In agreement with previous results on lut2 mutant 14, lut2.1 plants showed similar organ size compared to WT plants, but a slightly lower Chl content per fresh weight and leaf surface. When analyzed for their pigment composition [see Additional file 1] it appeared that the Chl a/b ratio was higher in lut2.1 with respect to WT as was the Chl/Car ratio. Lutein was completely absent from the mutant; a strong compensatory increase of violaxanthin was observed. WT dark-adapted plants did not contain any antheraxanthin or zeaxanthin which were, instead, found in lut2.1 leaves to low, but detectable amounts 14. When exposed to strong light for 20 min, lut2.1 plants accumulated A+Z to levels approx 3 times higher than WT. In agreement with previous results 14, the quantum yield of PSII photochemistry (F_v/F_m chlorophyll fluorescence ratio) was not significantly different in lut2.1 with respect to WT. However, we found that the fluorescence quantum yield of Chl in dark-adapted plants was always lower in lut2.1 with respect to WT [see Additional file 2]. This observation suggests that some kind of constitutive thermal dissipation mechanism, resulting in the quenching of chlorophyll fluorescence, is activated in lut2.1 chloroplasts. According to 11, NPQ was higher in WT with respect to lut2.1 leaves [see Additional file 6]. The two genotypes differ for the initial rate of qE, which is much slower in lut2.1. The PSII antenna size was determined by measuring the half time in the rise of chlorophyll fluorescence in the presence of the photosynthetic electron transport inhibitor DCMU 15. The half time was 65 ms in WT vs. 81 ms in lut2.1, suggesting that the functional antenna size was 20% smaller in the mutant [see Additional file 2]. These results support suggestions by Lokstein et al. 12 based on different methods.

The antenna sizes of PSI and PSII adapt to light quality by phosphorylating LHCII. Upon phosphorylation, this complex is disconnected from the PSII reaction center and diffuses to PSI complexes, where it increases light harvesting and electron transport capacity. This mechanism has been called state transition (see 16 for a review). We assayed the capacity for performing State I – State II transitions by measuring the increase in oxygen evolution when a far red light was superimposed to a background of blue-green light (Emerson effect). The state transition phenomenon was clearly visible in WT, with the Emerson effect being low in state II (ca. 5.5%, indicating an almost even distribution of the blue-green light energy between PSI and PSII) and high in state I (ca. 30%, indicating a strong imbalance in light energy in favor of PSII). In lut2.1, the change in the Emerson effect was very small, indicating that the capacity for change in antenna size of PSI through state I – state II transitions was severely impaired (Table 1). To our knowledge, this is the first evidence for a specific need of lutein in the mechanism of state transitions.

Reduced stability of LHCII trimers has been previously reported in the lut2 mutant 12. Such phenotype could be, in principle, due to the altered pigment composition, or to altered protein composition of the complexes, or both. Thus, we decided to further these observations using sucrose density gradient fractionation of solubilized thylakoids, followed by SDS-PAGE of the fractions, and HPLC analysis of the pigment content of the fractions. The results of the fractionation are shown in Figure 1A. Five bands are visualized in the WT: Band 1 is yellow and contains free carotenoid pigments; band 2 contains the minor antenna complexes CP24, CP29 and CP26, and LHCII monomers; band 3 contains LHCII trimers; band 4 contains the LHCII-CP29-CP24 complex; band 5 contains the PSII core complex; and band 6 the PSI-LHCI complex. Mutant thylakoid membranes show the complete absence of band 3 (trimeric LHCII), while band 2 (monomeric LHCII) is much more represented than in WT. Upon normalization to the Chl content of the PSII core complex band, the Chl content associated to Lhc proteins in band 2+3 is lower in lut2.1 by approx. 10%, in agreement with the smaller functional PSII antenna size indicated by our fluorescence measurements and a previous report 12, while that associated to the PSI-LHCI complex is unchanged. SDS-PAGE analyses show that band 2 from lut2.1 contain the same polypeptides as the corresponding band from WT, although the relative amount of the Lhcb1-3 polypeptides, components of LHCII, is increased (Figure 1B). Overall, the data confirm that LHCII is present in the lut2.1 mutant but its aggregation state is monomeric rather than trimeric 12.

HPLC analyses of bands 2 and 3 (Table 2) indicate that V, A and Z are associated to the Lhcb proteins in band 2 of lut2.1, while in WT only V, N and L are found in bands 2 and 3.

We then asked if the lack of lutein and its substitution by violaxanthin in Lhc proteins, per se, was the actual reason for LHCII monomerization in lut2.1. In order to verify this point, we used recombinant Lhcb1 protein, overexpressed in bacteria, for reconstitution with different xanthophyll species plus Chl a and Chl b. Refolded proteins were then separated from free pigment by Ni²⁺ column chromatography and fractionated by sucrose gradient ultracentrifugation in order to resolve different aggregation states. The results (Figure 1C) indicate that Lhcb1 reconstituted with a mix containing all pigments, as well as the complex with lutein only, did produce trimers. Conversely, if violaxanthin was supplied in the absence of lutein, a violaxanthin-binding complex was obtained which did not produce trimers. For the first time, our measurements show that the binding of lutein per se is sufficient for LHCII trimerization, and that violaxanthin cannot substitute for lutein in this function.

Lutein binds to specific sites within LHCII complexes 17, termed sites L1 and L2, while neoxanthin binds to site N1 and V+A+Z to the external site V1 18. Different binding sites provide slightly different protein environments, which are reflected in different shifts of the absorption maxima of the bound xanthophylls 19 (see legend to Table 3). Thus, it is possible, by applying a spectral deconvolution analysis, using spectral forms of Chl and carotenoids in protein environment 20 to deduce the protein environment in which a carotenoid is bound. The complete data set for spectral deconvolution is given [see Additional file 8], while relevant results are summarized in Tab. 3.

Since, in LHCII monomers from lut2.1, lutein is completely substituted by violaxanthin, we asked if this xanthophyll occupies the same sites L1 and L2 occupied by lutein in the WT. We used for this analysis IEF-purified LHCII proteins, in which the external V1 site is empty 19. The results are summarized in Table 3. The low amplitude Viola spectral form at site L2 (492 nm) 18 is maintained in lut2.1 with a 4-fold higher amplitude, meaning that this site is now completely occupied by violaxanthin. A new violaxanthin spectral form, with a similar amplitude and an unusually high red-shift (505 nm) appears at site L1. Neoxanthin spectral forms and energy transfer are instead unaltered in lut2.1 with respect to WT. Both violaxanthin spectral forms in lut2.1 show high efficiency of energy transfer (80–90%) to Chl a. Since energy transfer is strongly influenced by the pigments' mutual distance and orientation, these data strongly suggest that the two violaxanthins occupy, in lut2.1, the L1 and L2 sites. The unusually high red-shift and energy transfer efficiency of Viola at site L1 is probably due by the "unnatural" binding of this pigment at this site, normally occupied by lutein.

In order to identify a possible effect of the altered pigment composition on the stability of Lhc proteins, we measured the heat denaturation dependence of the major CD signal at 492 nm 2122 [see Additional file 7]. In band 2 from lut2.1 and WT, two inflection points showing essentially the same values were found, suggesting that both LHCII and minor Lhcb complexes had, on the average, the same stability to heat denaturation, irrespective of whether they bound violaxanthin or lutein. In order to distinguish between the contributions of individual Lhc gene products to the above determination, the band 2 from WT and lut2.1 was fractionated by preparative IEF and the fractions analyzed for polypeptide composition [see Additional file 9]. Fractions containing the same Lhcb apoproteins, as determined by SDS-PAGE, were analyzed for their stability to heat denaturation and their pigment composition [see Additional file 3]. It clearly appeared that not only LHCII, but also other Lhcb proteins folded correctly and showed unaltered stability when violaxanthin was substituted for lutein. The Chl a/b ratio was significantly lower in LHCII isoforms from lut2.1 with respect to WT, while IEF bands with less acidic pI, enriched in minor Lhc proteins, were less affected in their Chl a/b ratio.

Strong illumination of chlorophyll-proteins in the presence of oxygen leads to ³Chl* formation, which reacts with molecular oxygen forming ¹O₂*. Singlet oxygen causes bleaching of Chl with kinetics inversely dependent on the efficiency of chlorophyll triplet quenching by bound xanthophylls. The photobleaching behavior of pigment-proteins from sucrose gradient bands (Figure 1A) was determined as previously described 9. The results are shown in Figure 2A. The highest resistance was found in WT band 3, containing trimeric LHCII, while band 2, containing mostly minor Lhcbs, was more prone to photobleaching in agreement with previous findings 23. In the case of band 2 from lut2.1, the resistance to photobleaching was, surprisingly, only slightly higher than in the case of WT, although the LHCII content was much higher (the LHCII/minor antennae ratio was 2.5 in band 2 from WT and 3.7 in lut2.1, see Figure 1B). This suggests that either the presence of violaxanthin, rather than lutein, within these proteins, or the monomerization of LHCII, caused a decreased resistance to photobleaching. To clarify this point, we analyzed the photobleaching behavior of monomeric LHCII from WT and lut2.1 purified by IEF (Figure 2B). The lut2.1 complex was clearly more sensitive to photobleaching than that from WT. An increase in resistance to photobleaching was detected in trimeric LHCII from WT with respect to the monomeric form, thus indicating that trimerization per se contributes to photoprotection. Into LHCII, site L1 was shown to be essential for ³Chl* quenching and consequently for protection from photobleaching in the presence of oxygen, while site L2 had little relevance in this respect 9; therefore, we conclude that the reduced resistance to photobleaching is due not only to the monomerization of LHCII subunits, but also to the substitution of lutein in site L1 by violaxanthin.

In order to further substantiate this conclusion, we performed direct measurements of the kinetics of carotenoid triplet formation and triplet chlorophyll quenching by time-resolved spectroscopy of lutein- vs. violaxanthin-containing monomeric Lhcb1 proteins. Time-resolved absorbance changes were recorded, subsequently to chlorophyll excitation at 650 nm. Consistent with previous results 9 recombinant proteins binding violaxanthin showed faster photobleaching than those binding lutein (not shown). The data shown in Figure 3 refer to in vitro reconstituted, recombinant proteins. ³Car* formation and decay can be followed as the changes in absorbance at 505 nm, while ¹Chl* gives a negative signal at 440–460 nm (panels B and D, see Experimental Procedures for a detailed discussion of the spectral deconvolution procedure). Spectra measured on lutein- and violaxanthin-containing Lhcb1 gave similar half-times for ³Car* decay (2–2.5 μs) but evidenced a rise-time for violaxanthin triplet (~50 ns) slower than for lutein (~20 ns). Analysis of purified monomeric LHCII proteins purified from WT and lut2.1 membranes by IEF yielded similar results (data not shown).

The biochemical data suggest a deficit in the efficiency of photoprotection at the level of Lhcb proteins, particularly LHCII, in the lut2.1 mutant, caused by the substitution of lutein with violaxanthin in site L1. It can thus be expected that growth at high light intensity may reveal additional features of the lut2.1 phenotype. WT and lut2.1 plants were grown for 3 weeks in control conditions (120 μmol m^-2 s^-1) at 21°C and then either exposed to high light (1400 μmol m^-2 s^-1) or grown at the same light intensity for three additional weeks (Figure 4A). After treatment, leaves were analyzed for pigment composition [see Additional file 4] and thylakoid protein composition (Figure 4B–C). Growth in high light produced damages consisting into reddening and bleaching of older leaves. The damages were more pronounced in mutant plants.

Thylakoid membranes were isolated from low- and high-light grown plants and analyzed by SDS-PAGE (Figure 4B–C). The relative abundance of thylakoid proteins was evaluated by densitometry of Coomassie-stained gels upon identification of individual selected bands by immunoblotting with specific antibodies (not shown). Both WT and lut2.1 thylakoids showed a decrease in the LHCII/PSII ratio in high light, as evaluated by the level of the 33 kDa oxygen evolving complex 1 polypeptide (Figure 4B). WT plants decreased their content in Lhcb1+2 polypeptides upon growth in high light by 15% with respect to control plants while other Lhcb proteins were marginally affected. lut2.1 plants showed a similar effect, but the amplitude of the decrease in LHCII was much higher, suggesting that mutant plants over-react to increasing light by degrading their major antenna complex and thus avoiding photoinhibition (Figure 4C).

In agreement with previous results 12 the Chl a/b ratio increased in WT and lut2.1 with respect to control conditions, the amplitude of the change being higher in the mutant. lut2.1 had increased Chl a/b ratios even in control conditions. Growth in high light decreased the Chl/Car ratio in WT and lut2.1. WT plants did not contain any A+Z in low light, and low levels in high light conditions. lut2.1 plants contained low, but detectable levels of A+Z in low light conditions 14, and their increase in high light was 8 times higher than in WT plants. Although the increase in A+Z was the highest, all carotenoid species increased their relative amount with respect to Chls. This effect was stronger in lut2.1 with respect to WT plants [see Additional file 4].

Our results strongly suggest that lut2.1 plants are affected in their capacity to prevent photooxidation of their antenna system, due to the lower efficiency of violaxanthin, with respect to lutein, in quenching ³Chl*. Growth in low temperature conditions should enhance the amplitude of photodamage 24. We have thus evaluated the effect of growing plants at 4°C at either low (20 μmol m^-2 s^-1) or high light conditions (800 μmol m^-2 s^-1). The experiment was performed on WT, lut2.1, npq1 (previously shown to have a decreased resistance to oxidative stress under light stress conditions 6) and the double mutant npq1lut2.1. While WT and lut2.1 are able to increase A+Z content at the expense of Viola upon light treatment, npq1 and npq1lut2.1 plants cannot [see Additional file 1].

Plants were grown at 120 μmol m^-2 s^-1, 21°C for three weeks (t_o) and then transferred at 4°C at either low light or high light for three additional weeks. In low light, none of the genotypes showed an evident stress effect, while in high light, plants were affected to different extents (Figure 5): in WT, older leaves showed photobleaching accompanied by accumulation of anthocyanin, an indicator of stress in Arabidopsis 2526. These symptoms were much stronger in npq1 and lut2.1 mutants, extending to the younger leaves, while many of the older leaves were almost completely bleached. Consistently with previous reports 27, the npq1lut2.1 genotype was more light-sensitive than either npq1 or lut2.1, suggesting that the lack of zeaxanthin exacerbates the photodamage induced by the lack of lutein. More quantitative analyses were performed on detached leaves, choosing leaves that remained green over the entire period of the experiment [see Additional file 5].

Plants of WT and mutants, grown in standard conditions (120 μmol m^-2 s^-1) were treated for 30 hours at high light and low temperature (1100 μmol m^-2 s^-1, 8 hours light photoperiod, 8°C). Following stress, the level of photoinhibition was assayed by chlorophyll fluorometry (F_v/F_m) (Figure 6A), while lipid peroxidation was quantified by measuring leaf chemiluminescence 2829 (Figure 6B). Our results clearly show that the highest levels of lipid peroxidation and photoinhibition were obtained in the npq1lut2.1 genotype, in accordance with evidences obtained on C. reinhardtii lor1npq1 double mutant 30; npq1 had intermediate levels and lut2.1 did not show a significant difference from WT. Similar results were obtained in a shorter experiment in which detached leaves, floating in water at 10°C, were treated at high light (1100 μmol m^-2 s^-1) for 20 h (data not shown).

We analyzed 20000 independent T-DNA insertion lines of Arabidopsis thaliana (accession Wassilewskija-2) available from the Institut National de la Recherche Agronomique (INRA, Versailles, France). Four hundred seed pools, each representing one row or one column of the grid (100 lines), were prepared and then combined into 80 superpools, each representing 500 lines.

The presence of T-DNA insertion in the lycopene ε-cyclase gene (lyec) was assessed by PCR amplification on DNA from each of the superpools, followed by a nested PCR. The first (10 cycles: 94° 2'; 65°-1°/cycle 30"; 72° 2'. 35 cycles: 94° 15"; 55° 30"; 72° 1') was performed by using the lyec specific primers 5'-AGTTAGTCGACGTTTGCTCCATG-3' and 5'-CAATGGTAATAGGCTTGTCATC-3' and the T-DNA specific primers 5'-CTACAAATTGCCTTTTCTTATCGA-3' and 5'-CTGATACCAGACGTTGCCCGCATAA-3' The nested PCR (35 cycles: 94° 45"; 56° 45"; 72° 1') was performed by using the lyec specific primers 5'-GAGGAGGTAAAGTATGGTTCCAC-3' and 5'-CTCTCTCCAAACATGCTCAATAC-3' and the T-DNA specific primers 5'-CATGTACATCAAGCTTATCGATAC-3' and 5'-TACGAATATCTGCATCGGCGAAC-3'. One mutant line, containing a T-DNA insertion in the sixth exon of the gene, was identified in the superpools and then in the pools, and kindly provided by INRA. To identify homozygous lines, PCR analyses was performed using primers 5'-AAGCTTCTTCCGTACTTTC-3' and 5'-CAATCGTAAACAATATAAGCG-3', flanking the site of insertion, and the T-DNA specific primer 5'-CATGTACATCAAGCTTATCGATAC-3'. This mutant will be indicated as lut2.1 in order to avoid confusion with the original lut2 mutant 14 solely to indicate that it is made in Wassilewskija ecotype. After completion of this work we have obtained the same mutation in Columbia ecotype and verified that it showed the same behaviour as the lut2.1 with respect to sensitivity to light stress.

WT plants of Arabidopsis thaliana ecotype Wassilewskija and mutants npq1 57, were obtained from the Arabidopsis Stock center. Genotype npq1 lut2.1 was obtained by crossing single mutant plants. Plants were grown for three weeks in controlled conditions (~120 μmol m^-2 s^-1, 21°C, 8 h light/16 h dark). For long term treatment, 3 weeks old seedlings were exposed (a) to light conditions of 120 or 1400 μmol m^-2 s^-1 for 3 weeks at 21°C, and (b) to light conditions of 20 or 800 μmol m^-2 s^-1 for 3 weeks at 4°C. Short-term high light treatment was performed for 20 minutes at 1200 μmol m^-2 s^-1.

Chlorophyll fluorescence from intact leaves or from leaf discs was measured with a PAM-2000 fluorimeter (Walz), as previously described 37. The maximal quantum yield of PSII photochemistry was measured in dark-adapted leaves from the maximal fluorescence level (F_m) and the initial level (F_o): (F_m-F_o)/F_m = F_v/F_m. Variable fluorescence was induced in leaf discs, infiltrated with DCMU 2.5 10^-5 M, with a red light of 8 μmol m^-2 s^-1 produced by a light emitting diode. The half-time of the fluorescence rise was taken as a measure of the functional antenna size of PSII 15. Photosynthetic O₂ evolution was measured with the photoacoustic method, as described by Havaux et al., 37 (for a review, see 58). The Emerson enhancement (E) of O₂ evolution was determined in state I or in state II by adding a continuous far-red light (>715 nm, 34 W m^-2) to the modulated blue-green light (obtained with a BG38 Schott filter; photon flux density 24 μmol m^-2 s^-1). E (%) = [(Φ (+FR) - Φ (-FR))/Φ (-FR)] × 100, where Φ (+FR) is the amplitude of oxygen evolution signal in the presence of the far-red light and Φ (-FR) is the signal measured with the modulated exciting light only. State II was reached by illuminating leaves with blue-green light for 10 min, and state I was obtained after 10-min illumination with far-red light.

Non-photochemical quenching of chlorophyll fluorescence was measured with a PAM 101–103 fluorimeter (Walz). NPQ was calculated according to the following equation 59: NPQ = (F_m-F'_m)/F'_m, where F_m is the maximum Chl fluorescence from dark-adapted leaves and F'_m the maximum Chl fluorescence under actinic light exposition.

Unstacked thylakoid membranes were isolated from dark-adapted leaves as previously described 39.

Membranes corresponding to 500 μg of chlorophylls were washed with 5 mM EDTA and then solubilized in 1 ml with 0.6% α-DM, 10 mM HEPES pH 7.5. Solubilized samples were then fractionated by ultracentrifugation in a 0.1–1 M sucrose gradient containing 0.06% α-DM, 10 mM HEPES pH 7.5 (22 h at 280,000 × g, 4°C).

Monomeric Lhcb proteins were further fractionated by flat-bed isoelectric focusing at 4°C as previously described 60.

The pigments were extracted either from whole leaves, thylakoid membranes and isolated antenna complexes with 80% acetone, then separated and quantified by HPLC 61 and by fitting of the spectrum of the acetone extract with the spectra of individual pigments 21.

SDS-PAGE analyses was performed with the Tris-Tricine buffer system as previously described 62. Gel images were acquired using a Bio-Rad GS710 scanner. The picture was then analysed with GEL-PRO ANALYZER software (Media Cybernetics Inc., MD, USA) that quantifies the staining of the bands as IOD (optical density integrated on the area of the band).

The construct over-expressing Lhcb1 was obtained as described 63, except for a sequence coding for a His₆ tail inserted at the 3' end before stop codon. In vitro reconstitution of LHCII with altered xanthophyll composition was performed as previously described 18. Purification of reconstituted holocomplexes was performed by a Ni²⁺ chelating column (Pharmacia Source 15S) as described 9. Reconstitutions were accomplished with a mix of xanthophylls as follow: LHCII Viola, 100% violaxanthin; LHCII Lute, 100% lutein; the Chl a/b ration in the mixture was 2.3. In order to perform in vitro trimerization of reconstituted complexes, we followed a method previously described 64 with some modifications. Ni²⁺ column and bound reconstituted LHCII were washed with trimerization buffer: 0.1 mg/ml PG, 0.06% β-DM, 0.2 M NaCl, 20 mM phosphate buffer pH 7.5, 10 mM imidazole. Pigment-protein complexes were collected by washing column with eluting buffer: 0.5 M imidazole, 20 mM phosphate buffer pH 7.5, 0.2 M NaCl, 0.06% β-DM; LHCII trimers were separated from monomers by sucrose gradient ultracentrifugation 19.

Steady state spectra

Time-resolved spectroscopy:

This was analyzed as described 21 by following the decrease of the 492 nm CD signal detected by a Jasco 600 spectropolarimeter at increasing temperatures, from 20 to 80°C (scan rate 1°C/minute and step 0.2°C). Protein thermal stability was measured as denaturation T°_1/2, determined by inflection point of CD denaturation curves.

The kinetics of antennae photobleaching was measured as described 18 but with a higher light intensity of ca 6000 μmol m^-2s^-1 and sample cooling at 10°C. Initial and maximal absorbance was 0.6.

Photoxidative stress was induced in whole plants by a strong light treatment at low temperature. Whole Arabidopsis plants were exposed to high light (1100 μmol photon m^-2 s^-1 with a photoperiod of 8 h) at low temperature (7°C/8°C, day/night air temperature), as described previously 66.

Photoinhibition of PSII was measured by chlorophyll fluorometry (F_v/F_m ratio) with a PAM-2000 fluorimeter (Walz). Photoxidative stress was measured by thermoluminometry with a custom-made apparatus that has been described 7. The amplitude of the TL peak at 135°C was used as an index of lipid peroxidation.