The structural basis for light harvesting in organisms producing phycobiliproteins

Abstract

Cyanobacteria, red algae, and cryptophytes produce 2 classes of proteins for light harvesting: water-soluble phycobiliproteins (PBP) and membrane-intrinsic proteins that bind chlorophylls (Chls) and carotenoids. In cyanobacteria, red algae, and glaucophytes, phycobilisomes (PBS) are complexes of brightly colored PBP and linker (assembly) proteins. To date, 6 structural classes of PBS have been described: hemiellipsoidal, block-shaped, hemidiscoidal, bundle-shaped, paddle-shaped, and far-red-light bicylindrical. Two additional antenna complexes containing single types of PBP have also been described. Since 2017, structures have been reported for examples of all of these complexes except bundle-shaped PBS by cryogenic electron microscopy. PBS range in size from about 4.6 to 18 mDa and can include ∼900 polypeptides and bind >2000 chromophores. Cyanobacteria additionally produce membrane-associated proteins of the PsbC/CP43 superfamily of Chl a/b/d-binding proteins, including the iron-stress protein IsiA and other paralogous Chl-binding proteins (CBP) that can form antenna complexes with Photosystem I (PSI) and/or Photosystem II (PSII). Red and cryptophyte algae also produce CBP associated with PSI but which belong to the Chl a/b-binding protein superfamily and which are unrelated to the CBP of cyanobacteria. This review describes recent progress in structure determination for PBS and the Chl proteins of cyanobacteria, red algae, and cryptophytan algae.

Introduction

Cyanobacteria, red algae, and cryptophytes produce 2 types of proteins for light harvesting: phycobiliproteins (PBP) and chlorophyll-binding proteins (CBP). PBP are a superfamily of water-soluble proteins found in a subset of oxygenic phototrophs, specifically in most members of the cyanobacteria and in members of eukaryotic rhodophyte, glaucophyte, and cryptophyte algae (Sidler 1994; Bryant and Canniffe 2018). PBP owe their brilliant colors to linear tetrapyrrole chromophores known as bilins (Ledermann et al. 2017; Bryant et al. 2020a), and they principally act as light-harvesting antenna complexes (LHC) for PSII, which has fewer chlorophyll (Chl) a molecules (∼35 Chl a) than PSI (∼90–95 Chl a molecules) (Jordan et al. 2001; Umena et al. 2011; Kato et al. 2022a; Gisriel 2024). However, under some conditions, energy can also be transferred from PBP to PSI (Dong et al. 2009). The ratio of energy transferred to PSI or PSII can be adjusted by cells to optimize the overall rate of photosynthesis and thereby the growth rate (Ho et al. 2017b).

Published studies on PBP began early in the 19th century when Esenbeck (1836) described a brilliantly blue-colored, red-fluorescent, photo-labile, water-soluble pigment, “saprocyanin,” that could be released by the cyanobacterium Oscillatoria sp. Kützing (1843) renamed this pigment “phykokyan,” meaning algal blue pigment (Greek: phykos = seaweed; kyanos = sky blue). Kützing additionally isolated a water-soluble red pigment, “phykoerythrin” (Greek: phykos = seaweed erythros = red), from several red algae. Sorby (1877) employed differential heat denaturation in sucrose solutions of algal extracts to provide the first evidence for a third major PBP, which he named allophycocyanin (AP) (Greek: allos = other, hence other algal blue pigment). Tandeau de Marsac (2005) has summarized many other details concerning the long history of studies of this important protein superfamily and early observations on the complexes they form.

PBP descend from a single gene encoding an ancestral bilin-binding antenna protein, which evolved into the current superfamily through many rounds of gene duplication and divergence (Sidler 1994; Apt et al. 1995; Zhao and Qin 2006). Major PBP subfamilies include red-colored phycoerythrins (PE; λ_max ∼560 nm), fuchsia-colored phycoerythrocyanins (PEC; λ_max ∼590 nm), blue-colored phycocyanins (PC; λ_max ∼620 nm), and aqua-colored allophycocyanins (AP; λ_max ∼650–710 nm) (Glazer 1989; Bryant and Canniffe 2018). Most of these subfamilies contain variants, such as allophycocyanin-B (AP-B) or Rhodophytan PC (R-PC), that can have specialized functions in assembly, light-harvesting, and/or energy transfer to the 2 photosystems or specifically occur in either cyanobacteria (e.g. C-PE) or rhodophytes (e.g. R-PE). The fundamental structural unit of all PBP is a heterodimeric (αβ) protomer, commonly referred to as the “monomer.” Monomers comprise 1 member of the α subunit family and 1 subunit from the β subunit family (Bryant and Canniffe 2018). Each subunit carries at least 1 and up to 3 bilin chromophores, which are typically bound covalently to the protein through a single thioether linkage to the side chain of a conserved Cys residue (Sidler 1994; Ledermann and Aras 2017). Depending on the protein concentration and the physicochemical conditions, protomers can oligomerize to form toroid-shaped (αβ)₃ trimers or (αβ)₆ hexamers (Bryant and Canniffe 2018), although a unique, far-red–absorbing variant of AP that forms helical nanotubes has recently been described (Gisriel et al. 2023b; see below). In the presence of assembly-directing proteins known as linker proteins, some of which are pigmented and also known as γ subunits in red algae, toroids of most PBP can form cylindrical stacks that can further oligomerize to form multiprotein complexes known as phycobilisomes (PBS) (Glazer 1989; Sidler 1994; Bryant and Canniffe 2018). In red algae the linker protein gene family has undergone extensive duplications and diversification to expand this family of proteins, which probably has contributed to the enlargement and modification of red algal PBS (Lee et al. 2019).

Six structural classes of PBS have been described since their initial description in red algae in the 1960s (Gantt and Conti 1965; 1966a, 1966b; 1968). They include structures described as hemiellipsoidal (Gantt and Conti 1965, 1966a, 1966b; Gantt and Lipschultz 1972), block-shaped (Gantt and Lipschultz 1980), hemidiscoidal (Gantt and Conti 1968; Bryant et al. 1979), bundle-shaped (Guglielmi et al. 1981), paddle-shaped (Jiang et al. 2023), and bicylindrical far-red light (FRL) AP cores (Gisriel et al. 2024) (Fig. 1). Two additional PBP complexes formed from single types of PBP have also been described: rod-shaped CpcL-PC-FNR complexes (Watanabe et al. 2014; Zheng et al. 2023), and a FRL-absorbing AP (ApcD4-ApcB3) that assembles as helical nanotubes with the Chl a-binding protein IsiX (Gisriel et al. 2023b). Cyanobacterial hemidiscoidal PBS have a core substructure comprised of 2, 3, 5, or even 7 cylindrical stacks of AP trimers together with some minor AP variants (Bryant and Canniffe 2018; Jiang et al. 2023). The core is usually surrounded on 2 sides by 6 to 8 peripheral rods, which are cylindrical stacks of PC together with PE or PEC if present (Bryant and Canniffe 2018). Rod-shaped CpcL-PC-FNR complexes are smaller, only ∼840 kDa with ∼54 bilin chromophores (Fig. 1), and they associate with PSI by the CpcL rod linker protein that has a C-terminal transmembrane α-helix (Watanabe et al. 2014; Zheng et al. 2023). Hemidiscoidal PBS, which can range in size from about ∼4.5 to ∼7 mDa and carry 300 to 500 bilin chromophores, are the most common class of PBS in cyanobacteria (Bryant et al. 1979; Zheng et al. 2021). The largest PBS are the block-shaped and hemiellipsoidal PBS of red algae, which can have masses of up to ∼14–18 mDa and carry more than 2000 bilin chromophores (Zhang et al. 2017; Ma et al. 2020; Dodson et al. 2023). The larger block-shaped and hemiellipsoidal PBS of red algae have up to 14 peripheral rods, and literally all nooks and crannies are filled by PE subunits.

Figure 1.

Types of PBS and PBP complexes in cyanobacteria and red algae. The figure provides the source organism, the number and type(s) of bilin chromophores, the approximately molecular mass, and the dimensions. For the bundle-shaped PBS of G. violaceus, a cartoon scheme is shown because no detailed structure is yet available. The gray circles for the paddle-shaped PBS of A. panamensis represent 4 mobile PC hexamers whose exact positions have not been determined. The images for PBS and PBP complexes were taken from the following PDB files (https://www.rcsb.org/): hemiellipsoidal, 7EXT; PDB 6KGX; block-shaped, PDB 5Y6P; hemidiscoidal, paddle-shaped is a composite of PDB 8IMI, 8IMJ, 8IMK, 8IML, 8IMM, 8IMN, and 8IMO; bicylindrical core FRL-AP, 8UHE; rod-shaped CpcL-PC-FNR complex, 8HFQ; and helical FRL-AP, 8DDY. For the sources of the other information, see text.

In addition to antenna complexes made up from PBP, cyanobacteria and PBP-producing algae also utilize antenna complexes that are made of Chl-binding proteins. In cyanobacteria, iron starvation induces the formation of the Chl a-binding protein, IsiA, which forms rings or arcs that surround PSI complexes (Guikema and Sherman 1983; Burnap et al. 1993). Plants and green algae do not produce PBP, but they produce a wide range of LHC that bind carotenoids, Chl a, and Chl b (CAB superfamily) and that form supercomplexes with PSI and PSII (Green et al. 1991; Green and Durnford 1996). In the eukaryotic algae that produce PBP, PSI complexes primarily associate with Chl a-binding LHCI proteins, which are members of the CAB superfamily, rather than PBS to increase the absorbance cross-section of PSI.

Beginning about 10 years ago, the resolution revolution that empowered biological structure determination by cryogenic transmission electron microscopy (cryo-EM) began to transform structural biology profoundly (Kühlbrandt 2014). Photosynthesis research is a field that was poised to benefit enormously from this technique for several reasons: (1) photosynthetic complexes can easily be isolated in large amounts; (2) photosynthetic complexes are mostly membrane associated, and though well characterized biochemically and spectroscopically, they are very difficult to crystallize for traditional x-ray crystallography; and (3) photosynthetic complexes have sizes and shapes that facilitate particle alignment for structure determination. These properties have facilitated numerous studies investigating the structures of photosynthetic proteins in recent years. This review will briefly summarize progress in the determination of antenna structures in oxygenic phototrophs that produce PBP. The determination of high-resolution structures for the enormous block-shaped PBS of red algae, which have nearly 1000 protein subunits and more than 2000 bilin chromophores, is an especially noteworthy achievement (Sui 2021), but other advances are also quite notable as well. This review will focus on recent structural advances for PBS and PBP complexes but will mostly exclude recent structural studies on reaction centers, which will be described elsewhere (Gisriel 2024). A recent review (Shen 2022) has described Chl-binding antenna structures associated with PSI supercomplexes in a number of organisms, some of which (e.g. diatoms, green algae) are not included here. The following topics are addressed:

• Cyanobacteria and red, glaucophyte, and cryptophyte algae utilize a variety of PBP complexes for light harvesting but also utilize CBP for light harvesting.

• Six structural classes of PBS have been described: hemiellipsoidal, block-shaped, hemidiscoidal, bundle-shaped, paddle-shaped, and FRL allophycocyanin-containing bicylindrical cores; 2 additional PBP antenna structures have been reported. Relatively high-resolution structures are presently available for examples of all 8 except bundle-shaped.

• The hemiellipsoidal and block-shaped PBS of red algae have similar structures with masses of 13 to 18 mDa, comprising 700 to 900 individual protein subunits and 1500 to 2000 bilin chromophores.

• Hemidiscoidal PBS are the most common type found in cyanobacteria. They have masses of 4.5 to 7.5 mDa, comprising about 200 to 300 polypeptides and 300 to 500 bilin chromophores.

• The relict, paddle-shaped PBS of Anthocerotibacter panamensis have unique heptacylindrical, allophyocycanin cores that are surrounded by 16 phycocyanin hexamers, none of which form rod-like stacks as observed in other PBS.

• Although PBS mostly harvest energy for PSII, rod-shaped and FRL-absorbing allophycocyanin helical nanotubes together with IsiX act as antenna complexes for PSI.

• Cyanobacteria also have membrane-intrinsic, Chl a-, b-, and/or d-binding antenna proteins (CBP) that can be stress-induced (e.g. iron stress–induced IsiA) or constitutively expressed when PBP are reduced or absent. These CBP belong to the CP43/PsbC protein superfamily that also includes PsbC and PsbB of PSII.

• Red algae and cryptophytes have membrane-intrinsic antennae proteins that bind Chls that serve as LHC for PSI. However, these proteins belong to the Chl a/b (CAB) superfamily that is structurally unrelated to the CP43/PsbC superfamily employed in cyanobacteria.

Phycobilisomes

PBS were first described in ultrastructural studies by Gantt and Conti as thylakoid-associated granules in chloroplasts of the unicellular red algae Porphyridium cruentum (Gantt and Conti 1965, 1966a, 1966b) and Porphyridium aerugineum (Gantt and Conti 1968). The demonstration that these granules were assemblages of PBP followed their initial observation (Gantt and Conti 1966a, 1966b). Shortly thereafter, PBS were shown to occur on the stromal surfaces of the thylakoids of several cyanobacteria (Gantt and Conti 1969; Edwards and Gantt 1972; Wildman and Bowen 1974). Lefort-Tran (Lefort 1965; Bourdu and Lefort 1967) was the first to observe similar thylakoid-associated granules on cyanelle thylakoids of glaucophytes, including Cyanophora paradoxa. Interestingly, cryptophytan PBP do not assemble into PBS; instead, these PBP are localized in the thylakoid lumen, but their organization remains the subject of speculation (Gantt et al. 1971; Spear-Bernstein and Miller 1989; Rathbone et al. 2023). Soon after PBS were first identified in cells by transmission electron microscopy of thin sections, methods were developed for the isolation of most of the major structural classes of PBS, and the terms used to describe their structures were introduced: hemispherical/hemiellipsoidal (Gantt and Lipshcultz 1972, 1974), hemidiscoidal (Gantt and Conti 1969; Koller et al. 1977, 1978; Bryant et al. 1979), block-shaped (Gantt and Lipshultz 1974), and bundle-shaped (Guglielmi et al. 1981) (Fig. 1). Rod-shaped CpcL-PC-FNR complexes and helical FRL-AP-IsiX complexes were described later (Watanabe et al. 2014; Gisriel et al. 2023b), and paddle-shaped PBS have only very recently been described (Jiang et al. 2023). Remarkably, in the past 8 years, relatively high-resolution structures of all types of PBS and PBP complexes except the bundle-shaped PBS of Gloeobacter violaceus have been solved by single particle analysis by cryo-EM and/or cryogenic electron tomography (cryo-ET) (Fig. 1).

Hemiellipsoidal PBS: P. cruentum

As noted above, PBS were discovered in ultrastructural studies of the unicellular red alga, P. cruentum, by Gantt and Conti (1965, 1966a, 1966b). Porphyridium purpureum has been widely studied more recently, but P. cruentum is probably the same or at least a very closely related species (Li et al. 2022a). Historically, P. cruentum was the subject of many studies in which the PBS constituents were purified and characterized (e.g. Glazer and Hixson 1975; 1977; Ley et al. 1977), and its PBS were studied by electron microscopy numerous times (e.g. Arteni et al. 2008). Thus, it was fitting that one of the first PBS to have its structure determined by cryo-EM at reasonably high resolution is the PBS of P. purpureum. In 2020, the structure of the 14.7-MDa, hemiellipsoidal PBS of P. purpureum was reported at a global resolution of 2.82 Å, although the core region had a somewhat higher resolution of 2.68 Å (Fig. 2; Ma et al. 2020). This PBS has approximate dimensions of 610 Å (length) by 390 Å (height) by 380 Å (thickness), accounting for the overall hemiellipsoidal outline. Because of the relatively high resolution and in combination with previous biochemical studies of the proteins, the identities of all 1598 bilins could be assigned unambiguously. There are 120 phycocyanobilin, 1430 phyerythrobilin, and 48 phycourobilin chromophores in the P. purpureum PBS (Fig. 1). The structural model includes 706 protein subunits comprising 528 PE subunits, 72 PC subunits, 46 AP subunits, and 60 linker proteins (including γ-subunits of PE, which are chromophorylated L_R linkers). The phycourobilin content of the P. purpureum PBS is much lower than that for Griffithsia pacifica PBS (Zhang et al. 2017), because phycourobilins are only bound to γ-subunits of PE in the former case, while all PE β-subunits in the latter case additionally bind both phycourobilin and phycoerythrobilin. This is presumably an adaptation to the enrichment in green light in the niches where G. pacifica normally occurs (Xie et al. 2021).

Figure 2.

Structure of the hemiellipsoidal PBS of P. purpureum (PDB 6KGX). A) face and side views of the PBS showing the PE containing peripheral rods as solid cylinders that are labeled Ra/Ra′ through Rg/Rg′. Individual PE hexamers are labeled Ha/Ha′ to Hd/Hd′. Note that some PE rods (e.g. Rf/Rf” and Rg/Rg’) are not attached directly to the AP-containing core (blue; dashed oval) but are attached to PE in other peripheral rods (red arrows). B) Scheme showing the overall structural organization of the P. purpureum PBS. Adapted with permission from Extended Data Fig. 4 of Ma et al. (2020), copyright 2020 Springer Nature.

The core of the P. purpureum PBS is comprised of 20 ApcA, 22 ApcB, 2 ApcD, 2 ApcF, and 2 ApcE polypeptides, which combine to form 3 cylinders (Ma et al. 2020). Each of the 2 lower cylinders of the core are stacks of 3 trimers, arranged as a face-to-face hexamer combined with a trimer. The upper core cylinder that is bound by the C-terminal REP (repeat; Pfam00427) domain of the 2 ApcE polypeptides in the core is formed from a pair of back-to-back AP trimers. Thus, the cores of red algal PBS are smaller than the cores found in most cyanobacterial hemidiscoidal PBS because they are missing 4 AP trimers (see below). The AP core is surrounded by 14 peripheral rods. Three peripheral rods are bound on each side of the core in a pattern reminiscent of many hemidiscoidal PBS (see below). Each of these 6 rods terminates at its core-proximal end with a hexamer of R-PC, a variant of PC in which the β−153 phycocyanobilin is replaced by phycoerythrobilin (Glazer and Hixson 1975; Ducret et al. 1994). Eight of the peripheral rods contain only PE, and 4 of these are attached directly to the lower core cylinders. Additionally, 8 PE hexamers are individually attached to the peripheral rods that only contain PE. The peripheral rod organization of the P. purpureum PBS is surprisingly similar to that of the block-shaped PBS of G. pacifica (Zhang et al. 2017), except that the rods generally are 1 PE hexamer shorter in the P. purpureum PBS. This difference reduces the overall width of the PBS and contributes to the hemiellipsoidal appearance. A curious feature of these PBS is that there are 2 individual (αβ) monomers and 10 PE β-subunits (PebB) distributed throughout the structure, which fill in what would otherwise be empty spaces among the rods, core, and the extra PE hexamers. Nature seems to stuff all available spaces with chromophore-binding polypeptides in these PBS.

P. purpureum exhibits photoacclimation to changes in the growth light intensity. When grown in medium light (350 µmol photons m⁻² s⁻¹), the cells are a pale pinkish color, but when cells are grown in low light (50 µmol photons m⁻² s⁻¹), the cells have a deeper bordeaux color (purplish-red, hence the species epithet, purpureum) (Dodson et al. 2023). Consistent with the appearance of the cells to the eye, cells grown at low light have fourfold more Chl a per cell and much more PBP per cell as well. Additionally, the absorbance and fluorescence properties of the cells were slightly different from cells grown at a higher irradiance level. The PBS from cells grown at higher light intensity were smaller than those grown under low-light intensity: 13.6 vs 14.7 mDa. Specifically, 4 PE hexamers found at the ends of 2 peripheral rods from low-light grown cells were missing in the PBS from high-light grown cells (Dodson et al. 2023). This change reduces the total bilin content of the PBS by ∼8.8% (132 phycoerythrobilins and 8 phycourobilins). Spectroscopic measurements indicated that the increased pigment content did not compromise energy transfer efficiency. The PBS from low-light-grown cells reveal subtle differences in the number of closely coupled chromophores, which might offset any loss of efficiency brought about by increasing the number of chromophores in the PBS (Dodson et al. 2023).

Cryo-ET is a powerful method for studying native protein complexes in their cellular setting (Young and Villa 2023). The photosynthetic apparatus of red algal chloroplasts is very highly ordered and is an ideal subject for cryo-ET after cryo-focused ion-beam milling. Li et al. (2021) applied this method to study the chloroplast thylakoid membranes of P. purpureum. They obtained structural models for a native PBS-PSII supercomplex at 14.3-Å resolution and for a larger PBS-PSII supercomplex containing 2 PBS-PSII supercompexes at 15.6 Å resolution. Because of its large size, the PBS interacts closely with a PSII tetramer (i.e. 2 PSII dimers) as well as PSII monomers from interspacing PSII dimers from either side of the tetrameric units; in the larger supercomplex with 2 PBS, spacer PSII dimers are inserted before and after the PSII tetramer that interacts with 1 PBS. Evidence for 3 previously undescribed connector proteins that link PBS to PSII was obtained. The structural complexity suggests that there will likely be several energy transfer pathways from the PBS to PSII and between PSII dimers as well.

In a more recent cryo-ET study, You et al. (2023) reported the structures of 2 types of supercomplexes from P. purpureum that also contain PSI and its LHCI antenna complexes (Fig. 3). The structure of a single PBS-PSII tetramer flanked by 2 PSI-LHCI complexes was solved at a global resolution of 3.3-Å resolution. Some of the complexes that had originally been modeled as PSII dimers in fact turned out to be PSI-LHCI complexes when higher resolution was achieved, which clearly demonstrates that the interpretation of low-resolution structures must be considered with skepticism. This larger complex has dimensions of 550 Å (length) by 590 Å (width) by 500 Å (height) and has a mass of about 16.7 mDa. A double PBS-PSII-PSI-LHCI complex was also determined at a global resolution of 4.3 Å. This complex contains 2 PBS, 10 PSII dimers (2 PSII tetramers with a spacer PSII dimer between them), and 2 PSI-LHCI complexes at both ends of the complex. The PSI complexes are monomeric and there are 8 LHCI subunits that surround that portion of the PSI monomer complex that is not directly involved in binding to PSII. This larger complex has a mass of more than 33 mDa and has the same height as the single complex (500 Å) but is 900 Å long and 620 Å wide. Amazingly, the PBS contains 1434 phycoerythrobilin, 48 phycourobilin, and 120 phycocyanobilin chromophores; each PSII dimer contained 70 Chl a, 4 heme molecules, 20 carotenoids, and 54 lipid molecules; and each PSI-LHCI complex contained 187 Chl a, 21 carotenoids, 2 zeaxanthins, and 4 lipids. Thus, the single PBS-PSII-PSI-LHCI complexes have more than 2100 bilin and Chl chromophores for light harvesting, and the double complex has nearly 4300 bilin and Chl chromophores. The previously identified “connector” molecule is associated with the stromal surface of PSII and plays a role in PBS binding to PSII. It binds in close proximity to the PsbY subunit of PSII, which had previously only been observed in cyanobacteria; PsbY is also bound near the connector and Cyt b₅₅₉ (PsbE-PsbF) (You et al. 2023).

Figure 3.

Arrangement of PBS, PSI, and PSII in P. purpureum chloroplast thylakoid membrane. A) Model at left showing the PBS-[LHCI]₂-[PSII]₄ supercomplex (PDB 7Y5E) from P. purpureum. Model at right showing [PBS]₂-[LHCI]₂-[PSII]₁₀ supercomplex from P. purpureum.B) Schematic diagram showing additional details of the [PBS]₂-[LHCI]₂-[PSII]₁₀ supercomplex (PDB 7Y7A) from P. purpureum. Adapted with permission from You et al. (2023), Figure 1, copyright 2023 Springer Nature.

The 2 PBS in the PBS-PSII-PSI-LHCI supercomplexes make direct contact with PSII but do not make direct contact with PSI nor with one of the PSII dimers. Four proteins, L_RC2, L_RC3, L_PP1, and L_PP2, stabilize the binding of the PBS to PSII (You et al. 2023). The first 2 proteins are rod-core linkers in the PBS, while the latter 2 proteins are novel PBS-PSII linkers. The C-terminal regions of L_RC2 and L_RC3 have extensive binding interactions with the stromal surface of PSII. The N terminus of L_PP1 contains PSI and PSII binding motifs, while the C-terminal binds to the PBS. L_PP2 is a homolog of ApcG, a recently identified core linker in the Synechocystis sp. PCC 6803 (hereafter Synechocystis 6803) PBS (see below), and this protein binds to both the membrane-facing side of the PBS as well as the stromal surface of PSII. The PBS core associates with 1 PSII monomer of the dimer via L_RC2 and L_PP1 and binds to the second monomer of PSII in the dimer via L_RC3 and L_PP2.

In cyanobacteria, ApcD is required for efficient energy transfer from PBS to PSI (Zhao et al. 1992), and cross-linking studies suggest that it comes in close proximity to PSI in PBS-PSII-PSI supercomplexes (Liu et al. 2013). The role of ApcD in energy transfer in red algae may be different because the PBS core makes no direct contact with PSI, at least not in the 2 types of complexes characterized by You et al. (2023). Instead, it appears that energy transfer to PSI in red algae proceeds by a spillover mechanism from the PBS to PSII to PSI. This would also presumably be true for the spacer PSII dimers that do not make direct contact with the PBS. These structural findings are largely in agreement with the model for absorbance cross-section and energy transfer processes described for red algae (e.g. Butler 1978; Ley and Butler 1980).

Block-shaped PBS: Griffithsia pacifica

Block-shaped PBS were isolated from the red alga G. pacifica, and their structure was determined by single-particle cryo-EM to a global resolution of 3.5 Å (Zhang et al. 2017). G. pacifica typically lives at depths of ∼20 m, where green light (∼498 nm) is most prevalent, whereas P. purpureum/P. cruentum grows near the water surface (Xie et al. 2021). G. pacifica PBS are larger than the hemiellipsoidal PBS of P. cruentum (Ma et al. 2020), having a mass of ∼18.0 mDa; they are approximately 680 Å (length) by 390 Å (height) by 450 Å (thickness). This larger size results because the peripheral rods in the PBS of G. pacifica are generally longer by 1 PE hexamer and because of other minor differences such as differences in linker protein content because of the longer peripheral rods (see below). As a result of the longer peripheral rods, profile views from the top or bottom of these PBS have an overall rectangular outline; hence their description as “block-shaped.” In addition to being larger and having a larger absorbance cross-section because of the presence of more bilin chromophores, some of the phycoerythrobilin chromophores of PE in the G. pacifica PBS are replaced by phycourobilin, which absorbs maximally at 495 nm. This adaptation allows the absorbance of the G. pacifica PBS to better match the blue-green light available in the niche occupied by this alga.

The G. pacifica PBS comprises 862 polypeptides and 2048 bilin chromophores. There are 46 AP (α+β) subunits, 528 PE (α+β) subunits, and 72 (α+β) PC subunits in the peripheral rods; 144 PE subunits in individual hexamers and subunits; and 70 linker proteins to provide the scaffold for assembly. The AP core is a simplified version of the cores found in hemidiscoidal PBS (see below), and its structure is essentially identical to that in P. purpureum described above. The overall arrangement of the peripheral rods in the G. pacifica PBS is very similar to that in the P. purpureum PBS. As in the P. purpureum PBS, there are 14 peripheral rods, 6 of which terminate with an R-PC hexamer proximal to the core. The other peripheral rods contain only PE and linker proteins to join them to the PBS. Additionally, there are 10 individual PE hexamers as well as some individual PE α and β subunits distributed throughout the periphery of these PBS. Interestingly, these rods, hexamers, and subunits containing only PE do not make direct contact with cores or with PC, so their energy transfer to the cores is dependent on contacts with PE in other rods. In total, 12 additional PE hexamers are added to the structure—1 hexamer is added to each of 6 peripheral rods on both sides of the PBS.

Hemidiscoidal PBS: Synechococcus sp. PCC 7002, Nostoc sp. PCC 7120, Synechocystis 6803, and Thermosynechococcus vulcanus

The first studies on hemidiscoidal PBS, which were incorrectly described as “disc-shaped,” were conducted with the red alga Rhodella violacea (e.g. see Koller et al. 1977, 1978; Mörschel et al. 1977). Obviously, discs have a circular outline (e.g. computer discs, dinner plates, flying saucers, etc.), but these PBS have a semicircular outline and are thus hemidiscoidal. R. violacea PBS have a tricylindrical AP core surrounded on 2 sides by 3 peripheral “tripartite units,” which were shown to contain what can now be described as 1 (αβ)₆ hexamer of PC and 2 (αβ)₆ hexamers of PE (Koller et al. 1978). The nature of the AP core in these PBS was more poorly characterized, and it is not clear how many (αβ)₃ trimer equivalents make up each of the 3 cylinders in R. violacea. Shortly after the initial studies on R. violacea, the structures of hemidiscoidal PBS from several cyanobacteria were described by negatively staining isolated PBS and imaging them using transmission electron microscopy (Bryant et al. 1979; Glazer et al. 1979). Interestingly, the cores of PBS from Synechococcus sp. PCC 6301 (hereafter Synechococcus 6301; closely related to Synechococcus sp. PCC 7942; hereafter Synechococcus 7942) were found to have only 2 core cylinders, whereas PBS from the other cyanobacteria examined at that time had cores assembled from a pyramidal stack of 3 AP cylinders (Bryant et al. 1979; Glazer et al. 1979). Somewhat later, it was recognized that some cyanobacterial PBS actually have “pentacylindrical” cores (Glauser et al. 1992; Ducret et al. 1996, 1998). These structures have 2 half cylinders of AP added to a tricylindrical AP core. It was also recognized that the overall structure of hemidiscoidal PBS was determined by the domain structure of ApcE, the core-membrane linker PBP (L_CM). ApcE has an N-terminal domain that assembles as an α-type PBP subunit in each of the 2 lower cylinders of PBS cores. The C-terminal domain of ApcE has either 2, 3, or 4 REP (Pfam00427) domains. These domains can stabilize the binding of 2 PBP trimers to form an (αβ)₆ hexamer. When ApcE has only 2 REP domains, the resulting PBS has only 2 core cylinders, as in Synechococcus 6301/7942 (Glazer et al. 1979; Bryant and Canniffe 2018). When ApcE has 3 REP domains, the resulting PBS has 3 core cylinders arranged as a pyramidal stack, as in Synechococcus sp. PCC 7002 (hereafter Synechococcus 7002) PBS (Zheng et al. 2021). When ApcE has 4 REP domains, the resulting PBS has a pentacylindrical core (as in Noctoc sp. PCC 7120 (hereafter Nostoc 7120) or Mastigocladus laminosus PCC 7603 (Glauser et al. 1992; Ducret et al. 1998).

Over a period of about 20 years, Glazer and coworkers used classical biochemical methods to dissect the structure of the hemidiscoidal PBS from Synechococcus 6301 and Synechocystis sp. PCC 6701, the results of which were nicely summarized in reviews (e.g. Glazer 1985, 1989). Using genetic approaches, including the characterization of null mutants for structural subunits of PBS and site-specific mutations, others identified the components of the PBS of Synechococcus 7002 (summarized in Bryant 1991) and other cyanobacteria. Only ApcG was not identified in this way (see Domínguez-Martín et al. 2022 and below). Although the structural models built from biochemical and genetic analyses did not have the resolution of structures determined by cryo-EM, it is important to note that the models built from biochemical and genetic methods have recently been confirmed by structural analyses at high resolution by cryo-EM.

Chang et al. (2015) reported the first low-resolution (21 Å) model for the hemidiscoidal PBS of Nostoc 7120. This structure was determined by single-particle analysis of negatively stained PBS images, and aspects of the structure were further inferred from biochemical and spectroscopic approaches and by studying PBS isolated from mutants lacking specific rod-core linker polypeptides. In some cases, the PBS were still attached to PSII complexes, but the resolution (∼34 Å) of the model built from those complexes was too low to provide specific structural insights.

Zheng et al. (2021) determined the structures of 2 hemidiscoidal PBS: Synechococcus 7002, which have a tricylindrical AP core, at 3.5-Å global resolution; and Nostoc 7120, which have a pentacylindrical AP core substructure, at 3.9-Å global resolution (Fig. 4). The Synechococcus 7002 PBS is ∼450 Å (width) by 300 Å (height) by 220 Å (depth); the structural model contains 6 AP hexamer equivalents in the core, 12 PC hexamers in the peripheral rods, and 288 total phycocyanobilin chromophores (Figs. 1, 4, A and C). Each peripheral rod comprises 2 PC (αβ)₆ hexamers and 3 rod linkers, CpcG, CpcC, and CpcD. CpcG is a rod-core linker (L_RC) that joins the peripheral rods to the core; the rod linker (L_R) CpcC connects the 2 PC hexamers in the rods, and CpcD is the rod-terminating linker (L_RT) that controls the rod length (Bryant et al. 1990; de Lorimier et al. 1990a, 1990b). The derived structural model shows all rods having the same length, that is, 2 PC hexamers (Zheng et al. 2021), but in reality, a few rods have only 1 PC hexamer, some have 3 or more, but most have 2 hexamers (Bryant et al. 1990; de Lorimier et al. 1990a, 1990b). Each core cylinder has 4 trimeric AP toroids that are organized as 2 hexamers, and each core cylinder terminates at each end with the small core linker (L_C), ApcC, which is inserted into the central cavity of the terminal toroid (Fig. 4A). Each lower core cylinder has 4 different trimeric AP assemblies that are arranged in an anti-parallel manner in the 2 basal cylinders. The first trimer has 2 ApcA subunits, 1 ApcD subunit, 3 ApcB subunits, and 1 ApcC core linker. ApcD binds one of the terminal emitter chromophores in the core, with emission at about 680 nm. The second trimer has the N-terminal domain of ApcE (behaving as an α-type subunit), 2 ApcA subunits, 2 ApcB subunits, and 1 ApcF subunit. ApcF and ApcE are not part of the same (αβ) protomer, but instead ApcF is paired with ApcA but is closely appressed to the chromophore binding site of ApcE, which in turn is partnered with ApcB in a second protomer. The third and fourth trimers are composed of ApcA and ApcB, and they only differ by the presence of ApcC in the central cavity of the terminal toroid. They form a face-to-face hexameric unit that is bound back-to-back with the hexamer that contains the terminal emitter chromophores. The upper cylinder comprises 4 trimeric complexes (ApcA-ApcB)₃, arranged as pairs of trimers bound face-to-face to make the 2 hexamers that then bind back-to-back. The outer-most trimer at each end of the cylinder has ApcC bound, and each hexamer is connected and stabilized by the third REP domain of ApcE (Fig. 4A).

Figure 4.

Structures of 2 cyanobacterial hemidiscoidal PBS. A) Schematic view of the Synechococcus 7002 PBS. B) Space-filling structural model of the Synechococcus 7002 PBS (PDB 7EXT) with tricylindrical core in face view (left) and bottom view (right). C) Schematic view of the Nostoc 7120 PBS. D) Space-filling structural model of the Nostoc 7120 PBS (PDB 7EYD) with pentacylindrical core in face view (left) and bottom view (right). In the face view, red arrows designate AP hexamers that represent the fourth and fifth “cylinders” of the pentacylindrical core. For B and D, note the close proximity of ApcD (magenta), the N-terminal domain of ApcE (red) and ApcF (orange); ApcD and ApcE harbor the terminal emitter chromophores. Note that the peripheral rods of the Nostoc 7120 PBS can be extended by the attachment of phycoerythrocyanin hexamers to the core-distal ends of some rods under some growth conditions (e.g. low light intensity). Figure adapted from Zheng et al. (2021), Fig. 1, Creative Commons Attribution License (CC BY, https://creativecommons.org/licenses/by/4.0/).

The petH gene encodes ferredoxin:NADP⁺ reductase (FNR), which reduces NADP⁺ using electrons transferred from PSI by soluble [2Fe-2S]-ferredoxin. Most but not all cyanobacterial PetH proteins have a domain at their N terminus that is highly similar to the rod-terminating linker protein, CpcD (Schluchter and Bryant 1992). On average, about 1 FNR is bound to a single PBS, although there are 6 equivalent potential binding sites (Gómez-Lojero et al. 2003). It has not been possible to visualize the small amount of PetH associated with the peripheral rods (see the discussion of CpcL-PC-FNR rod complexes below). However, mass spectrometry and cryo-EM has shown that the long-form of FNR, with a CpcD-like domain at its N terminus, is associated with hemidiscoidal PBS as well as CpcL-PC-FNR rod complexes in Synechocystis 6803 (Liu et al. 2019; Zheng et al. 2023). The structural studies for CpcL-PC-FNR rod complexes demonstrated that the enzymatic domain of FNR is too mobile to be visualized by cryo-EM, but the CpcD-like N-terminal domain was clearly visible in the structural model (Zheng et al. 2023). These results confirmed the original proposal of Schluchter and Bryant (1992) concerning the localization of PetH in the PBS.

Many studies over the years have shown that PBS can occur in regularly spaced rows on the surface of the thylakoid membranes (e.g. Gantt and Conti 1969; Edwards and Gantt 1972; Bryant et al. 1979). A recent cryo-ET study of Synechocystis 6803 confirmed that PBS could form well-aligned rows of 4 to about 20 PBS; these tightly packed PBS formed parallel rows on the membrane surface. PBS were relatively rare on the plasma membrane surface and were most abundant on thylakoid-membrane facing surfaces, although PBS were also observed on the surfaces of membranes facing the plasma membrane and on the surface of membranes facing the cytoplasm (Rast et al. 2019). This contrasts with ribosomes, which were much more numerous on the cytoplasm-facing surface of the thylakoids—even more so than in the cytoplasm itself. When the structural model for Synechococcus 7002 PBS is docked using the spacing information derived from the tomography studies on Synechocystis 6803, one finds that in the tightly packed rows of PBS that the nearest phycocyanobilin chromophores in the cores of adjacent PBS are separated by ∼40 Å (Zheng et al. 2021). This distance would be compatible with inter-PBS excitation energy transfer. Thus, the close packing of PSII and PBS in rows could link multiple PBS and multiple PSII dimers into much larger photosynthetic units, similar to those in P. purpureum chloroplasts (You et al. 2023). These rows become most prevalent under low light conditions, when the PBP and PBS content of cells is highest (Ho et al. 2017).

In addition to the pentacylindrical core substructure, Nostoc 7120 PBS differ from those of Synechococcus 7002 by having 8 peripheral PC rods instead of 6 (Zheng et al. 2021) (Fig. 4C-4D). These PBS are somewhat larger, with dimensions of ∼540 Å (length) by 320 Å (height) by 210 Å (depth); the structural model contains 348 phycocyanobilin chromophores. The core is very similar to that of Synechococcus 7002 except that 2 half cylinders, that is, an AP (αβ)₆ hexamer, are bound by the fourth REP domain of each of the 2 copies of ApcE in the core (Fig. 4C). These half cylinders are not arranged parallel to the other core cylinders, as originally suggested (Ducret et al. 1996), but instead are bound to the end surface of the core perpendicular to the other core cylinders (Chang et al. 2015; Zheng et al. 2021). Nostoc 7120 and M. laminosus each have a small multigene family that encodes 4 rod-core linker proteins, 3 of which attach peripheral PC rods to the PBS core (Bryant et al. 1991; Glauser et al. 1992; Ducret et al. 1996, 1998). Two copies of the CpcG4 (L_RC4) linker attach rods to the sides of the core at the bottom, and 2 copies of this linker attach 2 rods to the upper core cylinder. A CpcG2 linker (L_RC2) attaches a PC hexamer to each of the 2 half-cylinders in the core. Finally, CpcG1 is also used to attach a second peripheral rod to each of the half-cylinders. As modeled, the Nostoc 7120 PBS contains 8 hexamers of AP and 14 hexamers of PC per PBS. However, under low-light conditions, the rods can contain additional PC hexamers, and PEC hexamers can be added as well (Bryant et al. 1976; Bryant and Canniffe 2018).

The 6.2-MDa PBS of Synechocystis 6803 have been studied by single-particle cryo-EM, which produced the highest resolution structure of a hemidiscoidal cyanobacterial PBS to date. Dominguez-Martín et al. (2022) solved the structures of 3 conformations of these PBS at global resolutions ranging from 2.7 Å to 3.5 Å. The Synechocystis 6803 PBS are ∼550 Å (width) by 360 Å (height) by 190 Å (depth), and each PBS binds ∼396 phycocyanobilin chromophores. The AP core is tricylindrical and is formed by a pyramidal stack of 3 AP cylinders with each comprising 4 toroid-shaped (αβ)₃ trimers as described above for Synechococcus 7002 and Nostoc 7120. The Synechocystis 6803 PBS core comprises 80 protein subunits that collectively bind 72 phycocyanobilin chromophores. Two copies of a previously undescribed, 10-kDa core linker, ApcG (sll1873), were found in association with each of the 2 bottom cylinders of the core. ApcG also interacts with PSII through its N-terminal region; together with the PB-loop of ApcE, ApcG forms a protrusion on the basal surface of the PBS that faces the thylakoid surface (Domínguez-Martín et al. 2022; Liu 2023). Growth of an apcG deletion mutant was impaired under limiting light conditions. Studies with a phospho-mimicking variant of ApcG suggest that this protein is important in distributing energy to the 2 photosystems and that it may play a role in regulating energy transfer from PSII to PSI (Espinoza-Corral et al. 2023).

Each peripheral PC rod of the Synechocystis 6803 PBS contains 40 polypeptides and 54 phycocyanobilin chromophores (Domínguez-Martín et al. 2022). Quantitative mass spectrometry suggested that the binding of the intrinsically disordered C-terminal extensions of the 6 rod-core linkers (CpcG) to the AP core affects the structural integrity of the core-membrane interface formed by ApcE and ApcG (Liu 2023). In the 3 different conformational classes observed by Domínguez-Martín et al. (2022), the 2 top peripheral rods appear in different positions relative to the bottom and side rods that do not appear to be mobile. The upper peripheral rods can both be up (up-up conformation), both be down (down-down conformation), or 1 rod can be up and 1 can be down (up-down conformation). Because of this rod mobility, peripheral rods were processed individually to a resolution of 2.1 Å. The bottom peripheral rods are attached to core cylinders B1 and B2, and the side and top rods are both attached to the top (T) cylinder of the core. When both top PC rods are in the down conformation, the height of the PBS is reduced to 260 Å. Model building indicates that only the up-up conformation is compatible with the close packing that is required to make rows of PBS on the surface of the thylakoid membrane. This raises the interesting possibility that the conformation of the peripheral rods could be actively altered to control the connectivity of the PBS antenna complexes to PSII dimers in the membranes. It also cannot be ruled out that the differential configuration of rods in the Synechocystis 6803 PBS structures are an artifact of single-particle analysis due to the removal of the PBS from their normal cellular environment. This will certainly be a topic for further investigation.

In addition to the structures of 3 conformations of Synechocystis 6803 PBS, Domínguez-Martín et al. (2022) determined the structure of the PBS in the quenched state with the bound, red form of orange carotenoid protein (OCP^R). The structure shows that OCP^R binds to both sides of the core as a dimer, which is formed through formation of a complex of the C-terminal domains of OCP. Interestingly, OCP^R is not bound near the terminal emitters but is bound to what might be considered a sensor module from the energy transfer point of view. OCP^R binds to ApcA-ApcB monomers located in the top and bottom core cylinders in the AP hexamer units that do not contain any terminal emitters. The carotenoid (canthaxanthin in the recombinant protein; echinenone in the natural protein) molecule that was bound to the N-terminal domain acts as the quencher. The closest phycocyanobilin to the canthaxanthin carotenoid moiety of OCP^R is ∼27 Å away. Modeling of energy transfer and quenching showed that a key feature that makes the quenching efficient is that OCP^R binds at 2 sites as a dimer.

Hemidiscoidal PBS of the thermophilic cyanobacterium Thermosynechococcus vulcanus have also been characterized structurally (Kawakami et al. 2021, 2022). These PBS are similar in many respects to those of Nostoc 7120 (Zheng et al. 2021). The T. vulcanus PBS were initially isolated and studied by negative staining and later by single-particle cryo-EM. In the latter study, the PBS core and PC rods were reconstructed to 3.7-Å and 4.2-Å global resolution, respectively. These PBS appear to be unstable, and the peripheral rods exhibited extensive dissociation under the isolation conditions employed, even after glutaraldehyde cross-linking. Thus, the cores and rods were studied separately. The T. vulcanus PBS has a pentacylindrical core, which is consistent with the fact that the ApcE subunit in this organism has 4 REP (Pfam00427) domains. In an earlier publication, the authors stated that these PBS have 8 peripheral PC rods (Kawakami et al. 2021), but in the subsequent article the authors state there are only 6 peripheral PC rods (Kawakami et al. 2022). Considering that the ApcE subunit has 4 REP domains, the resulting PBS are predicted to be similar to those of Nostoc 7120 and M. laminosus PBS with 8 peripheral rods, and this seems even more likely given that there are also 3 rod-core linker (L_RC) proteins as in those other cyanobacteria. The core structure differs from that of all other hemidiscoidal PBS from cyanobacteria because the bottom cylinders contain only 3 trimeric complexes. This could be similar to red algal PBS (see above), but the upper core cylinder in red algae has only 2 AP trimers whereas this cylinder in the T. vulcanus PBS has 4 AP trimers. Furthermore, ApcD is placed in a totally different position than is the case for all other PBS. The placement of ApcD seems questionable because of the relatively low resolution, and it seems highly unlikely that the core would be compositionally identical to other cyanobacteria but structurally so different. The authors state that they did not assign an ApcD subunit in the deposited coordinates in the PDB database. Further studies will be required to determine whether the observed differences are real or are artifacts caused by dissociation or possibly issues in the interpretation of the cryo-EM data.

Cyanelles of the glaucophyte alga Cyanophora paradoxa clearly have a cyanobacterial origin and have even retained a peptidoglycan wall layer (Pfanzagl et al. 1996). C. paradoxa cyanelles have regular arrays of hemidiscoidal PBS (Giddings et al. 1983). The cyanelle genome of C. paradoxa encodes the α and β subunits of AP and PC as well as ApcE; the sequence of ApcE suggests that C. paradoxa PBS should have tricylindrical AP cores (Stirewalt et al. 1995). The remaining linker genes are encoded in the nucleus (Watanabe et al. 2012), but the exact functions of these linkers are not known at this time. There are 4 rod-core linkers, CpcG1, CpcG2, CpcK1, and CpcK2; 2 core linkers, ApcC1 and ApcC2; and a single rod-terminating linker, CpcD. The CpcK1 linker has 2 REP (Pfam00427) domains and thus could join 2 hexamers of PC. The CpcK2 linker is likely to be similar based on its molecular weight, but the gene for it is incomplete and thus one cannot be sure what the missing sequence might do. The reason for the duplications of the other linkers is not clear, although it seems likely they function like their homologs in cyanobacteria. At the present time it is not possible to propose a model for the C. paradoxa PBS with confidence. Future cryo-EM and/or cryo-ET efforts may reveal structural details.

A subject of intense interest but limited progress concerns the molecular details of how hemidiscoidal PBS bind to PSII and PSI, because this should help to explain the pathways of excitation energy transfer from PBS to reaction centers in the membrane. However, it has proven difficult to isolate supercomplexes containing PBS and PSI, PSII, or both. A low-resolution study described the connection of PBS to PSII for Nostoc 7120, but the extremely low resolution achieved in that study did not reveal important structural details (Chang et al. 2015). Megacomplexes from Synechocystis 6803 that contained a PBS, a PSII dimer, and a PSI trimer were isolated and studied by cross-linking and mass spectrometry (Liu et al. 2013). Although informative in identifying some structural constraints, this study only provided some initial insights into how PBS dock on photosystem complexes. A preprint describing more details for one type of megacomplex has recently appeared (Zhang et al. 2023). In this study, Spirulina platensis cells were grown in low-intensity orange light, which should promote high cellular levels of PBP and PSI. Long rows of PBS and PSII should be present in these cells, and a structure at ∼3.5-Å resolution of 2 closely packed PBS in association with 3 PSII dimers was obtained by cryo-ET. When finalized and fully published, this study will undoubtedly provide new insights into how PBS dock onto PSII, but it likely represents only 1 type of interaction because no PSI complexes were observed and binding of PBS to isolated, single PSII dimers is likely to be different. Rapid progress in the application of cryo-ET recently is occurring; this is a powerful new method for in situ structure determination of photosynthetic complexes in their functional states (Young and Vella 2023).

Bundle-shaped PBS: Gloeobacter violaceus

Gloeobacter violaceus PCC 7421 (G. violaceus 7421) was the first unicellular, freshwater cyanobacterium to be described that lacks thylakoid membranes (Rippka et al. 1974). It was recovered from wet limestone rock scrapings from the Kernwald area of Oberwald canton, Switzerland. Cultures of this organism are bright purple or violet because of the relatively low Chl a content of the cells, which contain large amounts of PBP, including AP, PC, and an unusual PE that carries a phycourobilin chromophore that is more typically found in marine and red algal PE (Bryant et al. 1981). For many years, G. violaceus 7421 was the only described strain of the genus Gloeobacter, which is an important taxon due to its early divergence from all other cyanobacteria in phylogenetic trees as well as the unusual and likely ancestral properties of Gloeobacter sp. However, 2 studies in 2013 suggested that similar organisms might be more common than previously understood. Saw et al. (2013) described Gloeobacter kilaueensis, which was recovered from an epilithic biofilm in a lava cave in Kīlauea Caldera, Hawai′i. Furthermore, Mareš et al. (2013) obtained multiple Gloeobacter sp. isolates and suggested that similar organisms are likely to occur on wet-rock surfaces worldwide.

Guglielmi et al. (1981) isolated the PBS from G. violaceus 7421 and found them to be bundles of ∼6 cylindrical rods attached to a core-like structure that was presumed to contain AP (Fig. 1). The rods were stacked hexameric (αβ)₆ discs of PC and PE. This model was refined by studies of Koyama et al. (2006) and Krogmann et al. (2007), who further suggested that the core was pentacylindrical like that found in the cores of hemidiscoidal PBS from Nosoc 7120. Presently, no high-resolution structure exists for any bundle-shaped PBS. However, Wang et al. (2023) have recently described the development of a genetic system for G. violaceus PCC 7421, and they described the deletion of the cpeBA genes encoding the α and β subunits of PE as well as the deletion of the glr2806 gene encoding a linker protein. PBS in the mutant lacking PE have only 3 layers of PC hexamers in each rod. In the glr2806 mutant, 2 hexamers are missing from the core substructure, showing that this linker protein is associated with the AP core and not the peripheral rod bundle. With the development of this genetic system as well as strong interest in resolving questions about the roles of some of the unusual linker proteins, it seems likely that a more refined model or perhaps a structure for a bundle-shaped PBS might become available soon.

Paddle-shaped PBS: Anthocerotibacter panamensis

Anthocerotibacter panamensis is a recently described member of the Gloeobacteria that was not isolated from a wet-rock surface. Instead, it was isolated from a hornwort that was collected in Panama (Rahmatpour et al. 2021). However, it is still unclear whether A. panamensis is an epiphyte, a contaminating soil organism, or an unusual symbiont, because unlike this isolate, most cyanobacterial symbionts of hornworts are filamentous and fix dinitrogen. A. panamensis does not synthesize PE but produces PC and 2 isoforms of AP, and the cells lack thylakoid membranes like G. violaceus. Analysis of the complete genome revealed that this cyanobacterium has a unique combination of traits; it lacks circadian clock genes and has one of the most reduced set of genes for photosystems and PBS components known among cyanobacteria. Although capable of oxygenic photosynthesis under a wide range of light intensities, A. panamensis grows more slowly than most cyanobacteria and may be adapted to a low-light lifestyle and niche.

PBS of A. panamensis were isolated, and their structure was determined by single-particle cryo-EM to a global resolution of 2.9 Å (Jiang et al. 2023; Fig. 5). Surprisingly, this PBS is unlike any previously described and has several unique and possibly ancestral properties. The overall structure has been described as “paddle-shaped” (Fig. 1), but it could also be described as resembling a Christmas tree of AP decorated with 6 PC hexamers as ornaments and 2 chains of tinsel formed from 2 CpcN linkers that each bind 5 additional PC hexamers. This 2-fold symmetric PBS has approximate dimensions of ∼500 Å (height) by 273 Å (width) by 129 Å (depth) and has a unique, heptacylindrical core that surprisingly lacks ApcD and ApcF. One subdomain of the core resembles the pentacylindrical core of Nostoc 7120 PBS (Zheng et al. 2021); the second core subdomain comprises 2 additional AP cylinders connected to the top of the lower structure by 2 copies of a novel linker, ApcH, which has 2 REP domains with similarity to the REP domains in the core-membrane linker PBP, ApcE. Each upper cylinder is made up of 4 (αβ)₃ trimers formed from an AP variant, ApcA1-ApcB1, that is different from the AP variant, ApcA2-ApcB2, found in the pentacylindrical core substructure. Each end of the upper 2 cylinders is terminated by a core linker, ApcC1, and the ends of each of the 5 cylinders in the lower section are terminated by a variant core linker, ApcC2. Unlike hemidiscoidal PBS described above, the A. panamensis PBS do not contain PetH (FNR) (Jiang et al. 2023).

Figure 5.

Structure of the paddle-shaped PBS from A. panamensis. A) Schematic model of subunit organization and interactions. B) Structure with associated dimensions (created from PDB files 8IMI, 8IMJ, 8IMK, 8IML, 8IMM, 8IMN, and 8IMO). Note the heptacylindrical core (cyan and purple) and the 16 PC hexamers (green) that surround it. Also note that none of the PC hexamers form stacks as observed for all other PBS that contain PC. The fourth and fifth PC hexamers attached by CpcN (Rt4, Rt5, Rt4′, and Rt5′) are not depicted in Panel B. Panel A is adapted from Jiang et al. (2023), Figure 3, Creative Commons Attribution License (CC BY, https://creativecommons.org/licenses/by/4.0/).

The most surprising aspect of the A. panamensis PBS is that it lacks any peripheral rods formed from stacked PC hexamers (Fig. 5; Jiang et al. 2023). Instead, each side of these PBS is decorated with 1 PC hexamer attached by a CpcG linker; 2 unstacked PC hexamers attached by CpcJ, which is a novel rod-core linker; and a string of 5 unstacked PC hexamers attached to each upper cylinder by a second novel rod-core linker, CpcN. Thus, 16 PC hexamers surround a core of 24 trimer equivalents of AP. As in other PBS, one of the α-subunits in each bottom core cylinder is replaced by the N-terminal phycocyanobilin-binding domain of ApcE, but the functionally specialized subunits ApcD and ApcF are not present. The absence of PC stacking at the periphery suggests that excitation energy transfer from PC to AP might be inefficient in this PBS, and calculated rates and efficiencies of energy transfer based on the structure suggested that this is indeed the case for the PC hexamers in the chains assembled by CpcN. This is further substantiated by the steady-state fluorescence emission spectrum of these PBS at low temperature, which shows substantial fluorescence emission from PC despite its proximity to AP in the cores.

Rod-shaped CpcL-PC-FNR PBP complexes from Synechocystis 6803 and other cyanobacteria

The cpcBACDEFG1G2G3G4 operon of Nostoc (Anabaena) 7120 encodes a small, multigene family of rod-core linker proteins, CpcG1-4 (Bryant et al. 1991), and 3 of those genes homologous to cpcG genes encode proteins that are found in the hemidiscoidal PBS of this cyanobacterium (Glauser et al. 1992). This finding and other observations led to the realization that previous models for the structure of Nostoc 7120 PBS had been incorrect. In Synechococcus 7002 and Synechocystis 6803, a single gene product, CpcG/CpcG1, is the rod-core linker that attaches 6 peripheral, PC-containing rods to the AP core of PBS (Bryant et al. 1990; Bryant 1991; Zheng et al. 2021). A re-examination of the Nostoc 7120 PBS led to the realization that these PBS had 8 peripheral rods, not 6 as had previously been assumed (Glauser et al. 1992; Ducret et al. 1996, 1998). Furthermore, the cores of the hemidiscoidal PBS of Nostoc 7120 and Mastigocladus laminosus contain 2 half-cylinders of AP (each equivalent to 1 hexameric (αβ)₆ unit) yielding a pentacylindrical core (Glauser et al. 1992; Ducret et al. 1996, 1998). The CpcG4 rod-core linker attaches 4 PC rods to the core, and the CpcG1 and CpcG2 rod-core linkers each attach 2 peripheral rods to the core. This logically leads to the question: what is the role of CpcG3? CpcG3 is closely related to CpcG2 of Synechocystis 6803 and Synechococcus 7002. CpcG2 is a paralog of CpcG/CpcG1 in those strains and is not a component of their hemidiscoidal PBS. CpcG3 from Nostoc 7120 and CpcG2 from Synechocystis 6803 and Synechococcus 7002 differ from other members of the CpcG family by having a span of hydrophobic amino acids at their C termini, which is predicted to form an α-helical segment sufficiently long to anchor these proteins to the thylakoid membrane. A mutational study in Synechocystis 6803 revealed that the PC associated with CpcG1 mostly transferred energy to PSII, whereas the PC associated with CpcG2 preferentially transferred energy to PSI (Kondo et al. 2007). Furthermore, Synechocystis 6803 CpcG2 was tightly associated with the thylakoid membrane fraction, unlike CpcG/CpcG1, which was found in the soluble protein fraction after dissociation of PBS at low ionic strength (Kondo et al. 2007). Considering the unique properties of CpcG2 (CpcG3 in the case of Nostoc 7120) and to reduce confusion, CpcG2/CpcG3 was renamed CpcL (Watanabe and Ikeuchi 2013). Watanabe et al. (2014) subsequently described the isolation of a supercomplex between a PSI tetramer (dimer of dimers) and a rod-shaped CpcL-PC complex that some have referred to as the CpcL-PBS (Watanabe et al. 2014). Single particle analysis of the negatively stained complexes provided a model with a resolution of ∼17 Å. We believe that the term PBS is inappropriate for this structure, which is obviously a PBP complex but does not refer to a structure with a core and multiple classes of PBP as is the case for all other types of PBS. The CpcL-PC rod structure is structurally analagous to a PBS substructure, namely the peripheral rods, but CpcL allows these rods to bind to the thylakoid membrane and PSI rather than the AP core of the PBS. For this reason, we will refer to the rod-shaped structures as CpcL-PC-FNR complexes (Fig. 1).

The interactions between PBS and the photosystems have generally been considered to be weak, and complexes of PBS with PSI and/or PSII have been uncommon, although some examples have been reported (e.g. Liu et al. 2013; Watanabe et al. 2014; Chang et al. 2015). A recent study of CpcL-PC-FNR complexes and PSI in Nostoc 7120 found that these 2 complexes formed a very tight interaction after iron starvation (Shimizu et al. 2023). These authors were able to isolate stable complexes of both PSI monomers and PSI dimers with CpcL-PC-FNR complexes, but they were unable to isolate similar complexes containing CpcL-PC-FNR complexes and PSI tetramers under iron-starvation conditions. This differs from the results of Watanabe et al. (2014), who isolated tetrameric complexes of PSI with bound CpcL-PBS from iron-replete cells. This difference suggests that there must be some important differences between the PSI complexes under iron-replete and iron-starvation conditions, but what those differences might be is presently unclear.

Zheng et al. (2023) recently isolated CpcL-PC-FNR complexes from an AP-less mutant of Synechocystis 6803 and determined its structure by single-particle cryo-EM to a global resolution of 2.6 Å resolution (Fig. 6). The complexes are rod-shaped with a mass of ∼840 kDa, an average length of 160 Å, and a diameter of ∼100 Å. There was some heterogeneity in the length, but on average there were 3 PC hexamers per complex. The CpcL-PC-FNR complexes had an absorbance maximum at 620 nm and an emission maximum at low temperature of 670 nm when excited at 590 nm. In addition to the 3 hexameric (αβ)₆ PC complexes, there are single molecules each of CpcL, CpcC1, CpcC2, and PetH/FNR. The PetH found in FNR that has a CpcD-like domain at its N terminus (Schluchter and Bryant 1992); only the CpcD-like domain of FNR could be modeled because of the flexibility and mobility of the enzymatic domain of PetH (Fig. 6). Thus, this complex is analogous to a peripheral rod of the hemidiscoidal PBS of Synechocystis 6803 except that CpcG1 is replaced by CpcL anchoring the protein to the thylakoid membrane and PSI instead of the AP core of the PBS, and CpcD is replaced by the CpcD-domain of PetH (Zheng et al. 2023; Dominguez-Martín et al. 2022). This structure validates the original suggestion of Schluchter and Bryant (1992), confirmed by Gomez-Lojero et al. (2003), concerning the nature of the interaction of PetH with PBS. Presumably, the low occupancy of FNR at any of the 6 to 8 possible binding sites in hemidiscoidal PBS, and the flexible connection of the CpcD-domain to the enzymatic domain of PetH, has thus far precluded the direct observation of PetH in association with the peripheral rods of hemidiscoidal PBS.

Figure 6.

Structure of rod-shaped CpcL-PC-FNR complexes from Synechocystis 6803. A) Side-view of the CpcL-PC-FNR complexes (PDB 8HFQ). Subunits α-PC (lilac) and β-PC (blue) form 3 (αβ)₆ hexamers. B) Linker proteins CpcL (green), CpcC1 (magenta), CpcC2 (aqua), and the N-terminal domain of FNR/PetH (orange) are shown using cylindrical helices within the hexamers (transparent). C) Linkers only shown in space-filling representation. Note that only the CpcD-like domain of FNR/PetH could be modeled in the structure. Presumably, the catalytic domain of FNR/PetH (red glow) is flexible and could therefore not be resolved in the cryo-EM experiment. The structure of spinach FNR was taken from PDB 1FNC. D) Schematic representation of CpcL-PC-FNR complexes and PSI in the thylakoid membrane. PC hexamers are shown in blue, and the PSI complex is shown in green. The 2 domains that occupy the central cavities of the hexamers, Pfam00427and Pfam01383, are shown as vertical and horizontal rectangles, respectively. PC trimers are numbered sequentially from bottom to top. Adapted from Zheng et al. (2023), Figure 1, Creative Commons Attribution License (CC BY, https://creativecommons.org/licenses/by/4.0/).

CpcL-PC-FNR rod complexes carry 54 phycocyanobilin chromophores. One of the PC β subunit chromophores that interacts extensively with CpcL and is near where PSI would be bound has been suggested to be the terminal emitter of this complex for energy transfer to PSI (Zheng et al. 2023). That phycocyanobilin exhibits a more planar conformation and is similar to those found in other PBP that absorb longer red wavelengths like the phycocyanobilin of the α-subunits of AP-B (Peng et al. 2014) and FRL-absorbing AP (Soulier and Bryant 2021; Soulier et al. 2022; Gisriel et al. 2023a, 2023b, 2023c). This suggestion has very recently been confirmed by Guo et al. (2024), who showed that the red-shifted spectroscopic properties of 1 PC β subunit phycocyanobilin arises from its specific interaction with conserved residue Q57 of CpcL. Finally, Li et al. (2022b) recently reported that tethering of PetH to CpcL-PC-FNR rod complexes is essential for photoheterotrophic growth of Synechococcus 7002. Regarding this last point, Yamanaka and Glazer (1983) long ago reported that heterocysts of an Anabaena sp. contained rod-shaped PC complexes, which contained no AP and which strongly resembled the CpcL-PC-FNR complexes described here. Heterocysts lack PSII but use PSI to produce reductant and ATP for nitrogen fixation, and these PSI complexes would certainly benefit from the additional absorbance cross-section provided by the chromophores of CpcL-PC-FNR complexes. Watanabe et al. (2014) reported that CpcL-PC-FNR rod complexes are enriched 4-fold in heterocysts relative to vegetative cells. It seems likely that this antenna complex plays an important role in light harvesting in heterocysts which are depleted in hemidiscoidal PBS and PSII.

Helical FRL-AP–IsiX

Many cyanobacterial genomes include an operon or sometimes a cluster of 3 genes that are upregulated during an acclimation mechanism to low-light environments called low-light photoacclimation, or LoLiP. These genes include isiX, apcD4, and apcB3. IsiX is a paralog of the Chl a-binding protein, IsiA (Soulier et al. 2020; Soulier et al. 2022). IsiA is induced in response to iron starvation and forms rings around PSI in cyanobacteria (Toporik et al. 2019; Akita et al. 2020; Cao et al. 2020; see below). IsiX is induced in response to low light conditions, and based on its homology to IsiA, was recently suggested to bind to the periphery of PSI (Gisriel et al. 2023c). ApcD4 and ApcB3 form a FRL-absorbing AP variant. It was recently determined that the structural basis for FRL absorbance in ApcD4-ApcB3 lies in the α-subunit (ApcD4) where the protein environment surrounding the phycocyanobilin causes its pyrrole rings to adopt a coplanar conformation (Gisriel et al. 2023b). In general, the conditions under which the isiX, apcD4, and apcB3 genes are expressed, any specific complex they might assemble, and what their roles are in photosynthesis is unknown (and unstudied) in most cyanobacteria. An exception to this is Thermostichus spp. (formerly Synechococcus sp.). Thermostichus spp. occur in phototrophic microbial mat communities associated with the effluent channels of the alkaline siliceous hot springs, e.g. Mushroom and Octopus Springs, in Yellowstone National Park, WY, USA. Ward and coworkers (Becraft et al. 2015; Nowack et al. 2015; Olsen et al. 2015) showed that there are high-light and low-light adapted ecotypes of Thermostichus spp. in these mats, and that the low-light ecotypes have and co-express the apcD4-apcB3-isiX genes together with a gene encoding a cyanobacteriochrome photoreceptor, LcyA. Soulier et al. (2022) showed that the oligomeric form of the recombinant protein from Thermostichus sp. heterologously produced in Synechococcus 7002 has absorbance maxima at ∼621 nm and ∼709 nm with a fluorescence emission maximum at 714 nm at 77 K when excited at 590 nm (Soulier et al. 2020; Soulier and Bryant 2021; Gisriel et al. 2023b).

The recombinant ApcD4-ApcB3 oligomers were studied by single-particle cryo-EM, and their structure was determined at a global resolution of 2.89 Å (Gisriel et al. 2023b). Surprisingly, unlike all other PBP that have been studied structurally, ApcD4-ApcB3 occurs in solution as helical nanotubes with a diameter of about 112 Å and an inner diameter of 43 Å (Fig. 7). The length of the nanotubes is undefined but was estimated to be in the range of 10 to 13 monomers, which would be similar to the 12 protomers in the AP-containing core cylinders in PBS. Soulier et al. (2022) showed that this FRL-AP was tightly associated with thylakoid membranes and that detergent solubilization of the membranes was required to release the FRL-AP. Moreover, a complex containing ApcD4, ApcB3, and IsiX was detected by electrophoresis on native gels. IsiX has a C-terminal extension of 127 amino acids that are not predicted to form any folded ultrastructural element (Gisriel et al. 2023b). It was proposed that this unstructured polypeptide region might be inserted into the nanotube (or the nanotube could assemble around it), and that this could provide a means to limit the length of the helical nanotubes to maximize their efficiency for light harvesting. Energy transfer calculations showed that the efficiency of excitation energy transfer to a terminal chromophore decreased markedly as the helices become longer. Because IsiA binds to the periphery of PSI, it seems logical to propose that IsiX binds Chl a molecules, which could be verified in the complexes mentioned above, in a manner similar to IsiA. The entire IsiX-ApcD4-ApcB3 complex could then dock onto the periphery of PSI under low light conditions. The role of the LcyA protein, if any, in the expression of the apcD4-apcB3-isiX operon is unclear, although the 4 genes are cotranscribed. LcyA has 2 GAF domains, one of which binds a phycocyanobilin chromophore that could perceive and respond to light. The second GAF domain has a motif predicted to bind a [4Fe-4S] cluster. This GAF domain was indeed shown to bind such a cluster that was sensitive to oxygen, and while suggestive, it is not known whether LcyA is controls the expression of the other genes in this cluster (Soulier et al. 2022).

Figure 7.

Proposed interaction of FRL-AP (ApcD4-ApcB3)_∼12 nanotubes, IsiX, and PSI based on Gisriel et al. (2023b). A) Cartoon showing the membrane plane view of how a PSI trimer interacts with IsiX bound with a FRL-AP nanotube. The transmembrane domain of IsiX was suggested to bind in a position similar to an IsiA-binding site. The C-terminal extension of IsiX binds the helical FRL-AP nanotube, one possible orientation (standing normal to the membrane) of which is shown based on similarity to the CpcL-PC-FNR rod complexes. The C-terminal extension may serve as a molecular ruler to limit the length of the nanotube and establish its optimal length for energy transfer to IsiX or PSI. B) View of the cartoon from the stromal side of the membrane. Note that each PSI monomer has a stromal domain near its center, and the 3 proposed IsiX-FRL-AP binding sites are bound between each PSI monomer, thus avoiding steric clashes between the FRL-AP nanotube and the stromal domains of PSI monomers.

A major unresolved question concerning the FRL-absorbing AP variants had been how the AP α-subunits of such proteins shift the absorbance of phycocyanobilin so far to the red compared with PC or AP. This answer is not simple, as there are multiple effects. One trivial mechanism that occurs for a few cases is that the phycocyanobilin is not bound covalently to the protein. This increases the number of conjugated double bonds in phycocyanobilin by one and shifts the absorbance by about 40 nm to the red. As first suggested by Peng et al. (2014), another major mechanism for bathochromic shifting of the absorbance occurs by rigidly binding the chromophore in an extended conformation in which the pyrrole rings are brought into a highly coplanar conformation. In particular, the conformation of pyrrole ring A is more parallel and coplanar with the remaining 3 rings of the chromophore (Gisriel et al. 2023a, 2023b) than in other PBP subunits. The electrostatic environment of the chromophore is also suggested to play a role. In any event, a key aspect of shifting the absorbance to the red occurs upon oligomerization. The key event is the binding of the β-subunit from an adjacent protomer to an α-subunit chromophore binding site. This causes a conformational change in pyrrole ring D of the phycocyanobilin on the α-subunit that shifts its absorbance to the red. The binding also greatly reduces the inhomogeneous broadening and thereby narrows the absorbance band at ∼710 nm, which lead to a very small Stokes shift for the fluorescence emission (Gisriel et al. 2023a). Very similar effects are observed in the bicylindrical cores that occur in the AP variants that are formed in FRL in FaRLiP cyanobacteria (Gisriel et al. 2023c; Zhou et al. 2024).

Bicylindrical FRL-AP cores: Synechococcus 7335

FRL photoacclimation, or FaRLiP, is an acclimation process that occurs widely among mostly terrestrial cyanobacteria (Gan et al. 2014, 2015; Zhang et al. 2019; Antonaru et al. 2020, 2023; Ko et al 2024). This process occurs when FaRLiP-capable cyanobacteria grow in light in which wavelengths from 700 to 800 nm (i.e. FRL) are predominant, and once the process is completed, cells are then able to utilize FRL to perform oxygenic photosynthesis. This acclimation process results in a complete remodeling of their photosynthetic apparatus (Gan et al. 2014). Cells replace some Chl a (∼8% to 10%) by synthesizing Chl d and Chl f, which are FRL-absorbing Chls, and the pigment binding subunits of PSI and PSII are replaced by FRL-specific paralogs (for a review, see Gisriel 2024). This process involves the expression of a conserved cluster of 20 genes (Gan et al. 2014; Gan et al. 2015; Chen et al. 2019), including 5 genes encoding AP variants that absorb FRL. Six genes encode paralogs of PSI subunits, 5 genes encode paralogs of PSII subunits, and 5 genes encode variants of phycobiliproteins. A red/far-red photoreceptor (RfpA) that is a sensor histidine kinase activated by FRL (Gan et al. 2014), a CheY-type phosphate shuttle protein (RfPC), and a response regulator/transcriptional activator (RfpB) control the transcription of the genes in the cluster (Zhao et al. 2015). Finally, a distant “super-rogue” paralog of PsbA, ChlF, is responsible for the synthesis of Chl f (Ho et al. 2016). Ho et al. (2017a) isolated a core complex containing the FRL-absorbing AP variants from the FaRLiP-competent strain, Synechococcus sp. PCC 7335 (hereafter Synechococcus 7335). The light absorbed by PBP is critical for this process, and mutants unable to synthesize any one of the AP variants cause complete loss of all other FRL-absorbing pigments. The mutant cells fail to accumulate Chl d and are unable to grow in FRL (Bryant et al. 2020b). Gan et al. (2014) hypothesized that the FRL-absorbing AP variants would form bicylindrical cores, reminiscent of the PBS cores found in Synechococcus 6301/7942 (Glazer et al. 1979) but with FRL-absorbing AP variants. Ho et al. (2017a) showed that isolated cores indeed formed bicylindrical complexes that contained the products of the 5 apc genes in the FaRLiP-specific gene cluster (ApcD2, ApcD3, ApcD5, ApcB3, ApcE2) as well as ApcF and ApcC also found in PBS produced by cells grown in visible light. It was suggested that ApcD3 might be the functional equivalent of ApcD of hemidiscoidal PBS as a terminal emitter together with the chromophore bound by ApcE2 (Ho et al. 2017a; Herrera-Selgado et al. 2018). In Synechococcus 7335, neither of these phycocyanobilins are covalently bound, which would cause a bathochromic shift in the absorbance and fluorescence emission of these 2 chromophores. For the other AP variants in the bicylindrical cores, the red-shifted absorbance is in part due to the coplanar conformation of the phycocyanobilin chromophores on the α-subunits as described above in the preceding section (Gisriel et al. 2023a). In particular, pyrrole rings B, C, and D are highly coplanar, which can lead to a bathochromic shift of the absorbance of phycocyanobilin (Peng et al. 2014; Soulier and Bryant 2021; Gisriel et al. 2023b, 2023c; Guo et al. 2024).

Gisriel et al. (2024) isolated FRL-AP core complexes from Synechococcus 7335 cells that had been grown in FRL for 2 months. The complexes were unstable and partially dissociated after concentration and further purification on a second sucrose gradient and subsequent size-exclusion chromatography to reduce the levels of contaminating, non-pigmented proteins. The resulting complexes had absorbance maxima at 710 and 650 nm with a fluorescence emission maximum at 77 K of 716 nm with a shoulder at 730 nm. The complexes were subjected to single-particle cryo-EM analysis, and the structure was solved to a global resolution of 2.78 Å (Fig. 8). These complexes were cylindrical in shape with a length of ∼85 Å and diameter of ∼95 Å. Each contained 19 polypeptide subunits: 8 ApcB2 (β-type subunit), 6 ApcD5 (α-type subunit), 1 ApcE2 (α-type subunit with 2 repeat (REP) linker domains; L_CM, the core-membrane linker), 1 ApcD3 (α-type subunit), 1 ApcD2 (α-type subunit), 1 ApcF (β-type subunit), and 1 ApcC (core linker). The PBP subunits were arranged as 3 disc-shaped (αβ)₃ trimers. The first and second trimers were arranged face-to-face into a hexamer that included ApcC, ApcD3, ApcE2, ApcD2, and ApcF, and the third trimer was joined to this hexamer-like unit by the second REP domain of ApcE2 in a tail-to-tail manner. Gisriel et al. (2024) provide a lengthy discussion of why this complex is likely missing the fourth trimer that would be composed solely of ApcD5 and ApcB3 together with ApcC. The addition of a fourth FRL-AP trimer would increase the cylinder length to ∼105 Å. Overall, each cylinder of the bicylindrical core has a nearly 1-to-1 correspondence to the organization of the bottom core cylinders in hemidiscoidal PBS as described above (Figs. 3 and 8). The exceptions are that all phycocyanobilin-bearing subunits except ApcF are variants optimized for FRL absorbance, and 1 ApcA molecule is replaced by ApcD2, which is positioned to possibly assist in stabilizing the side-by-side binding of the 2 cores cylinders on the FRL-PSII complex. The phycocyanobilins bound to the ApcD2 subunits are in very planar configurations that red-shift their absorbance, as found for the FRL-absorbing ApcD4 subunits in the helical nanotubes described above, and the phycocyanobilins bound to ApcD3 and ApcE2 lack the thioether linkage to the protein causing an extension in their conjugated system and thus red shifted absorbance (Gisriel et al. 2024). Considering that FRL-PSII binds only 2 Chl d molecules and 8 Chl f molecules per dimer (Gisriel et al. 2022, 2023d), the addition of twelve strongly FRL-absorbing phycocyanobilins that provide additional absorbance cross-section in the FRL region of the solar spectrum should substantially increase the rate of photosynthesis when the incident light is greater than 700 nm. Gan et al. (2014) found that the rate of oxygen evolution from cells acclimated to FRL was much greater than that of cells grown in visible light when the actinic light was greater had wavelengths above 700 nm.

Figure 8.

Superposition of the FRL-AP core complex determined by Gisriel et al. (2024) onto a typical hemidiscoidal PBS and corresponding subunit arrangement compared with a visible light-absorbing AP cylinder. A) Superposition of 2 FRL-AP cylinders (colors; PDB 8UHE) onto the bottom AP cylinders from a hemidiscoidal PBS (grey, PDB 7EXT). The red boxes denote the AP trimers missing in the FRL-AP core structure. B) Organization of subunits in the FRL-AP core complex compared with those in a visible light (VL)-absorbing AP cylinder. The red glowing FRL-AP subunits correspond to those missing in the structure. Panel B adapted from Gisriel et al. (2024), Figure 2B, Creative Commons Attribution License (CC BY, https://creativecommons.org/licenses/by/4.0/).

Chl antenna proteins in cyanobacteria and red algae

In the green chloroplast lineage that leads from green algae to higher plants, membrane-bound Chl a/b-binding proteins form light-harvesting supercomplexes with PSI (LHCI) and PSII (LHCII). They form a superfamily of diversified proteins, the Chl a/b-binding protein (CAB) family, that vary somewhat structurally and functionally, but all bind Chl a and carotenoids and many can bind Chl b or Chl c. Red algae and cryptomonads have LHCI proteins of the CAB superfamily that bind Chl a and carotenoids and are only associated with PSI. The glaucophyte, C. paradoxa, lacks LHC proteins altogether, but interestingly, it produces dimeric and tetrameric PSI complexes like some cyanobacteria (Watanabe et al. 2011). The structure of the tetrameric form of C. paradoxa differs somewhat from that of tetrameric PSI complexes from Nostoc 7120 (Kato et al. 2019; Zheng et al. 2019; Kato et al. 2022b) and Chroococcidiopsis sp. TS-821 (Li et al. 2019; Semchonok et al. 2022). Cyanobacteria also contain membrane-associated, Chl-binding proteins, but these proteins are not related to the CAB superfamily. Instead, these proteins have 6 transmembrane α-helices and are members of the PsbC (CP43) superfamily. In most PBP-containing cyanobacteria, the only Chl a-binding antenna protein is the iron starvation–induced protein, IsiA (IsiX, a paralog of IsiA, is also sometimes present; see above). In some cyanobacteria that lack or have highly reduced amounts of PBP, such as Prochlorococcus spp. and Acaryochloris marina, these proteins were known as prochlorophyte Chl-binding (PCB) proteins (Chen et al. 2005a, 2005c). Chen et al. (2008) suggested referring to all members of the CP43 superfamily as Chl-binding proteins (CBP) and made a detailed suggestion of new names for the proteins of this family. Unfortunately, the suggested names do not reasonably allow translation into valid gene names, so only the class name, CBP, will be used here. Chloroxybacteria is an alternative name for organisms that previously were called prochlorophytes; the latter term is no longer used because these organisms are not evolutionary predecessors of green algae. Chloroxybacteria usually contain a small family of CBP proteins, some of which associate with PSI and some with PSII, just as some but not all cyanobacteria with PBP produce members of a small family of IsiA-like CBP proteins (e.g. Shen et al. 2016; Nagao et al. 2023). A number of studies of Chl-binding antenna protein complexes in cyanobacteria and red algae have recently been performed using single-particle analysis and cryo-EM. Some highlights of these studies follow.

Iron-starvation induced protein A (IsiA)

Iron starvation of Synechococcus 7942 causes a large decrease in the PBP content, known as chlorosis, and a blue-shift in the Q_y absorbance maximum of the total cellular Chl (Guikema and Sherman 1983). These changes are also accompanied by the appearance of a large amount of a membrane-associated Chl-protein complex (Jia et al. 2021). Other studies identified an operon, isiAB, which is induced by iron starvation and encodes flavodoxin (IsiB) and a Chl a-binding protein, IsiA, which is paralogous to PsbC (CP43) (Laudenbach and Straus 1988). It then only remained to demonstrate that IsiA was the protein component of the membrane-associated Chl complex (Burnap et al. 1993). IsiA is not solely induced under iron-starvation conditions, but it can also be produced in response to high-light stress and oxidative stress (Havaux et al. 2005; Singh et al. 2005). Modeling suggested that, like PsbC, IsiA should have 6 transmembrane α-helices and that it should bind ∼13 Chl a molecules (Bibby et al. 2001b). A number of studies investigated the structure of IsiA-PSI complexes by single-particle EM analysis of negatively stained PSI-IsiA complexes (Bibby et al. 2001a, 2001b; Boekema et al. 2001; Kouril et al. 2005). These low-resolution structural studies revealed that IsiA produces an 18-membered ring that surrounds a PSI trimer and double-ringed complexes with 35 IsiA molecules that surround a PSI monomer. Still larger complexes have sometimes been observed in Thermosynechococcus elongatus; an outer ring with 25 IsiA molecules surrounding an 18-membered ring and a PSI trimer have been reported (Chauhan et al. 2011). Multiple studies have also commented on the presence of incomplete rings and arcs around various PSI complexes (e.g. Zhao et al. 2020).

Single-particle cryo-EM analyses have recently been used to determine the structures of 3 IsiA-PSI complexes: Synechocystis 6803 at 3.09-Å resolution (Toporik et al. 2019), T. vulcanus at 2.7-Å resolution (Akita et al. 2020), and Synechococcus 7942 at 2.9-Å resolution (Fig. 9; Cao et al. 2020). In the latter study, a structure of the IsiA-PSI supercomplex with the flavodoxin electron acceptor bound to the stromal subunits was also determined at 3.3-Å resolution. The IsiA proteins in these 3 complexes are about 77% to 78% identical and 86% to 89% similar in amino acid sequence, so it is not surprising that they exhibit very similar structures in which an 18-membered ring of IsiA surrounds a PSI trimer. Leptolyngbya sp. JSC-1 produces an interesting complex in which a monomeric PSI complex binds 6 IsiA paralogs (Shen et al. 2016). One of the IsiA-like subunits has a PsaL-like domain at its C terminus, which apparently blocks the formation of PSI dimers and tetramers. Nagao et al. (2023) have reported the structure of the orthologous complex from Nostoc 7120 at a global resolution of 2.62 Å.

Figure 9.

Structure of trimeric PSI-IsiA₁₈ supercomplex from Synechococcus 7942 (PDB 6KIG; Cao et al. 2020). A) Two views of the structure shown with a transparent surface and Chl tetrapyrrole rings (green). Note that the Chls form a stromal and a lumen ring. The stromal ring is more continuous while the lumenal ring is somewhat punctate due to clustering of the Chls. Also note that within a PSI monomer, similar stromal and lumenal rings also occur (Bryant and Canniffe 2018). B) Two views of the complex shown in cylindrical helical representation where PSI is colored yellow and IsiA subunits are colored individually. C) Close up view of an IsiA monomer. In addition to the tetrapyrrole rings of the 17 Chl a molecules (green), the 4 β-carotenes (orange) are shown.

The Synechocystis 6803 IsiA₁₈-PSI₃ supercomplex is nearly circular with threefold symmetry and has a diameter of 300 Å in and a height of 110Å (Toporik et al. 2019). This IsiA₁₈-PSI₃ complex binds 591 Chl a molecules, which are divided almost equally between the IsiA ring (306 Chls) and the PSI trimer (285 Chls). Each IsiA subunit binds 17 Chl a molecules, 13 of which are in similar positions as in PsbC, and 4 carotenoids, which were modeled as β-carotene, although previous studies suggested that there are only 2 β-carotenes, 1 echinenone, and 1 zeaxanthin (Ihalainen et al. 2005). The Chls are distributed asymmetrically: 11 of the 17 Chls occur in a ring near the stromal surface and 6 form a cluster nearer the lumenal surface of the membrane. The Chls on the stromal side forms a continuous ring around the PSI trimer, while the Chls on the lumenal side are arranged in small independent clusters. However, the lumenal Chls may be responsible for excitation energy transfer to PSI; distances to PSI Chls range from 13 to 24 Å on the lumenal side, although the majority are less than 20 Å. On the stromal side, the shortest Chl to Chl distance is 18.5Å, but most of the distances between Chls in the IsiA ring and the PSI trimer are in the 21 to 25 Å range (Toporik et al. 2019). The carotenoids are all located in very close proximity to Chls, and thus they should be able to participate in light harvesting as well as photoprotection.

A recent further study of the Synechocystis 6803 IsiA₁₈-PSI₃ supercomplex examined the issue of structural heterogeneity of the IsiA ring relative to the PSI trimer (Harris et al. 2023). It was found that very large translation movements, as large as 20 to 30 Å, of the ring subunits can occur in these complexes. Nevertheless, single-molecular studies of energy transfer within individual IsiA₁₈-PSI₃ complexes remained more or less constant, with very rapid energy transfer occurring for all conformations (remarkably, sometimes up to three-fold faster than the ensemble average). These results suggest that the evolutionary design of this antenna system can mitigate rather large-scale structural perturbations, which results in robust and efficient energy transfer (i.e. antenna function) within a flexible membrane system. The specific arrangement and density of Chls leads to efficient energy transfer no matter what conformation of the 2 components might be.

The Synechococcus 7942 IsiA₁₈-PSI₃ supercomplex is very similar to that of Synechocystis 6803 (Cao et al. 2020). The ring of IsiA molecules was rotated about 1.5° relative that of Synechocystis 6803, which leads to small shifts of 3 to 4 Å in the positions of individual IsiA subunits. It was concluded that the position of the ring might not be specific, but the ring conformation relative to PSI is not fixed (Harris et al. 2023). Each PSI complex was found to bind 1 additional Chl molecule, so this complex binds 594 Chl a rather than 591 as for Synechocystis 6803. Each IsiA binds 17 Chl a and 4 carotenoids that were identified as mostly β-carotene and some zeaxanthin. An interesting observation is that the PSI complexes in the IsiA₁₈-PSI supercomplex contain PsaK1. Some cyanobacteria have 2 psaK genes, psaK1 and psaK2, and in Synechocystis 6803, PsaK2 was found in the PSI complexes from cells grown under normal, nutrient-replete conditions (Malvath et al. 2018; Cao et al. 2020). T. elongatus only has a single psaK gene orthologous to psaK2. However, PsaK1 is found in IsiA₁₈-PSI supercomplexes of Synechococcus 7942 and Synechocystis 6803. The presence of PsaK1 in the IsiA₁₈-PSI complexes could be an adaptive mechanism to improve energy transfer between IsiA and PSI under iron-starvation conditions.

The T. vulcanus IsiA₁₈-PSI supercomplex binds 585 Chl a and 138 carotenoids (modeled as β-carotene), and, like the other 2 structures, IsiA subunit binds 17 Chl and 4 carotenoids (Akita et al. 2020). The T. elongatus PSI complexes had 93 Chl a and 22 carotenoids per monomer (279 per trimer), and the IsiA subunits bind 306 Chl a and 72 carotenoids. Two of the 4 Chls that are specific to each IsiA subunit compared with PsbC are probably involved in energy transfer from IsiA to PSI, and 1 Chl may be involved in IsiA-to-IsiA energy transfer. Thus, the additional Chl binding sites that evolved in IsiA appear to have been added to make IsiA more functional as a light-harvesting antenna protein. Femtosecond time-resolved fluorescence measurements indicate that energy migration to P700 and trapping is very rapid and do not support the idea IsiA acts primarily as an energy quencher rather than playing an important role in light harvesting under iron starvation conditions. The preponderance of evidence suggests that, when it is present in the IsiA₁₈-PSI supercomplexes, IsiA functions as a very efficient antenna complex for PSI (Chauhan et al. 2011; Akita et al. 2020).

Leptolyngbya sp. JSC-1, also known as Marsacia ferruginose (Brown et al. 2010), is a filamentous, nonheterocystous cyanobacterium that requires elevated iron concentrations for optimal grow. In spite of this observation, this cyanobacterium produces 5 different IsiA-like proteins (Shen et al. 2016). One gene cluster, isiA-isiB-isiC, is similar to those of many other cyanobacteria. The second gene cluster (isiA2-isiA3-isiA4-cpcG2-isiA5) contains 4 isiA-like genes and cpcG2 (now known as cpcL). Transcript levels for the isiABC operon increase about 300-fold under iron starvation conditions, and those for the second cluster increase about 500-fold (Shen et al. 2016). Analysis of complexes containing PSI or PSII provided evidence that 4 of the proteins (IsiA1, IsiA2, IsiA3, and IsiA4) associate with PSI under iron-starvation conditions. IsiA5 was not found in complexes containing PSI but instead seemed to be associated with PSII. IsiA4 is actually a fusion protein in which the N-terminal domain is homologous to IsiA and the C-terminal domain is homologous to PsaL. Single-particle analysis of the photosystem complexes produced under iron starvation provided evidence that the IsiA-like proteins were organized as arcs or rings surrounding PSI and PSII (Zhao et al. 2020).

Nagao et al. (2023) have recently characterized a supercomplex of IsiA homologs with a monomeric PSI complex in Nostoc 7120 by single-particle cryo-EM. This cyanobacterium also has a small multigene family of isiA homologs, which occur as an operon of 4 isiA-like genes, all of which become highly expressed under iron limitation conditions (Nagao et al. 2021). The PSI monomer contains PsaA, PsaB, PsaC, PsaD, PsaE, PsaF, PsaI, PsaJ, PsaM, PsaX, and an unidentified protein or proteins that occupy the position of PsaK from other structures. Nostoc 7120 has 3 proteins with similarity to PsaK, and it is not clear which 1 (or multiple) proteins occur in the monomer complex.

The PSI monomer is enclosed by a hook-shaped arc of 6 IsiA subunits that cover 2 sides of the end of the monomer formed by PsaA. The first IsiA subunit is a fusion of an IsiA domain and a PsaL domain, which accounts for the absence of PsaL in the monomer, because the fusion protein binds to the site normally occupied by PsaL. Nagao et al. (2023) named this subunit IsiA2-1, and it is homologous to IsiA4 in Leptolyngbya sp. JSC-1 (Shen et al. 2016). The PsaL domain binds 3 Chl a molecules and 2 carotenoids; the IsiA domain was modeled as binding 3 carotenoids and fourteen Chl a; 9 occur on the stromal side and 5 on the lumenal side of the subunit. Of the remaining 5 IsiA binding sites, only 2 of the proteins in those sites could be identified. These proteins binding to sites 4 and 5 were identified as being the products of the same gene, isiA1, but had different numbers of Chl a molecules assigned to them. Site 4 had 17 Chl a like other IsiA complexes described above, but only 10 Chl a molecules could be assigned to the subunit in site 5 (Nagao et al. 2023). The density map for this position was weaker than for site 4, which could reflect differences in occupancy or a difference in the selectivity of the site for a particular IsiA variant. The proteins bound at the remaining 3 sites, sites 2, 3, and 6, also exhibit weak densities in the electrostatic potential map, and could not be assigned to particular gene products, but it is likely that they include products of each of the remaining 2 genes, isiA3 and isiA4. The protein bound in site 2 binds at least 8 Chl a, that in site 3 binds at least 1 Chl a and 1 carotenoid, and that in site 6 binds at least 11 Chl a molecules. Further details concerning the identities of the subunits in the remaining sites and the numbers and types of pigments they bind will require a higher resolution structural model.

Nostoc 7120 continues to produce PsaL under iron starvation conditions, and tetrameric PSI complexes were observed under those conditions (Nagao et al. 2023). This suggests that the PsaL domain of the subunit bound in IsiA site 1 must compete with PsaL for binding to the PSI monomer. It is not clear what advantage is achieved by having 2 forms of PSI, but it may be related to interactions with the hemidiscoidal PBS as well as the rod shaped CpcL-PC-FNR complexes. The latter can certainly associate with a PSI tetramer, but it can probably not bind to an IsiA-PSI complex. Thus, cells might retain some PC to bind to CpcL to act as an antenna for the PSI tetramer and use the Chl-binding IsiA proteins as the antenna for PS monomers.

Red algal PSI-LHC1 supercomplexes

Red algae produce Chl a-binding proteins that belong to the CAB protein superfamily, which includes the light-harvesting Chl proteins of green algae and plants (Green and Durnford 1996). In red algae, the CAB proteins form a small multigene family, the lhcr genes, light-havesting Chl proteins from the red algal lineage, whose products form complexes with PSI (Kirilovsky and Büchel 2019; Kato et al. 2024). Structural studies have been conducted with PSI-LHCI supercomplexes from 3 red algae: Cyanidioschyzon merolae (Antoshvili et al. 2019), Cyanidium caldarium (Pi et al. 2018; Kato et al. 2024; Nagao et al. 2023b), and P. purpureum (You et al. 2023). Interestingly, 3 different methods have been employed for these 3 cases: x-ray crystallography, single-particle cryo-EM, and cryo-ET, respectively. Although the 3 supercomplexes share obvious similarities, there are nonetheless some significant differences. At present, it is not clear whether these differences arise from differences in complex stability during isolation or from biological differences of the 3 organisms. It should also be noted that the structure of PSII from the thermoacidophilic red alga C. caldarium has been determined at 2.76-Å resolution (Ago et al. 2016).

The PSI-LHCI complexes of red algae have very similar PSI monomers at their cores, comprising 12 subunits: Psa A, B, C, D, E, F, I, J, K, L, M, and O, but PsaX of some cyanobacteria is absent. PsaO, which occurs as 2 variants in C. caldarium, binds to PsaA between PsaK and PsaL and does not occur in cyanobacteria because it would interfere with oligomer formation. Otherwise, the other core subunits are structurally similar to the homologous proteins of cyanobacterial PSI. The core monomer binds 97 Chl a, 20 β-carotenes, 2 zeaxanthins, 3 [4Fe-4S] clusters, 2 phylloquinones, and 4 lipids (Kato et al. 2024). In C. caldarium and C. merolae, the PSI monomer is surrounded by 1 or 2 arcs of LHCI subunits, which collectively bind up to ∼57 Chl a molecules and 21 carotenoids (mostly zeaxanthin and some β-cryptoxanthin (Kato et al. 2024; Nagao et al. 2023b). Three of the LHCI subunits bind in the vicinity of PsaA, PsaJ, and PsaF, while the remaining 2 LHCI subunits are bound to the opposite side of the monomer near PsaB, PsaI, PsaL, and PsaM. This overall arrangement is very similar to the structure observed for 1 complex isolated from C. merolae; however, in C. merolae, a second form of the PSI-LHCI complex is observed in which the 2 subunits on the PsaB-side of the PSI monomer are missing (Fig. 10), suggesting that these subunits are more loosely bound to the PSI monomer (Pi et al. 2018). There are hints that this could be physiologically important, but no definitive answer is yet available on this issue. In C. merolae, the LHCI subunits are reported to bind 11 to 13 Chl a and 5 zeaxanthin (Pi et al. 2018). In C. caldarium, it is possible to isolate the LHCI complex separately from the PSI monomers (Nagao et al. 2023). The 5 LHCI subunits are encoded by 3 lhcr genes in both thermoacidophilic red algae. Each Lhcr subunit has 3 transmembrane α-helices and binds 9 to 13 Chl a and 4 or 5 zeaxanthins (Kato et al. 2024). In C. merolae, the LHCI subunits are reported to bind 11 to 13 Chl a and 5 zeaxanthin (Pi et al. 2018; Antoshvili et al. 2019). Whether these compositional differences are real or products of loss during isolation and/or resolution differences is currently uncertain.

Figure 10.

Structure of a PSI monomer with 5 LHCI subunits from C. merolae (PDB 5ZGB; Pi et al. 2018). PSI monomer (yellow) showing the binding positions of 5 LHCI subunits (other colors). The tetrapyrrole rings of Chls are shown in green. The dotted circles with question marks indicate the approximate binding positions of 3 additional LHCI subunits that are observed in the in situ structure of PSI in P. purpureum obtained by cryo-ET (You et al. 2023). Whether these subunits have been lost during isolation or are specific to P. purpureum is presently uncertain.

The PSI-LHCI complex of P. purpureum has not been studied in isolation but instead was characterized in situ as part of a massive PSI-PSII-LHC supercomplex by cryo-ET (You et al. 2023), which eliminates doubts concerning loss of subunits. The monomeric PSI core is similar to that described above but has 1 additional subunit, PsaR, which previously had not been associated with red algal PSI complex but is a known component of PSI in diatoms (You et al. 2023). PsaR binds to PsaB in a position similar to the binding site for PsaK on PsaA. The LHCI complex in P. purpureum contains the 5 subunits discussed above for C. merolae and C. caldarium, but it additionally exhibits 2 important differences. Firstly, the complex contains 3 additional Lhcr subunits that seem to be the products of genes that are not found in other red algae and that bind to PSI via PsaR, which has not yet been observed in other red algae. Secondly, Lhcr1 is replaced by a unique, three-helix Chl a-binding protein known as RedCap that forms a monophyletic lineage within the red algal CAB protein family and their relatives (Sturm et al. 2013). With the additional 3 Lhcr subunits, the LHCI subunits form a continuous eight-membered arc, with RedCap at one end and then an arc of 7 Lhcr subunits, that wraps around the PSI monomer on 3 sides, leaving only the PsaL-PsaO-PsaK side of the monomer uncovered. That open side of the PSI complex forms the interface for binding to the PsaW-PsbB-PsbH-PsbX surface of a PSII tetramer (You et al. 2023).

Kato et al. (2024) have discussed an evolutionary scheme that accounts for the similarities and differences among the various PSI-LHCI complexes of red algae and the other red lineage organisms (cryptophytes, diatoms). The results obtained with red algal PSI-LHCI complexes are important, because they provide a baseline for understanding the related complexes found in other red-lineage organisms, including cryptophytes, diatoms, haptophytes, and dinoflagellates (see Falkowski et al. 2004). Although the LHCI proteins of these organisms do not bind Chl b like their homologs in the green lineage, there is generally much heterogeneity with respect to binding Chl a, Chl c, and especially carotenoids (Kirilovsky and Büchel 2019).

Although green algae are not a subject of this review, 1 recent study of the green alga Prasiola crispa, is worthy of note here for multiple reasons (Kosugi et al. 2023). This aerial green alga forms layered, mat-like colonies under the severe terrestrial conditions found in Antarctica, and like cyanobacteria that grow in mats, organisms in the lower layers of the mat receive light highly enriched in FRL. P. crispa produces an undecameric LHC, Pc-frLHC, that forms a toroid-shaped ring with 11-fold symmetry. Although this complex acts as an antenna for PSII, the subunit of this complex is not a member of the LHCII family but rather is a Chl-binding protein with similarities to the subunits of LHCI complexes of red algae. The Pc-frLHC complex absorbs maximally at 671 nm and 706.5 nm, similar to the FRL-absorbing antenna proteins produced during FaRLiP by cyanobacteria. The Pc-frLHC subunits have 4 transmembrane α-helices, and each subunit binds 10 Chl a, 1 Chl b, and 1 molecule each of the carotenoids violoxanthin and loroxanthin. The 120-Å ring opening is too small to accommodate a PSII complex, so presumably PSII binds to the exterior surface of the rings. Pc-frLHC is the first eukaryotic light-harvesting protein that forms rings and evidently this structure has convergently evolved to resemble other antenna ring structures like those of IsiA (see above) and the LH1 and LH2 complexes of purple bacteria (Bryant and Canniffe 2018).

Cryptophyte light-harvesting apparatus

Cryptophytes, also known as cryptomonads, are a group of single-celled, biflagellated eukaryotic algae that, like diatoms, obtained their chloroplasts from an ancestral red alga by secondary endosymbiosis. This conclusion is assured because they have retained a portion of the red algal nucleus as a structure known as the nucleomorph (Douglas et al. 2001). Like diatoms, they also contain membrane-intrinsic LHCI proteins that bind Chls a and c as well as the unusual carotenoid alloxanthin. As described above, red algae also produce LHCI proteins from the CAB family that bind Chl a, but unlike diatoms, red algae do not produce Chl c or alloxanthin. Cryptophytes have PBP like red algae and glaucophytes, but no AP has been found in them and their PBP do not form PBS. Instead, it has long been known that the PBP in cryptophytes are densely packed in the thylakoid lumen (Gantt et al. 1971; Spear-Bernstein and Miller 1989).

Cryptophyte PBP are usually designated by the prefix Cr and followed by a class designation, either PE or PC, a number representing the absorbance maximum (e.g. CrPE555), and they are unique within the family of PBP. Each CrPBP is an α₂β₂ heterotetramer comprised of 2 (αβ) protomers; the α-subunits are unique and are unrelated to any other known proteins, whereas the β-subunits are highly similar to the β-subunit of PE that occurs in red algal PBS. Additionally, although it is often stated that any given cryptophyte produces a single PBP (e.g. Glazer and Wedermayer 1995), which are described as CrPE if red-colored and CrPC if blue-colored, this is probably not quite the case for 2 reasons. Firstly, the α-subunits are encoded by a small nuclear gene family of related but non-identical sequences so that an organism can have about twenty or more different α-subunits. Furthermore, the 2 α-subunits in the typical heterotetramer are usually the products of different genes. Secondly, while a given cryptophyte may have a predominant PBP, either PE or PC, there are minor spectral types present in cells with modified properties. For example, Hemiselmis andersenii (Ha) produces HaPE555, HaPE565, and HaPE645; these 3 spectral types occurred in the ratio 5:1:1, respectively (Rathbone et al. 2023). X-ray crystallography has shown that the cryptophyte PBP can adopt 2 different structures: a closed form (Wilk et al. 1999; Doust et al. 2004) and an open form (Harrop et al. 2014). The latter has so far only been shown to occur in members of the genus Hemiselmis (Michie et al. 2023). Rathbone et al. have reported 4 new structures for HaPE555 at resolutions ranging from 1.57 to 1.95 Å; all have the open quaternary structure with 2 distinct but nearly identical α-subunits (Rathbone et al. 2023). The protein packing in the crystals show molecules that are organized into filaments and 2-dimensional rafts. A structure for HaPE560 was determined at 1.45 Å-resolution, and while it is an open form with the same bilin chromophore composition as HaPE555, structural differences caused by an inserted loop in the α-subunits produces a new quaternary structure, termed the open-brace form. The structure of the closed form of HaPE645 was solved at 1.49 Å. A puzzling finding is that the chromophore attached to Cys82 in the 2 β-subunits of this protein are different: 1 is phycocyanobilin and 1 is phycoerythrobilin. It is an interesting conundrum to imagine how 2 identical polypeptides can have different chromophores attached, but this may be controlled by the N terminus of the α-subunits, which are different (the heterotetramer is actually α₁α₂β₂). PBP lyases can attach and remove bilins from folded proteins, which presumably might already have bound α-subunits and thereby be modified by them. Changes in the chromophore content of the β-subunits of Hemiselmis pacifica PC577 and Proteomonas sulcata PE545, but the mechanism by which this chromophore change occurs is unknown (Spangler et al. 2022).

Rathbone et al. (2023) estimate that only 3 to 4 PE molecules would be required to span the width of the thylakoid lumen, which was determined in their study to be 12.7 ± 2.9 nm under their growth conditions; however, this width can increase under low irradiance levels and under such conditions ∼10 to 13 PE molecules would be required. Their working model for the organization of the PBP in the lumen suggests that HaPE645 is tethered to the thylakoid membrane and that HaPE555 and HaPE560 form a tightly packed matrix in the lumen, perhaps organized as rods or rafts as seen in the crystal packing. It will be interesting to see if cryo-ET can be applied to determine at long last the structural organization of proteins in cryptophytes.

Although no structural information currently exists for any cryptophyte PSII complex, a very recent study has determined the structure of an 885-kDa PSI-LHCI supercomplex from Chroomonas placoidea at 2.7-Å resolution (Zhao et al. 2023). The LHCI antenna complex comprises fourteen alloxanthin-Chl a/c-binding proteins, designated as ACP1, which together with a specialized subunit, ACP-1S, surround a PSI monomer. In total, this supercomplex binds 373 Chls and carotenoid molecules and contains 2 proteins that have not been found in other related structures, ACP-1S and an unknown protein (Unk1). The supercomplex has dimensions of 205 × 195 × 110 Å, and it binds 254 Chl a molecules, 20 Chl c molecules, 59 alloxanthins, 25 α-carotene molecules, 12 crocoxanthin molecules, 3 monadoxanthin molecules, 3 [4Fe4S] clusters, and 39 lipid and detergent molecules. The monomeric PSI core complex comprises 14 subunits: Psa: A, B, C, D, E, F, I, J, K, L, M, R, O, and Unk1 (Zhao et al. 2023). The 9 subunits in bold font occur in all PSI complexes from organisms that evolve oxygen. PsaK is absent in members of the Gloeobacteria and diatoms, and PsaM is absent in plants. PsaO does not occur in cyanobacteria or diatoms, and PsaR occurs in diatoms but not cyanobacteria. The cryptophyte PSI complexes are also missing PsaG, PsaH, and PsaN, which are found in PSI in green algae and plants, and is also missing PsaX, which is uniquely found in some but not all cyanobacteria. The PSI monomer binds 100 Chl a but no Chl c, and it shares all of those binding sites with the PSI complexes of the PSI complexes of red algae and diatoms. The monomer core also binds 26 carotenoids: 19 α-carotenes, 5 alloxanthins, and 2 crocoxanthins. The cryptophytan PsaR subunit binds 1 Chl a and 2 carotenoids, and is similar in properties to PsaR in diatoms. PsaK and PsaO are similar to their homologs in red algae. PsaK binds 2 Chl a and 2 carotenoids, and it provides a docking site for 2 subunits of ACP1, subunits 1 and 11. PsaO binds 3 Chl a and 2 alloxanthins, one of which could play a role in quenching excess light. The Unk1 protein binds to the lumenal side of the PSI monomer near PsaB, PsaM, and ACP1-7. The 14 ACP1 antenna subunits are not identical and are the products of different genes. In the current model, they bind 8 to 13 Chl a molecules, and all but ACP1-8 binds 1 or 2 Chl c molecules. Each ACP1 subunit also binds 5 or 6 carotenoids, which includes 3 to 5 alloxanthin molecules. The novel subunit ACP1-S is an exception as this 20.2-kDa subunit binds only 3 Chl a, 1 Chl c, and 2 α-carotenes. For further details and in-depth comparisons to PSI complexes of red algae and diatoms, the reader should consult Zhao et al. (2023).

Photosynthetic apparatus in Acaryochloris spp. and chloroxybacteria

The initial description of A. marina in 1996 marked the first example of an organism that could utilize FRL to perform oxygenic photosynthesis (Miyashita et al. 1996; for a review of some general properties of A. marina, see Loughlin et al. 2013). Although the first isolate was derived from a marine asicidian species, subsequent studies showed that Chl d-producing cyanobacteria also could be found growing on the under-surfaces of fronds of a marine red alga (Murkami et al. 2004). Further studies have shown that Acaryochloris spp. can be found as epiphytes on diverse red algae, as ectosymbionts of tunicates, as epiliths, and in many other marine and freshwater environments (Kühl et al. 2005; Loughlin et al. 2013; Miller et al. 2022). A. marina produces mostly Chl d and only a very small amount of Chl a (∼1% to 10% of total Chl depending on the strain and the growth conditions) for its photosynthetic apparatus (Miyashita et al. 1996; Murakami et al. 2004). One of the most studied isolates, A. marina MBIC11017, has genes to produce PC along with multiple linker proteins related to CpcG2/CpcL, that can probably tether PC rods to membrane-bound photosystem complexes as described above (Swingley et al. 2008). Surprisingly, most other strains of A. marina lack the genes to produce PBP. Interestingly, an early diverging member of the genus, Acaryochloris thomasi, was isolated as a benthic epilith from rocks along the coast of France near Roscoff (Partensky et al. 2018). Unlike all other isolates of the genus Acaryochloris, A. thomasi completely lacks Chl d but instead has Chl a and Chl b in a ratio similar to that found in chloroxybacteria. A. thomasi has genes to produce 7 CBP antenna proteins; however, this strain also produces PC that likely forms rod-shaped antenna complexes similar to the rod-shaped CpcL-PC-FNR complexes described above. It has been suggested that the ancestors of Acaryochloris spp. had lost the ability to produce PBP and that the 2 strains that have regained PBP, specifically PC, obtained the capacity to produce it again by horizontal gene transfer (Ulrich et al. 2021). Overall, genomic analyses indicate that horizontal gene transfer has been very important in the evolution of the members of this genus (Ulrich et al. 2021; Miller et al. 2022). In A. marina MBIC11017, all genes required to produce PC (hemO, pcyA, cpcB, cpcA, cpcC, cpcD, cpcE, cpcF, and cpcG2, are encoded on a large plasmid, pREB3 (Swingley et al. 2008; Ulrich et al. 2021). This strain also has a gene similar to apcB, which encodes the β-subunit of AP, and additionally has a gene very distantly related to apcA but lacking the conserved chromophore binding site. However, no AP has yet been convincingly demonstrated to occur in any strain of Acaryochloris spp. It has been suggested that this ApcA paralog may play a role in the assembly of PSII (Partensky et al. 2018; Ulrich et al. 2021). Finally, a recent study of spectral differences among strains of A. marina indicates that there are 3 groups of strains reflecting variants that absorb at shorter, intermediate, and longer wavelengths within the far-red range (700 to 800 nm) (Ulrich et al. 2024). These variants presumably arose after the evolution of Chl d production to accommodate different FRL-enriched environmental niches (Ulrich et al. 2021).

Chloroxybacteria, also sometimes referred to as green oxyphotobacteria, include members of the genera Prochloron, Prochlorothrix, and Prochlorococcus (and now A. thomasi as just described above). All members of these genera produce Chl a, Chl b, or divinyl-Chl a and Chl b in the case of Prochlorococcus spp., and carotenoids including zeaxanthin and α-carotene but not β-carotene. Most chloroxybacteria have small multigene families that encode CBP proteins, which are members of the CP43 superfamily with 6 transmembrane α-helices. Members of this protein family are widely distributed in cyanobacteria but have not been found as antenna proteins in eukaryotes to date. Most chloroxybacterial strains except Prochlorococcus strain MED4 have 1 CBP that is strongly induced by iron-stress and produces a ring-like antenna complex for PSI. This has been shown to be the case for A. marina, in which an 18-membered ring of PcbC surrounds trimeric PSI as described above for other cyanobacteria undergoing iron stress (Chen et al. 2005a; Li et al. 2018). Of course, in A. marina, the complex binds Chl d rather than Chl a as found for IsiA described above for typical cyanobacteria. It is also the case for Prochlorococcus strain SS120 (Bibby et al. 2001b). Prochlorococcus sp. strain MIT 9313 has only 2 genes encoding CBP/Pcb proteins, and PcbB is induced by iron stress and produces an 18-membered ring that surrounds a trimeric PSI complex. The other, PcbA, is constitutively expressed and produces 4-membered arcs that can bind to 1 or 2 sides of a PSII dimer (Bibby et al. 2003a). Prochlorococcus strain MED4 has only a single pcb gene, pcbA, which produces an antenna complex for PSII, while the SS120 strain, a low-light ecotype, has 8 pcb genes, 1 of which, pcbC, is induced more than 20-fold under iron limitation conditions and presumably makes the antenna for PSI as previously suggested (Bibby et al. 2001b, 2003a). Prochloron didemni produces a CBP antenna comprising 2 arcs with 4 Pcb proteins that surround a PSII dimer (Bibby et al. 2003b). Finally, A. marina PSII associates with 16 CBP arranged as 2 double arcs with 8 Pcb molecules, which form a large Chl d supercomplex surrounding a PSII tetramer (Chen et al. 2005b). Prochlorothrix hollandica can also produce an 18-membered ring of CBP/Pcb ring that surrounds a PSI trimer, so this is a trait found in most chloroxybacteria (Bumba et al. 2005). Although there are 2 other Pcb proteins produced by P. hollandica, stable complexes with PSII have not yet been reported (Herbstová et al. 2010).

The A. marina PSI complex has been isolated as a monomer and as a trimer, and the structure of the trimer has been determined by 2 groups (Hamaguchi et al. 2021; Xu et al. 2021). Although the results were similar for both studies, significantly higher resolution, 2.58 Å vs 3.3 Å, was achieved by Hamaguchi et al. (2021). The complex comprises 11 subunits: Psa: A, B, C, D, E, F, J, K, L, M), and Psa27, a functional and structural homolog of PsaI. Each monomer in the trimer binds significantly fewer Chls than the PSI complex of T. elongatus (71 vs 95 Chl a). The monomer appears to have differences in the Chl complement of long-wavelength absorbing pigment oligomers because the absorbance and 77-K fluorescence emission properties of the complex are altered compared with the trimeric state. As anticipated from spectroscopic studies, the P740 special pair is a dimer of Chl d and its C13 epimer, Chl d′; the complex binds 70 Chl d, 1 Chl d′, 2 pheophytin a molecules (at the A₀ positions), 2 phylloquinones, 12 α-carotenes (and no β-carotene), and 3 [4Fe-4S] clusters. A small amount of Chl a, equivalent to 1 Chl a per Chl d′, was detected (Hamaguchi et al. 2021; Nagao et al. 2023a); however, the location of the single Chl a has not yet been determined.

In spite of the fact that antenna supercomplexes in chloroxybacteria and Acaryochloris spp. have been extensively studied over the past 25 years, surprisingly, no high-resolution structure for any antenna complexes formed by a member of the CP43/CBP family of antenna proteins for Acaryochloris spp. had been published until very recently. As noted above, structures have recently appeared from 2 groups for PSI complexes, and it is reasonable to expect that complexes with bound Pcb/CBP antenna proteins will soon follow. Importantly, structures of a PSII dimer with 2 arcs of 4 Pcb/CBP antenna proteins bound at 3.3-Å resolution, and of a PSII tetramer with 16 Pcb/CBP proteins bound at 3.6-Å resolution have just been published (Shen et al. 2024). The megacomplex is quite large, accounting for 80 protein subunits, 624 cofactors, and a mass of ∼1.9 mDa; the dimensions of the megacomplex are ∼340 Å (length)×200 Å (width) and 90 Å (height). The antenna complexes are arranged as a pair of 4-membered arcs on 2 sides of the PSII tetramer. Four different protein subunits, PcbC2, PcbA3, PcbA6, and PcbA2, are found in each of the 4 arcs; each arc binds 69 Chl d and 14 zeaxanthin molecules. The PSII core comprises 15 polypeptide subunits, 37 Chl d, 2 pheophytin a, 1 heme, 1 non-heme iron, and 11 α-carotene molecules. It will be quite interesting to see the differences required to alter IsiA to accommodate either Chl d instead of Chl a, although the details of Chl d binding will likely require higher resolution than is currently available for this first structure.

Concluding remarks

PBP are versatile light-harvesting proteins that can greatly expand the absorbance cross-sections of cells that produce them by filling the gap between the Soret (blue) and Q_y (red) absorbance bands of Chls. They also provide an advantage because they are water soluble and can fit chromophores into a 3-dimensional volume in the cytoplasm instead of the largely 2-dimensional space in the thylakoid membrane, which must also provide space for the photosystems and other membrane proteins, such as ATP synthase, transporters, and electron transport complexes. A disadvantage of PBP antenna structures is that they are very expensive energetically for the cell to build, requiring large amounts of extremely valuable resources (reduced carbon, nitrogen, and sulfate; and ATP) that cells require for growth. PBP are much more expensive to produce than Chl proteins in requiring about 80 to 160 amino acids to bind a single bilin chromophore, depending on whether the protein is PE or AP. Chl proteins used in photosynthesis require only ∼15 to 20 amino acids to bind 1 Chl, not even accounting for the carotenoids these proteins also bind. This certainly provides a strong selection against using PBP as antenna complexes and may partly explain why PBS were lost by many lineages during evolution (see Green 2019).

Because Chl proteins reside in the thylakoid membrane (or worse still in the cytoplasmic membrane when there are no thylakoid membranes), it becomes a severe problem to keep adding more and more CBP/CAB proteins to a membrane to act as antenna for a photosystem. The photosystems or reaction centers become embedded in a sea of Chls that are mostly similar in energy, which precludes an efficient biased random-walk for energy transfer to the special pair. As more subunits are added to the periphery of such complexes, the distance and time for energy transfer to the special pair must increase. This can limit the overall efficiency of photon utilization by an organism. There must be some practical limit beyond which adding more subunits to a supercomplex becomes counterproductive. In this respect, PBP that use the water-soluble space adjacent to the membrane provide an obvious advantage.

To counteract these distance effects, Chl proteins have densely packed Chls so that interpigment distances are short, enabling very rapid energy transfer. It is interesting that evolution has converged on using ring-like arrangements of Chls to facilitate rapid delocalization of excitation energy. Proteins of the CP43 superfamily (e.g. IsiA and IsiX) and proteins of the CAB superfamily (e.g. LHCI proteins) of antenna complexes in cyanobacteria and algae, respectively, have converged on rings of Chls as a consistent theme in the structure of light-harvesting complexes. Rings and partial rings (arcs) are reminiscent of the simpler types of LH1 and LH2 light-harvesting complexes found in purple bacteria (Bryant and Canniffe 2018). Double-layer rings of Chls near the stromal and lumenal surfaces of IsiA (Fig. 9), in the core antenna domains PSI and PSII sometimes within subunits of these proteins emphasizes the benefits of rings of pigments in antenna design during evolution.

The last decade, and especially the last 5 years, have provided remarkable advances in the structural biology of LHC in cyanobacteria and red algae. What was once unimaginable—the idea that one could determine relatively high-resolution structures for objects as large and complex as PBS—has now become so routine that a recent study comparing the structures of 13- to 15-mDa-PBS was performed to describe the structural and compositional changes that occur in red algal PBS in response to different physiological conditions (Dodson et al. 2023). The realization that FRL can be used to drive oxygenic photosynthesis has also reenergized research in photosynthesis (summarized in Elias et al. 2024; Gisriel 2024). This topic partly emerged from ecophysiological investigations conducted on microbial mats and stromatolytes. An important lesson from these studies is the importance of “letting Nature be the guide.” While it is obvious that model organisms and ex situ/in labo studies have greatly enhanced our understanding of photosynthesis and many other processes, it is still important to remember that nonmodel organisms and in situ studies are often the surest route to the discovery of new processes and novel innovations in photosynthesis, and indeed in science in general.

Funding

Research by the authors reported in this publication was supported by the National Science Foundation grant MCB-1613022 to D.A.B. and the National Institute of General Medical Sciences of the National Institutes of Health under Award Number K99GM140174 to C.J.G. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Data availability

No new data were generated or analysed in support of this research.