Excitation energy transfer kinetics of trimeric, monomeric and subunit-depleted Photosystem I from Synechocystis PCC 6803

Photosystem I is the most efficient photosynthetic enzyme with structure and composition highly conserved among all oxygenic phototrophs. Cyanobacterial Photosystem I is typically associated into trimers for reasons that are still debated. Almost universally, Photosystem I contains a number of long-wavelength-absorbing ‘red’ chlorophylls (Chls), that have a sizeable effect on the excitation energy transfer and trapping. Here we present spectroscopic comparison of trimeric Photosystem I from Synechocystis PCC 6803 with a monomeric complex from the ΔpsaL mutant and a ‘minimal’ monomeric complex ΔFIJL, containing only subunits A, B, C, D, E, K and M. The quantum yield of photochemistry at room temperature was the same in all complexes, demonstrating the functional robustness of this photosystem. The monomeric complexes had a reduced far-red absorption and emission equivalent to the loss of 1.5–2 red Chls emitting at 710– 715 nm, whereas the longest-wavelength emission at 722 nm was not affected. The picosecond fluorescence kinetics at 77 K showed spectrally and kinetically distinct red Chls in all complexes and equilibration times of up to 50 ps. We found that the red Chls are not irreversible traps at 77 K but can still transfer excitations to the reaction centre, especially in the trimeric complexes. Structure-based Förster energy transfer calculations support the assignment of the lowest-energy D ow naded rom http://pndpress.com /bchem j/article-oi/10.1042/BC J20210021/1/bcj-2021-0021.pdf by gest on 29 M arch 2021 Bchem al Jornal. This is an Acepted M ancript. ou re encuraged to se he Vrsion of R eord tat, w en puished, w ill relace his vesion. he m st up-tote-version is avilable at https://drg/10.1042/BC J210021

state to the Chl pair B37/B38 and the trimer-specific red Chl emission to Chls A32/B7 located at the monomer-monomer interface. These intermediate-energy red Chls facilitate energy migration from the lowest-energy states to the reaction centre.

Introduction
Photosystem I (PSI) is a membrane-bound pigment-protein complex essential for all oxygenevolving photosynthetic organisms. It has a highly conserved structure and composition [1][2][3].
PSI consists of 11-14 protein subunits coordinating 94-96 chlorophylls (Chls), about 24 carotenoids (Cars) and other cofactors [4][5][6]. In cyanobacteria, PSI is usually found in a trimeric or monomeric form, though recently, tetrameric forms have been reported in heterocyst-forming species [7]. The oligomerization state may change towards the monomeric form under different growth conditions, such as light, temperature and available nutrients [8][9][10]. In PSI of the model cyanobacterium Synechocystis PCC6803, 86 Chls are associated with the two largest central subunits of the heterodimeric core, PsaA and PsaB, while 9 Chls are bound to the smaller subunits F, J, K, L, and M [6,11]. PsaL, a 16-kDa hydrophobic protein subunit of the PSI core, was found to be essential for the formation of trimers [12]. The rest of the small subunits, however, could each be deleted without observing any severe growth defects in cells [12,13]. A minimal PSI complex from Synechocystis, consisting only of A, B, C, D, E, K and M subunits, has been recently obtained and crystalized; however, no high-resolution structure could be resolved, casting some doubt on the intactness of the complex [6].
PSI of almost all organisms contains long-wavelength Chl forms, dubbed 'red' Chls, absorbing light at wavelengths longer than the absorption of the reaction centre (RC) Chls P700 [14,15], and thus broadening the absorption spectral range. The red Chls have unusual spectroscopic properties, such as large bandwidth, Stokes shift and electron-phonon coupling, that result from the mixing of excitonic and charge-transfer (CT) states [16,17]. The number of red Chls and their spectral properties vary among species and oligomeric forms of PSI. Deconvolution analyses of the absorption spectra of PSI of Synechocystis PCC 6803 [18] and spectral holeburning studies [19,20] have indicated that ~5 red Chls are present in trimers and ~3 in monomers. It has been recognized that the red forms originate from strong Coulombic Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20210021/905591/bcj-2021-0021.pdf by guest on 29 March 2021 interactions between the Chls and several candidate pigments (or pigment groups) have been proposed [4,[21][22][23][24][25][26], but these assignments remain largely hypothetical.
Although the red Chls account for only a small fraction of the total absorption cross-section [27], they have sizeable impact on the dynamics of energy transfer (ET) and trapping, as the ET from the red Chls to the RC is energetically uphill and thermally activated [28,29]. At cryogenic temperatures, the thermal energy is not sufficient for the uphill ET and the long-wavelength pigments act as (pseudo)traps increasing the excited-state lifetime in this region and dominating the fluorescence emission [28,30]. Energy equilibration between the bulk antenna Chls and the low-energy pigments absorbing below 700 nm is shown to take 2-4 ps in PSI from Synechocystis sp. PCC 6803 at room temperature (RT) [18,[31][32][33][34][35] significantly faster than the effective photochemical trapping time of ~20 ps. Recently, significantly slower lifetimes of ET have been uncovered in the PSI core from higher plants at cryogenic temperature [36]. Very few lowtemperature time-resolved fluorescence studies have been reported for cyanobacterial PSI and only for thermophilic species, containing a greater number of red Chls [37,38].
Because of its high quantum efficiency, relative abundance and structural and functional stability, cyanobacterial PSI is an attractive candidate for the development of bio-hybrid devices for solar energy conversion and biosensor applications [39][40][41]. Engineering a 'minimal' PSI with reduced protein content, requiring less cellular resources to produce and a smaller footprint but retaining full functionality can evidently be advantageous for such developments. Here we report a comparative investigation of PSI trimers and monomers from Synechocystis PCC 6803 as well as subunit-depleted 'minimal' PSI ΔFIJL with a focus on the pigment spectral forms affected by the oligomerization state and subunit depletion and on the excitation dynamics at room and low temperature. Previously, it has been shown that monomers exhibit differences in absorption, emission, and circular dichroism (CD) spectra compared to trimers [10,42]. We examined these difference in detail to determine the pigment groups and interactions underlaying them. While trimers and monomers showed similar trapping times at RT [18,32], specific differences are observed in the kinetics of equilibration with the red Chls at 77 K. We confirm that the subunitdepleted PSI is fully functional with indistinguishable picosecond excitation kinetics.

Thylakoid membranes preparation
Thylakoid membranes were isolated following the protocol of Zakar et al. [43]. One-week old cells were harvested and washed in buffer A (20 mM MES/NaOH, pH 6.4, 10 mM MgCl 2 , 10 mM CaCl 2 ) at RT by centrifugation for 5 min at 6000 g. The pelleted cells were resuspended in buffer A supplemented with 25% glycerol, 0.5 mM phenylmethanesulfonyl fluoride and 1 mM benzamidine and broken with glass beads (≤ 106-μm) using a Mini bead beater (Precellys Evolution) equipped with dry ice cooling compartment. After breaking, the suspension was centrifuged (4°C, 6 min, 4000 g) to remove the cell debris and remaining glass beads. Then the supernatant was further centrifuged for 30 min at 30000 g, 4°C and the pellet was solubilized in buffer A and used for further experiments.

Photosystem I preparation
PSI samples were isolated from freshly prepared thylakoid membranes following the protocol of Vajravel et al. [42]. The thylakoid membranes were solubilized by incubating with 2% n-dodecyl β-D-maltoside (β-DDM) at 4 °C for 30 min. The suspension was centrifuged for 15 min at 10000 g to remove the unsolubilized material. The supernatant was then loaded on a stepwise (6 steps, 0.2-0.9 M) sucrose gradient containing 20 mM HEPES (pH 7) and 0.05% of β-DM followed by centrifugation at 220,000 g for 17-18 h at 4 °C. The gradient fractions containing PSI trimers (from WT) and monomers (from WT, ΔpsaL and ΔFIJL mutant) were collected by a syringe. The samples were washed in a medium containing 0.03% β-DDM, concentrated using Amicon Ultra filters (Millipore) and stored at −80 °C until use. Additional experiments were performed on PSI minimal suspension obtained by incubating the crystals [6] in a buffer medium containing 0.03 % β-DDM (while agitating).

Pigment analysis
The pigment composition of all samples (thylakoid membranes, purified complexes) was determined by fitting the acetone extract spectra in the visible range from 350 to 750 nm with the spectra of the individual pigments. Absorption spectra of pure Chl a (Sigma-Aldrich) and Cars extracted from the thylakoid membranes were recorded in 80% acetone. Then these spectra were used as a reference to obtain a fit of the spectra of the acetone extracts of PSI samples. Molar absorption coefficients were taken from Lichtenthaler [44]. For analysis of the Chls and Cars composition by high-performance liquid chromatography, the pigments were extracted in pure acetone and the quantification was performed using a Shimadzu Prominence HPLC system as described in Zsiros et al. [45]. The pigments were identified according to their retention time and absorption spectrum and quantified by integrated chromatographic peak area recorded at the wavelength of maximum absorbance for each kind of pigments using the corresponding molar decadic absorption coefficient [46].

Absorption and circular dichroism spectroscopy
Absorption and circular dichroism spectra in the range of 350-750 nm were recorded at RT with a Thermo Evolution 500 dual-beam spectrophotometer and a Jasco J-815 spectropolarimeter, respectively. The samples were diluted in 20 mM Tricine buffer (pH 7.5) with 0.03% β-DDM to an absorbance of one at the red maximum. The measurements were performed in a standard glass cell of 1-cm optical path length with 1 nm (absorption) or 2 nm (CD) spectral bandwidth.

Structure-based modelling of the excitation dynamics
Semi-quantitative calculations of the dynamics of ET in PSI were performed using Förster theory as described in Byrdin et al. [23] with the assumptions and limitations therein and with modifications as follows. Distances, relative orientations and electronic couplings between Chls are calculated using atomic coordinates in the trimeric PSI structure of Synechocystis PCC 6803 [6]. Excitonic interactions are neglected, except for the RC Chls. All bulk Chls are assigned gaussian absorption lineshapes with central wavelengths normally distributed within the Q y absorption band and uniform widths (σ = 80 cm⁻¹) along with two vibronic/Q x sidebands shifted by 1100 and 2300 cm⁻¹. The fluorescence spectra mirror the absorption spectra with a Stokes shift of 140 cm⁻¹. Separate spectral parameters are set for the RC and linker Chls as well as for the designated red Chls. The absorption/emission spectra are calculated for 1000 random realizations and Förster ET rates are estimated for each Chl pair based on the distance, orientation and spectral overlap. Additional decay rates of 0.5 ns⁻¹ and 1 ps⁻¹ are imposed for all antenna Chls and for P 700 , respectively. The population kinetics is calculated by inverting the transfer rate matrix. Fluorescence decays at different wavelengths are calculated from the time-dependent population and the fluorescence emission spectra. Simulated fluorescence data are generated by convolving the disorder-averaged decays with a model IRF and adding Poisson noise.

Absorption and CD spectra
The absorption spectra of trimeric and monomeric PSI complexes isolated from Synechocystis  Table S1), in agreement with previous studies [18].
The L subunit of PSI binds two Cars and the F, I and J subunits bind five more [5,6].
Accordingly, the monomeric PSI complexes, especially ΔFIJL, have decreased absorption around 500 nm. The HPLC analysis showed 15-30% lower content of carotenes, echinenone and zeaxanthin in the ΔFIJL PSI ( Supplementary Fig. S1, Table S2). For a more precise quantification we performed spectral decomposition of the absorption spectra of acetone pigment extracts from the three PSI types using purified Chl a and total Cars extracted from the thylakoid membranes ( Supplementary Fig. S2). From the fits we estimate 24, 22 and 17 Cars in PSI from WT, ΔpsaL and ΔFIJL, respectively. The values exactly match the expected number of Cars, confirming that no additional pigments or subunits were lost during the isolation of PSI. The CD spectra of the monomeric PSI types were recorded to test for possible changes in the pigment exciton interactions (Figure 2). In the red region, the amplitude of the negative excitonic CD band (688 nm) diminished by 25% in both monomeric PSI types with no distinguishable difference among the two. In the blue-green region, the CD spectrum of PSI WT shows an intense broad band with a maximum at 505 nm, originating mainly from β-carotenes. It was previously observed that this band is sensitive to the oligomerization of PSI [10,42] and can be used as a fingerprint of trimeric PSI in vivo as well as in isolated complexes [48]. The band intensity was 58% smaller in ΔpsaL complexes relative to WT PSI, whereas the absorption in this region is hardly affected (as ΔpsaL PSI contains just two Cars less than WT). The exact same result was obtained from monomers isolated from WT cells as well ( Supplementary Fig. S3).
Even more striking is the 85% loss of CD 505 in the minimal PSI, where the total Car content is reduced by 29% (Supplementary Fig. S2). Virtually identical CD spectra were recorded from the thylakoid membranes of WT, ΔpsaL and ΔFIJL (Supplementary Fig. S4)ruling out the possibility that the differences are due to the PSI isolation procedure.

Fluorescence kinetics at RT
The RT fluorescence decay curves over a 2-ns time period were subjected to global analysis with a three-exponential model. The main fluorescence decay lifetime (with 98% amplitude) of trimeric PSI isolated from WT Synechocystis was 25±1 ps (Table 1)   The RT emission spectrum of PSI represents a thermal quasiequilibrium between all emitting Chlsbulk and redand can reveal the different red spectral forms and their abundance. Note that the steady-state emission spectrum measured from trimeric and monomeric PSI [32], is not suitable for such quantification because of the very high relative fluorescence yield of the free Chls and PSII. We decomposed the 25-ps DAES into two 'bulk' Chl pools and three red pools emitting at 707, 715 and 720 nm (see Supplementary Table S3 and     To test how the oxidation state of the RC may affect the fluorescence decay kinetics, measurements were also performed with P 700 pre-oxidized by preillumination in the presence of 1 mM potassium ferricyanide. Under these conditions the average fluorescence lifetime in the farred region was longer, mainly due to the increased relative amplitude of the 1 ns decay component. Conversely, in samples that were frozen in darkness with electron donors (ascorbate and phenazine methosulfate) to increase the proportion of open RCs, the average lifetimes at 730 nm were shorter ( Table 2). These results indicate that excitations at the longest wavelength Chls can be transferred to P 700 and used for photochemistry at 77 K.

Induced CD of carotenoids
Of the 24 carotenoids per monomer in the trimeric PSI [6], only 20 were resolved in the monomeric PsaL His PSI [11] two are missing from the L subunit and two in subunit A may be unresolved in the lower-resolution structure. Five more Cars are coordinated by the F, I, and J subunits and therefore a total of at least seven Cars should be absent in the minimal PSI. In principle, several other Cars located between these and the 'core' subunits, could also be lostfor instance, hydroxyechinenone I4021but this does not seem to be the case if we rely on the quantification of pigment acetone extracts. It is remarkable then, that the loss of seven Cars results in >80% reduction of the CD intensity at 505 nm. Clearly, only a few Cars are responsible for the CD.
Cars in solution rarely exhibit optical activity but they can acquire CD when bound to proteins [50,51]. There are two types of induced CDone is when the achiral molecules adopt different conformations such that atom displacement breaks the molecular symmetry, and another is when  S7). Interestingly, the majority of them are xanthophylls (B4011, F4016, I4020, M4021), such as zeaxanthin and echinenone, but also several carotenes and 9-cis carotenes (A4019, B4018).
Despite a very minor loss of Cars in ΔpsaL PSI (only 2-3 Cars), it has a greatly reduced CD spectrum. The CD difference spectrum shows that the missing Car is red-shifted (512 nm maximum), which may indicate a contribution from xanthophylls like echinenone that have more conjugated -bonds compared to carotene [52]. The I subunit, shown to be lost in other psaLdeficient monomers [6], binds echinenone or other xanthophylls at position I4020this Car might contribute to the CD band at 511 nm. The F/J subunits bind only three Cars, including the 9-cis zeaxanthin F4016, that contribute 30% of the total CD. A more comprehensive analysis comparing PSI from different cyanobacterial strains augmented with structure-based calculations [53] could validate the assignment of the Car CD.

Red Chls in trimeric PSI
Trimeric PSI from different cyanobacterial species has been shown to contain additional red absorption that is not observed in monomers [18,20]. Upon comparison of the absorption as well as emission spectra of trimeric and monomeric PSI from Synechocystis, we determined that trimers contain an equivalent of 1.5-2 additional red Chls (based on equal oscillator strength).
These red Chls emit with a maximum at 707-708 nm at RT (Figure 3) and 710-715 nm at 77 K ( Figure 4)in contrast, the lowest-energy states peak at 722-724 nm. Since a high-resolution crystal structure of PSI from a thermophilic cyanobacterium has been available, several groups have attempted to identify the Chls responsible for the red absorption/emission, assuming that they need to be strongly excitonically coupled [4,21,23,26]. One strongly coupled pair that has been promoted as a likely candidate for the lowest-energy red Chls is A32/B7, located at the monomer-monomer interface, close to the I/L subunits. As our time-resolved fluorescence shows, the longest-wavelength emission is hardly affected in monomeric PSI. Hence, if Chls A32/B7 give rise to the trimer-specific red absorption, they cannot harbour the lowest-energy state, as previously suggested [21,23]. We propose that A32/B7 forms a state emitting around 710 nm in the trimer but is blue-shifted in the monomeric complex due its altered environment (that can modify the pigment excitonic coupling as well as the charge-transfer character of the excited state). The lowest-energy states emitting at 722-724 nm remain to be located on one or more Chls outside the monomerization region, e.g. A38/A39, B31/B32, B37/B38 [23][24][25]54].

Excitation dynamics in PSI
The deletion of the four subunits F, I, J, L from PSI results in discernible changes in its spectroscopic characteristics but the overall excitation trapping rate is unaffectedit remains around 25 ps at RT in PSI with open RC (where we have used external electron donors to quickly re-reduce P 700 ). This means that despite the changes in the energetics of the antenna, the quantum yield of photochemistry of the minimal PSI is the same as in the WT trimeric complex. These results illustrate the functional robustness of this remarkable enzyme. Nevertheless, we observed differences in the low-temperature fluorescence kinetics of monomeric complexes that parallel the reduction in far-red emission.
Energy equilibration between the bulk antenna and the red Chls in PSI is thought to occur considerably faster than the photochemical trapping, at least at RT, resulting in trapping of a quasiequilibrated state [27,55]. Equilibration times in the order of 3-4 ps have been detected by time-resolved absorption and emission spectroscopy [18,33,34]. Our RT time-resolved fluorescence data are in line with thisalthough we are not able to resolve lifetimes shorter than ~10 ps, it is evident that the bulk-red equilibrium is reached well before trapping, which occurs with an effective lifetime of 25 ps. At 77 K, however, the red Chls are populated on different timescales spanning up to 50 ps. ET slows down at lower temperatures, on the one hand because of the reduced spectral overlap and, on the other, because of the restricted uphill transfer.
Significant slowdown of ET at 77 K has been shown in different photosynthetic complexes [see 56 and refs. therein, 57] and also modelled [58,59]. Another possible explanation for the slow spectral changes is that protein dynamics lowers the energy of the red states shifting the emission to longer wavelengths. CT states are associated with large reorganization energies [19,20] and protein motions are generally more temperature-dependent than ET and frozen at cryogenic temperatures [60]. If this is indeed the case, then any assignment of spectral forms to red Chls

Quenching by open and closed RCs
Global analysis of the time-resolved fluorescence data at 77 K resolved three far-red components with lifetimes from ~200 ps to 1.3 ns (Figure 4), resembling previous results on PSI from T.
elongatus [38] as well as the PSI core from higher plants [36]. It has been previously postulated that excitations on red Chls can be quenched by the RC (P 700 ) either in its neutral or oxidized (P 700 + ) state [14,23,38]. The broad near-infrared absorption of P 700 + allows it to efficiently quench excitations on the redmost Chls in Arthrospira (Spirulina) platensis and T. elongatus [37,38,61]. In the present work, however, we found that P 700 + is a less efficient quencher for the red Chls. The different behaviour is probably because of the shorter-wavelength emission of the red Chls in Synechocystis, and, consequently, larger spectral overlap with the absorption of P 700 and smaller overlap with P 700 A notable result is that the quenching of red states at 77 K is 30-40% faster in trimers ( Figure 5) than in monomers, showing that the oligomeric state of PSI has a measurable effect on the excitation dynamics, at least at low temperature. The differences appear to hold irrespective of the oxidation state of the RC (  [24,62]. We tentatively set the same absorption wavelengths (see Supplementary table   S4) Table S5).
The time-dependent population of the three red Chl states confirms that B37/B38 is the longestlived, whereas A32/B7 rapidly decays (Supplementary Figure S9). This is not surprising as the former is the lowest-energy form and A32/B7 is closer and more strongly coupled to P 700 . In monomers, the decay of the B31/B32 and B37/B38 pairs is slower than in trimers. The main reason for the slowdown is the lack of the red state at A32/B7, which can be verified by comparing the population dynamics in monomers with and without a red state on A32/B7 (Supplementary Figure S10). In other words, the intermediate-energy red state on A32/B7 acts as a bridge between the Chls emitting at longer wavelengths (especially B37/B38) and P 700 , increasing the overall probability for excitations on the red Chls to be transferred to the RC and used for photochemistry. Obviously, this depends on the mutual arrangement of the red Chls and the RC. The pair A32/B7 located on the donor side of the complex, close to P 700 , is well positioned to mediate energy transfer from the longest-wavelength form B37/B38 on the stromal side. Figure 7 illustrates the assigned red Chl pairs and the main routes of effective uphill ET between them according to the theoretical calculations. Note that the scheme is only qualitative as the actual ET rates will depend on the exact spectra and transition dipole orientations of the red Chls as well as on whether the RC is open or closed.