Structural determinants for red-shifted absorption in higher-plants Photosystem I

Far-Red; Lhca; Light-Harvesting; Low-energy absorption; Photosynthesis; Photosystem I; Red 33 Forms 34 SUMMARY 35 - Higher plants Photosystem I absorbs near-infrared light through long -wavelength chlorophylls, enriched under 36 vegetation canopies, to enhance photon capture. Far-red absorption originates from chlorophyll s pairs within the 37 Lhca3 and Lhca4 subunits of the LHCI antenna , known as the “red cluster” composed of chlorophylls a603 and 38 a609. 39 - We used reverse genetics to produce an Arabidopsis mutant devoid of red-shifted absorption, and we obtained 40 high-resolution cryo-EM structures from purified PSI-LHCI complexes in both wild-type and mutant plants. 41 - Computed excitonic coupling values suggested a possible contribution of additional nearby pigment molecules, 42 namely chlorophyll a615 and violaxanthin in L2 site, to far-red absorption. Therefore, we investigated the structural 43 determinants of far-red absorption and analyzed the spectroscopic effects of these additional pigments by producing 44 further Arabidopsis transgenic lines. 45 The two experimental structures were used for quantum mechanics calculations , revealing that excitonic 46 interactions alone cannot explain far-red absorption, while charge transfer states were needed for accurate spectral 47 simulations. 48 - Our findings demonstrate that the molecular mechanisms of light-harvesting under shaded conditions rely on very 49 precise tuning of chromophore interactions, an understanding of which is crucial for designing light -harvesting 50 complexes with engineered absorption spectra 51 52

53 Photosynthetic organisms use solar radiation as their primary energy source to convert carbon dioxide and water 54 into oxygen and sugars. The light reactions of oxygenic photosynthesis involve two multimeric pigment -binding 55 protein complexes: photosystems (PS) I and II, which work in series. The initial step of light harvesting is catalyzed 56 by PSII, which is responsible for water splitting and oxygen evolution. Meanwhile, PSI mediates the reduction of 57 NADP+ to NADPH, serving as a temporary storage for reducing power (Croce & van Amerongen, 2020). Both PSs 58 share a common structure that includes a core complex housing the reaction centers (RC) where charge separation 59 reactions occur and an antenna system (LHC) that enhances the light-absorbing cross-section of the complexes (Pan 60 et al., 2020). Despite these similarities, there are striking differences in their spectral properties, with PSI exhibiting 61 a red -shifted absorption profile (Rivadossi et al. , 1999) . The red -most PSI absorption arises from special 62 chlorophyll (chl) pairs that absorb at energies below that of the RC P700, referred to as “red chl forms” or simply 63 “red forms” (RF) (Croce & Van Amerongen, 2013). 64 These pigments extend the absorption capacity into the near-infrared spectrum, providing advantages under canopy 65 or in dense culture conditions where most visible wavelengths are absorbed by the upper leaf layers (Martínez-66 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 3 García et al., 2010) while infrared radiation is strongly enriched, resulting in a far-red to red ratio higher than 5 67 (Park & Runkle, 2017) . Thus, although RF contribute only a small percentage of the total absorption (Gobets & 68 Van Grondelle, 2001), they enjoy preferential excitation. Moreover, RF are highly effective in energy transfer and 69 trapping, as approximately 80% of the PSI excitation transits through these pigments to reach P700 (Croce et al., 70 1998). Since these low-energy chls are highly populated, they may play crucial roles in photoprotection and/or the 71 concentration of excitation energy (Rivadossi et al., 1999; Jennings et al., 2003; Carbonera et al., 2005). However, 72 their exact physiological role(s) remain partially understood (Jennings et al., 2013). 73 RF are present in nearly all types of PSI complexes, and their occurrence is closely related to the availability of 74 infrared radiation. In cyanobacteria (e.g., Arthrospira platensis), these pigments are localized within the core 75 complex and facilitate the absorption of far-red photons, resulting in PSI fluorescence emission peak s at 720-725 76 nm (Karapetyan et al., 1997). In green algae (e.g., Chlamydomonas reinhardtii) and mosses (e.g., Physcomitrium 77 patens), red chls are associated with the LHCI antenna system, resulting in a PSI-LHCI fluorescence emission peak 78 wavelength ranging between 715-722 nm (Mozzo et al., 2010; Gorski et al., 2022) (Fig. 1). Upon land colonization, 79 environmental niches rich in far-red radiation becomes widespread under canopy, leading to a greater association 80 of RF with LHCI, positioned distal to P700 (Croce et al., 1998). Indeed, the PSI-LHCI fluorescence emission of 81 flowering plants shifts towards the red spectrum, peaking at 730 nm or even beyond (Akhtar & Lambrev, 2020). 82 83 84 85 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 4 86 Figure 1. Evolution of low -absorbing spectral features in Viridiplantae . (a) Time tree of representative 87 Viridiplantae species and their divergence times (in millions of years, MYA). The most red -shifted fluorescence 88 emission peak of each species' experimental spectra is reported and colored according to the color code of panel b. 89 Note that the PSI core complex emits at ~720 nm, which can be considered the baseline level for RF associated 90 with antenna proteins. The emergence of RF can be traced back to approx. 489 -403 MYA. (b) Low temperature 91 (77K) fluorescence emission spectra of representative species from green algae (C. reinhardtii), mosses (P. patens), 92 full sun land plant angiosperms: Z. mays (Poaceae) and A. thaliana (Brassicaceae); shade plants: F. verschaffeltii 93 (Acanthaceae) and A. comosus (Bromeliaceae); and sea grasses (P. oceanica and C. nodosa). Shade plants thrive 94 in far-red enriched light while seagrasses live in the absence of far-red radiation. The areas of the spectra associated 95 with emission from PSII-LHCII, the PSI core, or the PSI-LHCI supercomplex are colored differently. (c) Multiple 96 alignments of the Lhca4 (corresponding to Lhca8 for C. reinhardtii and Lhca2b for P. patens and Lhca3 protein 97 sequences from C. reinhardtii (Cr), P. patens (Pp), P. oceanica (Po), C. nodosa (Cn), A. thaliana (At), Z. mays 98 (Zm), A. comosus (Ac), and F. verschaffeltii (Fv). Residues binding red form chlorophylls (chl a603 and chl a609) 99 are colored in red and shown in red rectangles. Coordinating residues for chl a615 are colored green for species 100 with reported PDB structure. 101 102 In higher plants, LHCI binds the PSI core complex as functional heterodimers: Lhca1 -Lhca4 and Lhca2 -Lhca3, 103 both of which exhibit similar spectral properties and emit far-red fluorescence at approximately 730 nm (Wientjes 104 & Croce, 2011) . In vitro reconstitution of recombinant Lhca complexes (rLhca) showed that the far -red shift is 105 primarily caused by the Lhca4 and Lhca3 components of the dimers (Croce et al., 2002; Castelletti et al., 2003), 106 and is linked to a specific sequence substitution unique to these two LHCs, where an Asn serves as the binding 107 residue for chl a603 instead of a His residue, as seen in all other LHC proteins (Jansson, 1999; Morosinotto et al., 108 2003) (Fig. 1c). Indeed, the substitution of Asn with His (a603-NH substitution) resulted in the complete loss of 109 the far-red absorption and emission forms in both rLhca3 and rLhca4 complexes (Morosinotto et al., 2003). It was 110 hypothesized that the presence of Asn induces strong excitonic interaction s between the chl a603–a609 pair (A5 111 and B5 according to Kühlbrandt ’s nomenclature (Kühlbrandt et al. , 1994) ) leading to a charge transfer (CT) 112 character (Romero et al., 2009), and ultimately resulting in the formation of the RF (Morosinotto et al., 2003). 113 Complementing koLhca4 Arabidopsis lines with mutant Lhca4, carrying either His or non -pigment-binding 114 residues at the chl a603 ligand position (Li et al., 2023) resulted in a blue shift of ~3 nm. 115 However, accumulating evidence suggests that the substitution of Asn vs. His as the ligand of chl a603 alone cannot 116 account for the variability in far-red spectral properties found in nature (Li et al., 2023; Elias et al., 2024). For 117 instance, in the Lhca2/a4/a9 subunits of C. reinhardtii, where the Chl a603 ligand is Asn, emission peaks range 118 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 5 between 690 and 717 nm, with significantly lower absorption beyond 700 nm compared to higher plants Lhca4 119 (Mozzo et al., 2010). Moreover, the a603-HN (H111N) mutation, although introducing a significant red-shift in 120 the PSII antenna complex Lhcb4, did not achieved the 43 nm shift observed in the chl a603-NH mutant of Lhca3 121 or Lhca4 (687 nm instead of ~730 nm) (Morosinotto et al., 2003; Guardini et al., 2020; Sardar et al., 2024), despite 122 the high structural similarity between the antenna complexes. 123 In addition to the arrangement of relevant chls, the surrounding microenvironment can be important for optimizing 124 low-energy spectral forms. Structural models of PSI -LHCI reveal a chromophore cluster comprising three chls - 125 chl a603, chl a609, chl a615 - and two xanthophylls (xan): violaxanthin in site L2 and a lutein (lut) (Qin et al., 126 2015). Notably, chl a615 is present only in the Lhca3 and Lhca4 subunits - the red-shifted subunits - while it is 127 absent in all other LHC proteins, regardless of whether they serve PSI or PSII. This observation raised the question 128 of its potential role in forming excitonic interactions (Melkozernov & Blankenship, 2003; Wientjes et al., 2012). 129 In this study, we investigated the structural and biophysical determinants of the red chl forms in Arabidopsis 130 thaliana PSI-LHCI by combining in vivo site-directed mutagenesis and single-particle cryogenic electron 131 microscopy (cryo -EM). We obtained two high -resolution structures of the PSI -LHCI supercomplexes from 132 Arabidopsis thaliana wild-type (WT) and the a603-NH mutant genotype, which lacks RF. We then computed the 133 excitonic interactions within PSI-LHCI pigments, focusing on the interaction between the L2 xan and the chl a603–134 a609 pair in forming low-lying energy spectral states. 135 136

137 Plant materials 138 The Arabidopsis thaliana koLhca3 koLhca4 mutant was generated by crossbreeding koLhca3 and koLhca4 NASC 139 insertional lines, following the methodology outlined by (Bressan et al., 2016). Complementation of the koLhca3 140 koLhca4 mutant was achieved through Agrobacterium tumefaciens -mediated transformation, as described by 141 (Zhang et al., 2006), resulting in a603-NH, a615-HA, a615-HI and A3WT-A4WT lines. 142 Purification and characterization of PSI-LHCI samples 143 To purify PSI -LHCI supercomplexes, Arabidopsis thaliana plants were grown for approximately 6 weeks in a 144 phytotron (150 μmol photons m -2 s-1, 23°C, 70% relative humidity, 8/16 h of day/night). Prior to the isolation 145 procedure, Arabidopsis leaves were dark -adapted for 60 minutes at 4°C. Unstacked thylakoid membranes were 146 prepared as described in (Bassi et al., 1985). Thylakoid membranes were resuspended to a chl concentration of 1 147 mg/mL in 10 mM HEPES pH 7.5 and solubilized by adding an equal volume of 2% dodecyl -β-D-maltoside (β-148 DM). The samples were vortexed for 30 seconds and incubated on ice for 10 minutes, and the insoluble material 149 was removed by centrifugation at 20,000 xg for 10 minutes at 4°C. The supernatants were fractionated by sucrose 150 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 6 gradient ultracentrifugation at 284,000 xg at 4°C for 18 hours (Beckman SW40 Ti rotor) or at 141,000 xg at 4°C 151 for 30 hours (Beckman SW28 Ti) (Fig. S1). Bands containing PSI-LHCI were harvested using a Hamilton syringe. 152 For cryo-EM preparations, PSI-LHCI samples were concentrated to a final volume of ~800 μL, and sucrose was 153 removed by dialysis overnight against a solution containing 10 mM HEPES pH 7.5 and 0.05% β-DM. Finally, the 154 samples were concentrated to a chl concentration of ~1,5-2 mg/mL. 155 Purification PSI-core and LHCI 156 PSI-core complex and LHCI were purified from WT and mutant lines as described by (Croce et al., 1998; Wientjes 157 & Croce, 2011) with some modifications. Briefly, PSI-LHCI from sucrose gradient ultracentrifugation were diluted 158 in 10 volumes of 5 mM tricine (pH 7.8) and centrifuged for 3 h at 411,000 xg, using a T-865.1 rotor, Sorvall. Pellets 159 were resuspended in a buffer containing 5 mM Tricine (pH 7.8) and 0.05% β -DM. The chl concentration was 160 adjusted to 0.3 mg/mL and the samples were solubilized by adding 1% β -DM and 0.5% Zwittergent -16. The 161 samples were kept on ice with gentle agitation for 25 min, then rapidly frozen in liquid nitrogen and slowly thawed 162 to enhance the yield in detached LHCI. Solubilized complexes were fractionated by sucrose gradient 163 ultracentrifugation at 485,000 xg at 4°C for 6 hours, using a Beckman SW60 Ti rotor. 164 Spectroscopy and pigment analysis 165 Absorption spectra were recorded at room temperature ( RT, 22°C) using an SLM -Aminco DW -2000 166 spectrophotometer in a buffer containing 10 mM HEPES pH 7.5 and 0.05% β -DM for native complexes or 80% 167 acetone buffered with Na2CO3 for pigment extracts. 168 Emission and excitation fluorescence spectra were recorded at cryogenic temperatures (77 K) using a Jobin–Yvon 169 Fluoromax-3 spectrofluorometer. Samples were diluted in a buffer containing 50% w/v glycerol, 10 mM HEPES, 170 pH 7.5, 0.05% β-DM, and excited at 440 nm. 171 Emission spectra of intact leaves were recorded at 77 K using an Ocean Insight SR -6NVN500-50 172 spectrofluorometer. 173 CD spectra were recorded at 4°C on a Jasco J1500 spectropolarimeter. 174 The pigment composition of LHCI complexes was assessed from the deconvolution of acetonic spectra as described 175 in (Chazaux et al., 2022). HPLC pigment separation and quantification was performed according to (Gilmore & 176 Yamamoto, 1991) by using an Agilent 1260 Infinity II HPLC system. 177 Cryo-EM sample preparation and data acquisition 178 The PSI supercomplex at a chl concentration of ~1.5 -2 mg/mL was vitrified with a Mark IV Vitrobot (Thermo 179 Fisher Scientific). 3 μL of the sample were applied to a Quantifoil R 0.6/1Cu 300 -mesh grid previously glow -180 discharged at 30 mA for 30 seconds in a GloQube (Quorum Technologies). Immediately after sample application, 181 the grids were blotted in a chamber at 4 ºC and 100% humidity and then plunge-frozen into liquid ethane. 182 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 7 Vitrified grids were transferred to a Talos Arctica (Thermo Fisher Scientific) operated at 200 kV and equipped with 183 a Falcon 3 direct electron detector (Thermo Fisher Scientific). 2716 and 2601 movies were acquired, for WT and 184 a603-NH mutant, respectively, at a nominal magnification of 120’000x, corresponding to a pixel size of 0.889 185 Å/pixel, in electron counting mode, with a nominal defocus range of -0.8 to -2.4 μm and with a total dose of 40 e-186 /Å2 equally distributed on 40 frames. The cryo -EM experiments were conducted at the NoLimits Center of the 187 University of Milan. 188 Cryo-EM data processing and image reconstruction 189 All image processing and reconstruction steps were performed using CryoSPARC v4.4 (Punjani et al., 2017). The 190 experimental workflows are outlined in Fig . S2 and S3. After patch motion correction and CTF estimation, 2313 191 and 2257 manually curated micrographs were used for initial particle picking for WT and a603-NH mutant, 192 respectively. 193 For AtPSI-WT, after a first round of reference -free autopicking and 2D classification, the best classes were used 194 for template-based autopicking, resulting in an initial number of 575,271 particles. Particles were extracted with a 195 box size of 448 px, binned 2X2, and subjected to several rounds of 2D classification. The best 2D classes (282,376 196 particles) were used for ab initio reconstruction (3 classes) and heterogeneous refinement, resulting in an initial 197 map at 3.74 Å resolution. After 3D classification and local and global CTF refinement, the best particles were re -198 extracted at full size and used for a final round of non-uniform refinement (Punjani et al., 2020), obtaining a final 199 density map at 3.13 Å resolution. 200 For AtPSI-a603-NH, 531,976 particles generated from a first round of reference -free autopicking were subjected 201 to 2D classification, and the best classes were used to train the neural network in Topaz (Bepler et al., 2019) on a 202 subset of 157 micrographs. The trained model was then used to pick the entire dataset, resulting in a total number 203 of 115,589 particles. These particles were extracted with a box size of 448 px, binned 2X2, and after several rounds 204 of 2D classification, the best particles were selected for ab initio reconstruction (3 classes) and heterogeneous 205 refinement, resulting in an initial 4.19Å resolution map. After two rounds of 3D classification, particles 206 corresponding to the best 3D class were re -extracted without downscaling and, following local and global CTF 207 refinement, further subjected to homogeneous and non-uniform refinement, yielding a final map with a resolution 208 of 3.29 Å. 209 Before model building, the maps were sharpened with a global B factor of 83.7 Å2 and 91.8 Å2 for WT and a603-210 NH mutant, respectively. The resolution was estimated by the “gold -standard” Fourier Shell Correlation 211 (FSC=0.143) criterion. Local resolution estimation was performed as implemented in CryoSPARC on the 212 unsharpened map. The final EM maps (colored according to the local resolution) and the FSC curves are shown in 213 Fig. S2 and S3 (panels D and E). 214 215 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 8 Model building and refinement 216 The high-resolution structure of the P. sativum PSI-LHCI complex (PDB code 7DKZ) was used as the starting 217 model. After initial docking of the model in the map with USCF ChimeraX (Goddard et al., 2018), each protein 218 chain was independently rigid -body fitted and refined with Phenix -refine (Adams et al., 2010). The amino acid 219 sequences of the different subunits were manually mutated to match the sequences of A. thaliana using COOT 220 (Emsley & Cowtan, 2004), and the resulting model underwent several rounds of real -space refinement in Phenix 221 and manual rebuilding in COOT. Ligands and water molecules were modelled when unambiguously identified in 222 the density map and refined to a reasonable B factor. Ligand restrains for refinement were generated with eLBOW 223 (Moriarty et al., 2009). The stereochemical quality of the final model was assessed with MolProbity (Chen et al., 224 2010). Data collection and model refinement statistics are summarized in Table S3. High-resolution figures were 225 prepared with ChimeraX and PyMOL (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC). 226 227 Energy transfer and excitonic coupling calculations 228 We used a point-dipole approximation (PDA) to compute the excitonic couplings, as described by (van Amerongen 229 & van Grondelle, 2001; M üh et al. , 2010; Liguori et al. , 2015; Sen et al. , 2021). The chl point dipoles were 230 positioned at the geometric centers of the four nitrogen atoms of the chlorin ring (Friedl et al. , 2022) 231 232 The chl Q Y and Q X transition dipole moments were aligned parallel to the nitrogen ND -NB and NC -NA axis, 233 respectively, as in (van Amerongen & van Grondelle, 2001; Georgakopoulou et al., 2007; Liguori et al., 2015). 234 The carotenoids (car) point dipole was placed on the C15 atom, with transition dipole moments (transition 235 S2 ← S0) oriented parallel to the central part of the polyene chain (atoms C11 -C33), as modelled in 236 previous studies (Georgakopoulou et al., 2007; Liguori et al., 2015). The QY transition dipole moments were used 237 for couplings between chls (chl a–chl a, chl a–chl b, and chl b–chl b), while couplings between chls and cars were 238 computed with the QX transition dipole moments for chls (Croce et al., 2001; Polívka & Frank, 2010). 239 The excitonic couplings between two pigments i and j, in the PDA were calculated using the following formula 240 (cm-1): 241 242 where f 1 is the local field correction factor, and are the module of the transition dipole moment of 243 pigments i and j, εr is the relative dielectric constant, here equal to 2.4 (Gobets & Van Grondelle, 2001; Liguori et 244 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 9 al., 2015) is the module of the distance between the center of the dipole moment vector and is the 245 orientation factor between chl i and j 246 247 with and are the normalized transition dipole moment vector and is the normalized distance vector. 248 Dipole moment values were taken as 4 D, 3.4 D, and 4.5 D for chl a, chl b, and cars, respectively (van Amerongen 249 & van Grondelle, 2001; Georgakopoulou et al., 2007; Liguori et al., 2015) In the case of ε r = 2.4, (f 12μ2)/εr was 250 calculated to be 17.6 D2 for chl a (van Amerongen & van Grondelle, 2001). The calculations were run via homebuilt 251 codes using Python 3.8. 252 253 Excited state calculations 254 We performed QM/MM optimizations and polarizable QM/MM (Bondanza et al., 2020) excited-state calculations 255 of the Chls in the Lhca4 structure from the present work. Before QM/MM calculations, the structure was refined in 256 a pure MM protocol as detailed in the Supplementary Information. All chls were optimized independently, except 257 for a603-a609 which were optimized together. Excited-state calculations were performed for all chls in Lhca4 and 258 for the a603-a609 dimer; control calculations were performed by also including L2 Vio with chls a603 and a609. 259 We employed a diabatization procedure to extract CT energies and couplings from the dimer calculations (Nottoli 260 et al., 2018). Finally, we built an exciton model considering all QY states of chls and the CT states within the a603-261 a609 dimer, which we used to simulate absorption spectra of Lhca4 (Sláma et al., 2023). Detailed computational 262

are reported in the SI. 263 Statistics 264 Statistical analyses were performed in OriginPro using One -way analysis of variance (ANOVA), means were 265 separated with Tukey’s post hoc test at a significant level of P < 0.05 (see figure legends for details). Error bars 266 represent the standard deviation. 267 268 269

270 The koLhca3 koLhca4 Arabidopsis thaliana genotype was complemented with mutant isoforms of Lhca3 and 271 Lhca4, in which the Asn responsible for chl a603 binding was replaced with a His ( a603-NH mutant) in both 272 antenna subunits (Asn99 in AtLhca4 and Asn103 in AtLhca3). We observed a shift in the 77K fluorescence 273 emission spectrum of the leaves from 736 to 724 nm (Fig. S4). 274 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 10 PSI-LHCI supercomplexes from both WT and a603-NH mutant lines were purified by sucrose gradient 275 ultracentrifugation (Fig. S1). RT absorption spectra confirmed the loss of far -red spectral forms in the a603-NH 276 mutant complex, which peaked at 704 nm and extended to 750 nm. This was accompanied by an increased 277 absorption ranging from 650 to 700 nm, peaking at 679 nm (Fig . 2a). Low temperature (77K) emission spectra 278 revealed a 13 nm blue shift in the a603-NH mutant PSI supercomplexes compared to WT, with emission peaks at 279 721 nm (Fig. 2b). Emissions around 720 nm can be attributed to the PSI core complex (Bassi & Simpson, 1987), 280 leading us to conclude that the a603-NH mutation effectively abolished Lhca-associated RF in vivo, consistent with 281 previous reports (Morosinotto et al., 2003; Li et al., 2023). 282 283 284 Figure 2. Spectral analysis of PSI-LHCI and LHCI from A. thaliana WT and a603-NH (a, c) RT absorption 285 and (b, d) 77K fluorescence emission spectra of PSI-LHCI (a, b) and LHCI (c, d) from A. thaliana WT and a603-286 NH. The WT minus NH difference spectra are shown as black lines in the corresponding plots. The amplitude of 287 the absorption difference spectra was magnified by a factor of 3, while the fluorescence difference spectra were 288 multiplied by a factor of 0.25 in order to plot them on the same axis. Key wavelengths are indicated in nm above 289 the respective peaks. Experiments were repeated independently twice, with similar results. 290 291 To assess the impact of the a603-NH mutation on the spectral properties of LHCI complexes compared to PSI-292 LHCI supercomplexes, we purified LHCI heterodimers (Lhca1 -Lhca4 and Lhca2-Lhca3). LHCI dimers isolated 293 from WT plants exhibited a 77K emission maximum at 728 nm, consistent with (Wientjes & Croce, 2011) . In 294 contrast, LHCI dimers from the a603-NH line showed an emission peak at 689 nm, highlighting a significant blue 295 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 11 shift of approximately 39 nm (Fig . 2d), which aligns with observations made in vitro (Morosinotto et al., 2003). 296 Similarly, RT absorption spectra of isolated LHCI showed the loss of a broad absorption tail in the a603-NH mutant, 297 covering the range from 680 to 730 nm and peaking at 698 nm (Fig . 2c). Conversely, the PSI core purified from 298 both WT and a603-NH lines exhibited unchanged optical properties, maintaining a peak at 720 nm (Fig. S5). 299 300 Structures of PSI-LHCI from WT and a603-NH plants 301 To investigate the structural determinants underlying the blue-shifted absorption/emission caused by the a603-NH 302 mutation, we determined the cryo-EM structures of the A. thaliana PSI-LHCI WT and a603-NH supercomplexes. 303 The final reconstructions, at 3.13 Å and 3.29 Å resolution, respectively, revealed well-defined density maps for 304 both the PSI core and the LHC subunits (Fig . S6), allowing the construction of accurate models for the two 305 complexes. 306 The structure of the individual subunits and the positioning of the chromophores within the AtPSI-LHCI WT 307 supercomplex were very similar to those observed in other land plants (Qin et al., 2015; Mazor et al., 2017; Huang 308 et al., 2021; Iwai et al., 2024; Nelson, 2024) (Fig. S7-S12, Table S1). In both AtPSI-LHCI WT and a603-NH 309 structures, the primary difference in pigment composition between the different Lhcas was found in the chl a/b ratio 310 (Fig. S13). Specifically, Lhca1/a2/a3 each bound 14 chl, while Lhca4 bound 15 chls. 311 Chl a615 was coordinated by His168 and His151 from helix C in Lhca3 and Lhca4, respectively. These pigments 312 showed clear density maps in both structures, allowing for accurate modelling of the chls ( including the chlorin 313 rings and parts of the phytol tails) and the xan molecule (Fig. S14). An additional xan molecule (lut) was located 314 close to chl a615 in Lhca4 only. Since this lut was absent in the red-emitting Lhca3, it was not considered a potential 315 component of the red-emitting cluster. 316 317 In the a603-NH mutant structure, the bulkier His side chain at position 103 (Lhca3) and 99 (Lhca4) caused a ~0.7 318 Å displacement of the chl a603 chlorin ring (Fig. 3, S14, S15, S16, S17), clearly visible in the corresponding density 319 maps (Fig. S14). The chlorin ring and part of the phytol tail of chl a603 were well -defined in both the WT and 320 a603-NH density maps, allowing the reliable assignment of the position of chl a603 in both structures (Fig. S14b, 321 c). Surprisingly, the distance between the chlorin ring centers of chl a603 and chl a609 remained essentially 322 unchanged across both WT and mutant proteins (9.3 Å in WT vs. 9.1 Å in the a603-NH for Lhca3, and 9.3 Å vs. 323 9.2 Å for Lhca4). In comparison, the distance between the centers of the closest pyrrolic rings of the two chls (ring 324 C) increased from 4.2 to 4.4 Å in Lhca3 and from 4.3 to 4.6 Å in Lhca4 (Fig. S15). 325 At the same time, the distance between chl a603 and violaxanthin in L2 increased from 5.2 Å in WT to 5.9 Å in 326 a603-NH, and from 5.3 Å to 6.0 Å in both Lhca3 and Lhca4, representing a change of about 12%. 327 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 12 We then computed excitonic coupling (EC) values, which take into account the relative spatial orientation between 328 each pair of pigments in the cluster. The EC between chl a603 and Vio decreased from 534 cm-1 and 524 cm-1 for 329 Lhca3 and Lhca4 in the WT to 375 cm -1 and 384 cm -1 in the a603-NH mutant, representing a change of 27-29% 330 (Fig. 3, S18). In contrast, the introduction of the a603-NH mutation resulted in only slight alterations of the EC 331 values between the chl a603 and chl a609 pair (6-8 cm-1). Furthermore, while the EC values between the chl pairs 332 were in the limited range of 87 cm-1 and 101 cm-1 for all Lhca, the EC values for the WT chl a603–Vio L2 coupling 333 were significantly higher for Lhca3 and Lhca4 compared to Lhca1 and Lhca2, which do not show red -shifted 334 absorption. Notably, the a603-NH mutation decreased the ECs of Lhca3 and Lhca4 to values comparable to those 335 of the ‘non-red’ subunits. Collectively, these data suggest that the xan ligand might play a role in tuning the 336 absorption properties of Lhca3 and Lhca4 toward low energy levels. 337 338 339 Figure 3. Structural superposition and excitonic coupling analysis of Lhca4 and Lhca3 WT and a603-NH. 340 (upper part) Structures of WT and a603-NH red clusters of Lhca4 and Lhca3 from A. thaliana PSI-LHCI. WT 341 structures are colored in red and orange, while a603-NH mutant structures are colored in blue and cyan. Lines are 342 drawn between the pigments to highlight the excitonic couplings reported in the table (lower part, values in cm-1). 343 The thickness of the line is proportional to the EC value. 344 345 346 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 13 Effect of Xanthophyll occupancy of site L2 on low-energy absorption 347 To investigate the role of the xan at the L2 site of Lhca3 and Lhca4 in modulating RF -inducing interactions, we 348 compared the fluorescence excitation spectra of WT and a603-NH PSI-LHCI, highlighting the contribution of 349 different wavelengths and associated chemical species to the ir far-red fluorescence emissions. In the 455-505 nm 350 region, with the major contribution from cars (Ashenafi et al., 2023), the a603-NH mutant showed a systematically 351 lower signal compared to WT (Fig. 4a), implying a change in the efficiency of energy transfer from xan to far-red 352 chl. 353 The difference between absorption and excitation (Abs-Exc) plots in Fig. S19a provides insight into the extent of 354 energy absorbed and then transferred to chl a at each wavelength. In the 475 -505 nm region, the Abs minus Exc 355 values for the a603-NH mutant were larger than those of the WT, suggesting stronger energy dissipation (i.e., lower 356 conversion to chl a excited state) and a reduced overall contribution of cars to far-red fluorescence. 357 Changes were also observed in the 560-680 range, which is associated with contributions from chl (Ashenafi et al., 358 2023). The a603-NH mutant also showed a lower signal compared to the WT. Specifically, the highest peak and 359 the area under the curve showed a decrease of 4% and an 8%, respectively, in the a603-NH vs. the WT. 360 361 362 363 Figure 4. Spectral analysis of PSI -LHCI and isolated LHCI from A. thaliana WT, a603-NH, and npq2 364 plants. 365 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 14 Low-temperature (77K) fluorescence excitation spectra of PSI-LHCI from A. thaliana WT, (a) a603-NH, (b) and 366 npq2 plants. Far-red fluorescence emission (at 760 nm for WT and npq2, and 740 nm for the a603-NH mutant) was 367 followed by exciting the samples from 425 to 690 nm. Spectra were normalized to the peak value. Shaded areas 368 indicate spectral regions where carotenoids and chls absorption is dominant and are colored in blue and red, 369 respectively. Key wavelengths corresponding to absorption/emission peaks are indicated in nm above the respective 370 peaks. c-d) CD (4°C) spectra of purified LHCI complexes from WT, a603-NH and npq2 mutants. Spectra are 371 normalized to the same absorption in the Q Y region. The difference spectra are shown as black lines in the 372 corresponding plots. The values of the fluorescence excitation difference spectra have been magnified by a factor 373 of 2 for better visualization. Experiments were repeated independently twice, with similar results. 374 375 We proceed to investigate the role of xan in promoting RF. Since the removal of xan in site L2 is not feasible due 376 its essential role in protein folding and stability (Dall’Osto et al., 2013), we analyzed the effect of altering L2 377 occupancy. To this aim, we isolated the PSI -LHCI supercomplex from the npq2 genotype, which has zeaxanthin 378 (zea) replacing vio (Ballottari et al., 2014) owing to the inactivation of zeaxanthin epoxidase (Niyogi et al., 1998) 379 (Fig. S15). The low-temperature fluorescence emission spectra revealed a ~2 nm blue shift in the npq2 PSI-LHCI 380 compared to the WT supercomplex (Fig. S20b). 381 We then recorded fluorescence excitation spectra to assess the different contributions to fluorescence emission 382 within the wavelength range of 450 to 520 nm for both the npq2 genotypes and the WT (Fig. 4b). As a result of 383 the vio→zea exchange, we observed a reduction in the contribution of xan to far-red emission, particularly evident 384 in the 510-540 nm range, similar to what observed for the a603-NH mutant (Fig. S19). The xan exchange also led 385 to a slight change in the chl a/b ratio, accounting for enhanced absorption at 465 nm and 652 nm (Fig . S20a), 386 consistent with previous observations (Ballottari et al., 2014) (Fig. S20d). 387 To analyze pigment-pigment interaction in WT and mutant LHCI complexes, we recorded CD spectra in the visible 388 region (350-750 nm). In the QY region, the spectra of all genotypes displayed signature peaks typical of LHC (i.e., 389 -/+/-), indicating a similar and conserved structural conformation and pigment organization (Mozzo et al., 2008). 390 The CD spectra of the WT and a603-NH complexes revealed a large difference in the far-red region (λ > 700 nm), 391 where the mutant showed a markedly reduced (-) signal associated with the excitonic interaction responsible for 392 the RF (Fig. 4c). The difference spectrum (black line) evidenced the disappearing of a (-) low-energy band in a603-393 NH LHCI (690-730 nm), which was compensated by a high-energy (+) band appearing at 684 nm. This conservative 394 signal aligns with the loss of excitonic interactions between chl a603 and a609, as previously reported (Morosinotto 395 et al., 2003; Wientjes et al., 2012). 396 In the Soret region, interpreting the CD spectra is complicated due to the superimposition of signals from chls and 397 cars. However, we detected notable differences between 465-530 nm when comparing both the a603-NH and npq2 398 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 15 samples with the WT. These differences can be attributed to genuine changes in pigment-pigment interactions, 399 variations in Xan composition (vio vs. zea) or minor losses of cars during purification (Fig. S20d). 400 Our observation suggested a minor, yet significant, role of the xan at L2 in interacting with the chl a603-a609 pair, 401 thus contributing to modulating the overall absorption of the cluster towards longer wavelengths. 402 403 Excited state calculations 404 To model the effect of the a603-NH mutation on the red-shifted absorption at the molecular level, we performed 405 structure-based polarizable quantum mechanics/molecular mechanics (QM/MM) calculations of excited states on 406 the Lhca4 WT and a603-NH mutant models, including charge-transfer excitations (Fig. 5a). In the spectra simulated 407 with the standard exciton model (noCT in Fig. 5c, Fig. S21), neither the WT nor the a603-NH exhibited a signature 408 of the RF observed in the experiments, notably the broad band peaking at >700 nm. Conversely, by adding the CT 409 contribution for the a603-a609 dimer in the calculations, we observed a low energy band ( ∼710 nm) in the 410 simulated spectrum of the WT (Fig. 5b), resembling the experimental observations in the isolated rLhca4 (Wientjes 411 et al., 2012). In contrast, virtually no change in the spectrum was observed for the mutant (Fig. 5c). This indicates 412 that the major contribution to the WT/ a603-NH absorption shift resides in the different extent of the coupling of 413 CT states to local excitations between chl a603-a609, supporting a previous proposal (Wientjes et al., 2012; Sláma 414 et al., 2023). 415 A secondary effect can be noticed in the WT spectrum simulated without CT states (Fig. 5c), namely the 416 appearance of a red-shifted shoulder, corresponding to a603-a609 absorption. While this shoulder is clearly more 417 blue-shifted than the red band at ~710 nm, it indicates red-shifted site energies for a603-a609, in contrast with the 418 a603-NH mutant. Nonetheless, only with CT states does the model predict the significantly red-shifted band 419 observed in the experiment. 420 We also assessed the electronic interactions between the chls dimer and L2 vio. To this end, we computed the 421 excited states for two supermolecules, one formed by a603 and a609 (dimer), and the other formed by the same 422 two chls and the vio (trimer). We investigated the energy of the lowest excited state responsible for fluorescence 423 emission, to explore the effect on the electronic structure of including or excluding vio (Fig. 5d). The small change 424 observed (15 to 50 cm -1) is consistent with the varying treatment of the car (as point charges in the dimer vs. an 425 active molecule in the trimer), indicating that Vio in L2 does not effectively mix its orbitals with the two chls and 426 therefore does not significantly contribute to CT excitation within the a603-a609 pair. The change in the difference 427 between the lowest dimer and trimer energy states in WT and a603 -NH (13 and 44 cm -1, respectively) suggests a 428 very minimal involvement of the car in L2 in tuning the low -energy absorption of the cluster, even smaller than 429 what our experimental results indicate (Fig. 4). 430 431 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 16 432 Figure 5. (a) Schematic depiction of local excitations (left) and charge transfer excitations (right). In local 433 excitations, electron transitions occur within orbitals of the same molecule; in charge-transfer excitations electrons 434 are promoted from orbitals localized on one molecule to orbitals localized on another molecule. (b) Comparison of 435 WT and a603-NH absorption spectra in the simulations (left) and experiments for isolated Lhca429 (right). (c) 436 Effect of adding CT excitations to the exciton model in the a603-NH mutant (left) and WT (right). (d) Energy of 437 the lowest excited state in the a603-a609 dimer and in the a603-a609-vio trimer. The experimental spectra of WT 438 and a603-NH are adapted with permission from Ref. (Wientjes et al., 2012). Copyright © 2012 Elsevier B.V. All 439 rights reserved. 440 441 442 The role of the extra chromophore chl a615 443 Lhca3 and Lhca4 bound an additional chl molecule, a615 (referred to as chl a617 in (Qin et al., 2015)), which was 444 coordinated by a His residue located on either the third or fourth turn of the C helix of Lhca3 and Lhca4, respectively 445 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 17 (Fig. 3). This pigment was absent in Lhca1 and Lhca2, leading to speculation about its involvement in RF formation 446 (Melkozernov & Blankenship, 2003) (Fig. 3), as the positioning of chl a615 allowed for favorable dipolar coupling 447 with chl a609. Notably, the EC value between chl a615 and chl a609 was 79 cm-1 for Lhca3, whereas it dropped to 448 19 cm-1 in Lhca4 due to the differing orientation of the chlorin ring in relation to chl a609. Although His residues 449 were also present in the second turn of the C helix of Lhca1 and Lhca2, the lack of electronic density from cryo -450 EM suggested that chl a615 was absent in these subunits. 451 An additional lut molecule was located near the chlorin ring of chl a615 in Lhca4 (Fig. S22a), positioned to allow 452 strong dipolar coupling (492 cm -1) with this chl. This lut had previously been observed only in the cryo -EM 453 structure of Z. mays PSI-LHCI (PDB 5ZJI (Pan et al., 2018)) and in the X -ray structure of P. sativum PSI-LHCI 454 (Qin et al., 2015). The xan was situated at the interface between Lhca1 and Lhca4, in contact with both subunits , 455 and one of its hydroxyl groups formed a hydrogen bond with the carbonyl oxygen of Ser210 in Lhca1 (Fig. S22b). 456 Consequently, it might got lost during the purification of monomeric Lhca4. 457 To investigate the potential role of chl a615 in far -red light absorption, we generated mutants lacking this 458 chromophore in both Lhca3 and Lhca4 by substituting the His-binding residue with non-binding Ala (a615-H→A) 459 or Ile (a615-H→I) (Remelli et al., 1999; Guardini et al., 2022). 460 When recording 77 K emission spectra from intact leaves of genotypes with and without chl a615, we observed a 461 small blue -shift from 736 nm to 733 ± 1 nm (Fig . 6a, b). However, in the isolated PSI -LHCI complex , both 462 genotypes showed the same emission at 734 nm (Fig. 5c). The ∼2-3 nm red -shift in leaf samples compared to 463 isolated supercomplexes can be attributed to self -absorption effects (Weis, 1985) . The a603-NH mutant and 464 koLhca3 koLhca4 exhibited a blue-shifted λmax to ~724 nm, similar to the emission of the PSI core complex (Croce 465 et al., 1998). We concluded that chl a615 does not play a role in forming RF. To explain the 3 nm shift observed 466 in leaf emission with and without chl a615, we analyzed the pigment-protein organization of thylakoids by sucrose 467 gradient ultracentrifugation. Fig. S23 compares the fractionation patterns from solubilized thylakoids from the WT 468 and chl a615-less mutants. In addition to the lowest (higher MW) band containing the fully assembled PSI-LHCI 469 supercomplex, a prominent green band containing the PSI-core complex was present in both a615 mutant lines, 470 while it was very faint in the WT (Wientjes et al., 2009). Consistently, the upper band (with the lowest MW) was 471 enriched in the mutants compared to the WT. We interpreted these results as indicating a de-stabilization of the 472 dimeric Lhca1-Lhca4 and Lhca2-Lhca3 dimers consequent to the missing chl a615. This was further confirmed by 473 a second sucrose gradient fractionation upon treating the isolated PSI-LHCI with Zwittergent-16, a procedure that 474 allows the isolation of LHCI dimers from the PSI core (Fig. 6d). The pattern from the WT yielded both monomers 475 and dimers of Lhcas, while only monomers were observed in both chl a615 mutant lines. These results support the 476 hypothesis that chl a615 plays a key role in stabilizing the dimeric Lhca complexes. 477 478 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 18 479 Figure 6. Spectral characteristic of different chl-binding mutants of A. thaliana. (a) Fluorescence emission 480 spectra (λexc in nm) measured on leaf extracts for different genotypes of A. thaliana: WT, koLhca3 koLhca4 lines 481 complemented with WT sequences of Lhca3 and Lhca4 (A3WT-A4WT), mutants lacking chl a615 (a615-HA and 482 a615-HI), koLhca3 koLhca4 (koA3A4), and a603-NH mutant (a603-NH), and normalized to the λ max. (b) Barplot 483 of the peak emission wavelength (λ max), measured on the same genotypes. The values on the individual bars 484 represent the mean λ max in nm, the error bars correspond to the standard deviations (n = 5 - 10). Values that are 485 significantly different (ANOVA followed by Tukey’s post-hoc test at a significance level of P < 0.05) are marked 486 with different letters. c) PSI-LHCI Fluorescence emission spectra (λexc in nm) measured on the WT, a603-HA and 487 a603-HI genotypes. d) Sucrose gradient fractionation of solubilized PSI -LHCI from WT, a603-HA and a603-HI 488 plants. Five pigment-containing bands were resolved and identified as: free-pigments, monomeric LHCA, dimeric 489 LHCA, PSI -core complex, and PSI -LHCI supercomplex. Experiments in panels c and d were repeated 490 independently twice, with similar results. 491 492 493 494 495 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 19

496 497 Figure 7. Structural superposition and excitonic coupling analysis (values in cm -1) of the red cluster pigments 498 from F. verschaffeltii (PDB 8WGH), Z. mays (PDB 5ZJI), A. thaliana, P. patens (PDB 7XQP), and C. reinhardtii 499 (PDB 7ZQC). The analysis focused on ( a) Lhca4 or the corresponding subunit, i.e. Lhca8 for C. reinhardtii and 500 Lhca2b for P. patens (Yan et al., 2021; Gorski et al., 2022) and (b) Lhca3. 501 502 A major trend in the evolution of PSI -LHCI within the green lineage has been the progressive increase in the 503 amplitude of the far-red tail of the absorption spectra. This change is consequent to the appearance of low-energy 504 absorption forms in LHCI, developed around 489-403 MYA, likely in response to the far -red enriched radiation 505 made available by the development of vegetation canopies (Fig. 1). 506 Notably, these variations in spectral properties occurred despite the structural similarities among their LHC subunits 507 (Iwai et al., 2024). This can be observed through the structural superposition of the red clusters from representative 508 C3 (A. thaliana and F. verschaffeltii), C4 (Z. mays), green algae (C. reinhardtii), and mosses (P. patens) Lhca4 and 509 Lhca3 (Fig. 7), or the corresponding homologous subunits. The substitution of Asn, coordinating chl a603, with a 510 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 20 His residue is a structural feature known to disrupt the excitonic interactions that generate the low-energy states in 511 At-rLhca3 and At-rLhca4 (Morosinotto et al., 2003; Wientjes et al., 2012; Li et al., 2023). However, the relationship 512 between Asn ligand and the presence of Lhca -associated RFs appears overly simplistic; indeed, in higher plants 513 which exhibit the strongest far-red emission (i.e., A. thaliana, Z. mays, and F. verschaffeltii), chl a603 is coordinated 514 by Asn residues in both Lhca3 and Lhca4, while in C. reinhardtii chl a603 is coordinated by a His residue in Lhca3 515 and Lhca8 ( the Lhca4 homologue) . Notably, the Asn ligand is present in other loosely bound Lhca subunits 516 (Stauber et al., 2009; Su et al., 2019) which are located in more distal positions relative to the PSI core complex 517 (Huang et al., 2021). Moreover, P. patens, which occupies an intermediate position both evolutionary and in terms 518 of absorption properties, has an Asn residue that coordinates chl a603 in Lhca3, but not in Lhca2b, its Lhca4 519 structural homologue. 520 Interestingly, Asn remains the coordinating residue for chl a603 in the seagrasses P. oceanica and C. nodosa, 521 despite their loss of RF after returning to the marine environment (Fig. 1). This observation reinforces the idea that 522 the nature of the chl a603 ligand is not the sole factor determining the presence of RF. 523 The PSI-LHCI a603-NH supercomplex showed a significant absorption reduction in the >700 nm region, along 524 with a ~13 nm blue -shifted fluorescence emission at cryogenic temperatures (Fig . 2). This shift is notably larger 525 than that observed in genotypes with the a603-NH mutation limited to the Lhca4 antenna (Li et al., 2023). In the 526 WT PSI-LHCI (Wientjes et al., 2009), the terminal emitter is the antenna complex , which emits at at 734 nm. In 527 contrast, the a603-NH mutant has the PSI core complex as the emitter, at 721 nm (Ihalainen et al., 2003; Akhtar & 528 Lambrev, 2020). This difference complicates the assessment of the real extent of the emission blue-shift caused by 529 the mutation in LHCI. This was possible upon purification of LHCI from WT and a603-NH mutant (Fig . 2), 530 allowing for the measurement of a 39 nm blue-shift between the peak emissions. This finding aligns with previous 531 studies on in vitro reconstituted complexes, which reported a shift of ~45 nm (Morosinotto et al., 2003; Wientjes 532 et al., 2012). 533 By comparing the cryo-EM structures of the WT and a603-NH mutant PSI-LHCI, we detected a ~0.7 Å shift in the 534 chlorin ring of chl a603, induced by the Asn→His substitution in Lhca3 and Lhca4 structures (Fig . 3, S13 ). 535 Although this structural difference is subtle, its impact on the coupling with charge-transfer (CT) states that underlie 536 RF formation could be considerable, given the sensitivity of these interactions to distance (Cupellini et al., 2018). 537 To connect the effect of this structural shift with the observed spectroscopic properties, we performed EC 538 calculations using the point-dipole approximation (Liguori et al., 2015). While the Asn→His substitution changed 539 the separation between the chlorin rings of chl a603 and chl a609, it also increased the distance between the chl 540 a603 chlorin ring and the neighboring vio L2. This change led to a substantial change in the chl a603–vio L2 541 coupling value, which dropped from ~530 cm-1 in the WT to ~380 cm-1 in the a603-NH mutant for both Lhca3 and 542 Lhca4 subunits (Fig. 3). 543 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 21 We noticed that the chl a603–vio L2 coupling value serves as a good indicator of the presence of RF (Fig . 3, 6, 544 S22). Species lacking RF (C. reinhardtii) showed values 500 cm -1. Interestingly, P. patens, which has 546 red absorption properties that are intermediate between those of plants and algae, has EC values >500 cm-1 only in 547 Lhca3. Moreover, the strength of the chl a603–vio L2 coupling correlates well with the fluorescence emission λmax 548 (Fig. 1). 549 550 Based on this structural and computational analysis, we investigated the influence of the xan in position L2 on low-551 energy absorption, by applying two investigation strategies. First, we compared 77 K excitation spectra of the a603-552 NH and WT PSI-LHCI (Fig. 4) and observed a systematically reduced intensity (by 6%) in the 455-505 nm region 553 of the a603-NH PSI-LHCI supercomplex, an effect that can be attributed to the contribution of cars (Ashenafi et 554 al., 2023). Second, we analyzed the PSI-LHCI supercomplex from the npq2 genotype, in which the L2 vio binding 555 site in PSI-LHCI supercomplex is occupied by zea (Ballottari et al., 2014). The structural difference between the 556 two xan caused a ~2 nm blue shift in npq2 compared to WT 77 K fluorescence emission spectra and a concomitant 557 reduction in the amplitude of the red -emission tail at 752 nm (Fig . S20). Interestingly, in the 480-510 nm region, 558 where xan absorption is prominent, the PSI-LHCI from npq2 showed a lower contribution to far-red fluorescence 559 emission compared to WT PSI-LHCI (Fig. 4b), consistent with the amplitude of the 516 nm CD signal (Fig. 4d). 560 Using QM/MM analysis of the excited states of the dimer a603–a609 and the trimer a603–a609–vio(L2) on the 561 WT and a603-NH experimental structures of Lhca4 we show that the local rearrangement of the pigments a609–562 a603 due to the a603-NH substitution impacts the electronic overlap between orbitals (Fig . S25), leading to a 563 decrease in CT couplings that are responsible for the low-lying red-states in the WT (Fig. 5b). 564 Excited state calculations showed that the standard excitonic model was insufficient to explain the red -shifted 565 absorption of the pigment cluster, while the introduction of CT states was needed to best simulate the experimental 566 absorption spectra (Fig. 5c). 567 While the inclusion of vio(L2) in the computed CT states has only a minor contribution to the far -red absorption 568 (Fig. 5c). 569 Our structural, spectroscopic and multiscale excited -state calculations analysis, suggests that while the origin of 570 the low-energy absorption in PSI-LHCI and of RF lies in the CT states of the chl a603 –a609 pair, the xan in L2 571 may play a minor role in modulating the energy transfer towards the chl a603 –a609 pair, thus contributing to the 572 extension of the absorption of far-red photons. 573 One additional chl, chl a615 was located close to the “red cluster” specific of the Lhca3 and Lhca4 isoforms of 574 higher plants (Melkozernov & Blankenship, 2003) . Chl a615 exhibited EC values with chl a609 similar to the 575 canonical red pair chl a603–a609, especially in the Lhca3 isoforms due to the more favorable orientation (Fig. 7). 576 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 22 Upon s ite-directed mutagenesis to selectively remove chl a615 in both Lhca3 and Lhca4 , low-temperature 577 fluorescence emission spectra allowed ruling out the involvement of Chl a615 in RF. Nevertheless, this experiment 578 showed that chl a615 in Lhca4 was needed for stabilization of the Lhca1-Lhca4 dimer and/or its assembly with the 579 PSI-core (Fig. 6d), (Fig. S23). Taken together, our detailed analysis of the large 3-chl-1-xan pigment assembly in 580 the LHCI domain expressing RF highlights the role of the charge-transfer states within the a603–a609 pair (Sláma 581 et al., 2023). Xan L2 effect was determined to be of minor importance despite undergoing the largest structural 582 change when comparing WT vs a603-NH Lhca3 and Lhca4. Finally, we report on two chl a615 and one lut ligand, 583 which, although not contributing to the formation of RF, were proven to substantially contribute to the assembly of 584 LHCI dimerization and assembly with PSI-core complex. 585 Further research will be necessary to better clarify the contribution of the protein environment surrounding the “red 586 cluster,” for example, by comparing high-resolution structures of red-shifted and blue-shifted Lhcas such as from 587 F. verschaffeltii and seagrasses. 588 Tuning chl absorption towards spectral regions where photons are available, especially under dense canopies, is a 589 privileged strategy for enhancing crop photosynthetic yield. The detailed knowledge of structure-function relations 590 within LHC antenna proteins will contribute to the rational engineering of crops (Ort et al., 2015; Cutolo et al., 591 2023). 592 593 ACKNOWLEDGMENT 594 RB acknowledges financial support from the European Research Council (ERC Advanced Grant 101053983 -595 GrInSun). Part of this work was carried out at Unitech NOLIMITS, an advanced imaging facility established by 596 the University of Milan. The authors acknowledge Dr. Gabriele Procaccini from Anton Dohrn Experimental Marine 597 Station (Naples) for providing P. oceanica and C. nodosa plants. 598 COMPETING INTERESTS 599 None declared. 600 AUTHOR CONTRIBUTIONS 601 R.B., S .C. and L .D. conceived the work and designed the experiments. Z .G., V .F.P. and A .A. carried out the 602 construction of mutants and performed their biochemical and spectroscopical characterization. Z.G., S.C. and A.A. 603 carried out the preparation of the samples for cryo-EM. D.M.V.B. e A.C-S. conducted cryo-EM sample preparation 604 and data collection. S.C. and D.M. analyzed cryo-EM data and reconstructed PSI-LHCI structures. D.M. performed 605 bioinformatics analysis and EC calculations. E .B., C .J., L .P-G., L .C. and B .M. carried out quantum chemical 606 calculations and analyses. D.M., S.C., and Z.G. wrote the original text draft, and all authors discussed the results 607 and contributed to drafting the manuscript. 608 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 23 609 DATA AVAILABILITY 610 Sequence data from this article can be found in the Arabidopsis Genome Initiative under accession numbers 611 At1g61520 (LHCA3), At3g47470 (LHCA4) and At5g67030 (ZE). The KO lines were obtained in the NASC under 612 stock numbers N876497 (koLhca3), N679009 (koLhca4) and N60000 (npq2). 613 The Cryo-EM maps and coordinates have been deposited in the EMDB and wwPDB, respectively: PSI–LHCI WT 614 (cryo-EM map, EMD-51219; consensus refinement map; PDB: 9GBI), PSI–LHCI a603-NH mutant (cryo-EM map, 615 EMD-51227; PDB: 9GC2). 616

617 Adams PD, Afonine P V, Bunkóczi G, Chen VB, Davis IW, Echols N, Headd JJ, Hung L -W, Kapral GJ, 618 Grosse-Kunstleve RW, et al. 2010. PHENIX: a comprehensive Python-based system for macromolecular structure 619 solution. Acta Crystallographica Section D: Biological Crystallography 66: 213–221. 620 Akhtar P, Lambrev PH . 2020. On the spectral properties and excitation dynamics of long -wavelength 621 chlorophylls in higher-plant photosystem I. Biochimica et Biophysica Acta - Bioenergetics 1861: 148274. 622 van Amerongen H, van Grondelle R. 2001. Understanding the energy transfer function of LHCII, the major light-623 harvesting complex of green plants. The Journal of Physical Chemistry B 105: 604–617. 624 Ashenafi EL, Nyman MC, Shelley JT, Mattson NS. 2023. Spectral properties and stability of selected carotenoid 625 and chlorophyll compounds in different solvent systems. Food Chemistry Advances 2: 100178. 626 Ballottari M, Alcocer MJP, D’Andrea C, Viola D, Ahn TK, Petrozza A, Polli D, Fleming GR, Cerullo G, 627 Bassi R. 2014. Regulation of photosystem I light harvesting by zeaxanthin. Proceedings of the National Academy 628 of Sciences of the United States of America 111: 2431–2438. 629 Bassi, R, Hinz U, Barbato R . 1985. The role of the light harvesting complex and photosystem II in thylakoid 630 stacking in the chlorina-f2 barley mutant. Carlsberg Res.Commun. 50: 347–367. 631 Bassi R, Simpson DJ . 1987. Chlorophyll‐protein complexes of barley photosystem I. European Journal of 632 Biochemistry 163: 221–230. 633 Bepler T, Morin A, Rapp M, Brasch J, Shapiro L, Noble AJ, Berger B. 2019. Positive-unlabeled convolutional 634 neural networks for particle picking in cryo-electron micrographs. Nature methods 16: 1153–1160. 635 Bondanza M, Nottoli M, Cupellini L, Lipparini F, Mennucci B . 2020. Polarizable embedding QM/MM: the 636 future gold standard for complex (bio) systems? Physical Chemistry Chemical Physics 22: 14433–14448. 637 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 24 Bressan M, Dall’Osto L, Bargigia I, Alcocer MJP, Viola D, Cerullo G, D’Andrea C, Bassi R, Ballottari M . 638 2016. LHCII can substitute for LHCI as an antenna for photosystem i but with reduced light -harvesting capacity. 639 Nature Plants 2: 1–10. 640 Carbonera D, Agostini G, Morosinotto T, Bassi R. 2005. Quenching of chlorophyll triplet states by carotenoids 641 in reconstituted Lhca4 subunit of peripheral light -harvesting complex of photosystem I. Biochemistry 44: 8337–642 8346. 643 Castelletti S, Morosinotto T, Robert B, Caffarri S, Bassi R, Croce R . 2003. Recombinant Lhca2 and Lhca3 644 subunits of the photosystem I antenna system. Biochemistry 42: 4226–4234. 645 Chazaux M, Schiphorst C, Lazzari G, Caffarri S . 2022. Precise estimation of chlorophyll a, b and carotenoid 646 content by deconvolution of the absorption spectrum and new simultaneous equations for Chl determination. Plant 647 Journal 109: 1630–1648. 648 Chen VB, Arendall WB, Headd JJ, Keedy DA, Immormino RM, Kapral GJ, Murray LW, Richardson JS, 649 Richardson DC . 2010. MolProbity: all -atom structure validation for macromolecular crystallography. Acta 650 Crystallographica Section D: Biological Crystallography 66: 12–21. 651 Croce R, van Amerongen H . 2020. Light harvesting in oxygenic photosynthesis: Structural biology meets 652 spectroscopy. Science (New York, N.Y.) 369: 933. 653 Croce R, Van Amerongen H. 2013. Light-harvesting in photosystem I. Photosynthesis Research 116: 153–166. 654 Croce R, Morosinotto T, Castelletti S, Breton J, Bassi R . 2002. The Lhca antenna complexes of higher plants 655 photosystem I. Biochimica et Biophysica Acta - Bioenergetics 1556: 29–40. 656 Croce R, Müller MG, Bassi R, Holzwarth AR. 2001. Carotenoid-to-chlorophyll energy transfer in recombinant 657 major light -harvesting complex (LHCII) of higher plants. I. Femtosecond transient absorption measurements. 658 Biophysical Journal 80: 901–915. 659 Croce R, Zucchelli G, Garlaschi FM, Jennings RC . 1998. A thermal broadening study of the antenna 660 chlorophylls in PSI- 200, LHCI, and PSI core. Biochemistry 37: 17355–17360. 661 Cupellini L, Caprasecca S, Guido CA, Müh F, Renger T, Mennucci B. 2018. Coupling to charge transfer states 662 is the key to modulate the optical bands for efficient light harvesting in purple bacteria. The Journal of Physical 663 Chemistry Letters 9: 6892–6899. 664 Cutolo EA, Guardini Z, Dall’Osto L, Bassi R . 2023. A paler shade of green : engineering cellular chlorophyll 665 content to enhance photosynthesis in crowded environments. New Phytologist 293(5): 1567–1583. 666 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 25 Dall’Osto L, Piques M, Ronzani M, Molesini B, Alboresi A, Cazzaniga S, Bassi R. 2013. The Arabidopsis nox 667 mutant lacking carotene hydroxylase activity reveals a critical role for xanthophylls in photosystem I biogenesis. 668 Plant Cell 25: 591–608. 669 Elias E, Oliver TJ, Croce R . 2024. Oxygenic Photosynthesis in Far -Red Light: Strategies and Mechanisms. 670 Annual Review of Physical Chemistry 75: 231–256. 671 Emsley P, Cowtan K. 2004. Coot: model-building tools for molecular graphics. Acta crystallographica section D: 672 biological crystallography 60: 2126–2132. 673 Friedl C, Fedorov DG, Renger T . 2022. Towards a quantitative description of excitonic couplings in 674 photosynthetic pigment--protein complexes: Quantum chemistry driven multiscale approaches. Physical Chemistry 675 Chemical Physics 24: 5014–5038. 676 Georgakopoulou S, van der Zwan G, Bassi R, van Grondelle R, van Amerongen H, Croce R . 2007. 677 Understanding the changes in the circular dichroism of light harvesting complex II upon varying its pigment 678 composition and organization. Biochemistry 46: 4745–4754. 679 Gilmore AM, Yamamoto HY. 1991. Resolution of lutein and zeaxanthin using a non-endcapped, lightly carbon-680 loaded C18 high-performance liquid chromatographic column. Journal of Chromatography A 543: 137–145. 681 Gobets B, Van Grondelle R. 2001. Energy transfer and trapping in photosystem I. Biochimica et Biophysica Acta 682 - Bioenergetics 1507: 80–99. 683 Goddard TD, Huang CC, Meng EC, Pettersen EF, Couch GS, Morris JH, Ferrin TE. 2018. UCSF ChimeraX: 684 Meeting modern challenges in visualization and analysis. Protein science 27: 14–25. 685 Gorski C, Riddle R, Toporik H, Da Z, Dobson Z, Williams D, Mazor Y . 2022. The structure of the 686 Physcomitrium patens photosystem I reveals a unique Lhca2 paralogue replacing Lhca4. Nature Plants 8: 307–687 316. 688 Guardini Z, Bressan M, Caferri R, Bassi R, Dall’Osto L . 2020. Identification of a pigment cluster catalysing 689 fast photoprotective quenching response in CP29. Nature Plants 6: 303–313. 690 Guardini Z, Gomez RL, Caferri R, Dall’Osto L, Bassi R . 2022. Loss of a single chlorophyll in CP29 triggers 691 re-organization of the Photosystem II supramolecular assembly. BBA - Bioenergetics 1863: 148555. 692 Huang Z, Shen L, Wang W, Mao Z, Yi X, Kuang T, Shen JR, Zhang X, Han G. 2021. Structure of photosystem 693 I-LHCI-LHCII from the green alga Chlamydomonas reinhardtii in State 2. Nature Communications 12: 1–14. 694 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 26 Ihalainen JA, Rätsep M, Jensen PE, Scheller HV, Croce R, Bassi R, Korppi-Tommola JEI, Freiberg A. 2003. 695 Red spectral forms of chlorophylls in green plant PSI - a site-selective and high-pressure spectroscopy study. The 696 Journal of Physical Chemistry B 107: 9086–9093. 697 Iwai M, Patel-Tupper D, Niyogi KK. 2024. Structural Diversity in Eukaryotic Photosynthetic Light Harvesting. 698 Annual Review of Plant Biology 75: 1–34. 699 Jansson S. 1999. A guide to the Lhc genes and their relatives in Arabidopsis. Trends in Plant Science 4: 236–240. 700 Jennings RC, Zucchelli G, Croce R, Garlaschi FM . 2003. The photochemical trapping rate from red spectral 701 states in PSI -LHCI is determined by thermal activation of energy transfer to bulk chlorophylls. Biochimica et 702 Biophysica Acta - Bioenergetics 1557: 91–98. 703 Jennings RC, Zucchelli G, Santabarbara S . 2013. Photochemical trapping heterogeneity as a function of 704 wavelength, in plant photosystem i (PSI-LHCI). Biochimica et Biophysica Acta - Bioenergetics 1827: 779–785. 705 Karapetyan N V, Dorra D, Schweitzer G, Bezsmertnaya IN, Holzwarth AR. 1997. Fluorescence spectroscopy 706 of the longwave chlorophylls in trimeric and monomeric photosystem I core complexes from the cyanobacterium 707 Spirulina platensis. Biochemistry 36: 13830–13837. 708 Kühlbrandt W, Wang DN, Fujiyoshi Y . 1994. Atomic model of plant light -harvesting complex by electron 709 crystallography. Nature 367: 614–621. 710 Li X, Zhu L, Song J, Wang W, Kuang T, Yang G, Hao C, Qin X . 2023. LHCA4 residues surrounding red 711 chlorophylls allow for fine -tuning of the spectral region for photosynthesis in Arabidopsis thaliana. Frontiers in 712 Plant Science 13: 1–12. 713 Liguori N, Periole X, Marrink SJ, Croce R. 2015. From light-harvesting to photoprotection: Structural basis of 714 the dynamic switch of the major antenna complex of plants (LHCII). Scientific Reports 5: 2–11. 715 Martínez-García JF, Galstyan A, Salla-Martret M, Cifuentes-Esquivel N, Gallemí M, Bou-Torrent J. 2010. 716 Regulatory components of shade avoidance syndrome. In: Advances in botanical research. Elsevier, 65–116. 717 Mazor Y, Borovikova A, Caspy I, Nelson N . 2017. Structure of the plant photosystem I supercomplex at 2.6 Å 718 resolution. Nature plants 3: 1–9. 719 Melkozernov AN, Blankenship RE . 2003. Structural Modeling of the Lhca4 Subunit of LHCI -730 Peripheral 720 Antenna in Photosystem I Based on Similarity with LHCII. Journal of Biological Chemistry 278: 44542–44551. 721 Moriarty NW, Grosse-Kunstleve RW, Adams PD. 2009. electronic Ligand Builder and Optimization Workbench 722 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 27 (eLBOW): a tool for ligand coordinate and restraint generation. Acta Crystallographica Section D: Biological 723 Crystallography 65: 1074–1080. 724 Morosinotto T, Breton J, Bassi R, Croce R. 2003. The nature of a chlorophyll ligand in Lhca proteins determines 725 the far red fluorescence emission typical of photosystem I. The Journal of biological chemistry 278: 49223–49229. 726 Mozzo M, Dall’Osto L, Hienerwadel R, Bassi R, Croce R . 2008. Photoprotection in the antenna complexes of 727 photosystem II: role of individual xanthophylls in chlorophyll triplet quenching. J.Biol Chem. 283: 6184–6192. 728 Mozzo M, Mantelli M, Passarini F, Caffarri S, Croce R, Bassi R . 2010. Functional analysis of Photosystem I 729 light-harvesting complexes (Lhca) gene products of Chlamydomonas reinhardtii. Biochimica et Biophysica Acta 730 (BBA)-Bioenergetics 1797: 212–221. 731 Müh F, Madjet ME-A, Renger T. 2010. Structure-based identification of energy sinks in plant light -harvesting 732 complex II. The Journal of Physical Chemistry B 114: 13517–13535. 733 Nelson N . 2024. Investigating the Balance between Structural Conservation and Functional Flexibility in 734 Photosystem I. International Journal of Molecular Sciences 25: 5073. 735 Niyogi KK, Grossman AR, Björkman O . 1998. Arabidopsis mutants define a central role for the xanthophyll 736 cycle in the regulation of photosynthetic energy conversion. Plant Cell 10: 1121–1134. 737 Nottoli M, Jurinovich S, Cupellini L, Gardiner AT, Cogdell R, Mennucci B. 2018. The role of charge-transfer 738 states in the spectral tuning of antenna complexes of purple bacteria. Photosynthesis research 137: 215–226. 739 Ort DR, Merchant SS, Alric J, Barkan A, Blankenship RE, Bock R, Croce R, Hanson MR, Hibberd JM, 740 Long SP, et al. 2015. Redesigning photosynthesis to sustainably meet global food and bioenergy demand. 741 Proceedings of the National Academy of Sciences of the United States of America 112: 8529–8536. 742 Pan X, Cao P, Su X, Liu Z, Li M. 2020. Structural analysis and comparison of light-harvesting complexes I and 743 II. Biochimica et Biophysica Acta - Bioenergetics 1861: 148038. 744 Pan X, Ma J, Su X, Cao P, Chang W, Liu Z, Zhang X, Li M . 2018. Structure of the maize photosystem I 745 supercomplex with light-harvesting complexes I and II. Science 360: 1109–1113. 746 Park Y, Runkle ES. 2017. Far-red radiation promotes growth of seedlings by increasing leaf expansion and whole-747 plant net assimilation. Environmental and Experimental Botany 136: 41–49. 748 Polívka T, Frank HA . 2010. Molecular factors controlling photosynthetic light harvesting by carotenoids. 749 Accounts of chemical research 43: 1125–1134. 750 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 28 Punjani A, Rubinstein JL, Fleet DJ, Brubaker MA. 2017. cryoSPARC: algorithms for rapid unsupervised cryo-751 EM structure determination. Nature methods 14: 290–296. 752 Punjani A, Zhang H, Fleet DJ . 2020. Non-uniform refinement: adaptive regularization improves single-particle 753 cryo-EM reconstruction. Nature methods 17: 1214–1221. 754 Qin X, Suga M, Kuang T, Shen JR . 2015. Structural basis for energy transfer pathways in the plant PSI -LHCI 755 supercomplex. Science 348: 989–995. 756 Remelli R, Varotto C, Sandonà D, Croce R, Bassi R. 1999. Chlorophyll binding to monomeric light-harvesting 757 complex. A mutation analysis of chromophore -binding residues. Journal of Biological Chemistry 274: 33510–758 33521. 759 Rivadossi A, Zucchelli G, Garlaschi FM, Jennings RC. 1999. The importance of PS I chlorophyll red forms in 760 light-harvesting by leaves. Photosynthesis Research 1999 60:2 60: 209–215. 761 Romero E, Mozzo M, Van Stokkum IHM, Dekker JP, Van Grondelle R, Croce R. 2009. The origin of the low-762 energy form of photosystem I light-harvesting complex Lhca4: mixing of the lowest exciton with a charge-transfer 763 state. Biophysical journal 96: 35–37. 764 Sardar S, Caferri R, Camargo FVA, Capaldi S, Ghezzi A, Dall’Osto L, D’Andrea C, Cerullo G, Bassi R, 765 Dall’Osto L, et al. 2024. Site-Directed Mutagenesis of the Chlorophyll -Binding Sites Modulates Excited -State 766 Lifetime and Chlorophyll-Xanthophyll Energy Transfer in the Monomeric Light -Harvesting Complex CP29. The 767 Journal of Physical Chemistry Letters 15: 3149–3158. 768 Sen S, Mascoli V, Liguori N, Croce R, Visscher L . 2021. Understanding the relation between structural and 769 spectral properties of light-harvesting complex II. The Journal of Physical Chemistry A 125: 4313–4322. 770 Sláma V, Cupellini L, Mascoli V, Liguori N, Croce R, Mennucci B . 2023. Origin of Low-Lying Red States in 771 the Lhca4 Light-Harvesting Complex of Photosystem I. Journal of Physical Chemistry Letters 14: 8345–8352. 772 Stauber EJ, Busch A, Naumann B, Svatos A, Hippler M . 2009. Proteotypic profiling of LHCI from 773 Chlamydomonas reinhardtii provides new insights into structure and function of the complex. Proteomics 9: 398–774 408. 775 Su X, Ma J, Pan X, Zhao X, Chang W, Liu Z, Zhang X, Li M. 2019. Antenna arrangement and energy transfer 776 pathways of a green algal photosystem-I–LHCI supercomplex. Nature Plants 5: 273–281. 777 Weis E . 1985. Chlorophyll fluorescence at 77 K in intact leaves: characterization of a technique to eliminate 778 artifacts related to self-absorption. Photosynthesis research 6: 73–86. 779 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 29 Wientjes E, Croce R. 2011. The light-harvesting complexes of higher-plant Photosystem I: Lhca1/4 and Lhca2/3 780 form two red-emitting heterodimers. Biochemical Journal 433: 477–485. 781 Wientjes E, Oostergetel GT, Jansson S, Boekema EJ, Croce R . 2009. The role of Lhca complexes in the 782 supramolecular organization of higher plant photosystem I. Journal of Biological Chemistry 284: 7803–7810. 783 Wientjes E, Roest G, Croce R . 2012. From red to blue to far -red in Lhca4: How does the protein modulate the 784 spectral properties of the pigments? Biochimica et Biophysica Acta - Bioenergetics 1817: 711–717. 785 Yan Q, Zhao L, Wang W, Pi X, Han G, Wang J, Cheng L, He Y -K, Kuang T, Qin X, et al. 2021. Antenna 786 arrangement and energy-transfer pathways of PSI -LHCI from the moss Physcomitrella patens. Cell discovery 7: 787 10. 788 Zhang X, Henriques R, Lin S -S, Niu Q -W, Chua N -H. 2006. Agrobacterium -mediated transformation of 789 Arabidopsis thaliana using the floral dip method. Nature Protocols 1: 641–646. 790 791 SUPPORTING INFORMATION 792 Additional Supporting Information may be found online in the Supporting Information section at the end of the 793 article. Supplemental methods. Figure S1: Sample preparation and characterization. Figure S2: Cryo-EM data 794 processing workflow for AtPSI-LHCI WT. Figure S3: Cryo-EM data processing workflow for AtPSI-LHCI a603-795 NH. Figure S4: Spectral characteristic of different of A. thaliana WT and a603-NH mutant. Figure S5: Absorption 796 and emission spectra of PSI core complex of A. thaliana WT and a603-NH mutant. Figure S6: Atomic models of 797 PSI-LHCI subunits and selected ligands superimposed on cryo -EM maps. Figure S7: Top and side view of the 798 PSI-LHCI WT supercomplex adn pigment organization. Figure S8: Positions of ligands in the PSI -LHCI WT of 799 A.thaliana. Figure S9: Lipid molecules found in PSI -LHCI supercomplexes. Figures S10: Superposition of the 800 PSI WT from A. thaliana (PDB 9GBI, white) and from P. sativum (PDB 7DKZ) Figure S11: Global superposition 801 RMSD of the AtPSI-WT (PDB 9GBI) structure with PSI-WT structures from PDB 8J7B, 8JZA (both from cryo -802 em data), and 7DKZ. Figure S12: Position of chl, carotenoids and lipid molecules in PSI-LHCI supercomplexes. 803 Figure S13: Pigment content of the LHCI antenna subunits of A. thaliana WT (PDB 9GBI). Figure S14: Atomic 804 models of the “red cluster” pigments in Lhca3 and Lhca4. Figure S15: Pigment distances in the “red cluster” of 805 Lhca3 and Lhca4. Figure S16: Top and side view of superimposed structures of chl a603, chl a609, chl a615, 806 Violaxanthin L2 and Lutein. Figure S17: Top and side view of superimposed structures of chl a603, chl a609, chl 807 a615, Violaxanthin L2 and Lutein. Figure S18: Excitonic coupling absolute values between pigments in LHCI WT 808 and a603-NH. Figure S19: Difference spectrum between RT absorption and 77 K excitation spectra in the 425-690 809 nm region of PSI-LHCI from A. thaliana WT and a603-NH. Figure S20: Absorption and emission spectra of PSI-810 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint 30 LHCI complex of A. thaliana WT and and npq2 mutant, absorption spectra and pigment composition of LHCI 811 complexes of A. thaliana WT and and npq2 mutant. Figure S21: Site energies of Lhca4 chls for WT and a603 -812 NH mutant. Absorption spectra of WT and mutant simulated using the simple exciton model. Figure S22: 813 Structural superposition of Lhca4 WT from A. thaliana (PDB 9GBI,), P. sativum (PDB 5L8R), Z. mays (PDB 814 5ZJI). Cryo-em map for Lut over chl a615 in Lhca4. Figure S23: Sucrose gradient fractionation of thylakoid 815 membranes of WT and a615 mutant lines. Figure S24: Superposition of Lhca4, Lhca2 from A. thaliana WT (PDB 816 9GBI) and Lhcb4 from P. sativum (PDB 5XNL) and excitonic couplings between pigments pairs in the red cluster. 817 Figure S25: Visual representation of the overlap between LUMO orbitals of chls a603 and a609 as computed for 818 the WT and a603-NH structures. Figure S26: Sequence alignment of the sequences of the synthetic genes encoding 819 for Lhca3/a4_WT and Lhca3/a4_a615 -HA and Lhca3/a4_a615 -HI. Table S1: AtPSI-603-NH structural model. 820 CLA, chl a; CHL, chl b; BCR, β -carotene; LUT, lutein; XAT, violaxanthin; DGD, digalactosyl -diacyl glycerol 821 (DGDG); LMG, 1,2-distearoyl-monogalactosyl-digliceride; LHG, 11,2-dipalmitoyl-phosphatidyl-glycerol; LMT, 822 dodecyl-β-D-maltoside; PQN, phylloquinone; SF4, Fe4 -S4 cluster. Table S2: List of the primers used to obtain 823 and characterize the a603-NH mutant lines. Table S3: Cryo-EM data collection, refinement, and validation 824 statistics. Supplemental References. 825 826 827 828 .CC-BY-NC 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 6, 2025. ; https://doi.org/10.1101/2025.05.05.652163doi: bioRxiv preprint