Gallium ferredoxin as a tool to study the effects of ferredoxin binding to photosystem I without ferredoxin reduction

Abstract

Reduction of ferredoxin by photosystem I (PSI) involves the [4Fe–4S] clusters F_A and F_B harbored by PsaC, with F_B being the direct electron transfer partner of ferredoxin (Fd). Binding of the redox-inactive gallium ferredoxin to PSI was investigated by flash-absorption spectroscopy, studying both the P700⁺ decay and the reduction of the native iron Fd in the presence of Fd_Ga. Fd_Ga binding resulted in a faster recombination between P700⁺ and (F_A, F_B)⁻, a slower electron escape from (F_A, F_B)⁻ to exogenous acceptors, and a decreased amount of intracomplex Fd_Fe reduction, in accordance with competitive binding between Fd_Fe and Fd_Ga. [Fd_Ga] titrations of these effects revealed that the dissociation constant for the PSI:Fd_Ga complex is different whether (F_A, F_B) is oxidized or singly reduced. This difference in binding, together with the increase in the recombination rate, could both be attributed to a c. −30 mV shift of the midpoint potential of (F_A, F_B), considered as a single electron acceptor, due to Fd_Ga binding. This effect of Fd_Ga binding, which can be extrapolated to Fd_Fe because of the highly similar structure and the identical charge of the two Fds, should help irreversibility of electron transfer within the PSI:Fd complex. The effect of Fd binding on the individual midpoint potentials of F_A and F_B is also discussed with respect to the possible consequences on intra-PSI electron transfer and on the escape process.

Introduction

Photosystem I (PSI) is a light-dependent oxidoreductase which oxidizes plastocyanin or cytochrome c ₆ on the lumenal side of the thylakoid membrane and reduces ferredoxin (Fd) or flavodoxin on its stromal side (Grotjohann and Fromme 2005; Jensen et al. 2007). Photoinduced charge separation and stabilization in PSI eventually leads to the formation of P700⁺, a dimer of chlorophyll a molecules, and (F_A, F_B)⁻, F_A and F_B being [4Fe–4S] clusters harbored by the PsaC subunit of PSI (Golbeck 2003). The core of PSI exhibits pseudo C₂-symmetry and supports double-branched electron transfer (ET) involving the two phylloquinones PhQ_A and PhQ_B in the A- and B-branches, respectively (Srinivasan and Golbeck 2009). The [4Fe–4S] cluster F_X, which is located at the pseudo-symmetry axis, is reduced by the reduced phylloquinones whereas the final steps of ET involve sequentially F_X, F_A, and F_B (Diaz-Quintana et al. 1998; Vassiliev et al. 1998). Cluster F_B is the redox partner of Fd, a soluble highly acidic 11 kDa protein with a [2Fe–2S] cluster (Fukuyama 2004). Reduced Fd must then dissociate from PSI before it can transfer an electron to a number of partners that are essential for assimilatory pathways of chloroplasts and cyanobacteria (Knaff 1996).

Crystal structures of PSIs (Amunts et al. 2010; Jordan et al. 2001) and Fds (references in Mutoh et al. 2015) have been reported whereas the structure of the PSI:Fd complex is not available yet. Moreover the X-ray structure of a gallium-substituted Fd (Fd_Ga) containing a [2Ga–2S] has been recently obtained and was found to be highly similar to the structure of the natural iron protein Fd_Fe (Mutoh et al. 2015). Fd_Ga is diamagnetic, contrary to Fd_Fe, and this property has been used to identify the different Fd residues involved in the interaction with PSI and ferredoxin-NADP⁺-oxidoreductase (FNR) by NMR (Mutoh et al. 2015).

Functional studies by flash-absorption spectroscopy of Fd reduction by PSI have revealed kinetic complexity with several phases of intracomplex ET (Sétif and Bottin 1995), the origin of which is not clear (Sétif 2001). These studies also revealed the involvement of the 3 stromal PSI subunits, PsaC, PsaD, and PsaE, in Fd binding (references in Sétif et al. 2002). However, the dependence of the Fd to PSI-binding parameters (dissociation constants K_d) upon the redox states of PSI and Fd is not known. A better knowledge in this area is necessary to understand the energetics of complex formation and intracomplex ET and in fine to characterize Fd turnover.

In this work with flash-absorption spectroscopy, we used the property of Fd_Ga to be redox-inactive for probing selectively the effects of its binding to PSI in the absence of ET. Fd_Ga was prepared using Fd_Fe from Thermosynechococcus elongatus (Th. elong., Mutoh et al. 2015) and was tested with cyanobacterial PSIs from Th. elong. and Synechocystis sp. PCC6803 (Syn. 6803). It is shown that the detailed characterization of these effects can teach us thermodynamic and kinetic information about the “real” system with Fd_Fe by extrapolating the effects of Fd_Ga binding without ET to the effects of Fd_Fe binding before ET. We thus used a non-functional protein to get insight into the functional PSI/Fd system.

Materials and methods

Biological materials

Fd_Fe and Fd_Ga from Th. elong. were prepared as previously described (Mutoh et al. 2015). Fd_Fe from Syn. 6803 was purified as previously described (Barth et al. 2000). All PSIs were purified as previously described (Kruip et al. 1993), in the trimeric form for WT from both cyanobacteria and in the monomeric form for the PSI E105Q_PsaD mutant (Bottin et al. 2001). The PSI and Fd concentrations were estimated by assuming absorption coefficients of 7.7 mM⁻¹ cm⁻¹ for P700⁺ at 800 nm (Cassan et al. 2005) and 9.7 mM⁻¹ cm⁻¹ at 422 nm (Tagawa and Arnon 1968), respectively. Except where specifically indicated, measurements were made with PSI from Th. elong. The calculation of the concentration of colorless Th. elong. Fd_Ga was obtained from the titration of absorption changes due to Fd:FNR complex formation (Figures SI1/2), from which an absorption coefficient of 5.1 mM⁻¹cm⁻¹ at the 277 nm maximum was determined.

In vitro flash-absorption spectroscopy

Laser flash-induced absorption change measurements were made at 22 °C and at pH 8 in open 1-cm square cuvettes (aerobic conditions) as previously described (Sétif 2015). P700⁺ decay was studied at 800 nm, where P700⁺ from cyanobacterial PSI absorbs maximally in the infra-red region (Cassan et al. 2005). Most measurements of Fd reduction were made at 580 nm, a wavelength which has been extensively used in previous studies (Sétif and Bottin 1994, 1995). This wavelength has two advantages: (a) PSI has a weak absorption, which allows us to avoid relatively easily the actinic effects of the measuring light; (b) it is an isosbestic point for the formation of carotenoid triplets which may obscure the absorption signals in the μs time range. Kinetics of Fd_Fe reduction was obtained after subtraction of a control measurement made on a reference cuvette without Fd_Fe (Sétif 2015; Sétif and Bottin 1995). This subtraction procedure allows us to observe, on a short time scale, only the reduction of Fd from (F_A, F_B)⁻ corresponding to fast PSI:Fd intracomplex ET (<300 µs after the flash). The kinetics at both wavelengths were recorded with a DC-0.3 MHz bandwidth. This leads to some distortion of the fast subµs kinetics of Fd reduction at 580 nm (Fig. 1). All flash measurements were performed in Tricine 20 mM pH 8.0, in the presence of 5 mM MgCl₂, 30 mM NaCl and 0.03% β-dodecyl-maltoside, with 1 to 2.5 mM sodium ascorbate and 8 to 25 µM 2,6-dichlorophenolindophenol (DCPIP) as slow exogenous electron donors to P700⁺.

Calculations and fits

Fd_Fe reduction difference signals at 580 nm were integrated between 0 and 60 µs after the flash and the integrated signals were used for data fitting. In the absence of Fd_Ga, a simple binding equilibrium was assumed, giving a quadratic equation in [PSI:Fd_Fe]. When Fd_Ga is present, the data were analyzed in the frame of a competitive binding model (presence of two complexes PSI:Fd_Ga and PSI:Fd_Fe exclusive one from the other), for which an exact mathematical expression is available (Wang 1995). Fits were performed using MS-Excel Solver. 3-D plots (Figure SI5) and numerical simulations (Figure SI9) were made with Mathematica 6.0.

P700⁺ decay signals at 800 nm were fitted with 2 exponential components using Origin 7.5 (OriginLab Corp., Northampton, MA). From each biexponential fit, k_r and k_e values were calculated from Eq. (1) in Results (see also Fig. 3A). For each titration using a series of Fd_Ga concentrations, the initial measurement without Fd_Ga was used to derive k_rF and k_eF. The k_r and k_e values in the presence of Fd_Ga were fitted using Eqs. 3, 4 of Results and a simple binding equilibrium between Fd_Ga and PSI, with the three best-fit parameters k_rC, k_eC, and K_{d_red}. This fit was performed using MS-Excel Solver.

Results

Fd_Ga inhibits reduction of Fd_Fe by PSI

The kinetics of Fd_Fe reduction by the terminal acceptor (F_A, F_B) of PSI within the PSI:Fd complex has long been studied by flash-absorption spectroscopy in the 460–600 nm region (Sétif and Bottin 1994, 1995). They are characterized by the presence of 3 different first-order components with t_1/2 of c. 0.5, 15, and 100 µs (Sétif 2001). During the titration of the Fd reduction signal at increasing Fd concentrations, the relative amplitudes of the three components remain constant, i.e., the shape of the first-order kinetics is not modified, whereas their total amplitude increases, as expected for an increase in the proportion of PSI-binding Fd. This allowed the dissociation constant K_d of the PSI:Fd complex to be calculated. Such titrations are shown in Fig. 1 at 580 nm both in the absence (part A) and the presence (0.69 µM, part B) of Fd_Ga, with all proteins from Th. elong. At a given Fd_Fe concentration, the signal in the presence of Fd_Ga is smaller than in its absence, showing that Fd_Ga inhibits Fd_Fe reduction by inhibiting the formation of the PSI:Fd_Fe complex. The signal amplitudes of both titrations are plotted in Fig. 1C and were fitted assuming a simple binding equilibrium between PSI and Fd_Fe. This resulted in K_d = 0.74 µM in the absence of Fd_Ga (red) and an apparent K_d = 1.78 µM in the presence of Fd_Ga (blue), in line with the inhibition of PSI:Fd_Fe complex formation by Fd_Ga.

Fig. 1

Titration of the amount of PSI:Fd_Fe complex in the absence (A and red squares in C) and in the presence (B and blue circles in C) of Fd_Ga. The signals measured at 580 nm in A and B correspond to fast first-order intracomplex reduction of Fd_Fe by (F_A, F_B)⁻ with amplitudes proportional to the amounts of PSI:Fd_Fe complex. These amplitudes were plotted in part C and fitted with models of binding equilibria. 32 measurements were averaged for each trace and a reference signal measured in the absence of Fd_Fe was subtracted. PSI, Fd_Fe, and Fd_Ga (0.69 µM) are from Th. elongatus. A and B: the signal amplitudes increase with successive additions of Fd_Fe (0, 0.29, 0.58, 1.16, 2.29, 4.53, and 8.98 µM from top to bottom traces). PSI concentration: 0.175 µM down to 0.165 µM. Fd_Ga concentration: 0.69 µM down to 0.65 µM. C: the signals measured in A and B were integrated between 0 and 60 µs and the integrated signals were used to derive the proportions of PSI-binding Fd_Fe by best-fit analyses (continuous curves). Data without Fd_Ga (red) were fitted assuming a simple binding equilibrium between PSI and Fd_Fe, leading to K_d(PSI:Fd_Fe) = 0.74 µM. Two procedures were used to best fit the data with Fd_Ga (blue), giving superimposable curves: Firstly, a simple binding equilibrium between PSI and Fd_Fe led to K_d(PSI:Fd_Fe) = 1.78 µM; secondly, a competitive binding equilibrium model was used assuming a fixed K_d(PSI:Fd_Fe) of 0.74 µM, leading to K_d(PSI:Fd_Ga) = 0.44 µM

The above signals are difference signals using reference cuvettes containing no Fd_Fe (without Fd_Ga for Fig. 1A and with Fd_Ga for Fig. 1B). It was also checked that there was no difference between the signals of the two reference cuvettes. A similar null difference recorded at a high signal-to-noise ratio is shown in Figure SI3. The absence of a Fd_Ga-induced signal was also checked at two other wavelengths, 480 and 540 nm, where the proportions of the 3 first-order components are different (Sétif and Bottin 1995). Such an absence of signal difference at short time due to Fd_Ga was a priori expected from the fact that Fd_Ga is redox-inactive. However, the possibility exists that Fd binding to PSI leads either to a change in (F_A F_B) absorption or to a change in the kinetics of ET from $\text{F}_{\text{A}}^{-}$ to F_B, these kinetics possibly occurring in the microsecond time domain (Sétif 2001). In both cases, such a Fd-binding effect could have been mistakenly attributed to Fd reduction. The present data show that Fd_Ga has no such effect and this can be extrapolated to Fd_Fe, assuming that Fd_Ga and Fd_Fe binding to PSI have similar effects on (F_A, F_B) (as supported by data below). In turn, this implies that the fast intracomplex kinetics of Fd_Fe reduction is not spoiled by a signal arising from (F_A, F_B) disturbation and can be entirely attributed to Fd reduction itself.

Fd_Ga binds to PSI competitively with Fd_Fe; evidence for a single Fd-binding site in PSI

Fd_Fe and Fd_Ga have very similar structures (Mutoh et al. 2015). Moreover the [2Fe–2S] cluster of oxidized Fd_Fe and the [2Ga–2S] of Fd_Ga exhibit the same charge with metals in the + III oxidation state. Due to these similarities, it is anticipated that Fd_Ga binds to PSI at the same site than Fd_Fe, i.e., the binding is competitive. The data of Fig. 1B were fitted in a competitive binding model (exact formula from Wang 1995), by fixing the K_d for PSI:Fd_Fe at its value of 0.74 µM obtained without Fd_Ga. The best fit was obtained for K_d(PSI:Fd_Ga) = 0.44 µM (Fig. 1C), a value almost half that of K_d(PSI:Fd_Fe). Such a two-fold difference was observed as well with two other PSI preparations from Syn. 6803, WT and the E105Q single-site mutation of the PsaD subunit of PSI (see Table 1). The choice of studying this mutant was made as it was found to exhibit a significant increase of its affinity for Fd (Bottin et al. 2001) and thus may be useful as a control of Fd_Ga binding.

Although the previous analysis with a competitive binding model gives satisfactory results, we examined further whether there could be a second Fd-binding site on PSI which may have escaped observation up to now because, inter alia, it was not functional for fast ET. It is anticipated that competition experiments may reveal such a secondary site if, when occupied, it modifies the properties of the functional main site by changing either the kinetics of first-order ET or the binding affinity to this site. For this purpose, we firstly compared the intracomplex kinetics of Fd_Fe reduction in the absence and presence of Fd_Ga at the three wavelengths of 480, 540, and 580 nm. We found only minor differences in the shape of the first-order kinetics which appear mostly after 80 µs after the flash (Figure SI4). Such minor differences may be attributed to the differences in the second-order diffusion-limited Fd_Fe reduction, the contributions of which are different whether Fd_Ga is present or not. Secondly we performed, by flash-absorption spectroscopy at 580 nm, competition experiments with different Fd_Ga concentrations of c. 0.34, 0.67, 1.35 µM (data in Fig. 1 are a subset of this full dataset). We also included an experiment by changing [Fd_Ga] at a fixed Fd_Fe concentration of 9 µM. When globally fitting this full dataset with a competitive binding model, we obtained an almost perfect fit (Figure SI5) with K_d(PSI:Fd_Fe) and K_d(PSI:Fd_Ga) of 0.76 and 0.44 µM, respectively. Moreover, best fits of partial datasets (at fixed Fd_Ga concentrations or at [Fd_Fe] ≈ 9 µM) with the same model gave very similar values of K_d(PSI:Fd_Ga) (between 0.40 and 0.45 µM, Figure SI6). This analysis shows that our observations are fully consistent with a competitive binding model and makes it unlikely that there is a second Fd bound to PSI, at least when probed with [Fd] <9 µM.

Table 1 Dissociation constants K_ds (in µM) for binding of Th. elong. Fd_Ga and Fd_Fe to different PSIs

Fd_Ga accelerates recombination between P700⁺ and (F_A, F_B)⁻ and slows down electron escape from (F_A, F_B)⁻

We studied the P700⁺ decay following charge separation in the presence of a slow exogenous electron donor at varying Fd_Ga concentrations. This was measured at 800 nm on a slow time scale where two different decay components are observed (Fig. 2). As the Fd_Ga concentration increases (from top to bottom curves), the decay kinetics becomes faster and faster: the initial decay is faster and the amplitude of the faster/slower component increases/decreases with [Fd_Ga]. The biphasic decay pattern can be easily explained by the following reactions: a charge recombination (rate k_r) between P700⁺ and (F_A, F_B)⁻ which competes with electron escape (rate k_e) from (F_A, F_B)⁻ to an exogenous acceptor (oxygen or another oxidized species present in the sample, see further below); when P700⁺ is left alone without any recombination partner because of escape, it decays more slowly via reduction by the exogenous donor (rate k_don). The P700⁺ decay can be easily calculated according to the previous reactions, leading to Eq. (1):

$$[\text{P}{{700}^{+}}(t)]=\frac{[\text{PSI}]\times {{\text{k}}_{\text{r}}}{{\text{e}}^{-({{\text{k}}_{\text{r}}}+{{\text{k}}_{\text{e}}}+{{\text{k}}_{\text{don}}})\text{t}}}+{{\text{k}}_{\text{e}}}{{\text{e}}^{-{{\text{k}}_{\text{don}}}\text{t}}})}{({{\text{k}}_{\text{r}}}+{{\text{k}}_{\text{e}}})}$$

(1)

Fig. 2

P700⁺ decay at varied Fd_Ga concentrations. Flash-induced absorption changes measured at 800 nm with Th. elong. PSI trimers. The decay accelerates with [Fd_Ga] (0, 0.29, 0.59, 1.16, 2.32, 4.6, 6.9, 9.1, 17.9, and 35.2 µM from top to bottom trace). All kinetics were normalized to a PSI concentration of 0.19 µM (signal amplitudes corrected for dilution). Each trace is the average of 8 experiments

A kinetic example is shown in Fig. 3A, where the two exponential decay components are described in terms of rates and amplitudes. From Eq. (1), the initial P700⁺ decay rate is:

$$\left( \text{d[P}{{700}^{+}]}/\text{dt} \right)\left( \text{t}=0 \right)\text{ }=\text{ }-\left( {{k}_{r}}+\text{ }{{k}_{don}} \right)\text{ }\times \text{ [PSI]}$$

(2)

Fig. 3

Analysis of P700⁺ decay in terms of recombination and escape process. A. A kinetic example is shown with rates and amplitudes of the two decaying exponential phases corresponding to Eq. (1). k_r and k_e are the recombination and escape rates, respectively, whereas k_don is the rate of P700⁺ reduction by exogenous electron donors. B and C. Recombination and escape rates were calculated as a function of the Fd_Ga concentration after fitting each trace of Fig. 2 by a biexponential decay function and using Eq. (1). The P700⁺ decay in the absence of Fd_Ga was used to derive the rates k_rF and k_eF for free PSI (5.2 and 6.6 s⁻¹, respectively). The model used to fit the kinetics in the presence of Fd_Ga is explained in the text and the best-fitting parameters (continuous lines) are the following: k_rC = 14.9 s⁻¹, k_eC = 1.4 s⁻¹, K_{d_red} = 1.45 µM

Equation (2) reveals that a change in the initial P700⁺ decay rate must be attributed to a change in the recombination rate k_r (k_don = constant). One can then conclude from Fig. 2 that k_r increases with Fd_Ga. Moreover the ratio A_fast/A_slow between the amplitudes of fast and slow phases equals k_r/k_e (Eq. (1) and Fig. 3A). Whether this is due not only to an increase in k_r with [Fd_Ga] but also to a decrease in k_e cannot be decided without a quantitative analysis done as follows. All kinetics of Fig. 2 were fitted by a biexponential decay with almost perfect fits: $\text{P}{{700}^{+}}\left( t \right)\text{ }=\text{ }\left[ \text{PSI} \right]\text{ }\left( {{A}_{fast}}\exp \left( -{{k}_{fast}}\times \text{ }t \right)\,+\,{{\text{A}}_{slow}}\exp \left( -{{k}_{slow}}\times \text{ }t \right) \right)\text{ with }{{A}_{fast}}+\text{ }{{A}_{slow}}=\text{ }1$. K_slow was found to be c. constant, as expected from a Fd_Ga-independent reduction of P700⁺ by an exogenous electron donor $({{k}_{\text{slow}}}={{k}_{\text{don}}}\text{)}$. With the two equalities: ${{k}_{r}}+\text{ }{{k}_{e}}=\text{ }{{k}_{\text{fast}}}-\text{ }{{k}_{\text{slow}}}\text{ and }{{A}_{\text{fast}}}/{{A}_{\text{slow}}}=\text{ }{{k}_{r}}/{{k}_{e}}$, one can calculate k_r and k_e for each Fd_Ga concentration, resulting in the plots of Fig. 3B, C. These plots show that Fd_Ga both accelerates recombination and slows down escape.

The dissociation constant for the PSI:Fd_Ga complex deduced from P700⁺ decay is larger than that found from competition experiments

The P700⁺ decay data were further analyzed within Scheme 1. This scheme assumes that, in the presence of Fd_Ga, PSI can be in two states, either free or binding Fd_Ga, and that the recombination and escape rates of the two states differ (k_rF(ree) and k_rC(omplex), k_eF(ree) and k_eC(omplex)). Moreover it was assumed in our analysis that the binding equilibrium K_{d_red} (“red” indicates that (F_A, F_B) is singly reduced) between the two forms is fast as compared to the k_r and k_e rates. This assumption is supported by the following arguments: the second-order rate constant k_on of Fd_Fe association to PSI with (F_A, F_B)⁻ (both proteins from Syn. 6803) has been previously measured at a value of 3.5 × 10⁸ M⁻¹s⁻¹ (Sétif and Bottin 1995) and a similar value has been measured for both proteins from Th. elong. (data not shown); if one assumes a similar value for Fd_Ga, the association rate is 87 s⁻¹ at the lowest value of [Fd_Ga] that was used in the experiment described in Fig. 2 (0.29 µM). Moreover, with a K_d in the µM range (see below), one can calculate a dissociation rate constant k_off = k_on × K_{d (1 µM)} = 350 s⁻¹. These rates are much larger than the k_r and k_e rates that were calculated (Fig. 3B, C). The present system is therefore under the fast exchange regime, where the recombination and escape rates are population-averaged values of the rates corresponding to free and complexed PSI, according to the equations:

Scheme 1

Reactions involved in the decay of the charged pair P700⁺ (F_A, F_B)⁻ with PSI-binding or -not-binding Fd_Ga

$${{k}_{r}}=\text{ }{{k}_{rF}}\times \text{ }\left( \text{proportion of free PSI} \right)\text{ }+\text{ }{{k}_{rC}}\times \text{ }\left( \text{proportion of complexed PSI} \right),$$

(3)

$${{k}_{e}}=\text{ }{{k}_{eF}}\times \text{ }\left( \text{proportion of free PSI} \right)\text{ }+\text{ }{{k}_{eC}}\times \text{ }\left( \text{proportion of complexed PSI} \right),$$

(4)

where the proportions of PSI in the two states depend on the dissociation constant K_{d_red}.

The data of Fig. 3B, C were best fitted using the above formulas (continuous curves) and gave the following best-fit parameters:

k_rF = 5.2 s⁻¹, k_eF = 6.6 s⁻¹, k_rC = 14.9 s⁻¹, k_eC = 1.4 s⁻¹, K_{d_red} = 1.45 µM. This experiment was repeated another time with the same PSI preparation from Th. elong. and was also made with the same two types of Syn. 6803 PSIs that were studied for competition experiments (Tables 2, 3). In a control experiment (Figure SI7), we also observed no effect of Fd_Ga addition on the P700⁺ decay with a PSI mutant having a very low affinity for Fd_Fe [mutant R39Q of the PsaE subunit of PSI (Barth et al. 2000)].

Table 2 Effect of Fd_Ga addition on the kinetic characteristics of P700⁺ decay in isolated PSI (Figs. 2, 3, Eqs. 1, 3, 4)

Table 3 Properties of Fd_Ga binding to photosystem I, characterized with (F_A, F_B) being either oxidized or singly reduced

From these data, it appears that for Th. elong. PSI, Fd_Ga binding induces a 2.7–2.9 fold increase in k_r and a 3–5 fold decrease in k_e. Comparable effects were obtained for the 2 types of Syn. 6803 PSI (Table 2). Moreover, the dissociation constants K_{d_red} are 2.4 to 5.2 larger than the K_ds(PSI:Fd_Ga) measured from competition experiments (3.3–3.6 larger in the case of Th. elong. PSI, Table 3). This apparent discrepancy in K_ds will be discussed further below.

Electron escape from (F_A, F_B)⁻. O₂ is not the only electron acceptor

Whereas all flash-absorption experiments described above were made under aerobic conditions, we also measured P700⁺ decay at 800 nm in the absence of Fd_Ga with screw-cap cuvettes after prolonged incubation and agitation of PSI samples under a N₂ atmosphere in a glove box (data not shown). The slow phase decreased in amplitude indicating that oxygen can accept electrons from (F_A, F_B)⁻ as previously observed (Jordan et al. 1998), a reaction which may contribute to the water–water cycle/Mehler reaction (Asada 2000). However, it was impossible to completely eliminate the slow phase even after extensive incubation in nitrogen, in line with previous observations (Jordan et al. 1998; Rousseau et al. 1993). This suggests that at least one other acceptor is involved.

This was confirmed by the following observation that the slow phase amplitude increases with the amount of sodium ascorbate and DCPIP (both being added for reducing slowly P700⁺ between two consecutive flashes), indicating that oxidized forms of ascorbate and/or DCPIP could as well accept electrons from (F_A, F_B)⁻. The effect of ascorbate/DCPIP concentrations was carefully studied during a Fd_Ga titration at 800 nm with both the WT and the E105Q_PsaD mutant of Syn. 6803: In Table 2, the two experiments for each of the two PSIs were made at concentrations of 1 mM ascorbate/10 µM DCPIP (first line) and 2.5 mM ascorbate/25 µM DCPIP (second line). In each case, it is observed that escape is significantly larger at the highest ascorbate/DCPIP concentrations whereas recombination is not significantly affected. Moreover the Fd_Ga effects (ratios of k_r and k_e) were not dependent upon the ascorbate/DCPIP concentrations, suggesting that the acceptor site is the same for both oxygen and the exogenous radical(s) and that both acceptors are similarly affected by Fd_Ga binding. Studying in greater detail the escape process is certainly worthwhile but is outside the main scope of the present work.

(F_A, F_B) single reduction decreases the affinity of Fd_Ga for PSI, meaning that Fd binding decreases the midpoint potential of (F_A, F_B)

It should be first noted that the competition experiments made at 580 nm correspond to (F_A, F_B) being oxidized (hence the notation K_{d_ox} that is used further below and in Table 3). This is illustrated by Scheme 2 where equilibrium between the three states, free PSI, PSI:Fd_Fe, and PSI:Fd_Ga is established in darkness, i.e., with oxidized (F_A, F_B) whereas fast first-order Fd reduction (bold arrow in Scheme 2) only concerns PSI:Fd_Fe. When a short flash is given, Fd is reduced by (F_A, F_B)⁻ within the complex faster than the time needed for its dissociation or for Fd association to free PSI, so that there is little possibility that the equilibrium between the three above states is substantially modified before intracomplex Fd reduction takes place (slow exchange regime).

Scheme 2

Fd_Fe fast reduction (thick arrow) occurring from a preestablished dark equilibrium involving the PSI:Fd_Fe and PSI:Fd_Ga complexes

Therefore the difference between K_d determinations at 580 and 800 nm can be attributed to the fact that they correspond to different redox states of (F_A, F_B), with the PSI:Fd_Ga complex being destabilized by (F_A, F_B) reduction (K_{d_red} > K_{d_ox}). This complex destabilization is thermodynamically equivalent to (F_A, F_B)⁻ destabilization by Fd_Ga binding to PSI, as illustrated by the thermodynamical cycle of Scheme 3.

Scheme 3

Thermodynamical cycle relating Fd-binding affinities in 2 different redox states of (F_A, F_B) and midpoint potentials of (F_A, F_B) in Fd-binding vs Fd-non-binding states

From this cycle, Eq. (5) can be derived, where E_mF and E_mC are the midpoint potentials for single reduction of (F_A, F_B) in the free and complexed forms of PSI, respectively.

$$\Delta {{\text{E}}_{\text{m}}}=\text{ }{{\text{E}}_{\text{mC}}}-\text{ }{{\text{E}}_{\text{mF}}}=\text{ }\left( \text{RT/F} \right)\text{ }\times \text{ }\ln \left( {{\text{K}}_{\text{d }\_\text{ ox}}}/{{\text{K}}_{\text{d}\_\text{red}}} \right)$$

(5)

A Fd_Ga-binding shift of −30 mV can hence be calculated for E_m(F_A, F_B) of Th. elong. with values of −23 and −40 mV obtained in the cases of Syn. 6803 PSIs (Table 3).

The fact that K_{d_ox} and K_{d_red} are different means that the model that we used for fitting the P700⁺ decay data is oversimplified. Indeed we assumed in our analysis, for it to be tractable, that the ratio between the two PSI populations (free/complexed) remains constant during the whole P700⁺ decay (Eqs. 3, 4). Strictly speaking, this cannot be true if K_{d_ox} ≠ K_{d_red} and this should moreover lead to a non-purely biexponential P700⁺ decay. Firstly, it can be noted that we obtained no evidence for a significant deviation from biexponential decay. Secondly and more importantly, we tested whether our simplifying model could lead to significant errors in the best-fit parameters resulting from our analysis. This approach, described in detail in SI (Figure SI8/9/10), shows that the relative errors made on the three parameters prone to potential errors (k_rC, k_eC, and K_{d_red}) are less than 2% and in turn validates our simplified model.

The Fd_Ga-induced acceleration of recombination can be attributed mostly to the shift of E_m(F_A, F_B)

Due to the large distance between P700 and (F_A, F_B) (edge-to-edge distance ≈ 40 Å), charge recombination between P700⁺ and (F_A, F_B) is not a direct process and involves thermal repopulation of intermediate acceptors (Brettel and Leibl 2001), most probably via the A-side phylloquinone PhQ_A which has a higher E_m than PhQ_B (Srinivasan and Golbeck 2009) (Scheme 4). Assuming that the two states $\text{P}{{700}^{+}}\text{PhQ}_{\text{A}}^{-}\left( {{\text{F}}_{\text{A}}},\text{ }{{\text{F}}_{\text{B}}} \right)$ and P700⁺PhQ_A(F_A, F_B)⁻ are in quasi-equilibrium, the recombination rate k_r between P700⁺ and (F_A, F_B)⁻ depends on the free energy difference between the two radical pairs according to Eq. (6):

Scheme 4

Energetics and reactions involved in the recombination reaction between P700⁺ and (F_A, F_B)⁻ in the Fd-binding vs Fd-non-binding states. The free energies in eV are calculated assuming midpoint potentials of 0.42 V for P700⁺/P700 (value for Th. elong., Nakamura et al. 2011), −0.54 V for (F_A, F_B)/(F_A, F_B)⁻, and −0.67 V for $\text{Ph}{{\text{Q}}_{A}}/\text{PhQ}_{A}^{-}$ (Santabarbara et al. 2005)

$${{k}_{\text{r}}}={{k}_{rPhQA}}\times \text{ }(\exp \left( \Delta \text{G}/\text{RT} \right)/\left( 1\text{ }+\text{ }\exp \left( \Delta \text{G}/\text{RT} \right) \right)\text{ }={{k}_{rPhQA}}\times \text{ }(\exp \left( \Delta \text{G}/\text{RT} \right)\,\text{with}\,\,\Delta \text{G}/\text{RT }<<-1,$$

(6)

where k_rPhQA is the rate of direct recombination between P700⁺ and $\text{PhQ}_{\text{A}}^{-}$

One has:

$$\Delta G = {\text{ }}F{\text{ }} \times {\text{ }}\left( {E_{{\text{m}}} \left( {PhQ_{{\text{A}}} } \right){\text{ }}{-}{\text{ }}E_{{\text{m}}} \left( {F_{{\text{A}}} ,{\text{ }}F_{{\text{B}}} } \right)} \right)$$

(7)

with E_m(PhQ_A) being the midpoint potential of PhQ_A when all other acceptors are oxidized.

The negative shift ∆E_m of E_m(F_A, F_B) due to Fd_Ga binding is therefore expected to accelerate the recombination reaction, as observed. If Fd_Ga has no effect on E_m(PhQ_A), k_rPhQA is not modified by Fd_Ga binding and, from Eqs. (6) and (7), it can be easily calculated that the ratio between k_rF(ree) and k_rC(omplex) depends solely on ∆E_m according to Eq. (8):

$$\ln \left( {{k}_{rF}}/{{k}_{rC}} \right)\text{ }=\Delta{{E}_{m}}/\left( \text{RT}/F \right)\Leftrightarrow \Delta {{E}_{m}}=\text{ }\left( \text{RT}/F \right)\text{ }\times \ln \left( {{k}_{rF}}/{{k}_{rC}} \right)$$

(8)

Comparison of Eqs. (5) and (8) then shows the following equality between ratios of recombination rates and dissociation constants:

$$k_{\rm{rC}}/k_{\rm{rF}}=K{\rm{d\_red}}/K{\rm{d\_ox}}$$

(9)

For easier comparison, the ratios given in Tables 2 and 3 are duplicated in Table 4 with their standard deviations. For WT PSIs from both cyanobacterial species, the 2 ratios are quite similar, although marginally higher for K_{d_red}/K_{d_ox}. This indicates that the effect of Fd_Ga binding on the recombination can be attributed mostly to the negative shift of E_m(F_A, F_B). The case may be slightly different for the Syn. 6803 E105Q_PsaD mutant as the effect on recombination appears to be smaller than that on affinity. This difference in ratios may be attributed to the fact that Fd_Ga binding could also have an effect on E_m(PhQ_A) in the mutant (small upward arrow beside the question mark in Scheme 4). According to our data, a small negative shift of c. 10 mV on E_m(PhQ_A) would decrease the difference between ∆G_F(ree) and ∆G_C(omplex) leading to a smaller effect on the recombination reaction.

Table 4 Comparison of ratios involving either Fd_Ga binding for 2 different redox states of (F_A, F_B) (singly reduced vs oxidized) or binding-dependent recombination rates between P700⁺ and (F_A, F_B)⁻

In a control experiment made to test indirectly whether Fd_Ga could have additional effects on other intermediate acceptors, we also checked with 2 different PSIs (WT from Th. elong. and the E105Q_PsaD mutant from Syn. 6803) that Fd_Ga binding does not modify the yield of charge separation (Figure SI11).

Discussion

Comparison of different PSIs and different Fds

In the present study, we characterized in vitro the effect of Th. elong. Fd_Ga addition to PSI, and compared the binding properties of Fd_Ga to those of Fd_Fe. We also extended this study to WT PSI from Syn. 6803, as this system was the most extensively studied in the past. Moreover we studied the E105Q_PsaD PSI mutant from Syn. 6803, which exhibits a higher affinity for Fd than WT PSI (highest affinity of all cyanobacterial PSIs yet characterized; Sétif et al. 2002), with the idea that the properties of Fd_Ga binding might differ in that case. Indeed the mutant was the only case where Fd_Ga addition significantly shifts the E_m of the phylloquinone PhQ_A involved in charge recombination, in addition to the main effect on the E_m of (F_A, F_B). This effect was however smaller (≈−10 mV) than that on (F_A, F_B). Besides this, the only significant difference between the mutant and WT from Syn. 6803 resided in the recombination rate between P700⁺ and (F_A, F_B) in free PSI, with a two-fold slower rate in the mutant. This suggests that recombination, and therefore most probably E_m(F_A, F_B) is sensitive to the negative charge carried by E105. Apart from these effects, the mutant behavior was essentially identical to those of both WT PSIs (similar effects of Fd_Ga binding on recombination and escape rates; no change in the yield of charge separation).

From Table 1, it appears that Fd_Ga binds to PSI 1.6 to 2.0 times better than Fd_Fe. As Fd_Ga and Fd_Fe have very similar structures and possess the same charge, such a consistently observed difference may have two origins, either a subtle difference between the two Fd structures or a different partial charge localization within the cluster or between the cluster and the cysteine ligands.

Competitive binding of Fd_Ga and Fd_Fe vs single binding site

Conformational flexibility of the PSI:Fd complex has been invoked e.g., on the basis of ET kinetics (Sétif and Bottin 1995) or of experimentally supported calculations (Cashman et al. 2014). Different Fd global orientations have even been proposed with Fd binding either to the side (Fromme et al. 1994; Jolley et al. 2005; Lelong et al. 1996; Sétif et al. 2002) or more to the top (Cashman et al. 2014; Jolley et al. 2005; Ruffle et al. 2000) of the stromal ridge made by the three extrinsic PsaC, PsaD, and PsaE subunits. From these reports, the question can be asked whether PSI can bind 2 Fds together.

The present kinetic data support competitive binding of Fd_Ga and Fd_Fe to PS I, thus making unlikely the possibility that 2 Fds can bind together to PSI. Moreover Fd_Ga binding was tested with two different Syn. 6803 single-site-directed mutants that were previously shown to be the most strongly disturbed in Fd_Fe reduction: mutant R39Q of the PsaE subunit binds Fd_Ga very poorly, as judged by the unmodified kinetics of P700⁺ decay (Figure SI7), in line with the strong loss of affinity for Fd_Fe (Barth et al. 2000); mutant E105Q of the PsaD subunit binds Fd_Ga with a sixfold better affinity than the WT from Syn. 6803, similarly to the affinity increase for Fd_Fe (K_{d_ox}, Table 1). These measurements with mutants also support the idea that Fd_Ga and Fd_Fe bind to PSI in a very similar way.

Effects of Fd binding on the individual E_ms of F_A and F_B

The present results show that Fd_Ga binding induces a shift of −23 to −40 mV in the midpoint potential for single reduction of (F_A, F_B) (E_m(F_A, F_B)) in different PSIs. Assuming that the same effect occurs upon Fd_Fe binding, we will use in the following the notation Fd instead of Fd_Ga, unless inappropriate.

During the recombination reaction, the two states ($\text{F}_{\text{A}}^{-}$, F_B) and (F_A, $\text{F}_{\text{B}}^{-}$) are in fast equilibrium which allowed us to consider (F_A, F_B), as a whole, as a single electron carrier. However, forward electron transfer from F_X to F_A to F_B to Fd_Fe is dependent upon the individual E_ms of F_A and F_B and therefore it is useful to know the individual Fd-binding shifts of these E_ms (∆E_m(F_A) and ∆E_m(F_B)). Unfortunately, our data are not helpful in this respect, and we will discuss this topic below only to highlight the range of possibilities compatible with both the present state of knowledge and our data.

As shown by the equations given in SI, ∆E_m(F_A, F_B) will depend upon ∆E_m(F_A) and ∆E_m(F_B) but also on the difference in E_ms of F_A and F_B in free PSI (∆E_mBA = E_m(F_B) −E_m(F_A)). It will be supposed in the following that F_A is the cluster of highest E_m in free PSI, in line with most titration experiments by EPR, a technique which allows $\text{F}_{\text{A}}^{-}$, $\text{F}_{\text{B}}^{-}$, and ($\text{F}_{\text{A}}^{-}$, $\text{F}_{\text{B}}^{-}$) to be distinguished (Evans et al. 1974; Heathcote et al. 1978; Ke et al. 1973). In these reports, E_m(F_A) was c. −550 mV, while the second reduction corresponding to the state ($\text{F}_{\text{A}}^{-}$, $\text{F}_{\text{B}}^{-}$) was titrated at E_m ≈ −590 mV. This E_m is that of F_B reduction when F_A is prereduced and is not E_m(F_B) that we are presently interested in (F_B reduction with oxidized F_A; redox couple (F_A, F_B)/(F_A, $\text{F}_{\text{B}}^{-}$)). Because of their close proximity, it is extremely likely that the interaction potential between F_A and F_B (shift of E_m(F_B) due to reduction of F_A) is significant and may be even larger than ∆E_mBA. If mostly of electrostatic nature, this interaction potential should be negative and could account for up to −40 mV between the two titration waves measured by EPR. In summary, this means that E_m(F_B) has never been measured but it can be conservatively assumed that E_m(F_B) is lower than E_m(F_A) by a value comprised between a few mV and 40 mV. The difference between E_m(F_A, F_B) and E_m(F_A) is thus plotted in Fig. 4A as a function of ∆E_mBA, with ∆E_mBA comprised between −40 and 0 mV (equation (X) given in SI). E_m(F_A, F_B) is therefore above E_m(F_A) by an amount which is comprised between 5 mV (when E_m(F_B) is 40 mV lower than E_m(F_A)) and 18 mV (entropy-like factor of (RT/F) × ln(2)) when E_m(F_B) = E_m(F_A).

Fig. 4

Difference in midpoints potentials E_ms for the terminal PSI acceptors F_A and F_B. A Difference between E_m(F_A, F_B) [(F_A, F_B) considered as a single electron carrier] and E_m(F_A) as a function of the difference ∆E_mBA between the individual E_ms of F_B and F_A (equation (X) of SI). B Family of curves, all assuming that Fd_Ga binding induces a −30 mV on E_m(F_A, F_B) and each one corresponding to a different value of ∆E_mBA. Each curve plots the Fd_Ga-binding-induced shift of F_B (E_m(F_B)) as a function of the same quantity for F_A (equation (Z) of SI)

In Fig. 4B, we plotted curves corresponding to the couple of [∆E_m(F_A), ∆E_m(F_B)] values which gives ∆E_m(F_A, F_B) = −30 mV, as found with Th. elong. PSI, with the different curves corresponding to different ∆E_mBA values between −40 and 0 mV. We also assumed that ∆E_m(F_A) and ∆E_m(F_B) are comprised between −60 mV and 0 mV. All curves cross at a single point (−30 mV, −30 mV), as expected from the fact that identical shifts on E_m(F_A) and E_m(F_B) should shift E_m(F_A, F_B) by the same value. With ∆E_mBA = 0 (F_A and F_B of identical E_m in free PSI, black curve in Fig. 4B), couples of values are symmetrically distributed across the diagonal, meaning that there is no preferential influence of ∆E_m(F_A) or ∆E_m(F_B) on ∆E_m(F_A, F_B). By contrast, when E_m(F_B) is well below E_m(F_A), the range of ∆E_m(F_B) values is larger than the range of ∆E_m(F_A) values. This means that ∆E_mBA is mainly determined by ∆E_m(F_A) or in other words, that a given shift of E_m(F_A) has a larger effect on ∆E_m(F_A, F_B) than a similar shift of E_m(F_B).

What could be the respective effects of Fd binding on E_m(F_A) and E_m(F_B)? As F_B is closer to the PSI surface than F_A (Jordan et al. 2001) and is the direct ET partner of Fd (Diaz-Quintana et al. 1998; Vassiliev et al. 1998), it can be hypothesized that it is more strongly disturbed than F_A by Fd binding (|∆E_m(F_B)| > |∆E_m(F_A)|), as highlighted by the hashed area in Fig. 4B. Then the “global” shift of E_m(F_A, F_B) should favor Fd reduction within the complex whereas the individual effects on E_m(F_A) and E_m(F_B) should slow down reduction of F_B from $\text{F}_{\text{A}}^{-}$. Fd binding would therefore make the slightly uphill ET from F_A to F_B presumably present in free PSI even more uphill. As this situation appears to be quite compatible with a high rate of Fd reduction, it means that ET from F_A to F_B is nevertheless very fast in the presence of Fd because of the small distance separating the two clusters (Sétif 2001).

Considering again (F_A, F_B) as a whole as a single electron carrier, the negative shift of E_m(F_A, F_B) in the PSI:Fd complex should help in the irreversibility of ET between (F_A, F_B) and Fd within the complex. However, this point cannot be discussed quantitatively as long as the shift of E_m(Fd) induced by complex formation remains known. Whereas work is underway to attempt to characterize this shift, it can be noted that a negative shift of E_m(Fd) has been observed in complexes with FNR, with values ranging from −90 mV (Batie and Kamin 1981) to −22/−15 mV (Pueyo et al. 1992; Smith et al. 1981). Such a shift is in accordance with a lower solvent accessibility in the complex, as water solvation is considered to increase the midpoint potential of exposed iron–sulfur clusters (Stephens et al. 1996).

Effect of Fd_Ga binding on electron escape from (F_A, F_B)⁻

It has been previously reported that electron escape is c. 10-fold smaller after selective destruction of F_B (Diaz-Quintana et al. 1998). This effect can be attributed to the fact that escape is much less efficient from $\text{F}_{\text{A}}^{-}$ than from $\text{F}_{\text{B}}^{-}$. A decrease of electron escape with Fd_Ga binding may thus be attributed to a decreased proportion of $\text{F}_{\text{B}}^{-}$versus $\text{F}_{\text{A}}^{-}$ in singly reduced (F_A, F_B), in accordance with the idea discussed above that Fd_Ga binding may induce a larger negative shift on E_m(F_B) than on E_m(F_A). However, the electron escape decrease can as well be attributed to a decreased access to oxygen (or alternative electron acceptors) in the complex. For distinguishing these two alternative possibilities, it would be e.g., useful to compare the [O₂] dependence of escape in PSI with and without Fd_Ga, a task which is outside the scope of the present study. Whatever the explanation, using Fd_Ga binding as a tool to limit electron escape and thus to accumulate reduced acceptors under conditions of PSI multiple turnover may prove useful to mimic, in vitro, situations where the acceptor side of PSI is overreduced in vivo.

Fd_Ga as an inhibitor

The present work provides clear evidence that Fd_Ga binding to PSI can lead to significant effects, including competitive binding with Fd_Fe. Fd_Ga can thus be considered as an inhibitor of PSI function, whose effects on PSI are not fully explored yet. It can be anticipated that the inhibitory role of Fd_Ga will be of great use to identify and characterize in vitro other Fd-dependent processes.