Direct energy transfer from photosystem II to photosystem I confers winter sustainability in Scots Pine

Seasonal changes in Photosystem II- and Photosystem I-related functional activities

To monitor photosynthetic performance, we recorded several PSII and PSI parameters during three consecutive growth seasons (2015–2016, 2016–2017, and 2017–2018) along with concomitant changes in daily air temperature and solar radiation (Fig. 1a–c). For simplicity, in the main figures we only present (here onwards) data from 2017 to 2018, which we divided into five distinct seasons, based on weather parameters: Summer (S, June–Aug), autumn (A, Sept–mid Nov), winter (W, mid Nov–mid Feb), early spring (ES, mid Feb–mid Apr), and late spring (LS, mid Apr–June). Data from the other two growth seasons are provided in Supplementary Figs. 1–3.

We found characteristic seasonal patterns in maximum PSII fluorescence (Fm) and maximum quantum efficiency of PSII (expressed as Fv/Fm), in accordance with earlier reports^8,10. Fv/Fm was highest in S, fell with reductions in ambient temperatures during A and W (Fig. 1d, e), and was lowest in ES (63% lower than in S), when low temperatures coincided with rises in solar irradiance (Fig. 1a–c; see also Supplementary Fig. 2a). During LS, Fv/Fm gradually increased with increasing temperatures and peaked in S. The seasonal changes recorded in Fv/Fm were found to be mostly affected by changes in Fm rather than Fo (basic intrinsic fluorescence) (Fig. 1d). For a deeper understanding of PSII performance, the fraction of absorbed light energy utilized by PSII photochemistry¹⁵ (ΦPSII) was measured (Fig. 1f). During W ΦPSII decreased significantly and reached minimal values (19% of S values) in ES (Supplementary Fig. 2b).

Furthermore, we tried to quantify the amount of absorbed light energy thermally dissipated by non-photochemical quenching (NPQ)³. The component of NPQ that play a crucial role under fluctuating light conditions is the fast component (ΔpH-PsbS dependent-qE³ or zeaxanthin dependent-qZ⁴), which increased with increasing light intensities in all the seasons. In S and A, the fast component did not reach a stationary phase even at 1500 µmol m⁻² s⁻¹ illumination, while in W and LS samples a stationary phase was reached at 500 µmol m⁻² s⁻¹ and in ES already at 300 µmol m⁻² s⁻¹ (Fig. 1g). Most importantly, in ES inducible steady-state NPQ was much smaller and overall slower due to the smaller amplitude of the fast component (~50% less than in S) in ES (Fig. 1h). This is due to the fact that ES needles have already developed static NPQ. Instead, the quantum yield of non-regulated and/or constitutive loss (ΦNO) of energy was high during ES (Fig. 1i) (Supplementary Fig. 2d). This strongly suggests that this static NPQ is the fraction of absorbed light energy neither going to drive photochemistry (Φ_PSII) nor thermally dissipated by rapid regulated NPQ processes (qE/qZ).

To confirm that this static quenching is a manifestation of a “sustained mode of quenching” we artificially relaxed the ES needles (hereafter, ER) for 48 h in low light (80 µmol m⁻² s⁻¹) with 18/6 h photoperiod. During this recovery, we observed no significant changes in Fv/Fm for 6–8 h (less than a 10% increase), a modest increase (30%) after 24 h (Fig. 1j), and almost complete (95%) recovery to S levels after 48 h. Fo did not change significantly throughout the recovery period (Fig. 1j), but changes in Fm followed the same recovery dynamics as Fv/Fm (Fig. 1j), although maximal fluorescence (Fm) was much higher in S than in 48 h ER samples. ΦPSII did not change significantly until 12 h of recovery but recovered to 50% and 90% of S levels after 24 and 48 h, respectively (Fig. 1k). qE/qZ-dependent fast NPQ kinetics followed similar patterns to ΦPSII (Fig. 1l). It should be noted that light-dependent induction of the fast component of NPQ (qE/qZ) was slower in the ER samples than in S samples, even after 48 h of recovery, which appears to be mainly due to photoinhibition.

Seasonal changes in the partitioning of absorbed light energy within PSI showed that Y(I) was highest in S, declined in A and W, and reached minimal values in ES (Fig. 2a). Although it decreased, Y(I) was much less affected than the PSII activity (ΦPSII) during the cold periods (Fig. 1f) (Supplementary Fig. 3a). Y(NA) followed the same seasonal pattern, with minimum values registered in ES and a steady recovery until S (Fig. 2b) (Supplementary Fig. 3b). In contrast to Y(NA), Y(ND) considerably increased during the cold periods and peaked (at 69% higher than S values) in ES, then gradually declined to minimum values in S (Fig. 2c, Supplementary Fig. 3c). This suggests that during early spring most of the absorbed light energy is dissipated by oxidized P700 (P700⁺) to prevent photoinhibition caused by over reduction of the acceptor side as iron-sulfur clusters get damaged¹⁶. During the first 3 h of the recovery period, ΦPSI did not significantly change, but it gradually returned to S levels over 48 h (Fig. 2d). Y(NA) did not change much during recovery (Fig. 2e). Y(ND) increased slightly during the first 3 h, but then gradually decreased, and ER samples did not quite reach S values. Based on these results, we conclude that the restoration of PSI light use efficiency takes longer than 48 h (Fig. 2f).

Sustained non-photochemical quenching dominates in early spring needles

To elucidate the mechanism of “sustained quenching” observed in pine during the early spring periods, ultrafast time-resolved fluorescence measurements on intact pine needles were performed using the setup and approach described previously^17,18. In addition, a special chamber was designed to keep the rotational cuvette with the sample at −20 °C (Supplementary Fig. 4). This chamber was used in order to maintain the “native” sustained quenched state in pine needles. Fluorescence decay traces were collected from pine needles in the same states as reported in the previous subsection: S, ES, ER: Summer needles (S) do not contain any “sustained quenching”, while E.spring needles (ES) contain the highest levels of “sustained quenching”. To resolve the mechanisms of this quenching, fluorescence decay traces of S needles in the original state (dark-adapted, Fig. 3a, Summer dark) were compared to the needles collected in ES when sustained quenching was present (Fig. 3a, Early spring) and after the quenching in the needles had been relaxed at room temperature (ER) (Fig. 3a, E. spring recovered). To understand whether sustained quenching observed in ES occurs via a similar quenching mechanism to light-induced quenching, S needles exposed to HL for 30 min (SQ) were also measured (Fig. 3a, Summer quenched). As shown by the direct comparison of the decay curves (Fig. 3a), the fluorescence in SQ needles is much shorter-lived as compared to that of S needles. However, the fluorescence decays of ES needles are still pronouncedly shorter-lived than that of the SQ samples, thus characterizing the “sustained quenching” in the ES samples as the most pronounced quenching at all detection wavelengths and under all conditions. In contrast, ER and S samples showed very similar fluorescence decays, indicating that the ES sample recovered quite (although not completely, vide infra) well from sustained quenching within 48 h of recovery treatment.

Global analysis of the fluorescence kinetics of pine needles

To get the first hints on the mechanism(s) underlying sustained quenching in early spring global analysis¹⁹ was performed on all datasets: ES, S, SQ (Fig. 3b), and ER (Supplementary Figs. 5–8) needles. Six decay-associated spectra (DAS) were required to fit the fluorescence kinetics in all four cases. Three DAS were tentatively assigned to PSI: 11–16 ps, 50–60 ps, 150–180 ps (Fig. 3b, Supplementary Figs. 5, 6). Their spectra and lifetimes are reminiscent of previously reported PSI-related DAS in other plant species in vivo^17,18,20. The fastest component represents energy equilibration between Chl_red and Chl_blue^21,22. The second DAS in S needles peaks around 685–690 nm with a broad emission in the 700–720 nm region. Therefore, it mainly represents PSI core decay in combination with the 710–720 Chls of Lhcas. The third DAS (185 ps) peaks at 680 nm with some contribution at 730 nm, and therefore, it should be primarily assigned to LHCII and some Chl_red in PSI. In quenched states (SQ and ES, in particular) these two DAS have substantially higher contribution around 680 nm and, as a result, the reconstructed steady-state PSI spectra have higher emission around 680–685 nm as compared to the unquenched states (S and ER see Supplementary Fig. 5). It indicates that upon quenching PSII and/or LHCII kinetics also contribute to these PSI-related components. In S and ER states, the remaining three DAS peak at 682 nm and therefore, are assigned to PSII (Fig. 3b and Supplementary Fig. 9). Their lifetimes are 0.5–0.6 ns, 1.5–2.0 ns, and 4 ns, similar to what was resolved previously for PSII in the closed state^17,18. However, in ES two considerably shorter-lived PSII-related DAS were observed with lifetimes of 0.3 ns and 0.8 ns. This demonstrates that PSII kinetics is indeed quenched strongly in ES needles. Additionally, we resolved a DAS with 8 ps lifetime in the ES sample. The “oscillations” in the spectra giving rise to the sharp bands are due to the mixing of the contributing components, representing excitation energy transfer (EET) from Chl a pools emitting at ~680 nm to those emitting at 690 nm and 700–730 nm, respectively. This suggests that there occurs efficient EET between PSII and/or LHCII and PSI in the ES needles. However, these processes could be unambiguously resolved only by target modeling.

In a nutshell, global analysis results suggest that sustained quenching in the ES state involves both PSII/LHCII and PSI through a direct energy transfer. This mechanism provides much stronger quenching than the high-light-induced quenching in the summer needles (SQ).

Target modeling of the fluorescence kinetics of unquenched pine needles

To resolve the mechanism(s) responsible for the “sustained quenching” in pine needles in detail we performed target analysis²³ of all the kinetic data. In target analysis, a specific kinetic model is tested using a linear time-invariant compartmental model. Each compartment in Fig. 3 is presented by a square box. Each box belongs to any of the three different components: PSI, PSII, LHCII quenched. Fluorescence emitting compartments are presented by colored boxes. The red flashes with numbers directing to that compartments represent the so-called excitation vector, providing the percentage of excitations directed to each compartment and should roughly illustrate the relative number of Chl a (i.e., relative Chl a antenna size) in it. Non-colored boxes represent non-emitting compartments, like radical pairs (RP) in RCs and CT states, which are contributing to the overall observable kinetics and thus have to be included. Target modeling is required to describe the real time-dependent concentrations of each compartment, rates between them (arrows in Fig. 3, ns⁻¹), and spectra (species associated spectra (SAS)) of the interacting components. Note, that the global analysis, presented in the previous subsection, is equivalent to a number of noninteracting, parallelly decaying compartments and gives only the first indications of the processes involved. It is a mere mathematical analysis, not considering any physical description or information on the studied system. By testing kinetic models that had been validated previously on isolated subcomplexes^24,25,26 it was possible to separate the PSI from the PSII kinetics. Therefore, our assignment of PSI and PSII is based on a combination of the used kinetic model, resolved rates as well as SAS. The target model and the excitation vector presented in Fig. 3c resulted in the best fitting quality (Supplementary Fig. 7). In S needles (Fig. 3c, Summer dark) the PSI kinetics was described with two emitting compartments (PSI red, Ant/core) and one non-emitting radical pair (RP, Fig. 3c). The resolved rates and SAS are strongly reminiscent of the ones reported for the PSI-LHCI complex²⁶. Kinetics from the PSII antenna/RC complex was described by one emitting compartment (Ant/core) and two non-emitting radical pairs (RP1 and RP2, Fig. 3c) as reported previously^17,24,25,27. According to their SAS and rates, the emitting species represents the equilibrated excited state of the PSII-LHCII super-complex. In our analysis, to achieve a very good fit, two pools of PSII complexes with different decay rates were required (pool 1, pool 2, Fig. 3c, see the detailed comparison of the residuals plots in Supplementary Fig. 7) for the S sample. The two pools differ in the rate constants of charge separation in a manner that is typical for PSII particles with different antenna sizes (see^18,19 for discussion and explanation), which result in different average lifetimes (Supplementary Table 3). In the S sample, PSI with its antenna accounts for 30% of the total absorption cross-section at the excitation wavelength (662 nm), while the two PSII pools account for 40% and 30% of the total absorption cross-section. No internal quenching related to any NPQ process of these PSII compartments was observed.

The target model for the ER needles is rather similar to that of S needles (Supplementary Fig. 5). The main difference is that the smaller PSII pool (with a very small absorption cross-section of ca. 10%) shows an unusually small charge separation rate. The most likely explanation of this phenomenon is a small percentage of PSII RCs has been photoinhibited during the 48 h relaxation process from the sustained quenching state, which agrees with NPQ measurements.

Direct energetic transfer between PSII and PSI is the dominating component of sustained quenching

As for the S samples, also the description of the ES sample (Fig. 3c) required a three-compartmental target model for PSI and the presence of two PSII pools, each described by a three-compartmental model. The two PSII pools differ in the rate constants of charge separation in a similar manner as in S samples. The kinetics are also very similar to that of the S samples, however with a slightly increased charge separation rate. This can be interpreted as a consequence of the partial reduction in the physical antenna size. This goes in line with an increase in the Chl a/b ratio in ES as compared to S samples (Supplementary Fig. 9, Supplementary Table 2). Resolved antenna SAS (Ant/Core compartment of PSI, Fig. 3) of PSI in ES samples contains stronger emission in the 680 nm region, as compared to S needles. That is why we suggest that some LHCII strongly coupled to PSI contribute to it, like recently reported²⁰.

The striking feature is the presence of a very strong contribution of a direct energy transfer from PSII to PSI. Without allowing for this direct energy transfer, the data could not at all be fitted adequately (Supplementary Figs. 6, 7). This process has a transfer rate of 4.4 ns⁻¹. A small compartment representing quenched and functionally detached LHCII was also needed for a satisfactory description of the ES samples (See more details on this component in the next section). Overall, the target analysis shows that direct energy transfer from PSII to PSI complexes provides the dominant quenching mechanism of PSII complexes in ES needles causing the pronounced “sustained quenching”.

Comparison of sustained quenching and light-induced quenching

To model the fluorescence kinetics of SQ needles again three-compartmental models for PSI and PSII were required. However, unlike in previous cases, one single PSII pool was sufficient to describe the PSII kinetics.

Besides PSII and PSI, one additional component was needed to satisfactorily describe the fluorescence kinetics in the SQ samples. The rates, spectra, and average lifetime (0.4 ns, Supplementary table 3) of these compartments are strongly reminiscent of what was reported previously for highly quenched LHCII aggregates²⁸, and the quenched LHCII complexes under NPQ conditions in wild type Arabidopsis¹⁷ and in Arabidopsis mutants depleted of the reaction centers²⁹. We, therefore, assign this highly quenched component to the quenched LHCII antenna energetically disconnected from both PSII and PSI. However, the addition of this quenched LHCII compartment was not sufficient to fully describe the fluorescence kinetics (Supplementary Fig. 7). The species-associated spectra (SAS) and kinetics of PSII still contained a substantial amount of 690–700 and 730 nm emission, entirely uncharacteristic for PSII super-complexes, but characteristic of PSI complexes^22,26. Interestingly, also the SAS of the PSI antenna/core complex contains a 680 nm emission, which is characteristic of PSII. This strongly suggests—as can be directly deduced from the spectral properties—the presence of direct energetic contact between PSI and PSII (Fig. 3c). Introducing such a connection in the target model indeed resulted in a considerably better fit (χ² decreased from 1.198 to 1.093, Supplementary Fig. 6). Also, the PSII and PSI kinetics became well-separated in such a model. The resolved direct energy transfer rate from PSII to PSI is quite high (2.3 ns⁻¹), implying that a part of energy absorbed by PSII is actually funneled to PSI. However, this rate of direct energy transfer in SQ needles is only about half of that in ES needles, suggesting that unlike in cold-induced sustained quenching, upon light illumination, spillover is not the major process. The contribution of the quenched antenna, on the other hand, plays a major role.

Efficiency of direct energy transfer quenching in different conditions

By evaluating the kinetic information provided in Fig. 3c one can get a quantitative estimation of the degrees of light use efficiencies (for photosynthesis) on the one hand, and for photoprotection/quenching on the other hand, present in both PSII and PSI in the different situations (c.f. Supplementary Fig. 8, Supplementary Table 4). In S needles 70% of the total absorbed energy drives PSII charge separation and only 30% PSI charge separation. In the SQ needles the percentage of energy flowing into the PSII reaction center drops to 7.1%, while it increases to 67% for PSI. Upon strong light illumination, most of the PSII reaction centers will be closed, and a fraction of that total energy flowing through the RC can produce harmful reactive species like ROS^30,31. Since the percentage of energy flow into PSII decreases 10 times (70% in S and 7.1% in SQ) in SQ as compared to S, this means that the potential of oxidative damage is reduced by a factor of 10 by the quenching process. This factor is increased by a further factor of approx. 2 due to the large pool of functionally detached and quenched LHCII. So, in SQ needles the overall quenching provides a protection factor of 20 for PSII compared to S needles. In the sustained quenched ES samples this effect is much more extreme: Only 1.5% of the total absorbed energy in the system flows through the PSII reaction center potentially causing oxidative damage (assuming all other factors are equal). Thus, in total, “sustained quenching” in ES samples provides a protection factor of approx. 40 to PSII (double of SQ) compared to S needles. This huge protection factor explains well why “sustained quenching” is so effective enabling pine needles to survive the harsh winter and early spring conditions.

Massive destacking of grana membranes in the winter

In accordance with early reports^32,33, we observed strong seasonal changes in chloroplast ultrastructure, including loss of grana stacks during early spring (Fig. 4a). Ultrastructural changes derived from morphometric analysis of electron micrographs (Fig. 4a) indicate that average numbers of grana per chloroplast steadily decreased from autumn to winter and reached minimal values in early spring, corresponding to just 68% of S values (Supplementary Table 1). More strikingly, the number of appressed thylakoid layers/granum dramatically declined from A (4.97) to ES (2.72) and rose back in S (6.50) (Supplementary Table 1). These changes corroborate the shift in grana stacking illustrated in Fig. 4b. In S and A, 3–6 layered grana stacks accounted for 60–70% of total stacks. In contrast in ES samples, 70–75% and 15–20% of the grana had only two and three layers, respectively (Fig. 4b). These changes in ES samples were accompanied by a transient doubling of the number of lipid globules (plastoglobuli) per cell (Supplementary Table 1) when the chloroplast structure deviated most strongly from the S state.

To obtain a deeper understanding of the thylakoid plasticity in ES needles we also subjected needles collected at specific early spring and summer dates to EM analysis after artificially induced recovery under different light conditions (Fig. 5a). When both S and ER samples were exposed to HL for 30 (SQ1 for S samples, ERQ1 for ER samples) and 60 min (SQ2 for S samples, ERQ2 for ER samples) similar destacking of grana (as of ES samples) have been observed (Fig. 5b). These changes in grana structures recorded in ER and S samples (Fig. 5c, d) following short term HL exposure strongly corroborate the high dynamic plasticity of the thylakoid membrane, which appears to be essential for acclimation to the harsh boreal winters.

Changes in pigment and protein composition of ES, ER, SQ, and S samples

We also analyzed pigments of the pine needles used for spectroscopic measurements and found substantial differences in pigment contents between S and ES needles (Supplementary Table 2). In S and ES samples the Chl a/b ratios were 2.8 and 3.4, respectively, indicating that S needles had higher relative amounts of Lhcbs and other Chl b-binding proteins. The Chl/carotenoid ratio was also substantially lower in ES samples than in S samples, indicating that they had higher amounts of carotenoids (mainly lutein and violaxanthin/zeaxanthin). We observed no significant difference in total Chl pigment contents between S and SQ samples.

For quantifying the core and antenna proteins, isolated thylakoids were subjected to SDS-PAGE and were immunoblotted against specific antibodies (Supplementary Fig. 9). This revealed that ES samples contain a lower abundance of core (PsaD, PsbD) and antenna (Lhcb2, Lhca4) proteins than in S, in line with earlier reports³⁴ (Supplementary Fig. 9). ER samples contained lower amounts of PsbD and Lhcb2 compared to S, which might reflect on the minor amount of photoinhibition seen in fluorescence data.

Source: Ecology - nature.com