Identification of an OPR protein involved in the translation initiation of the PsaB subunit of photosystem I

Authors


(e-mail jean-david.rochaix@unige.ch).

Summary

Genetic analysis of mutants deficient in the biosynthesis of the photosystem I complex has revealed several nucleus-encoded factors that act at different post-transcriptional steps of chloroplast gene expression. Here we have identified and characterized the gene affected in the tab 1-F15 mutant, which is specifically deficient in the translation of the photosystem I reaction center protein PsaB as the result of a single nucleotide deletion. This gene encodes Tab 1, a 1287 amino acid protein that contains 10 tandem 38–40 amino acid degenerate repeats of the PPPEW/OPR (octatricopeptide repeat) family, first described for the chloroplast translation factor Tbc2. These repeats are involved in the binding of Tab 1 to the 5′-untranslated region of the psaB mRNA based on gel mobility shift assays. Tab 1 is part of a large family of proteins in Chlamydomonas that are also found in several bacteria and protozoans, but are rare in land plants.

Introduction

The primary reactions of photosynthesis in algae and plants occur in the thylakoid membrane and are mediated by the two photosystems PSI and PSII, and their light-harvesting systems. They are connected in series through the electron transport chain, which includes the plastoquinone pool, the cytochrome b6f complex and plastocyanin. Absorbed light excitation energy is used by the two photosystems to create charge separations across the membrane, with PSII oxidizing water at one end of the chain and PSI reducing ferredoxin at the other end, followed by the reduction of NADP+ to NADPH. Electron transport is coupled with proton translocation across the membrane, giving rise to a proton gradient that is used by the ATP synthase to generate ATP. Both ATP and NADPH drive the Calvin–Benson cycle that promotes the assimilation of CO2 and the formation of sugars.

Each of the photosynthetic membrane complexes contains numerous subunits of nuclear and chloroplast genetic origin (Eberhard et al., 2008; Stern et al., 2010). The mRNAs of nucleus-encoded subunits are translated on cytoplasmic ribosomes, and their products are imported into chloroplasts, whereas the mRNAs of chloroplast-encoded subunits are translated on chloroplast ribosomes, and finally assembled together with their nucleus-encoded partners into functional photosynthetic complexes. Whereas the composition, structure and function of these complexes are well understood, the mechanisms underlying their coordinate assembly are still largely unknown.

A genetic analysis of numerous mutants of Chlamydomonas, Arabidopsis and Zea mays (maize) that are deficient in photosynthesis or in pigmentation has revealed a large set of nuclear loci that are involved in different post-transcriptional steps of chloroplast gene expression, such as RNA processing and stability, RNA editing, splicing, translation and assembly of the complexes in the thylakoid membrane (Eberhard et al., 2008; Stern et al., 2010). These loci encode factors that act mostly in a gene-specific manner in Chlamydomonas, whereas in land plants some of them are involved in the expression of several chloroplast genes, and a loss of these factors generally introduces more pleiotropic effects (Eberhard et al., 2008; Stern et al., 2010).

Photosystem I acts as a light-driven plastocyanin ferredoxin oxidoreductase (Nelson and Yocum, 2006). The PSI core complex consists of the two large reaction center polypeptides PsaA and PsaB and the small PsaC protein, which are encoded by the chloroplast genome. These three proteins bind most of the redox co-factors of PSI involved in electron transport, as well as many chlorophyll, carotenoid and lipid molecules. In addition, the complex contains nine or 10 subunits, PsaD–PsaN, which are generally nucleus-encoded except for PsaJ, which is chloroplast-encoded. In this study we have characterized tab 1-F15, a PSI-deficient nuclear mutant of Chlamydomonas. It is specifically affected in the translation of the two PSI reaction center polypeptides PsaA and PsaB, based on protein pulse-labeling experiments (Girard-Bascou et al., 1987). Further analysis with a chimeric chloroplast transgene, in which the psaB 5′ untranslated region (5′-UTR) was fused to a reporter gene, revealed that the reporter gene was expressed in the wild-type but not in the mutant nuclear background, indicating that this mutant is specifically affected at the level of initiation of translation of the chloroplast psaB mRNA (Stampacchia et al., 1997). Because PsaB appears to be the major anchor protein for the assembly of PSI, it is particularly important to identify the factors required for its expression and integration into the thylakoid membrane. Besides tab 1-F15, another mutant tab 2-F14 is also specifically unable to translate the psaB mRNA. The Tab 2 locus has been identified, and codes for a protein that is highly conserved in plants and cyanobacteria (Barneche et al., 2006; Dauvillée et al., 2003). Here, we show that Tab 1 encodes a large protein with 10 tandem 38–40 amino acid degenerate repeats with RNA binding properties. These octatrico peptide repeats (OPRs) are also found in other proteins involved in post-transcriptional steps of chloroplast gene expression. Together these proteins form a new family of proteins, and they are also present in other unicellular organisms and some bacteria, but are rare in land plants.

Results

Isolation of the nuclear gene deficient in the PSI mutant tab 1-F15

Previous work has shown that the tab 1-F15 mutant is specifically deficient in the translation of the PsaB subunit of PSI (Girard-Bascou et al., 1987; Stampacchia et al., 1997). To understand how the corresponding factor acts, a first step was to isolate its gene. The tab 1-F15 mutant was transformed with an indexed cosmid library of Chlamydomonas (Zhang et al., 1994) by selecting for photoautotrophic growth or for growth on acetate-containing medium under 60 μmol photons m−2 s−1. PSI mutants are light sensitive and unable to grow under these conditions. Only a single cosmid, called c35D7, could be identified that complemented the mutant (see Experimental procedures for details). Analysis of this cosmid revealed that it contained a 6-kb genomic NotI fragment that was considerably rearranged based on sequence comparison with the corresponding genomic locus. We therefore used an 889-bp NotI-NdeI fragment from c35D7 to rescreen the cosmid library. A new cosmid of 40 kb, called c102A3, was identified that rescued the F15 mutant with a high yield by transformation. To identify the gene affected in F15, dubbed Tab 1, the cosmid was subcloned in smaller fragments that were then tested for their ability to rescue the mutant by transformation on minimal medium. A 5.4-kb NotI fragment complemented the mutant and was used as a probe for isolating 3.6 kb of cDNA from a Chlamydomonas cDNA library, which was also able to rescue the mutant. A BLAST search of the Chlamydomonas genome with this cDNA as a probe revealed that the Tab 1 gene is located on chromosome 7 of the algal genome (Merchant, 2007) (http://genome.jgi-psf.org/chlamy/chlamy.info.html; GenBank, HQ683741).

Tab 1 encodes a PPPEW/OPR protein

The sequence of the full-length 4.8-kb Tab 1 open reading frame was assembled from the partial 3.6-kb cDNA and the 1.2-kb genomic region corresponding to the 3′ end of the gene, which was covered by an expressed sequence tag (EST) sequence (estExt_fgenesh2_pg.C_160172). Comparison of the genomic and cDNA sequence revealed five introns within the 6-kb Tab 1 gene with a 3′-UTR of 658 nucleotides (Figure 1). Data on the mapping of Tab 1 transcripts were also obtained from the ∼5 million-plus EST reads from Joint Genome Institute, University of California (JGI-UCLA) and Genoscope (http://genomes.mcdb.ucla.edu). In this way the 3′ end was confirmed and the 5′-UTR was estimated to be close to 280 nucleotides. High-throughput RNA sequencing (RNA-seq) revealed that Tab 1 is expressed at very low levels: ∼10-fold less than the average Chlamydomonas gene (http://genomes.mcdb.ucla.edu).

Figure 1.

 Structure of the Tab 1 gene and its product. The Tab 1 gene is shown with exons (black boxes) and introns (lines). The 5′ and 3′ UTRs are indicated as open boxes at both ends of the gene. The region encoding the PPPEW/OPR repeats is indicated by open yellow boxes at the top. Relevant restriction sites are indicated: St, StuI; A, AflII; X, XmnI; C, ClaI; S, SmaI; N, NotI. The Tab 1 protein amino acid sequence is shown below. Sites corresponding to the five intron insertions are indicated by wedges. The 10 PPPEW/OPR repeats arranged in tandem are indicated alternatively in red and blue and underlined. The tab 1-F15 mutation creates a stop codon at L670, highlighted in yellow. Residues highlighted in green (L299 and P952) indicate the ends of the fragment fused to the PsaD transit peptide for testing in vivo complementation. Residues highlighted in blue (E390 and F747) indicate the ends of the fragment used for the in vitro RNA binding experiments. Residues E771 and P866, highlighted in gray, encompass the end of the region that is essential for Tab 1 activity (see Table S2).

The Tab 1 gene encodes a protein of 1287 amino acids, with a predicted molecular weight of 129.093 (Figure 1) (GenBank ADY68544.1). It differs from the predicted protein Cre07.g348800.t1.2 in Augustus u10.2, a gene prediction set based on the Chlamydomonas reinhardtii genomic sequence, EST and RNA-seq data (http://genomes.mcdb.ucla.edu), by the presence of an additional exon (exon 2, encoding 53 amino acids; Figure 1). The protein is predicted by targetp to be localized in the chloroplast, with a putative pre-sequence of 27 amino acids. The protein has a high content of A (21%), S (10.7%), L (9.7%), P (9.0%) and G (8.6%). A BLAST search revealed that Tab 1 shares common features with Tbc2, a nucleus-encoded protein that is required for the translation of the psbC mRNA (Auchincloss et al., 2002), and with the chloroplast splicing factor Raa1 (Merendino et al., 2006). This common feature of Tab 1 is the presence of 10 tandem 38–40 amino acid PPPEW repeats located between residues 400 and 782 (Figures 1 and 2a). Repeats of this sort were first identified in Tbc2 (Auchincloss et al., 2002), and they are also present in Raa1 (Merendino et al., 2006) and Tda1 (Eberhard et al., 2011). A BLAST search of the C. reinhardtii nuclear genome with Tab 1, Tbc2 and Raa1 as queries revealed a total of 15 proteins that were significantly related to one of these proteins. These proteins were then analyzed for the presence of repeated sequence motifs. A multiple sequence alignment of these repeats was performed to derive a consensus sequence named PPPEW (Figure 2b) (for details, see Experimental procedures). Searching the entire Chlamydomonas proteome with the PPPEW profile identified 39 additional proteins with several repeats besides Tab 1, Tbc2, Raa1 and Tda1 (Figure 2c). These repeats were found in several copies, often arranged in tandem, in the 43 proteins (Figure 2c). Searching the UniprotKB/Swissprot database for proteins containing similar motifs with the mast search tool (http://meme.sdsc.edu/meme/intro.html) indicated that this motif is mainly found in protozoans, such as apicomplexans, and in a small family of proteins in the parasitic alpha-proteobacterium Coxiella burnetii. Another feature of Tab 1 that is shared with numerous Chlamydomonas proteins is the presence of several stretches of four or more Ser, Arg, Gln, Ala and Gly residues, which are interspersed throughout the protein.

Figure 2.

 PPPEW/OPR repeats of Tab 1 (a) Alignment of the 10 PPPEW/OPR repeats of Tab 1 (color scheme: clustalx, using 30% identity threshold). Note that in the last repeat two amino acids have been deleted. (b) Consensus PPPEW/OPR sequence derived from the repeats of 43 proteins from the PPPEW family. (c) The PPEW/OPR repeat family in Chlamydomonas reinhardtii. Numbers in parenthesis indicate the number of amino acids. The // symbol indicates that the protein sequence has been cut to fit the page.

Tab 1 lacks a single nucleotide in the tab 1-F15 mutant

Sequencing of the entire cDNA of Tab 1 in tab 1-F15 by reverse transcription polymerase chain reaction (RT-PCR) revealed a single base deletion within one of the PPPEW/OPR repeats that changes Leu670 to a stop codon (Figures 1 and S2). The 5.4-kb NotI genomic fragment that complements tab 1-F15 lacks the 187 amino acid C-terminal region of Tab 1, which is apparently not essential for the function of the protein. To further test which part of Tab 1 is essential for photoautotrophic growth, the 4.5-kb NotI fragment was digested with several restriction enzymes that cut at different sites within the Tab 1 gene. The digested DNA was then used to assess the transformation yield, which remained high for enzymes cutting downstream of the PPPEW/OPR repeats (Table S2). Digestion with ClaI, which cuts at a single site corresponding to P866, located 84 residues downstream of the last OPR repeat, did not significantly affect the transformation yield, suggesting that the remaining C-terminal 421 residues of Tab 1 are not essential (Figure 1). However, digestion with XmnI, which cleaves at a unique site corresponding to E771 within the last OPR repeat, significantly diminished the transformation yield, indicating that the region of Tab 1 essential for its activity ends between E771 and P866. These results suggest that the entire set of OPR repeats is required for the full activity of Tab 1. The protein contains several hydrophobic regions, some of which are predicted to be transmembrane domains by some, but not all, prediction programs.

Tab 1 accumulates to low levels and is highly unstable

In order to determine the level of Tab 1 an antibody was produced against a recombinant subfragment of Tab 1 (see Experimental procedures). Immunoblot analysis of total wild-type cell extracts with Tab 1 antiserum revealed two bands at 180 and 185 kDa, which were undetectable in tab 1-F15 and which are significantly larger than the predicted mass of 130 kDa (Figure 3). It is not clear whether the smaller form results from proteolytic cleavage or from authentic processing of the protein. The abnormal migration of Tab 1 may result from its long hydrophobic stretches. After separation of the soluble and insoluble fractions by centrifugation of the total cell extract, most of Tab 1 was found in the latter fraction, suggesting that it is associated with membranes or with the nucleoid (Figure 3). Attempts to localize the protein in the chloroplast after subcellular fractionation failed because the protein was highly sensitive to proteolysis, even in the presence of a cocktail of protease inhibitors. The number of Tab 1 molecules per cell was estimated at 12 000, and the molar ratio between Tab 1 and PsaD, a representative PSI protein, was 1/500 (see Appendix S1 for details; Figure S3).

Figure 3.

 Tab 1 is associated with the insoluble cell fraction. Total cell extracts (T) from wild-type and tab 1-F15 were prepared and separated into pellet (I) and supernatant (S) fractions. Proteins were fractionated by PAGE and immunoblotted with Tab 1 antibodies. The two TAB 1 bands at 180 and 185 kDa are indicated with arrowheads. I, insoluble fraction; S, soluble fraction; T, total proteins.

Because Tab 2 is also required for translation initiation of the psaB mRNA in C. reinhardtii, it is possible that this protein may interact with Tab 1 (Dauvillée et al., 2003).

However, attempts to show an interaction between these two proteins by co-immunoprecipitation failed. If these two proteins were part of the same complex the level of one of these proteins might be decreased in the absence of the other. Contrary to this expectation, the loss of Tab 2 in the tab 2-F14 mutant did not significantly affect the level of Tab 1.

Tab 1 is an RNA binding protein

The fact that Tab 1 is involved in the translation of the psaB mRNA and that its target is located within the psaB 5′-UTR raises the possibility that it is a RNA binding protein. To test this possibility, the Tab 1 cDNA was fused to a T7 promoter, but no recombinant protein could be detected in Escherichia coli. To circumvent this problem we fused the first nine PPPEW/OPR repeats of Tab 1 to the maltose binding protein (MBP-Tab 1; for details see Experimental procedures). Under these conditions ∼50% of the recombinant protein produced in E. coli was soluble. Incubation of increasing quantities of MBP-Tab 1 with 32P-labeled psaB 5′-UTR revealed efficient binding of the protein to the RNA, as determined by electrophoretic band shift assays (Figure 4a). We estimate the Kd of this interaction close to 33 nm based on the data shown in Figure 4(a). To test the specificity of binding, competition experiments were performed with an excess quantity of unlabelled psaB 5′-UTR or with an unrelated RNA originating from the pKS plasmid. Competition of RNA binding was observed with a 100-fold excess of psaB RNA, but not with KS RNA, indicating that Tab 1 shows RNA binding specificity (Figure 4b). To test the specificity amongst different chloroplast mRNAs, competition experiments were performed with the psbB and psbC 5′-UTRs (Figure 4c). Whereas the psbB 5′-UTR did not compete with the psaB 5′-UTR, the psbC 5′-UTR competed very efficiently, indicating that the Tab 1 RNA binding properties cannot account for the specificity of this protein for psaB mRNA translation. Further competition experiments with oligo rA, rG, rU and rC showed that only oligo rU competed with the binding of psaB RNA (Figure 4e). The sequence of the psaB 5′-UTR is indeed U-rich (43%), and contains U-rich tracts. However, one has to take into account that one of the PPEW/OPR repeats and all of the flanking protein regions were missing from the assayed protein, which might alter the RNA binding properties of Tab 1.

Figure 4.

 Tab 1 binds to the psaB 5′-UTR. (a) Electrophoretic mobility shift assay (EMSA) of recombinant PPEW-Tab 1 (E390–F747) and 32P-labeled psaB 5′-UTR (373 nucleotides). The increasing concentration of recombinant protein used is indicated in nm. U, unbound RNA probe; B, bound RNA probe. (b) RNA binding competition experiments. EMSA with 100 nm PPEW-Tab 1 and increasing quantities of unlabeled psaB 5′-UTR was performed as indicated. KS refers to a transcript from the Bluescript vector (150 nt). (c) RNA binding competition experiments with 200 nm PPEW-Tab 1 and increasing quantities of unlabeled psaB (373 nt), psbB (147 nt) and psbC (372 nt) 5′-UTR RNA. (d) RNA binding competition experiments with 100 nm PPEW-Tab 1 and increasing quantities of oligo rA, rG, rC and U, as indicated. The autoradiograms were quantified with a phosphorimager and the results are shown below. Numbers refer to the excess molar quantities of competitor RNA as compared with the labeled psaB 5′-UTR. P, 32P-labeled psaB 5′-UTR. The RNA binding experiments were repeated three times and gave rise to similar patterns.

Association of Tab 1 mRNA with large complexes

Because Tab 1 is likely to be involved in the initiation of translation of the psaB mRNA we compared the polysome profiles of Tab 1 in the wild type and tab 1-F15 mutant. After sucrose density centrifugation of the polysomes, RNA was isolated from the different fractions and hybridized with a labeled psaB probe (Figure 5). Comparison of the psaB mRNA distribution between the wild type and tab 1-F15 revealed a shift towards smaller sizes in the mutant. However, a significant portion of psaB mRNA was still associated with large complexes in tab 1-F15 (Figure 5, fractions 5–9). A similar observation was made by Eberhard et al. (2011) with atpA mRNA in the tda1-F54 mutant, which is deficient in the initiation of translation of this mRNA. These authors noticed that atpA mRNA is associated with large non-polysomic complexes that co-sediment with polysomes but are more resistant to EDTA. We therefore examined the sedimentation of psaB RNA in wild type and tab 1-F15 in the presence of EDTA. A drastic shift towards the top of the gradient occurred (Figure 5, fractions 2–4), indicating that the RNA protein complexes detected are all sensitive to EDTA. To further check the specificity of the tab 1-F15 mutation, the polysome profile of psbD mRNA was determined and shown to be very similar in the wild type and tab 1-F15 (Figure 5).

Figure 5.

 Comparison of polysome profiles in wild-type and F15-Tab 1 cells. Soluble cell extracts were fractionated by sucrose density gradient centrifugation in the presence of Mg2+ or EDTA, as indicated. RNA was isolated from the different fractions shown, blotted on filters and hybridized with psaB and psbD probes. The direction of sedimentation is indicated with an arrow.

Discussion

Tab 1 is involved in the translation of the PsaB photosystem I subunit

The origin of chloroplast genomes can be traced to an early symbiotic event in which a cyanobacterium invaded a eukaryotic cell and gradually lost its genetic autonomy by transferring the majority of its genes to the nucleus of the host. Chloroplast genomes therefore represent remnants of cyanobacterial genomes, and code for a limited set of chloroplast proteins involved in photosynthesis, plastid gene expression and in a few other functions. However, expression of these genes requires a surprisingly large number of nucleus-encoded proteins (Barkan and Goldschmidt-Clermont 2000, Eberhard et al., 2008). As an extreme example, trans-splicing of the psaA gene of C. reinhardtii requires more than 14 nucleus-encoded factors (Goldschmidt-Clermont et al., 1990). Several factors specifically required for the processing/ stability of chloroplast mRNAs have been identified (Stern et al., 2010). Similarly, nucleus-encoded proteins involved in the translation of specific mRNAs have been characterized (Eberhard et al., 2008). Coordination of the accumulation of the subunits of a photosynthetic complex occurs mainly through two distinct mechanisms. First, subunits made in excess that cannot assemble are rapidly degraded. Second, coordination of synthesis is achieved through the control of epistasy of synthesis (CES) process, mainly studied in C. reinhardtii, in which unassembled subunits directly or indirectly inhibit their own synthesis (Choquet et al., 1998; Choquet and Vallon, 2000; Wostrikoff and Stern, 2007). In the case of PSI the first subunit to be inserted into the thylakoid membrane is PsaB (Nelson and Yocum, 2006). In the absence of PsaB, PsaA cannot assemble and inhibits its own translation (Wostrikoff et al., 2004). Similarly, in the absence of PsaA, the PsaC subunit cannot assemble and inhibits its own translation. In this way synthesis of the subunits is dependent on the ordered assembly of the complex. Because the first subunit to be integrated within the thylakoid membrane is PsaB, it is particularly important to identify and characterize the factors required for the assembly of this anchor protein of PSI that acts as a scaffold for the other subunits. In this study we have identified Tab 1 as a factor required for translation of the psaB mRNA. Based on the previous finding that expression of a chimeric gene containing the psaB 5′-UTR fused to a reporter coding sequence is dependent on Tab 1 (Stampacchia et al., 1997), this factor probably acts mainly at the level of initiation of translation. We previously characterized the Tab 2 protein that is also specifically required for the initiation of translation of PsaB (Dauvillée et al., 2003). One possibility is that Tab 2 and Tab 1 act in a concerted way, although we were unable to obtain any evidence for an interaction between these two proteins.

An unusual property of Tab 1 is its high sensitivity to proteolysis. The full-size protein could only be detected after SDS solubilization of total cell extracts. Several attempts to confirm the chloroplast localization of Tab 1 by cellular fractionation resulted in the complete loss of the protein in all cases.

Tab 1 binds psaB mRNA

One problem associated with TPR and PPR proteins is that when they are expressed as recombinant protein in E. coli they are often insoluble, making the analysis of their biochemical functions difficult. Although RNA binding was reported recently for the HCF107 protein that contains half-a-tetratricopeptide repeat motifs (HAT motifs), which are variants of TPR repeats (Hammani et al. 2012), in most cases chloroplast TPR proteins could not be produced in soluble form. A few PPR proteins have been expressed in soluble form, and their RNA binding properties could be examined. Examples include the PPR protein HCF152 from Arabidopsis involved in chloroplast RNA processing (Meierhoff et al., 2003; Nakamura et al., 2003) and PPR10 from maize, which binds to specific sites on chloroplast mRNAs, defines the position of their 5′ and 3′ termini and protects them against exonucleolytic degradation (Pfalz et al., 2009). We did not succeed in expressing the full size Tab 1 protein in soluble form in E. coli. The presence of several long Ala stretches are expected to make the protein highly insoluble and prone to aggregation. Even the part of the protein containing the PPPEW/OPR repeats was insoluble when expressed as recombinant protein in E. coli. To test its RNA activity we therefore fused this domain to the maltose binding protein and could show that it has RNA binding activity with some specificity towards the psaB 5′-UTR, at least compared with an unrelated RNA. Competition experiments with other chloroplast RNAs, however, revealed that although Tab 1 does not bind to the psbB 5′-UTR, it also binds efficiently to the psbC 5′-UTR. Both the psaB and psbC 5′-UTRs are U-rich and have a similar base composition (76 and 80% AU, respectively). Although the Tab 3 protein appears to bind preferably U-rich RNA based on the competition with oligo U, this cannot explain its specificity towards the psaB mRNA because the psbB 5′-UTR is even more U-rich (92% AU). It is possible that its specific binding to the psaB 5′-UTR requires the cooperation of other proteins such as Tab 2, which is also necessary for translation of the same message (Dauvillée et al., 2003). In the case of Nac2, a nucleus-encoded factor involved in the stability/processing of the psbD mRNA of C. reinhardtii, its association with RBP40 was required for the specific binding of the Nac2-RBP40 complex to the psbD 5′-UTR (Schwarz et al., 2007).

Polysome profiles of psaB mRNA in tab 1-F15

Although the polysome profile of psaB mRNA in the tab 1-F15 mutant displays a shift towards the monosome peak, compared with the wild type, a significant fraction of psaB mRNA remains associated with large complexes that are sensitive to EDTA (Figure 5). Eberhard et al. (2011) also found that in the case of the tda1 mutant deficient in the initiation of translation of the atpA mRNA, this RNA remained associated with large complexes that are insensitive to EDTA. The question arises therefore whether the large complexes seen in Figure 5 for psaB mRNA represent polysomes or ribonucleoprotein complexes that could be involved in mRNA storage. It is also possible that Tab 1 could have a dual function in translation, and be involved in both translation initiation and elongation. A deficient initiation coupled to a slow-down in elongation would have a compensatory effect, and could partly stabilize polysomes. Because the mutation in Tab 1 creates a stop codon at residue 670, this would give rise to a truncated product of 69 000 kDa that may still demonstrate partial activity. Although such a protein was undetectable in tab 1-F15 (Figure 3), we cannot exclude that trace quantities of this protein could give rise to some residual translational activity. Finally, we cannot exclude that Tab 1 is not absolutely required for the initiation of translation of the psaB mRNA, but that in its absence the residual level of translation is too low to support photoautotrophic growth.

Tab 1 belongs to the PPPEW/OPR protein family

A striking feature of Tab 1 is the presence of 10 PPPEW/OPR repeats, which constitute 30% of the protein. The remaining part does not contain any known motif but is characterized by several long stretches of Ala residues. Amongst the nucleus-encoded proteins involved in chloroplast gene expression, many are repeat-containing proteins (Schmitz-Linneweber and Small, 2008). Proteins of this sort are widespread in nature and the atomic structure of several representatives have been determined: alpha-catenin, with 12 ARM repeats of 42 amino acids (Huber et al., 1997); the A subunit of protein phosphatase 2A, with 15 HEAT repeats of 39 amino acids (Groves et al., 1999); Pumilio, with eight Puf repeats of 36 amino acids (Edwards et al., 2001); and protein phosphatase 5, with three TPR (tetratricopeptide repeats) of 38 amino acids (Das et al., 1998). These structural studies have revealed that the repeats exist as tandem helical repeats, and that they form a superhelix with an extended surface that is believed to be involved in protein–protein interactions or in RNA recognition, as shown for Pumilio, a translational repressor of the hunchback mRNA in Drosophila (Edwards et al., 2001). The best-characterized chloroplast repeat-containing proteins are the PPR (penta trico peptide repeats) proteins of land plants. These proteins are part of a large family that includes 450 members (Schmitz-Linneweber and Small, 2008; Small and Peeters, 2000). Although PPR proteins have not yet been crystallized, it is likely that they have a structure similar to the TPR proteins. These PPR proteins are involved in several aspects of chloroplast RNA metabolism, including translation, RNA processing and RNA editing. Some of the PPR proteins and at least one protein with HAT repeats have been shown to bind to specific RNA sequences (Nakamura et al., 2003; Pfalz et al., 2009; Hammani et al. 2011). With all these proteins, the superposition of helical domains could form an interaction surface with RNA, similar to the case of Pumilio. In contrast to plants Chlamydomonas contains only 11 PPR protein and ∼80 TPR protein genes (Eberhard et al., 2011). Instead, the PPPEW/OPR family has greatly expanded in this alga. In this respect the RNA binding activity of Tab 1 revealed in this work is important, as it shows that OPR proteins are directly involved in chloroplast RNA metabolism. Besides translation (Auchincloss et al., 2002; Eberhard et al., 2011), these proteins are also required for RNA processing and splicing (Merendino et al., 2006). It is likely that amongst the other 39 PPPEW/OPR proteins of unknown function identified in this work, several of them participate in specific post-transcriptional steps of chloroplast, and perhaps mitochondrial, gene expression in C. reinhardtii, a possibility that can be tested through reverse genetics. This assigned function of OPR proteins is compatible with the large number of nucleus-encoded factors identified through genetic means, which are involved in the expression of specific chloroplast genes. An important problem is to elucidate the molecular basis of the sequence-specific RNA binding activities of the OPR proteins. The crystal structure of the analogous PUF repeats supports a model in which each repeat of the protein binds a single or several nucleotides of its target sequence (Edwards et al., 2001; Wang et al., 2001, 2002). A new challenge will be to elucidate the code of the OPR proteins that connect the succession and amino acid sequence of the repeats with the nucleotide sequence of the RNA target.

Experimental procedures

Strains and media

Wild-type C. reinhardtii, tab 1-F15 (Stampacchia et al., 1997) and UVM4 (Neupert et al., 2009) mutant cells were grown in Tris-acetate-phosphate (TAP) medium, as described by Harris (1989). For some experiments minimal medium High salt minimum medium (HSM) was used.

Genomic complementation

The mutant strain tab 1-F15 was rescued by transformation with an ordered cosmid library (Purton and Rochaix, 1994; Zhang et al., 1994), and transformants were selected on either HSM or TAP medium under light, under conditions that block the growth of the tab 1-F15 mutant. Cells were treated with autolysin to remove the cell wall, and were then transformed using the glass-bead method (Kindle, 1990).

Oligonucleotides and plasmid constructions

The oligonucleotides used in this work are listed in Table S1.

A Tab 1 cDNA NotI fragment of 4.2 kb was isolated from a C. reinhardtii cDNA library and cloned in Bluescript KS. The cDNA was inserted into pGenD between the NdeI and EcoRI sites downstream of the PsaD promoter and 5′-UTR (Fischer et al., 1996). Because this cDNA was not complete, the authentic coding sequence was reconstituted from the adjacent genomic DNA that encodes the C-terminal 185 amino acids of Tab 1. The HA tag consisting of three copies of the HA peptide YPYDVPDYA was inserted in frame in the unique SacI site of Tab 1, yielding p958. The paromomycin expression cassette (Sizova et al., 1996) was inserted upstream of the PsaD promoter in p958, yielding p1001. The 5.4-kb genomic fragment of Tab 1 was inserted in pSL18 with the HA tag at the SacI site, yielding p925.

To test whether the 10 PPPEW/OPR repeats are sufficient for rescuing the tab 1-F15 mutant, the following plasmids were constructed. The unique NaeI site of the pGenD2 plasmid (Fischer et al., 1996) was changed into a HpaI site, and the paromomycin cassette was inserted into the BamHI site, yielding plasmid p909 that includes the transit peptide of PsaD. Upon digestion of this plasmid with HpaI and EcoRI, a tag (Tag2c) containing His6, HA and StrpII was inserted, yielding plasmid p1034. The 1964-bp XbaI-PvuI fragment from the Tab 1 cDNA containing all of the repeats was then inserted into the HpaI site of p1034, yielding p1036.

To produce recombinant protein for RNA binding studies, the region containing the 10 PPPEW repeats was amplified by PCR using the oligonucleotides TAB 1-5′RVb and TAB 1-3′SalI as primers, and the product was inserted in the pST4 plasmid cut with EcoRV and SalI, downstream of the maltose binding protein gene driven by the T7 promoter (pST4-MBP-psaB). Moreover, a Tab 1 gene fragment corresponding to amino acids 390–743 of Tab 1 was amplified with primers TAB 1-5′seq and TAB 1-3′seq (Table S1), and inserted into the NdeI site of the pet28a vector (Novagen, now EMD Millipore, http://www.emdmillipore.com). This yielded a recombinant protein of 44 kDa that was used for preparing antibodies against Tab 1.

RNA binding experiments

Because the intact Tab 1 protein is insoluble, a shorter version (amino acids 390–747) fused to the maltose binding protein was produced in E. coli with the pST4-MBP-psaB plasmid (Figure S1). The soluble cell extract was loaded on an amylose resin (New England Biolabs, http://www.neb.com), washed with 20 mm Tris-HCl, pH 7.4, 0.2 M NaCl, 10 mm beta-mercaptoethanol, 1 mm EDTA and eluted with washing buffer containing 10 mm maltose, giving rise to four fractions (Figure S1). Fraction 2 was used for the RNA binding experiments. Between 0.8 and 4.8 μg of purified recombinant Tab 1 protein (corresponding to a concentration of 6.6–200 nm) were incubated with 32P-labeled psaB 5′-UTR transcript (3 fmol psaB RNA in vitro, transcribed from pet28-psaB) in 30 μl of 50 mm Tris-HCl, pH 8.0, 100 mm KCl, 6 mm MgCl2, 15% glycerol (v/v), 1 μg yeast tRNA, 16 U RNasin for 20 min on ice (Barneche et al., 2006). Competition studies were performed by pre-incubating the competitor RNA for 5 min before the addition of labeled RNA. In some experiments 10 U of T1 RNase were added to the reaction. Samples were electrophoresed on a 5% polyacrylamide gel (29 : 1) overnight at 50 V using Tris-borate-EDTA buffer. Gels were dried and exposed to −70°C overnight and quantified with a phosphorimager.

Polysome purification

Preparation and size fractionation of polysomes were performed as described by Barkan (1988) and Yohn et al. (1996), with slight modifications. Chloramphenicol (100 μg ml−1) was added to the culture 10 min prior to collecting cells at 2 × 106 ml−1 from a 400-ml culture that was split into two equal parts. Each culture was centrifuged, and the pellet was frozen and ground with a mortar and pestle in liquid nitrogen. One extract was resuspended in polysome extraction buffer [200 mm Tris-HCl, pH 9, 200 mm KCl, 35 mm MgCl2 hexahydrate, 25 mm ethylene glycol-bis(b-aminoethylether)-N,N,N9,N9-tetraacetic acid (EGTA), 0.2 M sucrose, 1% Triton X-100, 2% polyoxyethelene-10-tridecyl-ether], with 0.5 mg ml−1 heparin, 0.7% beta-mercaptoethanol, 100 μg ml−1 chloramphenicol, 0.5% Na-deoxycholate and Protease Inhibitor Cocktail (Sigma-Aldrich, http://www.sigmaaldrich.com). In the other extract MgCl2 was replaced by 20 mm EDTA. Following a 20-min centrifugation at 7700 g at 4°C, the supernatant was layered onto linear 15–60% sucrose density gradients (40 mm Tris–HCl, pH 9, 20 mm KCl, 10 mm MgCl2, 0.5 mg ml−1 heparin, 0.7% beta-mercaptoethanol-mercaptoethanol, and 100 μg ml−1 chloramphenicol. In the EDTA gradients MgCl2 was replaced with 1 mm EDTA. Centrifugation was at 4°C (Beckman SW 40; Beckman Coulter, https://www.beckmancoulter.com) for 3.5 h at 256 000 g.

Bioinformatic analysis

The original repeats of Tab 1 were identified using meme (http://meme.sdsc.edu/meme/intro.html). The same software was used for identifying repeats in the 15 Tab 1-related proteins identified by BLAST. A multiple sequence alignment of the identified repeats was built using mafft in mode G-INS-i (Katoh et al., 2005). After manual curation of the multiple sequence alignments, a generalized profile was built using pftools (pfmake in mode semiglobal and substitution matrix blosum45; Sigrist et al., 2002), and calibrated on a shuffled version of UniProtKB/Swiss-Prot (release 34). The generalized profile was named PPPEW.

The PPPEW profile was used to search the full Chlamydomonas proteome (Merchant, 2007) (http://genome.jgi-psf.org/Chlre4/Chlre4.home.html), with pfsearch (pftools) combined with ftrep, a program using a signal processing technique (autocorrelation) to promote low-scoring matched repeats if in phase with proximal matches (L. Cerutti, A. Bohne, J.D. Rochaix and O. Vallon, unpublished data).

Acknowledgements

We thank Nicolas Roggli for drawings and photography, Silvia Ramundo for technical advice and Michel Goldschmidt-Clermont for his critical reading of the article. This work was supported by grant 31003A_133089/1 from the Swiss National Foundation.

Ancillary