Introduction

The PBS structure

Phycobilisome (PBS) is a large, multi-subunit light-harvesting protein complex with a mass of several MDa found in cyanobacteria, glaucocystophytes, and rhodophytes (Watanabe and Ikeuchi 2013). PBSs are composed of phycobiliproteins (phycobilin chromophore binding proteins, PBP) and of several colorless linker proteins, which serve for the PBS structure assembling and also for optimization of chromophores spectral properties (Füglistaller et al. 1987; Liu et al. 2005; Nganou et al. 2016). PBS consists of the core subcomplex mainly composed of allophycocyanin (APC) and of a number of peripheral rods composed of phycocyanin (PC) or phycocyanin and phycoerythrin (Bryant 1991; Sidler 1994; Watanabe and Ikeuchi 2013).

The structure of the PBS core varies for different organisms. In Anacystis nidulans (Synechococcus sp. PCC 6301), the PBS core is composed of two APC cylinders parallel to each other (Lundell and Glazer 1983b; Anderson and Eiserling 1986). The tricylindrical PBS cores were found in various species of Synechocystis (Bryant et al. 1979; Arteni et al. 2009), Synechococcus (Bryant et al. 1979; Mörschel and Schatz 1987), and Thermosynechococcus elongatus (Barber et al. 2003), as well as in Nostoc flagelliforme (Yi et al. 2005) and Calotrhix sp. (Rosinski et al. 1981), and some red algae like Rhodella violacea (Reuter et al. 1990) or Galdieria sulphuraria (Stadnichuk et al. 2011). In this type of the core, three APC cylinders form a triangle structure: two tightly packed parallel basal cylinders contact the thylakoid membrane surface and one upper cylinder is located in parallel with them (Fig. 1). Core structures of PBSs from Anabaena sp. PCC7120 (Glauser et al. 1992; Ducret et al. 1998; Chang et al. 2015), Mastigocladus laminosus (Glauser et al. 1992), Thermosynechococcus vulcanus (David et al. 2014), and Gloeobacter violaceus (Guglielmi et al. 1981; Koyama et al. 2006) are composed of five cylinders: three cylinders are arranged in the same way as in the tricylindrical core and two short (one APC hexamer) cylinders are perpendicular to them (Chang et al. 2015). The unique PBSs of Acariochloris marina (Marquardt et al. 1997; Hu et al. 1999) are composed of only one mixed PC–APC cylinder.

Fig. 1
figure 1

Schematic side (A) and bottom (B) views of the tricylindrical PBS core (\({\text{D}_\text{3h}}\) symmetry group). Arrows indicate two-fold (L2) and three-fold (L3) symmetry axes. Two-fold axes lay in the mirror plane (not shown). Chirality of the APC trimer and TEs location break the symmetry of the PBS core down to C2 with only one vertical L2 axis remaining (A). Numbers correspond to the APC trimers composition: #1: \((\alpha \beta )_3\), #2: \((\alpha \beta )_3\) \({\text{L}_\text{C}}\), #3: \((\alpha \beta )_2\)(ApcD/\(\beta\))\({\text{L}_\text{C}}\), and #4: \((\alpha \beta )_2\)(\({\text{PBL}_\text{CM}}\)/ApcF). The mutual arrangement of trimers #3 and #4 is under discussion

Full size image

Each core cylinder is formed by two hexamers of APC \(\alpha \beta\) heterodimers each carrying two chromophores. Every basal cylinder contains one copy of each minor APC subunit: ApcD, or \(\alpha^{\text{B}}\) APC (Lundell and Glazer 1981); ApcE, or \({\text{L}_\text{CM}}\) (Capuano et al. 1991); and ApcF, or \(\beta ^{18}\)-subunit (Gingrich et al. 1983; Gindt et al. 1994). ApcD and ApcE carry long-wave chromophores and serve as independent terminal emitters (Lundell and Glazer 1983b; Mimuro et al. 1986). According to mutational analysis, the ApcD chromophore was reported to serve as an energy donor to PSI, while ApcE serves as a donor to PSII (Zhao et al. 1992). These observations are in good agreement with the results of protein cross-linking experiments that revealed interaction between ApcD subunit and PSI, and between chromophore carrying domain of \({\text{L}_\text{CM}}\) (\({\text{PBL}_\text{CM}}\)) and PSII (Liu et al. 2013).

The \({\text{PBL}_\text{CM}}\) substitutes \(\alpha\)APC subunit, while \(\beta ^{18}\) polypeptide substitutes \(\beta\)APC subunit in the same APC trimer (Lundell and Glazer 1983c; Anderson and Eiserling 1986). Furthermore, it is very likely that \({\text{PBL}_\text{CM}}\) and \(\beta ^{18}\) form a heterodimer (Lundell and Glazer 1983c). \({\text{PBL}_\text{CM}}\)/\(\beta ^{18}\) heterodimer is always located in the middle trimer of the PBS core basal cylinder, while ApcD is always a part of one of the edge trimers (Lundell and Glazer 1983b; Gingrich et al. 1983; Anderson and Eiserling 1986).

It is still unclear whether the ApcD-carrying trimer contacts the \({\text{PBL}_\text{CM}}\)-carrying trimer or not, and this problem was widely discussed (Yamanaka et al. 1982; Lundell and Glazer 1983b; Anderson and Eiserling 1986; Gindt et al. 1994). Due to the similarity between ApcA and ApcD (Lundell and Glazer 1981), the single-particle EM also does not provide any information on their mutual arrangement (Chang et al. 2015). Nevertheless, it is often accepted that ApcD and \({\text{PBL}_\text{CM}}\) are located in the adjacent trimers of the basal cylinder (Watanabe and Ikeuchi 2013; Kirilovsky 2015; Kirilovsky and Kerfeld 2016). At the same time, the protein cross-linking technique did not reveal any links between ApcD and ApcE or ApcF (Liu et al. 2013; Tal et al. 2014; Harris et al. 2016), which favors the remote arrangement of ApcD and \({\text{PBL}_\text{CM}}\) in the basal cylinder.

There are two types of the linker polypeptides in the PBS core structure: \({\text{L}_\text{C}}\) (ApcC) and mentioned above \({\text{L}_\text{CM}}\) (Watanabe and Ikeuchi 2013). The multi-domain \({\text{L}_\text{CM}}\) polypeptide ties up all the PBS core components and plays an important role in PBS association with the thylakoid membrane (Capuano et al. 1993; Watanabe and Ikeuchi 2013; Liu et al. 2013). The \({\text{L}_\text{C}}\) linker is located in each edge APC trimer of the core cylinders (Lundell and Glazer 1983c; Gingrich et al. 1983; Anderson and Eiserling 1986). According to the X-Ray data, the \({\text{L}_\text{C}}\) linker is located deeply inside the APC trimer and should not be considered as a capping protein terminating the APC trimers association to the core cylinders (Reuter et al. 1999).

There are four APC trimer types in the PBS core: #1: \((\alpha \beta )_3\), #2: \((\alpha \beta )_3\) \({\text{L}_\text{C}}\), #3: \((\alpha \beta )_2\)(ApcD/\(\beta\))\({\text{L}_\text{C}}\), and #4: \((\alpha \beta )_2\)(\({\text{PBL}_\text{CM}}\)/ApcF), as it was demonstrated for bicylindrical (Lundell and Glazer a, b, c) and tricylindrical cores (Gingrich et al. 1983; Anderson and Eiserling 1986). The upper cylinder does not contain any minor long-wavelength components and always has a composition of 2:1:1:2. Composition of the basal cylinders is more complex, as they contain terminal emitters (TEs). The order of trimers in the first basal cylinder would be 2:1:4:3 if ApcD and ApcE are located in the adjacent trimers, or 2:4:1:3 in case of separated TEs (Gindt et al. 1994). Then, the second cylinder can have the same (2:1:4:3 or 2:4:1:3) as well as inverse (3:4:1:2 or 3:1:4:2) arrangement (Fig. 1B). C\(_2\) symmetry of the PBS core detected through the single-particle EM implies antiparallel arrangement of APC trimers in basal cylinders (Arteni et al. 2009; Chang et al. 2015). Thus, the arrangement of the trimers in basal cylinders is (2:1:4:3, 3:4:1:2) or (2:4:1:3, 3:1:4:2). In the current study, we demonstrate that the second option is more probable according to the analysis of circular dichroism (CD) spectral data (Lundell and Glazer 1983b), results of protein cross-linking (Liu et al. 2013), and due to the peculiarities of PBS interaction with the photosystem II (PSII).

Crystallization of PBSs, in contrast to various individual PBPs, is significantly hindered by its soft structure and by the presence of long unstructured loops in the \({\text{L}_\text{CM}}\) polypeptide (Capuano et al. 1991). As a result, no fine crystal structure of the PBS core was yet obtained. The structures of recently crystallized PBSs were not resolved due to random positioning of the core components in the crystals (David et al. 2014). At the same time, the structures of entire PBSs of N. flagelliforme (Yi et al. 2005) and then of Anabaena sp. PPC 7120 (Chang et al. 2015) were investigated using a single-particle EM that allowed authors to develop molecular models of entire PBS with tri- and pentacylindrical cores. The detailed molecular model of pentacylindrical PBS was obtained by Chang et al. (2015) by the docking of crystal structures into the cryo-EM images. In our work, we use an alternative model of tricylindrical PBS core obtained on the basis of the structure of APC crystal lattices (Zlenko et al. 2016b).

EM observations

PBSs serve as an antenna of PSII (Melis 1991; Mimuro et al. 1999; Kirilovsky 2015). At the same time, a part of the energy absorbed by PBPs is transferred to PSI (Wang et al. 1977; Mullineaux 1994; Rakhimberdieva et al. 2001; Dong et al. 2009). Depending on illumination conditions, photosynthetic organisms can promptly redistribute the energy of absorbed light between the photosystems (Murata 1969; Bonaventura and Myers 1969; Mullineaux 1994). The mechanism of short-term adaptation to unbalanced light absorption by PSI and PSII is known as state transitions. A PSI-specific light adjusted state is called “State 1” and PSII-specific light-adapted state is known as “State 2” (Mullineaux and Allen 1990; Delphin et al. 1995; Fork and Mohanty 2012). In cyanobacteria, state transitions are accompanied by distinct thylakoid membrane reorganization. In State 1, the EF particles (PSII dimers) are arranged in rows clearly visible in the freeze-fracture EM images, while the transition to State 2 leads to the randomization of their positioning (Giddings et al. 1983; Olive et al. 1986, 1997).

The presence of PBSs on the thylakoid membrane surface significantly affects both the formation of the PSII dimer rows and thylakoids structure (Olive et al. 1997; Folea et al. 2008). Different EM data show that, on the thylakoid membrane surface, PBSs are arranged in rows parallel to the PSII dimer rows, and each PSII dimer contacts exactly one PBS (Lefort-Tran et al. 1973; Lichté and Thomas 1976; Giddings et al. 1983; Mörschel and Mühlethaler 1983; Arteni et al. 2009). The center-to-center distance between the EF particles in the rows was reported to be in the range from \(\sim10\) nm (Lichté and Thomas 1976; Giddings et al. 1983) to \(\sim\) 12 nm  (Mörschel and Mühlethaler 1983; Mörschel and Schatz 1987; Westermann and Wehrmeyer 1995; Kuhl et al. 1999; Dekker et al. 1988). The maximum reported value was 14 nm (Nilsson et al. 1992). At the same time, in the native-like rows of the dimers found in the crystal lattice of PSII from Th. elongatus, the observed distance equals 11.6 nm (Hellmich et al. 2014). According to the results of the transmission EM observations of solubilized photosynthetic membranes from Synechocystis sp. PCC 6803, the precisely measured PSII dimers row pitch equals 12.2 nm (Folea et al. 2008). Thus, the pitch of the PSII dimers row seems to be close to the value of 12.2 nm (Folea et al. 2008; Arteni et al. 2009; Hellmich et al. 2014).

In the early papers, the thickness of the PBSs from red alga Rhodella violacea (Mörschel et al. 1977) and cyanobacterium A. nidulance (Khanna et al. 1983) was reported to be \(\sim\) 7 nm that is far below the values obtained in the later studies. According to the single-particle EM, the thickness of the tricylindrical PBS core is 15.5 nm for Synechocystis sp. PCC 6803 (Arteni et al. 2009) or 11–12 nm for N. flagelliforme (Yi et al. 2005) and Th. vulcanus (David et al. 2014). The thickness of the pentacylindrical PBS core from Anabaena sp. PCC 7120 equals 14.4 nm (Chang et al. 2015). According to the transmission EM observations of individual PBSs, the core thickness equals 10–11 nm (Wildman and Bowen 1974; Yamanaka et al. 1980) or \(\sim\) 14 nm (Bryant et al. 1979; Williams et al. 1980; Anderson and Eiserling 1986). Thus, the thickness of the isolated PBS core seems to vary between two values of \(\sim\) 11 and \(\sim\) 15 nm. The upper bound estimate of the PBS core thickness is approximately 3 nm greater, that the common PSII dimers row pitch and than the PBS row pitch that was also reported to be \(\sim\) 12 nm (Mörschel and Mühlethaler 1983; Mörschel and Schatz 1987; Giddings et al. 1983). This mismatch becomes crucial in the case of one-by-one arrangement of PBS and PSII in the rows observed in EM images (Lichté and Thomas 1976; Giddings et al. 1983; Mörschel and Mühlethaler 1983) and the origin of this problem is still under discussion (Bryant et al. 1979; Mörschel and Mühlethaler 1983; Yi et al. 2005; Arteni et al. 2009; Hellmich et al. 2014).

The difference in PBS core thickness and PBS/PSII rows pitch cannot be explained by the variability of the structure among the species. Indeed, there are multiple examples of such mismatches in the same species: in Synechococcus sp., the thickness of the PBS core was reported to be \(\sim\) 14 nm (Bryant et al. 1979), while the row pitch was \(\sim\) 12 nm (Mörschel and Mühlethaler 1983; Dekker et al. 1988; Kuhl et al. 1999); in Synechocystis sp., the core thickness was \(\sim\) 14 nm (Bryant et al. 1979; Williams et al. 1980; Anderson and Eiserling 1986; Arteni et al. 2009), while the pitch of PSII rows was 12.2 nm (Folea et al. 2008). Moreover, the core thickness was reported to be different in Anabaena genus: \(\sim\) 10 nm (Wildman and Bowen 1974) or 14.4 nm (Chang et al. 2015).

Bryant et al. (1979) explained the mismatch between the PBS core thickness in different studies by the resolution of PBS core halves in thin sections (Wildman and Bowen 1974; Lichté and Thomas 1976; Mörschel et al. 1977). However, in the later studies, the row pitch and PBS core thickness were still reported to be significantly different (of about 12 nm and 10–15 nm, respectively, as mentioned above). These facts, together with the recent excellent EM images of PSII rows in the thylakoid membrane (Folea et al. 2008) and of PBSs (Yi et al. 2005; Arteni et al. 2009; Chang et al. 2015), make the mistake in EM measurements to be a very unlikely reason of the observed discrepancy.

The PBS core thickness assessment

The thickness of isolated PBSs varies in the range from 10 to 15 nm (see above). At the same time, the results of PSII dimers and PBSs row pitch measurements give approximately the value of 12 nm in all the studies (see above), and thus look more reliable. According to the X-ray data, the width of a single cyanobacterial (Brejc et al. 1995; Reuter et al. 1999; Murray et al. 2007; McGregor et al. 2008; Sonani et al. 2015; Peng et al. 2014; Marx and Adir 2013) and red algal (Liu et al. 1999) APC trimer is about \(\sim\) 3 nm. Thus, if we assume that the PBS core thickness is 14–15 nm, there would be a gap of \(\sim\) \(1\) nm between the APC trimers in the core cylinder or a gap of \(\sim\) \(3\) nm between two hexamers. Otherwise, the APC trimers would be a bit swollen and their structure would be different from the structure observed in crystals.

The gap between the APC hexamers was observed in the cores of isolated PBSs obtained from Lyngbya–Plectonema–Phormidium PCC 7409 (Bryant et al. 1979). However, this observation have not been confirmed by more recent results showing that the trimers in the core are distributed uniformly and that there is no gap between the APC hexamers (Yamanaka et al. 1980; Yi et al. 2005; Arteni et al. 2009; Chang et al. 2015). Note that even 1-nm gap is very large in the molecular scale and has to be filled with water that would decrease the energy transfer efficiency in the PBS core. The presence of the water-filled gaps between the APC trimers or hexamers would mean that there is no direct protein–protein interaction between them. Thus, the gap between the APC trimers or hexamers contradicts the CD data obtained by Lundell and Glazer (1983b) for A. nidulans that reveal such interaction.

Loosening of the APC hexamers structure was observed by X-Ray crystallography under some special crystallization conditions, but this effect was caused by the APC trimers segregation, rather than by the trimers swelling (Marx and Adir 2013). The core swelling seems to be in fewer contradiction with the CD spectral data (Lundell and Glazer 1983b), than the proposition that there are the water-filled gaps between the trimers. The core swelling is also in agreement with the EM observations of the individual PBSs (Yamanaka et al. 1980; Arteni et al. 2009). Nevertheless, the thickness of the swollen PBS core (\(\sim\) 14 nm) is very large and such cores could not be packed in the row in one-by-one arrangement with PSII dimers (Bryant et al. 1979; Yi et al. 2005; Arteni et al. 2009; Hellmich et al. 2014).

In the model of Anabaena sp. PCC 7120 with a PBS core of 14.4 nm thick developed by Chang et al. (2015), the gaps between the APC trimers are absent, but the trimers are tilted with respect to the core cylinder axis. This tilt is clearly seen in the EM images (Chang et al. 2015), but at the same time, it was not observed either in N. flagelliforme (Yi et al. 2005) or in Synechocystis sp. PCC 6803 (Arteni et al. 2009) and was never reported elsewhere earlier. Beyond that, according to this model, there are no \({\text{L}_\text{C}}\) linkers inside the edge APC trimers of the upper core cylinder, while in the tricylindrical cores this linker proteins are located in the upper cylinder (Gingrich et al. 1983; Anderson and Eiserling 1986).

Finally, taking into account very high methodological level of the recent single-particle EM investigations of PBSs (Arteni et al. 2009; Chang et al. 2015), the only way to resolve the contradiction between the PBS core thickness and the PBS row pitch is to propose that the structures of PBSs dissociated from thylakoid and of PBSs packed in the row are slightly different. PBS core structure can become a little bit more loose in solution as compared to its structure in the row. In this case, the swelling of the free PBS can explain the low crystallization ability of the entire PBS as well as the core.

The modeling strategy

The mean backbone RMSD between the APC trimers among the different species of cyanobacteria and red algae lay below 1 Å that means that they are almost identical (Zlenko et al. 2016b). Thus, we can assume that tricylindrical cores of PBSs from different species have almost the same structure. Using this assumption, earlier we have developed a crystal structure-based model of the APC packing in the most common tricylindrical PBS core (Zlenko et al. 2016b). The obtained model does not contain either \({\text{L}_\text{C}}\) linkers or REP domains of \({\text{L}_\text{CM}}\) and could be used only for the analysis of APC mutual arrangement and packing. The APC molecules are the most bulky components of the core, while the Pfam_01383 (Reuter et al. 1999) and Pfam_00427 (Gao et al. 2011) linker domains lie completely inside the PBP trimer’s cavity. Thus, the dense packing of the components in the PBS core and also of the cores in the row requires the dense packing of APC trimers. The obtained model core implies the densest possible packing of APC trimers and maximal theoretical efficiency of energy transfer inside the core. Moreover, all the protein–protein interfaces in the proposed model precisely reproduce those from crystal lattices of APC from different species (Brejc et al. 1995; Liu et al. 1999; McGregor et al. 2008).

The developed model of PBS core can be used for reconstruction of PBS and PSII arrangement in the coupled rows. The cyanobacterial photosynthetic apparatus is very flexible and PBSs and photosystems can form several other supramolecular complexes besides rows. The PSII dimers can form 2D crystal-like domains in the solubilized thylakoid membranes of Synechocystis sp. PCC 6714 (Folea et al. 2008). The similar 2D domains were observed for PSII dimers and hemielipsoidal PBSs in the red alga Porphyridium cruentum (Arteni et al. 2008). The PBS–PSII–PSI supercomplexes were recently isolated from Synechocystis sp. PCC 6803 (Liu et al. 2013) and their structure was characterized by protein cross-linking technique. The supercomplexes of tetrameric PSI and the PC-cylinders were found in Anabaena sp. PCC 7120 (Watanabe et al. 2014). As compared to other complexes, the structure of coupled rows imposes significant restrictions in the possible mutual orientation of PBS and PSII (Bald et al. 1996), as the PBS cores must be tightly packed to provide a pitch of 12.2 nm (Folea et al. 2008). These restrictions allow us to analyze spatial arrangement of PBS core and PSII dimer theoretically.

The pitch of 12.2 nm obtained by Folea et al. (2008) is accepted to be the closest to the native PSII rows state (Arteni et al. 2009; Hellmich et al. 2014), and we also took 12.2 nm as a reference value. While PBS structure in solution seems to be a bit swollen, the crystal-based model (Zlenko et al. 2016b) fits for modeling of squeezed state of PBS cores in the rows. Indeed, the core model obtained by Chang et al. (2015) has a thickness of 14.4 nm that makes it incompatible with the PBS rows structure. At the same time, the thickness of the PBS core in our model (Zlenko et al. 2016b) is 12.4 nm that is only 2 Å greater than the PSII dimers row pitch measured by Folea et al. (2008). Here, we reconstructed the mutual arrangement of the PBS core and PSII dimer analyzing both spatial matching of the PBS cores and PSII dimers, and proposed efficiency of energy transfer on the basis of our crystal-based PBS core model (Zlenko et al. 2016b). We proposed the most probable positions of terminal emitters in the core and identified the chlorophyll molecules of PSII gathering excitation energy from the PBS. Although we considered the most common tricylindrical core, we believe that all of the obtained results can be extrapolated to bi- and pentacylindrical cores of other cyanobacteria.

Results and discussion

PBS core symmetry

APC trimers in the PBS core scheme are usually depicted as flat disks with a hole in the center (Fig. 1). If we assume that PBS core contains only bulk APC molecules, then it would have a symmetry group \({\text{D}_\text{3h}}\) (Fig. 1A). In such a model, the main three-fold symmetry axis goes between the cylinders in parallel with them (Fig. 1, L\(_3\)), and three two-fold axes go through the center of each cylinder perpendicular to the main axis (Fig. 1, L\(_2\)). Two-fold axes lay in the mirror plane that dissects the structure between the APC hexamers. However, this simple model does not take into account the real shape of APC trimer and its chirality.

According to the X-Ray data, the shape of APC trimers is more complex than a flat disk. Native APC trimers are chiral and are always “right-twisted” (Fig. 2A), while the “left-twisted” trimers (Fig. 2B) are never observed (Brejc et al. 1995; Liu et al. 1999; Reuter et al. 1999; Murray et al. 2007; McGregor et al. 2008; Marx and Adir 2013; Peng et al. 2014; Sonani et al. 2015). As a result, in the real PBS core the mirror plane disappears, and the symmetry breaks down from \({\text{D}_\text{3h}}\) to D\(_3\). Due to the localization of TEs and antiparallel APC disks arrangement in the basal cylinders (Gindt et al. 1994; Arteni et al. 2009; Marx and Adir 2013; Chang et al. 2015), the symmetry of PBS core further breaks down from D\(_3\) to C\(_2\) that was observed by Arteni et al. (2009) and Chang et al. (2015). Three-fold symmetry axis disappears and there remains only one two-fold symmetry axis that goes through the center of the upper cylinder between the basal cylinders (Fig. 1A, vertical L\(_2\) axis).

Fig. 2
figure 2

Concave side view of the APC trimer from Spirulina platensis (1ALL, Brejc et al. 1995). The experimentally observed right-twisted (A) and hypothetical left-twisted (B) structures. Left-twisted trimer was obtained by mirroring of the right-twisted trimer. The protein surfaces are shown in gray, and the chromophore molecules are shown with purple sticks. Black arrows indicate the APC tip and the RCP (N-terminal domain of OCP) binding site (Stadnichuk et al. 2013, 2015; Leverenz et al. 2015; Zlenko et al. 2016a)

Full size image

PBS core arrangement

In each APC trimer of basal cylinders, only one \(\alpha\)APC subunit located close to the thylakoid membrane (“bottom” \(\alpha\)APC subunit) can be substituted with one of the TE molecules (Zlenko et al. 2016b). It was clearly proved that \({\text{PBL}_\text{CM}}\) is located in one of the middle APC trimers of basal cylinder, while ApcD is located in one of the edge APC trimers (Lundell and Glazer 1983b; Anderson and Eiserling 1986). Thus, TE subunits can be located in the adjacent APC trimers, or can be separated by the \((\alpha \beta )_3\) APC timer (Fig. 1B). But, there is no clear evidence for one of these alternatives to be preferable (Gindt et al. 1994). The problem of mutual arrangement of TEs is even more complicated, as in two basal cylinders of the PBS core there are eight bottom \(\alpha\)APC subunits, and four of them are always substituted with TE molecules (Adir 2005; Watanabe and Ikeuchi 2013).

Let us consider the bottom view of PBS core with cylinders oriented along the vertical axis (Fig. 3). Taking into account the C\(_2\) symmetry of the PBS core and antiparallel arrangement of APC trimers in basal cylinders (Arteni et al. 2009; Chang et al. 2015), there exist two alternative core structures with the adjacent (Fig. 3, left) and two core structures with the remote (Fig. 3, right) TEs location. Due to the chirality of the APC trimer, core structures with adjacent TEs located on downward (Fig. 3A) and upward (Fig. 3B) diagonals are not equivalent to each other. The same is true for the core structures with remote TEs location (Fig. 3C, D). The difference appears due to the existence of two possible locations of bottom \(\alpha\)APC subunits with respect to the core middle plane (Fig. 3, dash-dotted line in general layout). Bottom \(\alpha\)APC subunits are located either close to the core middle plane (“internal” position, even numbers in Fig. 3) or close to the core edge (“external” position, odd numbers in Fig. 3).

Fig. 3
figure 3

General view of the PBS core bottom surface (Zlenko et al. 2016b) with different possible locations of TEs. Two basal cylinders (denoted as left and right) are shown in light-gray and are oriented vertically. Bottom \(\alpha\)APC subunits that can be substituted by \({\text{PBL}_\text{CM}}\) and ApcD are colored green and red, respectively. There are four possible mutual arrangements of TEs in the core shown in small images (A–D). \({\text{PBL}_\text{CM}}\) and ApcD can be located either in the adjacent (A and B) or in the remote (C and D) trimers. Position of each TE is determined as internal (even numbers) or external (odd numbers) with respect to the middle plane of the core (vertical dash-dotted line in general view), and by its location in the bottom core projection on downward (solid line in A, C, and D) or upward (dashed line in B, C, and D) diagonals

Full size image

Since APC trimers have their concave sides facing down in the first (Fig. 3, positions 1 and 8) and third (Fig. 3, positions 3 and 6) layers, the APC tips are always located on the \(\alpha\)APC subunit right side in these layers. Similarly, the concave side of APC trimers facing up in second (Fig. 3, positions 2 and 7) and fourth (Fig. 3, positions 4 and 5) layers, the APC tips are located on the left side of \(\alpha\)APC subunits. As a result, in the case of adjacent TEs arrangement, orientation of the tips in the TE subunits located on downward diagonal (Fig. 3A) is inverse as compared to TEs located on upward diagonal (Fig. 3B). In the case of remote TE location, all TEs have internal (Fig. 3C) or external (Fig. 3D) position, and thus have their tips oriented outwards or inwards the core, respectively.

If we consider PBS as a part of PBS/PSII supercomplex, then internal and external TE locations also differ by TE chromophore accessibility from the surrounding medium. \(\alpha 84\)-phycobilin chromophores in any phycobiliprotein trimer are localized near the protein surface, in the cavity at the bottom of the tip (Fig. 2A). Consequently, in the case of internal TE location with the tips oriented outwards the core (Fig. 3, even numbers), the chromophores appear to be close to the outside surface of the PBS core and PSII dimer supercomplex, while the chromophores of external TEs with the tips oriented inwards the core (Fig. 3, odd numbers) are located deep inside the supercomplex. As a result, the bulk solution agents can access the TE chromophores only in the case of the TE internal location.

One of the important agents interacting with \({\text{PBL}_\text{CM}}\) subunits of the PBS core (Stadnichuk et al. 2012, 2013, 2015; Zlenko et al. 2016a), is the orange carotenoid protein (OCP) and its proteolytically cleaved N-terminal domain known as the red carotenoid protein (RCP). Both, OCP (Rakhimberdieva et al. 2004; Kirilovsky et al. 2014) and RCP (Kerfeld 2004; Leverenz et al. 2014) are known as PBS quenchers. According to the molecular modeling results (Stadnichuk et al. 2015; Zlenko et al. 2016a), the OCP binding site becomes localized on the APC trimer tip (Fig. 2A). RCP binding site was supposed to be located in the cavity at the bottom of the APC tip (Leverenz et al. 2015) that correlates well with the position of the N-terminal domain of OCP molecule in the model of OCP/APC supercomplex (Stadnichuk et al. 2015; Zlenko et al. 2016a). Thus, both OCP and RCP binding sites on the \({\text{PBL}_\text{CM}}\) surface are easily accessible only in the case of internal location of \({\text{PBL}_\text{CM}}\) observed in the PBS core arrangements A and C (Fig. 3).

The only difference between A and C core arrangements is the ApcD subunit location: in the adjacent (Fig. 3A) or in the remote (Fig. 3C) trimers. In the lack of direct evidence, we can make some conclusions about the mutual arrangement of ApcD and \({\text{PBL}_\text{CM}}\) analyzing CD spectral data obtained for Synechococcus sp. PCC 6301 PBS (Lundell and Glazer 1983b). Positive CD signal of ApcE chromophore (\(\lambda =672\) nm) in the corresponding APC trimer appears to be completely insensitive to the ApcD-containing trimer binding to the “18S-particle,” that was observed during the “tertiary subcore particles” assembling (Lundell and Glazer 1983a, b). At the same time, binding of \((\alpha \beta )_3\) APC trimer to the “tertiary subcore particles” leads to 20% increase of CD signal of ApcE chromophore that was observed during the process of entire PBS assembling (Lundell and Glazer 1983b). This difference indicates that the \({\text{PBL}_\text{CM}}\)-containing trimer interacts most probably with the \((\alpha \beta )_3\) APC trimer rather than with ApcD-containing trimer. The effect is rather weak and the described difference cannot be directly attributed to the interaction of \({\text{PBL}_\text{CM}}\) with \((\alpha \beta )_3\) APC trimer since entire PBS assembling is accompanied by peripheral rods binding, which may affect CD spectra. The ApcD and \({\text{PBL}_\text{CM}}\) location in the remote trimers does not contradict to the results of protein cross-linking experiments, as the links between ApcE or ApcF and ApcD were not found (Liu et al. 2013; Tal et al. 2014; Harris et al. 2016).

Summarizing our observations, we can conclude that the core arrangement C is more probable than other alternatives (Fig. 3C). First, this structure includes ApcE chromophore accessible to interaction with OCP and RCP from the surrounding medium. Second, the TE subunits location in the remote trimers correlates well with the available CD spectral data (Lundell and Glazer 1983b).

Rows of PBS cores

It is generally accepted that in rows the PBS cores are stacked face-to-face with cylinders parallel to the row direction (Fig. 4, top view), which implies the densest packing of PBSs (Bald et al. 1996; Kuhl et al. 1999; Arteni et al. 2009). Indeed, any other packing would increase the core projection on the row axis and thus would increase the row pitch. Nevertheless, the molecular details of face-to-face contact between the PBSs are unclear. Here we will focus on the PBS cores packing in the row, since the thickness of the core seems to be greater than the thickness of the PBS peripheral part, according to the single-particle EM images (Arteni et al. 2009).

Fig. 4
figure 4

The model of coupled rows of the PBS cores and the PSII dimers for their symmetrical arrangement: “top,” “side,” and “front” projections. Each projection is drawn for three PBS cores and three PSII dimers. In side and top views, the middle PBS core is shown in blue, and in the top view, all the PBS cores are shown transparent. In the side view, the middle PSII dimer is shown in yellow, while in the top view, all PSII dimers are shown in yellow and outlined for clarity. Phycobilins and chlorophylls are colored pink and green, respectively, and not shown in top view for clarity. The distances from ApcE (green circle) and ApcD (red circle) phycobilin to the nearest chlorophyll (blue circle) are indicated for A and C core structures (Fig. 3). The ApcD (red circles) and ApcE (green circles) positions are indicated in top view for two selected basal cylinders. The angle between the APC trimers plane and the face plane of the PSII dimer is about \(35^\circ\) (top view). The intersection of these planes is a two-fold symmetry axis both of the PBS core and of the PSII dimer (vertical L\(_2\) axis in Fig. 1A). The insertion represents the position of \({\text{L}_\text{C}}\) linkers (red) in the interface between two adjacent cores according to Reuter et al. (1999). Only one of three possible mutual orientations of \({\text{L}_\text{C}}\) molecule is shown

Full size image

The structure of the developed earlier PBS core model (Zlenko et al. 2016b) provides a natural way of the PBS cores face-to-face stacking: for the adjacent PBS cores it should be the same as stacking of the adjacent APC hexamers inside the core (Fig. 4). Indeed, packing and orientation of APC trimers in the first and third layers of the model core are the same, as well as they are the same in the second and fourth layer of trimers (Fig. 3). Thus, protein surfaces turned to the middle of the core (between the hexamers) and to its external surface are the same, and the hexamer–hexamer interface in the middle of the core can be naturally considered as the interface between the different PBS cores.

In our model of the core (Zlenko et al. 2016b), the linker domains are absent, but they are present in the real PBS core and can influence the PBS cores interaction in the row. In early studies, it was proposed that the \({\text{L}_\text{C}}\) linker terminates the growth of basal cylinders in the same way as its homolog \({\text{L}_\text{R}}\) rod linker (Füglistaller et al. 1987) terminates the growth of the peripheral rods. However, the rows of PBSs and PSII dimers can be easily and quickly assembled and disassembled during state transitions (Olive et al. 1986; Kirilovsky 2015). This fact indicates that the adjacent cores in the row cannot be linked by the REP domains of \({\text{L}_\text{CM}}\) polypeptide and all REP domains of both ApcE polypeptides are localized inside the individual PBS cores. Thus, the mechanism of the growth restriction of the peripheral rods cannot occur in the PBS rows, since there are no linker polypeptides between the individual cores. On the other hand, \({\text{L}_\text{C}}\) linkers could mechanically restrict the adjacent cores interaction by pushing them away from each other. According to the X-Ray data, the \({\text{L}_\text{C}}\) subunit is located deeply inside the APC trimer cavity and does not protrude outside the trimer surface (Reuter et al. 1999). The shortest distance between the \({\text{L}_\text{C}}\) molecules inserted into the edge trimers of the adjacent cores in the proposed model is of about 1 nm (Fig. 4, insertion) that makes negligible their influence on the cores interaction.

The tight packing of PBSs should allow energy transfer along the row and its more effective utilization (Mörschel and Mühlethaler 1983; Kuhl et al. 1999) according to the calculated interchromophore distances. In the proposed model of stacked PBS cores, the rates of energy transfer between the chromophores inside the PBS core are approximately the same as between the cores. Indeed, the distances between the chromophores and their orientations are nearly the same in two adjacent hexamers inside the core cylinders and between two adjacent cores. Thus, the obtained model of PBS cores row can be considered as a giant “supercore,” providing extremely fast redistribution of the absorbed light energy among the PSII dimers. Such energy transfer between the pigment system II units can occur under different light conditions in cyanobacteria (Mimuro and Fujita 1978).

It is important to outline that the interaction between the PBSs and PSII dimers and state transition processes depend on the action of other proteins (Emlyn-Jones et al. 1999) that may also mediate rows assembling or disruption. Bald et al. (1996) have proposed that specific lipids surrounding of PSII dimers in the membrane could act as a kind of “glue” mediating and stabilizing the rows formation. The influence of such factors is not reflected in the developed molecular model that provides information only on the APC packing in the coupled rows of the PBS cores and PSII dimers.

Coupled rows of PBS cores and PSII dimers

In the built model of the PBS cores row (Zlenko et al. 2016b), the center-to-center distance between the cores is 12.4 nm. This value is only 2 Å greater than the reference center-to-center distance between the PSII dimers obtained by Folea et al. (2008). At the same time, the thickness of PSII dimers observed in the crystal lattice is only 11.6 nm (Hellmich et al. 2014) and 11.7 nm according to the single-particle EM observations of PSII double dimers (Kuhl et al. 1999). Previously, this discrepancy was fixed by the scaling of the EM image (Hellmich et al. 2014). However, we find it much more reliable to introduce additional space of 8 Å between PSII dimers in our model. Indeed, such separation can occur in the membrane due to the interaction of PSII dimers with surrounding lipids that were removed during the crystallization process. Thus, in our model, we consider a row of PBS cores described above, coupled with a row of PSII dimers moved away from each other at a distance of \(\sim8\) Å (Fig. 5) as compared to the crystal lattice (Hellmich et al. 2014). So, both rows have a pitch of 12.4 nm.

Fig. 5
figure 5

Top view of two coupled PSII dimers in the crystal lattice (A) and in the model (B) rows. Protein surfaces are shown in light-gray, chlorophyll molecules are shown in green. The center-to-center distances between PSII dimers (row pitch) and the minimal distances between the chromophores are indicated. Arrows mark the location of CP43 and CP47 proteins and reaction center (RC). Crystal structure (A) was taken from PDB ID 4PJ0 (Hellmich et al. 2014)

Full size image

As it was reported earlier by Bald et al. (1996) and Arteni et al. (2009), the planes of the APC trimers are not parallel to the face planes of the PSII dimers in the rows. The angle between these two planes depends on the definition of the PSII dimer face plane. We chose it as a plane dissecting the dimer in its broadest place, and got the value of \(35^\circ\) (Fig. 4, top view), while the alternative choice of the PSII dimer face plane gives the value of \(17^\circ\) (Arteni et al. 2009). However, this difference appears only due to the formal PSII dimer face plane definition and does not reflect any real difference between the PBS and PSII orientation. It is important to note that in the coupled rows the angle of mutual orientation of PBSs and PSII dimers is strictly defined by the PSII row structure (Hellmich et al. 2014) and face-to-face PBS cores packing (see above). Mutual orientation of the PBS and PSII dimer in a single complex lacks such restrictions, and in our previous work (Zlenko et al. 2016b), we have reported a slightly different angle. The angle between PBS and PSII in the single PBS/PSII supercomplex reported by Chang et al. (2015) differs by \(25^\circ\) from the angle shown in Fig. 4 and also mentioned by Bald et al. (1996), Kuhl et al. (1999), and Arteni et al. (2009). For such an angle projection of the PBS with thickness of 20 nm and width of 49 nm (Chang et al. 2015) equals 29 nm, which is incompatible with the reported row pitch for pentacylindrical PBSs of about 12 nm (Mörschel and Mühlethaler 1983).

Mutual arrangement of the row of PBS cores and the row of PSII dimers is determined by two parameters: the distance between the rows and their relative shift in the row direction. Indeed, the angle between the rows of PBSs and PSII dimers in the membrane plane is restricted, PBS core and PSII rows must be parallel to each other and the plane of the basal cylinders of the PBS core must be parallel to the membrane plane. In order to choose the correct shift value, we moved the model row of PBS cores along the row of PSII dimers and calculated the minimal distance between TE phycobilins and the nearest chlorophyll of PSII dimer. The distances were determined for all possible eight positions of TEs in the PBS core (Fig. 3, general layout), the resulting plot is given in Fig. 6.

The TE chromophores position inside TE subunits is very close to the position of phycobilins in the bulk APC subunits (McGregor et al. 2008; Peng et al. 2014; Tang et al. 2015; Zlenko et al. 2016a). Thus, due to the strong periodical structure of the rows, the distance curve is the same for any TE in the internal (or external) position (Fig. 3). Since one PSII dimer interacts with one PBS core composed of two layers of APC hexamers, the pitch of possible internal (or external) TE positions is twice smaller than the PSII dimers row pitch. Thus, the distance curves for TEs located on upward and downward diagonals (Fig. 3) are half-period-shifted against each other (Fig. 6).

Fig. 6
figure 6

The distance from TE chromophores to the nearest chlorophyll of the PSII dimer (distance curves) in coupled rows of the PBS cores and the PSII dimers. Data shown separately for TEs of the left and right basal cylinders (Fig. 3), which are symmetrical due to the C\(_2\) symmetry of the PBS core and the PSII dimer. Distance curves for TEs in internal (red) and external (blue) positions in each cylinder are plotted on two axes for TEs located on upward and downward diagonals (Fig. 3, A–D). The zero points of the relative shift axes correspond to the symmetrical arrangement of the PBS cores and the PSII dimers shown in Fig. 4. The positive direction of axes corresponds to the shift of the PBS cores to the right (side view in Fig. 4)

Full size image

As mentioned previously, we denote by letters A, B, C, and D the corresponding arrangements of TEs in the PBS core given in Fig. 3. The minimal distance curves for internal \({\text{PBL}_\text{CM}}\) in structures A and C coincide and correspond to the red curve on the “downward diagonal” axes in Fig. 6. Similarly, the distance curves for external \({\text{PBL}_\text{CM}}\) in structures B and D are the same (blue curve on the “upward diagonal” axes in Fig. 6). The distance curves for internal ApcD in structures B and C are the same as for internal \({\text{PBL}_\text{CM}}\) in structures A and C, but shifted by half a period (Fig. 6, red on “upward diagonal” axes). Similarly, the distance curve for external ApcD in structures A and D is the same as for external \({\text{PBL}_\text{CM}}\) in structures B and D, but also shifted by half a period (Fig. 6, blue on “downward diagonal” axes). Then, due to C\(_2\) symmetry of rows both of PBS cores and of PSII dimers, the distance curves for symmetrical TEs located in different basal cylinders of the PBS core (for instance, positions 1 and 5 or 2 and 6 in Fig. 3, general layout) can be obtained from one another by the reflection with respect to the zero point (Fig. 6, left and right cylinder plots).

Let us consider the distance curves for TEs located on downward and upward diagonals (Fig. 3). Interchromophore distance curves for both internal and external TEs have a significant decrease around the zero point for TEs located on the downward diagonal (Fig. 6, “downward diagonal” axes). At the same time, the distance curves for TEs located on the upward diagonal reach their maximum values around the zero point (Fig. 6, “upward diagonal” axes). The zero points in Fig. 6 correspond to the symmetrical arrangement of PBS and PSII rows (Fig. 4, top view), confirmed by EM images (Mörschel and Mühlethaler 1983). Thus, the shortest distance between TE phycobilins and PSII chlorophylls is 45–48 Å for the PBS core with TEs located on upward diagonal, and 36–38 Å for the core with TEs located on downward diagonal. This difference can be explained by the location of CP43 and CP47 proteins inside the PSII dimer. In PSII row, CP43 antenna is shifted laterally (Fig. 5) and appears to be close to TEs located on the downward diagonal. At the same time, CP47 protein is shifted medially (Fig. 5) and appears to be close to TEs located on the upward diagonal, but in this case the distance between TE phycobilin and the nearest PSII chlorophyll is larger. As a result, according to Förster mechanism (Förster 1948), the rate of energy transfer to PSII from downward diagonal TEs is approximately 5 times higher than from upward diagonal TEs that makes the location of TEs on the downward diagonal much more probable.

The core configuration with both TEs located on the downward diagonal seems to be preferable as it provides the shorter interchromophore distance. However, the roles of ApcE and ApcD chromophores in energy transfer to PSII are not equal that should result from their arrangement, since emission spectra of ApcE and ApcD are very similar (Mimuro et al. 1986). It is well known that knockout of apcD gene does not significantly affect the energy transfer from PBS to PSII (Gindt et al. 1992, 1994; Ashby and Mullineaux 1999; Kuzminov et al. 2014). There is only a small decrease of the energy transfer rate in ApcD-deficient mutant as compared to wild-type cells. Thus, the distance from ApcD chromophore to the nearest chlorophyll of PSII should be greater than the distance from ApcE chromophore.

If both TEs are located on the downward diagonal (Fig. 3A, and Fig. 7, left), the rate of energy transfer from ApcD to PSII would be approximately two times higher than that from \({\text{PBL}_\text{CM}}\) to PSII. In this case, according to Förster mechanism of energy transfer, the deletion of apcD gene would lead to the dramatic decrease of the energy transfer rate by \(\sim\)70%. On the other hand, if we suppose that ApcD subunit is located on the upward diagonal, while \({\text{PBL}_\text{CM}}\) is located on the downward diagonal (Fig. 3C, and Fig. 7, left), the distance from ApcD chromophore to the nearest CP47 chlorophyll would be about 46 Å (Fig. 6, red curve on “upward diagonal” axes). In this case, the decrease of the energy transfer rate in ApcD-deficient mutant would be about 20%. Exactly the same value was found experimentally for Synechocystis sp. PCC 6803 lacking ApcD (Ashby and Mullineaux 1999). Thus, the arrangement of TEs shown in Fig. 1B and in Fig. 3C seems to be preferable both from the structural point of view, and according to the analysis of CD data (Lundell and Glazer 1983b) discussed above.

If we accept the PBS core to have the structure given in Fig. 3C, then the main part of energy gathered by PBS would be funneled to PSII through ApcE. In this case, the optimal distance of \(\sim\) 39 Å between phycobilin of ApcE and PSII CP43 chlorophyll would be observed if PBS cores row was shifted \(\sim\) 15 Å to the right (Fig. 6, red curve on “downward diagonal” axes of left cylinder). However, this is not the case, because one should take into account positions of both \({\text{PBL}_\text{CM}}\) subunits of the core. Since both PBS core and PSII dimer have C\(_2\) symmetry (Fig. 4, top view), the distance curve for \({\text{PBL}_\text{CM}}\) located in position 2 (Fig. 6, red curve on “downward diagonal” axes of left cylinder) would be a reflection of the distance curve for \({\text{PBL}_\text{CM}}\) located in position 6 (Fig. 6, red curve on “downward diagonal” axes of right cylinder). As a result, the distance to the nearest chlorophyll from the first and second \({\text{PBL}_\text{CM}}\) phycobilins would be \(\sim\) 39 and \(\sim\) 47 Å, respectively. The relative gain in the energy transfer rate from the first \({\text{PBL}_\text{CM}}\) chromophore (\(\sim\) 1.6 times) is smaller than the loss in the rate from the second \({\text{PBL}_\text{CM}}\) (\(\sim\) 2.0 times), according to Förster mechanism of energy transfer (Förster 1948). Thus, the symmetrical mutual arrangement of PBS and PSII in the rows seems to be more probable, than asymmetric (Fig. 4). The symmetrical arrangement of PBS cores and PSII dimers in the rows was however postulated earlier considering the two-fold symmetry of both complexes (Bald et al. 1996; Kuhl et al. 1999).

The mutual arrangement of PBS core and PSII dimer was recently investigated by protein cross-linking technique (Liu et al. 2013). There were found five cross-links between ApcE and the PSII dimer: \(^{87}\)K:ApcE-\(^{227}\)K:PsbB (CP47), \(^{317}\)K:ApcE-\(^{23}\)K:PcbD (D2), \(^{523}\)K:ApcE-\(^{35}\)K:PsbI, \(^{523}\)K:ApcE-\(^{457}\)PcbC (CP43), and \(^{584}\)K:ApcE-\(^{457}\)K:PsbC (Liu et al. 2013). Three of the mentioned lysines (\(^{317}\)K, \(^{523}\)K, and \(^{584}\)K) are located beyond the \({\text{PBL}_\text{CM}}\) domain and thus are not presented in our model. Only one of them is located inside the \({\text{PBL}_\text{CM}}\) domain (\(^{87}\)K:ApcE), presented in the coiled PB-loop (Capuano et al. 1991) that is also absent in our model. Nevertheless, the cross-linking between \(^{87}\)K:ApcE and \(^{227}\)K:PsbB is highly possible in our model according to the estimated length of the PB-loop. Thus, the model of Liu et al. (2013) correlates very well with our results. The only difference concerns the choice of structure B (Fig. 3B) instead of C (Fig. 3C) by Liu et al. (2013) resulting from smaller distance between the PB-domain and CP47 for core structures with the \({\text{PBL}_\text{CM}}\) located on the upward diagonal (Fig. 3). This contradiction seems to be negligible, since the cross-linking is possible in the case of both B and C core structures. The results of Liu et al. (2013) demonstrate that ApcD subunit interacts only with PSI, and does not interact with PSII. Thus, the ApcD subunit should be located aside the PSII dimer (Bald et al. 1996) that corresponds to our model (Fig. 4, red circles in top view).

Biological conclusions

The developed model implies the following organization of coupled rows of PBS cores and PSII dimers. Each PBS core is located on the top of PSII dimer so that their C\(_2\) symmetry axes coincide (Fig. 4, top view, intersection of the planes). ApcD and \({\text{PBL}_\text{CM}}\) subunits are located in the remote APC trimers that is in agreement with their functioning as the independent TEs (Lundell and Glazer 1983b; Mimuro et al. 1986; Ashby and Mullineaux 1999). PBS cores are stacked face-to-face and the interface between them is the same as between the hexamers inside the core. Such “supercore” structure serves as a “pipeline” (Kruip et al. 1997) allowing energy transfer along the PBS row and its redistribution among the PSII dimers, as it was proposed earlier (Mörschel and Schatz 1987; Kuhl et al. 1999). In the State 1 adapted to the PSI-specific light, the PSII reaction centers require some additional energy influx to compensate for the increased PSI activity (Kirilovsky 2015). The proposed model provides a possible mechanism of enhancement of energy transfer from PBSs to PSII reaction centers observed in State 1 (Olive et al. 1986; Mullineaux and Allen 1990; Delphin et al. 1995; Ashby and Mullineaux 1999) and can also explain the increase of PSII absorption cross-section during the State 2 to State 1 transition (Mullineaux and Allen 1988).

The model of coupled rows gives some insights into the location of chlorophyll molecules that gather the energy from the PBS. The minimal distance between phycobilin and chlorophyll chromophores is observed for ApcE and CP43. The distance curve for TEs in external position (odd numbers in Fig. 3) has two local minima aside from the zero point (blue curves in Fig. 6), while the distance curve for internal TEs (even numbers in Fig. 3) has a minimum in a \(+15\) Å point. These minima correspond to the TE chromophores approaching one of the two special chlorophyll molecules of CP43 (Fig. 4, side view and Fig. 7, left). These chlorophylls have a specific location inside the PSII: their centers of mass come closer to the cytoplasmic surface of the thylakoid membrane by approximately 5 Å (Fig. 7, left) and at the same time they occupy the most lateral position in the row (Fig. 4, blue circles in the front view). We observed the same chlorophyll molecules shifted to the cytoplasm in all the cyanobacterial PSII crystal structures available from PDB database. Thus, it is very likely that these two chlorophyll molecules serve as gates for energy funneling from PBS TEs to the PSII.

Fig. 7
figure 7

Side view of the PBS core basal cylinders attached to the PSII dimer in coupled rows (left) and the same complex with attached RCP (right). Protein surfaces are colored light-gray (PBS core), yellow (PSII dimer), and red (RCP). Chromophores are shown with blue (phycobilins), green (chlorophylls), and orange (carotenoid) sticks. Chromophores only for the front basal cylinder of the PBS core are shown for clarity. CP43, CP47, and RC chromophores are marked with arrows. ApcE and ApcD chromophores are indicated with green and red circles, respectively. Two possible ApcD positions for the core structures A (right circle) and C (left circle) are indicated in accordance with Fig. 3. Blue circle indicates the suggested RCP binding site. The RCP structure was taken from 4XB4 (Leverenz et al. 2015)

Full size image

The next important feature of our model concerns the interaction of the PBS with fluorescence quenching agents (OCP and RCP). The very recent findings of OCP photoactivation under high light conditions show that C-terminal domain releases carotenoid molecule and dissociates from N-terminal domain, which in turn fastens to the PBS core and quenches its fluorescence (Liu et al. 2016; Maksimov et al. 2016; Zhang et al. 2016). The proteolitically cleaved N-terminal domain of OCP (RCP) can quench the PBS fluorescence itself (Kerfeld 2004; Leverenz et al. 2015). To examine the possibility of RCP interaction with the coupled rows of PBS cores and PSII dimers, we used a simple rigid-body docking procedure (Macindoe et al. 2010).

In our model there is a cavity between the PSII dimer and PBS core (Fig. 7, blue circle), whose shape and size coincides very well with the shape and size of RCP molecule (Fig. 7). The protrusion on the RCP surface formed by 91–96 and 163–165 amino acid residues interacts with a cleft between the APC trimers forming a hexamer, while the main part of the protein interacts with a wide groove between two APC hexamers of basal cylinder (Fig. 7). The observed RCP position in the resulting PBS/PSII/RCP tertiary complex in respect to APC trimer is very close to positions proposed earlier for RCP (Leverenz et al. 2015) and N-terminal domain of OCP (Zlenko et al. 2016a). The proposed location of the interaction site does not contradict the results of cross-linking experiments showing interaction of OCP with ApcB and \({\text{PBL}_\text{CM}}\) subunits of the PBS core (Zhang et al. 2014; Harris et al. 2016). At the same time, the exact OCP binding site location is under discussion (Zhang et al. 2014; Leverenz et al. 2014, 2015; Stadnichuk et al. 2015; Harris et al. 2016; Zlenko et al. 2016a; Kirilovsky and Kerfeld 2016). In the observed supercomplex, the ApcE chromophore plays a role of the phycobilin closest to 3’-hydroxyechinenone of RCP (Stadnichuk et al. 2012). The determined distance between RCP and ApcE chromophores of \(\sim\) 24 Å (Fig. 7, right) is almost the same as was reported earlier (Stadnichuk et al. 2015; Zlenko et al. 2016a) that confirms the reliability of the proposed model of the interaction of the OCP and PBS.

Methods

The arrangement of PBS core rows was reconstructed using the recently developed model of the PBS core (Zlenko et al. 2016b). PSII rows were modeled using the X-ray structure of PSII and PSII native-like rows from Th. elongatus (4PJ0, Hellmich et al. 2014). Analysis of interaction of RCP molecule with the model of coupled rows of PBS cores and PSII dimers was made using the RCP structure from Synechocystis sp. PCC 6803 (4XB4, Leverenz et al. 2015).

Analysis of possible mutual orientations of the PBS cores row and the PSII dimers row was made using in-house python utilities. The rows of PBS cores and PSII dimers were oriented parallel to each other, with the plane of basal cylinders parallel to the membrane plane. Then, the PBS cores row was moved along the PSII dimers row with a step of 0.5 Å. At every step the PBS cores row was moved apart from the PSII row (in the normal direction to the membrane plane) and then was brought back until the protein surfaces contact. The distance between PBS and PSII chromophores was calculated at every step. We used the distance between the centers of mass of chromophores as an assessment of the distance between the chromophore molecules.

All manipulations with atomic coordinates and data visualization were made using PyMOL Molecular Graphics System (version 1.7 Schrödinger, LLC), using standard and supplementary in-house utilities. The possible RCP binding site on the side surface of the coupled rows of the PBS cores and the PSII dimers was determined using rigid-body protein docking procedure, controlling both proteins shape and electrostatic properties (Macindoe et al. 2010). In eight of ten top docking results, the RCP molecule is located inside the cavity between the PBS core and the PSII dimer (Fig. 7, blue circle) oriented as shown in Fig. 7b. The suggested RCP/OCP binding site is in accordance with the previous model of the single OCP/PBS complex (Stadnichuk et al. 2013; Leverenz et al. 2015; Zlenko et al. 2016a).