The plastid redox insensitive 2 mutant of Arabidopsis is impaired in PEP activity and high light-dependent plastid redox signalling to the nucleus

Authors

  • Peter Kindgren,

    1. Umeå Plant Science Centre, Department of Plant Physiology, Umeå University, S-901 87 Umeå, Sweden
    2. Australian Research Council Centre of Excellence in Plant Energy Biology, The University of Western Australia, 35 Stirling Highway Crawley 6009 WA, Australia
    Search for more papers by this author
  • Dmitry Kremnev,

    1. Umeå Plant Science Centre, Department of Plant Physiology, Umeå University, S-901 87 Umeå, Sweden
    Search for more papers by this author
    • These authors contributed equally.

  • Nicolás E. Blanco,

    1. Umeå Plant Science Centre, Department of Plant Physiology, Umeå University, S-901 87 Umeå, Sweden
    Search for more papers by this author
    • These authors contributed equally.

  • Juan de Dios Barajas López,

    1. Umeå Plant Science Centre, Department of Plant Physiology, Umeå University, S-901 87 Umeå, Sweden
    Search for more papers by this author
    • These authors contributed equally.

  • Aurora Piñas Fernández,

    1. Umeå Plant Science Centre, Department of Plant Physiology, Umeå University, S-901 87 Umeå, Sweden
    Search for more papers by this author
  • Christian Tellgren-Roth,

    1. Department of Immunology, Genetics and Pathology, Rudbecklaboratoriet, Uppsala University, 751 85 Uppsala, Sweden
    Search for more papers by this author
  • Ian Small,

    1. Australian Research Council Centre of Excellence in Plant Energy Biology, The University of Western Australia, 35 Stirling Highway Crawley 6009 WA, Australia
    Search for more papers by this author
  • Åsa Strand

    Corresponding author
    1. Umeå Plant Science Centre, Department of Plant Physiology, Umeå University, S-901 87 Umeå, Sweden
      (fax +46 90 786 6676; e-mail asa.strand@plantphys.umu.se).
    Search for more papers by this author

(fax +46 90 786 6676; e-mail asa.strand@plantphys.umu.se).

Summary

The photosynthetic apparatus is composed of proteins encoded by genes from both the nuclear and the chloroplastic genomes. The activities of the nuclear and chloroplast genomes must therefore be closely coordinated through intracellular signalling. The plastids produce multiple retrograde signals at different times of their development, and in response to changes in the environment. These signals regulate the expression of nuclear-encoded photosynthesis genes to match the current status of the plastids. Using forward genetics we identified PLASTID REDOX INSENSITIVE 2 (PRIN2), a chloroplast component involved in redox-mediated retrograde signalling. The allelic mutants prin2-1 and prin2-2 demonstrated a misregulation of photosynthesis-associated nuclear gene expression in response to excess light, and an inhibition of photosynthetic electron transport. As a consequence of the misregulation of LHCB1.1 and LHCB2.4, the prin2 mutants displayed a high irradiance-sensitive phenotype with significant photoinactivation of photosystem II, indicated by a reduced variable to maximal fluorescence ratio (Fv/Fm). PRIN2 is localized to the nucleoids, and plastid transcriptome analyses demonstrated that PRIN2 is required for full expression of genes transcribed by the plastid-encoded RNA polymerase (PEP). Similarly to the prin2 mutants, the ys1 mutant with impaired PEP activity also demonstrated a misregulation of LHCB1.1 and LHCB2.4 expression in response to excess light, suggesting a direct role for PEP activity in redox-mediated retrograde signalling. Taken together, our results indicate that PRIN2 is part of the PEP machinery, and that the PEP complex responds to photosynthetic electron transport and generates a retrograde signal, enabling the plant to synchronize the expression of photosynthetic genes from both the nuclear and plastidic genomes.

Introduction

Chloroplasts, like mitochondria, evolved from free-living prokaryotic organisms that entered the eukaryotic cell through endosymbiosis. The gradual conversion from endosymbiont to organelle during the course of evolution has been accompanied by a dramatic reduction in genome size, as the chloroplasts lost most of their genes to the nucleus and the endosymbionts became dependent on their eukaryotic host. The plastid genomes of current land plants encode, over a wide range of species, 75–80 proteins (Timmis et al., 2004), whereas the number of proteins in the chloroplast is estimated to be between 3500 and 4000 proteins (Soll and Schleiff, 2004). About 19% of the Arabidopsis nuclear genes are of cyanobacterial origin (Martin et al., 2002), and an elaborate import system is required to transport these proteins back into the plastids after synthesis in the cytosol (Soll and Schleiff, 2004). The genes remaining in the chloroplast genome are photosynthesis related, or encode housekeeping components such as components of the plastid gene expression machinery (rRNA, a complete set of tRNA and some ribosomal proteins; Wakasugi et al., 2001). The chloroplast genes of higher plants are transcribed by at least two types of RNA polymerases: one is the nuclear-encoded plastid RNA polymerase (NEP), a T3–T7 bacteriophage type that predominantly mediates the transcription of the housekeeping genes (Hedtke et al., 1997, 2000; Puthiyaveetil et al., 2010); the other type, plastid-encoded RNA polymerase (PEP), is a eubacterial-type multi-subunit enzyme. True to its cyanobacterial origin, PEP has a catalytic core comprised of RpoA, RpoB, and RpoC1 and RpoC2. The core subunits of PEP are encoded by the plastid genome, but the promoter specificity factors, or sigma factors, of this enzyme are encoded by nuclear genes (Hanaoka et al., 2003). Photosynthesis-related genes such as psbA, psbD and rbcL are transcribed by PEP (Allison et al., 1996; DeSantis-Maciossek et al., 1999). Promoters of most chloroplast operons encoding components involved in photosynthesis have the -10 and -35 cis-elements similar to those found in bacterial promoters. The PEP enzyme recognizes these cis-elements. The NEP enzyme recognizes the YRTA motif, which can also be found upstream of several genes with PEP promoters, indicating that these genes can be transcribed by both polymerases (Pfannschmidt and Liere, 2005). The initiation of chloroplast development in light and the activation of the photosynthetic reactions are accompanied by the repression of NEP activity and an increase of PEP-mediated transcription (Hanaoka et al., 2005).

Isolations of the transcriptionally active chromosome (TAC) and the soluble RNA polymerase (sRNAP) revealed that in addition to the core components of PEP, a large number of other proteins are required for chloroplast transcription. As many as 40–60 proteins appear to be present in the TAC from chloroplasts. Using mass spectroscopy, 35 components were identified from Arabidopsis and mustard TACs, and 18 of those components, called pTACs, were novel proteins (Pfalz et al., 2006). In addition, TAC and sRNAP preparations from pro-plastids, chloroplasts and etioplasts have different protein compositions (Reiss and Link, 1985; Pfannschmidt and Link, 1994; Suck et al., 1996). Photosynthetic activity has a strong effect on PEP-dependent plastid gene expression, and it has been proposed that redox signals from the thylakoid membrane are linked to plastid gene expression via complex networks of phosphorylation events (Steiner et al., 2009). The ancestral symbiont sensor kinase CSK has been suggested to link photosynthesis with gene expression in chloroplasts (Puthiyaveetil et al., 2008). Furthermore, the phosphorylation of SIG1 is regulated by redox signals and selectively inhibits transcription of the psaA gene (Shimizu et al., 2010). In addition, a plastidial thioredoxin z (TRXz) was identified as a component of TAC and soluble RNPase, suggesting that TRXz could be involved in the redox regulation of PEP activity (Pfalz et al., 2006; Arsova et al., 2010; Schroter et al., 2010).

Although it is true that the chloroplast is dependent on the nucleus for its function, it is also becoming clear that the plastids produce multiple retrograde (organelles → nucleus) signals at different times of their development, and in response to changes in the environment, that orchestrate major changes in nuclear gene expression (Fernandez and Strand, 2008). The photosynthetic apparatus is composed of proteins encoded in both the nuclear and the chloroplastic genomes. To ensure that all the photosynthetic complexes are assembled stoichiometrically, and to enable their rapid reorganization in response to changes in the environment, the activities of the nuclear and chloroplast genomes must be closely coordinated through intracellular signalling. In order to achieve this, retrograde control has evolved to coordinate the expression of nuclear genes encoding organellar proteins with the metabolic and developmental state of the plastids and mitochondria, through signals emitted from the organelles that regulate nuclear gene expression (Rodermel, 2001).

Multiple microarray experiments have revealed that, in addition to PEP-dependent plastid gene expression, the expression of photosynthesis-associated nuclear genes is also affected by changes in chloroplast redox status (Rossel et al., 2002; Kimura et al., 2003; Vanderauwera et al., 2005; Kleine et al., 2007). Three different groups of redox-derived signals have been proposed to control the expression of nuclear-encoded photosynthesis genes: (i) oxidation–reduction (redox) state of components in or coupled to the photosynthetic electron transport chain (PET); (ii) reactive oxygen species (ROS); (iii) metabolite exchange between the chloroplast and the cytosol (Nott et al., 2006; Fernandez and Strand, 2008; Pogson et al., 2008). In the redox-imbalanced mutants (rimb), the expression of the nuclear gene encoding the antioxidant enzyme 2-cys-peroxiredoxin (2-CPA) is uncoupled from the redox state of the photosystem I (PSI) acceptor side (Heiber et al., 2007). However, the identities of the RIMB loci are still unknown, and no other component with a defined role in the redox-mediated retrograde signalling has so far been identified. Thus, despite extensive work, the mechanisms by which redox changes in the chloroplast are sensed, and how this information is transduced to the nucleus to initiate a genetic response, have remained elusive. By using a forward-genetics approach we identified a chloroplast component involved in redox-mediated retrograde signalling from the chloroplast. The plastid redox insensitive 2 (prin2-1) mutant expresses a defective version of PRIN2, a plastid protein localized to the nucleoids. The prin2-1 and the allelic prin2-2 show significantly lower expression levels of PEP-dependent plastid genes, compared with the wild type and a genome-uncoupled phenotype, in response to excess light and the inhibition of PET. These results link redox regulation of photosynthesis-associated nuclear genes to PEP activity in the chloroplast, suggesting that components associated with the PEP complex respond to photosynthetic electron transport, and generates a retrograde signal to regulate the expression of nuclear-encoded photosynthesis genes.

Results

PRIN2 is required for excess light signalling, and is localized to plastid nucleoids

The repression of LHC expression in response to excess light requires plastid signals (Escoubas et al., 1995; Fey et al., 2005a,b; Bräutigam et al., 2007; Kleine et al., 2007). To isolate mutants with impaired retrograde signalling following exposure to excess light, we used the LUCIFERASE reporter gene system, linked to the promoter of LHCB1.1. Ten-day-old M2 seedlings from a seed pool mutagenized by ethane methyl sulfonate (EMS) were grown under continuous light (100 μmol photons m−2 sec−1) before exposure to high light (3 h; 1000 μmol photons m−2 sec−1, HL). The HL conditions used result in a highly reduced redox status and an inhibition of the photosynthetic electron transport (as demonstrated by the photosynthetic parameters, qP and qL; Figure S1) and a strong repression of LHCB1.1 expression. Putative plastid redox-insensitive (prin) mutants that exhibited no or reduced repression of LUC activity following HL treatment, compared with the control, were selected for further analysis. M3 seedlings were re-screened and endogenous LHCB1.1 expression was checked with semiquantitative real-time polymerase chain reaction (rtPCR). Mutants that also demonstrated insensitive endogenous LHCB1.1 expression following HL treatment were selected as true plastid redox-insensitive mutants. In this mutant screen we isolated the prin2-1 mutant.

We mapped PRIN2 to a ∼200 kb region between marker CER425765 and T19D16-T7-2 on the upper arm of chromosome 1 using map-based cloning (Figure 1a). Using SOLiD™ System Sequencing a mutation in At1g10522 was identified in prin2-1 (Figure 1a). The identified mutation was a G → A base substitution in position 524 from the translation start, and it creates a premature stop codon, resulting in a protein that is six amino acids shorter than the wild-type version of PRIN2 (Figure 1b). The PRIN2 protein has no described function in Arabidopsis. Analysis of the PRIN2 sequence revealed that it is a plant-specific protein with no homology to any other Arabidopsis proteins, and it contains no domains with known function (Figure 1b). TargetP predicts that PRIN2 is localized to chloroplasts (Emanuelsson et al., 2007). The predicted mature form of the protein is highly similar between different plant species (Figure 1b). The PRIN2 gene is primarily expressed in green tissue (Figure S2). Seedlings and mature leaves show higher expression than flowers and roots. The expression pattern suggests that PRIN2 is expressed in tissues with mature chloroplasts, but also in seedlings during chloroplast development. Furthermore, the expression profile suggests that PRIN2 is light regulated (Figure S2). PRIN2 expression is also slightly upregulated in response to HL (Figure S2).

Figure 1.

 Positional cloning of PRIN2 and alignment of the PRIN2 protein.
(a) prin2-1 was mapped by analyzing 512 F2 individuals derived from a prin2-1 (Col) × Ler cross with molecular markers. The interval linked to the mutation was sequenced and the base substitution identified. The premature stop codon caused by the prin2-1 mutation is highlighted.
(b) Alignment of PRIN2 plant proteins. Residues that are identical to Arabidopsis PRIN2 are in yellow, 50% identical residues among all proteins in blue and similar residues in green. The predicted transit peptide is shaded in red.

A genomic fragment of the predicted open reading frame (ORF) of At1g10522 and the corresponding cDNA were both able to rescue the prin2-1 phenotype (Figure 2b), confirming that the identified mutation in At1g10522 caused the phenotype. A T-DNA insertion line (GABI_772D02, prin2-2) for PRIN2 was retrieved, and the lack of PRIN2 transcript confirmed that prin2-2 was a null allele (Figure 2d). Seedlings of the prin2-1 and prin2-2 mutants contained significantly less chlorophyll and tetrapyrroles compared with the wild type (Table 1). However, the chlorophyll a : chlorophyll b ratio was normal, suggesting that light-harvesting complex of photosystem II (LHCII) formation is active (Table 1). Photosynthetic parameters were measured in 10-day-old seedlings at the growth irradiance (Table 1). A reduced capacity for photochemistry (ΦPSII) was observed in the prin2 alleles, probably as a consequence of damaged and/or non-fully functional photosynthetic electron transport (PET), as indicated by the lower variable to maximal fluorescence ratio (Fv/Fm) in the mutants compared with the wild type (Table 1). However, the prin mutants were able to absorb and use light energy to reduce the primary quinone acceptor of PSII (QA) (Table 1). Both mutant lines showed a clear reduction in growth rate compared with the wild type (Figures 2b and S3). The prin2-1 mutant was isolated as a redox-insensitive mutant with impaired repression of LHCB1.1 in response to HL. When the prin2-2 mutant was included in our analysis, it demonstrated a similar response to HL (Figure 2e). The misregulation of LHB1.1 expression following HL exposure was also observed in rosette plants (Figure 2e). Thus, the prin2 mutants demonstrated a genome-uncoupled phenotype in response to excess light. To exclude the possibility that the HL-induced cryptochrome 1 (cry1)-mediated pathway is impaired in the prin2 mutants, we investigated the expression of ELIP1, the major marker gene for the cry1-mediated pathway, following exposure to HL (Kleine et al., 2007). The induction of ELIP1 was similar to the induction shown in the wild type following exposure to HL, suggesting that the cry1-mediated pathway is not affected in the prin2 mutants (Figure S4).

Figure 2.

 Impaired LHCB1.1 repression in response to high light (HL) in the prin2 mutants.
(a) Position of prin2-1 mutation and the T-DNA insertion of line GABI_772D07 (prin2-2) in PRIN2 (At1g10522). Primer positions used in (c) and (d) are shown.
(b) Phenotype of prin2 mutants and complementation of prin2-1 with a PRIN2 wild-type copy. Two-week-old seedlings and 4-week-old plants are shown.
(c) Genomic DNA from wild type, HZ prin2-2 and HM prin2-2 were confirmed for T-DNA insertion. Primer combinations are shown in (a).
(d) Expression of PRIN2 was checked in wild type, prin2-1 and prin2-2 plants. Expression of RCE1 (At4g36800) was used as a control.
(e) 7-day-old seedlings or 4-week-old rosette plants grown in 150 μmol photons m−2 sec−1 (CL) and exposed to 1000 μmol photons m−2 sec−1 (HL) for 3 h. Bars represent repression after 3 h of HL compared with CL for each genotype. LHCB1.1 (At1g29920) expression was related to the expression of RCE1 (At4g36800). Data represent mean (±95% CI) from at least three independent biological replicates. Significant differences were calculated with a Student’s t-test (*P < 0.0001).

Table 1.   Determination of chlorophyll fluorescence parameters, chlorophyll and tetrapyrrole content in wild type, prin2-1 and prin2-2 seedlings
 WTprin2-1prin2-2
  1. Each point represents the mean (±SD) of at least seven replicates. The chlorophyll fluorescence parameters were measured 3–6 h into the light period, and the leaf was dark-incubated for 10 min prior to dark-adapted measurements (Fv/Fm). Light-adapted determinations were performed after 4 min of exposure to 150 μmol photons m−2 sec−1.

Fv/Fm0.822 ± 0.0100.722 ± 0.0440.634 ± 0.025
ΦPSII0.543 ± 0.0470.458 ± 0.0530.380 ± 0.026
Y(NO)0.337 ± 0.0240.358 ± 0.0270.361 ± 0.033
NPQ0.391 ± 0.0460.460 ± 0.0480.781 ± 0.063
qL0.351 ± 0.0400.527 ± 0.0250.609 ± 0.056
qP0.697 ± 0.0510.752 ± 0.0320.762 ± 0.046
Chlorophyl a (nmol g−1 FW)614.5 ± 23.7232.0 ± 15.9172.4 ± 9.1
Chlorophyl b (nmol g−1 FW)180.9 ± 12.973.2 ± 6.551.5 ± 6.3
Chlorophyl a/chlorophyl b3.4 ± 0.33.2 ± 0.23.4 ± 0.3
Mg-ProtoIX (pmol g−1 FW)169.9 ± 51.964.9 ± 21.713.0 ± 1.1
Mg-ProtoIX-ME (pmol g−1 FW)123.7 ± 38.521.9 ± 12.817.2 ± 2.1

To confirm the predicted chloroplast localization in vivo, we constructed a PRIN2 yellow fluorescence protein fusion protein (PRIN2:YFP). The YFP signal clearly overlapped with the autofluorescence from chlorophyll, confirming the in silico prediction of a plastid localization (Figure 3). Furthermore, the YFP signal co-localized in the chloroplast with a PEND:CFP construct. PEND binds chloroplastic and nuclear DNA (Sato et al., 1993; Arsova et al., 2010), indicating that PRIN2 was localized to the chloroplast nucleoids (Figure 3). To further confirm this, PRIN2 was also shown to co-localize with pTAC12, a component of plastid transcriptionally active chromosomes (Figure S5) (Pfalz et al., 2006).

Figure 3.

 Subcellular localization of PRIN2. The PRIN2:YFP fusion protein is localized to the chloroplast and co-localizes with PEND:CFP. The PRIN2:YFP and PEND:CFP constructs were transformed into Arabidopsis protoplasts and visualized using confocal microscopy. PEND binds chloroplastic and nuclear DNA, and the strong blue CFP signal represents the nucleus. The autofluorescence from chlorophyll is in red. Scale bars: 10 μm.

The prin2 mutants demonstrated lower expression levels of genes transcribed by PEP

To investigate if expression of plastid-encoded genes was impaired in the prin2 mutants, analysis of the complete plastid transcriptome was performed in prin2-1, prin2-2 and wild-type seedlings grown under control conditions (150 μmol photons m−2 sec−1, CL). Compared with the wild type, both prin2 mutants showed very low expression levels of plastid-encoded genes associated with photosynthesis (Figure 4; Table S1). The plastid psbA and psaA genes showed reduced expression levels in the prin2 alleles, compared with the wild type, and also in rosette plants (Figure S6), demonstrating that PEP activity is impaired throughout the life cycle of the prin2 mutants. In contrast to the photosynthesis genes, increased levels of mRNAs of plastid housekeeping genes and the genes encoding PEP subunits were observed in the prin2 mutants (Figure 4). Other mutants with impaired PEP activity, such as plastid transcriptionally active chromosome 2 (ptac2), chloroplast biogenesis 19 (clb19), yellow seedlings 1 (ys1) and Δrpo, show expression profiles similar to the prin2 mutants (Hess et al., 1994; Pfalz et al., 2006; Chateigner-Boutin et al., 2008; Zhou et al., 2009). Thus, the plastid gene expression profile suggests that PEP activity is strongly impaired in the prin2 mutants. In addition, network analysis of expression patterns by mutual rank (MR) coefficient demonstrated that PRIN2 is co-expressed with several of the PTAC genes and other components of plastid transcriptionally active chromosomes (Table S2). To generate the co-expressed gene networks, the three genes having the highest MR are connected with each other (Kinoshita and Obayashi, 2009; Obayashi and Kinoshita, 2009), and the genes are sorted and ranked according to their MR (Table S2).

Figure 4.

 Plastid gene expression in the prin2 mutants. Seven-day-old seedlings were grown in 150 μmol photons m−2 sec−1 and sampled. Using real-time PCR, the expression of 78 chloroplast genes was investigated. The genes were classified by their chloroplastic function. All expression data represent mean (±95% CI) from at least three independent biological replicates. Light-grey bars represent prin2-1 and grey bars represent prin2-2. The expression shown is compared with wild-type levels.

LHCB and RBCS expression levels were released in the prin2-1 gun1-1 and prin2-2 gun1-1 double mutants

We created double mutants between the prin2 mutants and gun1 to investigate the relation between PRIN2 and GUN1. GUN1 is a pentatricopeptide repeat (PPR) protein involved in retrograde communication, and GUN1 is proposed to function downstream of several plastid signals, including the PGE signal triggered by inhibited plastid gene expression (Koussevitzky et al., 2007). GUN1 was shown to localize to the chloroplast nucleoids (Koussevitzky et al., 2007), indicating that, like PRIN2, GUN1 is also associated with sites of active transcription in the plastid. However, the expression of the plastid-encoded psbA and psaA genes in gun1-1 was similar to the wild type (Figure 5a), as was expression of all other plastid transcripts (Figure S7). Furthermore, the low expression levels of psbA and psaA in the prin2 mutants were not enhanced or rescued by the gun1-1 mutation in the double mutants (Figure 5a). No differences were observed between the single prin2 mutants and the prin2 gun1-1 double mutants, confirming that the gun1-1 mutation has no effect on plastid gene expression in seedlings.

Figure 5.

 The gun1 mutation rescues the suppressed LHCB and RBCS expression in prin2. Expression levels of photosynthesis genes encoded in the plastid (a) (psaA, AtCg00350; psbA, AtCg00020; rpoB, AtCg00190), in the nucleus (b) (LHCB1.1, At1g29920; LHCB2.4, At3g27690; RBCS, At1g67090) and in 7-day-old seedlings grown in control light conditions (CL, 150 μmol photons m−2 sec−1). Relative expression was calculated using RCE1 (At4g36800) as a reference gene. Data represent means (±95% CI) from at least three independent biological replicates.

Expression of LHCB1.1, LHCB2.4 and RBCS was investigated in seedlings grown under control conditions. The expression of the photosynthesis-associated genes in the gun1 mutant was similar to that in wild type, whereas the expression level in both prin2 alleles was low compared with the wild type (Figure 5b). We used the prin2 gun1-1 double mutants to test if the expression levels observed in the prin2 mutants could be rescued by a dysfunctional GUN1. In seedlings of the prin2-1 gun1-1 and prin2-2 gun1-1 double mutants, the expression of LHCB1.1, LHCB2.4 and RBCS indeed reverted to wild-type levels (Figure 5b). Thus, GUN1 appears to be involved in the regulation of the expression of nuclear encoded photosynthesis genes only, and not in the expression of plastid-encoded genes.

PRIN2 is specific to the HL-mediated retrograde signalling pathway

The prin2 mutants displayed a genome-uncoupled (gun) phenotype in response to HL. We investigated whether the prin2 mutants also demonstrated a classical gun phenotype when grown on norflurazon (Susek et al., 1993). Both prin2 mutants showed slightly reduced repression of LHCB expression compared with wild type when grown on norflurazon (Figure 6a). However, the gun phenotype in the prin2 mutants was not as strong as the phenotype observed in gun5 and gun1 when grown on norflurazon (Figure 6a). The gun phenotype on norflurazon has been linked to reduced flux through the tetrapyrrole biosynthetic pathway and accumulation of specific tetrapyrroles (Larkin et al., 2003; Strand et al., 2003; Nott et al., 2006; Ankele et al., 2007; Mochizuki et al., 2008). The prin2 mutants showed significantly lower levels of tetrapyrroles (Table 1), and the weak gun phenotype following norflurazon treatment of prin2-1 and prin2-2 could be an effect of impaired tetrapyrrole biosynthesis (Mochizuki et al., 2008; Zhang et al., 2011). In contrast to gun5, gun1 demonstrates impaired LHCB repression when exposed to the plastid translation inhibitor chloramphenicol (Koussevitzky et al., 2007). The repression of LHCB in the prin2 mutants following chloramphenicol treatment was similar to that in wild type and the gun5 mutant (Figure 6b). The chloramphenicol treatment therefore revealed a response that was specific to the gun1 mutant.

Figure 6.

 The prin2 mutants show impaired retrograde signalling following growth on norflurazon. Expression levels of LHCB1.1 (At1g29920) on (a) norflurazon and (b) chloramphenicol in 7-day-old seedlings compared with seedlings grown in control conditions. Relative expression was calculated using RCE1 (At4g36800) as a reference gene. Data represent means (±95% CI) from at least three independent biological replicates.

A functional PEP enzyme is required for correct LHCB expression in response to HL

Following exposure to HL, prin2 mutants showed misregulation of LHCB1.1 expression (Figure 2e). The plastid transcriptome analysis revealed that the expression profiles of the prin2 mutants were similar to those of mutants with impaired PEP activity (Pfalz et al., 2006; Chateigner-Boutin et al., 2008; Zhou et al., 2009). To examine if a functional PEP enzyme is required for the correct expression of nuclear-encoded photosynthesis genes in response to HL, we investigated the response to HL in ys1-1 (Zhou et al., 2009) and tic40 mutants (Chou et al., 2003). The YS1 gene encodes a PPR protein responsible for editing site 25992 in the rpoB transcript, and ys1-1, lacking this editing event, has impaired PEP activity (Zhou et al., 2009). The ys1-1 mutant is only mildly affected and is able to sustain autotrophic growth, like the prin2 alleles but unlike any of the mutants totally lacking PEP activity (Pfalz et al., 2006; Arsova et al., 2010). Thus, the ys1 mutant was selected as a complement to the prin2 alleles in this experiment. TIC40 is primarily involved in the import of chloroplast membrane proteins, and the mutation does not directly affect PEP activity, even though chloroplast development is severely affected (Figure S8) (Chiu and Li, 2008). The response to HL was also investigated in gun1 and gun5.

Wild type, prin2-1, prin2-2 and the collected mutants were grown for 7 days in CL and then exposed to HL for 3 h. Following HL exposure, LHCB1.1 and LHCB2.4 expression was strongly repressed in the wild type (Figure 7). The repression of LHCBs following HL treatment in gun1 and gun5 was similar to the repression observed in the wild type (Figure 7), suggesting that GUN1 and GUN5 are not involved in this retrograde signalling pathway. Furthermore, tic40 demonstrated repression of LHCB1.1 and LHCB2.4 similar to wild type (Figure 7). In contrast, the mutants with impaired PEP activity, ys1-1, prin2-1 and prin2-2, demonstrated reduced repression of LHCB following exposure to HL compared with the wild type (Figure 7). Thus, the mutant analyses confirmed that PEP activity is important for these characteristic changes in LHCB expression in response to excess light.

Figure 7.

 Mutants with impaired plastid-encoded RNA polymerase (PEP) activity show reduced repression of LHCB genes following exposure to high light levels (HL). Expression levels of LHCB1.1 (At1g29920) and LHCB2.4 (At3g27690) in 10-day-old seedlings after 3 h of HL treatment (1000 μmol photons m−2 sec−1). Expression levels are compared with the respective control and relative expression was calculated using RCE1 (At4g36800) as a reference gene. Data represent means (±95% CI) from at least three independent biological replicates.

To further investigate the mechanism involved in the PEP-dependent HL-triggered plastid signal, we exposed our prin2-1 gun1-1 and prin2-2 gun1-1 double mutants to HL. No difference in LHCB expression could be observed between the single prin2-1 and prin2-2 mutants and the prin2-1 gun1-1 and prin2-2 gun1-1 double mutants following exposure to HL (Figure 7). Thus, the double mutants also demonstrated a genome-uncoupled phenotype in response to HL. Unlike the prin2 single mutants, the prin2-1 gun1-1 and prin2-2 gun1-1 double mutants showed wild-type LHCB expression in the control (Figures 5b and 7), so our results suggest that impaired PEP activity alone is responsible for the misregulation of LHCB expression following HL exposure (Figure 7). The misregulation of LHCB expression in response to HL was maintained throughout the life of the plant in the prin2 single mutants and the prin2 gun1-1 double mutants, and in ys1-1, as demonstrated by the results from the rosette plants (Figure S6). This regulation is independent of GUN1, both in seedlings and in mature rosette plants.

The PEP-dependent HL-triggered plastid signal is linked to PET activity

To investigate if the HL-triggered plastid signal impaired in the rin2 mutants is linked to PET activity, we exposed seedlings to two different inhibitors of photosynthetic electron transfer: 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) and 2,5-dibromo-3-methyl-6-isopropyl-benzoquinone (DBMIB). DCMU blocks the flow of electrons from PSII to PQ, leaving PQ oxidized, and DBMIB inhibits electron transfer from PQ to the cytochrome b6f complex, resulting in a more reduced intersystem. Wild-type, prin2-1 and prin2-2 seedlings were treated with the inhibitors, and LHCB expression was determined and compared with control conditions (Figure 8), PET activity was monitored by chlorophyll a fluorescence to confirm the inhibition of PET (Figure 9, Figure S9). In the wild type treatment with DCMU resulted in induction, and treatment with DBMIB resulted in the repression of LHCB1.1 and LHCB2.4 expression (Figure 8a,b). This is consistent with previous results for these genes in response to DBMIB and DCMU (Escoubas et al., 1995; Bräutigam et al., 2007). Similarly to the HL response, the prin2 alleles were shown to be insensitive to treatment with DCMU and DBMIB regarding the regulation of LHCB1.1 and LHCB2.4 (Figure 8a,b). The expression of psbA was also insensitive to the treatment with the inhibitors in the prin2 mutants (Figure 8a,b). We clearly demonstrate that in the prin2-2 mutant both HL exposure and treatments with the inhibitors (DCMU and DBMIB) generate a change in the redox status of PET (Figure 9). All the measured photosynthetic parameters (ΦPSII, NPQ and qL) confirmed the expected changes in redox status of PET following HL exposure and inhibitor treatments both in the wild type and in the prin2 mutant (Figure 9). This suggests that the HL-triggered PRIN2-dependent plastid signal is linked to PET activity, and to the redox status of PQ and PET.

Figure 8.

 The plastid-encoded RNA polymerase (PEP)-dependent signal triggered by high light (HL) is linked to photosynthetic electron transport (PET) activity. Expression levels of LHCB1.1 (At1g29920), LHCB2.4 (At3g27690) and psbA (AtCg00020) in 10-day-old seedlings treated with inhibitors of photosynthetic electron transport: (a) 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) and (b) 2,5-dibromo-3-methyl-6-isopropyl-benzoquinone (DBMIB). Relative expression was calculated using RCE1 (At4g36800) as the reference gene. Data represent means (±95% CI) from at least three independent biological replicates. (c) The variable to maximal fluorescence ratio (Fv/Fm) following 6 h of exposure to HL and after recovery at CL for another 6 h. Data represent means (±SDs) from at least three independent biological replicates.

Figure 9.

 Effect of 2,5-dibromo-3-methyl-6-isopropyl-benzoquinone (DBMIB) and 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) on photosynthetic electron transport (PET) activity. Ten-day-old seedlings of Col (a, c and e; circles) and prin2-2 (b, d and f; triangles) were treated with DBMIB (dark-grey symbols), DCMU (light-grey symbols) and water (empty symbols), and chlorophyll fluorescence parameters φPSII, NPQ and qL were recorded using growth light intensity as actinic light (AL). Measurements conducted with high light (HL; 1000 μmol photons m−2 sec−1) are displayed with closed black symbols. Data represent means (±SDs) of at least seven measurements from independent plants.

Analysis of Fv/Fm following HL exposure showed that the prin2 mutants were severely affected by the HL treatment, as demonstrated by a strong photoinactivation of PSII and a drop in Fv/Fm (Figure 8c). The ability to recover following exposure to HL was also severely impaired in the mutants compared with the wild type, especially in the prin2-2 mutant (Figure 8c).

Discussion

The prin2-1 mutant was isolated in a screen designed to identify mutants with impaired regulation of expression LHCB1.1 in response to high redox pressure and inhibition of PET activity in the chloroplasts. The prin2 mutant alleles exhibit a clear genome-uncoupled phenotype in response to HL, and this phenotype is maintained throughout the life of the prin2 mutant plants. PRIN2 encodes a protein localized to the plastid nucleoids that promotes PEP activity, linking redox regulation of LHCB expression to PEP activity in the chloroplast. PRIN2 co-localizes with PTAC12 (Figure S5), which is one of 18 novel plastid proteins that were identified in transcriptionally active chromosomes (Pfalz et al., 2006). pTAC2, pTAC6 and pTAC12 were shown by analysis of their corresponding Arabidopsis T-DNA insertion lines to be essential for chloroplast development and autotrophic growth. The phenotypes and plastid gene expression profiles of the prin2 mutants resemble those of the ptac mutants (Pfalz et al., 2006). All show lower expression levels of class-I genes with PEP promoters, compared with the wild type, whereas transcripts from class-III genes with NEP promoters accumulate, compared with the wild type (Figure 4) (Pfalz et al., 2006). The expression profiles observed in the ptac and prin2 mutants are similar to those reported for the Δrpo mutant and other PEP-deficient mutants (Hess et al., 1994; Allison et al., 1996; Hajdukiewicz et al., 1997). The lower expression levels of PEP-dependent photosynthesis-related plastid genes in the prin2 mutants and the localization to the nucleoids suggest that PRIN2 is a component of the plastid transcription machinery, and that it plays a role in the control of plastid gene expression (PGE).

It is clear that PRIN2 is required for the correct expression of PEP-dependent genes, although PRIN2 has not been identified as a component of the TACs. However, preparations of TAC and sRNAP from different types of plant material have different protein composition, suggesting that the components associated with PEP change in response to developmental signals and changes in the environment (Reiss and Link, 1985; Pfannschmidt and Link, 1994; Suck et al., 1996). The large number of proteins associated with TAC and the PEP complex, and the variation in the composition of the complex, suggests that regulation of plastid gene expression is both complex and sophisticated. It is possible that PRIN2 has a regulatory role and is not a permanent core component of the PEP complex. The phenotype of the prin2 mutants also suggests that PRIN2 is not an essential component of the TAC, because the prin2 mutants are able to survive the seedling stage, unlike ptac mutants (Figure S3). Furthermore, PRIN2 is a plant-specific protein, suggesting that PRIN2 evolved relatively late compared with the eubacterial core subunits of the PEP complex. Proteins that evolved late are often involved in regulatory functions that coordinate different processes in the cell (Pfannschmidt and Liere, 2005).

Plastid gene expression (PGE) is essential during chloroplast development and for the initiation of LHCB expression. The prin2 mutant alleles showed constitutively low expression levels of PEP-dependent genes compared with the wild type (Figures 4, 5a and S6), and repressed LHCB expression compared with wild type under control conditions in seedlings (Figure 5b). This effect on LHCB expression in seedlings is consistent with the known effects of inhibitors of plastid transcription and translation: they inhibit the induction of expression of genes encoding light-harvesting complex apoproteins (LHCs) and the small subunit of Rubisco (RBCS) (Oelmuller et al., 1986; Rapp and Mullet, 1991; Sullivan and Gray, 1999). This effect on nuclear gene expression was released in the prin2 gun1-1 double mutants (Figure 5b). The gun1 mutant, in contrast to the other gun mutants (gun2-5), displays a strong genome uncoupled phenotype when treated with inhibitors of plastid translation (Figure 6b) (Koussevitzky et al., 2007). GUN1 is a member of the P subfamily of PPR proteins. In addition to the PPR motifs, GUN1 also has a small mutS-related (SMR) domain (Koussevitzky et al., 2007). There are five other Arabidopsis proteins with a domain structure similar to GUN1, and one of those is pTAC2 (Pfalz et al., 2006). GUN1-YFP and pTAC2-CFP were shown to co-localize (Koussevitzky et al., 2007), indicating that like PRIN2 and pTAC2, GUN1 is also associated with sites of active transcription in the plastid. However, in contrast to ptac2 and prin2, the gun1 mutant demonstrated normal plastid gene expression (Figure S7). Our results demonstrate that GUN1 mediates the prin2-induced repression of LHCB and RBCS expression, without being involved in the expression or regulation of plastid genes.

Plastid gene expression (PGE) is under tight redox control (Pfannschmidt and Liere, 2005). Imbalances in the redox state of PET occur when the absorption and transformation of light by the photochemical reactions of photosynthesis either exceed the capacity to use the photosynthetic electrons for reductive C, N and S metabolism, and/or exceed the capacity of the photosynthetic apparatus to dissipate the excess energy as non-photochemical quenching (Huner et al., 1998; Pfannschmidt, 2003). A thioredoxin (TRXz) was identified as a component of the TAC (Pfalz et al., 2006), and is suggested to be involved in the redox regulation of PEP (Arsova et al., 2010). The redox state of TRXz could regulate kinases that, in turn, regulate PEP activity during dark-to-light transitions (Arsova et al., 2010). Phosphorylation of sigma factors, as well as phosphorylation of PEP itself, has been shown to regulate plastid gene expression (Baginsky et al., 1997; Shimizu et al., 2010). In a yeast-two-hybrid (Y2H) screen, two fructokinase-like proteins (FLNs) were identified as TRXz targets. The FLN1, FLN2 and TRXz interactions were confirmed in planta (Arsova et al., 2010), and in support of the interaction, both FLN1 and FLN2 were found as components of TAC (Pfalz et al., 2006). The involvement of TRXz and the kinases imply that phosphorylation works synergistically with thiol modifications to regulate PEP activity (Steiner et al., 2009). Our results suggest that PRIN2 is involved in the light-regulated control of PEP activity and transcription of chloroplast photosynthesis genes (Figure 9). However, PRIN2 does not appear to be directly linked to the proposed TRXz/FLN-mediated thiol modification and phosphorylation loop, as no direct interaction between TRXZ, FLN1 and PRIN2 could be detected in a Y2H assay (Figure S10). In addition, PRIN2 contains no putative kinase or phosphatase domains.

The misregulation of LHCB1.1 and LHCB2.4 expression in response to HL exposure in the prin2 mutants is linked to the redox state of PET (Figure 8, 9). The redox state of PET influences a large number of processes in the plant other than plastid gene expression, such as altering the excitation distribution between photosystems through state transitions (Bellafiore et al., 2005), nuclear gene expression (Pfannschmidt, 2003; Fernandez and Strand, 2008), and also plant growth and morphology (Gray et al., 1997). In addition, the redox state of PET has been suggested to act as a sensor of cellular energy status (Huner et al., 1998), although the mechanisms by which information about the redox changes to PET are transduced to the nucleus to initiate changes to the transcriptome have remained elusive. Accumulation of ROS could potentially influence the response in the prin2 mutants, with the different ROS species activating distinct signalling pathways (Fernandez and Strand, 2008). For example, the prin2-1 mutants show a clear misregulation of ASCORBATE PEROXIDASE2 (APX2) in response to HL (Figure S4). APX2 expression has been shown to be regulated by both redox changes in PET and H2O2 (Karpinski et al., 2003). The high light exposure and treatments with the inhibitors used in this study also generate ROS accumulation. As a consequence it is very difficult to discriminate between redox changes and H2O2 accumulation, and to determine what is the source of the signal. Detailed experiments using light of specific wavelengths (Pfannschmidt, 2003), and/or to monitor the ROS levels, could possibly elucidate the exact mechanisms involved in regulating the PEP complex. Furthermore, the plant hormone abscisic acid (ABA) has been found to accumulate following HL treatment, and to be involved in retrograde signalling and the regulation of HL-responsive genes, including LHCB (Koussevitzky et al., 2007; Rossel et al. 2002). ABA biosynthesis is dependent on functional chloroplasts, and the biosynthesis could possibly be impaired in the prin2 mutants. It has also been suggested that there is an interaction between sugar and retrograde signalling pathways controlling the expression of LHC genes via the plastid redox state (Oswald et al., 2001). Thus, other components could be involved as the PRIN2 pathway is not necessarily linear, and PRIN2 could be part of a wider network of signalling that includes light, ROS and sugar signalling.

Our results suggest that a fully functional PEP complex is important for the correct LHCB expression in response to redox changes to PET. Interestingly, this redox-triggered PEP-dependent signal is not mediated by GUN1. Reduced repression of LHCB expression, compared with the wild type, following exposure to HL was found to be specific to chloroplast mutants with impaired PEP activity (Figure 7), and the genome-uncoupled phenotype in response to HL was also maintained in mature plants in those mutants (Figure S6). During seedling development, LHCB expression is not induced if PEP activity is inhibited. This regulation of nuclear gene expression is mediated by GUN1. During the response to HL a different aspect of the PEP activity, requiring PRIN2 but not GUN1, is communicated to the nucleus to regulate LHCB expression in response to changes in the redox status of PET. Thus, the status of the PEP enzyme links the functional state of the chloroplast to the nucleus, enabling the plant to synchronize the expression of photosynthetic genes from the nuclear and chloroplast genomes.

Experimental Procedures

Plant material and growth conditions

Seedlings of Arabidopsis thaliana were grown on full-strength MS medium, including 2% sucrose. All genotypes are in the Colombia ecotype. For the isolation of prin2-1, seedlings were grown in continuous light (100 μmol photons m−2 sec−1). The T-DNA insertion line, prin2-2 (GABI_772D02), was ordered from CeBiTech (Bielefeld University), and the insertion was confirmed with PCR-based genotyping (Table S4). Seedlings and rosette plants were grown in long-day conditions (16-h light/8-h dark) at 150 μmol photons m−2 sec−1, unless otherwise stated. Exposure to HL (3 h, 1000 μmol photons m−2 sec−1) started in the middle of the light period, and the control was sampled at the same time point. Protoplasts were isolated from 14-day-old seedlings following the method described by Zhai et al. (2009). Inhibitor concentrations were used as follows: 0.5 μm norflurazon, 200 μg ml−1 chloramphenicol, 50 μm DCMU for 4 h, 100 μm DBMIB for 6 h, and the DBMIB was reapplied every second hour according to the method described by Pfannschmidt et al. (2001). Ten-day-old WT and mutant seedlings were transferred from MS plates to plates with Whatman filters soaked with corresponding inhibitors, or water for the control conditions. Once there, seedlings were also sprayed with the inhibitor at the given concentration at the beginning of the experiment (3 h into the light period).

Isolation, positional cloning and complementation of prin2-1

The regulatory part of the LHCB1.1 promoter (199 bp) was fused to the LUCIFERASE gene in the AtM Domega vector, and the construct was transformed into A. thaliana using the floral-dip method (Clough and Bent, 1998). Seeds from a stable line (6-3) with a single LHCB1.1::LUC insertion were EMS mutagenized. About 15 000 7-day-old seedlings were screened for high LUC activity after HL exposure using a Typhoon scanner (GE Healthcare, http://www.gelifesciences.com). The prin2-1 mutant was back-crossed three times to abolish other mutations. A mapping population was generated by crossing prin2-1 with ecotype Landsberg erecta, and 512 F2 seedlings with the prin2-1 phenotype were used for high-resolution mapping according to (Lukowitz et al., 2000). The exact position of the prin2-1 mutation was determined using SOLiD™ System Sequencing (Uppsala Genome Centre). For prin2-1 mutant complementation, a PCR fragment of 540 bp containing the complete coding sequence of At1g10522 was amplified from Col-0 cDNA (forward, 5′-attB1- ATGGCTTCAATGCACGAAGCTC-3′; reverse, 5′-attB2- CTAATCAGTGCCGGTCCATTCC-3′). Using Gateway technology, the fragment was subcloned into pDONR207 (Invitrogen, http://www.invitrogen.com) and later cloned into the binary vector pH2GW7 (Karimi et al., 2002). prin2-1 plants were then transformed with pH2GW7/At1g10522 using the floral-dip method (Clough and Bent, 1998).

SOLiDTM sequencing and mutation identification

A 30-μg portion of DNA was used to construct SOLiD3 mate-pair libraries according to the manufacturer’s instructions. DNA was sheared into fragments of about 2.5 kb by HydroShear (Genomic Solutions), and end-repaired using End Polishing Enzyme 1 and 2. Cap adaptors (5′-pACAGCAG-3′ and 5′-CATGTCGTCp-3′) are ligated to both ends of the fragments. Next, the adapter-ligated DNA sample was separated on a 0.8% agarose gel, and DNA fragments of ∼2.5 kb in length were recovered and purified. The sample was circularized using a biotinylated internal adaptor, nick translated with Escherichia coli DNA polymerase 1 and digested with T7 exonuclease and S1 nuclease. Digested DNA was end-repaired using End Polishing Enzyme 1 and 2 and bound on streptavidin beads. P1 (5′-CCACTACGCCTCCGCTTTCCTCTCTATGGGCAGTCGGTGAT-3′ and 5′-ATCACCGACTGCCCATAGAGAGGAAAGCGGAGGCGTAGTGGTT-3′) and P2 adaptors (5′-AGAGAATGAGGAACCCGGGGCAGTT-3′ and 5′-CTGCCCCGGGTTCCTCATTCTCT-3′) are ligated to the fragments. The library was further nick-translated followed by PCR-based amplification, and released from the beads. PCR products were separated on a 4% agarose gel and a library band of 250–350 bp was recovered. Throughout the library preparation procedure, DNA was purified and concentrated with QIAquick columns (Qiagen, http://www.qiagen.com) after each enzymatic reaction and PCR. Emulsion PCR was performed according to the manufacturer’s manual (SOLiD3 System Templated Bead Preparation Guide; Applied Biosystems, http://www.appliedbiosystems.com), before SOLiD sequencing. Subsequently, 50-nt mate-pair sequences were collected on the AB SOLiD3 instrument. Reads were mapped to the reference and paired using corona_lite 0.40r2.0. Single nucleotide polymorphisms (SNPs) were identified using the consensus SNP pipeline from the same software suite. The analysis of all SNPs for EMS-induced substitutions (G → A or C → T) revealed one change within 100 kb both up- and down-stream of the physical mapping region containing no heterozygous SNPs: a G → A transition at position 3 471 276 (TAIR9) within At1g10522.

Fusion protein constructs and microscopy analysis

For subcellular localization of PRIN2, the full-length cDNA, including the predicted transit peptide, was amplified from Col-0 cDNA with primers: forward, 5′-CATGACCGGTAATGGCTTCAATGCACGAAGCTC-3′ and reverse, 5′-CATGACCGGTGCATCAGTGCCGGTCCATTCCAGT-3′. The amplified fragment was cloned in the pEYFP vector (Seidel et al., 2005). Protoplasts were transformed as described by Zhai et al. (2009). YFP and CFP signals were monitored using an SP2 confocal laser scanning system equipped with an inverted microscope (Leica, http://www.leica.com).

HPLC analysis and chlorophyll determination

Chlorophyll content was determined as previously described (Porra et al., 1989). HPLC analyses were performed according to the method described by (Mochizuki et al., 2008). Leaf material was homogenized in acetone: 0.1 m NH4OH (8:2, v/v). Column eluent was monitored by UV detection and tetrapyroles were identified and quantified using authentic standards. Mg-protoIX and Mg-ProtoIX methylester were purchased from Frontier Scientific (http://www.frontiersci.com).

PSII photochemistry

In vivo chlorophyll fluorescence was measured using a modulation fluorometer PAM 101–103 (Walz, http://www.walz.com) from the adaxial side of excised leaf material. The nomenclature of van Kooten and Snel (1990) was used for the parameters of Chl fluorescence.

RNA isolation, cDNA synthesis and real-time PCR

Total RNA was isolated using the Plant RNA Mini Kit (Omega, http://www.omega.com). cDNA was synthesized using the iScript cDNA Synthesis Kit (Bio-Rad, http://www.bio-rad.com). cDNA was used in an iQ SYBR Green Supermix reaction (Bio-Rad). At least three biological replicates were used for each data point and all reactions were performed in technical triplicate. RT-PCR was run in a CFX96 real-time system (Bio-Rad) and monitored by using the cfx manager (Bio-Rad) (Table S3). Data were analyzed by using LinRegPCR (Pfaffl, 2001; Ramakers et al., 2003). Plastid transcriptome analysis was performed on 7-day-old seedlings, following the method of Chateigner-Boutin et al. (2008).

Acknowledgements

We thank Dr F. Börnke for the PEND:CFP construct and the Arabidopsis stock centre for the T-DNA insertion lines. Tatjana Kleine and Jannice Örnmark are acknowledged for their help with the mutant screen. This work was supported by grants from STINT, the Swedish research foundation, VR and the FFL-grant from Foundation for Strategic Research, SSF (ÅS) and the Australian Research Council (IS).

Ancillary