The structure of PSI-LHCI from Cyanidium caldarium provides evolutionary insights into conservation and diversity of red-lineage LHCs

Significance

Photosynthetic red-lineage organisms originate from red algae, which represent an evolutionary intermediate between cyanobacteria and other red-lineage eukaryotic algae. Numerous phylogenetic studies have shown the conservation and diversity of LHCs (light-harvesting complexes) in the red lineage, and structural studies have shown differences in the number, sequences, and pigment compositions of LHCs in the PSI-LHCI (photosystem I-LHC) supercomplexes. However, a correlation between the structural and phylogenetic findings of red-lineage LHCs is unclear. In this study, we solved a structure of the PSI-LHCI supercomplex from the red alga Cyanidium caldarium RK-1 (NIES-2137) by cryoelectron microscopy and revealed the orthologous relationships of LHCs by phylogenetic analysis. Our study opens a broad avenue for investigating the evolution of the red-lineage LHCs.

Abstract

Light-harvesting complexes (LHCs) are diversified among photosynthetic organisms, and the structure of the photosystem I-LHC (PSI-LHCI) supercomplex has been shown to be variable depending on the species of organisms. However, the structural and evolutionary correlations of red-lineage LHCs are unknown. Here, we determined a 1.92-Å resolution cryoelectron microscopic structure of a PSI-LHCI supercomplex isolated from the red alga Cyanidium caldarium RK-1 (NIES-2137), which is an important taxon in the Cyanidiophyceae. We subsequently investigated the correlations of PSI-LHCIs from different organisms through structural comparisons and phylogenetic analysis. The PSI-LHCI structure obtained shows five LHCI subunits surrounding a PSI-monomer core. The five LHCIs are composed of two Lhcr1s, two Lhcr2s, and one Lhcr3. Phylogenetic analysis of LHCs bound to PSI in the red-lineage algae showed clear orthology of LHCs between C. caldarium and Cyanidioschyzon merolae, whereas no orthologous relationships were found between C. caldarium Lhcr1–3 and LHCs in other red-lineage PSI-LHCI structures. These findings provide evolutionary insights into conservation and diversity of red-lineage LHCs associated with PSI.

Oxygenic photosynthesis in cyanobacteria, algae, and land plants converts solar energy into chemical energy concomitant with the evolution of oxygen molecules (1). The light-energy conversion takes place in photosystem I and photosystem II (PSI and PSII, respectively), two multisubunit pigment–protein complexes that perform light harvesting, charge separation, and electron transfer reactions (1). In addition to PSI and PSII, light-harvesting complexes (LHCs) participate in the acquisition of sunlight and transfer of excitation energy to the photosystem cores. PSI and PSII are thus associated with LHCI and LHCII, respectively, to form PSI-LHCI and PSII-LHCII supercomplexes (1).

LHCs are highly diversified among photosynthetic organisms in terms of the protein sequences and pigment compositions of chlorophylls (Chls) and carotenoids (Cars) (1–3). The differences of LHCs cause color variations in photosynthetic organisms, which can be classified into green and red lineages (4). The green lineage includes green algae and land plants, whereas the red lineage includes red algae, diatoms, haptophytes, cryptophytes, and dinoflagellates (4). The structures of LHCs and their association patterns with the photosystem cores have been revealed by structural studies, especially using cryoelectron microscopy (cryo-EM) (3, 5). In the red lineage, the number, sequences, and pigment compositions of LHCIs have been found to differ greatly among the PSI-LHCI structures of red algae (6–8), a diatom (9, 10), and a cryptophyte (11).

Red algae are grouped into a distinctive photosynthetic lineage including unicellular and large multicellular taxa (12) and represent an evolutionary intermediate between cyanobacteria and red-lineage algae (4). Two types of PSI-LHCI structures from different red algae have been reported. The red alga Cyanidioschyzon merolae belongs to the Cyanidiophyceae (Cyanidiophytina) (13–16), and its PSI-LHCI structure contained three to five LHCI subunits (6, 7). In contrast, the red alga Porphyridium purpureum belongs to the Porphyridiophyceae (16, 17), and its PSI-LHCI structure contained seven LHCI subunits and one red lineage chlorophyll a/b-binding-like protein (RedCAP) (8) which is included in the LHC protein superfamily (18, 19). Thus, the number and binding sites of LHCIs in PSI-LHCI differ remarkably between the two types of red algae.

Cyanidium caldarium is a unicellular red alga belonging to the Cyanidiophyceae (13–16) and lives in thermo-acidic environments (20). Isolation and characterization of C. caldarium PSI-LHCI supercomplexes have been reported (21, 22). Gardian et al. suggested that the C. caldarium PSI-LHCI has 0 to 8 LHCIs per monomeric PSI core in response to growth-light conditions (21). However, the overall structure of the C. caldarium PSI-LHCI is still unknown.

In this study, we solved a 1.92-Å resolution structure of a PSI-LHCI supercomplex purified from C. caldarium RK-1 (NIES-2137) by cryo-EM single-particle analysis. The structure shows a PSI-monomer core and five LHCI subunits. The arrangement of LHCIs in the C. caldarium PSI-LHCI structure was very similar to that of the C. merolae PSI-LHCI structure. Based on the structural comparisons together with phylogenetic analysis of the red-lineage LHCs bound to PSI, we discuss the molecular evolution of LHCs from red algae to diatoms and cryptophytes.

Results and Discussion

Overall Structure of the C. caldarium PSI-LHCI Supercomplex.

The PSI-LHCI supercomplexes were purified from C. caldarium RK-1 (NIES-2137) and characterized in our previous study (22). Cryo-EM images of the PSI-LHCI supercomplex were obtained by a JEOL CRYO ARM 300 electron microscope operated at 300 kV. The final cryo-EM map was determined with a C1 symmetry at a resolution of 1.92 Å (SI Appendix, Figs. S1 and S2 and Table S1), based on the “gold standard” Fourier shell correlation (FSC) = 0.143 criterion (SI Appendix, Fig. S2A).

The atomic model of PSI-LHCI was built based on the resultant cryo-EM map (Materials and Methods and SI Appendix, Fig. S2 and Tables S1–S3). The structure reveals a monomeric PSI core associated with five LHCI subunits (Fig. 1 A and B). Three of the five LHCI subunits are located near PsaA, PsaJ, and PsaF, whereas the remaining two subunits are positioned near PsaB, PsaI, PsaL, and PsaM (Fig. 1B). The five LHCIs were named LHCI-1 to 5 (Fig. 1A). The PSI core contains 97 Chls a, 20 β-carotenes (BCRs), 2 zeaxanthins (ZXTs), 3 [4Fe-4S] clusters, 2 phylloquinones, and 4 lipid molecules, whereas the LHCI subunits contain 57 Chls a and 21 ZXTs (SI Appendix, Table S3).

Fig. 1.

Overall structure of the PSI-LHCI supercomplex from C. caldarium RK-1 (NIES-2137). Structures are viewed from the stromal side (Left) and the direction perpendicular to the membrane normal (Right). Only protein structures are shown, and cofactors are omitted for clarity. The LHCI (A) and PSI-core (B) subunits were labeled and colored differently. (A) The five LHCI subunits were labeled as LHCI-1 to 5 (red) with their gene products indicated in parentheses (black).

Structure of the C. caldarium PSI Core.

The PSI-monomer core has 12 subunits, PsaA, PsaB, PsaC, PsaD, PsaE, PsaF, PsaI, PsaJ, PsaK, PsaL, PsaM, and PsaO (Fig. 1B), and their arrangements are similar to that of the C. merolae PSI-LHCI structures (6, 7). The sequences of PSI subunits were based on a draft genome data of C. caldarium RK-1 (NIES-2137) obtained in this study. Among the PSI genes, two types of PsaO, PsaO-1 and PsaO-2, were found; however, we cannot distinguish between the two PsaOs in the PSI-LHCI structure. This is because the structure of PsaO was modeled from Y64 to Y146 (SI Appendix, Fig. S3A), whose sequence is the same between PsaO-1 and PsaO-2 (SI Appendix, Fig. S3B).

The number and arrangement of Chls in the C. caldarium PSI-LHCI (SI Appendix, Fig. S3C) are virtually identical to those in the C. merolae PSI-LHCI (7). In contrast, BCR852 of PsaB in the C. caldarium PSI-LHCI (SI Appendix, Fig. S3D) was not found in the C. merolae PSI-LHCI (7). ZXT304 of PsaF and ZXT105 of PsaJ were identified according to our criterion (Materials and Methods and SI Appendix, Fig. S3E), and they correspond to BCR304 of PsaF and BCR105 of PsaJ, respectively, of the C. merolae PSI-LHCI (7).

ZXT has a hydroxyl group in each end ring, whereas BCR does not have any hydroxyl groups in either of the rings. In the C. caldarium PSI-LHCI structure, no characteristic interactions, such as hydrogen-bond interactions, are found around the end rings of PsaF-ZXT304 and PsaJ-ZXT105 (SI Appendix, Fig. S3 F and G). Because the protein and cofactor structures around PsaF-ZXT304 and PsaJ-ZXT105 in the C. caldarium PSI-LHCI are similar to those around PsaF-BCR304 and PsaJ-BCR105 in the C. merolae PSI-LHCI (7), it is suggested that the corresponding Cars are exchangeable between ZXT and BCR. However, it should be noted that the two BCRs of the C. merolae PSI-LHCI may be ZXTs, because the C. merolae PSI-LHCI structures were reported at low resolutions (3.6 to 4.0 Å) (6, 7), which could lead to a mis-assignment of the Cars. Further improvement of the C. merolae PSI-LHCI structure is required for identification of the Car molecules of PsaF and PsaJ, in order to gain structural insights into an exchangeable mechanism of Cars in photosynthetic pigment–protein complexes.

Structure of the C. caldarium LHCIs.

The five LHCI subunits in the PSI-LHCI structure were identified using three Lhcr genes, so that LHCI-1 and LHCI-3 were Lhcr1, LHCI-2 and LHCI-4 were Lhcr2, and LHCI-5 was Lhcr3 (Fig. 1A). These arrangements correspond to those in the C. merolae PSI-LHCI (7). Two types for each gene of Lhcr1, Lhcr2, and Lhcr3 were found in the draft genome data of C. caldarium RK-1 (NIES-2137); they were named as Lhcr1-1/Lhcr1-2, Lhcr2-1/Lhcr2-2, and Lhcr3-1/Lhcr3-2, respectively. The structures of LHCI-1 and LHCI-3 were modeled from Q41 to P212 (SI Appendix, Fig. S4 A and B); in this region, the sequence of Lhcr1-1 is consistent with that of Lhcr1-2 (SI Appendix, Fig. S4C). The structures of LHCI-2 and LHCI-4 were modeled from Q34 to F216 (SI Appendix, Fig. S4 D and E). The 40th amino acid of Lhcr2-1 is alanine, whereas that of Lhcr2-2 is threonine (SI Appendix, Fig. S4F). Based on the cryo-EM density map, the 40th amino acid of LHCI-2 and LHCI-4 was provisionally modeled as alanine. The structure of LHCI-5 was modeled from A63 to V213 (SI Appendix, Fig. S4G); in this region, the sequence of Lhcr3-1 is consistent with that of Lhcr3-2 (SI Appendix, Fig. S4H).

For the assignments of each LHCI subunit, we focused on characteristic amino acid residues among the three Lhcrs. For LHCI-1 and LHCI-3, Lhcr1 was identified because its amino acid residues Y61 and R62 are different from the corresponding residues in Lhcr2 and Lhcr3 (SI Appendix, Fig. S5 A, B, and D). For LHCI-2 and LHCI-4, Lhcr2 was identified because its amino acid residues G68 and F69 are different from the corresponding residues in Lhcr1 and Lhcr3 (SI Appendix, Fig. S5 A, C, and E). For LHCI-5, Lhcr3 was identified because its amino acid residues Y73 and L74 are different from the corresponding residues in Lhcr1 and Lhcr2 (SI Appendix, Fig. S5 A and F).

LHCI-1 binds 9 Chls a and 4 ZXTs (Fig. 2A). The axial ligands of the central Mg atoms of Chls within LHCI-1 are mainly provided by main and side chains of amino acid residues (SI Appendix, Table S4). In contrast, LHCI-3 contains 11 Chls a and 5 ZXTs (Fig. 2C). The axial ligands of the LHCI-3 Chls are summarized in SI Appendix, Table S4. The number of Chls and ZXTs in LHCI-1 is less than those in LHCI-3, albeit with the same gene product between the two subunits. This may be due to weaker densities for Chls and ZXTs in LHCI-1 than those in LHCI-3, leading to the inability of pigment assignment according to our criterion (Materials and Methods). Alternatively, it may be possible that some Chls may be originally absent in LHCI-1. Because LHCI-1 and LHCI-3 are located near PsaB and PsaA, respectively (Fig. 1), the different pigment content between LHCI-1 and LHCI-3 may result in variations in the excitation-energy transfer pathways from LHCIs to PSI and/or among LHCIs.

Fig. 2.

Structure of LHCIs. (A–E) Structures of LHCI-1 to LHCI-5 depicted as transparent cartoon models and arrangements of Chls and ZXTs. Chls and ZXTs are colored green and orange, respectively. Only rings of the Chl molecules are depicted. (F and G) Comparison of Chls (F) and ZXTs (G) between C. caldarium (cyan) and C. merolae (orange). The C. caldarium PSI-LHCI structure was superimposed with the C. merolae PSI-LHCI structure (PDB: 5ZGB). The structures are viewed from the stromal side. Chls and ZXTs are shown as sticks. Only rings of the Chl molecules are depicted. Numbers 1 to 5 (red) indicate the sites of LHCI-1 to 5, respectively, in the C. caldarium PSI-LHCI structure.

LHCI-2 and LHCI-4 each bind 13 Chls a and 4 ZXTs (Fig. 2 B and D), whereas LHCI-5 binds 11 Chls a and 4 ZXTs (Fig. 2E). The Chl ligands of LHCI-2, LHCI-4, and LHCI-5 are mainly provided by amino acid residues (SI Appendix, Table S4). The RMSDs of the structures between LHCI-3 and the other four LHCIs range from 0.74 to 1.55 Å for 302 Cα atoms (SI Appendix, Table S5).

The arrangement of Chls and ZXTs in the LHCI subunits of C. caldarium PSI-LHCI is virtually identical to those in the C. merolae PSI-LHCI (7) (Fig. 2 F and G); however, their numbers are slightly fewer in the C. caldarium PSI-LHCI than those in the C. merolae PSI-LHCI (7) (Fig. 2 F and G). The pigment molecules of a604/a605/ZXT615/ZXT616 of LHCI-1, a605 of LHCI-3, and a202/ZXT216/ZXT218 of LHCI-5, were observed in the C. merolae PSI-LHCI structure but not in the C. caldarium PSI-LHCI structure (Fig. 2 F and G). These results imply different excitation-energy-transfer mechanisms of LHCIs between C. caldarium and C. merolae.

The differences in the pigment compositions between C. caldarium and C. merolae PSI-LHCI structures may be due to their species difference. Alternatively, it may originate from experimental problems such as the preparation conditions and structural analyses of PSI-LHCI supercomplexes between C. caldarium in our study and C. merolae in Pi et al. (7). The pigments may be lost during preparation in one species, or may be mis-assigned in one study due to limited resolution.

Structural Varieties of the C. caldarium PSI-LHCI Supercomplexes.

Gardian et al. suggested that the C. caldarium PSI-LHCI supercomplexes have 0 to 8 LHCIs based on their negative-stain electron microscopy single-particle analysis (21). Although the PSI-LHCI structures solved by these authors are at a much lower resolution, the binding property of LHCIs may differ between our study and Gardian et al. (21) for at least two reasons. The first one is growth conditions: We cultured C. caldarium RK-1 (NIES-2137) at a photosynthetic photon flux density of 30 µmol photons m⁻² s⁻¹ at 30 °C with bubbling of air containing 3% (v/v) CO₂, whereas Gardian et al. cultured C. caldarium under light irradiances of 20 or 200 μmol photon m⁻² s⁻¹ at 42 °C with air bubbling (21). The growth temperature and additional CO₂ may contribute to changes in the antenna sizes of PSI-LHCI. The second one is that the C. caldarium strain used may differ between the two studies. We used the RK-1 (NIES-2137) strain; however, Gardian et al. did not show the name of strain (21). To address these issues, it will be necessary to solve a variety of PSI-LHCI structures from C. caldarium grown under various conditions.

Structural Comparisons of the C. caldarium LHCIs with Those of Other Red Algae.

We compared binding sites of LHCIs in the PSI-LHCI structures between C. caldarium and C. merolae (Fig. 3A). The five LHCI subunits of LHCI-1 to LHCI-5 in the C. caldarium PSI-LHCI structure (red) are located at the same positions of Lhcr1*, Lhcr2*, Lhcr1, Lhcr2, and Lhcr3, respectively, in the C. merolae PSI-LHCI structure (cyan) (7) (Fig. 3A). The gene products of each LHCI subunit are shown in Fig. 3B, and multiple sequence alignments of the LHCI proteins located at the same positions in the PSI-LHCI structures of C. caldarium and C. merolae are shown in SI Appendix, Fig. S6 A–C. The C. caldarium Lhcr1–3 show high sequence similarities (89 to 91%) to the C. merolae Lhcr1–3, respectively. These results provide evidence for strong conservation of the binding sites of the individual LHCIs to PSI and their genes between C. caldarium and C. merolae. It should be pointed out that in C. merolae, two types of PSI-LHCI have been reported, namely, one has three LHCI subunits associated at one side of the PSI core, whereas the other one has five LHCI subunits (7). In C. caldarium, we only observed the five LHCIs-type PSI-LHCI but the three LHCIs-type particles were not observed. This may be due to a low-light intensity for the growth of the C. caldarium cells, or alternatively, only the five LHCIs-type PSI-LHCI exists in C. caldarium.

Fig. 3.

Structural comparisons of the C. caldarium LHCIs with those of other red algae. (A) Superposition of the C. caldarium PSI-LHCI structure with the C. merolae PSI-LHCI structure (PDB: 5ZGB). The LHCI subunits of C. caldarium and C. merolae were colored red and cyan, respectively. The structures are viewed from the stromal side. (B) Correlation of the names of LHCI subunits in the PSI-LHCI structures with their genes between C. caldarium and C. merolae. The names of LHCI subunits and their genes are derived from Pi et al. (7) for C. merolae. (C) Superposition of the C. caldarium PSI-LHCI structure with the P. purpureum PSI-LHCI structure (PDB: 7Y5E). The LHCI subunits of C. caldarium and P. purpureum were colored red and blue, respectively. The structures are viewed from the stromal side. (D) Correlation of the names of LHCI subunits in the structures with their genes between C. caldarium and P. purpureum. The names of LHCI subunits and their genes are derived from You et al. (8) for P. purpureum.

The C. caldarium LHCIs of LHCI-1 to LHCI-5 (red) exist at the same positions of RedCAP, Lhcr2, Lhcr5, Lhcr4, and Lhcr3, respectively, in the P. purpureum PSI-LHCI structure (blue) (8) (Fig. 3C). The gene products of each LHC subunit are shown in Fig. 3D, and multiple sequence alignments of the LHCI proteins located at the same positions in the PSI-LHCI structures of C. caldarium and P. purpureum are shown in SI Appendix, Fig. S7 A–E. The amino acid sequence of C. caldarium Lhcr1 has a low similarity of 35% with that of P. purpureum RedCAP. Since RedCAPs are one of the LHC protein superfamily but are distinct from the LHC protein family (18, 19), the binding site at LHCI-1 in the C. caldarium PSI-LHCI structure may be diversified between Cyanidiophyceae and Porphyridiophyceae. In contrast, sequence alignments between C. caldarium Lhcr2 and P. purpureum Lhcr2, between C. caldarium Lhcr1 and P. purpureum Lhcr5, between C. caldarium Lhcr2 and P. purpureum Lhcr4, and between C. caldarium Lhcr3 and P. purpureum Lhcr3 show similarities of 52 to 64%.

Structural Comparison of the C. caldarium LHCIs with the Diatom Chaetoceros gracilis FCPIs.

Diatoms have unique LHCs, fucoxanthin Chl a/c-binding proteins (FCPs), whose bindings to PSI have been shown by cryo-EM single-particle analysis (9, 10). Here, we compared binding sites of the C. caldarium LHCIs with the C. gracilis FCPIs (9) (Fig. 4A). The four LHCI subunits, LHCI-2 to LHCI-5, in the C. caldarium PSI-LHCI structure (red) are located at the same positions of Fcpa1, Fcpa5, Fcpa6, and Fcpa7, respectively, in the C. gracilis PSI-FCPI structure (orange) (9), whereas the C. caldarium LHCI-1 site is empty in the C. gracilis PSI-FCPI structure (Fig. 4A). The gene products of each LHCI subunit are shown in Fig. 4B, and multiple sequence alignments of the LHCI proteins and FCPIs located at the same positions in the C. caldarium PSI-LHCI and C. gracilis PSI-FCPI structures are shown in SI Appendix, Fig. S8 A–D. The results of sequence alignments between C. caldarium Lhcr2 and C. gracilis Lhcr1, between C. caldarium Lhcr1 and C. gracilis Lhcr5, between C. caldarium Lhcr2 and C. gracilis Lhcr6, and between C. caldarium Lhcr3 and C. gracilis Lhcr7 show similarities in the range of 45 to 50%.

Fig. 4.

Structural comparisons of the C. caldarium LHCIs with the diatom C. gracilis FCPIs and the cryptophyte C. placoidea ACPIs. (A) Superposition of the C. caldarium PSI-LHCI structure with the C. gracilis PSI-FCPI structure (PDB: 6L4U). The C. caldarium LHCI and C. gracilis FCPI subunits were colored red and orange, respectively. The structures are viewed from the stromal side. (B) Correlation of the names of LHCI and FCPI subunits in the structures with their genes between C. caldarium and C. gracilis. The names of FCPI subunits and their genes are derived from Nagao et al. (9) for C. gracilis. (C) Superposition of the C. caldarium PSI-LHCI structure with the C. placoidea PSI-ACPI structure (PDB: 7Y7B). The C. caldarium LHCI and C. placoidea ACPI subunits were colored red and purple, respectively. The structures are viewed from the stromal side. (D) Correlation of the names of LHCI and ACPI subunits in the structures with their genes between C. caldarium and C. placoidea. The names of ACPI subunits and their genes are derived from Zhao et al. (11) for C. placoidea.

It is known that antenna sizes of FCPIs in the C. gracilis PSI-FCPI supercomplexes are altered in response to growth conditions, especially CO₂ concentrations and temperatures (23). This has been shown by the presence of two different PSI-FCPI structures, one with 16 FCPI subunits (9) and the other with 24 FCPI subunits (10) (SI Appendix, Fig. S9A). Different from our PSI-FCPI structure with 16 FCPIs (9), each FCP subunit in the PSI-FCPI structures with 24 FCPI subunits (10) was identified using transcriptome data of C. gracilis as well as FCP sequences from other diatom species. The amino acid sequences of 16 FCPIs observed in our PSI-FCPI structure (9) were different from those in the PSI-FCPI structures with 24 FCPI subunits (10). Moreover, the gene names of the 16 FCPIs differ between our structure (9) and the other (10), which were discussed previously (10, 24). As our PSI-FCPI structure (9) was built according to the genome data of C. gracilis (24), we performed further data analysis and discussion of the C. caldarium LHCI-2 to LHCI-5 using the structural data of our PSI-FCPI structure (9) (Fig. 4 A and B) in the present study.

It is particularly noteworthy that the PSI-FCPI structure with 24 FCPIs showed a FCPI subunit located at the C. caldarium LHCI-1 site (SI Appendix, Fig. S9A). The FCPI subunit was built using an FCP sequence of fc13194 in the diatom Fragilariopsis cylindrus (SI Appendix, Fig. S9B). This gene is a RedCAP as discussed in Kumazawa et al. (24). Here, we searched this gene in the C. gracilis database, and obtained a gene like RedCAP in C. gracilis, which was termed as CgRedCAP. A sequence alignment between fc13194 and CgRedCAP (SI Appendix, Fig. S9C) shows a similarity of 50%. Thus, the FCPI subunit located at the same position of the C. caldarium LHCI-1 may be a RedCAP (Gene ID: g6493.t1) in C. gracilis.

Structural Comparison of the C. caldarium LHCIs with the Cryptophyte Chroomonas placoidea ACPIs.

Cryptophytes have unique LHCs, alloxanthin Chl a/c-binding proteins (ACPs), which have been shown to bind to the PSI core to form a PSI-ACPI supercomplex by cryo-EM single-particle analysis (11). We compared the binding sites of the C. caldarium LHCIs with the C. placoidea ACPIs (Fig. 4C). The C. placoidea PSI-ACPI structure showed 14 ACPI subunits in total, which is much more than LHCIs found in the red algae. The five LHCI subunits of LHCI-1 to LHCI-5 in the C. caldarium PSI-LHCI structure (red) are located at the same positions of ACPI-8, ACPI-7, ACPI-3, ACPI-2, and ACPI-1, respectively, in the C. placoidea PSI-ACPI structure (purple) (11) (Fig. 4C). The gene products of each LHCI subunit are shown in Fig. 4D, and multiple sequence alignments of the LHCI proteins and ACPIs located at the same positions in the C. caldarium PSI-LHCI and C. placoidea PSI-ACPI structures are shown in SI Appendix, Fig. S10 A–E. The results of sequence alignments between C. caldarium Lhcr2 and C. placoidea ACPI-7, between C. caldarium Lhcr1 and C. placoidea ACPI-3, between C. caldarium Lhcr2 and C. placoidea ACPI-2, and between C. caldarium Lhcr3 and C. placoidea ACPI-1 show similarities of 49 to 59%. As for ACPI-8, a sequence alignment between C. caldarium Lhcr1 and C. placoidea ACPI-8 shows a low similarity of 37%. Compared with RedCAPs of P. purpureum and C. gracilis, ACPI-8 has sequence similarities of 61 and 56%, respectively (SI Appendix, Fig. S11), indicating that ACPI-8 is a RedCAP.

Phylogenetic Relationships among the Structurally Known Red-Lineage LHCs.

We further examined evolutionary relationships of the C. caldarium LHCI-1 to LHCI-5 with other red-lineage LHCs in the PSI-LHCI structures including PSI-FCPI and PSI-ACPI by phylogenetic analysis (Fig. 5A). At the site of C. caldarium LHCI-1, the C. caldarium Lhcr1 (CcLhcr1) and C. merolae Lhcr1 (CmLhcr1) belonged to the same clade in the phylogenetic tree (Fig. 5A), whereas P. purpureum, C. gracilis, and C. placoidea have no Lhcr but RedCAP (Fig. 5A). Since RedCAPs were phylogenetically distinct from the LHC protein family but assumed to serve as a light harvesting antenna and belonged to the LHC protein superfamily (18, 19), the LHC subunits at the C. caldarium LHCI-1 site can be divided into two groups: C. caldarium/C. merolae Lhcr1s and P. purpureum/C. gracilis/C. placoidea RedCAPs.

Fig. 5.

Phylogenetic analysis and correlation of LHCs in PSI-LHCI structures among red-lineage algae. (A) Unrooted maximum-likelihood tree of the LHCI proteins, FCPIs, and ACPIs whose binding sites to PSI are conserved between C. caldarium and other red-lineage algae, except for RedCAPs. The phylogenetic tree was inferred by IQ-TREE 2 using the LG+I+G4 model and the trimmed alignment of 21 sequences with 268 amino acid residues. Black circles at branches indicate ≥95% ultrafast bootstrap support (1,000 replicates). The names of red-lineage algae were written in front of each protein name; for example, CcLhcr1-2 means Lhcr1-2 of C. caldarium. Cm, C. merolae; Pp, P. purpureum; Cg, C. gracilis; Cp, C. placoidea. (B) Correlation of the LHCI proteins, FCPIs, ACPIs, and RedCAPs in the structures with their binding positions based on the C. caldarium LHCI-1 to 5. Each protein was explained in Figs. 3 and 4, and SI Appendix, Fig. S9; however, RedCAPs of C. gracilis and C. placoidea were interpreted in the present study.

At the site of C. caldarium LHCI-2, CcLhcr2 and CmLhcr2 belonged to the same clade in the phylogenetic tree, but Lhcr2 of P. purpureum (PpLhcr2), Lhcr1 of C. gracilis (CgLhcr1), and ACPI-7 of C. placoidea (CpACPI-7) did not exist in the same clade of CcLhcr2 (Fig. 5 A and B). In addition, PpLhcr5, CgLhcr5, and CpACPI-3 were not included in the same clade of CcLhcr1 at the site of C. caldarium LHCI-3, and PpLhcr4, CgLhcr6, and CpACPI-2 were not grouped into the same clade of CcLhcr2 at the site of C. caldarium LHCI-4 (Fig. 5 A and B). At the site of C. caldarium LHCI-5, CcLhcr3 and CmLhcr3 belonged to the same clade in the phylogenetic tree, but PpLhcr3, CgLhcr7, and CpACPI-1 did not exist in the same clade of CcLhcr3 (Fig. 5 A and B). Thus, our phylogenetic tree strongly indicates that the C. caldarium Lhcr1, Lhcr2, and Lhcr3 are orthologous to the C. merolae Lhcr1, Lhcr2, and Lhcr3, respectively, at each site of LHCI-1 to 5 in the C. caldarium PSI-LHCI structure. However, the orthologous relationships of CcLhcr1–3 with LHCs in the P. purpureum PSI-LHCI, C. gracilis PSI-FCPI, and C. placoidea PSI-ACPI structures cannot be supported at any sites of the C. caldarium LHCI-1 to 5 in our phylogenetic tree.

Sturm et al. reported that RedCAPs exhibited a unique molecular family distinct from LHC proteins (19), although both of them belong to the LHC protein superfamily (18, 19). In fact, a phylogenetic tree including LHCI proteins, FCPIs, ACPIs, and RedCAPs suggests that RedCAPs are distinct from the LHC protein family including LHCIs, ACPIs, and FCPIs (SI Appendix, Fig. S12). However, it should be noted that this tree hardly provides information about the evolutionally history of RedCAPs and LHC protein family, because their structures with three-transmembrane helices seem to be developed from an ancestral single-helix protein of the LHC protein superfamily in different processes (18, 19).

Evolution of the Red-Lineage LHCs Bound to PSI.

Structural comparisons of LHCs in the C. caldarium PSI-LHCI structure with the structures of other red-lineage algae reveal conservation of the binding positions of C. caldarium LHCI-1 to LHCI-5 (Figs. 3 and 4 and SI Appendix, Fig. S9). In contrast, our phylogenetic analysis showed conservation and diversity of the structurally known LHCs among the red-lineage algae (Fig. 5), even though both diatoms and cryptophytes are thought to evolve from red algae (4). It is known that RedCAPs are grouped into one of the ancestral LHC protein family in the red lineage (18, 19) and they appear to be already present in an ancestral red alga prior to the divergence between Rhodophytina and the monophyletic Cyanidiophyceae including the order Cyanidiales, Cyanidioschyzonales, and Galdieriales (16, 18, 19, 25). Galdieria sulphuraria is classified into the order Galdieriales and retains RedCAP in its genome, whereas no RedCAPs have been found at least in the genome of C. merolae (26), which belongs to the order Cyanidioschyzonales (16).

Based on the structural and phylogenetic findings, we propose a schematic model for evolution of the red-lineage LHCs that bind to PSI (Fig. 6). An ancestral red alga appears to have RedCAP (18, 19), which has been found in the PSI-LHCI structures including PSI-FCPI and PSI-ACPI from P. purpureum, C. gracilis, and C. placoidea. This suggests that the binding property of RedCAP in the PSI-LHCI structures including PSI-FCPI and PSI-ACPI are conserved among the Porphyridiophyceae, diatoms, and cryptophytes, as well as an ancestral red alga. In contrast, the PSI-LHCI structures of C. caldarium and C. merolae have Lhcr1 instead of RedCAP at the LHCI-1 site (Red subunits in Fig. 6), suggesting that the two red algae choose Lhcr1 to associate with PSI after the divergence from the order Galdieriales. Since the classification of RedCAPs is different from that of the LHC protein family including CcLhcr1 and CmLhcr1 in the LHC protein superfamily (18, 19), it is highly possible that the C. caldarium and C. merolae PSI-LHCIs possess LHCs with completely different properties at the LHCI-1 site in the process of evolution from an ancestral red alga.

Fig. 6.

Evolutionary scheme of red-lineage LHCs in the PSI-LHCI structures. Structures are viewed from the stromal side. Gray, PSI; red and orange, the LHCI-1 site; blue, the LHCI-2 to 5 sites; green, other LHCs. The red and orange subunits indicate Lhcr1s and RedCAPs, respectively. A detailed explanation of this scheme is described in the main text.

Yoon and coworkers have shown that both the order Galdierales and Cyanidiales/Cyanidioschyzonales lose massive genes during evolution of the Cyanidiophyceae (27). Such evolutionary adaptation may result in the loss of RedCAP and the reduced number of LHC genes after the divergence of Galdierales from other Cyanidiophyceae. At the LHCI-1 site, the space generated by the loss of RedCAP is filled by CcLhcr1 and CmLhcr1 in the C. caldarium and C. merolae PSI-LHCIs, respectively. Here, we define this evolutionary event as neolocalization: a phenomenon in which a structural defect caused by a gene loss is complemented or modified by the product of another existing gene. This may have taken place at the LHCI-1 site during evolution from an ancestral red alga to the order Cyanidiales and Cyanidioschyzonales.

At the sites of LHCI-2 to 5 (blue subunits in Fig. 6), prominent orthologous relationships were found only for CcLhcr1–3 and CmLhcr1–3 both structurally and phylogenetically. The unique LHC features of i) clear orthology between the order Cyanidiales and Cyanidioschyzonales, ii) neolocalization at the LHCI-1 site, and iii) diversity of RedCAP and LHC protein families in the red lineages provide an evolutionary model for red-lineage LHCs. Thus, our structural and phylogenetic findings together with the evolutionary scheme of LHCs will open a broad avenue for studying the conservation and diversity of LHCs associated with PSI among red-linage algae.

It is important to note that the number of species in our phylogenetic analysis is too limited because we focused on the red-lineage algae whose PSI-LHCI structures are determined. The limited species may prove to be challenging in approaching molecular evolution of LHCIs in the red lineage. However, this study employing a combination of structural and phylogenetic analyses is a unique approach to unravel evolutionary conservation and diversity of LHCs, which is remarkably different from previous phylogenetic analyses of LHCs. To establish this evolutionary perspective more firmly, further studies are needed to solve the structures of PSI-LHCI supercomplexes from various red-lineage algae.

Materials and Methods

Cryo-EM Data Collection and Image Processing.

Three-microliter aliquots of the C. caldarium PSI-LHCI (2.83 mg of Chl mL⁻¹) were applied to gold presputtered Quantifoil R0.6/1 Cu 200 mesh grids. The grids were blotted with a filter paper and plunged into liquid ethane cooled by liquid nitrogen, followed by transfer into a CRYO ARM 300 electron microscope (JEOL) equipped with a cold-field emission gun operated at 300 kV. In total 17,650 image stacks were collected. After image processing, the final PSI-LHCI structure was reconstructed from 228,449 particles. The overall resolution of the cryo-EM map was estimated to be 1.92 Å by the gold-standard FSC curve with a cutoff value of 0.143 (28) (SI Appendix, Fig. S2A). Details are described in SI Appendix.

Amino Acid Sequences of PSI and LHCI Proteins.

In this study, we generated the draft genome of C. caldarium RK-1 (NIES-2137) independently. This is because the amino acid sequence of PsaA protein in the C. caldarium RK-1 (NIES-2137) strain was different from those in the C. caldarium RK-1 strain reported previously (29) (SI Appendix, Fig. S13). The detailed protocol is described in SI Appendix.

Model Building and Refinement.

Two types of the cryo-EM maps were used for the model building of the PSI-LHCI supercomplex: One was a postprocessed map, and the other was a denoised map using Topaz version 0.2.4 (30). Each subunit of the homology models constructed using the Phyre2 server (31) was first manually fitted into the two maps using UCSF Chimera (32), and then, their structures were inspected and manually adjusted against the maps with Coot (33). Each model was built based on interpretable features from the density maps with the contour levels of 2.5 and 2.0 σ in the denoised and postprocessed maps, respectively. The PSI-LHCI structure was refined with phenix.real_space_refine (34) and Servalcat (35) with geometric restraints for the protein-cofactor coordination. The statistics for all data collection and structure refinement are summarized in SI Appendix, Tables S1 and S2. Details are described in SI Appendix.

Phylogenetic Analysis.

Amino acid sequences of LHCs were aligned using mafft-linsi v7.490 (36). The alignment was trimmed using ClipKit v1.4.1 with smart-gap mode. The phylogenetic trees of Fig. 5A and SI Appendix, Fig. S12 were inferred using IQ-TREE 2 (37) with the model selected by ModelFinder (38). LG+I+G4 model was selected from the alignment for Fig. 5A and the selected model of SI Appendix, Fig. S12 was the Q.pfam+I+G4 (39). The trees were visualized by iTOL v6 (40). Ultrafast bootstrap approximation was performed with 1,000 replicates (41).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Atomic coordinate and cryo-EM maps for the reported structure have been deposited in the Protein Data Bank under an accession code 8WEY (https://www.rcsb.org/structure/8WEY) (42) and in the Electron Microscopy Data Bank under an accession code EMD-37480 (https://www.ebi.ac.uk/emdb/EMD-37480) (43). All other data are included in the manuscript and/or SI Appendix.