Marcin
Sarewicz
,
Małgorzata
Dutka
,
Rafał
Pietras
,
Arkadiusz
Borek
and
Artur
Osyczka
*
Department of Molecular Biophysics, Faculty of Biochemistry, Biophysics and Biotechnology, Jagiellonian University, Kraków, Poland. E-mail: artur.osyczka@uj.edu.pl; Tel: +48 12 664 63 48
First published on 26th August 2015
Here, comparative electron spin–lattice relaxation studies of the 2Fe–2S iron–sulphur (Fe–S) cluster embedded in a large membrane protein complex – cytochrome bc1 – are reported. Structural modifications of the local environment alone (mutations S158A and Y160W removing specific H bonds between Fe–S and amino acid side chains) or in combination with changes in global protein conformation (mutations/inhibitors changing the position of the Fe–S binding domain within the protein complex) resulted in different redox potentials as well as g-, g-strain and the relaxation rates (T1−1) for the Fe–S cluster. The relaxation rates for T < 25 K were measured directly by inversion recovery, while for T > 60 K they were deduced from simulation of continuous wave EPR spectra of the cluster using a model that included anisotropy of Lorentzian broadening. In all cases, the relaxation rate involved contributions from direct, second-order Raman and Orbach processes, each dominating over different temperature ranges. The analysis of T1−1 (T) over the range 5–120 K yielded the values of the Orbach energy (EOrb), Debye temperature θD and Raman process efficiency CRam for each variant of the protein. As the Orbach energy was generally higher for mutants S158A and Y160W, compared to wild-type protein (WT), it is suggested that H bond removal influences the geometry leading to increased strength of antiferromagnetic coupling between two Fe ions of the cluster. While θD was similar for all variants (∼107 K), the efficiency of the Raman process generally depends on the spin–orbit coupling that is lower for S158A and Y160W mutants, when compared to the WT. However, in several cases CRam did not only correlate with spin–orbit coupling but was also influenced by other factors – possibly the modification of protein rigidity and therefore the vibrational modes around the Fe–S cluster that change upon the movement of the iron–sulphur head domain.
The family of 2-iron-2-sulphur proteins is divided into two classes according to the average value of g-tensor principal components (gav): ferredoxins with gav ∼ 1.96, and Rieske clusters with gav ∼ 1.91.7 The EPR spectra of clusters reveal a clear rhombic or close to axial symmetry. Variations in g values observed for the Rieske clusters have been explained on the basis of a ligand-field model of antiferromagnetically coupled iron ions.8 Modifications of g values are particularly associated with subtle changes in the ligand field around the Fe2+ ion which is coordinated by histidines. Additionally, the microheterogeneity of the protein environment around the cluster is a source of spread in the ligand field which is manifested as a g-strain effect in the EPR spectra of the clusters in frozen glassy solutions. A statistical theory for the g-strain tensor has been developed to describe asymmetry in the line shapes of the iron–sulphur cluster spectra linking the g-strain with rigidity of the protein core that binds the cluster.9,10
The spin–lattice relaxation rate (T1−1) quantitatively describes how fast the energy absorbed by the spin is transferred to the phonons' bath and its value depends on the coupling strength of the relaxing paramagnet to the surrounding environment as well as on the vibrational modes of the lattice. Commonly, the vibrational characteristics of the glassy frozen protein solution, constituting the lattice is described by the Debye model.11 Temperature dependence of T1−1 of proteins containing the [2Fe–2S] cluster has revealed the existence of direct (one-phonon), second-order Raman and Orbach relaxation processes.12–14 The second-order Raman process is recognised as a ubiquitous relaxation mechanism for all paramagnetic centres and its efficiency depends on several factors including the extent of spin–orbit coupling and rigidity of the protein that influences the vibrational modes of a paramagnetic centre.15,16 In general, the larger the spin–orbit coupling (reflected as a deviation of the g-value from ge) and the less rigid the structure, the higher the efficiency of the Raman process.17 The Orbach mechanism requires the existence of a low-lying excited state which, in the case of iron–sulphur clusters, originates from antiferromagnetic coupling between high-spin iron ions.12,18
The relaxation rates of iron–sulphur clusters have so far mainly been measured for small proteins such as ferredoxins. In these cases, assigning changes in parameters of relaxation to specific structural details of proteins has generally been complicated due to the fact that the comparison involves different proteins coming from different organisms. In this study, we analysed T1−1 of a reduced Rieske cluster [2Fe–2S]2+ (Fe–S) of a large membrane protein complex – cytochrome bc1 (Fig. 1A).19–21 This protein provides a convenient model to compare T1−1 of the same metalloprotein exposed to a variety of structural changes associated with modifications of both the local environment of the cluster and the global environment at the level of the whole protein. By the term “local modifications”, here we understand the changes in the structure of hydrogen bonds formed with the cluster or its ligands, while by “global modifications” we understand the changes in the position of the whole iron–sulphur protein head domain (ISP-HD) in relation to other subunits of cytochrome bc1.
Fig. 1 Crystal structure of cytochrome bc1 and structural details of the ISP. (A) A ribbon model of a native cytochrome bc1 dimer from Rhodobacter capsulatus (PDB ID: 1ZRT20) containing three catalytic subunits: cytochrome b – green, cytochrome c1 – red, and ISP – magenta. (B) A model of one monomer of cytochrome bc1 with overlaid different positions of the ISP-HD. The Fe–S cluster is shown in b-, c- and intermediate positions as yellow-red spheres. The sticks represent hemes c1, bL and the inhibitor stigmatellin. The positions of the ISP-HD were taken from the structures in the PDB database: 1BCC,221BE3,641BGY64 and 1ZRT.20 (C) The local environment of the Fe–S cluster at the b position in the presence of stigmatellin that forms the hydrogen bond with cluster-liganding H156. Yellow dashed lines indicate the H bonds and magenta dotted lines represent the bonds between cluster ligands and iron ions of the cluster. Green and magenta ribbons represent cytochrome b and ISP subunits. |
The Fe–S cluster in cytochrome bc1 is embedded in a water-exposed extrinsic domain (ISP-HD) that naturally undergoes a large-scale movement, inherent in the catalytic cycle of the enzyme. The movement allows the cluster to transfer electrons between the catalytic Qo site in cytochrome b and heme c1 on the cytochrome c1 subunit.22–26 At one extreme (the so-called “b position”), the ISP-HD is in close contact with the Qo site and the Rieske cluster interacts with the substrate bound at the site, while at the other extreme (the so-called “c position”) the ISP-HD is close to heme c1 (Fig. 1B). It has been shown in previous studies that the redox potential of the cluster varies significantly between these two positions – it is generally more than 100 mV higher at the b position, compared to the c position.25,27
In isolated non-inhibited wild-type (WT) cytochrome bc1, the ISP-HD is distributed between b and c position.26 However, the average position of the ISP-HD can be modified by introducing specific mutations that affect the motion and also by addition of specific inhibitors of cytochrome bc1.26,28,29 Therefore, the properties of the Fe–S cluster of cytochrome bc1 can be studied under a variety of conditions in which the position of the cluster can be controlled.
Here, we performed a detailed analysis of the spin–lattice relaxation rate of the Rieske cluster of cytochrome bc1 isolated from Rhodobacter capsulatus under a variety of conditions to determine the basic parameters of the mechanisms that govern the relaxation between 5 K and 120 K particularly, the Orbach energy and Debye temperature. We compared the effect of point mutations S158A and Y160W that remove the specific H bonds in the proximity of the cluster. We also investigated the effect on the efficiency of the Raman process of changes in the global distribution of positions of the ISP-HD modified by using specific inhibitors and mutations.
The measurements of the enzymatic activity of the different forms of cytochrome bc1 were performed as described in ref. 33 using 20 μM horse cytochrome c, 20 μM decylubihydroquinone and 10–100 nM of the enzyme.
Equilibrium redox titrations of the Fe–S cluster were performed on the isolated protein at pH 8. To minimise the pH shift upon freezing the bicine buffer was used.34 The following redox mediators were applied: tetrachlorohydroquinone, diaminodurol, 1,2-naphtoquinone-4-sulfonate, phenazine methosulphate, phenazine ethosulphate, 1,2-naphtoquinone. Ambient redox potential was adjusted by addition of small aliquots of potasium ferricyanide or sodium dithionite solutions. For each point of ambient potential a small portion of the sample was anaerobically transferred to an EPR tube and frozen in liquid nitrogen. Spectra of the titrated protein were subsequently measured by CW EPR at 20 K and the redox potential was determined by fitting the Nernst equation to the data points (Fig. S1, ESI†).
Prior to EPR measurements, all samples of cytochrome bc1, except for those prepared for redox titration, were reduced with sodium dithionite to keep all the protein cofactors in reduced state. After reduction, samples were frozen and stored in liquid nitrogen until used.
The measured IR curves were non-exponential, indicating the distribution of relaxation rates or spectral diffusion. To evaluate the possible effect of spectral diffusion on the recorded IR trace we performed measurements in which the first inverting π pulse was replaced with a saturating picket-fence of microwave pulses, or with a single saturating long-pulse. However, in all cases the shape of recovery curves remained unchanged. The experimental IR curves were mathematically modelled using a stretched exponent function:
The average spin–lattice relaxation rate was then calculated using the equation:
For tds-treated WT, S158A and Y160W and non-inhibited S158A, WT, Y160W, and +2Ala, S158A +1Ala IR curves were measured at gz, gy and gx transitions. In all cases the obtained T1 values depended on their position in the Fe–S spectrum, proving the anisotropic character of the relaxation. In the presence of tds, the spin–lattice relaxation rate increased in the order: (1/T1)gx < (1/T1)gy < (1/T1)gz, while for non-inhibited enzymes a different relationship occurred: (1/T1)gy < (1/T1)gx < (1/T1)gz. The origin of these changes in the relaxation anisotropy upon addition of tds is not known but probably it could be associated with the formation of the hydrogen bond between the inhibitor and the cluster liganding histidine 156. In further analyses only T1 determined at gy transition was considered.
A model of “powder” EPR spectrum simulation, adequate for frozen solutions of metalloproteins (with S = 1/2) was developed by Hagen.37 In our simulations we implemented the basic foundations of Hagen's algorithm: the g-strain tensor was defined in the g-values space, and its components (up to six) determined an orientation-dependent Gaussian linewidth γG(θ,φ) of the resonance transition at geff = g(θ,φ). In addition, the tensor L (expressed in terms of g-values space) was incorporated into the model to account for the presence of anisotropic homogeneous broadening. The principal values of the L tensor were used to calculate the effective value of the Lorentzian linewidth γL(θ,φ) using an expression analogical to that used for the Gaussian linewidth γG(θ,φ). The subsequent lineshape for a given resonance line resulted from convolution of the Gaussian and Lorentzian broadening mechanisms which for practical reasons was approximated by the pseudo-Voigt profile.38,39 The simulated EPR spectrum was created as a superposition of all the calculated signals originating from particular centres that were randomly oriented with respect to the applied magnetic field. This was performed as an orientation-weighted sum over a sphere represented by a SOPHE grid,40,41 which typically consisted of 1000 orientations. Each spectrum was transformed from the g-value domain to the magnetic field domain using proper renormalization formulas. All numerical simulations of the EPR spectra were implemented in a Matlab environment, using the original procedures (available upon request). The values of g, g-strain and L were determined for all examined cytochrome bc1 samples as the “best fit” parameters. Initially, the simulated spectra were fitted using the Simplex method and then further optimised using the Levenberg–Marquardt algorithm (LMA). For temperatures where T2 = T1, based on the fitted values of the L tensor, the spin–lattice relaxation rates for each line i = x, y, z were calculated using the formula:
In the case of multicomponent spectra the average relaxation rate was calculated as the weighted mean value of relaxation rates obtained from the L tensor for each spectral component.
To compare spectral anisotropy between different samples we introduced a phenomenological parameter A, that provides the average (over the whole EPR spectrum) measure of spin–orbit coupling. A is defined as:
As shown in Table 1, S158A and Y160W decreased Em by ∼133 and ∼100 mV, respectively. This is reflected in the lower enzymatic activity and the slower growth rate of bacterial cells under photosynthetic conditions.‡ The changes in the Em of S158A and Y160W are associated with the removal of hydrogen bonds in the vicinity of the cluster as anticipated from inspection of the crystal structure of analogous mutants in Rhodobacter sphaeroides and biochemical studies of the mutated ISP from other organisms.46–49
Form | ΔEm,pH8 [mV] | Photosynthetic growth | Turnover rate of cytochrome c reduction [s−1] ± st. error |
---|---|---|---|
WT | 0 | +++ | 140 ± 5 |
+1Ala | +13 | +++ | 144 ± 7 |
+2Ala | +89 | − | 4.2 ± 0.2 |
S158A | −133 | + | 18.1 ± 0.6 |
Y160W | −103 | ++ | 11.9 ± 0.2 |
Table 1 shows that +1Ala and +2Ala also influence the Em, as reported earlier.27 When the ISP-HD is in the b position, the redox potential of Fe–S is higher compared to the potential of Fe–S in other positions:27 thus, in +2Ala, in which almost the entire population is in the b position, the redox potential is elevated by ∼90 mV, while in +1Ala, in which the population at the b position is larger than in the WT but less than is the case for +2Ala, the potential is elevated by ∼20 mV. The increase in the redox potential upon shifting the ISP-HD to the b position (+1Ala or +2Ala) appears to be independent from the effect of hydrogen bonding as it is also observed for mutants with the shifted ISP-HD and lacking specific H bonds (see example S158A vs. S158A +1Ala in Fig. S1, ESI†). The origin of the change in redox potential of Fe–S related to the change in position of the ISP-HD is still not clear,25,50 but it is likely related to distortions of the active site upon binding the ISP-HD to the Qo site that modifies the ionisation potential of the cluster.4,51
Inhibitors of the Qo catalytic site included tds, ato, fam and myx.52,53 The first three inhibitors (tds, ato, and fam) fix the ISP-HD at the b position, which can occur with (tds and ato) or without (fam) formation of a hydrogen bond between the inhibitor and cluster-liganding H156 residue. Myxothiazol has the opposite effect; it shifts the domains out of the b position.54
The effect of point mutations S158A and Y160W on T1 was initially tested in the presence of tds to eliminate the possible influence from changes in the distribution of ISP-HD positions on T1. The relaxation rate for these samples increased in the order of S158A + tds < Y160W + tds < WT + tds (4.5 × 103, 6.1 × 103 and 11.3 × 103 s−1, at 12 K respectively) (Fig. 2B). This effect must result from modifications in the local protein structure near the Fe–S cluster binding site, in particular the lack of specific hydrogen bonds in the mutants.
Simulation of CW EPR spectra at 25 K allowed us to estimate the principal g tensor and g-strain values that define the inhomogeneous line broadening and the number of components in the particular CW spectrum (see Table S1 in the ESI†). Introducing off-diagonal elements of the g-strain tensors did not improve the fits thus in all cases only principal g-strain values were considered.
The use of one spectral component was sufficient to simulate spectra only for samples in which the ISP-HD was fixed at the b position due to the presence of mutation (+2Ala, S158A +2Ala) or inhibitor tds (WT + tds, S158A + tds, Y160W + tds). This observation suggests that tds and +2Ala mutants arrest the ISP-HD at the b position in such a way that the population of the Fe–S clusters is uniform and single-component. Other inhibitors that also restrict the motion of the ISP-HD (ato, fam) required the introduction of two components, with clearly different g-tensors, to reproduce the experimental spectra. In the presence of ato, addition of ∼18% of the second component was needed, while in the presence of fam, the two components contributed almost equally. The difference in the number of components between tds and ato or fam can be explained by the fact that ato and fam are not as strong inhibitors as stigmatellin and that the hydrogen bond between the inhibitor and ISP-HD is weaker (ato)55–57 or not formed at all (fam).52,58,59 The remaining spectra also required more than one component to obtain good fits but generally two components were sufficient to reproduce all experimental spectra. Fig. S2 (ESI†) shows the inadequacy of selected single-component fits.
Point mutations S158A and Y160W led to an increase in g values and a decrease in both average spectral anisotropy and g-strain values of the Fe–S cluster, when compared to WT. This effect is associated with the changes in the electronic configuration due to modifications of hydrogen bonds around the Fe–S cluster. The addition of myx to the samples led to only small changes in g values and g strain. The representative results of spectral simulations for the WT, S158A and Y160W, in combination with tds and myx are shown in Fig. 3.
In several cases, the fitting of the CW spectra using a model with isotropic Lorentzian broadening led to unacceptably poor fits (see examples for WT and WT + myx shown in Fig. 4). Given that, and also taking into account the anisotropy of IR rates, we concluded that the broadening of the lines must be described by the anisotropic Lorentzian linewidth. Indeed, the anisotropic model of Lorentzian broadening significantly improved the fits (Fig. 4). Clearly, at higher temperatures the relaxation follows the anisotropic trend observed at low temperatures and this anisotropy is even more pronounced.
In Fig. 5, two representative curves of the relaxation rate as a function of temperature 5–120 K are shown. In the range 10–25 K, 1/T1 increases with approximately a sixth power of temperature, indicating domination of the second-order Raman process. Above 80 K this dependence is exponential, suggesting the contribution from the Orbach process, which starts to dominate upon an increase in temperature. At the lowest temperatures (below 10 K), a contribution from the direct process can be recognised from the observation that IR curves are frequency-dependent (the relaxation rates measured at the Q band are higher than those measured at the X band). This frequency-dependence becomes undetectable upon an increase in the temperature as other mechanisms prevail (data not shown). Fig. 5 also shows representative fits of the sum of these three mechanisms to the experimental data points yielding parameters shown in Table 2. For all forms of cytochrome bc1 investigated here, the Debye temperature falls within the range 100–120 K (with an average value of 107 K), which is in agreement with values reported for frozen glassy solutions of proteins.15 The Orbach energy falls within the range ∼350–550 K with an average value of ∼450 K (∼300 cm−1), which also is in the same order as previously reported values for ferredoxins.14,60
Group | Sample | C Dir [s−1 K−1] | C Ram [s−1 K−9] × 10−10 | θ D [K] | C orb [s−1 K−3] | Δ orb [K] |
---|---|---|---|---|---|---|
A | WT | 24.0 ± 3.2 | 1.59 ± 0.16 | 108 ± 4 | 966 ± 130 | 353 ± 29 |
WT + tds | 24.8 ± 2.9 | 1.24 ± 0.11 | 103 ± 3 | 612 ± 59 | 398 ± 25 | |
WT + myx | 33.9 ± 5.7 | 1.86 ± 0.22 | 109 ± 5 | 1052 ± 225 | 406 ± 46 | |
WT + ato | 40.3 ± 9.4 | 0.91 ± 0.17 | 111 ± 6 | nd | nd | |
WT + fam | 23.0 ± 2.3 | 0.98 ± 0.07 | 104 ± 3 | nd | nd | |
WT + ant + myx | 22.7 ± 6.8 | 1.96 ± 0.24 | 110 ± 5 | nd | nd | |
+1Ala | 27.1 ± 2.9 | 1.39 ± 0.12 | 102 ± 3 | 907 ± 144 | 505 ± 28 | |
+2Ala | 31.8 ± 4.0 | 0.913 ± 0.090 | 108 ± 3 | 444 ± 43 | 410 ± 25 | |
B | S158A | 14.3 ± 2.4 | 0.441 ± 0.066 | 118 ± 5 | 234 ± 49 | 447 ± 48 |
S158A + tds | 14.8 ± 1.2 | 0.411 ± 0.031 | 109 ± 3 | 364 ± 24 | 465 ± 16 | |
S158A + myx | 16.0 ± 1.8 | 0.77 ± 0.08 | 101 ± 3 | nd | nd | |
S158A +1Ala | 17.1 ± 2.0 | 0.71 ± 0.06 | 99 ± 3 | 345 ± 26 | 523 ± 16 | |
S158A +2Ala | 21.5 ± 2.3 | 0.455 ± 0.046 | 108 ± 3 | 222 ± 22 | 470 ± 30 | |
C | Y160W | 20.6 ± 3.3 | 1.09 ± 0.13 | 108 ± 4 | 4290 ± 1077 | 554 ± 64 |
Y160W + tds | 14.4 ± 1.2 | 0.599 ± 0.041 | 107 ± 3 | 469 ± 34 | 425 ± 18 | |
Y160W + myx | 23.1 ± 4.0 | 1.15 ± 0.13 | 104 ± 4 | nd | nd |
Comparison of the WT with S158A and Y160W indicates that the Orbach energy increases in these mutants (353 K for WT vs. 447 and 554 K in S158A and Y160W, respectively), suggesting an increased antiferromagnetic coupling between Fe ions in the cluster. This effect is likely to be related to subtle changes in the geometry of the Fe–S cluster particularly in terms of the Fe–S distance and/or Fe–S–Fe angle.61 Inhibitors and mutations affecting the motion of the ISP-HD also influence the Orbach energy, but at this point it is difficult to find a clear tendency for these changes (Table 2).
The point mutation around the Fe–S cluster (S158A or Y160W) has a dominant effect on the relaxation rates thus, the influence of other factors differentiating the samples was further examined within 3 groups: A – encompassing WT variants with the presence or absence of different inhibitors and +1Ala, +2Ala insertions; B – S158A variants with or without inhibitors or +1Ala, +2Ala insertions; and C – Y160W variants with or without inhibitors. The values of CRam′ were compared within each of these groups.
Such analysis revealed the following regularities. When the ISP-HD is fixed at the b position due to the presence of tds, ato, fam or mutation +2Ala, the efficiency of the Raman process decreases. On the other hand, the inhibitors that shift the ISP-HD distribution out of the b position (myx, and myx + ant) cause the CRam′ to increase in comparison to non-inhibited enzymes. These results suggest that the relaxation efficiency of the Raman process for the Fe–S cluster in cytochrome bc1 generally correlates with an increase in the average distance of the ISP-HD from the b position.
The efficiency of Raman relaxation generally increases with an increase in spin–orbit coupling, which is proportional to the difference between g-values and ge = 2.0023. It is also known that for paramagnetic centres with comparable spin–orbit coupling the efficiency is smaller for structures that are more rigid. Therefore, to estimate possible contributions from spin–orbit coupling and protein rigidity, we analysed the dependence of CRam′ on the average spectral anisotropy A (Fig. 6).
The experimental data, depicted in Fig. 6, confirmed the expected general tendency for an increase in CRam′ with growing spin–orbit coupling. However, close inspection of the data in Fig. 6 reveals several cases for which an increase in spectral anisotropy does not lead to an increase in CRam′ (for example CRam′ in S158A +1Ala = S158A = Y160W + tds, S158A +2Ala = S158A + tds or +2Ala = WT + fam = WT + ato) or an increase in CRam′ is not associated only with a change in spectral anisotropy (for example, CRam′ in S158A + tds < S158A < S158A + myx, WT < WT + myx < WT + ant + myx). There is even a case for which CRam′ decreases while spectral anisotropy increases (compare CRam′ in a series +1Ala > Y160W + myx > WT + tds > Y160W > WT + ato).
Considering the black points in Fig. 6 (all describing different forms of S158A), one can predict that of the two forms of the Fe–S cluster with the ISP-HD arrested at the b position (S158A +2Ala and S158A + tds) the one, with the inhibitor tds, is probably more rigid. On the other hand, the Fe–S core becomes less rigid as the population of the cluster is shifted out of the b position (CRam′ in S158A + tds < S158A < S158A + myx). A similar trend can be observed when the ISP-HD in the WT is shifted out of the b position (compare CRam′ for WT<WT + myx < WT + ant + myx – blue points in Fig. 6). This suggests that the rigidity of the Fe–S cluster changes depending on its position, and it is generally larger when Fe–S is in contact with the catalytic Qo site (b position). In addition, in the b position the rigidity appears to be modified by the presence and the type of inhibitor that occupies the site (compare for example: +2Ala, WT + fam and WT + ato). This is probably related to the capability of an inhibitor to create a hydrogen bond with cluster-liganding H156.
Interestingly, Fig. 6 shows examples of CRam′ values that are negatively correlated with spectral anisotropy. This is best seen for the series +1Ala, Y160W + myx, WT + tds, Y160W and WT + ato. To explain this result, we propose that the effect of an increase in spin–orbit coupling is compensated or even dominated by the progressive increase in the rigidity of the protein.
Removing the intramolecular H bond in S158A or Y160W leads to a decrease in CRam′ which may result from changes in vibrational modes and/or spin–orbit coupling. While at this stage separating contributions from these two factors is difficult, one can expect that, if the effect of changes in vibrational modes is significant, then the intramolecular H bond provides an additional channel to transfer vibrations from the surroundings to the paramagnetic centre (Fe–S). On the other hand, an addition of the intermolecular H bond in the presence of the inhibitor may lead to an increase in the rigidity of the Fe–S binding core, probably due to the introduction of an additional constraint to the motion when two subunits of cytochrome bc1 (cytochrome b and ISP) are in close contact and are additionally stabilised by interactions with the bound inhibitor. However this interpretation should be treated with caution, as at this level we have no data on the vibrational modes of the lattice for the different variants of the cytochrome bc1.
The spin–lattice relaxation rate of the Fe–S cluster revealed the existence of significant anisotropy, which becomes more pronounced as temperature increases. Above 80 K, it is so large that simulation of the CW EPR spectra of the cluster must include the anisotropic L tensor which defines different magnitude of Lorentzian broadening for the gz, gy, and gx transitions.
The temperature dependence of the relaxation rates can be explained by assuming the presence of direct (below 5 K), Raman (dominating between ∼10–25 K) and Orbach (dominating above ∼80 K) processes. The presence of these mechanisms in the Rieske cluster of cytochrome bc1 is in agreement with mechanisms described for spin–lattice relaxation of other iron–sulphur clusters in small proteins such as ferredoxins.18,63
The average Orbach energy reflecting antiferromagnetic coupling of the Rieske cluster in cytochrome bc1 was approximately 450 K (∼300 cm−1) which is similar to other iron–sulphur proteins. We found that eliminating specific H bonds between the protein and sulphur atom of the cluster or sulphur atom of the cluster-liganding C133 leads to the general increase in the Orbach energy and may suggest that distances or angles between cluster atoms undergo modifications resulting in stronger spin–spin exchange coupling between Fe3+ and Fe2+ ions of the Fe–S cluster.
The efficiency of the Raman process (CRam) generally increased exponentially with an increase in spectral anisotropy A (related to the spin–orbit coupling). Nevertheless, for some samples characterised by rather large spectral anisotropy, the expected increase of CRam′ was not observed. This lack of correlation can be interpreted as a reflection of changes in the rigidity of the protein that leads to modification of vibrational modes of the Fe–S cluster, that overlays with the effects of changes in spin–orbit coupling. This suggests that upon formation or breaking of inter- or intramolecular hydrogen bonds and/or modification of interactions of the ISP-HD with other subunits of cytochrome bc1 the flexibility of the surroundings for the vibrational motions of the Fe–S cluster changes. It can be speculated that these changes contribute to the overall settings of the electrochemical properties of the cluster and may influence electron transfer between Fe–S and its redox partners during the reactions catalysed by cytochrome bc1.
EPR | Electron paramagnetic resonance |
Fe–S | Iron–sulphur |
ISP | Iron–sulphur protein subunit |
ISP-HD | Movable head domain of iron–sulphur protein subunit |
tds | Tridecyl-stigmatellin |
ato | Atovaquone |
fam | Famoxadone |
myx | Myxothiazol |
ant | Antimycin |
E m | Redox midpoint potential |
IR | Inversion recovery |
CW | Continuous wave |
θ D | Debye temperature |
LMA | Levenberg–Marquardt algorithm |
Footnotes |
† Electronic supplementary information (ESI) available: Fig. S1: results of equilibrium redox titration of the Fe–S cluster for different forms of cytochrome bc1; Fig. S2: selected results of simulations of X-band CW EPR spectra of the Rieske cluster recorded at 25 K using a single-component model; Fig. S3: temperature dependence of the spin–lattice relaxation rate (1/T1) of the Rieske cluster for the tds-inhibited WT in linearized form for Raman and Orbach processes; Table S1: values of g and g-strain components determined from fitting the two-component model to X-band CW EPR spectra (25 K) for different forms of cytochrome bc1. See DOI: 10.1039/c5cp02815a |
‡ Rb. capsulatus obligatorily requires functional cytochrome bc1 to support photosynthetic growth of the bacteria in the absence of oxygen; therefore growth under these conditions reflects the general activity of the enzyme. |
This journal is © the Owner Societies 2015 |