Sinjini
Bhattacharjee
a,
Srilatha
Arra
b,
Isabella
Daidone
*b and
Dimitrios A.
Pantazis
*a
aMax-Planck-Institut für Kohlenforschung, Kaiser-Wilhelm-Platz 1, 45470 Mülheim an der Ruhr, Germany. E-mail: dimitrios.pantazis@kofo.mpg.de
bDepartment of Physical and Chemical Sciences, University of L'Aquila, Via Vetoio (Coppito 1), 67010 L'Aquila, Italy. E-mail: isabella.daidone@univaq.it
First published on 9th April 2024
Photosystem II (PSII), the principal enzyme of oxygenic photosynthesis, contains two integral light harvesting proteins (CP43 and CP47) that bind chlorophylls and carotenoids. The two intrinsic antennae play crucial roles in excitation energy transfer and photoprotection. CP43 interacts most closely with the reaction center of PSII, specifically with the branch of the reaction center (D1) that is responsible for primary charge separation and electron transfer. Deciphering the function of CP43 requires detailed atomic-level insights into the properties of the embedded pigments. To advance this goal, we employ a range of multiscale computational approaches to determine the site energies and excitonic profile of CP43 chlorophylls, using large all-atom models of a membrane-bound PSII monomer. In addition to time-dependent density functional theory (TD-DFT) used in the context of a quantum-mechanics/molecular-mechanics setup (QM/MM), we present a thorough analysis using the perturbed matrix method (PMM), which enables us to utilize information from long-timescale molecular dynamics simulations of native PSII-complexed CP43. The excited state energetics and excitonic couplings have both similarities and differences compared with previous experimental fits and theoretical calculations. Both static TD-DFT and dynamic PMM results indicate a layered distribution of site energies and reveal specific groups of chlorophylls that have shared contributions to low-energy excitations. Importantly, the contribution to the lowest energy exciton does not arise from the same chlorophylls at each system configuration, but rather changes as a function of conformational dynamics. An unexpected finding is the identification of a low-energy charge-transfer excited state within CP43 that involves a lumenal (C2) and the central (C10) chlorophyll of the complex. The results provide a refined basis for structure-based interpretation of spectroscopic observations and for further deciphering excitation energy transfer in oxygenic photosynthesis.
CP43 is closest to the D1 protein, which hosts the branch of RC pigments that are active in charge separation26 and that also accommodates the Mn4CaOx cluster of the oxygen-evolving complex (OEC), the site of water oxidation. In addition to its role in EET, CP43 plays a pivotal role in maintaining the overall structural integrity of the RC and contributes to the stabilization of the OEC itself by providing a direct manganese-coordinating ligand (Glu354) as well as crucial second-sphere functionality (Arg357). As an essential core antenna complex, CP43 has been the subject of numerous studies that attempted to elucidate its spectroscopic properties and excitonic structure.13,27–31 CP43 contains 13 embedded Chl a pigments and 4 β-carotenes. Fig. 2 depicts the spatial arrangement of the CP43 chlorophylls and indicates the labeling used in the present work, which follows the numbering recommended by Müh and Zouni.18Table 1 describes distinctive characteristics of each chlorophyll in terms of their axial ligation and the hydrogen-bonding interaction at the 131-keto group. Additionally, to facilitate comparisons with previous studies that follow structure-specific numbering of CP43 chlorophylls, Table 1 lists the corresponding numbering of the CP43 chlorophylls in selected PSII crystallographic models. As shown in Fig. 2, the chlorophylls are distributed in three layers, with four Chl molecules present in the lumenal layer (C1–C4), eight in the stromal layer (C5–C9, C11–C13), and one (C10) positioned in the center of the lipid bilayer. This arrangement is reminiscent of the distribution of chlorophylls in the CP47 antenna.13,32
![]() | ||
Fig. 2 The pigments of the CP43 complex with their labels: (a) side view, indicating also positions of carotenoids; (b) stromal (“top”) view, depicting also proximal pigments of the reaction center that belong to the D1 (PsbA) chain; (c) lumenal (“bottom”) view. Chlorophyll C10 is shown in both panels on the right to aid orientation. C5 is the closest CP43 chlorophyll to D1 pigments PheoD1 and ChlzD1 (ca. 21–22 Å, see ESI† for details). |
Site | Axial | Keto H-bond | Location | 2AXT33 | 3BZ1 (ref. 34) | 3WU2-A35 | 3WU2-B | 4IL6 (ref. 36) |
---|---|---|---|---|---|---|---|---|
C1 | Hisε237(α) | H2O | L | 33 | 474 | 501 | 902 | 501 |
C2 | Hisε430(α) | Tyr297 | L | 34 | 475 | 502 | 903 | 502 |
C3 | Hisε118(α) | — | L | 35 | 476 | 503 | 904 | 503 |
C4 | H2O(α) | H2O, LMG-519 | L | 37 | 477 | 504 | 905 | 504 |
C5 | Hisε441(α) | H2O, Arg449 | S (close to D1) | 41 | 478 | 505 | 906 | 505 |
C6 | Hisε251(α) | — | S | 42 | 479 | 506 | 907 | 506 |
C7 | H2O(α) | Hisδ164 | S | 43 | 480 | 507 | 908 | 507 |
C8 | Hisε444(α) | — | S (close to D1) | 44 | 481 | 508 | 909 | 508 |
C9 | Hisε53(β) | Ser275 | S | 45 | 482 | 509 | 910 | 509 |
C10 | Hisε56(α) | — | Center (close to PsbK) | 46 | 483 | 510 | 911 | 510 |
C11 | Asn39(α) | H2O, Arg41 | S (close to PsbK) | 47 | 484 | 511 | 912 | 511 |
C12 | Hisε164(α) | H2O | S | 48 | 485 | 512 | 913 | 512 |
C13 | Hisε132(α) | Tyr131, LMT 102 | S | 49 | 486 | 513 | 914 | 513 |
Spectroscopic characterizations of intrinsic antenna complexes CP43 and CP47 are often performed on samples extracted from PSII. In the case of isolated CP43 samples thus only 13 chlorophylls contribute to the spectra, facilitating analysis and fitting of the spectra to some extent. However, isolated samples depart from the native structure and possibly lack the structural integrity of the PSII-complexed system, potentially introducing inconsistencies in the resulting data sets. Additionally, it is possible that isolated preparations or certain treatments may result in deformation or even loss of one or more chlorophylls, as suggested for CP47.32
The CP43 core antenna in PSII is anticipated to possess two quasi-degenerate low-energy “trap” states,31,37 inferred from hole-burning (HB) studies,38 other spectroscopic investigations39,40 and structure-based calculations.27–29 However, the exact assignment of these low energy excitonic states remains a topic of active debate in literature. Shibata et al. reported the presence of two red-shifted pigment domains within the PSII core complex emitting at 685 nm and 695 nm,16 the former assigned to CP43. Previously, Hughes et al. claimed that both states are localized on one chlorophyll, but one is excitonically coupled to other states.37 On the other hand, Raszewski, Renger and Müh have associated one CP43 trap state with a localized exciton in the lumenal layer whereas the other trap state was concluded to be on a delocalized domain in the stromal layer.29,41
The kinetics of EET are also debated, with different groups arriving at significantly different values for the transfer times from the CP43/CP47 antennae to the RC compared to the rate of primary charge separation (CS) at the RC.15,16,42,43 The conclusions depend on the distinct assumptions and theoretical models employed. For example, Holzwarth and co-workers reported timescales of a few ps for EET to the RC based on transient absorption and fluorescence kinetics,44,45 whereas Renger arrived at estimates approximately an order of magnitude slower for the forward and half of that for the backward EET process.41,46 This also led to alternate EET mechanisms being proposed, namely the “exciton-radical pair equilibrium” (ERPE) model30,44 and the “transfer-to-trap limited” (TTTL) model.45,47 It has also been suggested that CS and EET may occur on the same timescale.14 More recently, Yang et al. investigated the EET dynamics of the PSII core complex using two-dimensional electronic-vibrational (2DEV) spectroscopy19 and suggested that C5 (current labeling) in CP43 and the peripheral D1 chlorophyll known as ChlzD1 likely form the pathway for energy transfer from CP43 to the RC. The results were consistent with the TTTL model in the sense that EET from CP43 to ChlzD1 was found to be faster than subsequent EET to other D1 pigments, a step which thus constitutes the kinetic bottleneck.
Spectroscopic studies towards determining possible EET pathways in photosynthetic light harvesting complexes (LHCs) still face two main challenges.17,48,49 First, the closely spaced pigments exhibit significant excitonic couplings rendering it impossible to make a direct correlation between the absorption bands and individual pigments. Second, the highly congested excitonic manifold makes it non-trivial to assign site energies to specific Chl molecules. That is why theoretical approaches that complement experimental data (e.g. spectral densities, absorption and fluorescence spectra) by simulating the structures and estimating the site energies of pigment–protein complexes have long played an important role in the study of antenna complexes.25,28,29,50–58 Past investigations employed quantum mechanics/molecular mechanics (QM/MM) and molecular dynamics (MD) simulations based on predetermined potential energy surfaces (PES) derived from DFT,59 or “on-the-fly” PES obtained from the semi-empirical DFTB approach.60 However, it remains challenging even with approximate QM methods to perform state of the art simulations on groups of chlorophyll pigments in systems such as CP43 and CP47 while simultaneously taking fully into account the short- and long-range effects of the protein matrix. Furthermore, even in existing theoretical studies there have been discrepancies regarding the relative ordering of site energies and identity of the low energy trap states.38,39 In two investigations, Müh, Renger and co-workers identified three red chlorophylls in the isolated CP43 protein.28,29 The excitonic couplings and local transition energies of chlorophylls were computed using Poisson–Boltzmann (PB) electrostatics in both cases, albeit the latter study was based on the high-resolution crystal structure of PSII.28,29 Saito et al. evaluated EET coupling between Chls in the PSII monomer based on the QM/MM diabatization scheme.61 In a more recent computational study, Sarngadharan et al. determined the site energies and excitonic couplings of CP43 chlorophylls27 using time-dependent long-range corrected density functional tight binding approach (TD-LC-DFTB) and QM/MM MD trajectories. All the above theoretical studies seemed to have reached a consensus that the red-most chlorophylls in CP43 are delocalized and likely belong to two separate “domains” of the transmembrane region. It is also important to note that all or most of the theoretical studies are based on the isolated CP43 protein, without the remaining PSII matrix. Moreover, axial ligation to the central Mg atom, pigment–protein interactions (e.g., H-bonding to the keto group), and pigment–pigment interactions (e.g. in Chl dimers) are known to directly influence excited state properties of chlorophylls,32,62,63 it is therefore crucial to treat these interactions at the QM level along with the chromophore, an aspect which has been overlooked in several past studies.
Here we utilize a comprehensive large-scale QM/MM approach employing range-separated time-dependent density functional theory (TD-DFT) to investigate the low-energy excited states for all CP43 chlorophylls. The calculations explicitly account for interactions of the pigments with the complete membrane-embedded all-atom PSII monomer. Beyond individual pigments, we apply the same methods on pigment dimers in order to study coupled pairs and to investigate the presence of possible charge-transfer (CT) states.62,64,65 Crucially, our work incorporates the perturbed matrix method (PMM)66–69 that enables us to leverage the full information from long-timescale molecular dynamics simulations for the extraction of site energies and excitonic couplings. In addition to a new set of site energies and excitonic couplings, the results identify a previously unknown low-lying state with significant charge transfer character among CP43 chlorophylls, which may have important implications for the functional role of the protein. Combined with a refined analysis of the contributions of the different monomeric and dimeric pigment groups to the first exciton, the present results provide a detailed map of static and dynamic properties of the CP43 pigments and contributes to the improved understanding of this essential photosynthetic antenna.
Ĥ = Ĥ0 + ![]() | (1) |
![]() ![]() | (2) |
In case of the interacting chromophores, i.e., a set of interacting QCs, we may consider the possible excitation coupling occurring among the QCs.87 In the present case, the interactions between the electronic excitations localized on each Chl are considered, with the exception of C2–C10 and C7–C9 that were considered as dimers due to their electronic coupling (see above). To this aim, the perturbed Hamiltonian operator for the 11 QCs (9 single chlorophylls and 2 dimers) is considered and, in matrix notation, is expressed as follows:
![]() ![]() | (3) |
![]() | (4) |
Finally, by diagonalizing the excitation matrix, the perturbed excitation energies and eigenstates (i.e., the exciton states) are obtained. In the present case we utilized 3000 frames, corresponding to the final 30 ns of a previously performed 200 ns-long MD trajectory,64,91 saved at intervals of 10 ps. This specific time frame was previously employed for the redox potential calculations65 and was selected to ensure a reasonably converged root mean square deviation (RMSD) of the entire system with respect to the starting, crystallographic structure. In principle, there is no computational constraint on the number of frames that can be utilized within the MD-PMM framework, as it solely depends on the length of the MD trajectory.
The PMM approach has previously been employed in the study of chlorophylls and pheophytins of the PSII-RC.65 This application demonstrated very good agreement with experimentally-derived thermodynamic parameters such as reduction potentials, and with experimental kinetics properties such as rate constants of primary electron transfer within the RC. Here, we further validated the MD-PMM approach by computing the absorption line shape of CP43 at room temperature (see Fig. S2†). This was accomplished using our previously-established procedure, which is outlined in the ESI.† Notably, there is a close match with the experimental line shape. Specifically, the experimental exciton splitting at 293 K (that manifests as a slight shoulder at low energy) is accurately reproduced (12 nm in the computed spectrum versus 13 nm in the experimental one). The total bandwidth is slightly overestimated (34 nm versus the experimental 24 nm), most probably due to an overestimation of the inhomogeneous broadening provided by the perturbation calculated along the MD trajectory.
The geometries of all individual chlorophylls were optimized using DFT and QM/MM as described in the Methodology. Following complete optimizations, we computed the vertical excitation energies of each CP43 chlorophyll in vacuo (also referred to commonly as “in the gas phase”) using TD-DFT on the QM/MM optimized geometries. Table 2 provides a comprehensive overview of the vertical excitation energies of the lowest excited state (S1 or Qy), and the protein electrochromic shifts of each CP43 chlorophyll. We find that the in vacuo first excited state energies lie within the range of 1.868–1.966 eV (see Table 2). This demonstrates that intrinsic structural features of the pigment itself, like the macrocyclic ring curvature and the nature of axial ligation, play already major roles in differentiating the site energies of each chlorophyll. We subsequently performed excited state calculations on the CP43 protein excluding the rest of the PSII monomer, so that we can compare the effect of having an “isolated” CP43 chain as opposed to a complete PSII monomer (we emphasize that this is not equivalent to simulating an experimentally extracted CP43 antenna because we are not simulating here the conformational state of the latter, a problem that presents a distinct challenge). The results show that the local protein electrostatics already shift the Chl site energies toward the red or blue regime relative to their gas-phase values (Table 2). Overall, the site energies for the isolated CP43 lie in the range of 1.850–1.998 eV. The six chlorophylls C1, C2, C3, C5, C8 and C9 are red-shifted, whereas the remaining C4, C6, C7, C10, C11, C12 and C13 chlorophylls are blue-shifted with respect to the gas-phase excitation energies (Fig. 3). Interestingly, 3 out of the 6 red-shifted chlorophylls are on the lumenal side of the protein (Fig. 2c). These findings are consistent with the stromal versus lumenal (membrane-transverse) trends obtained for the pigments in the RC and the CP47 antenna of PSII.32,62 Chlorophylls C5 and C1 are the pigments that shift most to the red due to protein matrix electrostatics.
Site | TD-DFT gas-phase | TD-DFT|QM/MM isolated CP43 | TD-DFT|QM/MM full PSII monomer | PMM | ||||||
---|---|---|---|---|---|---|---|---|---|---|
E | f osc | E | f osc | Shift | E | f osc | Shift | E mean | Std. dev. | |
C1 | 1.868 | 0.22 | 1.850 | 0.20 | −18 | 1.846 | 0.19 | −22 | 1.873 | 0.005 |
C2 | 1.879 | 0.22 | 1.870 | 0.20 | −9 | 1.867 | 0.20 | −12 | 1.894 | 0.010 |
C3 | 1.868 | 0.22 | 1.866 | 0.20 | −2 | 1.861 | 0.20 | −7 | 1.867 | 0.005 |
C4 | 1.913 | 0.23 | 1.927 | 0.19 | 14 | 1.928 | 0.17 | 15 | 1.945 | 0.014 |
C5 | 1.966 | 0.23 | 1.944 | 0.22 | −22 | 1.938 | 0.25 | −28 | 1.915 | 0.011 |
C6 | 1.895 | 0.22 | 1.954 | 0.21 | 59 | 1.970 | 0.21 | 75 | 1.963 | 0.026 |
C7 | 1.886 | 0.23 | 1.897 | 0.20 | 11 | 1.885 | 0.21 | −1 | 1.890 | 0.008 |
C8 | 1.893 | 0.22 | 1.883 | 0.23 | −10 | 1.915 | 0.23 | 22 | 1.895 | 0.005 |
C9 | 1.897 | 0.23 | 1.881 | 0.23 | −16 | 1.881 | 0.22 | −16 | 1.912 | 0.008 |
C10 | 1.901 | 0.22 | 1.939 | 0.23 | 38 | 1.955 | 0.23 | 54 | 1.942 | 0.013 |
C11 | 1.916 | 0.22 | 1.998 | 0.22 | 82 | 1.983 | 0.24 | 67 | 1.964 | 0.020 |
C12 | 1.922 | 0.22 | 1.938 | 0.21 | 16 | 1.926 | 0.21 | 4 | 1.925 | 0.010 |
C13 | 1.917 | 0.23 | 1.930 | 0.22 | 13 | 1.961 | 0.22 | 44 | 1.937 | 0.018 |
The identity of the red-shifted chlorophylls remains consistent when we inspect the full results for the complete PSII monomer, but the effect of the protein matrix is now significantly more pronounced. The global PSII protein electrostatics further stabilize the site energies of chlorophylls C1, C2, C3 and C5, with C5 still being the most red-shifted pigment. This result is significant for two reasons. First, C5 (C505 in 3WU2) is located on the periphery of CP43 and is in direct contact with the D1 (PsbA) subunit of PSII. Second, recent two-dimensional electronic-vibrational (2DEV) spectroscopy measurements on the PSII core complex19 suggested that chlorophyll C5 along with the peripheral chlorophyll ChlzD1 likely mediate EET from CP43 to the PSII-RC. The electrochromic shift on chlorophyll C9 (16 meV) remains invariant in the isolated CP43 and in the PSII monomer. This is expected as well because it is deeply buried within the CP43 protein matrix and is the pigment least affected by structural deformation of CP43 upon isolation. Another interesting result is seen for C8, the only pigment to show opposite shifts in the intact PSII monomer compared to isolated CP43. C8 is also located close to the D1 (PsbA) subunit, making it susceptible to conformational changes during extraction from the native PSII. The D1 electrostatics induce an additional blue-shift in the excitation energy of C8. The remaining blue-shifted chlorophylls (C6, C11, C13) are mostly located in the stromal layer of CP43, except for C10 which is in the middle of the transmembrane region. C6 and C13 lie on the periphery of CP43 while C10 and C11 are located close to the PsbK subunit (Fig. 2b). It is suggested that the salt bridge between D2-R233 and CP43-E29 may affect the interaction between C11 and the protein, thus lowering its site energy slightly in the intact PSII compared to extracted samples.28
Based on the results on CP43 chlorophylls in the intact PSII monomer, the computed site energies lie in the range of 1.846–1.983 eV while the second excited states (S2/Qx) range from 2.260–2.468 eV (see Table S1†). Moreover, the energy gap between the first two excited states is also seen to vary among the different domains, with the lowest being for C1 and C2 in the lumenal domain, and the highest being for site C8 in the stromal layer close to PsbA (D1). The Qx–Qy energy gap is modulated by both intrinsic (macrocyclic ring curvature) and extrinsic components (axial ligation, H-bonding, protein electrostatics), as also seen for the case of CP47 and RC chlorophylls.32,92
It is noted that the differences in site energies of the CP43 chlorophylls do not exclusively arise from protein electrostatics, as significant shifts are observed for the pigments in their QM/MM optimized geometries when the excited state calculations are performed in vacuo. We analyzed the contribution of these other effects through excited state calculations on the pigments in vacuo, but without the influence of the axial ligation and H-bonding interactions. Differences in the Qy energy with that of optimized Chl a give the contribution of the QM/MM geometry optimization on the site energies. The results are summarized in Fig. S3.† Our results suggest that protein electrostatics still have the dominant contribution to site energy shifts, but there are significant contributions from the QM/MM geometry optimizations (i.e., intrinsic structural differentiation), especially contributing to the blue-shift of Chls C4 and C9–C13. The axial ligands do not affect the results significantly, but only slightly red-shift the S1 state in each case. One important finding is that CP43-R449, which is H-bonded to a water bridging the keto group of C5, individually blue-shifts its site energy to a great extent but overall, the protein matrix overcompensates and tunes it in the opposite direction. These findings resemble past observations on chlorophylls of CP47, which also had a similarly wide distribution of site energies even in the absence of protein electrostatics.32 Results on both antennae are in this respect different from results on the PSII RC, where all 4 Chls (ChlD1, ChlD2, PD1 and PD2) had essentially the same site energies in the absence of the protein electrostatics.62 Given that the influence of the axial ligands is not a decisive factor, it seems that the most plausible explanation for the difference is that the protein matrix allows greater flexibility for the CP43/CP47 chlorophylls, whereas the RC pigments need to be more rigid to allow for efficient electron transfer.
An effective approach for assessing the spectral density of a chromophore within a photosynthetic pigment–protein complex (PPC) involves simulating the fluctuations in site energy resulting from environmental effects.50,53,93 However, current limitations arise from the substantial size and the intricacy of these complexes and from the impossibility of simulating a membrane embedded PPC immersed in a water box while applying continuously QM methods for meaningful sampling.94–96 Therefore, to account for the effect of conformational changes of the PSII protein environment on the individual site energies, the site energies of individual Chls were also calculated by means of the MD-PMM procedure. The energies are obtained at each time frame of the MD simulation over a specific interval. The corresponding site energy distributions are reported in Fig. S4.† It can be observed that the distributions of the individual pigment molecules differ in their peak positions, widths, and skewness, although all distributions are roughly Gaussian in shape. The corresponding mean site energies are also reported in Table 2 and compared with the results obtained by means of the TD-DFT calculations. Fig. 3b shows that the two sets of site energies have the same overall trend and that also the quantitative agreement is quite high, with maximum absolute deviations (MAD) of 20–30 meV. The maximum deviations are seen for pigments C1, C2, C5, C9 and C13.
In one line of investigation, Raszewski et al. associated the “red” chlorophyll (at 685 nm) with C4 (or C504 in 3WU2) in the lumenal layer of CP43.41 In subsequent studies relying on high-resolution PSII structures,28 the other “trap” state was suggested to be a delocalized excited state involving multiple pigments (C7, C9, and C11) located in the stromal layer. These assignments of trap states were derived largely from theoretical simulations of optical spectra that employed computed excitonic couplings and adapted site energies. A somewhat different perspective was presented by Shibata et al.,16 who reported the presence of two red-shifted chlorophyll domains within PSII-cc, emitting at 685 nm and 695 nm respectively, based on time-resolved fluorescence spectroscopy. In their study, only the 685 nm band was attributed to either C9 or C7 sites within CP43. A detailed comparison of the chlorophyll site energies obtained from different literature sources is provided in Table S3 of the ESI.† It is important to note that the results presented in the current work are not directly comparable to site energies obtained by approaches that involve fitting to various experimental data sets and different theoretical methods. Nevertheless, we can compare the relative trends in site energies with respect to the chlorophyll with the lowest site energy.
Based on the site energies (obtained here using QM/MM and the PMM methodologies, see Table 2 and Fig. 3b), the identity of the red chlorophylls in CP43 differs from the previous assignments in literature (see Table S3†). We identify the red-most chlorophyll to be in the lumenal layer of CP43, with major contributions from sites C1 (C501 in 3WU2) followed by C3 and C2. The recent computational studies by Sarngadharan et al.27 reflect the same trend of the red most chlorophylls being on C1 and C3 in the lumenal layer, consistent with the current results (Table 2). The second lowest site energies are on chlorophylls C9 and C7 from the stromal layer. This is in line with the work by Jankowiak et al. on non-photochemical HB studies who first reported that the two “trap” states38 are localized in different layers of the thylakoid membrane. Another important finding by Hughes et al. identified that at least one of the red chlorophylls has the 131-keto group hydrogen bonded to a protein residue.37 Based on the 1.9 Å crystal structure, both chlorophylls C9 and C7 are H-bonded to C-Ser275 and C-His164 respectively while C1 is H-bonded to a H2O via the 131-keto group. By contrast C11 (C511 in 3WU2), a proposed site for the red most chlorophyll in some assignments, is seen to possess a significantly high first excitation energy based on our calculations. This Chl has a unique axial ligation to the sidechain of PsbC-N39 and is in close contact with PsbK (see Fig. 2) and the N-terminal loop region of CP43 exposed to bulk water. Based on the PMM results in Table 2, the site energy of C11 has a significantly high standard deviation, which denotes that changes in the local protein environment induce large fluctuations in the computed site energies. There have also been discrepancies regarding the site energy of chlorophyll C4 (C504 in 3WU2). Some studies based on refinement fits of optical spectra identified C4 to have a low excitation energy,28 but our findings as well as previous structure-based simulations on the CP43 subunit do not agree with this assignment. However, it is also important to note that the inclusion of inter-pigment interactions, particularly electronic and excitonic couplings between specific chlorophylls or groups of chlorophylls may change the relative trends in site energies.
At this point it is important to consider the above results in relation to the D1 protein. The peripheral chlorophylls ChlzD1 and ChlzD2 are discussed in the context of EET from the CP43 and CP47 core antennae, respectively, to the RC. Even though these two pigments are bound to the D1 and D2 proteins, they are thought to be functionally associated to the CP43/CP47 core antennae.18 Notably, ChlzD1 is in van-der-Waals contact (4.2 Å distance) with a β-carotenoid (CarD1) to which it can be excitonically coupled, and is close to the D1 active branch RC pigments (Fig. 2, ca. 21 Å), and hence assumed to play a direct role in EET between CP43 and the RC. The recent study by Yang et al. reported a faster EET from CP43 to ChlzD1 compared to EET from ChlzD1 to other D1 pigments,19 concluding that the latter process is the actual rate limiting step in the overall EET pathway, consistent with previous studies by Renger and co-workers.42 Nguyen et al. performed similar studies on D1D2Cytb559 complexes and identified a distinct excitonic state prior to primary charge separation (Trap*) likely belonging to one of the Chlz pigments.97 The above findings are not in agreement with the theoretical study of Hsieh et al. who reported that the Chlz sites do not mediate EET into the RC based on molecular dynamics simulations of the PSII core complex.98 Here, we explicitly compare the excited state energetics of the peripheral D1 chlorophyll ChlzD1 with the CP43 chlorophylls, including the full effect of PSII protein electrostatics (QM/MM) and conformational changes (PMM). Based on our results, ChlzD1 exhibits a lower Qy (S0–S1) excitation energy (1.839 eV) than all CP43 chlorophylls. Even with the inclusion of the dynamics based on PMM calculations, we find that ChlzD1 has a consistently lower mean site energy (1.855 eV) than all other CP43 chlorophylls. Based on our findings, we cannot explicitly conclude if ChlzD1 can mediate EET from CP43 to the RC. Also, we do not locate the lowest site energy on the nearest CP43 pigment, C5. Nevertheless, one possible implication could be that the low-lying excited state on the peripheral ChlzD1 may act as a protective “trap” state to quench excess excitation energy from the RC, and avoid photodamage of the D1 protein.
The calculated excitonic coupling values are listed in Table 3. The largest couplings mainly involve the following pigment pairs: C2–C4, C8–C9, C9–C11, C7–C9, C5–C7, C10–C11, C6–C7 and C2–C10. Except for the C5–C9 dimer, all other couples exhibit relatively small center-to-center distances (see Table S5†). The chlorophylls C7, C9 and C10 are the most strongly coupled, i.e., they strongly interact with at least three other chlorophyll pairs with significantly high coupling values. The largest excitonic couplings (i.e., with absolute values greater than 100 cm−1) are found for the C2–C4, C8–C9, C9–C11 and C7–C9 pairs.
C2 | C3 | C4 | C5 | C6 | C7 | C8 | C9 | C10 | C11 | C12 | C13 | |
---|---|---|---|---|---|---|---|---|---|---|---|---|
C1 | 10.89 | 21.90 | 3.09 | −3.07 | −12.22 | 21.05 | 7.15 | 10.89 | 2.70 | −2.52 | 0.63 | −0.11 |
C2 | 8.14 | 154.07 | −8.67 | −16.15 | −1.01 | 16.94 | 23.26 | 73.09 | 5.05 | 12.33 | −10.27 | |
C3 | −36.43 | 4.48 | −0.45 | 7.88 | −4.39 | −11.26 | 23.04 | 8.19 | 25.02 | 5.71 | ||
C4 | −3.40 | 4.57 | −2.08 | −1.53 | −0.97 | −34.68 | −4.46 | −4.72 | 5.09 | |||
C5 | 25.46 | −98.32 | −5.90 | −31.38 | −1.18 | −0.81 | −1.55 | 5.03 | ||||
C6 | −77.69 | −1.14 | −32.26 | −12.14 | −4.85 | −4.15 | 10.23 | |||||
C7 | 17.84 | −112.59 | −4.61 | −9.06 | −34.46 | 26.34 | ||||||
C8 | 130.37 | 66.94 | 1.78 | 2.11 | −5.16 | |||||||
C9 | 16.20 | −117.64 | 36.43 | −6.15 | ||||||||
C10 | −97.23 | 19.40 | −11.76 | |||||||||
C11 | 25.24 | −13.91 | ||||||||||
C12 | −48.50 |
The only exception in our estimated excitonic coupling values compared to those in previous literature is seen for the C2–C10 chlorophyll pair which yields a high excitonic coupling value of 73 cm−1. Moreover, the location of the C2–C10 is such that it connects the two layers of pigments in CP43 (Fig. 2c). A high excitonic coupling constant therefore implies that this dimer may play a role in EET between the stromal and lumenal layers of CP43. Similar arguments were made by Saito et al. in recent computational studies of EET coupling in the PSII-cc.61 It is to be noted that excitonic coupling values in closely spaced pigments are highly sensitive to the approximations used to estimate them.101 For instance, here the point dipole approximation is seen to predict a large positive coupling for the C2–C10 pair, but the TrEsp method in past investigations reported small coupling values for the same pair.27,29
It is noted that the couplings reported here are slightly higher in magnitude compared to those reported in previous theoretical studies,27,29 but are of the same magnitude as the values estimated based on PDA by Ishikita and co-workers on CP43 (ref. 61) and those by Grondelle and co-workers on the PSI-Lhca4 complex.95 This trend may be attributed to several effects, namely the overestimation of gas-phase transition dipoles by TD-DFT,41 or the fact that in the current implementation of the PMM procedure the perturbation does not include polarization effects of the environment,50,96,97 which may be responsible for a screening of the couplings. Consequently, in order to account for delocalized excited states within strongly coupled groups of pigments, one likely needs to employ more accurate quantum chemical methodologies.
The four Chl pigments C1–C2–C3–C4 constitute the lumenal layer of CP43 with an average Mg–Mg distance of 12.3 Å (see Fig. 2c). As discussed in Section 3.1 based on our TD-DFT and PMM results on individual pigments, we identified C1 and C3 to possess the lowest site energies in the “intact” CP43 antenna. C2 and C4 are the most closely arranged pigments in this layer and our findings suggest that this pigment pair has a strong positive coupling constant (154 cm−1), consistent with previous studies.27,29,61 However, based on our calculations, their individual site energies are not isoenergetic and our calculations (Table S6†) on the C2–C4 dimer do not reveal any delocalized exciton. The energy corresponding to the lowest excited state on C2 remains unaffected, but the exciton localized on C4 is only slightly red-shifted (29 meV).
The C7–C9 pair is in the stromal layer of CP43 with an Mg–Mg distance of 11.4 Å. Interestingly, both lowest excited states S1 (1.868 eV) and S2 (1.891 eV) are superpositions of the Qy transitions of C9 and C7 (analogous to the PD1PD2 special pair in the RC).62 The NTO pairs for the corresponding S1 and S2 transitions are shown in Fig. 4. The delocalized lowest excited state (S1) for this dimer is seen to possess an even lower excitation energy while the S2 state has a slightly higher energy compared to the similar Qy energies of the individual pigments C9 and C7 (1.881 and 1.885 eV respectively). The further red-shift of the lowest excited state in the dimer allows us to conclude that C9 and/or C7 may contribute majorly and almost equally to the red trap state in the stromal layer of CP43. This is in line with previous assignments of the lowest excitonic states made by Renger and coworkers.28,29 Based on the S0 → S1 excitonic couplings reported by Müh et al.,28 Saito et al.61 and Sarngadharan et al.,27 and those obtained in this work using the MD-PMM approach, both C7 and C9 are part of a strongly coupled group of chlorophylls in the stromal layer. The low energy excitation profiles of the remaining Chl pairs in the cytoplasmic layer (C5–C7, C6–C7, C8–C10, C9–C10, C9–C11, C10–C11, C12–C13) majorly constitute localized excitations corresponding to the Qy and Qx transitions on individual pigments.
One motivation for the study of pigment dimers is the identification of possible charge transfer states.102–109 In photosynthetic pigment–protein complexes CT states may participate in either spectral-tuning103 or photoprotection through the rapid quenching of excess excitation energy.104 It has been suggested that the latter process likely involves an excitation energy transfer from an excited Chl monomer to a strongly coupled Chl leading to a rapid non-radiative decay process via short-lived intermediate CT states.105 Fleming and co-workers reported CT quenching in the minor antenna complexes of PSII (CP29, CP26 and CP24).106,107 Ramanan et al. have reported the presence of mixed excitonic-CT states in Chl heterodimers in the low-energy manifold of the LHCII complex.108 More recently, Ostroumov et al. reported the presence of far-red emitting Chl–Chl CT states as intermediates in the excited state quenching of LHCII.109 Stark spectroscopy and more recent computational studies on another light harvesting protein, the PSI-Lhca4 complex demonstrated that a distinctive red-shifted emission originates from the mixing of the lowest exciton state with a CT state of an excitonically coupled dimer.110 Intermolecular charge-transfer (CT) states among photosynthetic pigments (ChlD1+PheoD1− or PD1+PheoD1−) directly control primary charge separation and charge recombination processes in the PSII-RC,26,63,65 but the presence of CT states in CP43 or CP47 core antenna proteins of PSII has not been reported. The present TD-DFT calculations on Chl dimers enable us to approach this question for the case of CP43.
The result that stands out concerns the C2–C10 pair. These chlorophylls (Fig. 5, Mg–Mg distance 10.2 Å) constitute a stacked dimer with parallel ring planes located approximately in the center of the CP43 transmembrane region connecting the stromal and lumenal layers (see Fig. 2c). Our results on the “crystal-like” configuration (snapshot 1) indicate that the first excited state for this dimer is localized on C2 (S1, 1.843 eV), a likely candidate for the lowest excitation energy in CP43. Most importantly, we find the second S2 (1.943 eV), and third (S3, 2.001 eV) excited states of the C2–C10 dimer to have C2 → C10 CT character mixed with the lowest excited state of C10. A detailed analysis of the NTOs involved in the transition i.e., the precise decomposition of the transition based on the NTO coefficients, shows that the second (S2) and third (S3) excitations are represented by a “delocalized” donor orbital within the C2–C10 dimer, with the acceptor NTO localized on C10, resulting in substantial CT character. The NTOs for the S2 state are shown in Fig. 5.
Until now we looked into the excited state properties of the Chl dimers of CP43 based on a single “crystal-like” structural configuration of PSII (snapshot 1). Now we investigate how the dynamics of PSII influence the CT excitation character in the C2–C10 pair. For this purpose, we performed the same type of excited state calculations, each with individually optimized QM/MM geometries, on nine additional structurally independent snapshots (snapshots 2–10) obtained from unbiased production MD with consecutive intervals of 5 ns. The QM/MM-TDDFT results on the C2–C10 pair are provided in Table S7.† Fig. S5 and S6† depict the NTOs and difference densities respectively for the lowest CT state of the pair, which shows that the effect of the protein matrix is similar both in the crystal-like conformation of the protein and in the selected MD snapshots. The relative order of site energies remains the same, i.e., Chl C2 has a lower site energy than C10, but the C2 → C10 CT character is distributed differently among the S1, S2 and S3 states depending on the protein configuration. Specifically, in the crystal-like snapshot both S2 and S3 states have CT character mixed with local excitation on C10 while in half of the selected MD snapshots the S3 state has dominant C2 → C10 CT character (see Table S7†). Interestingly we also find that the S1 state has some CT character mixed with the local excitation on C2 for two of the examined protein configurations. Our findings thereby demonstrate that the conformational dynamics of PSII tune the extent of LE–CT mixing of the C2–C10 pair in its lowest excited states and thus allows the C2 → C10 CT states to span an energy range that can bring considerable CT character as low as 1.81 eV. Such dynamic evolution of CT character has also been demonstrated in the low-energy excitation profile of RC pigments.26,62,64,65 The C2–C10 dimer is not the only Chl pair in CP43 to possess a CT state; our results locate higher-energy excited states (above 3 eV) with pure CT character for the C9–C11, C9–C10, C8–C10, C10–C11 pairs (Table S6†), however the C2–C10 pair is unique in having a very low-lying CT state, essentially interleaved with the lowest locally excited states of the whole system.
Understanding the molecular mechanism of formation of low-lying CT states in LHCs has crucial implications not only for mechanisms of light harvesting and EET but also for photoprotection. Recently, Sláma et al. employed multiscale modelling approaches to show that the low-lying red states and red-shifted fluorescence bands in PSI-Lhca4 originate from the interplay of exciton and CT states within a Chl pair.110 Such energetically low-lying mixed excitonic-CT states have also been assigned to Chl heterodimers in the excitonic manifold of the major plant light harvesting complex LHCII based on 2DES studies.108 Mixed exciton-CT states may also act as an energy sink and thus determine pathways of EET in normal light conditions, or in some cases, may participate as intermediate trap states for excitation energy quenching in excess light.109 Raszewski and Renger41 in their seminal work on the PSII core complex antennae proposed that CP43/CP47 may switch from a “light-harvesting” mode for open RCs to a “photoprotective” mode for closed RCs. Therefore, our identification of a C2 → C10 CT state coupled to the excitonic manifold of C10 may have the following functional implications: (a) facilitate EET within CP43 from the lumenal to the stromal layer, (b) spectral tuning of the “low-lying” red chlorophylls in CP43, or (c) act as intermediate “trap” states for quenching of excess excitation energy away from the RC. The latter is particularly relevant for CP43 owing to its close proximity to the RC pigments and the D1 protein.
The lowest exciton shows two subpopulations that depend on fluctuations in the protein conformation, which implies two major conformational basins with slightly different excitonic energies. The analysis of the contribution of the pigments and dimers used as basis for the excitonic coupling calculations shows that the C7–C9 and C2–10 dimers contribute the most. Minor contributions arise from Chls C1, C3, C5, C8 and C12. Most importantly, two rather distinct groups of coupled pigments participating to the two subpopulations mentioned above, can be identified: the one at lower excitonic energy (peaked at around 1.852 eV) characterized by the contributions from the C7–C9 dimer, C12 and, as minor contribution, C5 and the other (peaked at around 1.860 eV) characterized by contributions from the C2–C10 dimer, C3 and, as minor contribution, C1 (the C8 contribution is not mentioned because it provides a similar contribution to the two groups). The weight of the contributions of the different monomeric and dimeric pigments to the first exciton is highlighted in Fig. 6. The C7–C9 dimer provides the major contribution (with a weight of around 0.8) to the most red-shifted trap, along with C12 (with a weight of around 0.12–0.15) and C5 (with a weight of around 0.08–0.05) (see Fig. 6c). Instead, The C2–C10 dimer provides the major contribution (with a weight in the range of 0.6–0.8) to the second trap, along with C3 (with a weight in the range of 0.10–0.30) and C1 (with a weight of around 0.05–0.10) (see Fig. 6d). In one investigation, Müh et al.28 reported that two degenerate low-energy exciton transitions represent the lowest excited states of the two domains in the lumenal (containing C2 and C4) and stromal layers (containing C5 and C7–C11). Our results here support a similar scenario, in the sense of two excitonic domains in the lumenal and stromal layer, but with different contributions from the different pigments. It would nevertheless be interesting to have an estimate of the EET kinetics from the second trap (C2–C10, C3, C1) to the first (mostly red-shifted) trap (C7–C9, C12, C5), but this remains beyond the scope of the current work.
Based on our results, 3 out of the 4 Chls with the lowest site energies i.e., C3, C7, C9 (Fig. 3b) are located in the center of each layer. Interestingly, these pigments show strong excitonic coupling with each other group of chlorophylls (Table 3). For instance, C9 is strongly coupled to C7 (−112.59 cm−1) which is excitonically coupled to C5 (−98.32 cm−1). We found that this pair possesses delocalized excited states (Fig. 4) in the low-energy regime, and contributes almost equally and majorly to the first excitonic state of CP43 (Fig. 6c). This identity of the low-energy trap state agrees with earlier assignments made by Müh et al.28 and Shibata et al.16 suggesting that chlorophylls C9 and C7 belong to the same excitonic domain, as also seen in this work. Although some studies claim that the low energy sinks in the CP43/CP47 core antenna should be in close proximity to the RC pigments to facilitate efficient EET, this argument remained controversial because some simulations suggested that EET from the stromal layer to the RC is equally efficient as EET from the lumenal layer.41 Towards this, our assignment of a coherent excitonic domain comprising C7, C9 and C5 in the stromal layer has important functional implications towards EET from CP43 to the RC because C5 is indeed the closest CP43 pigment to peripheral ChlzD1 as well as the RC pigments (ChlD1 and PheoD1). This finding can be considered to be in agreement with recent studies by Yang et al.19 because we identify C5 as part of the lowest excitonic domain in C43 and therefore as a pigment that can mediate energy transfer from CP43 to the RC.
Our calculation of the global excitonic states reveals another distinct excitonic domain comprising chlorophylls C1, C3 and the C2–C10 dimer (Fig. 6d). Our TD-DFT results on C2–C10 indicated a low-lying CT state mixed with the excited state of C10 (Fig. 5). While recent computational studies have investigated the role of mixed excitonic-CT states towards the spectral tuning of “red” fluorescence states in PSI-Lhca4,110 the involvement of a CT state in spectral tuning of the “red” states in PSII core antenna complexes have never been reported. Based on our current results, the C2–C10 interaction red-shifts the Qy energy of C2 by ca. 24 meV, and most importantly this pair also forms a coherent excitonic domain along with the other “red” pigments C3 and C1 (Fig. 3b). Both results imply that the C2 → C10 CT state likely plays a role in the spectral tuning of the “red” trap state within the lumenal layer of CP43.
Finally, we reported global excitonic states involving all 13 Chls in CP43 by means of the MD-PMM procedure. Our findings led us to conclude that the lowest excited state is not localized on the same chlorophyll site at each system configuration but its nature depends on the dynamics of the protein matrix. Specifically, we find two red-shifted excitonic domains, one in each layer of the thylakoid membrane. The lower energy excitonic state has contributions from C7–C9, C12 and C5. Although unique, this identity of the low-energy trap state aligns in several ways with past assignments.16,19,28 The concept of domains has been previously discussed in the literature, but our results present a new perspective. We demonstrate that the identity of these domains is influenced by conformational motions, and their relative energy can be thus modulated through conformational dynamics. This is particularly significant at room temperature, where conformational dynamics play a crucial role.
The coherent excitonic domain in the stromal layer (comprising C7, C9 and C5) has functional implications for EET from CP43 to the RC because C5 is the closest CP43 pigment to peripheral ChlzD1 as well as the RC pigments.19 The other excitonic domain involves the C2–C10 dimer along with C3 and C1 in the lumenal layer. Most importantly, the mixed excitonic-CT state on C2–C10 may play a role in the spectral tuning of the “red” pigments in the lumenal layer of CP43 or act as intermediate for quenching of excess excitation energy in PSII (photoprotection). Overall, this study establishes a refined basis for future kinetic modelling of EET pathways as well as for structure-based interpretation of spectroscopic properties of CP43, and contributes to an improved understanding of light harvesting and excitation energy transfer in oxygenic photosynthesis.
Footnote |
† Electronic supplementary information (ESI) available: Fig. S1–S6, Tables S1–S8, example ORCA input file, Cartesian coordinates of optimized pigments. See DOI: https://doi.org/10.1039/d3sc06714a |
This journal is © The Royal Society of Chemistry 2024 |