J Biol Inorg Chem (2000) 5 : 682±691 DOI 10.1007/s007750000152

OR IG IN AL A RT IC L E Patrick Bertrand ´ François Dole ´ Marcel Asso Bruno Guigliarelli

Is there a rate-limiting step in the catalytic cycle of [Ni-Fe] hydrogenases?

Received: 3 April 2000 / Accepted: 14 June 2000 / Published online: 11 October 2000  SBIC 2000

Abstract The question of the existence of a rate-limiting step in the catalytic cycle of Ni-Fe hydrogenases was taken up by using the sets of data available in the case of two specific enzymes: the hydrogenase from Thiocapsa roseopercisina, in which isotope effects have been systematically investigated over a wide pH range, and the enzyme from Desulfovibrio fructosovorans, for which the activities and the redox properties have been studied in two different forms, the wild type and the P238C mutant. When these data are analyzed in the light of appropriate kinetic models, it is concluded that electron transfer and proton transfer are rate limiting in the H2 uptake and H2 evolution reactions, respectively. This proposal is consistent with the data available from other Ni-Fe enzymes. Keywords Hydrogenase ´ Metalloenzyme ´ Nickel enzyme ´ Iron-sulfur center ´ Catalytic cycle

Introduction Hydrogenases are enzymes involved in the production and consumption of molecular hydrogen by microorganisms, which are classified as [Ni-Fe] and Fe-only hydrogenases according to the metal composition of their active site [1, 2]. Despite detailed structural information provided by recent X-ray crystal studies, the catalytic mechanism of [Ni-Fe] hydrogenases is not yet well understood. Some uncertainty remains about the nature of the various steps which take place at the active site itself: on the one hand, the respective role of the two metal sites of the dinuclear [Ni-Fe]

P. Bertrand ()) ´ F. Dole ´ M. Asso ´ B. Guigliarelli Laboratoire de BioØnergØtique et IngØnierie des ProtØines, UPR CNRS 9036, IBSM et UniversitØ de Provence, 31 Chemin Joseph Aiguier, 13402 Marseille Cedex 20, France E-mail: [email protected]

center is still controversial [3, 4, 5, 6, 7]. On the other hand, the influence of the selenocysteine present in some [Ni-Fe] hydrogenases on the catalytic mechanism has not yet been elucidated [2, 6]. The precise role of the Fe-S centers in the catalytic cycle is not well understood either. Although the quasi-linear arrangement of the Ni-Fe and Fe-S centers suggests an evident electron transfer chain connecting the active site to electron donors and acceptors, the location of the high-potential [3Fe-4S]+,0 center between the two low-potential [4Fe-4S]2+,+ centers casts doubts on the efficiency of this electron transfer system [8]. Recent experiments carried out in Desulfovibrio fructosovorans hydrogenase have shown that converting the [3Fe-4S]+,0 center characterized by E90=+65 mV into a [4Fe-4S]2+,+ center characterized by E90= ±250 mV results in only modest changes in both the H2 uptake and H2 evolution activities, suggesting that the [3Fe-4S]+,0 center might not play a redox role in this enzyme [9]. Although the complete elucidation of these points obviously requires numerous studies, a first step may consist in examining whether there is a rate-limiting step in the catalytic cycle of hydrogenases. This question arises naturally because this catalytic cycle involves necessarily several steps of very different nature, such as the diffusion and the heterolytic cleavage of the H2 molecule, proton and electron transfers, and interactions with the redox partners. As a consequence, the existence of a rate-limiting step has already been put forward by several authors [10, 11]. Some authors have modeled the reaction mechanism by assuming that the activation energy measured in the H2 uptake reaction catalyzed by the Azotobacter vinelandii enzyme with methylene blue as electron acceptor [11] corresponds to the H2 activation step which takes place at the [Ni-Fe] center [3, 4]. However, a recent study carried out with the enzyme from Chromatium vinosum has shown that the H2 uptake activity measured by electrochemical techniques is

683

much higher than that probed with dyes [12]. In the present work, we show that a quantitative analysis of the available data can yield useful information about the existence and the nature of rate-limiting steps.

Materials and methods The set of rate equations describing the kinetic behavior of the system shown in Fig. 3 was numerically integrated by using the fourth-order Runge-Kutta method, as previously described [13]. The convergence of the method was checked by varying the integration step and by verifying that the steady-state values did not depend on the initial populations.

Results Numerous data concerning the activity of [Ni-Fe] hydrogenases in the D2/H+ exchange, H2 evolution, and H2 uptake reactions have been reported in the literature. They were obtained, however, with enzymes which were either activated or non-activated, and by using various electron donors and acceptors. In order to handle a set of data which is as homogeneous as possible, we have selected in Table 1 those obtained with the following conditions: the enzyme was pre-activated prior to the activity measurements, and the electron donor/acceptor was methyl viologen at saturating concentration (1 or 2 mM). In Table 1, one unit of specific activity (1 U) is equal to 1 mol H2 exchanged, evolved, or oxidized per minute and per mg enzyme. On the basis of a molecular weight of 90,000, one unit corresponds to 1.5 H2 molecules activated per second, and to a flux of 3 electrons and 3 protons passing through each enzyme molecule per Table 1 Activities displayed by [Ni-Fe] hydrogenases. One unit of specific activity (1 U) is equal to 1 mol H2/min per mg activated enzyme. The electron donor/acceptor was methyl viologen

second in the H2 evolution and H2 uptake reactions. In the case of the hydrogenases from Thiocapsa roseopersicina, Azotobacter vinelandii, Desulfomicrobium baculatum, and Desulfovibrio gigas, the pH profiles of the activities have been studied and both the maximum activity and the optimum pH are given in Table 1. From the data quoted in Table 1, it appears that the specific activities of various [Ni-Fe] hydrogenases measured under similar conditions can vary by more than an order of magnitude. It is currently difficult to analyze these variations. Firstly, significant differences are sometimes observed between activity values reported by different authors for the same enzyme, which is probably due to the different techniques used in the experiments. Secondly, some key parameters expected to govern the activity, like the redox potentials of the various metal centers, have not yet been determined for some hydrogenases. In a first step, we have therefore considered only the sets of data available for two specific hydrogenases: the enzyme from T. roseopercisina in which isotope effects have been systematically investigated over a wide pH range, and the enzyme from D. fructosovorans for which the activities and the redox properties have been studied in two different forms, the wild type and the P238C mutant. The data obtained with other enzymes have been examined in a second step. The T. roseopersicina hydrogenase: comparison of activities displayed in a series of reactions Isotope effects have been studied quantitatively over a wide pH range in the various reactions catalyzed by T. roseopersicina hydrogenase [10]. This comprehensive at saturating concentration throughout, except in the case of the H2 uptake reaction with the D. baculatum enzyme in which the electron acceptor was benzyl viologen

Species

D2/H+ exchange

H2 evolution

H2 uptake

Ref.

Thiocapsa roseopersicina

max. 220 U, pH 5.5, 30 C, vH2/vHD=0.2±0.45a

max. 65 U, pH 4.0, 30 C

max. 55 U, pH 9.5, 30 C

[10]

Azotobacter vinelandii

max. 34 U, pH 5.0, 42 C, vHD=0

max. 32 U, pH 5.0, 42 C

±

[11, 14]

Desulfomicrobium baculatumb

max. 350 U, pH 4.0, 32 C, vH2/vHD=0.25±1.3c

max. 430 U, pH 4.0, 32 C

max. 120 U, pH 7.5, 32 C

[15]

Desulfovibrio gigas

max. 120 U, pH 8.5, 32 C, vH2/vHD=0.3±0.5d ±

max.: n.r.,f pH 4.5, 32 C

max.: n.r.,f pH 8.0, 32 C

[15, 16]

700 U , pH 5.7; 350 U , pH 7.8; 30 C

420 U, pH 7.8, 25 C

[17, 18]

Desulfovibrio fructosovorans

±

305 U, pH 8.0, 37 C

205 U, pH 8.0, 37 C

[19]

Desulfovibrio fructosovorans

WTe: 223 U; P238C: 175U; pH 5.5, 30 C

WT: 65 U; P238C: 104 U; pH 7.6, 30 C

WT: 330 U; P238C: 205 U; pH 8.5, 30 C

[9]

a

d

b

e

pH range 3±8.5 Cytoplasmic enzyme c pH range 2±8.5

pH range 5±10 Wild-type enzyme f Not reported

684

Fig. 1 Elementary kinetic steps used to analyze the catalytic cycle of Ni-Fe hydrogenases. M stands for the active site and B is a nearby base

set of data can be conveniently analyzed by using elementary kinetic steps S1, S±1, S2, and S±2 which are defined in Fig. 1. Steps S1 and S2 correspond to the diffusion and heterolytic cleavage of H2 and D2, respectively, and S±1 and S±2 to the reverse reactions. By using these elementary modules, the catalytic cycles of the six reactions catalyzed by the T. roseopersicina enzyme can be described as shown in Fig. 2. The activity measured at pH 7.2 and 30 C, expressed as the number of H2, D2, or HD molecules taken up or evolved per second, is given for each reaction. In the following, these numbers are analyzed by using the modular scheme of Fig. 2. Let us consider first reactions 3 and 4 describing H2 and D2 uptake, respectively, which display identical activities at pH 7.2 and 30 C (Fig. 2). Since the rate of HD evolution is equal to 165 s±1 in reaction 1, the rate of the S2 step is greater than this value, which implies that S2 is not rate limiting in reaction 4. From reaction 2, we learn that the rate of the S1 step is greater than 22 s±1. Actually, the rate of S1 is expected to be larger than or equal to that of S2 (Fig. 1), so that S1 is not expected to be rate limiting in reaction 3 either. It follows that reactions 3 and 4 can be rate limited either by H+/D+ transfer or by electron transfer to methyl viologen. The absence of any isotope effects strongly suggests that electron transfer to methyl viologen is rate limiting in the H2 and D2 uptake reactions catalyzed by the T. roseopersicina enzyme at pH 7.2. The fact that the H2 uptake activity of this enzyme is larger at pH 9.0 than at pH 7.2 [10] is consistent with this interpretation, since the [4Fe-4S]2+,+ centers of [Ni-Fe] hydrogenases are known to be more reducing at basic pH [20]. However, a weak isotope effect of about 20% is observed at pH 9.0, indicating that electron transfer is not completely rate limiting at this high pH value [10].

Let us focus now on reactions 5 and 6, describing H2 and D2 evolution, respectively. According to reaction 1, the rate of the S±1 step is greater than 66 s±1, so that this step is not rate limiting in reaction 5. Likewise, the data obtained in reaction 2 indicate that S±2 is not rate limiting in reaction 6. The strong isotope effect displayed in reactions 5 and 6 suggests that H+ and D+ uptake are rate limiting at pH 7.2 in the T. roseopersicina enzyme. As expected for reactions limited by proton (deuteron) uptake steps, the rates of reactions 5 and 6 are higher at acidic pH than at neutral pH [10]. The isotope effect is smaller at pH 4.1 than at pH 7.2 [10], suggesting that the proton uptake rate becomes sufficiently high at low pH values to be comparable to that of another step. Coming back to the exchange reactions 1 and 2, the strong isotope effects (Fig. 2) and the similar pH dependence to that of the H2 evolution reaction [10] both suggest that proton and deuteron uptake are rate limiting in the D2/H+ and H2/D+ exchange reactions, respectively. The D. fructosovorans hydrogenase: analysis of the activities obtained with the wild-type and P238C mutant forms The activities displayed by the wild-type and P238C mutant forms of this hydrogenase are given in Table 1, and the values of the redox potentials of the iron-sulfur centers are given in Fig. 3a. In the H2 uptake reaction with methyl viologen, the activities measured at pH 8.5 and 30 C with the wild-type and P238C mutant forms correspond to kcat (methyl viologen) values equal to 1000 s±1 and 620 s±1, respectively. The catalytic cycle therefore takes place in the millisecond time scale for this enzyme. Although conversion of the [3Fe-4S]+,0 center into a [4Fe-4S]2+,+ center was accompanied by a decrease of 315 mV of the redox potential in the P238C mutant, the two forms of the enzyme were found to exhibit similar activities in both the H2 evolution and H2 uptake reactions (Table 1 and Fig. 3a). The same effect was observed when the physiological acceptor cytochrome c3 was used in the H2 uptake reaction (Fig. 3a). Two interpretations can be proposed to explain the weak sensitivity of the activity with respect to the nature and the redox potential of the median iron-sulfur center: either this center is not involved in the intramolecular electron transfer, or it plays a redox role but the electron transfer between the proximal and distal Fe-S centers is not rate limiting in the catalytic cycle of the two forms [9]. The latter hypothesis can be tested by evaluating the efficiency of the electron transfer system made of the three Fe-S centers. With that aim, the scheme corresponding to reaction 3 in Fig. 2 was expanded by using the kinetic model shown in Fig. 3b to mimic the H2 uptake reaction. In this model, the intramolecular

685 Fig. 2 Modular description of the catalytic cycle of the six reactions catalyzed by the Ni-Fe hydrogenases. The elementary steps S1, S±1, S2, and S±2 are defined in Fig. 1. M stands for the active site, B is a nearby base, and MV stands for methyl viologen. The activities measured at pH 7.2 and 30 C with the enzyme from T. roseopersicina, expressed as the number of H2, D2, or HD molecules consumed or produced per second, are given in parentheses. They were taken from [10]. The value of the vD2/vHD ratio in reaction 2 was kindly communicated by N. Zorin

electron exchanges within the system made of the Fe-S centers and the intermolecular transfer between this system and the acceptor are considered explicitly. k is the rate constant of the electron flux delivered by the active site. Its value is determined by the rates of several steps: diffusion and heterolytic cleavage of the H2 molecule, proton release and electron transfer from the active site to the proximal Fe-S center. k9 is the rate constant characterizing the reduction of the acceptor by the distal Fe-S center. The intramolecular electron exchanges between the Fe-S centers are described by the rate constants k1, k±1, k2, and k±2, and the ratios k1/k±1 and k2/k±2 are determined by the differences between the redox potentials of the proximal and median clusters and of the median and distal clusters, respectively. Since the redox properties of the

Ni-Fe center and of the distal and proximal [4Fe-4S]2+,+ centers are not significantly altered by substituting Cys for Pro238, the values of k and k9 are expected to be similar in the wild-type form and in the P238C variant. In contrast, the driving forces governing the electron exchanges between the Fe-S centers are greatly changed by the mutation (Fig. 3a), so that the ratios k1/k±1 and k±2/k2, which are both equal to 1.1”107 in the wild-type form, are dramatically decreased to 1.8”102 in the P238C variant. The system made of the three Fe-S centers possess eight redox states. The populations of the eight redox species are related by a set of rate equations involving the rate constants defined in Fig. 3b. For a given set of rate constants and initial populations, these equations can be numerically integrated to obtain the time

686 Fig. 3 a Comparison of the redox properties and H2 uptake activities ( pH 8.5, 30 C) of the wild-type and P238C mutant forms of the Ni-Fe hydrogenase from D. fructosovorans. The data were taken from [9]. b Kinetic model used to evaluate the efficiency of the electron transfer system made of the Fe-S centers in Ni-Fe hydrogenases. Electron transfers which are favored by large driving forces are indicated by bold arrows

Fig. 4a,b Time course of the populations of selected redox species involved in the kinetic model of Fig. 3b. Integration of the rate equations was carried out as explained in Materials and methods, by using the following values of the rate constants: k=k9=2000 s±1, k1=k±2=6”107 s±1, k±1=k2=3”105 s±1. These numbers were chosen so as to give ratios k1/k±1 and k±2/k2 equal to those deduced from the redox potentials measured in the P238C mutant form. The value of k±1 and k2 is that evaluated in the

text for the mutant form. The calculations were done by assuming that the Fe-S centers are either a all oxidized or b all reduced at t=0. The following symbols are used for the various species: (±±) POX, MOX, DOX; (± ±) PRED, MRED, DRED; (. . .) POX, MRED, DOX; (± . ±) POX, MRED, DRED; (- - - - -) PRED, MOX, DOX. Dred is the sum of the populations of redox species in which center D is reduced

687

course of the various populations. An example is given in Fig. 4: after a transient regime, the various populations ultimately reach steady-state levels which depend on the rate constants values but not on the initial populations. In the model shown in Fig. 3b, reduction of the acceptor is assumed to take place only via the distal Fe-S center so that the rate of reduction of the acceptor can be written: d‰Ared Š=dt ˆ kcat ‰hydrogenaseŠ

…1†

with kcat ˆ k0 ‰Dred Š, where [Dred] is the sum of the steady-state populations of redox species in which center D is reduced. In the most general case, the expression giving [Dred] in terms of the rate constants defined in Fig. 3a is quite complicated. This expression greatly simplifies, however, when some rate constants are much larger than the others. This situation is very likely met in the case of D. fructosovorans hydrogenase: as mentioned above, the redox potential values are such that k1 and k±2 are much larger than k±1 and k2 in both the wild-type and P238C mutant forms of the enzyme. Besides, the intramolecular electron transfers corresponding to k1 and k±2, which are favored by large driving forces, are expected to be much faster than both the intermolecular transfer corresponding to k9 and the rate constant k which is determined by several steps (see above). In these circumstances, the quantity kcat given by Eq. 1 no longer depends on the ªfastº rate constants k1 and k±2, and is simply given by: …kcat †

1

ˆ …k†

1

‡ …k0 †

1

‡ …k 1 †

1

‡ …k2 †

1

…2†

It is useful to write Eq. 2 in the form: kcat ˆ

kMAX cat 1 ‡ kMAX cat =k12

…3†

with kMAX ˆ cat

kk0 k 1 k2 ; k12 ˆ k ‡ k0 k 1 ‡ k2

…4†

As noted previously, substituting Cys for Pro238 is expected to alter slightly the values of k and k9, so MAX are expected in the two that similar values of kcat forms of the enzyme. In contrast, the values of k±1 and k2 are certainly greatly modified by the conversion of the [3Fe-4S]+,0 cluster into a [4Fe-4S]2+,+ cluster. Equation 3 shows that similar values of kcat can nevertheless be observed in the two forms provided k12 be MAX . On the basis of the kcat values much larger than kcat measured with D. fructosovorans hydrogenase and methyl viologen, this requires that k12 is larger than about 600±1000 s±1 in both the wild-type and the P238C mutant forms. The plausibility of this requirement can be evaluated by calculating k±1 and k2 on the basis of the high-temperature limit of the expression given by electron transfer theories [21]:

kET ˆ 2p=h…Tab †2 …4pkB T†

1=2

exp… …
G ‡ l†2 =4lkB T † …5†

where Tab is the electronic factor, l the reorganization energy, and DG the driving force. By using for Tab an empirical expression which accounts for the rate constants measured in photosynthetic reaction centers and for l an arbitrary value of 0.7 eV, k2 was estimated at 5”104 s±1 in the wild-type form of D. fructosovorans hydrogenase [22]. We have carried out a more realistic evaluation by using the rate constant of 5”106 s±1, which has been measured in the case of the electron exchange between two [4Fe-4S]2+,+ clusters in clostridial ferredoxins [23]. In these proteins, the center-tocenter distance between the Fe-S centers is equal to 12 Š as in [Ni-Fe] hydrogenases, and the best hypothetical electron transfer pathway connecting the Fe-S centers is very similar to that proposed for [Ni-Fe] hydrogenases [8, 23], which suggests similar electronic factors. However, the redox potentials of the two Fe-S clusters are equal in clostridial ferredoxins [23], so that the driving force DG governing the transfer equals zero. In contrast, the electron exchanges corresponding to k±1 and k2 are characterized by unfavorable DG values equal to +0.405 eV and +0.130 eV in the wild-type and P238C mutant forms of the enzyme, respectively (Fig. 3a). The rate constant expected in the enzyme can be evaluated by assuming that the values of Tab and l are identical in the enzyme and in the ferredoxin. The result depends on the reorganization energy l of the system, which is presently unknown. To our knowledge, reorganization energies characterizing electron transfers involving [4Fe-4S]2+,+ centers have been reported only for the enzyme trimethylamine dehydrogenase, for which l values in the 1.4±2.4 eV range have been proposed [24, 25]. Considering l values in the 0.5±2 eV range, which includes most of the available data obtained in metalloproteins [26], yields k±1 and k2 values in the narrow range 2.8±3.3”105 s±1 in the P238 mutant form. These numbers are much larger than the experimental value kcat=600 s±1, which demonstrates that a fast, non-limiting electron transfer can take place between the proximal and distal [4Fe-4S] centers in the mutant enzyme, despite the unfavorable driving force of +0.13 eV. In contrast, in the case of the wild-type protein, the very unfavorable driving force of +0.405 eV leads to much smaller values of k±1 and k2, in the 55±650 s±1 range. According to this estimate, the efficiency of the electron transfer system made of the Fe-S centers would not be great enough to ensure a non-limiting step in the H2 uptake reaction with the wild-type protein. The activities measured in H2 evolution experiments (Table 1) can be analyzed by using a similar kinetic model to that shown in Fig. 3b in which, however, the electron transfers corresponding to k and k9 are replaced by the reverse reactions. Since the electron transfer system made of the three Fe-S centers is thermodynamically symmetric, this model leads to

688

expressions equivalent to Eqs. 3 and 4. The values of k±1 and k2 evaluated above should therefore be compared to kcat=200 s±1 and 300 s±1, corresponding to the H2 evolution activities measured at pH 7.6 and 30 C in the wild-type and mutant forms, respectively (Table 1). As found in the case of the H2 uptake experiment, k±1 and k2 are large enough to ensure kcat"kcat MAX in the mutant form. The case of the wild-type protein is less clear-cut, although the highest limit of 650 s±1 would be sufficient to ensure that electron transfer between the Fe-S centers is not rate limiting in the H2 evolution reaction, despite the highly unfavorable driving force of the transfer between the [3Fe-4S] center and the [4Fe-4S] centers. Actually, the preceding conclusions rest on several hypotheses which are not necessarily valid. Firstly, the evaluation of k±1 and k2 was based on kinetic data obtained in clostridial ferredoxins. This could be an inappropriate reference if the electronic factors and the reorganization energies were found to differ significantly in the ferredoxin and in the hydrogenase. For example, the kcat values calculated for the wild-type and the mutant proteins in the H2 uptake reaction would differ by only 30% if the electronic factor were three-fold larger in the enzyme than in the ferredoxin and if the reorganization energy were equal to 2 eV. Secondly, the driving force of the electron transfers shown in Fig. 3b could differ from the DG values deduced directly from the experimental redox potentials if the redox centers were coupled by significant redox interactions. It should be noted that the preceding analysis was based on the assumption that the redox potential of each Fe-S cluster is not dependent on the redox state of the other clusters. Actually, the complete description of the eight microscopic redox states involves nine redox parameters which can be reduced to seven independent parameters, namely the three microscopic redox potentials eP, eM, and eD of the proximal, median, and distal Fe-S clusters, respectively, and four interaction potentials (Fig. 5). In the case of D. fructosovorans wild-type hydrogenase, the results of the titration performed at redox equilibrium [9] show that eM=+65 mV is higher than eP and eD, and that (eP+iPM) and (eD+iMD) are equal to ±340 mV. The electron exchange between the proximal and the distal clusters can be considered either through the three microscopic one-electron reduced states, or through the three microscopic two-electron reduced states (Fig. 5). In the first case, the ratio of the rate constants used in the model of Fig. 3b involves only the three redox potentials eP, eM, and eD. For instance, the ratio k2/k±2 is given by exp(F(eD±eM)/RT), which is not favorable to the electron transfer, as shown above. In the second case, the electron transfer involves the two following steps: k01

k02

k

k

Pred Dred „ Mred Dred Pred Mred „ 0 0 1

2

…6†

and the ratio of the rate constants are strongly dependent on the interaction potentials values: k91/k9±1=exp(F(eD+iPD±eM±iPM)/RT) and k92/k9±2=exp (F(eM+iMD±eP±iPD)/RT). These expressions show that the electron exchange between the proximal and the distal centers is favored by anticooperative redox interactions between neighboring clusters (iPM and iMD<0) and by a cooperative redox interaction between distant clusters (iPD>0). For instance, considering that the system is symmetric (iMD=iPM) and taking iPM=±60 mV and iPD=+60 mV, which are similar values to those measured in some multicenter Fe-S proteins [27, 28], leads to eP=eD=±280 mV and to an increase of the k91/k9±1 and k9±2/k2 ratios by more than three orders of magnitude, changing from 0.92”10±7 in the absence of redox interactions to 1.2”10±4. This shows that even moderate redox interactions, which cannot be spectroscopically detected in the course of redox titrations performed at equilibrium, can strongly affect the kinetics of intramolecular electron exchange. From the preceding discussion, it appears that despite the apparently unfavorable DG values deduced from potentiometric experiments, the electron transfer between the [3Fe-4S] center and the [4Fe-4S] centers may be fast enough to be non-limiting in both the H2 uptake and H2 evolution reactions. In these circumMAX (Eq. 3), stances the activity would be given by kcat so that the steps corresponding to either k or k9 (or to the reverse reactions in the H2 evolution experiment) could be rate limiting (Fig. 3b). Since k depends on the diffusion and heterolytic cleavage of H2 as well as on proton transfers steps, it could be sensitive to isotope effects. To our knowledge, such effects have not yet been studied in the case of D. fructosovorans

Fig. 5 Schematic representation of the microscopc redox states in a system containing three redox centers. The reduced proximal, median, and distal clusters are labeled P, M, D, respectively. Bold arrows correspond to transitions involving experimentally detectable redox species

689

hydrogenase. Since k9 is the rate constant characterizing the reduction of the electron acceptor by the distal [4Fe-4S] center, it depends on the nature of the acceptor. The fact that the H2 uptake activity measured with a saturating concentration of acceptor increases four-fold when methyl viologen is replaced by cytochrome c3 (Fig. 3a) and six-fold when it is replaced by methylene blue [19] strongly suggests that reduction of the acceptor is rate limiting in this reaction. This conclusion is consistent with the analysis carried out in the preceding section, which indicates that electron transfer is rate limiting in the H2 uptake experiment carried out with the T. roseopersicina enzyme.

Discussion and conclusion We first consider the problem raised by the efficiency of the electron transfer system made of the three Fe-S centers in [Ni-Fe] hydrogenases. From a kinetic point of view, this system is quasi-symmetric. As a consequence, if electron transfer through this system were rate limiting in both the H2 uptake and H2 evolution reactions, similar activities would be measured in both experiments. The inverse pH profiles of these activities (Table 1) show that there are actually rate limited by different kinetic process. Therefore, if electron transfer through this system takes place during the catalytic cycle, it cannot be rate limiting in both reactions. The data obtained by substituting Cys for Pro238 in D. fructosovorans hydrogenase bring more detailed information: comparing the redox properties and the H2 uptake and H2 evolution activities of the wild-type and P238C forms shows that electron transfer through this system is not rate limiting in either reaction. This requires that, in this system, the smallest rate constant is greater than 600±1000 s±1 in both forms of the enzyme. From a quantitative analysis based on data obtained in clostridial ferredoxins and on the existence of moderate redox interactions, it was concluded that this condition could be fulfilled despite the apparently unfavorable redox potentials measured in the wild-type enzyme. Therefore, the [3Fe-4S] center cannot be currently excluded from the electron-transfer chain connecting the active site to electron donors and acceptors on the sole basis of its high redox potential. An experiment aimed at better understanding the role of the [3Fe-4S]+,0 center in [NiFe] hydrogenases was conducted in the enzyme from Azotobacter vinelandii, in which two cysteine residues expected to coordinate the [3Fe-4S] center were individually replaced by serines [14]. By comparison with the wild-type form, the H2 uptake and H2 evolution activities measured in whole cells were found to decrease in the mutants by three orders of magnitude and one order of magnitude, respectively [14]. Although these results suggest that the [3Fe-4S] center plays an important role in this enzyme, the fact that

the metal centers were characterized neither in the wild-type nor in the mutant forms of this protein precludes a more detailed interpretation. Analyzing the set of data available in the case of the T. roseopersicina hydrogenase by using the modular scheme of Fig. 2 brings useful information about the catalytic cycle of this enzyme. Firstly, this analysis shows that steps S1 and S±1, corresponding to the diffusion and heterolytic cleavage of H2 (Fig. 1), are rate limiting neither in the H2 uptake nor in the H2 evolution reaction at pH 7.2 and 30 C (Fig. 2). Recent experiments carried out with a crystal of D. fructosovorans hydrogenase have revealed the existence of hydrophobic channels connecting the active site to the protein surface [29]. These channels, which are largely conserved in all [Ni-Fe] hydrogenases, could play the role of H2 reservoirs, ensuring fast H2 diffusion. Secondly, the data obtained with the T. roseopersicina enzyme strongly suggest that, at pH 7.2 and 30 C, proton uptake and electron transfer are rate limiting in the H2 evolution and H2 uptake reactions, respectively. It is interesting to examine whether the data available for other [Ni-Fe] hydrogenases are consistent with this interpretation. We first observe that the pH profile of the various activities is similar in all [Ni-Fe] hydrogenases: the H2 evolution activity reaches a maximum at pH~4±5 and the H2 uptake activity is maximum at pH~7.5±9.5 (Table 1), suggesting common rate-limiting steps in all these enzymes. In the case of the Azotobacter vinelandii hydrogenase, the rates of D2/H+ exchange and H2 evolution with methyl viologen were found to be equal and to be characterized by the same activation energies over a large pH range [11]. These findings, which were observed in the membrane-bound as well as in the purified enzyme, indicate that the two reactions share the same rate-limiting step. According to Fig. 2, this step could be either the proton uptake or the S±1 step. Since the rate of S±1 is not expected to depend strongly on pH, the most likely candidate is the proton uptake step. We have already mentioned that, in this enzyme, the H2 uptake activity was much more altered than the H2 evolution activity when Cys residues expected to coordinate the [3Fe-4S] center were replaced by Ser. Similar effects were observed when the target of the Cys-to-Ser substitution was the proximal [4Fe-4S] center. In this case, the specific activities in the H2 uptake and H2 evolution reactions were found to decrease in the mutant by factors equal to 45 and 4.5, respectively [14]. All these observations are consistent with the former reaction being more sensitive to the presence of the Fe-S centers. Unfortunately, the effect of these mutations on the insertion and the redox properties of these centers was not further studied. Although Desulfomicrobium baculatum hydrogenase was initially considered as containing only two [4Fe-4S]2+,+ clusters [16], recent experiments have shown that this protein actually accomodates three [4Fe-4S]2+,+ clusters [30], giving an arrangement of

690

metal centers very similar to that found in the P238C mutant of the D. fructosovorans enzyme [9]. As in the case of the A. vinelandii enzyme, the rate of H2 evolution with methyl viologen was found to match closely that of D2/H+ exchange over a wide pH range [15], suggesting a common rate-limiting step in both reactions. A peculiarity of this enzyme is that the vH2/vHD ratio exceeds unity in the D2/H+ exchange reaction at pH greater than 5 [15]. A more detailed analysis, which could also illuminate the role of the selenocysteine present in D. baculatum, must await the determination of the redox potentials of the three Fe-S centers. We now consider the D. gigas enzyme, which has been studied by several groups. In this enzyme, the pH profiles of the H2 evolution and H2 uptake activities are similar to those observed in other [NiFe] enzymes, but the maximum D2/H+ exchange activity occurs at pH 8.5 whereas it occurs in the pH range 4±5.5 in other [Ni-Fe] enzymes (Table 1). This peculiar pH dependence suggests that proton uptake is not rate limiting in the D2/H+ exchange reaction with this enzyme. The H2 evolution and H2 uptake activities of the D. gigas enzyme were shown to depend on the redox state of the enzyme, and their variations could be fitted to Nernst's equation with n=1 and Em (pH 7) values equal to ±350 mV [17] and ±340 mV [31], respectively. Unfortunately, these values correspond approximately to the redox potential of the Ni-C/Ni-SI couple [20] as well as to that of the [4Fe-4S]2+,+ clusters [20, 32], so that it cannot help identifying a rate-limiting step. A clue about the importance of electron transfer in the H2 uptake reaction may come from experiments in which the Fe-S centers of the D. gigas enzyme were destroyed after incubation with Cu or Hg salts [33]: the decrease of the H2 uptake activity was found to be about four-fold larger than that of tritium/proton exchange [33], suggesting that electron transfer might be rate limiting in the former reaction. Further insight into the nature of the rate-limiting step can be gained by noting that, as already mentioned in the case of the D. fructosovorans enzyme, the H2 uptake activity of D. gigas hydrogenase is strongly dependent on the nature of the electron acceptor: activity values measured at pH 7.0 with saturating concentrations of methyl viologen and benzyl viologen were found to be equal to 80 U (at 30 C) and 1200 U (at 25 C), respectively [31, 33]. Likewise, H2 uptake activities measured at pH 7.6 and 25 C with saturating concentrations of methyl viologen and D. gigas cytochrome c3 were found to be equal to 420 U and 670 U, respectively [18, 34]. All these findings are consistent with the reduction of the acceptor being rate limiting in the H2 uptake reaction catalyzed by Ni-Fe hydrogenase. A similar conclusion was deduced from an electrochemical study carried out with the enzyme from Chromatium vinosum [12]. It is interesting to note that electron transfer to the acceptor has also been proposed to be rate limiting for Feonly hydrogenases [35].

The data available for [Ni-Fe] hydrogenases seem to be consistent with the general scheme deduced from the data concerning the T. roseopersicina enzyme. A more detailed analysis should await the full characterization of the metal centers present in some of these enzymes, as well as results given by complementary methods, such as systematic studies of isotope effects, new site-directed mutagenesis experiments, and theoretical modeling. Acknowledgements We thank Drs. C. Hatchikian and M. Rousset for helpful discussions.

References 11. Adams MWW (1990) Biochim Biophys Acta 1020:115±145 12. Albracht SPJ (1994) Biochim Biophys Acta 1188:167±204 13. Pavlov M, Siegbahn PEM, Blomberg MRA, Crabtree RH (1998) J Am Chem Soc 120:548±555 14. Pavlov M, Blomberg MRA, Siegbahn PEM (1999) Int J Quantum Chem 73:197±207 15. Niu S, Thomson LM, Hall MB (1999) J Am Chem Soc 121:4000±4007 16. De Gioia L, Fantucci P, Guigliarelli B, Bertrand P (1999) Int J Quantum Chem 73:187±195 17. De Gioia L, Fantucci P, Guigliarelli B, Bertrand P (1999) Inorg Chem 38:2658±2662 18. Volbeda A, Charon M-H, Piras C, Hatchikian EC, Frey M, Fontecilla-Camps JC (1995) Nature 373:580±587 19. Rousset M, Montet Y, Guigliarelli B, Forget N, Asso M, Bertrand P, Fontecilla-Camps JC, Hatchikian EC (1998) Proc Natl Acad Sci USA 95:11625±11630 10. Zorin NA, Dimon B, Gagnon J, Gaillard J, Carrier P, Vignais PM (1996) Eur J Biochem 241:675±681 11. McTavish H, Sayavedra-Soto LA, Arp DJ (1996) Biochim Biophys Acta 129:183±190 12. Pershad HR, Duff JLC, Heering HA, Duin EC, Albracht SPJ, Armstrong FA (1999) Biochemistry 38:8992±8999 13. Bertrand P, Asso M, Mbarki O, Camensuli P, More C, Guigliarelli B (1994) Biochimie 76:524±536 14. McTavish H, Sayavedra-Soto LA, Arp DJ (1995) J Bacteriol 177:3960±3964 15. Lespinat PA, Berlier Y, Fauque G, Czechowski M, Dimon B, Le Gall J (1986) Biochimie 68:55±61 16. Teixera M, Fauque G, Moura I, Lespinat PA, Berlier Y, Prickril B, Peck HD, Xavier A, Le Gall J, Moura JJG (1987) Eur J Biochem 167:47±58 17. Fernandez VM, Aguirre R, Hatchikian EC (1984) Biochim Biophys Acta 790:1±7 18. Fernandez VM, Rua ML, Reyes P, Cammack R, Hatchikian EC (1989) Eur J Biochem 185:449±454 19. Hatchikian EC, Traore AS, Fernandez VM, Cammack R (1990) Eur J Biochem 187:635-643 20. Cammack R, Patil DS, Hatchikian EC, Fernandez VM (1987) Biochim Biophys Acta 912:98±109 21. Marcus RA, Sutin N (1985) Biochim Biophys Acta 811:265±322 22. Page CC, Moser CC, Chen X, Dutton PL (1999) Nature 402:47±52 23. Kyritsis P, Huber JG, Quinkal I, Gaillard J, Moulis J-M (1997) Biochemistry 36:7839±7846 24. Falzon L, Davidson VL (1996) Biochemistry 35:12111±12118 25. Wilson EK, Huang L, Sutcliffe MJ, Mathews FS, Hille R, Scrutton NS (1997) Biochemistry 36:41±48 26. Bertrand P (1991) Struct Bonding 75:1±48 27. Guigliarelli B, Asso M, More C, Augier V, Blasco F, Pommier J, Giordano G, Bertrand P (1992) Eur J Biochem 207:61±68

691 28. Cavazza C, Guigliarelli B, Bertrand P, Bruschi M (1995) FEMS Microbiol Lett 130:193±200 29. Montet Y, Amara P, Volbeda A, Vernede X, Hatchikian EC, Field MJ, Frey M, Fontecilla-Camps JC (1997) Nat Struct Biol 4:523±526 30. Garcin E, Vernede X, Hatchikian EC, Volbeda A, Frey M, Fontecilla-Camps JC (1999) Structure 7:557±566 31. Lissolo T, Pulvin S, Thomas D (1984) J Biol Chem 259:11725±11729

32. Texeira M, Moura I, Xavier AV, Moura JJG, LeGall J, DerVartanian DV, Peck HD, Huynh BH (1989) J Biol Chem 264:16435±16450 33. Fernandez VM, Hatchikian EC, Cammack R (1985) Biochim Biophys Acta 832:69±79 34. Nivi›re V, Hatchikian EC, Bianco P, Haladjian J (1988) Biochim Biophys Acta 935:34-40 35. Verhagen MFJM, Wolbert RBG, Hagen WR ( 1994) Eur J Biochem 221:821±829