article published online: XX XX 2009 | doi: 10.1038/nchembio.xxx
Relating diffusion along the substrate tunnel and oxygen sensitivity in hydrogenase
Q1
Pierre-Pol Liebgott1, Fanny Leroux1,2, Bénédicte Burlat1,2, Sébastien Dementin1, Carole Baffert1,2, Thomas Lautier3–5, Vincent Fourmond1, Pierre Ceccaldi1,2, Christine Cavazza6, Isabelle Meynial-Salles3–5, Philippe Soucaille3–5, Juan Fontecilla-Camps6, Bruno Guigliarelli1,2, Patrick Bertrand1,2, Marc Rousset1 & Christophe Léger1*
In hydrogenases and many other redox enzymes, the buried active site is connected to the solvent by a molecular channel whose structure may determine the enzyme’s selectivity with respect to substrate and inhibitors. The role of these channels has been addressed using crystallography and molecular dynamics, but kinetic data are scarce. Using protein film voltammetry, we determined and then compared the rates of inhibition by CO and O2 in ten NiFe hydrogenase mutants and two FeFe hydrogenases. We found that the rate of inhibition by CO is a good proxy of the rate of diffusion of O2 toward the active site. Modifying amino acids whose side chains point inside the tunnel can slow this rate by orders of magnitude. We quantitatively define the relations between diffusion, the Michaelis constant for H2 and rates of inhibition, and we demonstrate that certain enzymes are slowly inactivated by O2 because access to the active site is slow.
T
unnels dedicated to guiding gaseous substrates likely occur in many redox enzymes that use or produce molecular oxygen, H2, N2 or CO. In certain bi-functional enzymes where a catalytic intermediate is transferred between distant active sites, the existence of a functional tunnel could be inferred from the observation that the intermediate is never released to the solvent1. When the putative tunnel connects a buried active site to the outside, it may be identified using X-ray crystallography or molecular dynamics calculations2,3. However, measuring rates of diffusion along these tunnels is often impractical because the movement of the ligand does not alter the spectroscopic signatures of the sample, unless the active site is a heme and the kinetics of CO or O2 binding and release can be monitored using transient absorbance measurements. Only in the case of small heme-proteins such as hemoglobins and myoglobins could internal ligand pathways be thoroughly studied using an integrated approach combining time-resolved spectroscopy, time-resolved crystallography of photoproducts, site-directed mutagenesis and molecular dynamics4–6. In contrast, regarding large metalloenzymes where a tunnel connects a deeply buried active site to the solvent, experimental results on the kinetics of intramolecular diffusion are scarce7,8. Hydrogenases, the enzymes that catalyze the biological conversion between dihydrogen and protons, are classified as ‘NiFe’ or ‘FeFe’ according to the metal content of their dinuclear active site. In both cases, the latter is connected to the solvent by a hydrophobic tunnel2. Figure 1a shows the overall structure of Desulfovibrio fructosovorans NiFe hydrogenase and the hydrophobic cavities that connect the solvent and the deeply buried NiFe active site. The most conserved tunnel is about 30 Å long and goes through the small subunit. A bottleneck at the end of
this tunnel, near the Ni ion, is shaped by the backbone of residue Glu25 and the lateral chains of Val74 and Leu122 (Fig. 1b). Hydrogenases are drawing much attention for their mechanism of action and potential uses, either as H2-oxidation catalysts in biofuel cells and other biocatalytic devices9–11, or for H2 production in photoelectrochemical biofuel cells or by photosynthetic microorganisms12,13. Yet the fact that most hydrogenases are inhibited by oxygen hinders further developments, and much effort is currently devoted to understanding the molecular bases of oxygen sensitivity. Understanding how the structures of the tunnels in hydrogenases determine the diffusion rates and possibly the selectivity of the enzymes with respect to substrates and inhibitors of similar size is therefore crucial. From the comparison between the sequences and biochemical properties of homologous NiFe hydrogenases, investigators have proposed that the size of the amino acids at positions 122 and/or 74 (Fig. 1b) may determine the accessibility of the active site and therefore the resistance to O2 (ref. 14). This hypothesis has been variably supported by the results of mutagenesis studies. Both Ralstonia eutropha and Rhodobacter capsulatus RH (regulatory hydrogenase) become oxygen sensitive when the two conserved bulky amino acids, Phe122 and Ile74, that guard the entrance of the active site are replaced with smaller residues, leucine and valine, respectively; this mimics the gas tunnel of “standard” oxygensensitive NiFe hydrogenases, such as those from D. fructosovorans and Allochromatium vinosum15,16 (we use D. fructosovorans amino acid numbering throughout). However, the L122F, V74I and L122F V74I mutants in R. eutropha MBH (membrane-bound hydrogenase) have slightly greater affinity for O2 than the wild-type (WT) enzyme17; no simple relation was observed between the Michaelis constants for H2 (Km) and the apparent inhibition constants for O2
Centre National de la Recherche Scientifique, Unité Propre de Recherche 9036, Unité de Bioénergétique et Ingénierie des Protéines, Institut Fédératif de Recherche 88, Institut de Microbiologie de la Méditerranée, Marseille, France. 2Aix Marseille Université, Marseille, France. 3Université de Toulouse; Institut National des Sciences Appliquées, Université Paul Sabatier, Institut National Polytechnique; Laboratoire d’Ingénierie des Systèmes Biologiques et des Procédés, Toulouse, France. 4Institut National de la Recherche Agronomique, Unité Mixte de Recherche 792, Ingénierie des Systèmes Biologiques et des Procédés, Toulouse, France. 5Centre National de la Recherche Scientifique, Unité Mixte de Recherche 5504, Toulouse, France. 6Laboratoire de Cristallographie et de Cristallogenèse des Protéines, Institut de Biologie Structurale J.P. Ebel, Commissariat à l’Energie Atomique, Centre National de la Recherche Scientifique, Université Joseph Fourier, Grenoble, France. *e-mail:
[email protected] 1
nature CHEMICAL BIOLOGY | Advance online publication | www.nature.com/naturechemicalbiology
1
article
Nature chemical biology doi: 10.1038/nchembio.xxx
a
or a single number that can quantify it: certain hydrogenases react reversibly with O2, although not in a competitive manner, while others are oxidized to a mixture of inactive states that reactivate distinctively under reductive conditions15–22. The fact that all these aspects of the reactivity contribute to making a hydrogenase more or less oxygen tolerant complicates the quantitative comparison between enzymes and/or mutants. Determining which of the parameters that characterize oxygen tolerance depend on the properties of the tunnel should help to clarify the diffusion and inhibition mechanisms. We do so hereafter, by characterizing and comparing a number of hydrogenases and hydrogenase mutants, and establishing clear and quantitative relations between amino acid sequence, Michaelis constants, and rates of diffusion and inhibition. We shall demonstrate that substituting amino acids whose side chains point inside the tunnel can greatly affect the rates of CO and O2 diffusion to and from the active site. Obstructing the tunnel increases the enzyme’s resistance to O2, on condition that diffusion becomes slow enough that it limits the rate of inhibition. The mutations have the same relative effects of the rates of diffusion of H2, CO and O2, but hydrogen transport is so fast that it fully limits the turnover rate of H2 oxidation in only one of the NiFe hydrogenase mutants we studied.
b NiFe
L122
V74
NiFe
E25
Figure 1 | Structure of D. fructosovorans NiFe hydrogenase depicting the “dry” hydrophobic cavities. The structure is from Protein Data Bank entry 1YQW. (a) The large and small subunits are shown as dark and light blue ribbons, respectively. Also shown are the active site, the chain of Fe-S clusters that wires the active site to the redox partners, and a grid delineating internal regions accessible to a probe of 1 Å radius. (b) Close up showing the access to the active site as the surface of the atoms that tile the end of the dry tunnel. Smaller, red spheres indicate the position of ordered water molecules in nearby “wet” cavities. Spheres in the background depict the Ni and Fe ions. Their ligands and residues Leu122, Val74 and Glu25 are shown as sticks. The side chains of Val74 and Leu122 Q23 define the surface of the tunnel that is shown in orange.
(KiO2) of these MBH mutants, and whether the change in oxygen sensitivity is a consequence of oxygen access to the active site being modified is unknown. Last, an observation of ours was that the L122F V74I double mutation in D. fructosovorans does not increase the oxygen tolerance of the enzyme18, despite the fact that it narrows the tunnel and slows access to the active site8. We also emphasized that the FeFe hydrogenase from Clostridium acetobutylicum reacts much more slowly with oxygen than the homologous enzyme from Desulfovibrio desulfuricans19, but whether this results from differQ3 ences at the active site or differences within the tunnel is unknown. There may be two reasons why no clear idea has emerged from these studies. First, the question of whether diffusion in the tunnel of hydrogenase relates to oxygen sensitivity could not be addressed until we proposed methods for determining rates of intramolecular diffusion8. Second, there is not a single definition of oxygen sensitivity,
RESULTS Initial screening of D. fructosovorans NiFe hydrogenase mutants
Table 1 | Kinetic properties of the hydrogenases Enzymes
kcatapp (s−1)a
Km (matm (H2))b
kcat (s−1)c
kinCO (s−1 mM (CO)−1)d
koutCO (s−1)e
kinO2 (s−1 mM (O2)−1)f
koutO2 (s−1)g
WT Df NiFe H2ase
750 ± 90
10 ± 5
760 ± 100
63,000 ± 42,000
500 ± 490
32 ± 3
n/ah
L122F V74I
800 ± 30
50 ± 20
840 ± 50
5,200 ± 3,100
1.30 ± 0.05
36 ± 4
n/ah
V74D
270 ± 90
50 ± 20
280 ± 100
2,100 ± 150
1.7 ± 0.5
24 ± 3
n/ah
L122A V74M
600 ± 30
100 ± 50
660 ± 60
1,490 ± 340
0.20 ± 0.05
45 ± 4
0.0015i
V74F
720 ± 15
120 ± 60
800 ± 60
1,150 ± 75
0.40 ± 0.05
52 ± 5
0.0007i
L122M V74M
590 ± 30
210 ± 40
715 ± 60
220 ± 75
0.015 ± 0.003
30 ± 4
0.0005i
V74N
490 ± 15
300 ± 70
640 ± 55
185 ± 31
0.8 ± 0.2
21 ± 3
n/ah
V74W
375 ± 60
280 ± 50
485 ± 95
130 ± 31
0.0033 ± 0.0003
28 ± 7
n/ah
V74E
225 ± 30
500 ± 200
340 ± 90
78 ± 21
0.8 ± 0.2
9.5 ± 1.2
n/ah
V74M
460 ± 15
300 ± 80
600 ± 60
19 ± 6
0.0040 ± 0.0005
6.5 ± 1.5
0.0004i
V74Q
210 ± 15
3,000 ± 1,000
860 ± 270
5.4 ± 1.1
0.10 ± 0.03
4±1
n/ah
WT Ca FeFe H2ase
32,000
800 ± 300
57,000 ± 9,000
16 ± 1
0.015 ± 0.002
2.5 ± 0.4
0.30 ± 0.05
WT Dd FeFe H2ase
56,000l
270 ± 50
71,000 ± 3,000l,k
4,400 ± 260
0.030 ± 0.005
40 ± 8
0.15 ± 0.05
j
j,k
Turnover rate under one atm of H2, at 30 °C, in 100 mM Tris/HCl buffer at pH 8, 50 mM MV (see Methods for details). bMichaelis constant for H2 determined in PFV experiments at pH 7. We could detect no significant effect of varying the temperature8. cMaximal turnover rate, extrapolated to infinite concentration of H2 using kcat = kcatapp × (1 + Km/[H2])8. dSecond-order rate constant for binding of CO, defined by equation (11), measured at or extrapolated to 40 °C (NiFe hydrogenases) or 30 °C (FeFe hydrogenases), and corrected for the protective effect of H2 (equation (12)). eFirst-order rate constant for CO release, defined by equation (11), measured at or extrapolated to 40 °C (NiFe hydrogenases) or 30 °C (FeFe hydrogenases). fSecond-order rate constant for binding of O2, defined by equation (13), measured at 40 °C (NiFe hydrogenases) or 30 °C (FeFe hydrogenases). gFirst-order rate constant for O2 release, defined by equation (13), measured at 40 °C (NiFe hydrogenases) or 30 °C (FeFe hydrogenases). hThis enzyme exhibits no reactivation phase when oxygen is flushed out. iRegarding NiFe hydrogenases, since koutO2 is very small, neither koutO2 nor kinact 2 in equation (13) could be accurately determined. jThis is the average of four different preparations. The value of kcatapp varies between preparations (±30%), but the values of the other five parameters do not. Experimental conditions: 37 °C, 1 atm of H2, 100 mM phosphate buffer at pH 7.2, 20 mM DTT, 20 mM MV (this is smaller than the Michaelis constant of 60 mM36). kThis error accounts only from the uncertainty on the value of Km. lValues from ref. 32. Experimental conditions: 30 °C, 1 atm of H2, 50 mM Tris/HCl buffer at pH 8, 1 mM MV. a
2
Q30
We constructed and purified a number of single and double mutants at positions 122 and/or 74: Val74 replaced with aspartate, phenylalanine, asparagine, tryptophan, glutamate, methionine, glutamine, isoleucine and histidine; Leu122 replaced with methionine, cysteine and phenylalanine; L122F V74I, L122A V74M and L122M V74M. The production yield was in the range of 0.5–1.0 mg Q4 of purified protein per liter of culture for the WT and all mutants, except for V74D and V74E, whose production was slightly lower (0.3 mg l−1). We screened the mutants using protein film voltammetry to estimate the rates of binding and release of the competitive inhibitor CO (see below and ref. 8), and we selected the mutants with the slowest inhibition kinetics: V74D, V74F, V74N, V74W,
nature chemical biology | Advance online publication | www.nature.com/naturechemicalbiology
article
Nature chemical biology doi: 10.1038/nchembio.xxx V74E, V74M, V74Q, L122F V74I, L122A V74M and L122M V74M. These mutants were assayed for H2 oxidation in the presence of 50 mM oxidized methyl viologen to measure the turnover rate app under 1 atm of H2 (kcat ), and we used the Michaelis constants for H2 determined by protein film voltammetry22 to deduce the values of the maximal turnover rate at infinite concentration of H2 (kcat), app using kcat = kcat (1 + Km/[H2]). Table 1 collects these data. Although certain mutations increase the value of Km by up to two orders of magnitude (for example, from 10 matm of H2 for the WT to about 3 atm for V74Q), the maximal turnover rates of nearly all mutants fall into the narrow range of 650–800 s−1; only the V74D and V74E variants have about half of the maximal activity of the WT enzyme.
EPR characterization of the mutants
We used electron paramagnetic resonance (EPR) spectroscopy to detect possible modifications induced by the L122A V74M, V74Q and V74E mutations (see Supplementary Methods for experimenQ5 tal details). This completes the previous EPR and FTIR characterizations of the V74M, L122M V74M and L122F V74I mutants presented in refs. 8,18. In as-prepared (air-oxidized) samples of the WT enzyme, the Ni appears as a superposition of two EPR signals (NiA and NiB), whose relative proportions vary between different preparations. The NiA signature (g = 2.31, 2.23, 2.01) predominates in the spectra of the three mutants (Fig. 2a). Upon partial reduction with dithionite, the Ni in the WT enzyme exhibits the so-called NiC signal at g = 2.19, 2.14, 2.01, which is split at 6K by the magnetic interaction with the proximal reduced [4Fe-4S] cluster23. We could detect this signal, albeit as traces, in reduced samples of the L122A V74M (Fig. 2b) and L122F V74I (ref. 8) mutants. In contrast, in partially reduced samples of the V74E and V74Q mutants, new, low-intensity signatures of the Ni (at g = 2.245, 2.104, 2.042 and 2.217, 2.091, 2.049, respectively) replaced the usual signal. At low temperature, these signals are broadened (V74E) or split (V74Q) by a magnetic interaction with the reduced, proximal [4Fe-4S] cluster (Fig. 2b). However, it is not clear whether these new species, which account for less than 0.05 spin per molecule, represent catalytic intermediates. For all mutants, the spectral contributions of the Fe-S clusters in both oxidized and reduced samples showed no difference with the WT enzyme. In summary, in the reduced state, the so-called NiC signature is either essentially unaffected (in V74M and L122M V74M; ref. 18), detected as traces (L122F V74I (ref. 8) and L122A V74M, Fig. 2b), or replaced with other low-intensity species (V74Q, V74E, Fig. 2b). The latter observation suggests that introducing a polar amino acid near the Ni perturbs the magnetic characteristics of the NiC state, but the magnitude and shape of the Ni signal in the reduced state are not correlated with the kcat value of the mutants. Regarding asprepared, air-oxidized samples, the relative proportions of the two major EPR-active states (NiA and NiB) change between different preparations of the same enzyme, and we observed no correlation between the ratio of NiA/NiB and the oxygen sensitivity of the different mutants, as determined below using protein film voltammetry. However, we note that the EPR signatures of the oxidized forms of all tunnel mutants characterized so far are essentially identical, suggesting that the mutants react with oxygen to form the same inactive species.
Inhibition kinetics of the NiFe hydrogenase mutants
In protein film voltammetry, the enzyme is adsorbed onto an electrode, electron transfer is direct and the activity is measured as a current20,22. This is useful for following the changes in activity that result from the enzyme being exposed to inhibitors. A particular method consists in monitoring the H2-oxidation current after the concentration of CO or O2 first suddenly increases when an aliquot of solution saturated with inhibitor is injected in the electrochemical
a
g scale
2.4
2.3
2.2
2.1
2.0
WT
AM V74E V74Q
280
300
320
340
B (mT)
b 2.5 2.4 2.3 2.2
g scale 2.1 2.0
1.9
1.8
1.7
WT AM V74E V74Q
300
350
400
B (mT) Q24 Figure 2 | (a,b) EPR characterization of WT, L122A V74M (“AM”), V74E and V74Q forms of D. fructosovorans NiFe hydrogenase in the oxidized (a) and reduced (b) states. Microwave power 10 mW, modulation amplitude 1 mT, T = 100 K (a) or 6 K (b). In b, we indicate the coupling patterns due to the magnetic interaction with the proximal [4Fe-4S] cluster. The narrow line at g = 2 arises from redox mediators.
cell, and then slowly returns to zero as the buffer is flushed by a stream of H2 (ref. 21 and section 2.4.2 in ref. 22). The concentration of H2 is nearly constant, and the inhibitor concentration follows an exponential decay. The concentration of inhibitor need not be independently measured because its change against time is defined by the amount of inhibitor that is injected and by the time constant of the decay, which is determined a posteriori by fitting the change in current. If binding and/or release of the inhibitor is slow, the change in activity is delayed from the time of injection and the rates of inhibition can be measured8,21,22,24. We have shown previously that the electrochemical measurements of the rate of CO release from hydrogenase correlate fairly well with the results of independent measurements carried out with the enzyme in solution, rather than adsorbed onto an electrode8. We used the electrochemical approach presented in ref. 8 to determine the rate of inhibition by CO (kinCO) and the rate of CO release (koutCO) of the selected NiFe hydrogenase variants. For those that are most slowly inhibited by CO (V74Q, V74M, V74E, V74N, V74W, V74F and L122M V74M), the rates could be measured at 40 °C. For L122F V74I, V74D and L122A V74M, we determined the rates at a lower temperature, over a T range of about 15 °C, and we extrapolated the data to 40 °C using Arrhenius plots. Table 1 shows that the rate of inhibition by CO of the V74M and V74Q mutants is more than three orders of magnitude slower than that of the WT (19 ± 6 and 5.4 ± 1.1 s−1 mM(CO)−1, for V74M and V74Q, respectively).
nature CHEMICAL BIOLOGY | Advance online publication | www.nature.com/naturechemicalbiology
3
article
0.85
f
0.95
0.5
0.80 40 0.19
60
80
Ca
100 120
d
0.18
6 4 2 0 h 1.0
1.0 0.98
Dd i/i(0)
V74Q V74M L122M–V74M WT
g
1.5 1.0 0.5 0 1.0
[O2] (µM)
0.90
1.0
b 1.0
i/i(0)
e [CO] (µM)
c
6 4 2 0
i/i(0)
[O2] (µM)
a
Nature chemical biology doi: 10.1038/nchembio.xxx
0.5
0.96 0.5
0 5 10 15
0.17 0 0
200 400 Time (s)
600
0.16 200 250 300 350 400
0
0
100 200 300 Time (s)
0
0
100 200 Time (s)
Figure 3 | The inhibition by O2 of D. fructosovorans NiFe hydrogenase selected mutants, and the reaction with CO and O2 of the FeFe hydrogenases from C. acetobutylicum and D. desulfuricans. C. acetobutylicum, Ca; D. desulfuricans, Dd. The plain lines in b show the change in NiFe hydrogenase activity (current i normalized by its value i(0) just before the inhibitor is added); the dashed lines are the fits to the equations derived in ref. 19; the parameters are given in Table 1. (a) The change in oxygen concentration plotted against time, reconstructed from the amount of O2 injected and the time constant of the exponential decay; the latter is calculated from fitting the change in current8,19,21. (c,d) Enlarged views. E = +200 mV versus SHE, T = 40 °C, pH 7, electrode rotation rate = 2 kr.p.m. (f,h) FeFe hydrogenase activities plotted against time. (e,g) The changes in CO or O2 concentration plotted against time. The dashed lines are the fits to the equations derived in refs. 8,24 (f) and ref. 19 (h) (the parameters are given in the main text, Results section and in Table 1). Q25 The inset in h is a close up of the C. acetobutylicum data. E = −150 mV (f) or +200 mV (h), T = 30 °C, pH 7, = 3 kr.p.m.
To illustrate how the mutations impact oxygen sensitivity, we show in Figure 3a–d the effect of repeatedly injecting aliquots of aerated solution on the activity of four selected NiFe enzymes. The experiment revealing the inhibition by O2 must be carried out at relQ6 atively high electrode potential (+200 mV versus SHE, pH 7), otherwise oxygen reduction contributes to the current and decreases the concentration of O2 that is experienced by the enzyme. At such high electrode potential, hydrogenases are reversibly oxidized to an inactive state18–21. Therefore, the current decreases even under anaerobic conditions, and adding oxygen further increases the rate of inactivation. The resistance increases in the order WT < L122M V74M < V74M < V74Q (Fig. 3b). The V74M and V74Q mutants react with oxygen more slowly than the other enzymes (Fig. 3c), although the effect is small in comparison to the 1,000-fold decrease in kinCO. The L122M V74M enzyme partly reactivates after oxygen has vanished18 (Fig. 3d), despite the fact that the very oxidizing conditions used in this experiment drive the anaerobic formation of the inactive state Q7 called NiB (refs. 20–22). To compare the mutants, we fitted the data in Figure 3b using the model introduced in ref. 19, according to which the enzyme inactivates in two concurrent processes: an oxygen-independent oxidation and the two-step reaction with oxygen, whereby O2 binds with a bimolecular rate constant KinO2 to form an adduct that either dissociates (with rate koutO2) or irreversibly inactivates (equation (13) in Methods). We set koutO2 to zero when the inactivation was fully irreversible, and let it free for the mutants that exhibit a reactivation (V74M, V74F, L122M V74M and L122M V74A). As an example, the V74M and V74Q mutations slow kinO2 fivefold and eightfold, from 32 ± 3 in the WT to 6.5 ± 1.5 and 4 ± 1 s−1 mM(O2)−1, respecQ8 tively (Table 1).
Q9
Inhibition kinetics: C. acetobutylicum and D. desulfuricans FeFe hydrogenases
Figure 3 also compares the behaviors of the FeFe hydrogenases from C. acetobutylicum (in red) and D. desulfuricans (in green), when exposed to CO (Fig. 3e,f) or O2 (Fig. 3g,h). Carbon monoxide is a competitive inhibitor of FeFe hydrogenases, and we have shown previously that it protects against inactivation by O2, suggesting that both inhibitors target the active site19. 4
Figure 3f shows that all things being equal, the clostridial enzyme is much less inhibited by CO than the D. desulfuricans enzyme, and by fitting the data (dashed lines) we found that binding of CO to D. desulfuricans hydrogenase is indeed more than two orders of magnitude faster than to C. acetobutylicum hydrogenase (kinCO = 4,400 ± 260 compared to 16 ± 1 s−1 mM−1, at 30 °C). As is the case with CO, Figure 3h shows that the clostridial enzyme is less sensitive to O2. We found that the enzyme from D. desulfuricans reacts with O2 more than ten times more quickly than that from C. acetobutylicum (kinO2 = 40 ± 8 compared to 2.5 ± 0.4 s−1 mM−1).
Modeling
By using the above described methods, we could characterize each NiFe hydrogenase mutant by a set of five kinetics parameters whose interdependence will now be quantitatively examined. Figure 4 shows the values of kcat, Km, koutCO and kinO2 plotted as a function of kinCO, the observable parameter that was most affected by the mutations. Logarithmic scales are used for all parameters but kcat. Q10 The algebra is detailed in the Supplementary Results.
The relation between koutCO and kinCO
We first focus on the log-log plot of koutCO against kinCO (Fig. 4c). Each square corresponds to a NiFe hydrogenase form, and most of the data points are close to a line of slope 1, showing that the dissociation constant KdCO = koutCO/kinCO is constant for these mutants. We further describe the kinetics of inhibition on the basis of a threeCObinding CO state model similar to that used for interpreting CO kinetics CO CO k [CO] k k [CO] k + 1 + 2 + 1 + 2 in myoglobin25: G E-CO E + CO E + CO G E-CO CO CO CO CO k k k k CO CO CO CO CO −1 −1 −1 CO −1 COk CO CO CO [CO] k +1k k[CO] +2k k+ CO CO k+ [CO] k[CO] [CO] kk+CO 2k+2 1 + 1k+1 [CO] + 1 G 2 [CO] + 2k E-CO k+CO2 E E+ CO E + CO E-CO + 1 + 2 +G 1 E+ CO + CO + CO G G G E-CO E-CO CO CO E-CO EE + CO G E-CO E + CO G CO CO CO CO CO k-CO k-CO E-CO (1) k 1k k-CO 1k CO CO k-CO k k k 1 1 1 -1k -1 -1k -1k -1 k-CO1 -1 -1 → →-1
← ←CO molecule where E + CO is a state where the is in the solvent, G is a geminate (transient) state where the ligand has passed the bottleneck in the tunnel, and E-CO represents the final, inactive bound state. k+1CO[CO] and k−1CO are the first-order rate constants relative to the diffusion step—that is, the formation and dissociation of the geminate state; K1CO = k−1CO/k+1CO denotes the corresponding equilibrium
nature chemical biology | Advance online publication | www.nature.com/naturechemicalbiology
article
kcat (s–1)
WT NiFe
L122F V74I
V74F L122A V74M V74D
V74M
V74Q
a
V74E V74W V74N L122M V74M
Nature chemical biology doi: 10.1038/nchembio.xxx
O
O
500
Km (m atm H2)
104
co –1 kout (s )
O
O2
Slop
e –1
102
102
pe
*
1
*
Slo
(2a) (2b)
The meaning of equation (2a) is that the bimolecular rate of inhibition by O2 is determined by the slower of two processes—namely the diffusion of O2 toward the active site (k+1O2) or the rate of reaction of O2 at the active site, with an apparent bimolecular rate constant kinO2max. Furthermore, the observation in Figure 4d that the mutants that are most slowly inhibited react with either ligand at about the same rate (the data points are close to the y = x line) suggests that: O CO (3) k+12 = kin
1
1
In the Supplementary Results, we show that equation (3) holds when two conditions are satisfied: (i) CO in the geminate state binds more quickly to the active site than it is released to the solvent (k+2CO >> k−1CO). This condition should be easily fulfilled in the case of mutants whose tunnels are obstructed (hence k−1CO is small), and implies that the measured bimolecular rate constant for CO inhibition kinCO is determined by the rate of crossing the 74-122 gate:
* 10–2
e
y=
x
102
Lin
kin02 (s–1 mM–1)
O
kin2 max = k+22 / K1
b
d
O
1 / kin2 = 1 / k+12 + 1 / kin2 max
1,000
0
c
inhibition kinetics have the same values for CO and O2. Hence, we describe the kinetics of inhibition by O2 using a two-step model like that for CO (equation (1), from which we derive in the Supplementary Results the rate of inhibition kinO2):
CO CO k+1 = kin
101
101
102
103 co kin (s–1 mM–1)
104
105
Figure 4 | Kinetic properties of selected NiFe hydrogenase mutants. (a–d) Maximal turnover rate (kcat, in a), Michaelis constant (Km, in b), rate of CO release (koutCO, in c) and rate of inhibition by O2 (kinO2, in d), all plotted against the rate of binding of CO (kinCO). In a, the horizontal gray line marks the range of kcat value observed for most mutants. The dashed line in b has slope −1; the plain line is the best fit to equation (9). In c, the dashed line has slope +1, and the dashed and solid horizontal arrows indicate the 12-fold and 30-fold changes in kinCO resulting from replacing a carboxylate with an amide at position 74, or from increasing the length of the side chain of residue 74 by one CH2. In c, the WT data point is not within the limits of the graph; in that case, CO inhibition is so fast that the error on the value of kinCO introduced by the extrapolation to 40 °C is very large. In d, the dashed Q26 line depicts y = x; the plain line is the best fit to equation (6). Q11 constant. Similarly, k+2CO, k−2CO and K2CO = k−2CO/k+2COcharacterize
bond formation and breaking at the active site. Below, we shall use similar notations for O2 and H2. In this framework, the dissociation constant KdCO is the product of K1CO and K2CO (see Supplementary Equation 2 in the Supplementary Results), and the fact that KdCO is constant (Fig. 4c) strongly suggests that most mutations affect neither one nor the other (we consider it unlikely that the mutations have exactly inverse effects on K1CO and K2CO). Because they leave K1CO unchanged, these mutations slow the rates of diffusion to and from the active site (k+1CO and k−1CO) in the same manner. In contrast, the three mutants pointed to by asterisks in Figure 4c bind CO less tightly than the WT, suggesting that either K1CO or K2CO is greater.
The relation between kinO2 and kinCO
Figure 4d shows that the variations of kinO2 and kinCO are correlated. This suggests that some of the rate constants that determine the
(4)
(ii) The rates of diffusion of O2 and CO are very similar, which is consistent with the finding that in myoglobin, NO, CO and O2 diffuse at the same rate25. O
CO k+1 = k+12
(5)
Substituting equation (3) in equation (2a) gives O
O
CO 1 / kin2 = 1 / kin + 1 / kin2 max
(6)
Equation (6) with kinO2max = 32 s−1 mM(O2)−1 describes fairly well the entire dataset in Figure 4d (plain line).
The relation between Km and kinCO
Finally, we determine the effect of the mutations on the rate of subCOCO COCO between strate diffusion along the tunnel by interpreting the relation k1 [CO] k2 k [CO] k + 1 + 2 + + +CO CO G E2ECO+and G E-CO E-CO the Michaelis constants for H the rate of inhibition by CO CO CO CO CO CO k [CO] k + 1 + 2 k k k k −1 −1 −1 −1 (Fig. 4b). We start from a simple model G of catalysis E + CO E-CO CO k−CO k 1 −1 H2 H2 H2 H2 k H2[H 2 ] H k + 1 2 + k [ H k ] [ H ] k [ H ] cat k k+E [H+1 ] 1 2 ] k k 2 EH + + -k k 1 +1 2 k2+1 2[+H cat cat cat + ¾ ¾ ® E + 2 H + 22eecat cat E E → 2¾ → EH EH EH ¾ ® ¾ ¾ + 2+ EH+ 22+ H 2+ e+ 2-eEH ® E +2® 2H + e+ HEH EE ¾ ¾ ® +® H + e2(7) E EH ¾ ¾ E 2H 2 2¾ 2E H H 2 2 k-Hk12-k12 -12kk--H112k-12
2
→ ←← ←terms of the maximal turnover rate at infinite and we express Km in concentration of H2 (kcat) and the rates of diffusion of H2 along the tunnel k+1H2 and k−1H2: H k +k 2 K m = cat H -1 (8) k+12 Regarding hydrogenases8,17 and various other redox enzymes26–28, mutations intended to block the substrate-access channel were found to increase the Michaelis constant; our model explains under which conditions the Michaelis constant depends on the rate of substrate diffusion. Equation (8) shows that provided the mutation has no effect on kcat (Fig. 4a), an x-fold increase in Km reveals the decrease of the rate of diffusion (k+1H2) by a factor of x or more. Indeed, the Michaelis constant equates K1H2 if H2 diffuses very quickly (k−1H2 >> kcat), whereas it is proportional to the reciprocal of k+1H2 if k−1H2 is small. Regarding the latter limiting case, the observation in
nature CHEMICAL BIOLOGY | Advance online publication | www.nature.com/naturechemicalbiology
5
article
Nature chemical biology doi: 10.1038/nchembio.xxx
Figure 4b that for the mutants whose diffusion is most slowed, Km is proportional to the reciprocal of kinCO demonstrates that the rates of diffusion of H2 and CO are proportional to each other. Therefore, we note that α is the ratio of the two (α = k+1H2/k+1CO) and we use Q12 equation (4) to rewrite equation (8) as follows:
k /a H K m = catCO + K1 2 kin
(9)
The fit of the data in Figure 4b (plain line) returns K1H2 = 15 matm(H2) and kcat/α = 30 atm(H2) s−1 mM(CO)−1. Using the average value of kcat = 725 ± 75 s−1 and the Henry constant for H2 (0.78 mM Q13 atm−1 at 30 °C), we obtain α = 31 ± 3 mM(CO) mM(H2)−1.
DISCUSSION
We have used protein film voltammetry to determine the overall rate constants for CO and O2 binding and release in two homologous FeFe hydrogenases and in a series of NiFe hydrogenase mutants whose tunnel was partially obstructed but that react with O2 to form Q14 the same inactive species (Fig. 2a). The mutations increased the Km for H2 (Fig. 4b), but they did not affect the maximal rate of H2 oxidation (kcat), except when a carboxylic function was introduced in the tunnel near the active site (in V74E and V74D, Fig. 4a). We interpret the simple relations deduced above (equations (6) and (9)), which exist between the rate of inhibition by CO and the other independent observables, to clarify the relation between rate of diffusion in the tunnel and resistance to oxygen. Our analysis shows that the bimolecular rate of inhibition by the competitive inhibitor CO (the abscissa in Fig. 4) represents the rate of diffusion of CO to the active site (see equation (4)). The effect of the mutations that keep the side chain apolar simply correlates with the Van der Waals volume of the amino acid: decreasing the bulk of the position 122 residue increases the diffusion rate (compare in Fig. 4 the values of kinCO for V74M and L122A V74M; or L122F V74I and the WT). The effect of changing the size and polarity of the residue at position 74 can be appreciated by comparing the V74E, V74D, V74Q and V74N mutants two at a time: increasing the length of the position 74 residue side chain by one CH2 (D74E or N74Q) slows Q15 the diffusion rate about 30-fold. Replacing a carboxylic acid with an Q16 amide, keeping the Van der Waals volume constant (E74Q or D74N), slows diffusion by a factor of about 12 (the continuous and dashed horizontal arrows in Fig. 4c depict these variations of rate). The two contributions, size and polarity, are independent of each other, and therefore the single replacement of Asp74 with glutamine decreases the rate of diffusion by more than two orders of magnitude. Our data confirm that bulky amino acids slow down diffusion by filling the tunnel, which is consistent with experimental results obtained with myoglobin mutants4. Overall, there seems to be no experimental evidence supporting the mechanism according to which large residues would speed up transport by promoting the formation of transient cavities6. It is noteworthy how much the up to 104-fold variation in diffusion rates in hydrogenase mutants contrasts with the much smaller effects usually observed in kinetic studies of ligand binding to globins, possibly because in the latter, fluctuations or alternate pathways level the effect of the mutations. Yet we cannot exclude that fluctuations and/or electrostatics play a role in defining the diffusion kinetics in hydrogenase also. Indeed, comparing the WT, L122F V74I, L122M V74M and V74M mutants, we have previously noted the absence of correlation between diffusion rate and diameter of the bottleneck in the tunnel, calculated from the static structures (Protein Data Bank entries 1YQW, 3CUR, 3CUS and 3H3X)8,18. Moreover, the large retardation resulting from the isosteric replacement of a carboxylate with an amide may reveal an electrostatic effect, such as the stabilization of a water molecule that is part of the barrier to ligand entry, as observed in certain myoglobin mutants4.
6
Figure 4d compares the overall rates of inhibition by CO and O2 for all NiFe hydrogenase mutants studied here. For the WT NiFe hydrogenase and most mutants, the rate of inhibition by CO is orders of magnitude faster than the rate of aerobic inhibition. Considering that CO and O2 diffuse within the protein at about the same rate (equation (5)), this observation implies that the rate of inhibition by O2 is limited by the reaction at the active site. Mutations such as L122F V74I and L122M V74M decrease the rate of intramolecular diffusion by blocking the tunnel, but this has no effect on the overall reaction with O2 because the diffusion process (k+1O2 in equation (2a)) does not limit the rate of inhibition. However, other mutations decrease the rate of diffusion by so much (four orders of magnitude for V74M and V74Q) that this step becomes slower than the reaction of O2 at the active site, and this decreases the overall rate of inhibition by O2. This important result is pictured in Supplementary Figure 1 of the Supplementary Results. A major prediction of our model, and a test of the above statements that CO and O2 diffuse toward the active site at the same rate whereas CO reacts at the active site more quickly than does O2, is that there should be no enzyme for which the rate of inhibition by O2 is faster than the rate of inhibition by CO. None of the hydrogenases we have characterized disobeyed this rule. This includes a number of D. fructosovorans NiFe hydrogenase mutants not discussed here: V74H, V74I, L122M, L122F, L122A, L122C, V117M E22M, E25D (ref. 29), H184C (ref. 30) and P238C (ref. 31), and also the FeFe hydrogenases from C. acetobutylicum and D. desulfuricans. Regarding the latter, comparing the bimolecular rates of inhibition by CO and O2 (Fig. 3e,f) shows that the clostridial enzyme is inhibited much more slowly than the enzyme from D. desulfuricans. The clostridial enzyme reacts relatively slowly with CO and O2, and the two reaction rates are similar (16 and 2.5 s−1 mM−1, respectively); by analogy with the NiFe hydrogenase variants, this suggests that diffusion along the tunnel of the clostridial hydrogenase limits the rate of inhibition by O2. We are engineering this enzyme to determine Q17 which of the residues that line the hydrophobic tunnels of the two hydrogenases2 may be responsible for this. We conclude that the properties of the tunnel do not affect aerobic inactivation in D. desulfuricans FeFe hydrogenase and in the WT form and many mutants of D. fructosovorans NiFe hydrogenase (including L122F V74I) because the rate of inhibition is not limited by the rate of intramolecular diffusion. This is unlike the situation in C. acetobutylicum FeFe hydrogenase and in the V74M, V74Q and V74E mutants of D. fructosovorans NiFe hydrogenase. Modifying the tunnel changes the rate of transport of all diatomic molecules, including H2. Indeed, the observation in Figure 4b that Km is proportional to 1/kinCO when CO inhibition is slow indicates that the rates of diffusion of H2 and CO are proportional to each other (see the demonstration of equation (9)). This shows that H2 and CO access the active site via the same path, which is consistent with the earlier independent observation of ours that the rates of release of CO and H2 are correlated8. That is not to say that H2 and CO diffuse at the same rate: in the two mutants whose diffusion rates are most affected, the pseudo first-order rate of inhibition by CO under 1 atm of CO (5 and 20 s−1 for V74Q and V74M, respectively; Table 1) is much slower than the apparent value of kcat under one atm of H2 (kcatapp = 210 ± 15 and 460 ± 15 s−1, respectively), which incorporates H2 binding. By interpreting the relation between Km and kinCO, we found that H2 diffuses about 30 times faster than CO. Therefore, the mutations in the tunnel slow CO and oxygen access, but diffusion of H2 remains fast enough that they decrease the apparent turnover rate under 1 atm of H2 (kcatapp in Table 1) no more than fourfold. Regarding the V74Q mutant, from the value of kinCO = 5.4 ± 1.1 s−1 mM(CO)−1 and the Henry constant for H2, we calculate k+1H2 = 130 ± 40 s−1 atm(H2)−1; the fact that this value nearly equates the turnover rate under 1 atm of H2 (kcatapp = 210 ± 15 s−1) is a clear indication that substrate diffusion limits the rate of H2 oxidation in this mutant.
nature chemical biology | Advance online publication | www.nature.com/naturechemicalbiology
Figure 3c clearly shows that the V74M and V74Q mutations increase the enzyme’s resistance to oxygen by slowing the rate of inhibition. It is therefore noteworthy that the L122M V74M mutant is more resistant than the WT enzyme (gray line in Fig. 3b) despite the fact that it reacts with oxygen at the same rate (Fig. 3c). This is because the double mutant slowly reactivates after it has been exposed to oxygen (Fig. 3d); this process has small amplitude, but it is significant because under the very oxidizing conditions used in these experiments, one expects nothing but the inactivation of the enzyme20–22. So far, we have observed this unexpected reactivation with the V74M, V74F, L122M V74M and L122A V74M mutants. We showed18 that the phenotype of the V74M and L122M V74M mutants is not a consequence of a modification of the structure of the active site, but rather reveals subtle changes in the kinetics of the reaction with O2. In any case, Figure 3a–d clearly illustrates the distinct strategies that can be used for improving resistance to oxygen, namely blocking the gas tunnel to slow oxygen access (this is operational for the V74Q, V74M and V74E mutants) or modifying the chemistry at the active site. Combining beneficial modifications will certainly be the key to designing oxygen-tolerant hydrogenases.
METHODS
Protein purification and microbial strains, plasmids and growth conditions. The FeFe hydrogenase from D. desulfuricans was purified as described32, except that fully anaerobic conditions were used throughout. The purification procedure of C. acetobutylicum FeFe hydrogenase was described previously33. The E. coli strain DH5α, plasmids and growth conditions used in this study were described previously34. The NiFe hydrogenase–deleted D. fructosovorans strain MR400 [hynnpt hynABC] was grown anaerobically at 37 °C in basal medium, consisting of (each in g l−1) NH4Cl, 0.5; NaCl, 1.0; Na2SO4, 4.7; KH2PO4, 1.3; K2HPO4, 1.3; fructose, 5.4; yeast extract, 0.1; cysteine-HCl, 0.4; and 1 ml trace element solution SOS (ref. 34), at pH 6.8. The basal medium was boiled under 1 atm of Ar, then cooled and autoclaved. Prior to inoculation, 2.5 mM Na2S, 10 mM NaHCO3, 1 mM CaCl2 and 1 mM MgCl2 were added. Kanamycine at 50 mg l−1 was present routinely, and gentamycin 20 mg l−1 was added only when cells harbored the plasmid pBGF4. Site-directed mutagenesis, production and purification of the proteins carrying the strep tag II sequence at the N terminus of the large subunit were described previously8. The procedures for assaying hydrogen oxidation are described in the Supplementary Methods. Electrochemistry. All electrochemical experiments were carried out in a glove box
Q18 filled with N2, with the equipment described previously21.The preparation of the
enzyme films is described in the Supplementary Methods. The electrochemical data were analyzed with an in-house program called SOAS35. For measuring the Km for H2, we used a method discussed previously8,21,22. While the activity is measured at −160 mV, pH 7, the cell solution initially saturated with H2 is degassed by bubbling argon at t ≥ t0. The substrate concentration is [H2] = [H2]0 exp(−(t − t0)/τ), and the value Km is determined by fitting the change in current to i(t ) =
Q19
i max 1+
Km [H2 ]0
exp
t −t 0 t
(10)
For determining the kinetics of CO binding and release8,22,24, the activity is measured after small aliquots of a buffer saturated withCO and kept in a capped serum bottle are injected into the cell solution, which is 〉ushed with H2. “One atm of CO” in Table 1 refers to the concentration of CO in a solution that is equilibrated with 1 atm of CO at 25 °C. We assume that the active enzyme converts to an inhibited form with a pseudo first-order rate constant kCOin,app [CO]0 exp(−t/τ), where [CO]0 CO CO denotes the initial rate of inactivation after CO is injected. The first-order rate k [CO] k + 1 + 2 of E + CO G E-CO activation is kCOout. CO CO CO CO [CO ] ] CO k COkin,app [kCO ][CO
k− k 1 −1
in,app in,app kin,app [CO ] active inactive active active inactive inactive CO CO CO active kout CO kout kout inactive
kout
(11)
→
The current is proportional to the fraction of active enzyme. Its fit to equation (54) in ref. 22 returns kCOout, kCOin,app and← τ. Because CO (unlike O2) is a competitive inhibitor6,9, the maximal value of the inhibition rate constant is obtained by correcting kCOin,app using equation (55) in ref. 22: [H ] CO CO kin = kin,app × 1 + 2 Km
(12)
For studying the inhibition by O2, the method is the same except that we inject aliquots of a solution saturated with either oxygen or air at 25 °C (refs. 19,21). We assume that the active enzyme inactivates in a first-order oxygen-independent
article
COCO
COCO
k [CO] k k1 [CO] k2 + + + 1 + 2 EE++CO CO GG COCO E-CO E-CO COCO COCO CO CO k1 [CO] k2 k k k k k k +[CO] 1 + 2 + + −1 −1 −1 −1 CO G E-CO ECOE++CO G E-CO CO CO CO CO CO k [CO] k process (rate kinact 1) and an oxygen-dependent reaction that consists in k1reversible k k−k1 −1 + 1 + −21 − 19 E + CO G E-CO oxygen binding followed by irreversible inactivation : CO CO k − →→ k 1 −1
Nature chemical biology doi: 10.1038/nchembio.xxx
k
k O2 [O ]2 2 O2 OO
k
2[O in k kk inact inact 2 kk ] 2inac kin [O2k]inact 2 k1inact kinact kkinact [[2OO ]] tive kkinact kin kinact kinact 2 in in [O in in 1[O2 ] inact 1 active 1 kinact inact 2 inactive 11 inact 22 2k inactive ¬1¾ ¾¾ ¾ ¾¾ ® k2inO] [OO kinact 2 ] 1¾¾ inact 2 active t (13) inactive inactive ¬ ¬ ¾¾ ¾¾ ¾ ¾ active active inactive inac inac ¬ ¾¾ ¾ ¾ inactive ¾ ¾¾ ¾¾ ® ® ¬ ¾¾ ¾ active ¾¾¾ ® ina t¾¾ ive tive ive inactive inactive ive ti inactive ¬ ¾ active inac ¾ ¾¾ ® t¾ inactive 2 inactive ¬ ¾¾ active inac ¾ ® t2ive inactive ← ← → → inac inac O k O2 OO inactive ¬¾¾ ¾ active ® inactive 2 2 inactive ¾¾¾ out O2k k k k O2 k O2 k 2 O2
out
kout
out out out
O2
O2
out
→ ←← For WT NiFe hydrogenase and the mutants that exhibit no reactivation phase after O O2 is 〉ushed out, we used the above model, ← setting k out = kinact 2 = 0.
2
out
Q20
The bimolecular rates of inhibition in units of s−1 per atm of CO or O2 were converted to units of s−1 per mM using the Henry constants of 0.96 and 1.25 mM atm−1, respectively.
Q21
Accession codes. Protein Data Bank: The structures discussed in the text were deposited as part of previous studies as entries 1YQW, 3CUR, 3CUS and 3H3X.
Q22
Received 28 May 2009; accepted 26 October 2009; published online XX XX 2009
References
1. Maynard, E.L. & Lindahl, P.A. Evidence of a molecular tunnel connecting the active sites for CO2 reduction and acetyl-CoA synthesis in acetyl-coa synthase from Clostridium thermoaceticum. J. Am. Chem. Soc. 121, 9221–9222 (1999). 2. Fontecilla-Camps, J.C., Volbeda, A., Cavazza, C. & Nicolet, Y. Structure/ function relationships of [NiFe]- and [FeFe]-hydrogenases. Chem. Rev. 107, 4273–4303 (2007). 3. Cohen, J., Kim, K., King, P., Seibert, M. & 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 13, 1321–1329 (2005). 4. Nienhaus, K., Deng, P., Olson, J.S., Warren, J.J. & Nienhaus, G.U. Structural dynamics of myoglobin: ligand migration and binding in valine 68 mutants. J. Biol. Chem. 278, 42532–42544 (2003). 5. Ruscio, J.Z. et al. Atomic level computational identification of ligand migration pathways between solvent and binding site in myoglobin. Proc. Natl. Acad. Sci. USA 105, 9204–9209 (2008). 6. Cohen, J. & Schulten, K. O2-migration pathways are not conserved across proteins of a similar fold. Biophys. J. 93, 3591–3600 (2007). 7. Salomonsson, L., Lee, A., Gennis, R.B. & Brzezinski, P. A single-amino-acid lid renders a gas-tight compartment within a membrane-bound transporter. Proc. Natl. Acad. Sci. USA 101, 11617–11621 (2004). 8. Leroux, F. et al. Experimental approaches to kinetics of gas diffusion in hydrogenase. Proc. Natl. Acad. Sci. USA 105, 11188–11193 (2008). 9. Vincent, K.A. et al. Electrocatalytic hydrogen oxidation by an enzyme at high carbon monoxide or oxygen levels. Proc. Natl. Acad. Sci. USA 102, 16951–16954 (2005). 10. Vincent, K.A. et al. Enzymatic catalysis on conducting graphite particles. Nat. Chem. Biol. 3, 761–762 (2007). 11. Karyakin, A.A. et al. The limiting performance characteristics in bioelectrocatalysis of hydrogenase enzymes. Angew. Chem. Int. Ed. 46, 7244–7246 (2007). 12. Hambourger, M. et al. [FeFe]-hydrogenase-catalyzed H2 production in a photo-electrochemical biofuel cell. J. Am. Chem. Soc. 130, 2015–2022 (2008). 13. Ghirardi, M.L., Dubini, A., Yu, J. & Maness, P.-C. Photobiological hydrogenproducing systems. Chem. Soc. Rev. 38, 52–61 (2009). 14. Volbeda, A., Montet, Y., Vernède, X., Hatchikian, E.C. & Fontecilla-Camps, J.C. High-resolution crystallographic analysis of Desulfovibrio fructosovorans NiFe hydrogenase. Int. J. Hydrogen Energy 27, 1449–1461 (2002). 15. Buhrke, T., Lenz, O., Krauss, N. & Friedrich, B. Oxygen tolerance of the H2-sensing [NiFe] hydrogenase from Ralstonia eutropha H16 is based on limited access of oxygen to the active site. J. Biol. Chem. 280, 23791–23796 (2005). 16. Duché, O., Elsen, S., Cournac, L. & 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. 272, 3899–3908 (2005). 17. Ludwig, M., Cracknell, J.A., Vincent, K.A., Armstrong, F.A. & Lenz, O. Oxygen-tolerant H2 oxidation by membrane-bound [NiFe]-hydrogenases of Ralstonia species: coping with low-level H2 in air. J. Biol. Chem. 284, 465–477 (2009). Q27 18. Dementin, S. et al. Introduction of methionines in the gas channel makes [NiFe] hydrogenase aero-tolerant. J. Am. Chem. Soc. 131, 10156–10164 (2009). 19. Baffert, C. et al. Hydrogen-activating enzymes: activity does not correlate with oxygen-sensitivity. Angew. Chem. Int. Ed. 47, 2052–2055 (2008). 20. Vincent, K.A., Parkin, A. & Armstrong, F.A. Investigating and exploiting the electrocatalytic properties of hydrogenases. Chem. Rev. 107, 4366–4413 (2007). 21. Léger, C., Dementin, S., Bertrand, P., Rousset, M. & Guigliarelli, B. Inhibition and aerobic inactivation kinetics of Desulfovibrio fructosovorans NiFe
nature CHEMICAL BIOLOGY | Advance online publication | www.nature.com/naturechemicalbiology
7
article
Q28
Q29
Nature chemical biology doi: 10.1038/nchembio.xxx
hydrogenases studied by protein film voltammetry. J. Am. Chem. Soc. 126, 12162–12172 (2004). 22. Léger, C. & Bertrand, P. Direct electrochemistry of redox enzymes as a tool for mechanistic studies. Chem. Rev. 108, 2379–2438 (2008). 23. Guigliarelli, B. et al. Structural organization of the Ni and [4Fe-4S] centers in the active form of Desulfovibrio gigas hydrogenase. Analysis of the magnetic interactions by electron paramagnetic resonance spectroscopy. Biochemistry 34, 4781–4790 (1995). 24. Almeida, M.G. et al. A needle in a haystack: the active site of the membranebound complex cytochrome c nitrite reductase. FEBS Lett. 581, 284–288 (2007). 25. Rohlfs, R.J., Olson, J.S. & Gibson, Q.H. A comparison of the geminate recombination kinetics of several monomeric heme proteins. J. Biol. Chem. 263, 1803–1813 (1988). 26. Riistama, S., Puustinen, A., Verkhovsky, M.I., Morgan, J.E. & Wikstrom, M. Binding of O2 and its reduction are both retarded by replacement of valine 279 by isoleucine in cytochrome c oxidase from Paracoccus denitrificans. Biochemistry 39, 6365–6372 (2000). 27. Chen, L., Lyubimov, A.Y., Brammer, L., Vrielink, A. & Sampson, N.S. The binding and release of oxygen and hydrogen peroxide are directed by a hydrophobic tunnel in cholesterol oxidase. Biochemistry 47, 5368–5377 (2008). 28. Barney, B.M., Yurth, M.G., Dos Santos, P.C., Dean, D.R. & Seefeldt, L.C. A substrate channel in the nitrogenase MoFe protein. J. Biol. Inorg. Chem. 14, 1015–1022 (2009). 29. Dementin, S. et al. A glutamate is the essential proton transfer gate during the catalytic cycle of the NiFe hydrogenase. J. Biol. Chem. 279, 10508–10513 (2004). 30. Dementin, S. et al. Changing the ligation of the distal [4Fe4S] cluster in NiFe hydrogenase impairs inter- and intramolecular electron transfers. J. Am. Chem. Soc. 128, 5209–5218 (2006). 31. Rousset, M. New shuttle vectors for the introduction of cloned DNA in Desulfovibrio. Plasmid 39, 114–122 (1998). 32. Hatchikian, E.C., Forget, N., Fernandez, V.M., Williams, R. & Cammack, R. Further characterization of the [Fe]-hydrogenase from Desulfovibrio desulfuricans ATCC 7757. Eur. J. Biochem. 209, 357–365 (1992). 33. Fourmond, V. et al. Correcting for electrocatalyst desorption and inactivation in chronoamperometry experiments. Anal. Chem. 81, 2962–2968 (2009).
8
34. Rousset, M. et al. [3Fe-4S] to [4Fe-4S] cluster conversion in Desulfovibrio fructosovorans [NiFe] hydrogenase by site-directed mutagenesis. Proc. Natl. Acad. Sci. USA 95, 11625–11630 (1998). 35. Fourmond, V. et al. SOAS: a free software to analyse electrochemical data and other one-dimensional signals. Bioelectrochem. 76, 141–147 (2009). 36. Demuez, M. et al. Complete activity profile of Clostridium acetobutylicum [FeFe]-hydrogenase and kinetic parameters for endogenous redox partners. FEMS Microbiol. Lett. 275, 113–121 (2007).
Acknowledgments
This work was funded by the Centre National de la Recherche Scientifique, Commissariat à l’Energie Atomique, Agence Nationale de la Recherche, the University of Provence and the City of Marseilles, and supported by the Pôle de Compétitivité Capénergies. The Groupe de Recherche 2977 (“Bio-hydrogène”) provided the publication fees for this article.
Author contributions
P.-P.L. designed mutants of the NiFe enzyme, performed mutagenesis, carried out protein purification, solution assays and electrochemical measurements, and analyzed data, with the support of M.R. and C.L. F.L. performed electrochemical measurements on several forms of the NiFe hydrogenase (WT, L122M V74M, V74M, L122F V74I). B.B. characterized by EPR the NiFe hydrogenase mutants, with the support of B.G. S.D. designed mutants of the NiFe enzyme, performed mutagenesis, carried out protein purification and solution assays, and interpreted studies, with the support of M.R. C.B. performed the electrochemical characterization of the two FeFe hydrogenases and analyzed the data. T.L. purified the FeFe hydrogenase from C. acetobutylicum and assayed its activity, with the support and advice of I.M.-S. and P.S. V.F. contributed to modeling. P.C. characterized the V74W mutant, and analyzed the data. C.C. purified the FeFe hydrogenase from D. desulfuricans, with the support of J.F.-C. P.-P.L., S.D., B.B., C.B., M.R., B.G., P.B. and C.L. co-designed research. S.D., P.B. and C.L. conceptualized, analyzed and interpreted all studies and co-wrote the manuscript.
Additional information
Supplementary information is available online at http://www.nature.com/ naturechemicalbiology/. Reprints and permissions information is available online at http://npg.nature.com/reprintsandpermissions/. Correspondence and requests for materials should be addressed to C.L.
nature chemical biology | Advance online publication | www.nature.com/naturechemicalbiology
Relating diffusion along the substrate tunnel and oxygen sensitivity in hydrogenase Pierre-Pol Liebgott, Fanny Leroux, Bénédicte Burlat, Sébastien Dementin, Carole Baffert, Thomas Lautier, Vincent Fourmond, Pierre Ceccaldi, Christine Cavazza, Isabelle Meynial-Salles, Philippe Soucaille, Juan Fontecilla-Camps, Bruno Guigliarelli, Patrick Bertrand, Marc Rousset & Christophe Léger EDITORIAL SUMMARY AOP: Determining rates of gas diffusion along enzyme channels has proven difficult. Comparing the rates of O2- and CO-mediated inhibition of wild-type and mutant hydrogenases reveals the quantitative impact of amino acid changes on diffusion rates Q2 and suggests design strategies for creating oxygen-tolerant hydrogenases. NiFe
L122
V74
E25
Substrate (H2) or inhibitors (CO, O2)
QUERY FORM Nature Chemical Biology Manuscript ID
[Art. Id: 276]
Author Editor Publisher AUTHOR:
The following queries have arisen during the editing of your manuscript. Please answer queries by making the requisite corrections directly on the galley proof. It is also imperative that you include a typewritten list of all corrections and comments, as handwritten corrections sometimes cannot be read or are easily missed. Please verify receipt of proofs via e-mail
Query No.
Nature of Query
Q1
Please carefully check the spellings of all the author names and affiliations.
Q2
Editorial summaries are short statements written by the editor who handled your manuscript that will appear with your graphical abstract in the online and printed versions of your paper. We request that you read over the current summary for scientific accuracy. These statements are limited to 40-50 words, and the editor will take your comments into consideration in finalizing the editorial summary.
Q3
Sentence correct as edited?
Q4
Sentence correct as edited?
Q5
Please spell out.
Q6
Please define ‘SHE’.
Q7
Sentence correct as edited?
Q8
Sentence correct?
Q9
Please cut to fit on one line.
Q10
Sentence correct as edited?
Q11
Sentence correct?
Q12
Sentence correct as edited?
Q13
Sentence correct as edited?
Q14
Sentence correct?
Nature Publishing Group
QUERY FORM Nature Chemical Biology Manuscript ID
[Art. Id: 276]
Author Editor Publisher AUTHOR:
The following queries have arisen during the editing of your manuscript. Please answer queries by making the requisite corrections directly on the galley proof. It is also imperative that you include a typewritten list of all corrections and comments, as handwritten corrections sometimes cannot be read or are easily missed. Please verify receipt of proofs via e-mail
Query No.
Nature of Query
Q15
Sentence correct as edited?
Q16
OK?
Q17
Sentence correct as edited?
Q18
Correct as edited?
Q19
Should this be ‘flushed’? Character has been corrupted.
Q20
‘flushed’?
Q21
Sentence correct?
Q22
Section added per journal style. Correct? Please edit description of structures as necessary.
Q23
Legend correct as edited?
Q24
Please provide a brief overall title sentence that describes the figure as a whole without referring to specific parts or panels.
Q25
Please revise legend so that each panel is described separately, and panels are described in sequential order.
Q26
Please state nature of error bars.
Q27
Year correct as edited per PubMed?
Q28
Ref. correct as edited per PubMed?
Q29
First author initials correct as edited per PubMed?
Q30
Please cut to fit on one line. Nature Publishing Group
