Kinetic and Spectral Resolution of Multiple Nonphotochemical Quenching Components in Arabidopsis Leaves^,

Abstract

Using novel specially designed instrumentation, fluorescence emission spectra were recorded from Arabidopsis (Arabidopsis thaliana) leaves during the induction period of dark to high-light adaptation in order to follow the spectral changes associated with the formation of nonphotochemical quenching. In addition to an overall decrease of photosystem II fluorescence (quenching) across the entire spectrum, high light induced two specific relative changes in the spectra: (1) a decrease of the main emission band at 682 nm relative to the far-red (750–760 nm) part of the spectrum (Δ F₆₈₂); and (2) an increase at 720 to 730 nm (Δ F₇₂₀) relative to 750 to 760 nm. The kinetics of the two relative spectral changes and their dependence on various mutants revealed that they do not originate from the same process but rather from at least two independent processes. The Δ F₇₂₀ change is specifically associated with the rapidly reversible energy-dependent quenching. Comparison of the wild-type Arabidopsis with mutants unable to produce or overexpressing the PsbS subunit of photosystem II showed that PsbS was a necessary component for Δ F₇₂₀. The spectral change Δ F₆₈₂ is induced both by energy-dependent quenching and by PsbS-independent mechanism(s). A third novel quenching process, independent from both PsbS and zeaxanthin, is activated by a high turnover rate of photosystem II. Its induction and relaxation occur on a time scale of a few minutes. Analysis of the spectral inhomogeneity of nonphotochemical quenching allows extraction of mechanistically valuable information from the fluorescence induction kinetics when registered in a spectrally resolved fashion.

One of the most important photoprotective mechanisms against high-light (HL) stress in photosynthetic organisms is the nonphotochemical quenching (NPQ) of excitation energy, which is mostly due to thermal deactivation of pigment excited states in the antenna of PSII. There exist a number of literature reviews on the subject (Demmig-Adams and Adams, 1992; Horton et al., 1996; Horton and Ruban, 1999, 2005; Niyogi, 1999, 2000; Müller et al., 2001; Golan et al., 2004; Krause and Jahns, 2004). Chlorophyll (Chl) fluorescence, and in particular pulse amplitude-modulated (PAM) fluorometry as introduced by Schreiber et al. (1986), has become by far the dominant technique to measure NPQ in leaves, chloroplasts, and intact microorganisms (Krause and Weis, 1991; Govindjee, 1995; Maxwell and Johnson, 2000; Krause and Jahns, 2003; Schreiber, 2004), more recently often combined with specific NPQ mutant studies (Golan et al., 2004; Kalituho et al., 2006, 2007; Dall'Osto et al., 2007). In this technique, periodic saturating light pulses are applied, superimposed on the continuous actinic irradiation applied to induce NPQ, in order to transiently close the PSII reaction centers (RCs). Since the photochemistry contribution (photochemical quenching) is thus brought to zero, the method allows us to follow the dynamics of the NPQ development and relaxation by fluorescence in a relatively simple manner (Krause and Jahns, 2003, 2004).

Mostly based on its relaxation kinetics, NPQ has been divided technically into the three kinetic components qE, qT, and qI, for the rapid, middle, and slow phases of relaxation (Horton and Hague, 1988), initially attributed to energy-dependent quenching, state transitions, and photoinhibitory quenching (Quick and Stitt, 1989). The rapidly forming and reversible part of NPQ, qE, is the most thoroughly studied. It is well established that this type of quenching is a finely regulated process in which the main governing factors are the proton gradient across the chloroplast thylakoid membrane, Δ pH (Wraight and Crofts, 1970; Briantais et al., 1979), the xanthophyll cycle (i.e. conversion of violaxanthin to antheraxanthin and zeaxanthin [Zx]; Demmig et al., 1987; Demmig-Adams, 1990; Demmig-Adams and Adams, 1992), and the action of the PsbS protein (Funk et al., 1995; Li et al., 2000, 2004; Niyogi et al., 2005). The actual molecular mechanism is still unknown, although there is no shortage of hypotheses and proposed quencher candidates: energy transfer from Chl to Zx in the major light-harvesting complex (LHCII; Frank et al., 2000); electron transfer from a carotenoid to Chl forming a Zx-Chl or lutein-Chl charge-transfer state (Holt et al., 2005; Avenson et al., 2009); direct or indirect quenching by the PsbS protein (Li et al., 2000; Niyogi et al., 2005); energy transfer from Chl to lutein in LHCII (Horton et al., 1991; Ruban et al., 2007) linked to the aggregation of or a conformational change in LHCII; and last but not least, a far-red (FR) light-emitting quenched Chl-Chl charge-transfer state formed by the aggregation of LHCII (Miloslavina et al., 2008). Quenching in the PSII RC has also been proposed (Weis and Berry, 1987; Finazzi et al., 2004; Huner et al., 2005; Ivanov et al., 2008) as an additional type of Zx-independent quenching. Alternatively, it has been suggested that quenching by lutein can complement the Zx-dependent quenching (Niyogi et al., 2001; Li et al. 2009). Johnson et al. (2009) have recently given support to the notion that both Zx-dependent and Zx-independent quenching originate from the same PsbS-dependent mechanism, which is modulated by Zx (Crouchman et al., 2006).

While the rapidly relaxing phase qE is now well characterized in its dependence on the various factors, the much slower qT and qI phases are still controversial, and each of them may have contributions from more than one mechanism. The qI component has been traditionally attributed to photoinhibition of PSII (Somersalo and Krause, 1988), associated with coordinated degradation and repair of the photosystem (Powles and Björkman, 1982; Kyle, 1987; Krause, 1988; Aro et al., 1993; Long et al., 1994; Murata et al., 2007). Lately, though, it is more widely accepted that under most conditions the photoinhibition is low and qI, like qE, is a result of thermal deactivation of excited states. Different hypotheses have been put forward to account for its seeming irreversibility: persistent transmembrane Δ pH (Gilmore and Yamamoto, 1992), stable protonation of proteins (Horton et al., 1994), accumulation of inactive PSII reaction centers (Briantais et al., 1992; Schansker and van Rensen, 1999), or stable binding of Zx to CP29 (Färber et al., 1997). The connection of the qT phase with state transitions has been doubted as well, and in fact it is now thought that the fraction of energy redistributed from PSI to PSII under high-light conditions is negligible (Walters and Horton, 1991, 1993) and that the qT must have a different origin or that it has erroneously been ascribed as NPQ (Schansker et al., 2006).

Along with the large amount of contradictory evidence on the nature and location of the NPQ quenching site(s), the question of whether the light-induced reversible NPQ represents one single mechanism of deexcitation located in a single site brought about by the combined action of PsbS and Zx (Johnson et al., 2009) or whether it comprises several parallel and largely independent mechanisms acting on different parts of the PSII antenna has not been finally answered. One way to answer this question might be to carefully examine the spectral properties of NPQ-related fluorescence changes. Quenching in different locations of the PSII antenna or with different mechanisms might give rise to a differential quenching in various parts of the PSII antenna that might affect the PSII fluorescence spectra in different ways. This appears possible, since the various pigment-protein complexes of the photosynthetic apparatus have slightly different absorption and emission spectra (Holzwarth, 1991; Holzwarth and Roelofs, 1992). However, in the vast majority of modulated Chl fluorescence instrumentation, including the most widely used PAM fluorometer (Schreiber et al., 1986), the signal is integrated over a broad wavelength range, usually covering the whole range of 710 nm or greater. This integration over the long-wave part of the spectrum has several undesirable consequences and is associated with the unnecessary loss of available information. For example, the fluorescence of PSII peaks in the region of 680 to 685 nm, whereas beyond 700 nm, the PSII fluorescence intensity drops to less than 20% of its peak intensity. In contrast, the fluorescence of intact PSI complexes is dominant in the region above 710 nm (Haehnel et al., 1982; Karukstis and Sauer, 1983; Holzwarth et al., 1985; Holzwarth, 1986; Slavov et al., 2008). Thus, the widely used instrumentation measures the NPQ parameters in a region with reduced PSII contribution and relatively high PSI contribution to total fluorescence, despite the fact that NPQ is generally considered to be primarily a PSII phenomenon. Only in a few studies has the fluorescence in the red and the FR region been separated in order to evaluate the contribution of PSI and its influence on the NPQ parameters (Genty et al., 1990; Peterson et al., 2001). NPQ might also shift the fluorescence properties of the PSII antenna complexes or give rise to entirely new fluorescing components (Miloslavina et al., 2008). This would remain undetected if the NPQ fluorescence changes are not resolved in the spectral domain. It follows from these considerations that a great deal of insight into the NPQ mechanisms and locations may be gained if the spectral dimension is added to the NPQ fluorescence characterization. Among the many advantages of such an approach, one would then be able to distinguish whether NPQ simply leads to a uniform decrease of PSII fluorescence across the emission range or whether this decrease is nonuniform, localized in specific pigment protein complexes, and/or whether new fluorescing species are actually being produced in the NPQ process.

The HL-induced NPQ effects on the leaf fluorescence spectra have often been studied also at low temperature, where the differentiation between pigment sites is better (Krause et al., 1983; Demmig and Björkman, 1987; Ruban and Horton, 1994). However, the possibility to resolve the kinetics of NPQ development and relaxation is largely lost when performing the measurements at low temperatures. The 77 K spectra of leaves and thylakoid membranes are characterized by three main peaks, F₆₈₅, F₆₉₅, and F₇₃₀, believed to originate predominantly from Chl a in CP47 of PSII, a specific Chl in CP43 of PSII, and PSI, respectively (Satoh and Butler, 1978; van Dorssen et al., 1987; Andrizhiyevskaya et al., 2005; Komura et al., 2007). Fluorescence from the major LHCII peaks at 680 nm (Rijgersberg and Amesz, 1978) and from the PSII reaction center Chls at 683 nm (Roelofs et al., 1993; Andrizhiyevskaya et al., 2005). Low-temperature studies on the effects of HL irradiation are confined to the changes in the FR-to-red fluorescence ratio, which are the result of the quenching of PSII fluorescence or energy redistribution between the photosystems (state transitions). Ruban and Horton (1994) have shown that photochemical quenching in Guzmania is maximal at 688 nm, whereas nonphotochemical processes quench preferentially at 683 and 698 nm.

In this study, we undertook a detailed investigation of the NPQ-associated spectral changes in the fluorescence spectra of Arabidopsis (Arabidopsis thaliana) measured at room temperature (RT) and at 77 K. It follows from the above discussion that deeper insight into the mechanisms of NPQ processes may be gained by combining the kinetic and the spectral information of the fluorescence changes occurring in NPQ. For this purpose, we developed a multiwavelength spectrometer with parallel detection, allowing us to follow the entire time-dependent fluorescence spectra of leaves during the induction and relaxation phases of NPQ with high sensitivity.

Specific questions to be addressed in this study are the following. Are there more than one NPQ processes and NPQ sites? Are these processes occurring in a linked fashion or are they independent? How do they depend on the various cofactors known to affect NPQ, in particular regarding the roles of PsbS and Zx? Using this novel approach of adding the spectral information to the NPQ fluorescence changes, we discovered specific spectral changes associated with different NPQ components. By comparing the effects measured on various NPQ mutants of Arabidopsis, it is possible to assign these NPQ components to specific quenching processes. The results provide evidence that the total NPQ is a combination of several parallel and largely independent processes, likely occurring at different locations in the photosynthetic apparatus.

RESULTS

77 K Fluorescence Spectra

Fluorescence emission spectra measured at 77 K from dark-adapted wild-type Arabidopsis leaves and from leaves irradiated for 30 min with 600 μ mol photons m⁻² s⁻¹ red light (620 nm) are shown in Figure 1. To facilitate detailed comparison of their shape, all spectra were normalized. Since NPQ is supposed to alter the properties of PSII rather than PSI, we wanted to use PSI fluorescence as a reference for normalization to better visualize the changes in PSII. According to the notion that at shorter wavelengths the emission originates from PSII and at longer wavelengths from PSI, we normalized the spectra at 760 nm, near the FR tail of the spectra, where the contribution from PSII is minimal and that from PSI is maximal. More precisely, the spectra were normalized to the averaged intensity between 755 and 765 nm in order to further reduce the already low noise in the data (the signal-to-noise ratio at 760 nm is about 300:1) and thus avoid inaccurate normalization. The 77 K fluorescence emission spectra of wild-type leaves show peaks at 683, 691, and 730 nm. The relative ratio between them has been reported to be strongly dependent on the leaf anatomy, particularly the thickness and the Chl content, which determine its optical properties, and also on the angle of excitation and detection (Weis, 1985). However, for leaves at similar physiological states at a fixed orientation, we found the spectral differences to be insignificant. Following HL treatment of the wild-type leaves, the spectra revealed characteristic and reproducible differences: a decrease of fluorescence in the PSII wavelength range (less than 700 nm) and an increase in the FR range (greater than 710 nm). The light-minus-dark difference spectra (Fig. 1B) show two negative peaks, at 683 and 691 nm, matching the peaks of the emission spectra, and a positive band appearing around 727 nm (i.e. somewhat blue shifted to the major PSI emission peak; 730 nm). In some cases, a negative difference band at 715 nm was also observed (only a shoulder in Fig. 1B).

Figure 1.

A, 77 K fluorescence emission spectra, normalized at 760 nm, of dark-adapted Arabidopsis wild-type leaves (solid line) and leaves preilluminated for 30 min with red light (photon flux density 600 μ mol m⁻² s⁻¹). The curves represent averages from four to six measurements on different leaves. The se is shown for selected wavelengths. B, Difference spectrum (light minus dark). rel.u., Relative units.

Whereas the negative difference bands are undoubtedly associated with quenching of the PSII antenna fluorescence (F₆₈₅ and F₆₉₅), the origin of the positive 730-nm band is less clear. In order to check whether its appearance corresponds to the kinetics of NPQ, leaves were rapidly frozen in the light at different times of HL irradiation and 77 K fluorescence spectra were recorded. Furthermore, spectra were taken after keeping preilluminated leaves for different times in darkness. The resulting time-dependent changes of the ratio F₇₃₀/F₇₆₀ are shown in Figure 2. The ratio rose within the first several minutes of irradiation toward a maximum value and declined back after a few minutes of darkness. This is a typical behavior for energy-dependent quenching (qE). However, because these measurements have to be carried out on different leaves, the level of accuracy is not high enough to quantitatively compare the kinetics of the 77 K fluorescence changes at different wavelengths.

Figure 2.

Dependence of the fluorescence ratio F₇₃₀/F₇₆₀ on the duration of irradiation with 600 μ mol m⁻² s⁻¹ red light (A) or on the time during redarkening (B) of wild-type Arabidopsis leaves preilluminated for 15 min. The parameter is calculated from fluorescence emission spectra registered at 77 K after rapid freezing of the leaves. Each value represents an average of two to six leaves, and error bars represent se.

RT Fluorescence Spectra

RT measurements allow for a far better quantification of the relative changes in the fluorescence spectra before and after illumination when they are registered from the same leaf without changing its orientation in the optical path. To exploit this possibility, we constructed a special dual-light-emitting diode (LED) instrument, based on an Ocean Optics USB2000 CCD spectrometer (see “Materials and Methods”). Since the spectrometer is able to register a full emission spectrum with high signal-to-noise ratio in less than 10 ms, the instrument can follow fast light-induced spectral changes with remarkable sensitivity. The dual-LED mode enables probing of the fluorescence spectra and kinetic development from closed PSII RCs either under variable actinic light or in the absence of it. This allows us, on the one hand, to correlate the spectral changes with kinetic components of NPQ and, on the other hand, to also discriminate the processes using different excitation wavelengths if necessary.

The NPQ kinetics, measured with a single actinic/excitation source, is presented in Figure 3 as the time and wavelength dependence of the NPQ parameter, calculated as NPQ = F(t₀)/F(t) − 1 (Briantais et al., 1979; Bilger and Björkman, 1990). Along the time axis, the figure shows the known kinetics of NPQ formation in the wild-type Arabidopsis and the mutants npq4 and npq1. The PsbS-deficient npq4 plants generate much less NPQ as compared with the wild type, and the kinetics lacks the fast NPQ phase, whereas in npq1, which cannot form Zx, most of the NPQ is formed in the fast initial phase, attributed to the action of PsbS. Along the wavelength axis, the figure reveals the significant nonhomogeneity of the NPQ parameter with respect to the detection wavelength. A distinctive “valley” in the surface plot between 700 and 730 nm is observed for wild-type leaves but is lacking in the npq4 mutant and is smaller in npq1.

Figure 3.

Three-dimensional plots of the NPQ parameter [NPQ = F(t₀)/F(t) − 1] of leaves of Arabidopsis wild-type (A), npq4 (B), and npq1 (C) plants as a function of emission wavelength and irradiation time. Actinic/excitation light was 620 nm, and photon flux density was 1,000 μ mol m⁻² s⁻¹. Fluorescence spectra were recorded every 2 s using the actinic light as excitation source. [See online article for color version of this figure.]

The normalized (to the average intensity at 745–755 nm) emission spectra measured in dual-LED mode from a wild-type leaf in the dark-adapted state, after 30 min of irradiation with 600 μ mol photons m⁻² s⁻¹ red light and after 5 min of redarkening, are shown in Figure 4. The RT spectra are characterized by a major PSII band with a single maximum at 685 nm and a lower intensity broad band in the 710- to 740-nm range, which corresponds to the vibrational tails of the two photosystems and to the “red” Chls of PSI. The most pronounced HL-induced effect on the spectral shape is, as expected, the decrease of the main PSII band. The light-minus-dark difference spectra have a negative peak at 682 nm. In the FR region, the spectra of light-adapted leaves (dashed line in Fig. 4A) show a well-resolved increase above the dark-adapted spectra. In the light-minus-dark difference spectra, this is observed as a positive band with a maximum at 720 nm. Five minutes after switching off the actinic light, the main fluorescence band recovered part of its intensity (dotted line in Fig. 4A). Interestingly, at the same time, the observed increase around 720 nm disappeared completely. This is, to our knowledge, the first indication that the two light-induced effects (i.e. the decrease of F₆₈₂ and the increase of F₇₂₀) are not always matched. The effect is clearly demonstrated by the difference spectra in Figure 4B. The light-minus-redark difference spectrum (dashed line) shows that F₇₂₀ is a fast-relaxing component and the redark-minus-dark spectrum (dotted line) shows the complete absence of the F₇₂₀ difference band after 5 min of dark, the only difference remaining in the F₆₈₂ band.

Figure 4.

A, RT fluorescence emission spectra, normalized at 750 nm, of wild-type Arabidopsis leaves in the dark-adapted state after 30 min of illumination (actinic light of 620 nm, photon flux density of 600 μ mol m⁻² s⁻¹) and after 5 min of redarkening. Additional pulses of blue light (0.2 s, photon flux density of 1,500 μ mol m⁻² s⁻¹) were applied every 60 s to detect the fluorescence spectra from closed PSII reaction centers. rel.u., Relative units. B, Relative difference spectra (in percentage relative to the emission at 750 nm): light minus dark, light minus redark, and redark minus dark. The plots show results of a single leaf measurement. Error bars in B represent the se at 682 and 720 nm for nine measured leaves.

We compared the described light-induced spectral changes in several Arabidopsis NPQ mutants: npq4, which lacks the PsbS protein (Li et al., 2000); L17, which overexpresses PsbS (Li et al., 2002); npq1, which is unable to convert violaxanthin to Zx; npq2, which lacks violaxanthin and neoxanthin but accumulates Zx (Niyogi et al., 1998); and stn7, a mutant lacking the Stn7 protein kinase and thus unable to phosphorylate LHCII and to undergo state 1-state 2 transitions (Bellafiore et al., 2005). Figure 5 shows the light-minus-dark difference spectra for the different mutants, irradiated for 30 min with 600 μ mol photons m⁻² s⁻¹ red light. In all six genotypes, the magnitude of the relative F₆₈₂ decrease seemed comparable, although there were marked dissimilarities in the F₇₂₀ difference band. Most notably, this band was absent in the npq4 mutant and, conversely, was significantly enhanced in the PsbS overexpressor L17 compared with the wild type

Figure 5.

Comparison of light-minus-dark fluorescence difference spectra (30 min of HL), obtained as in Figure 4, for Arabidopsis wild-type (w.t.), npq4, and L17 (A) and stn7, npq1, and npq2 (B) plants. Error bars represent the largest se values at 682 and 720 nm, corresponding to six to nine leaves measured individually for each mutant.

The kinetics of induction and relaxation of NPQ, recorded in dual-LED mode with saturating pulses every 60 s, are shown in Figure 6 for the different mutants. Note that the scaling on the vertical axis is different because the NPQ values largely differed between specific mutants (e.g. L17 typically shows stronger NPQ than the wild type, whereas in the npq1 and npq4 mutants, NPQ is weak). In all mutants during the slow phase of NPQ induction, the increase of the NPQ parameter was significantly smaller in the FR region than at 682 nm. This behavior is well known and attributed to PSI fluorescence in the FR region, which is not quenched (Genty et al., 1990; Pfündel, 1998). Furthermore, in all mutants but npq4 during the first minutes of irradiation, the two curves representing the NPQ parameter at 720 and 750 nm diverge, forming a gap between each other, which remained constant with further increase of time under illumination but quickly disappeared after turning off the actinic light. This is in line with the already shown rapidly reversible light-induced increase of F₇₂₀. In npq4, there are no detectable differences in the NPQ parameter at 720 and 750 nm at any time, and in npq1, the difference is very small.

Figure 6.

Time courses of the light-induced formation and subsequent dark relaxation of the NPQ parameter [NPQ = F(t₀)/F(t) − 1] calculated at three wavelengths in Arabidopsis wild-type (w.t.) plants and the mutants L17 (PsbS overexpressor), npq4 (PsbS less), npq1 (Zx deficient), npq2 (Zx accumulator), and stn7 (Stn7 less) irradiated at 600 μ mol m⁻² s⁻¹ photon flux density red light for 30 min (white bars). The fluorescence was recorded every 60 s by applying a 0.2-s pulse (460 nm, photon flux density of 1,500 μ mol m⁻² s⁻¹).

As a quantitative measure of the spectral changes described above, we use the fluorescence ratios F₆₈₂/F₇₅₀ and F₇₂₀/F₇₅₀. Since the shape of the detected spectrum can slightly vary from leaf to leaf due to anatomical and hence light-scattering and reabsorption differences, we compare only the HL-induced changes in these ratios (i.e. the light-minus-dark differences −Δ F₆₈₂ = −[(F₆₈₂/F₇₅₀)_light − (F₆₈₂/F₇₅₀)_dark] × 100 and Δ F₇₂₀ = [(F₇₂₀/F₇₅₀)_light − (F₇₂₀/F₇₅₀)_dark] × 100). The time courses of these changes are shown in Figure 7 for the different Arabidopsis genotypes. The most important conclusion from these plots is that Δ F₆₈₂ and Δ F₇₂₀ have substantially different time courses. The ratio increases rapidly in the first few minutes of irradiation and then continues to grow at a slower rate. Clearly, F₆₈₂ is quenched even in the absence of PsbS and without operation of the violaxanthin cycle. After switching the actinic light off, Δ F₆₈₂ quickly drops to a transient minimum but does not relax completely and even continues to increase in the absence of background actinic illumination. This transient drop is negligible in npq4 and npq1 and stronger in the L17 mutant.

Figure 7.

Time-dependent changes in the relative fluorescence differences −Δ F₆₈₂ (white circles) and Δ F₇₂₀ (black circles) during illumination (620 nm, photon flux density of 600 μ mol m⁻² s⁻¹) and subsequent redarkening of leaves of Arabidopsis wild-type (w.t.) plants and the mutants L17 (PsbS overexpressor), npq4 (PsbS less), npq1 (Zx deficient), npq2 (Zx accumulator), and stn7 (Stn7 less). Fluorescence was excited by saturating pulses applied every 60 s. The parameters were calculated as the light-minus-dark difference in the ratios F₆₈₂/F₇₅₀ and F₇₂₀/F₇₅₀ (Δ F₇₂₀ was multiplied by 5 for clarity). The curves represent the results from a single leaf measurement. se values corresponding to six to nine measured leaves are shown as error bars.

Unlike Δ F₆₈₂, the time course of the Δ F₇₂₀ spectral change lacks the slow phase of increase in HL. Furthermore, the light-induced Δ F₇₂₀ relaxes almost completely within several minutes after switching the continuous light off. The lifetime of the relaxation of Δ F₇₂₀ in the wild type was about 2 min. The magnitude of Δ F₇₂₀ was very sensitive to the various NPQ-related mutations. Lack of the PsbS protein completely inhibited the light-induced Δ F₇₂₀ changes. The effect was reversed in the PsbS overexpressor, where the ratio increased to a value higher than in the wild type. The Zx-deficient mutant npq1 and the Zx-accumulator npq2 both exhibited the HL-induced Δ F₇₂₀, but in npq1 it was significantly reduced and appeared to have a more complex kinetics. In npq2, the fluorescence spectral change was enhanced but in the dark it recovered more slowly than in the wild type, which is characteristic for the kinetics of NPQ recovery of this mutant (Niyogi et al., 1998). The absence of the Stn7 kinase did not result in substantial changes in any of the observed parameters. More importantly, the inability to undergo state transitions was not accompanied by absent or smaller Δ F₇₂₀ change.

The values of the parameters Δ F₆₈₂ and Δ F₇₂₀ obtained after 30 min of irradiation are summarized in Table I for all measured plant genotypes, together with the technical Stern-Volmer-type parameters NPQ, qE, and qI (Krause and Jahns, 2003), detected at 682 nm. The most striking result is the very strong correlation between Δ F₇₂₀ and the rapidly forming and relaxing quenching component qE, whose contribution is defined as the part of total quenching that relaxes after 10 min of darkness. The studied mutants have a largely different capacity for qE, ranging from 0 (in npq4) to about double the wild-type value in L17. In all mutants, the light-induced Δ F₇₂₀ remarkably followed the qE. In contrast, all mutants were able to produce qI (i.e. the part of quenching that did not relax in 5 min), and the variations in this parameter and in Δ F₆₈₂ were rather small. It is important to stress that while qE and (especially) the Δ F₇₂₀ fluorescence change can be attributed to a particular (PsbS-dependent) photochemical/physical mechanism of quenching, the qI parameter cannot be associated with a single mechanism and is purely a technical term.

Table I. NPQ parameters and fluorescence ratios in Arabidopsis leaves.

NPQ = F_dark/F_light − 1, qE = F_dark/F_light − F_dark/F_redark, qI = F_dark/F_redark − 1, Δ F₆₈₂ = [(F₆₈₂/F₇₅₀)_light − (F₆₈₂/F₇₅₀)_dark] × 100, Δ F₇₂₀ = [(F₇₂₀/F₇₅₀)_light − (F₇₂₀/F₇₅₀)_dark] × 100, where F_dark, F_light, and F_redark correspond to the fluorescence of dark-adapted leaves, after 30 min of HL irradiation, and after 5 min of subsequent redarkening, respectively. Average values ± se are shown (n = 6–9).

Genotype	NPQ682	qE682	qI682	−Δ F682	Δ F720
Wild type	2.9 ± 0.1	2.1 ± 0.4	0.9 ± 0.2	30 ± 2	4.4 ± 0.2
stn7	3.4 ± 0.1	2.8 ± 0.1	0.6 ± 0.0	27 ± 3	5.4 ± 0.3
npq4	1.4 ± 0.1	0.0 ± 0.0	1.3 ± 0.1	29 ± 4	0.3 ± 0.2
L17	6.8 ± 0.5	5.6 ± 0.5	1.3 ± 0.2	35 ± 3	10.4 ± 0.5
npq1	1.5 ± 0.1	0.5 ± 0.1	1.1 ± 0.1	27 ± 2	1.9 ± 0.3
npq2	2.3 ± 0.1	1.6 ± 0.2	0.8 ± 0.2	31 ± 5	6.7 ± 0.4

The behavior of the F₆₈₂/F₇₅₀ ratio was strongly dependent on the measurement protocol, and particularly the frequency of the saturating pulses. When saturating pulses were applied at 3-min intervals (Fig. 8A), the changes in F₆₈₂/F₇₅₀ were significantly smaller as compared with the changes measured with a 1-min pulse interval. The slow phase and particularly the slow increase after switching the actinic light off were basically abolished. The NPQ parameter and the magnitude of Δ F₆₈₂ and Δ F₇₂₀ are shown in Figure 8B for 1-, 3-, and 6-min pulse intervals. The pulse repetition rate has impact on the NPQ and on the Δ F₆₈₂ change, but the Δ F₇₂₀ change is independent from it.

Figure 8.

A, Time-dependent changes in the relative fluorescence difference −Δ F₆₈₂ during illumination and redarkening of leaves of wild-type (w.t.), L17, and npq4 plants. Conditions are as in Figure 7, except that saturating pulses were applied at 3-min intervals. B, NPQ parameter at 682 nm, −Δ F₆₈₂, and Δ F₇₂₀ obtained after 30 min of HL irradiation of wild-type leaves with additional saturating pulses every 1, 3, or 6 min. Error bars represent se (n = 3–9).

DISCUSSION

Spectral Changes Detected at 77 K

The normalized fluorescence emission spectra of HL-treated leaves showed the well-established decrease of fluorescence in the PSII region compared with the PSI region shown by many authors (Krause et al., 1983; Kyle et al., 1983; McTavish, 1988). In addition to this change, a characteristic light-induced change, not previously reported, was found in the FR region: the increase of F₇₃₀ relative to F₇₆₀. There are two possible explanations to interpret this effect: (1) the negative and the positive relative difference bands have the same common origin (i.e. quenching of PSII fluorescence); and (2) there are different mechanisms or sites that are responsible for the different positive/negative light-induced spectral changes. The latter hypothesis brings two further possibilities: (1) the relative increase of F₇₃₀ is a consequence of the quenching of fluorescence components with maxima outside the difference band; and (2) F₇₃₀ is a genuine new fluorescence emission component resulting from the HL-induced formation of a new FR light-emitting excited-state species. The 77 K fluorescence data alone did not provide sufficient evidence to distinguish between these possibilities. However, the results cannot be explained within the usual concept that at low temperature the fluorescence in the FR region originates from PSI alone. The observed increase of F₇₃₀ relative to F₇₆₀, if regarded as a consequence of PSII quenching, would only be possible if PSII fluorescence had a broader vibrational tail extending more to the FR than PSI (i.e. if the ratio F₇₆₀/F₇₃₀ of the pure PSII spectrum were higher than that of the pure PSI spectrum). A state 1-state 2 transition also cannot explain the observed effect. Both the quenching of PSII and a potential state transition could explain the negative F₆₈₃ and F₆₉₁ difference band but not the positive F₇₃₀ band demonstrated in the normalized spectra.

Different Mechanisms Are Responsible for the HL-Induced Spectral Changes

The hypothesis that the positive difference band in the FR region discovered under NPQ conditions has a different mechanism of origin than the quenching of the main PSII fluorescence band was tested by a series of RT experiments in which the time course of light-induced formation and dark relaxation was followed and compared for each fluorescence difference band. The reasoning is that in case these difference bands have a common origin and location in the PSII antenna, they should have the same kinetics and the same spectral characteristics; conversely, differences in the kinetics of formation or relaxation would be an indication that the underlying mechanisms or sites of action are different for these bands. Although at RT the spectral changes in the FR region are smaller than at 77 K, the positive difference band at 720 nm can be resolved clearly and with high precision in our experiments, to reveal that it shows drastically different kinetics of induction and recovery, as compared with the quenching of PSII detected at 682 nm.

The red (682-nm) and FR (720-nm) spectral changes not only differ in their time course during irradiation and subsequent redarkening but also are affected in a distinctly different manner by mutations. For example, the presence or absence of PsbS has a strong impact on the FR change only but no statistically significant effect on the 682-nm changes. The 682-nm changes depend significantly on the applied frequency of saturating light pulses, yet the 720-nm changes are insensitive to it. Therefore, we have to conclude that there are at least two independent processes responsible for the fluorescence time changes. One only leads to a relative decrease of the main PSII fluorescence band at 682 nm. The other quenches the PSII fluorescence as well but concomitantly develops a specific new fluorescence in the 720- to 730-nm range.

The FR Fluorescence Change Is Related to qE

We attempted to investigate the particular biological process underlying the newly found spectroscopic feature of NPQ, Δ F₇₂₀. Although the increase of FR (greater than 700 nm) relative to red fluorescence (680–690 nm) has been commonly interpreted as a result of a state transition (Butler and Kitajima, 1975; Mawson and Cummins, 1986; McTavish, 1988; Walters and Horton, 1991), this cannot explain the Δ F₇₂₀ change for two reasons: first, the relative increase of the F₇₂₀ to F₇₅₀ tail of PSI fluorescence has no obvious explanation; and second, the amount of energy redistribution from PSII to PSI under HL conditions used in our experiments is generally believed to be very low, if present at all (Walters and Horton, 1991). Without relying on any such assumptions, however, we can completely exclude a state transition hypothesis, since in the stn7 mutant of Arabidopsis, which is unable to undergo state transitions (Bellafiore et al., 2005), the Δ F₇₂₀ band is unaffected. Therefore, the Δ F₇₂₀ spectral change is definitely not associated with a state 1-state 2 transition.

The time course of the HL-induced Δ F₇₂₀ spectral change, which reaches a stationary state in a few minutes of irradiation and reverses completely within a few minutes in darkness, clearly associates it with the energy-dependent type of quenching (qE) and distinguishes it from the so-called “sustained” quenching (qI). The magnitude of the relative difference Δ F₇₂₀ is strongly correlated with the value of the qE parameter in all examined mutants. The Δ F₇₂₀ band is thus a spectral marker for the mechanism underlying qE, and as such it opens new possibilities to probe qE specifically and independently from other quenching components (e.g. photoinhibition or others).

The most striking feature of the HL-induced FR fluorescence change is its complete absence in the PsbS-deficient mutant npq4 and its enhancement in the PsbS overexpressor. This demonstrates the direct role of the PsbS protein and further confirms the association of PsbS with qE but not with qI. Since the Δ F₆₈₂ change was not inhibited by the lack of PsbS (on the contrary, it seems to be slightly enhanced), it becomes clear that Arabidopsis plants have at least two different mechanisms for the rapidly inducible NPQ: a PsbS-dependent one and a PsbS-independent one. Furthermore, it follows that the function of PsbS can be exerted independently from the xanthophyll cycle, since the rapid light-induced Δ F₇₂₀ is detected also in the npq1 and npq2 mutants. However, when Zx cannot be formed, the amplitude of the spectral change is significantly smaller, which is consistent with the lower ability of the npq1 mutant to generate both qE and Δ F₇₂₀. It must be also noted that double mutants containing the npq4 mutation (e.g. the npq1npq4 and the npq2npq4 mutants) also did not have any detectable Δ F₇₂₀ (data not shown), confirming the compulsory role of PsbS for qE and the relation of Δ F₇₂₀ with qE.

Could the observed FR fluorescence change have a different and more trivial origin? The measured fluorescence time changes could be in principle influenced by changes in leaf absorbance and reflectance. But such changes would affect the whole fluorescence intensity and would not primarily give rise to distinct spectral changes of the fluorescence. These effects would thus be compensated by the normalization procedure. Changes in the effective absorption might also lead to different reabsorption and thus affect the spectral shape of the fluorescence emission. Clearly, leaves have a strong reabsorption of fluorescence in the range below 695 nm. However, there is little or no absorption in the FR region; thus, such artifacts can be excluded. Another source of error could be light-scattering changes (e.g. associated with Δ pH-induced thylakoid swelling). However, light scattering primarily affects the measurement of absorbance but has a much smaller, and almost independent from the optical arrangement, effect on fluorescence. Furthermore, such light-scattering changes could hardly give rise to a distinct spectral change of the fluorescence in a relatively narrow range. More importantly, though, it is extremely unlikely that effects such as Δ pH-induced thylakoid swelling and others, that in general might give rise to such scattering changes, would follow the well-known qE dependence on PsbS content, as does the observed Δ F₇₂₀ fluorescence change. Therefore, while we cannot totally exclude some disturbances due to possible optical artifacts, it is extremely unlikely that they could in fact explain the majority of the observed Δ F₇₂₀ spectral fluorescence change. Rather, the dependence of this change on the different PsbS mutants strongly suggests that it is indeed qE related.

Photophysical Origin of the FR Spectral Change

The actual photophysical origin of the HL-induced changes in the FR region of the fluorescence spectrum cannot be determined from our steady-state fluorescence measurements alone. However, recent ultrafast fluorescence data on isolated LHCII aggregates and intact leaves of Arabidopsis acquired in our laboratory (Miloslavina et al., 2008; Holzwarth et al., 2009) indicate the likely origin of this NPQ-related emission component. Time-resolved fluorescence spectra of HL-treated leaves at RT revealed a new PsbS-dependent antenna fluorescence component not present in the dark-adapted leaves (Holzwarth et al., 2009). Remarkably, the emission spectrum of this fluorescence decay component showed a strong enhancement in the FR region, in the range 700 to 730 nm, as compared with normal PSII fluorescence. This component was absent in PsbS-deficient plants, and its amplitude was enhanced in the PsbS overexpressor L17. It can be imagined that both the HL-induced FR decay component and the Δ F₇₂₀ change detected here in the steady-state spectra presented originate from the same mechanism.

Interestingly, when LHCII forms aggregates, or higher order oligomers, in vitro, it exhibits enhanced fluorescence in the FR region, particularly strong at low temperatures (Ruban and Horton, 1992; Mullineaux et al., 1993; Vasil'ev et al., 1997) but also at RT (Miloslavina et al., 2008). Because oligomerization of LHCII leads to strong fluorescence quenching (Ide et al., 1987; Ruban and Horton, 1992; Mullineaux et al., 1993), it has actually been proposed as a mechanism for NPQ in vivo (Horton et al., 1991, 2005; Ruban et al., 2007). An essential finding in this direction is that the spectrum and lifetime of the detached antenna component measured in intact quenched leaves under NPQ conditions closely match the FR fluorescence of LHCII aggregates in vitro (Miloslavina et al., 2008). Furthermore, when these LHCII aggregates are cooled to 77 K, a new FR fluorescence located in the range 700 to 730 nm appears (Ruban and Horton, 1992; Mullineaux et al., 1993; Miloslavina et al., 2008). Thus, the FR fluorescence at low temperature in LHCII aggregates also parallels the Δ F₇₂₀ band observed here. We thus suggest that the qE-associated PsbS-dependent FR spectral change has the same origin as the FR emission of quenched LHCII aggregates.

Flash-Induced Quenching Component

Whereas the Δ F₇₂₀ band can be ascribed to a specific mechanism of quenching (i.e. the PsbS-dependent qE), this does not hold true for the Δ F₆₈₂ band, which reflects any processes that quench PSII fluorescence and will depend on both the qE and qI components of NPQ. The F₆₈₂/F₇₅₀ ratio can be then expected to decrease gradually with the slower formation of Zx and not to reverse immediately after turning off the actinic light because of the remaining qI. However, this cannot explain two important observed effects: (1) the slow changes in the F₆₈₂/F₇₅₀ ratio in the npq1 and npq2 mutants, in which the Zx content does not change; and (2) the rise of the ratio observed after switching off the actinic light. These changes are obviously induced by the short saturating flashes only and accumulate with each flash, provided that the time interval between flashes is short. The total irradiation dose from the flashes in our measurements is negligibly small compared with the background actinic light: only 0.4% of the total number of incident photons at a flash interval of 60 s. This rules out the possibility that the fluorescence changes are simply an additive effect of the measuring and actinic light. Therefore, the flash-induced changes reflect a specific process and appear to be triggered by the high turnover rate of PSII during the saturating flash. A key observation is that the flash-induced Δ F₆₈₂ depends on the flash repetition rate. This suggests that these flash-induced changes are reversible on a time scale of a few minutes. At shorter intervals between flashes, the time is not sufficient for these changes to relax and the effects are cumulative. This behavior shows that there is a reversible component of NPQ, independent from both PsbS and Zx, that can be induced by a short pulse of strong light and relaxes on a time scale of several minutes. This then represents an additional independent mechanism of NPQ. It might be responsible for the quenching observed in mutants having neither PsbS nor xanthophyll cycle activity, such as the npq1npq4 and npq2npq4 double mutants. Based on our data, we cannot tell at present what the actual underlying quenching mechanism might be or whether it takes place in the antenna or the RC of PSII. We hypothesize, however, that this quenching might be located in the RC of PSII, thus relating it to RC quenching that has been proposed in the literature (Finazzi et al., 2004; Ivanov et al., 2008).

It must be stressed that the flash-induced quenching component severely distorts the measured NPQ relaxation kinetics when measuring flashes are spaced at short intervals (i.e. shorter than 5 min) and, more importantly, that it contributes significantly to qI, interpreted usually as irreversible or slowly reversible quenching, whereas in fact it is also a quenching component relaxing on a relatively short time scale of about 5 to 10 min. Thus, in our measurements, by far the largest parts of the slow induction and relaxation phases, generally called the qI phase in the literature, are genuine reversible NPQ quenching components and are not related to damage or photoinhibition. In fact, our experiments strongly suggest that true photoinhibition is minor under the conditions applied here even for HL irradiation up to 1 h on intact leaves.

CONCLUSION

We have been able with the help of a newly designed multiwavelength fluorometer for registering spectrally resolved fluorescence and NPQ induction kinetics to dissect and distinguish three contributions to NPQ. We also discovered spectral features specific to the different components of NPQ. We demonstrate three separate and independent mechanisms or sites of action of NPQ. The first one is the rapidly inducible and rapidly relaxing PsbS-dependent qE mechanism, which gives rise to a relative FR fluorescence increase around 720 nm. This spectral signature of qE suggests that a new emitting species is produced together with the qE quenching induction. The second, slower NPQ quenching process occurs essentially independently from PsbS but can be correlated to the formation of Zx. It does not give rise to new FR-fluorescing species but only decreases PSII fluorescence uniformly across the spectrum. Finally, a third quenching process, independent from both PsbS and Zx, was detected that appears to be reversible in the dark on an approximately 5-min time scale but is completely unrelated to the mechanism of qE and is triggered by a high excitation or turnover rate in PSII. All of these three NPQ mechanisms are reversible over the time scales and excitation intensities used in our experiments. Any nonrelaxing quenching contributions appear to be minor.

MATERIALS AND METHODS

Plant Material

Arabidopsis (Arabidopsis thaliana ecotype Columbia 0) wild-type and mutant plants were grown in soil at a light intensity of 150 μ mol photons m⁻² s⁻¹ and a constant temperature of 20°C under long-day conditions (14 h of light/10 h of dark). Leaves from 5- to 6-week-old plants were used for all experiments. The following mutants were used: npq1, defective in the violaxanthin deepoxidase (Niyogi et al., 1998); npq2 (also termed aba1-6), defective in the Zx epoxidase (Niyogi et al., 1998); npq4, PsbS deficient (Li et al., 2000); L17, PsbS overexpressing (Li et al., 2002); and stn7, defective in the LHCII kinase (Bonardi et al., 2005).

77 K Fluorescence Measurements

In all experiments, plants were dark adapted for at least 1 to 2 h. Prior to the measurements, leaves were detached, moistened, and placed between two glass plates, where they were irradiated at RT for certain times using a high-power red (620-nm) LED (Philips Lumileds Lighting) at an intensity of 600 to 1,000 μ mol m⁻² s⁻¹. The leaves were then immediately frozen in liquid nitrogen and placed in an optical cryostat (Oxford Instruments). Fluorescence spectra were recorded in a homemade system based on a USB2000 CCD spectrometer (Ocean Optics) using a blue (465-nm) LED as an excitation source, filtered with a 550-nm long-pass glass filter.

Dual-LED Fluorescence Spectrometer for RT Measurements

For measurements of NPQ generation and relaxation at RT, a special dual-LED fluorescence setup was devised, schematically illustrated in Figure 9A. The optical setup is assembled in a dark casing. Two high-power LEDs (Luxeon Star-O; Philips Lumileds Lighting), one red (620 nm) and one blue (460 nm), are focused with lenses onto the leaf holder. Short-pass 630-nm filters (Thorlabs) are placed in front of the LEDs to remove long-wavelength emission. The fluorescence is collected via another lens at an angle of 45 ° with the leaf surface. Scattered light is filtered through a custom-made 660-nm long-pass interference filter (LayrTech). The fluorescence is focused onto a fiber-optics guide connected to a CCD spectrometer (USB2000; Ocean Optics). The data-acquisition protocol is programmed in Microsoft Visual Studio using an interface library from Ocean Optics. Data analysis is done in MATLAB. In our experiments, the red LED was used as continuous actinic light for the induction of NPQ and the blue LED provided periodic short pulses used at the same time to close the reaction centers and as excitation (measuring) source. For this purpose, a fast-response, electronically switched direct current power supply was built, ensuring pulse rise/decay time of less than 50 μ s. The triggering signal for the pulses is supplied by the spectrometer and programmatically synchronized with the fluorescence detection. Fluorescence spectra are recorded with each given saturating pulse as depicted in Figure 9B. Before the pulse, a background spectrum is recorded. The LED is triggered and, after a predefined delay, the actual pulse-excited fluorescence spectrum is recorded, the LED is then switched off, and the background signal is subtracted by software. This way, the output signal represents only the fluorescence excited by the saturating pulses but not by the actinic light. After switching off the actinic light, the relaxation of NPQ can be monitored, provided that a sufficient time separation is allowed between the measuring pulses to prevent quenching induction.

Figure 9.

A, Schematic representation of the dual-LED fluorescence setup for registering fluorescence emission spectra of leaves during the formation and relaxation of NPQ. Solid lines represent electrical connections, and dotted lines represent the optical path. Numbers are as follows: 1, regulated direct current power supply; 2, excitation LED; 3, actinic LED; 4, short-pass filter (630 nm); 5, lens; 6, leaf holder; 7, long-pass filter (660 nm); 8, fiber optics; 9, casing; 10, USB2000 spectrometer; 11, triggering signal; 12, computer interface. B, Timing diagram of a typical measurement cycle showing the recording of the background fluorescence, turning on of the saturating pulse (SP), and recording of the fluorescence spectrum.

The actual measurement conditions used in the reported dual-light experiments were as follows. Blue measuring/saturating pulses of 200 ms duration and approximate intensity of 1,500 μ mol photons m⁻² s⁻¹ were given every 60 s. Fluorescence detection was started 30 ms after the pulse onset. After the first pulse (to acquire maximum PSII fluorescence in the dark-adapted state), the red actinic light with intensity of 600 μ mol photons m⁻² s⁻¹ was switched on.

The setup can also be used in a single-LED mode, where the same LED is used as actinic and excitation source. This allows for following faster induction changes registering a full spectrum every few milliseconds. Thus, in effect, the described setup provides most of the functionality of both conventional PAM and direct detection fluorometers, although with full spectral resolution. The time-dependent spectra shown in Figure 3 were acquired in single-LED mode using red light (1,000 μ mol photons m⁻² s⁻¹) and a 2-s recording interval.

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers AF134131, AF370251, and AF281655.