Research Article

Switching sides—Reengineered primary charge separation in the bacterial photosynthetic reaction center

Philip D. Laible, Deborah K. Hanson, James C. Buhrmaster, Gregory A. Tira, Kaitlyn M. Faries, Dewey Holten, and Christine Kirmaier
PNAS January 14, 2020 117 (2) 865-871; first published December 31, 2019; https://doi.org/10.1073/pnas.1916119117

  1. Edited by Donald R. Ort, University of Illinois at Urbana–Champaign, Urbana, IL, and approved December 4, 2019 (received for review September 16, 2019)

Significance

Photosynthetic organisms use pigment–protein complexes called reaction centers (RCs)—effectively nature’s solar cells—to convert the energy of sunlight into charge-separated species that power life processes. Whether from plants, algae, or photosynthetic bacteria, RCs feature 2 quasi-mirror-image arrangements of protein and pigment cofactors. In RCs from purple photosynthetic bacteria, photon-induced electron transfer (ET) reduces the A-branch bacteriopheophytin (BPh) with near-unity quantum yield, while ET to the symmetry-related B-branch BPh is suppressed. Amino acid changes made here in the Rhodobacter sphaeroides RC nearly completely reverse this design. Specifically, we report RCs wherein ET to the B-branch BPh occurs with 90% yield. We draw insight regarding architectural and mechanistic foundations of the primary photon-driven charge-separation events of photosynthesis.

Abstract

We report 90% yield of electron transfer (ET) from the singlet excited state P* of the primary electron-donor P (a bacteriochlorophyll dimer) to the B-side bacteriopheophytin (HB) in the bacterial photosynthetic reaction center (RC). Starting from a platform Rhodobacter sphaeroides RC bearing several amino acid changes, an Arg in place of the native Leu at L185—positioned over one face of HB and only ∼4 Å from the 4 central nitrogens of the HB macrocycle—is the key additional mutation providing 90% yield of P+HB−. This all but matches the near-unity yield of A-side P+HA− charge separation in the native RC. The 90% yield of ET to HB derives from (minimally) 3 P* populations with distinct means of P* decay. In an ∼40% population, P* decays in ∼4 ps via a 2-step process involving a short-lived P+BB− intermediate, analogous to initial charge separation on the A side of wild-type RCs. In an ∼50% population, P* → P+HB− conversion takes place in ∼20 ps by a superexchange mechanism mediated by BB. An ∼10% population of P* decays in ∼150 ps largely by internal conversion. These results address the long-standing dichotomy of A- versus B-side initial charge separation in native RCs and have implications for the mechanism(s) and timescale of initial ET that are required to achieve a near-quantitative yield of unidirectional charge separation.

Conversion of light energy to chemical potential takes place in membrane-spanning reaction center (RC) proteins of photosynthetic bacteria, algae, and plants. Two transmembrane subunits harbor 2 quasisymmetric A and B chains, or branches of (bacterio)chlorophyll, (bacterio)pheophytin, and quinone. Fig. 1 shows the arrangement of the L, M, and H polypeptides and cofactors in RCs from Rhodobacter sphaeroides (1). P is a bacteriochlorophyll (BChl) dimer; BA and BB are monomeric BChls; HA and HB are bacteriopheophytins (BPhs); QA and QB are ubiquinones; and Fe is a nonheme iron. In bacterial RCs and photosystem II of plants and algae, only one branch of cofactors is used for photon-driven charge separation; in the bacterial RC, the A branch is active, and the B pathway is unused. The standard model for the free energies of the A- and B-side charge-separated (CS) states relative to P* in wild-type (WT) RCs from R. sphaeroides is shown in Fig. 2A. Initial charge separation in <10 ps and ∼100% yield occurs by a 2-step process: P* → P+BA− → P+HA−. Several models for the time constants of these 2 electron-transfer (ET) steps have been presented (2⇓⇓–5). Both theory and experiments have shown that TyrM210 is key to positioning P+BA− slightly (∼70 meV) below P* in free energy, while symmetry-related PheL181 does not provide such stabilization of P+BB−, which is thought to be ∼100 to 200 meV above P* (Figs. 1 and 2A) (6⇓⇓⇓⇓⇓⇓–13). State P+HB− also is higher in free energy than its A-side counterpart (P+HA−) and may be only ∼70 meV below P* (14). Thus, normal electron flow in the WT RC is P* → P+BA− → P+HA− → P+QA− → P+QB−; states P+BB− and P+HB− are not utilized.

Fig. 1.
Fig. 1.

(A) R. sphaeroides crystal structure (PDB ID code 1PCR) with polypeptides L (gray), M (green), and H (tan) and embedded cofactors: BChls (green), BPhs (purple), quinones (black), carotenoid (orange), and nonheme Fe (brown). Note that one helix of the L polypeptide lies near the B-side cofactors, and one helix of the M polypeptide lies near the A-side cofactors. (B) RC cofactors (gray) and the native amino acids (blue) that are substituted in 2 RC templates, bg1 and bg2, that differ only in having the native Phe at L216 (bg1) versus Trp at L216 (bg2). Using these 2 templates, the work here focuses on RCs bearing Arg in place of the native Leu at L185 (red).

Fig. 2.
Fig. 2.

(A and B) General model for the relative free energies of P* and CS states in WT RCs (A) and in Arg-bg1 and Arg-bg2 mutant RCs (B). P* is ∼1.4 eV above the ground state. (C–E) Model state diagram for the major decay processes of 3 P* populations in Arg-bg1 and Arg-bg2 RCs: P1*, τ ∼4 ps (C); P2*, τ ∼20 ps (D); and P3*, τ ∼150 ps (E). Solid arrows indicate dominant processes, and dashed arrows reflect competing processes that have much lower yields.

Nearly quantitative redirection of the native initial charge-separation process to give P+HB− in R. sphaeroides RCs is reported here. This is achieved with an amino acid substitution near HB—Arg replacing the native Leu at L185 (red in Fig. 1B)—in RCs also bearing a set of well-studied template, or background, substitutions of native amino acids shown in blue in Fig. 1B. The template mutations give a “starting point” advantage to B-side ET over A-side ET (15⇓–17). TyrM210 near BA is changed to Phe, and PheL181 near BB is changed to Tyr. Consequently, compared to the WT RC, the free energy of P+BA− is higher, and the free energy of P+BB− is lower (Fig. 2B versus Fig. 2A). A third substitution, LeuM214 changed to His, results in replacement of HA by a BChl denoted βA. P+βA− (Fig. 2B) is higher in free energy than P+HA− in WT (Fig. 2A) and close to the position of P+BA− in WT (18, 19). We have previously shown that these 3 mutations in R. sphaeroides RCs result in ∼20% yield of P* → P+HB− and that additionally replacing ThrM133 with Glu increases the P+HB− yield to ∼60% (17). These 4 mutations are the core of 2 RC backgrounds used here, denoted bg1 and bg2. Both backgrounds also bear 2 mutations near QA (Fig. 1) that provide a double block on binding this cofactor (20, 21). The only difference between bg1 and bg2 is that in bg2, a Trp replaces the native Phe at L216 located between HB and QB. Using both bg1 and bg2 templates, an Arg replaces the native Leu at L185, resulting in the Arg-bg1 and Arg-bg2 mutants studied here. These are part of 2 sets of mutants (to be reported elsewhere) where all amino acids have been substituted for LeuL185 in the bg1 and bg2 backgrounds. We show that ArgL185 makes P+HB− a better trap, supporting initial charge separation via a <10-ps 2-step mechanism in a portion of the P* population and via an ∼20-ps superexchange mechanism in another portion—together providing near quantitative conversion of P* to P+HB−.

Results

All results were obtained on RCs isolated and purified by using standard techniques and solubilized in 10 mM Tris (pH 7.8)/0.1% Deriphat 160-C buffer (SI Appendix). For all studies, QB was displaced from the RC by added terbutryn and QA by the mutational blocks noted above. The absence of both QA and QB enables high-repetition-rate ultrafast transient absorption (TA) experiments. Studies were performed on multiple biological replicate (BR) RCs for the Arg-bg1, Arg-bg2, bg1, and bg2 mutants. A BR is an RC obtained from an independently grown culture of cells originating from an archival frozen stock that came from a single colony carrying a plasmid of known sequence. Additionally, for some BRs of Arg-bg1, Arg-bg2, bg1, and bg2 RCs, repeat TA measurements were performed, where the same aliquot of RCs was reinterrogated following a freeze/storage/thaw cycle.

The yield of ET to HB in Arg-bg1, Arg-bg2, and template bg1 and bg2 RCs was determined via TA spectroscopy and was based on comparison to the ∼100% yield of ET from P* to HA in the W(M252)V mutant (denoted WTV), RCs which lack QA but otherwise are WT. In Fig. 3A, the P+HA− spectrum of WTV RCs (black) shows the maximal bleaching of the Qx absorption manifold of HA, with Qx(0,0) and (1,0) vibronic bands at ∼542 and ∼507 nm, respectively. The HA anion has a broad absorption band with a peak at ∼665 nm. The orange, red, and blue spectra show comparable results for formation of P+HB− from 3 BRs of Arg-bg2 RCs. To make the comparisons shown in Fig. 3A, the ∼0.3-ps P* spectrum of each Arg-bg2 BR was first normalized to the P bleaching at ∼600 nm in the ∼0.3-ps P* spectrum of WTV. The normalization factor was then applied to the TA spectrum of the Arg-bg2 BR acquired at a (longer) time, where the amount of P+HB− was maximal. For each Arg mutant, the yield of P+HB− relative to the ∼100% yield of P+HA− in WTV was then determined by comparing the integrated magnitude of maximal bleaching of the HB Qx(0,0) and (1,0) vibronic manifold to that for WTV RCs (Fig. 3 A, Inset and SI Appendix, Fig. S1). The same P+HB− yields were obtained by comparing the peak amplitude of the HB anion band (664 nm) of the mutants versus the HA anion band (665 nm) of WTV (Fig. 3A).

Fig. 3.
Fig. 3.

(A) Ultrafast TA spectra for one BR of purified RCs of WTV and 3 BRs of Arg-bg2 and yields of P+HB− (Arg variants) or P+HA− (WTV). A, Inset shows the integrated Qx bleaching manifold for Arg-bg2 BR13 (gold) and WTV BR3 (gray). (B) P+HB− yield as a function of culture density of the cells from which RCs were isolated, for the BRs of Arg-bg1 and Arg-bg2 (green) and templates bg1 and bg2 (black). Three of the points for Arg-bg1 and 9 of the points for Arg-bg2 are the average values from pairs of repeat TA experiments, as listed in SI Appendix, Table S1. The average error in the yield of P+HB− determined from these 12 pairs of repeat measurements is ±3%. SI Appendix, Table S1 Graphic reproduces B, includes the individual error bars, and labels the points with their respective BR numbers given in SI Appendix, Table S1. (C) pH of R. sphaeroides cultures (7.05 at inoculation) as a function of culture density. C, Inset is a space-filling model of a portion of the R. sphaeroides RC protein (PDB ID code 1PCR) showing native LeuL185 (red) along with cofactors HB (cyan), BB (black), and QB (silver), including their phytyl chains (not shown in Fig. 1). An expanded version of C, Inset is given in SI Appendix, Fig. S64.

The largest Qx integration value from 4 BRs of WTV RCs was taken to reflect 100% yield of ET from P* to HA. The other 3 BRs of WTV gave values of 97%, 97%, and 93% (SI Appendix, Table S1). This narrow range indicates high precision of the TA experiments and Qx bleaching integration method and/or only small natural variations of WT RC samples. Note that charge separation in WT R. sphaeroides RCs has a near-unity absolute quantum yield (22, 23). A 3- to 4-ps time constant for P* ET to BA (2⇓–4) in the native RC competing against 100- to 200-ps P* ET to HB (15, 19, 24, 25) and 100- to 200-ps P* internal conversion (16) determines that P* in the WT RC decays ∼2% by ET to HB, ∼2% to the ground state, and ∼96% by ET to HA (via BA). As noted above, we took a value of 100% for A-side ET in WT in order to simplify comparison with the mutants.

Larger variability of the P+HB− yield was found for 3 initial BRs of Arg-bg2 RCs (75%, 87%, and 60%), ultimately prompting measurements on 15 BRs of RCs from Arg-bg2 and 6 BRs of Arg-bg1 (SI Appendix, Table S1). Additionally, for 9 BRs of Arg-bg2 and 3 BRs of Arg-bg1, repeat TA measurements were performed on the same aliquot of RCs after a freeze/storage/thaw cycle (SI Appendix, Table S1). The reproducibility of the yield of P+HB− determined from the 12 pairs of repeat TA experiments on BRs of Arg-bg1 and Arg-bg2 is ±3%. The sum total of results on the BRs (and repeat measurements) reveals that the highest (average ∼90%) P+HB− yields are obtained for Arg-bg2 and Arg-bg1 RCs from midstationary phase cultures, and the lowest (60 to 70%) yields are obtained from late stationary phase cultures (SI Appendix, Table S1), which have higher culture density (Fig. 3B) and pH (Fig. 3C). Changes in RC spectral and kinetic properties with conditions (e.g., detergent and temperature) are well known (26⇓⇓–29). However, to our knowledge, dependence of RC photochemistry on culture density has not been reported. The underlying origin of these observations is not known at this time and is beyond the scope of this work.

The native Leu at L185 is positioned near one face of HB (Fig. 3C) and is C2-symmetry related to LeuM214 (Fig. 1B). One model of an Arg at L185 (in PyMOL) in the R. sphaeroides Protein Data Bank (PDB) ID code 1PCR structure (1) centers the Arg Nη atom ∼4 Å from each of the 4 pyrrolic nitrogen atoms of HB and ∼6 Å from the ring-V keto group of HB. Fig. 3A shows that the positions of the Qx(0,0) and Qx(1,0) bands of HB in the ArgL185 mutants are blue shifted compared to those of HA and do not vary as a function of P+HB− yield (or culture density).

The inherent (native) differences between the positions of the Qx absorption bands of HA and HB have been described (17, 30⇓⇓⇓–34). In brief, the A-side native GluL104 (thought to be neutral) forms a hydrogen bond to the ring-V keto group of HA. Removing the hydrogen bond to HA (substituting Leu for Glu at L104) blue shifts the Qx bands of HA by ∼10 nm (30). On the B side, the residue that is related by C2 symmetry to GluL104 is ThrM133, which is changed to Glu here in bg1, bg2, Arg-bg1, and Arg-bg2 RCs. Glu substituted at M133 is also thought to be neutral and make a hydrogen bond to HB because this substitution is shown in prior work (17, 31⇓⇓–34) to red shift the Qx bands of HB by ∼7 nm (SI Appendix, Table S2). We find here that the presence of ArgL185 results in a blue shift of the Qx bands of HB by ∼7 nm in Arg-bg1 and Arg-bg2 RCs compared to bg1 and bg2 RCs (SI Appendix, Figs. S2 and S3 and Table S2). Similarly, the presence of ArgL185 in an analog that lacks GluM133 also gives a blue shift of the Qx band of HB (by ∼5 nm) compared to the control RC with the native LeuL185 (SI Appendix, Table S2). At this time, we have no other independent information for bg1, bg2, or the Arg variants as to whether or when there is, or is not, any hydrogen bond to the ring-V keto group of HB.

Discussion

Overview of the Analysis and Model.

It is clear that an Arg at L185 influences the absorption properties of HB, that this effect is invariant with P+HB− yield (Fig. 3A), and that the high ∼90% yield of P+HB− is found for RCs isolated from cells harvested at lower culture density (Fig. 3B). Global analysis was performed on the ultrafast TA datasets (SI Appendix, Figs. S4–S11) of all 36 BRs of RCs from WTV, TrpL216-WTV, bg1, bg2, Arg-bg1, and Arg-bg2. The global analysis explored 5 to 9 exponentials (depending on sample) to obtain decay-associated difference spectra (DADS) (SI Appendix, Figs. S12–S57). The results were input into target analysis to obtain species-associated difference spectra (SADS) (SI Appendix, Figs. S58 and S59).

Even for WT RCs, it is well known that P* decay is not fit well by a single exponential. For Arg-bg1, Arg-bg2, bg1, and bg2 RCs, P* decay requires (minimally) 3 kinetic components with average lifetimes of P1* ∼4 ps, P2* ∼20 ps, and P3* ∼150 ps (SI Appendix, Fig. S60). Representing P* decay by 3 exponentials is the simplest distillation of what is likely to be a distribution of populations. Fig. 2 C–E gives a model for 3 populations having different free-energy orderings of the initial CS states with respect to an associated P* (P1*, P2*, and P3*). Further, each state in the populations is viewed as having an intrinsic range of free energies, as depicted by the vertical size of the boxes in Fig. 2 A and B (but not shown in Fig. 2 C–E). Distributions can derive from both static and dynamic properties of the protein and cofactors and both have been invoked to explain multiexponential P* decay profiles in WT and mutant RCs (4, 7, 8, 33, 35⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓–46).

Fig. 4A shows a correlation between a higher yield of P+HB− and a shorter amplitude-weighted average lifetime of P1*, P2*, and P3*. Insight into this correlation is provided by Fig. 4B and SI Appendix, Table S3. As the P+HB− yield decreased from ∼90 to ∼60%, the proportion of P3* (∼150-ps lifetime) increased from ∼10 to ∼25%. An ∼5% decrease in the abundance of P1* (∼4-ps lifetime) and an ∼10% decrease of the abundance of P2* (∼20-ps lifetime) accounted for an ∼15% increase in the amount of P3*. The results are the same for the Arg-bg1 and Arg-bg2 variants.

Fig. 4.
Fig. 4.

P* decay properties in RCs from the 15 BRs of Arg-bg2 (filled symbols) and 6 BRs of Arg-bg1 (open symbols). A shows the yield of P+HB− versus the amplitude-weighted average (AWA) P* lifetime. B gives the fractional contribution of the 3 P* decay components (comp.) (Fig. 2 C–E) versus the yield of P+HB−. A plots column 5 versus column 6 of SI Appendix, Table S3. B plots column 6 versus columns 2 (red, P1*), 3 (blue, P2*), and 4 (black, P3*) of SI Appendix, Table S3. The reproducibility of the yield of P+HB− determined from the 12 pairs of repeat TA experiments on BRs of Arg-bg1 and Arg-bg2 is ±3% (SI Appendix, Table S1).

We hypothesize that the presence of ArgL185 results in net stabilization of P+HB− (Fig. 2). Arg is considered normally to be protonated (pKa = 12.7) (47). For example, in the RC, conserved and symmetry-related ArgL135 and ArgM164 residues have their Nη atom ∼7.0 Å from the ring-V keto group (and ∼13 Å from the central Mg) of PB and PA, respectively. Both of these Arg residues are thought to be protonated, because replacing either Arg with Leu lowers the P/P+ potential by 15 mV, and replacing either by Glu lowers it by 35 mV (48). The average P oxidation potential from a sampling of Arg-bg1 and Arg-bg2 RCs is ∼25 mV higher than for bg1 and bg2 RCs (SI Appendix, Table S4). The very close proximity of ArgL185 and HB was noted above. If ArgL185 is protonated, the positive charge could make HB significantly easier to reduce and (by inference, more than) offset the apparent increase in P oxidation potential. However, we have no direct knowledge, and draw no conclusion, as to whether ArgL185 is protonated in Arg-bg1 and Arg-bg2 RCs, or whether the protonation state of ArgL185 might vary among BRs of Arg-bg1 and Arg-bg2 RCs. Potential differences in water access to HB or other structural or electrostatic effects associated with residues other than ArgL185 also could affect the free energy of P+HB−. Regardless, Fig. 3B and our modeling analysis (Figs. 2 and 4) imply that stabilization of P+HB− by some means occurs to a greater degree or in a greater percentage of the RCs in samples with the higher P+HB− yields.

Global and Target Analysis.

Using 2 representative BRs of Arg-bg2 and one BR of bg2, here we describe the distinctive DADS associated with P1*, P2*, and P3* that underlie the model in Fig. 2 C–E. The ∼4-ps DADS (Fig. 5 A and E) show characteristics consistent with 2-step P1* → P+BB− → P+HB− ET. These included features (with appropriate signs) for P+ (1,250 nm), a BChl anion (1,030 nm), both BChl and BPh anions (600 to 700 nm), BChl bleaching (600 nm), and BPh bleaching (529 nm for Arg-bg2 or 536 nm for bg2). In contrast, the ∼20-ps DADS (Fig. 5 B and F) are dominated by characteristics for P2* → P+HB− conversion by superexchange ET. These include features for formation of HB bleaching and HB− and P+ absorption, along with substantial diminution of features for the BChl anion (see also below). The ∼150-ps DADS (Fig. 5 C and G) are comprised of features associated with P3* internal conversion to the ground state (decay of P bleaching at ∼600 and 830 to 870 nm). Such features are basically not present for P1* and are present for P2*, but to a much smaller extent. The ∼150-ps average P3* lifetime for the R. sphaeroides mutants studied here (SI Appendix, Table S5) is similar to the ∼180- to 200-ps P* internal conversion time measured for the R. capsulatus DLL mutant RC where no ET occurs (16, 49) and for other R. capsulatus mutant RCs where P* → ground state competes effectively with various ratios of A- versus B-side ET (15, 19).

Fig. 5.
Fig. 5.

(A–C and E–G) Comparison of visible (A–C) and near-infrared (E–G) DADS for the 3 P* decay components for Arg-bg2 BR13 (orange), Arg-bg2 BR3 (blue), and bg2 BR4 (black) RCs: ∼4-ps DADS (A and E), ∼20-ps DADS (B and F), and ∼150-ps DADS (C and G). (D and H) DADS associated with P+HB−, which forms with a yield of 0.91 (Arg-bg2 BR13), 0.61 (Arg-bg2 BR3), and 0.73 (bg2 BR4); the DADS for bg2 is the sum of 3- and 12-ns subcomponent spectra.

The different decay modes for P1*, P2*, and P3* indicated by the distinctive DADS underlie the 3-population model and differences between the relative free energies of the states seen in Fig. 2 C–E, respectively. A 3-population distillation is, again, a simplification of a distribution of P* populations. Results of target analysis indicate that the first step of P1* → P+BB− → P+HB− ET has a time constant of ∼4 ps and that the time constant for the second step is similar or slightly longer. Not shown in Fig. 2 C–E, but included in the model for target analysis (SI Appendix, Fig. S58), is a small (∼10%) amount of P+HB− forming from P+BB− on the 20- to 30-ps timescale. For simplicity, this avenue of forming P+BB− was modeled as arising from P2* rather than introducing a fourth P* population. Potential contributions of both a “fast” and a “slow” 2-step process have been modeled for initial A-side ET in WT RCs (4). These points reinforce the complexity of the photochemistry and the DADS. Target analysis (SI Appendix, Fig. S58) returned nearly identical SADS (SI Appendix, Fig. S59) for P1*, P2*, and P3* that have the expected characteristics, notably, including absence of features for CS states, illustrating the consistency of the analysis.

The DADS associated with decay of the other products of P* decay—P+βA− (SI Appendix, Figs. S61 and S62) and P+HB− (Fig. 5 D and H)—show the expected features. The yield of P+βA− is 5% or less for Arg-bg2 RCs and 15 to 20% for bg2 RCs. P+βA− has an ∼1-ns lifetime and appears to decay in part by thermal repopulation of P* (SI Appendix, Fig. S58). Because QB is not present in the samples studied here, P+HB− decays by charge recombination partly to give the ground state (80 to 85% yield) and partly to give the triplet excited state (PR) in 15 to 20% yield, which in turn decays by energy transfer to the carotenoid to give CarT (SI Appendix, Figs. S61–S63). The P+HB− decay is fit adequately by a single exponential with a time constant of ∼12 ns for Arg-bg1 and Arg-bg2 RCs, but is dual-exponential with time constants of 3 and 12 ns for bg1 and bg2 RCs. The virtual absence of the ∼3-ns P+HB− decay component in Arg-bg1 and Arg-bg2 RCs can be explained by stabilization of P+HB− by ArgL185 (Fig. 2B versus Fig. 2A), reducing quantum/thermal mixing with P* and/or P+BB−. We have discussed such mixing previously for various R. capsulatus mutants (34, 50, 51). Analogous mixing of A-side state P+HA− (52⇓⇓–55) or state P+βA− (18, 19, 56, 57) with P+BA− and/or P* has also been discussed.

Collectively, our results show that replacing Leu185 with Arg has more than one effect that supports B-side charge separation. 1) ArgL185 complements the template mutations (notably, changing TyrM210 to Phe) in significantly slowing ET to the A side (SI Appendix, Table S5). This effect can be traced to the ∼25-mV higher P/P+ potential when L185 was Arg versus the native Leu (SI Appendix, Table S4). This will elevate P+BA− and P+βA− in free energy by ∼25 meV. 2) While the B-side CS states will be similarly elevated in free energy by the increased P oxidation potential, our results indicate (as mentioned above) that there is a larger local stabilizing effect of Arg on the free energy of P+HB−. Specifically, the presence of ArgL185 makes HB (and perhaps BB) easier to reduce. This underlies the increased yield of P+HB− due to a larger contribution of 2-step ET and an increased superexchange rate constant as well. A pronounced local effect on HB is reasonable in view of PyMOL models that replace LeuL185 with Arg in the R. sphaeroides crystal structure, which indicates that ArgL185 could be only ∼4 Å from each of the 4 central pyrrolic nitrogen atoms of HB. 3) Replacing Leu by Arg at L185 slows P+HB− charge recombination to ∼12 ns. This is the longest time constant for this process found to date and is comparable to the ∼15-ns A-side P+HA− charge recombination time (when QA is absent) (58, 59). Slower P+HB− charge recombination and the F(L216)W mutation in the bg2 template both should improve P+HB− → P+QB− ET.

Unifying RC Distributions, Relaxations, and Complex Dynamics.

The electronic structure of P and how it is perturbed by excitation may link many complexities of RC dynamics. The logic behind this follows. Protein motions afford a fluctuating environment for the RC cofactors, temporally modulating their electronic structures and redox properties. At a snapshot in time, these will differ among RCs in a sample. At the instant of excitation, the ensemble will be mapped onto a P* potential energy surface(s), and subsequent protein-cofactor motions can alter the distribution of populations. Protein motions may limit ET rates (40). They may also contribute to free-energy relaxations of P+HA− ranging from <1 to >10 ns when ET to QA is blocked (52, 53, 55, 60, 61). Complex kinetics were also found for P* → P+BA−→ P+HA− and P+HA− → P+QA− in WT RCs, implying multiple populations (4, 37). P* populations that exhibit different photochemistry, such as seen here, also have been described for various mutants (9, 33, 44, 62). These findings, the data here, and the model in Fig. 2 C–E can be unified by considering a trigger that abruptly alters the fluctuating ensemble of protein/cofactor interactions/energetics at the instant of excitation of P, over and above effects due simply to adding excitation energy.

A change in the physical/electronic structure of P upon excitation might be such a trigger. This will simultaneously affect the free-energy ranges (i.e., the vertical sizes of the boxes in Fig. 2 A and B) of all of the electronic states of the RC (Fig. 2 C–E) because they all involve P. The electronic structure of P can be viewed in analogy to the properties of strongly coupled tetrapyrrole dimers (63⇓⇓⇓⇓–68). The highest occupied molecular orbital (HOMO) of P is an antibonding MO at higher energy than the HOMOs of the constituent BChls (PA and PB) due to interactions between the 2 BChls. (This makes P easier to oxidize than a BChl monomer.) The dimer lowest unoccupied molecular orbital (LUMO) is a bonding combination of the LUMOs of PA and PB. Excitation of P to P* promotes an electron from the antibonding HOMO to the bonding LUMO and shifts electron density from the outward faces of PA and PB into the region between PA and PB. This increases the interaction between PA and PB, which reduces the PA−PB distance and raises the HOMO in energy. This increase in HOMO energy makes P* easier to oxidize over and above the general effect that an excited state of a molecule is easier to oxidize than its ground state. Because P* is easier to oxidize (than it would have been had the PA−PB distance not been reduced upon excitation), the free energies of all of the CS states will be lower than they were at the PA−PB distance in the ground state. Ensuing oxidation of P* to P+ via ET removes the electron in the bonding LUMO, shifting the PA−PB distance to a value intermediate between those for P* and P, raising the P/P+ potential and free energies of all of the CS states, again potentially affecting the vertical sizes of the boxes in Fig. 2 A and B. Any of these effects, starting at excitation of P to P*, could promote protein dynamics, some of which might contribute to dielectric relaxations in response to charge separation (55, 61).

Motions that alter the interactions and distance between the 2 macrocycles of P have been proposed to be important for the spectral and redox properties of P and for P* ET dynamics (45, 65, 69⇓⇓–72). Such motions also likely underlie the much shorter (100 to 200 ps) time constant for P* internal conversion compared to that (∼5 ns) for excited monomeric BChl* (73). A related view is that fluctuations of distance between the 2 macrocycles of P alter the admixture of the charge-resonance and exciton (or charge-transfer and locally excited) configurations of P*. Thus, the electronic structure of P and how it is perturbed by excitation may link many complexities of RC dynamics. This includes the nature of the distributions that are central to our analysis of how the very high, ∼90%, yield of B-side charge separation is attained.

Two-step P* → P+BB−→ P+HB− ET mimics P* → P+BA−→ P+HA− charge separation on the A side in the native RC. The ∼90% yield of P+HB− is achieved by comparable contributions of this 2-step process and slower (∼20 ps) P* → P+HB− superexchange. A residual ∼10% of P* decay occurs mainly by internal conversion to the ground state. These results have many interesting consequences that should provoke additional experiments and theoretical analysis. It is possible that RCs employing comparable contributions of the 2 mechanisms for A-side charge separation may have existed during evolution of the photosynthetic apparatus. If such RCs (analogous to those studied here) had been a stepping stone, our results show that further tuning of the relative free energies of the CS states to convert the ∼50% superexchange population into a “2-step” population would gain only the last 5 to 10% of the near-unity yield of A-side charge separation.

Materials and Methods

Complete descriptions of sample preparation, static and time-resolved optical spectroscopy, and analysis of TA datasets are given in SI Appendix.

Data Availability.

All data needed to support the conclusions of this manuscript are included in the main text and SI Appendix. R. sphaeroides strains are available upon request.

Acknowledgments

We thank Dr. David Bocian for helpful discussion and Abby Gibbons for measurements of pH and densities of cultures. This work was supported by the US Department of Energy, Office of Basic Energy Sciences Grant DE-CD0002036 (to C.K. and D.H.) and associated Field Work Proposal (P.D.L.). Argonne, a US Department of Energy Office of Science laboratory, is operated under Contract DE-AC02-06CH11357.

Footnotes

  • ↵1To whom correspondence may be addressed. Email: kirmaier{at}wustl.edu.

  • Author contributions: P.D.L., D.K.H., D.H., and C.K. designed research; D.K.H., J.C.B., G.A.T., K.M.F., and C.K. performed research; J.C.B., K.M.F, D.H., and C.K. analyzed data; and D.H. and C.K. wrote the paper.

  • The authors declare no competing interest.

  • This article is a PNAS Direct Submission.

  • This article contains supporting information online at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1916119117/-/DCSupplemental.

Published under the PNAS license.

View Abstract
Back to top
Article Alerts
Email Article
Citation Tools
Request Permissions
Switching sides—Reengineered primary charge separation in the bacterial photosynthetic reaction center
Philip D. Laible, Deborah K. Hanson, James C. Buhrmaster, Gregory A. Tira, Kaitlyn M. Faries, Dewey Holten, Christine Kirmaier
Proceedings of the National Academy of Sciences Jan 2020, 117 (2) 865-871; DOI: 10.1073/pnas.1916119117
Digg logo Reddit logo Twitter logo Facebook logo Google logo Mendeley logo
Proceedings of the National Academy of Sciences: 117 (2)
Table of Contents

Submit

Article Classifications

  • Physical Sciences
  • Biophysics and Computational Biology

Jump to section

You May Also be Interested in

Similar Articles