Structure of a unique PSII-Pcb tetrameric megacomplex in a chlorophyll d–containing cyanobacterium

Abstract

Acaryochloris marina is a unique cyanobacterium using chlorophyll d (Chl d) as its major pigment and thus can use far-red light for photosynthesis. Photosystem II (PSII) of A. marina associates with a number of prochlorophyte Chl-binding (Pcb) proteins to act as the light-harvesting system. We report here the cryo-electron microscopic structure of a PSII-Pcb megacomplex from A. marina at a 3.6-angstrom overall resolution and a 3.3-angstrom local resolution. The megacomplex is organized as a tetramer consisting of two PSII core dimers flanked by sixteen symmetrically related Pcb proteins, with a total molecular weight of 1.9 megadaltons. The structure reveals the detailed organization of PSII core consisting of 15 known protein subunits and an unknown subunit, the assembly of 4 Pcb antennas within each PSII monomer, and possible pathways of energy transfer within the megacomplex, providing deep insights into energy transfer and dissipation mechanisms within the PSII-Pcb megacomplex involved in far-red light utilization.

The structure of a tetrameric PSII-Pcb from Acaryochloris marina is solved, providing insights into far-red light utilization.

INTRODUCTION

Oxygenic photosynthesis uses solar energy to drive oxidation of water and reduction of carbon dioxide (CO₂), resulting in the generation of molecular oxygen and carbohydrates indispensable for survival and evolution of almost all life forms on Earth. The light reactions in photosynthesis are catalyzed by photosystem I (PSI) and PSII, two multi-subunit membrane-protein complexes located in the thylakoid membranes of plants, algae, and cyanobacteria (1, 2). Among these two photosystems, PSII catalyzes the light-driven oxidation of water into molecular oxygen and the reduction of plastoquinone, thereby converting light energy into chemical energy (3, 4). After initial light induced charge separation between the primary electron donor (special pair) and primary electron acceptor (pheophytin), the electron is transferred via the primary and secondary quinone acceptors, Q_A and Q_B (3–5), to the plastoquinone pool to be used in subsequent reactions in the thylakoid membrane.

The mature PSII supercomplex exists typically in a homodimeric form composed of a reaction center (RC) core and peripheral antenna protein subunits (1). The structure of the PSII core is rather conserved during evolution among different photosynthetic organisms, whereas the light-harvesting antenna proteins and the pigments they bind, as well as supramolecular organization of photosynthetic apparatus exhibit a remarkable diversity in various photosynthetic organisms mainly as a result of adaptation to different light environments (6–8). In cyanobacteria and red algae, a water-soluble phycobiliprotein complex phycobilisome associates to the stromal side of the PSII core and serves as the antenna system to harvest light energy (9, 10), whereas, in higher plants, green algae, and diatoms, the peripheral antenna system is composed of variable numbers and species of transmembrane light-harvesting complex II (LHCII) proteins or fucoxanthin-chlorophyll (Chl) a/c–binding proteins (11–15).

Acaryochloris marina is a unique cyanobacterium that contains a large amount of unusual Chl, Chl d, in addition to Chl a found in other photosynthetic organisms (16–19). Although both Chl d and Chl a share a similar structure, Chl d has a formyl group at the C-3 position of the porphyrin ring I, instead of a vinyl group seen in Chl a. This structural difference causes a redshift of the Q_Y maximum absorption of Chl d by around 30 nm relative to that of Chl a (20–22). Consequently, the maximum absorption of A. marina cells is observed at around 710 to 720 nm (22). As the major pigment, Chl d is bound to the proteins of the PSI and PSII core complexes, where it not only serves as a light-harvesting pigment (23) but also may be involved in primary charge separation in the RCs (17, 18, 23–30). Intriguingly, the structure of A. marina PSI RC has been determined by cryo–electron microscopy (cryo-EM) recently, which reveals a unique type I RC where the primary electron donor P740 is a special pair of Chl d, and pheophytin a (Pheo a) serves as the primary electron acceptor (24, 25). On the other hand, the nature of the primary electron donor of PSII from A. marina has been controversial due to the absence of the structure of PSII from A. marina (26–32).

In addition to water-soluble phycobiliproteins that act as light-harvesting antenna in PSII of A. marina (33, 34), like prochlorophytes (35), A. marina contains two types of pcb gene families, pcbA and pcbC, which encode six-transmembrane, Chl d–binding antenna proteins PcbA and PcbC, respectively (36, 37). These Pcb proteins are homologous to CP43 and CP47 of PSII and the iron-stress–induced protein A (IsiA) in some cyanobacteria (38–41) but are structurally different from the Chl a/b–binding antenna proteins of higher plants and green algae. Previously, a PSII-Pcb megacomplex from A. marina has been isolated, and a two-dimensional (2D) projection map has been obtained, indicating that this megacomplex is mainly composed of two PSII dimers flanked by eight Pcbs (42), which is similar to the structure observed in one PSII dimer surrounded by 10 Pcb antennas from Prochloron didemni (35). However, detailed knowledge regarding the composition of the protein subunits and pigments of the megacomplex, the precise nature of the cofactors involved in the electron transfer reaction, the assembly of Pcb subunits and PSII core, and the energy transfer within this complex are elusive because of the absence of a high-resolution structure of the complete PSII-Pcb megacomplex.

To understand the molecular structure of PSII-Pcb in A. marina, we purified the PSII-Pcb megacomplex from A. marina and resolved the structure using single-particle cryo-EM. 3D cryo-EM density maps of a PSII-Pcb tetramer and a PSII-Pcb dimer were obtained at an overall resolution of 3.6 and 3.3 Å, respectively. Our structure reveals the characteristics and molecular assembly of PSII-Pcb megacomplex in A. marina and provides deep insights into the arrangement of protein subunits and pigments as well as energy transfer pathways within this large complex. This provides a structural basis to understand the molecular mechanism of adaption of the Chl d–type cyanobacteria to far-red light and low light.

RESULTS

Overall structure of the tetrameric PSII-Pcb megacomplex

The tetrameric PSII-Pcb megacomplex was isolated from a Chl d–containing cyanobacterium A. marina (Materials and Methods and Fig. 1A) and characterized by size exclusion chromatography, electrophoresis, absorption and fluorescence spectroscopy, pigment quantification (Fig. 1, B to G), and negative-staining electron microscopy (fig. S1A). The SDS–polyacrylamide gel electrophoresis (PAGE) and peptide composition of the sample isolated by the sucrose density gradient (SDG) centrifugation contained the subunits of PSII core and Pcb antenna and also the protein subunits PsaA/PsaB of PSI, indicating that this sample has a contamination of PSI (Fig. 1C). The room-temperature absorption spectrum of the isolated sample showed a maximum absorbance of the Q_y peak at 707.1 nm (Fig. 1D), which was redshifted by around 30 nm relative to Chl a due to the presence of Chl d. The fluorescence emission spectrum at 77 K showed the maximal fluorescence peak of 730 nm, suggesting that the excitation energy is transferred to the Chl d–containing PSII and PSI RC cores (Fig. 1E). Five types of pigments [zeaxanthin (Zea), Chl d, Chl a, α-carotene (α-Car), and Pheo a] were identified according to their characteristic absorption spectra and elution times in the high-performance liquid chromatography (HPLC) (Fig. 1, F and G). Among them, Chl d was the major pigment similar to that reported in the previous studies (27), and the ratio of Chl d to Chl a is 91. While Zea was detected as the major carotenoids, the purified sample also contained a small amount of α-Car, whereas no β-Car present in most of other cyanobacteria was found (Fig. 1F).

Fig. 1. Sample purification and biochemical characterization of the PSII-Pcb tetramer megacomplex from A. marina.

(A) Separation of the thylakoid membranes solubilized by n-dodecyl-β-d-maltoside (β-DDM) by sucrose density gradient (SDG) centrifugation from A. marina. (B) Size exclusion chromatographic elution profile of the PSII-Pcb–enriched fraction isolated by SDG from A. marina. Elution was performed at 4°C with a Superose 6 Increase 10/300 GL column (flow rate of 100 μl min⁻¹) and monitored by absorption at 280 nm. mAU, artifical unit. (C) SDS–polyacrylamide gel electrophoresis (PAGE) analysis of the PSII-Pcb tetrameric megacomplex–enriched sample purified from A. marina. The proteins labeled were identified by mass spectrometry (MS) analysis (see source data 1). Lane 1, marker; lane 2, PSII-Pcb after size exclusion chromatography (20 μg of Chl d). (D) Room-temperature absorption spectrum of the PSII-Pcb tetrameric megacomplex–enriched sample. (E) Low temperature (77 K) fluorescence emission spectrum of the PSII-Pcb tetrameric megacomplex–enriched sample excited at 398 nm. (F) HPLC analysis of pigments of the PSII-Pcb megacomplex–enriched sample. Five major pigment peaks are eluted from the PSII-Pcb tetramer megacomplex and are identified as zeaxanthin (Zea), chlorophyll d (Chl d), chlorophyll a (Chl a), α-carotene (α-Car), and pheophytin a (Pheo a), respectively, based on their elution time and characteristic absorption spectra. (G) Room-temperature absorption spectra of Chl a from PSII of T. vulcanus and Zea/Chl d from PSII-Pcb tetramer–enriched sample of A. marina. The pigment analysis shown in (F) were conducted more than three times, and all showed the same results as shown here.

To solve the structure by cryo-EM, we collected 11,708 cryo-EM micrographs and picked up 1,146,757 particles for subsequent data processing (fig. S1). After 2D and 3D classifications, a PSII-Pcb tetramer density map was obtained at an overall resolution of 3.6 Å upon imposing a C2 symmetry (fig. S1 and table S1). Upon masking one dimer of the PSII-Pcb tetramer, a local resolution of 3.3 Å was obtained for a PSII-Pcb dimer (fig. S1D). According to these density maps, the structure of the PSII-Pcb complex is built, and the statistical parameters are shown in table S1.

The PSII-Pcb megacomplex is a tetramer with dimensions of ~340 Å in length, 200 Å in width, and 90 Å in height (Fig. 2A). Two PSII-Pcb dimers are arranged side by side and associate with each other by interactions of both PSII core subunits and Pcb antennas. Each PSII-Pcb dimer is a homodimer with a twofold symmetry, similar to that found in algae and higher plants previously (Fig. 2B) (11–13). Each PSII monomer contains 15 core subunits and four Pcb antenna subunits (named as PcbA2, PcbA6, PcbA3, and PcbC2) (Fig. 2C). Among these Pcb subunits, PcbA2 and PcbA6 bind directly to the CP43 side of the PSII core, whereas PcbA3 and PcbC2 are attached to the CP47 side through PsbH. In the interstice between two Pcb subunits (between PcbA6 and PcbA3 of the adjacent PSII monomer) and CP47, a previously unidentified subunit with a single transmembrane helix was found, but its identity could not be determined due to the limited resolution. This subunit was hence constructed with poly-alanines without side chains and named unknown transmembrane protein (UTP). It may be involved in the association of the Pcb antennas with the PSII core.

Fig. 2. Overall structure of the PSII-Pcb tetramer from A. marina.

(A) Cryo-EM density map of the PSII-Pcb tetramer. Protein subunits are depicted in different colors. (B) Overall structure of a PSII-Pcb dimer with a top view and a side view, respectively. Each subunit is represented as cylindrical cartoons and colored differently as in (A). (C) Structure of a PSII-Pcb monomer viewed from the stromal side, with each subunit colored the same as in (A).

In addition to the protein subunits, we identified 106 Chls, 14 Zeas, 11 α-Cars, one b-type cytochrome, two pheophytins, one nonheme ion, one bicarbonate ion, one plastoquinone, and a large number of lipids in each PSII-Pcb monomer (fig. S2 and table S2). Carotenoids are modeled as Zeas and α-Cars according to the HPLC results as well as previous biochemical analysis (28, 43). Because of the similarity between the formyl group of Chl d and the vinyl group of Chl a (21), we could not distinguish the pigments between Chl d and Chl a clearly. Thus, all Chls were modeled as Chl d. In total, 80 subunits and 624 cofactors were identified in the whole PSII-tetramer of A. marina, which gives rise to a total molecular weight of ~1.9 MDa.

Structure of the PSII core of A. marina

The structure of the PSII core from A. marina is highly similar to that of other cyanobacteria, green algae, and higher plants reported previously (5, 11–13). In each monomer, 15 PSII core subunits, including four large transmembrane subunits (D1, D2, CP43, and CP47), and 11 low–molecular weight transmembrane subunits (PsbE, PsbF, PsbH, PsbI, PsbK, PsbL, PsbM, PsbT, PsbX, Ycf12, and PsbZ) were identified (fig. S2). Notably, PsbJ, PsbY, and the three extrinsic subunits—PsbO, PsbV, and PsbU—were not found in the density map, suggesting that they are loosely bound and have been dissociated from the PSII core during purification. PsbJ is located between PsbK and PsbF, and the absence of this subunit may lead to the loss of Q_B, Mn₄CaO₅ cluster, and extrinsic subunits (PsbO, PsbV, and PsbU) (5, 44, 45). PsbY is located in the vicinity of PsbE and is easily lost during the process of sample purification (46). Upon superposition on the basis of CP47, PsbE, PsbF, Ycf12, and PsbZ in A. marina were found to be shifted by 3.7 Å compared with their structures in Thermosynechococcus vulcanus [Protein Data Bank (PDB) code: 3WU2] (5) and Synechocystis sp. PCC 6803 (PDB code: 6WJ6) (46). This shift is possibly caused by the traction of PcbA2 and PcbA6. On the other hand, cofactors including 37 Chls (one additional Chl in CP47 and one in CP43 are found, see below), two pheophytins, 11 α-Cars, one b-type cytochrome, one nonheme ion, one bicarbonate ion, one plastoquinone, and 13 lipids were identified. No electron density corresponding to the Mn₄CaO₅ cluster was observed in each PSII core, indicating the loss of the Mn₄CaO₅ cluster in the isolated sample.

Three copies of the genes encoding the D1 and D2 subunits, respectively, are present in A. marina (47, 48), and we found that the D1 and D2 subunits in the structure are encoded by psbA2 and psbD1 genes with their UniProt numbers corresponding to A5A8K9 and B0C1V6, respectively (fig. S3). This assignment is consistent with the mass spectrometry (MS) results of the SDS-PAGE bands (Fig. 1C and source data 1). The assignment of psbA2 gene in the structure is also confirmed by the sequence alignment as three residues of Asp¹⁷⁰, Glu³³³, and Asp³⁴² serving as ligands for the Mn₄CaO₅ cluster are completely conserved in psbA2 and other species but are changed to Glu¹⁷⁰, Ser³³³, and Thr³⁴² in the psbA1 gene (fig. S3A). PsbA1 belongs to a divergent paralog “rogue D1” family and its encoding gene psbA1 is inactive under the current A. marina culture conditions (48, 49). Comparison of the PSII core subunits among A. marina, T. vulcanus, and Synechocystis 6803 (Fig. 3A and figs. S4 and S5A) revealed that the C-terminal region of D1 beyond residue Ala³³⁶ in A. marina is invisible. It is unlikely that the C-terminal peptide is cut in the D1 subunit of A. marina; therefore, it is most likely that the C-terminal region become flexible due to the loss of the Mn₄CaO₅ cluster, leading to the invisibility of the cryo-EM map of this region.

Fig. 3. Comparison of PSII core structures among A. marina, T. vulcanus (PDB code 3WU2), and Synechocystis 6803 (PDB code 7N8O), and arrangement of cofactors of the electron transfer chain in A. marina PSII.

(A) Comparison of the PSII core structure among the three cyanobacterial species, with the C-terminal region of D1 enlarged in the right side. The PSII cores of T. vulcanus and Synechocystis 6803 are colored in black and gray, respectively, and the subunits of A. marina PSII core are colored differently as in Fig. 2. (B) Structural comparison of the CP43 subunit among the three species of cyanobacteria. One extra Chl molecule (Chl d517) and a lipid molecule (MGDG518) were found in CP43 of A. marina. (C) Structural comparison of the CP47 subunit among the three species of cyanobacteria. The CP47 subunit of A. marina has one extra Chl molecule (Chl d623) and a lipid molecule (PG624). In addition, the AB loop and EF loop of the A. marina CP47 subunit differ from those of the other species. (D) The arrangement of major cofactors in the electron transport chain of A. marina PSII. The phytol tails of the Chls and pheophytins have been removed for clarity. The coloring scheme is the same as in Fig. 2. The distances between electron transfer cofactors are given in angstroms. (E) The arrangement of the special pair (P_D1 and P_D2) and accessory Chls (Chl_D1 and Chl_D2) in the A. marina PSII core.

In the CP43 subunit of A. marina, a Chl d molecule (Chl d517) ligated by His²⁵⁵ and a lipid molecule (MGDG518) were newly found compared with that of other cyanobacteria (Fig. 3B). Notably, Chl d517 is situated between the PcbA2/PcbA6 antenna and the PSII core, which could potentially facilitate the energy transfer from PcbA2/PcbA6 to the PSII core and enhance the interaction between Pcb antennas and the PSII core (Fig. 3B). In CP47 of A. marina, an extra AB loop and a smaller EF loop were found, which differ from those of other cyanobacteria (Fig. 3C). In addition, CP47 of A. marina contains an extra lipid molecule (PG624) and a Chl d molecule (Chl d623), and the Chl d623 is ligated by His⁹⁵ within the extra AB loop. This pigment is located between the PcbA3/UTP and the PSII core, which may mediate the connection between the PSII core and PcbA3 antenna (Fig. 3C). The predicted structure of PsbO of A. marina using the Alphafold showed a similar structure as that of T. vulcanus and Synechocystis 6803, but there is an additional alpha helix in the N terminus of PsbO from A. marina compared to the PsbO in the crystal structure of T. vulcanus PSII (PDB code 3WU2) and the cyo-EM structure of Synechocystis 6803 PSII (PDB code 7N8O) (fig. S5B). At the lumenal side, spatial conflict was found between the EF loop of CP47 and loop^β8-β9 of the A. marina PsbO structure predicted by AlphaFold. This might hinder the binding of PsbO with PSII, which may be one of the reasons for the unstable binding of PsbO in A. marina (fig. S5B).

In the PSII RC, most of the cofactors existed in other oxygenic organisms were found in the core of A. marina, with the exception of the Mn₄CaO₅ cluster and Q_B (Fig. 3, D and E). The relative distances between cofactors in the PSII RC are highly conserved among the structures of A. marina, T. vulcanus, and Synechocystis 6803 (fig. S5C). We modeled P_D1/P_D2, Chl_D1/Chl_D2, and Chl_ZD1/Chl_ZD2 as Chl d based on the pigment analysis and previous studies with purified PSII (27, 28), although there is a possibility that one of them may be Chl a. P_D1/P_D2 at the interface of D1 and D2 are coordinated by His¹⁹⁸ of D1 and His¹⁹⁶ of D2, respectively. Two accessory Chls d (Chl_D1 and Chl_D2) are located close to P_D1 and P_D2, respectively, and the two Chls d (Chl_ZD1 and Chl_ZD2) are symmetrically distributed on the peripheral of the D1 and D2 proteins (Fig. 3, D and E). Toward the stromal side, a pair of Pheo a molecules (Pheo_D1 and Pheo_D2) were identified (Fig. 1F and fig. S2). After the primary electron acceptor, one plastoquinone (Q_A), one non-haem iron, and one bicarbonate ion were identified in the similar positions as those reported previously (5).

Comparison of the core structures between A. marina and T. vulcanus (5) showed that, while amino acid residues surrounding P_D1, Chl_D1, Pheo_D1/Pheo_D2, Q_A, and the non-haem iron are completely conserved, the amino acid residues surrounding P_D2 and Chl_D2 are largely changed (fig. S5, D to I). In the vicinity of P_D2, His¹⁹⁶_D2 coordinates Mg of Chl_D2 in A. marina PSII, which is similar to His¹⁹⁷_D2 in the T. vulcanus PSII, whereas Trp¹⁹¹_D2, Phe¹⁸⁵_D2, and Phe²⁰⁶_D1 of T. vulcanus PSII are replaced by Leu¹⁹⁰_D2, Ala¹⁸⁴_D2, and Leu²⁰⁶_D1 in the A. marina PSII, respectively (fig. S5J). In the protein environment of Chl_D2, Vla²⁰²_D1, Ile¹⁷⁷_D2, and Phe¹⁷⁸_D2 in the vicinity of Chl_D2 in A. marina PSII are similar to Vla²⁰²_D1, Ile¹⁷⁷_D2, and Phe¹⁷⁸_D2 in T. vulcanus PSII, whereas Gln¹⁹⁹_D1, Phe¹⁷³_D2, and Phe²⁰⁶_D2 of T. vulcanus PSII are changed to Met¹⁹⁹_D1, Trp¹⁷²_D2, and Leu²⁰⁶_D2 in the A. marina PSII (fig. S5K). These structural changes may bring effects on the redox potential of the P_D1/P_D2 and Chl_D1/Chl _D2 pairs, although these changes are occurred mainly between hydrophobic/hydrophobic residues, and no hydrogen bond interactions between these key Chls and the surrounding amino acid residues are found.

Structural features of the Pcb antennas

Sixteen Pcb antennas are found in the structure of the PSII-Pcb tetramer of A. marina, with each PSII core binding four different Pcb antenna subunits (PcbA2, PcbA6, PcbA3, and PcbC2). It has been reported that the Pcb antenna is a CP43-like protein homologous to the six-transmembrane Chl-binding IsiA protein (38–41). Nine protein sequences homologous to CP43 are identified in the A. marina genome upon Blast analysis of the CP43-like proteins, which are named B0C2V8, B0C576, B0C6I0, B0C009, B0C012, B0C3E5, B0C011, B0C3E6, and A8ZMK2 in the UniProt database and ABW26174.1, ABW26316.1, ABW26401.1, ABW28356.1, ABW28359.1, ABW28644.1, ABW28358.1, ABW28645.1, and ABW32413.1 in the National Center for Biotechnology Information (NCBI) database, respectively. A phylogenetic tree was constructed for the nine proteins of A. marina combined with other homologous subunits from various algae. On the basis of this analysis, A8ZMK2 is grouped as an IsiA-type antenna; B0C2V8, B0C012, B0C576, B0C6I0, B0C3E5, and B0C009 are classed as PcbA-type antenna; and B0C011 and B0C3E6 belong to the PcbC-type antenna (fig. S6, A and B). The types of Pcb antennas are determined according to the density of each subunit in the high-resolution structure of the PSII-Pcb megacomplex (fig. S6, C to F). PcbA2 and PcbA6 were identified as B0C576 (Fig. 4A and fig. S6D) and B0C3E5 (Fig. 4B and fig. S6C) by comparing the structural features of residues Thr²⁵⁴ and Leu²⁷⁵ of PcbA2 as well as residues Leu²⁶¹, Val²⁷⁵, and Trp²³³ of PcbA6, respectively. PcbA3 and PcbC2 were assigned as B0C6I0 and B0C3E6 by comparing the structural features of residues Tyr²²⁰ and Leu¹⁰⁴ of PcbA3 (Fig. 4C and fig. S6E) as well as residues Ile²⁶⁴ and Phe¹¹³ of PcbC2, respectively (Fig. 4D and fig. S6F). These results indicate that PcbA2, PcbA6, and PcbA3 are PcbA-type antennas, whereas PcbC2 is a PcbC-type antenna. The assignment of these antenna subunits is also consistent with the MS results of the SDS-PAGE bands (Fig. 1C and source data 1).

Fig. 4. Structures of four Pcb antennas and their comparisons.

(A to D) Structures of the four Pcb antennas. The colors of PcbA2, PcbA6, PcbA3, and PcbC2 are yellow green, cyan, deep orange, and brown, respectively. (E) Superposition of all four Pcb monomers, with the regions having different structures circled by black boxes and shown in the following panels. (F) Differences between the C-terminal and N-terminal regions of the four Pcb antennas. (G) Differences of pigment distributions around helix A and helix F of the four Pcb antennas. PcbA2 has an extra Chl d518 around helix A, whose density map is represented by gray meshes. (H) Differences between the four Pcb antennas in the EF loop. (I) Differences between the four Pcb antennas in the DE loop. (J) Differences of Car of four Pcb antennas around helix A. (K) Differences between the four Pcb antennas in the CD loop.

Both PcbA- and PcbC-type antennas in the PSII-Pcb megacomplex contain six transmembrane helices, with their N-terminal and C-terminal regions located on the stromal side (Fig. 4, A to E), which resembles the IsiA protein induced by iron deficiency in some cyanobacteria (38–41). This arrangement is thus different from the typical structures of LHCII seen in green algae and higher plants, which has three transmembrane helices (11–13). In addition, the Pcb antennas bind mainly Chl d, Zea, and lipid molecules. Different numbers of cofactors are found in various Pcb subunits. PcbA2 contains 19 Chls d, three Zeas, and one lipid; both PcbA6 and PcbA3 contain 18 Chls d, four Zeas, and one lipid molecule; and PcbC2 has 14 Chls d and three Zeas but without any lipid molecules.

Although the transmembrane helix regions of the four Pcb antennas are relatively conserved, there are some structural differences among these Pcb antennas. This may lead to variations in the modes of assembly and connection between the Pcb antennas and the PSII core, as well as in the pigment binding, light capture, and energy transfer processes. The C-terminal region of PcbA6 is longer than that of the other three antennas, whereas the C-terminal region of PcbC2 is the shortest compared with other three antennas. The N-terminal tail of PcbA2 is the shortest compared with the other three Pcb antennas in PSII-Pcb megacomplex (Fig. 4F). Among pigments located between the Pcb antenna and PSII core, the PcbA2 antenna has an additional Chl d molecule (Chl d518) near helix A, whereas a Chl d molecule (Chl d504) and a Zea molecule (Zea521) observed in PcbA2, PcbA6, and PcbA3 are absent around helix F of the PcbC2 antenna. The absence of these pigments in PcbC2 may affect the energy transfer from PcbC2 to the PSII core (Fig. 4G). PcbC2 lacks an EF loop and associated Chl d517 in the lumenal side, which are present in PcbA2, PcbA6, and PcbA3 and may be crucial for the interactions between different Pcb antennas (Fig. 4H). At the stromal side, PcbC2 has an extra DE loop compared to the other three Pcb antennas, which is able to mediate the connection between PcbC2 and PcbA2 of the adjacent PSII-Pcb dimer supercomplex (Fig. 4I). Moreover, the head group of Zea522 in PcbC2 is shifted closer to Chl d505, and a Chl d molecule (Chl d516) is absent around helix A of PcbC2 compared to the other three Pcb antennas (Fig. 4J). PcbC2 contains an extra Chl d (Chl d518) around the CD loop but misses two Chls (Chl d515 and Chl d519) identified in other three Pcb antennas (Fig. 4K). These Chl molecules are located in the interface between the Pcb antennas, and their absence may reduce the energy transfer efficiency among different Pcb antennas.

Comparisons of the protein sequences of four Pcb antennas in A. marina with the IsiA protein sequence of Synechococcus sp. PCC 7942 showed that, while the PcbA-type antennas (PcbA2, PcbA6, and PcbA3) had a similar sequence to IsiA, the PcbC-type antenna (PcbC2) had a rather different sequence from that of IsiA (fig. S7A). This is consistent with the reported phylogenetic relationship between CP43 and Pcb proteins and support the idea that PcbA and IsiA are more closely related to each other than to PcbC (50). The structures of the PcbA-type antenna are also highly similar to that of IsiA (fig. S7B), although the lengths of the C-terminal region of PcbA6 and the N-terminal region of PcbA2 are different from that of IsiA. In addition, PcbA-type antennas contain an extra Chl (Chl d519) around the CD loop, which may enhance the energy transfer efficiency among Pcb antennas. On the other hand, there are remarkable differences between the structures of the PcbC-type antenna and IsiA (fig. S7C). The PcbC2 antenna has fewer Chls and carotenoid molecules than IsiA around helix F, IsiA has a longer C-terminal tail, and the positions of the N-terminal region of IsiA and the PcbC2 antenna are different. Furthermore, PcbC2 lacks an EF loop observed in the lumenal side of IsiA but contains a long DE loop at the stromal side and a CD loop at the lumenal side, both of which may enhance its interactions with PcbA2 of the adjacent PSII-Pcb supercomplex.

Assembly of the tetrameric PSII-Pcb megacomplex

Tight association between the Pcb antennas and PSII core are essential for formation of a stable and integrated PSII-Pcb tetrameric megacomplex (Fig. 5). At the stromal side, hydrophobic interaction is found between Leu³⁴⁴ at the C-terminal region of PcbA3 and Val²²⁶ in helix D of CP47. Trp³¹ and Trp³² in the N-terminal region of PcbC2 form hydrophobic interactions with Phe²²³ in helix D of CP47 and Va^l31 of PsbH, respectively (Fig. 5A). At the lumenal side, Ile³²⁸ in helix F of PcbC2 is involved in hydrophobic interactions with Phe¹¹ in the transmembrane helix of PsbX. These results indicate that PcbA3 and PcbC2 are mainly associated with CP47, PsbH, and PsbX in the PSII core via hydrophobic forces. Moreover, LHG624_CP47 and α-Car101_PsbH are located between PcbC2 and the PSII core, which may help to enhance the interactions between the Pcb antennas and PSII core (Fig. 5B). PcbA2 is located at the CP43 side, and Ile²⁹² on the EF loop of PcbA2 forms a hydrophobic interaction with Trp²¹⁶ on the CD loop of CP43. In addition, residue Arg⁵² in helix A of PcbA6 forms a hydrogen bond with the head group of a lipid molecule (LMG518) in CP43 (Fig. 5C). The C-terminal tail of PcbA2 extends into the stromal side of CP43 and forms broad and tight hydrophobic interactions with the DE loop of CP43 (Fig. 5D). At the lumenal side, the aromatic residues and a hydrophilic residue on helix F and EF loop of PcbA2 form hydrophobic as well as hydrophilic interactions with an aromatic and a hydrophilic residue on helix C of CP43, respectively (Fig. 5E).

Fig. 5. Interactions among different subunits of the PSII-Pcb tetramer of A. marina.

The overall structure of the PSII-Pcb tetramer is shown in the middle, where interaction areas are boxed and shown in the panels represented by the letter written within the boxes. (A) Interface between PcbA3/PcbC2 and CP47/PsbH of the PSII core at the stromal side. (B) Interactions between PcbC2 and PsbX. (C) Interactions between PcbA2/PcbA6 and the CP43 subunit of PSII core at the lumenal side. (D) Interactions between the C-terminal region of PcbA2 and CP43. (E) Interactions between PcbA2 and CP43 at the lumenal side. (F) Connection between PcbA6 and PcbA3 through the UTP subunit. (G) Interactions of PcbA2 and PcbC2 between the adjacent PSII-Pcb dimers at the stromal side. (H) Interactions of PcbA2 and PcbC2 between the adjacent PSII-Pcb dimers at the lumenal side. (I and J) Interactions of PsbE/PsbF/Ycf12 with PsbZ between the adjacent PSII-Pcb dimers. The interaction interface was shown in a surface mode (I), and the amino acid residues at the interaction interface were shown in a stick mode.

The Pcb antennas interact with each other extensively to ensure that they are integrated and attached to the periphery of the PSII core. At the stromal side, the C-terminal tail of PcbA2 extends to a region of PcbA6 in a direction parallel to the membrane plane and forms ionic interactions with the N-terminal region of PcbA6. Moreover, the residues Leu³²⁹, Leu³³³, Met³³⁶, and Phe³³⁸ in helix F of PcbA2 form extensive hydrophobic interactions with Ile¹⁰⁸ and Leu¹¹² in helix B of PcbA6 (Fig. 6A). At the lumenal side, residues Phe²⁹⁰ and Val²⁸⁸ in the EF loop of PcbA2 form hydrophobic interactions with Leu⁴⁷ and Phe⁴⁸ of helix A and Phe⁹⁰ of helix B of PcbA6. Tyr²⁶⁸ of helix E of PcbA2 forms aromatic-aromatic interactions with Phe⁹⁰ of PcbA6. In addition, there are a number of aromatic residues in the interface between helix E and CD loop of PcbA2 and helix B of PcbA6, and these loops form extensive and close hydrophobic interactions as well as aromatic-aromatic interactions (Fig. 6A).

Fig. 6. Interactions among the Pcb antennas within a PSII-Pcb tetramer megacomplex of A. marina.

(A) Interactions between PcbA2 and PcbA6. Main structures of the two subunits are depicted in a transparent surface mode. Interfaces of different areas are circled by boxes and enlarged in the separate panels. Connections between different residues are denoted by black dotted lines, and their distances are depicted in angstroms. (B) Interactions between PcbA6 and PcbA3. (C) Interactions between PcbA3 and PcbC2.

Interactions between PcbA6 and PcbA3 from the adjacent PSII-Pcb monomer contribute to the formation of the PSII-Pcb dimer. An unknown protein UTP with a single transmembrane helix was found between PcbA6 and PcbA3. This UTP interacts with the C-terminal region of PcbA6 at the stromal side and with helix A, EF loop, and Chl d504 of PcbA3 at the lumenal side, respectively (Fig. 5F). Therefore, this UTP serves as a linker to connect PcbA6 and PcbA3 antennas tightly between two adjacent PSII-Pcb monomers of one PSII dimer, which may stabilize the PSII dimer and the peripheral Pcb antennas. Moreover, at the stromal side, cation-π interactions are formed between Arg²³⁶ on DE loop of PcbA6 and Phe¹²³ in BC loop of PcbA3 (Fig. 6B). The unique and elongated C-terminal tail region of PcbA6 folds back and interacts with multiple regions of PcbA3. Specifically, Glu³⁴⁷_PcbA6 forms a hydrogen bond and an ionic interaction with Gln²⁶ of helix A and Lys¹¹⁶ in the BC loop of PcbA3, respectively. The residues Phe³⁴⁰ and Phe³⁴¹ in the C-terminal region of PcbA6 form aromatic-aromatic interactions with Trp¹³ of the N-terminal region of PcbA3. The Leu³²³ and Leu³²⁷ in helix F of PcbA6 form hydrophobic interactions with Leu¹⁰⁸ of helix B in PcbA3. At the lumenal side, Pro²⁸² and Leu²⁸⁴ in EF loop of PcbA6 form hydrophobic interactions with Phe⁹⁰ of helix B and Leu⁴⁷/Phe⁴⁸ of helix A of PcbA3. Thus, there are many aromatic residues in the interface between PcbA6 and helix B of PcbA3, resulting in a wide range of hydrophobic interactions between them, which may facilitate the association of the two PSII-Pcb monomers (Fig. 6B).

As a PcbC-type antenna, PcbC2 has a structure different from the other three Pcb antennas; hence, the interaction between PcbA3 and PcbC2 is also different from those observed between other Pcb antennas (Fig. 6C). At the stromal side, a hydrophobic interaction is formed between Leu³⁴⁴ in the C-terminal of PcbA3 and Trp³¹ in the N-terminal of PcbC2. A few residues, such as Leu and Phe, are distributed between helix F of PcbA3 and helix B of PcbC2; they form extensive hydrophobic interactions. The residue Tyr³³⁰ of PcbA3 is capable of forming cation-π interaction with Arg¹³¹ of PcbC2. At the lumenal side, residues Ala²⁸² and Leu²⁸⁴ in the EF loop of PcbA3 form hydrophobic interactions with Phe¹⁰⁸ in helix B and Val⁶⁵ in helix A of PcbC2, respectively. A number of aromatic residues are distributed between CD loop, helix E and helix F of PcbA3, and helix B of PcbC2, forming close hydrophobic interactions between them (Fig. 6C). Thus, the four Pcb antennas mainly form hydrophobic interactions among them, which contributes to their compact arrangement and enhances the stability of the PSII-Pcb dimer structure in A. marina.

Interactions between PcbA2 of one dimer and PcbC2 from another dimer, along with the interactions among small subunits of the PSII core, promote formation of the PSII-Pcb tetramer and ensure its stability. At both the stromal and lumenal sides, PcbA2 and PcbC2 interact closely at the interface between adjacent PSII-Pcb dimers. At the stromal side, the aromatic residues in helix B of PcbA2 and Chl d513_PcbA2 interact with the nonpolar residues in DE loop of PcbC2, strengthening the connection between two subunits (Fig. 5G). The interface between helix B of PcbA2 and CD loop of PcbC2 also contains a number of aromatic and nonpolar residues, ensuring them to form broad and extensive hydrophobic interactions (Fig. 5H). The major interactions between the adjacent PSII core dimers are also found at the stromal side (Fig. 5, I and J). A variety of nonpolar residues, including Leu, Ile, Met, and Ala, are located at the interface between PsbE and Ycf12/PsbZ of the two adjacent PSII core dimers, and the hydrophobic interactions formed between them may stabilize the PSII-Pcb tetramer. At the lumenal side, residue PsbE-Leu⁴² of one PSII core dimer forms a hydrophobic interaction with Val³_PsbZ of the adjacent PSII core dimer. In summary, interactions between the Pcb antennas and the PSII core ensure the integrity and stability of the PSII-Pcb tetramer in A. marina.

Pigment arrangement and possible energy transfer pathways in the PSII-Pcb tetramer

As excitation energy transfer (EET) between Chls depends on not only their distances but also the relative orientation of the pigments, and the pigments are arranged mainly in two layers, the stromal and lumenal layers, we discuss the EET in the two layers based on a calculation of Förster resonance energy transfer (FRET) network and the center-to-center distance between Chls. For the FRET calculation, the energy transfer rate between Chl d molecules with Mg-Mg distance less than 50 Å in the PSII-Pcb dimer and tetramer complex was estimated (Fig. 7). Analysis of the FRET network in the PSII-Pcb tetramer of A. marina reveals a number of possible EET pathways for the far-red light capturing and transfer from the peripheral Pcb antennas to the PSII RC (Fig. 8). The EET between the Pcb antennas and the PSII core and between the adjacent PSIIs is mainly achieved by pigments from PcbA2, PcbA3, and PcbC2, as PcbA6 does not interact directly with the PSII core (Fig. 8A).

Fig. 7. Structure-based calculations of FRET networks within the PSII-Pcb dimer and tetramer complex from A. marina.

(A and B) Arrangement of Chl molecules and the Förster resonance energy transfer (FRET) networks enabling efficient FRET processes with lifetimes of less than 10 ps within the PSII-Pcb dimer (A) and tetramer (B), respectively. (C and D) Organization of Chl molecules and the FRET networks enabling efficient FRET processes with lifetimes of less than 100 ps within the PSII-Pcb dimer (C) and tetramer (D), respectively. The network of Chl molecules in the PSII-Pcb dimer and tetramer complex are viewed from the stromal side. Chl molecules are shown as blue spheres. FRET processes between adjacent Chls are shown as lines, and the relative FRET rates are indicated with the line width.

Fig. 8. Pigment distribution and possible EET pathways in the PSII-Pcb tetramer of A. marina.

(A) Pigment arrangement and possible EET pathways of the entire PSII-Pcb tetramer viewed perpendicular to the membrane plane. Chl arrangement at the stromal side is shown in the top side, and the Chl arrangement at the lumenal side is shown in the bottom side. The colors of Chl d are the same with the colors of corresponding subunits. The EET pathways from the peripheral Pcb antennas to RC and among the Pcb antennas are indicated by yellow arrows. (B to D) The interfacial Chls between Pcb antennas and the PSII core or Pcb antenna. The pigment interface between PcbA2 and CP43 (B); between CP47, PcbA3, and PcbC2 (C); and between PcbA2 and PcbC2 (D). Only pigments located in the EET pathways with close distance are shown in the figure, and other pigments are represented by transparent sticks.

At the interface between PcbA2 and CP43 (Fig. 8B), the newly discovered Chl d molecule (Chl d517) of CP43 is located in the middle of the lipid layer, and the Mg-to-Mg distance (D_center) from Chl d516_PcbA2 and Chl d517_PcbA2 to Chl d517_CP43 is 17.28 and 13.57 Å, respectively. At the lumenal side, there is a pair of parallel Chls, Chl d504_PcbA2 and Chl d518_PcbA2, which are absent in the other Pcb antennas. The D_center between these Chl pairs and Chl d503_CP43 is 25.30 and 23.27 Å, respectively. At the stromal side, D_center between Chl d511_PcbA2 and Chl d513_CP43 is 22.24 Å. Among these pigment pairs, the Chl d517_PcbA2 and Chl d517_CP43 have the highest FRET rate with a k_FRET of 1.216 ps⁻¹ (table S3). Thus, the energy is mainly transferred from PcbA2 to the RC by the newly found Chl d517 of CP43. Notably, the FRET rate analysis revealed an efficient EET between Chl d504_PcbA2 and Chl d518_PcbA2, which are parallel to each other with a plane-to-plane distance of around 4 Å and a k_FRET of 16.835 ps⁻¹ (Fig. 8A and table S3).

PcbA3 and PcbC2 interact with CP47, and the pigments located between their interfaces may promote energy transfer (Fig. 8C). At the stromal side, D_center from Chl d508_PcbC2 and Chl d511_PcbC2 to Chl d610_CP47 is 21.66 and 21.13 Å, respectively. Chl d602_CP47 forms a pigment cluster with Chl d517_PcbA3 and Chl d502_PcbC2 at the lumenal side, and D_center from Chl d517_PcbA3 and Chl d502_PcbC2 to Chl d602_CP47 is 18.32 Å and 15.60 Å, respectively. The pigment pairs between Chl d508_PcbC2 and Chl d610_CP47 and between Chl d517_PcbA3 and Chl d602_CP47 have a k_FRET of 0.100 and 0.135 ps⁻¹, respectively (table S3). Thus, both Chl d602 and Chl d610 in CP47 may mediate an efficient EET from the PcbA3/PcbC2 antenna to the PSII core. The energy transfer pathways between PcbA2 and PcbA6 and between PcbA6 and PcbA3 are similar, as the internal pigment arrangements in PcbA2, PcbA6, and PcbA3 are nearly identical (Fig. 9), although PcbA2, PcbA6, and PcbA3 exhibit minor differences in their protein structures. The k_FRET between Chl d505_PcbA2 and Chl d513_PcbA6 with a D_center of 13.76 Å at the stromal side and that between Chl d517_PcbA2 and Chl d504_PcbA6 with a D_center of 14.44 Å at the lumenal side are 0.573 and 0.310 ps⁻¹, which suggest that these two pigment pairs may contribute to the efficient EET between PcbA2 and PcbA6 (Figs. 8A and 9 and table S3). In addition, a pigment cluster is formed between Chl d504_PcbA6, Chl d517_PcbA2, and Chl d517_CP43 at the lumenal side, which further enhance the efficient capture and transfer of energy from the Pcb antennas to the PSII RC via the PcbA2 antenna. On the other hand, EET between PcbA3 and PcbC2 is different from those between PcbA2 and PcbA6 and between PcbA6 and PcbA3 due to the different pigment arrangement in PcbC2. At the stromal side, the pigment pair between Chl d505_PcbA3 and Chl d513_PcbC2 with a D_center of 13.34 Å has a larger k_FRET compared to that between Chl d519_PcbA3 and d501_PcbC2 with a D_center of 23.73 Å at the lumenal side, indicating that Chl d505_PcbA3 and Chl d513_PcbC2 may mediate the efficient EET between PcbA3 and PcbC2 (Figs. 8A and 9 and table S3). Thus, the excitation energy transfers efficiently among the four Pcb antennas and ultimately to the PSII RC through the PcbA2, PcbA3, and PcbC2 antennas.

Fig. 9. Pigment arrangement of different Pcb antennas in the PSII-Pcb megacomplex of A. marina.

The colors of PcbA2, PcbA6, PcbA3, and PcbC2 are the same as those in Fig. 2. Only pigments in the EET pathways with close distance are shown in the figure, and other pigments are represented by transparent sticks.

In the PSII-Pcb tetramer, two PSII-Pcb dimers are in close contact, and the association between adjacent PSII-Pcb dimers is mainly mediated by PcbA2 and PcbC2. EET between two PSII-Pcb dimers is achieved by these two antennas, rather than the PSII cores of these dimers. D_center between Chl d513_PcbA2 and Chl d506_PcbC2 at the stromal side is 18.28 Å and the k_FRET between these two Chls is 0.250 ps⁻¹, whereas D_center between Chl d518_PcbA2 and Chl d518_PcbC2 at the lumenal side is 24.13 Å and the k_FRET between them is 0.008 ps⁻¹ (Fig. 8D and table S3). Therefore, EET between adjacent PSII-Pcb dimers may be mediated mainly by the Chl d513 of PcbA2 and Chl d506 of PcbC2.

DISCUSSION

A. marina is a unique cyanobacterium able to synthesize Chl d as its major pigment and, thus, can use far-red light efficiently. The PSII-Pcb tetramer structure solved in this study provides a near-atomic resolution structure of a giant PSII-Pcb megacomplex. This structure consists of two PSII core dimers flanked by eight symmetric, Chl d–binding Pcb proteins on each side, which shows a unique supramolecular organization. This complex has a total molecular weight of ~1.9 MDa, representing the largest PSII protein supercomplex resolved from cyanobacteria, algae, and higher plants so far. The existence of such a large PSII-Pcb supercomplex has been reported previously but at lower resolutions, so the exact numbers and identities of subunits were not identified (42). The current high-resolution cryo-EM density map allows identification of the PSII core subunits and Pcb proteins unambiguously. It should be noted that A. marina lives in marine environments with a low visible light intensity but high near-infrared intensity (16, 47). The cells used in this study are cultured in a liquid medium under relatively low white light (20 μmol photons m⁻² s⁻¹) before isolation of the PSII-Pcb megacomplex. The PSII-Pcb tetramer structure revealed in this study shows some structural features distinct from that of other organisms where Chl a and Chl b are the dominant pigments, and the Pcb proteins function as peripheric antenna for PSII. It was already shown that Pcb proteins form an 18-subunit light-harvesting antenna ring around the PSI trimeric RC complex in the prochlorophyte Prochlorococcus (51), and functional association of 8 or 10 Pcb proteins with the PSII dimeric RC was shown for Prochlorococcus (51) and Prochloron (35), respectively.

In the PSII core of A. marina, 15 protein subunits were identified; they are similar to those of well-known Chl a–containing cyanobacteria (5, 46). A. marina has three psbA and three psbD genes encoding the D1and D2 subunits, respectively (47, 48). We have successfully identified the isoforms of these subunits in the structure based on the current cryo-EM map. D1 is modeled as a product of psbA2, where six–amino acid residues D1-Asp¹⁷⁰, D1-Glu¹⁸⁹, D1-His³³², D1-Glu³³³, D1-Asp³⁴², and D1-Ala³⁴⁴ involved in binding of the Mn₄CaO₅ cluster in other oxygenic cyanobacteria are conserved (fig. S3A). Together with the conserved CP43-Glu³⁵⁴ in the subunit of CP43, all the seven–amino acid residues critical for the binding of the Mn₄CaO₅ cluster are conserved in A. marina. This is consistent with the previous functional studies showing that the water-oxidizing complex of A. marina closely resembles that of Chl a–containing cyanobacteria and higher plants (52–55), indicating the conservation of the water-oxidizing complex in different oxygenic organisms.

The primary donor of PSII is modeled as Chl d in A. marina, although we could not distinguish Chl d from Chl a from the cryo-EM map. This is consistent with previous reports (28–31) and also consistent with the extremely low Chl a content (91 Chls d per Chl a) detected from the HPLC analysis, which equals to nearly one Chl a per two Pheo a (0.996 Chls a per two Pheo a) molecules (Fig. 1F). Assuming one Chl a per two Pheo a molecules in the PSII-Pcb structure, the ratio of Chl d to Chl a (105 Chls d per Chl a) observed in the PSII-Pcb structure is higher than that obtained from HPLC analysis. This could be due to presence of partial PSII-Pcb complexes missing one or two Pcb subunits in the sample or PSI contamination. Because of the limited resolution as well as the small difference between the structures of Chl a and Chl d, however, our current data cannot resolve the position and function of the trace amount of Chl a unambiguously. This Chl a may be existed in the RC as that has been suggested by other studies (30) or be associated with Pcb proteins and is not an essential component of the PSII RC (23). There is also a possibility that this Chl a may be located at the interface between Pcb subunits and PSII core like the previously unidentified Chls found in the current structure.

Unexpectedly, some remarkable changes are found in the amino acid residues surrounding P_D2 and Chl_D2 of the A. marina PSII in comparison with that of T. vulcanus PSII (fig. S5). These amino acid residues may be involved in redox tuning of the P_D1/P_D2 and Chl_D1/Chl _D2 pairs, resulting in the alteration of the coupling between P_D1 and P_D2 as well as Chl_D1 and Chl _D2, although it is not clear how such changes would influence the redox potentials of P_D2 or Chl_D2. This could reflect the adjustment of the energetics for an efficient photochemistry reaction to use far-red light with Chl d as the photoactive pigment in A. marina PSII. Thus, the electron transfer energetics and energy trapping efficiency in A. marina PSII may be somewhat different from the other organisms (56), enabling it to carry oxygenic photosynthesis under far-red light. In addition, structural comparison showed that the amino acid residues surrounding the formyl group of Chl d in P_D1, Chl_D1, and Chl_D2 in A. marina PSII have been changed compared to that in T. vulcanus PSII and are very similar with that of FaRLiP PSII from Synechococcus 7335 (fig. S8) (57). Notably, the formyl group of Chl d in Chl_D1 of A. marina PSII forms hydrogen bond interaction with the surrounding Tyr rather than Thr as that observed in FaRLiP PSII (fig. S8) (57).

On the other hand, Pheo a is identified as the primary electron acceptor in A. marina, which is consistent with the previous spectroscopic studies (28–30, 52). Furthermore, the amino acid residues surrounding Pheo_D1/Pheo_D2 and Q_A are completely conserved with those of T. vulcanus (5). This indicates the conservation of the accepter side in A. marina PSII, although a positive shift of the redox potential value of 124 mV for Em (Pheo a/Pheo a⁻) and 59 to 66 mV for Em (Q_A/Q_A⁻) relative to that of Synechocystis 6803 was observed in previous photochemical redox titration studies (55, 58).

One of the distinct features of the PSII-Pcb tetramer is the binding of transmembrane Pcb proteins as the major light-harvesting protein. The Chl d–binding Pcb protein is a unique LHC protein found in A. marina, which may be similar to the Pcb proteins found in Chl b containing cyanobacteria like Prochlorothrix and Acaryochloris thomasi (RCC1774). This suggests that these cyanobacteria may use a similar strategy to adapt to their light environments (50, 59). Each PSII-Pcb monomer contains four Pcb proteins acting as the light-harvesting system, which increases the overall antenna cross section to enhance the light-harvesting capacity of PSII. The well-defined cryo-EM map allowed us to identify the isoform types of all Pcb subunits in the whole structure, which shows the binding of both PcbA- and PcbC-type Pcb antennas. This is consistent with previous studies that both PcbA and PcbC may be associated with the PSII core (42), despite the report that PcbC may form a complex with PSI under iron stress conditions (60). Intriguingly, the IsiA protein, A8ZMK2, supposed to be induced under iron limitation and form a complex with PSI, is not observed in the PSII-Pcb structure (43). All these Pcb subunits contain six transmembrane helices, and both of their N-terminal and C-terminal regions are located on the stromal side (Fig. 4E). This is different from the typical structures of LHC protein found in green algae and higher plants, which has three transmembrane helices (11–13).

Detailed comparison shows some obvious differences between the PcbA-type and PcbC-type antennas in both the structure and pigment arrangement. PcbA antennas have similar structure and pigment arrangement as the IsiA, whereas the PcbC antenna exhibits some different structures and binding sites of pigment. This is consistent with previous phylogenetic analysis (50). In addition, the A. marina Pcb antennas contain a number of Zea similar to that of the Lhcr antennas in red algae PSI but different from the LHC antennas from other organisms (61), suggesting that the Pcb antennas may play an important role in photoprotection in A. marina. Overall, the existence of different types of Pcb proteins and their specific structural features suggest that Pcb antennas are important and essential for the supramolecular organization of PSII and photoacclimation of A. marina to shade habitats and far red light-enriched conditions.

Because A. marina has a unique Pcb arrangement unlike green algae and higher plants, the assembly between Pcb antennas and PSII core are different from the structures of PSII-LHCII seen in green algae and higher plants. In the PSII-Pcb structure, all the Pcb antennas exist as monomers, so the monomeric Pcb antennas are directly bound to the PSII core through the intrinsic light-harvesting systems CP43 and CP47. A previously unidentified Chl molecule was found in each of the CP43 and CP47 subunits, which likely facilitates the connection between the PSII core and Pcb antennas and enables the efficient energy transfer between Pcb and PSII core. An unidentified protein was found at the antenna interface of the PSII-Pcb dimer, which acts like a hook to connect the Pcb antennas of two PSII monomers and ensures the stability of the dimer within a PSII-Pcb tetramer. Furthermore, strong interactions are found between the PSII core and Pcb antenna subunits of two PSII dimers, supporting the binding of two PSII dimers in the PSII-Pcb tetramer structure.

Intriguingly, during purification of the PSII-Pcb megacomplex, negative-staining electron microscope images indicated the presence of PSII-Pcb polymers, which was composed of multiple PSII-Pcb dimers connected to each other mainly by the interactions of Pcb antennas (fig. S9). This indicates that the PSII-Pcb supercomplexes likely form neatly arranged regions in the thylakoid membrane. The PSII-Pcb tetramer megacomplexes reported in the present study are purified from a minor band in the SDG, which may reflect a partial PSII-Pcb complex of larger megacomplexes present in vivo. There may be more Pcb antennas bound loosely at the outside of the present PSII-Pcb megacomplex in vivo, and these antenna subunits may have been lost by solubilization with the detergent during purification. It is also possible that PSII may form larger complexes consisting of multiple PSII-Pcb tetramers in vivo, as observed in fig. S9. The presence of such a condensed PSII region in vivo has been reported in the previous studies (34, 42). In addition, we do not know whether the tetrameric organization of PSII has any functional advantages over dimeric PSII reported for many other organisms. The tetrameric PSII and neatly arranged array of PSII in thylakoid membranes may improve the efficiency of light energy capture and energy distribution supporting Chl d–PSII functions, and these condensed arrays may also improve the PSII stabilization in vivo. Further studies are needed to elucidate the detailed organization of PSII complex and their association with Chl-binding light-harvesting protein complexes in the thylakoid membranes.

In summary, the current PSII-Pcb tetramer structure of A. marina provides important insights into the PSII function of far-red light utilization compared to the previous low-resolution structure. The Pcb protein composition and its molecular arrangement as well as the changes in the protein configuration and amino acid residues surrounding some of the cofactors involved in electron transfer are revealed. Different from the PSII-LHCII supercomplex of higher plants and green algae, the PSII-Pcb tetramer has a unique supramolecular organization and Chl d pigment network, which results in a larger antenna size and pathways of EET totally different from those of higher plants and green algae with Chl a as the main pigment. This may be crucial for A. marina to maintain efficient energy utilization and to avoid photoinhibition in a specific light environment where far-red light dominates.

MATERIALS AND METHODS

Purification of the PSII-Pcb megacomplex

A. marina MBIC11017 were cultured in a 10-liter flask bubbled with 3% (v/v) CO₂ at 28°C under continuous illumination with white fluorescent lamp at a low light intensity (20 μmol photons m⁻² s⁻¹) as previously described by Chen et al. (23). Cells were harvested by centrifugation (5000g for 10 min, 25°C). The PSII-Pcb tetramer megacomplex was prepared under dim green light at 4°C according to Chen et al. (42) with modifications. For thylakoid membrane preparation, the harvested cells were suspended in AM Buffer-1 [15% glycerol, 50 mM MES-NaOH (pH 6.5), 10 mM CaCl₂, and 5 mM MgCl₂] containing 1 mM benzamidine, 6-aminocaproic acid, and phenylmethylsulfonyl fluoride and broken by a pressure disruptor (AH-D150) at 1100 bar for two times. Then, the suspension of crushed cell was centrifuged at 2000g for 10 min to remove unbroken cells and large fragments of cell debris. The supernatant was centrifuged at 40,000g for 1 hour, and the precipitate was washed twice with AM Buffer-1 to remove phycobiliproteins and other soluble proteins. The precipitated thylakoid membrane was suspended in AM Buffer-2 [500 mM betaine, 50 mM MES-NaOH (pH 6.5), 10 mM CaCl₂, and 5 mM MgCl₂] and stored at −80°C before use.

For isolation of the PSII-Pcb megacomplexes, thylakoid membranes at a concentration of 1 mg Chl d ml⁻¹ were solubilized with n-dodecyl-β-d-maltoside (β-DDM) by adding 20% (w/v) β-DDM to a final concentration of 0.9% (w/v) on ice for 10 min. The solubilized membranes were centrifuged at 15,000g for 10 min to remove unsolubilized thylakoid membranes. The supernatant was slowly loaded into a continuous SDG prepared by freeze-thawing of a 0.7 M sucrose solution containing 50 mM MES-NaOH (pH 6.5), 10 mM CaCl₂, 5 mM MgCl₂, and 0.012% w/v β-DDM in the centrifuge tubes and centrifuged at 29,000 rpm (SW40Ti rotor) for 21 hours at 4°C with a Beckman Coulter Optima XPN-100 centrifuge. A narrowband of the target protein under the PSI trimer band of the SDG was carefully collected using a syringe and placed on ice. Then, the collected sample was concentrated to a concentration of 1 mg of Chl d ml⁻¹ using a 100-kDa cutoff membrane concentrator. For size exclusion chromatographic elution of the PSII-Pcb fraction, the sample isolated by SDG was injected into a gel filtration column (GE, Superose 6 Increase 10/300 GL) in a buffer containing 50 mM MES-NaOH (pH 6.5), 10 mM CaCl₂, 5 mM MgCl₂, 150 mM NaCl, 0.012% w/v β-DDM, and 0.1% digitonin and eluted. The peak fractions were collected for further biochemical and cryo-EM analysis. The concentration of Chl d was determined according to the method described in the previous analysis (62).

Characterization of the PSII-Pcb megacomplex

Protein composition of the purified PSII-Pcb megacomplex was analyzed by electrophoresis using a gel containing 16% polyacrylamide and 7.5 M urea (63). The gel was stained with Coomassie brilliant blue (CBB) R-250. For MS measurement, CBB-stained bands were cut out from the gel and digested using sequencing-grade modified trypsin, and the resultant peptides were extracted and separated by an analytical column, with a 60-min gradient elution at a flow rate of 0.30 μl min⁻¹ with the EASY-nLC 1000 system, and were directly interfaced with the Thermo Orbitrap Fusion mass spectrometer. The analytical column was a homemade, fused silica capillary column (75-μm inside diameter, 150-mm length; Upchurch, Oak Harbor, WA) packed with C18 resin (300 Å, 5 μm; Varian, Lexington, MA). The tandem MS (MS/MS) spectra from each LC-MS/MS run were searched against the selected database using Proteome Discovery searching algorithm (version 1.4).

Ultraviolet (UV) absorption spectrum was measured by a UV-visible spectrophotometer (UV-2700, Shimadzu, Japan) at room temperature, with sampling interval of 0.1 nm and scanning range of 250 nm to 800 nm. Fluorescence emission spectra were measured at 77 K with a fluorescence spectrometer (F-4500, Hitachi, Japan) equipped with a xenon lamp source, and the spectra were recorded at a wavelength range from 600 to 800 nm with the excitation wavelength of 398 nm. The slit widths of both excitation and emission were set at 5.0 nm.

Pigment composition of the PSII-Pcb megacomplex was analyzed by HPLC as described previously (43). The sample after size exclusion chromatography was mixed with 100% cold methanol to extract pigments, and the extract was centrifuged at 14,000g for 10 min under dim light. The supernatant was injected into a C18 reversed-phase column (C18 5u, Alltima, UK) in a Waters e2695 separation module equipped with a Waters 2998 photodiode array detector. The mobile phase was 100% methanol, and the detection wavelength was 400 nm. Pigments were identified on the basis of their absorption spectra and elution times.

Cryo-EM sample preparation and data collection

A 2.5-μl purified PSII-Pcb was applied to holey carbon grids covered with a graphene oxide film (Quantifoil R1.2/1.3, Au, 300 mesh), waited for 60 s, blotted for 3.0 s at a humidity of 100% and 4°C, and then rapidly plunged into pre-cold liquid ethane with a Vitrobot Mark IV (Thermo Fisher Scientific). All cryo-EM data were collected on a Krios microscope (Thermo Fisher Scientific) operated at 300 kV, equipped with Gatan energy filter (Gatan BioQuantum 1967) and k3 summit direct detector (Gatan) set to a slit width of 20 eV. The microscope was carefully aligned before data acquisition, including the coma-free alignment to minimize the beam tilt. Automated data acquisition was carried out with EPU software (Thermo Fisher Scientific) through faster acquisition mode with Aberration-free imaging shift alignment. Movies were taken in the super-resolution mode at a nominal magnification ×81,000, corresponding to a physical pixel size of 1.07 Å, and a defocus range from −1.0 μm to −2.5 μm. Each movie stack was dose-fractionated to 32 frames with a total exposure dose of 50 e⁻/Ų and exposure time of 2.86 s.

Cryo-EM data processing

A total of 11,708 movies were binned 2 × 2, dose-weighted, and motion-corrected using MotionCor2 (64). Parameters of contrast transfer function (CTF) were estimated by Gctf (65). Around 3000 particles were manually picked from a subset of images using EMAN2 (66), which were subjected to 2D classification using Relion v4.0 (67). Averaged images of good classes were selected and served as picking-template using Gautomatch (developed by K. Zhang; www2.mrc-lmb.cam.ac.uk/download/gautomatch-053/). These averaged images were combined to make an initial model with EMAN2. After template picking, all data processing was carried out using Relion v4.0 (67).

Using the template-pick algorithm, 1,146,757 particles were picked and subjected to two rounds of 2D classification, from which 824,322 particles were selected from good classes and subsequently subjected to two rounds of 3D classification. Subsequently, 132,346 particles were selected and subjected to 3D refinement. After 3D auto-refinement, CTF refinement, particles were lastly reconstructed using a C2 symmetry, which resulted in a density map with an overall resolution of 3.6 Å. Because another C2 symmetry axis was found inside the asymmetric unit, we further imposed a C2 symmetry to the asymmetric unit to improve the resolution. After imposing the internal symmetry, the resolution of an asymmetric unit was improved to 3.3 Å based on Fourier shell correlation coefficient of 0.143 (68).

Model building and refinement

During construction of the atomic model of the PSII-Pcb tetramer, the PSII core structure of T. vulcanus (PDB code 3WU2) was initially docked into the density map using UCSF Chimera (69), and the amino acid sequences of each chain of the T. vulcanus PSII core were manually mutated to the corresponding sequences of each chain from A. marina in Coot (70). The individual protein chains and amino acid residues as well as cofactors of the initial model of the PSII-Pcb were further adjusted and processed manually on the basis of the cryo-EM map with Coot. The Pcb antennas were mainly modified based on the structure of IsiA (PDB code 6KIG) (40). The IsiA protein is docked into the density map and then adjusted to conform to the amino acid sequences of Pcb from A. marina. The type of Pcb antenna is identified according to the amino acid residues of each subunit in the density map. An unknown protein located between PcbA6 and PcbA3 is modeled with poly-alanines. Subsequently, the structure of the PSII-Pcb megacomplex was refined in real space against the cryo-EM map by Phenix (71) with geometry and secondary structure constraints. The statistics for data collection and structure refinement are summarized in table S1. All the figures in this article were generated using UCSF ChimeraX (72).

Sequence alignment, phylogenetic analysis, and the BLAST analysis

Sequence alignment and phylogenetic analysis were performed with MEGA X (73). Evolutionary relationships were analyzed using the neighbor-joining method with 1000 bootstrap replication (74). The evolutionary distances were computed using the p-distance method and are expressed in the units of the number of base differences per site. All ambiguous positions were removed for each sequence pair (pairwise deletion option). Evolutionary analyses were conducted in iTOL (75). The BLAST analysis was carried out with a query sequence of A. marina CP43 against the genome of A. marina in the UniProt database. Nine CP43 like sequences (PcbA1, PcbA2, PcbA3, PcbA4, PcbA5, PcbA6, PcbC1, PcbC2, and IsiA) with corresponding UniProt numbers (B0C2V8, B0C576, B0C6I0, B0C009, B0C012, B0C3E5, B0C011, B0C3E6, and A8ZMK2) and corresponding NCBI accession numbers (ABW26174.1, ABW26316.1, ABW26401.1, ABW28356.1, ABW28359.1, ABW28644.1, ABW28358.1, ABW28645.1, and ABW32413.1) were found and used for construction of the phylogenetic tree.

Network analysis of FRET rates

On the basis of the FRET theory (76), FRET rate constants of Chl pairs were calculated with the definition k_FRET = Cκ²/(n⁴r⁶), where C is a factor derived from the spectral overlap integral between the fluorescence and absorption spectra of donor and acceptor Chls, κ is the dipole orientational factor, n is the refractive index, and r is the distance between donor and acceptor dipoles. The C value of 32.26 was applied for Chl d→Chl d interaction as in a previous study (13, 77), and the n value of 1.55 was used (78). In our calculation, the dipole orientational factor is defined as $\mathrm{\kappa }=3({\widehat{\mathbf{u}}}_{\mathrm{A}}·\widehat{\mathbf{r}})({\widehat{\mathbf{u}}}_{\mathrm{D}}·\widehat{\mathbf{r}})-{\widehat{\mathbf{u}}}_{\mathrm{A}}·{\widehat{\mathbf{u}}}_{\mathrm{D}}$, where ${\widehat{\mathbf{u}}}_{\mathrm{A}}$ and ${\widehat{\mathbf{u}}}_{\mathrm{D}}$ are the dipole unit vectors of acceptor and donor Chls, derived from the coordinates of ND and NB atoms, respectively. The distance r was calculated from the coordinates of two magnesium atoms in the Chl pairs, and $\widehat{\mathbf{r}}$ is the unit vector that marks the direction between them. FRET rate constants were calculated using our in-house Bash script, and the network analysis was performed and visualized using the NetworkX 3.1 and Matplotlib 3.7.2 modules implemented in Python 3.10.

Supplementary Materials

This PDF file includes:

Figs. S1 to S9

Tables S1 to S3

Legend for source data 1

Other Supplementary Material for this manuscript includes the following:

Source data 1