Translational regulation of chloroplast genes. Proteins binding to the 5'-untranslated regions of chloroplast mRNAs in Chlamydomonas reinhardtii.

We have examined the effects of illumination, carbon source, and levels of chloroplast protein synthesis on trans-acting proteins that bind to the leaders of five representative chloroplast mRNAs. The accumulation of these five chloroplast mRNAs and the proteins they encode were measured in cells grown under identical conditions. Extracts from all cell types examined contain a minimum set of six chloroplast 5'-untranslated region (UTR)-binding proteins (81, 62, 56, 47, 38, and 15 kDa). Fractionation results suggest that multiple forms of the 81-, 62-, and 47-kDa proteins may exist. A 36-kDa protein was found in all cells except those deficient in chloroplast protein synthesis. Binding of the 81-, 47-, and 38-kDa proteins to the rps12 leader is effectively competed by the atpB or rbcL 5'-UTRs, indicating that the same proteins bind to all three leaders. In contrast, these three proteins do not bind to the nuclear-encoded alpha-1 tubulin leader, which bound novel proteins of 110, 70, and 43 kDa. Cis-acting sequences within the 5'-UTRs of two chloroplast mRNAs (rps7 and atpB) have been identified which are protected from digestion by RNase T1 by extracts enriched for the 81-, 47-, and 38-kDa proteins.

Regulation of chloroplast mRNA translation represents an important determinant of plastid gene expression (1,2). Abundant genetic evidence exists in the green alga Chlamydomonas reinhardtii that nuclear-encoded factors are required for the stability, processing, and translation of chloroplast-encoded mRNAs (for reviews, see Refs. [1][2][3][4]. In this alga, the lightregulated translation of the psbA mRNA encoding the D1 protein of photosystem II correlates with the binding of a 47-kDa protein(s) to a 30-nucleotide sequence within its 5Ј-untranslated region (UTR) 1 starting at position Ϫ60 5Ј to the AUG codon. This sequence ends immediately adjacent to the putative Shine-Dalgarno (SD) sequence which begins at position Ϫ30 nucleotide (5). Binding of the 47-kDa protein(s) to the 5Ј-UTR is thought to lead to the formation of an active translation complex which is inactivated in the dark by phosphorylation of a 60-kDa member of this complex that has only minor RNA contact (6). Additionally, binding of these trans-acting proteins to psbA mRNA is postulated to respond to the redox potential in the chloroplast established by photosynthesis (7). In contrast, binding of a 46-kDa protein to the psbC 5Ј-UTR in the recessive nuclear mutant F64 has been correlated with the failure to translate psbC mRNA encoding P6, the 43-kDa chlorophyll a-binding core subunit of photosystem II (8).
In addition to light-mediated translational regulation, as seen in the case of psbA mRNA, translation of plant and algal chloroplast mRNAs may respond to other environmental and physiological regulatory signals. When chloroplast protein synthesis is reduced, C. reinhardtii preferentially translates mRNAs for chloroplast-encoded ribosomal proteins (r-proteins), while translation of mRNAs for photosynthetic proteins is severely diminished (9). Furthermore mRNAs for chloroplast-encoded r-proteins appear to be constitutively transcribed, accumulate early during chloroplast development in land plants and are preferentially loaded on polysomes (10,11). Nutrients, especially carbon source, can act as central regulatory signals controlling physiology, metabolism, cell cycle, and development in land plants and green algae (for reviews, see Refs. 12 and 13). In maize cells in culture, acetate represses transcription of seven nuclear-encoded photosynthetic genes (14). Similarly, in Chlamydomonas and other "acetate flagellates" this reduced carbon source has been shown to repress the expression of the nuclear-encoded rbcS and cab genes (15,16). Much less is known about the effect of carbon source on chloroplast gene expression.
In this study we have attempted to broaden our understanding of the participants required for translational regulation in chloroplasts of C. reinhardtii by examining the spectrum of proteins capable of binding to the 5Ј-UTRs of representative chloroplast-encoded mRNAs specifying photosynthetic and rproteins (1)(2)(3)(4).
Northern Analysis-Wild type and mutant strains of C. reinhardtii grown as described above were harvested by centrifugation (8,000 ϫ g, 10 min, 4°C) and frozen at Ϫ70°C, and the pellets were thawed by the addition of 4 M guanidine isothiocyanate, 25 mM sodium citrate (pH 7), 0.5% sarcosyl, and 0.1 M 2-mercaptoethanol for RNA extraction as described elsewhere (29). RNA was size fractionated on a 1.2% agarose formaldehyde gel (23) and blotted to a Magna NT membrane (MSI Scientific). The blots were hybridized under stringent conditions (30) with cloned probes for the atpB (P-130), psbA (P-269), rbcL (P-266), rps7 (P-395), and rps12 (P-175) genes labeled with [ 32 P]dATP (DuPont NEN) using a random priming kit (Boehringer Mannheim) and washed under the same conditions. Antisera-The ␤-subunit of ATP synthase from wild type C. reinhardtii (CC-125) was overexpressed in Escherichia coli from a 2.1kilobase pair NcoI-BamHI fragment containing the atpB coding sequence cloned into pET3a8c (P-305) (31). Similarly, r-protein S12 from wild type C. reinhardtii (CC-125) was overexpressed from a 550-base pair NcoI-SmaI fragment containing the rps12 gene cloned into pET3a8c (P-304). Overexpressed proteins were eluted from SDS-polyacrylamide gels and used for immunization of rabbits (32). Other polyclonal antibodies used were raised against r-protein S7 isolated from chloroplast ribosomes of C. reinhardtii (CC-125), a synthetic peptide corresponding to residues 58 -86 deduced from the D1 sequence of tobacco and spinach, the Rubisco holoenzyme from a C3-plant, and ␤-tubulin of C. reinhardtii.
Protein Isolation and Immunoblotting-Wild type and mutant strains of C. reinhardtii were grown as described above. Cells were harvested, washed once with water, pelleted again, resuspended at 10 8 cells/ml in protein loading buffer (30 mM Tris-HCl (pH 6.8), 1% SDS, 0.1 M dithiothreitol, 10% glycerol, 0.01% bromphenol blue), incubated at 100°C for 1 min, and centrifuged for 5 min. Protein corresponding to 10 6 cells was electrophoresed on 10 -17% SDS-polyacrylamide gradient gels, electroblotted to nitrocellulose (Schleicher & Schuell, BA85), and blocked with Tris-buffered saline, plus 5% w/v non-fat dry milk. Proteins were detected using the Renaissance chemiluminescence detection kit (DuPont NEN), and various exposures of the gel to Hyperfilm-ECL (Amersham Corp.) were quantified by digitizing the image and analyzing it using NIH Image software (Version 1.55). A linear relationship between signal and protein concentration was shown to exist in multiple exposures using a dilution series of protein extracts from mixotrophically grown cells probed with the antibody against the ATPase ␤ subunit.
In Vitro RNA/Protein Binding Experiments-Wild type and mutant strains of C. reinhardtii were grown as described above, and protein extracts were prepared essentially according to the protocol of Danon and Mayfield (5). Cells from 6-liter cultures at 1-3 ϫ 10 6 cells/ml were pelleted, frozen in liquid nitrogen, and stored at Ϫ70°C until processed. Cell pellets were thawed in low salt extraction buffer (10 mM Tris-HCl (pH 7.5), 10 mM NaCl, 10 mM MgCl 2 , 5 mM ␤-mercaptoethanol), and the following protease inhibitors from Sigma were added: 1 g/ml each of leupeptin, chymostatin, and pepstatin, 2 g/ml antipain, 10 g/ml benzamidine HCl, and 76 g/ml phenylmethylsulfonyl fluoride. These cells were broken at 4000 p.s.i. in a French press, and the cell extract was centrifuged for 1 h at 200,000 ϫ g (Beckman 70Ti rotor, 52,000 rpm) at 4°C. The S-200 supernatant was immediately applied to a 50-ml heparin-Actigel column (Sterogene, Bioseparations) equilibrated with buffer A (20 mM Tris-HCl (pH 7. h at 37°C. 1 unit of RNase-free DNase I (Sigma) was added, and the reaction was incubated for an additional 10 min at 37°C. Under these conditions RNAs were labeled to specific activities of 5 ϫ 10 8 to 2 ϫ 10 9 cpm/g. Unlabeled transcripts used in competition experiments were synthesized as above except that all four unlabeled ribonucleotides were included, and the reactions were scaled up to a 100-l final volume. The reactions were phenol-chloroform-extracted, and the RNA probes were separated from unincorporated nucleotides on Sephadex G-25 spun columns (23). All 5Ј-UTR probes were derived from the clones described above. Total RNA was isolated as described above from E. coli cells (strain XL1-Blue) grown in LB medium to mid log phase (0.65 A 600 ) and from C. reinhardtii mutant strains CC-373 and CC-744 grown mixotrophically to mid log phase (1-3 ϫ 10 6 cells/ml). These RNAs were used as alternative competitors to E. coli tRNA (Calbiochem) in certain UV cross-linking experiments.
For the gel mobility shift assay, pooled or individual heparin-Actigel column fractions (7 g) were preincubated for 10 min at room temperature with 5 units of RNasin in the presence of 3 mM MgCl 2 in a total volume of 5 l and then added to 20 g of E. coli competitor tRNA and 32 P-labeled rps12 leader (15 pM) in a final volume of 15 l. After 15-min incubation at room temperature, 2 l of xylene cyanol were added, and the mixtures were loaded onto a 15 cm ϫ 15-cm 5% (49:1 acrylamide: bisacrylamide) native polyacrylamide gel containing 1 ϫ TBE and electrophoresed in TBE buffer at 25 mA for 2 h until the dye marker was about 2.5 cm from the bottom of the gel. The gel was then fixed in 10% methanol, acetic acid, dried, and exposed to x-ray film (Kodak XAR5) at Ϫ70°.
The conditions for UV cross-linking were described previously (5,33). Binding reactions (15 l) were performed as follows: 7 g of protein from individual heparin-Actigel column fractions was preincubated in the presence of 3 mM MgCl 2 and 0.5 units of RNasin in a volume of 5 l for 10 min at 22°C. E. coli tRNA (0.1 g) as a nonspecific competitor and [ 32 P]UTP-labeled chloroplast 5Ј-UTR probe (about 15 pM) were added to give a final volume of 15 l. After 15 min at 22°C, the binding reactions were placed on ice and cross-linked with 254-nm UV irradiation of 1.0 J/cm 2 using a Stratalinker (Stratagene). The RNA transcripts were digested with 10 g of RNase A (Sigma) for 30 min at 55°C, boiled for 1 min in protein loading buffer, separated on 15 cm ϫ 15-cm, 7.5-17% SDS-polyacrylamide gels, dried, and exposed to x-ray film (Kodak XAR5 or Fuji RX) at Ϫ70°using intensifying screens (DuPont Chronex). In competition experiments, E. coli tRNA or rRNA (100 -300fold molar excess) or total RNA (50 -100-fold mass excess) from the C. reinhardtii strains CC-373 and CC-744 were preincubated with protein extracts prior to the addition of [ 32 P]UTP-labeled leaders. Unlabeled 5Ј-UTRs from the chloroplast atpB and rbcL genes or the polylinker sequence from the pBluescript KS ϩ plasmid (0 -2.5 nM) were also included in certain binding reactions as specific competitors prior to addition of the labeled 5Ј-UTR RNA. Analysis of UV cross-linking reactions for each column fraction on the autoradiograms allowed us to define binding proteins of the same molecular weight which differ in 2 B. Randolph-Anderson, unpublished results. their elution profile from the heparin-Actigel column due either to protein modifications or to differences in their amino acid sequence.
Filter Binding Assay-The RNA binding assay used here was similar to a nitrocellulose filter binding assay described previously (34). Binding reactions were carried out as described above for UV cross-linking with the following modifications. Reactions (50 l) contained [ 32 P]UTPlabeled rps12 probe (14 pM) and pooled heparin-Actigel protein extracts (CC-105 Ϫspec, fractions 10 -12) at concentrations of 10 Ϫ4 to 10 4 g/l for saturation binding studies. RNasin (10 units) was added to all reactions to inhibit RNase activity. After incubation (15 min at 25°C) the 50-l samples were filtered through 0.45-m pore size nitrocellulose disks (Millipore, HAWP, 25 mm) and washed twice with 1 ml of binding buffer. Filters were dried, and the retained radioactivity was determined by scintillation counting.
Mapping of Protein Binding Sites-For RNase T1 protection gel shift assays, 15 l of the binding reaction mixture containing 0.5 fmol of [ 32 P]UTP-labeled atpB or rps7 5Ј-UTR and 7 g of pooled heparin-Actigel column fractions were incubated for 10 min at 25°C, treated with 10 units of RNase T1 (U. S. Biochemical Corp./Amersham Corp.) for 5 min at 25°C and electrophoresed on a 5% native acrylamide gel in 1 ϫ TBE buffer. Gel-retarded bands were visualized by autoradiography, excised, and eluted from the acrylamide in 0.5 M NH 4 Ac, 10 mM MgAc, 0.1 mM EDTA (pH 8.0), 0.1% SDS at 37°C overnight. For subsequent RNase T1 mapping of the protected fragments, the eluted samples were phenol-extracted, ethanol-precipitated, and resuspended in a minimum volume (5 l) of diethyl pyrocarbonate-treated water to which was added 20 units of RNase T1. The samples were incubated at 22°C for 10 min, denatured, and analyzed on 20% denaturing acrylamide gels.
Digitization and Quantification of Autoradiographs-All figures were prepared from digitized autoradiographs (300 dots/inch) using a Sharp flatbed scanner (model JX-320) and Canvas TM Version 3.5 software. Western and Northern blot autoradiographs were quantified using NIH Image Version 1.55. UV cross-linking competition gels were imaged using a Molecular Dynamics PhosphorImager and quantified using ImageQuant Version 3.3 software.

Effects of Different Environmental and Physiological Conditions on the Accumulation of Representative Chloroplast Proteins and
Their mRNAs-To provide a base line for interpreting the spectrum of proteins binding to the leaders of chloroplast mRNAs, we characterized the expression of five representative chloroplast genes under the growth conditions from which the protein extracts were prepared. Genes examined include rps7 and rps12 encoding ribosomal proteins S7 and S12, and rbcL, psbA, and atpB, encoding LSU (the large subunit of Rubisco), D1 (the photosystem II reaction center protein), and ␤ (the catalytic subunit of ATP synthase). The rbcL, psbA, and atpB genes of C. reinhardtii are monocistronic transcription units (35)(36)(37), while the two r-protein genes examined are contained within polycistronic messages: rps7 is the 5Ј member of a dicistronic unit containing atpE (38), and rps12 is the 3Ј-terminal gene in a four-gene operon containing in order: psbJ, atpI, and psaJ (27,39,40). Fig. 1 compares the steady state levels of mRNA and protein for the psbA, rbcL, atpB, rps7, and rps12 genes found in cells grown under different conditions of 1) illumination (mixotrophic ϭ light ϩ acetate versus heterotrophic ϭ dark ϩ acetate), 2) carbon source (mixotrophic ϭ light ϩ acetate versus phototrophic ϭ light ϩ CO 2 ), and 3) levels of chloroplast protein synthesis under mixotrophic growth (ac-20 cr-1 versus wild type; spr-u-1-27-3, ϩ versus Ϫ spec). Representative Northern blots and immunoblots for each of the five chloroplast genes are shown in Fig. 1A and the values of replicate experiments are quantified in Fig. 1B as a percentage of the values for phototrophically grown wild type cells with the range of variation indicated. To verify equivalent protein loading for each extract, the accumulation of the nuclear-encoded ␤-tubulin protein is shown (Fig. 1A).
Unlike angiosperms and algae such as Euglena, which do not maintain a differentiated chloroplast in the dark, heterotroph-ically grown wild type cells of C. reinhardtii synthesize chlorophyll and contain well developed thylakoids stacked into grana (18,41). Accumulation of LSU, ␤, S7, and S12 proteins in C. reinhardtii was found to be essentially unaffected by dark versus light growth in the presence of acetate. In contrast, heterotrophically grown cells appear to accumulate only 24% of D1 found in mixotrophically grown cells and 40% of that in phototrophically grown cells. While levels of psbA and rps7 mRNA show mostly minor variations between these growth conditions, accumulation of the rps12, rbcL, and atpB mRNAs in mixotrophically grown cells is 30 -50% of that in heterotrophically or phototrophically grown cells. Thus levels of protein accumulation are not directly coupled to steady state levels of mRNA in the chloroplast of this alga.
Growth of C. reinhardtii in the light on acetate has been reported to reduce the abundance of the nuclear-encoded cabII-1 transcript, encoding a chlorophyll a/b-binding protein and to modify the ratio of the two nuclear transcripts encoding the small subunit (SSU) of Rubisco compared to phototrophically grown cells (16,42). We found that acetate also differentially altered the expression in light grown cells of the D1 and LSU photosynthetic proteins encoded by the chloroplast rbcL and psbA genes. Mixotrophically grown cells accumulated about 2-fold higher levels of D1 protein and 1.5-fold less LSU protein compared to phototrophically grown cells. No appreciable changes were observed in accumulation of the S7, S12, or ␤ proteins under these conditions. Accumulation of mRNAs encoding the photosynthetic proteins LSU and ␤ were reduced 2.2-5-fold in mixotrophically versus phototrophically grown cells (Fig. 1). In contrast, accumulation of the psbA mRNA encoding the D1 protein was only slightly reduced by the presence of acetate in the medium. Again little or no correlation was seen between accumulation of a specific mRNA and its cognate protein.
We also evaluated the effects of reduced chloroplast protein synthesis on the accumulation of chloroplast-encoded photosynthetic and r-proteins and their mRNAs. Two mutant strains deficient in chloroplast protein synthesis were utilized: 1) the nuclear ac-20 cr-1 double mutant which results in a permanent deficiency of chloroplast ribosome monomers (18), and 2) the chloroplast 16 S rDNA mutant, spr-u-1-27-3, which when grown mixotrophically in the presence of spectinomycin, accumulates nearly wild type levels of chloroplast ribosomes, but is deficient in chloroplast protein synthesis (9). Mixotrophically grown cells of ac-20 cr-1 were found to be severely deficient in the chloroplast encoded D1, LSU, and ␤ photosynthetic proteins, but accumulated nearly wild type levels of chloroplast encoded r-proteins S7 and S12 (Fig. 1). Levels of rps7, rps12, and rbcL mRNAs detected in ac-20 cr-1 (Fig. 1) were equal or greater than those in wild type cells grown on acetate in the light, whereas the psbA and atpB messages were at least 60% of the values for mixotrophically grown wild type cells.
The chloroplast mutant spr-u-1-27-3 grown on spectinomycin accumulated no D1 or LSU and greatly reduced ␤, but nearly normal levels of r-proteins S7 and S12 (Fig. 1). Accumulation of atpB and rbcL mRNAs under these conditions was greater than in wild type or in the mutant grown in the absence of spectinomycin, whereas psbA mRNA was greatly reduced. In contrast levels of rps7 mRNA were equal to those in cells with normal chloroplast protein synthesis although the rps12 mRNA exhibited a reduction. Our results with both mutants are consistent with the hypothesis of class-specific translational regulation of ribosomal versus photosynthetic proteins under conditions of reduced chloroplast protein synthesis (1,9).
In Vitro Gel Retardation and UV Cross-linking Analysis of the 5Ј-UTR-binding Proteins-To identify trans-acting proteins which might mediate the differential accumulation of these chloroplast encoded proteins, heparin-Actigel column fractions enriched for nucleic acid binding proteins from S-200 extracts of the aforementioned cells were tested for the presence of proteins that either shift the mobility of chloroplast mRNA leaders in gel retardation assays, or UV cross-link to these leaders. Column fractions derived from spr-u-1-27-3 grown in the presence of spectinomycin were incubated with 32 P-labeled 5Ј-UTRs of the rps12 and atpB mRNAs, and protein binding was analyzed by retardation of the RNA fragments' migration on native gels. This assay revealed bands with retarded mobility in reactions with a contiguous subset of gradient fractions eluting at intermediate KOAc concentrations (about 0.4 -0.8 M). Unshifted bands were seen in reactions with either high or low salt-eluting fractions or in control reactions containing either leader alone or leader ϩ boiled extract ( Fig. 2A). Since the degree of band retardation was fraction dependent, multiple protein interactions are likely to be involved. To quantify the amount of protein extract required to saturate a 5Ј-UTR binding reaction, a filter binding assay was performed (see "Materials and Methods"). The rps12 5Ј-UTR was synthesized in vitro and radiolabeled by incorporation of [␣-32 P]UTP. Purified RNA was incubated in the presence of increasing concentrations of protein (Fig. 2B, fractions 10 -12), and the resulting RNA protein complexes were collected on nitrocellulose filters. Bound radioactivity, reflecting the number of RNA-protein complexes formed, was determined by scintillation counting. Under these conditions saturation binding occurs at a protein concentration of ϳ1 mg/l (Fig. 2B). In all subsequent UV cross-linking and competition experiments, a subsaturating concentration of protein (7 g/l) was used to ensure a high degree of specificity in the RNA-protein interactions. To characterize further the individual proteins in the column fractions with band shifting activity, we analyzed these fractions for the presence of proteins that UV cross-link to the leader sequences of the chloroplast psbA, rbcL, atpB, rps7, and rps12 genes. The leader of the nuclear ␣-1 tubulin mRNA was included as a control to verify the chloroplast specific nature of these binding proteins.
A high degree of reproducibility was found in the pattern of 5Ј-UTR binding proteins between individual UV cross-linking experiments done with the same extract and sequential preparations of in vitro transcribed leaders of the same chloroplast gene. Duplicate heparin-Actigel fractions prepared from separate cultures of the same genotype (spr-u-1-27-3, ϩspec) or from phenotypically similar genotypes (spr-u-1-27-3, Ϫspec and wild type) also contained the same sets of 5Ј-UTR binding proteins. Since the chloroplast occupies a large percentage of cell volume in C. reinhardtii (17), chloroplast proteins would be expected to comprise a large proportion of the total protein in cell extracts and hence predominate among total proteins in the S-200 preparations.
The presence of 100-fold molar excess E. coli tRNA over labeled 5Ј-UTR in the binding reactions minimized binding of nonspecific proteins from the pooled heparin-Actigel columns to the chloroplast 5Ј-UTRs. Increasing the tRNA concentration or adding total E. coli RNA to 250-fold molar excess compared to labeled probe had no detectable effect on the binding of the 81-, 47-, and 38-kDa proteins to the psbA (Fig. 3) or atpB (data not shown) chloroplast leaders, whereas addition of 300-fold molar excess tRNA or total E. coli RNA resulted in a slight reduction in the binding of these proteins to both leaders. None of the proteins bound to the 5Ј-UTR from the nuclear ␣-1 tubulin gene of C. reinhardtii (Fig. 3). Instead, the ␣-1 tubulin leader bound three new proteins of 110, 70, and 43 kDa which were also not competed off by excess tRNA or total RNA from E. coli. Thus we believe the purification of whole cell lysates on heparin-Actigel columns and UV cross-linking is a valid approach for isolating and characterizing proteins that bind specifically to 5Ј-UTRs of chloroplast mRNAs.
Individual lanes in the autoradiographs of UV cross-linking gels (Figs. 4 and 5) represent binding reactions with sequentially eluted column fractions from extracts of cells grown under the conditions specified. Cross-linking of a given protein to any one of the five leaders tested is evidence that the protein is present in the particular extract and fraction analyzed. Absence of binding of a protein to a specific leader may result from either protein:protein interactions or structural features within a given leader which may inhibit binding. Variability in signal intensities of UV cross-linked proteins between leaders for a given extract arises in part because of differences in specific activity of the leader probes, differences in base composition (i.e. number of [ 32 P]UTP residues) of the individual leaders within the protein binding domain and/or differences in exposure times during autoradiography. The gels shown in Figs. 4 and 5 were exposed in an attempt to normalize the intensity of the 81-kDa protein, and exposures of varying lengths were used to verify the results presented.
Spectrum of 5Ј-UTR Binding Proteins Present in Cells Grown under Different Environmental Conditions-Individual fractions from heterotrophically, mixotrophically, and phototrophically grown wild type cells and from mixotrophically grown cells of two mutants under conditions of reduced chloroplast protein synthesis were assayed by UV cross-linking to 32 Plabeled leaders of the five representative chloroplast mRNAs transcribed in vitro. Extracts from wild type cells grown under all three conditions contain seven proteins (81, 62, 56, 47, 38, 36, and 15 kDa) which consistently UV cross-link to one or more of the chloroplast leaders (Table I). Three of these proteins (81, 62, and 47 kDa) are observed to occur either as doublets or are found in noncontiguous heparin-Actigel column fractions (Figs. 4 and 5). These observations suggest either that unique proteins of these molecular masses undergo modifications (e.g. phosphorylation) that alter their physical properties or that several unrelated proteins of similar molecular masses are present in the extracts.

Effects of Illumination on the Presence of UTR Binding Proteins in Wild
Type-The effect of light versus dark growth in the presence of acetate on the pattern of proteins binding to the 5Ј-UTRs of individual chloroplast mRNAs is shown in Fig. 4. Although the same sets of proteins appear to be present in extracts of both mixotrophically and heterotrophically grown cells, differences are apparent in their binding patterns to the 5Ј-UTRs of individual chloroplast genes. For example, the 47-kDa protein present in extracts of dark grown wild type cells FIG. 2. A, autoradiogram of a gel retardation assay demonstrating the presence of proteins in heparin-Actigel column fractions from an S-200 extract of C. reinhardtii that bind to the 5Ј-UTR of the chloroplast rps12 mRNA. 32 P-Labeled transcripts for the rps12 5Ј-UTR were incubated either alone (control), in the presence of pooled column fractions heated to 100°C for 10 min (boiled extract), or in the presence of individual column fractions (heparin-Actigel column fractions). The S-200 extract was prepared from the spr-u-1-27-3 mutant grown in the absence of spectinomycin. All reactions contained a ϳ100-fold mass excess E. coli tRNA to compete for any nonspecific RNA binding proteins present in the column fractions. Each lane represents a gel retardation assay with proteins from a single fraction eluted with an increasing potassium acetate (KOAc) gradient. B, concentration dependent binding of heparin-Actigel-purified proteins to the rps12 5Ј-UTR. Filter binding assay was conducted by incubating uniformly radiolabeled rps12 5Ј-UTR (14 pM) in the presence of increasing concentrations (1 ϫ 10 Ϫ4 to 1 ϫ 10 4 g/l protein) of pooled heparin-Actigel column fractions from the above extracts. Subsaturating amounts of pooled heparin-Actigel column fractions (7 g/l) from phototrophically grown wild type cells corresponding to those represented in Fig. 4 were UV cross-linked to [␣-32 P]UTP-labeled psbA and ␣-1 tubulin 5Ј-UTRs (15 pM). E. coli tRNA was added as competitor prior to addition of labeled probe at concentrations of 100-, 250-, and 300-fold molar excess over the ␣-32 Plabeled 5Ј-UTR RNA (lanes 1-3). Reactions in lanes 4 and 5 contained a 150-and 200-fold excess of E. coli rRNA, respectively. Apparent molecular masses of the bands in kilodaltons were estimated compared to prestained molecular mass standards. The uppermost band in the ␣-1 tubulin panel (*) may represent either a high molecular mass ␣-1 tubulin-specific protein, or one or more ␣-1 tubulin-specific proteins that failed to enter the resolving gel.
UV cross-links to the psbA leader in three contiguous fractions. However, the signal for the middle fraction is weakest, suggesting that two forms of the 47-kDa protein may exist. Comparable fractions from light grown cells show strongest cross-linking in the two fractions eluting at highest salt concentrations. Thus, only the high salt eluting form of the 47-kDa protein may cross-link to the psbA leader in extracts of phototrophically grown cells. While the binding patterns for the 81-, 62-, and 56-kDa proteins to the psbA leader are quite similar in extracts from mixotrophically and heterotrophically grown cells, the 36-kDa band is much stronger in the latter extract. In the case of the rbcL leader, binding of the 56-kDa protein appears to be reduced in the extract from heterotrophic cells compared to that from mixotrophic cells. Even though an 81-kDa protein is present in three fractions in extracts of mixotrophically grown cells that cross-links to the psbA and rbcL 5Ј-UTRs, an 81-kDa protein from only one of these fractions cross-links to the atpB leader (Fig. 4).

FIG. 4. Autoradiograms of SDS gels showing 5-UTR-binding proteins present in extracts of wild type cells grown heterotrophically, mixotrophically, or phototrophically that bind to leaders of representative chloroplast mRNAs.
Proteins in heparin-Actigel column fractions from cell extracts were UV cross-linked to 32 P-labeled RNAs (synthesized in vitro) corresponding to the psbA, rbcL, atpB, rps7, and rps12 5Ј-UTRs. Each lane within a panel represents a binding reaction with proteins from a single fraction eluted with an increasing potassium acetate (KOAc) salt gradient. Each UV cross-linking reaction was repeated at least two times. Apparent molecular masses are indicated in kilodaltons compared to prestained molecular mass standards.
In light grown cells the 47-kDa protein binding to the atpB leader is observed predominantly in two fractions. Absence of a strong 47-kDa protein band from the low salt eluting fraction of the light grown cells is especially evident. A 36-kDa protein that bound to the rps7 and rps12 leaders was much more prominent in extracts from light than dark grown cells. In extracts of heterotrophically and mixotrophically grown cells, the psbA leader cross-links six of the seven proteins, whereas the rps7 leader binds only three or four (Fig. 4). Leaders of the rbcL, atpB, and rps12 mRNAs bind an intermediate number of proteins. Qualitative differences in the binding of the 47-and 15-kDa proteins to specific leaders in individual column fractions are also evident.
Effects of Carbon Source on the Presence of UTR Binding Proteins in Wild Type-Comparison of S-200 extracts from phototrophically versus mixotrophically grown cells also revealed the same basic set of seven proteins that cross-react with one or more of the five chloroplast 5Ј-UTRs tested ( Table   I). As described above for the comparison of heterotrophically versus mixotrophically grown cells, a number of differences were found in the binding pattern of individual column fractions from phototrophically grown cells to specific chloroplast leaders (Fig. 4). In the case of all five UTRs tested, the signals for the 47-kDa protein were much more intense from mixotrophic than from phototrophic cells relative to the 81-kDa binding protein in the same gel lanes. However, binding of the 38-kDa protein was enhanced in the extract from phototrophically grown cells for all five UTRs examined. In extracts of both mixotrophically and phototrophically grown cells, binding of the 47-kDa protein to the rps7 5Ј-UTR is severely reduced, and binding of the 81-kDa protein occurs only in one fraction in the extract from mixotrophic cells.  Reducing chloroplast protein synthesis over seven-to eightcell generations by mixotrophic growth of the spr-u-1-27-3 mutant in spectinomycin also affects the spectrum of proteins which binds to the five chloroplast leaders. As in the case of ac-20 cr-1, six of the seven UTR-binding proteins seen in mixotrophically grown wild type cells are present in the S-200 extract from the spr-u-1-27-3 (ϩspec) cells and all seven are seen in the spr-u-1-27-3 (Ϫspec) cells (Fig. 5). Binding of the 36-kDa protein to all five chloroplast leaders is greatly diminished or absent in extracts from the ϩspec cells. The reduction in chloroplast protein synthesis in spr-u-1-27-3 does not appear to increase the intensity of the 62-kDa band relative to the 81and 47-kDa bands as seen in ac-20 cr-1. Binding of the 56-kDa protein to the rps7 and rps12 leaders may be selectively reduced in the ϩspec extracts. Unique proteins of 60, 45, and 29 kDa UV cross-link specifically with the atpB leader in a single low salt fraction of the ϩspec extract (data not shown). A new 109-kDa protein also cross-links in the spr-u-1-27-3 (ϩspec) extract to the atpB leader (Fig. 4) in addition to a unique form of the 15-kDa protein eluting at high salt in both ϩ and Ϫspec extracts of this mutant (data not shown). The significance of the four novel bands specific for the atpB leader present in extracts from spr-u-1-27-3 grown under conditions of reduced chloroplast protein synthesis is unknown. Qualitative differences in the UV cross-linking patters of extracts from ac-20 cr-1 and spr-u-1-27-3 (ϩspec) may reflect the somewhat "leaky" nature of chloroplast protein synthesis phenotype in spr-u-1-27-3 under the latter condition compared to the more stringent phenotype of ac-20 cr-1 with its large reduction in chloroplast ribosomes. Reduction in the amounts of chloroplast synthesized photosynthetic proteins accumulated in spr-u-1-27-3 (ϩspec) is dependent upon the concentration of spectinomycin used as well as the number of generations the cells are grown in the presence of the antibiotic.

UTR Binding Proteins Present under Conditions of Reduced
Competition Experiments Demonstrate That the Same Transacting Proteins Bind to the 5Ј-UTRs of Several Different Chloroplast mRNAs-To determine whether the trans-acting proteins bound to the chloroplast-encoded 5Ј-UTRs examined were either message-specific, or general factors common to all five chloroplast leaders, we carried out competition experiments in the presence of increasing amounts of unlabeled competitor RNA under equilibrium conditions. A 50-and 100-fold molar excess of unlabeled competitor RNA from a 248-nucleotide polylinker region of pBluescript KS ϩ (Stratagene) was preincubated with a heparin-Actigel column fraction enriched for the 38-, 47-, and 81-kDa proteins prior to the addition of labeled RNA (15 pM) and irradiation with UV light. This nonspecific RNA probe was unable to compete for the binding of the 81-or 47-kDa proteins present in spr-u-1-27-3 (Ϫspec) extracts that cross-linked to the rps12 5Ј-UTR (data not shown). However, binding of 81-, 47-, and 38-kDa proteins to the rps12 leader was eliminated when unlabeled atpB or rbcL 5Ј-UTR was used as competitor RNA (Fig. 6). The relative binding affinities of these proteins were estimated by comparing the competitor concentrations resulting in a 50% reduction in cross-linking. Using the atpB leader as competitor, the binding affinities for the 81-, 47-, and 38-kDa proteins are estimated to be 0.22 nM, 0.32 nM, and 1.52 nM, respectively. This corresponds respectively to 7.1-, 10.3-, and 49-fold molar excess of unlabeled competitor over labeled rps12 leader RNA. When the rbcL leader is used as the competitor, binding affinities of 0.1 nM, 0.14 nM, and 0.48 nM are estimated for the 81-, 47-, and 38-kDa proteins corresponding to 3.2-, 4.5-, and 15.5-fold molar excesses. These results indicate that the rbcL leader has a roughly 3-fold higher affinity for the 81-, 47-, and 38-kDa proteins compared to the atpB leader. Furthermore the data suggest that binding of the 38-kDa protein to the rps12 leader is enhanced when the extract is partially depleted of the 81-and/or 47-kDa protein(s) as indicated by an initial increase in its binding prior to being competed by higher concentrations of either the atpB or rbcL leader RNA (Fig. 6). The atpB leader also efficiently competed for the binding of 81-, 62-, and 47-kDa proteins with labeled rbcL, psbA, and rps7 leaders (data not shown). These competition experiments strongly suggest that the 81-and 47-kDa proteins exist as single species in the column fraction assayed and that the trans-acting proteins in the fractions studied which crosslinked to the chloroplast-encoded messages examined are not gene-specific, but recognize all five chloroplast leaders tested.
Competition experiments utilizing total RNA isolated from mixotrophically grown cells of atpB (ac-u-c-2-21 (⌬atpB)) and psbA (ac-u-␤ (⌬psbA)) deletion mutants which lack mRNAs for these two chloroplast genes were also performed. Pooled column fractions containing the 81-, 47-, 38-, 36-, and 15-kDa proteins were cross-linked to labeled atpB or psbA leaders in the presence of a 50-and 100-fold mass excess of competitor RNAs relative to the labeled probe. If binding of a particular protein to a leader is a gene-specific event, then unlabeled competitor RNA from a deletion mutant will not compete the binding of that protein from the labeled 5Ј-UTR of the gene deleted in the competing mutant strain. Total RNA from the ⌬psbA strain competes the binding of the 81-, 47-, 38-, and 36-kDa proteins to the atpB leader more strongly than does ⌬atpB RNA (data not shown). As observed previously (Fig. 6),

TABLE I Number and molecular mass of transacting proteins present in heparin-Actigel-purified extracts of cells grown under different environmental conditions which UV cross-link to
chloroplast-encoded 5Ј-UTRs Growth conditions are denoted as: mixotrophic (M), phototrophic (P), heterotrophic (H), absence (Ϫ), and presence (ϩ) of spectinomycin under mixotrophic conditions. Proteins which bind to at least one of the five leaders examined (atpB, rbcL, psbA, rps7, and rps12) for each extract (F), those which do not bind to any leader (E), and those with weak binding ( ) are shown. the 81-kDa protein is the most efficiently competed of the three proteins using either ⌬atpB or ⌬psbA RNA. Binding of the 47-kDa protein to the atpB leader was less affected by the presence of 100-fold excess ⌬atpB RNA than any of the other proteins, suggesting one of the multiple forms of this protein may be specific for this leader. Presence of excess ⌬atpB or ⌬psbA RNA also preferentially reduced cross-linking of the 81-kDa protein, and to a lesser extent the 47-kDa protein, to the psbA leader. Since these competition experiments were done with pooled column fractions, we cannot rule out the possibility that dynamic associations between proteins in different fractions affect the overall UTR binding pattern of the pooled mixture to a particular chloroplast leader, due to protein:protein interactions.
Mapping of the Binding Site(s) for the 47-and 81-kDa Proteins to the 5Ј-UTRs of atpB and rps7 mRNAs-The two 5Ј-UTRs analyzed show no sequence identity to one another with the exception that they are all (A ϩ U)-rich (70 -84%) and contain putative Shine-Dalgarno (SD) sequences (atpB (GGAGG) Ϫ81 to Ϫ85, rps7 (GGA) Ϫ114 to Ϫ116). In light of the fact that we find 81-and 47-kDa trans-acting proteins binding to all 5Ј-UTRs examined in the absence of any primary sequence homology, we have begun to define their binding sites on the atpB and rps7 leaders in relation to putative secondary structures. Gel retardation and RNase T1-mapping experiments were conducted, and putative secondary structures were generated by the Zuker m-fold program (Fig. 7) (43,44). 32 P-Labeled RNA probes for the atpB and rps7 5Ј-UTRs were incubated with a heparin-Actigel column fraction (Fig. 6, Ϫspec) which contains primarily 47-and 81-kDa cross-linking signals but a weak 38-kDa signal, and the complexes were resolved by native PAGE (Fig. 7, A and D). In the absence of protein extract, the probe was completely degraded by treatment with RNase T1 (Fig. 7, A and D). Incubation of the labeled RNA with the same protein extract and subsequent partial digestion with RNase T1 results in a protected (gel-retarded) complex (Fig. 7,  A and D, *). The protected complex was eluted from gel slices, phenol/chloroform-extracted, and ethanol-precipitated.
To identify the RNA protected by the 81-, 47-, and 38-kDa proteins, full-length transcripts of the atpB and rps7 5Ј-UTRs were digested to completion with RNase T1, and the fragments were resolved by denaturing PAGE (Fig. 7, B and E, lane 1). Analysis of the RNase T1-protected sequences for both the atpB and rps7 leaders on the same denaturing gel resolved several fragments for each leader (Fig. 7, B and E, lane 2) which were resistant to further hydrolysis by RNase T1 (data not shown). Protected fragments of 35, 34, and 22 nucleotides with longer exposure were resolved for the atpB 5Ј-UTR (Fig. 7, B, lane 2). Protected fragments of 30, 28, 20, and 19 nucleotides were detected for the rps7 leader (Fig. 7, E, lane 2). Alignment of the atpB and rps7 fragments protected from RNase T1 digestion on the primary sequence and the putative secondary structures predicted by the Zuker m-fold algorithm (44) are shown in Fig.  7, C and D. Due to the high A ϩ T content of these leaders they probably fold into a variety of conformations with the native state of the mRNA being an equilibrium mixture of many different conformations. Further experiments will be necessary to determine the specific structures of these leaders as well as the specific binding sites for the individual trans-acting factors. DISCUSSION Our aim has been to characterize proteins present in cells of C. reinhardtii grown under different environmental conditions that bind to the 5Ј-UTRs of chloroplast mRNAs and affect their translation. In this way we hoped to identify proteins that interact with all chloroplast mRNAs (core proteins) and those which might be specific for a group of genes with related functions (e.g. chloroplast protein synthesis versus photosynthesis) or for cells grown in a specific environment (light versus dark, acetate versus CO 2 ). Previous studies of proteins that regulate translation of chloroplast mRNAs in this alga have focused strictly on trans-acting regulatory proteins required for expression of specific chloroplast genes (1,45).
Environmental Effects on Organelle Gene Expression in Chlamydomonas-Translation of many chloroplast-encoded mRNAs specifying photosynthetic proteins is light regulated in higher plants which accumulate protochlorophyllide and form etioplasts in the dark (45-47). In contrast, wild type C. reinhardtii forms chlorophyll containing chloroplasts in the dark (18). In this study, we observed a 4-fold increase in accumulation of D1 protein in C. reinhardtii cells grown mixotrophically compared to heterotrophically grown cells, although equivalent levels of psbA mRNA were present in both the light and dark (Fig. 1). This confirms previous observations that the synthesis of D1 is translationally regulated (5,6,(45)(46)(47). In contrast to D1, the proteins involved in ATP production (␤), carbon fixation (LSU), as well as r-proteins S7 and S12 accumulate to equivalent levels in C. reinhardtii cells grown in either light or dark on acetate (Fig. 1).
In C. reinhardtii and other chlorophytes, transcription of nuclear-encoded chloroplast proteins involved in light harvesting and carbon fixation appear to be repressed by acetate (13,16,42). We find that mixotrophically grown cells of C. reinhardtii accumulate 2-fold more D1 protein than phototrophically grown cells, with no change in the level of accumulation of the psbA mRNA (Fig. 1). Elevated levels of the D1 protein in mixotrophically grown cells may result from decreased turnover in response to the reduction in the rate of O 2 evolution observed compared to that of phototrophically grown cells (48). In contrast, LSU levels decrease 2-fold in cultures grown in the light or dark in the presence of acetate as a carbon source FIG. 6. Competition experiments with unlabeled atpB or rbcL 5-UTRs. Two heparin-Actigel column fractions (7 g/l) from spr-u-1-27-3 (Ϫspec) were pooled and preincubated with increasing concentrations of unlabeled atpB (A) or rbcL (B) competitor 5Ј-UTR RNA prior to addition of [␣-32 P]UTP-labeled rps12 RNA (15 pM). The samples were UV cross-linked and treated with RNase A, and the proteins were resolved on SDS-polyacrylamide gels. Radioactivity associated with each individual protein was quantified on a Molecular Dynamics Phos-phorImager and plotted as a function of competitor concentration (C and D). Apparent molecular masses are indicated (kilodaltons) compared to prestained molecular mass standards. compared to phototrophically grown cells. While accumulation of both rbcL and atpB messages was reduced under mixotrophic conditions, only the level of LSU protein accumulation was coordinately reduced and ␤ subunit protein remained virtually constant (Fig. 1). Clearly the expression of these key chloroplast genes is largely being modulated post transcriptionally in response to carbon source. In contrast, accumulation of r-proteins S7 and S12 and their respective mRNAs are much less affected by carbon source or illumination during growth.
Effects of Reduced Levels of Chloroplast Protein Synthesis on Organelle Gene Expression-We earlier demonstrated that synthesis of the chloroplast encoded LSU and ATP synthase ␣ proteins is selectively blocked under conditions of reduced chloroplast protein synthesis (9). The results reported here extend this list to include ATP synthase ␤ subunit and the photosys-tem II D1 protein (Fig. 1). Since no diminution in mRNA levels accompanies the reduced accumulation of these photosynthetic proteins (Fig. 1), their regulation appears to be post transcriptional. Our previous pulse-chase experiments (9) suggested that regulation of ATPase ␣ and LSU was occurring at the translational level. As found earlier for r-protein L-2 (9), we also observe that synthesis of r-proteins S-7 and S-12 preferentially continues under conditions of reduced chloroplast protein synthesis. Furthermore, all or most of the remaining chloroplast-encoded r-proteins must also continue to be synthesized, since the ribosome monomers made by the spr-u-1-27-3 mutant grown in the presence of spectinomycin accumulate in nearly normal amounts (49). exist of nuclear-encoded trans-acting proteins that bind to the 5Ј-UTRs of individual chloroplast and mitochondrial mRNAs in a gene-specific manner (1), only one universal binding protein (p40) has been reported for yeast mitochondria and none, to the best of our knowledge, for the chloroplast 5Ј-UTRs (16). In this study we show that extracts from wild type cells contain at least seven proteins (81, 62, 56, 47, 38, 36, and 15 kDa) that cross-link to 5Ј-UTRs of several different chloroplast mRNAs ( Table I). The 81-, 47-, and 38-kDa proteins clearly do not bind to the 5Ј-UTR from the nuclear encoded ␣-1 tubulin mRNA (Fig. 3B). Furthermore, UV cross-linking of the 81-, 47-, and 38-kDa proteins to the chloroplast psbA leader is not competed by excess cold tRNA or total RNA from E. coli (Fig. 3A).
Experiments in which the unlabeled 5Ј-UTR of atpB or rbcL was competed against each of the labeled 5Ј-UTRs (data shown in Fig. 6 for only rps12) strongly suggest that the same 81-and 47-kDa proteins present in the particular column fraction analyzed bind to all leaders. However, since 81-and 47-kDa species are frequently found in more than one fraction and the 47-kDa protein clearly migrates as a doublet in several cases (Figs. 4 and 5), we cannot rule out the possibility that several forms of these proteins exist. Indeed experiments in which total unlabeled RNA from the ⌬atpB and ⌬psbA mutants was competed against the labeled atpB and psbA leaders strongly suggest the existence of at least two 47-kDa species, one of which is specific for atpB and the other for psbA (Fig. 7). Two other proteins may also occur in modified forms. The 56-kDa protein present in spr-u-1-27-3 (Ϫspec) extracts which binds to the rps12 leader elutes in a single high salt fraction compared to elution of the same protein over three fractions in other extracts (Fig. 5). The 15-kDa protein which elutes in a low salt fraction in most extracts occurs in a single high salt fraction binding to the rbcL leader in extracts of spr-u-1-27-3 (Ϫspec) (data not shown).
In extracts from cells with reduced chloroplast protein synthesis (ac-20 cr-1 and spr-u-1-27-3 ϩspec), which preferentially translate mRNAs for chloroplast encoded r-proteins (9, 18, 49), a 36-kDa protein present in wild type cells is missing or greatly reduced. This suggests that the 36-kDa protein could be synthesized on chloroplast ribosomes or required for translation of photosynthetic protein mRNAs. The relative enrichment of the 62-kDa protein relative to the 81-and 47-kDa proteins in the extracts from ac-20 cr-1 (Fig. 5) may be related to the permanent deficiency of chloroplast ribosomes in this mutant strain.
The several differences observed in the binding patterns of these proteins between leaders of specific r-protein and photosynthetic protein mRNAs will require additional study before they can be interpreted in terms of the hierarchical model for control of translational regulation in chloroplasts that we postulated recently (1). Based on our current observations, one can postulate that the ubiquitous 81-and 47-kDa proteins (or a subset of their various forms), may serve to mark the chloroplast mRNAs for translation. Association of other proteins with this complex in a gene or class-specific fashion (1), may yield a complex competent for translation initiation.
Identification of Cis-acting Sequences-We have identified cis-acting sequences within the 5Ј-UTRs of atpB and rps7 mRNAs which bind fractions enriched primarily for the 81-and 47-kDa trans-acting factors. In both cases sequences adjacent to the putative SD sequence are protected from RNase T1 digestion. These results are consistent with the binding of a 47-kDa protein to the psbA leader, involving sequences just 5Ј to the SD sequence within a putative stem loop (5,50). In the case of the psbA message, deletion of the SD sequence (GGAG) abolishes psbA translation and D1 accumulation (50). However, not all chloroplast genes contain the canonical GGAG sequence, and of these, the petD (GGA) SD sequence has been shown not to function in translation (51). None of the 5Ј-UTRs we have analyzed show significant sequence homology to one another, but all are (A ϩ U)-rich (70 -84%). The latter suggests that the mRNA in vivo is in dynamic equilibrium as a population of secondary structures.
In conclusion, our results emphasize the importance of defining the 5Ј-UTR-binding proteins present in different cell extracts and their spectrum of interactions with the 5Ј-UTRs of different chloroplast genes. Either cloned gene(s) or antibody probes will be necessary to resolve the conundrum that 46 -47-kDa trans-acting proteins have been reported to be specific for regulating expression of the chloroplast psbA (5-7) or psbC (8) genes, whereas our data suggest that one or more 47-kDa species bind ubiquitously to all chloroplast leaders examined. Only in this way can trans-acting factors that bind generally to chloroplast 5Ј-UTRs be distinguished from the putative gene specific factors that have so far been characterized (1,4).