Fluorescence lifetime snapshots reveal two rapidly reversible mechanisms of photoprotection in live cells of Chlamydomonas reinhardtii

Abstract

Photosynthetic organisms avoid photodamage to photosystem II (PSII) in variable light conditions via a suite of photoprotective mechanisms called nonphotochemical quenching (NPQ), in which excess absorbed light is dissipated harmlessly. To quantify the contributions of different quenching mechanisms to NPQ, we have devised a technique to measure the changes in chlorophyll fluorescence lifetime as photosynthetic organisms adapt to varying light conditions. We applied this technique to measure the fluorescence lifetimes responsible for the predominant, rapidly reversible component of NPQ, qE, in living cells of Chlamydomonas reinhardtii. Application of high light to dark-adapted cells of C. reinhardtii led to an increase in the amplitudes of 65 ps and 305 ps chlorophyll fluorescence lifetime components that was reversed after the high light was turned off. Removal of the pH gradient across the thylakoid membrane linked the changes in the amplitudes of the two components to qE quenching. The rise times of the amplitudes of the two components were significantly different, suggesting that the changes are due to two different qE mechanisms. We tentatively suggest that the changes in the 65 ps component are due to charge-transfer quenching in the minor light-harvesting complexes and that the changes in the 305 ps component are due to aggregated light-harvesting complex II trimers that have detached from PSII. We anticipate that this technique will be useful for resolving the various mechanisms of NPQ and for quantifying the timescales associated with these mechanisms.

The pigment-protein complexes responsible for light harvesting in photosynthesis are highly dynamic and are able to alter their function in response to their environment (1, 2). In very low light, photosynthetic organisms maximize energy transfer from the chlorophylls in the light-harvesting antenna to those in the special pairs of the reaction centers of photosystems I and II (PSI and PSII), where charge separation occurs (3, 4). However, the capacity for photosynthesis in plants saturates at low light intensities (5). Light that is absorbed by chlorophylls in the light-harvesting antennae but that cannot be used for charge separation presents a significant hazard to PSIIs of oxygen-evolving organisms, because it can lead to the formation of high-energy species that can damage proteins and inhibit photosynthesis (5).

To protect itself from photodamage and prevent the loss of useful photosynthesis during its repair (6), PSII performs a collection of regulatory mechanisms, called nonphotochemical quenching (NPQ) (1, 2), in which excess absorbed light is dissipated harmlessly as heat. In high light, before an NPQ quenching site has turned on, a high population of PSIIs contains closed reaction centers. This closure occurs because electron transfer in the reaction center from the plastoquinone Q_A to Q_B is slow relative to light absorption, transfer, and trapping in PSII (3). PSIIs with closed reaction centers have a fluorescence lifetime greater than 1 ns (7–10) and are susceptible to damage. NPQ mechanisms turn on in response to a feedback signal triggered by the high light conditions over a timescale of seconds to tens of minutes. Once an NPQ quenching site has turned on in PSII, the lifetime of the excitation decreases well below 1 ns (7–10), and PSII is protected. Each mechanism of NPQ has a unique timescale for induction and for the lifetime of PSII once the NPQ quenching site associated with that mechanism has turned on. Measurements of NPQ as photosynthetic organisms adapt* to high light are typically done using pulsed amplitude modulated (PAM) chlorophyll fluorescence (11), which is a measurement of the fluorescence yield and thus does not distinguish between different mechanisms of NPQ. Transient absorption spectroscopy (12) and time-resolved fluorescence (7–10) have revealed changes in the quenching of chlorophyll excitation, but only by measurement before and after light adaptation. Picosecond-resolved spectroscopic measurements or snapshots of the photosynthetic organism during light adaptation would distinguish between populations of PSIIs undergoing different NPQ quenching processes.

The major, rapidly reversible component of NPQ is called qE (1, 2). It is triggered by a pH gradient across the thylakoid membrane. While qE quenching sites are thought to occur in the light-harvesting complexes of PSII in Arabidopsis thaliana, which mechanisms are involved in qE quenching remains highly controversial (2). To measure both the lifetimes and the amplitudes associated with different qE mechanisms, we developed an apparatus that allowed us to directly measure the onset and relaxation of the qE-associated changes of the excited state lifetime of PSII in live cells of the green alga Chlamydomonas reinhardtii.

Results and Discussion

PAM Fluorescence Trace of C. reinhardtii.

Fig. 1 shows a PAM fluorescence trace for wild-type C. reinhardtii grown photoautotrophically in high light (400 μmol photons m^-2 s^-1) until the late log phase of growth, and dark adapted for 20 min before measurement. Actinic light was applied at T = 0 s, where T is the time axis along which light adaptation occurs. During the first 0.3 s of actinic light illumination (Inset), the fluorescence yield increased as the plastoquinone pool became fully reduced, and PSII reaction centers became saturated (13). The induction of NPQ quenching pathways caused the fluorescence yield to rapidly decrease for 0.3 s < T < 2.5 s, decrease much less for 2.5 s < T < 10 s, and plateau by T = 20 s at a yield slightly lower than the yield before actinic light was applied. The actinic light was turned off after T = 20 s, and the fluorescence yield dropped below its value before actinic light was turned on. This decrease in fluorescence yield occurs because oxidation of the plastoquinone pool opened PSII reaction centers while NPQ quenching sites were still turned on. NPQ turned off rapidly as the algae adapted to the darkness over the next 60 s, suggesting that the NPQ observed here is qE quenching.

Fig. 1.

PAM fluorescence trace of wild-type C. reinhardtii. The actinic light intensity during the light induction period between T = 0 s and T = 20 s was 1,000 μmol photons m^-2 s^-1. Inset shows the fluorescence yield during the first 3 s after the actinic light is applied. The time resolution was 0.11 s.

The PAM measurement allowed us to qualitatively describe the dynamics as algae adapted to high light. However, it is possible that different qE processes contributed to the changes in fluorescence yield in this organism and that the amplitudes and fluorescence lifetimes of these processes were averaged out. To determine the lifetimes and amplitudes of qE quenching processes, we measured fluorescence decays of C. reinhardtii at different points on the T axis as the algae induced qE in high light for 20 s and as the algae turned off qE in an ensuing 60 s of darkness.

Measurement of Time-Resolved Fluorescence Decays During Light Adaptation.

NPQ is typically measured at different time points along a PAM trace by saturating pulses of light that close all PSII reaction centers (11). We constructed an apparatus whereby a similar strategy could be applied to measuring fluorescence decays as algae adapted to high light. The apparatus consists of a conventional single photon counting (SPC) setup with the addition of an actinic light source and three shutters in front of the excitation, actinic, and detection beams. The apparatus was built such that actinic light could be applied to the sample, with short periods in which the sample would interact with the laser to measure the time-resolved fluorescence while the PSII reaction centers remained saturated (Fig. 2A). To record an adequate number of fluorescence counts for the shortest increments in the adaptation trajectory, the experiment was also performed using the pulsed laser as both an actinic and a measuring light source (Fig. 2B). The repetition rate of the laser (75.7 MHz) and the pulse energy at the sample (5 pJ) were chosen to keep PSII reaction centers closed and maximize fluorescence collection while avoiding singlet-singlet and singlet-triplet annihilation conditions (SI Text). If singlet-triplet annihilation was a significant factor, the results from the actinic light treatment and the laser light treatment would be different. In fact, measurements on the same sample using the two strategies gave very similar results, which validated the use of the laser as both the actinic light source and the measuring light (Fig. S1, Fig. S2).

Fig. 2.

Schematic of measurement. (A) In the actinic light measurement, the sample remained in the dark until time T = 0 s, when the actinic light treatment began. Measurement of the fluorescence lifetime occurred periodically by closing the shutter to the actinic light beam and opening those in front of the pulsed laser and the detector (thin rectangles). The actinic light shutter was closed at T = 20 s and a measurement was made a time T_relax later. (B) In the laser light treatment, laser light was applied for 0 s < T < 20 s. Fluorescence acquisition periods were ΔT = 0.08 s long and spaced 0.2 s apart. The first F(t,T) was collected at T = 0.1 s to avoid loss of counts for the first measurement due to the opening time for the shutter. A measurement was made at a time T_relax as in the actinic light measurement. A representative fluorescence lifetime trace corresponding to a collection period at the beginning of the 20 s induction time is shown. The data were smoothed with a moving average filter with a span of 25 bins for visual clarity.

The apparatus measures fluorescence decays F(t,T) at different points along the T axis, where t corresponds to the arrival time of fluorescence photons after excitation of the sample with a laser pulse. The measurement of F(t,T) using the actinic light source began, at T = 0, by opening the shutter in front of the actinic light beam. To make a fluorescence lifetime measurement during qE induction, when 0 s < T < 20 s, the actinic light shutter was closed, shutters in front of a pulsed laser and a detector were opened, and a trigger was sent to the SPC board to initiate collection and binning of fluorescence counts for a time ΔT. After ΔT, to resume light adaptation, the laser and detector shutters were closed, and the actinic light shutter was reopened. This strategy prevents actinic light from overloading the detector and allows for the collection of fluorescence decays during the qE induction period. To measure fluorescence decays during the qE relaxation period, the actinic light shutter was closed at T = 20 s. At a time T_relax later the laser and detector shutters were opened, and the SPC board was triggered to collect and bin counts for ΔT, after which the laser and detector shutters were closed.

Ideally, the collection time, ΔT, would be set such that F(t,T + ΔT) ≈ F(t,T). As seen in Fig. 1, Inset, the fluorescence yield changes every 0.11 s during the first 3 s of the actinic light treatment, so ΔT should be less than that time. However, approximately 1,000 fluorescence counts in the maximum bin must be obtained during ΔT to allow accurate fitting of the fluorescence decays. A compromise between these two considerations and the electronic limitations of the photon counting board allowed us to collect photons for ΔT = 0.08 s every 0.2 s on the T axis, provided the measurement F(t,T) was repeated on more than 10 different aliquots from an algal culture to increase the number of counts in the maximum bin to approximately 1,000. This ΔT was too small to use the actinic light source as in the strategy described above because the shutters had a open/close time of approximately 40 ms. Even if faster shutters were used, the sample would be exposed to the laser for 40% of the light adaptation period. For the purposes of measuring light induction of qE in C. reinhardtii, we used the laser light as both an actinic light source and as a measuring light.

At T_relax, the time for which the algae are in darkness after the high light treatment, we wanted to measure a fluorescence decay with the PSII reaction centers closed and without any qE induced by the laser during the measurement. However, as seen in Fig. 1, Inset, it takes approximately 0.3 s of exposure of dark-adapted cells to high light before the reaction centers are closed. qE quenching turns on immediately after saturation of the reaction centers. Because these changes may occur at different rates depending on T_relax, we averaged the fluorescence curves collected between 0.2 and 1.0 s within ΔT. Selection of this interval resulted in the observation of full reversal of changes in the fluorescence lifetime caused by the 20 s laser light treatment. Only one measurement at T_relax was done per sample to avoid the influence of the laser on any measurements at later T_relax times.

Fig. 3 shows the results of the experiment described in Fig. 2B. The qualitative dynamics of the fluorescence decays curves matched those seen in the PAM fluorescence trace in Fig. 1. The fluorescence decay time increased in the first 0.3 s due to the closing of all reaction centers in the excitation volume. The fluorescence lifetime decreased substantially in the next 2 s, decreased less from 2.5 to 10.1 s, and barely decreased from 10.1 to 19.5 s. Laser illumination was stopped after 20 s, and the lifetime returned to its initial value 60 s after the light treatment.

Fig. 3.

Normalized fluorescence decays from emission at 680 nm from the measurement described in Fig. 1B. The data were smoothed with a moving average filter with a span of 20 bins for visual clarity. The direction of the black arrow indicates increasing T. Blue represents early T and red represents late T within the timespan specified in each panel.

Amplitudes and Lifetimes from the Time-Resolved Fluorescence Decays.

To quantify the changes seen in Fig. 3 over T, fluorescence counts were summed over all collected wavelengths for each F(t,T), and the resulting fluorescence decays were normalized and globally fit using the method of least squares to the equation

[1]

where the ith component has an amplitude A_i(T), such that . Three lifetime components were required to give a reasonable fit, as judged by the χ² value being between 0.8 and 1.2 and uncorrelated residuals for each F(t,T). Each lifetime and its corresponding amplitude do not necessarily correspond to one physical process and its absorption cross-section in the thylakoid membrane. Rather, each lifetime and its amplitude are possibly the result of a sum of the amplitudes of several processes in the membrane with approximately the same lifetime. To calculate the uncertainty in the fitted parameters, we performed bootstrapping (SI Text) on the fits, which was repeated 500 times to obtain 68% confidence intervals for each of the fitted parameters.

The lifetimes from the global fit were 64 (63, 65) ps, 305 ps (304, 320), and 1,000 (990, 1,074) ps. The 68% confidence intervals indicated in parenthesis were taken as an average over several time points. For simplicity, we will refer to these components as the 65 ps, 305 ps, and 1 ns components with amplitudes A_65 ps(T), A_305 ps(T), and A_1 ns(T), respectively.

Because variable fluorescence is associated only with PSII in the thylakoid membrane (3) we associated any changes in the amplitudes of these three components with respect to T with changes in excitation trapping in PSII. A_65 ps(T) and A_305 ps(T) are shown in Fig. 4 A and B, respectively. A_65 ps(T) + A_305 ps(T) and A_1 ns(T) are shown in Fig. 4C. The 68% confidence intervals of the amplitudes at selected points are indicated in this figure by error bars.

Fig. 4.

Changes in the amplitudes of the fluorescence lifetime components over T. (A) Amplitude of the 65 ps component [A_65 ps(T), green circles], (B) amplitude of the 305 ps component [A_30 5ps(T), violet circles], and (C) Amplitude of the 1 ns component [A_1 ns(T), blue circles] and the sum of the 65 ps and 305 ps components [A_65 ps(T) + A_305 ps(T), red circles]. The error bars indicate 1 standard deviation in the uncertainty of that parameter and are shown for T = 0.3 s, 1.9 s, 7.9 s, 17.7 s, and 50 s. Insets in A and B show the A_65 ps(T) and A_305 ps(T) for 0.3 s < T < 3 s. The lines in Insets are shown to indicate the changes in the amplitude.

A_65 ps(T) (Fig. 4A) decreased slightly for the first 0.3 s of light illumination (Inset), then rose from approximately 0.61 at T = 0.3 s to approximately 0.68 when T = 1 s and remained constant within error for the remainder of the light adaptation period. Based on this plot, no definite conclusions on the dynamics of this component could be made for 20 s < T < 80 s because of the uncertainty in the amplitude. A_305 ps(T) (Fig. 4B) also decreased for the first 0.3 s of illumination (Inset) to approximately 0.16 but subsequently increased for the remainder of the saturating light treatment, reaching approximately 0.25 at T = 20 s. Forty-five seconds after the light treatment ended, the amplitude decreased back to the value at T = 0.3 s and appeared to decrease to an even lower value by 60 s in darkness. Finally, A_1 ns(T) increased in the first 0.3 s of light illumination, decreased substantially from 0.3 s < T < 3 s, and plateaued by 20 s in actinic light (Fig. 4C). It increased over 60 s in the darkness back to A_1 ns(T = 0.3 s).

We ascribe the changes in amplitudes to changes in the quantities of PSIIs with such lifetimes. Coarse-grained models of energy transfer and trapping in PSII (14) as well as measurements of the fluorescence lifetime (7–10) suggest that PSIIs with a site that can trap excitation energy, such as a qE site and/or an open reaction center, have a lifetime of less than 700 ps. PSIIs with closed reaction centers and no access to an energy trapping site have a lifetime of 1 ns or greater. Therefore, we associate the decrease in A_65 ps(T) and A_305 ps(T) in the first approximately 0.3 s of illumination with a decrease in the number of PSIIs with access to an open reaction center as the plastoquinone pool becomes fully reduced. The subsequent increase in A_65 ps(T) and A_305 ps(T) with 0.3 s < T < 20 s is an increase in the number of PSIIs with closed reaction centers and access to a qE quenching site. The decrease in A_305 ps(T) and, to a lesser extent A_65 ps(T), from 20 s < T < 80 s we ascribe to the decrease in the number of PSIIs with closed reaction centers and access to a qE quenching site as qE sites turn off in the darkness. The changes in A_1 ns(T) are changes in the number of PSIIs with access to neither a qE site nor an open reaction center. Because, for T > 0.3 s, the changes in A_305 ps(T) are clearly reversible and those for the A_65 ps(T) may be reversible, we are observing one and possibly two lifetimes associated with qE switching on and off in live cells of C. reinhardtii.

To test if the increases in both A_65 ps(T) and A_305 ps(T) are linked to a transmembrane pH gradient and thus to qE, we measured time-resolved fluorescence decays after 150 s of an actinic light treatment and after adding 100 μM of the ionophore nigericin to the sample after the light treatment (Fig. S3). Nigericin removes the pH gradient across the thylakoid membrane and is typically used to determine whether changes in fluorescence lifetime are due to qE (7). The fluorescence decays from the two measurements were globally fit to three exponential decays, and the results are shown in Table 1. The amplitudes of both short components (in this fit, 70 and 330 ps) decrease upon the addition of nigericin, suggesting that the changes in both A_65 ps(T) and A_305 ps(T) in Fig. 4 are due to qE. We associate the remaining amplitude of the 70 ps component with PSI quenching (9, 15, 16), and we ascribe the remaining amplitude of the 330 ps component partially to PSI quenching and partly to detached, phosphorylated light-harvesting complex II (LHCII) trimers (17).

Table 1.

Amplitudes and lifetimes for wild-type cells adapted to high light (600 μmol photons m^-2 s^-1) for 150 s (light adapted) and treated with nigericin after the high light treatment (nigericin)

	Amplitudes
Condition	70 ps	330 ps	1.2 ns	(ps)
Light adapted	0.51	0.43	0.06	250
Nigericin	0.35	0.30	0.35	544

The fluorescence was collected at 680 nm and 68% confidence intervals of the fits as calculated by bootstrapping were within ± 0.01 of the amplitudes shown.

Although A_65 ps(T) was not unambiguously reversible (Fig. 4A), the fact that this component was linked to the pH gradient and that other measurements indicate that this component’s amplitude changes are reversible (Fig. S2A) lead us to believe that this component is reversible and attributable to qE.

We fit A_65 ps(T) and A_305 ps(T) from T = 0.3 s, when the plastoquinone pool was fully reduced, to the end of the light treatment at T = 20 s to a simple rise curve with the form

[2]

The results of the fitting are shown in Table 2. The 95% confidence intervals of the rise time, τ_rise, for the two components do not overlap. These results give evidence that the changes in A_65 ps(T) and A_305 ps(T) in Fig. 4 are due to two different qE triggering processes.

Table 2.

Fits for the increase in A_65 ps(T) and A_305 ps(T) for 0.3 s < T < 20 s (Fig. 4 A and B)

Component	τrise (s)	a	c
65 ps	0.68 (0.45, 1.47)	0.07 (0.05, 0.10)	0.60 (0.58, 0.62)
310 ps	3.18 (2.27, 5.26)	0.07 (0.06, 0.09)	0.18 (0.16, 0.20)

The data were fit to Eq. 2. The 95% confidence bounds are shown in parenthesis.

Interpretation and Prospects.

To interpret the origin of the two qE components observed, we turn to the current hypotheses for the molecular mechanisms underlying qE, which have been put forth primarily for plants. There is a consensus that the protonation of the PsbS protein as a result of a pH decrease in the thylakoid lumen is thought to be the trigger for the qE process and that the xanthophylls zeaxanthin and lutein are necessary for quenching in vivo (1, 2). However, hypotheses differ on the role of xanthophylls and the location of the qE quenching site. In a series of papers, Avenson et al. (18) and Ahn et al. (19) have proposed that charge-transfer quenching of chlorophyll excitations by xanthophylls in the minor complexes of PSII (CP24, CP26, and CP29) is an important contribution to qE in A. thaliana. In contrast to this model, Ruban et al. (2) propose that low lumen pH leads to the aggregation of LHCII, the major light-harvesting complex of PSII. The aggregated LHCII trimers undergo a conformational change (20) and quench chlorophyll excitations by energy transfer to the S1 state of lutein and subsequent relaxation (21). Holzwarth and coworkers have proposed that quenching occurs in both detached LHCII trimers and the minor complexes of PSII (9). C. reinhardtii lacks PsbS but Bonente et al. suggested that the LHCSR3 protein plays a similar triggering role (22). In addition, unlike PsbS, LHCSR3 binds pigments, and fluorescence lifetime measurements indicated that the protein could perform pH-dependent quenching (22).

Changes in fluorescence lifetimes in response to high light in C. reinhardtii have been reported by Holub et al. for large T values (23). However, there have been no measurements on the quenching mechanisms associated with qE in C. reinhardtii. There were, however, recent fluorescence lifetime imaging measurements (17) of a second component of NPQ, called state transitions, which involves the detachment of phosphorylated LHCII trimers from the PSII supercomplex and subsequent reattachment of some of the LHCII to PSI (24). Iwai et al. (17) observed a 250 ps component for detached, phosphorylated, and aggregated LHCIIs in C. reinhardtii and suggested that this lifetime could be the same for detached, aggregated LHCIIs due to qE.

Turning to the data in Fig. 4, Table 1, and Table 2, A_65 ps(T) increases much more rapidly under saturating light than does A_305 ps(T) after the reduction of the plastoquinone pool, reaching its maximum value at T ≈ 1.3 s. Kinetic modeling of the transient absorption data showing formation of a cation radical in thylakoids from A. thaliana suggested that the time for chlorophyll excitations to be quenched in the minor complexes is approximately 30 ps (25). A possible interpretation of the changes in A_65 ps(T) for T > 0.3 s is that they are due to changes in the number of PSII supercomplexes that can perform the minor complex/charge-transfer mechanism. Because of the slower rise time of A_305 ps(T) following reduction of the plastoquinone pool, we suggest that the changes in A_305 ps(T) for T > 0.3 s correspond to changes in the number of aggregated LHCIIs that are detached from PSII supercomplexes and that this lifetime may be related to the 250 ps lifetime reported by Iwai et al. The protonation of LHCSR3 may cause it to become a quencher along with the minor complexes as well as a trigger for LHCII aggregation.

These tentative interpretations of the two components can be tested by applying this technique to different mutants of C. reinhardtii (26–28). Our interpretation for the two short lifetimes may also be tested by numerical modeling of energy transfer and trapping in PSII and LHCII aggregates (29, 30). The rise times of A_65 ps(T) and A_305 ps(T) may be representative of the time for the pH to drop in the lumen, the time for a rearrangement to happen in the thylakoid membrane (31, 32), or some combination of the two.

More generally, the technique described here can easily be applied to studying other photosynthetic organisms. The addition of the fluorescence lifetime time axis t to measurements of the adaptation of photosynthetic organisms to varied light conditions should help enable the elucidation of the molecular mechanisms of NPQ.

Concluding Remarks

We have developed an apparatus that can measure time-resolved fluorescence snapshots of photosynthetic organisms as they adapt to varying light intensities. Our data on wild-type C. reinhardtii suggest that there may be two mechanisms involved in the rapidly reversible component of NPQ known as qE. We speculate that one mechanism involves quenching in detached LHCII complexes whereas a second occurs in the minor complexes and/or the LHCSR3 protein. These ideas can be tested by application of the method developed here to a wide variety of photosynthetic mutants both in C. reinhardtii and in other model organisms, by extension of the snapshot technique to time-resolved absorption spectroscopy, and by numerical modeling of energy transfer and trapping in PSII and the dynamics of the thylakoid membrane.

Materials and Methods

Growth and Sample Preparation of C. reinhardtii.

Cells of the 4A^- strain were grown in low light (40 μmol photons m^-2 s^-1) with a reduced carbon source [tris-acetate phosphate (TAP) medium] for 1 to 2 d until exponential growth was reached. Approximately 1 × 10⁵ cells from this culture were used to inoculate 50 mL of minimal [high salt (HS)] media. The HS culture was grown in high light (400–440 μmol photons m^-2 s^-1) in air, with ambient CO₂ as the only carbon source. Cultures that were in the late-logarithmic growth phase, between 3 and 5 × 10⁶ cells mL^-1, were used for experiments. Before performing fluorescence measurements on the samples, the cells were dark incubated for 30–50 min. The sample was prepared in a 1 cm pathlength cuvette. The cuvette contained algae diluted to a concentration of 0.5 OD at 410 nm with 500 μL of 0.1% agarose to suspend the algae and water to 1.5 mL.

PAM Fluorescence Measurements.

The maximum efficiency of PSII (F_v/F_m) was measured as described in Peers et al. (26) and was approximately 0.6 for all measurements. The measurement shown in Fig. 1 was done as described in the text.

Fluorescence Lifetime Apparatus.

Time-resolved fluorescence measurements were acquired using a time-correlated single photon counting (TCSPC) apparatus. A commercial mode-locked oscillator (Mira 900F; Coherent) pumped by a diode-pumped, frequency-doubled Nd:YVO4 laser (Verdi V-10; Coherent) generated approximately 150 fs pulses with a repetition rate of 76 MHz and was tuned to 820 nm with an FWHM of 12 nm. This output was frequency doubled to 410 nm using a 1-mm-thick beta barium borate crystal. The pulse energy at the sample was 10 pJ. Fluorescence emission was sent through a polarizer set at the magic angle. The fluorescence was then sent through a spectrograph (Newport; 77400-M) and detected with a microchannel plate photomultiplier tube (MC-PMT) (Becker-Hickl; PML-16C). The MC-PMT could discriminate the fluorescence photons into 16 channels, separated 12 nm apart from each other. Each channel contained the fluorescence decay for photons at that wavelength, ± 1 nm. The MC-PMT was controlled using the DCC-100 detector control (Becker-Hickl), with the gain set to 90%. The photons were collected into 3.6 ps wide bins using a Becker-Hickl SPC-630 counting board. The FWHM of the instrument response function was 150 ps.

Shutter Apparatus.

The sample can interact with the laser and/or a white actinic light source (Schott KL1500 LCD), both of whose access to the sample is gated by a controllable shutter. The path of fluorescence photons to the detector is also gated by a shutter. The shutters were constructed using a rotary solenoid (LEDEX; part no. 810-282-330) which was connected to a 12-V power supply (Circuit Specialists, Inc.; 3645A). To control the shutters, a 5-V square wave was sent from the computer through a data acquisition card (National Instruments; PCI-6229) to three relays (Potter & Brumfield; JWS-117-1). A 5-V square wave from the computer triggers the TCPSC board to start a measurement and collect photons (33).

Measurement of qE Induction and Relaxation.

As depicted in Fig. 2, the laser was used to induce changes in the sample. The laser illuminated the sample for 20 s, and time-resolved fluorescence decays were collected for 80 ms every 200 ms. After 20 s of laser exposure, the laser shutter was closed for a relaxation time. The measurement sequence was repeated for 12 aliquots from the same culture, and the fluorescence decay at each collection time was averaged over all aliquots. After the relaxation time ended, the laser shutter was opened for 1.6 s, during which eight fluorescence decays were collected every 200 ms. The relaxation times used were 1, 2, 5, 7, 10, 14, 18, 22, 26, 30, 45, and 60 s. Of the eight fluorescence decays collected in the 1.6 s of the relaxation measurement, the second to the fifth fluorescence traces were summed to give the relaxation trace. Because more counts were collected in the measurement of relaxation than in measurement of the induction, each relaxation lifetime came from the average of two aliquots. Fluorescence decays were generated by adding counts at each collection time over all aliquots and over channels 6 (644 nm) to 12 (716 nm). Time resolution was reduced to 12.2 ps per bin by summing counts every four bins of the raw data. Measurement of changes in the lifetime due to a white actinic light source is described in SI Text. The measurement of the fluorescence lifetime of light-adapted and nigericin-treated algae is described in Fig. S3. The data were analyzed as described in the main text. A more detailed description of the data analysis is in SI Text.