The PsbS Protein Controls the Organization of the Photosystem II Antenna in Higher Plant Thylakoid Membranes*

The PsbS subunit of photosystem II (PSII) plays a key role in nonphotochemical quenching (NPQ), the major photoprotective regulatory mechanism in higher plant thylakoid membranes, but its mechanism of action is unknown. Here we describe direct evidence that PsbS controls the organization of PSII and its light harvesting system (LHCII). The changes in chlorophyll fluorescence amplitude associated with the Mg2+-dependent restacking of thylakoid membranes were measured in thylakoids prepared from wild-type plants, a PsbS-deficient mutant and a PsbS overexpresser. The Mg2+ requirement and sigmoidicity of the titration curves for the fluorescence rise were negatively correlated with the level of PsbS. Using a range of PsbS mutants, this effect of PsbS was shown not to depend upon its efficacy in controlling NPQ, but to be related only to protein concentration. Electron microscopy and fluorescence spectroscopy showed that this effect was because of enhancement of the Mg2+-dependent re-association of PSII and LHCII by PsbS, rather than an effect on stacking per se. In the presence of PsbS the LHCII·PSII complex was also more readily removed from thylakoid membranes by detergent, and the level of PsbS protein correlated with the amplitude of the psi-type CD signal originating from features of LHCII·PSII organization. It is proposed that PsbS regulates the interaction between LHCII and PSII in the grana membranes, explaining how it acts as a pH-dependent trigger of the conformational changes within the PSII light harvesting system that result in NPQ.

Light harvesting antennae in plant photosynthesis increase the rate at which absorbed photons excite the photosynthetic reaction centers. In the case of photosystem II (PSII) 3 in higher plants, light harvesting is carried out by a complex assembly of chlorophyll protein antenna complexes (LHCII) located in the grana membranes of the chloroplast (1). The main components are trimeric LHCIIb and monomeric CP24, CP26, and CP29. These are associated together with the dimeric PSII reaction center core complexes to form LHCII⅐PSII supercomplexes (2). The minimal unit of the supercomplex comprises two cores (C 2 ) and two strongly bound LHCIIb trimers (S 2 ), together with two copies of CP26 and CP29, referred to as the C 2 S 2 LHCII⅐PSII supercomplex (3). Two further trimers bound with moderate strength (M 2 ) together with CP24 make up the most predominant supercomplex, C 2 S 2 M 2 . LHCII⅐PSII supercomplexes are segregated from photosystem I, associating together in the lateral plane of the thylakoid membrane, sometimes forming highly ordered semi-crystalline domains. Transverse associations between the outer stromal surfaces of these proteins result in membrane appression and the characteristic grana stack. Although this highly conserved macro-organization of the PSII antenna has been rationalized in terms of efficient capture and utilization of light (4), it can be argued that it is the enabled structural and functional flexibility that is probably more important. Changes in interaction between subunits of the macrostructure allow the regulation of photosynthetic light harvesting (5).
The PSII antenna also contains a number of proteins of relatively low abundance, which have been implicated as playing roles in the regulatory processes (6). One of these proteins, PsbS, has been shown to play a key role in nonphotochemical quenching (NPQ). NPQ describes the processes of energy dissipation, induced under excess light conditions that provide photoprotection of the thylakoid membrane from the potentially damaging effects of the unwanted absorbed light energy (7,8). The main component of NPQ is a feedback regulatory mechanism that is induced in response to the build up of the thylakoid ⌬pH and is frequently referred to as qE (9). Mutants of Arabidopsis deficient in PsbS show greatly inhibited levels of qE (10), and overexpression of the psbS gene results in an enhancement of the maximum extent of qE (11). However, the mechanism of action of PsbS has not been determined. Mutation of glutamate residues on the putative lumen-facing surface of the protein inhibits qE formation suggesting that PsbS is involved in sensing the ⌬pH (12). It is further proposed that PsbS provides the site of quenching; binding of zeaxanthin stimulated by protonation followed by interaction with a chlorophyll⅐protein complex that allows chlorophyll-carotenoid energy transfer and energy dissipation (13). Alternatively, it has been suggested that PsbS acts by modulating a quenching process that is intrinsic to a chlorophyll⅐protein complex (14,15). Evidence to support this suggestion comes from the observation that the stimulation of qE by increased levels of PsbS does not require zeaxanthin (16). It is known that complexes such as LHCII can switch between conformational states with different extents of energy dissipation both in vitro and in vivo (17,18). An indirect modulating role for PsbS, which may still include ⌬pH sensing and/or zeaxanthin binding, implies that it controls this conformational switching in LHCII.
Previous work has suggested that NPQ is influenced by the organization of the light harvesting antenna system in the grana membranes of the chloroplast. Thus, depletion of the minor complex CP24 resulted in a partial inhibition of qE and a disruption of the C 2 S 2 M 2 LHCII⅐/PSII macro-organization (19). Similarly, qE is also partially inhibited in lutein-deficient mutants, which display a decreased stability of the trimeric LHCII structure that is such an intrinsic part of the macroorganization (20). Although alternative explanations of these data can be offered, the question is nevertheless raised whether the function of PsbS could be to provide dynamic regulation of this macrostructure, controlling interactions between LHCII components, modulating their change in conformation. There is already some circumstantial evidence to support this view. In one study, PsbS is shown to associate either with LHCII or the PSII core, depending upon pretreatments, which lead to the monomerization of PsbS dimers (21). In later work, PsbS was found to be widely distributed in the membrane and associated with a number of different complexes (22). It was also found that the PsbS-deficient npq4 mutant displays alterations in PSII function that are not related to NPQ (23). Here, a new direct approach has been taken to test whether PsbS has a role in the organization of the PSII antenna. We show that PsbS specifically controls the association between LHCII and the PSII core.
Room temperature fluorescence analyses of leaves and thylakoids were carried out using a PAM101 chlorophyll fluorimeter (Heinz-Walz). For leaves, plants were dark-adapted for 1 h and illuminated with 1000 mol m 2 s intensity. Reversible NPQ (qE) was calculated as (Fm/FmЈ)-(Fm/FmЉ) using maximum fluorescence values in the dark-adapted state (Fm), during actinic illumination (FmЈ) and after 10-min relaxation in the dark (FmЉ). Thylakoids were assayed with continuous stirring at a chlorophyll concentration of 10 g/ml in presence of 10 M 3-(3,4-dichlorophenyl)-1, 1-dimethylurea (DCMU). Low temperature fluorescence spectra of thylakoid membranes were recorded using a Jobin Yvon FluoroMax-3 spectrophotometer. The chlorophyll concentration was 10 g/ml. For fluorescence emission spectra, excitation was defined at 435 nm with a 5-nm spectral bandwidth. The fluorescence spectral resolution was 1 nm. For fluorescence excitation spectra, fluorescence was detected at either 693 or 732 nm with a 4-nm spectral band-width. The excitation spectral resolution was 1 nm. Spectra were normalized as indicated in the legend to Fig. 4.
CD spectra were recorded between 400 and 750 nm at room temperature in a J810 dichrograph (Jasco) as described previously (19). Four spectra recorded for each sample were averaged to improve the signal-to-noise ratio.
PAGE and Western blotting of thylakoid membranes were carried out as described by Jansson et al. (25) and data analyzed as in (19). Fast protein liquid chromatography (FLPC) was performed on detergent-solubilized thylakoid membranes as described by Ruban et al. (24).
For electron microscopy, thylakoid membrane samples were fixed in 3% glutaraldehyde phosphate buffer and stained in uranyl acetate. Electron microscopy was performed on a FEI Tecnai transmission electron microscope operating at 80 kV.

Re-organization of LHCII⅐PSII during Thylakoid Re-stacking
Is Regulated by the Level of PsbS-In the presence of sufficient [Mg 2ϩ ] isolated thylakoid membranes exhibit a high PSII fluorescence yield. Reduction in [Mg 2ϩ ] results in a dramatic decrease of fluorescence, which is associated with the disruption of the LHCII⅐PSII macrostructure and eventual unstacking of the grana membranes (26,27). Subsequent addition of Mg 2ϩ causes an increase in fluorescence as PSII is re-organized and grana stacks return. These changes were used as an assay system to probe the effect of PsbS on the organization of PSII, specifically to explore its role in LHCII-PSII interaction. Thylakoid membranes from wt, npq4-1 (deficient in PsbS), and L17 (overexpression of PsbS) plants were resuspended in a medium free of bivalent cations. Fig. 1A shows the decrease of fluorescence yield during the measurement using chloroplasts from wt plants. At a minimum fluorescence level (stage 1) MgCl 2 was added back in steps of 0.1 mM, resulting in Mg 2ϩ -titration curves for the fluorescence yield increase. At a final concentration of 1.2 mM MgCl 2 the fluorescence level reached its maximum, which was about 85% of that of the control thylakoid membranes. Fig. 1B shows the plots of the relative fluorescence change against the MgCl 2 concentration in wt, npq4-1, and L17 thylakoids. Although each of the samples showed MgCl 2 -dependent fluorescence increases, the titration curves were strikingly different. The npq4-1 PsbS-deficient mutant showed a higher requirement for MgCl 2 for the restoration of fluorescence than the wt, whereas the L17 PsbS overexpresser needed less. The shapes of the titration curves of the three genotypes were conspicuously different; thus, the titration curve of npq4-1 was strongly sigmoidal, whereas in L17 the sigmoidicity was greatly reduced. The data were fitted to the following equation: Y ϭ Y max C n /(C 0 n ϩ C n ), where Y ϭ FmЈ/Fm, normalized to the range of 0 -1, Y max is theoretical maximum of Y, C ϭ MgCl 2 concentration, C 0 ϭ MgCl 2 concentration at C ϭ 0.5 Y max , and n ϭ sigmoidicity parameter. The sigmoidicity parameters were 4.28 Ϯ 0.09, 2.58 Ϯ 0.13, and 1.64 Ϯ 0.004 and C 0 values 0.655 Ϯ 0.020, 0.460 Ϯ 0.016, and 0.321 Ϯ 0.019 (means Ϯ S.E.) for the titration curves of npq4-1, wt, and L17, respectively. These data are clearly dependent on the amount of expressed PsbS; the greater the amount of this protein the less sigmoidal the response is and the higher the Mg 2ϩ requirement.
In these three lines the extent of qE is dependent on the amount of PsbS (10,11). The question arises about the structural requirements of the effect of PsbS on Mg 2ϩ -dependent re-organization of PSII and its function in qE. To answer this question, different npq mutants were studied in which PsbS had point mutations. Mutations in the putative transmembrane helices lead to the partial loss of function of PsbS resulting in reduced qE (28). Three different mutants were studied: npq4-5, npq4-6, and npq4-7, where the mutations are S104L, A97T, and P185L, respectively. All three lines had highly reduced qE ( Fig.  2A), and their expression levels of PsbS were also lower than that of the wild-type plants (B). The function of PsbS in qE can also be altered by mutation of two glutamate residues on the lumenal side in positions 122 and 226 (12). A single glutamate to glutamine change in either of these sites (E122Q, E226Q) resulted in a reduction of qE ( Fig. 2A). A double mutant carry-ing mutations in both positions (E122Q,E226Q) had a highly reduced qE, similar to the PsbS-deficient npq4-1 mutant. These mutants were generated by expression of site-directed mutant genes and consequently were predicted to have high levels of PsbS protein; in fact, the mutants had 3-5 times more PsbS protein than the wt. Comparing Fig. 2, A and B, it can clearly be observed that the level of qE in these mutated lines did not correlate with the expression level of PsbS. Fig. 2C shows the lack of any correlation between qE and the sigmoidicity parameters obtained from the titration curves. In complete contrast, there was a clear correlation between the expression level of PsbS and the sigmoidicity parameter (Fig. 2D) irrespective of its functionality in qE. The level of the PsbS protein per se appears to determine the differences in kinetics of the Mg 2ϩ -induced fluorescence increase in the wt plants, the L17 overexpresser, and all the npq4 mutants. The structural requirements of PsbS for these two roles are clearly distinct. It should be pointed out that all mutants displayed the same ratio of variable to maximum fluorescence (Fv/Fm) (data not shown) indicating that they all had the same intrinsic efficiency of light harvesting.
PsbS Controls the Re-association of LHCII and PSII rather than Grana Stacking per se-Mg 2ϩ -dependent re-organization of the thylakoid membrane during re-stacking is composed of at least two processes: the changes in interaction between PSII complexes and PSI modulating the spillover of excitation energy and the association of LHCII to PSII core complexes (27,29,30). The summation of both processes gives rise to the increase in fluorescence and the formation of grana stacks.  Experiments were carried out to determine which of these events was controlled by the level of PsbS in the thylakoid membrane. Fig. 3, A-C shows the electron micrographs of wt thylakoids in different stacking states. The micrograph of control membranes at 5 mM MgCl 2 (Fig. 3A) clearly shows grana stacks (arrows). In the absence of MgCl 2 (Fig. 3B) long unstacked lamellae can be observed (position 1 in Fig. 1A), whereas after titration with 1.2 mM MgCl 2 newly organized grana stacks can be observed (position 2 in Fig. 1A) (Fig. 3C). Samples were taken of thylakoids of L17 and npq4-1 at 0.4 mM MgCl 2 (position 3 in Fig. 1B) at which point there was a maximum difference in fluorescence, the former being at about 70% of its maximum, whereas for the latter the increase was only 10%. As expected, in the npq4-1 thylakoids there was no evidence of restacking (Fig.  3D). However, it was found that there were also no detectable stacks in the L17 thylakoids (Fig. 3E). Thus the PsbS-dependent fluorescence increase was not associated with restacking; rather it must be because of regulation of the association of LHCII and PSII.
Fluorescence spectroscopy was used to test this hypothesis for the role of PsbS. Fig. 4 shows the 77 K fluorescence emission spectra of unstacked (u) and stacked (s) thylakoid membranes from L17 (A) and npq4-1 (B). The four main bands appearing in the emission spectra are the LHCII band (680-nm shoulder), PSII bands (at 685 and 693 nm), and a PSI band (at 732 nm). Compared with the stacked thylakoids, in the membranes lacking Mg 2ϩ ions (position 1 in Fig. 1A), the PSII bands decreased relative to the emission at 732 nm. The same features were found in both L17 and npq4-1. However, at 0.4 mM MgCl 2 (m), whereas in L17 there was a significant increase in PSII fluorescence; in npq4-1 the spectrum was almost identical to that of the unstacked membranes. The excitation spectra for PSI fluorescence showed that the absorption cross-section of PSI was larger in the unstacked membranes than the restacked ones, in both samples (Fig. 4, C and D). The difference spectra, stackedminus-unstacked normalized at the position of the long wavelength PSI absorption (705-715 nm region) have bands at 650, 660, 670, and 676 nm, characteristic of PSII (see the close resemblance between the stacked-minus-unstacked spectrum and the excitation spectrum of a PSII "BBY" preparation), indicating increased energy transfer from PSII to PSI. However, at 0.4 mM MgCl 2 , for both L17 and npq4-1 the excitation spectra were almost identical to those of the unstacked membranes, i.e. the PsbS-dependent increase in PSII fluorescence was not accompanied by a decrease in PSI cross-section. The excitation spectra for PSII fluorescence (693-nm band) show a decrease in cross-section in the unstacked thylakoids compared with the restacked ones in both L17 and npq4-1 (Fig. 4, E and F). At 0.4  Fig. 2A). C, restacked wt thylakoid membranes (stage 2 in Fig. 2A). Arrows indicate grana stacks. D, npq4-1 thylakoids at 0.4 mM MgCl 2 (stage 3 in Fig. 2B). E, L17 thylakoids at 0.4 mM MgCl 2 (stage 3 in Fig. 2B). Scale bar, 0.5 m.  npq4-1 (B). C and D, excitation spectra for fluorescence at 732 nm (PSI) of thylakoid membranes at 0 (red, unstacked, u), 5 mM (blue, stacked, s), and 0.4 mM MgCl 2 (black, m) for L17 (C) and npq4-1 (D), normalized to 705 nm, the position of the PSI absorption band; u-s are the differences between u and s spectra (green); in (C) PSII is the fluorescence excitation spectrum of a PSII BBY preparation (745 nm vibronic satellite detection) (gray). E and F, excitation spectra for fluorescence at 693 nm (PSII) of thylakoid membranes at 0 (red, unstacked, u), 5 (blue, stacked, s), and 0.4 mM MgCl 2 (black, m) for L17 (E) and npq4-1, normalized to the intensities of 693 nm emission relative to 732 nm, obtained from Fig. 4, A  and B. (F); m-u are the differences between m and u spectra (green); in C, LHCII is the fluorescence excitation spectrum of isolated LHCII trimers (detection at 680 nm) (gray). FEBRUARY 15, 2008 • VOLUME 283 • NUMBER 7 mM the spectrum resembles the stacked spectrum for L17 but the unstacked spectrum for npq4-1. The difference spectrum, 0.4 mM-minus-unstacked, has no detectable features for npq4-1, but for L17 it has bands at 433, 441, 472, 486, 495, and 511 nm, all of which are characteristic of LHCII (31), i.e. the PsbS-dependent increase in PSII fluorescence is accompanied by an increase in energy transfer from LHCII. It is concluded that PsbS stimulates the Mg 2ϩ -dependent association of LHCII to PSII and not the dissociation of PSII from PSI.

Regulation of the Photosystem II Antenna by PsbS
Changes in PsbS Level Are Associated with Altered Organization of LHCII⅐PSII in Grana Membranes-CD spectroscopy was used to investigate the macro-organization of the thylakoid membranes in vivo (19,32). Excitonic CD bands arising from the short range interactions of the chromophores at (Ϫ) 653 nm, (ϩ) 435 nm, and (ϩ) 448 nm represent the smaller peaks in the spectrum (Fig. 5A). The large bands at around (ϩ) 685-690 nm and (ϩ) 500 nm are the so-called psi-type bands originating from the long-range interaction of the chromophores in large, chiral macrodomains (33). A smaller negative band at around 670 nm has the same origin. These CD bands are sensitive to the macro-organization of the thylakoid membrane (19,34). Qualitatively, the spectra of npq4-1, wt and L17 are very similar (Fig.  5A) but there are small but significant differences, particularly with respect to this negative band at 670 nm. The wt-minus-npq4-1 spectrum shows that the wt exhibits a larger psi-type CD bands than the mutant, with peaks at (ϩ) 685 nm and (Ϫ) 670 nm (Fig. 5B). In the spectrum of L17 the psi-type bands are larger than in the wt. Thus, a difference spectrum L17-minus-npq4-1 shows further enhancement, especially of the 670-nm band. The amplitude of the difference CD 685nm Ϫ CD 670nm is 4.29 Ϯ 0.11, 4.97 Ϯ 0.11, and 5.66 Ϯ 0.18 for npq4-1, wt, and L17, respectively (ϮS.E.; the differences between npq4-1 and wt and between L17 and wt are significant with Ͼ99.9% confidence level by t test, n ϭ 54 and 65, respectively). Clearly, the amplitude of the psi-type CD bands correlates with the amount of PsbS in the membrane.
The results of solubilization of grana membranes by mild detergent treatment have previously been used to assess the stability of the domains of LHCII⅐PSII in the grana membrane (35). Isolated stacked thylakoids were solubilized with 0.7% ␣-DM and the resultant macromolecular complexes separated by gel filtration using fast protein liquid chromatography (Fig.  6A). Several fractions were separated, the identities of which have been previously documented (24,35). The first fraction contains the largest particles, which consist of PSII-enriched membrane fragments (I). This is followed in the gel filtration profile by the LHCII⅐PSII supercomplexes (II). In the npq4-1 mutant the ratio between the amplitudes of peak I and peak II is higher than that of the wt, whereas in the L17 PsbS overexpresser this ratio is lower than the wt. These differences were also observed when 1.0% ␣-DM was used for solubilization (Fig.  6B). There was a strong shoulder arising from peak I in npq4-1, but in the wt this shoulder was much weaker and in L17 not detectable. Thus, the increase in level of PsbS correlated with the increase in extent to which LHCII⅐PSII supercomplexes were removed from the grana membranes.

DISCUSSION
The PsbS protein is an essential component of the regulatory machinery that allows the thylakoid membrane to dynamically respond to changes in the rate of light absorption by the reversible switching between the light harvesting and light dissipation modes of the PSII antenna. However, its mechanism of action has not been established. Previous work has focused on the notion that PsbS is the site of energy dissipation by a mechanism, which involves direct interaction of PsbS-bound zeaxanthin with antenna chlorophyll (12,13). The main thrust of the experimental support for this model is the evidence that zeaxanthin is a direct quencher of excitation energy (36) and that rapidly reversible NPQ is inhibited when PsbS is absent (10). Here we have provided support for an alternative explanation for the mode of action of PsbS; i.e. it controls interactions between the components of the PSII light harvesting antenna in FIGURE 5. Circular dichroism spectroscopy of npq4-1, Col-0, and L17 leaves. A, CD spectra of leaves of npq4-1 (1), wt (2), and L17 (3) at room temperature. The spectra are averages of more than 20 independent measurements. B, difference CD spectra in the red region (620 -730 nm). Solid line, wt-minus-npq4-1; dashed line, L17-minus-npq4-1. Spectra were first normalized on the difference between CD at 620 nm minus CD at 650 nm, the amplitude of the chlorophyll b band. the grana membrane, which modulate the extent of a quenching process intrinsic to an antenna complex (14,15).
New evidence for a role of PsbS in the organization of LHCII⅐PSII in the grana membrane came from observations made on restacking of unstacked membranes. The concentration of PsbS exerted a dramatic effect on the Mg 2ϩ -dependence for the increase in fluorescence as well as the shape of the Mg 2ϩ titration curve. Unstacking of thylakoid membranes results in the randomization of the distribution of chlorophyll protein complexes that are highly segregated in the stacked membranes (27). In the unstacked state LHCII is dissociated from PSII (27,37), and there is an increase in energy transfer from PSII to PSI (29,30). Although it has previously been assumed that the most part of the fluorescence decrease upon unstacking arises from the latter process (27), the fact that we observe a significant proportion of fluorescence increase without restacking indicates that the association of LHCII with PSII is a major factor determining the PSII fluorescence yield. This association is dependent upon Mg 2ϩ and is largely a spontaneous process of self-assembly. The sigmoidal shape of the titration curve suggests a cooperative process, as the subunits of the antenna are added to the cores, and domains of LHCII⅐PSII supercomplexes are formed. Clearly this process is not dependent upon PsbS because the restoration of high fluorescence and membrane stacks is found in the npq4-1 mutant. However, PsbS has a marked facilitating effect, rather like a molecular chaperone. The change in shape of the titration curve to a hyperbolic one in the L17 overexpresser suggests that PsbS provides a template for the assembly process; groups of LHCII components are organized by PsbS allowing concerted assembly into the macrostructure (Fig. 7).
The influence of PsbS on the organization of PSII persists in the stacked membranes. The amplitude of the psi-type CD bands correlated with the amount of PsbS protein. This CD signal has been correlated with Mg 2ϩ -dependent grana stacking (34) and with the accumulation of LHCII during chloroplast development (32). More specifically, it was recently shown to be associated with the presence of the C 2 S 2 M 2 macrostructure of the thylakoid membrane (19). The differences between L17, wt, and npq4-1 are only 7-8% of the CD signal. Clearly, in the absence of PsbS there is not a large-scale alteration in grana organization. Rather, it is suggested that the molecular interactions giving rise to the CD signal are enhanced by PsbS. Evidence to support differences in interactions in grana membranes also came from the observation of the differing sensitivity of the thylakoid membranes to solubilization by detergent. In the L17 membranes, PSII supercomplexes are more easily released from the membranes.
This organizing function of PsbS does not depend upon the protein being active in qE. Point mutations and site-directed mutations in PsbS, which eliminate or reduce qE have no effect on this process. Although it could be argued that this indicates that these structural effects of PsbS are secondary ones, not related to NPQ, with the primary effect of PsbS being as providing the quenching site, we would argue oppositely. Thus, the ⌬pH-dependent action of PsbS can be depicted as acting upon the structure of PSII antenna domain, triggering a change in LHCII interactions (Fig. 7). There is increasing evidence that qE proceeds via conformational change in the subunits of the PSII antenna (17,18), and that interactions between these subunits are an integral part of the process (5,15,19). We therefore propose that protonation of PsbS drives the concerted conformational change in LHCII and thus the formation of qE. The two functional modes of PsbS function in Fig. 7 are hence both reflections of its LHCII-organizing properties.