Introduction

Pigment–protein complexes in photosynthetic organisms convert light energy into chemical energy. In purple anoxygenic bacteria, reaction centers (RCs) embedded in the membrane perform the primary photochemistry (Blankenship et al. 1995). The RC from Rhodobacter sphaeroides consists of three protein subunits and several cofactors (see e.g., Allen et al. 1987; Yeates et al. 1988; Ermler et al. 1994; Stowell et al. 1997; Camara-Artigas et al. 2002). The core L and M subunits surround the cofactors that are divided into two distinct branches related by an approximate two-fold symmetry axis that runs from the center of P to the non-heme iron (Fig. 1). After light excites the primary electron donor, a bacteriochlorophyll a (BChl a) dimer, P, an electron is transferred to a series of electron acceptors along the A branch of cofactors and then to the secondary quinone, QB, creating the charge-separated state P•+QB •−. After reduction of P•+ by an exogenous cytochrome c 2, P can be excited again, leading to the transfer of a second electron to QB •− in a process that is coupled to the uptake of two protons. The generated hydroquinone QBH2 then carries the electrons and protons to the cytochrome bc 1 complex in a cycle that generates the proton gradient needed for the creation of energy-rich compounds.

Fig. 1
figure 1

(a) Cofactors in the bacterial photosynthetic RC from Rb. sphaeroides (PDB entry 1M3X; Camara-Artigas et al. 2002). (b) Structure of the primary donor of the RC from Rb. sphaeroides with the two BChl a molecules PL and PM (phytyl chain truncated), and the three mutated residues His L168, Asn L170, and Asn M199 (PDB entry 1M3X; Camara-Artigas et al. 2002). (c) Molecular structure of bacteriochlorophyll a (BChl a) with IUPAC numbering; the two methyl groups 21 and 121 and the β-protons 7, 8, 17, and 18 are indicated

Full size image

The two BChls that form P overlap at the ring A position with a separation distance of 3.5 Å (see e.g., Allen et al. 1987; Yeates et al. 1988; Ermler et al. 1994; Stowell et al. 1997). Due to the close contact, the two BChls are electronically coupled and the wavefunction of the unpaired electron is distributed over the conjugated systems of both macrocycles. This has been shown by some of the earliest spectroscopic measurements on the RC, in which a dimeric structure was postulated for the primary donor (“special pair hypothesis”)(Norris et al. 1971; 1975; Feher et al. 1975). Electron paramagnetic resonance, EPR, and its advanced multiple resonance methods (ENDOR/TRIPLE) are well-suited for the detailed characterization of the electronic structure of P•+ by mapping the spin density distribution over the conjugated system. In wild type, the distribution is asymmetric with more of the spin density being located on the L-side of P (PL) than the M-side (PM)(Geßner et al. 1992; Lendzian et al. 1993; Rautter et al. 1994; 1995; 1996; Artz et al. 1997; Müh et al. 2002; Lubitz et al. 2002). Due to the large number of protons in the BChl macrocycle (Fig. 1c) that interact with the unpaired electron of P•+, the EPR spectrum shows just a single, unresolved line with a linewidth ΔB pp (peak-to-peak) of 9.6 G (Norris et al. 1971; McElroy et al. 1972; Feher et al. 1975). The linewidth is reduced as compared to that of monomeric BChl a •+ (~14 G at room temperature) due to the dimeric character of P•+ (Norris et al. 1971; 1975; McElroy et al. 1972; Feher et al. 1975; Lendzian et al. 1993).

Details of the spin density distribution can be obtained by determination of the hyperfine couplings (hfcs) via electron nuclear double resonance, ENDOR (Kurreck et al. 1988; Möbius et al. 1982). If the radical–protein complex rotates fast enough to average out all anisotropic contributions of the hfc (and g) tensors only isotropic interactions remain. This results in a single line pair from each set of equivalent protons in the ENDOR spectrum (and a single line in the Special TRIPLE spectrum), which directly yields the isotropic hfcs (Kurreck et al. 1988; Lendzian et al. 1981). It has been shown that for non-aggregated RCs (molecular weight 100 kDa) in detergent containing buffer at 25°C the molecular tumbling is fast enough to average out the g anisotropy and all hfc anisotropies of the proton coupling tensors in P•+ (Lendzian et al. 1981).

Since ENDOR-in-solution experiments suffer from sensitivity problems (Kurreck et al. 1988; Möbius et al. 1982; Plato et al. 1981), Special TRIPLE is usually used. This technique employs one microwave and two radio frequencies, the latter are symmetrically swept around the nuclear Larmor frequency of the respective nucleus being probed (here 1H). With respect to ENDOR, the method has a higher resolution and is less sensitive to the balance of electron and nuclear relaxation rates (Kurreck et al. 1988; Möbius et al. 1982; Plato et al. 1981). For these reasons, Special TRIPLE has a significant advantage when investigating P•+, which gives a weak signal and provides congested spectra. In a series of ENDOR and TRIPLE studies of P•+ in RCs both in liquid solution and single crystals, several hfcs have been resolved and unambiguously assigned (Geßner et al. 1992; Lendzian et al. 1993; Artz et al. 1997; Rautter et al. 1994; 1995; 1996; Müh et al. 2002). In general, for samples in liquid solutions, the technique of Special TRIPLE is well suited to obtain high-quality spectra that can be used to gain detailed insight into the spin and charge distribution within P•+. These techniques have also been used to investigate the effect of a number of different mutations in bacterial photosynthetic RCs (Artz et al. 1997; Rautter et al. 1995; 1996; Müh et al. 1998; 2002; Lubitz et al. 2002).

In general, the surrounding protein environment has been found to play a critical role in determining the properties of the electronic states of P (Allen and Williams 2006; Williams and Allen 2008). In wild type, there is one hydrogen bond between His L168 and the acetyl group of ring A (PL) (Fig. 1b). Mutants with the number of hydrogen bonds to the conjugated system of P ranging from zero to four have midpoint potentials from 410 to 765 mV, compared to 505 mV for wild type (Lin et al. 1994). These mutants also show significant shifts in the spin density distribution over the two halves of P (Rautter et al. 1995; Artz et al. 1997; Müh et al. 2002). The shifts of the P/P•+ midpoint potential and spin density are correlated and provided the basis for detailed theoretical models of the electronic structure of P•+ (Müh et al. 2002; Reimers and Hush 2003; 2004).

In addition to hydrogen bonds, electrostatic interactions have been shown to influence the energy of P•+. These interactions have been probed by insertion or removal of ionizable residues at several different residue positions located ~10–15 Å from the primary donor (Williams et al. 2001; Johnson and Parson 2002; Johnson et al. 2002). Several of the mutants showed a pronounced effect on the P/P•+ midpoint potential and thus also on the primary electron transfer (Williams et al. 2001; Haffa et al. 2002; 2003; 2004). The amino acid residue Asn M199 is located 8.5 Å from P (this is the closest distance from the oxygen or nitrogen atoms of the side chain to the conjugated atoms of P) (Fig. 1b). At pH 8, substitution of Asn M199 with Asp in the ND(M199) mutant was found to result in a 48-mV decrease in the midpoint potential compared to wild type. The replacement of Asn L170, which is located at a comparable distance on the symmetry related side (Fig. 1b), with Asp in the ND(L170) mutant resulted in a 44-mV lowering of the midpoint potential at pH 8 compared to wild type while a 75-mV decrease was observed for the mutation of His L168, which is hydrogen-bonded to the acetyl group of PL, to Glu in the HE(L168) mutant. The effect of having two alterations, His L168 to Glu and Asn L170 to Asp in the HE(L168)/ND(L170) mutant, was more pronounced with a decrease of 127 mV in the midpoint potential. The P/P•+ midpoint potential was found to be pH dependent in these mutants. For example, the P/P•+ midpoint potential for the ND(M199) mutant decreased by 53 mV as the pH was increased from 6.0 to 9.5 (Williams et al. 2001). The mutants were found to have initial electron transfer times ranging from 1.8 to 2.9 ps compared to 3.1 ps for wild type at pH 8 (Haffa et al. 2002). Use of 850 nm light to directly excite P resulted in formation of the charge-separated state P•+QA •− in all mutants. However, use of light at shorter wavelengths of 390, 740, or 800 nm, produced a long-lived charge-separated state consisting of the oxidized M-side BChl and reduced M-side bacteriopheophytin, B •+B H •−B , rather than a state involving P•+(Haffa et al. 2003). For the HE(L168)/ND(L170) double mutant, initial electron transfer following 390 nm excitation was strongly pH dependent, with primarily A-side transfer at pH 7.2 but formation of the B •+B H •−B state dominating at pH 9.5 (Haffa et al. 2004). In this work, the effect of the electrostatic interactions on the properties of P/P•+ in these mutants is investigated by EPR and ENDOR/TRIPLE measurements.

Materials and methods

Rhodobacter sphaeroides wild type 2.4.1 was grown under photosynthetic conditions. The RCs isolated from these cells were purified as previously described (van Mourik et al. 2001).

Cultures of Rb. sphaeroides wild type containing a hepta-histidine tag (WT-H7) and the four mutants, ND(L170), HE(L168), ND(M199), and HE(L168)/ND(L170), were grown under non-photosynthetic conditions (Williams et al. 2001). For isolation of these RCs, a hepta-histidine tag at the carboxyl terminal region of the M-subunit was used as described previously (Goldsmith and Boxer 1996). After purification, the RCs were placed in 15 mM tris(hydroxymethyl)-aminomethane pH 8, 0.025% lauryl dimethylamine oxide, and 1 mM EDTA. Before measurement, the samples were either kept in this buffer (for measurements at pH 8.0) or exchanged by repetitive concentration/dilution using 30 kDa Centricon or Microcon filters into 2-(N-morpholino)ethanesulfonic acid (MES) pH 6.5 or (2-[N-cyclohexylamino]ethane sulfonic acid (CHES) pH 9.5. Finally, the samples were concentrated to an OD802 of 80–130.

CW X-band EPR measurements were performed with a Bruker ESP 300 spectrometer at room temperature using a rectangular cavity with optical access (TE102, ER 4102ST, Bruker), using a capillary with 1 mm inner diameter. The radical cation P•+ was created via continuous illumination with white light in situ, using heat-absorbing glass and water filters.

CW X-band Special TRIPLE measurements were done on the same spectrometer at 288 K. A home-built ENDOR cavity was used, similar to the one previously described (Zweygart et al. 1994), but with a nitrogen gas cooling system. The cation radical P•+ was created in situ as described above. The data analysis was performed using home-written routines in Matlab™, similar to the program used before (Tränkle and Lendzian 1989). In several cases, a baseline was recorded under identical conditions (with the magnetic field off-resonant and subtracted) under the assumption that possible drifts and artifacts would be the same in both cases.

Q-band EPR and ENDOR measurements in frozen solution were done on a Bruker Elexsys E580 spectrometer at 80 K. For frozen solution experiments, sucrose (60%) was added to all samples. A home-built resonator was used (Silakov et al. 2007), similar to the one described previously (Sienkiewicz et al. 1996). A Davies-type pulse ENDOR experiment (Davies 1974) was performed as described previously (Epel et al. 2006).

Results

X-band EPR measurements

Measurements using the X-band EPR spectrometer were performed for both wild-type RCs and the four mutants, ND(L170), HE(L168), ND(M199), and HE(L168)/ND(L170), in liquid solution. In all cases, the spectrum was a single unresolved line centered at g close to g e (see Fig. 2 for an example).

Fig. 2
figure 2

Comparison of CW X-band EPR spectra of light-induced P•+ in RCs from Rb. sphaeroides wild type with hepta-histidine tag (WT-H7) (red line) and from ND(L170) (blue line) at pH 8.0

Full size image

For wild-type RCs at pH 8.0, the spectrum was simulated using a Gaussian function with a linewidth ΔB pp (peak-to-peak) of 9.6 G (±0.2 G) at g = 2.0026 in agreement with published data of this radical in RCs from Rb. sphaeroides 2.4.1 (see for example Feher et al. 1975; Norris et al. 1971; Artz et al. 1997). The spectrum of the four mutant RCs at pH 8.0 were fitted yielding the same g-value and different Gaussian linewidths. For all of the mutants, the EPR linewidth was increased relative to wild type. The linewidth is smallest for the ND(M199) mutant (10.1 G), followed by the HE(L168) mutant (10.2 G), with the ND(L170) mutant and the double mutant HE(L168)/ND(L170) having the most pronounced increase (11.0 G). The increase in linewidth for the mutants indicates a stronger asymmetry of the dimer than found for wild type (Lendzian et al. 1993). However, the mutations may also cause local effects like spin redistributions within the BChl macrocycles or change the geometry of the BChl macrocycles. Since the hfcs of the β-protons at positions 7, 8, 17, and 18 (Fig. 1c) are strongly dependent on the geometry of the respective hydrated rings (Rautter et al. 1995), the EPR linewidth may be changed even without a spin redistribution between the two halves of the dimer. More definitive conclusions can, therefore, only be drawn if the resolution is increased significantly, e.g., by double and triple resonance experiments, yielding the individual nuclear hyperfine coupling constants.

X-band CW 1H Special TRIPLE measurements

P•+ in Wild-Type RCs

Figure 3 compares the Special TRIPLE spectra of WT 2.4.1 (bacteria grown photosynthetically) and WT-H7 (hepta-histidine tag, grown non-photosynthetically) at pH 8.0. The WT 2.4.1 spectrum is identical to that observed before (Geßner et al. 1992; Artz et al. 1997; Müh et al. 2002). The assignment of lines and hfcs (Table 1) follows that of our earlier work (Geßner et al. 1992; Lendzian et al. 1993). Most pronounced are the resonances of the protons of the four (freely rotating) methyl groups (positive hfcs)Footnote 1 and the two β-protons (L-side, positive hfcs). As an indicator for the spin density distribution in the BChl macrocycle, the hfcs of the β-protons at the positions 7, 8, 17, and 18 are less suited, since they are sensitive to the dihedral angle of the respective rings that can easily change (Käss et al. 1994; Rautter et al. 1995). The two spectra show some very small but distinct differences of the proton hfcs. Based upon previous studies, the shifts are unlikely to arise from a difference in the carotenoid composition, due to incorporation of spheroidene and spheroidenone in cultures grown under anaerobic and aerobic conditions, respectively, or differences in the preparations (Geßner et al. 1992; Rautter et al. 1994). The ENDOR/TRIPLE spectrum is sensitive to electrostatic interactions as indicated by the large changes observed upon introduction of hydrogen bonds or use of zwitterionic detergents (Rautter et al. 1995; Müh et al. 1998; 2002). Thus, the most likely cause for the small spectral shift is addition of electrostatic interactions due to the presence of the hepta-histidine tag at the carboxyl terminus region of the M-subunit. For the discussion concerning the mutants, since the changes are very small, the two wild-type samples can be considered to be basically equivalent.

Fig. 3
figure 3

1H-Special TRIPLE spectra (X-band) of light-induced P•+ from RCs from Rb. sphaeroides wild type 2.4.1 (WT 2.4.1) (black line) and from wild type with hepta-histidine tag (WT-H7) (red line) at pH 8.0. The isotropic hyperfine couplings a iso are directly obtained from the Special TRIPLE frequency by ν ST = a iso/2 (for details see Lendzian et al. 1993). Assignments of the lines to molecular positions of the L- and the M-half of the BChl-dimer are given (cf. structure in Fig. 1c)

Full size image
Table 1 1H hfcs [MHz] of P•+ in wild-type reaction centers from Rb. sphaeroides and mutants at pH 8.0 with (tentative) assignments, ratios and sums of hfcs, and EPR linewidths
Full size table

P•+ in mutant RCs

Since the mutants show pronounced pH dependences of the P/P•+ midpoint potential and electron transfer rates, the spectra were measured at three different pH values, 6.5, 8.0, and 9.5. The wild type showed no spectral changes at the pH values of 8.0 and 9.5. Differences in the spectra of the mutants compared to wild type should be predominately due to the substitution of the amino acid residue, excluding any spatial structural changes of P/P•+. Based upon previous studies (Haffa et al. 2002; 2003; 2004; Williams et al. 2001), comparison of the spectra of the mutants at different pH values should show the effect of changes in the protonation, or charge, of the introduced residue. At any given pH, the deprotonated and protonated forms of the residue will be in equilibrium with a ratio determined by the pK a value. If the protonation and deprotonation process is fast compared to the EPR/TRIPLE timescale, only an averaged single species with a shifted spin density distribution will be observed. If the protonation and deprotonation process is slower, two different species associated probably with different spectral properties are expected. Considering that already the Special TRIPLE spectra of one species is rather complicated due to the large number of coupled protons, it becomes difficult to disentangle these spectra containing contributions from two P•+ species. Upon a pH change, either the magnitude of the hfcs will change (L-side and M-side hfcs in opposite direction) or the portion of one species will increase while the other one will decrease.

The assignment of the hfcs in P•+ spectra of mutant RCs has been greatly aided by determining the magnitudes of the four large methyl hfcs, two from each side of the dimer (PL and PM). We have previously measured and analyzed a large number of mutant RCs (Rautter et al. 1995; 1996; Artz et al. 1997; Müh et al. 2002; Lubitz et al. 2002) and the ratio between these hfcs on the respective halves has always been similar, except for mutations that lead to rotation of the acetyl groups of P. In addition, the sum of these four hfcs was found to be constant at ~14 MHz. The spectra of the four mutants are discussed individually below. For the ND(L170) and ND(M199) mutants the respective hfcs are given in Table 1.Footnote 2

ND(L170) mutant

The Special TRIPLE spectrum of ND(L170) RCs at pH 8.0 is shown in Fig. 4 in comparison with the spectrum of WT-H7 at pH 8.0. The P•+ spectrum of the mutant RCs shows two intense, well-resolved signals from β-proton hfcs that are much larger than those in wild type with the two largest methyl group hfcs also larger than found in wild type. Since the ratio between these methyl group hfcs is 1.37, which is typical for the two methyl groups on PL, the strongly coupled β-protons must belong to the L-side, too. In addition, there are several less intense signals overlapping with the methyl groups that are probably due to β-protons. A broader peak around 1.4 MHz is observed that probably arises from several protons, including the stronger coupled methyl group of the M-side. The smaller methyl group is expected to be ~2.4 times smaller and is out of our detection range.

Fig. 4
figure 4

1H-Special TRIPLE spectra (X-band) of light-induced P•+ from RCs from Rb. sphaeroides wild type with hepta-histidine tag (WT-H7) (red line) and from the mutant ND(L170) (blue line) at pH 8.0. The isotropic hyperfine couplings a iso are directly obtained from the Special TRIPLE frequency by ν ST = a iso/2 (for details see Lendzian et al. 1993). Assignments of the lines to molecular positions of the L- and the M-half of the BChl-dimer are given (cf. structure in Fig. 1c)

Full size image

The spectrum from this mutant at pH 8.0 looks very different from that of wild type and resembles the spectra of the heterodimer mutants. In the heterodimer mutants, the exchange of His L173, which coordinates the central Mg of PM, to Leu results in the incorporation of bacteriopheophytin in place of PM (Bylina and Youvan 1988) with most of the spin density being located on PL (Nabedryk et al. 2000; Schulz et al. 1998; Rautter et al. 1995). Hence, it has to be concluded that in P•+ of ND(L170) RCs most of the spin density (86%) is located on PL, which is attributed to the presence of the charged Asp at position L170. Similar electrostatic effects have been reported earlier for mutant RCs (Johnson et al. 2002). An increase of the pH to 9.5 did not alter the spectrum very much (data not shown). If we assume that the residue is completely deprotonated, the pK a should be under 8.0. At pH 6.5, the RCs became unstable and no spectra could be obtained.

ND(M199) mutant

The Special TRIPLE spectra of ND(M199) RCs at pH 6.5, 8.0, and 9.5 are shown in Fig. 5. At pH 8.0, the two large β-proton hfcs are shifted to higher values compared to wild type and a third strongly coupled β-proton is visible. Four intense and narrow lines are present that are assigned to methyl groups. Assuming both larger methyl hfcs belong to the L-side and the two smaller hfcs to the M-side, ratios of 1.79 and 1.57 are calculated, respectively, which are both very different from the values of 2.4 and 1.4 found for wild type and most mutants (Rautter et al. 1995). However, an assignment of the hfcs with 6.32 and 2.59 MHz to one side yields a ratio of 2.44 that would fit very nicely to the M-side but the remaining two lines yield a ratio of 2.18 that does not fit to the L-side at all. The assumption that the signal at 2.59 MHz represents an overlap of L-side and M-side methyl hfcs signals solves this problem, as the ratio of 3.54/2.59 is equal to 1.37, which is the expected ratio for the L-side (Table 1). This assumption leaves the smallest signal of 1.62 MHz unassigned.

Fig. 5
figure 5

1H-Special TRIPLE spectra (X-band) of light-induced P•+ from RCs from Rb. sphaeroides mutant ND(M199) at pH 6.5 (green), 8.0 (red), and 9.5 (black). The isotropic hyperfine couplings aiso are directly obtained from the Special TRIPLE frequency by ν ST = a iso/2 (for details see Lendzian et al. 1993). Assignments of the lines to molecular positions of the L- and the M-half of the BChl-dimer are given (cf. structure in Fig. 1c)

Full size image

The pH dependence for the P/P•+ midpoint potential of this mutant between pH 6.5 and 9.5 was well described using the Henderson–Hasselbalch equation with a pK a of 7.9 (Williams et al. 2001). Consequently, we can expect at pH 8.0 a contribution of two different species, one protonated and the other deprotonated (if the rate constants are slower than the time resolution of the TRIPLE experiment, see discussion above). Comparison with the spectra at pH 9.5 and 6.5 shows that some lines change intensity. This pH difference seems to indicate the presence of two species that could be associated with the protonated and the deprotonated state of the Asp residue. The high pH form (deprotonated) has more spin density on PM and the low pH form (protonated) is similar to wild type with a dominant PL spin density. A species with several lines similar to those of wild type can indeed be found in the spectrum of this mutant at pH 6.5 (already present with lower intensity at pH 8.0) (see Fig. 5).

HE(L168) and HE(L168)/ND(L170) mutants

Special TRIPLE spectra of HE(L168) RCs were recorded at pH 8.0 and 6.5 (data not shown). In comparison to wild-type, the spectrum at pH 8.0 is changed with regard to the signals of the β-protons and the methyl group protons. His L168 forms a hydrogen bond to the oxygen of the acetyl group at position 3 in the BChl macrocycle of the L-side. Hence, the mutation may alter the hydrogen bonding and the acetyl group may undergo a change in its orientation. This leads not only to a shift of spin density between the two halves of P but also results in a redistribution of the spin density within the BChl macrocycle (Rautter et al. 1995). Previous measurements of the spectrum of the mutant HF(L168) were interpreted with the dimer becoming nearly symmetric with slightly more spin density (57%) on the PM half of the dimer. In addition, it cannot be excluded that protonated glutamic acid at lower pH may form a hydrogen bond in contrast to its deprotonated form. The RCs of the double mutant HE(L168)/ND(L170) could be measured only at pH 8.0 (data not shown) due to degradation of the sample at other pH values that seriously limited the signal-to-noise. The problems with the assignment discussed for the mutant HE(L168) apply here, too. Due to these different possible influences and the limited quality of the spectra, no assignments have been made for either of these mutants and they are not discussed below.

Pulsed Q band ENDOR measurements

Experiments in frozen solution of wild type and mutant RCs were performed in addition to liquid solution with the aim of corroborating the hfc data. The advantage of frozen solution is better sample stability and larger sample volume leading to better intensities. In frozen solution, all anisotropic contributions are no longer averaged out. Frozen solution ENDOR thus delivers additional information, but the resolution is strongly decreased in these spectra. Due to their small anisotropy, the methyl groups give fairly strong and narrow signals in such spectra. In wild type, only the two methyl groups with the largest couplings could be simulated, and in the mutants studied in this work only the one with the largest methyl hfc. The deduced isotropic hfcs (Table 1) are the same as those obtained from liquid solution experiments within error. Thus, the frozen ENDOR measurements fully support our Special TRIPLE measurements in the liquid state.

Discussion

In earlier work, it has been shown that the spin density distribution of the primary donor radical cation P•+ in bacterial RCs is a very sensitive probe for structural and electrostatic changes of the dimer and its surrounding. The spin density shifts have for example been correlated with the redox potential of P/P•+ and the electron transfer rates (Rautter et al. 1996; Müh et al. 2002; Lubitz et al. 2002). In the present work, it was shown that even a His-tag attached to the RC leads to a small change of the P•+ characteristics. In the mutants, the effects are much larger.

Two of the mutants, ND(L170) and ND(M199), are located at symmetry related locations that are ~8.5 Å away from P (Fig. 1b). The effects of these mutations should be primarily electrostatic, with the non-protonated form (charged Asp residue) being predominant at high pH (Williams et al. 2001). For ND(L170), the spin density was found to be shifted to the L-side (86% on PL) compared to 68% for wild type. In the case of ND(M199), the spin density was shifted in the opposite direction with only 41% of the spin being on the L-side of P. For the ND(M199) mutant, the ratios of the methyl group hfcs and the pH dependence are reasonable if we assume that the signal at 2.59 MHz arises from two methyl groups. In these spectra, lines from a second species are evident with different intensities at different pH values. These spectral differences indicate a pH-dependent equilibrium between two species with a pK a value close to 8 as found in measurements of the pH dependence of the P/P•+ midpoint potential (Williams et al. 2001).

Such behavior is consistent with the energies of P shifting in response to charges on these two amino acid residues. A negatively charged residue on M199 should destabilize PM, and hence make the two halves more symmetric resulting in a decrease in the spin density on PL. Likewise, a negatively charged residue on L170 should destabilize the energy of PL making the two halves more asymmetrical resulting in an increase of the spin density on PL. These effects are opposite to those observed for the hydrogen bonding mutants (Artz et al. 1997; Rautter et al. 1995; 1996; Müh et al. 2002; Lubitz et al. 2002), as the introduction of a hydrogen bond to the conjugated system of P can be thought of as introducing a net partial positive charge.

The changes in spin density distribution can be directly related to the change in the energy of one of the BChls (Müh et al. 2002). The spin-density ratios, ρ L/ρ M, are 6.1 and 0.7 for ND(L170) and ND(M199), respectively, compared to 2.1 for wild type. These ratios correspond to energy differences between PL and PM of +150 and −45 meV for ND(L170) and ND(M199), respectively, as compared to +60 meV for wild type. Thus, the two mutations both increase the energy of the nearest cofactor, PL for ND(L170) and PM for ND(M199), by nearly the same amount of 90 and 105 meV, respectively. Since in both cases the energy of PL or PM is increasing, the midpoint potential should decrease. For the ND(M199) mutant, the midpoint potential was measured to decrease by 73 mV relative to wild type at pH 9.5, where Asp M199 is expected to be fully ionized (Williams et al. 2001). The extent of the midpoint potential is comparable but not exactly matching the predicted relationship based upon the hydrogen bonding mutants (Müh et al. 2002; Reimers and Hush 2003; 2004). By comparison, spin density ratios of 3.1 and 1.6 were observed for mutants in which Arg was replaced with Glu at the symmetry related positions L135 and M164, respectively (Johnson et al. 2002). For the L135 and M164 mutants, the changes in energy of PL and PM are smaller than those observed for L170 and M199, indicating that the electrostatic interactions are dependent on the distance and orientation relative to P. The results from the mutations at these four residues show that the energies of PL or PM can be preferentially changed depending on the placement of a charged residue.

The amount of B-side electron transfer after excitation at 390 nm has been observed to be altered in the HE(L168)/ND(L170) mutant in a pH-dependent manner that has been interpreted as arising from the presence of ionizable amino acids residues (Haffa et al. 2004). At pH 7.2, electron transfer occurs along the A-branch resulting in the charge-separated state P•+QA •−. At pH 9.5, excitation leads to electron transfer involving the B branch of cofactors and results in the state B •+B H •–B . The present ENDOR/TRIPLE measurements are consistent with the proposal that the switch to B-side electron transfer is due to electrostatic interactions involving the cofactors and the introduced substitutions. The results indicate that the energies of PL and PM change by about 100 meV due to these charges. The comparable distances of L170 to P and BB, 9.0 and 10.5 Å respectively, suggests that B-side electron transfer occurs at least partially by a decrease of the energy of BB •+ by 100 meV, thus favoring formation of B •+B H •–B (Haffa et al. 2004). In general, these data are not only consistent with the idea that B-side electron transfer can be manipulated by the introduction of charges that favor formation of the B-side charge-separated states but also provide a means to quantify the energies of these states.