The 5′ untranslated region of the chloroplast psbA mRNA, encoding the D1 protein, is processed in Chlamydomonas reinhardtii. Processing occurs just upstream of a consensus Shine-Dalgarno sequence and results in the removal of 54 nucleotides from the 5′ terminus, including a stem-loop element identified previously as an important structure for D1 expression. Examination of this processing event in C. reinhardtii strains containing mutations within the chloroplast or nuclear genomes that block psbA translation reveals a correlation between processing and ribosome association. Mutations within the 5′ untranslated region of the psbA mRNA that disrupt the Shine-Dalgarno sequence, acting as a ribosome binding site, preclude translation and prevent mRNA processing. Similarly, nuclear mutations that specifically affect synthesis of the D1 protein specifically affect processing of the psbA mRNA. In vitro, loss of the stem-loop element does not prohibit the binding of a message-specific protein complex required for translational activation of psbA upon illumination. These results are consistent with a hierarchical maturation pathway for chloroplast messages, mediated by nuclear-encoded factors, that integrates mRNA processing, message stability, ribosome association, and translation.
Photosystem II is comprised of proteins encoded within both the nuclear and chloroplast genomes. The expression of these proteins is regulated in response to light at the level of transcription, RNA processing, message stability, translation, and protein turnover (reviewed in Gillham et al., 1994; Mayfield et al., 1995; Rochaix, 1996). For many chloroplast-encoded proteins, translation appears to be the rate-limiting step of gene expression. Chloroplast mRNA levels remain relatively unchanged throughout dark/light transitions whereas protein synthesis rates increase dramatically upon illumination (Fromm et al., 1985; Deng and Gruissem, 1988; Malnoë et al., 1988). Genetic and biochemical evidence suggests that nuclear-encoded factors mediate this light-regulated translation of chloroplast-encoded proteins (Jensen et al., 1986; Kuchka et al., 1988; Rochaix et al., 1989; Girard-Bascou et al., 1992; Wu and Kuchka, 1995; Yohn et al., 1996; Zerges et al., 1997; Yohn et al., 1998b). These nuclear factors typically interact with specific elements within the 5′ untranslated regions (UTRs)1 of the chloroplast mRNAs (Rochaix et al., 1989; Sakamoto et al., 1994a; Zerges and Rochaix, 1994; Stampacchia et al., 1997; Zerges et al., 1997).
The D1 protein, a core component of Photosystem II, is encoded by the chloroplast psbA gene. Throughout the light phase of photosynthesis, the D1 protein is subject to photodamage and is rapidly turned over (reviewed in Barber and Andersson, 1992). D1 inactivation is compensated by a 50–100-fold increase in the rate of D1 protein synthesis in response to light without a corresponding increase in psbA mRNA levels (Fromm et al., 1985; Klein et al., 1988; Malnoë et al., 1988; Krupinska and Apel, 1989). The use of reporter gene constructs in tobacco has demonstrated that the psbA 5′ UTR is sufficient to confer light-dependent translational regulation in vivo (Staub and Maliga, 1994). An in vitro translation system derived from tobacco chloroplasts has identified critical regulatory elements for D1 synthesis in the psbA 5′ UTR including potential ribosome binding sequences (RBS), an AU-box, and to a lesser extent, an upstream stem-loop element (Hirose and Sugiura, 1996). A stem-loop element has also been mapped within the 5′ UTR of the spinach psbA mRNA (Klaff and Gruissem, 1995; Klaff et al., 1997). This element encompasses a putative RBS, an endonucleolytic cleavage site for mRNA decay (Klaff, 1995), and sequences recognized by stromal proteins (Klaff et al., 1997; Alexander et al., 1998).
In the unicellular green algae Chlamydomonas reinhardtii, similar elements within the 5′ UTR have been identified as important for psbA expression. Among these elements is a stem-loop structure immediately upstream of a consensus Shine-Dalgarno (SD) sequence. Mutational analysis of the stem-loop region has shown that this element serves a role in psbA expression, although the exact nature of this role remains unresolved (Mayfield et al., 1994). Deletion of the SD sequence prevents ribosome association with the psbA mRNA and synthesis of the D1 protein, consistent with its proposed function as an RBS (Mayfield et al., 1994).
A complex of proteins, thought to serve as light-dependent translational activators, specifically recognizes the C. reinhardtii psbA 5′ UTR (Danon and Mayfield, 1991). The binding activity of this complex is modulated in response to changes in photosynthetic activity via a redox switch (Danon and Mayfield, 1994; Kim and Mayfield, 1998). The principal RNA-binding (RB) protein in this complex is a chloroplast-localized poly(A)-binding protein (cPABP) homologue (Yohn et al., 1998a). Several nuclear mutants have been isolated in which the loss of the cPABP is accompanied by the absence of D1 protein synthesis due to a block in psbA mRNA association with polyribosomes (Yohn et al., 1996; Yohn et al., 1998b). A previous study suggested that this complex binds the stem-loop element upstream of the SD sequence (Danon and Mayfield, 1991). A model has been proposed (summarized in Mayfield et al., 1995) in which the stem-loop element acts as a translational attenuator in the dark, preventing ribosome association with the SD sequence. Upon illumination, and subsequent increase in photosynthetic activity, binding activity of the protein complex increases and stimulates translation initiation by disrupting the repressive stem-loop element or by providing a platform for ribosome association.
However, examination of psbA transcripts from C. reinhardtii indicated that in vivo the majority of this message lacks sequences upstream of the SD sequence including the stem-loop element (Erickson et al., 1984; Nickelsen et al., 1994; Shapira et al., 1997). In this study, we investigate the differential accumulation of psbA mRNAs containing different 5′ termini. These different 5′ UTRs likely result from the processing of the 90-nucleotide (nt) 5′ UTR to generate a new 5′ terminus 36 nt upstream of the initiation codon. Processing of the psbA 5′ UTR is shown to be closely correlated with ribosome association. In the absence of a competent and accessible SD sequence, D1 protein is not synthesized and the psbA 5′ UTR is not processed. Nuclear mutations that block D1 translation, in conjunction with reduced association of psbA mRNA with ribosomes, also reduce psbA processing. However, removal of the stem-loop element as a consequence of processing does not prohibit the binding of the nuclear-encoded protein complex to this psbA 5′ UTR in vitro, nor does it preclude dynamic light-dependent translational regulation mediated by the psbA RB complex. Based on these observations, we propose a model for psbA mRNA maturation in which 5′ end formation does not serve as a prerequisite for initiation complex formation but rather processing of the 5′ UTR occurs in conjunction with the early stages of ribosome assembly at the RBS.
Materials And Methods
Cell Growth Conditions
Unless otherwise noted, all C. reinhardtii strains were grown at 25°C under constant light in complete media (Tris-acetate-phosphate; Harris, 1989) to a density of ∼106 cells/ml. Cells were harvested by centrifugation at 4°C for 5 min at 4,000 g, washed with H2O, and pelleted again at 4,000 g for 5 min at 4°C. Cell pellets were frozen in liquid N2 and stored at −70°C.
Total and polyribosome-associated RNAs were prepared according to published protocols (Barkan, 1988; Cohen et al., 1998). To prepare membrane-associated RNA, the ground frozen cell pellet was twice extracted with extraction buffer lacking the detergents Triton X-100 and polyoxyethelene-10-tridecyl-ether to remove soluble RNA. The insoluble pellet was then resuspended in extraction buffer containing these detergents and used to prepare polyribosome-associated RNA.
Primer Extension Analysis
Approximately 1 pmol of 5′-radiolabeled DNA oligonucleotide complementary to positions +47 to +32 of the psbA coding region (5′-CGAGCCCATAGGCTAG-3′) or positions +66 to +49 of the psbD coding region (5′-AAGCCAGTCATCAGCGTC-3′) was annealed to 5–25 μg RNA in a 7-μl mixture containing 50 mM Tris (pH 8.3), 60 mM NaCl, and 10 mM DTT by slow cooling from 80°C to room temperature. The reaction mixture was incubated at 42°C for 30 min after addition of 8 μl reaction mixture (50 mM Tris, pH 8.3, 60 mM NaCl, 10 mM DTT, 15 mM MgCl2, 1.25 mM each dNTP, and 0.2 U AMV reverse transcriptase; Life Sciences Inc., St. Petersburg, FL). The RNA was degraded by addition of 20 μl degradation buffer (50 mM Tris, pH 7.5, 0.1% SDS, and 7.5 mM EDTA, pH 8.0) and 3.5 μl 3 M KOH followed by incubation at 90°C for 3 min then 42°C for 25 min. The sample was neutralized upon addition of 6 μl acetic acid and 10 μl 3 M NaOAc (pH 5.3), EtOH precipitated, and resuspended in gel loading buffer. Products were separated by electrophoresis in a 7.5% polyacrylamide/8 M urea gel and quantified by PhosphorImager analysis. Samples were loaded to achieve approximately equal signals and do not reflect the relative amounts of psbA mRNA in each strain of C. reinhardtii.
Analysis of psbA 5′ UTR Length by Northern Analysis after RNase H Cleavage
100 pmol of a DNA oligonucleotide complementary to positions +171 to +156 of the psbA coding region (5′-TGGCGGAGCAGCGATG-3′) was annealed to 5 μg wild-type total RNA or 15 μg hf261 total RNA in a 3.5-μl mixture containing 0.5 μl buffer (200 mM Tris, pH 7.5, 200 mM KCl, 1 mM EDTA, pH 8.0, and 1 mM DTT) and 0.1 U PRIME RNase inhibitor (5 Prime → 3 Prime, Inc., Boulder, CO) by slow cooling from 80°C to room temperature. RNA cleavage was initiated upon addition of 0.5 μl 100 mM MgCl2, 0.5 μl 0.2 U/μl PRIME RNase inhibitor, and 0.5 μl RNase H (2.2 U/μl; GIBCO BRL, Gaithersburg, MD). Reactions were incubated at 37°C for 30 min. Products were separated by electrophoresis in a 5% polyacrylamide/8 M urea gel. RNA blotting and hybridization with a random-primed, [32P]-labeled psbA 5′ cDNA fragment (HindIII-ClaI) were performed as described previously (Mayfield et al., 1994).
Gel Mobility Shift Assay
Approximately 1 μg total protein from C. reinhardtii, partially purified by heparin-agarose chromatography according to Cohen et al. (1998), was incubated for 10 min at room temperature with 0.5 U PRIME RNase inhibitor in a total volume of 8 μl dialysis buffer (20 mM Tris-HCl, pH 7.5, 100 mM KOAc, 0.2 mM EDTA, pH 8.0, 2 mM DTT, 20% glycerol) supplemented with 4 mM MgCl2. To each sample, 0.08 pmol of in vitro transcribed [32P]-labeled RNA (beginning with two G residues added 90 bases [−90] or 36 bases [−36] upstream of the start codon and ending 171 bases downstream of the start codon), 20 μg of wheat germ tRNA (Sigma), and 3 μg of FuD7 (a C. reinhardtii strain lacking the psbA gene) total RNA were added and the reaction was incubated at room temperature for 10 min. Where indicated, 8 pmol unlabeled in vitro transcribed RNA was also added as a competitor. RNA–protein complexes were separated in a 5% nondenaturing polyacrylamide gel.
UV Cross-linking Assays
UV cross-linking assays were performed as gel mobility shift assays with the following alterations. Approximately 5 μg heparin-agarose purified protein was incubated with 1.0 pmol of internally labeled RNA (−90 or −36) generated by in vitro transcription using both [α-32P]ATP and [α-32P]UTP. After complex formation, the reactions were irradiated with short-wave UV light for 1 h at 4°C. The RNA was digested at 55°C for 45 min after addition of urea, EDTA (pH 8), and RNase A to final concentrations of 3 M, 3.75 mM, and 0.5 mg/ml, respectively. Labeled proteins were separated by SDS-PAGE and visualized using a PhosphorImager.
RNA Affinity Chromatography (RAC) and Immunoblot Analysis
RAC was performed as described by Cohen et al. (1998). In brief, 125 μg of in vitro transcribed RNA (−90 or −36) was coupled to 100 μl amino gel 1702 (Sterogene Bioseparations Inc., Carlsbad, CA). 1 ml heparin-agarose purified protein was passed over the resulting columns, washed, and recovered by elution with a high salt buffer. Equal quantities of RAC proteins were mixed with 2× sample buffer (5% SDS, 5% β-mercaptoethanol, 400 mM Tris, pH 6.8, 10% sucrose), heated to 65°C for 5 min, and separated by SDS-PAGE. The gels were either stained with Coomassie blue or electroblotted to nitrocellulose (Schleicher and Schuell, Keene, NH) in 10 mM CAPS (pH 11)/10% methanol (Mayfield et al., 1994). Filters were treated with rabbit polyclonal antisera specific for RB38, RB47, or RB60.
Multiple psbA 5′ UTR Exist In Vivo
S1 mapping of the psbA mRNA from C. reinhardtii previously identified two 5′ termini, one 90 nt upstream of the start codon and a more pronounced terminus 36 nt from the start codon (Erickson et al., 1984). Multiple 5′ termini in this assay may reflect multiple transcription start sites, posttranscriptional cleavage events, or RNA modifications or structural elements which result in poor hybridization of the radiolabeled probe. We used primer extension analysis to examine the 5′ termini of the psbA 5′ UTR. A radiolabeled oligonucleotide complementary to the 5′ coding region of psbA was annealed with total RNA prepared from wild-type C. reinhardtii. Reverse transcriptase was used to extend the probe to the 5′ terminus of the message. Extension products were separated by PAGE and quantified by PhosphorImager analysis. As shown in Fig. 1, 95% of the psbA message contains a 5′ terminus 36 nt upstream of the start codon whereas the remaining psbA mRNAs feature 5′ UTRs of 90 nt in length. Similar results were observed previously by primer extension analysis (Shapira et al., 1997).
The two predominant extension products could reflect different 5′ termini or may indicate partial blocks in primer extension resulting from strong structural elements within the 5′ UTR. To resolve this issue we used RNase H to cleave the psbA mRNA at the site of hybridization to a complementary oligonucleotide to generate 5′ psbA fragments of a size that could be resolved by PAGE and subsequently visualized by Northern analysis. The size of the wild-type psbA 5′ UTR can be compared with the 5′ UTR from the hf261 strain, a nuclear mutant in which only the 90-nt psbA 5′ UTR was observed by primer extension analysis (see Fig. 6). As shown in Fig. 2, the 5′ fragment of the psbA message from hf261 is longer than the predominant wild-type psbA 5′ fragment, confirming that the primer extension products reflect 5′ UTRs of differing size. Therefore, the vast majority of wild-type psbA transcripts from C. reinhardtii begin 36 nt upstream of the start codon, adjacent to the RBS (Fig. 3 A).
Neither primer extension analysis nor Northern analysis can distinguish between 5′ termini generated by transcription initiation and termini generated by posttranscriptional mRNA processing. However, the data will show that the appearance of the 36-nt 5′ UTR is most closely associated with posttranscriptional events, suggesting that this terminus most likely results from mRNA processing. This conclusion is further supported by the observation that partial deletion of a putative promoter element required for transcription initiation of the of the −36 terminus (Erickson et al., 1984) had no affect on the appearance of the 36-nt 5′ UTR (Loop-del [Mayfield et al., 1994]; data not shown).
Processing Is Independent of the Primary mRNA Sequence at the Cleavage Site
To identify RNA elements required for processing, site-specific mutations were engineered within the 5′ UTR of the psbA gene which was then reintroduced into the C. reinhardtii chloroplast genome by particle bombardment (Mayfield et al., 1994). One of these constructs, Alter, replaced the four uridine nucleotides just upstream of the RBS with adenines (Fig. 3,B). In addition to disrupting the stem-loop element, this mutation changed the primary sequence at the processing site. This mutation results in a 95% reduction in D1 protein synthesis and an 80% reduction in D1 protein accumulation (Mayfield et al., 1994). However, primer extension analysis of total RNA isolated from this strain demonstrates that cleavage of the psbA 5′ UTR is identical in extent and location to processing in wild-type C. reinhardtii (Fig. 4), indicating that processing is not a function of the primary RNA sequence at the cleavage site.
Processing of the psbA 5′ UTR Is Dependent upon the Presence of a Competent and Accessible RBS
Many chloroplast-encoded mRNAs have a sequence resembling a prokaryotic-like SD that serves as an RBS. However, because these sequences tend to vary in size and location relative to their Escherichia coli counterparts, the relevance of these RBSs in chloroplasts has been questioned (Fargo et al., 1998). Deletion of a putative SD sequence in the 5′ UTR of the C. reinhardtii psbA mRNA (RBS-del) eliminated psbA mRNA polyribosome association (Yohn et al., 1996) and D1 synthesis in vivo (Mayfield et al., 1994), suggesting that this sequence is a functional RBS. Additional mutations designed to disrupt the RBS have been introduced into the psbA 5′ UTR (Fig. 3,B). These mutations either eliminated the RBS (RBS-del, RBS-Alt), sequestered the RBS in an extended stem-loop structure (RBS-paired), or changed the location of the RBS by replacing the wild-type RBS with another RBS placed upstream of the stem-loop element (RBS-upstream). All of these changes block ribosome association and abolish D1 protein synthesis in vivo, providing further evidence that this SD sequence functions as an authentic RBS (Mayfield et al., 1994; Bruick, R.K., and S.P. Mayfield, unpublished results). Total RNA was prepared from each of these mutants and primer extension analysis was used to characterize the processing of the psbA 5′ UTR. As shown in Fig. 5, each of these mutants possess only one 5′ terminus corresponding to the unprocessed psbA 5′ UTR. These data show that a functional RBS, required for ribosome association and psbA translation, is also required for normal processing of the 5′ UTR.
Though processing of the psbA 5′ UTR requires a SD sequence, and presumably initial ribosome association, processing is not dependent upon D1 protein synthesis. For example, a deletion within the psbA 5′ UTR that positions the RBS 11 nt upstream of the start codon (RBS-11, Fig. 3,B) reduces psbA association with large polyribosomes and eliminates D1 protein synthesis in C. reinhardtii while enhancing translation of the same message in E. coli (Bruick, R.K., and S.P. Mayfield, manuscript in preparation). As demonstrated by a primer extension assay, the psbA 5′ UTR from this mutant is still cleaved to the same extent and in the same position relative to the RBS as in the wild type (Fig. 5). The exact location of the processing site was confirmed by RNA sequencing (data not shown). Thus, we imagine that the RBS in this construct remains accessible for recognition by factors involved in the early assembly of a translation initiation complex before processing but that this mutant RBS lacks the ability to specify the correct initiation codon.
Processing of the psbA 5′ UTR Is Reduced in Nuclear Variants Deficient in D1 Expression
Several nuclear mutations have been characterized in C. reinhardtii that specifically affect expression of the D1 protein (Yohn et al., 1996; Yohn et al., 1998b). In each of these mutants (F35, hf149, hf233, hf261, hf859, and hf1085), D1 protein synthesis is specifically lacking. While the primary defects of these mutations are not known, psbA association with polyribosomes is reduced or absent in the majority of these strains, suggesting that these mutations affect translation initiation (Yohn et al., 1996; Yohn et al., 1998b). Total RNA was prepared from each of these mutants and primer extension analysis of the psbA 5′ UTR was performed (Fig. 6). Each of these nuclear mutations has a pronounced effect on the amount, but not the position, of processing of the psbA 5′ UTR. In each case, the ratio of transcripts featuring a 90-nt 5′ UTR (−90) to processed transcripts (−36) is increased compared with the wild type, with some mutants lacking any detectable psbA terminus 36 nt upstream of the initiation (hf261, hf1085).
Just as these nuclear mutants specifically affect translation of the psbA mRNA, the effect on mRNA processing also appears to be psbA specific. In addition to the psbA 5′ UTR, the 5′ UTR of the psbD mRNA (encoding the D2 protein) is also processed in C. reinhardtii (Rochaix et al., 1984). Primer extension analysis using a radiolabeled oligonucleotide complementary to the psbD 5′ coding region was performed on total RNA prepared from the wild type and the hf261 strain (Fig. 6). Whereas processing of the psbA 5′ UTR is completely absent in hf261, processing of the psbD 5′ UTR remains unaffected in this mutant as does D2 protein synthesis (Yohn et al., 1998b).
Processing of the psbA 5′ UTR Is Not Dependent upon Illumination
To determine if the psbA 5′ UTR is differentially processed in response to environmental conditions, total RNA was isolated from C. reinhardtii grown under continuous light or continuous darkness. Primer extension analysis indicated no difference in the relative amounts of the 5′ termini of the psbA mRNA in response to growth under continuous light or in constant darkness (Fig. 7). A previous study reported no change in psbA processing throughout the 18 h after a shift from low light to high light growth conditions (Shapira et al., 1997). These results suggest that cleavage of the psbA 5′ UTR is not directly related to light-dependent translational regulation.
Processing of the psbA 5′ UTR Is Not Dependent upon mRNA Association with Large Polyribosomes
During growth under constant light, only 35% of psbA mRNA is associated with large polyribosomes (Yohn et al., 1996). Polyribosome-associated RNA was isolated and processing of the psbA 5′ was analyzed by primer extension analysis. The extension products of psbA mRNA associated with polyribosomes were indistinguishable from those observed in total RNA (Fig. 7). Because the D1 protein is inserted into the membrane during, or soon after, synthesis, we examined the subset of polyribosome-associated psbA mRNA associated with thylakoid membranes. If this subset of mRNAs comprises the pool of functionally relevant psbA message, we might observe a difference in the amount of processed psbA 5′ UTR in this fraction as compared with total mRNA. Again, primer extension analysis indicated no difference in the relative amount of processed psbA 5′ UTR between membrane-associated mRNA and other RNA populations (Fig. 7).
Removal of the Stem-Loop Element from the psbA 5′ UTR Does Not Affect Binding of the Light-regulated Protein Complex In Vitro
The stem-loop element upstream of the RBS had been identified previously as a binding site for a protein complex involved in the dynamic regulation of psbA translation in response to light (Danon and Mayfield, 1991). However, as a consequence of processing, >95% of the steady-state levels of psbA message lacks this stem-loop element in vivo. To determine if the processed psbA 5′ UTR was still capable of being recognized by the protein complex, a gel mobility shift experiment was used to compare complex binding to the two psbA 5′ UTRs in vitro. Heparin-agarose purified proteins from wild-type C. reinhardtii were incubated with in vitro transcribed RNAs corresponding to either the 90-nt psbA 5′ UTR (−90) or the processed 5′ UTR (−36). Complex binding is reflected by the appearance of an RNA–protein complex with reduced mobility in a nondenaturing polyacrylamide gel. As shown in Fig. 8, both RNAs are bound by protein. Furthermore, each RNA is able to compete the binding of the complex to the other RNA when added in excess as a cold competitor of the labeled RNA for the protein complex. The possibility of the −90 RNA being processed to the −36 RNA in vitro before complex formation is excluded by the observation that the migration of the −90 RNA–protein complex is retarded relative to the −36 RNA–protein complex.
Previous experiments have identified the primary psbA binding protein to be RB47, the cPABP (Danon and Mayfield, 1991; Yohn et al., 1998a). UV cross-linking of both radiolabeled transcripts (−90 or −36) complexed with heparin-agarose purified proteins resulted in the specific labeling of RB47 (Fig. 9). The lower intensity of RB47 labeling using the −36 transcript may be due to fewer radiolabeled residues in close proximity to the bound cPABP for the shorter RNA fragment or may reflect slightly lower binding affinity of the 36-nt 5′ UTR compared with the 90-nt 5′ UTR.
RAC was used to further characterize complex binding to the processed 5′ UTR. RNA transcripts corresponding to the entire (−90) and processed (−36) psbA 5′ UTRs were immobilized on solid supports. Heparin-agarose– purified C. reinhardtii lysate was passed over each support and bound proteins were eluted with high salt. An identical set of proteins was eluted from both columns as observed by Coomassie blue staining (Fig. 10,A). Western analysis with antibodies specific for the cPABP (RB47), the chloroplast-localized protein disulfide isomerase (RB60) (Kim and Mayfield, 1998), and a 38-kD component of the psbA binding complex (RB38) further demonstrated that both psbA 5′ UTRs are recognized by an identical protein complex in vitro (Fig. 10 B). Taken together, these results predict that removal of the stem-loop element from the psbA 5′ UTR does not prevent the binding of the psbA-specific protein complex nor does it preclude the dynamic regulation of translation in response to light.
In both C. reinhardtii and higher plants, the expression of nuclear and chloroplast genes is tightly regulated to allow for a rapid response to fluctuations in environmental conditions. The regulation of chloroplast-encoded genes is dependent upon sequences within their 5′ UTRs that influence message stability, translation, and possibly message localization (Gillham et al., 1994; Mayfield et al., 1995; Rochaix, 1996). Mutational analysis of the 5′ UTR of the psbA mRNA has shown that RNA elements, including a consensus SD sequence and an upstream stem-loop element, mediate D1 expression (Mayfield et al., 1994). In this study, primer extension analysis of the psbA mRNA from a variety of C. reinhardtii mutants has revealed a correlation between processing of the 5′ UTR and association of the mRNA with ribosomes. Alterations to the psbA 5′ UTR that disrupt the RBS prevent RNA processing. Nuclear mutations that specifically block D1 protein synthesis, in part by reducing ribosome association, specifically reduce psbA mRNA processing. The proximity of the processing site to an accessible and competent RBS, and the influence on processing of nuclear-encoded factors that affect ribosome association, suggest a relationship between ribosome association and processing of the 5′ UTR of the psbA mRNA.
In contrast to psbA messages from C. reinhardtii, generally only one 5′ terminus, corresponding to the unprocessed 5′ UTR from C. reinhardtii, has been observed for the psbA mRNA isolated from higher plants (Zurawski et al., 1982; Link and Langridge, 1984; Sugita and Sugiura, 1984; Boyer and Mullet, 1986; Boyer and Mullet, 1988; Hanley-Bowdoin and Chua, 1988; Liere et al., 1995). Nevertheless, processing of mRNAs is a common phenomenon in higher plant chloroplasts, in which most messages are transcribed as polycistronic transcripts. For example, the petD mRNA, which is transcribed as a dicistronic message downstream of petA, contains elements within the 5′ UTR that are required for accumulation of the monocistronic petD mRNA. Failure to process the petD 5′ UTR results in the loss of translation of the petD message (Sakamoto et al., 1993; Sakamoto et al., 1994a,b).
Processing of the 5′ UTR of the rbcL transcript, the chloroplast mRNA encoding for the large subunit of Rubisco, is also correlated with ribosome association and protein synthesis. In C. reinhardtii, for example, changes in large subunit synthesis in response to light levels coincide with changes in processing of the 5′ UTR (Shapira et al., 1997). rbcL transcripts with longer 5′ UTRs (beginning 168 nt upstream of the start codon rather than 93 nt) are not associated with polyribosomes and probably are not translationally competent. Recovery of large subunit synthesis is accompanied by an increase in the relative level of processed mRNAs. In mature barley leaves, inhibition of large subunit protein synthesis by methyl jasmonate is accompanied by a change in the predominant rbcL 5′ UTR, resulting from alternative processing of the primary transcript. Once again, the longer 5′ UTR is not associated with polyribosomes (Reinbothe et al., 1993). These observations further suggest that in both C. reinhardtii and higher plants, processing of 5′ UTRs coincides with ribosome association of those mRNAs. Thus, although it is often assumed that 5′ end formation occurs before, and is necessary for proper translation initiation, the available data also support the possibility that processing of mono- and polycistronic 5′ UTRs may be a consequence of ribosome association, rather than a prerequisite for it.
Although processing of the C. reinhardtii psbA 5′ UTR appears to be linked with translation initiation at the level of small ribosomal subunit recognition of the SD sequence association, no correlation was observed between processing and the dynamic regulation of translation in response to illumination. When C. reinhardtii is grown in the dark rather than the light, D1 protein synthesis is reduced in conjunction with reduced binding activity of the cPABP complex. Nevertheless, no difference in processing was observed between psbA mRNA isolated from C. reinhardtii grown in the light or in the dark, nor is there a direct linear relationship between the amounts of light-regulated translational activator proteins, such as the cPABP, and processing in the nuclear mutants. Therefore, processing is dependent upon at least the initial stages of ribosome assembly at the SD sequence, which occurs before, and independent of, dynamic light-dependent translation. However, some proteins may be common to both of these processes.
Previous studies have demonstrated that mutations introduced within the psbA 5′ UTR upstream of the RBS can disrupt D1 expression. Typically, these mutants accumulate normal amounts of psbA mRNA but synthesize D1 protein at <5% of the wild-type level (Mayfield et al., 1994). Nevertheless, the psbA 5′ UTR in each of these mutants is processed normally. Processing within these transcripts removes the mutated RNA upstream of the RBS and results in a steady-state pool of psbA transcripts that is virtually indistinguishable from the wild type. Therefore, sequences upstream of the RBS must influence D1 expression before processing and ribosome association, possibly by acting on steps such as formation of a preinitiation complex or perhaps even in localization of the psbA mRNA to specific regions of the chloroplast (Rochaix, 1996). The D1 protein must be synthesized in close proximity to the thylakoid membrane, and translational activators, including the cPABP, have been shown to be associated with a specific membrane fraction within the chloroplast (Zerges and Rochaix, 1998). Perhaps these upstream sequences act in concert with translational activator proteins to target the psbA mRNA to membrane-associated ribosomes.
Sequences upstream of the processing site could also influence chloroplast gene expression at the level of message stability, acting before the early stages of ribosome association. For example, the 5′ UTR of the chloroplast psbD mRNA from C. reinhardtii is processed in a manner similar to that of the psbA mRNA (Rochaix et al., 1984; Erickson et al., 1986; Nickelsen et al., 1994). Analysis of nuclear mutants that destabilize the psbD mRNA has indicated that the sequences upstream of the processing site in the psbD 5′ UTR may mediate this mRNA stability by interacting with nuclear encoded factors (Nickelsen et al., 1994). Furthermore, a deletion engineered within the psbD 5′ UTR designed to initiate psbD transcription at the processing site resulted in a lack of psbD mRNA accumulation despite a wild-type rate of transcription of the mRNA (Rochaix, 1996). An analogous chloroplast deletion engineered to initiate psbA transcription at the processing site also resulted in failure to accumulate any psbA mRNA (data not shown). These data show that sequences upstream of the RBS can influence message stability, likely acting during the early stages of mRNA maturation. Interestingly, an endonucleolytic cleavage site associated with mRNA degradation has been mapped within the spinach psbA 5′ UTR in vitro. Similarities exist between this endonucleolytic cleavage site and the processing site within the C. reinhardtii psbA 5′ UTR. Both cleavage sites are located immediately upstream of a consensus RBS. In spinach, this site also defines the upstream boundary of the core RNA element recognized by an RB complex that includes the ribosomal protein S1 (Klaff, 1995; Alexander et al., 1998). The close proximity of this cleavage site in spinach psbA mRNA to the putative RBS and the site of ribosome assembly may reflect an intimate relationship between translation initiation, mRNA processing, and message stability. This relationship is also observed for psbA mutants in C. reinhardtii (RBS-del, RBS-Alt, RBS-paired) in which the loss of ribosome binding due to the absence of a SD sequence results in the absence of processing and a reduction in psbA mRNA accumulation (Mayfield et al., 1994).
Originally, mutations to the upstream stem-loop element of the psbA mRNA were thought to influence D1 translation by altering the binding site of the light-regulated cPABP complex as the stem-loop element was found to be protected by protein binding in a T1-gel mobility shift assay (Danon and Mayfield, 1991; Mayfield et al., 1994). However, gel shift assays, UV cross-linking, and RAC assays presented here all indicate that both the 90-nt and processed psbA 5′ UTRs are recognized by an identical set of psbA-specific RB proteins. Thus, dynamic light-regulated translation via binding of these activator proteins should still be possible with the truncated psbA mRNA.
Based on these and other results, we propose a model for psbA maturation and translation in the C. reinhardtii chloroplast. The psbA mRNA is transcribed with a 5′ UTR extending at least 90 nt upstream of the start codon. The 5′ UTR is rapidly processed to the −36 transcript but not before the upstream RNA elements influence D1 expression. These upstream elements could affect expression at a number of key steps including RNA stability, formation of a preinitiation complex, or mRNA localization. Processing is mediated by proteins, including ribosomal subunits and possibly other initiation factors, that recognize the RBS during the early stages of ribosome association. The nuclease responsible for processing has not yet been identified but appears to be closely associated with a ribosomal subunit or with a protein complex required for ribosome association. Interaction of the ribosomal subunits and associated proteins with the processed 5′ UTR may act to stabilize the message. Processing may also initiate an mRNA degradation pathway affecting the half-life of the transcripts. Whereas removal of the upstream sequences may be necessary to make the message competent for high levels of translation, processing alone is not sufficient for translation of the message as evidenced by mutations, both chloroplast and nuclear, that affect psbA translation without eliminating processing. Upon illumination, the binding activity of a set of psbA-specific RB proteins, including the cPABP, is enhanced in response to increased photosynthetic activity. Binding of this protein complex stimulates translational activity, probably acting via recognition of the psbA 5′ UTR through a stretch of adenine residues between the RBS and the start codon (Bruick, R.K., and S.P. Mayfield, manuscript in preparation). While the molecular basis underlying each of these events remains unknown, we have begun to define distinct RNA elements within the psbA 5′ UTR that are recognized by trans-acting factors to orchestrate message stability, mRNA processing, general translational competency, and dynamic translational activation.
We thank Christopher Yohn and Amybeth Cohen for critical reading of this manuscript and for providing membrane-associated RNA (A. Cohen).
This work was supported by funds from the National Institutes of Health to S.P. Mayfield (GM54659). R.K. Bruick was supported by a National Science Foundation (NSF) predoctoral fellowship and an NSF Graduate Research Traineeship in Macromolecular Structure and Plant Biology.
Abbreviations used in this paper
Address correspondence to Stephen P. Mayfield, Department of Cell Biology and The Skaggs Institute for Chemical Biology, The Scripps Research Institute, 10550 North Torrey Pines Rd., La Jolla, CA 92037. Tel.: (619) 784-9848. Fax: (619) 784-9840. E-mail: email@example.com