Structural basis and evolution of the photosystem I–light-harvesting supercomplex of cryptophyte algae

Abstract Cryptophyte plastids originated from a red algal ancestor through secondary endosymbiosis. Cryptophyte photosystem I (PSI) associates with transmembrane alloxanthin-chlorophyll a/c proteins (ACPIs) as light-harvesting complexes (LHCs). Here, we report the structure of the photosynthetic PSI–ACPI supercomplex from the cryptophyte Chroomonas placoidea at 2.7-Å resolution obtained by crygenic electron microscopy. Cryptophyte PSI–ACPI represents a unique PSI–LHCI intermediate in the evolution from red algal to diatom PSI–LHCI. The PSI–ACPI supercomplex is composed of a monomeric PSI core containing 14 subunits, 12 of which originated in red algae, 1 diatom PsaR homolog, and an additional peptide. The PSI core is surrounded by 14 ACPI subunits that form 2 antenna layers: an inner layer with 11 ACPIs surrounding the PSI core and an outer layer containing 3 ACPIs. A pigment-binding subunit that is not present in any other previously characterized PSI–LHCI complexes, ACPI-S, mediates the association and energy transfer between the outer and inner ACPIs. The extensive pigment network of PSI–ACPI ensures efficient light harvesting, energy transfer, and dissipation. Overall, the PSI–LHCI structure identified in this study provides a framework for delineating the mechanisms of energy transfer in cryptophyte PSI–LHCI and for understanding the evolution of photosynthesis in the red lineage, which occurred via secondary endosymbiosis.


Introduction
Oxygenic photosynthesis, one of the most important types of metabolism that produces atmospheric oxygen and organic matter, plays a fundamental role in driving evolution (Dekker and Boekema 2005;Nelson and Junge 2015). Through oxygenic photosynthesis, cyanobacteria, algae, and plants capture sunlight energy and convert it into chemical energy to drive almost all life activities. The major components responsible for light-driven photosynthetic electron transport are photosystem I (PSI) and photosystem II (PSII); these multi-subunit pigment-protein supercomplexes reside in thylakoid membranes (Mullineaux and Liu 2020). PSI transfers electrons derived from the PSII-mediated oxidation of water to ferredoxin, producing reducing power and the energy needed for CO 2 assimilation. The PSI reaction center core is generally surrounded by lightharvesting complexes (LHCs), which bind with numerous pigments including chlorophylls (Chls) and carotenoids, forming a PSI−LHCI supercomplex to increase the absorption cross-section of PSI (Nelson and Junge 2015;Qin et al. 2015Qin et al. , 2019Mazor et al. 2017;Pi et al. 2018;Antoshvili et al. 2019;Su et al. 2019;Suga et al. 2019;Caspy et al. 2020;Nagao et al. 2020;Xu et al. 2020;Yan et al. 2021;Gorski et al. 2022;Naschberger, Fadeeva, et al. 2022;Naschberger, Mosebach, et al. 2022).
Cryptophytes are a phylum of single-celled biflagellate eukaryotic algae that function as significant primary producers in ecologically diverse habitats (Shalchian-Tabrizi et al. 2008;Stiller et al. 2014;Zimorski et al. 2014;Kim et al. 2017;Abidizadegan et al. 2021). In parallel with the evolution of green lineages, cryptophytes and diatoms obtained the specialized photosynthetic organelles plastids from an ancestral red alga via secondary endosymbiosis. Whereas red algae contain extrinsic antenna phycobilisomes and transmembrane (TM) LHCs but lack Chl c, and diatoms possess Chl a/c-containing LHCs but lack phycobiliproteins, cryptophytes utilize both phycobiliproteins and LHCs, which contain Chl a/c 2 as well as alloxanthins as the major carotenoids, thereby designated alloxanthin Chl a/c-binding proteins (ACPIs). ACPIs play an role in extending the spectral region of captured light (Pennington et al. 1985;Schagerl and Donabaum 2003;Chen et al. 2007;Janssen and Rhiel 2008;Kereiche et al. 2008;Takaichi 2011;Takaichi et al. 2016;Greenwold et al. 2019), pointing to their unique position in the evolution of red-lineage plastids. The structures of PSI-LHCI supercomplexes in red algae and diatoms, known as PSI-LHCR and PSI-FCPI (fucoxanthin-chlorophyll a/c-binding proteins), respectively, have recently been solved (Pi et al. 2018;Antoshvili et al. 2019;Nagao et al. 2020;Xu et al. 2020). However, the architecture and the pigment arrangement of cryptophyte PS−LHCI supercomplexes remain unclear.
Here, we report the cryogenic electron microscopy (cryo-EM) structure of PSI-ACPI, an 885-kDa membranespanning photosynthetic protein supercomplex, from the cryptophyte Chroomonas placoidea at 2.7-Å resolution. The structure reveals that PSI-ACPI comprises 14 ACPI subunits that form 2 antenna layers surrounding a PSI core containing a total of 373 pigments. In addition, 2 protein subunits were identified that are not present in other previously characterized PSI-LHCI complexes. Our study provides important insight into the molecular mechanisms of light capturing and energy transfer in cryptophyte PSI-ACPI. In addition, we provide evidence for the evolutionary variations of PSI-LHCI supercomplexes in the red lineage.
The PSI core contains 100 Chl a, 19 α-Car, 5 Alx (A850, B843, O204, O205, R203), and 2 Cro (K104, L204) molecules (Supplemental Table S2). The cryptophyte PSI core possesses all the Chl-binding sites identified in the red algal and diatom PSI cores (Supplemental Fig. S6). A new Chl-binding sites (PsaB/a841) were found in the cryptophyte PSI core, which is absent in the red algal and diatom PSI cores (Supplemental Fig. S6, B and E). The cryptophyte PSI core lacks 1 carotenoid-binding site identified in red algal PsaA, as does the diatom PSI core. However, 6 new carotenoidbinding sites evolved in the PsaJ/M/O/R subunits compared with the red algal PSI core, 3 of which were identified in PsaJ and PsaO but are also absent in the diatom PSI core (Supplemental Fig. S6, C and F). Alx and Cro replace α-Cars or β-Cars in the PSI cores of other oxyphototrophs (Supplemental Fig. S6F), which is similar to the diatom PSI core, in which several β-Cars are replaced by Ddx and Fx (Xu et al. 2020).
Cryptophyte PsaR binds to 1 Chl a and 2 carotenoids and shares a high similarity with diatom PsaR in terms of amino acid sequence, structure, and location within PSI-LHCI (Fig. 2, A and B;Supplemental Figs S7A and S8B); both have low sequence similarity to PsaG of green-lineage organisms (Xu et al. 2020). In addition, cryptophyte PsaR possesses a longer N-terminal loop at the stromal side, a shorter helix αA at the lumenal side, and a shorter loop between αA and αB relative to diatom PsaR, which may enhance its binding with PsaB (Supplemental Fig. S8, A and D). Like diatom PsaR (Xu et al. 2020), cryptophyte PsaR also mediates the association of peripheral antenna components and energy transfer (Supplemental Fig. S8, E to G). Cryptophyte PsaK and PsaO are highly similar to their counterparts located at the same positions in the red algal PSI core (Pi et al. 2018;Antoshvili et al. 2019) (Fig. 2, C to E;Supplemental Figs S6A, S7, C, D, and S9, A and E). PsaK binds to 2 Chl a and 2 carotenoids and mediates the binding of the PSI core with ACPI-1 and ACPI-11 via its long loop at the stromal side, which is absent in red algal PsaK ( Fig. 2C; Supplemental Fig. S9, A to C). The extended N-terminal loop of cryptophyte PsaK facilitates its association with PsaA (Supplemental Fig. S9, A and D). PsaO binds to 3 Chl a and 2 Alx molecules, and a new carotenoid-binding site (Alx205) was identified close to Chl a201/202 (Fig. 2E), suggesting its role in energy quenching. The extended terminal loops of PsaO and PsaL enhance the association of PsaO with the core (Supplemental Fig. S9, E to I). Although PsaK and PsaO coexist in red algae, green algae, and plants (Qin et al. 2015(Qin et al. , 2019Mazor et al. 2017;Pi et al. 2018;Antoshvili et al. 2019;Su et al. 2019;Caspy et al. 2020;Yan et al. 2021;Gorski et al. 2022), the PsaK/O-mediated binding of antennas to the PSI core at the PsaK-PsaO side was only identified in cryptophyte PSI-ACPI (Supplemental Fig. S10).
A distinctive feature of the cryptophyte PSI core is the presence of a extrinsic subunit Unk1 at the lumenal side close to PsaB, PsaM, and ACPI-7 (Figs 1 and 2, F and G). Its amino acid residues could not be assigned due to its low-resolution density map, and its structure was built as poly-alanines. Unk1 lacks ligands and contains 4 parallel α-helices close to the lumenal side of PsaB (Fig. 2, F and G). Its long loop between helices binds to the interface of PsaB, PsaM, and ACPI-7 (Fig. 2G). The low-resolution densities suggest the loose association of Unk1 with the PSI core. We found that the densities assigned to Unk1 do not fit the structures of the cryptophyte phycocyanin β subunit (PDB: 4LMS), PsbQ (PDB: 3LS0), Psb27 (PDB: 2Y6X), or the PsnL2 or PsnL3 subunits of the NDH-1 complex (PDB: 7WFF), which all possess 4 parallel α-helices. The exact structure and role of Unk1 merit further investigation.

Structures of ACPI subunits
All ACPI apoproteins contain 3 major TM helices (αA, αB, αC) and a short amphipathic helix αE between αA and αB, and 11 ACPIs (2 to 4, 6 to 11, 13 to 14) possess an amphipathic helix αD at the C-terminal region ( Fig. 3A; Supplemental Figs S11 and S12), resembling those of other LHCs (Qin et al. 2015(Qin et al. , 2019Mazor et al. 2017;Pi et al. 2018;Antoshvili et al. 2019;Su et al. 2019;Suga et al. 2019;Caspy et al. 2020;Nagao et al. 2020;Xu et al. 2020;Yan et al. 2021;Gorski et al. 2022). The C-termini of ACPI-1/5/12 are shorter and that of ACPI-9 is longer than those of other ACPIs (Supplemental Figs S11 and S12). Since the cryo-EM densities of the ACPI-10 and ACPI-13 subunits are identical, we used the same sequence from the transcriptome to build their models. To better illustrate the structure, we assigned them ACPI-10 and ACPI-13 based on their distinct positions and binding sites in the PSI-ACPI supercomplex (Fig. 1A). Two highly conserved Glu-Arg pairs form salt bridges between the αA and αC helices in all ACPIs to stabilize the ACPI structure (Supplemental Fig. S12).
The ACPI-S subunit has a molecular mass of 20.6 kDa and is located between ACPI-2/3/4 and ACPI-12/13/14; it binds to 3 Chls a, 1 Chl c, and 2 α-Car molecules, thereby presumably serving as a special antenna subunit (Fig. 3, B and C; Supplemental Fig. S13A). An ACPI-S homolog was only found in one other cryptophyte species, Guillardia theta Figure 1. Overall structure of the cryptophyte PSI-ACPI supercomplex. A) The PSI-ACPI supercomplex viewed from the stromal side. Core subunits are shown in surface view and their names labeled. LHCs are shown in cartoon form and labeled as ACPI. The PSI core subunit and the subunit lying between the ACPs are labeled "Unk1" and "ACPI-S", respectively. B) The PSI-ACPI supercomplex viewed from the lumenal side. C) Side view of the PSI-ACPI supercomplex. D) Two layers of ACPIs in the PSI-ACPI supercomplex. Outer layer, pink; the PSI core, gray; ACPI-S and Unk1 are shown in red and blue, respectively.
(Supplemental Fig. S7E). ACPI-S has 1 TM helix at the interface between ACPI-3, ACPI-4, ACPI-13, and ACPI-14, as well as long-terminal loops and an amphipathic helix at the C-terminal region parallel to the lumenal membrane surface. Its structure is very different from that of other ACPIs: its TM helix is shorter than those of other ACPIs, and its coordinated Chls have distinct binding positions from those of other ACPIs (Supplemental Fig. S13B). The overall structure of ACPI-S stretches across ACPI-2/3/4/12/13/14 via its long Nand C-terminal loops, forming interactions with PsaF, ACPI-3/4, and ACPI-12/13/14 and resulting in slight shifts of ACPI-4 and ACPI-5 in PSI-ACPI (relative to FCPI-8 and FCPI-9 in diatom PSI-FCPI) ( Phylogenic analysis revealed that ACPIs belong to the Lhcr family, similar to red algal LHCRs and diatom Lhcr-type FCPIs, except that the ACPI-8 subunit forms a group with FCPI-1 (Supplemental Fig. S15). Lhcr-type ACPIs can be categorized into Group I (ACPI-2/3/4/7/11/14) and Group II (ACPI-1/5/ 6/9/10/12/13), which mainly differ in the structures of their AE loops between αA and αE (Supplemental Fig. S16, A and B). ACPI-3/7/11/14 in Group I share similar structures containing a short αB helix, which is consistent with the structures of red algal Lhcr1 and diatom FCPI-7 (Supplemental Fig. S16A). ACPI-1/2 structurally resemble the corresponding red algal Lhcrs (Lhcr3/2). However, the structures of ACPI-1 and its counterpart diatom FCPI-5 differ in both the loop and helix regions due to the loss of PsaK in diatom PSI-FCPI (Supplemental Fig. S16B) (Xu et al. 2020). ACPI-4 shows some structural differences with its diatom counterpart FCPI-8, especially in the loop regions, facilitating its interactions with adjacent subunits (Supplemental Figs S13, C, D, S16A, and S17A). ACPI-S and FCPI-19 share similar positions within the supercomplexes, whereas the TM helix of ACPI-S is further from the PSI core than the TM helix αB of FCPI-19 (Supplemental Fig. S13D). These differences may result in the shifting of ACPI-4 compared with FCPI-8. ACPI-5 shifts relative to its diatom counterpart FCPI-9 and differs from FCPI-9 in its loop regions, particularly the N-terminal loop (Supplemental Fig. S16B), which favors the binding of FCPI-21, a region with no homolog in cryptophyte PSI-ACPI (Xu et al. 2020). ACPI-7 has a longer C-terminal loop than red algal Lhcr2* and diatom FCPI-11 at the same position, which helps stabilize the binding of ACPI-7 (Supplemental Figs S16A and S17B). The structure of ACPI-8 is markedly distinct from those of other ACPIs and red algal Lhcr1* but is comparable with its counterpart FCPI-1 in diatoms (Supplemental Figs S16C and S18, A and D) (Pi et al. 2018;Xu et al. 2020). ACPI-8 also possesses a unique N-terminal helix, which is absent in diatom FCPI-1 (Supplemental Fig. S18D).

Associations between the PSI core and ACPIs
The associations between adjacent ACPIs within the same layer are mainly formed by the BC loop (between Αb and Αc) of ACPI-n and the N-terminal loop of ACPI-(n + 1), except for ACPI-7/8 and ACPI-11/1, which have greater intersubunit distances (Supplemental Fig. S21). The interactions between outer and inner ACPIs are mainly mediated by ACPI-S via their loops at both membrane surfaces (Supplemental Figs S13 and S14). An ACPI-S homolog is absent in diatom PSI-FCPI, and diatom FCPI-17/18/19/20 (which share similar positions with the outer ACPI-12/13/ 14 subunits) could readily be detached from the inner FCPI ring (Nagao et al. 2020;Xu et al. 2020), further suggesting the important role of ACPI-S in stabilizing ACPIs. In addition, the outer ACPI BC loops and N-terminal loops also form interactions with the BC loops of inner ACPIs, reinforcing interlayer interactions (Supplemental Fig. S21). The inner ACPIs associate with the PSI core mainly via their loop regions, as do LHCIs from other oxyphototrophic species (Supplemental Fig. S21). The ACPI-1/2/3/8 subunits have similar binding sites with the PSI core as their counterparts in red algal PSI-LHCR and diatom PSI-FCPI (Supplemental Fig. S20, A and E; Movie 1) (Pi et al. 2018;Xu et al. 2020). ACPI-8 binds with the PSI core via its AB loop (Supplemental Fig. S17, C and E), which is consistent with diatom FCPI-1 (Xu et al. 2020). ACPI-4/5/6 bind to the PSI core via PsaR, as observed in diatom PSI-FCPI (Supplemental Figs S8, E to G and S20E) (Nagao et al. 2020;Xu et al. 2020). The Unk1 subunit mediates the interactions between ACPI-7 and PsaB, resulting in the shift of ACPI-7 outward from the PSI core compared with Lhcr2* and FCPI-11 (Supplemental Fig. S20, C, D, I, and J; Movie 1 and 2). Compared with their FCPI counterparts, the locations of ACPI-9/10/11/12/13/14 are greatly shifted, presumably due to the presence of ACPI-S in cryptophyte PSI-ACPI and the absence of PsaK and PsaO in PSI-FCPI (Supplemental Fig. S20, G and H; Movie 1). Apart from the finding that the PSI-ACPI supercomplex possesses 14 ACPI subunits, we also observed 1 class of PSI-ACPI that contains 11 ACPIs but not ACPI-9/10/11, suggesting relatively weak associations of ACPI-9/10/11 to the PSI core (Supplemental Figs S2 to S4).

Pigment arrangement in ACPIs
The 14 ACPIs contain 151 Chl a, 19 Chl c, 54 Alx, 10 Cro, 3 Mon, and 4 α-Car molecules (Supplemental Table S2). The Chl a/c ratio is 7.95, which is higher than that in diatom FCPI antennas (6.82) (Xu et al. 2020). The Chl c content per ACPI is 1.36, which is similar to that of FCPI (1.40) (Xu et al. 2020). The Chl/Car ratio is 2.39, which is comparable to that (2.44) for red algal LHCRs but greater than that (1.94) for diatom FCPIs (Pi et al. 2018;Xu et al. 2020); these results suggest that the capacity for blue-light absorption by ACPIs is likely lower than that of FCPIs.
Each Lhcr-type APCI apoprotein binds to 11 to 15 Chls, among which up to 3 are Chls c and the rest are Chls a (Supplemental Table S2). The Chl-binding sites 301 to 312 are conserved in many Lhcr-type ACPIs (  Table S4; Movie 3), whereas ACPI-3/4/7/11/14 lack the 311 sites due to the absence of the histidine ligand (Supplemental Fig. S12). In contrast, other 7 Chl-binding sites (313 and 316 in ACPI-9, 314 and 318 in ACPI-3, 315 in ACPI-1/5/6/9/12, 317 in ACPI-7, and 319 in ACPI-12) are absent in red algal LHCRs, while Chls 314 to 317 of ACPIs share similar binding sites with Chls in diatom FCPIs (Supplemental Fig. S22, A and B). Chls 314/317 coordinated at the interface of adjacent ACPI subunits and Chls 316/318 at the interface of ACPIs and core subunits might mediate energy transfer between adjacent subunits (Supplemental Figs S23, A to D and S24A); Chl c313 binds to the C-terminal region of ACPI-9 and forms a Chl pair with Chl c308 (Supplemental Fig. S23E); Chl a315 binds to the special AE loop in most Group II ACPIs (Supplemental Fig. S23F); and Chl a319 binds to the C-terminal region of ACPI-12 (Supplemental Fig. S23G).
Unlike other ACPIs, ACPI-8 has only 8 Chl-binding sites (303 to 307, 309 to 310, 314), most of which are shifted due to deviations in protein structure, except for site 304 (Supplemental Fig. S18, A and B). A Chl-binding site 314 was identified in ACPI-8 that was not present in FCPI-1 (Supplemental Fig. S18E) (Xu et al. 2020). The pigments associated with ACPI-S are located at the interface between the ACPI inner and outer layers (Supplemental Fig. S23H), suggesting that ACPI-S functions in energy transfer between ACPIs.

Energy transfer within the PSI-ACPI supercomplex
Our structural analysis allowed us to propose the possible excitation energy transfer (EET) pathways in cryptophyte PSI-ACPI based on the close Chl-Chl distances, as described in previous studies (Qin et al. 2015(Qin et al. , 2019Pan et al. 2018;Pi et al. 2018;Antoshvili et al. 2019;Su et al. 2019;Suga et al. 2019;Nagao et al. 2020;Xu et al. 2020;Yan et al. 2021). Chls in ACPIs can be divided into a stromal layer and a lumenal layer based on their spatial binding positions (Supplemental Fig. S24C). The Chl a305/a306 pairs are similar to the Chl pairs at the same locations in other LHCIs (Morosinotto et al. 2003;Qin et al. 2015Qin et al. , 2019Mazor et al. 2017;Pi et al. 2018;Antoshvili et al. 2019;Su et al. 2019;Suga et al. 2019;Caspy et al. 2020;Nagao et al. 2020;Xu et al. 2020;Yan et al. 2021;Gorski et al. 2022). These are potential red-shifted Chls that might have lower energy in ACPIs. All Chl a305/a306 pairs are located at the interfaces between inner ACPIs and the PSI core as well as between outer and inner ACPIs at the stromal side (Fig. 4E), suggesting they might mediate EET from ACPIs to the PSI core. In addition, the Mg-Mg distances between Chl a305/a306 pairs and Chl a311/a312 in ACPI1-7/9 to 14 and Chl a309 in ACPI-8 are approximately 13 Å. This short distance may provide the structural basis for efficient EET between the stromal and lumenal Chl layers (Supplemental Fig. S24C).
Apart from vertical EET, EET in PSI-ACPI can be categorized into 3 possible lateral pathways: (i) between ACPIs within the same layer, (ii) from outer to inner ACPIs, and (iii) from inner ACPIs to the PSI core (Supplemental Fig. S24D). At the Figure 4. Arrangement of the pigments and possible energy transfer pathways in the PSI-ACPI supercomplex. A) Typical Chl sites in ACPI-2 viewed from the stromal side. Its 12 Chl sites are conserved in most ACPIs. Chls a and Chls c are colored green and blue, respectively. B) Nineteen Chl sites of all ACPIs viewed from the stromal side. The 303, 308, 310, and 315 sites can bind both Chl a (green) and Chl c (blue). C) Typical carotenoid sites in ACPI-2 viewed from the stromal side. Its 5 carotenoid sites are conserved in most ACPIs. D) Eight carotenoid-binding sites of all ACPIs viewed from the stromal side. Alloxanthin (Alx), crocoxanthin (Cro), monadoxanthin (Mon), and α-carotene (α-Car) are colored orange, brown, magenta, and gray, respectively. The 401, 403, and 404 sites can bind different types of carotenoids. Stromal-side view of the energy transfer pathways from the inner ACPIs to PSI core (black arrows), between the outer and inner layer ACPIs (blue double-headed arrows), and within the ACPI layers (magenta double-headed arrows) at the stromal E) and luminal F) layers. Zoom-in images of the boxed areas are shown in (G). The Mg-to-Mg distances shorter or longer than 20 Å are indicated with solid or dotted arrows, respectively. A1 to A14 represent ACPI-1 to ACPI-14. AS represents ACPI-S. Chl pairs are highlighted as sticks. The interfacial Chls supporting energy transfer between the outer and inner ACPI layer and between the inner ACPIs and PSI core are shown as spheres. G) Zoom-in views of the squared areas in (E) and (F). The Mg-to-Mg distances between Chls are labeled in Å. stromal side, EET within the same ACPI layer is primarily mediated by the Chl a305/a306 pairs, Chl a304 of 1 ACPI, and Chl a301 of the adjacent ACPI (Fig. 4, E and G; Supplemental Fig. S25A). At the lumenal side, Chl a/c310 is closely associated with Chl a308 in adjacent ACPI, facilitating EET within the same ACPI layer (Fig. 4, F and G; Supplemental  Fig. S25A). Chls a603 and a601 of ACPI-S mediate stromal and lumenal EET between ACPI-3 and ACPI-4, respectively (Fig. 4, E to G; Supplemental Fig. S25B). In addition, EET between ACPI-4 and ACPI-5 is mediated by ACPI-4/Chl c310 and ACPI-5/Chl a312 at the lumenal side due to the larger distances between their Chls at the stromal side (Fig. 4, F and G). EET between ACPI-7 and ACPI-8 is mainly mediated by ACPI-7/Chl c317 and ACPI-8/Chl a307 at the stromal side, and EET between ACPI-8 and ACPI-9 is primarily mediated by ACPI-8/Chl a310 and ACPI-9/Chl a312 at the luminal side (Fig. 4, E to G). EET between ACPI-1 and ACPI-11 is likely less efficient due to the large gap between these structures (Fig. 4
From the inner ACPIs to the PSI core, Chl a305/a306 pairs mediate EET from ACPIs to the PSI core at the stromal side ( Fig. 4E; Supplemental Tables S5 and S6). At the lumenal side, the EET pathways are mostly mediated via Chl a312 ( Fig. 4F; Supplemental Tables S5 and S6). As ACPI-7 shifts outward from the PSI core, the EET between ACPI-7 and the PSI core is mediated by the Chl a841 in PsaB ( Fig. 4F; Supplemental Fig. S6, Tables S5 and S6). EET between ACPI-10 and the PSI core is facilitated by the coupling of ACPI-10/Chl a311 and PsaO/Chl a202 ( Fig. 4F; Supplemental Tables S5 and S6). The main EET pathways between ACPI-6/8/9 and the PSI core are through other Chls rather than Chl a312 ( Fig. 4F; Supplemental Tables S5 and S6). EET from ACPI-2/8 to the PSI core mainly occurs at the lumenal side, given the larger Chl-Chl distances (≥25 nm) at the stromal side (Fig. 4, E to G). ACPI-5 has a larger distance from the PSI core than other ACPs, and its EET to the core is mediated by adjacent ACPIs (Fig. 4, E and F; Supplemental Table S5).

The structure of cryptophyte PSI-ACPI provides insights into the evolution of red lineage PSI-LHCI
Our structural analysis revealed that cryptophyte PSI-ACPI shares common architectural features with red algal PSI-LHCR and diatom PSI-FCPI. Cryptophyte PSI-ACPI also possesses unique characteristics in terms of its protein organization, pigment association, and EET pathways. The cryptophyte PSI core contains PsaK and PsaO that are homologous to those of red algae but are absent in the diatom PSI core. PsaR is present in both the cryptophyte and diatom PSI cores but is absent in the red algal PSI core. PsaS was identified in the diatom PSI core but is absent in the cryptophyte and red algal PSI cores. The cryptophyte PSI core also contains the Unk1 subunit, which was not identified in the PSI cores of red algae, diatoms, or any other oxyphototrophs. Moreover, the number of antenna subunits of cryptophyte PSI-ACPI (14 ACPIs) lies between those of red algal PSI-LHCR (5 LHCRs) and diatom PSI-FCPI (24 FCPIs) (Pi et al. 2018;Nagao et al. 2020;Xu et al. 2020). Consistently, the variability of the number of PSI peripheral antennas has also been detected among evolutionarily distinct green-lineage photoautotrophs (Pan et al. 2020;Suga and Shen 2020;Bai et al. 2021). Most ACPIs share similar structures with red algal LHCRs, whereas ACPI-4 and ACPI-8 structurally resemble diatom FCPI-8 and FCPI-1, respectively. The arrangements of ACPI-1/2/3/7/8 are similar to those of their counterparts in red algal PSI-LHCR and diatom PSI-FCPI, and ACPI-4/5/6 share similar positions with their counterparts in diatom PSI-FCPI (due to the presence of PsaR). In contrast, the organizations of ACPI-9/10/11/12/13/14 are unique to cryptophyte PSI-ACPI, which are mediated by PsaK, PsaO, and ACPI-S (Supplemental Fig. S20; Movie 1).
Cryptophyte ACPI-1/2/3/7 share similar EET pathways with their counterparts in red algal PSI-LHCR and diatom PSI-FCPI, and ACPI-6/8 share similar EET pathways with FCPI-10/1 of diatom PSI-FCPI (Pi et al. 2018;Nagao et al. 2020;Xu et al. 2020). ACPI-9/10/11 possess unique EET pathways to the PSI core mediated by PsaK and PsaO (Fig. 4, E to G). The ACPI-S-mediated EET from outer ACPIs to inner ACPIs is unique to cryptophyte PSI-ACPI (Fig. 4, E to G). Moreover, the EET pathways via Chl a312 of ACPI-1/4/7 to the PSI core are absent in diatom PSI-FCPI due to the lack of the corresponding Chls in FCPI-5/8/11 (Nagao et al. 2020;Xu et al. 2020). The EET pathways between ACPI-4 and the PSI core are distinct from those of PSI-FCPI, given the shift of FCPI-8, the absence of Chl a312, the binding of the Chl a413, and the shift of Chl a405 in FCPI-8 (Supplemental Fig. S25F) (Xu et al. 2020). The binding of Unk1 leads to the organizational shift of ACPI-7 outward from the PSI core, which may result in less efficient EET from ACPI-7 to the PSI core compared to red algae and diatoms (Supplemental Fig. S20; Movies 2 and 3). ACPI-9/Chl a316 is a specific Chl that mediates EET from ACPI-9 to PsaO, which is absent in FCPI-2 at the similar position in diatom PSI-FCPI (Fig. 4, F and G).
The structural similarity and variations of red-lineage PSI-LHCI supercomplexes highlight the intermediate state of cryptophyte PSI-ACPI between red algal and diatom PSI-LHCI and provide important insights into the secondary endosymbiosis of red-lineage oxyphototrophs (Kooistra and Medlin 1996;Medlin et al. 1997;Shalchian-Tabrizi et al. 2008;Stiller et al. 2014;Zimorski et al. 2014) (Fig. 5). During the evolution of the red-lineage PSI-LHCI supercomplex, cryptophyte PsaR was integrated into the red algal PSI core and facilitated the binding of ACPI-4/5/6; Unk1 binds to PsaB and ACPI-7 at the lumenal side; ACPI-S binds to the inner ACPIs and mediates the association of outer ACPIs (ACPI-12/13/14) with inner ACPIs. ACPI-9/10/11 may then associate with the resulting cryptophyte PSI-ACPI intermediate structure. During secondary endosymbiosis from cryptophytes to diatoms, it is presumed that the loss of PsaK, PsaO, Unk1, and ACPI-S led to conformational shifts of the corresponding FCPIs and facilitated the integration of other FCPIs to generate a large PSI-FCPI supercomplex. The flexible loop structures of diatom FCPIs may reduce the stability of the PSI-FCPI supercomplex (Nagao et al. 2020;Xu et al. 2020), likely providing the structural basis for the photoacclimation of diatoms, allowing them to survive in specific ecological niches.
In summary, the cryo-EM structure of cryptophyte PSI-ACPI reveals the specific protein organization and pigment arrangement of the PSI core and associated antennas, providing insight into the fundamental mechanisms of light harvesting and energy transfer in PSI-ACPI. Our characterization of cryptophyte PSI-ACPI sheds light on the structural variations of PSI-LHCI in the red lineages and highlights the intermediate state of cryptophyte PSI-ACPI between red algal PSI-LHCR and diatom PSI-FCPI during secondary endosymbiosis.

Purification of PSI-ACPI
Chroomonas placoidea T11 (a gift from Prof. Chen Min, College of Chemical and Biological Sciences and Engineering, Yantai University, Shandong, China) was Figure 5. Possible evolutionary development of red-lineage PSI-LHCI supercomplexes. Cryptophyte PSI-ACPI represents an intermediate PSI-LHCI complex between the red algal PSI-LHCR and diatom PSI-FCPI complex. Briefly, during the evolution of the red-lineage PSI-LHCI supercomplex, PsaR subunit bound to the red algal PSI core and mediated the binding of ACPI-4/5/6 in cryptophyte PSI-ACPI, providing the structural basis for the association of ACPI-S, which mediated the association of outer ACPIs (ACPI-12/13/14) with inner ACPIs. Unk1 bound to cryptophyte PSI core at the lumenal side. ACPI-9/10/11 associated with the cryptophyte PSI core, forming an ACPI ring around the core. The PsaK, PsaO, Unk1, and ACPI-S subunits were lost during evolution from cryptophytes to diatoms, resulting in the conformational shifts of the corresponding FCPIs and facilitating the integration of other FCPIs to generate a large PSI-FCPI supercomplex. cultured in F/2 medium under continuous light (40 μmol photons m −2 s −1 ) (LED, T5, 23 W, 5,000 K) at 22 °C with continuous bubbling of air. Purification was performed at 4 °C under dim light. Cells in the logarithmic phase were pelleted by centrifugation at 6,000 × g for 10 min and washed with MES1 buffer (25 mM MES-NaOH, pH 6.5, 1.0 M betaine, 10 mM MgCl 2 ) followed by another centrifugation at 6,000 × g for 10 min. The cell pellets were resuspended in MES1 and broken down with glass beads (diameter 212 to 300 μm) (Casella et al. 2017;Zhao et al. 2020). Unbroken cells were removed by centrifugation at 3,000 × g for 10 min, and thylakoid membranes were collected by centrifugation at 21,000 × g for 30 min. The thylakoid membranes were washed with MES2 buffer (25 mM MES-NaOH, pH 6.5, 1.0 M betaine, 1 mM EDTA), resuspended in MES3 buffer (25 mM MES-NaOH, pH 6.5, 1.0 M betaine, 10 mM NaCl, 5 mM CaCl 2 ) at 0.3 mg mL −1 Chl a, and solubilized with 0.9% (w/v) n-dodecyl-α-D-maltopyranoside (α-DDM) (Anatrace, USA) for 5 min on ice. The mixture was centrifuged for 30 min at 21,000 × g, and the supernatant was loaded onto a discontinuous sucrose gradient (10% to 30%) with an interval of 2% in MES3 buffer containing 0.02% α-DDM, followed by centrifugation at 230,000 × g for 20 h (Beckman SW41 rotor). The PSI-ACPI band was collected and further purified by size-exclusion chromatography (GE; Superose 6 Increase 10/300 GL) in MES4 buffer (25 mM MES-NaOH, pH 6.5, 0.5 M betaine, 50 mM NaCl, 5 mM CaCl 2 , 0.02% α-DDM). The peak fractions were collected and concentrated using an Amicon Ultra 100 kDa cutoff filter (Millipore) at 4,000 g.

Characterization of PSI-ACPI
Absorption spectra were measured at room temperature using a UV-Vis Spectrophotometer (UV-1900, Shimadzu). The PSI-ACPI supercomplexes were denatured as described (Ma et al. 2017) and separated by 8% to 16% SDS-PAGE.
For mass spectrometry, Coomassie Brilliant blue-stained bands were cut out from the gel, reduced with dithiothreitol, alkylated with iodoacetamide, and digested using sequencing-grade modified trypsin, and the resulting peptides were extracted. The peptides were separated through a reverse phase trap column (nanoViper C18, 100 μm × 2 cm, Thermo Fisher) connected to the C18-reversed phase analytical column (75 μm × 10 cm, 3 μm resin, Thermo Fisher) with an EASY-nLC 1000 System, which was coupled to a Q Exactive mass spectrometer (Thermo Fisher). The MS/MS spectra from each LC-MS/MS run were searched using MASCOT engine (version 2.4) against the selected database using the Proteome Discovery searching algorithm (version 1.4).
Pigment composition was analyzed by HPLC (Shimadzu, Japan) using a C18 reversed-phase column (Waters, Ireland) with a Shimadzu photodiode array detector. Pigment extraction and elution were performed as described previously (Xu et al. 2020). The elutes were detected at 445 nm with a wavelength detection range of 300 to 800 nm. As reported previously (Pennington et al. 1985;Schagerl and Donabaum 2003;Roy et al. 2011;Takaichi et al. 2016), the major pigments of cryptophytes are chlorophyll c (Chl c), alloxanthin (Alx), monadoxanthin (Mon), crocoxanthin (Cro), chlorophyll a (Chl a), and α-carotene (α-Car). Pigments were identified based on the characteristic absorption peaks of their absorption spectra and elution profiles (Wright and Jeffrey 2006;Roy et al. 2011). Three major pigments (Chl a, Chl c, and Alx) were purified by HPLC from pigment extracts of C. placoidea thylakoid membranes and used as standards to measure the pigment contents in the samples based on the absorption peak area of the HPLC profile.

Sequence analysis of PSI-ACPI from C. placoidea
Total RNA was extracted from C. placoidea and subjected to transcriptome sequencing by BioMarker (BMK). Sequencing libraries were generated using a NEBNext Ultra RNA Library Prep Kit for Illumina (NEB, MA, USA) following the manufacturer's recommendations. The first-strand cDNA was synthesized using random hexamer primers and RNase H, and the second-strand cDNA was synthesized using DNA polymerase I and RNase H. Terminal repair, A-tailing, and adapter addition were performed to prepare the cDNA for hybridization. The cDNA fragments ∼240 bp in length were selected using the AMPure XP system (Beckman Coulter, Beverly, USA). The cDNA was treated with USER Enzyme (NEB), and PCR was performed to obtain the final cDNA library. After clustering of the index-coded samples, the library preparations were sequenced on an Illumina HiSeq 2000 platform, and paired-end reads were generated. The transcriptome was assembled de novo based on the left.fq and right.fq using Trinity (Grabherr et al. 2011). Sequences of the PSI core and ACP subunits were identified via comparison to sequences in the National Center for Biotechnology Information databases. The sequences of the PSI core subunits are completely identical to those in C. placoidea strain CCAP978/8.

Phylogenetic analysis
Sequence alignments in Supplemental Figs S7 and S12 were performed with CLC Sequence Viewer 8.0 and ESPript 3.0 (Robert and Gouet 2014). Protein sequences used to produce phylogenetic tree in Supplemental Fig. S15 were imported in MEGA X and aligned with MUSCLE (default parameters; Supplemental File S1). The alignment was used to produce a phylogenetic tree with MEGA X (Kumar et al. 2018) (Supplemental File S2). The evolutionary history was inferred using the Neighbor-Joining method (Saitou and Nei 1987). The optimal tree with a sum of branch length = 13.99626746 is shown. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1,000 replicates) are shown next to the branches (Felsenstein 1985). The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the Poisson correction method (Zuckerkandl and Pauling 1965) and are expressed as the number of amino acid substitutions per site. This analysis involved 38 amino acid sequences. All positions with <95% site coverage were eliminated, i.e. fewer than 5% alignment gaps, missing data, and ambiguous bases were allowed at any position (partial deletion option). The final dataset contained 107 positions.

Cryo-EM data collection
An aliquot of 4 μL of PSI-ACPI sample at a Chl concentration of 2.0 mg mL −1 was applied to a freshly glow-discharged holey carbon grid (Quantifoil Au R2/1, 200 mesh) with continuous carbon support. The grid was blotted for 2 s at 100% humidity at 8 °C with a force level of 2 and immediately plunged into liquid ethane cooled by liquid nitrogen with Vitrobot Mark IV (Thermo Fisher, USA). The grids were loaded into a 300 kV Titan Krios G 3i microscope (Thermo Fisher) equipped with a K3 BioQuantum direct electron detector (Gatan, USA) for data acquisition. A total of 9,688 movie stacks were automatically recorded using EPU (Thermo Fisher) (Thompson et al. 2019) at a total dose for a stack of 50 e − Å −2 in a defocus range of −1.0 to −1.8 μm. A superresolution mode was used at a nominal magnification of ×81,000 corresponding to a pixel size of 0.53 Å with the energy filter slit set to 20 eV.

Data processing
All movie stacks were corrected by MotionCor2.1 (Zheng et al. 2017) with dose weighting. CTF parameters for each movie were estimated by CTFFIND-4 (Rohou and Grigorieff 2015). Image processing was mainly performed using cryoSPARC 3.1.1 (Punjani et al. 2017). After automatic particle picking and reference-free 2D classifications, 211,375 particles were selected, with obvious junk excluded from the particle set. The selected particles were used as templates for the template-picking procedure in cryoSPARC. The template-picked particles were processed by reference-free 2D classifications to remove bad particle images, and 582,601 particles were selected. After combining the 2 particle sets and removing duplicated particles, the remaining 412,630 particles were 3D classified into 5 classes, among which 2 classes with 118,810 and 133,521 particles, respectively, were subjected to homogeneous refinement. After 3D nonuniform refinement and sharpening, global (per-group) CTF refinement and local (per-particle) CTF refinement were performed. The overall resolutions of the maps of cryptophyte PSI-11ACPI and PSI-14ACPI were 2.71 and 2.66 Å, respectively. To improve the resolution of the cryo-EM density maps, particle subtraction was performed, followed by local refinement targeting the core complex and peripheral LHCIs (distinguished by LHCI-a, LHCI-b, and LHCI-c), resulting in final resolutions of 2.60, 2.98, 2.88, and 2.93 Å for PSI-11ACPI and 2.53, 2.76, 2.77, and 2.83 Å for PSI-14ACPI. The resolution was estimated based on the gold-standard Fourier shell correlation 0.143. The local resolution of the cryo-EM density map was generated using ResMap (Kucukelbir et al. 2014).

Model building and refinement
For model building of the cryptophyte PSI-14ACPI supercomplex, the structure of red algal PSI-LHCR (PDB: 5ZGB) (Pi et al. 2018) was manually placed and rigid-body fitted into the 2.66 Å resolution cryo-EM map of PSI-14ACPI with UCSF Chimera (Pettersen et al. 2004). The amino acid sequences were then mutated to their counterparts in C. placoidea obtained from transcriptome sequencing, except for PsaM, whose sequence could not be found in the transcriptome sequence data and were modeled using sequences of C. placoidea CCAP978/8 (PsaM: YP_009420403.1), as other PSI core subunits of these 2 strains share the same sequences. PsaR was mutated from PsaR of C. gracilis PSI-FCPI (PDB: 6LY5) (Xu et al. 2020). The subunit Unk1 (chain X in PDB file) was constructed with polyalanines. Due to its lowresolution density map, a suitable sequence could not be identified from transcriptome sequences. All ACPIs were identified based on the best match of the amino acid sequence with the cryo-EM density map and were mutated from red algae LHCRs, except for ACPI-8, which was mutated from diatom FCPI-1 (Xu et al. 2020). De novo model building was performed on the ACPI-S subunit. Chl c was assigned as described in previous reports (Wang et al. 2019;Nagao et al. 2020). Chl a and Chl c were distinguished by the density map corresponding to the phytol chain for Chl a and the planarity of C-18 1 , C-18, C-17, and C-17 1 resulting from the C-18=C-17 double bound for Chl c. Each residue and cofactor was manually checked and adjusted with COOT (Emsley et al. 2010). The geometrical restraints of pigments were generated from the Grade Web Server. To build the model of the PSI-11ACPI supercomplex, the PSI-14ACPI formation was initially fitted into the 2.71 Å resolution cryo-EM map, and the 3 additional ACPIs (9/10/11) were removed. The structures of the PSI-14ACPI and PSI-11ACPI supercomplexes were refined via real-space automatic refinement against the cryo-EM map by Phenix (Adams et al. 2010). Manual correction and automatic real-space refinement were carried out iteratively until the structure matched the cryo-EM density map to the maximum extent.

Accession numbers
The cryo-EM map and atomic coordinates have been deposited in the Protein Data Bank and the Electron Microscopy Data Bank under accession numbers EMD-33659 and 7Y7B for the PSI-14ACPI supercomplex structure and EMD-33683 and 7Y8A for the PSI-11ACPI supercomplex structure.

Supplemental data
The following materials are available in the online version of this article.
Supplemental Figure S1. Preparation and characterization of PSI-ACPI from C. placoidea.
Supplemental Figure S2. Cryo-EM data processing for the PSI-ACPI supercomplex.
Supplemental Figure S3. Evaluation of cryo-EM map quality.
Supplemental Figure S4. Comparison of the structures and pigment sites of 2 types of cryptophyte PSI-ACPI supercomplexes.
Supplemental Figure S5. Cryo-EM density maps and structures of the ACPI-S, Unk1 subunits, and pigment molecules.
Supplemental Figure S6. Comparison of the PSI cores of the cryptophyte C. placoidea, red alga C. merolae and diatom C. gracilis.
Supplemental Figure S7. Comparison of the sequences of PSI-ACPI from C. placoidea (C.p.) with the corresponding sequences from other algae.
Supplemental Figure S8. Location of the PsaR subunit and its interactions with surrounding subunits.
Supplemental Figure S9. Structures of the PsaK, PsaO, and PsaL subunits and their interactions with surrounding subunits.
Supplemental Figure S11. Structures of individual ACPI subunits.
Supplemental Figure S13. Pigment association and location of ACPI-S.

Supplemental
Supplemental Figure S16. Structural comparison of the ACPI structures and between ACPIs and Lhcrs from the red alga C. merolae and FCPIs from the diatom C. gracilis.
Supplemental Figure S17. Location of ACPI-4 viewed from the stromal side (A), interactions of the C-terminal loop of ACPI-7 with ACPI-6 (B), and interactions between ACPI-8 and PSI core subunits at the lumenal side (C-E).
Supplemental Figure S19. The structures of 3 adjacent ACPIs used as the building modules.
Supplemental Figure S20. Structural comparison of PSI-ACPI of the cryptophyte C. placoidea with that of the red alga C. merolae and the diatom C. gracilis.
Supplemental Figure S21. Interactions between ACPIs in the same layer (black circle), between outer ACPIs and inner ACPIs (red circle), and between inner ACPIs and PSI core (blue circle) at the stromal side (A) and lumenal side (B).
Supplemental Figure S22. Comparisons of pigment arrangements in ACPIs with those in red alga Lhcrs and diatom FCPIs.
Supplemental Figure S24. Arrangement of the pigments in PSI-ACPI.
Supplemental Figure S25. EET between ACPIs and from ACPI to the PSI core.
Supplemental Table S1. Cryo-EM data collection, refinement, and validation statistics.
Supplemental Table S3. Comparison of the protein subunits in PSI-LHCI of cryptophytes, cyanobacteria, red algae, diatom, green algae, and land plants.
Supplemental Table S4. Pigment binding sites in ACPIs. Supplemental Table S5. Possible EET pathways from ACPIs to the PSI core.
Supplemental Table S6. EET pathways from inner ACPIs to the PSI core.
Supplemental File S1. Sequence alignment file for the phylogenetic analysis in Supplemental