PsbI Affects the Stability, Function, and Phosphorylation Patterns of Photosystem II Assemblies in Tobacco*

Photosystem II (PSII) core complexes consist of CP47, CP43, D1, D2 proteins and of several low molecular weight integral membrane polypeptides, such as the chloroplast-encoded PsbE, PsbF, and PsbI proteins. To elucidate the function of PsbI in the photosynthetic process as well as in the biogenesis of PSII in higher plants, we generated homoplastomic knock-out plants by replacing most of the tobacco psbI gene with a spectinomycin resistance cartridge. Mutant plants are photoautotrophically viable under green house conditions but sensitive to high light irradiation. Antenna proteins of PSII accumulate to normal amounts, but levels of the PSII core complex are reduced by 50%. Bioenergetic and fluorescence studies uncovered that PsbI is required for the stability but not for the assembly of dimeric PSII and supercomplexes consisting of PSII and the outer antenna (PSII-LHCII). Thermoluminescence emission bands indicate that the presence of PsbI is required for assembly of a fully functional QA binding site. We show that phosphorylation of the reaction center proteins D1 and D2 is light and redox-regulated in the wild type, but phosphorylation is abolished in the mutant, presumably due to structural alterations of PSII when PsbI is deficient. Unlike wild type, phosphorylation of LHCII is strongly increased in the dark due to accumulation of reduced plastoquinone, whereas even upon state II light phosphorylation is decreased in ΔpsbI. These data attest that phosphorylation of D1/D2, CP43, and LHCII is regulated differently.

has recently been shown for PsbJ and PsbL (22). Although the loss of these proteins has only a minor effect in Synechocystis sp. PCC6803, relatively dramatic changes of PSII function and stability have been noted in the corresponding tobacco mutants. Inactivation of PsbJ and PsbL in plants affects the Q A re-oxidation kinetics and the back-electron flow from plastoquinol to Q A , respectively (19,22). Similarly, inactivation of PsbH, PsbI, and PsbK in cyanobacteria and chloroplasts of algae resulted in quite diverse phenotypes (for details and review, see Refs. 1 and 2). It is, therefore, conceivable to conduct comparative studies on different evolutionary lineages in a phylogenetic context. Moreover, the biogenesis of PSII components differs in cyanobacteria and higher plants. Because of the dual genetic origin of the thylakoid system in higher plants, the biogenesis depends largely on factors encoded by the nuclear genome. For instance, HCF136, one of these factors, is essential for the biogenesis of the PSII core complex in higher plants, but its homologous protein appears to be largely dispensable in cyanobacteria (23,24).
The assignment of several LMW subunits including PsbI (4.8 kDa) to the x-ray structure of PSII still remains to be settled (25,26). Earlier studies of the cyanobacterial PSII x-ray structure placed the transmembrane helix of PsbI opposite to the dimerization axis and close to Chlz D2 (13), whereas the crystallographic studies located the PsbI protein at the periphery of PSII core, in close proximity to Chlz D1, the helices A and B of the D1 protein, and the helix VI of CP43 (17,18). PsbI was present in PSII reaction center (RC) preparations of both spinach and cyanobacteria (27). No knock-out strains for psbI are currently available from higher plants. Mutants lacking PsbI have been generated from Chlamydomonas reinhardtii (28) and Synechocystis (29). Inactivation of psbI in Synechocystis and Thermosynechococcus elongatus strain BP-1 caused some reduction of PSII activity, resulting in decreased oxygen evolution to 70 -80% of WT levels, but the mutants grew photoautotrophically. The Synechocystis ⌬psbI mutant was found to be slightly more sensitive to light than the corresponding WT (29). Deletion of psbI in C. reinhardtii caused a more severe effect compared with that in cyanobacteria (28). Although the C. reinhardtii ⌬psbI mutant grows photoautotrophically under low light, its growth rate was quite sensitive to high light. Both the amounts of PSII and the oxygen evolution activity in the mutant were found to be only 10 -20% of WT levels.
In an attempt to evaluate the roles as well as biogenetic and phylogenetic aspects of the enigmatic LMW PSII subunits, we have systematically inactivated individual genes using a transplastomic approach in tobacco. Here we demonstrate that PsbI exerts a crucial role in the stability of dimeric PSII and the intrinsic electron flow as well as in the phosphorylation of proteins of the core complex and the light-harvesting antenna of PSII in tobacco.

EXPERIMENTAL PROCEDURES
Knock-out Construct Strategy for ⌬psbI-In the tobacco plastome the psbI gene is located downstream of psbK, which also encodes a LMW PSII polypeptide (30 -32). In a transplastomic approach, the psbI gene of Nicotiana tabacum cv. Petit Havanna, 110 bp in length (nucleotide position 8398 -8508 on the plastid chromosome; GenBank TM /EBI accession number Z00044), was inactivated by replacing most of the gene with the terminator-less, chimeric amino glycoside 3Ј adenyl transferase (aadA) cassette conferring resistance to spectinomycin in reading frame orientation (7). For this, two PCR reactions were performed with primers psbI-1 (5Ј-GGA TCC AAA ATG CAA TTA TCT CTC C-3Ј) and psbI-2 (5Ј-AAG CTT GCA GCT GAA TTC TAC ACA ATC TCC AAG ATG-3Ј) and with primers psbI-3 (5Ј-GAA TTC AGC TGC AAG CTT ATC CCG GAC GTA ATC CTG G-3Ј) and psbI-4 (5Ј-ATC TCG AGA TTA CAA CTA TAA CAG GC-3Ј). PsbI-2 and psbI-3 generated overlapping products and introduced a diagnostic EcoRI restriction site. The third reaction was performed with primers psbI-1 and psbI-4 using the first PCR product as a template. Primers psbI-1 and psbI-4 introduced two flanking restriction sites, BamHI and XhoI, used for cloning into the vector pBluescript II KSϪ (Stratagene Inc., La Jolla, CA). The presence of a unique restriction site, BtrI, within the disrupted psbI gene allowed the insertion of the aadA cassette between the EcoRI and BtrI restriction sites. The details of the map construction, location of restriction sites, and primer annealing positions are shown in Fig. 1 A and B. The resulting construct (carrying the aadA cassette in the same polarity as psbI) was sequenced to verify correct copying of the gene and used for plastid transformation employed in a biolistic approach (33). Selection and culture conditions of the transformed material as well as the check for homoplastomy were carried out as described (7). Independent transformants were obtained displaying an identical phenotype (data not shown). The transformed material was first grown (12-h photoperiod at 25°C, 10 -20 mol m Ϫ2 s Ϫ1 light intensity) for 4 -5 weeks on Murashige and Skoog (34) medium supplemented with 3% sucrose, 0.8% agar, and 500 mg/liter spectinomycin. Before transferring the plants to the greenhouse, they were kept for 4 -5 weeks on Murashige and Skoog medium without sucrose. If not otherwise indicated, all analyses were carried out with young leaves of ϳ2-month-old plants grown in vitro and under greenhouse conditions (day 27°C, night 20°C), respectively. Tobacco lines carrying the aadA cassette in a neutral insertion site and referred to as RV plants were used as WT control plants (22).
SDS-PAGE and Immunoblot Analysis-Thylakoid membrane proteins of 3-4-week-old plants were isolated and solubilized for SDS-PAGE as described (7). Proteins separated by SDS-Tris-glycine-PAGE (15% acrylamide) (35) were electroblotted to polyvinylidene difluoride membranes (Amersham Biosciences), incubated with monospecific polyclonal antisera (7), and visualized by the enhanced chemiluminescence technique (Amersham Biosciences). Equal loading of proteins was always checked by zinc-imidazole staining of the gels before blotting.
Separation of Thylakoid Membrane Complexes by Sucrose Density Gradient Centrifugation-Thylakoid membranes used for the separation of protein complexes by sucrose density gradient centrifugation were isolated as described earlier (7).
Resolution of Isolated Thylakoid Membrane Complexes by Blue Native (BN)-PAGE and in Vivo Labeling-Labeling of thylakoid membrane proteins was performed by incubation of cotyledons with [ 35 S]methionine for 40 min as described (36).
BN-PAGE analysis was performed as described earlier (37). All solutions were supplemented with 10 mM NaF. Thylakoid membranes equivalent to 30 g of chlorophyll were solubilized with dodecyl-␤-D-maltoside (1% final concentration) and separated on a 4 -12% acrylamide gradient. After electrophoresis, lanes of the BN polyacrylamide gel were excised, denatured, and run in the second dimension in SDS-PAGE with 15% acrylamide and 4 M urea. Subsequently, the gels were silver-stained or used for immunoblotting. Individual spots were labeled according to their determination by mass spectrometry (38).
In Vitro Phosphorylation of PSII Proteins-Redox-dependent LHCII phosphorylation of WT and mutant thylakoids was carried out as described (39). Activation of the protein kinase in the dark was obtained by the addition of reduced duroquinol (1 mM) (40). Phosphorylation was terminated by the addition of denaturing sample buffer, and the sample proteins were resolved by SDS-PAGE. The extent of phosphorylation was detected with phosphothreonine antibodies (Zymed Laboratories Inc., Berlin, Germany; New England Biolabs, County Road, MA).
Chlorophyll a Fluorescence Induction Kinetics-Chlorophyll a fluorescence induction kinetics of tobacco WT and mutant leaves was measured using a pulse-modulated fluorimeter (PAM101, Waltz, Effeltrich, Germany) (22). Leaves were darkadapted for 5 min before the fluorescence measurements. The minimal (Fo) and maximal (Fm) fluorescence yield and the variable fluorescence (Fv), calculated as (Fm Ϫ Fo) as well as the ratio Fv/Fm, which reflects the potential yield of the photochemical reaction of PSII (41), were recorded at room temperature. Photochemical and non-photochemical quenching (qP and NPQ, respectively) were calculated as (Fm Ϫ FmЈ)/FmЈ) and (FmЈ Ϫ F)/(FmЈ Ϫ Fo), respectively (42).
Measurements of PSI Activity-Photosystem I activity was measured on leaves as absorption changes at 830 nm induced by far red light (⌬Amax) (730 nm; 12 watts m Ϫ2 ) and in the absence or presence of actinic light (⌬A) (650 nm, 20 and 250 mol of photons m Ϫ2 s Ϫ1 ) using the PSI attachment of PAM101 (Walz, Effeltrich, Germany) (43). The oxidation status of PSI at the light intensities indicated was expressed as the fraction ⌬A/⌬Amax. Thermoluminescence (TL) Measurements-TL measurements were performed using a home built apparatus (22). Thylakoid fractions were prepared by grinding leaves in a buffer containing 20 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 20 mM NaCl, and 100 mM sorbitol. Homogenized material was filtered through nylon micromesh and used immediately for measurements. Samples (200 l; 10 -15 g chlorophyll/sample) were dark-adapted on the TL stage at 20°C for 3 min and then rapidly frozen to Ϫ20°C. The samples were then excited with saturating flashes delivered by a xenon arc discharge lamp (EG&G, 0.05-microfarad capacitor, charged at 1000 V, 3 s at 70% light emission). TL was recorded upon heating the sample at a constant rate of 0.6°C s Ϫ1 . The herbicides DCMU (3-[3Ј,4Ј-dichlorphenyl]-1,1-dimethylurea) or Ioxynil (4-hydroxy-3,5-di-iodobenzonitrile) were added at concentrations of 10 and 5 M, respectively, to inhibit the electron transfer from Q A Ϫ to Q B . For measuring the B band (Q B Ϫ/S 2 ,S 3 recombination) oscillations, the dark-adapted samples were slowly cooled, and consecutive flashes (1-6 flashes, time interval 300 ms) were applied between 1 and 0°C followed by rapid cooling to Ϫ10°C.
Low Temperature Fluorescence Measurements-Low temperature (77 K) emission spectra were performed with thylakoids prepared from young dark-adapted leaves of WT and ⌬psbI plants (44). Thylakoid membranes (40 g chlorophyll/ ml) were transferred into a glass tube (0.7-mm internal diameter) for fluorescence measurements and immediately frozen in liquid nitrogen. Fluorescence was excited at 440 nm, and the emission was recorded between 650 and 800 nm. All spectra were recorded with a Jobin Yvon Spex Fluorolog spectrofluorometer (Horiba, France) equipped with a photomultiplier (R 374, Hamamatsu, Japan). Slits of 1 nm were used.
Photoinhibition-To determine the sensitivity of PSII to oxidative stress, leaves of WT and ⌬psbI plants grown under greenhouse conditions were exposed to 500 mol of photons m Ϫ2 s Ϫ1 , and the photoinactivation of PSII was measured as ⌬(Fv/Fm)/time. To estimate the PSII recovery process during the exposure to the high light treatment, leaf discs were exposed to a similar light treatment after preinfiltration with a FIGURE 1. Construct map for replacing most of psbI with the aadA cassette. A, the PCR-based site-directed mutagenesis strategy for disruption of psbI and for introduction of the restriction sites EcoRI, BamHI, and XhoI using 4 primer combinations (primers psbI 1, 2, 3, and 4) is indicated. The base-pair positions of the gene psbI (8398 -8508) and the PCR primers containing introduced restriction sites are indicted in brackets. Plastid DNA was used as the template for the first two PCR reactions, and the resulting products were used for the third PCR. The arrows indicate the transcriptional direction of the genes psbK, psbI, and trnS. BtrI, restriction site downstream of psbI. The resulting PCR product was ligated into the BamHI and XhoI sites of the transformation vector pBluescript KSϪ (Stratagene). B, the map shows the introduction of the EcoRI-and SmaI-digested aminoglycoside 3Ј adenyl transferase (aadA) selection cassette into the disrupted psbI gene at the BtrI and the introduced EcoRI restriction sites. P, 16 S rDNA promoter; R, ribosome binding site.
solution of D-threo-chloramphenicol (200 g ml Ϫ1 ) for 30 min before the light exposure. For control purposes leaf discs were incubated in water.
To assess the photoinhibition and the capacity to recover from photoinhibition, leaves were also exposed to 1500 mol of photons m Ϫ2 s Ϫ1 until a similar loss of activity was reached in both WT and ⌬psbI (measured as Fv/Fm ϭ 0.17), and subsequently incubated at low light (3 mol photons m Ϫ2 s Ϫ1 ) for up to 6 h, measuring the Fv/Fm level every 1 h.

RESULTS
Disruption of psbI in the tobacco plastid chromosome caused an increased light sensitivity, but young leaves appeared normal green in independent transformants. Comparable with ⌬psbZ but different from ⌬psbE, -F, -L, or -J (19, 7), ⌬psbI was capable of growing photoautotrophically on soil under greenhouse conditions. Initial fluorescence kinetic data of ⌬psbI suggested a defect in PSII (see below).  (Table 1).
To compare the efficiency of the electron flow between PSII and PSI with that between PSI and its final electron acceptors in intact leaves, the extent of PSI oxidation expressed as ⌬A/⌬Amax was monitored using absorption changes at 830 nm in the background of different intensities of actinic light in the steady state (20 and 250 mol of photons m Ϫ2 s Ϫ1 , 650 nm) ( Table 1). The results showed significantly higher levels of oxidized PSI in the steady state excited by both actinic light intensities in the mutant as compared with the WT ( Table 1). The results are indicative of a significantly lower rate of electron flow from PSII to the plastoquinone pool in the mutant relative to plastoquinol oxidation activity via PSI.
Levels of PSII Core Components Are Specifically Reduced in the ⌬psbI Mutant-To corroborate that inactivation of psbI did not affect the expression of genes for PSII antenna and other photosynthetic membrane complexes, immunoblotting using specific antisera against distinct thylakoid proteins was performed. The data obtained confirmed that the stationary protein levels of PSII antenna (LHCB1, CP29, CP26, and CP24), PSI (PsaF and LHCI), ATP synthase (␣ and ␤ subunits), and cytochrome b 6 f complex (cytochrome f ) were comparable with those of the WT (Fig. 2 and data not shown). Levels of the PSII core proteins D1, CP43, and CP47 as well as of the oxygen evolving complex protein PsbO were reduced to about 50% compared with the WT, indicating that the relative content of the PSII RC is lower in ⌬psbI (Fig. 2).
The Q A Midpoint Potential Is Affected in ⌬psbI as Measured by Charge Recombination-Charge recombination between the different oxidation states (S 1 -S 3 ) of the Mn 4 Ca complex at the electron donor side and reduced primary and secondary semiquinone acceptors of PSII Q A or Q B , respectively, serves as an indicator of the forward and back electron flow activity within photosystem II (45,46). During the recombination process P 680ϩ is generated that is reduced by electron flow from Q B Ϫ or Q A Ϫ, a process accompanied by luminescence (46,47). Back  electron flow uphill the redox potential requires energy input that can be supplied by heat and thus the glow generated by the charge recombination in darkness is termed thermoluminescence (TL) (48). The temperature at which the luminescence is maximal is related to the energy gap between the recombining pairs. The maximal TL signal generated by recombination of Q B Ϫ/S 2,3 pairs, the B band emission, occurs in tobacco thylakoids at about 35°C (22,49). To measure the recombination of the Q A Ϫ/S 2 pair, one has to block reduction of the Q B quinone during the excitation of the sample by a single turnover flash. This can be achieved by addition of electron flow inhibitors binding specifically at the Q B site such as the urea or phenol derived herbicides, DCMU or Ioxynil, respectively. Because back electron flow from Q A Ϫ to P 680ϩ presents a lower energy gap, the resulting TL signal, termed Q band, occurs at a lower temperature. In WT tobacco, the Q band in presence of DCMU occurs at about 15°C, whereas that obtained in presence of Ioxynil occurs at about 3°C ( Fig. 3) (22,50,51). This difference between the Q band temperature resulting from the recombination of the same charge separated pair (Q A Ϫ to P 680ϩ ) is ascribed to changes in the conformation of the Q B site upon binding of DCMU that affects the midpoint potential of the Q A site. Conversely, binding of Ioxynil to the Q B site is supposed not to affect the midpoint potential of the Q A (50). Thus, use of these herbicides may give information not only on their efficiencies to bind to the Q B site but also on the effect of their binding on the conformation of this site and the resulting effect on its interaction with the Q A site expressed as alterations in the Measurements of the TL emission of ⌬psbI thylakoids showed that the peak temperatures of the B band and that of the Q band induced by the addition of Ioxynil were the same, 35 and 3°C, respectively, for both mutant and WT control (Figs. 3, A  and B). However, the Q band was downshifted to 10°C in DCMU-treated mutant samples (Fig. 3B).
In light-exposed thylakoids, the Q B site quinone exhibits a binary oscillation between the quinone and semiquinone reduced states. The double-reduced quinone is protonated to quinol that leaves the site and is replaced by a quinone molecule from the plastoquinone pool. Upon transition from light to darkness, half of the PSII population is in the Q B Ϫ and half in the Q B states. In illuminated thylakoids, the stable S-states of the Mn 4 Ca complex exhibit a four steps oscillation, S 0 and three increasing oxidation steps, S 1 to S 3 . The states S 4 and S 4 Ј are highly unstable, extract 4 electrons from water, releasing dioxygen and returning to the S 0 state (52).
Upon transition to darkness, the population of PSII consists of the S 0 to S 3 states. However, during dark adaptation at 25°C for 3 min back electron flow from the Q B Ϫ population to that of the oxidized states of the Mn 4 Ca complex and, thus, charge recombination, will occur driven by the thermal energy and potential difference between the oxidized S 2 and S 3 states and Q B Ϫ. The S 1 state practically does not recombine (47). Under the experimental condition used, this will result in a final ratio of 75% S 1 , 25% S 0 and practically equal amounts of Q B and Q B Ϫ (47). Consecutive single turnover excitations of dark-adapted thylakoids leads to oscillations of the ratio of recombining Q B Ϫ/ S 3 :Q B Ϫ/S 2 pairs and respective light emission with a higher emission for the recombination of the Q B Ϫ/S 3 pair. Thus, the TL signal intensity of the B band oscillates with the number of single turnover exciting flashes given to dark-adapted thylakoids, with a period of 4, the maxima being at the second and sixth flash (46,47,52).
The changes in the properties of the Q B site of the ⌬psbI mutant may be exhibited not only in the TL measurements elicited by a single turnover flash in presence of DCMU but also under conditions of multiple excitations sustaining forward electron flow. Therefore, we tested whether the mutation affects the oscillation pattern of the TL signal with the number of excitation flashes due to a limited forward electron flow. Indeed, the WT exhibits the normal 2/6 oscillation pattern, whereas the mutant shows an unusual 1/5 oscillation type, indicating an interference with the electron transfer on either the donor or the acceptor side of PSII (Fig. 3C). These results indicate possible changes in both back and forward electron flow in the mutant, suggesting that the mutation affects the interaction between the Q A /Q B sites and, thus, possibly resulting in alteration of the Q A midpoint potential under continuous forward electron flow.
The Ratio PSII/PSI and Energy Transfer to the PSII RC Are Reduced in the ⌬psbI Mutant-The ratio of PSI to PSII (F PSI / F PSII ) and the functional connection of the LHCII antenna to the RC were monitored by 77 K fluorescence spectroscopy (Fig. 4, A and B). Thylakoid suspensions prepared from young mutant leaves, which displayed a relatively high ratio Fv/Fm of ϳ0.70, were used. A significantly increased ratio F PSI /F PSII appeared in all mutant samples tested as compared with WT (Fig. 4A). The data indicate an excess of PSI relative to PSII. When the signal intensities of the low temperature emission spectra were normalized to the CP43-related peak at 688 nm, leaves of the ⌬psbI mutant yielded a higher CP47-related fluorescence at 697 nm as compared with the WT (Fig. 4B). This is indicative of a decreased energy transfer to CP43 and a favored fluorescence emission from CP47 in the mutant. Mutant leaves exhibited an emission shoulder at 680 nm, which is missing in the wild type, indicating a partial dissociation of the outer LHCII antenna from the PSII RC (Fig. 4B). This may also reflect   the reduced amount of the core components relative to the antenna of PSII (Fig. 2).
Effect of the ⌬psbI Mutation on the Stability of Dimeric PSII and Higher Order PSII-LHCII Complexes-The presence and the relative content of chlorophyll-protein complexes have been investigated by separation of solubilized thylakoid membrane complexes in sucrose density gradients. The chlorophyllprotein banding patterns showed that the PSII-LHCII supercomplexes were below the limit of detection, whereas levels of the PSII monomer as well as of trimeric LHCII complexes associated with CP29 and CP24 antenna were predominant in mutant thylakoids (Fig. 5). A moderate increase was noted in the relative intensity of free LHCII monomers to that of the WT control sample. These results indicated the presence of unstable dimeric PSII-LHCII supercomplexes in the psbI mutant. To confirm these data native membrane complexes were separated by BN-PAGE followed by SDS-PAGE in the second dimension. In accordance with the data described, trimeric LHCII complexes, monomeric PSII and monomeric RC47 complexes lacking CP43, accumulated to higher levels at the expense of PSII dimers and PSII-LHCII supercomplexes in ⌬psbI (Fig. 6, A-C). The potential of ⌬psbI to assemble higher order PSII complexes was checked by immunological detection of the D1 protein; however, only minute amounts of dimeric PSII complexes could be detected in ⌬psbI upon prolonged exposure (Fig. 6B). Interestingly, when de novo synthesized native complexes were investigated by in vivo radiolabeling, substantial amounts of dimers and higher order PSII-LHCII supercomplexes were found to assemble in the mutant (Fig. 7). We, therefore, conclude that PsbI is not essential for an efficient assembly process but rather for the stability of dimeric PSII and PSII-LHCII supercomplexes. Other thylakoid protein complexes, such as PSI,ATP synthase, and cytochrome b 6 f complex, were found to be unaltered in size and abundance (data not shown and Fig. 6,  A and C).

Phosphorylation of PSII Core Proteins Is Remarkably Decreased and That of LHCII Is Reversely Regulated in psbI
Mutants-Light-induced phosphorylation of the LHCII is mediated by the redox state of the plastoquinol pool, whereas that of the RC has not been studied extensively. Photoautotrophically grown tobacco WT plants showed an increased phosphorylation of D1, D2, and CP43 with increasing light intensity from 30 to 200 mol m Ϫ2 s Ϫ1 and with increasing incubation time from 5 to 15 min.
DCMU treatment inhibits the light-induced phosphorylation of the RC proteins in the WT and, thus, resulted in a phosphorylation status similar to that of dark-adapted plants (Fig.  8A). Phosphorylation of PSII-RC proteins was almost equally distributed in WT supercomplexes, dimers, monomers, and RC47 complexes separated by BN-PAGE (Fig. 8B). However, in ⌬psbI plants phosphorylation of D1 and D2 was close to the limit of detection and that of CP43 was reduced to Յ5% in the dark and under all chosen light treatments (Fig. 8, A-C). We conclude that the responsible kinase does not phosphorylate RC proteins efficiently in the psbI mutants.
Phosphorylation of the LHCII antenna is barely detectable in darkness (state I) but increases with increasing light intensity at 650 nm from 10 to 40 mol photons m Ϫ2 s Ϫ1 (state II) in the WT (Fig. 8, C and D). High light treatment (500 mol of photons m Ϫ2 s Ϫ1 ) caused a decrease in LHCII phosphorylation in the WT (39). Strikingly, phosphorylation of LHCII was high in state I (dark) in the mutant and decreased already under low light (650-nm light at 10 mol m Ϫ2 s Ϫ1 ). Therefore, it is evident that the phosphorylation of LHCII is reversely regulated in ⌬psbI compared with the wild type upon dark/light changes. Phosphorylation of LHCII in the dark induced a re-distribution of light energy in favor of PSI in the mutant as revealed by calculation of the F PSI /F PSII ratio (Table 1).
Remarkably, compared with dark-adapted plants, light treatment also reduces phosphorylation of CP43 in the mutant but had no effect on the phosphorylation of the RC proteins D1 and D2 (Fig. 8C). To test whether the responsible kinase is unable to access the RC proteins or whether a reduced PSII activity caused dephosphorylation of the RC proteins in ⌬psbI, the plastoquinone pool was reduced in vitro by adding reduced duroquinone. The duroquinol-dependent activation of LHCII phosphorylation in the mutant was comparable with that of the WT (Fig. 8E). Moreover, phosphorylation of the RC proteins of duroquinol-treated samples was increased in the WT. Surprisingly substantial amounts of CP43 could be phosphorylated upon duroquinol treatment. On the other hand, phosphorylation of D1 and D2 remained barely detectable in the mutant (Fig. 8E). This implies that the phosphorylation of CP43 and that of D1/D2 can be differently regulated in plants and that structural changes of PSII caused the failure to efficiently phosphorylate RC proteins in the ⌬psbI mutant.
To test whether structural alterations or a reduced plastoquinone pool in the dark are also responsible for phosphorylation of the LHCII in ⌬psbI, PSI-specific far-red light (730 nm; 6 watts m Ϫ2 ) was applied to leaves for 15 min (Fig. 8F). It appears that LHCII proteins were dephosphorylated under these conditions, showing that indeed reduction of the plastoquinone pool causes phosphorylation of LHCII in the dark in ⌬psbI.
The psbI Mutant Shows an Increased Light Sensitivity-The lower rate of forward electron flow through PSII in ⌬psbI as well as the alteration of the Q B binding site suggest an increased back electron transfer and charge recombination in PSII, thus increasing the probability of damage of the PSII-RC through chlorophyll triplet formation, free radical production, and photo-oxidation (53-56). To verify this assumption, leaves of WT FIGURE 7. Two-dimensional gel electrophoresis of freshly labeled thylakoid membrane complexes. Pulse-labeled thylakoid membrane proteins were separated by two-dimensional gel electrophoreses. The proteins were transferred to a polyvinylidene difluoride membrane and analyzed by phosphorimaging (BAS2000 software package and the AIDA software package Version 3.25 beta; Raytest, Straubenhardt, Germany). Substantial amounts of PSII-LHCII supercomplexes and dimers were detectable. In agreement with previous reports, the antiserum from Zymed Laboratories Inc. hardly recognizes phosphorylated LHCII proteins (11). Phosphorylation of reaction center proteins CP43, D1, and D2 is induced by light (heterochromatic) in the WT but is almost absent in ⌬psbI. The immunoblot performed with antisera raised against the ATP synthase ␣ and ␤ subunits demonstrates equal loading. B, immunological analysis of the second dimension demonstrates that only traces of CP43 are detectable and that LHCII proteins are highly phosphorylated in dark-adapted mutants. C, a dilution series of the WT was chosen to estimate the lowered amount of phosphorylation of reaction center and antenna proteins in the mutant in the dark and in the light. D, the phosphorylation of LHCII in thylakoids of dark-adapted and red light-incubated plants showed a reverse regulation of the phosphorylation pattern in the mutant (state I ϭ dark, state II ϭ 40 mol of photons m Ϫ2 s Ϫ1 at 650 nm). E, phosphorylation of LHCII and reaction center proteins of dark-adapted thylakoids is significantly induced by reduced duroquinone in the WT. Duroquinol activates phosphorylation of the LHCII but only traces of CP43 in the mutant. F, oxidation of the plastoquinone pool was achieved by applying PSI-specific far-red (FR) light (6 watts m Ϫ2 ) for 15 min to leaves. This treatment induced dephosphorylation of LHCII in the mutant, indicating that the reduced plastoquinone pool caused phosphorylation of LHCII in the dark. and mutant plants were exposed to high irradiance (1,500 mol of photons m Ϫ2 s Ϫ1 ), and PSII photoinactivation was measured as a function of exposure time with and without the chloroplast translational inhibitor chloramphenicol. The results indicate that PSII photoinactivation is faster in the mutant as compared with the WT (Fig. 9A). To evaluate the cause of light sensitivity in ⌬psbI, the recovery rate of photo-damaged PSII was measured after photoinactivation. Within the first hour the recovery rate was almost identical in both cases. The mutant restored 75% of its original PSII quantum yield as compared with 88% for that of the WT after 6 h of recovery, indicating that ⌬psbI is primarily light-sensitive but is basically able to assemble photodamaged PSII complexes and, therefore, to recover from photoinhibitory injury (Fig. 9B).

Transplastomic psbI Knock-out Mutants in Tobacco Grow
Photoautotrophically-Stability of the PSII core complex in higher plants depends not only on the presence and subsequent assembly of D1, D2, CP43, and CP47 but also on several LMW proteins such as psbE, psbF, psbL, and psbJ, which are all encoded by the plastid chromosome (19,7,22). This is consistent with the succession of subunit integration during the assembly process that has been settled for the order given: cytochrome b 559 , D2, D1, CP47, and CP43 (57)(58)(59). It is, therefore, not surprising that deletion of either of those early assembled as well as structurally and functionally crucial components is deleterious to the photosynthetic process, since no functional PSII complexes are formed.
Limited information is available on LMWs (4, 7), including PsbI, which is also present in the PSII core complex (60). Unlike other components, which are present in core preparations (1-2), PsbI is not essential for photoautotrophic growth in Synechocystis and C. reinhardtii (28 -29). To understand the biogenetic and structural as well as functional aspects of PSII in higher plants more profoundly, we have generated tobacco knock-out plants lacking psbI and characterized the mutant using biochemical, spectroscopic, and fluorimetric approaches. The data obtained show that different from all other core components, PsbI is dispensable for the assembly of the RC core in higher plant thylakoids. Moreover, its loss even allows photoautotrophic growth, but its requirement is disclosed only under distinct light regimes. This differs from other LMWs, such as PsbE, F, L, and J (7), the latter two involved primarily in governing the redox potential of cofactors ligated by the D1/D2 heterodimer to ensure efficient charge separation and the following forward electron transfer through and out of PSII (19,22).
PsbI Confers Stability to Dimeric PSII-LHCII Supercomplexes-Analysis of PSII assemblies illustrates that only traces of PSII dimer and supercomplexes could be found in the absence of PsbI (Figs. 6 and 7). However, in vivo labeling experiments revealed that ⌬psbI mutants possess the potential to form various PSII-LHCII supercomplexes (Fig. 7D). Therefore, it is evident that the PsbI protein is essential for the stability of dimeric PSII and, depending on it, of PSII-LHCII complexes. In addition, PsbI is crucial for an efficient forward electron transport within photosystem II. In summary, PsbI exerts a dual function. It is less important for the basic assembly of this photosystem; however, it is required for the stability of higher order complexes of PSII and the proper functioning of PSII.
Although the precise localization of PsbI in the PSII assembly is still a matter of debate based on the crystal structure of the cyanobacterial PSII dimer and the proposed location of the individual LMW subunits, PsbI was assigned toward the outer side of the monomeric core (17)(18). At this position it may be involved in binding the antenna protein Lhcb4 (CP29), which in turn could interact with the LHCII trimers M and/or S (3). If this were true, dimer stability could be influenced by low mass subunits residing at two positions, centrally at the monomer interphase, such as PsbL (22), and peripherally, such as PsbI, with its interaction along the pseudo 2-fold axis of symmetry with the other monomeric core (17). Thus, PsbI could "bracket" two monomers forming a PSII dimer. Its absence would destabilize the dimer and, as a consequence, the interaction between PSII and the CP29-LHCII. This could explain the effects on the energy transfer to the core complex in the mutant. The low temperature fluorescence (77 K) analysis is consistent with a reduced rate of energy transfer to the core, especially to CP43, in the ⌬psbI mutant. Monomeric cores attached to LHCII proteins have never been observed in mixed populations of disrupted grana as recognized in cryoelectron micrographic images after quick and mild solubilization of the membranes (for review, see Ref. 3). Furthermore, antenna proteins tend to readily dissociate from monomerized PSII core complexes. Thus, it is conceivable that disruption of psbI affects primarily the stability of the dimer and only secondarily, the formation of supercomplexes.
Examination of the PSII 2 LHCII 8 ϩ 2 model (3) suggests that one of the functions of PsbI is to provide a flexible interface between the rather rigid structure of the PSII core, the fixed antennae, and the mobile LHCII. PsbI may serve as a structural buffer between the ever-changing positions and interaction strength of the peripheral antennae and the photochemical core. It may also play a direct role in stabilization of the dimeric form of PSII. Inactivation of PsbI not only alters energy transfer from the major antenna but also stability of structure/function of the PSII core dimer. The relatively pronounced effects in ⌬psbI attest the flexibility of the connection of the photosynthetic machinery with the "outer world." Increased Light Sensitivity of PSII in PsbI Mutants-Accumulation of monomeric PSII complexes in ⌬psbI may generate ␤-like PSII centers impaired in plastoquinone reduction (61) and, thus, be responsible for the slow reduction of oxidized P700, an increased sensitivity to photoinactivation, and a somewhat slower re-assembly of photodamaged PSII.
Inhibition of electron flow from the primary (Q A ) to the secondary (Q B ) quinone acceptor and the resulting accumulation of reduced Q A species are the principal events that initiate damage of PSII by photooxidative stress (62,54). Under high light conditions, the primary cause for photoinactivation of PSII is thought to be due to damage of the D1 protein (for review, see Ref. 55). Therefore, the sensitivity of PSII to high light and its ability to restore photochemical efficiency under conditions of low light was analyzed in mutant and WT. It appears that PSII is more rapidly degraded than repaired in ⌬psbI with increasing light intensity. Three hours of photoinhibitory light treatment were required to lose 75% of PSII quantum yield in the WT; however, a similar loss was noted already after 2 h in ⌬psbI, reinforcing an increased light sensitivity of its PSII (data not shown). Although the initial recovery rate is identical in mutant and WT, the overall capability to recover was lower in the mutant. The delayed repair of photoinactivated PSII in ⌬psbI suggests that PsbI also influences the recovery process, although this may be a secondary effect of the mutation. Furthermore, an increased light sensitivity of the mutant when synthesis of D1 is inhibited corroborates that D1 protein degradation occurs at substantially higher rates in ⌬psbI than in the WT.
The ⌬psbI Mutation Destabilizes the Q A Midpoint Potential-TL measurements, performed to check the effect of the mutation on the electron flow within PSII, indicated an alteration in the properties of the Q B binding site that affects the binding of ligands, resulting in changes of the midpoint potential of the Q B /Q A site. The effect of the mutation is expressed not only in the presence of DCMU occupying the Q B site but also in its absence as indicated by the alteration from the normal 2/6 (63) to the unusual 1/5 oscillation pattern. The above changes may result from an aberration of the theoretical 3:1 ratio of S 1 :S 0 states and/or the 1:1 occupancy of the state populations Q B Ϫ:Q B in dark-adapted samples (51,64). This in turn may reflect alterations in the midpoint potential of the Q A site.
The down-shift in the emission temperature of the DCMUinduced Q band indicates an accelerated charge recombination from Q A Ϫ to S 2 and, thus, a partial inhibition of back electron flow from Q B Ϫ and charge recombination via P 680ϩ that will affect the synchronization of the Q B Ϫ/S 2 transition of the PSII population. In conclusion, in the absence of PsbI the structural dynamics of the Q B binding site during light excitation may be destabilized, possibly affecting the Q A /Q B midpoint potential and, thus, back and forward electron flow of PSII. Therefore, the properties of electron transport appear modified by alterations in catalytic rather than changes in regulatory characteristics.
PsbI Is Required for Phosphorylation of PSII Core Proteins and a Reduced Plastoquinone Pool in the Dark Causes Phosphorylation of LHCII in the Dark in ⌬psbI-Phosphoproteins of the RC, i.e. D1, D2, and CP43, have been reported to depend on the protein kinase STN8 in Arabidopsis, whereas those of the LHCII depend on STN7 and STT7 in Arabidopsis and Chlamydomonas, respectively (9 -11). However, the direct targets of the kinases and whether the orthologous proteins in tobacco exert the same functions remain elusive.
Strikingly, phosphorylation of LHCII was regulated in response to light in the mutant, but regulation was reverse to the WT behavior when dark and light adapted probes were compared. Irrespective of chosen light conditions (10, 40 (state II), and 500 mol of photons m Ϫ2 s Ϫ1 , 650 nm), LHCII phosphorylation was abolished by light, indicating an imbalance of electron transport causing excitation re-distribution between the two photosystems as also shown by 77 K measurements (Table 1).
It has been suggested that cytochrome b 559 functions in the re-oxidation of the plastoquinone pool in dark-adapted leaves (65). Therefore, it is conceivable that PsbI also supports this function and that the plastoquinone pool remains predominantly reduced in the dark (Table 1). This may cause a redoxregulated activation of the corresponding kinase(s) in the dark in ⌬psbI. Our data unequivocally demonstrate that phosphorylation of the LHCII in the dark is not due to structural changes of PSII in the mutant but due to the reduced plastoquinone pool (Fig. 8F). Unlike WT, the increased re-oxidation rate of the plastoquinone induced by PSI as compared with the lower PSIIdependent reduction rate in ⌬psbI leads to oxidation of the plastoquinone pool under any light conditions. This explains the light-induced inactivation of the kinase and the dephosphorylation of LHCII.
Remarkably, ⌬psbI lost its capability to efficiently phosphorylate PSII-RC proteins. Phosphorylation of LHCII proteins is regulated and induced by duroquinol, but that of D1 and D2 cannot be induced under all chosen conditions. Therefore, we conclude that physiological responses cause the reverse regulation of LHCII phosphorylation but that structural alterations of PSII result in the loss of D1/D2 phosphorylation in the muta-tion. It is reasonable to assume that the PsbI protein allows a close contact of the kinase to the RC presumably by a direct interaction. It is also likely that the lack of phosphorylation of the core complex subunits D1, D2, and CP43 in ⌬psbI partially causes the increased light sensitivity since nuclear Arabidopsis mutants defective in phosphorylation of reaction center proteins show a slightly pale phenotype and a somewhat increased photosensitivity (10 -11). The fact that phosphorylation of CP43 can be induced by duroquinol but that of D1 and D2 cannot either implies that an earlier unidentified protein kinase could be involved in the phosphorylation of the PSII-RC and/or that access of the kinase to CP43 is favored as compared with D1 and D2.