Excitation landscape of the CP43 photosynthetic antenna complex from multiscale simulations

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.


Molecular Dynamics (MD) simulations
The initial structure of the Photosystem II (PSII) monomer used in this work is based on the high-resolution crystal structure of T. vulcanus (PDB ID: 3WU2). 1 The complete protein unit was embedded in a POPC lipid bilayer of dimension 176 × 176 Å 2 using Packmol-Memgen, 2 thoroughly solvated with a TIP3P water box (Figure S1) and neutralized with appropriate number of counterions in a 0.15M NaCl salt buffer.For the protein residues, waters and POPC bilayer we used the standard parameters from the AMBERff14SB, 3 TIP3P 4 and LIPID17 5 force fields, respectively.The partial charges and force field parameters for the organic cofactors (plastoquinones, carotenoids, structural lipids) were obtained using GAFF2, 6 bonded parameters for non-heme Fe and OEC were taken from earlier studies, [7][8][9] while those of the remaining cofactors (Chl a, Pheo, heme) were obtained from literature. 10,11 he nonbonded parameters for the metal ions were based on their respective oxidation states using data sets available for the TIP3P water model. 12For Na + and Cl -ions, we used the Joung-Cheatham parameters compatible with the TIP3P water model. 13,14 step-by-step minimization protocol was employed to remove energetically unfavorable geometric clashes in the system.In the heating phase, the system is first heated from 10 to 100 K in a succession of 5 ps in the NvT ensemble, followed by further heating from 100 to 303 K in the NpT ensemble for a total of 125 ps.The temperature during this step is maintained using the Langevin dynamics 15 with a collision frequency of 5 ps -1 .During the equilibration phase, the Cα atoms of amino acids were restrained with a force constant of 20 kcal mol -1 Å -2 .Subsequently, the restraints on the Cα atoms were systematically reduced (2 kcal mol -1 Å -2 /500 ps) and the system was further equilibrated for 65 ns to equilibrate the POPC bilayer.
Thereafter, production run was initiated without any restraints and the temperature and pressure set at 303 K and 1 atm, respectively.During the entire procedure, the temperature was controlled using Langevin dynamics with a collision frequency of 1 ps -1 and the system pressure was controlled using the Berendsen barostat 16 with anisotropic pressure scaling with a relaxation time of 2 ps.We employed the SHAKE algorithm 17 to constrain the bonds involving hydrogens, therefore a time step of 2 fs could be used.The electrostatic interactions were treated using the Particle Mesh Ewald (PME) approach 18 with a 10 Å cut-off.The AMBER20 package 19 was used to perform the energy minimizations and equilibration dynamics while the production MD simulations were performed in the GPU version of the pmemd module (pmemd.cuda). 20

Calculation of the absorption spectrum with the MD-PMM approach
For computing the absorption spectrum, after calculating the perturbed frequencies (ν) and transition dipoles (μj,i) for each exciton state over the frames of the MD trajectory, we evaluate the excitation energy distribution using a suitable number of frequency bins.Subsequently, we utilize the excitation energy distribution and the corresponding transition dipoles to determine the molar extinction coefficient ε0,i for the transitions from the ground to the ith excited state, thus providing the absorption spectrum according to: Here, νref, n(νref) and 0µ !,/ 0 are the frequency at the center of each bin, the corresponding number of MD frames and mean transition dipole square norm within the bin, respectively, ℏ = ℎ/2 with h the Planck constant,  ! is the vacuum dielectric constant, c is the speed of light and  -is the variance associated to the semiclassical intra-QC vibrations neglected in the evaluation of the unperturbed properties.In the present case, the value of  -(s = 10 -4 Hartree) has been estimated on the basis of the full width at half maximum (FWHM) of the experimental spectrum of chlorophyll a in toluene, 21 approximating the vacuum condition.The calculated spectral line shape in the Qy region is reported in Figure S2.Note that the width of the   Table S1.TD-DFT (ωB97X-D3(BJ)/def2-TZVP) excitation energies E (in eV) of the S2/Qx states and corresponding oscillator strengths for the CP43 chlorophylls computed in vacuo, with QM/MM on an isolated CP43 protein, and with QM/MM in the complete PSII monomer.The geometries in all cases are derived from QM/MM optimizations within the PSII monomer.Shifts are reported in meV with respect to the gas-phase values.

Figure S1 .
Figure S1.Side view and stromal (top) view of the molecular-mechanics (MM) model of the lipid bilayer bound PSII monomer used in this work (Na + and Cl -counterions not shown for clarity).
inhomogeneous broadening) is due to the explicit effect of the semiclassical fluctuations of the QCs and their environment along the MD trajectory and does not include any adjustable parameter.

Figure S2 .
Figure S2.Room-temperature absorption spectrum of the CP43 complex calculated with the MD-PMM approach.The spectrum has been shifted by 0.000457 Hartree to lower energies in order to match the position of the main experimental peak measured at 293 K.

Figure S3 .Figure S4 .
Figure S3.Decomposition of the electrostatic and non-electrostatic contributions (axial ligation, H-bonding, QM/MM geometry optimization) to the site energy shifts of the CP43 chlorophylls.

Figure S5 .Figure
Figure S5.Natural Transition Orbitals (NTOs) for the lowest excited state with significant CT character for the C2-C10 dimer in snapshots 2-10.

Figure S6 .
Figure S6.TD-DFT difference densities for the lowest excited states with CT character for the C2-C10 dimer in the 'crystal-like' (snapshot 1) and snapshots 2-10.

Table S2 .
TD-DFT Transition Dipole moments of the CP43 chlorophylls in vacuo.

Table S3 .
Comparison of CP43 chlorophyll site energies from the literature.All values are in eV.
Site Muh et al. 22 a Muh et al. 22 b Shibata et al. 23 c a Based on 3ARC

Table S4 .
Excitonic coupling constants (upper panel) and corresponding standard deviations (lower panel) between CP43 chlorophylls computed using the MD-PMM approach.All values are reported in cm −1 .

Table S6 .
QM/MM and TD-DFT (ωB97X-D3(BJ)/def2-TZVP) computed excitation energies (in eV) and corresponding oscillator strengths and NTO analysis for selected chlorophyll pairs in CP43.The geometries in all cases are derived from QM/MM optimizations of the pigments embedded inside the complete PSII monomer for snapshot 1. (LE: local excitations, CT: Charge Transfer states).

Table S8 .
Comparison of different range-separated functionals to compute the Chl a ionization potential (IP) and electron affinity (EA), the energy of the Highest Occupied Molecular Orbital (HOMO) for Chl and Chl -. 27 All energy values are in eV.HFX is the range of exact (Hartree-Fock) exchange included in each functional, and ω (in Bohr −1 )

Cartesian coordinates (in Å) of QM regions from the QM/MM geometry optimizations of the 13 CP43 chlorophylls and of ChlzD1 Note:
These coordinates are also provided as a separate plain-text file.