Distinctive and functional pigment arrangements in Lhcp, a prasinophyte-specific photosynthetic light-harvesting complex

Abstract

Light harvesting is essential for photosynthesis, and the diversity of light-harvesting systems enables photosynthetic organisms to acquire unique niches and thrive. Prasinophytes are marine green algae that diverge early in the evolution of photosynthetic eukaryotes and use a distinct light-harvesting complex known as Lhcp as their primary antenna. Lhcp consists of proteins and pigments unique to prasinophytes but shares some structural and functional features with the plant-type light-harvesting complex LHCII. Here, we use cryo-electron microscopy to determine the structure of Lhcp from the prasinophyte Ostreococcus tauri at 1.94 Å resolution, revealing all pigments responsible for light harvesting. The results show that the trimeric structure of Lhcp is stabilized by pigments, including a distinctive carotenoid identified as the cis-isomer of esterified antheraxanthin B. Comparison of Lhcp and plant-type LHCII reveals that while their core architecture is conserved, structural differences underlie their functional divergence. This work provides insight into the evolution of light-harvesting systems and highlights how structural diversity contributes to ecological adaptation.

The 1.94 Å cryo-EM structure of the Lhcp trimer from a prasinophyte Ostreococcus tauri reveals all pigments, including a unique carotenoid aiding trimer stabilization. Comparison with LHCII links structural differences to enhanced blue light absorption.

Introduction

Photosynthesis is a fundamental process that sustains biological systems on Earth by helping them utilize the abundant and inexhaustible supply of solar energy. Photosynthetic organisms have evolved light-harvesting complexes that allow them to adapt to the properties of the sunlight that is available to them in their particular ecological niches, especially from marine to terrestrial environments¹. On land, the photon density in the visible region shows almost no wavelength dependence, but underwater it differs dramatically, where most of the available light is in the blue-to-green region of the solar spectrum depending on depth and turbidity². Light-harvesting complexes have therefore evolved to optimize their ability to absorb such available photons. However, the major light-harvesting complexes of land plants absorb very little green light, the main component of terrestrial sunlight, and instead, since often they have to cope with excess light, they have also developed efficient mechanisms to switch between light harvesting and excess-energy quenching (photoprotection)³. Recent advances in structural biology have enabled high-resolution structural analysis of a wide range of organisms, providing great opportunities to elucidate diverse light-harvesting systems at the molecular level.

The major light-harvesting complex in land plants and green algae is the peripheral antenna that mainly supplies energy to photosystem II, called LHCII. In land plants and green algae, LHCII functions as a trimer, with each individual LHCII apoprotein binding 14 chlorophyll (Chl) molecules and four carotenoid molecules^4–7. Prasinophytes form a basal algal group in the green lineage⁸. The peripheral light-harvesting complex of Prasinophytes exhibits strong absorption at the 400–500 nm region⁹, allowing it to harvest the blue-green light that is available even in the deep ocean². Single-particle cryo-electron microscopy (EM) of the photosystem I supercomplex revealed that Lhcp forms a trimer, with each monomeric Lhcp apoprotein binding seven carotenoid molecules and 14 Chl molecules¹⁰. This trimeric Lhcp also transfers excitation energy mainly to photosystem II¹⁰. The Lhcp protein is rather different from those found in LHCII and forms a separate lineage within the Lhc family¹¹. Lhcp accumulates divinylprotochlorophyllide (Dvp), a precursor of chlorophyll that has been suggested to be present in primitive light-harvesting systems¹². In addition, Lhcp accumulates prasinophyte-specific carotenoids whose biosynthetic enzymes appear to have been lost as one moves up the green lineage¹³. In particular, three distinctive carotenoids have been utilized as key pigments for classifying Prasinophytes¹⁴: uriolide (Uri), micromonal (Mic), and a hydrophobic carotenoid whose chemical structure has yet to be identified and which is commonly referred to as unidentified M1 (Uid). Biochemical analyses¹¹ revealed that all these characteristic carotenoids were accumulated in the Lhcp complex. However, the relatively low resolution (2.9 Å) of the previous structure prevented the clear identification of these carotenoids in the Lhcp trimers¹⁰. Lhcp also contains both Chl a and Chl b. However, as Gisriel et al¹⁵. pointed out, a resolution of ~2.3 Å or higher is required to reliably distinguish between the two types of Chl molecule in a cryo-EM map. A high-resolution, 3D structural analysis of the Lhcp trimer is, therefore, required to unequivocally determine which types of Chl and carotenoid molecules are present in each binding site.

In this study, we elucidated the high-resolution cryo-EM structure of the Lhcp trimer from the prasinophyte marine alga Ostreococcus tauri. The structure of Lhcp is compared with that of LHCII, revealing functionally important differences.

Results

Structure of the Lhcp trimer with stoichiometric cofactors

The cryo-EM map was initially analysed using C₁ symmetry, and this revealed almost no differences among the three monomers. This was not unexpected since Lhcp is mainly found as a homotrimer of the Lhcp-2 polypeptide (UniProt ID: Q3B9U7)^10,11,16. The final structure of the Lhcp trimer was then fully resolved using C₃ symmetry, achieving an overall resolution of 1.94 Å (Fig. 1a–c; Supplementary Fig. 1 and Table 1). Each Lhcp monomer contained seven carotenoid molecules [two prasinoxanthin (Pra), and one each of Uri, neoxanthin (Neo), Mic, 7,8-dihydrolutein (Dhl), and Uid; Supplementary Fig. 2a–d], and 14 Chl molecules (seven Chl a, six Chl b, and one Dvp; Supplementary Fig. 2e). This pigment composition, as shown in Supplementary Fig. 2f, significantly revises the previously published cryo-EM model¹⁰ and corroborates the carotenoid composition reported in ref. ¹¹ (that demonstrated that the characteristic carotenoids (Uri, Mic, Dhl, and Uid) were present in amounts equal to that of Neo). However, the Chl a/b ratio found in the revised cryo-EM model is still inconsistent with the pigment composition reported in ref. ¹¹. Therefore, we reexamined the pigment composition using ultra-high performance liquid chromatography (UPLC), obtaining a Chl a/b ratio of 1.11 (Supplementary Fig. 2f). This corroborates the revised cryo-EM model.

Fig. 1. 3D structure of the Lhcp trimer.

a–c Full trimeric structure of Lhcp is shown from three different viewpoints as indicated. d, e Monomeric parts of Lhcp viewed from the peripheral and internal sides of the trimer are shown. Abbreviations: Chl chlorophyll, Dhl 7,8-dihydrolutein, Dvp Mg-2,4-divinylphaeoporphyrin-a₅ monomethyl ester, LMNG lauryl maltose neopentyl glycol Mic, micromonal Neo, 9’-cis-neoxanthin, Oax 8’-s-cis-3-octenoyl-antheraxanthin B, Pra prasinoxanthin, SQDG sulfoquinovosyl diacylglycerol, Uri uriolide. All structural models were created using ChimeraX⁵⁵.

Table 1.

Cryo-EM data collection, refinement and validation statistics

	#1 Lhcp (EMDB-64036) (PDB 9UC6)
Data collection and processing
Magnification	60,000
Voltage (kV)	300
Electron exposure (e–/Å2)	80
Defocus range (μm)	−700 to −2200
Pixel size (Å)	0.859
Symmetry imposed	C3
Initial particle images (no.)	5,408,616
Final particle images (no.)	1,291,024
Map resolution (Å)	1.94
FSC threshold	0.143
Refinement
Initial model used (PDB code)	8HG5
Model resolution (Å)	2.0
FSC threshold	0.5
Map sharpening B factor (Å2)	65.8
Model composition
Non-hydrogen atoms	8379
Protein residues	600
Ligands	69
B factors (Å2)
Protein	15.99
Ligand	14.43
R.m.s. deviations
Bond lengths (Å)	0.010
Bond angles (°)	1.698
Validation
MolProbity score	1.24
Clashscore	4.67
Poor rotamers (%)	0.22
Ramachandran plot
Favored (%)	99.00
Allowed (%)	1.00
Disallowed (%)	0.00

The apoprotein of Lhcp begins at the N-terminus on the stromal side, and contains three transmembrane helices (A, B, C) and two amphipathic helices (D and E) in the order B, E, C, A, D, and terminates at the C-terminus on the luminal side (Fig. 1d, e). Two salt bridges (Glu-181/Arg-66, and Arg-186/Glu-61) between helices A and B are conserved in comparison to LHCII (Supplementary Table 1). Although the three transmembrane helices are 3–5 residues shorter than those in LHCII, their relative positions within the trimer are well conserved, except for about six residues (from Gln-198 to Gln-203) on the luminal side of helix A are located about 18% (about 3 Å) more inward compared to LHCII (Supplementary Fig. 3a, b).

The improved cryo-EM map facilitated the revision of the Chl assignments for three sites among the 14 Chl molecules in each monomer: The assignment of two sites, 603 and 614, was revised from Chl a to Chl b, and vice versa for site 608 (Fig. 1d, e and Supplementary Table 2). Lhcp lacks Chl 601 found in LHCII but contains an additional Chl b near Chl b 605 as previously identified¹⁰; this site was renumbered as 620 (Supplementary Fig. 3c, d). Site 609 was occupied by Dvp, as previously described¹⁰. Therefore, compared with LHCII from land plants, Lhcp contained six conserved Chl a-binding sites (602, 604, 610, 611, 612, and 613) and three Chl b-binding sites (605, 606, and 607; Supplementary Table 2), with slight differences in the axial ligands of the two Chl a molecules at sites 611 and 612 and the orientation of the Chl b molecule at site 605 (Supplementary Table 2 and Supplementary Fig. 3c).

Uid was purified from a sucrose density gradient fraction primarily containing the Lhcp trimer (Supplementary Note 1). Mass spectrometry, UV-VIS absorption, and circular dichroism (CD) spectra revealed that the absolute structure of Uid is a 3-octenoyl ester of antheraxanthin B with an inverted 5,6-epoxide group compared with that of antheraxanthin (Supplementary Fig. 4). The cryo-EM map corroborated the unusual 8’-s-cis conformation of the 3-octenoyl ester of antheraxanthin B molecule in Lhcp (Hereafter, 8’-s-cis-3-octenoyl-antheraxanthin B will be referred to as Oax). Sixty-two water molecules were assigned based on their spherical electron density shape and hydrogen-bonded clusters in the cryo-EM map (Supplementary Fig. 5).

Like the land plant LHCII, in each monomer, three carotenoid molecules occupied the L1, L2, and N1 sites. Regardless of carbonyl groups or other structural factors, Pra (Pra-1) and Uri in the L1 and L2 sites, respectively, had their 3’-hydroxy-ε-ring directed towards the stromal surface of the complex and formed hydrogen bonds with amino acid residues Asp-164 and Ser-47, respectively (Supplementary Fig. 6). Neo bound to the N1 site with a similar geometry to that seen in the structure of LHCII, except that the 3-OH group is not hydrogen bonded because the highly conserved residue Thy-112 in land plant LHCIIs is Phe-105 in Lhcp (Supplementary Table 1). Four additional carotenoid-binding sites were identified in Lhcp (Supplementary Fig. 3e, f): an additional Pra molecule (Pra-2) between Pra-1 and Neo in an antiparallel direction compared with Pra-1; a Dhl molecule on the interior side of the trimer, crossing over Uri; an Oax molecule at the monomer–monomer interface on the lumenal side of the complex, parallel to the membrane plane; and a Mic molecule at a similar position to the V1 site carotenoid (violaxanthin, Vio) in LHCII, but with a significantly different binding style. Rotating the Mic approximately 90 degrees clockwise from the orientation shown in Fig. 1d reproduces the V1 site carotenoid in LHCII (Supplementary Fig. 3f). Compared to the V1 carotenoid, on the stromal side Mic has the 3’-hydroxy ε-ring closer to helix A, whereas on the luminal side the 3-hydroxy β-ring is pushed toward the surface of the trimer and closer to helix C of the neighbouring monomer.

The cryo-EM map also revealed a lipid-like structure at the lumenal side of the complex near Pra-1 (Supplementary Fig. 7a). Biochemical analysis comparing Lhcp processed with and without anion-exchange chromatography, which eliminates lipids attached to the hydrophobic surface of the pigment–protein complex¹⁷, revealed that Lhcp bound sulfoquinovosyl diacylglycerol (SQDG) (Supplementary Fig. 7b). Although the density of its head group remains unclear, we tentatively assigned this lipid as SQDG in the model (Supplementary Fig. 7c).

The unique carotenoids in Lhcp contribute to trimeric architecture stabilization

A distinctive feature of Lhcp is the presence of three Oax molecules immediately beneath the lumenal surface, arranged radially and parallel to the membrane plane (Fig. 2a–e). Near the 3-fold symmetry axis, these three Oax molecules interlock in a clockwise direction via van der Waals contacts between the 3-hydroxy-β-ionone head groups and the center of the conjugated chain in the C14–C15–C15’–C14’ region. This region, C15’–C14’–C13’, also contacts the Leu-200 and Ala-201 residues in helix A from the opposite side. The methyl group (C19’) of each Oax is in contact with two C-terminal Phe-232 residues coming from the same and adjacent monomers. The C8’–C9’–C10’ region of each Oax contacts the C-terminal Tyr-233 residue belonging to the anticlockwise-adjacent monomer. These interactions with amino-acid residues at the backside of the lumenal surface stabilize the unusual 8’-s-cis conformation of Oax and appear to pull the stromal end of helix A inwards (Supplementary Fig. 3a). Each Oax molecule contacts, in order from the head group to the acylated terminal group, a Dhl molecule belonging to the same monomer, a Dhl molecule and a Chl b607 molecule belonging to the clockwise-adjacent monomer, a Mic molecule belonging to the same monomer, and eventually Leu-126 and Ile-130 both belonging to the clockwise-adjacent monomer, thus establishing a trimeric backbone structure with the carotenoids as rigid conjugated rods.

Fig. 2. Hydrophobic interactions around Oax.

a Lumenal view of Lhcp trimer highlighting three hydrophobic carotenoids: Oax, Dhl, and Mic. b Same view showing only the Oax molecules. c Amino acid residues in van der Waals contact (less than 4 Å, indicated by blue dashed lines) with the adjacent Oax molecules are shown. d Side view of monomer-1 from the interior side of the trimer is shown, emphasizing the amino acid residues interacting with Oax. e Pigments interacting with the same Oax are shown: phytyl chain of each chlorophyll molecule is truncated for clarity. Abbreviations follow Fig. 1.

Lhcp lacks most of the well-conserved amino acids believed to stabilize the trimeric structure of LHCII¹⁸, which accumulate on the stromal and lumenal surfaces (Fig. 3). On the stromal surface (Fig. 3a, d, e), the N-terminal loop is significantly shorter in Lhcp than in LHCII, and Glu-37, the axial ligand for Chl a611, that substitutes for phosphatidylglycerol seen in LHCII, is located in this loop region (Fig. 3e). The hydrogen bond network, including Lys-183 in helix A and Tyr-35 adjacent to Glu-37 across the proline turn, appears to stabilize this region and maintain the position of Chl a611. On the lumenal surface, a well-conserved Trp residue in the C-terminal region (Trp-222 in spinach LHCII), which has been suggested to stabilize the amphipathic helix D via hydrophobic interactions with the adjacent monomer¹⁸, was absent in Lhcp (Fig. 3c, g). Instead, the hydrophobic surface of the short helix (from Val-222 to Ala-227) interacts with the acyl group of Oax and the 3-hydroxy-β-end group of Mic (Fig. 2d). Therefore, we propose that the hydrophobic interactions surrounding the Oax molecules are essential for stabilizing the trimeric architecture.

Fig. 3. Structure and sequence comparison between Lhcp and LHCII.

N- (a) and C-terminal (c) surfaces of superposed Lhcp (orange) and Spinacia oleracea LHCII (PDB:1RWT, green) are shown. b Side view of the superposition, aligned via A/B/C helices using CCP4MG. d–g Enlarged views of the regions highlighted in (a, c). h–k Corresponding amino acid sequence alignments for the regions shown in d–g. Key residues are marked in black for Lhcp and blue for LHCII.

Functional characterization of the Lhcp trimer

The Lhcp trimer, purified from O. tauri as described previously¹⁰, showed basic light-harvesting features, absorbing visible light and emitting from the red-most Chl a. Compared to LHCII, Lhcp exhibited enhanced absorption and energy transfer capacity in the blue-green region and a slight blue-shift in the Chl a emission peak (Fig. 4a). These features are shared with siphonous LHCII^19–21. The CD pattern of Lhcp differed significantly from that of plant LHCII²² and siphonous LHCII¹⁹, particularly in the Q_y region (Fig. 4b). The well-characterized CD pattern of trimeric LHCII follows a negative (681 nm)–positive (665 nm)–negative (651 nm) peak pattern due to coupling between the Q_y bands of Chl molecules²². In contrast, Lhcp exhibited a more complex pattern, with an additional positive (676 nm)–negative (671 nm) pair inserted between the negative (681 nm)–positive (665 nm) pair, and another additional positive (642 nm)–negative (635 nm) pair appearing on the shorter wavelength side of the negative (651 nm) peak, representing significantly altered coupling. The 77-K absorption spectrum of Lhcp enables better separation of these peaks than that seen in the room temperature spectrum in the 600–700 nm region (Fig. 4c; Supplementary Fig. 8a). The Q_y peaks at 676, 668, and 661 nm were empirically assignable to Chl a, and the peak at 643 nm was assigned to Chl b. The 632-nm peak could be attributed to the Q_y band of Dvp, as isolated Dvp exhibits Q_y absorption at 636 nm in pyridine¹². These results suggest that Dvp interacts with the Chls, resulting in the more complex CD pattern than that seen with LHCII. Also, at 77-K the broad room temperature absorption band in the green region separated into two distinct peaks at 518 and 537 nm (Fig. 4c). The overall energy transfer efficiency, calculated from the fluorescence excitation and the 1-T spectra, was slightly lower (~90%) at 518 nm than at 537 nm (~100%), probably due to different carotenoid molecules (Fig. 4a and Supplementary Fig. 8b).

Fig. 4. Spectroscopic characterization of the Lhcp trimer.

a Absorption (ABS), fluorescence excitation (FLE) and emission (FLU) spectra, along with energy transfer efficiency (η_EET). Blue arrows indicate η_EET at 518 and 537 nm. b Circular dichroism (CD) profile. c 77 K absorption spectrum; peak wavelengths were determined using a second-derivative method (see Supplementary Fig. 8a). d Decay-associated spectra (DAS) from nanosecond time-resolved absorption spectroscopy. The inset shows a 60-fold vertical magnification.

Photoexcitation of Chl a can produce a harmful triplet excited state (³Chl). In land-plants²³ and siphonous green algae²¹, carotenoids bound to LHCII efficiently quench ³Chl*. To investigate this photoprotective function of the carotenoids in Lhcp, nanosecond time-resolved absorption spectroscopy was used (Fig. 4d and Supplementary Fig. 9). Upon excitation of the Q_y band of Chl a, a bleaching at 400–500 nm and an increase in absorption at 500–600 nm were observed, corresponding to the ground-state bleaching and triplet excited-state absorption of carotenoids, respectively. This shows that the generation of triplet carotenoid (³Car*) upon excitation of Chl also takes place in Lhcp. The decay-associated spectrum of the 8.1-μs kinetic component indicated that the excited-state absorption the triplet carotenoids had two peaks at 506 nm and 534 nm, both with relatively wide bandwidths (Fig. 4d). Given the similarity of ³Car* excited-state absorption in various algal LHCs containing carbonyl carotenoids^21,24,25, this suggests that two distinct carbonyl carotenoids contribute to ³Chl* quenching in Lhcp. The decay-associated spectrum also contained a small, but non-zero, signal of a long-lived component (350 μs; Fig. 4d), ascribable to unquenched ³Chl*²⁶. The amount of unquenched ³Chl* was <5% of the total in the complex, reflecting the highly efficient quenching of ³Chl* in Lhcp.

Since triplet excitation energy transfer requires van der Waals contacts between the donor and acceptor molecules, the possible sites for efficient ³Chl* quenching in Lhcp were narrowed down to the Pra-1–a610–a612 and Uri–a602 domains (Fig. 5a, b), corresponding to the L1 and L2 domains of plant LHCII^3,18, respectively.

Fig. 5. Pigment clusters surrounding carbonyl carotenoid binding sites in Lhcp.

a Uri–L2 domain located on the side facing the trimer core. Pra–L1 domain (b) and Mic domain (c), located on the side facing the trimer periphery.

Plausible energy transfer pathway among Chl molecules based on excitonic couplings, comparison between Lhcp and LHCII

Energy transfer within LHCII has been described using excitonically coupled clusters of Chl molecules²⁷. Although the total number of Chl molecules remains the same in Lhcp, its Chl clustering differs significantly from those in plant LHCII. In Lhcp Chl a molecules are abundant in the stromal layer, while Chl b molecules are abundant in the lumenal layer, and in the stromal layer, the intermonomer interaction between Chl b601’-b609 found in plant LHCII is absent (Fig. 6a, b). To describe the flow of excitation energy in Lhcp, we calculated the excitonic coupling as described by Amerongen and Grondelle²⁸ with the new parameters introduced for Dvp-609 (Supplementary Note 2 and Supplementary Fig. 10).

Fig. 6. Chlorophyll clusters in the stromal and lumenal layers of the Lhcp trimer.

Chlorophyll (Chl) molecules in the stromal Chl layer (a) and lumenal Chl layer (b) of the Lhcp trimer are shown, viewed from the stromal and lumenal sides, respectively. Only the chlorin macrocycles are displayed. Mg–Mg distances between Chl molecules in asymmetric regions of the trimer are indicated with bold numbers.

On the stromal side, Chl b601 is absent, and the Mg–Mg distance between Chl a602 and Chl a611 is slightly shorter than that in LHCII (Fig. 6a). The Chl b608–b609–b601 cluster in LHCII is condensed in Lhcp, forming a large Chl a cluster involving a608–a610–a612–a611. The strong excitonic coupling between a608 and a610 (76 cm⁻¹) supported this assignment. The strongest excitonic coupling between a611 and a612 (151 cm⁻¹) suggested that a611–a612 dimer is the terminal emitter in Lhcp, as in LHCII^29,30. Strong excitonic coupling was also calculated between Dvp–609 and Chl b603 (−83 cm^–1) and between Chl b603 and a602 (−61 cm⁻¹), suggesting that the blue light (450–500 nm) absorbed by Dvp is primarily transferred to Chl a602 via Chl b603. Interestingly, the intermonomer excitonic coupling between Chl a602 and a602’ has a value in Lhcp (−5 cm⁻¹) that suggests Chl a602 may contribute to energy migration within the trimer.

An additional Chl b620 was detected on the lumenal side of Lhcp, and the orientation of b605 differed from that in LHCII. Chl b620 exhibited strong excitonic coupling (77 cm⁻¹) with the adjacent a604 with an Mg–Mg distance of 11.7 Å (Fig. 6b). The strong excitonic coupling (117 cm⁻¹) between b606 and a604 and the conserved Chl b cluster b605–b606–b607 suggests that the excitation energy absorbed by these Chl b molecules is transferred to Chl a604. Interestingly, the exciton coupling (24 cm⁻¹) between b607 and b603 is enhanced from that in LHCII (2 cm⁻¹), suggesting this b607/b603 coupling can contribute to the energy migration between the lumenal and stromal Chl clusters. At sites 613–614, the Chl a homodimer in plant LHCII becomes a Chl a–b heterodimer in Lhcp. Energy absorbed by Chl b614 is transferred to Chl a613 via strong excitonic coupling (−43 cm⁻¹). Note that the exciton coupling between Chl a613 of adjacent monomers (4 cm⁻¹) is significantly enhanced in Lhcp compared to LHCII (2 cm⁻¹), based on the inward position of the axial ligand of a613 (Glu-198) on the lumen side of helix A. This suggests that the energy transfer between Chl a613 of adjacent monomers proposed in plant LHCII^29,30 is preserved or even enhanced in Lhcp.

These calculated Chl–Chl couplings support the complex CD spectrum in the Q_y region. The CD pattern of negative (683 nm)–positive (663 nm)–negative (652 nm) bands can be assigned to the two conserved couplings of Chl a611–a612 and a604–b606, as in spinach LHCII²². The high-energy CD pattern of positive (642 nm)–negative (635 nm) bands was then assigned to the coupling between Dvp-609 and Chl b603, and the negative signal at 671 nm was attributed to the coupling between Chl a608–a610.

Carotenoids also exhibit a broad absorption in the blue region (400–500 nm) and eventually transfer their excitation energy to Chl a. Since the efficiency of this energy transfer depends on the distance and angle between the relevant transition dipoles, these important factors were compared for each carbonyl carotenoid and its proximal Chl molecule(s) (Supplementary Note 3 and Supplementary Table 3). All the unique carotenoids (including Pra-2, Dhl, Mic and Oax) have at least one partner Chl molecule with significant dipole-dipole interactions, which can facilitate the transfer of excitation energy to the terminal emitter Chl. Thus, an increased number of carotenoids per Chl molecule contributes to the enhanced blue light (400–500 nm) utilization by Lhcp.

Discussion

This contribution presents a much-improved structure of Lhcp trimer. This has allowed the previous assignment of the chlorophylls and their binding sites to be corrected, as well as the structures and locations of all the unusual carotenoids to be fully detailed.

The properties of any given organism’s light-harvesting system are expected to be optimised to fit the characteristics of the incident solar radiation available in that organism’s ecological niche. In the case of the marine alga O. tauri, this means being able to enhance the absorption of blue-green and green wavelengths of light. Whereas land plants get access to a full solar spectrum of typically high-intensity light. Their LHCII does not need to retain the ability to harvest blue-green and green light but does need to have evolved more extensive mechanisms for photoprotection, such as non-photochemical quenching. It is, therefore, interesting to compare Lhcp with land plant LHCII to elucidate the differences between these two light-harvesting complexes that facilitate these different imperatives.

Strikingly, especially when compared with land plant LHCII, the absorption of Lhcp extends into the green region due to the presence of carbonyl carotenoids (Supplementary Fig. 11). Each Lhcp monomer contains four carbonyl carotenoid molecules (Uri, Pra-1, Pra-2, and Mic) that are the potential contributors to this green light absorption (Fig. 5a–c). The relative absorption cross-section of the two green bands peaking at 518 and 537 nm was approximately 3:1 (Supplementary Fig. 8b), suggesting that three of the four molecules contribute to the former and one to the latter. The calculated energy transfer efficiency at 518 nm was slightly lower than at 537 nm, indicating that at least some of the carotenoids absorbing at 518 nm have less efficient singlet-singlet energy transfer to Chl. Accurately describing the energy transfer from carotenoids to chlorophyll remains challenging because the S₁ state of carotenoids is symmetrically forbidden, even though it is energetically close to the Q_y state of Chl^31–33. However, the traditional point-dipole approximation remains useful for assessing the structural factors behind the energy transfer between each pigment pair. Here, we approximated the transition dipole of the carotenoid to be located at the centre of C13 and C13’, with a strength of 3.0 Debye for carbonyl carotenoids (the reported value for the S₁-S₀ transition of peridinin³⁴.) Using the dipole strength values of Chls from Amerongen and Grondelle²⁸, the exciton coupling (V) of the carotenoid and proximal Chl was calculated by the following equation (see Supplementary Note 3 and Supplementary Tables 3–1):

$$V_{AB}=C\kappa /R_{AB}^3\left\{∣{\mu }_A∣∣{\mu }_B∣\right\}$$ 1

$$\kappa ={cos\theta }_{AB}\text{-}{3cos\theta }_A{cos\theta }_B$$ 2

where θ_AB is the angle between two dipoles and θ_A or θ_B is the angle of each dipole with respect to the axis joining the centres of the two dipoles. The exciton coupling between proximal carotenoids was also calculated similarly, but here the dipole strength of the S₂ state was assumed (Supplementary Tables 3–2). Uri and Pra-1 had one (a602) and two (a612 and a610) partner Chl a molecules with large couplings (V = −116, −306, and 192 cm^–1, respectively), whereas Mic and Pra-2 only had one Chl a partner (a608 and a611, respectively) with relatively small couplings (V = −16, and 15 cm^–1, respectively) (Fig. 5a–c and Supplementary Tables 3–1). Pra-2 is very close to Pra-1 and shows a large exciton coupling of V > 2500 cm^–1. As a consequence, Pra-2 appears to transfer excitation energy to Chl a via Pra-1. Therefore, as the energy transfer efficiency from Mic to the terminal emitter Chl a is expected to be low, it can be suggested that Mic is the major contributor to the 518-nm absorption band.

Interestingly, LHCII from the intertidal siphonous green algae also employs a few carbonyl carotenoids to enhance its green light absorption capacity while retaining the same basic structural framework of land plant LHCII^19,20. Siphonous LHCII also enhances blue-green (~480 nm) absorption by replacing Chl a with Chl b at two sites. In comparison to siphonous LHCII, Lhcp has a larger absorption cross section in the green (500–530 nm) and the blue (400–470 nm) regions (Supplementary Fig. 11). The former can be due to the additional carbonyl carotenoid, Pra-2 (see above). The latter results from both the increased number of carotenoids and the presence of the unusual pigment Dvp, which has a significant molar extinction coefficient at 450 nm³⁵.

The main centrally located Chl binding sites in the Lhcp trimer are highly conserved with respect to those seen in LHCII. However, the peripherally located Chl-binding sites at sites 601, 605, and 620 in Lhcp are not conserved in LHCII. Light harvesting requires enough pigments per unit area to capture sufficient photons while avoiding concentration quenching. This suggests that the arrangement of the conserved core Chl molecules is an optimal solution balancing these two opposing requirements. More generally, highly conserved pigment arrangements appear to be associated with conserved functions, e.g., the a610–a611–a612 cluster functions as a terminal emitter, and the L1 and L2 domains function as triplet quenchers. The former suggests that the terminal emitter is optimally positioned for efficient excitation energy transfer to adjacent pigment–protein complexes, and the latter domains have been optimized for highly efficient triplet quenching.

Although the positions of core Chl-binding sites are highly conserved, the type of Chl molecule that occupies each site can vary between the different types of LHCs. In Lhcp luminal Chl clusters accumulate Chl b molecules, including b620, whereas stromal Chl clusters contain only one Chl b molecule and one Dvp molecule and lack Chl b601, which is important for facilitating energy transfer between monomers in land plant LHCII³⁶. Therefore, blue-green light (480–500 nm) absorbed by the Chl b molecules in Lhcp is primarily transferred from the luminal cluster to the stromal cluster, ultimately reaching the terminal emitter. In contrast, in siphonous LHCIIs, Chl b is distributed equally in both stromal and luminal layers. Therefore, since Lhcp and siphonous LHCII are both highly efficient in transferring energy from Chl b to Chl a, they appear to employ different energy transfer pathways for harvesting blue–green (480–500 nm) light.

In land plant LHCII, a conformational change induced by lumenal acidification due to strong light has been suggested as a molecular mechanism for the rapid switch from the light-harvesting state to the energy-quenching state^3,18. Spectroscopic analysis has suggested that specific conformational changes in the a610–a611–a612 cluster and lutein at the L1 site are attributable to the energy-quenching state^37,38. Recently, Ruan et al.⁶ revealed that the structural basis of the conformational switch involves helix E changing from a 3₁₀ helix to an α-helix and a salt bridge forming between Lys-203 of helix A and Glu-207 of helix D at the lumenal surface. This induces helix B movement, leading to hydrogen bond formation between the Asp-54 of the B helices of the three monomers on the stromal side, shortening the distance between the L1 carotenoid and Chl a612. Interestingly, Lhcp does not contain these residues (Glu-207 and Asp-54 in LHCII, see Fig. 3i, k, respectively). Although the helix D in Lhcp also adopts a 3₁₀ helical conformation, the presence of an S–S bond between Cys-93 and Cys-98 immediately after helix D (Fig. 3f) inhibits hydrogen bond rearrangement from the 3₁₀-helix to an α-helix. Furthermore, the hydrophobic interactions underlying the unique carotenoid-based backbone structures of Lhcp are not expected to be affected by pH differences, as the hydrophobic interactions primarily reflect changes in entropy and are therefore largely unaffected by changes in ionic strength (proton concentration). In O. tauri, non-photochemical quenching comparable to that in land plants has been observed only when grown under high light conditions (400–800 μmol photons s^–1 m^–2)³⁹. Even if Lhcp trimer is involved in this non-photochemical quenching, it is unlikely to take place via the same mechanism proposed for land plant LHCII. However, the L1 site of Lhcp contains a carbonyl carotenoid (Pra), and the S₁ transition dipole moment of a carbonyl carotenoid was estimated to be significantly larger than that of typical non-carbonyl carotenoids³⁴. Consequently, the excitonic coupling between the S₁ state of this carotenoid and the Q_y band of Chl will be stronger in Lhcp. This suggests that the distance required for the quenching is shorter than in LHCII, and the possibility remains that the quenching may occur at the same site.

Various quenching sites have been proposed in the LHCII of land plants³. However, it is not possible to compare these proposed mechanisms directly to the case of Lhcp due to pigment replacement(s). But notwithstanding this, the a610–a611–a612 site could be a potential candidate. The difference between Lhcp and LHCII is that the axial ligand of a611 has changed from PG to glutamate (Glu-37), resulting in a change in the angle between the planes of the chlorin rings of a611 and a612 (see Supplementary Fig. 12). Ostroumov et al.⁴⁰ Proposed the quenching involving a charge-transfer state within the Chl a610–a612–a611. Although the large exciton coupling between a611 and a612 is preserved, the overlap of electron clouds–another important factor for generating the Chl-Chl charge transfer state⁴¹–is questionable when fixed in this configuration. Dynamic movement of the N-terminal loop of LHCII has been proposed to lead to a displacement of a611 relative to a612, resulting in significant fluctuations in the a611–a612 exciton coupling⁴². In Lhcp, the axial ligand of a611 is located within the N-terminal loop, and therefore, in principle, a change in the angle between a611 and a612 could also be caused by the dynamic motion of this loop, thereby functioning as a switch for non-photochemical quenching. Further investigations will be required to determine whether any of these potential possibilities are actually correct.

Methods

Sample preparations

Lhcp for cryo-EM was purified from O. tauri as follows: Thylakoid membranes were isolated using the following procedure. Cells were suspended in buffer A (25 mM HEPES-NaOH, pH 7.5, 0.33 M sucrose) and disrupted by two rounds of nebulization using BioNeb (Glas-Col, Terre Haute, IN, USA) at a pressure of 8 kgf cm^–2 with N₂ gas¹⁰. The obtained membranes were solubilized with 1.4% dodecyl-α-D-maltoside (α-DDM, Anatrace, Maumee, OH, USA). After solubilization, unsolubilized material was removed by centrifugation at 15,000 rpm for 1 min. The supernatant was then subjected to sucrose density gradient (SDG) ultracentrifugation (0.1/0.4/0.7/1.0/1.3 M sucrose in buffer B: 25 mM HEPES-NaOH, pH 7.5, 0.02% α-DDM) at 28.000 rpm for 20 h at 4 °C. The sucrose in the collected SDG fraction was removed by ultrafiltration using an Amicon Ultra-0.5 mL centrifugal filter unit (100 kDa cutoff) with buffer B. The detergent was then exchanged from α-DDM to lauryl maltose neopentyl glycol (LMNG) by three cycles of ultrafiltration using Amicon Ultra-0.5 (100 kDa), alternating between buffer C (25 mM HEPES-NaOH, pH 7.5) and buffer D (buffer C with 0.5% (w/v) LMNG).

Subsequently, the sample was repurified by ultracentrifugation through a graded sucrose layer (0.1–0.5 M) in buffer E (50 mM MES, 10 mM NaCl, 20 mM CaCl₂, 0.002% LMNG) at 50,000 rpm for 5 h at 4 °C using a swinging-bucket rotor (P55ST2, Hitachi-koki). The trimeric Lhcp fraction was collected, and the remaining sucrose was removed via ultracentrifugation using Amicon Ultra-0.5 (100 kDa) with buffer F (25 mM HEPES-NaOH at pH 7.5, 0.002% LMNG).

Lipid analysis

Lipid analysis was performed on Lhcp trimer samples purified under two different conditions: Lhcp trimer was isolated by two successive rounds of SDG, and the Lhcp trimer was further purified by anion exchange chromatography (AEC) followed by SDG. For AEC, a binary gradient was performed using the following solutions: eluent A, 25 mM Tris-HCl at pH 8.2 containing 0.03% β-DDM; and eluent B, eluent A containing 1 M NaCl. A linear gradient from 20% B to 40% B was applied during elution from 0 to 100 mL at a flow rate of 1.5 mL/min. Lipids were extracted from Lhcp trimer samples through the Bligh-Dyer method with minor modifications: Briefly, total lipids were extracted from Lhcp using a mixture of methanol and acetone (1:1, v/v), followed by the addition of an equal volume of chloroform and 1 M KCl. The chloroform layer was collected and dried under a stream of N₂ gas. Thin-layer chromatography (TLC) was performed using an HPTLC silica gel 60 plate (Merck, Darmstadt, Germany). Samples were spotted onto the plates, which were then developed in a saturated chamber using a solvent mixture of acetone:toluene:methanol:water (40:15:5:2, v/v/v/v)⁴³. Separated lipid bands were stained by immersing the dried plates in a 0.05% (w/v) Primulin dye solution in acetone:water (4:1, v/v) for 5 s, followed by air drying⁴⁴. The fluorescent bands were visualized under UV light at 365 nm using a hand-held UV lamp (Luv-4, As-One, Osaka, Japan).

Uid isolation and purification

Pigments were extracted from Lhcp trimer, isolated by SDG centrifugation, using a mixture of acetone and methanol (1:1, v/v). The solvent was evaporated under a stream of N₂ gas, and the dried extract was dissolved in acetone:hexane (10:90, v/v). The sample was injected into an HPLC system (CBM-20A, SPD-M20A, LC-20AD, DGU-20A_5R, Shimadzu, Kyoto, Japan) equipped with a COSMOSIL SL-II column (10 mm I.D. x 250 mm, Nacalai Tesque, Kyoto, Japan)⁴⁵. Pigments were separated by isocratic elution with acetone:hexane (20:80, v/v). Fractions containing the Uid were collected and subjected to multiple rounds of purification. The final purified pigment was characterized by UV-visible absorption spectroscopy, circular dichroism (CD), and mass spectrometry. The absolute chemical structure of Uid was determined to be 3-octenoyl-antheraxanthin B [Oax; (3S,5S,6R,3’R)-5,6-epoxy-5,6-dihydro-3’-hydroxy-3-octenoyloxy-β,β-carotene] (Supplementary Note 1).

Spectroscopic characterization

Absorption, fluorescence, fluorescence excitation, and CD spectra were recorded in 25 mM HEPES-NaOH at pH 7.5 containing 0.03% (w/v) α-DDM using a spectrophotometer (UV-1800, Shimadzu, Kyoto, Japan), a fluorometer (FP-8500, JASCO, Tokyo, Japan), and a CD spectroscope (J720W, JASCO, Tokyo, Japan), respectively. The pigment composition analyses were performed by UPLC using a Waters H-class system. Pigments were extracted from the samples with 90% acetone. The whole pigments were separated with a Cadenza CD-C18 UP, 150 × 2 mm, 3 μm column (Imtakt Corporation, Kyoto, Japan) with gradient shifting of three-solvent system, acetonitrile/isopropanol/water, from 65:15:20 (v/v/v) to 80:15:5 (v/v/v) at 6.5 min and to 50:50:0 (v/v/v) at 10 min, finally returning to 65:15:20 (v/v/v) at 15 min. The column temperature was 45 °C. The system was calibrated with commercial standards for only available pigments (DHI, Hoersholm, Denmark). Concentrations of Dhl and Dvp were estimated based on the response factor of lutein and Chl c₂, respectively, and Uri and Mic were estimated based on that of Pra using the values from Latasa et al.¹⁴. We used extinction coefficients from Roy et al.⁴⁶. The emission spectrum (500–850 nm region) with excitation at 475 nm and the fluorescence excitation spectrum (350–730 nm region) observed at 740 nm were recorded. The overall efficiency of excitation energy transfer was calculated by dividing the fluorescence excitation spectrum by the 1-T spectrum normalized at 675 nm. CD is expressed as the molar circular dichroism Δε = ε_L − ε_R = ΔA/(CL), where C is the molar concentration of Lhcp trimers calculated from the Chl a concentration calculated from 80% acetone extract using Porra’s equation⁴⁷ and L is the optical path length in cm. The 77 K absorption spectra of Lhcp were recorded on a spectrophotometer (Cary 60, Agilent Tech., Santa Clara, CA) equipped with the liquid-nitrogen cooling cryostat (Optistat DN, Oxford Instruments Holdings Inc., Oxford, UK).

Nanosecond time-resolved absorption spectroscopy

Nanosecond time-resolved absorption spectra of Lhcp were measured in the home-built laser system with a time resolution of 50 ns²³. Lhcp (OD at Q_y = 0.5) was excited by the pump pulses (672 nm, 5 mW, 3 × 5 mm²) provided by an optical parametric oscillator laser system (100 Hz, SpitLight EVO-OPO, InnoLas Laser, Germany). Absorption difference (ΔOD) was obtained by using laser-driven white light (LDLSEQ-1500, Energetiq Technology, Wilmington, MA) and Si-PIN photodiode S3071 (Hamamatsu Photonics, Japan) connected to a digital oscilloscope (WaveSurfer HDO-4054, LeCroy, Chestnut Ridge, NY). The obtained data were processed by using MATLAB R2019a (MathWorks) and open-source software Glotaran. Before measurement, the sample was incubated with 20 mM β-D-glucose, 5 U/mL glucose oxidase, and 5 U/ mL catalase as a final concentration for 15 min to remove the oxygen. A comparison of the steady-state absorption spectra before and after the measurement confirmed no degradation of the sample due to the measurement.

Cryo-EM data collection

We used the multiple blotting method⁴⁸ with modification to overcome low sample concentration. Quantifoil Cu R 1.2/1.3 holey carbon grids on both sides were performed by hydrophilic treatment using JEC-3000RC (JEOL, Japan), for 10 s at 10 mA. A 2.5 μL aliquot of Lhcp (~5 mg/mL) was applied to both sides of the grid and then manually blotted at room temperature. Quickly moving on the normal blotting operation, 2.5 μL of the sample was applied on the grid and frozen in liquid ethane using Vitrobot IV system (FEI) at 4 °C, 100% humidity, 3 s blotting time, 5 s waiting time, and blotting force –10. Data for Lhcp were collected on a CRYO ARM 300 II (CryoJEM-3300, JEOL, Japan), respectively, equipped with a cold field-emission electron detector camera (Gatan, USA). The Cryo-EM images were collected using Serial-EM⁴⁹. The holes were detected using YoneoLocker⁵⁰. Movie frames were recorded using a K3 camera at a nominal magnification of ×60,000 corresponding to pixel sizes of 0.859 Å (CRYO ARM 300 II) at the specimen level. The data were collected with a total exposure of 3 s fractionated into 40 frames, with a total of ~80 electrons Å^–2 in counting mode. In total, 6144 movies were collected with defocus values varying from 0.7 to 2.2 μm. Typical cryo-EM images averaged from motion-corrected movie frames are shown in Supplementary Fig. 1.

Cryo-EM data processing

A gain reference image was prepared with the relion_estimate_gain command in RELION 4.0⁵¹ using the first 300 movies. Images were processed using cryoSPARC ver. 4.2.1⁵². A total of 6144 movies of Lhcp were imported and motion corrected, and contrast transfer functions (CTFs) were estimated. A total of 5707 movies with maximum CTF resolutions greater than 5 Å were selected. The particles were automatically picked using a template picker job with a particle diameter at 200 Å using templates made from the cryo-EM map of marine green algal LHCII trimer (PDB ID: 9UBY). After particle extraction with 2× binning, two-dimensional (2D) classification into 50 classes was performed to select clear 2D class averages. Additionally, three rounds of 2D classification were performed to select 2,416,034 particles. Selected particles were extracted again with a box size of 0.859 pixels without binning and were subjected to homogenous refinement. After global/local CTF refinement and non-uniform refinement, global/local CTF refinement and non-uniform refinement were performed with C₃ symmetry, and a final map was reconstructed at 1.94 Å resolution (FSC = 0.143). The processing strategy is detailed in Supplementary Fig. 1.

Model building and validation

The atomic models of Lhcp were constructed with using WinCOOT 0.9.447⁵³, Phenix 1.19-41584⁵⁴, and both Chimera and ChimeraX⁵⁵ software packages. The coordinate of Lhcp (PDB ID: 8HG5) was directly docked as a template into the C₃ symmetry imposed cryo-EM map of Lhcp using Chimera such that the trans-membrane helices were roughly matched to the electrostatic potential profile of the cryo-EM map. Coordinates of Uri (Chemical Component Dictionary (CCD) ID: A1LYY), Dhl (A1L0E), Mic (A1LYX), Oax (A1LYW), and Dvp (A1L0F) are newly made based on the ChemDraw structure by Grobal Phasing Limited, Grade2 (https://grade.globalphasing.org/cgi-bin/grade2_server.cgi). The geometry-optimized model was globally refined in Phenix to achieve a global resolution of each electron potential map. The refinement statistics are summarized in Table 1.

Reporting summary

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

Author contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. S.S. designed the study; M.K., A.I., and E.K prepared the samples; S.S. purified the sample and performed cryo-EM data analyses; S.S. and T.M. measured EM micrographs, processed the EM data, and reconstructed the final EM map; T.M. and K.N. supervised the EM micrograph measurements and analyses; H.T. and G.K. supervised the structural model refinements; S.S. performed the structural analysis; S.S. performed lipid analysis; S.S. and R.F. performed the carotenoid determination; S.S., M.K., A.I., and E.K. performed the characterization; N.Y. performed spectroscopic analyses; J.Z. supervised the spectroscopic analysis; R.F., G.K., J.Z., K.N., and J.M. supervised the project; S.S., M.K., N.Y., and R.F. wrote the draft manuscript; S.S., N.Y., E.K., T.M., N.N., H.T., G.K., J.Z., K.N., J.M., R.F., and R.C. revised the final manuscript; all authors contributed to the interpretation of the results and improvement of the manuscript.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Lei Zheng and Laura Rodríguez Pérez. A peer review file is available.

Data availability

The atomic coordinates and structure factors have been deposited in the Protein Data Bank (PDB) under accession code PDB ID: 9UC6. The corresponding cryo-electron microscopy density map has been deposited in the Electron Microscopy Data Bank (EMDB) under accession code EMD-64036. The source data underlying Fig. 4 are provided in the Supplementary Data 1.xlsx file, available in the Supplementary Information section of this article. All other relevant data are available from the corresponding authors on reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-025-08977-x.