15478
J. Phys. Chem. B 2008, 112, 15478–15486
Major Mo(V) EPR Signature of Rhodobacter sphaeroides Periplasmic Nitrate Reductase Arising from a Dead-End Species That Activates upon Reduction. Relation to Other Molybdoenzymes from the DMSO Reductase Family Vincent Fourmond,†,‡ Be´ne´dicte Burlat,†,‡ Se´bastien Dementin,†,‡ Pascal Arnoux,‡,§,| Monique Sabaty,‡,§,| Se´verine Boiry,‡,§,| Bruno Guigliarelli,†,‡ Patrick Bertrand,†,‡ David Pignol,‡,§,| and Christophe Le´ger*,†,‡ Unite´ de Bioe´nerge´tique et Inge´nierie des Prote´ines, IBSM, UPR 9036, CNRS, 31 Chemin Joseph Aiguier, F-13402 Marseille Cedex 20, France, Aix-Marseille UniVersite´, 3 Place Victor Hugo, F-13333 Marseilles Cedex 3, France, Laboratoire de Bioe´nerge´tique Cellulaire, SBVME, IBEB, CEA, F-13108 Saint-Paul-lez-Durance, France, and Laboratoire de Biologie Ve´ge´tale et Microbiologie EnVironnementales, UMR 6191, CNRS, F-13108 Saint-Paul-lez-Durance, France ReceiVed: August 8, 2008; ReVised Manuscript ReceiVed: September 17, 2008
Enzymes of the DMSO reductase family use a mononuclear Mo-bis(molybdopterin) cofactor (MoCo) to catalyze a variety of oxo-transfer reactions. Much functional information on nitrate reductase, one of the most studied members of this family, has been gained from EPR spectroscopy, but this technique is not always conclusive because the signature of the MoCo is heterogeneous, and which signals correspond to active species is still unsure. We used site-directed mutagenesis, EPR and protein film voltammetry to demonstrate that the MoCo in R. sphaeroides periplasmic nitrate reductase (NapAB) is subject to an irreversible reductive activation process whose kinetics we precisely define. This activation quantitatively correlates with the disappearance of the so-called “Mo(V) high-g” EPR signal, but this reductive process is too slow to be part of the normal catalytic cycle. Therefore, in NapAB, this most intense and most commonly observed signature of the MoCo arises from a dead-end, inactive state that gives a catalytically competent species only after reduction. This activation proceeds, even without substrate, according to a reduction followed by an irreversible nonredox step, both of which are pH independent. An apparently similar process occurs in other nitrate reductases (both assimilatory and membrane bound), and this also recalls the redox cycling procedure, which activates periplasmic DMSO reductases and simplifies their spectroscopic signatures. Hence we propose that heterogeneity at the active site and reductive activation are common properties of enzymes from the DMSO reductase family. Regarding NapAB, the fact that we could detect no Mo EPR signal upon reoxidizing the fully reduced enzyme suggests that the catalytically active form of the Mo(V) is thermodynamically unstable, as is the case for other enzymes of the DMSO reductase family. Our original approach, which combines spectroscopy and protein film voltammetry, proves useful for discriminating the forms of the active site on the basis of their catalytic properties. This could be applied to other enzymes for which the question arises as to the catalytic relevance of certain long-lived, spectroscopically characterized species. Introduction Mononuclear molybdoenzymes form a large class of oxidoreductases that are involved in many essential redox processes, ranging from bacterial respiration to the biosynthesis of hormones in higher eukaryotes.1,2 The subclass called “DMSO reductase family” collects bacterial enzymes that house a Mobis(molybdopterin) cofactor (MoCo). Their oligomeric structures and cofactor contents vary greatly, from having the active site Mo as the only cofactor to having several hemes and iron-sulfur clusters in addition to the MoCo. They may specifically catalyze the reduction of nitrate, DMSO, chlorate or selenate, or the oxidation of arsenite, formate or nitrite, often in an oxo-transfer reaction coupled to the transfer of 2 electrons and 2 protons. * Corresponding author. E-mail:
[email protected]. Tel: +33 491164529. Fax: +33 491164578. † Unite ´ de Bioe´nerge´tique et Inge´nierie des Proteines, CNRS. ‡ Aix-Marseille Universite ´. § Laboratoire de Bioe ´ nerge´tique Cellulaire, CEA. | Laboratoire de Biologie Vege ´ tale et Microbiologie Environementales, CNRS.
Some of these substrates are toxic anions whose abundance in environments impacted by humans is an important issue of public health. These enzymes also catalyze important reactions in the global carbon, sulfur and nitrogen cycles. The monomeric DMSO reductases (DMSOR) from Rhodobacter sphaeroides and R. capsulatus are among the simplest enzymes in the DMSO reductase family. Being soluble, reasonably small, and housing the MoCo as the only chromophore, they are well suited for studies based on spectroscopic techniques (EPR, UV-vis, EXAFS, Raman) and crystallography.3-7 These investigations originally produced a surprisingly complex picture of the active site, whose structure could not be unambiguously defined in terms of number of oxo ligands and pterin coordination. Certain controversies were resolved in the early 2000s, when it became clear that the active site is actually heterogeneous in its as-prepared state, whereas its Raman and UV-vis signatures are much simpler once the sample has been “redox cycled”.3,4,6 This operation consists of reducing the enzyme by dithionite and reoxidizing it by either air or DMSO. Only Bray and co-workers established a link between redox
10.1021/jp807092y CCC: $40.75 2008 American Chemical Society Published on Web 11/11/2008
Mo(V) EPR Signature of Rhodobacter sphaeroides
J. Phys. Chem. B, Vol. 112, No. 48, 2008 15479
TABLE 1: EPR Signatures of Periplasmic Nitrate Reductases source
name
g1
A. Vinelandii NAS
very high-ga similar to high-g nitrateb
2.023 1.998
R. sphaeroides NapAB R. sphaeroides NapA
high-g (split)c high-g (split)c
E. coli NapA and NapAB
high-g (split)d
high-g split (high-g resting)e P. pantotrophus (formerly P. denitrificans and T. pantotropha) high-g nitratef NapAB low-g [split]h low-g [unsplit]i very high-gj high-g [azide]k high-g [thiocyanate]l
g2 1.998 1.989
g3
spin/mol
0.83 0.53
10% 7%
1.9987 1.9907 1.9810 1.9901 1.9984 1.9907 1.9810 1.9900
0.018 0.018
0.45 0.44
10-20% 60%
1.9971 1.9883 1.9811 1.9888
0.016
0.55
70-90%
1.9990 1.9988 1.9957 1.9969 2.0222 2.0074 2.008
0.018 0.0168 0.035 0.037 0.029 0.020 0.023
0.47 0.57 0.77 0.93 0.80 0.66 0.74
2.5-10%14 n.d.g 10-20% n.d. 40% n.d. 10%
1.9810 1.9820 1.9607 1.9594 1.9935 1.9879 1.985
2.004 1.989
anisotropy rhombicity g1-g3 (g1-g2)/(g1-g3) 0.030 0.017
1.9905 1.9892 1.9688 1.9622 1.9993 1.9940 1.991
1.993 1.981
gaV
1.9902 1.9903 1.9751 1.973 2.005 1.9961 1.994
Synechococcus NarB (NAS)
very high-gm high-g (split)n
2.0228 1.9983 1.9930 2.0047 1.9970 1.9902 1.9820 1.9897
0.030 0.015
0.82 0.45
10%-20% up to 100%
D. desulfuricans NapA
restingo low potentialp high potentialq nitrater turnovers cyanidet
n.d. 2.016 2.019 2.000 1.999 2.024
n.d. 0.052 0.059 0.018 0.017 0.029
n.d. 0.56 0.53 0.56 0.53 0.79
2), showing that a significant fraction of the spin density is localized on the sulfur ligands.2 This state is also said to be inactive.2,21 The most common and most intense Mo(V) signature is called “highg” and comes in various flavors. The “high-g resting” signal, also called “high-g split” or simply “ high-g”, is detected in intact cells and as-prepared NapAB samples of R. sphaeroides and P. pantotrophus,10,23 and after reduction of S. elongatus NarB.21 It has gav approximately equal to 1.99, it is rhombic [(g1 - g2)/(g1 - g3 ) 0.45], anisotropic (g1 - g3 ) 0.018), and split by the hyperfine interaction with two nonexchangeable protons. Depending on the enzyme and on the purification batch, it can account for 2.5-100% of the total Mo (Table 1). Yet,
15480 J. Phys. Chem. B, Vol. 112, No. 48, 2008 Richardson et al. first suggested14,20 that this species does not belong to the catalytic cycle because it is insensitive to a change in pH or to the addition of exogenous ligands (nitrate, nitrite, cyanide, azide and thiocyanate). Instead, under conditions of turnover with dithionite as reductant, they trapped the so-called “high-g nitrate” species (which has slightly lower anisotropy, 0.016, and higher rhombicity, 0.57) and proposed that this is the catalytically competent species, which reverts to the inactive, high-g form in the absence of substrate and reductant. However, subsequent time-resolved EPR experiments carried out under conditions of turnover with reduced methyl viologen (MV) could only detect the high-g species, and upon reconsideration the same authors suggested that this species, rather than the high-g nitrate species, is a long-lived catalytic intermediate.24 Despite its name, the “nitrate” signal is probably not from a nitratebound species, because it is observed in E. coli NapA even in the absence of substrate.18 In their studies of D. desulfuricans Nap, Moura et al. initially suggested that the high-g nitrate form arises from an active species25 until they observed that it is redox-inactive in potentiometric titrations22 (the corresponding species in E. coli NapA is not reduced by dithionite either).18 Instead, they isolated a high-g type species which was termed “high-g turnover” (gav 1.99, rhombicity 0.53, anisotropy 0.017) under catalytic conditions with MV as reductant.22 Later, they showed that this signal is the same regardless of whether 14Nor 15N-nitrate is used, suggesting that it is not a Mo-nitrate adduct either.13 Recent crystallographic data suggest that the Mo in D. desulfuricans Nap is coordinated by a nonproteic sulfur atom (rather than by an oxygen), but it is unknown if this coordination is that of an active species.13 Although most functional information has been tentatively gained from EPR, the relation between spectral signature and coordination has remained elusive and there is no definite proof that any of the observed Mo(V) signatures comes from a species that is actually involved in turnover. Yet the catalytic cycle has always been described in relation to either of the high-g signals (including in our own work26). Here, we study R. sphaeroides NapAB and we characterize in details the kinetics of a reductive activation processes that was first observed by Butt and co-workers.27 We show that it is slow, irreversible and correlates with the disappearance of the form of the MoCo that gives rise to the high-g signal. Therefore, in NapAB, this signature is that of a dead-end, inactive state, which activates under reducing conditions in a process that resembles the redox-cycling of periplasmic DMSO reductases. These results clarify the links between redox-cycling, active site heterogeneities, spectral signatures and changes in activity in a multicentered enzyme from the DMSO reductase family. Results Irreversible Reductive Activation. We use protein film voltammetry (PFV)26-32 to measure the activity of R. sphaeroides NapAB. In this technique, the enzyme is adsorbed onto an electrode whose potential (E) sets the thermodynamic driving force for the redox reactions, and the electrons required to reduce nitrate are transferred directly from the electrode to the enzyme. Hence, a negative current flows, whose absolute value is proportional to turnover rate. Even a fast change in activity, which may result from changing the electrode potential or from the enzyme activating, can be detected as a change in current. The electrode is always spun at high rate to avoid any limitation by mass transport.
Fourmond et al.
Figure 1. Steady state voltammograms for NapAB adsorbed at a graphite electrode (plain line). The dotted line shows a blank recorded with a bare electrode. [NO3-] ) 80 µM (≈Km), scan rate ν ) 20 mV/ s, electrode rotation rate ω ) 5 krpm, pH 6, 25 °C.
The typical voltammetric response of NapAB (Figure 1) indicates that the most reducing conditions do not produce the greatest turnover rates. Instead, the enzyme is most active in a certain potential window, and reversibly “switches off” at low potential. This is common to all enzymes from the DMSO reductase family for which PFV data are available, except arsenite oxidases.32 Explanations for this have already been offered and reviewed.26,28-31 The voltammogram in Figure 1 is representative of the steady state response obtained once the electrode potential has been repeatedly swept up and down. In preliminary experiments, we noticed small but systematic differences between the response of a newly formed film of enzyme on the first sweep to low potential and the response observed on all subsequent sweeps (Figure S1, Supporting Information). This change in activity is most easily detected and studied in chronoamperometry experiments, where the electrode potential is held at a certain value and the relaxation toward steady state after a potential step is detected as a current transient, as illustrated in Figure 2. The chronoamperogram a (black trace in Figure 2) shows the response of a fresh film of WT NapAB immerged into a solution of 1 mM nitrate (that is about 10 times Km) when the potential is stepped as indicated in the upper panel. When the potential is poised at E ) -160 mV (at t ) 40 s), the current is essentially constant (the small decay is attributed to film loss). On the step to -460 mV, at t ) 80 s, the activity first instantly decreases (the current becomes less negative) and then slowly increases before it stabilizes. This slow change in current demonstrates that the enzyme activates at low potential. We observed that the activation always proceeded with first-order kinetics (see the “methods” section) and the magnitude of the activation phase accounted for 20 ( 5% of the current reached after activation and was independent of E (this is the average of about 50 different runs). The instant decrease in activity when the potential is first stepped from -160 to -460 mV is not surprising considering the steady state profile in Figure 1. When this sequence of potential steps is repeated with the same film of enzyme (from t ) 480 s, trace b), the activity detected at E ) -160 mV is greater than at the same potential in the first experiment, as indicated by the downward arrow, and no further activation occurs on the second step to -460 mV, at t ) 560 s. Therefore, the activation proceeds only once, on the first step to -460 mV, and it is not reversed by taking back the enzyme to oxidizing conditions (+240 mV), hence its qualification as irreversible. The fact that the current at the end of trace b is slightly smaller than that reached at t ) 480 s reveals that some film desorption has occurred, but this does not question the above conclusions. Trace c was recorded with a fresh film dipped into a solution of 10 mM dithionite (in the absence of nitrate) for 10 min at
Mo(V) EPR Signature of Rhodobacter sphaeroides
J. Phys. Chem. B, Vol. 112, No. 48, 2008 15481
Figure 3. Activation rate constants ka relative to the irreversible activations of NapA and NapAB as a function of electrode potential, determined by fitting data such as those in Figure 2 (see the Methods). The lines are best fits to eq 1 with Ea0 in the range -240 to -270 mV.
SCHEME 1: Mechanism of the Activation Process
Figure 2. Chronoamperometric experiments demonstrating the irreversible reductive activation of NapAB. The top panel shows the sequence of potential steps which was applied to films of as-prepared WT NapAB (traces a and b), dithionite reduced NapAB (trace c), asprepared R392A NapAB (trace d) and NapA (trace e; in that case, film desorption is more pronounced). Vertical lines indicate the current scales. Horizontal dotted lines are for i ) 0. We replot these experiments in Figures 7, S2 and S3 (Supporting Information) to exemplify the data analysis procedure. [NO3-] ) 1 mM except for R392A-NapAB (trace d), [NO3-] ) 12 mM, because the Km of the R392A variant is 100 times greater than that of the WT enzyme. ω ) 5 krpm, pH 6, 25 °C.
room temperature, then rinsed with water and subjected to the same series of potential steps as above. The fact that no activation is detected on this chronoamperogram shows that dithionite reduction, like electrochemical reduction, has activated the enzyme, and that this activation does not require nitrate. The result of this experiment also makes it unlikely that the activation seen in trace a is an artifact resulting from the reorientation of the enzyme onto the electrode the first time it is taken to low electrode potential. Experiments carried out with the mutant NapAB-R392A confirm the above conclusion that the apparent activation is not linked to the electrode-enzyme interaction. The residue R392 is buried within NapA, in the vicinity of the active site (one of the guanidinium Nη is at 9.2 Å from the Mo); hence it should not contribute to the interaction between the protein surface and the electrode. In solution assays, NapAB-R392A exhibits regular maximal rate, but lower affinity for nitrate than the WT enzyme.17 Trace d in Figure 2 was obtained with a fresh film of NapAB-R392A tested as above. No activation could be detected, suggesting that it is either very fast or very slow, or that it has very small amplitude.33 We shall see in the next section that the latter possibility is true, but the important point here is that mutating a residue that is in the vicinity of the MoCo significantly affects the irreversible activation process. This suggests that the activation relates to a reaction that occurs close to the active site, as confirmed below, as opposed to an artifact of the electrochemical measurement. All the chronoamperograms in Figure 2 reveal that film desorption occurs, to some extent, and this results in the current
magnitude slowly decreasing over time. This is most visible for NapA (trace e), which forms the least stable films. For NapAB, film loss also appears to be faster at -160 mV than at -460 mV, and there may be some instantaneous film desorption associated with the potential jumps. In any case, the slow changes in current observed at -160 mV cannot result from the enzyme reverting to the initial, less active state: should it be the case, we would observe a reactivation at -460 mV following each poise at -160 mV. A series of experiments was carried out to examine how the activation rates measured from data such as those in Figure 2 depend on the experimental parameters of poising (electrode potential E, pH and substrate concentration). We describe how we precisely measured the activation rates ka in the methods section. Figure 3 collects the results, which we fit by assuming that the reductive activation is a Nernstian one-electron reaction (reduction potential E0a ) followed by a chemical step which determines the limiting activation rate klim a (Scheme 1):
ka )
klim a 0
1 + e(F/RT)(E-Ea)
(1)
The best fits in Figure 3 show that the midpoint potential of the sigmoids, E0a in eq 1, and the rates at low potential, klim a , do not significantly depend on pH in the range 6-7.5, and consistent with the above observation that activation occurs even in the absence of substrate, we could not detect an effect of nitrate concentration on the activation rates either (not shown). Experiments carried out with the monomer NapA (trace e in Figure 2) showed that it also irreversibly activates, and the amplitude of the activation is greater than for NapAB (≈70% versus 20%, see figure 7 below). Under the most reducing conditions, the activation of NapA is about 3 times faster than that of NapAB, but the sigmoidal increase in activation rate occurs at the same potential (diamonds in Figure 3). Therefore, the activation of NapAB is not a direct consequence of the reduction of the hemes, which are absent in NapA, or of the [4Fe4S] cluster, whose reduction potential is significantly lower in NapA than in NapAB (-240 mV versus -80 mV).10 Together with the above observation that the activation can be abolished by mutating an amino acid in the active site cavity, this suggests that it relates to the chemistry of the MoCo.
15482 J. Phys. Chem. B, Vol. 112, No. 48, 2008
Figure 4. (A) Mo(V) high-g EPR signature of NapAB. Microwave power: 1 mW. Modulation amplitude: 0.2 mT. Temperature: 55 K. (B) potentiometric titration curves of the [4Fe4S]+ and Mo(V) signals in NapA and NapAB, as indicated. The lines show the best fits to Nernstian sigmoids: for NapA, E0(MoV/MoIV) ) -210 mV, E0(FeS) ) -240 mV; for NapAB, E0(MoV/MoIV) ) -225 mV, E0(FeS) ) -80 mV.10,17 pH 7. (C) intensity of the Mo(V) signal of NapA against solution potential, in an experiment where the potential is held at -160 mV (see text). pH 7, 23 °C. The plain line shows the titration curve expected from the data in panel B. (D) intensity of the Mo(V) signal of NapA against time, in the same experiment as in panel C.
Overall, we conclude that the as-prepared forms of NapA and NapAB irreversibly activate the first time they are fully reduced. This proceeds according to an irreversible chemical reaction coupled to a one-electron reduction. The activation kinetics is affected neither by protonation nor by substrate binding, although this chemistry occurs at (or close to) the active site. Importantly, that the activation is much slower than turnover34 implies that the reductive activation process depicted in Scheme 1 is not part of the normal catalytic cycle. Comparison with the Results of EPR Potentiometric Titrations. Samples of NapAB prepared in air and analyzed by EPR exhibit the high-g signal (Figure 4A).10 This is the most intense signature of the MoCo in all preparations, but we have also detected the so-called “very high-g” signal (Table 1) in certain samples, and traces of the “low-g” signature in one sample (after reduction by dithionite). In potentiometric titrations with dithionite as the reductant, the disappearance of the Mo(V) signal can be fitted to the Nernst equation to determine the reduction potentials of the V/IV couple (Figure 4B), which is independent of pH in the range 6-9.31 Reversible oxidation of the Mo(V) to an EPR-silent Mo(VI) state occurs at very high potential (E0(VI/V) ) +550 mV at pH 7, in both NapA35 and NapAB).17 In contrast, we found that the MoCo is EPR-silent in many mutants. For example, we recently studied the R392A variant of NapAB.17 This mutant has a regular maximal rate in solution assays, and elemental analysis confirmed full Mo load. However, as described in ref 17, we could detect neither the high-g signal nor any other Mo signal upon titrating the mutant in the range -400 to +300 mV, whereas the EPR signatures of the hemes and the [4Fe4S] cluster are identical to those of the WT enzyme. The following observations question the catalytic competence of the high-g species. First, as described for P. pantotrophus
Fourmond et al. NapAB,14,20 the high-g signal of NapAB is insensitive to nitrate (not shown). Second, using the [4Fe4S] + signal as an internal standard, the high-g signal accounts for only 0.15 ( 0.05 spin/ molecule (average of four different preparations) whereas metal analysis and crystallography confirm full Mo occupancy).10,17 In NapA samples, the high-g signal has the same shape and redox properties but it is more intense (about 60 ( 5% of the total Mo, average of two preparations). Third, unlike the oxidation of Mo(V) to Mo(VI), the dithionite reduction of the MoCo to the EPR-silent form Mo(IV) is irreversible: we could observe no Mo(V) signal upon reoxidizing the sample with ferricyanide under either anaerobic conditions or air, whereas EPR confirmed that the [4Fe4S] cluster and the hemes had been reoxidized (not shown). To make sure that the enzyme had not been denatured over the course of the potentiometric titration with dithionite, we carried out PFV experiments using a sample of NapAB that had been “chemically poised”36 at E ) -342 mV (the Mo(V) accounted for
