A dominant nuclear mutation in Chlamydomonas identifies a factor controlling chloroplast mRNA stability by acting on the coding region of the atpA transcript

Authors


*For correspondence (fax 33 1 58 41 50 12; e-mail Wollman@ibbc.fr)

Summary

We have characterized a nuclear mutation, mda1-ncc1, that affects mRNA stability for the atpA gene cluster in the chloroplast of Chlamydomonas. Unlike all nuclear mutations altering chloroplast gene expression described to date, mda1-ncc1 is a dominant mutation that still allows accumulation of detectable amounts of atpA mRNAs. At variance with the subset of these mutations that affect mRNA stability through the 5′ UTR of a single chloroplast transcript, the mutated version of MDA1 acts on the coding region of the atpA message. We discuss the action of MDA1 in relation to the unusual pattern of expression of atpA that associates particularly short lived-transcripts with a very high translational efficiency.

Introduction

The expression of chloroplast genes coding for photosynthetic proteins is controlled by nuclear factors, the vast majority of which act post-transcriptionally. This nucleo-chloroplastic interaction has been studied extensively in the unicellular green alga Chlamydomonas reinhardtii, as well as in higher plants (for recent reviews, see Barkan and Goldschmidt-Clermont, 2000; Monde et al., 2000; Zerges, 2000). Chlamydomonas combines several advantages for such studies. It can grow heterotrophically, thereby allowing the isolation of photosynthetic mutants; it is particularly suitable for classical genetic approaches (Harris, 1989) and its chloroplast genome is easily transformable (Boynton et al., 1988).

In the past decade, the importance of the 5′ end of chloroplast mRNAs as targets for this nuclear control has been recognized: the targets are cis-acting elements essential for the maturation, nucleolytic resistance and translation of chloroplast mRNAs. In several instances, it was demonstrated that a 5′ end maturation process is a pre-requisite for the stable accumulation and translation of chloroplast mRNAs, and that specific 5′ UTR-directed nuclear factors play an essential role in this post-transcriptional regulation (Bruick and Mayfield, 1998; Drager et al., 1998; Nickelsen et al., 1994; Vaistij et al., 2000a). The characterization of the nuclear genes involved in these processes is still in its infancy. A few genes have been cloned, among which one is involved in the stability of psbB (Vaistij et al., 2000b) and one in the stability of psbD (Boudreau et al., 2000) The corresponding protein sequences and other data show an involvement not only in mRNA-protein interactions, but also in protein-protein interactions.

The 3′ UTR of chloroplast transcripts also contributes to their stabilization through the presence of stem-loop structures. This has been shown for instance for atpB (Stern et al., 1991) and psaB (Lee et al., 1996). The stem-loop structure blocks further trimming by a 3′-5′ exonuclease that acts after an endonucleolytic cut has primed the 3′ end maturation process (Stern and Kindle, 1993). A similar mechanism operates in spinach chloroplasts, where processing of the 3′ end of the petD mRNA has been studied in great detail (Hayes et al., 1996; Stern and Gruissem, 1987; Stern et al., 1989). However, with the exception of the nuclear mutation crp3 that affects 3′ end formation for several mRNAs (Levy et al., 1999), no genetic evidence for the existence of specific nuclear factors acting on a 3′ UTR has been reported. Biochemical studies have identified proteins that bind to the 3′ end in the case of the unprocessed petD transcript in spinach (Chen et al., 1995; Hayes et al., 1996; Yang et al., 1996), of psbA in barley (Memon et al., 1996) and of several unrelated mRNAs in mustard (Liere and Link, 1997; Nickelsen and Link, 1991, 1993). Several of these proteins also show endoribonuclease activity (Hayes et al., 1996; Nickelsen and Link, 1993; Yang et al., 1996), and thus may form part of a multiprotein RNA processing complex.

While the 5′ and 3′ untranslated regions have been frequently documented as controlling RNA stability, there have been no reports of cis-acting elements located in the coding regions of chloroplast genes that play such a role. On the other hand, such elements do exist in prokaryotes (Kulkarni and Golden, 1997), animal and fungal cells. For example, β-tubulin mRNA stability is controlled by co-translational recognition of an amino-terminal peptide (Bachurski et al., 1994), and a surveillance complex recognizes and degrades mRNAs which contain nonsense codons (reviewed in Czaplinski et al., 1999). However, these studies have not defined RNA-based stability or instability determinants within coding regions, but have rather shown that changes in the RNA lifetime are a secondary effect of other cellular regulatory phenomena.

In a previous study, we described three different nuclear mutants of C. reinhardtii affected in the expression of the atpA and atpB genes (Drapier et al., 1992). The two genes code for the α and β subunits of the chloroplast ATP synthase. One of these mutants, then named ncc1, showed unaltered rates of transcription of the atpA gene but displayed a decrease in the accumulation of the monocistronic atpA transcript; atpA is the first gene in the cluster atpA-psbI-cemA-atpH which is transcribed into four atpA-containing mRNA species (Drapier et al., 1998). In ncc1, atpA mRNA was still detectable by RNA filter hybridization analysis and its translation remained efficient enough for the ATP synthase complex to accumulate in the thylakoid membranes. These characteristics contrast with other Chlamydomonas nuclear mutants in which the affected chloroplast mRNA was undetectable, whether the target was atpB (Drapier et al., 1992), psbD (Kuchka et al., 1989), petD (Drager et al., 1998), psbB (Jensen et al., 1986; Monod et al., 1992) or psbC (Sieburth et al., 1991).

That the ncc1 mutation might have distinct characteristics from those identified previously could be predicted from the fact that it arose as a spontaneous mutation that did not alter photosynthesis. Therefore it was not identified through regular screening procedures that allow detection of mutants showing functional alterations (reviewed in Simpson and Stern, 2001). Indeed, S1 nuclease protection experiments strongly suggested that the destabilization of the atpA mRNA would not occur through its 5′ end (Drapier et al., 1992), in contrast to the mode of action of most other nuclear-encoded stabilizing factors (Nickelsen, 1998). Here we further characterize mutant mda1-ncc1 (m for maturation of the mRNA, d for ATP synthase complex, a for atpA) and show that it harbours a dominant mutation that destabilizes the atpA transcript through a target located in its coding region.

Results

The accumulation of the four atpA-containing transcripts is reduced in mda1-ncc1

The atpA gene is the first gene of a transcription unit that comprises four cistrons, atpA-psbI-cemA-atpH, in the chloroplast genome of Chlamydomonas (Figure 1a). Among the eight transcripts generated from this gene cluster, four contain the atpA sequence (Drapier et al., 1998). They are shown in Figure 1b by RNA filter hybridization with an intragenic atpA DNA fragment. It should be noted that the low levels of the tri- and tetracistronic atpA transcripts, each representing < 1% of the total atpA transcripts, precluded their detection in some experiments. The major phenotypic feature of the mda1-ncc1 nuclear mutant is the reduced accumulation of the monocistronic atpA transcript (Drapier et al., 1992), which can be readily seen in Figure 1b. In order to characterize more precisely the respective levels of atpA mRNAs in mda1-ncc1, we performed a dilution series of total RNA from a wild-type strain. This experiment provided evidence for a decreased accumulation of all forms of atpA transcripts in mda1-ncc1, albeit to very different extents (Figure 1c): it was by far more drastic for the monocistronic transcript, a 10- to 20-fold decrease, than for the polycistronic transcripts that displayed only a 4-fold reduction. The amounts of the other transcripts from the atpA-psbI-cemA-atpH transcription unit were not modified (data not shown). Since the atpA transcription rate is not altered in mda1-ncc1 (Drapier et al., 1992), we conclude that overall, the effects seen here are due to reduced RNA stability. However, altered processing efficiency of the longer transcripts into shorter ones may also play a role.

Figure 1.

Accumulation of atpA transcripts in wild-type and mda1-ncc1 mutant cells.

(a) Map of the atpA transcription unit. The three promoters of the unit are indicated by bent arrows, with dashed lines used for promoters that are not implicated in the transcription of atpA. The coding regions of the genes are represented as grey boxes.

(b) RNA filter hybridization analysis of wild-type and mda1-ncc1. Filters were hybridized first with an intragenic atpA probe then with a psbA probe as a loading control.

(c) Dilution series of total RNA from wild-type allowing a quantitative determination of the accumulation of each atpA transcript species for the mda1-ncc1 mutant. After hybridization with an intragenic atpA probe, the filter was re-hybridized with 25S rRNA as a loading control.

Mda1-ncc1 is a dominant mutation

Vegetative diploids (3 species: MDA1/MDA1, MDA1/mda1 and mda1/mda1) were prepared according to the usual complementation procedure between arg2 and arg7 mutations (for details, see Experimental procedures). Total RNA was extracted from the three types of cells and the atpA transcript profiles were determined by RNA filter hybridization with an intragenic atpA DNA fragment. The accumulation of atpA transcripts in the heterozygous MDA1/mda1 diploid, where both the wild-type MDA1 and mutant mda1 loci were present, was reduced to the same extent as in the homozygous mda1/mda1 diploid (Figure 2). The presence of a wild-type copy of the MDA1 gene, i.e. the presence of the wild-type nuclear gene product MDA1 in the heterozygous diploid cells, did not allow restoration, even partially, of the wild-type amount of atpA transcripts, except possibly for the low abundance tetracistronic transcript. After quantification we estimated a 10- to 20-fold decrease of the monocistronic transcript, similar to what is found in haploid mda1 mutant cells. Thus we conclude that mda1-ncc1 harbours a dominant mutation that affects the half-life of atpA mRNA. Consequently, the mda1-ncc1 allele is unlikely to be a null (loss-of-function) allele, and instead probably expresses a mutated version of the MDA1 protein product that acts on the atpA mRNAs (see Discussion).

Figure 2.

Accumulation of atpA transcripts in heterozygous vegetative diploids MDA1/mda1, relative to that in homozygous vegetative diploids MDA1/MDA1 or mda1/mda1. Experimental conditions are as described for Figure 1b.

The 5′ UTR of the atpA gene is not the target for the mutated nuclear factor MDA1

As a number of nuclear mutants of C. reinhardtii display a specific destabilization of chloroplast transcripts through their 5′ UTRs, we wondered if this was also the case with mda1-ncc1. We first used a strain whose chloroplast carried, besides the original atpA gene, a chimeric petA gene where the 5′ UTR and the first 75 bp of the coding sequence are derived from atpA (Figure 3a). The resulting strain, which was named transformant AFFF (Choquet et al., 1998), expresses the petA product, cytochrome f, from a chimeric mRNA of larger size than the original petA transcript (compare lanes AFFF and mda1 in Figure 3b).

Figure 3.

Accumulation of atpA transcripts in transformant AFFF or FAAA, in a wild-type or mda1-ncc1 nuclear context.

(a) Map of two constructs aimed at investigating the participation of the atpA 5′ UTR in the mda1-ncc1 phenotype. Transformant AFFF: the 5′ UTR of petA was replaced by that of the atpA gene (grey box with a bent arrow on the promoter site); note that 25 codons of the atpA coding sequence remain in the construct (black box). Transformant FAAA: the 5′ UTR of atpA was replaced by that of petA. R26 and R15 are the EcoRI restriction fragments, named after Rochaix (1978), where are located the chimeric genes.

(b) RNA filter hybridization analysis of a representative tetrad from an AFFF × mda1-ncc1 cross. Combined probes for atpA and petA intragenic fragments. Polycistronic atpA transcripts are poorly visible because of the numerous minor petA transcripts, the most abundant migrating between the mono- and dicistronic atpA transcripts (marked by an asterisk). A psbA probe was used as a loading control. The inferred genotype at the mda1 locus is given for the tetrad progeny (wt = MDA1; m = mda1 mutant).

(c) RNA filter hybridization analysis for a representative tetrad from an FAAA × mda1-ncc1 cross. Same experimental conditions as in Figure 1b and same tetrad labelling as in panel b.

In Chlamydomonas crosses, the chloroplast genome is uniparentally transmitted by the mt+ parent to tetrad progeny, whereas a nuclear mutation like mda1-ncc1 exhibits Mendelian segregation and is transmitted only to half of the progeny. Therefore, we crossed transformant AFFF, mt+ with mda1-ncc1, mt, in order to determine whether the tetrad products showed a 2 : 2 segregation in the amount of the AFFF mRNA, as was expected for the resident atpA transcripts. A filter blot containing RNA from a representative tetrad was hybridized with a mixture of an intragenic petA probe and an intragenic atpA probe to detect petA, AFFF and atpA mRNAs (Figure 3b). Members 3 and 4 of the tetrad bore the mda1-ncc1 nuclear mutation, since they showed a substantial decrease of monocistronic atpA mRNA. However, their content in chimeric AFFF mRNA was similar to that of the wild-type members 1 and 2 of the tetrad. We conclude that the chimeric AFFF mRNA does not contain the cis-elements involved in the mda1-ncc1-mediated destabilization of atpA transcripts.

Although the 5′ UTR of atpA was not sufficient to confer sensitivity of a chimeric transcript to the mda1-ncc1 mutation, it could still harbour some of the motifs involved in the destabilization process. If such was the case, a chimeric atpA transcript with a 5′ UTR distinct from the original atpA 5′ UTR should no longer be sensitive to the mda1-ncc1 mutation. To assess this hypothesis, we used transformant FAAA, in which the resident atpA gene was replaced by a chimeric version of the gene corresponding to a fusion of the 5′ UTR of petA with the coding region and 3′ UTR of atpA (see Figure 3a). After crossing FAAA, mt+ with mda1-ncc1, mt, we examined the abundance of atpA transcripts in the resulting tetrads. As shown in Figure 3c, we observed a 2 : 2 segregation in abundance of the FAAA transcript (in this representative tetrad, members 2 and 4 are mutant and members 1 and 3 are wild-type). Since the chimeric transcript was still sensitive to the mda1-ncc1 mutation, we concluded that the 5′ UTR of atpA was not required for the destabilization of atpA mRNA in a mda1-ncc1 context.

The 3′ UTR of the atpA gene is not the target of the mutated nuclear factor MDA1

In an earlier study we had suggested that atpA transcripts in the mda1-ncc1 mutant are principally degraded from their 3′ ends, based on comparative S1 nuclease protection experiments (Drapier et al., 1992). To test this hypothesis, we crossed mda1-ncc1, mt with the transformant Δ3, mt+Δ3 carries a deletion of most of the atpA 3′ UTR, from 62 bp to 375 bp downstream of the translation stop codon (see Figure 4a). If the target of the mutated nuclear factor MDA1 was in this deleted region, the four members of the tetrad should have the same atpA transcript profiles as the Δ3 parent, by virtue of their insensitivity to the presence or absence of the mda1-ncc1 mutation. Results for a representative tetrad are shown in Figure 4b. As previously shown (Drapier et al., 1998), only polycistronic atpA transcripts accumulate in transformant Δ3, since the processing site for monocistronic atpA mRNA is located within the deleted region. We observed a 2 : 2 segregation in the abundance of atpA transcripts, with the mutant members 1 and 4 showing a markedly reduced accumulation of atpA mRNA as compared to the wild-type members 2 and 3. Thus, the target for the mutated MDA1 factor is not located in the 3′ UTR of the atpA monocistronic transcript. We note that the dicistronic atpA transcript in Δ3 (i.e. the smallest atpA-containing mRNA in this strain) showed a much larger accumulation than the regular dicistronic transcript expressed in the presence of monocistronic atpA (strain mda1), but also exhibited a much larger relative decrease in the mda1-ncc1 context.

Figure 4.

Accumulation of atpA transcripts in transformants lacking the atpA-3′ UTR in wild type and mda1-ncc1 nuclear contexts.

(a) Position of Δ1 and Δ3 deletions are indicated on a map of the atpA gene cluster (same representation of the transcription unit than for Figure 1a); Δ3 eliminates the atpA-psbI intergenic region; Δ1 eliminates the intergenic atpA-psbI region as well as the whole psbI gene and part of the psbI-cemA intergenic region.

(b) RNA filter hybridization analysis for a representative tetrad from a Δ3 × mda1-ncc1 cross. Similar experimental conditions as in Figure 1b; tetrad labelling is the same as for Figure 3b.

(c) Transformant Δ1 was crossed with mda1-ncc1; RNA filter hybridization analysis for a representative tetrad is shown. Experimental conditions were as those described for Figure 1b; tetrad labelling is described in the legend to Figure 3b.

We then performed a similar analysis with transformant Δ1, mt+ which harbours a longer 3′ end deletion, beginning 7 bp downstream of the translation stop codon and extending 766 bp to a HindIII site in the psbI-cemA intergenic region (Figure 4a). In this strain, the 3′ end processing sites of both atpA and psbI are deleted and thus, the only atpA transcripts present are the tri- and tetracistronic forms (see lane Δ1(+) in Figure 4c. Although their accumulation is much greater in Δ1 than in wild-type cells, the total content of atpA mRNA, taking into account all atpA-containing transcripts, showed a 10-fold reduction. In Δ1, mt+ × mda1-ncc1, mt crosses, we observed a 2 : 2 segregation in the abundance of the tri- and tetracistronic transcripts (Figure 4c). They were hardly detectable in members 1 and 3 of the tetrad shown, but accumulated as in the Δ1 parent as well as in tetrad members 2 and 4. After quantification, we deduced that the sum of the tri- and tetracistronic transcripts was 15-fold lower in Δ1-mda1 progeny. This reduction was comparable to that of the monocistronic atpA transcript in mda1-ncc1 cells carrying a wild-type chloroplast (Figure 1c). Taken together, these deletion studies led us to the conclusion that the 3′ UTR of atpA does not contain the cis-acting targets for the mutated factor MDA1.

The target of the mutated nuclear factor MDA1 is located in the last 1360 bp of the atpA coding sequence

The experiments presented above strongly suggested the surprising conclusion that the atpA coding region is the target for the destabilizing action of the mutated MDA1 factor. To test this directly, we assembled another chimeric atpA gene in which its 5′ and 3′ UTRs were both replaced by their counterparts from the petA gene (Figure 5a); this yielded transformant FAFA. As shown in Figure 5b, only three atpA-containing transcripts are present in the transformant, since the 3′ end processing site for atpA had been deleted and replaced by the single site derived from petA. We crossed transformant FAFA, mt+ with the mda1-ncc1, mt and analysed the resulting tetrads for the presence of atpA-containing transcripts, as shown in Figure 5b. A 2 : 2 segregation was observed in their accumulation, with mda1-ncc1 mutant members 3 and 4 having nearly undetectable amounts of the atpA-containing transcripts. This experiment provided direct evidence that the target for mutated factor MDA1 was present within the coding region of atpA.

Figure 5.

Accumulation of atpA transcripts in transformant FAFA or ΔatpA in wild-type and mda1-ncc1 nuclear contexts.

(a) Schematic representation of the reporter gene FAFA in which the 5′ and the 3′ UTRs of atpA were replaced by those of petA (the promoter of petA is indicated by a bent arrow), and map of ΔatpA showing deletion of the 2 kb EcoRI-HindIII subfragment. R7, R8 and R15, EcoRI restriction fragments named after Rochaix (1978) where are located the chimeric genes.

(b) RNA filter hybridization analysis for a representative tetrad from a FAFA × mda1-ncc1 cross. Same experimental conditions as in Figure 1b; tetrad labelling similar to that in Figure 3b.

(c) RNA filter hybridization analysis for a representative tetrad from a.atpA × mda1-ncc1 cross. Same experimental conditions as in Figure 1b. No effect of the MDA1/mda1 locus was observed on the members of the tetrad progeny.

Because 76 bases of the coding region of atpA remained in the AFFF construct, which was insensitive to the presence of the mda1-ncc1 mutation (Figure 3a), we could further restrict the target for the mutated MDA1 factor to sequences downstream of position 77. We took advantage of another cross between strain ΔatpA, mt+ (Drapier et al., 1998) and mda1-ncc1, mt to investigate the contribution of the next 85 bases, since the deletion in ΔatpA begins at an EcoRI site 160 bp downstream of the ATG initiation codon (see Figure 5a). Figure 5(c) shows the abundance of the two atpA 5′ UTR-driven transcripts within a representative tetrad: they were present in comparable amounts in each of the four members. Therefore, critical elements of the target for the mutated factor MDA1 are located in the last 1360 bases of the atpA coding sequence.

The level of the expression of the α -subunit varies between strains expressing modified versions of the atpA gene

Since translation rates in Chlamydomonas chloroplasts cannot be predicted from the extent of chloroplast mRNA accumulation (Eberhard et al., 2002), we measured directly the rate of synthesis of the α-subunit in strains that showed an altered pattern of accumulation of atpA transcripts. To this end, we pulse-labelled chloroplast translates from whole cells with 14C acetate for 5 min in the presence of cycloheximide, which inhibits cytosolic translation. The patterns of chloroplast protein synthesis were similar in transformants Δ1 and Δ3, and wild-type cells, whereas those of the Δ1-mda1 and the Δ3-mda1 mutant strains were similar to that in the mda1-ncc1 single mutant. This is illustrated in Figure 6a for transformant Δ1 and the members of a tetrad resulting from a Δ1,mt+ × mda1-ncc1,mt cross (same observations for transformant Δ3 and members of a tetrad resulting from a Δ3,mt+ × mda1-ncc1,mt cross, data not shown). The rate of synthesis of the α-subunit decreased but that of the β subunit was stimulated in mutant Δ1-mda1 as it is in mda1-ncc1, thus inverting the ratio of labelling between the two CF1 subunits with respect to that in the strains possessing a wild-type nuclear genome.

Figure 6.

Expression levels of the α-subunit in transformants Δ1, FAAA and FAFA, in a wild-type or mda1-ncc1 nuclear context.

(a) The same tetrad as in Figure 4c (1 × mda1-ncc1) and the parental strains, together with the wild-type used as a control, were pulse-labelled for 5 min with 14C-acetate in the presence of an inhibitor of cytoplasmic translation, then solubilized in SDS and separated by polyacrylamide gel electrophoresis. Indicated on the left side are reference electrophoretic bands for LS (Rubisco large subunit, rbcL gene product), P5 (psbB gene product), P6 (psbC gene product) and cytf (petA gene product).

(b) Same experimental conditions and labelling as in panel a, the tetrad used is the one shown in Figure 3c (from the cross FAAA × mda1-ncc1).

(c)Immunoblot of whole cell protein extracts of daughter cells from FAAA × mda1 and FAFA × mda1 crosses. Asterisk indicates a band resulting from a cross reaction of the anti-β antibody with the β-subunit of the mitochondrial ATP synthase. The oxygen evolving enhancer protein 3 (OEE3) was used as a loading control.

The rate of synthesis of the α-subunit in FAAA was lower than in wild-type cells, and similar to that in mda1-ncc1 (Figure 6b). This is easily explained by the presence of 5′-cis-acting signals derived from the petA gene, upstream the atpA coding sequence. In 5 min pulses with 14C-acetate, these petA cis-acting signals drive the synthesis of cytochrome f, that incorporates far less 14C than the α-subunit in a wild-type strain, which is indicative of lower translation rates. The comparison of the rates of synthesis of the α and β subunits among the tetrad progeny of Figure 6b, is hampered by the lower protein labelling in members 2 and 4 as compared to members 1 and 3 of the tetrad. However, after quantification of the labelling of the α and β subunits relative to that in bands P5 and P6 (compare these bands along the tracks of Figure 6b), we observed that all members of a tetrad progeny from the FAAA × mda1-ncc1 cross showed (i) similar rates in α-subunit synthesis — close to that in the two parental strains but lower than that in the wild-type strain. They also displayed the same enhanced synthesis of the β-subunit as compared to that in the wild-type strain. We observed similar changes in the pattern of synthesis for the α and β subunits in FAFA × mda1-ncc1 crosses (experiment not shown). The mda1-ncc1 nuclear background in the daughter cells from crosses with FAAA and FAFA became apparent only when probing the steady-state protein levels with antibodies against the α and β subunits (Figure 6c): the α-subunit remained below detection in the mda1-ncc1 daughter cells, whereas the β-subunit was present in amounts comparable to that in mutant tda1-F54, which is impaired in the synthesis of the α-subunit (Drapier et al., 1992). Overall, these results establish that in the FAAA-mda1 and FAFA-mda1 background, the combined effects of a drop in atpA mRNA accumulation and a change is cis-acting signals for translation prevents functional assembly of the chloroplast ATP synthase. Accordingly these double mutant strains did not grow phototrophically (data not shown), allowing the mda1-ncc1 to be used as a dominant mutation preventing photosynthesis.

Mda1-ncc1 exhibits a dominance in the synthesis of the ATP synthase α and β subunits

To establish the dominance of mda1 at the translational level, we measured the rates of synthesis of the α and β subunits in the three types of vegetative diploids that we had isolated (Figure 2). The patterns of synthesis were similar in the homozygous diploids MDA1/MDA1 or mda1/mda1 (Figure 7) and in the corresponding haploids, showing a large decrease of the α/β ratio in mda1 context (Figure 6a,b). On the other hand, the labelling pattern of α and β subunits was similar in the homozygous mda1/mda1 diploid cells and the heterozygous MDA1/mda1 diploid cells, confirming the dominance of the mda1 mutation at the translational level.

Figure 7.

Rate of synthesis of the α-subunit in homozygous (MDA1/MDA1 or mda1/mda1) and heterozygous (MDA1/mda1) diploids. Same experimental conditions as in Figure 6a.

Discussion

The characterization of numerous nuclear mutants of Chlamydomonas impaired in the synthesis of a single chloroplast-encoded protein has defined distinct phenotypic classes, each one corresponding to a purported step in the regulation of chloroplast gene expression (Barkan and Goldschmidt-Clermont, 2000; Monde et al., 2000; Zerges, 2000). Nuclear factors acting either on RNA maturation or translation were thus identified genetically. The maturation mutants have been studied in great detail, particularly in the cases of nac2, mbb1 and mcd1, nuclear mutations that destabilize the psbD, psbB and petD mRNAs, respectively. In each case, the studies provided evidence that 5′ end processing is required for mRNA stabilization and translation, and that the mutated gene encoded a protein involved in these steps via an interaction with the cognisant 5′ UTR (Drager et al., 1998; Nickelsen et al., 1994; Vaistij et al., 2000a).

The nuclear mutant mda1-ncc1 here described does not belong to this well-described class, since the 5′ UTR of atpA plays no role in the mda1-ncc1-mediated destabilization process. atpA is the upstream cistron of a transcriptional unit comprising four genes, and the mda1-ncc1 mutation has a much stronger destabilizing effect on monocistronic atpA than on the polycistronic forms of the transcript. Since the four transcripts had identical 5′ ends, mutant mda1-ncc1 was a reasonable candidate for being altered in a nuclear factor involved in the 3′ end processing of monocistronic atpA, as suggested previously (Drapier et al., 1992). Initially we wondered whether mda1 might resemble crp3, another nuclear mutation that affects the processing of the 3′ ends of several transcripts, including atpA, psbI, cemA, and atpB (Levy et al., 1997). We found, however, that mda1 was not allelic to crp3 (which is recessive), and furthermore that mda1 is atpA-specific (Levy et al., 1999). Moreover, as shown here atpA-containing transcripts are still sensitive to the mda1-ncc1 mutation in transformants bearing extensive deletions at the 3′ end of the atpA coding region. Finally, the analysis of tetrads resulting from a cross mda1-ncc1 with a transformant that retains only the central part of the atpA gene provided direct experimental evidence that the cis-acting elements conferring sensitivity of the atpA transcript to the mda1-ncc1 mutation were located in its coding region. The mda1-ncc1 nuclear mutation has two unprecedented characteristics in the field of the chloroplast gene expression: (i) it disclosed a target in the coding region of a chloroplast gene that controls the half-life of the transcript. Coding region elements that contribute positively or negatively to the half-lives of transcripts have been described earlier in prokaryotes (Kulkarni and Golden, 1997) as well as in mammalian (Ross, 1995) and fungal (Hennigan and Jacobson, 1996) cells (ii) it is a dominant mutation whereas all other nuclear mutations altering chloroplast mRNA stability identified to date have been recessive. There are two major ways to explain the dominance of a mutation: the 'dominant negative hypothesis' or the 'gain of function hypothesis'.

The dominant negative hypothesis. Several copies of MDA1 would be present in a multimeric protein complex that confers nucleolytic resistance to the atpA mRNA This protection would occur through binding to some endonucleolytic cleavage site in the coding region as is the case of c-fos and c-myc, in mammalian system. The presence of even one mutated copy of MDA1 would be sufficient to prevent the protective binding of the protein complex to atpA mRNAs.

The gain of function hypothesis. MDA1 possesses a regulated nuclease activity directed against the atpA transcripts. The mutated version of MDA1 would have modified regulatory properties, for instance because of the loss of binding of a negative regulator and would incorrectly target the atpA transcripts. Alternatively, it could have acquired a new recognition site for the coding region of atpA. In this way it might be analogous to the endoribonuclease which cleaves c-myc mRNA in its wild-type form (Lee et al., 1998).

An extreme version of the 'gain of function' hypothesis is the recruitment by the mda1-ncc1 mutation of a nuclease active on atpA mRNA. In this scenario, the wild-type version of MDA1 would not recognize atpA transcripts. We think this is unlikely for two reasons. First, the de-repression of a nucleolytic activity would most likely target several chloroplast transcripts, whereas the mda1-ncc1 mutation is specific for atpA mRNA degradation, to the extent that protein pulse labelling and a limited number of filter blots allow that conclusion. Second, we have shown that atpA transcripts in a wild-type nuclear context are much more short-lived than all other chloroplast transcripts we have tested, an observation that suggests specific or enhanced sensitivity to a nucleolytic process (Eberhard et al., 2002). It is reasonable to propose that this process has merely been enhanced by the mda1-ncc1 mutation.

Neither of the two hypotheses that may account for dominance of mda1-ncc1 can explain by themselves the discrimination among atpA-containing mRNAs by the mutated MDA1 factor: monocistronic atpA transcripts were far more affected in mda1-ncc1 than any of the polycistronic atpA-containing transcripts. However, when the monocistronic form was missing because of a 3′-end modification, the dicistronic transcript became the major mda1-ncc1 target. In turn the tri- and tetracistronic transcripts became major targets when neither the mono- nor the dicistronic transcripts could be formed. Taken together, these results suggest that the most abundant atpA transcript is the major target of the mutated MDA1 factor. In parallel with their increased susceptibility to degradation in an mda1-ncc1 context, we observed that the polycistronic transcripts were increasingly engaged in translation: indeed, the tri- and tetracistronic transcripts that remained in strain Δ3 and represented as little as 10% of the total atpA transcripts in wild-type cells, were sufficient to drive the synthesis of the α-subunit at a wild-type rate. Since the target for mutated MDA1 is in the coding region of atpA, it is tempting to speculate that it destabilizes atpA-containing transcripts only during, or soon after, translation. In fact, the c-myc ribonuclease mentioned above, along with other known ribonuclease activities, are polysome-associated. The potential of association between mda1-ncc1 action and translation could be assessed in two sets of experiments. In the first, abundance of atpA transcripts would be probed in the double mutant mda1-tda1, resulting from a cross between mda1-ncc1 and the nuclear mutant tda1-F54, where translation of atpA mRNA is blocked (Drapier et al., 1992); and in the second, effects on atpA transcripts would be measured after treating mda1-ncc1 cells for up to 24 h with chloroplast translation inhibitors. In the latter case, we tested chloramphenicol, lincomycin and kasugamycin, and found no protective effect of the atpA mRNA in a mda1-ncc1 context (data not shown). In the former case, we still observed the destabilization of monocistronic atpA mRNA in the translation defective tda1-F54 mutant. However, these transcripts are still found in polysomal fractions in tda1-F54 (S. Eberhard, personal communication), which suggests that the mutation plays a role at the elongation step rather than at initiation. We would conclude from both sets of experiments that completion of atpA translation is not a prerequisite for the action of the mutated MDA1 factor. However, one should consider the possibility that the mda1-ncc1 mutation mediates degradation of polysome-bound atpA mRNA, perhaps after translation initiation has taken place and in response to the inevitable changes in RNA secondary structure that accompany ribosome passage.

Initially, cloning the MDA1 gene appeared as a most difficult task, since its mutated version in mda1-ncc1 had neither a selectable nor an easily screenable phenotype. However, we have been able to produce non-phototrophic strains harbouring the mda1-ncc1 mutation, where the petA promoter/5′ UTR has replaced that of atpA. Although we have at present no clues as to the molecular mechanism that prevents the subunit expressed in this genetic background from accumulating in a functional ATP synthase complex, these strains open the way to a cloning strategy for the MDA1 gene.

Experimental procedures

Strains and growth conditions

The wild-type strain of C. reinhardtii used here was derived from strain 137c; the ncc1 nuclear mutant was obtained from a spontaneous mutation that occurred in the wild-type strain (Drapier et al., 1992); the Δ3 and Δ1 strains bearing deletions in the 3′ UTR of the atpA gene, the ΔatpA strain bearing a deletion of the major part of the coding sequence of atpA and the entire psbI gene were constructed as described in Drapier et al. (1998); the construction of the strain AFFF bearing the petA gene under the control of the 5′ UTR of atpA was described in Choquet et al. (1998). All strains were grown on Tris-acetate-phosphate (TAP) medium (pH 7.2) at 25°C with a light intensity of 5.9 µmol photons m−2 s−1.

Plasmid constructions and chloroplast transformation

The pFAAA plasmid was constructed as followed: Plasmid p5FΔR, lacking any EcoRI restriction site, was created from plasmid p5F (Choquet et al., 1998): p5F was linearized by the unique cutter EcoRI, filled-in with Klenow enzyme and re-ligated on itself. A 644-bp fragment containing the 5′ part of the atpA coding sequence was amplified by PCR, using plasmid pATPA2 as a template and oligonucleotides ATPA5 (GCGGATCCATGGCAATGCGTACTCCA GAA BamHI, NcoI) and ATPA3 (GCTCTAGATACTTGAGCTACA GATGAAGCT XbaI) as primers. In the resulting fragment, a NcoI site has been created at the level of the initiation codon. This fragment was digested with XbaI and NcoI and cloned in the plasmid p5FΔR digested with the same enzymes, to yield plasmid p5F-atpA. In this new plasmid the 628 first nucleotides of the atpA coding sequence are fused in frame to the petA 5′ UTR. Plasmid p5FatpA was then digested by EcoRV (located upstream of the petA 5′ UTR) and EcoRI (located in the atpA coding sequence, 165 bp downstream of the initiation codon), and the 863 bp band was eluted. It was then cloned into the pATPA2 vector, first digested with AatII and filled-in with Klenow enzyme, then with EcoRI to result in plasmid pFAAA. Plasmid piWF comprising the complete petA sequence (Kuras and Wollman, 1994) was used as a template to amplify by PCR the 3′ UTR of the pet gene, using oligonucleotides 3FPac(5′-GCGTTAATTAATATTTAATTTTTTGTA GGGCTGCT) and 3FHpa(3′-CGCGTTAACTTCGCAACTGCCACTG ACG); the amplified fragment (491 bp), after digestion with PacI and HpaI, the two restriction sites underlined in the sequences of the oligonucleotides, was inserted into the vector pFAAA that had been previously digested with the same enzymes and was consequently missing the 366 bp within the atpA-3′ UTR. This created plasmid pFAFA, with the 5′ and 3′ UTRs of petA replacing those of atpA on either side of the atpA coding sequence ΔatpA cells were bombarded with tungsten particles coated with plasmids pFAAA or pFAFA as described in Kuras and Wollman (1994) and the transformants were selected by photosynthetic growth on minimal medium under bright light. The homoplasmy of the cells was checked by DNA filter hybridization.

Genetic analysis

For gametogenesis, cells were grown for 3–4 days on TAP plates containing one-tenth the usual amount of nitrogen. Mating, germination, and tetrad analysis were performed according to Harris et al. (1989). Diploid progeny from crosses were selected by incorporating the complementary auxotrophic mutations arg2 and arg7 into the parental strains (mda1-ncc1, arg2 and mda1-ncc1, arg7) and plating on TAP medium. After 12 days in low light, large colonies appeared on each plate that were comprised of larger cells that wild type haploid cells and displayed a mating type minus phenotype (12–20 colonies tested from each cross), as expected for diploid cells (Ebersold, 1967; Harris et al., 1989).

RNA analysis

Isolation of total RNA and RNA filter hybridization analysis were performed as described in Drapier et al. (1998). The probes were as follows: the intragenic atpA EcoRI-PstI fragment (947 bp) or the 5′ UTR of atpA to detect the atpA transcripts, the intragenic petA HindIII-AccI fragment (620 kb) to detect petA transcripts, the chloroplast R14 fragment according to the nomenclature of Rochaix (1978) to detect the psbA transcript or the 25S nuclear DNA fragment as loading controls for the RNA gels.

Protein analysis

Chloroplast proteins were pulse labelled in the presence of 14C-acetate and a cytosolic translation inhibitor (cycloheximide) and analysed by denaturing gel electrophoresis according to Drapier et al. (1992). Immunoblotting experiments were performed as in de Vitry et al. (1989).

Acknowledgements

D.B.S. was supported by a Guggenheim Fellowship and the Prix Georges Morel, and NSF awards MCB-0091020 and INT-9603351. F.A.W and D. D. were supported by the CNRS, UPR 1261. We gratefully acknowledge Yves Choquet for providing the FAAA strain and stimulating discussions.

Ancillary