Target and Specificity of a Nuclear Gene Product That Participates in mRNA 3′-End Formation in ChlamydomonasChloroplasts*

Chloroplast mRNA maturation is catalyzed by nucleus-encoded processing enzymes. We previously described a recessive nuclear mutation (crp3) that affects 3′-end formation of several chloroplast mRNAs in Chlamydomonas reinhardtii(Levy, H., Kindle, K. L., and Stern, D. B. (1997) Plant Cell 9, 825–836). In the crp3 background,atpB mRNA lacking a 3′-inverted repeat normally required for stability accumulates as a discrete transcript. The mutation also affects the atpA gene cluster; polycistronic mRNAs with psbI or cemA 3′-ends accumulate to a lower level in the crp3 background. Here, we demonstrate that the crp3 mutation also alters 3′-end formation of psbI mRNA and cemA-containing mRNAs. A novel 3′-end is formed in monocistronic psbItranscripts, and this is the only terminus observed when thepsbI 3′-untranslated region is fused to an aadAreporter gene. Accumulation of mRNAs with 3′-ends betweencemA and atpH, which is immediately downstream, was reduced. However, this sequence was not recognized as a 3′-end formation element in chimeric genes. The crp3 mutation was able to confer stability to three different atpB3′-stem-loop-disrupting mutations that lack sequence similarity, but are located at a similar distance from the translation termination codon. We propose that the wild-type CRP3 gene product is part of the general 3′ → 5′ processing machinery.

stability, for example, have been shown to interact with the 5Ј-untranslated region (UTR), 1 protecting the mRNA from 5Ј 3 3Ј exonucleolytic degradation (8 -10). In addition, point and deletion mutagenesis has been used to define potential sites of interaction between nuclear translation factors and the 5Ј-UTRs of petD (11), psbD (10), and psbA (12). These studies suggest a complex interplay between nuclear proteins and multiple cis-elements in chloroplast mRNAs.
In contrast to the 5Ј-UTR, whose functions in mRNA stability and translation are relatively well described, the role of the 3Ј-UTR in chloroplast gene expression is enigmatic. Although stem-loop-forming inverted repeat structures are commonly found in the 3Ј-UTR and stabilize RNAs in vitro and in vivo (13)(14)(15), the inverted repeats can be replaced in vivo by a polyguanosine sequence, which also forms a strong secondary structure (16). This suggests that at least for atpB mRNA, specific sequences are not required for formation of a stable 3Ј-end. Furthermore, the 3Ј-UTRs of various chloroplast genes are interchangeable both in tobacco (17) and in Chlamydomonas (18,19), suggesting that gene-specific regulation is not accomplished through elements in this region. Nonetheless, the stabilizing functions of 3Ј-UTRs are often orientation-dependent (18,19), and recent data from Chlamydomonas suggest that 3Ј-end formation may stimulate translation initiation in vivo (20).
To address the function of the 3Ј-UTR in more depth, we have taken a molecular genetic approach. Previously, a series of 3Ј-deletions were engineered downstream of the atpB gene. Chlamydomonas strains with UTRs that lacked the potential to form a stable 3Ј-stem-loop structure grew slowly under photoautotrophic conditions, were sensitive to high intensity light, accumulated a reduced amount of atpB mRNA that was heterogeneous in size, and accumulated a similarly reduced level of the ATPase ␤-subunit, the product of the atpB gene (15). Two types of phenotypic revertants were isolated from a prototypical strain of this series, atpB⌬26, by virtue of their ability to grow rapidly on minimal medium and their tolerance of high light. One class resulted from a dramatic amplification of the mutant atpB gene in the chloroplast, so that heterogeneous and unstable RNA accumulated to a high level and thus increased ATPase accumulation (21). A second type resulted from a mutation in a nuclear gene, which we have termed CRP3 (chloroplast RNA processing). This recessive mutation allows a discrete transcript to accumulate from the 3Ј-deleted copy of atpB, which results in increased accumulation of the ATPase ␤-subunit due to enhanced transcript stability and more efficient translation (20,22). Interestingly, the crp3 mutation also caused changes in other chloroplast transcripts, including accumulation of putative processing intermediates and altered * This work was supported by National Science Foundation Award MCB-9723274. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  stability of transcripts from the atpA gene cluster (22). Thus, CRP3 appears to encode a factor involved in the maturation of several chloroplast transcripts. In this report, we examine the specificity of suppression of RNA instability by crp3 and show that it has a target in the 3Ј-UTR. A possible role for CRP3 in wild-type cells is discussed.

EXPERIMENTAL PROCEDURES
Strains, Culture Conditions, and Genetic Analysis-The strains used in this study are shown in Table I. Cells were grown in TAP medium (23) under constant fluorescent lighting. Genetic crosses were performed using standard techniques (23).
Plasmids and Probes-We have previously described the structure and expression of the atpA gene cluster, which contains the atpA, psbI, cemA, and atpH genes (24). Plasmid pCEMA, which contains most of the cemA (ycf10) coding region, was generated by inserting a 1-kilobase EcoRI-HindIII fragment of the chloroplast EcoRI fragment 22 into pBluescript SK ϩ (Stratagene). This plasmid served as a template for amplifying the entire insert by PCR using T7 and T3 primers, and the PCR fragment was used as a probe in RNA filter hybridizations. The psbI coding region probe was inserted into pBluescript SK ϩ as a 296base pair PCR-amplified fragment using primers psbI1 and psbI3, covering from Ϫ35 relative to the translation initiation codon to ϩ 93 relative to the termination codon. The psbI-cemA intergenic region was amplified with primers psbI2 and psbI4, covering from Ϫ31 to ϩ405 relative to the psbI termination codon. The cemA-atpH intergenic region was amplified with primers cemA1 and cemA2, covering from Ϫ44 to ϩ207 relative to the cemA termination codon. Primers psbI2 and cemA1 had added SphI sites, and primers psbI4 and cemA2 had added XbaI sites. The psbI2-psbI4 and cemA1-cemA2 PCR products were inserted into pBluescript SK ϩ , and the orientation of the inserts was determined by XbaI digestion. Since pBluescript has one XbaI site in its polylinker, in one of the orientations, the PCR fragment could be excised as an XbaI fragment. This fragment was cloned into XbaI-digested pDAAD, which contains the aadA selectable marker downstream of the atpB gene (25). The orientation of the insertions was determined by digesting with SphI, as one site was present in the insert, and a second site was present in pDAAD between the aadA coding region and the rbcL 3Ј-UTR. For 3Ј-end mapping of atpB⌬24 and atpB⌬27 RNAs, the atpB 3Ј-regions from each strain were PCR-amplified using primers 9001 and 8901 (15) and inserted into the EcoRV site of T-tailed pBluescript SK (26).
Chloroplast Transformation-Chlamydomonas chloroplast transformation was carried out as described previously (15). Transformants were selected by their ability to grow on TAP plates containing 100 g/ml spectinomycin. Transformants were colony-purified, and the expected insertions were confirmed by DNA filter hybridizations and appropriate PCR amplifications (data not shown).
Isolation of Nucleic Acids, Filter Hybridizations, and Ribonuclease Protection-For whole-cell nucleic acid preparations, cells were grown in TAP medium, and RNA was isolated as described previously (22). RNA was size-fractionated on 6% formaldehyde and 1.1% agarose gels, transferred to nylon membranes, and hybridized with 32 P-labeled probes as described previously (22). The 3Ј-ends of atpB, psbI, and cemA transcripts were mapped using the plasmids described above and RNase protection as described previously (27). Gel imaging and quantification were performed using a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA).

RESULTS
Aberrant Processing of the psbI 3Ј-UTR in the crp3 Background-The Chlamydomonas atpA gene cluster contains four genes, which are transcribed into eight known mono-and polycistronic mRNAs, as shown in Fig. 1 (diagram) and described previously (24,28,29). The crp3 mutation was previously shown to affect the accumulation of mRNAs that terminate downstream of the psbI and cemA (ycf10) coding regions; the amount of psbI-terminated RNAs was moderately reduced, whereas the amount of cemA-terminated RNAs was strongly reduced (22). To examine the effect of the crp3 mutation on these RNAs in detail, we performed filter hybridizations using total RNA extracted from wild-type, crp3, or atpA⌬3 cells (24); in atpA⌬3, the psbI promoter has been deleted. Fig. 1A shows the results of probing with the psbI coding region. In wild-type cells, six RNAs were detected of which the most abundant was the monocistronic psbI message, transcript 7. In crp3 cells, it was expected that RNAs 2, 3, 6, and 7 might be affected since their 3Ј-ends are downstream of psbI or cemA. Indeed, RNA 2 was virtually undetectable; however, only a small change, if any, was seen in the accumulation of the bands corresponding to RNAs 3 and 6, which are similar in size. Most remarkably, a new species was observed that migrated slightly slower than RNA 7; this is marked as RNA 7* in Fig. 1A. This suggested that monocistronic psbI mRNA might be aberrantly processed in crp3 cells. This is supported by results from atpA⌬3, in which the psbI-proximal promoter has been deleted: neither RNA 7 nor the new transcript was observed. However, RNA 3 accumulated to a high level (it is shorter due to the atp⌬3 deletion). This suggests that crp3 selectively affects transcripts terminating downstream of psbI.
The difference in size between the wild-type and novel monocistronic psbI mRNAs could result from differences at the 5Јand/or 3Ј-end(s). The 5Ј-end of psbI mRNA was mapped by primer extension, using total RNA from wild-type and crp3 cells, as described previously (24), and no differences were observed between wild-type and crp3 cells (data not shown). The 3Ј-end of psbI mRNA was mapped by RNase protection, using a uniformly labeled antisense RNA spanning the intergenic region between psbI and cemA. As shown in Fig. 1B, two major protected bands were obtained using wild-type RNA. The longer fragment (readthrough) represents protection by RNAs 1, 2, 5, and 6, whereas the shorter fragment (wt) represents protection by RNAs 3 and 7. By comparison with a DNA sequencing ladder, the psbI 3Ј-end was mapped to 83 Ϯ 3 nucleotides downstream of the psbI termination codon (data not shown). In crp3, an ϳ50% reduction in the level of transcripts with the wild-type psbI 3Ј-end was observed (wt-crp3). In addition, a longer protected product that represents a novel psbI 3Ј-end accumulated. This product is ϳ100 nucleotides longer than the wild-type product, represents ϳ50% of the psbI 3Ј-ends, and corresponds to RNA 7* in Fig. 1A and possibly to a slightly longer RNA 3.
In an attempt to distinguish between the 3Ј-ends of RNAs 3* and 7*, we took advantage of the fact that in strain atpA⌬3, RNA 3 accumulates, but RNA 7 does not (Fig. 1A). The 3Ј-ends of psbI-containing transcripts were again mapped by RNase protection, using total RNA from atpA⌬3 and atpA⌬3-crp3; the latter strain was constructed by crossing atpA⌬3-mt ϩ to crp3mt Ϫ . As shown in the last two lanes of Fig. 1B, the wild-type 3Ј-ends downstream of psbI were also present in the atpA⌬3 strains, although at a lower level in the crp3 background. The novel psbI 3Ј-end, on the other hand, accumulated to a low, but detectable level in atpA⌬3-crp3 and must represent a modified version of transcript 3. The observation of a novel 3Ј-end partially explains the reduction in the atpA-psbI (RNA 3) level reported previously for crp3 and atpA⌬3-crp3 (atpA⌬3 was referred to as ⌬12 in this earlier paper (22)). However, the level of RNA 3* in the crp3 background appears to be less than expected from the amount of RNA 7*, suggesting an unexpected contribution of the 5Ј-end to the process of 3Ј-formation and/or RNA stability. Modified Processing or Stability of the cemA 3Ј-End-The cemA transcript does not accumulate in a monocistronic form, but only in polycistronic mRNAs transcribed from the atpA and psbI promoters (Fig. 1). Fig. 2 (diagram and A) highlights the four cemA-containing transcripts, which in wild-type cells are atpA-psbI-cemA-atpH (RNA 1), atpA-psbI-cemA (RNA 2), psbI-cemA-atpH (RNA 5), and psbI-cemA (RNA 6). In the crp3 background, a substantial decrease in the levels of RNAs 2 and 6 was observed; these are the two transcripts that terminate between cemA and atpH. When gels were run longer to increase resolution (see Fig. 4A), it was clear that the reduction in RNA 6 abundance was accompanied by an accumulation of a slightly larger species (RNA 6*) that, based on its estimated size, had its 3Ј-end within the atpH coding region. Furthermore, the reduction in RNA 6 was accompanied by an increase in RNA 5, consistent with a precursor-product relationship.
To discern whether 3Ј-end maturation downstream of cemA is modified in crp3 and to detect any residual RNA 2, RNase protection was performed using the uniformly labeled anti-sense RNA probe corresponding to the cemA-atpH intergenic region, as shown in Fig. 2 (diagram). Fig. 2B shows the results of this experiment. Two major species were protected by RNA from wild-type cells, the shorter one representing the 3Ј-ends of RNAs 2 and 6 and the longer one representing RNAs 1 and 5, FIG. 2. Analysis of cemA 3-end formation. A, total RNA from the indicated strains was subjected to gel electrophoresis and filter hybridization as described under "Experimental Procedures," using a cemA coding region probe or atpH as a loading control; atpH is unaffected by the crp3 mutation (24). RNA species are indicated by numbers and refer to the diagram (see legend to Fig. 1). B, RNase protection was performed on the cemA-atpH intergenic region using the uniformly labeled antisense RNA probe marked by a heavy line in the diagram. See the legend to Fig. 1 for additional details. The sizes of protected bands are 262 nucleotides (readthrough) and ϳ235 nucleotides (cemA 3Ј). wt, wild-type.

FIG. 1. Analysis of psbI 3-end formation.
A, total RNA from the indicated strains was subjected to gel electrophoresis and filter hybridization as described under "Experimental Procedures," using as probes psbI and psbA coding regions, the latter as a loading control. RNA species are identified by numbers as indicated in the diagram. RNAs denoted by asterisks (3*, 6*, and 7*) are unique to the crp3 background and are discussed under "Results." The diagram is adapted from Drapier et al. (24), and indicates the extent of the psbI promoter deletion in strain atpA⌬3. Bent arrows represent promoters, and RNA sizes are given in kilobases. B, RNase protection was performed on the psbI-cemA intergenic region using the uniformly labeled antisense RNA probe marked by a heavy line in the diagram and total RNA from the strains indicated at the top of the gel. Lane probe is probe alone; lane tRNA is a reaction containing 10 g of yeast tRNA instead of Chlamydomonas RNA. The protected bands marked readthrough, novel, and wt are discussed under "Results." They result from protection of the RNAs listed in parentheses. The sizes of the protected bands are 424 nucleotides (readthrough), ϳ180 nucleotides (novel); and ϳ80 nucleotides (wt). The mapped wild-type (wt) and novel 3Ј termini are indicated by vertical arrows between psbI and cemA. bp, base pairs. which contain atpH (readthrough). In the mutant background (crp3), the cemA 3Ј-end accumulated to a lower level, which was expected since RNA 6 was clearly reduced in abundance. To examine RNA 2, 3Ј-ends between cemA and atpH were mapped in strain atpA⌬3-crp3, which fails to accumulate RNAs 5 and 6 due to a deletion of the psbI promoter. In atpA⌬3 cells, the amount of probe protected by cemA 3Ј-ends was reduced relative to wild-type cells. This is consistent with our previous observation that whereas RNA 2 increases in abundance in atpA⌬3 cells relative to wild-type cells (Fig. 1A), this increase is not in proportion to the decline in RNA 6 (see Fig. 6 in Ref. 24). In atpA⌬3-crp3 cells, a further reduction in probe protected by cemA 3Ј-ends compared with the wild-type sibling (atpA⌬3) was observed however, the protected product was clearly present. This suggests that although not readily detectable on RNA filter blots, RNA 2 does accumulate to some extent in the crp3 background. However, unlike psbI, no heterogeneity in the size of the protected band was seen, suggesting that any processing that takes place between cemA and psbI occurs at the same site as in wild-type cells.
It should be noted that the size of the probe protected by RNA 8 would be only 31 nucleotides and therefore not visible in this experiment. Furthermore, the novel RNA 6*, which apparently has a 3Ј-end within the atpH coding region, would fully protect the probe and contribute slightly to the band labeled readthrough.
The psbI-cemA intergenic region confers correct or modified 3Ј-end processing to a chimeric mRNA, but the cemA-atpH region does not. To test whether the psbI-cemA and cemA-atpH intergenic regions contain information sufficient for the recognition of mRNA processing or stability functions altered in crp3, we tested the ability of these regions to serve as RNAprocessing sites in the context of a bacterial aadA reporter gene engineered to be expressed in Chlamydomonas chloroplasts (30). Fig. 3 (diagram) shows the configuration of these four constructs, in which the intergenic regions were inserted in the two possible orientations between the aadA coding region and the Chlamydomonas chloroplast rbcL 3Ј-UTR. The promoter and 5Ј-UTR were from the Chlamydomonas chloroplast petD gene. The original petD-aadA-rbcL cassette produces translationally competent mRNAs, which confer spectinomycin and streptomycin resistance to transformed cells (25). We rationalized that if the inserted intergenic sequences were sufficient to promote 3Ј-end formation, a shorter mRNA ending at the processing site would accumulate. Otherwise, a longer mRNA that represents processing at the rbcL 3Ј-end would accumulate. In either case, the mRNA was expected to be stable and to confer antibiotic resistance. Indeed, spectinomycin-resistant transformants were recovered when this construct was introduced into both the wild-type (CRP3) and crp3 mutant strains, and these transformants contained the desired DNA configurations.
Typical RNA accumulation patterns from the transformants and control strains are shown in Fig. 3. The middle panel shows results with an atpB probe. As expected, no discrete transcript accumulated in atpB⌬26, the wild-type recipient strain for the transformation (see "Experimental Procedures"), whereas a low amount of a discrete transcript was detected in the other recipient, atpB⌬26-crp3 (the crp3 mutation stabilizes the normally unstable atpB transcript). In all the transformants, atpB mRNA accumulation was wild-type since the atpB⌬26 deletion had been replaced with wild-type sequences during transformation (see diagram in Fig. 3). Fig. 3 (lower panel) shows results with a cemA probe. This probe provided verification of the nuclear background; lanes with a low level of RNA 2 and a high level of RNA 5 carry the crp3 mutation. Probing with aadA, as shown in the upper panel, revealed the transcripts produced from the chimeric genes. For psbI insertions downstream of the aadA coding region, different results were obtained depending on the orientation of the insertion. In the (ϩ)-orientation and in the wildtype background, a mRNA that corresponds to processing in the psbI 3Ј-UTR accumulated, indicating that the psbI-cemA intergenic region carries sufficient information to direct maturation of the psbI 3Ј-end. In the crp3 background, a slightly larger transcript accumulated. Based on its size, this transcript most likely terminates at the novel psbI 3Ј-end as in RNA 7* (Fig. 1A). However, there was little evidence for accumulation of aadA-psbI transcripts with the wild-type 3Ј-end. These results indicate that for the psbI 3Ј-end, the psbI-cemA intergenic region included in these constructs includes sequences sufficient for processing and stability in the crp3 background.
In transformants containing the psbI 3Ј-UTR in the (Ϫ)orientation, little or no RNA processed at the psbI 3Ј-end could be detected. Mainly the longer mRNA processed at the rbcL 3Ј-end accumulated, and the crp3 mutation had no effect. This indicates that the psbI 3Ј-UTR functions in an orientation-dependent manner, as has been reported previously for other 3Ј-UTRs in Chlamydomonas chloroplasts (18,19). The longer transcripts were also seen when the psbI 3Ј-UTR was in the (ϩ)-orientation, although in a reduced amount. This is consistent with inefficient transcription termination and incomplete processing. Transcription termination at Chlamydomonas  Fig. 1. For the aadA panel, the identities of each major hybridizing species are indicated schematically. Transcripts that contain the rbcL 3Ј-UTR are indicated schematically by two stem-loops. In the case where the first stem-loop is inverted, the psbI or cemA 3Ј-UTR was inserted in the (Ϫ)-orientation. Note that some readthrough transcripts are seen in the psbI ϩ strains. The identities of the lower molecular weight and more weakly aadA-hybridizing species have not been determined.
chloroplast 3Ј-inverted repeats has previously been reported to be inefficient (25). However, previous examples of chimeric constructs with two adjacent inverted repeats resulted in accumulation of only the shorter transcript, suggesting that the 3Ј-psbI processing site may be less efficient than those downstream of atpB (27).
For strains containing the chimeric constructs with the cemA-atpH intergenic region downstream of aadA, results were identical regardless of the orientation or nuclear background. In each case, only the longer mRNA, processed at the rbcL 3Ј-end, could be detected. Therefore, the cemA-atpH intergenic region is not sufficient for correct 3Ј-end maturation in this context. Possibly, additional sequences within the atpH coding region are required to confer processing activity. Indirect evidence for this is that the new RNA 6*, unique to crp3 and discussed above, may terminate within the atpH coding region, suggesting that a determinant recognized by crp3 resides in this region.
The crp3 Mutation Can Suppress Other 3Ј-UTR Deletions in the atpB Gene-Various deletions in the atpB 3Ј-UTR that eliminate part or all of the potential stem-loop structure cause mRNA heterogeneity and instability (15,21). The crp3 mutation was isolated as a suppressor of one of these deletions, atpB⌬26. To address the question of whether crp3 specifically suppresses the atpB⌬26 deletion, e.g. by a sequence-specific mechanism, we crossed atpB⌬26-crp3-mt Ϫ to two mt ϩ strains (atpB⌬24 and atpB⌬27) that have different deletion end points in the atpB 3Ј-UTR. Like atpB⌬26, these deletions remove part of the stem-loop structure; however, the sequence context surrounding the deletion is different. The 5Ј-deletion end points are shown schematically in Fig. 4 (diagram). Because the chloroplast genome is inherited from the mt ϩ parent, all four tetrad progeny would inherit the atpB⌬24 or atpB⌬27 deletion, whereas the mutant and wild-type alleles of CRP3 would segregate 2:2. PCR was used to verify that chloroplast DNA was indeed inherited from the mt ϩ strain, and at least six complete tetrads were obtained from each cross. Fig. 4A shows results from a representative tetrad of the cross atpB⌬27 ϫ atpB⌬26-crp3. A psbA probe (middle panel) served as a loading control, and a cemA probe (lower panel) identified the progeny carrying the crp3 mutation (a and b in this tetrad). An atpB probe (upper panel) showed a clear 2:2 segregation for the accumulation of a discrete atpB transcript, which was present only in the crp3 mutant progeny. Therefore, crp3 suppresses the partial absence of a 3Ј-stem-loop in the atpB⌬27 deletion. The intensity of the atpB transcript in atpB⌬27-crp3 suggests that suppression is even more effective than with atpB⌬26, perhaps because atpB⌬27 retains part of the stabilizing secondary structure. Fig. 4B shows results obtained in a parallel experiment with atpB⌬24. It is clear from the parental strain that atpB⌬24 accumulates a low amount of a wild-type sized atpB transcript (see also Ref. 21; Fig. 4B), although the deletion into the 3Јstem-loop is larger than the one in atpB⌬27. Because these strains were generated by bi-directional Bal31 deletion, we explain this difference by a fortuitous combination of flanking sequences. Accumulating atpB transcripts of representative tetrad progeny from the cross atpB⌬24 ϫ atpB⌬26-crp3 show a 2:2 segregation for an abundant, discrete transcript. As in the case of atpB⌬27, this phenotype is linked to crp3 as shown by the pattern of cemA hybridization (multiple complete tetrads were analyzed for each cross). These results clearly show that in the case of the atpB 3Ј-UTR, crp3 can suppress different deletion mutations and thus does not recognize a sequence specific to atpB⌬26.
The atpB mRNAs Have Different 3Ј Termini in the atpB⌬24, atpB⌬26, and atpB⌬27 Strains-The atpB transcripts in atpB⌬24-crp3, atpB⌬26-crp3, and atpB⌬27-crp3 all had similar mobilities on RNA gel blots. Since each had a different deletion in the atpB 3Ј-UTR, this suggested that 3Ј-end formation was occurring at different sequences, rather than at a common cryptic 3Ј-end maturation site downstream of all the 3Ј-deletion end points. To investigate this, the 3Ј-ends of atpB mRNAs in each of these strains were mapped by RNase protection. The sizes of protected fragments were determined using a DNA sequencing ladder, allowing us to estimate 3Ј-end locations with an error of approximately Ϯ5 nucleotides.
The results in Figs. 5 (A and B) show that two and one major atpB 3Ј termini were found in atpB⌬24-crp3 and atpB⌬27-crp3, respectively. These ends were also detected in the wild-type (CRP3) background, although at a much lower level; we previously reported similar results for atpB⌬26 (22). Thus, the recessive crp3 mutation appears to selectively stabilize minor RNA species also present in the unsuppressed (CRP3) siblings. In addition to the major protected species, numerous minor bands including the fully protected probe (minus vector sequences) were detected. These bands presumably correspond to species that are longer or shorter than the major discrete transcripts and can be seen in the filter hybridizations shown in Fig. 4. The existence of two major protected species for atpB⌬24-crp3 (Fig. 5B) suggests that two discrete transcripts should be seen on filter blots such as the one in Fig. 4B. However, the size difference between the two is only ϳ85 nucleotides, and they were not resolved on that gel.
The mapped 3Ј-ends of wild-type atpB mRNA and those of the deletion mutants in the crp3 background are presented in Fig. 5C. The 3Ј-ends in all of the deletion mutants are clustered in an ϳ195-nucleotide region Ͼ2 kilobases downstream of the atpB translation termination codon in wild-type cells. However, because of the extents of the deletions, the most proximal 3Ј-ends in atpB⌬24, atpB⌬26, and atpB⌬27 are ϳ70, 100, and  atpB⌬27 (B) deletions, total RNA from the indicated strains (a-d are tetrad products from the cross shown above them) was subjected to gel electrophoresis and filter hybridization as described under "Experimental Procedures," using an atpB coding region probe, psbA as a loading control, or cemA to determine crp3 genotype. The crp3 genotype is indicated as w (wild-type, no suppression) or m (mutant, suppresses atpB⌬26). RNA species in the cemA panel are indicated by numbers and refer to the diagram. The approximate locations of the 5Ј termini of the atpB 3Ј-deletions in each strain are indicated at the bottom center. wt, wild-type. 85 nucleotides downstream of the termination codon, respectively, as compared with 92-93 nucleotides in wild-type cells. Each of these ends and the more distal termini found in atpB⌬24 and atpB⌬26 map in very dissimilar sequences, ranging from 100% A ϩ U to 80% G ϩ C. Thus, a consensus sequence that might serve as a target for the crp3 product cannot be derived from these data. One possible explanation for this result is that the 3Ј-processing machinery in the mutant cells may be sterically hindered by an RNA structure near the end of the atpB coding region, which would be present in each strain. Such a structure would be expected to impede the activity of an exoribonuclease whose activity could be altered in crp3; 5Ј 3 3Ј exonucleolytic activity is the final step in atpB 3Ј-end maturation in wild-type cells (27).
The crp3 Mutation Is Non-allelic to Another Mutation That Affects Accumulation of mRNAs from the atpA Gene Cluster-Another Chlamydomonas nuclear mutant has been isolated, in which the accumulation of transcripts from the atpA gene is altered (31). In strains carrying ncc1, the accumulation of monocistronic atpA mRNA (RNA 4) is strongly reduced, although there is no net effect on the level of the atpA gene product, the ␣-subunit of the ATP synthase. Thus, like crp3, ncc1 does not cause a non-photosynthetic phenotype. Although ncc1 and crp3 affect different RNAs of the atpA gene cluster, the two mutations could represent alleles at the same locus or interact in some way. To test this possibility, ncc1-mt Ϫ was crossed with atpB⌬26-crp3-mt ϩ . Tetrad progeny of this cross were scored by RNA filter hybridizations using atpA and atpB probes, as shown in Fig. 6. A reduction in RNA 4 relative to RNA 3 was taken as evidence that the mutant allele of ncc1 was present, whereas a reduction in RNAs 2 and 3 and a discrete atpB transcript indicated that the strain carried crp3 (all progeny carry the atpB 3Ј-UTR deletion, atpB⌬26, contributed by the mt ϩ parent).
The simplest interpretation of the data is that crp3 and ncc1 represent independently segregating loci and that the two phenotypes are additive. The majority of the 10 complete tetrads obtained appeared to be tetratypes, such as the one shown in Fig. 6. Tetrad progeny a and d are phenotypically similar to the ncc1 and crp3 parents, respectively. Product c appears to have wild-type alleles at both loci and resembles atpB⌬26. Product b is most likely the double mutant, which exhibits reduced accumulation of RNAs 2-4, but accumulates a discrete atpB transcript, although RNA 2 is not well resolved in this experiment. These results indicate that at least two nuclear gene products participate in the maturation and/or stability of mRNAs transcribed from the atpA gene cluster.  5. 3 termini of atpB mRNAs in atpB 3-UTR deletion strains carrying the crp3 suppressor mutation. RNase protection assays were performed with uniformly labeled antisense RNAs corresponding to the atpB⌬24 and atpB⌬27 3Ј-UTRs in A and B, respectively. The major protected products are marked by arrows with the approximate distance between the 3Ј-end and termination codon indicated in nucleotides (nt). Lanes probe and tRNA are controls as described in the legend to Fig. 2B. The sizes of protected products were estimated using a DNA sequencing ladder. The band representing complete protection of the non-vector part of the probe is indicated as readthrough. The locations of atpB 3Ј termini in the different strains are indicated in C. Numbering is relative to the atpB translation termination codon, with the first base downstream of the codon being ϩ1. For the deletion mutants, the deletion end points are given numerically and are symbolized by dotted lines, and the schematic depicts the 5Ј terminus of the deletion relative to the stem-loop structure. The 3Ј-ends are marked by diagonal arrows. For the wild type (wt), 3Ј-ends were precisely mapped in a previous study (27) and are underlined within the surrounding sequence context. For the deletion mutants, the 10 nucleotides surrounding the estimated 3Ј termini are shown; the numbers indicate the mapped terminus Ϯ5 nucleotides.
FIG. 6. Independence of the two mutations ncc1 and crp3. Total RNA from the indicated strains was subjected to gel electrophoresis and filter hybridization as described under "Experimental Procedures," using a atpA coding region probe or atpB to determine crp3 genotype. The deduced or known genotypes at ncc1 and crp3 are shown as w (wild-type (wt) allele) or m (mutant allele). RNA species identified by numbers in the atpA panel refer to the diagram in Fig. 1.

DISCUSSION
Specificity of the crp3 Mutation-We have reported here a detailed analysis of a Chlamydomonas nuclear mutation that affects the maturation and/or stability of multiple chloroplast mRNAs. Although crp3 is a recessive and therefore probably a loss-of-function mutation, its primary phenotype is the appearance of new RNAs in the chloroplast. These include a discrete atpB transcript lacking a 3Ј-stem-loop, processing intermediates between the 3Ј-end of petD and trnR (22), as well as new 3Ј-ends for psbI-and cemA-containing mRNAs and the stabilization of additional atpB 3Ј-UTR deletions, as reported here. All of these new RNAs may be produced transiently in wildtype cells; thus, CRP3 might encode or activate a ribonuclease that is responsible for their degradation. Alternatively, CRP3 could encode an RNA-binding protein that is essential for correct 3Ј-processing and/or that regulates RNA stability; candidates for such factors have arisen from in vitro biochemical studies using chloroplast protein extracts (32)(33)(34)(35)(36). In two cases, these chloroplast RNA-binding proteins also have been shown to have ribonuclease activity (32,37), which is yet another possibility for CRP3.
Two features of crp3 distinguish it from most other nuclear mutations that affect Chlamydomonas chloroplast RNA metabolism. First, we have shown that the target of crp3 is the 3Ј-UTR, at least for psbI and most likely for atpB and petD as well (22). Other mRNA stability mutations analyzed in Chlamydomonas act on the 5Ј-UTR (8,9), although ncc1, which we determined is not allelic to crp3, appears to act on the atpA 3Ј-UTR (31). Although we have examined only a small number of mRNAs in the crp3 background, we have never seen any change in mRNA 5Ј termini. A second feature is that crp3 is not gene-specific, unlike other mutations affecting RNA stability in Chlamydomonas (8, 38 -41). It is likely that additional nuclear loci with functions and mutant phenotypes similar to crp3 will be found. Since crp3 was isolated as a suppressor of a lightsensitive growth phenotype caused by unstable RNA, this type of screen may prove useful in the future.
In vascular plants, and in particular in maize and Arabidopsis, nuclear mutants have been studied in which multiple chloroplast RNAs are affected. For example, crp1 in maize affects both RNA processing and translation (3), and the crs mutants generally affect splicing of numerous mRNAs (5). In Arabidopsis, several mutants have been isolated that display highly altered RNA accumulation patterns, which are difficult to explain by a single target for the product of the affected nuclear locus (4, 6, 7). These Arabidopsis mutants appear to be even more pleiotropic than crp3, although it is possible that other alleles of crp3 could have more variable phenotypes. The dissimilarity between the type of mutation most often isolated in Chlamydomonas and that most frequently found in vascular plants may simply be a consequence of the type of genetic screens available or the relatively small number of mutants that have been carefully studied to date, or it could reflect fundamental differences in gene expression strategies. It should be noted that nearly all mature Chlamydomonas chloroplast transcripts are monocistronic, whereas polycistronic transcripts are common in vascular plants. However, this may simply be a consequence of efficient processing of polycistronic transcripts in Chlamydomonas, as discussed further below, rather than a difference in chloroplast gene expression mechanisms.
Accumulation of New mRNA 3Ј Termini-Loss-of-function mutations in some bacterial ribonuclease genes cause phenotypes that may be related to crp3. For example, when transcripts of the dicistronic rpsO-pnp operon of Escherichia coli are examined in RNase III or RNase E temperature-sensitive mutants at the nonpermissive temperature, RNAs with new 5Ј or 3Ј termini accumulate (42). This occurs because the primary transcripts are rapidly processed in the wild-type background. Similar phenomena have been reported for other RNAs, e.g. puf and unc (43,44). Thus, crp3 may be a partial loss-of-function mutant, in which the accumulation of intermediates is a consequence of inefficient processing or degradation by the mutant ribonuclease. This in turn suggests that processing of primary transcripts may be common in Chlamydomonas chloroplasts and indeed may even be more efficient than in vascular plants, where partially processed RNAs accumulate readily (e.g. Refs. [45][46][47]. For example, the Chlamydomonas petD transcript matures rapidly, and its processing pathway can be discerned only from mutant strains (48,49). Furthermore, 3Ј-processing of transcripts such as atpB is efficient both in vivo and in vitro (27).
Definition of crp3 Targets-We attempted to define a molecular target for the crp3 product in two ways. First, we inserted two putative targets into a chimeric reporter gene and tested whether accumulation of the reporter gene transcripts responded to the crp3 genotype. As shown in Fig. 3, a psbI 3Ј-UTR sequence did respond as expected, whereas a cemA 3Ј-UTR sequence did not. This implies that in the case of cemA 3Ј-end formation, RNA structures formed through long-range interactions, rather than a specific primary sequence, may constitute the key elements. A second hint at a crp3 target came from its lack of specificity in suppressing atpB 3Ј-UTR deletions (Figs. 4 and 5). Since each RNA differed at its 3Ј-end but contained the same coding region, one interpretation of the ability of crp3 to allow stable mRNA accumulation is that an element within the atpB coding region determines the atpB 3Ј terminus. This would also account for the fact that each stable RNA in the crp3 background was of approximately the same length.