Light-associated and processing-dependent protein binding to 5' regions of rbcL mRNA in the chloroplasts of a C4 plant.

In amaranth, a C(4) dicotyledonous plant, the plastid rbcL gene (encoding the large subunit of ribulose-1,5-bisphosphate carboxylase) is regulated post-transcriptionally during many developmental processes, including light-mediated development. To identify post-transcriptional regulators of rbcL expression, three types of analyses (polysome heel printing, gel retardation, and UV cross-linking) were utilized. These approaches revealed that multiple proteins interact with 5' regions of rbcL mRNA in light-grown, but not etiolated, amaranth plants. Light-associated binding of a 47-kDa protein (p47), observed by UV cross-linking, was highly specific for the rbcL 5' RNA. Binding of p47 occurred only with RNAs corresponding to mature processed rbcL transcripts (5'-untranslated region (UTR) terminating at -66); transcripts with longer 5'-UTRs did not associate with p47 in vitro. Variations in the length of the rbcL 5'-UTR were found to occur in vivo, and these different 5' termini may prevent or enhance light-associated p47 binding, possibly affecting rbcL expression as well. p47 binding correlates with light-dependent rbcL polysome association of the fully processed transcripts in photosynthetic leaves and cotyledons but not with cell-specific rbcL mRNA accumulation in bundle sheath and mesophyll chloroplasts.

Post-transcriptional processes play a key role in regulating the expression of genes encoding photosynthetic enzymes such as ribulose-1,5-bisphosphate carboxylase (RuBPCase) 1 (1)(2)(3)(4)(5). In amaranth, a C 4 dicotyledonous plant, additional layers of regulation have been incorporated into the control of photosynthetic genes to achieve the specialized expression patterns required for this modified photosynthetic pathway (6). Post-transcriptional alterations in the expression of chloroplasts genes encoding the large subunit (LSU, encoded by rbcL genes) and nuclear genes encoding the small subunit (encoded by RbcS genes) of the amaranth RuBPCase occur in response to numer-ous developmental, metabolic, or environmental cues.
Two types of light-shift experiments show that rapid changes in RuBPCase gene expression are induced in response to changes in illumination, due to regulation occurring at two translational steps. First, when amaranth seedlings grown in complete darkness (etiolated) are transferred to light (light shift), RuBPCase synthesis is rapidly induced at the level of translational initiation (7,8). rbcL mRNAs accumulate in the absence or presence of illumination, but these associate with polysomes and are translated only in the light. The lightmediated initiation of translation is mostly specific for RuBP-Case, since transcripts encoding nonlight-regulated proteins are associated with polysomes regardless of the light conditions. Second, when light-grown seedlings are transferred to complete darkness (dark shift), synthesis of both RuBPCase subunits is rapidly repressed (7,9), even though both transcripts remain in association with polysomes. These findings indicate that during the dark shift, translation of both RuBP-Case subunits is regulated at the elongation step.
Post-transcriptional regulation of rbcL and RbcS gene expression has also been implicated during the initial stages of amaranth C 4 leaf development (10) and in association with changes in photosynthetic activity (11). For the plastid-encoded rbcL gene, post-transcriptional control of differential rbcL mRNA accumulation in isolated bundle sheath (bs) and mesophyll (mp) chloroplasts of mature amaranth leaves has been demonstrated, apparently acting through differential mRNA stability in the two plastid types (12).
There are many examples of plastid genes in plants and algae that are regulated at the levels of translation, control of mRNA stability, or RNA processing (8,9,(13)(14)(15)(16)(17)(18)(19)(20)(21)(22). In some cases, nuclear-encoded trans-acting proteins have been shown to interact specifically with cis-acting regions within the 5Ј-or 3Ј-UTRs of the plastid mRNAs, and these have been implicated in the regulation of one or more of these processes (19,(23)(24)(25)(26)(27)(28)(29)(30)(31)(32). A well characterized example of a plastid-encoded mRNA that is regulated by light at the level of translation is psbA, which encodes the D1 protein of photosystem II. In plastids of tobacco, spinach, and Chlamydomonas, changes in psbA translation have been correlated with binding of a protein complex at the 5Ј-UTR (23,28,29). An inverted repeat sequence within the Chlamydomonas psbA 5Ј-UTR interacts directly with a psbAspecific 47-kDa protein (a plastid form of poly(A)-binding protein). This binding is affected by redox potential and by interactions with an associated 60-kDa protein disulfide isomerase (23,(32)(33)(34)(35).
Little is known about control regions of the plastid-encoded rbcL transcripts or factors that interact with these regions, even though this gene encodes one of the most essential, abundant, and highly regulated proteins of photosynthesis. It has been shown that rbcL 5Ј-UTR can affect both translation rates and the stability of the transcript in plastids of Chlamydomo-nas (36) and tobacco (26,37), causing enhanced translation and more rapid degradation of a chimeric RNA in the light relative to the dark. The 5Ј portion of the coding region also appears to be involved in stabilizing the transcript (36). In Euglena chloroplast extracts, modifications to the rbcL 5Ј-UTR near the start codon, either deletions or the addition of secondary structure, reduce rates of translational initiation (38,39). In plastids of a higher plant, barley, methyl-jasmonate-induced processing of the rbcL 5Ј-UTR appears to play a role in regulating translation of the LSU polypeptide (40).
To identify components required for post-transcriptional rbcL regulation in the chloroplasts of a C 4 plant, we examined and characterized proteins capable of binding to the amaranth rbcL transcripts in response to changes in illumination. Because sequences involved in protein binding and post-transcriptional regulation have been identified within the 5Ј regions of several mRNAs (1-4), we concentrated on RNA/protein interactions occurring at or near the 5Ј-UTR. We show that specific, light-associated protein binding occurs in vivo and in vitro, and this binding correlates with light-mediated changes in LSU synthesis in both leaves and cotyledons. In addition, we demonstrate that binding of a 47-kDa protein is dependent on the length of the 5Ј-UTR, occurring only with RNAs containing 5Ј ends that correspond to mature rbcL transcripts. These results indicate that regulatory regions within or near the rbcL 5Ј-UTR interact with trans-acting binding proteins in a lightdependent manner, and that differential 5Ј mRNA processing in vivo may work in coordination with light activation of protein binding to regulate rbcL synthesis in the chloroplasts of a C 4 plant.

EXPERIMENTAL PROCEDURES
Plant Materials and Growth Conditions --For light-grown plants, seeds of Amaranthus hypochondriacus var. 1023 were germinated and grown in a growth chamber (Conviron, Asheville, NC) at 24°C with 14 h per day of illumination at an approximate intensity of 170 -200 E m Ϫ2 s Ϫ1 , or in a greenhouse bay equipped with supplemental sodium vapor lamps to provide 14 h per day of illumination (these two growth conditions gave identical binding activities). Light-grown (L) soluble chloroplast protein extracts, or total RNA, was prepared from fully expanded leaves of L plants, or for some experiments cotyledons were harvested from L seedlings 6 days after planting. To prepare etioplast (E) protein extracts or RNA, cotyledons were harvested from seedlings germinated and grown in total darkness for 6 days, as previously described (8). Dark-shifted (DS) plastid protein extracts or RNA were prepared from cotyledons of plants that were grown under normal illumination for 6 days and then transferred to darkness for 4 h before harvesting.
Bs and mp protoplasts were prepared from fully expanded green leaves and separated by sucrose density gradient centrifugation as previously described previously (12). Chloroplasts were isolated from the separated protoplasts and purified on Percoll gradients as described (12), and the separated plastids were used to prepare cell type-specific soluble chloroplast extracts.
Polysome Isolation-Polysomes used for heelprinting or for sucrose gradient analysis were isolated from leaves or cotyledons of L, E, or DS plants, as described previously (8,9).
RNA Heelprinting-Protein-protected RNA fragments were generated by modification of the heelprinting procedure of Wolin and Walter (41). Briefly, polysomes were treated with micrococcal nuclease to degrade RNA not protected by protein. The nuclease was inactivated by addition of EGTA, and the reactions were frozen at Ϫ80°C for 15 min. Protein-protected RNA fragments were pelleted through a 0.25 M sucrose cushion for 30 min at 70K in a TLA 100.3 rotor in a Beckman TL 100 ultracentrifuge at 4°C. The pelleted fragments were treated with proteinase K for 30 min at 37°C and then extracted twice with phenol/ chloroform. RNA fragments were precipitated with ethanol, pelleted, dried, and resuspended in RNase-free water.
For mapping the positions of protein-protected RNA fragments, the fragments were first annealed to single-stranded pRbl-1 DNA template, corresponding to the entire mRNA coding region of the amaranth rbcL gene (42), and primer extension was performed. Each annealing reaction contained 75 fmol of rbcL ssDNA, 150 fmol of 32 P-end-labeled primer, and 1 g of protected RNA fragments in hybridization buffer (33 mM Tris acetate, pH 7.7, 67 mM potassium acetate). The reactions were heated to 65°C for 5 min and then slowly cooled to 30°C for ϳ30 min. After annealing, extension reactions were carried out in the same buffer, with 20 mM magnesium acetate, 0.334 mM of each dNTP, 1 mM dithiothreitol, and 3 units of T4 DNA polymerase. After 30 min at 37°C, the extension products were extracted with phenol/chloroform/isoamyl alcohol, precipitated with ethanol, and resuspended in loading buffer (95% formamide, 10 mM EDTA, 0.1% bromphenol blue, 0.1% xylene cyanol). The samples were heated to 70°C for 5 min and fractionated on 8.3 M urea, 6% polyacrylamide gels, together with sequencing reactions of the pRbl-1 clone to map the locations of the protected fragments.
Preparation of Chloroplast-soluble Protein Extracts-Chloroplast extracts for binding assays were prepared from leaves or cotyledons according to published protocols (27,43). Briefly, total soluble proteins were prepared from a chloroplast lysate that had been clarified by centrifugation at 175,000 ϫ g for 3 h. The clarified supernatant was then passed over a DE52-cellulose anion exchange column, and the flow-through was precipitated with ammonium sulfate at 60% saturation. Precipitated protein was pelleted at 80,000 ϫ g in an SW 41 rotor for 30 min. The supernatant was discarded, and the pellet was resuspended in Buffer E containing 20 mM HEPES, pH 7.9, 60 mM KCl, 12.5 mM MgCl 2 , 0.1 mM EDTA, 2 mM dithiothreitol, and 17% glycerol. Soluble proteins in Buffer E were loaded onto a heparin-agarose column (Sigma) equilibrated with Buffer E. The sample was passed over the column twice, eluted with Buffer E, and the flow-through collected. Protein concentrations were determined by using a Bradford protein assay (Bio-Rad).
Generation and Preparation of 32 P-Labeled rbcL 5Ј RNA Transcripts-DNA fragments used for in vitro synthesis of rbcL 5Ј RNAs were polymerase chain reaction-amplified from pRBL1, using primers to specific regions of the UTR (42). Primers at the 5Ј end contained a T7 promoter. RNA corresponding to various regions of the rbcL 5Ј RNA were synthesized in vitro in 20-l reactions containing 1 g of DNA template, 10 mM dithiothreitol, 1 mM ATP, CTP, and GTP, 500 M UTP, 100 Ci of [␣-32 P]UTP in T7 transcription buffer (Epicentre Technologies), and 50 units of T7 RNA polymerase for 1 h at 37°C. After transcription, 1 unit of RNase-free DNase was added to the reaction, and after 15 min of incubation 1 l of 0.5 M aurintricarboxylic acid was added. Labeled transcripts were purified on 5% acrylamide, 7 M urea gels. The labeled RNA bands were cut out of the gels and crushed with a siliconized glass stirring rod. High salt buffer (0.5 M NaCl, 0.1 M Tris, pH 8.0, 20 mM EDTA) was added to the crushed gel slice, and the RNA was allowed to elute for 10 min at room temperature. The eluted samples were then phenol/chloroform/isoamyl alcohol-extracted and precipitated with an equal volume of isopropyl alcohol. The resulting RNA pellets were resuspended in RNase-free water and [ 32 P]UTP incorporation determined by liquid scintillation counting. Larger amounts of unlabeled RNA for use in competition analyses were made in a 10-fold increased reaction volume and with the addition of 10 mM MgCl 2 and 1 mM of all NTPs.
Gel Mobility Shift Assays-Gel mobility shift assays were performed according to the methods of Chen et al. (44), except without the addition of RNase T1. Competition analyses were performed using unlabeled rbcL RNA as self-competitor or heterologous yeast viral RNA (7z-AS, a 130-nt 3Ј-UTR sequence from a yeast double-stranded RNA virus, generously provided by Dr. J. Bruenn, State University of New York, Buffalo) (45). The probe and the unlabeled competitor RNAs were added to the reactions prior to the addition of the protein extracts, or protein was added to the unlabeled competitor and preincubated before addition of the labeled 5Ј RNA. Identical results were obtained from both types of competition reactions. Gels were visualized using a Phosphor-Imager equipped with ImageQuant version 4.2 software (Molecular Dynamics, Sunnyvale, CA).
UV Cross-linking Assay-UV cross-linking experiments were performed according to published methods (28,46) with some minor modifications. In all experiments, unless otherwise noted, 10 g of protein extract was incubated with 15 fmol (ϳ1.5-2.5 ϫ 10 5 total cpm) of in vitro transcribed radiolabeled RNA in a 20-l reaction volume containing 40 mM KCl, 10 mM MgCl 2 , 3 mM dithiothreitol, 0.05 mM EDTA, 8.5% (v/v) glycerol, 10 mM HEPES, pH 7.9, and at least 10 g of tRNA as a nonspecific competitor. Competition experiments with excess unlabeled RNAs were performed as described for gel mobility assays (relative concentrations of the competitors are detailed in the figure legends). After incubation for 10 min at 25°C, open microcentrifuge tubes containing the binding reactions were irradiated for 30 min with UV light (254 nm) using a UV cross-linker (Fisher) set to an energy level of 0.18 J/cm 2 at 25°C. The RNA was digested with 20 units of RNase T1 and 1 g of RNase A per sample at 37°C for 30 -45 min. For SDS-gel elec-trophoresis, 1 volume of 8% SDS, 3 M ␤-mercaptoethanol, 12.5% glycerol, and 0.001% bromphenol blue were added to each sample. The samples were boiled for 5 min, loaded onto 12.5% SDS-polyacrylamide gels, and separated at 30 mA for 3 h at room temperature. Gels were dried and analyzed using a PhosphorImager as described above.
Primer Extension and DNA Sequencing Reactions-A primer corresponding to the first 15 nt of coding region of the rbcL transcript (42) was 5Ј end-labeled with 32 P using polynucleotide kinase, hybridized to total RNA (isolated from tissues of plants grown under various illumination conditions), and extended using SuperScript II (Life Technologies, Inc.). The extension reactions were phenol/chloroform/isoamyl alcohol-extracted and precipitated with ethanol overnight. The reactions were resuspended, boiled, and loaded onto 6% polyacrylamide urea gels for analysis. 32 P-Labeled sequencing reactions were prepared using Sequenase (Amersham Pharmacia Biotech) with pRbl-1 DNA as a template and run in adjacent lanes to determine the size of the extensions. Gels were analyzed using a PhosphorImager.
Analysis of Polysome-associated Transcripts-Polysomes isolated from L seedlings were separated on 10 -50% sucrose gradients according to methods described previously (8,9). Gradient profiles were determined using an ISCO model 185 density gradient fractionator and model UA-5 absorbance/fluorescence detector. RNA was isolated from the gradient fractions (8) and analyzed by primer extension analysis as described above.

RESULTS
In Vivo Protein-protected RNA Fragments on rbcL Transcripts-The amaranth RuBPCase LSU polypeptide is synthesized only in light-grown (L) plants and not in dark-grown (etiolated, E) or dark-shifted (DS) plants. RNA heelprinting analysis was undertaken to determine whether there are corresponding light-associated differences in RNA-protein interactions occurring in vivo within the 5Ј-UTR or other 5Ј regions of the rbcL transcript. Polysome-associated RNAs purified from plants incubated under various light conditions were subjected to nuclease digestion, such that regions of the RNA that are associated with protein are protected. By visualizing the positions of protected RNA fragments, sites of RNA-protein interactions, as they occur in vivo, are revealed. Fig. 1 indicates the locations of several protected RNA fragments produced by the heelprinting reactions. Most of these originated from within the 5Ј-UTR of the rbcL mRNA, between the 5Ј terminus of the mature transcript at Ϫ66 (mapped by primer extension, data not shown) and the initiator AUG. For L plants (lane L), one of the most significant protected sites occurred adjacent to the 5Ј terminus, extending from Ϫ60 to Ϫ66 (lane L, band d). This protected fragment was clearly present in RNAs from L seedlings, but was observed only weakly in DS seedlings (lane DS), and not at all in E seedlings (lane E). There were several additional nuclease-protected fragments from further downstream in the 5Ј-UTR. One of these occurred with L plants at an AU-rich region (lane L, band c) but not with E or DS plants. Another protected fragment mapped to the Shine-Dalgarno sequence (lanes L and DS, band b) and was observed in RNAs from L and DS plants but not with E seedlings.
Several protected RNA fragments originated from within the coding region as well. One of these (*, lane L) occurred only in polysomal RNAs from L plants, and others occurred in RNAs from both L and DS plants (lanes DS and L). As with the 5Ј-UTR fragments, none of these coding region fragments occurred in RNAs isolated from E plants.
For Fig. 1, all of the RNase-protected fragments that were present in the polysomal RNA lanes, but not in the control lane (Fig. 1, lane C), were eliminated by treatment with protease (data not shown). Therefore, the protected RNA fragments originating at sites within the rbcL 5Ј-UTR and within the open reading frame were dependent on RNA-protein interactions.
Identification of Light-associated rbcL 5Ј RNA Binding Protein Activity by Gel Mobility Shift (GMS)-Soluble chloroplast protein extracts from L or E plants were purified by heparinagarose chromatography to enrich for nucleic acid binding proteins and incubated with in vitro synthesized 32 P-labeled rbcL 5Ј RNA (corresponding to the 5Ј end of mature processed rbcL RNA, terminating at Ϫ66, extending downstream through the UTR, and containing an additional 60 nucleotides of protein coding region). Incubation of the labeled 5Ј RNA with L plastid extract resulted in a well defined mobility shift band on nondenaturing polyacrylamide gels (Fig. 2, 5Ј LRP), which was not visible when this RNA was incubated with purified E plastid extracts (data not shown). Formation of the 5Ј LRP in L plastid extracts was protein-mediated, since it did not occur in control reactions with no protein extract, with extracts that had been boiled, or protease-treated (data not shown). Thus the GMS FIG. 1. Light-dependent protein-protected fragments at the 5 end of rbcL mRNA. The positions of RNA-binding proteins on polysome-associated rbcL mRNAs were determined by RNA heelprinting analysis as described under "Experimental Procedures." Arrows indicate protein-protected fragments and the corresponding region in the 5Ј-UTR region. E, 6-day-old etiolated seedlings. DS, 6-day L seedlings transferred to darkness for 4 h. L, 6-day-old light-grown seedlings. C, single-stranded template DNA alone (control). The sequence to the left represents the 5Ј-UTR of rbcL mRNA, from the initiator AUG up to and including the 5Ј-processing site at Ϫ66. The strong band near the top of the gel in all of the lanes represents a stop site that is not dependent on the presence of protein.
assays, like the heelprinting studies, indicated the presence of light-associated rbcL 5Ј RNA binding protein activity. As shown in Fig. 2, competition studies using unlabeled heterologous RNA of similar size to the rbcL 5Ј RNA (7z-AS, a 130-nt RNA from a yeast RNA virus) (45) indicated that formation of the 5Ј LRP complex in L extracts is specific. This complex could be competed with as little as 1ϫ molar excess of unlabeled self-competitor, whereas competition did not occur with up to a 10,000-fold excess of the yeast viral RNA.
Identification of Individual Proteins Binding rbcL 5Ј RNA by UV Cross-linking-When 32 P-labeled 5Ј RNA was incubated with extract from L or E plants and UV cross-linked, there were a number of binding proteins observed (Fig. 3A, 1st lane).
Most notable was a doublet protein band migrating at ϳ47 kDa (p47). When the rbcL 5Ј RNA was incubated with E extracts, no significant protein cross-linking to the RNA was observed (Fig.  3A, 3rd lane), indicating that p47 is a light-associated RNAbinding protein.
p47 binding activity in L extracts was usually observed as a doublet band. This could be due to modified forms of a single protein or two separate proteins of similar size that bind to rbcL 5Ј RNA. For the studies described here, p47 will be referred to as a single protein. The p47 doublet band was not observed in control reactions cross-linked in the absence of any protein extract (data not shown) or with L extracts in the presence of excess self-competitor (Fig. 3A, 2nd lane). The appearance of the light background smear in some competitor lanes shown here and in the following figures is due to incomplete RNase digestion of the radiolabeled RNA, in the presence of excess amounts of unlabeled transcript. The specificity of rbcL 5Ј RNA interactions with p47 in L extracts was addressed by competition assays (Fig. 3B). Self-RNA competed for cross-linking to p47 (as well as some minor bands) at 100ϫ molar excess, and competition was complete by 10,000ϫ. As with the GMS results, 7z-AS RNA did not compete for cross-linking even at very high molar excess (10,000ϫ or greater). In addition, the homopolymers poly(A), poly(U) (not shown), and the 5Ј-UTR of psbA did not compete for binding at even the highest concentrations. These findings indicate that the light-associated binding of p47 is specific to the rbcL 5Ј RNA.
Protein Binding to rbcL 5Ј RNA Is Dependent on the Length of the 5Ј-UTR-In dicots, rbcL mRNA is transcribed from a single highly conserved promoter region located 156 -185 nt upstream of the start codon (47). The primary transcript is constitutively processed such that the 5Ј end is within the range of Ϫ59 to Ϫ69, depending on the plant species. Both longer and shorter forms of rbcL mRNAs have been found in association with polysomes (3,48), and differential processing at the rbcL 5Ј end has been shown to affect LSU translation (40). These observations raise the possibility that differential processing of the rbcL 5Ј-UTR could affect the efficiency of its translation by altering RNA/protein interactions within the UTR.
The effects 5Ј-UTR length on p47 binding were examined by using rbcL 5Ј RNAs containing different regions of the 5Ј-UTR (Fig. 4A). The following RNAs had 5Ј termini occurring at different sites within the 5Ј-UTR and also contained rbcL coding region to ϩ60. These probe RNAs were Ϫ14 (which includes the Shine-Dalgarno sequence), Ϫ66 (corresponding in size and sequence to the processed mature rbcL mRNA), and Ϫ116 and Ϫ155 (composed of 116 and 155 nucleotides downstream of the start AUG codon, respectively). An additional construct, Ϫ155A, was similar to Ϫ155 but lacked the coding region and the Shine-Dalgarno sequence.
As shown in Fig. 4B, the predominant doublet band at 47 kDa was observed only with the rbcL 5Ј RNA terminating at Ϫ66. The longer Ϫ116 and Ϫ155 5Ј-UTR RNAs, both of which extended upstream and included all of the sequences found within the Ϫ66 RNA, showed no cross-linking to p47. The Ϫ14 RNA also did not cross-link to p47 but did show very strong cross-linking to one or more proteins of ϳ23 kDa. A crosslinked 23-kDa band of reduced intensity also occurred in the Ϫ66 lane (which included all of the sequences in the Ϫ14 RNA plus 52 additional upstream nucleotides). It is possible that this same smaller protein interacted with both RNA fragments, since it binds downstream of the p47-binding site. The longer 5Ј-UTR fragments also showed several less intense bands, some of which appeared to migrate at the same position as some of the bands in Ϫ66 and Ϫ14 RNA lanes. Interestingly, the Ϫ155A RNA, which lacked the Shine-Dalgarno and coding sequences, did not show any cross-linking, whereas the longest length UTR (Ϫ155) did cross-link to some proteins.
To determine whether differences in the length of the amaranth rbcL 5Ј-UTR occur in vivo, primer extension analysis was used. A primer corresponding to the 5Ј end of the rbcL coding region was 5Ј end-labeled and used in primer extension analysis with equalized amounts of total RNA isolated from L and E plants. Fig. 3C shows that there were two prominent 5Ј ends observable in both RNA samples as follows: one at Ϫ66, corresponding to the 5Ј end of mature processed rbcL 5Ј transcript, and the other at Ϫ174, corresponding in size to a typical dicot rbcL primary transcript (47). An additional prominent extension product was primarily in RNA isolated from L plants (Fig.  4C), where there was a clearly observable and abundant extension product terminating at Ϫ103. In L RNAs the ratio of transcripts terminating at Ϫ103 to Ϫ66 (determined by Phos-phorImager quantitation) was 1/3, indicating that a significant amount of rbcL mRNA accumulating in vivo contained a 5Ј-

FIG. 4. Light-dependent protein binding to different length rbcL 5-UTR RNAs and in vivo differences in the rbcL 5-UTR.
A, schematic representation of the 5Ј rbcL mRNA fragments used for protein binding analysis. Ϫ14, an RNA fragment that includes 14 nt upstream of the AUG translation start site plus 60 nt of coding region.
Ϫ66, an RNA corresponding in size and sequence to the processed mature rbcL mRNA, with a 66-nt long UTR upstream of the AUG plus 60 nt of coding region. Ϫ116, RNA with 116 nt upstream of the AUG plus 60-nt coding region. Ϫ155, RNA with additional 155 nt upstream of the mature rbcL mRNA-processing site plus 60 nt of coding region.
Ϫ155A, same as above with the Shine-Dalgarno and coding regions deleted. B, UV cross-linking of proteins to the various length 5Ј-UTR probes. C, 5Ј primer extension analysis of rbcL 5Ј-UTRs in vivo. Total mRNA isolated from cotyledons of etiolated (E) or light-grown (L) plants was hybridized with a 32 P-labeled primer corresponding to rbcL 5Ј coding sequence and extended with reverse transcriptase. Extensions were separated on a polyacrylamide urea gel, visualized, and quantified with a PhosphorImager. 2.8 times more total RNA was used for the D reactions than for the L reactions, to approximately equalize amounts of the rbcL template. D, protein binding to rbcL transcripts found in light-grown plants. B and C, 32 P-labeled RNA fragments corresponding to different regions of the 5Ј-UTR of rbcL were incubated with plastid extract in the absence (Ϫ) or presence (ϩ) of unlabeled self-competitor and UV cross-linked as described for Fig. 3.
FIG. 5. The 5-UTR and 5-coding region are required for p47 binding. 32 P-Labeled RNA fragments consisted of different regions of the rbcL 5Ј RNA probe, as diagrammed to the left of the gel panel. Ϫ66, the entire 5Ј RNA region, from Ϫ66 to ϩ60. UTR, only the UTR portion of the rbcL 5Ј RNA from ϩ1 to Ϫ66. CR, only the coding region from ϩ1 to ϩ60. These were incubated with plastid extract and UV cross-linked as described for Fig. 3. UTR that extended beyond the Ϫ66 processing site but was still significantly shorter than the primary transcript. In contrast to L plants, the Ϫ103 rbcL transcripts were greatly reduced in the E plants (Fig. 4C), where only trace amounts of the Ϫ103 product were present (Ϫ103/Ϫ66 ratio of Ͻ1/20). These findings indicate light-associated differences in the rate or mechanism of RNA processing within the 5Ј-UTR of the amaranth rbcL transcript, with a prominent intermediate transcript accumulating in the light.
In Vivo Differences in 5Ј-UTR Ends Correlate with Differences in Binding of p47 in Vitro- Fig. 4D shows that only one of three RNAs corresponding to in vivo rbcL mRNAs, the Ϫ66 transcript, cross-linked to p47. rbcL RNAs corresponding to the in vivo 5Ј-UTRs terminating at more upstream sites (Ϫ81, a very minor extension product in light RNAs, and Ϫ103) showed no p47 cross-linking. However, it is interesting to note that transcript corresponding to the in vivo Ϫ103 UTR, the most prominent intermediate 5Ј end in vivo, did show binding to a different sized protein, of ϳ50 kDa. Fig. 4 indicates that in L plastid extracts, sequences required for p47 binding occur within a 52-nt region of the 5Ј-UTR, between Ϫ14 and the Ϫ66 end of mature processed RNA. Furthermore, when additional 5Ј-UTR sequences were present on the transcript, binding of p47 was eliminated. The occurrence of longer rbcL 5Ј-UTRs in vivo, together with the finding that extra upstream UTR sequences prevent p47 binding in vitro, suggests that complete processing of the primary rbcL transcript to the Ϫ66 position may be necessary for this binding activity.
The 5Ј-UTR and Coding Region Are Required for p47 Binding-To determine whether the 5Ј-UTR alone is sufficient for p47 binding, probe RNAs were prepared that contained only the UTR portion of the rbcL 5Ј RNA (from ϩ1 to Ϫ66) or only the coding region (from ϩ1 to ϩ60). Fig. 5 shows that neither of these regions (UTR or CR, respectively) alone could be crosslinked to the p47 doublet. Only the 5Ј RNA probe (Ϫ66) containing both regions showed this binding activity.

Differences in rbcL 5Ј-UTR Length Do Not Correlate with
Changes in Polysome Association-Since p47 binding correlates with light-mediated activation of rbcL gene expression, and binding is not detected to rbcL mRNAs with 5Ј-UTRs extending beyond Ϫ66, it might be expected that differences in 5Ј-UTR length would be associated with differences in polysome association. Fig. 6A shows a sucrose gradient profile of polysomes obtained from L cotyledons. The distributions of rbcL mRNAs within different regions of the polysome gradient were similar to that reported for L seedlings in previous studies (8,9). Primer extensions from these L polysome-associated rbcL mRNAs (Fig. 6B) indicated that there were no differences in distribution for transcripts terminating at the various 5Ј locations. mRNAs ending at Ϫ66, Ϫ103, and the Ϫ174 primary transcripts all showed similar patterns of distribution along the polysome (fractions 2-5) and monosome (fractions 6 and 7) regions of the sucrose gradient.
In addition to the three most prominent 5Ј rbcL extension products, Fig. 6B shows that there were also some less abundant products on polysomes, notably those terminating at Ϫ131 and Ϫ155. Thus, whereas p47-kDa protein binding activity was observed only with the Ϫ66 mature transcripts in L plastid extracts, mRNAs with longer 5Ј-UTRs that are not capable of binding to this protein in vitro still appear capable of associating with polysomes of L seedlings in vivo.
p47 Binding and Patterns of rbcL Gene Expression-Possible correlations between 5Ј RNA binding activity and previously described post-transcriptional changes in rbcL gene expression were investigated using plastid extracts isolated from plants grown under different conditions of illumination or from separated bs and mp plastids (Fig. 7). UV cross-linking analysis revealed the typical p47 doublet band in extracts prepared from leaves as well as cotyledons of L plants (Fig. 7, light and 6-day  light cots, respectively), both of which function as photosynthetic tissues in amaranth seedlings (49). p47 binding was also apparent in DS plants (Fig. 7, 6-day dark-shift cots) but not in etiolated tissues (Fig. 7, etiolated).
Protein extracts from Percoll-purified chloroplasts that had been isolated from separated bs and mp protoplasts were used to determine whether differential binding of p47 correlates with post-transcriptional differences in rbcL mRNA accumulation occurring in the two plastid types (12). As shown in Fig. 7 (Bundle sheath and Mesophyll lanes), binding patterns for both of the highly purified plastid protein extracts were similar, although intensity of the cross-linked p47 band was approximately 4-fold lower in the mp extracts than in the bs extracts.
Is p47 Similar to Other Known Chloroplast 5Ј RNA-binding Proteins?-Although binding of p47 to amaranth rbcL RNA was not competed with amaranth psbA transcript, this binding activity does share some properties with light-associated proteins that interact with the 5Ј-UTR of psbA mRNA in other plant systems (1)(2)(3)(4). To distinguish the binding activities associated with these two plastid-encoded transcripts, we compared patterns of UV cross-linking to rbcL and psbA 5Ј RNAs in L plastid extracts. As shown in Fig. 8, the protein cross-linking patterns to these two RNAs were distinct. The psbA transcript cross-linked to two major proteins (Fig. 8, psbA lane1), one at ϳ48 kDa (just above the cross-linked proteins observed with rbcL 5Ј RNA) and the other at ϳ41-43 kDa. These are similar in size to two psbA 5Ј-UTR RNA-binding proteins identified in spinach (28,50). A third weaker band could also be seen at ϳ28 -30 kDa. This distinct protein cross-linking pattern was specific to the psbA 5Ј RNA, since the cross-linking could be competed with unlabeled self-competitor (Fig. 8, psbA, lane 2) but not with up to a 10,000ϫ molar excess of rbcL 5Ј RNA (Fig.  8, psbA, lane 3). Based on their different sizes and specificities of binding, it is evident that the major p47 rbcL 5Ј RNA-binding proteins are distinct from those that interact with psbA 5Ј RNA in amaranth chloroplasts. DISCUSSION In this study, three methods were used to identify amaranth rbcL mRNA-binding proteins that associate with rbcL 5Ј RNA only in plastid extracts from plants grown in the light and not in the dark, conditions that produce significant differences in synthesis of the LSU protein (6,7). Light-associated mRNA/protein interactions such as these, occurring at the 5Ј-UTR, are potential candidates for regulators of rbcL translation, processing, or stability in the plastids of this C 4 plant.
Proteins Bound to the 5Ј Regions of Polysome-associated rbcL mRNAs in Vivo-RNA heelprinting analysis revealed several RNA fragments that were protected from RNase digestion by protein association. Most of these were from L plants or from DS plants (when rbcL mRNAs are associated with polysomes (8,9)) and were greatly reduced or absent in mRNAs from E plants (when rbcL mRNAs are not on polysomes (8)). The RNase-protected fragments were isolated by centrifugation through two sucrose cushions, and relatively small complexes (6 S or less) do not pellet under these conditions (51). Thus any fragments identified by this heelprinting assay must be associated with ribosomes or with a large protein complex.
Two predominant fragments from within the 5Ј-UTR were associated only with RNAs from L plants, where the rbcL mRNAs are actively translated (8). One of these corresponded to the processed end of the amaranth rbcL mRNA, and another was from an AU-rich region at Ϫ21. Light-dependent binding to an AU-rich region within the 5Ј-UTR of psbA mRNA has been observed in other systems (23,29). RNase protection at the 5Ј end of processed chloroplast mRNAs, similar to that shown here for rbcL, has been reported for the ATPase synthase gene cluster of spinach (51). It is possible that binding of proteins at or near the 5Ј terminus of some plastid transcripts is required for their translation or stabilization in the light.
In addition to the L-specific fragments, there were two easily detected fragments located near the Shine-Dalgarno sequence that were present in RNAs from L as well as DS plants but were not observed with the E RNAs. Ribosome binding at the Shine-Dalgarno sequence of rbcL mRNA from barley chloroplasts has been observed using toeprinting analysis (52), and the protected fragments at the Shine-Dalgarno region may represent ribosome binding at this site. Other 5Ј RNA fragments detected in our assay occurred only in association with rbcL mRNAs from L plants and are potential candidates for light-dependent regulatory complexes.
Light-associated Protein Binding to the rbcL 5Ј RNA in Vitro-RNA heelprinting experiments show that there are in vivo differences in the occurrence of large protein complexes bound to specific 5Ј-UTR and coding regions of polysome-associated rbcL mRNAs in L and E plants. Similarly, the in vitro binding extracts indicated the presence of light-associated protein binding that is specific to the rbcL 5Ј region. This binding activity was visualized as a slow-migrating shifted band (5Ј LRP) on nondenaturing gels and as a 47-kDa protein doublet (p47) by direct UV cross-linking. Both binding activities occurred only in extracts prepared from L plants and not from E plants. The very high specificity of p47 for rbcL 5Ј RNA and its requirement for RNA corresponding to fully the processed transcript distinguish it from other plastid mRNA-binding proteins that have been found in chloroplasts (19,23,28,31,53,54).
Sequences necessary for p47 binding in L extracts occurred between Ϫ14 (relative to the initiator AUG) and the end of the processed rbcL 5Ј-UTR at Ϫ66. The addition of as few as 14 nucleotides corresponding to sequences upstream of the normal 5Ј-processing site completely eliminated detectable p47 binding in our in vitro assay. Secondary structures within the 5Ј-UTRs of prokaryotic and eukaryotic mRNAs can affect the binding of regulatory factors and affect their translation (16,40,(55)(56)(57), and a predicted secondary structure occurring within the rbcL-66 5Ј-UTR is a potential target for p47 binding. The additional 5Ј sequences could prevent binding of the 47-kDa binding proteins by interfering with normal secondary structure conformations. Alternatively, the extra 5Ј rbcL sequences could play a more active role in preventing binding of the 47-kDa proteins through direct interactions with additional regulatory factors. Several unique cross-linked bands were observed with RNAs containing sequences upstream of Ϫ66 (Fig. 4B). Proteins binding upstream of Ϫ66 could work by modulating structural changes in the p47 target sequence or by directly overlapping and blocking access to the target sequence.
L versus D amaranth seedlings showed differences in the length of the rbcL 5Ј-UTR, and in L plants all of the rbcL transcripts (full-length, intermediate, and mature) were found in association with polysomes that were presumably involved in synthesis of the LSU protein. It is likely that the intermediate length 5Ј-UTR transcripts result from post-transcriptional processing. The single rbcL promoter and 5Ј-leader regions are highly conserved among dicots, and there are no promoter-like sequences located between the primary transcript and the shortest mature transcript (47). The observation that intermediate processed rbcL transcripts are polysomeassociated indicates a relationship between light-associated p47 binding, light-associated transcript processing, and lightdependent rbcL translation in amaranth chloroplast, although the nature of this relationship remains to be determined. Other studies have linked mRNA processing at the 5Ј-UTR with polysome association in chloroplast. In barley, it has been shown that differential 5Ј-UTR processing, induced by the plant growth regulator methyl-jasmonate, regulates translational initiation (40). In Chlamydomonas plastids, ribosome association appears to be necessary for processing of the psbA 5Ј-UTR (58). Unlike the amaranth rbcL 5Ј RNA, where processing to Ϫ66 is necessary for p47 binding, processing of the psbA 5Ј-UTR did not affect binding of a specific mRNA-binding complex.
The 5Ј-UTR and coding region are each necessary, but not in themselves sufficient, for p47 binding to occur. p47 recognition could involve an RNA secondary structure in one region that is formed or stabilized by the presence of the other region. It is also possible that this binding requires recognition and contact with distinct sequences present in both regions. As shown in Fig. 1, protein-protected RNA fragments were derived from coding as well as noncoding regions, indicating that one or more proteins contact rbcL 5Ј RNA at multiple sites. A requirement for both UTR and non-UTR sequences in plastid RNAprotein binding and post-transcriptional gene regulation has been reported in other studies (36,50). p47 binding to rbcL 5Ј RNA occurred in plastid extracts from both L and DS plants and from leaves as well as cotyledons. Thus p47 binding activity was strictly correlated with polysome-association of fully processed rbcL mRNAs and was not specific to a developmental process associated with only one of these photosynthetic organs. Since p47 binding to rbcL 5Ј RNA occurred in both mp and bs plastid extracts, this protein cannot be directly implicated in post-transcriptional regulatory processes that determine C 4 -type cell-specific gene expression patterns characteristic of mature amaranth leaves (12). p47 may have an additional function(s) or interact with additional noncell type-specific mRNAs in this C 4 plastid type. Alternatively, although we have shown that bs/mp specificity is maintained for several plastid mRNAs during the incubation times required for protoplast purification and plastid purification (12), it is possible that lower levels of p47 cross-linking observed in the mp plastid extracts were due to some loss of bs specificity for the (likely nuclear) genes encoding p47 during preparation.
Taken together, our analysis of light-associated p47 binding to Ϫ66 rbcL 5Ј RNA suggests three possible functions for this protein. First, p47 could be responsible for translation-associated processing of rbcL mRNA to the Ϫ66 position. Immediate polysome association of the full-length transcript in L seedlings could be accompanied by cotranslational processing during protein synthesis, so that intermediate length transcripts are present only on polysomes in the light. Second, p47 could be responsible for capping and protecting the polysome-bound rbcL mRNAs from degradation following cleavage at the Ϫ66 site. 5Ј-UTRs of plastid mRNAs have been implicated in control of transcript stability in other systems (26,36,37,50,58). Third, p47 could be one of a small group of regulatory proteins that specifically bind to different length rbcL 5Ј-UTRs that accumulate in vivo (in this case the Ϫ66 mRNA), mobilizing these transcripts to polysomes. If this were the case, then other proteins, such as the 50-kDa protein that binds to the Ϫ103 rbcL RNA, could selectively activate the other rbcL transcripts. These possible functions are not exclusive; p47 could potentially have multiple roles in association with polysomes that are translating the rbcL mRNAs.