PSI of the Colonial Alga Botryococcus braunii Has an Unusually Large Antenna Size 1[OPEN]

PSI is an essential component of the photosynthetic apparatus of oxygenic photosynthesis. While most of its subunits are conserved, recent data have shown that the arrangement of the light-harvesting complexes I (LHCIs) differs substantially in different organisms. Here we studied the PSI-LHCI supercomplex of Botryococccus braunii, a colonial green alga with potential for lipid and sugar production, using functional analysis and single-particle electron microscopy of the isolated PSI-LHCI supercomplexes complemented by time-resolved ﬂ uorescence spectroscopy in vivo. We established that the largest puri ﬁ ed PSI-LHCI supercomplex contains 10 LHCIs ( ; 240 chlorophylls). However, electron microscopy showed heterogeneity in the particles and a total of 13 unique binding sites for the LHCIs around the PSI core. Time-resolved ﬂ uorescence spectroscopy indicated that the PSI antenna size in vivo is even larger than that of the puri ﬁ ed complex. Based on the comparison of the known PSI structures, we propose that PSI in B. braunii can bind LHCIs at all known positions surrounding the core. This organization maximizes the antenna size while maintaining fast excitation energy transfer, and thus high trapping ef ﬁ ciency, within the complex.

range of PSI beyond that of PSII and contribute significantly to light harvesting in a dense canopy or algae mat, which is enriched in far-red light (Rivadossi et al., 1999). The red forms slow down the energy migration to the RC by introducing uphill transfer steps, but they have little effect on the PSI quantum efficiency, which remains ;1 (Gobets et al., 2001;Jennings et al., 2003;Engelmann et al., 2006;Wientjes et al., 2011). In addition to their role in light-harvesting, the red forms were suggested to be important for photoprotection (Carbonera et al., 2005).
PSI-specific antenna proteins vary in type and number between algae, mosses, and plants. For example, the genomes of several green algae contain a larger number of lhca genes than those of vascular plants (Neilson and Durnford, 2010). The PSI-LHCI complex of plants includes only four Lhcas (Lhca1-Lhc4), which are present in all conditions analyzed so far (Ballottari et al., 2007;Wientjes et al., 2009;Mazor et al., 2017), while in algae and mosses, 8 to 10 Lhcas bind to the PSI core (Drop et al., 2011;Iwai et al., 2018;Pinnola et al., 2018;Kubota-Kawai et al., 2019;Suga et al., 2019). Moreover, some PSI-specific antennae are either only expressed, or differently expressed, under certain environmental conditions (Moseley et al., 2002;Varsano et al., 2006;Swingley et al., 2010;Iwai and Yokono, 2017), contributing to the variability of the PSI antenna size in algae and mosses.
The colonial green alga Botryococcus braunii (Trebouxiophyceae) is found worldwide throughout different climate zones and has been targeted for the production of hydrocarbons and sugars (Metzger and Largeau, 2005;Eroglu et al., 2011;Tasić et al., 2016). braunii PSI-LHCI supercomplexes. A, Suc density gradient of solubilized thylakoid membranes. The bands were assigned based on their protein content. B, SDS-PAGE of the PSI-LHCI supercomplexes purified by Suc gradient. The bands labeled in red have a M r not typical for PSI proteins, but they are unidentified due to the absence of the annotated genome. C, Fluorescence emission spectra (440 nm excitation) at 77 K were normalized to the maximum. D, Absorption spectra of the PSI-LHCI supercomplexes at 293 K and 77 K normalized to their integral. E, Second derivative in the Chl Q-region of the absorption spectrum at 77 K. Results were reproduced at least three times on different biological replicas. MWM, Molecular weight marker.
Here, we have purified and characterized PSI from an industrially relevant strain isolated from a mountain lake in Portugal (Gouveia et al., 2017). This B. braunii strain forms colonies, and since the light intensity inside the colony is low, it is expected that PSI in this strain has a large antenna size (van den Berg et al., 2019). We provide evidence that B. braunii PSI differs from that of closely related organisms through the particular organization of its antenna. The structural and functional characterization of B. braunii PSI highlights a large flexibility of PSI and its antennae throughout the green lineage. Table S1) were identified in the B. braunii genome by similarity searches using query sequences from Arabidopsis (Arabidopsis thaliana) and Chlamydomonas reinhardtii. The analysis showed that all the PSI core subunits present in these two organisms have clear homologs in B. braunii except for PsaN. On the other hand, B. braunii contains a gene homologous to PsaM, a subunit present in cyanobacteria and red algae but absent in Arabidopsis and C. reinhardtii.

PSI core subunits (Supplemental
To purify the PSI-LHCI supercomplex, the thylakoid membranes of B. braunii were mildly solubilized with n-dodecyl b-D-maltoside (b-DDM) and loaded on a Suc density gradient. It should be noted that the thylakoid membrane of B. braunii is more difficult to solubilize than that of other organisms and that milder detergent conditions do not permit the isolation of PSI-LHCI (see "Materials and Methods"; van den Berg et al., 2018). The band pattern upon centrifugation is shown in Figure 1A. The lower Suc gradient band (SGB) contained PSI-LHCI, as indicated by the presence of the characteristic PsaA-PsaB bands at high M r and of bands in the region corresponding to the M r of LHCIs (Fig. 1B). Individual LHCI proteins could not be identified by immunoblot with available LHCI antibodies (van den Berg et al., 2018) or by mass spectroscopy due to the lack of an annotated nuclear genome. The presence of PSI-LHCI in the SGB was confirmed by the absorption and fluorescence emission spectra. The emission spectrum at 77 K (Fig. 1C) had the main peak at 723 nm, which is typical for PSI, while the second peak at 678 nm suggests the presence of contamination with other photosynthetic proteins. The 77 K emission spectrum of the colonies (Fig. 1C) also showed a maximum at 723 nm, indicating that the red-most forms are  preserved in the purified complex. The second derivative of the 77 K absorption spectrum (Fig. 1, D and E) has the red-most minimum at 697.5 nm, which is likely the form responsible for the emission at 723 nm (Fig. 1C). The Stokes shift of 25.5 nm is similar to that of the red forms in Arabidopsis (Romero et al., 2009).
To determine the efficiency of excitation energy transfer and trapping in the PSI-LHCI supercomplex, time and spectrally resolved fluorescence measurements were performed using a streak camera setup. Fluorescence was collected in the 640 to 800 nm range upon preferential excitation of the LHCs (475 nm) or the PSI core (400 nm). Examples of streak camera images are presented in Figure 2, A and B. Four decay components were necessary to obtain a satisfactory fit of the data. The resulting decay-associated spectra (DAS) are shown in Figure 2C. The first component of 4 ps (excitation at 400 nm)/6 ps (excitation at 475 nm) mainly represents excitation energy transfer between Chls b and high-energy Chl a to low-energy Chls a. The second component of 20 ps is a combination of fast trapping (dominating the decay upon 400 nm excitation) and excitation energy transfer from high-to low-energy Chls a (dominating decay upon 475 nm excitation). The third component of 60 ps represents the main trapping time. The slowest component has a long lifetime (3,200 ps), a smaller amplitude (8% to 10%), and a maximum at 680 nm, which indicate that it is due to PSII/LHCII contamination (Fig. 2,C and D). Overall, the average fluorescence decay time of isolated PSI-LHCI supercomplexes from B. braunii is 48 ps upon 400 nm excitation, and 58 ps upon 475 nm excitation (Fig. 2D), which is 6 ps longer than for C. reinhardtii PSI-LHCI (Le Quiniou et al., 2015a).
To determine the M r of the isolated PSI-LHCI supercomplex, PSI SGB was loaded on a clear native (CN) gel. Four bands (CNB1-CNB4) were separated, one with M r similar to that of Arabidopsis PSI-LHCI ( Fig. 3A) and three with higher M r . SDS-PAGE confirmed that all four bands contained PSI-LHCI complexes ( Fig. 3B). All CN bands exhibit fluorescence maxima at 723 nm ( Fig. 3C) except for the lowest band (CNB4), the peak of which was blue shifted 3 nm. The absorption spectra of CNB1 to CNB4 (Fig. 3D) differ in the Chl b regions (450-500 nm and 640-660 nm), which show a relative decrease in amplitude going from the largest to the smallest complexes, suggesting differences in the number of LHCIs associated with the PSI core. In agreement with the spectra, pigment analyses show an increase of the Chl a/b ratio in the smaller PSI-LHCI complexes compared to the larger ones, again indicating differences in the LHCI content (Table 1). The Chl to carotenoid (car) ratio is the same in all complexes, but the CNBs with lower M r contain less neoxanthin and loroxanthin and more carotenes relative to Chl than the higher-M r bands (Table 1), again in agreement with a decrease in antenna size. Interestingly, the relative amount of lutein increases in the Table 1

. Pigment composition of the PSI-LHCI supercomplexes
Values are represented as the mean 6 SD of three biological replicas and are normalized to 100 Chls (a and b). CNB1 to CNB4 are numbered from highest to lowest in terms of M r . Loro, Loroxanthin; Neo, neoxanthin; Vio, violaxanthin; Lut, lutein; b-car, b-carotene, a-car, a-carotene.
smaller complexes, suggesting that the LHCI tightly associated with the core contains relatively more lutein than the more loosely bound complexes.
To investigate whether the difference in apparent M r between B. braunii and Arabidopsis PSI supercomplexes ( Fig. 3A) results in larger functional antenna size, we measured the P 700 oxidation kinetics in the largest complex purified (CNB1). Arabidopsis PSI-LHCI complexes were chosen as a control because of their high stability and well-defined antenna size. The kinetics were faster in the CNB1 complex than in PSI-LHCI and the PSI core of plants, which bind 156 and 100 Chls, respectively (Mazor et al., 2017), indicating that the functional antenna size of CNB1 corresponds to ;240 Chls ( Fig. 4). This result suggests that 10 LHCIs are associated with the complex (assuming 100 Chls in the PSI core and 14 Chls per LHCI).
To determine the structural organization of the largest purified PSI-LHCI supercomplex, the CNB1 complex was analyzed by single-particle electron microscopy of a negatively stained specimen. CNB1 was preferred over the SGB because of the lower contamination, and especially the absence of Suc. Image analysis of the whole data set resulted in distinct classes of particles. The six major averaged projection maps are shown in Supplemental Figure S2. The structural assignment of each class was obtained by fitting the projection maps with the plant PSI-LHCI structure (Fig. 5;Mazor et al., 2017). The number of bound LHCIs in each particle varies from 8 to 10. The fit indicates that the inner belt of the LHCI proteins contains four complexes, as in the plant PSI-LHCI (Fig. 5, LHCIs in orange), and two additional LHCIs at the PsaG-PsaH side of the core (Fig. 5, yellow). All other LHCIs bind at different positions and form the outer belt. One LHCI, which is present in all projections, is situated on the PsaF-PsaK side (Fig. 5, pink). Others (Fig. 5, cyan) bind to PSI either at the PsaK side (Fig. 5, A, C, and E) or at the PsaG side (Fig. 5, A and B). The final two LHCIs bind on the PsaG-PsaH side (Fig. 5D). Less abundant class averages are shown in Supplemental Figure S3.
Finally, to compare the properties of the purified PSI with that of PSI in vivo, we measured the trapping kinetics of PSI in vivo by time-resolved fluorescence spectroscopy directly on the colonies. The measurements were performed in oxic conditions with the PSII reaction centers closed with 3-(3,4-dichlorophenyl)-1,1dimethylurea (DCMU) and hydroxylamine (HA; state I), Figure 5. Structural models of PSI-LHCI supercomplexes isolated from B. braunii that feature unique antenna positions. A to E, Surface models of the resolved PSI-LHCI supercomplexes from B. braunii. The x-ray structure of the plant PSI-LHCI supercomplex (Protein Data Bank 5L8R; Mazor et al., 2017) and the structure of the Lhca1 protein from the same supercomplex were used to fit the EM maps. The PSI core complex is in green. LHCI color codes are as follows: orange, LHCIs that occupy 4 similar positions as in plant PSI-LHCI; yellow, LHCIs binding between PsaG and PsaH; pink, LHCIs binding near Lhca2-3; red, second row of LHCIs binding between PsaG and PsaH; cyan, additional LHCIs. Scale bar 5 100 Å .
to be able to disentangle the contributions of the two photosystems. The colonies were excited at 400 nm. The data were satisfactorily fitted with four components. The DAS are shown in Figure 6. Based on the spectra, the two main decay components of 119 ps and 1.6 ns can be attributed to PSI and PSII, respectively (Fig. 6). Note that the shape (but not the lifetime) of the PSI DAS in vivo is affected by reabsorption in the colony, similar to what is happening in plants (e.g. Chukhutsina et al., 2019), and therefore cannot be compared directly to that of the isolated complex. The PSI lifetime in the cells is considerably longer than the lifetime of the purified complex (119 versus 48 ps), suggesting that the antenna of B. braunii PSI in vivo is larger than that of the isolated complex. An energy transfer component from blue to red forms in PSI is also visible, and its lifetime is longer than in the isolated complex (42 ps versus 20 ps). This indicates that the additional antenna associated with PSI in vivo is less well connected with the rest of the complex. This loose connection has only a small effect on the trapping efficiency (97% versus 94% when considering a lifetime of 2 ns for the nonconnected antenna), but it does not allow us to use the lifetime to determine the exact size of the antenna in vivo, since in this case the lifetime does not scale with the number of pigments.
To estimate the antenna size of B. braunii, we then compared the P700 oxidation in the thylakoids of B. braunii and Arabidopsis. It was recently shown that in Arabidopsis, 1.4 LHCII trimers are associated with PSI-LHCI in vivo in dark-adapted plants (Chukhutsina et al., 2020). The data (Supplemental Fig. S4) show that also in the thylakoids, the antenna size of PSI in B. braunii is larger than in Arabidopsis, suggesting that the antenna of B. braunii is composed of ;17 LHC subunits (considering each of them as binding 14 Chls).

PSI-LHCI of B. braunii Shows a Unique Antenna Organization
In this work we have characterized the PSI-LHCI supercomplex of the colonial green alga B. braunii. The largest particle observed contains 10 LHCI proteins associated with the core, as is the case for the PSI-LHCI supercomplex of the green alga C. reinhardtii (Ozawa et al., 2018;Kubota-Kawai et al., 2019;Su et al., 2019), but the organization of the LHCIs differs. Moreover, in B. braunii, in addition to the complex containing 10 LHCIs, several other PSI-LHCI particles with a different number and organization of LHCIs were observed by electron microscopy. In total, the LHCIs were found to occupy 13 positions (Fig. 5). The four LHCIs in the inner belt are present in the PSI-LHCI of all plants and algae studied so far. Three more are in the outer belt observed in C. reinhardtii PSI-LHCI (Fig. 7A), corresponding to C. reinhardtii Lhca2, Lhca9, and Lhca5 (Suga et al., 2019). The additional LHCIs positioned on the PsaK side (Fig. 7B) and two of those on the PsaG side (Fig. 7C) largely overlap with the LHCIs observed in the structures of P. patens PSI-LHCI (Iwai et al., 2018;Pinnola et al., 2018). The final two LHCI positions in the outer belt on the PsaA side (Fig. 7D, red) are unique to B. braunii.
Although we cannot exclude that some of the small PSI-LHCI supercomplexes occur in vivo (Fig. 5), the different particles observed by electron microscopy (EM) are likely the result of a partial disassembly of the PSI-LHCI during purification. In addition, the timeresolved data show that the lifetime of PSI in the cells is far longer than that of the purified complexes (120 versus 48 ps), which indicates that the PSI antenna size is larger in vivo. Note that this is not the result of acclimation to low light as is the case for P. patens (Iwai and Yokono, 2017;Iwai et al., 2018;Pinnola et al., 2018), because in B. braunii the PSI antenna size remains identical under both low-and high-light growing conditions (van den Berg et al., 2019). Because PSI antenna size does not acclimate to the light intensity, it is also unlikely that PSI antenna size heterogeneity exists in the colony. It is also improbable that this difference in lifetime is due to state transitions, since the colonies were measured in the presence of oxygen with the PSII RCs closed with DCMU and HA, resulting in an oxidized plastoquinone pool, a condition that induces state I in C. reinhardtii (Nawrocki et al., 2016). The presence of red forms, which are known to slow down the excitation energy transfer (Jennings et al., 2003;Wientjes et al., 2011;Le Quiniou et al., 2015b), can also not be at the basis of the observed difference between the lifetimes in vivo and in vitro since the red Chl properties are similar in the purified complex and in the cells. The longer lifetime thus indicates that additional complexes are associated with PSI in vivo, increasing its antenna size. This additional antenna is transferring energy relatively slowly to the rest of the complex, indicating that it is less well connected with it than the antenna present in the purified complex. The loose functional connection is probably associated with a weaker structural interaction, which may explain why this part of the antenna is lost during purification. The fact that the purified PSI particle is smaller than the complex in vivo is not surprising, since purification can easily lead to loss of part of the antenna complexes. For example, the largest plant PSI complex purified so far only binds one LHCII trimer (Pan et al., 2018), whereas it is known that PSI binds more LHCIIs in the membranes (Benson et al., 2015;Bos et al., 2017;Chukhutsina et al., 2020). The same holds true for PSII, where the largest purified complex (C2S2M2; Su et al., 2017;Shen et al., 2019) only contains two LHCII trimers per core complex, while in vivo this number is larger and goes up to five (Anderson et al., 1995;Wientjes et al., 2013b).
It is tempting to speculate that all the LHCI positions observed in the different particles are occupied in vivo and that in B. braunii the PSI core is surrounded by two belts of LHCIs. This hypothetical model is presented in Figure 7D. Out of the 15 LHCIs shown in this model, 13 were observed in B. braunii, whereas the two in the outer belt on the PsaF side were observed in other organisms (Qin et al., 2019;Su et al., 2019) and might have been lost during purification in our study. In this model, the PSI core is surrounded by LHCIs, except at the PsaH/A side, which in plants and C. reinhardtii is the docking site for a LHCII trimer in state II (Drop et al., 2014;Pan et al., 2018).
The resulting PSI-LHCI complex is thus expected to have a larger antenna size than those of other green algae. In contrast to unicellular algae, colonial algae have to deal with constant internal shading by other cells in the colony (Beardall et al., 2009). The optical density of the colonies can be very high, and a large antenna system seems thus essential for the cells in the interior of the colony (van den Berg et al., 2019). In contrast to B. braunii hydrocarbon-producing strains, which often float near the water surface and experience high light intensities (Wake and Hillen, 1980), the extracellular polysaccharide (EPS)-producing strain used in this work does not float (Gouveia et al., 2017), and it grows faster under lower light regimes (García-Cubero et al., 2018), in an environment where a large antenna is beneficial. In this respect, it is important to note that our results show that despite the very large antenna, energy trapping in PSI-LHCI remains highly efficient.

Evolution of the PSI Supercomplex from Green Algae to Plants
Recently, several structures of PSI-LHCI complexes from various eukaryotic organisms have been resolved. The basic PSI unit, which corresponds to the PSI of vascular plants, is composed of the core and the four LHCIs located on the PsaG-PsaK side, and is conserved in all the green line organisms analyzed so far (Alboresi et al., 2017;Iwai et al., 2018;Pi et al., 2018;Pinnola et al., 2018;Kubota-Kawai et al., 2019;Qin et al., 2019;Su et al., 2019;Suga et al., 2019). The most striking structural difference between the PSI-LHCI structure of plants and algae/mosses is the presence of a second LHCI belt. In C. reinhardtii, Lhca5 was suggested to be essential for connecting the inner and the outer ring of LHCIs (Ozawa et al., 2018) and facilitating excitation energy transfer (EET) between the rings (Suga et al., 2019). Interestingly, this is the only LHCI position in the second belt that is conserved in P. patens (Iwai et al., 2018;Pinnola et al., 2018) and in multiple other algal PSIs (Qin et al., 2019;Su et al., 2019;Suga et al., 2019). Thus, Lhca5 might be a structural determinant for the association of the second belt, and it was possibly lost in vascular plants that instead favor a smaller PSI-LHCI complex (Neilson and Durnford, 2010). The binding affinity of Lhca2 and Lhca9 to the core is weak in C. reinhardtii (Drop et al., 2011;Su et al., 2019;Suga et al., 2019) but strong in Bryopsis corticulans (Qin et al., 2019) and B. braunii, as suggested by the fact that these positions are occupied in all the observed particles. This difference may be due to the presence of PsaM (Fig. 7D,  pink), which in B. corticulans is located in the proximity of Lhca 2 and Lhca9 (Qin et al., 2019) and is present in B. braunii but not in C. reinhardtii.
In conclusion, the results reported here show the high complexity and diversity in the composition and organization of the PSI antenna in algae. It is worth noting that PSI can afford a large antenna exactly because the extremely fast trapping time assures a very high quantum efficiency (Croce and van Amerongen, 2020).

Genomic Analyses of PSI in Botryococcus braunii
Query sequences from Arabidopsis (Arabidopsis thaliana) and Chlamydomonas reinhardtii (Supplemental Table S1) were obtained from the UniProt database and used for similarity searches with the BLASTP tool at the National Center for Biotechnology Information in the B. braunii genome (not annotated; Browne et al., 2017). The presence of homologs in the B. braunii genome was accepted at E-values ,10 25 and/or Bit scores .45.
Isolation and Purification of PSI Supercomplexes from B. braunii B. braunii strain CCALA778 was obtained from the Culture Collection of Autotrophic Organisms and cultured in 1 L modified CHU-13 medium (van den Berg et al., 2018) in 2 L Erlenmeyer flasks, bubbled with 5% (v/v) CO 2enriched air and shaken at 150 rpm under continuous illumination with white fluorescent light of 15 mmol photons m 22 s 21 . The culture was harvested in the logarithmic growth phase, and thylakoids were prepared as described previously (van den Berg et al., 2018). Solubilization with digitonin did not work on the B. braunii thylakoid membranes and established protocols for styrenemaleic acid copolymer did not yield usable membranes. a-DDM solubilization of the B. braunii thylakoid membranes yielded the same Suc band pattern as b-DDM solubilization, but with lower yield and more contaminating proteins. Therefore, mild b-DDM solubilization was used. To purify PSI-LHCI supercomplexes, thylakoids were solubilized at a Chl concentration of 2 mg mL 21 with 0.5% b-DDM (w/v) in 400 mM NaCl and 20 mM Tricine-NaOH (pH 7.8) for 20 min at 4°C, then loaded on a Suc density gradient obtained by freezing and thawing 0.5 M Suc, 20 mM Tricine-NaOH (pH 7.8), and 0.05% b-DDM (w/v) and centrifuged for 17 h at 240,000g at 4°C.

Steady-State Spectroscopy
The sample buffer used for all room temperature experiments was 0.5 M Suc, 20 mM Tricine (pH 7.8), and 0.05% b-DDM (w/v). In addition, for the 77 K experiments, the buffer contained 66% (w/w) glycerol. Sample OD at the maximum in the red region of the spectrum was adjusted between 0.8 and 1 for absorption and below 0.05 for fluorescence measurements. Absorption spectra were recorded with a Cary 4000 spectrophotometer (Varian). For 77 K measurements samples were cooled in a cryostat (Oxford Instruments). The 77 K absorption spectra were measured with a UV-2600 spectrophotometer (Shimadzu). Fluorescence emission spectra were recorded on a Fluorlog 3.22 spectrofluorimeter (Jobin-Yvon). For fluorescence emission spectra, the spectral bandwidths were 3 nm for excitation (440, 475, and 500 nm), and 1 nm for emission. An optical filter was placed before the detection path to block light ,600 nm.

Pigment Analyses
PSI complexes from eight pooled CNBs were eluted overnight in 10 mM Tricine-NaOH (pH 7.8) with 0.05% (w/v) b-DDM and then concentrated in concentrators with a 3 kD cutoff (Amicon Ultra, Millipore) at 7,500g. Pigments from Suc bands and eluted CNBs were extracted in 80% (v/v) acetone and analyzed by HPLC (System Gold 126, Beckman Coulter) with the previously described protocol (van den Berg et al., 2018).

Functional Antenna Size
The rate of P 700 oxidation that is directly proportional to the absorption cross section was measured directly on multiple pooled CNBs and in a separate experiment on thylakoids, both with similar, low optical density at 630 nm to avoid concentration-induced shading. A JTS-10 spectrometer was used (BioLogic) and absorption changes were monitored at 705 nm with a 10 mm interference filter (10 nm full width at half-maximum; Schott). Detecting light was filtered by 3-mm-thick Schott RG695 glass filters, while actinic light (light-emitting diodes peaking at 630 nm, ;60 mmol photons m 22 s 21 ) was turned off for ;300 ms during each detection to avoid artifacts. The gel pieces were incubated for 30 min in a solution containing 1 mM methyl viologen (Sigma) to prevent acceptor-side limitations, and 4 mM sodium ascorbate to prevent donor-side limitations. It was systematically verified that the rate of oxidation was at least one order of magnitude faster than the rate of P þ 700 reduction and, conversely, that the reduction proceeded to completion between illuminations and an identical quantity of P þ 700 was reached upon each oxidation.
The rates of P 700 oxidation in B. braunii were compared with that of known complexes of Arabidopsis (Wientjes et al., 2009). Curves were minimummaximum normalized before averaging and fitting. The rates were fitted with monoexponential function in OriginLab software. The measurements were averaged three to six times and performed on two independent CN gels; thylakoids were measured five times on two independent preparations.

Single-Particle Analyses
Multiple CNB1-containing PSI-LHCI supercomplexes were excised and placed in an Eppendorf tube with 60 mL of buffer (10 mM tricine-NaOH, 0.05% [w/v] b-DDM [pH 7.8]) at 4°C overnight with continuous stirring (Kou ril et al., 2014). Spontaneously eluted samples were used for the preparation of EM specimens by negative staining using 2% (w/v) uranyl acetate on glow-discharged carbon-coated copper grids. Approximately 7,300 images were collected using semi-automated GRACE software (Oostergetel et al., 1998) on a FEI Technai T20 microscope equipped with a LaB6 cathode, operating at 200 kV. Images of 2,048 3 2,048 pixels were recorded at 133.0003 magnification using a Gatan 4000 SP 4 K slow-scan charge-coupled device camera with a pixel size of 0.224 nm. From the selected EM micrographs, .55,800 particles (top-view PSI) were picked for single-particle analysis using the SCIPION image processing framework (de la Rosa-Trevín et al., 2016).

Time-Resolved Fluorescence
Picosecond time-resolved fluorescence measurements were performed with a streak camera setup as described previously (Le Quiniou et al., 2015a). Samples were placed in 10 mM Tricine (pH 7.8), 0.05% (w/v) b-DDM, and 0.5 M Suc at an optical density (OD) at 679 nm of 1.2 and measured in a 10 3 10-mm quartz cuvette (Hellma Analytics) at room temperture. During the measurements, the sample was stirred with a magnet bar (1,500 rpm). To minimize reabsorption, the laser beam (400 or 475 nm) was focused on the sample close to the cuvette wall and emission was collected at the right angle. To avoid singletsinglet annihilation, the pulse energy was reduced to 0.4 to 0.6 nJ (100 mW [400 nm] or 140 mW [475 nm] measured at the sample position). A power study confirmed the absence of annihilation (Supplemental Fig. S5). Colonies were measured in fresh culture media with 20 mM DCMU (Sigma) and 1 mM HA (Sigma) at an OD at 679 nm of 0.5 and a pulse energy of 0.13 nJ. Fluorescence was detected from 590 to 860 nm and 0 to 155 ps (time range [TR] 1; temporal response, 4-5 ps) and 0 to 1,500 ps (TR4; temporal response, 18 ps), and each dataset consisted of a sequence of images: 400 images of 10 s at TR1 and 100 images of 1 min at TR4. Image sequences were corrected for background, shading, and jitter (temporal drift between images within an image sequence) and finally averaged in HPD-TA 8.4.0 (Hamamatsu). These corrected datasets were binned to 2 nm, and zoomed between 640 and 800 nm in Glotaran 1.3 (Snellenburg et al., 2012). The datasets were analyzed globally with a sequential model in order to extract a minimum number of exponential components, n (with increasing lifetimes tn), that can satisfactorily describe the data (van Stokkum et al., 2004). The instrument response function was modeled as a gaussian with a full width at half-maximum of 4 ps (TR1) or 26 ps (TR4). Average lifetimes were calculated as S(A*t)/SA, with amplitude A taken as the area of the DAS over the 640 to 800 nm interval.

Supplemental Data
The following supplemental materials are available.
Supplemental Table S1. Overview of homologous protein sequences detected in the B. braunii genome with query sequences from Arabidopsis and C. reinhardtii.
Supplemental Figure S3. Structural models of two less abundant classes of PSI-LHCI supercomplexes isolated from B. braunii.
Supplemental Figure S4. Functional antenna size of PSI in B. braunii thylakoids compared to Arabidopsis thylakoids.
Supplemental Figure S5. Time-resolved fluorescence measurements of the PSI-LHCI complex isolated from B. braunii measured with a streak camera.

ACKNOWLEDGMENTS
Dr. Lijin Tian is acknowledged for his kind advice regarding the streak camera measurements. Maurits Dijkstra is acknowledged for his help with similarity search in the B. braunii genome.