Variable optical properties of light-harvesting complex II revisited

Understanding photosynthetic light harvesting requires knowledge of the molecular mechanisms that dissipate excess energy in thylakoids. However, it remains unclear how the physical environment of light-harvesting complex II (LHCII) influences the process of chlorophyll de-excitation. Here, we demonstrate that protein-protein interactions between LHCIIs affect the optical properties of LHCII and thus influence the total energy budget. Aggregation of LHCII in the dark altered its absorption properties, independent of the amount of prior light exposure. We also revisited the triplet excited state involved in light-induced fluorescence quenching and found another relaxation pathway involving emission in the green region, which might be related to triplet excited energy transfer to neighboring carotenoids and annihilation processes that result in photoluminescence. LHCII- containing liposomes with different protein densities exhibited altered fluorescence and scattering properties. Our results suggest that macromolecular reorganization affects overall optical properties, which need to be addressed to compare the level of energy dissipation.


Introduction
Major light-harvesting complex proteins of photosystem II (LHCIIs) form a trimeric structure in which chlorophyll (Chl) and carotenoid (Car) molecules (i.e., lutein, neoxanthin, and violaxanthin) are bound in a specific arrangement 1 . Photoexcitation of Chl generates the singlet excited state ( 1 Chl*), followed by ultrafast excitation energy transfer (EET) through neighboring Chls 35 within LHCII 2 . The excitation energy of 1 Chl* is eventually trapped at a photosystem reaction center (RC) where charge separation is induced and photosynthetic electron transport is activated.
Alternatively, 1 Chl* de-excitation can take place via emission of energy as fluorescence. Under high light (HL) conditions, saturation of photosynthetic electron transport generates a situation in which open RCs are not locally available, which results in an increased accumulation of 1 Chl*. This situation 40 considerably increases the yield of intersystem crossing from 1 Chl* to the triplet excited state ( 3 Chl*) 3 , which can then transfer the excited energy to a triplet ground state oxygen to produce singlet oxygen, a highly reactive species capable of photooxidative damage. To minimize 3 Chl* formation, LHCII proteins de-excite 1 Chl* by non-radiative relaxation mechanisms that are dependent on the interaction between neighboring pigments [4][5][6][7][8] . There are two different techniques generally used to 45 investigate these diverse 1 Chl* de-excitation mechanisms: i) ultrafast spectroscopy to observe the transient kinetics of EET among pigments on the order of picoseconds to nanoseconds; and ii) modulation fluorometry to observe the induction kinetics of Chl fluorescence quenching on the order reconcile the various mechanisms proposed in the literature.

The optical properties of LHCII are changed upon aggregation
We first used solubilized trimeric LHCIIs isolated from spinach thylakoid membranes to observe the kinetics of Chl fluorescence yields in the dark. As expected, the fluorescence yield stayed 70 constant because LHCII remained solubilized in detergent micelles during the modulation fluorometry measurement (Fig. 1a). The unchanged fluorescence yield also indicates that the measuring light (ML) was weak enough to cause no further accumulation of 1 Chl*. When we diluted the detergent concentration by putting the solubilized LHCII into the measuring buffer without the detergent, the fluorescence intensity gradually decreased (Fig. 1a), which reflects a typical process 75 of LHCII aggregation 11 . We confirmed that the intensity of ML did not influence the extent of the decreased fluorescence yield (Fig. 1b), suggesting that the gradual decrease of the fluorescence intensity is caused exclusively by the process of aggregation. Adding the detergent back to the buffer caused a rapid increase of the fluorescence intensity to the same level as observed in the solubilized LHCII (Fig. 1a), indicating that the process of aggregation causes no prolonged modification of 4 intrinsic properties of LHCII. When we removed the detergent completely from LHCII by dialysis, we observed much lower fluorescence intensity due to the highest degree of LHCII aggregation (Fig.   1a). Also, no change in the fluorescence yield was observed in the dialyzed LHCII, indicating that there was neither further aggregation nor change in intrinsic properties generated in the dark. These results suggest that the fluorescence intensity decreases due to the physical process of LHCII 85 aggregation independent of environmental light conditions.
Using the same LHCII samples, we measured the steady-state Chl fluorescence spectra at 77 K. We examined the results using two different ways of normalization: i) normalized by the emission at 530 nm from an internal standard (Alexa Fluor 430) to observe changes in fluorescence intensities of overall spectra and ii) at the main Chl fluorescence peak (Qy band) to observe changes 90 in spectral shapes. When the fluorescence spectra were normalized by the internal standard, the spectra of the aggregated and dialyzed LHCII samples exhibited a large decline at the Qy band compared to the solubilized LHCII (Fig. 1c). When the spectra were normalized at the Qy band, a shoulder of red-shifted emission appeared at ca. 720 nm in the aggregated LHCII as previously observed 8 and became more pronounced in the dialyzed LHCII (Fig. 1d, inset). This red-shifted 95 emission has been described to be due to changes in pigment interaction and/or local hydrophobic environment 12 , and it was suggested to be weak emission caused by Chl-Chl charge transfer states, which occur during a process of 1 Chl* de-excitation 8 , although the exact mechanism has recently been debated 13,14 . As discussed below, the relative fluorescence yields between each sample observed at room temperature and 77 K appeared to be different (i.e., the aggregated LHCII showed a lower yield 100 at 77 K than that measured at room temperature relative to the yield observed in the solubilized LHCII). In addition, when we measured absorption spectra, the dialyzed LHCII showed decreased absorbance at both Soret (400-470 nm) and Qy bands (Fig. 1e). Because we used the same Chl concentration of each sample (based on measurement of Chls extracted in acetone), these LHCII samples contain an equal amount of proteins and pigments. In fact, resolubilization of the dialyzed 105 LHCII increased the absorption, which was visibly noticeable (Fig. 1f). This change in absorption by aggregation is due to the optical distortion caused by two phenomena: light scattering and inhomogeneous distribution of pigments, the so-called sieve effect [15][16][17] . Theoretical and experimental studies have shown the complexity of the optical distortion that involves light scattering properties [17][18][19][20] . Because LHCII aggregates can scatter a considerable amount of light and also create the sieve 110 situation in which Chls are not distributed homogeneously within the sample 20 , the total probability of photoexcitation per volume becomes smaller than that of solubilized LHCIIs (i.e., the incident light is scattered and leaving the sample cuvette before being absorbed by Chls). Similar aggregationinduced changes in optical properties have been observed in non-photosynthetic proteins as well 21,22 .
Although we cannot exclude the involvement of other possible mechanisms (e.g. Chl concentration  2a). This light-induced quenching is considered to be caused by conformational changes within 125 LHCII 24 , which are suggested to promote singlet EET to lutein 6 (and zeaxanthin 4,5,7 , which is generated from violaxanthin by a violaxanthin de-epoxidase under HL conditions 25 , although zeaxanthin did not exist in any of our LHCII samples isolated after overnight incubation of leaves in the dark). In addition, in isolated LHCII without efficient energy trapping by RCs, the yield of 3 Chl* becomes significantly increased 3 , and the population of triplet excited state of Car ( 3 Car*) generated 130 by triplet EET from 3 Chl* also becomes significantly higher under HL conditions [26][27][28][29] . Despite the relatively long lifetime of 3 Chl* in a range of tens to hundreds of microseconds 30 , the interaction with Cars, especially within LHCIIs, drastically shortens the lifetime to the subnanosecond order due to efficient triplet EET from 3 Chl* to a neighboring Car [31][32][33] . The ultrafast triplet EET in LHCII has been suggested to have evolved to minimize the production of singlet oxygen in oxygenic photosynthetic 135 systems 32,33 . Triplet EET occurs via the Dexter mechanism, and Cars located in L1 and L2 sites (e.g. lutein) are considered to be the acceptors of the triplet excitation energy in LHCII 34- 36 . This implies that about a third of the total Chls in LHCII are able to undergo de-excitation by the 3 Car*, the socalled singlet-triplet (S-T) annihilation 28,29 . Thus, it is reasonable to consider that Chl fluorescence quenching in LHCII can occur through both singlet and triplet EET to lutein, especially in isolated 140 conditions under HL illumination.
On the other hand, the extent of the quenching observed in the aggregated LHCII induced by AL illumination was markedly larger than that observed in solubilized LHCII (Fig. 2a). Also, the extent of quenching increased with the intensity of AL (Fig. 2b). The resolubilization of the aggregated LHCII under AL caused a rapid increase in the fluorescence yield to a level similar to that 145 observed in the solubilized LHCII under the same light intensity (Fig. 2a). The dialyzed LHCII also showed additional quenching induced by AL illumination, even though no more aggregation process occurred (Fig. 2c). This result differentiates the effect of the aggregation process from that of light illumination, indicating that the light illumination caused the additional quenching within the aggregated structure. We did not observe a distinct difference in the aggregation-induced red-shifted 150 emission (ca. 720 nm) at 77 K between the AL condition and the dark (Fig. 2d, e as compared to Fig.   1c, d). Also, there was no additional optical distortion induced by HL (Fig. 2f). These observations suggest at least two possible explanations: i) the light-induced quenching occurs near the interface between each LHCII in the aggregate; and/or ii) the interaction of pigments at the interface allows an excitation energy migration among LHCIIs in the aggregate, increasing the connectivity to available 155 quenchers. The former possibility can be aligned with the aggregation-induced conformational changes 24 , which allow singlet EET to lutein 6 . The latter possibility is supported by the observations of annihilation processes, which occur significantly more in aggregated LHCIIs than in solubilized LHCIIs [37][38][39][40] . Such enhanced annihilation is also observed more in dimeric photosystem II complexes than in the monomeric complexes 41 . The quenching by annihilation processes, especially involving 160 the triplet excited state (i.e., S-T and triplet-triplet (T-T) annihilation), may explain the residual quenching after resolubilization, which is caused by the long-lived triplets that remain under HL conditions. Based on these observations, there are at least three different aspects that need to be taken into account when the level of quenching in LHCII is measured in vitro. First, the extent of 165 aggregation needs to be evaluated to determine the level of quenching. Not only does the concentration of detergents affect the extent of aggregation, but the solution pH also greatly affects micelle formation. Thus, the pH-shift treatment from neutral to acidic needs to be carefully analyzed because it would change the level of aggregation. Second, the amount of light exposure needs to be carefully determined. As shown in Fig. 2b, different AL intensities and durations of light exposure 170 generate different levels of quenching. Thus, the level of quenching needs to be determined by comparing before and after a certain amount of light exposure. Third, the involvement of triplet EET in quenching in aggregated LHCII needs to be considered. Previously, observations of triplet EET have been mostly done by ultrafast spectroscopy. However, AL treatments and saturation pulses generally used in modulation fluorometry would be strong enough to generate 3 Chl*. It is well known that the triplet excited state of Chls and Cars will cause different optical properties 3,34 . In the next section, we took a different approach by using microscopy techniques to examine changes in optical properties of LHCII induced by excess light energy.
A newly observed radiative process occurs in the green region simultaneously with Chl de-180 excitation In relation to the triplet EET described above, we observed a curious emission in the green region from LHCII. We solubilized thylakoid membrane proteins with a mild detergent and separated them by native protein gel electrophoresis. The fluorescence image of the gel indicated that the protein complexes showed emission signals from not only the Chl fluorescence region but also the green 185 region, especially more visible from LHC-containing bands (Fig. 3a). Intriguingly, when we illuminated a part of the native gel containing the aggregated LHCIIs for 1 min using a confocal laser scanning microscope, not only did Chl fluorescence yields decrease, but also the intensity of the green emission became stronger than that outside of the illuminated region (Fig. 3b). The same green emission was observed in chloroplasts under normal conditions in vivo (Fig. 3c). When a part of the 190 chloroplast was strongly excited for 0.1 second, the Chl fluorescence yield in the strongly excited region became lowered, while the green emission became stronger in parallel (Fig. 3d, e; Supplementary Video 1). It is also evident that the Chl fluorescence yields in the background gradually decreased, but the green emission in the background stayed almost unchanged (Fig. 3e).
The different kinetics in these regions can be explained by the different yields of 3 Chl* generated by The question is whether the coinciding rise in the green emission is related to the 3 Car* de-excitation or not. It is generally considered that carotenoids are non-emissive; however, a weak fluorescence was reported previously 42 . Especially when Cars contain more than around eight 205 conjugated double bonds, they exhibit emission from the second electronically excited singlet state (S2) 43 . The spectral deconvolution of LHCII indicates that the S2 energies of lutein at L1 and L2 sites and violaxanthin are 20250, 20050, and 20200 cm -1 , respectively 44 . The excited triplet state (T1) energy of β-carotene is estimated to be 8000-8900 cm -1 45,46 . Given the T1 energies of Chl a and b (10700 and 11200 cm -1 , respectively) 47 , the upconverted energies through T-T annihilation are similar 210 to the S2 energies of carotenoids (18700-20100 cm -1 , Fig. 3f). Moreover, it has been shown that T-T annihilation processes can generate photoluminescence 48 . It should also be noted that S-T annihilation between the lowest excited singlet state (S1) of Chl and T1 Car can also generate a higher excited triplet state (Tn) of Car, which has a higher energy level than that of S2 Car. The yield of intersystem crossing of Car is typically low; however, there is a mixing between excited states of Chl and Car in 215 the plant LHCII 49,50 which can alter the strength of spin-orbit coupling and emission intensity.
Although further investigation is certainly needed for understanding the relationship between the green emission dynamics and S-T and/or T-T annihilation processes, these results demonstrate that another radiative de-excitation pathway exists in isolated LHCII under HL conditions. 220

Aggregated LHCIIs in liposome environments show unique optical properties
To further investigate changes in optical properties of LHCII, we next used LHCII incorporated in liposomes, which provides the opportunity to study fluorescence quenching mechanisms under more native conditions 51,52 . We prepared LHCII-containing liposomes (proteoliposomes, or PLs) by detergent-mediated reconstitution (see Methods for details). We 225 removed by-products (e.g. smaller or larger PLs, empty liposomes, etc.) by sucrose gradient centrifugation and used only the PLs from the most abundant single band in the gradient ( Supplementary Fig. 1). We used the same amount of LHCII (100 µg Chl) incorporated in either 1.5 or 0.5 mg of liposomes to generate LHCII-PLs with low or high densities of LHCII per liposome, respectively, whose differences were visible by sucrose gradient (Supplementary Fig. 1d, e). 230 We examined Chl fluorescence properties of these two different LHCII-PLs in the same way as above. The fluorescence yields observed in these two PLs showed about half of the yield observed in the solubilized LHCII (Fig. 4a, b). The high-density PL showed a slightly lower yield than the low-density PL. Upon AL illumination, similar Chl fluorescence quenching was observed in both PLs, indicating that the light-induced fluorescence quenching of LHCII still occurs in lipid 235 membrane conditions. Interestingly, the red-shifted fluorescence emission shoulder measured at 77 K of the low-density PL was actually stronger than that of the solubilized LHCII (Fig. 4c), which was not observed in the high-density PL (Fig. 4d) or other LHCII samples (Figs. 1c, 2d). This could indicate a weakly aggregated state in liposome conditions. Similar to the other LHCII conditions, the fluorescence yield of the shoulder became smaller under AL illumination, indicating that the intensity 9 of total emission actually becomes smaller by the light-induced quenching. The fluorescence spectral shape was similar between the two PLs, although there was a slight shift of the peak between the two (Fig. 4e, f), which suggests different levels of aggregation occurring between the two PLs.
As shown above, Chl fluorescence yields observed in the low-and high-density LHCII-PLs were only slightly different from each other at room temperature (Fig. 4a, b). However, their 245 relative fluorescence yields observed at 77 K showed a distinct difference-the yield for the highdensity PL was about 80% lower than that of the low-density PL (Fig. 4c, d). This discrepancy is likely influenced by different levels of light scattering caused by different levels of aggregation between the two PLs. Another critical difference was the optical setup between modulation fluorometry and spectroscopy at 77 K. The signals observed by using a modulation fluorometer at 250 room temperature were directly collected from the sample cuvette through the light guide to the photodiode unit. On the other hand, the fluorescence measurement at 77 K was done using a spectrophotometer, in which the cryogenic sample dewar was located distantly from the photomultiplier tube with a 2-nm slit. Because of these differences in optics and the sample-detector distance, the signal detection appeared to be influenced by different levels of light scattering in the investigation is required to address these issues.
It should be noted that the absorbance spectra in this study were measured by using an integrating sphere, in which the sample was located in front of the entrance port. When the amount of backscatter (reflectance) is negligible, the total transmitted light collected within the sphere will provide true absorption spectra. However, the true absorption spectrum of dialyzed LHCII was not 265 similar to that of the solubilized LHCII (Figs. 1e, 2f), whereas that of the LHCII-PLs was similar to that of solubilized LHCII (Fig. 4g, h). This difference may indicate that the contributions of the sieve effect caused by LHCII aggregation are minimized by being situated in the lipid environments of PLs.
Thus, it is very likely that the aggregated LHCII in vitro and LHCII in lipid environments (e.g. PLs or thylakoid membranes) have different optical properties, which also affect their fluorescence 270 properties.

Variable optical properties of LHCII impact observed quenching mechanisms
In this study, we investigated fluorescence and absorption properties of LHCII in different physical conditions based on equal Chl concentrations. We conducted this study because 275 experimental light illumination (e.g. AL or laser) caused changes in fluorescence lifetime of LHCII samples in vitro as initially observed by Jennings et al. 53 . For example, when we used LHCII-PLs for time-correlated single-photon counting measurements, the fluorescence lifetime became shorter from ~1.7 to ~1.5 ns during the three consecutive measurements (a few seconds for each measurement).
When we illuminated the LHCII-PLs with AL at ~850 µmol photons m -2 s -1 for 10 min, the lifetime 280 became distinctly shorter to ~0.6 ns. This phenomenon is quite problematic because light-induced quenching will be induced differently depending on the amount of light exposure (e.g. Fig. 2b), which would be difficult to standardize among different experimental setups in different labs. Therefore, it will be essential to specify the amount of light exposure (i.e., wavelength, intensity, and duration) when investigating the level of quenching, especially with in vitro samples. 285 We also revisited light scattering and the sieve effect induced upon LHCII aggregation.   Table 1 summarizes possible mechanisms that decrease Chl fluorescence in different LHCII conditions. Although we cannot simply apply in vitro results to in vivo situations, we would like to consider the possible situations based on this study. It is evident that the solubilized LHCII shows the highest fluorescence yield among other physical environments examined here. Thus, the extent of quenching observed in the solubilized LHCII is smaller than that of the other LHCII samples 305 in vitro. Regardless of environmental light conditions, fluorescence intensities become lower upon aggregation of LHCII due to the decreased level of absorption caused by light scattering and the sieve effect. Although it appears to be fluorescence quenching, this is not related to Chl de-excitation mechanisms but simply results from altered physical conditions of LHCII. Under HL, quenching mechanisms are activated through singlet and triplet EET to other pigments (e.g. lutein). Singlet EET 310 has a faster rate constant than that of triplet EET, so the contribution of the quenching via singlet EET will be initially high. On the other hand, under a prolonged HL exposure (as frequently used in modulation fluorometry), the contribution of the quenching of 3 Car* generated by triplet EET increases. Moreover, we found an increase in the green emission upon Chl fluorescence quenching under HL, which might be related to S-T and/or T-T annihilation processes 43,48-50 and will need to be 315 taken into consideration for understanding the total energy budget.
As previously shown 51,52 , LHCII in liposome environments tends to form aggregates, which most likely represent native conditions in thylakoid membranes. The different concentration of LHCII per liposome still generates different levels of light scattering, which appears to affect fluorescence intensities depending on the measurement setups (Fig. 4). The light-induced quenching  56 . Nevertheless, it is more essential to consider collectively how different components have impact on overall quenching mechanisms rather than to focus exclusively on a specific mechanism, especially when we investigate the process of NPQ induction in vivo.
Furthermore, our study emphasizes that multiple processes that affect Chl fluorescence yields can 330 emerge as the level of protein-protein interactions and the amount of light exposure increase. To dissect how these multiple processes progressively arise under excess light conditions, it is necessary to take genetic approaches (such as the mutants lacking monomeric LHCII 55 and trimeric LHCII 57 ) and consider the spatial and temporal correlation between the different components that contribute to overall quenching mechanisms.

Isolation of LHCII.
We obtained spinach leaves from a local store a day before LHCII isolation, and leaves were kept moist with wet paper towels and incubated at 4 °C in the dark overnight. Isolation of LHCII was done as described previously 2 . Deveined leaves were homogenized in 25 mM Tricine-  Modulation fluorometry. Chl concentration of each type of LHCII sample was pre-determined by extracting Chl in 80% acetone as described previously 58 . The Chl concentration of each sample was adjusted to 5.0 µg Chl/mL in 25 mM HEPES-NaOH (pH 7.8) with 0.4 M sucrose. Chl fluorescence kinetics was measured using Dual-PAM-100 (Walz). The intensity of ML was set to 7 (~1 µmol photons m -2 s -1 ). For Fig. 1b, the intensity setting 2 (~0.5 µmol photons m -2 s -1 ) and 12 (~5 µmol 370 photons m -2 s -1 ) were also used. The intensity of AL (red light) was set to 15 (~1200 µmol photons m -2 s -1 ) with Emitter DUAL-E. For Fig. 2b, the intensity setting 5 (~100 µmol photons m -2 s -1 ) and 10 (~400 µmol photons m -2 s -1 ) were also used. Detector DUAL-DR was used to monitor Chl fluorescence emission (> 700 nm). The sample was continuously stirred using a magnetic stir bar. Emission wavelength measured was from 500 to 800 nm with a 2-nm slit size. Fluorescence emission for each sample was recorded consecutively three times to obtain average spectra.

385
Absorption spectroscopy. Chl concentration of each type of sample was pre-determined by using the extracted Chl in 80% acetone as described previously 58 . For measurements using an integrating sphere, the Chl concentration of each sample was adjusted to 50 µg Chl/mL in 25 mM HEPES-NaOH (pH 7.8) with 0.4 M sucrose. Total transmittance from 350 to 750 nm (1 nm step size) was recorded using a quartz cuvette with 1-mm pathlength located in front of the entrance port of the integrating 390 sphere installed in a LAMBDA 950 UV/Vis spectrophotometer (PerkinElmer). A blank was recorded in each measurement using 25 mM HEPES-NaOH (pH 7.8) with 0.4 M sucrose.
Native protein gel electrophoresis. Isolated thylakoid membranes were solubilized with 2% (w/v) α-DDM for 30 min on ice. Insoluble fractions were removed by centrifugation at 21,000 × g for 5 395 min at 4 °C. The solubilized fraction was subjected to clear native polyacrylamide gel electrophoresis as described previously 59 . The sample was applied to a 5-13% (w/v) acrylamide gradient gel with 25 mM imidazole-HCl (pH 7.0) as anode buffer and 50 mM Tricine-HCl (pH 7.0), 7.5 mM imidazole, 0.02% α-DDM, and 0.05% deoxycholate as cathode buffer. Fluorescence images of the gel were acquired using a Typhoon 9400 imager (GE Healthcare) using excitation at 488 nm. The signals from the green region and Chl fluorescence were measured using a 520 ± 20 nm bandpass filter and a 670 ± 30 nm bandpass filter, respectively.
Confocal laser scanning microscopy. The small piece of the native protein gel containing aggregated LHCII was observed by using a Zeiss LSM510 confocal laser scanning microscope with 405 a Plan Apochromat 63×/1.4 NA oil objective lens. The excitation was at 488 nm, and a 505-550 nm bandpass filter and a 650 nm longpass filter were used to observe the signals from the green region and Chl fluorescence, respectively. After an initial scan, a square region within the gel was selected and continuously scanned for 1 min (Fig. 3b). After that, the original image area was restored and scanned again to see the difference in the signal intensity between the green region and Chl  M.I. wrote the paper. All authors discussed the results and commented on the manuscript.  concentration (CMC). The aggregated LHCII was gradually formed by putting the solubilized LHCII in the buffer without α-DDM, whose final concentration became 0.0004% (w/v), which is below the CMC. The resolubilized LHCII was formed from the aggregated LHCII by adding 0.03% (w/v) α-DDM to the buffer at the point indicated with an arrow. The dialyzed LHCII was prepared by removing α-DDM completely from the solubilized LHCII by dialysis (see Methods for details and 635 Supplementary Fig. 1b). Chl concentration of each LHCII sample was adjusted to 5 µg Chl/mL. The intensity of ML was ~1 µmol photons m -2 s -1 . b, Chl fluorescence yields observed during aggregation measured using different intensities of ML. The number indicates the intensity in µmol photons m -2 s -1 . The kinetics are normalized at the highest yield of each. The inset shows the kinetics without normalization. c,d, Representative Chl fluorescence spectra measured at 77 K normalized at the 640 fluorescence peak (530 nm) of Alexa Fluor 430 (2 nM) added to the sample as an internal standard (c) or the main Qy peak (d). Excitation wavelength was 440 nm. The insets show the difference spectra from the spectra of the solubilized LHCII. Line colors are the same as in (a). e, Absorbance spectra measured using the solubilized LHCII and the dialyzed LHCII, adjusted to equal Chl concentration quantified by extraction in acetone. f, Photograph showing solubilized LHCII, dialyzed 645 LHCII, and the same dialyzed LHCII resolubilized with 0.03% (w/v) α-DDM. Chl concentration of each sample was adjusted to 5 µg Chl/mL.  The preparation of solubilized LHCII, aggregated LHCII, resolubilized LHCII, and dialyzed LHCII was done as described in Fig. 1a. Chl concentration of each LHCII sample was adjusted to 5 µg Chl/mL quantified by extraction in acetone. The light intensity of ML was ~1 µmol photons m -2 s -1 . An arrow in (a) indicates the time point when 0.03% (w/v) α-DDM was added to the aggregated LHCII for resolubilization. Open and closed arrowheads are the points when AL was turned on and 655 off, respectively. The number indicated in sample legend is the light intensity of AL in µmol photons m -2 s -1 . d,e, Representative Chl fluorescence spectra measured at 77 K normalized at the fluorescence peak (530 nm) of Alexa Fluor 430 (2 nM) added to the sample as an internal standard (d) or the main Qy peak (e). Excitation wavelength was 440 nm. The insets show the difference spectra from the spectra of the solubilized LHCII (Dark). f, Absorbance spectra measured using solubilized LHCII and 660 dialyzed LHCII before (Dark) and after AL illumination at 1200 µmol photons m -2 s -1 for 10 min. Each sample was adjusted to equal Chl concentration quantified by extraction in acetone. Spectra are shown without normalization. It should be noted that T1 Car can also induce S-T annihilation which quenches the red emission from S1 Chl (not shown in the diagram for simplicity).