The process of chloroplast biogenesis requires a multitude of pathways and processes to establish chloroplast function. In cotyledons of seedlings, chloroplasts develop either directly from proplastids (also named eoplasts) or, if germinated in the dark, via etioplasts, whereas in leaves chloroplasts derive from proplastids in the apical meristem and are then multiplied by division. The snowy cotyledon 2, sco2, mutations specifically disrupt chloroplast biogenesis in cotyledons. SCO2 encodes a chloroplast-localized protein disulphide isomerase, hypothesized to be involved in protein folding. Analysis of co-expressed genes with SCO2 revealed that genes with similar expression patterns encode chloroplast proteins involved in protein translation and in chlorophyll biosynthesis. Indeed, sco2-1 accumulates increased levels of the chlorophyll precursor, protochlorophyllide, in both dark grown cotyledons and leaves. Yeast two-hybrid analyses demonstrated that SCO2 directly interacts with the chlorophyll-binding LHCB1 proteins, being confirmed in planta using bimolecular fluorescence complementation (BIFC). Furthermore, ultrastructural analysis of sco2-1 chloroplasts revealed that formation and movement of transport vesicles from the inner envelope to the thylakoids is perturbed. SCO2 does not interact with the signal recognition particle proteins SRP54 and FtsY, which were shown to be involved in targeting of LHCB1 to the thylakoids. We hypothesize that SCO2 provides an alternative targeting pathway for light-harvesting chlorophyll binding (LHCB) proteins to the thylakoids via transport vesicles predominantly in cotyledons, with the signal recognition particle (SRP) pathway predominant in rosette leaves. Therefore, we propose that SCO2 is involved in the integration of LHCB1 proteins into the thylakoids that feeds back on the regulation of the tetrapyrrole biosynthetic pathway and nuclear gene expression.
After germination, seedlings have to establish chloroplasts to become autotrophic. Chloroplast biogenesis can be separated into two steps, the first occurring when seedlings are germinating in the dark that results in the development of etioplasts, and the second step commences when the emerging seedling reaches light and photosynthetic-active chloroplasts are formed. Etioplasts have a characteristic lattice-like structure, the prolamellar body, which requires carotenoids (Park et al., 2002), the chlorophyll precursor protochlorophyllide (pchlide), its reducing enzyme protochlorophyllide oxidoreductase (POR) and nicotinamide adenine dinucleotide (NADPH) for its formation (Gunning, 1965). Accumulation of pchlide needs to be regulated tightly because of its photosensitizing action upon illumination, which, if over-accumulated, results in the formation of singlet oxygen. Indeed, the fluorescent, flu, mutant was isolated based on the over-accumulation of pchlide that causes either early seedling death or a stunted growth and developing necrosis if subjected to extended periods of darkness (Meskauskiene et al., 2001). Because pchlide fluoresces in blue light and emits a bright red colour, this mutant was named fluorescent (Meskauskiene et al., 2001). Tetrapyrrole biosynthesis is regulated by direct binding of either heme or the FLU protein to glutamyl-tRNA reductase (GluTR or HEMA1) thus negatively regulating its activity (Meskauskiene and Apel, 2002).
Chloroplast biogenesis requires many different pathways to establish normal chloroplast function that include the import of plastid proteins that are encoded in the nucleus as well as the translation and transcription of chloroplast-encoded genes (Pogson and Albrecht, 2011). Furthermore, the internal membrane system, the thylakoids, needs to be established that harbours the photosynthetic apparatus. Thylakoids are formed from vesicles emerging from the inner envelope moving to the developing thylakoids. The vesicle-inducing plastid protein 1 (VIPP1) has been characterized as being involved in this process, as no vesicles are formed and thylakoid formation is severely impaired when this protein is missing (Kroll et al., 2001). Debate is ongoing as to whether these vesicles are used to incorporate the proteins of the photosystems into the thylakoids, with proponents that provide evidence that shows co-localization of these vesicles with light-harvesting chlorophyll binding (LHCB) proteins (Eggink et al., 2001). In addition, the signal recognition particle pathway (SRP) has been shown to target and integrate LHCB proteins to the thylakoids. LHCB proteins bind directly to SRP43 via their L18 motif and to SRP54 via one of their transmembrane domains (Jonas-Straube et al., 2001). Binding of FtsY and GTP to this transit complex targets and integrates LHCB proteins to the thylakoid membrane (Tu et al., 1999). Also, other essential proteins might be targeted to the thylakoid membrane. So, if photosystems cannot be assembled the mutations in the SRP pathway should be lethal to seedlings. But, interestingly, homozygous single and double mutants of this pathway can be propagated on soil (Rutschow et al., 2008), which indicated that at least one other pathway is partially redundant with the SRP and transport vesicle pathways.
Chloroplast biogenesis during seedling development has been studied by identifying mutations that disrupt the formation of chloroplasts after germination. These chloroplasts show the involvement of different biological processes, as well as the importance of functional chloroplast formation in seedling development and, ultimately, yield. A key class of mutations has been identified that specifically disrupts chloroplast biogenesis in cotyledons: sigma factor 6 (sig6), delayed greening 1 (dg1), white cotyledon (wco1), cyo1 (shi-o-u, means cotyledon in Japanese) and the snowy cotyledon (sco) loci (Yamamoto et al., 2000; Ishizaki et al., 2005; Albrecht et al., 2006, 2008, 2010; Ruppel and Hangarter, 2007; Shimada et al., 2007; Chi et al., 2010). These mutations have different functions: the sig6 and dg1 mutations have an impact on chloroplast gene transcription (Ishizaki et al., 2005; Chi et al., 2010); wco1 and sco1 both affect chloroplast protein translation (Yamamoto et al., 2000; Albrecht et al., 2006; Ruppel and Hangarter, 2007); sco3 encodes a protein of unknown function that is not located in chloroplasts but to the periphery of peroxisomes, associated with the cytoskeleton (Albrecht et al., 2010); and cyo1 and sco2 are allelic mutations in a gene that encodes a chloroplast protein disulphide isomerase that was shown to be capable of renaturation of denatured RNase A (Shimada et al., 2007; Albrecht et al., 2008). Intriguingly, despite showing bleached or chlorotic cotyledons, these mutants exhibit very few, if any, defects in the chloroplasts of rosette leaves; this situation suggests distinct complementary processes and/or proteins function in the two different types of photosynthetic leaves.
Chlorophyll is a prerequisite of thylakoid formation due to its requirement for integrating the light-harvesting chlorophyll binding (LHCB) proteins into this membrane system. Indeed, with the help of the chlorophyll b-less mutant in Chlamydomonas it was shown that if this pigment is missing, several LHCB proteins such as LHCB1, LHCB4 and LHCB6 are absent (Eggink et al., 2001). Furthermore, the import of these proteins into chloroplasts seems to require chlorophyll b because without this pigment the import of these proteins is impaired. At a low rate of chlorophyll biosynthesis only a small number of LHCB proteins were found within chloroplasts and only at the chloroplast envelope, the initial site of integration (Hoober et al., 2007).
A specific role of SCO2/CYO1 in chloroplast biogenesis has not yet been established. Thus far, it is not known which proteins require the function of SCO2 for folding, at what stages of plant development does SCO2 function, and how does its absence result in photobleaching of cotyledons but not rosette leaves? The following study describes the identification of SCO2/CYO1 protein-interacting partners in vitro and in vivo. We demonstrate how interaction with the light-harvesting complex proteins LHCB1 might have an impact on the regulation of chlorophyll biosynthesis and we propose an involvement of SCO2/CYO1 in the formation of thylakoids in developing chloroplasts.
SCO2 is co-expressed with genes encoding proteins of the tetrapyrrole biosynthesis pathway
two research groups independently identified sco2 and cyo1 mutant alleles in a protein disulphide isomerase by phenotypic screening, herein after only referred to as SCO2 (Shimada et al., 2007; Albrecht et al., 2008). Its requirement for chloroplast biogenesis in cotyledons and its protein disulphide isomerase activity was shown (Shimada et al., 2007; Albrecht et al., 2008), but nothing is known about its target proteins or its function in chloroplast biogenesis. To gain insight into possible function(s) or targets of SCO2, a co-expression analysis was performed (Figure 1). We used the gene expression trajectories of all genes contained in the AtGenExpress developmental baseline experiment (Schmid et al., 2005) to compute a co-expression matrix. All positively (0.8 < pxy < 1) or negatively (−0.8 < pxy < −1) correlated genes to SCO2 were extracted and analysed whether they had an already known function in the chloroplast (Figure 1a,b and Data S1 and S2). A total of 141 genes displayed positively (Figure S1), while 27 exhibited negatively correlated expression trajectories with SCO2 during plant development. When analysing the localization of the encoded proteins of the positively correlated genes, a highly significant portion (108; hypergeometric p <10−7) was shown or predicted to be targeted to the chloroplast, or dually targeted to both chloroplasts and mitochondria. As only one single gene of the 27 negatively correlated set of genes had an assigned function in the chloroplast, we focussed on the 141 positively correlated genes for our further analyses. In total, 108 of these 141 are targeted to or function in the chloroplasts. Of these genes, 57 are involved in transcription and translation of chloroplast proteins (Figure 1b,c) and a further seven are involved in chlorophyll biosynthesis.
We computed a gene regulatory network of the 141 genes that included SCO2, and in accordance with our expectation, the majority of the genes formed a tight hub with edges connecting almost every node. Interestingly, most genes surrounding this hub belong to a group of 27 genes with no relation to the chloroplast (Figure S3).
Seven of the co-regulated genes encode proteins directly involved in the tetrapyrrole biosynthesis pathway, including Glu-tRNA synthase, GSA, HEME1 and 2, HEMF1, HEMG and HO (Figures 1b and 2a). Additionally, the protein RIF10, also in the list of co-expressed genes with SCO2 (Data S1), regulates isoprenyl diphosphate – the basis of the phytol moiety of chlorophyll (Sauret-Gueto et al., 2006). Together, this finding indicates that SCO2 might influence or is involved in the regulation of the tetrapyrrole biosynthetic pathway. Therefore, the expression levels of six tetrapyrrole biosynthesis genes GSA, HEME1 and 2, HEMF1, HEMG and HO were analysed in sco2 and compared with its corresponding wild type (Ler) levels. For most of the analysed genes of either dark- or light-grown seedlings, a difference in transcript accumulation was not observed. Figure 2(b) indicates that HEME2 and HEMF1 transcripts are slightly down-regulated in the dark and HEMG transcript levels are reduced in light-grown seedlings (P < 0.05).
Mutations in sco2 affect the accumulation of the chlorophyll precursor protochlorophyllide in etiolated seedlings as well as in true leaves
We investigated the effect of SCO2 on the accumulation of the most abundant tetrapyrrole in etiolated seedlings, protochlorophyllide (pchlide) of sco2-1, flu (an overaccumulator of pchlide) and sco2-2flu (Figure 3a). In contrast to expectations based on the transcript levels of the tetrapyrrole biosynthesis genes, the sco2-1 mutation does affect the accumulation of pchlide, resulting in twice as much pchlide accumulated in etiolated seedlings as in wild type Ler (Figure 3b). As expected flu had hyper-accumulation of pchlide but, unexpectedly, sco2-2flu reduces the accumulation of this chlorophyll precursor to about a third of the level observed in the flu mutant in 4-day-old etiolated seedlings (Figure 3b). Nevertheless, the pchlide accumulation in sco2-2flu is still about four times higher compared with wild type (Ler) and twice as high as sco2-1.
Quantitative measurements of pchlide accumulation in leaves of 4-week-old plants kept for 16 h in the dark resulted in similar ratios of pchlide accumulation compared with Ler as observed in 4-day-old etiolated seedlings (Figure 3c). Thus, this phenotype of sco2 is not restricted to cotyledons but is also apparent in rosette leaves.
The mutation in sco2-2flu reduces or abolishes the singlet oxygen-induced cell death observed in flu
One major characteristic of the flu mutant is the induction of cell death due to the over-accumulation of pchlide and the resulting release of singlet oxygen after illumination (Meskauskiene et al., 2001). As the mutations of sco2 result in a phenotype with pale cotyledons (Shimada et al., 2007; Albrecht et al., 2008) and thus, effect the survival of seedlings, we examined the cell death rate in isolated protoplasts from sco2-1 and sco2-2flu and compared them to flu and Ler. Protoplasts of flu that have been kept in the dark for 24 h show an increased cell death rate of over 50% after 2 h exposure to light that increases to almost 100% after 24 h. In contrast, neither sco2-1 nor sco2-2flu show a change in cell death rate, in fact they behave like wild-type protoplasts (Figure 4a).
Mature flu plants, grown under long-day growth conditions with a 16 h light/8 h dark regime, exhibit a stunted growth of the bolting stems due to the release of singlet oxygen (Przybyla et al., 2008). To quantify the growth rate of bolting stems, laser measurements were performed of plants kept for 6 h in the dark before being exposed to light. Directly after illumination an increase in the rate of stem elongation was observed in Ler plants (Figure 4b). Although in mature plants the pchlide accumulation in sco2-1 is almost a third that of sco2-2flu (Figure 3c), both mutant lines exhibit very similar growth rates of bolting stems in the dark as well as in the light, although reduced compared with Ler (Figure 4b). The flu mutant shows a much reduced elongation of the bolting stem in the light after the exposure to dark as described previously (Przybyla et al., 2008). Thus, the mutation in sco2-2flu prevents the cell death phenotype observed in the flu mutant, a phenomenon that can be explained by reduction in pchlide accumulation.
SCO2 does not interact with FLU or HEMA1
Using the yeast two-hybrid system, Meskauskiene and colleagues showed that FLU interacts directly with the tetrapyrrole biosynthesis protein HEMA1, which is glutamyl tRNA reductase (Meskauskiene and Apel, 2002). As the mutations in sco2 affect the accumulation of pchlide in the mutant lines, we were interested if SCO2 interacts directly with either FLU or HEMA1. We tested this hypothesis using the yeast two-hybrid system. Our results showed that SCO2 does not interact with either protein in this system (Figure 5a).
SCO2 forms a homomer
The SCO2 protein disulphide isomerase is a DnaJ-like zinc finger domain that consists of one zinc finger (Shimada et al., 2007; Albrecht et al., 2008). However, DnaJ proteins have two sets of zinc finger domains. Therefore, we tested if SCO2 forms a homomer to obtain two sets of zinc fingers for its disulphide isomerase function and, indeed, yeast two-hybrid assays showed that SCO2 interacted with itself. From database analyses another protein was identified with a similar single DnaJ-like zinc finger domain and that was predicted to be located in the chloroplast (Figure S4). This protein, termed SCO2-like1, was also able to form a complex with SCO2 in yeast (Figure 5a). To confirm the interactions observed in yeast we transformed Arabidopsis cells with constructs that contained either SCO2 or the SCO2-like1 protein using the bimolecular fluorescence complementation (BIFC) system split yellow fluorescent protein (YFP) to monitor a possible interaction in planta. A strong YFP signal could be observed that showed the interaction of SCO2 with itself and, as expected, localized to the chloroplast. However, only background autofluorescence of plastids was visible when SCO2 and SCO2-like1 were used in BIFC experiments, which indicated that these two proteins do not interact in planta (Figure 5d). In fact, SCO2-like1 has recently been described as low quantum yield of photosystem II 1 (LQY1) and as being involved in the maintenance of the photosystem II under high light, rather than in the biogenesis of chloroplasts (Lu et al., 2011). Thus, in planta, SCO2 forms a complex with itself but not with another protein that contains a similar single zinc finger domain.
SCO2 is interacting with LHCB1 proteins
Protein disulphide isomerases, including SCO2, were shown to be involved in protein folding (Shimada et al., 2007). The question arises – which proteins are the targets of SCO2? A small-scale yeast two-hybrid screen was performed using a random primed cDNA library of 4-week-old Arabidopsis plants available at ABRC (CD-4). Fourteen clones were positively selected for interaction, of which nine contained fragments of the LHCB1-encoding cDNA. Two different LHCB1 genes, LHCB1.2 and LHCB1.5, were cloned and tested for interaction with SCO2. In yeast, both LHCB1 proteins interacted with SCO2 (Figure 5b), which was confirmed for LHCB1.5 in planta using the BIFC system (Figure 5d). In contrast, no interaction was observed between SCO2 and LHCB2, neither in yeast nor in planta (Figure 5b and Figure S5). To specify the location of SCO2 binding to the LHCB1 proteins, we compared the obtained LHCB1 protein interaction partners from the yeast two-hybrid screen in their peptide sequence with the known domains of LHCB1 proteins according to Heinemann and Paulsen (1999). Interestingly, all eight colonies that contained cDNA of LHCB1 from the random primed cDNA library encoded a short peptide of 70–127 amino acids in length and that covered the LHCB1.5 region from amino acids 62–130, which includes the first transmembrane domain (Figure 5c). Thus, SCO2 interacts specifically with LHCB1 proteins, both in vitro and in vivo. Transcripts of LHCB1.2, LHCB1.5, and to a lower extent also LHCB2.3, are already present in etiolated seedlings in both sco2-1 and Ler. LHCB2.3 is expressed predominately in true leaves, whereas LHCB1.5 is expressed similarly in etiolated and light-grown seedlings (Figure 6).
SCO2 is not part of the signal recognition particle protein targeting pathway
As sco2 is a chloroplast biogenesis mutant and SCO2 interacts with LHCB1 proteins, we examined whether SCO2 was involved in LHCB1 protein targeting to the thylakoids. LHCB proteins are targeted within the chloroplast to the thylakoids via the SRP pathway and require the SRP proteins, FtsY and SRP54, to be targeted to the thylakoids (DeLille et al., 2000). In fact, both proteins are also co-expressed with SCO2 (Data S1). The cDNA of both genes was cloned into the Gateway pACT3 vector or directly into pGBK.T7. Unfortunately, both proteins were transactivating in the yeast two-hybrid system in either combination and showed an ‘interaction’ with the empty vector on the triple selection medium (data not shown). Thus, the yeast two-hybrid system could not be used to analyse an interaction between SCO2 and these two proteins. Using the BIFC system, a YFP signal that confirmed an interaction with SCO2 was not observed for either of the two proteins, SRP54 and FtsY (Figure 5b and Figure S5).
sco2 chloroplasts are affected in thylakoid formation
While sco2, in the Ler background, has a pale green cotyledon and is viable in soil, the cyo1 allele, in Col background, is white and needs to be grown on Murashige Skoog (MS) medium that contained 1.5% sucrose to reach the true leave stage (Shimada et al., 2007; Albrecht et al., 2008). Chloroplast structure has been shown for cyo1 (Shimada et al., 2007). Because of the observed differences, we were interested if the chloroplast ultrastructure differed between the two allelic lines. The cyo1 plastids were described as being ‘smaller and abnormally shaped’ and as not developing ‘thylakoid membranes but instead containing a large number of electron-dense particles’ (Shimada et al., 2007). Our results showed that the ultrastructure of sco2 plastids differed dramatically from the ultrastructure described for plastids in cyo1. The chloroplasts in sco2-1 cotyledons of soil grown 10-day-old seedlings were either of normal shape or were globular with big vesicles that resembled elaioplasts (Figure 7a). Interestingly, both types of plastids can occur within one cell, although more frequently they were observed in different cells. A closer look at the ultrastructure of developed chloroplasts in sco2-1 showed that vesicles emerged from the inner envelope, primarily at the rounded ends of the elongated chloroplasts (see arrows on Figure 7b,c). These transport vesicles were not observed in any of the analysed wild-type chloroplasts. Thus, this observation seems to be specific for sco2. Previously, this type of vesicle has been observed in wild-type plants that had been treated at 4°C to slow down the formation or turnover of thylakoids (Morre et al., 1991). These vesicles were proposed to be transport vesicles that were involved in the formation of the thylakoid membrane and integration of the photosystems into the thylakoids by attracting and incorporating, among other proteins, the chlorophyll a/b binding proteins LHCB (Eggink et al., 2001). It seems that the loss of SCO2 perturbs the movement and/or fusion of the transport vesicles with the thylakoid membrane and results in an accumulation of these vesicles apparent at the rounded ends of the sco2 chloroplasts.
Accumulation of the photosensitizer, protochlorophyllide in sco2
In this manuscript we provide evidence for a direct involvement of the SCO2 protein disulphide isomerase in thylakoid formation, association with members of the LHCB1 protein family, and pchlide accumulation. Pchlide is a photosensitizer that, upon illumination, results in the release of the reactive oxygen species singlet oxygen (1O2), and leads to increased cell death (Meskauskiene et al., 2001). Furthermore, the reduction of pchlide to chlorophyllide a by protochlorophyllide oxidoreductase, POR, requires light for catalysis in angiosperms (Forreiter and Apel, 1993). Thus, tetrapyrrole biosynthesis needs to be regulated tightly in both light- and dark-grown tissues to prevent the release of singlet oxygen upon illumination (Eckhardt et al., 2004). In higher plants, both branches of the tetrapyrrole biosynthesis pathway, the heme as well as the chlorophyll branch, can directly negatively regulate the function of HEMA1 thereby lowering the biosynthesis of tetrapyrroles. Indeed, FLU directly interacts with HEMA1 (Meskauskiene and Apel, 2002). As over-accumulation of pchlide is evident in sco2 (Figure 3a), we investigated the nature of this regulation and its effects on the plant. SCO2 does not interact with FLU or HEMA1 and its inactivation in flu resulted in a lowering of pchlide levels. In fact, a second sco2 allele, sco2-2, was isolated from a second site suppressor of pchlide accumulation screen on flu. This finding indicated that SCO2 is involved directly or indirectly in the regulation of tetrapyrrole biosynthesis, but presumably in a different pathway than FLU. Another mutant isolated from a second site mutagenesis screen of flu was SOLDAT8 (singlet oxygen-linked death activator). This mutant also showed a reduction of pchlide accumulation to three-quarters of that of flu and a decreased cell death rate (Coll et al., 2009). SOLDAT8 encodes the chloroplast SIGMA FACTOR 6 and is involved in the regulation of plastid gene transcription. The extent to which there is any connection between SOLDAT8- and SCO2-mediated pchlide accumulation is unknown at this stage.
flu mutants accumulate high levels of pchlide in the dark and, after transfer to light, exhibit increased cell death and a stunted growth (Meskauskiene et al., 2001). While the sco2 and flu mutants all exhibited slower rates of stem growth after the dark–light transition, this effect was more pronounced in flu. The cell death rate of sco2 and sco2-2flu was monitored in dark-to-light exposed protoplasts and was comparable with wild type, not flu. This finding might be explained by the requirement for a minimum level of excess pchlide accumulation in the dark and, therefore, the release of a certain level of singlet oxygen upon illumination in order to induce cell death, which suggested dose dependency between pchlide accumulation, singlet oxygen release and cell death.
SCO2 and LHCB1 interactions: implications for thylakoid and photosystem assembly
LHCB proteins are transferred through the chloroplast envelope via the actin–Toc–Tic–VIPP1 complex and are directed from the inner envelope to the thylakoid membranes by transport vesicles (Jouhet and Gray, 2009). It has also been shown that the SRP pathway is involved in targeting LHCB proteins to the thylakoids. The pathway requires SRP43/SRP54/FtsY proteins for targeting to, and ALBINO3 for integration into, the membrane system, although it is still assumed that other unknown proteins are involved in this process prior to the interaction with ALBINO3 (Schunemann, 2007). Similarly, impaired chlorophyll biosynthesis, such as Chl b minus mutants, or low rates of chlorophyll biosynthesis result in the accumulation of only a small number of LHCB proteins in chloroplasts and only at the chloroplast envelope, the initial site of integration (Hoober et al., 2007).
In this study, we showed that SCO2 interacted with LHCB1 proteins in vitro and in vivo. Furthermore, chloroplast ultrastructural analyses revealed that the sco2 mutant contained an increased number of transport vesicles that could only be observed in wild-type chloroplasts when the transport to the thylakoids is impeded by low temperature (Morre et al., 1991). These envelope-derived vesicles were also detected in dark-grown Chlamydomonas chloroplasts after they had been transferred to light for a few minutes (Eggink et al., 2001). The necessity for transport vesicles for thylakoid formation has been verified in the Arabidopsis thaliana mutant that lacks VIPP1, a protein required for the formation of these vesicles (Kroll et al., 2001). Evidence suggested that the SCO2 protein seemed to be involved in this process of vesicle-derived thylakoid formation, which is compromised by the loss of SCO2 protein function. Our two hypotheses were that the sco2 mutation may either slow down the vesicle formation by delaying the integration of the LHCB proteins into these structures and/or it may affect the merging of these vesicles with the thylakoid membranes. The interaction of SCO2 with the LHCB1 proteins suggested that the affected LHCB1 integration into the transport vesicles caused the observed accumulation of these vesicles in the sco2 mutant (Figure 8).
The delayed integration of LHCB1 proteins into the thylakoid membranes seems to have an effect on the regulation of chlorophyll biosynthesis. LHCB1 proteins are already present at low levels in plastids of etiolated seedlings (Figure 6), and ensures that upon illumination these proteins can be loaded immediately with the newly converted chlorophyll, thereby limiting the oxidising effect of free tetrapyrrole in the chloroplast. If, as we propose, SCO2 is involved in the integration and folding of LHCB proteins at the initial location of integration, it would impact upon the integration of chlorophyll into folded LHCBs. Lack of this regulation or loading control in the sco2 mutants might result in feedback control on chlorophyll biosynthesis and result in the observed over-accumulation of the chlorophyll precursor pchlide. Different accumulation of pchlide in sco2-1 and sco2-2flu with flu had an epistatic effect on sco2 that enhanced further the hypothesis of at least two independent pathways for the regulation of the chlorophyll biosynthetic pathway, with SCO2 acting separately from the known direct FLU-mediated pathway. Interestingly, in Synechocystis, a specific level for chlorophyll regulation has also been identified; the loss of small chlorophyll-binding proteins (small CAB-like proteins, SCPs) resulted in down-regulation of pchlide levels in scpB− and scpE−Synechocystis lines (Xu et al., 2002). The authors hypothesized that SCPs regulate chlorophyll biosynthesis by sensing the availability of chlorophyll and increase the synthesis of these pigments if no binding occurs. Similarly, in Arabidopsis, the chlorophyll-binding protein EARLY LIGHT INDUCIBLE PROTEIN 2 (ELIP2) was described to be involved in the regulation of chlorophyll biosynthesis by regulation of the activity of 5-aminollevulinate synthesis and magnesium-proto IX chelatase activity. Overexpression of this protein results in reduced pchlide accumulation (Tzvetkova-Chevolleau et al., 2007). Furthermore, ELIP2 proteins are integrated into the chloroplast membrane system without the assistance of the SRP pathway (Kim et al., 1999). In sco2, the sensing mechanism for pigment binding might be disrupted due to misfolded proteins, which then results in the increased accumulation of pchlide in the dark. Two mutants described as affecting the retrograde signalling control from the chloroplast to the nucleus, genome uncoupled, gun4 and gun5, have mutations in a cofactor or a subunit of magnesium chelatase, which is the first enzyme after the branching point of chlorophyll biosynthesis (Mochizuki et al., 2001; Larkin et al., 2003). In fact, the presence of GUN4 gives a positive feedback to the synthesis of 5-aminollevulinate (Peter and Grimm, 2009). Thus, a control system at different levels, either during chlorophyll biosynthesis or by chlorophyll binding, while integrating chlorophyll-binding proteins into the chloroplast membrane system, seems to reduce the risk of free tetrapyrrole within the chloroplasts.
SCO2 is expressed ubiquitously in all plant tissues, both in seedlings and mature leaves. However, the chlorotic phenotype (10–20% of wild-type chlorophyll content) is visible predominantly only in the cotyledons. Conversely, the SRP54 mutant chaos (chlorophyll a/b binding protein harvesting-organelle-specific) was described to be pale green in both seedlings and mature plants and accumulates about 60% chlorophyll in all aerial tissues (Klimyuk et al., 1999). During germination, an increased demand for thylakoid formation is required for chloroplast biogenesis, which involves integrating proteins and lipids into the thylakoid membranes. Transport vesicles have been shown to be one of the ways of importing the fatty acids for thylakoid formation (Benning, 2009). Therefore, we hypothesize that in germinating seedlings the photosynthetic apparatus is assembled into the thylakoids predominantly by transport vesicles plus SCO2, facilitating a faster way to form thylakoids and photosystems during germination (Figure 8). This model requires the function of SCO2 in integration and folding LHCB proteins. In mature plants, SRP proteins are involved predominantly in targeting LHCB proteins to the already existing thylakoids, primarily to substitute defective LHCB proteins. This situation also explains why the sco2 phenotype, being defective in chloroplast biogenesis, is restricted to the embryonic leaves.
In Cytoscape, attributes on protein localization (Ferro et al., 2010) were retrieved from AT-CHLORO database ((http://www.grenoble.prabi.fr/at_chloro/) and added to each probeset. In addition, information on chlorophyll biosynthesis and biosynthesis-related genes and proteins was taken from PlantCyc/AraCyc (http://www.plantcyc.org/) (Mueller et al., 2003; Zhang et al., 2007; Karp et al., 2010). In addition, information on putative or validated functions of proteins in the chloroplast was taken from the Gene Ontology (GO) terms deposited in TAIR (http://www.arabidopsis.org). The P-values for significant overrepresentation of the GO term ‘chloroplast’ were calculated using the hypergeometric distribution. The GO-categories of the universe (all present genes on the ATH1 gene chip) were compared with the sampling distribution of the gene list.
Finally, co-expressed genes for SCO2 (AT3G19220, 257040_at) were identified from this matrix with Cytoscape as ‘first neighbours’.
Plant material and growth conditions
Plants were grown at 21°C either under continuous-light or under long-day conditions with 16 h light and 8 h dark at 100 μmol m−2 sec−1. For seedling analysis, seeds were surface sterilized and sown on 1× MS media (Sigma, http://www.sigmaaldrich.com) without sucrose. Growth-rate measurements were performed on 4-week-old plants grown under continuous light, starting to have a small bolting stem. Using a laser device as described by Przybyla et al. (2008), the growth rate of the bolting stem was measured after 6 h exposure to dark. At least three plants were measured per line. Transcript analyses with gene-specific primers on reverse-transcribed total RNA from 4-day-old etiolated or light-grown seedlings were performed as described previously (Albrecht et al., 2008).
Quantification of the precursor of chlorophyll, protochlorophyllide (pchlide) was done either on 10 seedlings grown for 4 days in the dark directly transferred into 80% acetone, or on 30 mg fresh weight of leaves from 4-week-old plants that were kept 16 h in the dark, then transferred into liquid nitrogen before light exposure under safe green light. Pchlide measurements of etiolated seedlings were performed by direct measurement in 80% acetone using a spectrophotometer at an excitation wavelength of 420 nm, measuring the emission wavelength between 600 and 700 nm. Because of the influence of chlorophyll in true leaves of 4-week-old plants, measurements of pchlide were performed using high pressure liquid chromatography (HPLC) as described previously (Przybyla et al., 2008).
Protoplast cell death analysis
The influence of singlet oxygen release on cell death in the different lines was analysed by quantifying the cell death rates of protoplasts isolated from 7-day-old seedlings. Protoplasts were exposed to 24 h dark before transfer to continuous light. After the indicated time points, cells were taken from the culture, dead cells stained with 0.04% Evans blue (Sigma) and the number of dead cells per culture compared with the total number of cells (Danon et al., 2006).
Protein interaction analysis
The identification of putative protein interaction partners was performed using the yeast two-hybrid system (Y2H). The full length cDNA of SCO2 was cloned into the pGBK.T7 vector using NcoI restriction sites. The cDNA was checked for correct insertion and sequencing, then the construct was transformed into the yeast strain AH109 together with the empty pGAD.T7 vector to check for transactivation. Yeasts that contained both vectors were not able to grow on selection medium that lacked tryptophane, leucine and histidine (Clontech, http://www.clontech.com/), so SCO2-pGBK.T7 is not transactivating. Thus, yeast that contained only SCO2-pGBK.T7 were transformed with a cDNA library (CD4, ABRC) from rosette leaves and checked for growth on selection medium that lacked all three amino acids. Transformation control with yeasts grown on selection medium that lacked only tryptophane and leucine was performed in parallel to check for presence of both vectors. Fourteen colonies were obtained from this screen, for which the insert in the pACT vector was sequenced using the pACT-F primer. Full-length cDNAs of LHCB1.2, LHCB1.5 and LHCB2.3 were cloned into vectors pGAD.T7 and pGBK.T7. All constructs were checked for transactivation, and showed that for both LHCB1 constructs in pGBK.T7 vector are transactivating. Thus, interaction of SCO2-pGBK.T7 was only analysed in Y2H with the LHCB-pGAD.T7 constructs. Full-length cDNAs of SRP54 and FtsY were cloned similarly. Additionally, cDNAs were cloned into the Gateway pENTR2B vector and the Gateway system was used for further cloning into BIFC vectors pSPYNE/pSPYCE. These were then transformed transiently into Arabidopsis suspension cells and analysed for a YFP signal. The small subunit of Arabidopsis ribulose biphosphate carboxylase (SSU) fused to the red fluorescent protein (RFP, kindly provided by Prof Whelan; Carrie et al., 2009) was used as a plastid localization control. The constructs were transformed into Arabidopsis cultured cells. Five μg of each of the split YFP and marker RFP plasmids were co-precipitated onto 1-μm gold particles and transformed using the biolistic PDS-1000/He system (Bio-Rad, http://www.bio-rad.com/). Particles were bombarded onto 2 ml of Arabidopsis cultured cells that rested on filter paper on osmoticum plates (2.17 g/L MS modified basal salt mixture, 30 g/L sucrose, 0.5 mg/L naphthalene acetic acid, 0.05 mg/L kinetin, 36.44 g/L mannitol). After bombardment, the cells were placed in the dark at 22°C. Fluorescence images were obtained 24 h after transformation using an Olympus BX61 epifluorescence microscope with excitation wavelengths of 490/500 nm (YFP) and 535/555 nm (RFP), and emission wavelengths of 515–560 nm (YFP) and 570–625 nm (RFP). Subsequent images were captured using the Cell® imaging software (http://www.microscopy.olympus.eu/microscopes/Software_cell_R_Software.htm). Only cells that contained the red fluorescence of the RFP were considered as transformed cells, as with the bombardment all three constructs were transferred into one single cell. Weak YFP-like signals in the other cells were used to compare the background with a real YFP signal of a positive protein–protein interaction. The positive interaction between SCO2 with SCO2 or LHCB1.5 could also be verified in Agrobacterium-mediated transformation of leaves (data not shown).
Cotyledons of 10-day-old seedlings were harvested and fixed in a primary fixation buffer (8.5 ml 0.1 m cacodylate buffer, 4% formaldehyde, 0.5 ml glutaraldehyde) for 2 h under vacuum. After washing with cacodylate buffer, the samples were stained for 1.5 h with 1% osmium-tetraoxide. A dehydration series with acetone was carried out before the samples were exposed to increasing concentrations of epon araldite. After hardening of the samples at 65°C for 2 days, the samples were cut into ultrathin sections using a diamond knife and transferred on copper grids for final staining with 5–6% uranium acetate (20 min) and lead citrate (15 min). Analysis was performed using the Hitachi H7100FA (125kV TEM, 1995; http://www.hitachi-hta.com) with digital camera.
Primers used in this manuscript are listed in Data S3.
We would like to thank Rasa Meskauskiene for providing the yeast two-hybrid clones for FLU and HEMA1. Thanks also to Mena Nater, Andre Imboden, and Barbara Maier for support and assistance with plants at the ETH Zurich.