Isolation of mutants with altered GPXHexpression

The expression of the GPXH gene is strongly induced by the increased production of ¹O₂ in the chloroplast. To identify components affecting ¹O₂-induced gene expression in C. reinhardtii, a mutant screen was performed using the GPXH-arylsulfatase (GPXH-ARS) reporter gene construct pYSn1 to search for clones with altered GPXH response [21]. The wild-type strain 4A⁺ transformed with pYSn1 was UV-mutagenized and colonies were grown on TAP plates in the dark. Then, a total number of 5500 clones were analyzed for their GPXH-ARS expression under medium light (ML) condition of 80 μmol photons m^-2 s^-1 in the presence or absence of 1 μM NR. Average GPXH-ARS expression was induced 6.0 ± 2.0 fold by NR treatment. A cutoff of 2.5-fold induction by NR and 2.2 fold higher expression was used to select for clones with reduced induction or increased basal expression, resulting in 22 GPXH-ARS induction deficient (gid) and 41 GPXH-ARS overexpressor (gox) mutants (Figure 1A). However, after retesting these clones, only six gid and 32 gox mutants could be confirmed.

The remaining 20 mutants, being either HL-sensitive and/or t-BOOH resistant, were then tested for the expression of the endogenous GPXH wild-type gene by qPCR. A significantly (P < 0.05) stimulated expression compared to the corresponding wild-type strain could be detected in seven clones ranging from 1.4- to 5.5-fold overexpression (Additional file 1). Even though there was a clear correlation (R² = 0.76) between GPXH expression and the expression of the reporter construct, all 20 mutants had a stronger overexpression of GPXH-ARS measured by enzyme activity than GPXH expression determined by qPCR, which might be the consequence of individual mRNA or the reporter enzyme stability (Figure 1B).

High doses of UV radiation as applied in this experiment induce multiple point mutations in the genome. To analyze whether defects in multiple genes might be responsible for the phenotype, we performed tetrad analysis of selected gox mutants by crossing them back to the strain 4A^- pYS1. Twelve independent tetrads were tested for segregation of high GPXH-ARS expression in each mutant. A clear 2:2 segregation of the wild-type and mutant phenotypes was found in strains 35H11 and 18F6 (Figure 2), indicating that mutations in a single nuclear gene is responsible for the increased GPXH expression in each case. The same was true for 21B4 except for one tetrad where one high, two medium and one low expressing progeny were found, suggesting that two closely linked mutations might be responsible for the high ARS activity phenotype. No consistent 2:2 segregation was found in backcrosses of strain 22D1 and 14A9. Whereas for 22D1 at least six tetrads resulted in either 3:1 or 1:2:1 segregations, for 14A9 the pattern of three tetrads differed from standard single allele segregation.

GPXH overexpression in goxmutants due to increased singlet oxygen production

Stimulated GPXH expression might either be due to a stimulated production of ¹O₂ or a constitutively active signaling pathway under LL condition. Sensitivity to HL intensity of several mutants indicates that the former might be the reason for high GPXH expression in some of the gox mutants. The formation of ¹O₂ was therefore measured with the fluorescent dye singlet oxygen sensor green (SOSG) allowing the specific quantification of ¹O₂ [26]. In order to detect significant amounts of ¹O₂ with the membrane impermeable SOSG in the wild-type strain, the cells had to be broken by freezing, and exposed to HL intensity (500 μmol photons m^-2 s^-1) for a period of 15 min. Increased ¹O₂ formation compared to wild-type (at least 1.8 fold) was detected in all but one of the HL-sensitive mutants (Figure 3 Additional file 2). Surprisingly, several mutants with normal resistance to HL also showed a stimulated ¹O₂ production even though the difference to wild-type was not always significant (P < 0.05).

Comparing ¹O₂ formation with GPXH overexpression in gox mutants, no direct correlation (R² = 0.04) between the two parameters could be found (Figure 3A). However, since the formation of ¹O₂ is a light intensity-dependent photoreaction and GPXH expression and ¹O₂ formation were quantified at different light intensities (80 and 500 μmol photons m^-2 s^-1), it is difficult to directly compare these parameters. We therefore measured GPXH expression in all 20 gox mutants at the highest possible light intensity at which the mutants could still survive for at least 24 h (250 μmol photons m^-2 s^-1) and which thus corresponds to HL conditions. Indeed, a much stronger stimulation of GPXH expression compared to wild-type could be detected in more than half of the mutants, resulting in a stronger correlation (R² = 0.48) between ¹O₂ formation and GPXH expression (Figure 3BAdditional file 1).

Increased levels of ¹O₂ in gox mutants are a consequence of either increased generation or lowered detoxification of ¹O₂. The former might result from a deficient photoprotection mechanism, such as a reduced capacity of NPQ. Therefore, NPQ was measured in the gox mutants acclimated to either LL (15 μmol photons m^-2 s^-1) or HL (250 μmol photons m^-2 s^-1) for at least 24 h. Four mutants (22D1, 18C2, 14A9 and 21B4) had significantly reduced NPQ under both LL and HL conditions (Figure 4A). Four other strains (15B10, 18G9, 14B5 and 14C11) only had lower NPQ under HL but not LL conditions, suggesting a light intensity dependent effect, and for three mutants (18F6, 14H8 and 18G9) a stimulation of NPQ even at LL conditions was found.

Energy-dependent quenching (qE) is one component of NPQ that requires synthesis of the xanthophylls, zeaxathin and antheraxanthin. However, these and other carotenoids as well as α-tocopherol are also important antioxidants involved in detoxification of ¹O₂. Pigment analysis of the 20 gox mutants acclimated to the same LL or HL intensity as for NPQ measurements revealed only small changes in pigment contents of few strains. Lutein was not altered in LL and only slightly higher in three mutants in HL conditions compared to the wild-type strain (Figure 4C). Similarly, α-tocopherol was not strongly affected in LL condition, except for 14H8, but was significantly reduced in 7 mutants after exposure to HL for 24 h (Figure 4D). For only one strain, 18F6, higher levels of α-tocopherol than in wild-type were detected under HL conditions, and this correlated with increased lutein and zeaxanthin levels in this mutant. Despite the important role of zeaxanthin and antheraxanthin in photoprotection and prevention of ¹O₂ generation, only moderate changes of these xanthophylls were detected in the mutants compared to the wild-type. Only one mutant (18G9) had a significantly reduced de-epoxidation of xanthophylls during HL exposure showing that this process seems still to be functional in all mutants (Figure 4B). Nevertheless, when grown in LL, five mutants had significantly reduced de-epoxidation states whereas other mutants had rather increased levels of zeaxanthin and antheraxanthin (Figure 4B and 5B).

GPXHexpression and singlet oxygen production negatively correlate with the xanthophyll de-epoxidation state

In order to analyze the relationship of all the parameters measured in the 20 gox mutants, linear correlation factors were calculated for every possible combination of parameters (Figure 5A). Not surprisingly, a strong positive correlation between antheraxanthin and zeaxanthin levels was found, which negatively correlated with violaxanthin under HL conditions, as expected from operation of the xanthophyll cycle. As already shown in Figure 3B, ¹O₂ formation correlated with GPXH expression under HL conditions. Both parameters also negatively correlated with antheraxanthin and zeaxanthin levels, especially in LL-grown cultures. Thus, when comparing ¹O₂ production at HL and xanthophyll levels at LL it was striking that all the mutants but one (14C11) had a high de-epoxidation state of xanthophylls or increased ¹O₂ production compared to wild-type (Figure 5B). This effect was less pronounced when ¹O₂ production was compared with de-epoxidation at HL because strongly stimulated de-epoxidation reduced the relative differences between the clones. Finally, antheraxanthin and zeaxanthin levels of HL-grown cultures weakly correlated with α-tocopherol and lutein contents under this light condition.

Hierachical clustering of the mutants with all measured parameters was performed to test whether there are groups of mutants with similar phenotypic pattern. These analyses revealed that one mutant (14H8) behaved differently from all other mutants (Figure 5C). Even though 14H8 was originally screened for high GPXH-ARS expression and was found to be HL sensitive, it did not show any stimulated GPXH expression or ¹O₂ production at HL intensities. All other mutants were divided into two major groups (I and II) mainly based on the de-epoxidation state of the xanthophyll pool. Group I had lower levels of antheraxanthin and zeaxanthin at LL than wild-type and most of these mutants showed high ¹O₂ production and increased GPXH expression. They could be further subdivided into two groups, where strain 18F6 and 18G9 belonged to one group (IB) with stimulated NPQ in LL conditions and no changes in α-tocopherol levels under any light intensity. Mutants of the other subgroup (IA), on the other hand, had strongly reduced α-tocopherol levels, especially in HL grown cultures, and either lowered or not changed NPQ compared to the wild-type strain.

The second group of mutants (II) with similar or slightly increased xanthophyll levels to wild-type had generally a much lower stimulation of ¹O₂ production and GPXH expression than mutants of group I. Exceptions were clones 14A9 and 21B4 from subgroup IIA with strong GPXH overexpression and stimulated ¹O₂ production, but these mutants showed strongly reduced NPQ under both LL and HL intensity that was probably responsible for these phenotypes. In the other subgroup (IIB) only two mutants (35H11 and 15B10) had significant GPXH overexpression and higher ¹O₂ production and except for lower α-tocopherol levels in HL for some mutants, little differences compared to wild-type were detected for most parameters.

No mutants with deficient GPXHinduction

Screening for mutants with altered GPXH expression resulted in the isolation of several high GPXH expression mutants but no strain with deficient or strongly reduced GPXH induction by ¹O₂ could be isolated. With a total number of 5500 UV-mutagenized clones tested we assume that the coverage of mutations in the genome should be high enough to hit at least one component of a putative ¹O₂ signaling cascade with a loss-of-function mutation. The relatively high UV dose resulting in the survival of only 0.4 to 2% of the cells and the fact that during segregation analyses at least two out of five gox mutants contained two mutations affecting the phenotype suggested a high mutation density in the screened population. Still, a second mutant screen for suppressor mutants of GPXH overexpression in strain 21B4 was performed using an additional plasmid containing the GPXH promoter in front of the nitrate reductase gene (data not shown). This enabled a direct selection for reduced expression under low light conditions by selecting for chlorate-sensitive clones. Out of a total number of 2 × 10⁶ cells surviving UV mutagenesis, 1060 clones were chlorate resistant, and of those, 15 clones also had low induction (< 2 fold) of the GPXH-ARS reporter construct by NR. However, all of these clones showed normal induction of the wild-type GPXH gene, indicating that under these screening conditions no gid mutants can be isolated. The conclusion that in strain 21B4 a constant increased ¹O₂ production is responsible for high GPXH expression might explain why no GPXH induction deficient mutants could be isolated, because GPXH might be essential for defense against ¹O₂-induced damage. This can be excluded for the original mutant screen because selection against light-sensitive mutants during the recovery phase after mutagenesis was prevented by growing the 5500 clones in the dark. Furthermore, none of the clones was light sensitive under ML conditions during the induction tests. Thus, even though steps were taken to minimize a negative selection against gid mutants, we cannot exclude that a functional ¹O₂ response of the GPXH and maybe other genes is essential for C. reinhardtii. On the other hand, it is also possible that several redundant signaling pathways form a complex network to activate GPXH expression, thereby hampering the isolation of gid mutants. Finally, it still might be that more dark-grown mutagenized clones had to be tested to find the desired mutants. A similar screen to isolate ¹O₂ responsive mutants in A. thaliana resulted in the identification of at least three mutants deficient in the upregulation of different ¹O₂-induced genes in the conditional flu mutant [27]. However, ¹O₂-signaling seems to have different cellular functions in A. thaliana, at least in seedlings, where it is part of a programmed cell death response and in C. reinhardtii where it seems to be involved in response to cytotoxic environmental stresses [20, 28].

GPXHoverexpression and correlation with singlet oxygen production, NPQ and pigment levels

In contrast to the lack of gid mutants, many gox mutants with a stimulated expression of the GPXH wild-type gene, especially under HL intensities, could be isolated (Figure 3). The light intensity-dependent increase of GPXH overexpression and the sensitivity to HL conditions indicated that in several mutants ¹O₂ formation might be enhanced and cause a photooxidative stress. Indeed, increased ¹O₂ formation was measured in all but one HL-sensitive as well as many HL-resistant gox mutants (Additional file 2), which nicely correlated with GPXH expression under HL conditions (Figure 3B). This indicates that in most or even all gox mutants, GPXH overexpression seems not to be caused by a constitutively active signal transduction pathway but by the increased production of ¹O₂ under the light conditions tested. The poor correlation between ¹O₂ production and GPXH expression under ML (Figure 3A), the clustering in at least five phenotypic distinct groups (Figure 5C) and the fact that three of the mutants (18F6, 21B4 and 35H11) have been mapped to different linkage groups (data not shown) indicates that mutations in different nuclear genes are responsible for the increased ¹O₂ production in the various mutants.

Singlet oxygen formation in photosynthetic organisms is caused by the conversion of excited chlorophylls into the triplet state and the reaction with molecular oxygen [3]. The organisms try to minimize this process by regulating the excitation pressure on the PSII reaction center chlorophylls, optimizing the electron flow in the photosynthetic electron transport chain and quenching chlorophyll triplet states and ¹O₂. Thus, ¹O₂ accumulation can result from either stimulated production, e.g. due to enhanced triplet chlorophyll formation, or lowered detoxification of the ROS due to defects in some light-induced protection mechanisms. Such a defect in protection mechanisms could be the reason for high ¹O₂ formation in most of the group I mutants of cluster analysis (Figure 5C) having reduced zeaxanthin and antheraxanthin levels which correlates with high ¹O₂ formation (Figure 5B). A well characterized mutants with reduced xanthophyll levels and defects in photoprotection is the npq1 lor1 double mutant lacking the carotenoids zeaxanthin, antheraxanthin and lutein. Similar to many mutants of group I, this mutant exhibits increased ROS production, GPXH expression and sensitivity upon HL but not LL illumination [16, 29]. However, comparison of npq1 and npq1 lor1 showed that reduced levels of zeaxanthin and antheraxanthin but normal levels of other carotenoids like lutein would probably not cause a very strong phenotype [16, 30]. Thus, reduced efficiency of more than one protection mechanism might be required to achieve increased ¹O₂ accumulation. Indeed, five of the group I mutants with reduced xanthophyll levels (13H11, 14B5, 14C11, 21E2, 22D1) also showed lowered α-tocopherol levels under LL and/or HL conditions (Figure 5C, group IA). Furthermore, levels of other components of the thylakoid membrane with ¹O₂-quenching capacity like plastoquinone and α-tocopherolquinone [31, 32] were not quantified but might also be affected in some of the mutants. Thus, reduced photoprotection by the combination of lower xanthophylls, α-tocopherol and maybe other deficiencies might be the cause for increased ¹O₂ production and GPXH expression in these mutants. This could either result from mutations in multiple loci, as found for clone 22D1 (Figure 3), or caused by mutations in a single gene encoding for a global regulator of cellular photoprotection mechanisms.

Even though deepoxidation of xanthophylls is involved in NPQ [33], these parameters did not correlate in our mutants (R² < 0.01) indicating that other factors also play an important role for NPQ. For example, two strains of group I (18F6 and 18G9) showed rather increased NPQ levels under LL conditions even though their xanthophyll levels were reduced, clustering them in a separate subgroup (IB) of group I (Figure 5C). Very high GPXH expression in 18F6 and 18G9 under ML conditions indicates stimulated ¹O₂ generation already at LL. We speculate that these mutants might have increased ¹O₂ production due to an enhanced energy transfer from the PSII reaction center to molecular oxygen. By this, quenching of excitation energy by molecular oxygen would increase NPQ but reduce the photosynthetic electron transport rate required for building up the proton gradient and activating the xanthophyll cycle. Thus, reduced xanthophylls would not be the cause but the consequence of increased ¹O₂ production. However, other effects of the mutations cannot be excluded which might also explain these phenotypes.

Contrary to 18F6 and 18G9, two mutants (14A9 and 21B4 in group IIA) had strongly reduced NPQ both under LL and HL conditions but similar or rather increased xanthophyll levels compared to wild-type showing that defects in qE-independent mechanisms seem to affect NPQ in these mutants. Photoinhibition (qI) stimulated by excess light should not be relevant under LL conditions, and a deficiency in state transition (qT) should not strongly affect NPQ at HL intensities. On the other hand, a qE-independent effect was also found in a C. reinhardtii mutant defective in two linked LHCSR3 genes [10]. However, segregation analysis of backcrosses of 14A9 and 21B4 with a wild-type strain revealed that both mutants have probably mutations in two different genes affecting GPXH expression (Figure 2) indicating that the phenotypes of these mutants could be caused by the combination of different defects.

In mutants with functional detoxification mechanisms, strong ¹O₂ accumulation after stimulated production of the ROS might be prevented by the induction of these detoxification mechanisms. This is supported by the negative correlation of ¹O₂ formation and deepoxidation of xanthophylls, where increased levels of zeaxanthin and antheraxanthin correlate with a low stimulation of ¹O₂ production (Figure 5B). Mutants with increased xanthophyll levels are mainly represented in the group IIB mutants of cluster analysis with rather few and weak phenotypic changes (Figure 5C). A general increase in antioxidant levels including zeaxanthin, antheraxanthin, lutein and α-tocopherol and a significant rise in NPQ was detected in strain 14H8, where no stimulated ¹O₂ production and GPXH expression could be measured any more. This shows that the various protection mechanisms can compensate each other and thus control the production of deleterious ROS. This is in agreement with data of various mutants lacking specific antioxidants: α-tocopherol-deficient strains of C. reinhardtii, Synechocystis sp. PCC6803 and A. thaliana all were very tolerant to photooxidative stress during HL conditions, and only under extreme conditions such as a combination of very HL and low temperature or chemical treatment a phenotype became visible [34–37]. It was suggested that the presence of other antioxidants such as zeaxanthin or increased levels of β-tocopherol can compensate for α-tocopherol deficiency [35, 37]. Conversely, the A. thaliana npq1 mutant, lacking zeaxanthin and antheraxanthin, accumulates higher amounts of α-tocopherol [38]. Thus, zeaxanthin, α-tocopherol and plastoquinol have overlapping functions in photoprotection and together prevent the formation of deleterious ¹O₂ under natural conditions [8, 32, 39–42]. However, when these protection mechanisms are overwhelmed, ¹O₂ starts to accumulate and damage cellular components. This is when defense genes like GPXH, which repair and remove damaged biomolecules, are required to survive the oxidative stress. Activation of genetic stress response without altering antioxidant levels by an acclimation to increased ¹O₂ production or the direct overexpression of the GPXH gene in C. reinhardtii were shown to increased resistance to oxidative stress by RB, NR and t-BOOH [20, 25]. Increased tolerance against t-BOOH was also found for 13 of the 32 gox mutants tested including all the group IIB mutants as well as strains 14B5, 14C11 and 35H11 (Table 1). Thus, increased expression of stress response genes like GPXH might explain the HL resistance of these mutants and shows the important role of GPXH and other defense genes in the photooxidative stress response of photosynthetic organisms.

Strains and growth conditions

The C. reinhardtii strain used to generate the population of mutants was 4A⁺ pYSn1, which is in a 137c (CC-125) background [43]. This strain was generated by co-transformation of 4A⁺ with the plasmids pYSn1 containing the GPXH-arylsulfatase reporter construct [21] and pBC1 containing the Streptomyces aminoglycoside 3'-phosphotransferase typeVIII encoding gene (aphVIII) for selection of transformants on paromomycin [44]. A near-isogenic mt^- strain (4A^- pYSn1) was obtained by crossing 4A⁺ pYS1 with 4A^- [43] and used to backcross mutants for segregation analysis.

All strains were grown heterotrophically in Tris-Acetate-Phosphate-medium (TAP) [45] either on 1.5% agar plates or in liquid cultures agitated on a rotary shaker (120 rpm) at 22°C and the light conditions indicated. For storage, mutants were kept on TAP agar plates in dim light.

Screening for GPXHexpression mutants

UV mutagenesis was performed in an UV Stratalinker™1800 (Stratagene, CA). Cells were grown heterotrophically to a density of 5 × 10⁶ cells ml^-1, and 20 ml were aliquoted into a sterile Pyrex^® petri dish (14 cm diameter) and exposed to 30 - 60 mJ cm^-1 of UV light. Cells were kept in the dark for 1 day immediately following UV treatment to prevent initiation of light-activated DNA repair mechanisms. Then the UV-mutagenized cells were spread on TAP plates and incubated in the dark until colonies appeared.

Relative GPXH-ARS expression was then calculated for each clone by dividing its ARS activity by the average control ARS activity. After the initial selection for altered ARS expression, the clones were rescreened three more times under the same exposure conditions to ensure reproducible changes in the selected mutants.

Testing for resistance phenotypes

Resistance to different oxidative stress conditions was tested by inoculating the clones in 150 μl TAP in a 96-well plate and incubation at ML for 2 days. Then 5 μl of each cultures was spotted on TAP plates containing the following chemicals: tert-butylhydroperoxide (t-BOOH: 100, 150, 200 and 250 μM), neutral red (NR: 4, 5, 8, 12 and 18 μM), rose bengal (RB: 2, 2.5, 3, 4 and 5 μM), metronidazole (MZ: 1, 2, 3, 5 and 8 mM), methyl viologen (MV: 0.5, 1, 1.5, 2 and 3 μM). The plates were incubated at either LL (t-BOOH, MZ and MV) or ML (NR and RB), together with a control plate without any chemical for 3 to 4 days depending on the light intensity. High light resistance was tested by exposing 5 μl of cultures on a TAP plate to 500 μmol photons m^-2 s^-1 (HL) for 3 days.

Segregation analysis

Six mutants (14A9, 18F6, 18G9, 21B4, 22D1 and 35H11) were analyzed genetically for segregation of their phenotypes by crossing them to strain 4A^- pYS1. The resulting zygospores were harvested and dissected as described in Harris (1989). For each mutant a total of twelve tetrads with four surviving cells each were analyzed for their GPXH-ARS expression pattern except for 18G9, which did not result in any viable progenies. GPXH-ARS expression analysis was performed as described above in three replicates per clone.

Singlet oxygen formation

Cultures grown under LL to 3 × 10⁶ cells ml^-1 were adjusted to 3 μg ml^-1 chlorophyll content, and 0.5 ml samples were frozen in liquid nitrogen to lyse the cells. After thawing, 100 μl of cell suspension were transferred to a black 96-well plate. To all the samples including two medium-only controls for background correction, 1 μl of a 1 mM Singlet Oxygen Sensor Green (SOSG) (Invitrogen/Molecular Probes) solution was added. Fluorescence levels were measured as peak height at 530 nm (excitation 480 nm) [26] after 0 and 15 min of exposure to HL conditions (500 μmol photons m^-2 s^-1) in a Tecan infinite^® 200 fluorescence plate reader. To confirm ¹O₂ detection by SOSG, fluorescence spectra of cultures exposed to the same light conditions in the presence of either 10 mM of the ¹O₂ quencher 1,4-diazabicyclo[2.2.2]octane (DABCO) or in medium containing 50% deuterium oxide (D₂O) were measured representatively for strain 22D1 in three independent replicates (Additional file 3). As expected, DABCO reduced the fluorescence signal of SOSG by quenching ¹O₂ whereas D₂O stimulated the signal because D₂O increases the ¹O₂ lifetime. Furthermore, no increased SOSG fluorescence could be measured in dark incubated algal samples compared to medium-only controls (background) confirming the detection of ¹O₂ by SOSG. Singlet oxygen production for each strain was determined by calculating the slope of the increase in background-corrected SOSG fluorescent signals during the 15 min of exposure, and relative values were calculated for each mutant by dividing its levels by the levels of ¹O₂ production in the wild-type strain.

RNA isolation and quantitative real-time PCR (qPCR)

Cells of 5 ml cultures grown to 3 × 10⁶ cells ml^-1 at the light condition indicated were harvested by centrifugation. Total RNA was isolated by the Trizol method as described earlier [46]. For qPCR experiments, 200 ng of individual total RNA was used in each 10 μl reverse transcription reaction with a reverse transcription kit (Applied Biosystems) according to the manufacturer's instruction.

Sequences of primers for qPCR were designed with the Primer Express™software (Applied Biosystems). qPCR reactions were performed on the ABI Prism^® 7000 or 7500 Sequence Detection System (Applied Biosystems) as described earlier [46]. Threshold cycle (C_t) values were determined for all reactions in the logarithmic amplification phase, and the average C_t value was calculated for each sample out of three technical replicates. Each reaction was confirmed to contain a single amplicon by gel electrophoresis and melting curves. Genomic DNA contamination in RNA samples was assessed by qPCR using RNA without reverse transcription as templates, and in all cases had at least six C_t-values higher than that of respective cDNAs. The efficiency for the amplification of each product was determined by serial dilutions of template cDNA and used to correct C_t values for variable amplification efficiencies. The C_t values of the CBLP gene were used as the internal references. Relative expression was calculated for each mutant compared to 4A⁺ pYS1 under the same growth conditions as an average with standard error out of three independent experiments.

Pigment analyses

For tocopherol and pigment analysis, cells from 1 ml samples were harvested by centrifugation and immediately frozen in liquid nitrogen. Pigments and α-tocopherol were extracted by vortexing in 300 μl of acetone for 1 min, and the extracts were filtered through 2 μm nylon filters. Pigments were fractionated and analyzed by high-performance liquid chromatography (HPLC) as described previously [42].

Determination of NPQ

Chlorophyll fluorescence was determined with an imaging pulse amplitude modulated (IPAM) chlorophyll fluorescence system (IMAG-MAX/l, Walz, Germany). Mutants were grown under LL (15 μmol photons m^-2 s^-1) or 250 μmol photons m^-2 s^-1 for 2 days up to a density of 3 × 10⁶ cells ml^-1 before the cells of 200 μl of culture were collected on a glass microfiber filter (Whatman, UK) by vacuum filtration. To keep filters moist they were transferred on a wet paper towel and cells were dark adapted for 20 min before maximal chlorophyll fluorescence level F _m was determined. Then actinic light of the same intensity as growth light was turned on for 15 min before F _m ' was measured to calculate average NPQ as (F _m -F _m ')/F _m ' of five independent replicates.

Cluster analysis

Hierarchical gene clustering was performed using the Cluster 3.0 software. Data were first log transformed and normalized by multiplying all values of a parameter with a constant scale factor so that the sum of the squares of the values for each parameter was 1.0. This resulted in a scale factor for GPXH expression and ¹O₂ formation of 7.1. For NPQ and pigment contents, which had similar maximal levels, a single scale factor (3.5) was calculated based on the parameter with the highest sum of squares (deepoxidation at LL). Clustering was performed based on average linkage and the results were visualized in a color-based expression pattern using the TreeView 1.60 software designed by the Eisen Lab http://rana.lbl.gov/.

Statistical analysis

Different parameters determined for individual mutant were analyzed for their significant differences to the same parameters measured in the wild-type strain using a paired Student's t-test. Significant differences at a p-value < 0.05 are indicated by a star. Correlations of different parameters were analyzed by linear regression and indicated by the square of the corresponding regression coefficient (R²).