INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

The primary photochemical event during photosynthesis in bacteriochlorophyll-containing organisms is a light-induced charge separation within a transmembrane protein complex called the reaction center (RC). In purple bacteria the protein complex contains at least three protein subunits referred to as L, M, and H, according to their respective degrees of mobility in sodium dodecyl sulfate-polyacrylamide gels. Associated with the L and M subunits are the cofactors, consisting of four bacteriochlorophylls (BChl), two bacteriopheophytins (BPhe), one atom of nonheme ferrous iron, and two quinones. Most investigations dealing with purple bacterial photosynthesis were performed on the RC of Rhodobacter (Rb.) sphaeroides, Rb. capsulatus, and Blastochloris (B., formerly called Rhodopseudomonas) viridis, respectively. (While Rb. sphaeroides and Rb. capsulatus contain BChl-a and BPhe-a as tetrapyrroles in B. viridis BChl-b and BPhe-b are incorporated.) The cofactors are arranged in two branches with an approximate C₂-symmetry called the A and B branch (for details of the crystal structure see Ermler et al. (1994) and Fig. 1).

View larger version (28K): [in this window] [in a new window]   FIGURE 1   Schematic view of the co-factors and the mutated amino acids in the RC form B. viridis.

The dynamics of the light-induced electron transfer steps have been investigated by time-resolved spectroscopy over the last two decades. Optical excitation leads to the transfer of an electron away from a pair of strongly coupled bacteriochlorophyll molecules, the so-called special pair P in ~2.2 ps (Breton et al., 1986; Dressler et al., 1991). Improved techniques yielded further details about the decay of the first excited electronic state P*: Deviations from a monoexponential decay (Du et al., 1992; Hamm et al., 1993; Jia et al., 1993; Ogrodnik et al., 1994; Beekman et al., 1995) and oscillatory contributions at low temperatures (Vos et al., 1991, 1992, 1993; Spörlein et al., 1998) were also reported. The arrival of the electron at the bacteriopheophytin H_A on the A branch is observed in the time range of 3 to 4 ps after electronic excitation (Breton et al., 1986; Fleming et al., 1988; Holzapfel et al., 1989, 1990). In a series of publications it has been established that the reaction does not directly lead to P⁺H and that the bacteriochlorophyll B_A acts as an intermediary electron carrier (Holzapfel et al., 1989, 1990; Dressler et al., 1991; Schmidt et al., 1993; Arlt et al., 1993).

For the RCs of Rb. sphaeroides and Rb. capsulatus, investigations of mutant or chromophore exchanged RCs yielded information on the energetics and the mechanism of photosynthetic electron transport (Bylina et al., 1988; Nagarajan et al., 1990, 1993; Gray et al., 1990; Finkele et al., 1990; Lauterwasser et al., 1991; Farchaus et al., 1993; Jia et al., 1993; Heller et al., 1995; Woodbury et al., 1995; Huber et al., 1998; Spörlein et al., 2000). For example, RCs with exchanged pheophytins allow to estimate the free energy difference between P* and P⁺B; a value of approximately 450 cm¹ was found in Rb. sphaeroides (Schmidt et al., 1993, 1994).

For B. viridis it has been shown that electron transfer times are in the same range as those observed in Rb. sphaeroides (Dressler et al., 1991). In RCs of B. viridis the first electron transfer step is similar, the second one slightly accelerated (0.65 ps instead of 0.9 ps) (Arlt et al., 1993). Considering the differences between B. viridis and Rb. sphaeroides RC in pigment (BChl-b versus BChl-a) and protein composition (only 50-60% sequence identity of the L- and M-subunits), the good agreement in the time constants is surprising.

While mutant RCs of Rb. sphaeroides and Rb. capsulatus have now been available for over a decade, the construction of a variety of mutant RCs of B. viridis has been achieved only in the midnineties (Laußermair and Oesterhelt, 1992; Dohse et al., 1995).

To study the energetics and transfer mechanism in RC of B. viridis in more detail, we used femtosecond spectroscopy at different temperatures together with site-directed mutagenesis. Most calculations dealing with the primary photosynthetic reaction are carried out on the basis of an improved structure of B. viridis (resolution of 2.3 Å (Lancaster and Michel, 1997)), and in some cases the structure of the mutant RC from B. viridis is now available (Kuglstatter et al., 1999; Lancaster et al., 2000). Therefore, our experimental results can be directly compared with theoretical calculations (Parson and Warshel, 1993).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

Native and mutant RCs of B. viridis were prepared as described previously (Ditta et al., 1985; Dressler et al., 1991; Laußermair and Oesterhelt, 1992). Electrochemical redox titrations were carried out with the set-up described in Wachtveitl et al. (1993b).

The transient absorption experiments in the subpicosecond time range were performed at room temperature by a laser-amplifier system based on a CPM dye laser (Schmidt et al., 1993). The basic features of the system are: a repetition rate of 50 Hz, an excitation wavelength _exc = 960 nm in the low energy Q_Y(P) band, probing pulses in the range between 660 and 1050 nm, and a width of the instrumental response function of 250 fs.

The experiments were performed with parallel polarization between pump and probe pulses. For the room temperature experiments, the sample was held in an optical cell with a 1-mm pathlength and stirred continuously. For measurements at cryogenic temperatures, the samples contained 52% (v/v) glycerol as a cryoprotector and 0.32 M benzyl viologen to prereduce the quinones thus avoiding the accumulation of long-lived photoproducts. The addition of glycerol and benzyl viologen slowed down the electron transfer dynamics by ~10% compared with measurements without glycerol and benzyl viologen. The concentration of the RCs was adjusted to yield a sample transmission of T  10% at 960 nm ( 50 µM). Approximately 15% of the RCs in the irradiated volume were excited. For modeling of the ultrafast reaction dynamics, nonadiabatic electron transfer (ET) theory (Marcus and Sutin, 1985; Bixon et al., 1991, 1995) is applied and the reactions are described by a rate equation system (Schmidt et al., 1995). The intermediates I_i and I_j of the ET processes are connected by microscopic rates _ij. The rates measured in the time-resolved experiments correspond to the eigenvalues k_i = 1/_i of the rate matrix. They are not identical with the microscopic rates when branching, backward rates, or recombination processes are involved (Schmidt et al., 1995). For the theoretical modeling of the transient absorption data a sum of exponentials weighted with amplitudes and convoluted with the instrumental response function is used.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

Investigated mutants

In this study we concentrated on RCs mutated around the special pair and the adjacent monomeric BChl-b pigments B_A and B_B. The positions of the exchanged amino acids within the RC are depicted in Fig. 1. The mutants can be divided into four groups.

RCs with a modified hydrogen bonding pattern to the special pair

RCs mutated at the positions L181 and M208, key positions located between P, B_A and H_A, and P, B_B and H_B, respectively

RC with a modified binding pocket of the bacteriochlorophyll B_A

RC mutated at position L162 located between the special pair P and the proximal heme c-559, which reduces the photooxidized special pair after primary charge separation

Steady-state characterization

Optical absorption spectra

Electrochemical redox titrations

Femtosecond spectroscopy

Time resolved experiments have been performed at probing wavelengths characteristic of the different ET steps (Dressler et al., 1991). For the modeling of the transient absorption changes we used time constants _i, i = 1, ... 4 connected to the different reaction steps (₁, decay of P* and ET from P* to B_A; ₂, ET from B_A to H_A; ₃, ET from H_A to Q_A; and ₄, (= ), long lasting absorption changes). The set of time constants for each mutant was determined globally from the data recorded at the different probing wavelengths. They are summarized in Table 2.

At certain probing wavelengths single reaction steps dominate the observed absorption changes: The population of the excited electronic level P* of the special pair can be investigated via its stimulated emission (gain) in the long wavelength region of the special pair Q_Y-band (1020-1050 nm). Fig. 2 presents some examples of the detected transient absorption changes of the mutant RC. In general, monoexponential fit functions (time constant ₁) do not reproduce all the details of the absorption changes related to the P* decay. Thus, a biexponential fit function was used (time constants _1A, _1B), as has been successfully applied in previous studies (Jia et al., 1993; Hamm et al., 1993). Except for the mutants M195YF, M208YF, and M208YL the amplitude a_1A of the faster decay component dominates. Both the relative amplitudes a_1A and a_1B (of the time constants _1A and _1B) and the relative values of the time constants themselves determine the deviation from a mono-exponential decay. The P* population N_P*(t) is given by:

(1)

Here a_1A and a_1B are the amplitudes of the two P* decay times _1A and _1B with the normalization N_P*(0) = 1 or a_1A + a_1b = 1. Often, the monoexponential decay constants ₁ are sufficient for a qualitative estimate for the lifetime of the excited electronic level P*. The data evaluation reveals the qualitative trend that larger deviations from a monoexponential function were observed in the mutants with a slower P* decay. This agrees well with results obtained on mutated RCs of Rb. capsulatus (Jia et al., 1993). The deviations from a monoexponential fit function are small for mutants showing a very fast decay (L168HF, L181FY, L168HF/M195YF).

View larger version (22K): [in this window] [in a new window]   FIGURE 2   Transient absorption data (circles, triangles, diamonds) of different mutants and WT RC at 1050 nm (mutant L168HF, where the Qy absorption peak is blueshifted at the corresponding wavelength 1020 nm). The solid curves are modeled using a biexponential decay. The modeled time constants 1A and 1B (see Eq. 1) and their amplitudes are given in Table 2. Note the linear scale for the delay times of <1 ps and the logarithmic scale at later delay times.

We find dominant decay times of P* (see Table 2) between 0.6 and 50 ps (1.8 ps is the value for WT). The removal of a hydrogen bond to the special pair generally leads to an acceleration of the P* decay (mutants L168HF, M195YF, and double mutant L168HF/M195YF). This double mutant shows the fastest P* decay with ₁ = 0.8 ps (monoexponential fit) or _1A = 0.6 ps (biexponential fit). Such a drastic acceleration by a factor 3 to 5 as compared with WT (monoexponential fit) was not observed for the corresponding mutants of Rb. sphaeroides (Woodbury et al., 1994; Murchison et al., 1993).

In qualitative agreement with the M208-mutants of Rb. capsulatus (Jia et al., 1993) and the M210-mutants of Rb. sphaeroides (Wachtveitl et al., 1998; Beekman et al., 1996; Finkele et al., 1990; Jia et al., 1993; Nagarajan et al., 1990) the replacement of the tyrosine (Y) at the position M 208 by a phenylalanine (F) or leucine (L) slows down the first electron transfer step by a factor of 6 to 10. In contrast, the introduction of an OH-group in the mutant L181FY accelerates the transfer twofold.

The temperature dependence of the P*-decay was measured for wild-type RCs, the mutants M195YF, L181FY, L168HF, M208YF, and for the double mutant L181FY/M208YF (for a summary of the data see Fig. 4 and Table 3). Whereas WT and the mutants M195YF, L181FY, and L168HF show an accelerated P*-decay at lower temperatures, the double mutant has a temperature-independent decay with a time constant _1a = 3 ps. For the mutant M208YF the dominant component _1b slows down from 50 ps at room temperature to 90 ps at 100 K. 

At probing wavelengths around 940 to 970 nm, i.e., at the center of the Q_Y(P)-band, the decay of P* and the recovery of the ground state absorption of P is measured. In WT RC (dotted lines in Fig. 3) the bleaching of the Q_Y(P) absorption remains constant apart from a small contribution of stimulated emission in the picosecond range, i.e., nearly all P* are converted to the P⁺ product. In mutants with a slow P* decay (₁ > 12 ps; e.g., M208YL, M208YF) a considerable part of the ground state absorption recovers on the picosecond-time scale (Fig. 3, B and C) pointing to a decreased formation of P⁺. Mutants with a shorter lifetime of P* (₁ < 10 ps) show the same behavior as WT RC (for example L181FY, Fig. 3 A). The levels reached after a delay time of 1 ns indicate that the quantum yield for P⁺Q formation of ~97% in WT RC (Trissl et al., 1990) is reduced to 65 ± 10% (M208YL) and 75 ± 10% (M208YF) in the mutants. Because the direct recombination times from the intermediate states P⁺H and P⁺Q in wild-type RC are in the nanosecond- and millisecond-time range, respectively (Ogrodnik et al., 1988), the reduced quantum yield after 1 ns must be due to recombination events via early intermediate states (see below).

View larger version (25K): [in this window] [in a new window]   FIGURE 3   Transient absorption data at 960 nm close to the peak of the P absorption band for three mutants and the WT RC of B. viridis (points). The solid curves are model functions using the time constants 1A and 1B in Table 2.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

In this paper we have studied the primary step of the ET reaction in WT RCs from B. viridis and in 12 mutants. The data are supplemented by the low temperature data on WT and five mutant RCs. The different mutations lead to a considerable change in the primary reaction dynamics. A speeding up of the ET to a reaction time at low temperatures of 250 fs was observed as well as a slowing down to time constants of 90 ps.

ET in photosynthetic RC is usually treated by the nonadiabatic theory, where three important parameters, the electronic coupling V, the reorganization energy , and the gain in free energy G determine the reaction rate. Several publications have evaluated the effect of mutations on the reaction dynamics. In most cases the mutation was assumed to predominantly influence the gain in free energy G (Jia et al., 1993; Bixon et al., 1995).

After a general description of ET theory we will describe the ET reaction in the reaction centers using the approach with constant  and V. However, we will show that the temperature-dependent ET rates cannot be explained within this model, and we suggest that the mutations strongly influence the electronic coupling.

Theoretical description of ET reactions

To obtain information on the molecular parameters of the first reaction step we will discuss the dominant component _1A of the biexponential decay function with the help of nonadiabatic electron transfer theory (Marcus and Sutin, 1985; Bixon et al., 1991). In the high temperature limit, the transfer rate constant k depends on the free energy difference G between the product and reactant states, and the temperature T according to the expression

(2)

in which FC is the Franck-Condon factor,  the reorganization energy, and V the electronic coupling matrix element (Marcus and Sutin, 1985). According to Eq. 2 the logarithm of the rate plotted as a function of the free energy difference G is a parabola (Marcus parabola): the rate k reaches a maximum for G =  (nonactivation case). The region with G <  is called the activated or normal region, whereas the region G >  is known as inverted region. The classical Marcus expression for the high temperature Franck-Condon factor is modified if the available excess energy is transferred into additional intramolecular vibrational modes with nonvanishing Franck-Condon factors. This model is required for the description of energetic situations where the first intermediate is well below the electron donor. These modes are considered via an effective high frequency mode _H. In this model the Franck-Condon factor in the high temperature limit k_BT is given by (Bixon et al., 1988):

(3)

in which _H is the frequency and S_H a measure for the coupling of the coupling strength of the high frequency mode. In the following we will use the values _H = 1500 cm¹ and S_H = 0.5 (Bixon et al, 1995).

Temperature dependent ET reactions were only recorded for mutants with small gain in free energy. Therefore we do not consider high frequency modes (single mode picture) and fit the temperature dependent data with the semiclassical Hopfield formula 4, which is simpler to evaluate but yields (for the range of parameters used here) a very good approximation of the exact equation (Hopfield, 1976; DeVault, 1984):
with

(4)

In the nonactivation case this equation is reduced to:

(5)

The single mode picture seems to be justified here, since temperature dependent data were not recorded for mutants with very low lying acceptor states, where high frequency modes have to be considered.

Modeling of the ET reactions, model A: room temperature rates of different mutants fitted with constant V and

The decay times of the dominant components _1A of the various mutants are used here to determine the relevant parameters of the primary electron transfer via Eqs. 2 and 3. Because these equations contain a series of reaction parameters, several assumptions are necessary to reduce the number of free parameters. At first we follow the most common approach used for the RCs of Rb. sphaeroides and Rb. capsulatus (Williams et al., 1992; Jia et al., 1993; Bixon et al., 1995; Zinth et al., 1998). It is supposed that the electronic coupling V and the reorganization energy  are the same for all mutants. Thus, the mutations act only on the free energy difference G, which can be deduced from the measured P/P⁺ midpoint potential. The variations of the Q_Y(P) absorption band are not considered when estimating the energy levels. This procedure is justified, since no large changes occur for mutants of group 2 to 4 and since it has been shown that the observed shifts of the group 1 mutants are not directly correlated with the energy level of P* (Mattioli et al., 1995; Wachtveitl et al., 1993a). In this model the only relevant parameter to be determined is the energy difference between P* and P⁺B in the wild-type RC. The absolute energetic positions of all other mutants follow from this value. The absolute energy difference G between G_mut(P*) and the first transfer product G_mut(P⁺I) of a mutant is given by:

(6)

The abbreviation I refers to the first acceptor in the ET. For a stepwise ET reaction the BChl-b B_A. Here C = G_WT(P⁺I) is the absolute energy difference between G(P⁺I) and G(P*) in WT and G the relative energy difference induced by the mutation. The relative changes G in the free energy of the first ET step can be estimated for several situations: if the energy level of P* and the molecular nature of the chromophores are not affected by the mutation, the standard free energy change G of a P⁺ containing intermediate state (P⁺I) can be estimated by the change of the P/P⁺ midpoint redox potential U_P/P⁺:

(7)

Here a decrease in redox potential with respect to the WT RC (U_P/P⁺ > 0) corresponds to an increased energy of the intermediate state P⁺I in the corresponding mutant. It should be noted that influences on the energy of the chromophore I can not be recorded by measuring the P/P⁺ redox potential. However, if a mutation leads to the incorporation of a different chromophore, the free energy change can be estimated by:

(8)

in which U_Chrom is the in situ difference of the midpoint redox potential of the exchanged chromophores. For those cases, where an estimate of the in vivo difference of the midpoint potential is not possible, it is necessary to use the in vitro value (e.g., in dimethylformamide). Only in the L153HL RC the mutation results in a chromophore exchange and a value of U_Chrom  0 has to be considered. Because BPhe-b in dimethylformamide has a substantially more negative one-electron reduction potential (0.5 V) than BChl-b (0.7 V) (Geskes et al., 1995), the insertion of BPhe-b into the B_A binding site drastically changes the energetics of the radical pair state P⁺B. According to Eq. 6 this corresponds to an energy difference G(P⁺B)  1600 cm¹. In this context, it must be noted that the absolute value of a redox midpoint potential strongly depends on the solvent used. However, the difference of the midpoint potentials of two chromophores, which is used here for the energy estimate, does not significantly depend on the solvent.

A reasonable estimate of the real free energy change G by the P/P⁺ midpoint potential is not possible for every mutant (Bixon et al., 1995). As a consequence we have to discard "poor" mutants. There are two types of mutated RC that are not considered for the energy estimates.

1) Double mutations are supposed to induce larger geometrical changes. For these mutants, it is not possible to maintain the hypothesis of a nearly unchanged electronic coupling V. We will see below that, even for the other mutants, changes in V may become important. 2) Mutants where the modification is not restricted to P/P⁺. Here the measured P/P⁺ potential is not sensitive to the changes on the other chromophores. The energy estimates within these mutant RCs cannot be accurate. This is the case for the M208 mutants (see below), where we determine G from the temperature dependence (Fig. 4). For L153HL RC, the energetic changes on B_A can be accounted for by the redox potential of the introduced bacteriopheophytin (Eq. 6).

View larger version (20K): [in this window] [in a new window]   FIGURE 4   Temperature dependence of WT and the mutants M208YF, L181FY, L168HF. Solid curves are calculated with the parameters given in Table 4.

The estimated relative energy changes G (according to Eqs. 7 and 8 of the selected mutants are shown in Table 1. These values, the observed decay constants _1A and Eqs. 2 and 6 yield a data set with 9 equations and the 3 unknown parameters , V₁ and C. A least squares fit produces a Marcus parabola (Eq. 2) with the following parameters:  = 600 cm¹ ± 200 cm¹, V = 30 cm¹ ± 8 cm¹, and C = 55 ± 250 cm¹.

Because G(P⁺H) is known to lie between 1600 cm¹ and 2000 cm¹ below G(P*) (Ogrodnik et al., 1994), one important consequence of a value C = G(P⁺I)_WT = 55 cm¹ is that the first electron transfer product cannot be P⁺H and must therefore be P⁺B.

The measured decay constants _1A versus the estimated energy differences are plotted together with the modeled Marcus parabola in Fig. 5 (broken curve). The minimum of the parabola is at G =  = 600 cm¹. Interestingly, this procedure places the WT RC in the activated region of the curve. On the other side, the introduction of a bacteriopheophytin instead of the bacteriochlorophyll at the site B_A in the L153HL mutant places the primary electron transfer reaction of this RC in the inverted region. The good agreement between most room temperature data points and the modeled parabola could be used as a justification for the assumption of nearly constant values of  and V for the selected "good" mutants.

View larger version (18K): [in this window] [in a new window]   FIGURE 5   Visualization of the reaction parameters deduced from the different mutants at room temperature using the assumptions of the model A. (Points) P* decay time of the dominant component plotted versus G(P+B) energies estimated by the midpoint potentials. The absolute energy scale was deduced from the fitting procedure. For M208YF, G was determined from the change of 1B with temperature (Table 4). (Solid line) Marcus parabola for one high frequency mode. (Broken curve) Single mode model. It should be noted that model A fails to explain the temperature dependence of the reaction rates (see text).

If the multimode model is used to fit the data, we obtain the solid curve in Fig. 5 a. A least squares fit using Eq. 3 and _H = 1500 cm¹ and S_H = 0.5 yields values of  = 600 cm¹ ± 200 cm¹, V = 37 cm¹ ± 10 cm¹, and C = 55 ± 250 cm¹. The most striking difference between the classical parabola and the multimode case is that in the latter case the mutant L153HL is much closer to the theoretical curve. In conclusion, both models yield reasonable results and can explain the room temperature data.

The ET parameters for most RCs, especially for the wild-type RC, correspond to a thermal activation of the primary ET reaction step (Eq. 2). From the reorganization energy and the free energy difference obtained in the classical Marcus case and in the multimode picture, we can estimate an activation energy E_A = (G(P⁺B)_WT + )²/4  180 cm¹ for the WT RC. In the single mode picture, this activation energy would slow down the primary ET steps significantly when the sample is cooled. For most other mutants the primary reaction is also activated, and a similar slowing down is expected. It should be noted that in the case of a significant gain in free energy (|G| , inverted region) the multimode model could avoid the slowing down of the primary ET reaction at lower temperatures for a specific choice of the reaction parameters. However, the small gain in free energy in the investigated mutants do not favor this explanation for ET to B_A. Structural reorganization of a network of water molecules associated with the slower rereduction of P⁺ by cytochrome were suggested to explain the temperature dependence of this slow ( 10⁶ µs) ET reaction (Ortega et al., 1998).

Even if we cannot fully rule out changes in reorganization energy  as an influence upon the primary reaction, the rigid environment of the primary electron carriers (Deisenhofer and Michel, 1989) requires an alternative explanation. Therefore, a temperature dependence of the reorganization energy  will not be discussed in the following.

Modeling of the ET reactions, model B: temperature dependence of the initial electron transfer reaction fitted with temperature independent parameters

The primary reactions (dominating component) of various RCs (WT, L181FY, M195YF, L168HF) are accelerated at low temperatures. Therefore, these reaction centers should have negligible activation energies E_A . Since this observation is in contradiction with the activation energies calculated above in section B, we have to consider alternative reaction parameters compatible with the experimentally observed temperature dependencies. To simplify the analysis, we first assume that the reorganization energy (the same value  = 600 cm¹ ± 200 cm¹ is used for all mutants) and the various electronic couplings are independent of temperature. The analysis yields a qualitative description of the temperature dependencies with  = 600,  = 150 cm¹ for all samples. Here was determined from the temperature dependence of the nonactivated mutants at low temperatures (<100 K), where no deviation from the theoretical model is found. The individual free energies and electronic couplings of the different mutants are given in columns 3 and 4 of Table 4. As expected, the results for G are inconsistent with the analysis according to model A (see Table 4, column 2). In addition, the electronic couplings vary considerably upon mutation. Even with these temperature-independent parameters the experimental data and model curves do not agree perfectly (see Fig. 4, at higher temperatures). Often the observed acceleration of the reaction rates toward low temperatures is higher than described even by the activationless ET theory. This discrepancy may result from the assumption of temperature-independent electronic couplings. Indeed, the experimental data of Fig. 4 may be well modeled if the couplings V decrease gradually between 150 and 300 K. 

M208YF is the only mutant studied that shows a slowing down of the ET reaction at lower temperatures. The simulation of this temperature dependence simply by using Eq. 4, however, is not successful, as a strong increase in reaction time at temperatures between 300 and 150 K requires a high activation energy, which would lead to an even stronger increase in reaction time at the lower temperatures. The leveling of the reaction time suggests that at low temperatures another reaction channel with   90 ps dominates the decay of P*. The fitting curve in Fig. 4 uses two decay channels for P*: the first one is thermally activated ET according to Eq. 4 with the parameters given in Table 4, whereas the second one is temperature independent with  = 90 ps. This additional decay path of P* may be connected with an internal conversion of P* to ground state P or to a direct superexchange ET to state P⁺H (Bixon et al., 1995).

Towards more consistent reaction models

In the discussion above several assumptions have been made to reduce the number of free parameters for the modeling of the ET reaction for different mutants and different temperatures. From the contradictory values of the reaction parameters, which result with simplifying model assumptions we can conclude that these assumptions are not adequate to describe photosynthetic ET. As a consequence, only qualitative conclusions may be drawn about the reaction parameters. For example, the low temperature data indicate that the electronic coupling for the first ET reaction is influenced by the temperature change. Considering this observation, it becomes evident that mutations with weak structural changes may also modify the electronic coupling V. Thus, the commonly used strategy underlying model A is inadequate, and the results on the energetics deduced from model A (see Fig. 5 for RC of B. viridis) have to be questioned. For the RCs of Rb. sphaeroides and Rb. capsulatus (Williams et al., 1992; Jia et al., 1993; Bixon et al., 1995; Zinth et al., 1998) similar arguments may apply.

Nevertheless, the analysis of the different mutated RCs at different temperatures yields an important qualitative insight to the reaction parameters, as only certain combinations of parameters are consistent with the observations.

Electronic coupling V

1

Reorganization energy

1

Gain in free energy G

Optimization of the primary reaction

For high quantum efficiencies  of the total photosynthetic reaction chain (it has been shown for WT B. viridis that  amounts to 97% (Trissl et al., 1990)) the primary reaction steps themselves have to occur with highest quantum efficiency _P* for initial charge separation. Quantum efficiency in the very first step is reduced if competing recombination channels (with rates ₁₀) reduce the probability of successful charge separation processes. For a first step with an ET rate _{12 P*} can be estimated to be: _P* = ₁₂/(₁₂ + ₁₀). When the ET rate ₁₂ is reduced (at a fixed recombination rate ₁₀), the quantum efficiency drops and the decay of P* (rate ₁ = 1/₁ = ₁₂ + ₁₀) is slowed down (Gray et al. 1990). The presented absorbance changes in the Q_Y(P) band of the M208 mutants at room temperature (Fig. 3) indicate that such a direct ground state recombination is important in the RC of B. viridis. The slow primary charge separation reaction in the M208YF and M208YL mutants is connected with an increase of the ground state recombination yield. The change in the long lasting absorbance observed in the special pair band (see Fig. 3) has been used to determine the quantum yield in the M208YL and M208YF mutants to be ~ 65% ± 10% and 75% ± 10%, respectively. These data point to a direct ground state recombination rate of ₁₀  1/(80 ps ± 30 ps). The additional decay channel used to explain the reaction time at low temperatures in the mutant M208YF of ₁ = 90 ps would be in full agreement with this finding. However, the treatment (by glycerol and benzyl viologen) of the samples used for the low temperature experiments may influence the absorption dynamics at long delay times. Therefore, we cannot finally decide if the limiting  = 90ps process necessary to fit the data of M208YF at low temperatures is related with internal conversion or with a direct superexchange ET to the BPhe-a H_A.

Considering the efficiency _P* of the first ET step, it is surprising that evolution did not lead to a RC corresponding to the mutant L168HF, where _P* is somewhat larger than in WT RC. In this context it should be taken into consideration that the efficiency of photosynthesis in a bacterium does not only depend upon the efficiency of the initial reaction. For a linear reaction model of the photo-induced reactions in photosynthesis, the overall efficiency of the total reaction chain depends on the forward and backward reaction of each step including the energy transfer from the light harvesting complex. To obtain an efficient energy transfer from the antenna to P* the energetic levels of the two pigment-protein complexes have to be adjusted for optimum excitation transfer. Apparently, this energy transfer is optimized for the absorption spectrum of WT RC, whereas for the fast-reacting mutant L168HF where the P-band (see absorption peak given in Table 1) lies at shorter wavelengths than in WT, the energy transfer efficiency from the low lying antenna to the RC should be reduced (Arlt et al., 1996a). As a consequence, the effect of the increased reaction speed, which would slightly improve the quantum efficiency for primary charge separation in the reaction center itself, (from 97% to 98.8%) is largely overcompensated by the loss due to poorer excitation transfer from the antenna.

	CONCLUSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

The large number of mutated RCs discussed in this paper has allowed us to investigate primary photosynthesis in B. viridis over a wide range of time constants from 250 fs to 90 ps. If hydrogen bonds to the special pair are removed, a drastic acceleration of the first transfer step is observed. On the other hand, we found a slowing down of the P* decay times in mutants, where tyrosine M208 was replaced by nonpolar amino acids, which agrees with the results on Rb. sphaeroides (Nagarajan et al., 1990; Finkele et al., 1990). These slowly reacting mutants have a reduced quantum efficiency for charge separation, which point to a recombination channel from P* in the 80 to 100-ps time domain. An evaluation of the whole data set within the framework of nonadiabatic theory reveals that the commonly made assumption of a constant electronic coupling V for all mutants and over the whole temperature range is not correct (in accordance with recent theoretical arguments Kolbasov and Scherz, 2000). One may therefore conclude that a simple relationship between mutation and a change in one reaction parameter cannot be given and that at least the electronic coupling is changed upon mutation.

Qualitatively, the reaction parameters for the wild-type RC are similar for B. viridis and Rb. sphaeroides with values of V = 20 to 35 cm¹,  = 400 to 800 cm¹, and nonactivated electron transfer G  . The investigations convincingly show that the primary reactions of B. viridis, i.e., a purple bacterium which uses BChl-b and BPhe-b as electron carriers and has accessed the ecological niche using light beyond 1000 nm, behaves very similarly to the other reaction centers such as Rb. sphaereoides where BChl-a and BPhe-a are used instead of BChl-b and BPhe-b. Not only are the reaction dynamics similar, but mutations also lead to similar changes in the different species. The results show that unique design criteria are used for the reaction centers of the different bacteria.