oxidize excessively-reduced compounds, evolving H2. Eventhough Fe- and NiFe-hydrogenases seem to be involved in proton reduction and H2 oxidation, respectively, the catalytic cycle is similar in the two enzymes, since we consider reversible reactions. In the H2uptake reaction, molecular hydrogen enters the active site by diffusing through hydrophobic channels (see below). H2-cleavage is carried-out by the NiFe or FeFe binuclear center, then the [FeS] clusters tansfer electrons from the active site to c-type cytochromes, physiological partners of hydrogenases in Dg. Artificial redox partner can be either a chromophoric compound or the electrode of the electochemical cell, in the case of protein film voltammetry. A network of protonable residues and organized water molecules, mostly unidentified, constitutes the proton pathway between active site and molecular surface. Finally different types of enzymatic assays can be performed on hydrogenase. Besides well known colorimetric assays using an electron acceptor such as methylene blue, methods have been developped to quantitatively analyze gas diffusion mechanisms and determine rates. [1][2]

1

2

C B A

Figure 1. Overview of hydrophobic cavities in (left) Fe-only and (right) NiFe-hydrogenases (pdb codes 1HFE and 1YQW, respectively). Large and small subunits are shown in blue and red, respectively. Active sites and iron-sulfur clusters atoms are represented as spheres.

Characterization of Gas channels : The cristallographer’s point of view Hydrophobic cavities were firstly identified in the crystal structure of Dg NiFe-hydrogenase. Fontecilla and coworkers have then emphasized the capacity of these cavities to bind xenon in the crystal by exposing thre latteer to a pression of xenon. The cavity shape resembles an almost planar capital ‘N’, upside-down ‘\/\’ (see Fig. 2). The right side of the ‘\/\’ (further called A) is located in the small subunit, ending on the vacant position of the Nickel atom, which suggests that substrate and inhibitors enter the active by this way. The diagonal part (B) is mainly located at the interface between the large and small subunits, the left side (C) running through the large one. It must be noted that residues around the channels are highly conserved in all Dg hydrogenases. More surprising is the existence of very similar channels in hydrogenases from other genera, such as the [NiFeSe] enzyme from Desulfomicrobium baculatum which presents only 35% sequence identity with the Dg enzyme. These experimental data suggested that such cavities should constitute a gas access pathway to the active site for substrate and inhibitors. Fontecilla and co-workers have also determined the crystallographic structure of FeFe hydrogenase from Desulfovibrio desulfuricans ATCC 7757 (DdHase), showing that a single hydrophobic channel runs from the surface to the buried active site and also points to the vacant coordination site of one of the two Fe atoms. Such

a channel is also present in hydrogenase I from Clostridium pasteurianum (CpIHase). It has to be noted that the cristallographic structures, even at high resolution, do not present water molecule inside the channels. “Empty’’ cavities, also called packing defects, are generally considered as destabilising the structure. However, in the Dg NiFe hydrogenase, two thirds of the residues lining the channels are part of secondary structure elements. This might favor the formation of cavities during the protein folding, since it is presumably harder for a highly ordered structure to collapse, which would remove internal cavities [1][2][3]. Crystallographic studies have proven the existence of hydrophobic cavities in hydrogenase. Furthermore, crystallography of xenon-saturated structures provides certainty for such cavities to be gas access patways. But because gas access to the active site through these channels was not obvious, computational approaches have been initiated to observe possible gas trajectories from the channels towards the active site. Contribution of Molecular Dynamics (MD) Although crystallography proved the existence of hydrophobic channels, it only offers a static view. Considering this, computational approaches are useful to study protein internal fluctuations and further observe predicted behaviour of gases diffusing inside the protein. In the case of hydrogenases, MD simulations have been run to evaluate preferrential trajectories for H2 and O2 toward the active site.

2

3

A

Regarding to the NiFe family, Soares and co-workers have performed several MD simulations in explicit solvent. Various replicates of H2 diffusion inside the protein were performed to improve the sampling of the system states. The MD trajectories showed that the A channel seems to be essential, since much more H2 molecules not only enter the protein through it but also access more often to the active site. Channel B is used to a lesser extent. Channel C is the least occupied, which is consistent since we observe that the channels A and B are directly connected with the active site, whereas C is not. To investigate the role of residues at the end of the channel, the valine residue ending the channel was mutated into alanine, then the same simulations were performed. Figure 2 shows statistics of H2 molecules, relatively to their distance to the active site. The mutant shows more H2 molecules in the region about 5 Å to the active site, which corresponds to the gap created by the mutation. We also observe less H2 molecules between 2 and 4 Å in the mutant that in the wt. This may be due to the presence of a trapped molecule in this region during the simulations conclude of the wild-type enzyme. These

Figure 3. A and B. VSAM (mesh) and trajectories of H2 (A, blue) and O2 (B, red) in the Fe-only enzyme CpI. Metal centers are represented as sticks.

Figure 2. MD trajectories inside a typical NiFe hydrogenase from Dg. Number of H2 molecules per picosecond according to their distance to the active site in the wild-type (A) and the mutant V67A (pdb code 2FRV).

results took the authors to that this valine residue may be a control point in the catalytic mechanism of NiFehydrogenase. Another important result of the Soares study is that all H2 molecules that reach the active site enter through the channels [4]. This consideration confirms the existence of specific gas pathways for H2 in hydrogenase, generally opposed to random diffusion in proteins of little molecules such as H2. Considering the Fe-only hydrogenase, the static structure presents only one identified hydrophobic cavity (Fig. 1). According to this, Shulten and co-workers have performed MD simulations to investigate possible gas pathways in the hydrogenase from Clostridium pasteurianum (CpI). Two types of experiments were done. They first carriedout temperature-controlled locally enhanced sampling (TLES) simulations to record H2 and O2 trajectories, then

they calculated internal cavities by the maximum volumetric solvent accessibility map (VSAM) method, using a larger radius for O2 than for H2. Figure 3 shows superimposed VSAM maps and TLES trajectories for H2

3

4

A

B

C

Figure 4. Closeups of gas channel near the NiFe active site in Desulfovibrio fructosovorans hydrogenase, wild type, FI and MM mutants. Large and small subunits are in magenta and blue, respectively

and O2 starting from the active site (A, B), and a slice through the superimposed VSAM of O2 and H2 (C). The obtained maps almost matched the TLES trajectories, so that no gas molecule was found, out of the VSAMs. The major information in Schulten’s paper is that gas access pathways appear to be entirely determined by the protein’s dynamical fluctuations, which proves that gases do not create new cavities by interacting with the protein, they rather merely insert into packing defects that arises spontaneously in the presence or absence of gas. Furthermore, these simulations show the existence of a second pathway from the active site to the surface. Therefore, intrinsic protein dynamics appear to be crucial in the mechanism of gas diffusion inside hydrogenase [5]. Do the residues ending the channel control the mechanism of gas diffusion ? Some NiFe-hydrogenases are insensitive to oxygen. These enzymes are named H2-sensors, or regulatory hydrogenases (RH), because they act as H2 sensors in a H2dependent signalling cascade triggering gene transcription of hydrogenases involved in energy-production processes. They are present in photosynthetic microoganisms that thrive in oxic environment. However the turnover of these enzymes is very low, which makes them useless for applications in engineering. Nevertheless, they can be useful for understanding mechanisms of O2-sensitivity. Structural data for RH are still not available. However, Fontecilla and co-workers have identified putative gas channel by sequences alignments of RH with standard NiFe-hydrogenases. The principal gas channel identified in the latter appears to be also present in RH (note). In the RH, the end of the putative channel consists of two bulky residues Phe and Ile, relatively to those found in standard NiFe-hydrogenases, Val and Leu (Fig. 2A). By considering this, it has been suggested that the insensitivity to oxygen of these enzymes would be attributable to apparent narrowness of the gas channel near the active site. To test this hypothesis, several research groups produced recombinant regulatory and standard NiFehydrogenases harboring typical residues of the standard or the regulatory enzyme, respectively. These recombined enzymes were then subjected to several types of enzymatic assays. As expected, specific H2-uptake activity of RH mutants was drastically affected in aerobic conditions, which is consistent with the role of the bulk residues to

block access to the active site by restricting the channel at the end of it (febs, 2005). Furthermore, studies based on mutated RH emphasized that their O2 insensitivity is based on limited access to the active site [6]. Concerning the mutated standard NiFe-hydrogenase, two mutated enzymes were produced, harboring either the ‘RH-like’ FI mutation, or the MM mutation which was made to significantly change the shape of the tunnel at this position. Figure 2 presents the entrance to the active site of the standard NiFe hydrogenase, either the wild-type or the two mutants FI and MM, showing that the mutations stop the channel, which was calculated with a 1.0 Å probe radius. Catalysis under aerobic conditions has not been reported for such recombinant enzymes, but two recent experimental approaches can provide kinetics data about the gas diffusion in hydrogenase. Obtaining kinetics of gas diffusion should be more informative about the mechanism of gas diffusion than only specific activity values and, furthermore, be useful for further applications. Also it has to be noted that, under anaerobic conditions, the turnover is very similar for the WT and the two mutated enzymes. The first method, that has been developped by A. Colbeau research team, is based on mass spectrometry measurements of the rates of D2, H2 and HD exchange in the absence of redox partner. Figure 5 shows typical results for D2 and HD concentration changes (plain and

Figure 5. Comparison of the kinetics of isotope exchange. The concentrations of [D2] and [HD] are plotted as plain and dashed lines, respectively. The experiments were started by adding the enzyme at t = 0. pH 7, T = 30°C, enzyme concentration : 2.7 µg/ml (WT), 5.6 µg/ml (FI), and 7.1 µg/ml (MM).

4

5 The slowing down of CO binding is obvious when the temperature is lowered. Data like those in Fig. 6 can be analyzed by using kinetic models so that inhibitor diffusion rates kin and kout can be obtained. In the WT enzyme, and at room temperature, binding of CO is fast, just below the diffusion limit. However, the rate constants for both forward and backward CO transport decrease by two orders of magnitude in the MM mutant. The FI mutant presents an intermediate phenotype. These experimental data demonstrate the existence of a specific path for gas access to the aactive site of NiFe-hydrogenase which coincides with the end of the tunnel found in the X-ray structure [7]. By using these two methods, Léger and co-workers reported, to the best of my knowledge, the first kinetics data on the gas diffusion in proteins. These emphasized the role of the channel size in the mechanism of gas diffusion in hydrogenase. It is not irrelevant thinking that it would be the same in other gas channels in proteins. Conclusion

Figure 6. Normalized current intensity of PFV experiments on wt and the FI and MM mutated NiFe-hydrogenase from Dg. Mixing time ≈ 0.1 s. Left shows the short-term change in current, whereas the end of the relaxation is shown in the right.

dashed lines, respectively) when the assay is initiated by adding the enzyme to a solution of D2. HD is only transiently present in the solution, and eventually D2 is quantitatively replaced with H2. This is in agreement with the reaction mechanism where HD is an intermediate : (i) D2 reversibly binds to the active site where it is heterotically cleaved ; (ii) H+ from the solvent substitutes for D+ ; and (iii) the D+ of the resulting HD species is eventually replaced with H+, generating H2. In case of a slowed transport of H2, D2 and HD in the channel, the probability that HD exits the enzyme before it further reacts to give H2 should decrease. We see on Fig. 5 that the two mutants, at higher concentration than the WT, show a lower HD production rate. This is consistent with a lowered diffusion rate along the channels. Secondly, C. Léger research team has developped a protein film voltammetry method (PFV) to probe kinetics of oxidative inhibition by gases. PFV is a technique where the enzyme is adsorbed onto an electrode that acts as a sink or a source of electrons, so that H2 oxidation activity is measured as a current. The high temporal resolution of this method allows for measuring the inactivation rate of the enzyme. The experiment consisted in injecting in the electrochemical cell aliquots of a solution saturated with CO, while the buffer is continuously degassed by bubbling H2. The transient change in activity is continuously monitored as the concentration of inhibitor decreases. Fig. 3 shows typical results from electrochemical experiments. In the case of the wild-type enzyme, at T = 40°C, the decrease activity after CO injection is faster than the mixing time. The recovery of activity exactly follows the decrease in CO concentration as the latter is flushed away.

Recent insights in the study of gas diffusion in hydrogenase revealed the importance of steric bulk at the end of the channel to slow down the inhibition rate of the enzyme. The experimental approaches I presented here, coupled to molecular biology for mutants production, is useful to study impact of slight modifications in the structure. Also computational studies enable to predict the impact of protein dynamics in the gas diffusion process, so that both computational and experimental approaches complement each other in understanding the mechanism of gas diffusion. All the studies carried-out on hydrogenases to make them discriminate O2 and H2 will be useful for the comprehension of the global mechanism of gas diffusion in enzymes. Applied research could then be of interest to the academics. Acknowledgements I thank Christophe Léger for helpful notes.

References [1]Fontecilla-Camps, J.C., High-resolution crystallographic analysis of Desulfovibrio fructosovorans [NiFe] hydrogenase, Int. J. of Hydrogen Energy 2002 ; 27 :1449-1461 [2] Fontecilla-Camps, J.C. F e-only hydrogenases: structure, function and evolution, Biochemistry, 2002 ; 91 :1-8 [3]Fontecilla-Camps, J.C., Gas access to the active site of Ni-Fe hydrogenases probed by X-ray crystallography and molecular dynamics, Nat Struct Biol, 1997 ; 4 :523-6 [4] Soares, C.M., Pathways of H2 toward the Active Site of [NiFe]-Hydrogenase, Biophys. J. , 2006 ; 91 :2035-2045 [5] Schulten, K., Finding Gas Diffusion Pathways in Proteins: Application to O2 and H2 Transport in CpI [FeFe]-Hydrogenase and the Role of Packing Defects Structure, 2005; 13: 1321–1329 [6] Colbeau, A., Enlarging the gas access channel to the active site renders the regulatory hydrogenase HupUV of Rhodobacter capsulatus O2 sensitive without affecting its transductory activity, FEBS J., 2005; 272: 3899-3908 [7]Léger, C., Experimental approaches to kinetics of gas diffusion in hydrogenase

5