Structural insights into assembly and environmental adaptation of photosystem I from a lichen green alga

Abstract

Photosystem I (PSI) of terrestrial plants typically contains an antenna system smaller than that of aquatic algae, presumably an adaptation to terrestrial high-light conditions. Moreover, green plants face multiple other stresses including drought when colonizing terrestrial environments and have evolved effective protective mechanisms. However, how exactly PSI antenna size is reduced and the precise nature of these photoprotective strategies remain to be elucidated. Lichens, composed of a mycobiont and a phycobiont, represent a symbiotic life form that thrives under terrestrial-like stress, with their phycobionts evolving in parallel with initial land-colonizing plants. Here, we report four structures of PSI complexes from a lichen phycobiont, Diplosphaera chodatii, at resolutions ranging from 3.0-Å to 4.1-Å, revealing the structural plasticity, the unique pigment composition and arrangement of DcPSI complexes. Our results provide insights into protective mechanisms of D. chodatii and potential structural strategies for reducing PSI antenna size to adapt to terrestrial habitats.

Land plant PSI binds fewer LHCIs than green algal PSI, likely for high-light adaptation. Here, the authors report four PSI structures with variable LHCI stoichiometry from a lichen green alga, providing insights into the PSI assembly, structural plasticity and high-light-adapted LHCI adjustments.

Introduction

Photosynthesis, the process by which plants harness sunlight to produce energy, is driven by an intricately interwoven cellular machinery composed of interacting molecular complexes. The absorption, transfer and conversion of light energy are carried out by photosystem I (PSI) and photosystem II (PSII) embedded in the thylakoid membranes¹. PSI is a multi-subunit protein-pigment complex and one of the most efficient photon energy conversion apparatuses found in nature. In addition, PSI participates in multiple regulatory steps involved in fine-tuning photosynthetic light reactions, including state transitions and cyclic electron transfer^2–5.

Eukaryotic PSI contains a core complex and a peripheral antenna system, with the latter being primarily composed of light-harvesting complexes I (LHCIs) encoded by multiple lhca genes. To date, high-resolution structures of PSI-LHCI have been reported for multiple green plant species^6–16. Comparison of these structures has shown that while their core complexes are highly conserved, both pigment composition and protein arrangement of peripheral LHCI antennae differ significantly, presumably due to their diverse living environments and different evolutionary stages¹⁷. For example, PSI complexes of several green algae, such as Chlamydomonas reinhardtii (Cr) and Bryopsis corticulans (Bc), bind up to ten Lhca proteins. These antennae are arranged in the shape of a parallel bilayer belt that is attached to the PSI core at the PsaF/PsaJ side, as well as a side belt associated with PsaG/PsaH. The side belt consists of two Lhca subunits, whereas the inner and outer layer of the bilayer belt contains four Lhcas each^6–8. Another green alga, Dunaliella salina, appears to have lost the outer Lhca layer of its PSI, which contains only the inner and side belts⁹. In comparison, PSI structures analyzed for land plants, such as Pisum sativum, Arabidopsis thaliana, Fittonia albivenis, and moss Physcomitrium patens have been shown to bind only the inner belt composed of four Lhca proteins^10–16. It has been widely accepted that land plants evolved from green algae that originally occupied aquatic environments¹⁸. While analysis of a number of PSI structures from different species suggested that land plants discarded the outer and side layer antennae of PSI during the transition from aquatic to terrestrial habitats, how exactly these antennae were lost has remained unanswered.

Viridiplantae are one of the three major lineages of eukaryotic phototrophs¹⁹ and comprised of Chlorophyta and Streptophyta, including green algae and land plants²⁰. Within Chlorophyta and Streptophyta, transitions from freshwater to terrestrial habitats occurred independently multiple times across a time span of 700-650 Million years ago (Mya)²⁰. Lichens, occurring in most terrestrial ecosystems, constitute mutualistic symbionts composed of mycobiont (a fungus) and phycobiont (a green alga or a cyanobacterium)²¹. Over 90% of lichens are associated with Chlorophyta as their phycobiont, while ~50% to 70% have formed symbiotic relationships with green algae of the Trebouxiophyceae²². Trebouxiophyceae is one of the desert lineages of green algae, arising from clades of aquatic freshwater organisms^23,24. The lichen phycobiont group in Trebouxiophyceae mainly occurred in the above-mentioned age of transitions from freshwater to terrestrial habitats within Chlorophyta and Streptophyta (700-650 Mya)²⁰, and distributes worldwide in various terrestrial environments similar to land plants, but are embedded in fungi to jointly form lichens^23,24. Therefore, the symbiotic life form of lichens is likely to represent an important step towards land colonization by eukaryotic organisms²⁵, especially by the green algae of Chlorophyta, which occurred in parallel to the evolutionary event of Streptophyta developing from the charophyte to the land plants^26,27. On the basis of these phenomena, lichen phycobionts may have faced similar terrestrial environmental stresses as land plants, and may therefore exhibit convergent adaptation. The photosystems of lichen phycobionts may possess structural features similar to those evolved by plants during their aquatic-to-terrestrial transition, representing an adaptive strategy parallel to that of terrestrial plants.

In addition, lichens are considered the pioneer organisms on Earth for their ability to inhabit extremely challenging environments such as deserts and thus establishing conditions for ecosystems to develop. Endocarpon pusillum is a common lichen species found in desert regions²⁸, and generally exists in microbiotic crust (MBC) form. The phycobiont of E. pusillum, namely Diplosphaera chodatii, is affiliated with Trebouxiophyceae (Chlorophyta)²⁰. One previous study showed that under drought stress, D. chodatii isolated from E. pusillum can survive independently in laboratory conditions for ~two months²⁹, suggesting that this lichen phycobiont has a strong protective regulatory capacity and good drought-tolerant activity. Several previous reports suggested that under stress conditions, algae protect their photosynthetic machineries by down-regulating their photosynthesis and activating non-photochemical quenching (NPQ), which reduces the reactive oxygen species generated^30–33. Furthermore, one recent study using Jatropha curcas, a drought-tolerant plant, demonstrated that under long-term drought conditions, NPQ occurs in PSI in addition to PSII³⁴. Moreover, it has been previously shown that the xanthophyll cycle carotenoid zeaxanthin (Zea) can bind to PSI-LHCI and play a role in PSI photoprotection³⁵. Another report suggested that lichen phycobionts are able to perform rapid changes of their photosynthetic apparatus to adapt to the environments³⁶. However, a more detailed understanding of the stress response and photoprotection of lichen phycobiont remains missing, primarily due to the lack of structural information on their photosystems. Here, we report the high-resolution structures of PSI complexes purified from the lichen phycobiont D. chodatii. Our structures reveal a high antenna plasticity and a dynamical assembly mode of D. chodatii PSI-LHCI (DcPSI-LHCI), providing a mechanistic understanding of its flexible acclimation to stress conditions, as well as shedding light on the changes occurring in PSI of plants transiting from a low-light (LL) aquatic to a high-light (HL) terrestrial environment.

Results

Overall structure

D. chodatii was first isolated from the crust lichen E. pusillum growing in arid desert (Supplementary Fig. 1a), and cultured in BBM medium in laboratory conditions³⁷. We purified four types of the DcPSI complexes from our cultured D. chodatii cells and solved their cryo-EM structures (Supplementary Figs. 1–3 and Supplementary Table 1). These complexes contain a monomeric PSI core bound to different numbers of peripheral LHCI subunits (Fig. 1). The four structures were designated according to the number of LHCIs, i.e., DcPSI-8LHCI, DcPSI-6LHCI, DcPSI-2LHCI and DcPSI-core, with resolutions of 3.0 Å, 4.1 Å, 3.8 Å and 3.5 Å, respectively. We first analyzed the DcPSI-8LHCI structure as it represents the most complete DcPSI structure with the highest resolution. DcPSI-8LHCI contains 13 PSI core subunits (PsaA-PsaM) and eight Lhca proteins, which are arranged in the shape of three belts, namely the four-Lhca inner belt, the two-Lhca outer belt and the two-Lhca side belt (Fig. 1a). Based on our high-resolution cryo-EM map (Supplementary Fig. 4) and our transcriptome sequencing result, we were able to determine the identity of all Lhca subunits in DcPSI-8LHCI structure. We named the eight Lhcas according to their counterparts in CrPSI-LHCI based on their similarity in sequence, structure and location within the complex (Supplementary Table 2). The DcPSI-8LHCI contains an inner antenna layer of Lhca1-a8-a7-a3, an outer layer of Lhca6-a5, and a side layer of Lhca2-a9, constituting four Lhca heterodimers (Lhca1-a8, Lhca7-a3, Lhca6-a5 and Lhca2-a9). In total, we modeled 21 protein subunits, 218 chlorophylls (Chls), 50 carotenoids, 19 lipids, 3 Fe₄S₄ clusters and 2 phylloquinones in the DcPSI-8LHCI structure (Supplementary Table 3).

Fig. 1. Overall structures of DcPSI-LHCI complexes.

a, b The structure of DcPSI–8LHCI viewed from the stromal side (a) and along the membrane plane (b). The inner, outer and side antenna belts are indicated in (a). c–e Stromal side view of the structure of DcPSI–6LHCI (c), DcPSI–2LHCI (d), DcPSI–core (e). PsaG, PsaH and Lhca antennae are shown in cartoon mode, and other PSI core subunits are shown in ribbon mode. Subunits are shown in different colors.

The four DcPSI structures closely resemble each other and are well superposed of their corresponding parts except several loop fragments and a small number of pigments located at the outermost regions (Supplementary Fig. 5a). Compared to DcPSI-8LHCI, DcPSI-6LHCI lacks the side belt composed of Lhca2-a9, as well as PsaG and PsaH, two core subunits that interact with Lhca9 and Lhca2, respectively (Fig. 1c and Supplementary Fig. 5a). DcPSI-core also lacks PsaG and PsaH, similar to DcPSI-6LHCI, and lacks binding of any antenna (Fig. 1e). Interestingly, we found that DcPSI-2LHCI binds only the side belt (Lhca2-a9), which appears to be the least stable Lhca belt in CrPSI⁷. Moreover, DcPSI-2LHCI contains core subunits identical to those in DcPSI-8LHCI, with stably bound PsaG and PsaH (Fig. 1d and Supplementary Fig. 5a). These structural observations strongly suggest that Lhca2 and Lhca9 are critical in stabilizing PsaH and PsaG in green algae. Our structural comparison showed that the overall structure of DcPSI-8LHCI is similar to PSI-LHCI from other species, except for the number of peripheral Lhca subunits. DcPSI-8LHCI contains an additional four Lhcas compared to land plant PSI-LHCI (Supplementary Fig. 5b), but lacks two Lhcas corresponding to Lhca1_out-a4 in the outer layer when compared to CrPSI-10LHCI (Supplementary Fig. 5c).

Differences in core subunits between DcPSI and PSI complexes from other species

While PSI core subunits are known to remain relatively conserved throughout the evolution of green lineages, we found minor differences when comparing the core structure of DcPSI with those of PSI from other species (Fig. 2). PsaM constitutes a core subunit present in cyanobacteria and several green algae, including D. chodatii, whereas the gene encoding PsaM has been lost in C. reinhardtii and most land plants. In DcPSI complexes, PsaM is located in a position similar to that in other PSI-LHCI structures^6,38, between PsaB and Lhca2, with one carotenoid (BCR101) that is located adjacent to PsaM and links PsaM to PsaB, PsaI and Lhca2 (Fig. 2b). While this carotenoid is also present in the Ostreococcus tauri PSI-Lhcp structure³⁸, we found that it is absent in other green algal PSI structures. The presence of PsaM and the carotenoid BCR101 may strengthen the Lhca2-a9 association with the PSI core, which may be one of the reasons for the existence of DcPSI-2LHCI complex, in which Lhca2-a9 are the only two antenna proteins preserved in the PSI complex. In comparison, this type of PSI has never been found in other species.

Fig. 2. Structure comparison of PSI core from D. chodatii and other green plants.

a Structural superposition of DcPSI-8LHCI (shown in transparent surface mode and DcPsaG/PsaH/PsaM also in cartoon mode, colored blue, magenta and salmon, respectively) with ZmPSI (PDB code: 5ZJI) viewed from the lumenal side. For clarity, only ZmPsaN is shown in pink ribbon and other subunits from ZmPSI are omitted. b PsaM binds a carotenoid molecule (BCR101) and is located between PsaB-PsaI and Lhca2, viewed from the stromal side. c Structural superposition of DcPSI-8LHCI and ZmPSI. DcLhca7 and DcLhca8 are shown in cartoon and transparent surface modes; ZmPsaN is shown in ribbon mode and clashes with DcLhca7 and DcLhca8. Other subunits of the two structural models are omitted for clarity. d Structural superposition of DcPSI-8LHCI (shown in transparent surface mode) with ZmPSI viewed from the lumenal side. PsaB, PsaG and Lhca1 from ZmPSI are shown in ribbon mode and other subunits are omitted. The additional chlorophyll (Chl 201) and lipid (Phosphatidyl glycerol, PG) molecules bound to ZmPsaG are shown as pink and white sticks, and also shown in the enlarged view. e Structural comparison of DcPsaH (magenta cartoon) with CrPsaH (PDB code: 7DZ7, pale-cyan ribbon) and ZmPsaH (pink ribbon), ZmPsaI (dark green) and ZmPsaL (teal) are also shown in ribbon mode. DcLhca2 is shown in transparent surface. Chlorophyll molecules of PsaH subunits are shown as sticks. Residues involved in the hydrogen bond interactions between ZmPsaH and ZmPsaL/ZmPsaI are shown as sticks, and the detailed interactions are shown in the zoom-in view on the right.

PsaN was observed to bind at the lumenal side of the PSI core in several eukaryotic PSI structures^16,38,39, but is absent in most PSI structures^6–8,12,13, suggesting that PsaN only loosely associates with other PSI subunits. In agreement with this suggestion, we failed to observe any density corresponding to PsaN in our PSI reconstruction. Comparison of the previously reported Zea mays PSI (ZmPSI) structure (PsaN-containing structure) with our DcPSI-8LHCI structure showed that the C-terminal tail and a middle loop of ZmPsaN clash with the C-terminal fragment of Lhca7 and Lhca8, respectively, in DcPSI-LHCI (Fig. 2c). Based on these findings, we propose that the binding of PsaN in DcPSI is unstable and requires conformational changes of loop regions in Lhca proteins and/or PsaN.

In both DcPSI-6LHCI and DcPSI-core, PsaG and PsaH are missing, presumably due to the absence of Lhca2-a9 (Supplementary Fig. 5a). The co-existence of PsaG/PsaH with Lhca2-a9 was also observed in an earlier study of CrPSI⁷. These results indicate that in green algae, Lhca2-a9 stabilizes the binding of PsaH and PsaG, which is consistent with our structural observation in DcPSI-LHCI that multiple salt bridges and hydrogen bonds are formed between PsaG and Lhca9, and between PsaH and Lhca2 (Supplementary Fig. 6a). However, land plant PSIs do not possess the side belt antennae (Lhca2-a9), yet PsaG and PsaH are present in all previously reported plant PSI structures^12,13,39, indicating that both subunits are more stable in land plant PSIs than in green algal complexes. Our structural comparison of PsaG from different green lineages showed that land plant PsaG possesses additional chlorophyll and lipid molecules compared to green algal PsaG (Fig. 2d). These molecules fill the interfacial space encircled by PsaB, PsaG and Lhca1, thus strengthening the interaction between PsaG and other subunits in land plant PSIs. The absence of these molecules (Supplementary Fig. 6b) may result in unstable binding of PsaG in green algae. Intriguingly, we found that compared to green algal PsaH (DcPsaH and CrPsaH), land plant PsaH binds two chlorophylls fewer (corresponding to Chls 202 and 203 in DcPsaH) (Fig. 2e). Both chlorophylls are located at the interface between PsaH and Lhca2, and may thus reinforce the interactions between PsaH and Lhca2 in green algal PSI (Fig. 2e). We also found that two hydrophobic residues in green algal PsaH (V56 and G100 in DcPsaH) are substituted by serines in land plants (S73 and S116 in ZmPsaH) (Supplementary Fig. 7). The two serine residues form additional hydrogen bonds with PsaL and PsaI (Fig. 2e), thus stabilizing the association of plant PsaH. Together, our structural analysis suggested that compared to green algal PsaG, the land plant counterpart binds additional chlorophyll and lipid molecules that may serve as molecular glue to stabilize its association with PsaB and Lhca1. In contrast, land plant PsaH binds fewer chlorophyll molecules than those in green algae, but develops more direct protein-protein interactions between PsaH and other PSI core subunits through a few amino acid substitutions in PsaH. These structural observations and comparisons may explain the stable binding of PsaG and PsaH in land plants.

Structural features and organization of DcLHCIs

When we compared our DcPSI-8LHCI structure with that of CrPSI-10LHCI for analyzing LHCI features, we found that three of the eight DcLhca proteins (Lhca1, Lhca3 and Lhca9) closely resembled their counterparts in CrPSI-LHCI (Supplementary Fig. 8a–c), and the other three (Lhca2, Lhca6 and Lhca7) adopt a protein conformation identical to their C. reinhardtii counterparts, but differ in the number of pigments bound compared to the corresponding CrLhca proteins (Supplementary Fig. 8d–f). DcLhca2 binds an additional carotenoid at the 628 site, but lacks another carotenoid BCR 623 present in CrLhca2 (Supplementary Fig. 8d). DcLhca6 contains two pigments fewer than CrLhca6, namely Chl 618 and BCR 622 (Supplementary Fig. 8e). DcLhca7 lacks Chl 615 and BCR 623 existing in CrLhca7, but contains Chl 619 that is exclusive to DcLhca7 (Supplementary Fig. 8f).

The remaining two DcLhcas (Lhca8 and Lhca5) differ from their counterparts in C. reinhardtii by both protein conformation and bound pigments. Compared to CrLhca8, DcLhca8 possesses a longer C-terminal tail (Fig. 3, Supplementary Fig. 9), which forms multiple hydrogen bonds with Lhca7 (Fig. 3b) and may hinder PsaN binding due to steric hindrance (Fig. 2c). Furthermore, we found that DcLhca8 binds an additional chlorophyll, Chl 617, which forms a Chl trimer with Chl 603-609 pair and is located at the Lhca1-a8 interface (Fig. 3a and Supplementary Fig. 9a). Notably, we found that Chl 617 is also present in the terrestrial plant counterpart, Lhca4 (Supplementary Table 2), and its coordinating residue (H138) is conserved in land plants (Supplementary Fig. 9b), suggesting that this feature may constitute a convergent adaptive strategy for D. chodatii and other land plants to terrestrial environment.

Fig. 3. Structure and organization of the double layered LHCI belt of D. chodatii PSI.

a Comparison of DcLhcas with CrLhcas (PDB code: 7DZ7) viewed from the lumenal side. DcLhcas are colored the same as in Fig. 1a, CrLhcas are shown in pale cyan. The long C-terminal tail of DcLhca8 was highlighted by an orange square. Pigment molecules different between the two structures are shown in spheres and labeled. b Detailed interactions between the C-terminal tail of Lhca8 and the lumenal helix E of Lhca7. The hydrogen bonds are indicated by black dashed lines. CrLhca8 and ZmLhca4 (PDB code: 5ZJI) are superposed on DcLhca8, and the C-terminus (C-ter) of the three Lhca proteins is indicated.

Compared to CrLhca5, DcLhca5 contains a longer BC loop, which provides an additional chlorophyll binding site (Supplementary Fig. 10a). This chlorophyll (Chl 605) is coordinated by the main chain carbonyl group of Y121 from the BC loop of DcLhca5 (Supplementary Fig. 10b), and located in a position similar to that of Chl 605 in plant LHCII (Supplementary Fig. 10c)⁴⁰, also at the outmost region of the complex (Fig. 3a). This location suggests that Chl 605 in DcLhca5 may increase the ability of the complex to absorb solar energy, thus slightly compensating for the loss of two outer Lhcas (Lhca1_out-a4) of DcPSI. Moreover, we observed an additional carotenoid (629 site) bound in the gap region between Lhca5-a6 and Lhca7-a3 (Fig. 3a).

In the DcPSI-8LHCI structure, adjacent Lhca proteins within each heterodimer are closely associated. The additional pigment molecules bound to DcLhca8 (Chl 617) and DcLhca5 (Chl 605 and BCR 629) further stabilize the DcLhca1-a8 and DcLhca5-a6 heterodimers respectively (Fig. 3a). The BCR629 also contributes to the stable association of the outer antenna dimer Lhca5-a6 with the inner belt. The long C-terminal tail of DcLhca8 reinforces the association between the Lhca1-a8 and Lhca7-a3 heterodimers, resulting in a highly stable inner Lhca belt in D. chodatii. These unique structural features of DcLhcas play an important role in the stability of the complex. In comparison, the pigment molecules present in CrPSI-LHCI but absent in DcLhca proteins may contribute to the stabilization of the outer Lhcas in C. reinhardtii PSI complex, particularly the Lhca1_out-a4 dimer. The Chl 618 and BCR 623 in CrLhca6 are positioned at the CrLhca6-CrLhca4 interface, and Chl 615 and BCR 623 in CrLhca7 are located in the gap region between the inner and outer Lhca layer (Fig. 3a). Their absence in DcLhca6 and DcLhca7 is likely related to the loss of the outer antennae corresponding to CrLhca1_out-a4 in D. chodatii.

Direct interactions between different Lhca belts and between Lhcas and the core are essential for the complex assembly (Supplementary Fig. 6a). The inner Lhca belt binds to the PSI core primarily through Lhca3, which forms multiple hydrogen bonds with core subunits. The outer layer attaches to the inner layer and is stabilized by interactions between Lhca5 and Lhca3, and between Lhca6 and Lhca7. The side layer Lhca9 strongly binds PsaG and PsaB, and Lhca2 interacts with PsaH and PsaM. In addition, several lipid molecules are located at the core-Lhca and Lhca-Lhca interfacial regions, further strengthening the association of these antenna belts and the core (Supplementary Fig. 6b).

Pigment arrangement and possible excitation energy transfer pathways

Our structure showed that each DcLhca binds 12–18 chlorophyll and 3–5 carotenoid molecules (Supplementary Table 4), with Lhca5 possessing the largest number of chromophores (18 chlorophylls and 5 carotenoids). The DcPSI-8LHCI contains 218 chlorophylls in total (Fig. 4a), with 121 chlorophylls located in the stromal side and 97 chlorophylls in the lumenal side (Supplementary Fig. 11). The energy transfer rate between neighboring Chls largely depends on their distances and the dipole orientations according to the Förster mechanism⁴¹. We observed in our structures multiple interfacial chlorophyll pairs composed of chlorophylls belonging to adjacent subunits, which allow for a large number of possible excitation energy transfer (EET) pathways. To analyze the EET network within the complex, we calculated the Förster Resonance Energy Transfer (FRET) rate between neighboring Chls in DcPSI-8LHCI structure (Fig. 4b). Our analysis showed that, except for Lhca2-a9, the energy transfer within all Lhca heterodimers is highly efficient. In particular, Chl 617 bound to DcLhca8 may further facilitate the energy equilibration within DcLhca1-a8 heterodimer. In addition, Chl 619 in DcLhca7 presumably increases the energy transfer efficiency between Lhca1-a8 and Lhca7-a3 dimers (Fig. 4b).

Fig. 4. Chlorophyll arrangement and potential energy transfer pathways in DcPSI-8LHCI.

a Arrangement of chlorophyll molecules in DcPSI-8LHCI, viewed from the stromal side. The potential red Chls in four LHCIs are shown in red. The phytol tail of Chls are omitted for clarity. b Structure-based calculation of FRET rates between neighboring Chls (shown as spheres) within DcPSI-8LHCI. Adjacent Chls with FRET lifetimes <40 ps are connected by black lines. Line width indicates the relative FRET rate of each chlorophyll-chlorophyll pair (the wider the line, the faster the rate). Three Chls (605 in DcLhca5, 619 in DcLhca7 and 617 in DcLhca8) absent from CrPSI-LHCI are shown in sphere mode in (a) and circled by dashed line in (b), and labeled.

The EET from the inner belt to the core occurs primarily through the terminal antennae Lhca1 and Lhca3, similar to that in other green plant PSI-LHCI complexes^13,42. Lhca5 and Lhca6, which are located in the outer belt, transfer the excitation energy to the PSI core via multiple EET pathways through the inner antennae. In comparison, the EET from the side belt to the core preferentially occurs through Lhca9, and the EET from Lhca2 to the core appears less efficient. Together, our analysis indicated that the EET from peripheral antennae to the core in D. chodatii occurs primarily through Lhca1, Lhca3 and Lhca9. This finding is consistent with our structural observation that these three Lhcas are nearly identical in different green algae (Supplementary Fig. 8a–c).

Potential regulative mechanisms adopted by DcPSI

Our structural analysis showed that DcPSI complexes are present in at least four types containing different numbers of LHCI antennae (Fig. 1), indicating that DcPSI complexes exhibit high antenna plasticity. To better assess the internal flexibility between different moieties of DcPSI-LHCI complexes, we performed multi-body analysis of DcPSI-8LHCI, with four defined bodies, namely the core moiety, the inner belt (Lhca1-a8-a7-a3), the outer belt (Lhca5-a6) and the side belt (Lhca2-a9) (Supplementary Fig. 12a). Our multi-body analysis showed that two major principal components (components 1 and 2) account for 50% of the total variances in the data (Supplementary Fig. 12b). In the first component, all three Lhca belts exhibit horizontal and vertical shifts relative to the core, with distances of 18-30 Å within the membrane and 10-12 Å along the membrane normal (Supplementary Fig. 12c). The second component shows a more pronounced vertical shift of the Lhca belts (Supplementary Fig. 12d). In both components, the outer and inner belts are synchronized when shifting vertically, and the outer belt moves only within the membrane plane relative to the inner belt, with distances of 5–12 Å (Supplementary Figs. 12e, f), suggesting that the two belts are relatively stable with each other. This observation is consistent with our structural results showing that the type of DcPSI-LHCIs containing only the inner belt antennae is lacking. The positional flexibility of the Lhca belts revealed by our multibody analysis may result in changes in distances and/or orientations of a number of interfacial chlorophyll pairs in DcPSI complexes, possibly enabling the DcPSI to dynamically regulate its energy transfer under different light conditions.

In addition, our 77 K steady-state fluorescence spectrum analysis of DcPSI-LHCI sample, D. chodatii cells and C. reinhardtii cells revealed that both D. chodatii cells and the PSI complex exhibit the fluorescence emission band peaking at 717 nm, ~2 nm red-shifted from that of C. reinhardtii cells (Supplementary Fig. 13), implying that DcPSI-LHCI may also contain long-wavelength chlorophylls (red-forms) similar to those in CrPSI-LHCI⁴³. Previous reports suggested that for the formation of the red-forms, an asparagine instead of a histidine is necessary to ligate one conserved chlorophyll (Chl 603)⁴⁴. We found that in the DcPSI-8LHCI structure, the Chl 603 in four Lhcas (Lhca1, Lhca2, Lhca8 and Lhca9) is coordinated by an asparagine, while it is coordinated by a histidine in the other four antenna proteins (Fig. 4a). These structural observations suggested that half of the antennae in DcPSI-LHCI possesses the structural potential for red-shifted states. This proportion aligns with that found in land plant PSIs, in which two of the four Lhcas constitute red antenna (Lhca3 and Lhca4)¹³, and is higher than that found in CrPSI, which contains three red Lhcas (Lhca2, Lhca4 and Lhca9) out of ten antenna proteins^7,45. Red chlorophylls are suggested to increase the absorption cross section of PSI and may also play a role in PSI photoprotection^43,46, accordingly, these observed structural features might potentially contribute to the light-harvesting and photoprotection of DcPSI.

DcPSI complexes constitutively bind Zea molecules

Our pigment analysis of DcPSI samples by High Performance Liquid Chromatography (HPLC) showed that in addition to the pigment molecules commonly present in other green plant PSI-LHCI complexes, namely Chl a, Chl b, neoxanthin (Neo), violaxanthin (Vio), lutein (Lut) and β-carotene (BCR), DcPSI complexes contain two additional types of pigments, which appeared as the third and fifth peaks (peak 3 and peak 5) in our HPLC result (Supplementary Figs. 14a, b). The retention times of these two peaks are similar to those of antheraxanthin (Ant) and Zea reported in a previous study⁴⁷. Moreover, the absorption spectrum of peak 5 is identical to that of the Zea standard sample (Supplementary Fig. 14c). Therefore, we speculated that our DcPSI sample contains Zea and possibly Ant molecules. Previous studies have demonstrated that under LL conditions, C. reinhardtii thylakoids do not contain Zea and Ant, but bind Vio molecules, which can be converted to Ant and then to Zea under HL conditions⁴⁸. Based on these results, we analyzed the pigment composition of thylakoid samples of C. reinhardtii cultured under HL and LL conditions using HPLC to verify the identities of peaks 3 and 5 in DcPSI sample. Our results showed that two peaks corresponding to peak 3 and peak 5 are present in HL cultured C. reinhardtii thylakoids, but absent in LL thylakoids (Supplementary Figs. 14b–d), confirming that they represent Ant (peak 3) and Zea (peak 5) molecules converted from Vio in C. reinhardtii under HL conditions. In comparison, these two peaks were observed in both LL and HL treated D. chodatii thylakoids as well as the DcPSI complexes purified from LL cultured cells (Supplementary Figs. 14a, b). Furthermore, compared to the LL thylakoids, the HL-treated D. chodatii thylakoids exhibited an obvious increase in Zea accumulation and a decrease in Vio content, suggesting that Zea is effectively converted from Vio under HL conditions. These results together indicate that DcPSI constitutively binds Zea molecules even under LL conditions, while HL illumination strongly promotes the conversion of Zea from Vio. The photoprotective role of Zea has been well established in previous studies^49,50. Our findings suggest that DcPSI complexes possess pigment features consistent with enhanced photoprotective capacity. Since the resolution of our cryo-EM density was not sufficiently high to allow unambiguous identification of different carotenoid species, we compared DcPSI-8LHCI structure with CrPSI-LHCI and BcPSI-LHCI structures (both of which do not bind Zea) to infer the potential binding positions of Zea molecules in DcPSI-8LHCI complex. We found that DcPSI-8LHCI possesses two carotenoids (629 of Lhca5 and 628 of Lhca2) that are not observed in other green algal PSI-LHCI structures (Supplementary Fig. 15a). In addition, the cryo-EM densities of these two carotenoids exhibited features similar to those of xanthophyll cycle carotenoids (Supplementary Fig. 15b). Based on these observations, we tentatively assigned these two carotenoids as Zea. Both putative Zea molecules are closely associated with several chlorophylls at short distances (Supplementary Figs. 15c-d), exhibiting structural features of photoprotective pigments⁵¹. However, it should be noted that the assignment of Zea in our structures remains preliminary and requires higher resolution structural data and further experimental verification.

Discussion

Here, we show that DcPSI exists in multiple forms, differing in the number of bound Lhcas. The DcPSI-8LHCI represents the largest PSI complex in D. chodatii, lacking two Lhcas in the outer layer compared to CrPSI-10LHCI (Supplementary Fig. 5c). This lack and the presence of smaller PSI forms are presumably not a result of a detachment of peripheral antennae from the complex occurring during cell growth and/or our preparation procedures. The solubilization conditions (0.8% α-DDM) and the purification process used in our study are considered to be mild, and have been successfully utilized in purifying CrPSI-10LHCI in a previous report⁷. In addition, earlier studies demonstrated that even harsher solubilization conditions (1% β-DDM) resulted in no loss of Lhca1_out-Lhca4 in PSI-LHCI samples^6,8. Moreover, all CrLhcas except CrLhca4 possess homologous proteins in D. chodatii with sequence identity of 53-71% based on our transcriptome sequencing data, whereas the potential Lhca4 homolog in D. chodatii (DcLhca4-like) showed less than 36% identity with CrLhca4 (Supplementary Fig. 16a). Furthermore, the structural model of the DcLhca4-like protein predicted by AlphaFold greatly differs from that of CrLhca4 in loop regions (Supplementary Fig. 16b). In contrast, the structures of other DcLhcas predicted by AlphaFold are highly similar to the corresponding CrLhcas (Supplementary Fig. 16c). These results suggested that D. chodatii may have lost or functionally diverged the gene encoding Lhca4 from its genome, resulting in the loss of the two outer Lhcas in the DcPSI, and the DcPSI-8LHCI structure constitutes the largest native form of PSI in D. chodatii. However, we cannot completely rule out the possibility that other smaller DcPSI forms may result from the degradation/dissociation of DcPSI-8LHCI during the growth/isolation process. Even so, the fact that these smaller DcPSI complexes could be obtained under such mild preparation conditions suggests that in vivo, the binding affinity of LHCI to the core varies among different PSI complexes. Some LHCIs in certain complexes bind to the core less strongly than those in the intact DcPSI-8LHCI, and therefore cannot survive the mild solubilization conditions, suggesting that DcPSI complexes exhibit high antenna plasticity in vivo.

Our structures show that while DcPSI complexes are largely similar to the PSI-LHCI from other green algae, they also exhibit structural features distinct from other green algal PSIs and similar to land plant PSIs, including the presence of Chl 617 in Lhca8 and the stoichiometry of potential red Lhcas. These features imply that D. chodatii may undergo two distinct evolutionary processes: orthologous evolution related to green algae, and convergent evolution paralleling land plants. This dual evolution might be explained by the fact that the selection pressure faced by D. chodatii is similar to that occurring during land colonization. One major stress faced by green plants in their land colonization is presumably the high light. As a result, these plants may have discarded a portion of their peripheral antennae to reduce the amount of absorbed solar energy, thus avoiding photodamage²³. Analysis of our DcPSI structures may provide clues about how green plants reduced their antenna size to adapt to a terrestrial environment. Our structures reveal that the outer layer of DcPSI-LHCI possesses only two Lhcas instead of four observed in other green algal PSI-LHCI complexes. The loss of two outer LHCIs in D. chodatii presumably occurred through the loss or change of lhca4 gene in D. chodatii genome. These results suggested that the green algae containing ten Lhcas in their PSI complexes may first discard two outer Lhcas (corresponding to CrLhca1_out-a4) when adapting to terrestrial environments, forming a PSI complex similar to the PSI-8LHCI found in D. chodatii. The remaining two outer antennae may have been rather unstable following the loss of Lhca1_out-a4, the next step might be the loss of the outer Lhca5-a6 dimer. In comparison, Lhca2-a9 may play a role in the environmental adaptation of green algae, as they appear to regulate the light harvesting capability of the PSI through reversibly binding to the PSI core in green algal species⁷. Therefore, Lhca2-a9 may have been lost following the disappearance of Lhca5-a6 (Fig. 5). This reduction in antenna size might represent a convergent evolutionary strategy adopted by Chlorophyta (lichen phycobionts) and Streptophyta (land plants) to cope with terrestrial environmental stresses.

Fig. 5. Schematic illustration of convergent antenna reduction strategies of different green lineages in the process of adapting to terrestrial environments.

The green and blue dashed lines represent parallel paths of evolution for plants in Chlorophyta and Streptophyta, respectively. The four PSI-LHCI structures are from green algae Bryopsis corticulans (PSI-10LHCI, PDB code: 6IGZ), Diplosphaera chodatii (PSI-8LHCI, PDB code: 9X4O) and Dunaliella salina (PSI-6LHCI, PDB code: 6SL5), and land plant Lathyrus oleraceus (PSI-4LHCI, PDB code: 4XK8). PSI core subunits are shown in light-blue, and LHCI subunits are shown in different colors.

In natural terrestrial environments, lichen phycobiont including D. chodatii must have encountered considerable stresses, such as high light and long periods of drought. D. chodatii may have developed various protective mechanisms critical for survival under such extreme conditions. First, the symbiotic life form of lichen allows the algae to be wrapped up by the lichen mycobiont, the organizational structure of lichen thallus precisely meets the requirement for a relatively weak light and also helps in water retention of the phycobiont. In addition, DcPSI may have reduced its antenna size by discarding Lhca1_out-a4 from the outer layer, thus reducing the amount of the light energy absorbed. Moreover, the high antenna plasticity of DcPSI complex may enable it to rapidly and dynamically adjust its light harvesting capacity in the constantly changing environments. Furthermore, our HPLC assessment indicated that DcPSI constitutively binds Zea, a xanthophyll cycle carotenoid crucial for photoprotection⁵². Compared to the LL thylakoids, Zea accumulation is obviously increased in D. chodatii thylakoids treated with HL, and these additionally produced Zea molecules may further bind to the PSI (and PSII), suggesting that DcPSI possesses structural potentials for enhanced photoprotective capacity, especially under stress conditions. These specific acclimatization strategies may enable D. chodatii to survive in the comprehensive stress environments, including high light and drought, and may also have been adopted by other terrestrial algae and land plants.

Methods

Cell culture

Diplosphaera chodatii cells were cultured in BBM medium³⁷ at 23 °C under continuous white light of ~40 µmol photons m^-2 s^-1 (LL conditions), and cells (LL-cells) grown in logarithmic phase were harvested through centrifugation. C. reinhardtii CC124 cells were cultured in TAP medium at 23 °C under LL conditions and harvested through centrifugation. For high light (HL) treatment, low light cultured D. chodatii and C. reinhardtii cells were further treated with HL illumination (500 μmol photons m^-2 s^-1) for 12 h before cell harvesting.

Sample purification and characterization

For preparation of the thylakoid membranes, the cells were resuspended in buffer A (0.3 M sucrose, 25 mM HEPES-KOH, pH7.5, 1 mM MgCl₂) and disrupted by French press (500 bar 4 times), then the lysate was centrifuged at 2000 g for 10 min. The supernatant was further centrifuged at 150,000 g for 20 min (HITACHI, P45AT rotor), and the pellets containing thylakoid membranes were resuspended in buffer B (400 mM Sorbitol, 15 mM NaCl, 10 mM MgCl₂). To obtain the DcPSI-LHCI supercomplex samples, thylakoid membranes were adjusted to chlorophylls concentration of 1.0 mg ml⁻¹ with 20 mM HEPES, pH 7.5, and mixed with an equal volume of 1.6% n-Dodecyl-α-D-maltopyranoside (α-DDM), then centrifuged at 20,000 g for 10 min at 4 °C. The supernatants were fractionated by sucrose density gradient ultracentrifugation at 222,000 g for 18 h (Beckman, SW41 rotor). Two green bands were collected and named band4 (the upper band, containing PSI-2LHCI and PSI-core complexes) and band5 (the lower band, containing PSI-8/6LHCI complexes) (Supplementary Fig. 1b). Both DcPSI-LHCI samples were concentrated to 1.5 mg ml⁻¹ in chlorophylls for cryo-EM specimen preparation. In addition, sodium dodecyl sulfate-polyacrylamide gel (SDS-PAGE) (Supplementary Fig. 1d) and absorption spectra (Supplementary Fig. 1c) of these supercomplex samples were performed⁵³. For SDS-PAGE analysis, a 10%-15% gradient separating gel was prepared for protein separation. A total of 30 μg (in chlorophyll) sample was loaded into each lane, and electrophoresis was performed in an ice-water bath at a constant current of 30 mA throughout the entire run. Absorption spectra of these supercomplex samples were measured using the ultraviolet-visible spectrophotometer (HITACHI U-3900, Japan). The sample with a chlorophyll concentration of ~0.02 mg mL^-1 was placed in a quartz cuvette with an optical path length of 1 cm, and the spectral scan range was set from 350 nm to 750 nm. Baseline calibration was performed with the corresponding sample buffer before measurements to eliminate background interference. Furthermore, 77 K steady-state fluorescence spectra of the band5 sample (containing PSI-8LHCI and PSI-6LHCI) as well as LL cultured D. chodatii and C. reinhardtii cells (Supplementary Fig. 13) were measured⁵⁴. All low-temperature fluorescence measurements were carried out on an FLS1000 fluorescence spectrometer (Edinburgh Instruments, UK) equipped with a liquid nitrogen cooling accessory to maintain a stable temperature of 77 K during data acquisition. The excitation wavelength was fixed at 430 nm for chlorophyll excitation.

High performance liquid chromatography (HPLC) and identification of Zeaxanthin

Pigment composition of band5 and band4 samples was analyzed by high performance liquid chromatography (HPLC)⁵⁵. For sample pretreatment, 10 μL of the sample at a concentration of ~1 mg mL^-1 (in chlorophyll) was mixed with 40 μL of acetone for pigment extraction. The mixture was thoroughly vortexed and centrifuged, and the supernatant was collected for subsequent HPLC detection. Pigment separation was performed using an Allsphere ODS-2 column with a 5 μm particle size (Lot No. 34/105). The mobile phase consisted of solvent A (Acetonitrile:Methanol:0.1 M Tris/NaOH buffer, pH 8.0 = 87:10:3, v/v/v) and solvent B (Methanol: Hexane = 4:1, v/v). A linear gradient elution from solvent A to solvent B was executed at a constant flow rate of 2 mL/min over a period of 9–12.5 min, and eluted pigments were monitored at a detection wavelength of 440 nm. The result showed that two additional peaks (peak 3 and peak 5), which were not observed in the CrPSI samples as reported previously⁷, are present in DcPSI, with the retention times similar to those of Ant and Zea, respectively⁴⁷. To further identify these two types of pigments, we extracted thylakoid membranes from D. chodatii and C. reinhardtii cells cultured under both LL and HL conditions. The pigment composition of these four types of thylakoid membranes were analyzed by HPLC and compared (Supplementary Fig. 14).

Amino acid sequencing

Fresh Diplosphaera chodatii cells were harvested and sent to the Beijing Genomics Institute (BGI) for high-throughput sequencing analysis. The sequences of most DcPSI-LHCI proteins were identified based on the analysis of the high coverage and high-precision transcription information. The majority of the DcPSI-LHCI subunits have amino acid sequences that are highly homologous to the corresponding subunits in CrPSI-LHCI. Altogether, we obtained the full-length sequences of nine PSI core subunits (PsaA/B/D/E/F/G/H/K/L) and seven LHCIs (Lhca1/a3/a5/a6/a7/a8/a9) from D. chodatii. The Lhca2 sequence appears to be partial and lacks ~30 residues at the C-terminus.

Grid preparation and data acquisition

Approximately 2.5 µl of the DcPSI–LHCI sample was applied to a glow-discharged holey carbon grid (GIG-C31213) and flash-frozen in liquid ethane at around 100 K using a semi-automatic plunge device (Thermo Fisher Scientific Vitrobot IV) with a blotting time of 2 s and blotting force 3 at 4 °C and 100% humidity. Sample screening was performed using a Talos F200C 200 kV electron microscope equipped with a Ceta camera (Thermo Fisher Scientific). Micrographs used for structure determination were collected on a 200 kV Arctica electron microscope (Thermo Fisher Scientific) equipped with K2 camera (Gatan) and operated at super-resolution mode. A total of 5514 and 4774 images were collected for band5 and band4 samples, respectively. The defocus range was set between 1.5 µm and 2.2 µm, and the magnification was 130,000. The micrographs were fractionated into 32 frames with a total dose of ~60 e^–Å^–2, and representative cryo-EM micrographs of band5 and band4 were shown in Supplementary Figs. 2a and 3a, respectively.

Image processing

Movie stacks were subjected to beam-induced motion correction⁵⁶. The contrast transfer function (CTF) parameters of each image were determined by program CTFFIND4⁵⁷. A small number of particles were picked using Laplacian-of-Gaussian method first, and used for 2D classification in RELION3.1⁵⁸. Then, the good 2D class was selected and used as reference for auto-picking. For band5 sample (containing PSI-8/6LHCI complexes) (Supplementary Fig. 2b), a total of 578,367 particles were picked and processed by 2D classification, and 211,744 particles were kept for further 3D classification. One class constituting the smaller PSI-6LHCI contains 24,782 particles, and was refined to a final resolution of 4.1 Å. The bigger PSI-8LHCI classes contain 69,264 particles, and are refined to an overall resolution of 3.0 Å. Further local refinement results in local maps with resolutions ranging from 2.8 to 3.4 Å (Supplementary Fig. 2b, c), and the composite map is used for model building. For band4 sample (containing PSI-2LHCI/core complexes), 20,609 and 44,940 particles were classified as PSI-2LHCI and PSI-core after 3D classification, and refined to a resolution of 3.8 and 3.5 Å, respectively (Supplementary Figs. 3b, c). The local resolution of these final maps was calculated using ResMap (Supplementary Figs. 2d, 3d)⁵⁹, and angular distribution of particles in the final 3D reconstructions are shown in Supplementary Fig. 2e and Fig. 3e.

Model building and refinement

To build the model of the DcPSI-8LHCI, CrPSI-LHCI-LHCII structure (PDB code: 7DZ7)⁶⁰ was manually fitted into the composite map using UCSF Chimera⁶¹ and the extra parts were removed. The amino acid sequences were then mutated to their counterparts in D. chodatii, which were obtained from our transcriptome sequencing data. Each individual subunit was identified based on the best match of the specific sequence to the cryo-EM density. Four core subunits were built using the sequences of C. reinhardtii (DcPsaC/I/J) or B. corticulans (DcPsaM) homologous proteins, because we failed to obtain their sequences in this study. The C-terminal sequence of Lhca2 was also missing in our transcriptome sequencing data, and this part was built using the C-terminal sequence of CrLhca2. The resulted complex structure was further refined using Phenix⁶² and manually adjusted in Coot⁶³. However, the current resolution of the reconstruction of the Lhca moiety does not allow an unambiguous identification of different carotenoids or to distinguish all Chl b from Chl a. Based on the cryo-EM density and the similarity between DcLhcas and the corresponding CrLhcas^6,8, we tentatively assigned 25 Chl b molecules in the DcPSI-8LHCI structure. The carotenoids in the DcPSI-LHCI structures were tentatively assigned to be the same carotenoids as in CrPSI-LHCI⁷, whereas two carotenoid molecules present in DcPSI-LHCI but absent in CrPSI-LHCI were tentatively assigned as Zea, as our HPLC result indicated that DcPSI-LHCI bind constitutive Zea molecules (Supplementary Figs. 14a, b). To build the model of DcPSI-6/2LHCI and DcPSI-core, DcPSI-8LHCI structure without the extra part was manually fitted into the corresponding maps and adjusted in Coot⁶³. All structures were refined in real space against the cryo-EM map using Phenix⁶² with geometry and secondary structure restraints. The geometries of structures were assessed using Phenix⁶² and MolProbity⁶⁴ (Supplementary Table 1). The figures for publication were prepared using PyMOL⁶⁵. Sequence alignment was performed using online UniProt Align⁶⁶ and ESPript 3.0⁶⁷.

Multi-body refinement and analysis

To evaluate the mobility of Lhca belts relative to the PSI core moiety in the DcPSI-8LHCI complex, the refined particles obtained after CTF refinement and Bayesian polishing procedures were subjected to multi-body refinement in RELION3.1⁶⁸. Four rigid bodies, corresponding to the PSI core, inner belt, outer belt and side belt were assigned by individual masks. Initial angular sampling rate, offset range and offset step were set to 7.5°, 6 pixels and 1 pixel, respectively. The top two eigenvectors accounted for 26.7% and 23.3% of variance in the rotations and translations of the four bodies. Each eigenvector comprising ten maps, bin001 and bin010 represents the two extreme positions of motion along a specific eigenvector. The structural models of the PSI core, inner belt, outer belt and side belt were fit into the bin001 and bin010 maps of the specific component in Chimera⁶¹.

FRET calculation

The FRET rate constant (k_FRET) was obtained based on the theory of FRET⁴¹, and the formula is k_FRET = (CK²)/(n⁴R⁶), where C is the factor calculated from the spectral overlap integral between the two chlorophylls, K is the dipole orientation factor, n is the refractive index and R is the distance between the central magnesium atoms of two chlorophylls. As estimated in ref. ⁶⁹, the C values for Chl a to Chl a, Chl a to Chl b, Chl b to Chl a and Chl b to Chl b energy transfer were 32.26, 1.11, 9.61 and 14.45, respectively. K² is defined as $K^2={\left[{\widehat{\mathrm{u}}}_{\mathrm{D}}\cdot {\widehat{\mathrm{u}}}_{\mathrm{A}}-3\left({\widehat{\mathrm{u}}}_{\mathrm{D}}\cdot {\widehat{\mathrm{R}}}_{\mathrm{DA}}\right)\left({\widehat{\mathrm{u}}}_{\mathrm{A}}\cdot {\widehat{\mathrm{R}}}_{\mathrm{DA}}\right)\right]}^2$, where ${\widehat{\mathrm{u}}}_{\mathrm{D}}$ and ${\widehat{\mathrm{u}}}_{\mathrm{A}}$ are the unit vectors of the transition dipoles of the donor and the acceptor chlorophylls derived from the vectors of the coordinates of NB and ND atoms, ${\widehat{\mathrm{R}}}_{\mathrm{DA}}$ is the unit vector along the direction connecting the ‘central magnesium atoms’ from the donor chlorophyll to the acceptor chlorophyll. The n value is 1.55. FRET rates were calculated computationally using Kim’s algorithm on the Python platform (Python v.3.8) and visualized using the interactive platform Gephi⁷⁰. The Python script used for FRET rate calculation is available at (https://doi.org/10.5281/zenodo.3250649).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

M.L. and X.W. conceived and supervised the project; X.S. performed the sample preparation and characterization, cryo-EM data collection and processing, model building and refinement; L.S., Z.L. and X.Z. cultivated D. chodatii cells and assisted in sample preparation; X.S. and M.L. analyzed the data; X.S., X.W. and M.L. wrote the manuscript; all authors discussed and commented on the manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

The project was funded by the National Key R&D Program of China (2025YFF0512900), National Natural Science Foundation of China (32171183 and 31930064), the Youth Innovation Promotion Association at the Chinese Academy of Sciences (Y2022038), the CAS Project for Young Scientists in Basic Research (YSBR-015) and the Space Application System of China Manned Space Program (KJZ-YY-WSM05).

Data availability

The cryo-EM density maps and atomic models generated in this study have been deposited in the Electron Microscopy Data Bank (EMDB) and the Protein Data Bank (PDB), respectively, with the following accession codes: DcPSI-8LHCI (EMDB accession code EMD-66561, PDB accession code 9X4O, DcPSI-6LHCI (EMDB accession code EMD-66562; PDB accession code 9X4P, DcPSI-2LHCI (EMDB accession code EMD-81136; PDB accession code 27GG, DcPSI-core (EMDB accession code EMD-66564, PDB accession code 9X4R, All other relevant data generated in this study are provided in the Supplementary Information and Source data files. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77656-0.