A thermophilic cyanobacterial strain, T. vulcanus was isolated from a hot spring in Yunomine, Japan (Koike and Inoue, 1983). Cells were grown photoautotrophically as batch culture in a BG11 medium (pH 7.0) at 50°C, and were continuously illuminated with a white fluorescent lamp at 50–100 μmol photons m⁻² s⁻¹ photon flux density (Shen et al., 2011). Cultures were aerated on a gyratory shaker operating at 100 rpm.

PSII CCs of T. vulcanus were isolated as described earlier (Shen and Inoue, 1993; Shen and Kamiya, 2000; Shen et al., 2011; Kawakami and Shen, 2018). During the final steps of the purification, dimers and monomers were separated from crude PSII by an anion exchange column, with a column buffer (30 mM MES-NaOH (pH 6.0), 3 mM CaCl₂, and 0.03% β-DDM) with a linear gradient of 170–300 mM NaCl. After collection of the monomer and dimer fractions PEG 1,450 was added at a final concentration of 13% to the samples which were then centrifuged to precipitate and concentrate them. Finally, they were suspended in a buffer containing 30 mM MES–NaOH (pH 6.0), 20 mM NaCl, and 3 mM CaCl₂ and stored in liquid nitrogen or at -80°C until use. For spectroscopic measurements, isolated PSII CCs were diluted in a reaction buffer (5% glycerol, 20 mM MES (pH 6.0), 20 mM NaCl and 3 mM CaCl₂).

For CD spectroscopy and fluorescence yield measurements PSII CCs were diluted to ~10 µg Chl mL⁻¹ in the reaction buffer, and for fluorescence emission spectroscopy it was diluted to ~1 µg Chl mL⁻¹. All experiments, except in section 3.1, were carried out on dimeric PSII CC. Stock solution of thylakoid lipids (MGDG, DGDG, SQDG and PG, Avanti Polar Lipids, Sigma) were prepared in methanol to a final concentration of 10 mg mL⁻¹. These lipids, according to Sigma, contain predominantly unsaturated (18:3) fatty acids – as typical for higher plant TMs (Siegenthaler and Murata, 1998). The PSII CC suspension was mixed in the dark with lipids at different concentrations (with methanol not exceeding 0.8% final concentration) along with 40 µM DCMU, which was dissolved in dimethyl sulfoxide (with a final solvent concentrations of 1%).

CD spectra in the range of 350–750 nm were recorded at room temperature with a Jasco J-815 spectropolarimeter (Jasco, Japan). The measurements were performed in a semi-micro quartz cell of 1 cm optical path length with 2 nm spectral bandwidth.

Chl-a fluorescence yields were measured using a Multi-Color (MC) PAM (Walz, Effeltrich, Germany). Fluorescence increments were induced by STSFs (Xe flashes, Excelitas LS-1130-3 Flashpac with FX-1163 Flashtube with reflector, Wiesbaden, Germany) of 1.5-µs duration at half-peak intensity. The frequency of the low intensity and non-actinic modulated measuring light was 1 kHz. This restricted the time resolution of our ChlF transition measurements to several milliseconds. Earlier, using DCMU-treated PSII CC of T. vulcanus exposed to repetitive STSFs, we have identified the formation and rapid recombination of the P₆₈₀ ⁺Pheo^- radical pair (Sipka et al., 2019). These recombination events evidently bring about, with a certain probability, the formation of triplet states, which under aerobic conditions relax in the time range of several microseconds (Lambrev et al., 2012; Laisk and Oja, 2020). These events do not destabilize the samples – as evidenced by double-STSF ChlF experiments presented in this paper, showing that doubling the flash intensity does not affect the fluorescence level at the millisecond time range [see also (Magyar et al., 2018; Sipka et al., 2021)].

The sample was placed on the sample holder of a thermoluminescence apparatus in order to control the temperature (Magyar et al., 2018). The timing of the flashes was controlled using a digital pulse generator (525 Six Channel Digital Pulse/Delay Generator, Berkeley Nucleonics, California, USA). The kinetic traces were recorded using MC-PAM’s own software.

The fluorescence transients were recorded after 10 min dark adaptation of the sample at room temperature, followed by an additional 5 min dark adaptation at 5°C, on which temperature all measurements were performed. At this temperature the recombination rate of the charge-separated state is very low, as reflected by thermoluminescence glow curves exhibiting the Q band above 20°C (Sipka et al., 2019); hence, allowing an easier determination of the magnitude of the Δτ-dependent increments between the fluorescence levels elicited by the first and second STSFs (see the schematic representation of the measuring protocol in Supplementary Animation S1 ).

The F ₁-to-F ₂ Chl-a fluorescence increments were determined by measuring the fluorescence level after the second STSF, delivered by different waiting times (Δτ) after the first flash, and were normalized to the F _m values, i.e., the maximal fluorescence levels, which were determined using saturating multiple-turnover long laser flashes. The (F ₂-F _o)/(F _m-F _o) values are plotted as a function of Δτ. For simplicity, the first data points obtained after simultaneously firing the two STSFs are given at Δτ = 10^-1 µs. The Δτ _1/2 half-rise time values, i.e., the Δτ values at 50% of the maximum of the F ₁-to-F ₂ increments, were determined using a logistic fit function.

Fluorescence emission spectra were measured at 77 K using an FP-8500 (Jasco, Japan) spectrofluorometer. Emission spectra in the range of 620–780 nm were recorded with excitation wavelength of 440 nm and excitation/emission bandwidth of 2.5 nm. The measurements were performed with 1 nm increments and 2 s integration time. Samples were cooled in a home-built accessory used with the FP-8500 spectrofluorometer. The spectra were corrected for the spectral sensitivity of the instrument using a calibrated light source (ESC-842, Jasco) as a reference.

The purity of PSII dimers and monomers was checked by blue native polyacrylamide gel electrophoresis on a 5–13% (w/v) gradient gel at 6°C according to (Schagger et al., 1994). The native gel was stained with Coomassie Blue G-250 to show all protein content of the samples. The amount of the samples loaded on each lane was 0.5 μg of Chl.

As revealed by X-ray crystallography of PSII CC of T. vulcanus, detergent molecules tend to accumulate in domains interconnecting the two monomers in the dimer (5 detergent molecules per monomer) (Umena et al., 2011; Yu et al., 2021). Thus, we tested if isolated monomers exhibit larger Δτ _1/2 values than dimers – given their potentially greater exposure to detergents than in dimers. However, as shown in Figure 1 , monomers and dimers display essentially the same half-waiting time values – rendering our hypothesis unlikely. These data are also consistent with our earlier finding that the stepwise Chl-a fluorescence rise, and the kinetics of the fast-fluorescence rise of monomeric and dimeric complexes are virtually indistinguishable (Sipka et al., 2021). Hence, the lipids ‘missing’ from the monomer-monomer interface are unlikely to be responsible for the increased Δτ _1/2 in PSII CC compared to the native systems ( Supplementary Material , Figure S1 confirms the oligomerization state of samples).

To test the effect of TM lipids separately on the rate-limiting step in the F ₁-to-F ₂ increment, we performed double-STSF induced fluorescence measurements on PSII CC of T. vulcanus, mixed with different TM lipids at different Chl:lipid ratios, and compared them with that of PSII CC dimers of the same batch. In a separate experiment, we checked that methanol – in which lipids were dissolved – exerted no effect at the given concentration (0.8% in the final volume) on Δτ _1/2. It displayed about the same half-waiting time, 1.5 ± 0.3 ms as the control, 1.6 ± 0.3 ms; methanol at 1.6% increased Δτ _1/2 to 3.51 ± 0.77 ms ( Supplementary Material , Figure S2 , Supplementary Table S1 ).

In earlier experiments (Magyar et al., 2022), PSII CCs were reconstituted into liposomes composed of TM lipids in proportions as in plants: 45% (w/v) MGDG, 30% DGDG, 15% PG, and 10% SQDG; at a final Chl:lipid ratio of 1:8. Here, we performed experiments to test the effects of individual lipid head group species on Δτ _1/2, using the same batch of PSII CC dimers. In this batch, the minimum Δτ _1/2 value of PSII CC embedded in liposomes was 0.5 ± 0.1 ms ( Supplementary Material , Table S1 ); this value is somewhat larger than in the native TMs and in the batch of PSII CC used earlier (Magyar et al., 2022). This suggests batch-to-batch variation in the lipid content of the isolated PSII CCs and/or in their capability of replenishing the missing lipid molecules.

In the following series of experiments, we added different isolated lipids and used different Chl:lipid ratios to test their effect on Δτ _1/2. The two non-ionic lipids, MGDG and DGDG exerted similar effects on the half-waiting time, which displayed slightly decreasing patterns towards higher lipid concentrations ( Figure 2 ). Upon gradually increasing the concentration of the externally added MGDG, from 1:0 to 1:8 Chl:lipid ratio, Δτ _1/2 of 1.6 ± 0.3 ms in the control gradually decreased to 0.8 ± 0.2ms ( Figure 2A ), indicating that more MGDG might be needed to occupy places in PSII CC responsible for the decrease of Δτ _1/2 in TMs. DGDG showed almost identical parameters with the exception of the measurement with the Chl:lipid ratio of 1:1, where Δτ _1/2 increased to 2.1 ± 0.3 ms ( Figure 2B ); these data suggest that DGDG has low ability to replace the detergent molecules around the site regulating the rate-limiting steps of PSII. Nonetheless, when applied at higher concentrations, it was capable of decreasing Δτ _1/2. It is to be noted that the slope (P) of the fitted curves became slightly steeper after the addition of these lipids ( Supplementary Table S1 ), a phenomenon which requires further investigation.

When the negatively charged lipids SQDG and PG were added, a larger decrease of the half-waiting times was observed ( Figure 3 ). With SQDG, this occurred only after applying 1:4 Chl:lipid ratio, reaching a value of 0.7 ± 0.2 ms ( Figure 3A ). In contrast, PG shortened Δτ _1/2 immediately after the addition of only one PG molecule per Chl, reaching about the same Δτ _1/2 value of 0.7 ± 0.2 ms, and no further decrease was attained by additional PG molecules ( Figure 3B ). It is also to be noted that in this case, the P values remained almost the same as in PSII CC ( Supplementary Table S1 ).

In these experiments, TM lipid molecules were added to PSII CC in different mixtures and proportions, keeping the same Chl:lipid ratio of 1:8 ( Figure 4 ). It was interesting to observe that PG and SQDG when combined with MGDG were less efficient than when they were applied alone (displaying Δτ _1/2 values of 0.93 ± 0.07 ms and 1.08 ± 0.27 ms vs. about 0.7 ms). It seems that in each case, the most abundant TM lipid species, MGDG, hindered rather than facilitated the PG-induced shortening of Δτ _1/2 – most probably because of competing with PG and/or SQDG to replace the detergent and occupy the space of the missing lipids. The same appeared to be true for DGDG. With lipid mixtures of MGDG : DGDG : SQDG : PG of 4:2:1:1, which is close to the lipid-species ratio of plant TMs, Δτ _1/2 remained about the same as in the case of MGDG : PG. Further decrease was achieved when all lipids were present in equal concentrations (MGDG : DGDG : SQDG : PG of 2:2:2:2). In this case, the half-waiting time value (0.8 ± 0.1 ms) assumed almost the same value as with PG alone.

The observation that PG seemed to be more effective than SQDG, suggests that PG is the most important ‘missing’ lipid from isolated PSII CC. There are indications in the literature that lipids can exchange each other (Güler et al., 1996; Yu et al., 2002; Kansy et al., 2014), thus, as both of these lipids are negatively charged, it is highly likely that in the absence of additional PG, SQDG might occupy its place. It must also be taken into account that non-ionic lipid molecules might have been replaced by anionic ones or vice versa; however, because of the different head groups of the lipids, this scenario is highly unlikely, unless these lipids are bound to PSII through hydrophobic interactions and not by ligation. It can be envisioned that excess amounts of the negatively charged lipids SQDG and PG replace neutral lipids at the binding sites both in the core-antenna complexes CP43 and CP47 and the D1/D2 heterodimer – which thus may influence the configuration of the local electric fields. Kansy et al. (2014) suggested similar changes owing to the negatively charged lipids with the possible explanation of the dissociation of CP43.

CD spectra of PSII in the visible wavelength region specifically probe excitonic couplings between pigments in the CC. No specific changes in the CD spectra of the PSII CC were observed upon the addition of either of the four thylakoid lipids, except for minor deviations that could be attributed to baseline and differential light scattering effects due to the sample turbidity in the presence of lipids ( Figure 5 ). These data are in good agreement with the absence of detectable changes in the CD spectra of PSII CC in solution and in reconstituted lipid membranes (Magyar et al., 2022).

Previously we have shown that reconstitution of PSII CC in liposomes prepared with a thylakoid lipid mixture brought about significant changes in the fluorescence emission spectra recorded at 77 K (Magyar et al., 2022). Particularly, the emission peak at 687 nm gained intensity relative to the 694 nm peak upon insertion of PSII in the lipid membranes. These bands originating from CP43 and CP47, respectively are denoted F685 and F695 (Andrizhiyevskaya et al., 2005). We observed the same qualitative change upon adding different thylakoid lipids to PSII CC ( Figure 6 ), which might originate from replacement of some of the detergent molecules in the periphery of CC (Umena et al., 2011; Suga et al., 2015). It is interesting to note that the changes were markedly different depending on the class of lipids used. SQDG exerted no significant effect on the fluorescence spectrum, whereas MGDG and DGDG decreased the relative intensity of the 694 nm peak. The negatively charged phospholipid PG had the strongest effect suppressing the emission at 694 nm relative to the short-wavelength band. Qualitatively, the spectral changes induced by the addition of lipids, especially PG, could be related to the difference in the fluorescence emission spectra of monomeric PSII CC compared to dimeric CC ( Supplementary Figure S3 ). The similarity might suggest that the lipids destabilize the PSII dimers. The changes in both cases are consistent with the hypothesis that structural destabilization of the CC exposing the red-shifted Chls in CP47 to a more polar environment alters their excited-state energy distribution, in effect decreasing the fraction of CC complexes possessing F695 states.

In our earlier study we showed that externally added lipids accelerated the rate of the PSII_C-to-PSII_L transition (Magyar et al., 2022), and they were proposed to reoccupy the sites taken by detergent molecules.

By using the cryo-EM structure of PSII CC in β-DDM obtained by Kato et al. (2021), we can try to identify possible lipid-binding sites occupied by detergent molecules that may be influencing the PSII_C-to-PSII_L transition dynamics ( Figure 7 ). As pointed out above (Section 3.1), the ‘missing’ lipids, replaced by detergents, in the monomer-monomer interface are highly unlikely to affect the rate-limiting steps in the PSII_C-to-PSII_L transition, as characterized by Δτ _1/2. Also, as proposed in Section 3.4, detergent molecules at the periphery of the CC, in the vicinity of CP43 and CP47 might contribute to the observed lipid addition-dependent changes in the 77 K emission spectra; however, these effects were not correlated with the lipid-induced shortening of Δτ _1/2.

A possible explanation of our observations is that reoccupying possible sites by lipids facilitates the light-induced reorganizations and shorten Δτ _1/2. Based on the scheme explaining the de-excitation pathways in PSII CC (Shibata et al., 2013) and the proposed mechanism involving the effects of prominently strong stationary and transient electric fields around the RC complex (Sipka et al., 2021), it is assumed that the site of the ‘missing’ and replenishable lipids are to be found near the cofactors of the RCs. For example, there is a DDM molecule interacting at about 27 Å distance from the primary donors that could be a possible candidate ( Figure 7 ). It is in close contact with native lipids filling a gap between the D1 protein and the peripheral low-molecular-mass (LMM) subunits, in close contact with PsbJ. If this binding site influences the structural reorganizations of PSII, it will raise the question of the role of the LMM subunits, particularly PsbJ, in the light adaptation process. However, it cannot be ruled out that other sites not identified in the structure are responsible for the lipid/detergent exchange effects on the light-induced reorganizations.

According to an alternative, non-conflicting explanation, the externally added lipids surrounding PSII CC can prevent or reduce the probability of dissociation of structural lipids in the vicinity of the RC. It is very likely that association and dissociation of lipids in the protein complexes depend on the lipidic environment of PSII CC. In the absence of externally added lipids, some of the lipid molecules of PSII CC may gradually dissociate from the protein complex and might be released into the detergent solution. Such a mechanism might affect the binding of the two PG molecules LHG410 and LHG627 which have their fatty acyl chains located close to the primary donor Chl molecules, at a distance of 12 and 16 Å, respectively (Kato et al., 2021). Restoring the molecular environment around the primary donor can thus also be responsible for the observed exogenous-lipid dependent decrease of Δτ _1/2.

In general, lipids, because of their high structural flexibility, may act as mechanical transducer molecules facilitating reorganizations in the matrix of RC. The fact that the negatively charged lipid, PG, appears to be the most efficient in accelerating Δτ _1/2 might indicate that electrostatic interactions play important roles in the assembly and structural dynamics of the RC complex of PSII.