The PSI in cyanobacteria is monomeric, trimeric or tetrameric, although it mainly exists as trimers in vivo. Almost 20 years ago, the structure of the prokaryotic PSI trimer from the cyanobacterium Thermosynechococcus elongatus was resolved at a high-resolution of 2.5 Å by X-ray crystallography (Fromme et al., 2001; Jordan et al., 2001), a great breakthrough in revealing the detailed molecular organization of PSI. Recently, more prokaryotic PSI structures with special functions from a variety of cyanobacterium species and physiological conditions were resolved by single particle cryo-EM analyses (Figure 1): functional monomeric PSI from T. elongatus (Netzer-El et al., 2018; Çoruh et al., 2021), tetrameric PSI from the heterocyst-forming cyanobacteria Anabaena sp. PCC 7120 (Kato et al., 2019; Zheng et al., 2019), the unique PSI reaction center that binds chlorophyll d in Acaryochloris marina (Hamaguchi et al., 2021; Xu et al., 2021) and the most red-shifted chlorophyll f in Halomicronema hongdechloris (Kato et al., 2020). The physiological functioning and evolutionary significance of the PSI tetramer prevalent in heterocyst-forming cyanobacteria has garnered much attention (Li et al., 2014, 2019; Semchonok et al., 2016). Structural analysis has uncovered unique monomer–monomer interactions in the PSI tetramer. Further, in cyanobacteria, the N-terminal, C-terminal and a middle region of PsaL each show marked variation between the trimer and tetramer of PSI, suggesting the spatial structure change of PsaL may have led to the tetramer formation (Figures 1E,F; Kato et al., 2019; Zheng et al., 2019). That formation might have been crucial for cyanobacteria to adapt to intense radiation, because a tetramer structure can increase the content of PSI-bound carotenoids, which may play a photoprotective role under high light incidence (Li et al., 2019). That study also supports the view that tetrameric PSI may be an intermediate in the evolution of cyanobacteria’s trimeric PSI to the monomeric PSI of plants and algae.

In cyanobacteria, each PSI core binds 12 protein subunits (from PsaA to PsaF, PsaI to PsaM, PsaX) to which 127 cofactors are non-covalently bound. Normally, cyanobacteria’s PSI lacks the peripheral antenna system of eukaryotic algae, but in some cases the hydrophilic phycobilidomes can be used for additional light-trapping antennas (Liu et al., 2013; Watanabe et al., 2014; Chang et al., 2015). Interestingly, under stressful conditions—especially in low-iron environments—cyanobacterial PSI express much iron stress-induced protein A (IsiA) that increase PSI’s effective cross-sectional absorption (Bibby et al., 2001; Boekema et al., 2001; Michel and Pistorius, 2004; Yeremenko et al., 2004). Like the PSII intrinsic antenna subunit CP43, the IsiA monomer has six transmembrane helixes and coordinates 17 chlorophyll molecules (Toporik et al., 2019; Cao et al., 2020). Electron microscopy studies showed that under iron starvation one or two IsiA rings encircled a trimeric PSI core to form a PSI-IsiA supercomplex. The cryo-EM structures of the PSI-IsiA supercomplex from the mesophilic cyanobacterium Synechocystis sp. PCC 6803 and Synechococcus sp. PCC 7942 showed that each PSI trimer is surrounded by 18 IsiA subunits, forming a closed ring (Figure 1D; Boekema et al., 2001; Toporik et al., 2019; Cao et al., 2020). The largest PSI-IsiA supercomplex consists of a PSI core trimer and 43 IsiA monomers having a double-ring shape (Yeremenko et al., 2004).

Unlike cyanobacteria’s PSI trimer or tetramer, eukaryotic PSIs in algae and plants are stably associated with LHC antenna subunits to form PSI-LHCI monomers. In eukaryotes, the crystal structure of PSI-LHCI in plants was first resolved at a 4.4-Å resolution from Pisum sativum (Ben-Shem et al., 2003), a monumental breakthrough for examining vascular plants’ PSI structure; this resolution has since been enhanced to 2.4 Å by X-ray crystallography (Mazor et al., 2015, 2017; Qin et al., 2015; Wang et al., 2021). In plants, each PSI core usually binds four LHCIs consisting of two separate dimers (i.e., proteins encoded by lhca1/lhca4 and lhca2/ lhca3) (Wientjes and Croce, 2011), which together form a semispherical belt attached to the PsaF/PsaJ side. Although plants reportedly have six Lhca genes (lhca1–6), both lhca5 and lhca6 are only expressed at sub-stoichiometric levels (Pan et al., 2020). Unlike the PSI-LHCI in vascular plants, the number of PSI peripheral antennas in eukaryotic algae varies considerably. More than one PSI-LHCI structure may exist among the algae (e.g., Busch and Hippler, 2011; Drop et al., 2011). Recently, the PSI-LHCI structure from a red alga (Cyanidioschyzon merolae) and a green alga (Chlamydomonas reinhardtii) provided strong evidence that, indeed, algae do harbor more abundant forms of PSI-LHCI structures. In C. merolae, a typical thermophilic, acidophilic red alga, each PSI core binds 3 or 5 LHCRs (Antoshvili et al., 2018; Pi et al., 2018), whereas in C. reinhardtii each PSI core can bind 8 or 10 LHCIs (Qin et al., 2019; Su et al., 2019; Suga et al., 2019). These studies further emphasize the great structural variability of PSI-LHCI in eukaryotic algae.

Under environmental selective pressure, eukaryotic algae gradually evolved via phagocytosis, and their PSI also evolved a peripheral antenna system with a stronger light-harvesting ability during the evolutionary process. In the red lineage, algae (Rhodophyta) have similar characteristics to cyanobacteria, namely an absence of flagella and the presence of phycobiliproteins in their plastids (chloroplasts) (Nozaki, 2005); thus, red algae are considered primitive and related to cyanobacteria. Consequently, PSI core subunits in red algae and cyanobacteria are mostly conserved. Nevertheless, compared with cyanobacteria’s PSI, red algal PSI evolved a PsaO subunit at the PsaL/A/K side (Tian et al., 2017), and lost the PsaX. Moreover, red algae evolved chlorophyll a-binding proteins, these solely associated with PSI (Figure 2; Wolfe et al., 1994a, b). Compared to cyanobacteria and vascular plants, red alga has the shortest PsaL structure, whose N-terminal region lacks the three small β-strands connecting PsaL to the core in cyanobacteria and vascular plants (Antoshvili et al., 2018; Pi et al., 2018), while the C-terminal region lacks the short α-helix contributing to cyanobacterial PSI trimer’s formation. This simplified PsaL structure and novel subunits PsaO and Lhcrs attached to PSI core surfaces may explain the formation of a monomeric PSI-LHCR supercomplex. The Lhcrs’ arrangement in red algae is very interesting, occurring on both sides of the PSI core in the PSI-LHCR structure from C. merolae (Pi et al., 2018). Evidently, this arrangement is better for capturing incoming light energy from different directions of the thylakoid membrane and for effectively transmitting it to the reaction center.

Oxyphototrophs in the green lineage mainly include green algae, mosses, and land plants. To date, PSI-LHCI’s structures—from three green algae (C. reinhardtii, Bryopsis corticulans, and Dunaliella salina), a moss (Physcomitrella patens) and two terrestrial plants (P. sativum and Zea mays)—has been determined, providing valuable insights into the evolution of PSI in this lineage (Ben-Shem et al., 2003; Qin et al., 2015, 2019; Su et al., 2019; Suga et al., 2019; Pan et al., 2020; Perez-Boerema et al., 2020; Wang et al., 2021; Yan et al., 2021). For the green algae, the structures of PSI-LHCI in B. corticulans living in marine water and C. reinhardtii living in freshwater are very similar, with only a slight difference: each PSI core binds 10 LHCI in B. corticulans whereas each PSI core can bind 8 or 10 LHCI in C. reinhardtii. Surprisingly, a simplified PSI-LHCI structure has been discovered from salt-tolerant D. salina, which contains only seven core subunits (PsaA–F and PsaJ) and four LHC proteins (Lhca1–4). Compared plants and green alga (C. reinhardtii and B. corticulans) binding with 14–16 subunits, D. salina forms the smallest known PSI-LHCI and may represent an acclimation response to hypersaline environments (Perez-Boerema et al., 2020).

Compared with the PSI core of cyanobacteria and red lineage organisms, green lineage oxyphototrophs evolved two significant membrane subunits: PsaG and PsaH (Figure 2B). The PsaG is homologous to PsaK and situated opposite of PsaK. The PSI-LHCI structure in plants and mosses suggests PsaG serves as an anchor for binding one Lhca at the first Lhca position. PsaH is located at the PsaL pole of the core complex and contains a single transmembrane helix and an additional helix encircling the PsaL subunit. Green lineage oxyphototrophs have retained their PSI monomer configuration albeit governed by the newly emerged PsaH subunit. The plant PSI features an extrinsic 10-kDa subunit PsaN located on the lumenal side close to PsaF, involved in the docking of plastocyanin (Amunts et al., 2007). Green lineage oxyphototrophs lack the PsaX of cyanobacteria and did not evolve the PsaR and PsaS seen in diatoms (Figure 2B).

A key feature of PSIs in green lineage oxyphototrophs is the differing amount of chlorophyll a/b proteins binding around the PSI core. Notably, the LHCI is about 2–3 times larger in unicellular green algae than vascular plants (Germano et al., 2002; Kargul et al., 2003). The structures of PSI-LHCI supercomplexes from C. reinhardtii suggest each PSI core can bind 8 or 10 LHCIs (Su et al., 2019; Suga et al., 2019). Eight LHCIs bind at the PsaF/PsaJ side, arranged into two layers as parallel half-rings. Each layer contains four LHCI proteins arranged in a crescent shape, forming the inner and outer LHCI belt. The additional two LHCIs are located near PsaB, at a similar position to those in the red algal PSI-LHCR. Interestingly, extra peripheral antennas are found at the PSI core’s PsaB side in both red and green algae. Hence, LHCIs tend to be arranged on different sides of the PSI core in eukaryotic algae during the primary endosymbiosis evolution, likely critical for their adaption to shifting light intensities in aquatic environments.

In the green lineage, bryophytes (i.e., liverworts, mosses, hornworts) diverged from the ancestor of seed plants and the structure and function of their PSI-LHCI represents PSI’s evolution from aquatic to terrestrial life (Falkowski, 2006; Iwai and Yokono, 2017). The PSI-LHCI supercomplex from the moss P. patens was structurally similar to land plants’ PSI-LHCI, except Lhca5 in the former instead of Lhca4 in the latter (Busch et al., 2013; Yan et al., 2021). Yet the P. patens genome codes for more diverse and redundant light-harvesting antenna proteins than those of green algae and land plants, suggesting complicated PSI-LHCI structures exist in P. patens (Busch et al., 2013; Iwai and Yokono, 2017). The PSI-LHCI from vascular plants—determined by X-ray crystallography at higher resolutions—is composed of 19 protein subunits and ca. 200 non-covalently bound cofactors (Ben-Shem et al., 2003). Hence, structure of PSI-LHCI gradually stabilized during its aquatic-to-terrestrial evolutionary transition.

Interestingly, in the green lineage, exists a short-term light-adaptation process to regulate the light-harvesting capacity of PSI and PSII, this termed “state transition”. During this process, a portion of LHCII is phosphorylated and dissociated from PSII and moved to PSI, forming a PSI-LHCI-LHCII supercomplex. To date, the molecular structure of PSI-LHCI-LHCII supercomplex has been reported for green algae, mosses, and land plants (Figure 2A; Iwai et al., 2015, 2018; Pan et al., 2018; Pinnola et al., 2018; Huang et al., 2021). The PSI-LHCI-LHCII supercomplex from C. reinhardtii has been determined to a 3.42 and 2.84-Å resolution by cryo-EM (Huang et al., 2021; Pan et al., 2021). Examination of the CrPSI-LHCI-LHCII supercomplex structure showed that each PSI core binds 10 LHCI proteins and two LHCII trimers, the latter forming an additional antenna belt at the PsaH/L/O side. In P. patens, a newly LHCII protein-Lhcb9 associated with PSI was found, forming a large PSI-LHCI-LHCII supramolecular complex, in which each PSI core may bind two rows of LHCI belt, one copy of Lhcb9, and a LHCII trimer (Iwai et al., 2018; Pinnola et al., 2018). The PSI-LHCI-LHCII supercomplex in plants (Z. mays) is smaller than that identified from C. reinhardtii and P. patens and it contains only one LHCI belt (Lhca1-Lhca4) and one LHCII trimer. The LHCII trimer in plants is located at a position similar to that in the PSI-LHCI-LHCII supercomplex of green alga and moss. However, the other LHCII trimer in CrPSI-LHCI-LHCII is specific to green alga and bridges the gap between the conserved LHCII trimer and Lhca2. In summary, compared with PSI of green algae and mosses, which has diverse structure of the peripheral light-harvesting complex, the plants’ PSI-LHCI and PSI-LHCI-LHCI evolved to become more stable, likely because vascular plants face stronger light exposure conditions after moving from an aqueous to terrestrial environment.

Diatoms and brown algae are derived from a secondary endosymbiotic event, and their chloroplasts are surrounded by four membranes (Veith et al., 2009). Recently, the organization and structure of the PSI-FCPI (fucoxanthin Chl a/c-binding proteins) supercomplex from the diatom Chaetoceros gracilis was reported at a 2.38 and 2.40-Å resolution, respectively, by cryo-EM (Nagao et al., 2020; Xu et al., 2020). These findings confirmed the PSI’s organization in red-lineage oxyphototrophs through a secondary endosymbiosis event. Compared that of PSI-LHCR in red algae, the structure of PSI in diatoms evolved two new subunits, PsaR and PsaS, but lost PsaK and PsaO (Figure 2B). PsaR is a unique membrane-spanning subunit phylogenetically close to PsaG in green lineage organisms, but it has a low homology with PsaG. Nevertheless, PsaR’s role may be similar to PsaG’s in these organisms: mediating the binding and energy transfer of FCPI to the PSI core. The diatom PSI core lost the PsaK that exists in all oxyphototrophs and PsaO typical of eukaryotic organisms, which may be related to the unique secondary endosymbiosis position of diatoms. The most striking difference between the red algal and diatom PSIs is the monomeric diatom PSI core being surrounded by 24 FCPI antennas (Nagao et al., 2020; Xu et al., 2020), amounting to 1.1 MDa (molecular weight), the largest known monomeric PSI-LHCI supercomplex of any photosynthetic organisms. Special fucoxanthin (Fx) chlorophyll a/c antenna proteins in diatoms can help to efficiently harvest blue-green light under water (400–550 nm), which is essential for their survival and fitness in aquatic environments (Wang et al., 2020).