Present address: Cereon Genomics, Cambridge, Massachusetts, USA.
5′ to 3′ exoribonucleolytic activity is a normal component of chloroplast mRNA decay pathways
Article first published online: 5 JAN 2002
The Plant Journal
Volume 19, Issue 5, pages 521–531, September
How to Cite
Drager, R. G., Higgs, D. C., Kindle, K. L. and Stern, D. B. (1999), 5′ to 3′ exoribonucleolytic activity is a normal component of chloroplast mRNA decay pathways. The Plant Journal, 19: 521–531. doi: 10.1046/j.1365-313X.1999.00546.x
- Issue published online: 5 JAN 2002
- Article first published online: 5 JAN 2002
- Received 26 May 1999; revised 28 June 1999; accepted 28 June 1999.
Molecular genetic studies have shown that determinants of chloroplast mRNA stability lie in both the 5′ and 3′ untranslated regions. While it is well-known that chloroplast mRNAs are unstable in the absence of certain nucleus-encoded factors, little is known of the decay mechanisms for chloroplast mRNA in wild-type cells. Here we used a poly(G)18 sequence, which impedes both 5′3′ and 3′5′ exoribonucleolytic RNA decay in vivo, to study the degradation pathway of petD mRNA in wild-type and mcd1 mutant chloroplasts of Chlamydomonas; the mcd1 mutant lacks a nucleus-encoded factor required for petD mRNA accumulation. Upon inserting poly(G) at positions –20, +25, +165 or +25/+165 relative to the mature petD 5′ end, mRNAs accumulate with 5′ ends corresponding to the poly(G) sequence, in addition to the normal RNA with its 5′ end at +1. We interpret these results as evidence for continuous degradation of petD mRNA in wild-type cells by a 5′3′ exoribonucleolytic activity. In the case of the –20 insertion, the accumulating RNA can be interpreted as a processing intermediate, suggesting that 5′ end maturation may also involve this activity. When examined in the mcd1 mutant background, petD mRNAs with the poly(G) 5′ ends, but not normal +1 ends, accumulated. However, no expression of SUIV, the petD gene product, was detected. Insertion of poly(G) at +165 in wild-type cells did not demonstrably affect SUIV accumulation, suggesting that ribosomal scanning does not occur upstream of this position. However, since neither poly(G) –20 nor +165 RNA could be translated in mcd1 cells, this raises the possibility that the MCD1 product is essential for translation.
Chloroplast mRNAs are dependent upon nucleus-encoded factors for their accumulation and have widely divergent half-lives, which can vary in response to environmental or developmental signals. Using transcription inhibition by tagetitoxin in barley ( Kim et al. 1993 ), actinomycin D inhibition in spinach ( Klaff & Gruissem 1991) or pulse–chase experiments in Chlamydomonas ( Hwang et al. 1996 ; Salvador et al. 1993 ), half-lives between 0.5 h and several days have been reported. To understand how chloroplast RNA stability is regulated requires a fuller knowledge of the relevant cis elements and trans factors, as well as RNA degradation pathways (reviewed in Rochaix 1996; Sugita & Sugiura 1996).
The first chloroplast RNA stability determinants were found in the 3′ untranslated region (UTR), which generally contains inverted repeats (IRs) having the potential to form stem–loop structures. Deletion of the IR leads to RNA instability in vitro ( Stern & Gruissem 1987) and in vivo ( Lee et al. 1996 ; Stern et al. 1991 ), presumably by failing to block processive 3′5′ exoribonucleases ( Drager et al. 1996 ). In Chlamydomonas, a nuclear suppressor of such a deletion was found ( Levy et al. 1997 ), suggesting that stabilization by the 3′ UTR includes the participation of protein factors, an hypothesis also supported by in vitro binding of chloroplast proteins to these sequences ( Memon et al. 1996 ; Yang et al. 1996 ).
Genetic studies in Chlamydomonas led to the identification of the 5′ UTR as another critical RNA stability determinant. This was demonstrated using a series of recessive nuclear mutants in which a specific chloroplast mRNA was unstable, and for mutants affecting petD and psbD, by showing that chimeric genes containing only the 5′ UTR of the affected mRNA responded to the nuclear genotype identically to the endogenous mRNA ( Drager et al. 1998 ; Nickelsen et al. 1994 ). Unlike the 3′ IR deletions, which caused accumulation of a reduced amount of heterodisperse mRNA ( Stern et al. 1991 ), the nuclear instability mutants result in a complete absence of transcript accumulation ( Drager et al. 1998 ; Drapier et al. 1992 ; Gumpel et al. 1995 ; Monod et al. 1992 ; Nickelsen et al. 1994 ). This suggests an absolute requirement for these nuclear factors as well as a highly efficient RNA degradation system recognizing, at least in two cases, an unprotected 5′ UTR.
Detailed analysis of the recessive mcd1-1 mutant, which fails to accumulate petD mRNA and its product, subunit IV (SUIV) of the cytochrome b6/f complex, revealed that, in mutant cells, petD mRNA is degraded by a 5′3′ exoribonucleolytic activity ( Drager et al. 1998 ). This could be shown by introducing a poly(G)18 sequence into the 5′ UTR of a chimeric petD–uidA gene (uidA is an E. coli gene encoding β-glucuronidase); this sequence forms a strong tertiary structure and blocked petD–uidA mRNA degradation in the mcd1 nuclear background. The stabilized mRNA was found to have poly(G) at its 5′ end, suggesting that the poly(G) blocked a processive exoribonuclease activity. We inferred that the product of the wild-type (WT) MCD1 gene served to protect the petD 5′ end from this activity. Similar results have recently been obtained for another Chlamydomonas mRNA, psbD, in the destabilizing nac2 nuclear background ( Nickelsen et al. 1999 ). Thus, it may be a common phenomenon that Chlamydomonas nucleus-encoded factors bind to chloroplast mRNA 5′ UTRs to protect them from exonucleases.
Several questions remained unanswered in our earlier studies of mcd1 that relate to fundamental questions of chloroplast mRNA metabolism. First, is the 5′3′ exoribonucleolytic degradation of petD mRNA in mcd1 a peculiarity of the mutant background, or is this a normal mRNA decay pathway in WT cells? Second, we could not exclude the possibility that the poly(G) tract served as a recognition site for an unknown endonuclease which cleaved at its immediate 5′ end, leading to the erroneous conclusion that an exonuclease activity was responsible for generating 5′ ends upstream of poly(G) tracts. Third, because poly(G) forms a complex tertiary structure ( Sundquist 1993), it might be possible to obtain insight into whether ribosome scanning is involved or required for petD translation initiation, by inserting poly(G) in different positions.
In the present work, we present evidence showing that 5′3′ exoribonucleolytic degradation is indeed a normal RNA decay pathway in Chlamydomonas cells, at least for petD. Furthermore, we conclude that poly(G) is not a recognition site for an endonuclease activity. Finally, we show that poly(G) insertions have position-dependent effects on translation, suggesting that petD mRNA has a complex translation initiation pathway, including a possible dependence on the MCD1 gene product.
Insertion of poly(G) cassettes into the petD 5′ UTR
The Chlamydomonas reinhardtii petD gene lies downstream of petA, which encodes cytochrome f, and can be transcribed either from the petA promoter or a petD-proximal promoter, which we have estimated to lie approximately 30 bp upstream of the mature 5′ end ( Sakamoto et al. 1994b ; Sturm et al. 1994 ). Mature petD mRNA has a 362 nt 5′ UTR, which contains three elements important for its function ( Sakamoto et al. 1994a ). Element I lies between nt 2 and 8 (the mature 5′ end is at nt 1), and is required for RNA stability, presumably through its interaction with the nucleus-encoded MCD1 protein (Higgs et al. unpublished data). Mutations in elements II (nt 197–212) and III (nt 317–330) cause defects in translation, but have no effect on RNA stability. We have proposed that transcription initiation at either the petA or petD promoter is followed by endonucleolytic cleavage, or a combination of endonucleolytic cleavage followed by 5′3′ exoribonucleolytic trimming, to generate the mature petD 5′ end.
To test the maturation and decay pathways of petD mRNA, poly(G)18 was inserted in several positions in the petD gene, as shown in Fig. 1(a). An insertion at position –20, relative to the mature 5′ end, was made to test whether exoribonucleolytic trimming of a primary transcript was required for mRNA maturation. If this hypothesis were correct, transcripts initiated at approximately –30 would be blocked from maturation in this strain. An insertion at +25 was made to test whether decay of petD mRNA could be blocked in mcd1 mutant cells, as it was with a chimeric petD–uidA reporter gene with a poly(G) insertion at +25 of the petD 5′ UTR. An insertion in a neutral (i.e. not in element I, II or III) site, +165, was made in order to verify the results obtained with the +25 insertion, and to provide a simple RNA filter hybridization assay for the presence of a putative decay intermediate with its 5′ end at poly(G) +165. Finally, a strain with insertions at both +25 and +165 was made, to test whether the poly(G) tract might in fact be recognized by an endonuclease, rather than an exonuclease as we had previously surmised.
To modify the petD gene of WT cells, the altered genes shown in Fig. 1(b) were inserted into a transformation vector containing a downstream selectable marker cassette ( Fig. 1a). The aadA selectable marker confers resistance to spectinomycin and streptomycin ( Goldschmidt-Clermont 1991). WT Chlamydomonas cells were transformed with each construct, and selection was carried out for antibiotic resistance. Because chloroplast transformation occurs via homologous recombination, the resident petD gene was replaced by a modified copy. The desired alteration of petD was monitored by PCR, and heteroplasmic transformants were streaked for single colonies. Finally, homoplasmic strains were isolated and verified by DNA filter hybridizations (data not shown).
The ability of each strain to carry out photosynthesis was tested by measuring chlorophyll fluorescence ( Bennoun & Béal 1997) and growth on minimal medium lacking a reduced carbon source. As indicated by (+) symbols at the right side of Fig. 1(b), strains pG–20 and pG165, as well as the WT control, displayed WT fluorescence induction kinetics and photosynthetic growth phenotypes, suggesting that the petD gene was expressed. In contrast, strains pG25 and pG25/165 displayed constitutively high chlorophyll fluorescence and an inability to grow on minimal medium, consistent with a block in photosynthetic electron transport. These preliminary results suggested that insertion of poly(G) at position +25, but not at positions –20 or +165, impeded petD mRNA and/or SUIV accumulation.
RNA and protein accumulation in poly(G) insertion strains
To examine directly the effects of inserted poly(G) sequences on petD mRNA and SUIV accumulation, RNA filter and immunoblot analyses were carried out, as shown in Fig. 2. Figure 2(a) shows accumulation of petD mRNA and that of a control RNA, from the chloroplast psbA gene. The WT strain accumulates a single 0.9 kb petD transcript, as previously reported ( Chen et al. 1993 ), and this amount was set as 100% relative to psbA. A negative control was RNA from the strain F16, which carries the nuclear mutant allele mcd1-1 ( Drager et al. 1998 ) causing petD mRNA instability. Like WT, strains pG–20 and pG25 appeared to show a single hybridizing band, although, as we show below, both accumulate two RNA species not resolvable in this gel. Strain pG–20 reproducibly accumulated slightly more petD mRNA than WT cells.
The two strains having poly(G) insertions at +165 accumulated a second transcript resolvable by agarose gel electrophoresis. In the case of strain pG165, the faster-migrating species constituted approximately one-third of the amount of the full-length band. The size of this second transcript was 0.7 kb, approximately the size expected if its 5′ end were at the +165 position. In pG25/165, a second transcript of 0.7 kb was also seen, although the amount was much less than in pG165, constituting 10–15% of the amount of the full-length petD mRNA. The accumulation of a significant amount of a putative +165 transcript is consistent with poly(G) at +165 blocking its degradation. However, if poly(G) were the site of an endonuclease cleavage, as opposed to an exonuclease digestion, we would have expected to see equal amounts of this RNA in strains pG165 and pG25/165. Because the amount of the 0.7 kb transcript in pG25/165 was greatly less than the amount seen in pG165, we concluded tentatively that petD mRNA can be degraded by a 5′3′ exonuclease activity in MCD1 (WT nucleus) cells.
Figure 2(b) shows the accumulation of SUIV relative to a control protein, the β-subunit of the F1F0 ATP synthase, as measured by immunoblot analysis. WT proteins were loaded in a dilution series to aid in quantification. As expected, F16 cells failed to accumulate SUIV, since they also do not accumulate petD mRNA. In contrast, strains pG–20 and pG165 both accumulated WT or even higher levels of SUIV, in general agreement with the RNA accumulation shown in Fig. 2(a). These results clearly indicate that the poly(G) insertion at +165 does not impede translation of SUIV. In contrast, SUIV accumulation in strains pG25 and pG25/165 was approximately 10% of the level in WT cells. While petD mRNA accumulation was reduced to 40–70% of the WT level in these strains, this cannot fully account for the SUIV level. Thus, we conclude that the +25 insertion reduces the translational efficiency of petD mRNA. Based on our previous results ( Chen et al. 1993 ), 10–20% of SUIV accumulation is sufficient to promote photosynthetic growth. Our estimates for pG25 and pG25/165, taken together with the phenotypic analysis diagrammed in Fig. 1, suggest that they are immediately below the threshold required for sustained growth on minimal medium. Interestingly, when the petD 5′ UTR with poly(G) at position +25 was fused to the E. coli uidA gene, no β-glucuronidase activity was detected ( Drager et al. 1998 ). This suggests that the translational efficiency of SUIV may be affected by sequences outside of the 5′ UTR, although we cannot exclude a role for differential protein stability between SUIV and β-glucuronidase.
Accumulation of petD-poly(G) mRNAs in the mcd1-1 background
The mcd1-1 mutation (and several other alleles at this locus) completely destabilize petD mRNA, and based on petD–uidA fusions we have proposed that the WT MCD1 product protects petD mRNA from exonucleolytic attack ( Drager et al. 1998 ). Since we had not tested this theory on intact petD mRNA, the poly(G)-containing strains offered an opportunity to do so. Therefore, the poly(G) insertion strains shown in Fig. 1 were crossed as the mt+ parent to mcd1-1 mt–. All tetrad progeny were expected to inherit the poly(G) insertion, as cpDNA is derived from the mt+ parent. However, the nuclear mcd1-1 allele is expected to segregate in a Mendelian fashion, i.e. 2:2 in tetrad progeny from a single zygote.
Figure 3 shows RNA filter hybridization results from representative progeny from crosses between mcd1-1 and the various poly(G) insertion strains. The results in Fig. 3(a), for strains pG25 and pG25/165, show that at the level of filter hybridizations, the progeny are indistinguishable (no complete tetrads were obtained for pG25 × mcd1-1). Although a petD transcript ending at +1 (which should accumulate only in MCD1 progeny) and one ending at pG+25 (which should accumulate in all progeny) would not be resolved in this gel system, the results suggest that as for the chimeric petD–uidA mRNA, poly(G) can protect petD mRNA in the mcd1 mutant background, since the presumed mutant progeny accumulate a near-WT level of at least some petD transcripts. In addition, the putative +165 transcript was unaffected by the nuclear genotype. This again suggests that RNAs terminated by a poly(G) tract do not depend on MCD1 for their stability.
Figure 3(b) shows results for pG165 × mcd1-1, and it is clear that in progeny 13 and 14, the full-length transcript (5′ end at +1) fails to accumulate. However, the +165 transcript was unaffected. The lack of full-length transcript was expected in the mutant progeny, since it would be unprotected by the WT MCD1 gene product. The accumulation of the +165 transcript suggests that, as for pG25, the pG+165 insertion can block a 5′3 exonuclease activity. Interestingly, the total amount of petD mRNA is substantially lower in the mcd1 mutant progeny. One interpretation is that the MCD1 protein is more efficient at protecting petD mRNA than is the poly(G) sequence, and/or that SUIV translation has a positive effect, in this context, on petD mRNA stability (see below).
Figure 3(c) shows the results of a cross between poly(G)–20 and mcd1-1. As shown in Fig. 2(a), the pG–20 insertion results in higher petD mRNA accumulation in a WT nuclear background; this can also be seen by comparing the first and third lanes of Fig. 3(c). Among the four tetrad progeny, two (11 and 13) accumulated a higher-than-WT amount of petD mRNA, relative to psbA, while two accumulated 10–20% of the WT level (12 and 14). In addition, the residual transcripts in progeny 12 and 14 migrated slightly more slowly than the +1 petD transcript, suggesting that they might have 5′ ends upstream of +1, perhaps at the poly(G) insertion site. This was confirmed by primer extension analysis as shown below, and indicates that these RNAs may be processing intermediates of petD mRNA trapped by the poly(G) insertion.
Accumulation of SUIV in petD-poly(G) × mcd1-1 backgrounds
The mcd1-1 strain accumulates neither detectable petD mRNA nor SUIV, as shown above. However, in each case where poly(G) was inserted into the petD 5′ UTR or immediately upstream, RNA accumulated in the mcd1-1 background, often to the same level as in WT progeny from the same tetrad ( Fig. 3). This raised the question of whether any of these RNAs might be translatable in the mcd1-1 background; pG–20 and pG165 were known to accumulate high amounts of SUIV in the WT background ( Fig. 2b). Although each complete tetrad for crosses involving pG–20 and pG165 segregated 2:2 for high chlorophyll fluorescence (data not shown), suggesting that photosynthetic electron transport was blocked in the mcd1 progeny, it was possible that some SUIV was accumulating, as it did in the non-photosynthetic pG25 and pG25/165 strains ( Fig. 2b).
Immunoblot analysis was performed to determine SUIV levels in tetrad progeny, as shown in Fig. 4 for pG–20 and pG165. Strain pG–20 was selected because the accumulating petD RNA in the mcd1-1 background contains the entire 5′ UTR, and therefore all known cis elements required for SUIV translation. Strain pG165 was selected because, in a WT nuclear background, SUIV accumulated in excess of the WT level, indirectly suggesting that the +165 transcript might be translatable. Figure 4(a) shows, however, that no trace of SUIV accumulated in the mcd1 progeny (12 and 14) of the cross pG–20 × mcd1-1, although a dilution series of WT protein revealed that as little as 1% should be detectable. Similarly, Fig. 4(b) shows that the mcd1 progeny of pG165 × mcd1-1 (21 and 22; a different tetrad than that used for the RNA analysis shown in Fig. 3b) failed to accumulate SUIV. These results argue that the MCD1 product is essential for translation, although we cannot exclude that RNA misfolding for pG–20, or missing cis elements in pG165, are actually responsible, since we do not know whether these RNA species are actively translated in the WT background.
5′ end mapping of petD mRNAs in poly(G) strains
The RNA filter hybridization results shown in Figs 2 and 3 allowed us to draw certain tentative conclusions about the structures of petD-poly(G) RNAs in WT and mcd1 mutant backgrounds. To map the mRNA 5′ termini accurately, we carried out primer extension or RNAse protection analyses, as shown in Figs 5 and 6. For pG–20, primer extension was used, as shown at the bottom of Fig. 5. The major band detected in each sample, with the exception of the yeast tRNA control and the mcd1 mutant strains (12 and 14), corresponds to the normal +1 5′ end of petD mRNA. However, to construct the pG–20 strain, a BglII linker was inserted at position +25 (Experimental procedures), causing this extension product to be 6 nt longer than that in untransformed WT cells (compare lanes WT and LS–20Bg). In each strain containing a pG–20 insertion, an additional extension product was seen, which mapped to position –32. This position corresponds to the 3′ end of the poly(G) insertion. A faint product is also visible at position –50 in progeny 12 and 14; this corresponds to the 5′ end of the poly(G) tract. Because the poly(G) tract inhibits the progress of reverse transcriptase (data not shown; see Nickelsen et al. 1999 ), it is likely that the RNAs giving a –32 extension product actually extend through the 5′ end of the poly(G) sequence. In any case, these longer RNAs probably correspond to the accumulating transcript in pG–20/mcd1-1 seen in Fig. 3(c). This experiment does not allow us to conclude where these transcripts are initiated, the two most likely possibilities being at the slightly upstream petD promoter, or at the petA promoter, with maturation of the petA mRNA 3′ end leading to a putative intermediate. The accumulation of a high level of the normal +1 transcript in the WT progeny suggests, as we have proposed previously ( Sakamoto et al. 1994b ), that the petD 5′ end can be created directly by an endonucleolytic event, or by an endonucleolytic cleavage between the promoter and the mature 5′ end (but downstream of –20), followed by 5′3′ exonucleolytic processing, which would be arrested in WT cells by the MCD1 product.
The 5′ ends of petD transcripts in pG165 and pG25/165 were mapped using protection by a 32P-uniformly labelled antisense RNA, with the strategy shown at the bottom of Fig. 6. Figure 6(a) shows results for tetrad progeny of the cross pG165 × mcd1-1. A major protected species labelled ‘FL’ is seen from RNA of the parental pG165 strain, as well as in two progeny (11 and 12; see Fig. 3b) of the cross. This species represents RNAs terminating at the normal petD +1 position, and thus 11 and 12 possess the WT MCD1 allele. In contrast, the mcd1-1 progeny 13 and 14 fail to accumulate the full-length species, but do accumulate a smaller species labelled ‘pG165.’ Based on its size and the RNA filter hybridization data shown above, this corresponds to a transcript with a 5′ end at the proximal end of the poly(G) tract. An identical pG165 product was protected from RNA of the MCD1 progeny, again corresponding in size and relative abundance to the smaller transcript in Fig. 3(b).
A similar experiment is shown in Fig. 6(b) for progeny of the cross pG25/165 × mcd1-1, except that one of the mutant progeny failed to grow. In this case, RNAs of the MCD1 progeny protect three species, with FL corresponding to the petD +1 position, pG25 corresponding to transcripts beginning at the poly(G) +25 position, and a ladder of bands corresponding to the poly(G) +165 position. As expected, the +1 protection product was absent from the mcd1-1 progeny; however, the RNAs terminating in poly(G) were as abundant as in the WT progeny. The results for the +25 position are identical to those obtained previously for a petD–uidA fusion with poly(G) at +25 ( Drager et al. 1998 ). In this experiment, however, the protection of transcripts around the poly(G) +165 position was revealed as a ladder of bands, unlike the results in Fig. 6(a) for pG165. The reason for this is unknown, as we cannot differentiate between an RNAse protection artefact and a genuine heterogeneity of RNA ends in these strains. Taken together, these results clearly illustrate the ability of poly(G) sequences to trap petD degradation intermediates in either WT cells or in the mcd1 mutant nuclear background.
The data presented here strongly support the notion that 5′3′ exoribonucleolytic decay is a major RNA degradation pathway in Chlamydomonas chloroplasts. Previously, the only evidence for this came from the ability of poly(G) sequences to block RNA degradation in certain nuclear mutant backgrounds ( Drager et al. 1998 ; Nickelsen et al. 1999 ). The existence of this pathway in the chloroplast, an endosymbiont of prokaryotic ancestry, contrasts with the apparent absence of 5′3′ exoribonucleases in present-day prokaryotes (reviewed in Belasco 1993). Thus, this function has probably been acquired from the host nucleus, as nucleus-encoded mRNAs, at least in yeast, are subject to 5′3′ decay (reviewed in Beelman & Parker 1995).
Enzymes that carry out 5′3′ exoribonucleolytic decay have been best characterized in the yeast Saccharomyces cerevisiae. Recent results have shown that Rat1p and Xrn1p are interchangeable 5′3′ exoribonucleases that differ only in their intracellular localization, with Rat1p acting in the nucleus and Xrn1p in the cytoplasm ( Johnson 1997). Typically, decay is initiated following deadenylation and decapping of the transcript ( Muhlrad et al. 1994 ). Homologues are also known in rat ( Shobuike et al. 1995 ) and humans (GenBank accession number AI536027). Xrn1p and Rat1p degrade most efficiently RNA substrates with a 5′ monophosphate ( Stevens & Poole 1995), consistent with decapping preceding mRNA decay. Chloroplast mRNAs are not capped in vivo, but in vascular plants numerous chloroplast transcripts have 5′ ends that can be capped in vitro, suggesting that they possess 5′ triphosphates and would be poor substrates for an Xrn1p/Rat1p-type activity. Interestingly, however, no successful in vitro capping of Chlamydomonas chloroplast transcripts has been reported, and indeed, 5′ processing appears to be an important facet of mRNA maturation ( Nickelsen et al. 1994 ; Sakamoto et al. 1994b ) and may generate monophosphate termini.
Evidence for 5′3′ exoribonucleolytic decay of nucleus-encoded transcripts in photosynthetic organisms is less direct than in yeast. For example, RNA filter hybridizations showed that apparent decay intermediates hybridized with probes from the 3′ part, but not the 5′ part, of oat phyA mRNA ( Higgs & Colbert 1994). In Chlamydomonas, it was shown that when a poly(G) sequence was inserted following the stop codon of the α1-tubulin mRNA, an apparent decay intermediate accumulated with poly(G) at its 5′ end; however, the upstream sequences could not be detected ( Gera & Baker 1998). These results, while not definitive, argue that 5′3′ exoribonucleolytic decay of nucleus-encoded transcripts is widespread in eukaryotes.
Here we have used a poly(G) sequence to trap RNA decay intermediates, a strategy first used in yeast ( Vreken & Raué 1992). While the accumulation of RNA fragments with 5′ poly(G) sequences, as shown here, is consistent with processive exonucleolytic digestion arrested by the poly(G), and 3′-poly(G) fragments can be interpreted as 3′5′ exonuclease products ( Drager et al. 1996 ), other possibilities must also be considered. One which we have addressed is that poly(G) might serve as a specific endonuclease cleavage site. Upon placing two poly(G) sequences in the same transcript (strain pG25/165), we showed that +25, rather than +165, was the most abundant 5′ end ( Fig. 6b). If endonuclease activity had occurred, then +165 would be expected to be more abundant than +25. Similar results were obtained upon engineering the Chlamydomonasα1-tubulin mRNA ( Gera & Baker 1998) or the yeast MFA2 transcript ( Muhlrad & Parker 1994), although in these cases very little, if any, of the RNA had a 5′ terminus at the second or subsequent poly(G) tract. This may suggest that in strain pG25/165, there is incomplete blockage of the putative exoribonuclease at the +25 position, or that endonuclease cleavage between +25 and +165 yields substrates resulting in pG165 products.
A second issue is whether the transcripts we have observed result from bona fide 5′3′ exoribonucleolytic activity, or from a series of endonucleolytic cleavages yielding a net 5′3′ directionality to decay. In E. coli, for example, it has been proposed that a wave of such cleavages follows the last translating ribosome in the case of the lac operon ( Cannistraro et al. 1986 ). RNAse E is a candidate for such an activity, since it has been reported to be a free 5′-end-dependent enzyme which interestingly, like Xrn1p and Rat1p, has a much higher activity on monophosphorylated transcripts than primary transcripts ( Mackie 1998). Although chloroplasts have been reported to contain an RNAse E-like activity ( Hayes et al. 1996 ), we cannot invoke a translational role in 5′3′ decay based on our data. This is because, first, petD mRNAs containing poly(G) in the +25 position are not efficiently translated ( Fig. 2) yet their degradation intermediates accumulate much like those from the pG165 transcript, which is translated, and second, because although none of the poly(G)-containing RNAs are translated in the mcd1 mutant background, this has no effect on the accumulation of 5′-poly(G) molecules. However, this does not rule out a role for an RNAse E homologue in petD mRNA decay.
Our data also raise the possibility of another role for this activity in chloroplasts, namely the 5′ maturation of petD mRNA. It can be hypothesized that the first step in petD mRNA processing is an endonucleolytic cleavage downstream of position –20, followed by exonucleolytic processing to generate the mature 5′ end at position +1. This processive activity would be arrested in WT cells by the MCD1 gene product, whereas in its absence the entire RNA is degraded. If endonucleolytic cleavage normally occurred upstream of position –20, we would have predicted that in strain pG–20, no petD mRNA with the mature +1 end would accumulate, and this was not the case. Thus, 5′3′ exoribonuclease might have multiple functions in chloroplasts.
Finally, our data address, at least indirectly, the interesting issue of whether the product of the nuclear MCD1 gene has an essential role in SUIV translation. Because Chlamydomonas mutants such as mcd1 and nac2, which prevents psbD mRNA accumulation ( Kuchka et al. 1989 ), as well as others (reviewed in Rochaix 1996), cause a complete absence of the target mRNA, no assessment of translational requirement can be made. On the other hand, both Chlamydomonas ( Drapier et al. 1992 ; Girard-Bascou et al. 1992 ; Stampacchia et al. 1997 ) and maize ( Barkan et al. 1994 ; McCormac & Barkan 1999) nuclear mutants affecting the translation of specific chloroplast proteins have little effect on RNA stability. Thus, it appeared that these two processes are controlled independently.
By inserting poly(G) into petD transcripts, we have enabled the accumulation of petD mRNA in the mcd1-1 nuclear background. In earlier experiments ( Drager et al. 1998 ), however, poly(G) was inserted exclusively at position +25 of a petD–uidA fusion, and no β-glucuronidase activity was observed, even in cells with a WT nucleus. In the present work, petD mRNAs with poly(G) at positions –20, +25, +165 and +25/165 were all shown to accumulate in the mcd1-1 background, but in no case was SUIV detectable. A priori, this could be considered as evidence that MCD1p is required for translation. However, the WT siblings of these mcd1-1 mutants also all contain petD transcripts with a 5′ end at +1, and it is possible that this is the sole translatable petD RNA species. We have been unsuccessful in trying to determine which transcripts are ribosome-associated in a WT nuclear background, a prerequisite to definitive interpretation of these data.
Interestingly, though, SUIV translation is normal in pG165/MCD1. If ribosome scanning had to occur through this region, one might expect that translation would be impeded. Although a eukaryotic mode of scanning probably does not operate in chloroplasts, certain forms of short-range scanning have been proposed (reviewed in Stern et al. 1997 ), or one could even envisage a novel mechanism such as discontinuous scanning or ‘shunting’, wherein secondary structures can be bypassed by ribosomes which initially bind upstream ( Fütterer et al. 1993 ; Yueh & Schneider 1996). A lack of scanning upstream of +165 also suggests that the reason for strongly reduced translation in pG25 is not an impediment to scanning. Because mutation of the +25 position (e.g. addition of a BglII site) also does not affect SUIV accumulation, we infer that the pG insertion may alter an essential RNA structure, although not in a way that interferes with binding of the putative MCD1 gene product.
Chlamydomonas strains, growth conditions and chloroplast transformation
The mcd1-1 mutant has been previously described ( Drager et al. 1998 ). Cells were grown in TAP medium ( Harris 1989) under constant light (70 μE m–2 sec–1). Chlamydomonas chloroplast transformants were generated as previously described ( Stern et al. 1991 ). The poly(G)-containing petD–aadA plasmids were introduced into chloroplasts of wild-type cells by particle bombardment, replacing the endogenous petD gene, and transformants were selected on TAP medium with spectinomycin (100 μg ml–1) under low light (1 μE m–2 sec–1). Transformants were repeatedly streaked for single colonies until the chloroplast DNA was homoplasmic for the introduced genes. The phenotypes of the transformants and progeny from crosses with the mcd1-1 mutant were determined, first by measuring the chlorophyll fluorescence transients ( Bennoun & Béal 1997), and second by plating strains on minimal medium to determine acetate dependence.
Construction of plasmids and probes
The poly(G)-containing petD plasmids were made by inserting the poly(G) tract into restriction endonuclease sites at positions –20, +25, and +165, with respect to the 5′ end of the mature petD transcript. Previously, a BglII site engineered at position +25 ( Sakamoto et al. 1994a ) and a NotI site engineered at position +165 (Higgs et al. unpublished data) were shown not to affect petD gene expression. To generate plasmid LS–20Bg, an intermediate in the generation of pG–20, site-directed mutagenesis ( Kunkel 1985) was used to add a NotI site to plasmid pD501 ( Sakamoto et al. 1994a ) at position –20; the addition of this site was tested and shown not to affect petD gene expression (data not shown). The 1.3 kb XhoI–PstI 5′ UTR fragments from these plasmids were inserted into the pBS SK+ vector (Stratagene), which had the NotI site in the polylinker sequence destroyed. Complementary oligonucleotides, containing the poly(G) tract and the appropriate restriction sites at the ends, were annealed and ligated into the unique NotI or BglII sites. In addition to the poly(G) sequence, the pG–20 plasmid also contained the BglII site at +25. Likewise, the pG25 plasmid contained the neutral NotI site at +165, and pG165 contained the neutral BglII site at +25. The pG25/165 plasmid was made by inserting a second poly(G) tract into the NotI site at +165 of the pG25 plasmid. The poly(G)-containing XhoI–PstI 5′ UTR fragments were used to replace the corresponding fragment of pD501 plasmid, which contains the downstream trnR gene and 2 kb of additional 3′ flanking chloroplast DNA. The aadA gene cassette ( Goldschmidt-Clermont 1991) was inserted as an EcoRV–SmaI fragment into the neutral HindIII site downstream of the trnR gene. The aadA gene has the Chlamydomonas atpA promoter and 5′ UTR, the E. coli aadA coding region, and the rbcL 3′ UTR; aadA was oriented such that it was transcribed in the same direction as petD and trnR.
A 496 bp fragment containing the petD 5′ UTR from the wild-type, pG25, pG165 and pG25/165 petD plasmids was PCR-amplified with primers WS11 ( Sakamoto et al. 1993 ) and WS12 ( Higgs et al. 1998 ), and ligated into the EcoRV site of the pBS SK+ vector (Stratagene). Antisense 32P-labelled RNA was in vitro-transcribed from XhoI-linearized plasmid DNA with T3 RNA polymerase (Promega), and gel-purified for use as a probe in RNAse protection assays.
RNA and protein accumulation measurements
Total cellular RNA was isolated from Chlamydomonas cells grown in TAP medium as previously described ( Drager et al. 1998 ). RNA was separated in 1.2% agarose–3% formaldehyde gels, blotted to a GeneScreen membrane (Du Pont), and hybridized with 32P-labelled petD and psbA DNA probes, as previously described ( Higgs et al. 1998 ). Radioactive bands were visualized and quantified with a Phosphorimager (Molecular Dynamics).
Total proteins were isolated and analysed on immunoblots by enhanced chemiluminescent (ECL) detection as previously described ( Higgs et al. 1998 ). The filter was also reacted with antibody raised against the Chlamydomonas oxygen-evolving enhancer protein 2 (OEE2) (1:10 000 dilution, a gift of F.-A. Wollman, Paris, France). Quantitative immunoblots were done as for ECL, except the ECL+ substrate (Amersham) was used and bands on the membrane were analysed using the blue fluorescence mode on the phosphorimager (Storm 840).
Primer extension and ribonuclease protection assays
Total cellular RNA was isolated as described above, and 10 μg was used for primer extension ( Higgs et al. 1998 ) reactions using primer WS5 ( Sakamoto et al. 1993 ). Products were analysed in a 7% sequencing gel and sized by comparison to corresponding petD (with added BglII site at +25) sequencing ladders.
Ribonuclease protection assays were conducted as previously described ( Drager et al. 1998 ), using the antisense probes described above. Protected fragments were size-fractionated in 5% sequencing gels, and sized by comparison to 32P-end-labelled low-molecular-weight RNA size standards (Gibco-BRL).
This paper is dedicated to the memory of Rob Drager. We thank members of the Stern laboratory for helpful comments and suggestions, and F.-A. Wollman for the gift of OEE2 antibody. This work was supported by NSF Grant MCB-9723274.
- 1994 A nuclear mutation in maize blocks the processing and translation of several chloroplast mRNAs and provides evidence for the differential translation of alternative mRNA forms. EMBO J., 13, 3170 3181. , , ,
- 1995 Degradation of mRNA in eukaryotes. Cell, 81, 179 183. &
- 1993. mRNA degradation in prokaryotic cells: An overview. In: Control of Messenger RNA Stability (Belasco, J.G. & Brawerman, G., eds). New York: Academic Press Inc, pp. 3 12.
- 1997 Screening algal mutant colonies with altered thylakoid electrochemical gradient through fluorescence and delayed luminescence digital imaging. Photosynthesis Res., 51, 161 165. &
- 1986 Specific endonucleolytic cleavage sites for decay of Escherichia coli messenger RNA. J. Mol. Biol., 192, 257 274. , ,
- 1993 Initiation codon mutations in the Chlamydomonas chloroplast petD gene result in temperature-sensitive photosynthetic growth. EMBO J., 12, 3627 3635. , ,
- 1996 A chloroplast transcript lacking the 3′ inverted repeat is degraded by 3′5′ exoribonuclease activity. RNA, 2, 652 663. , , ,
- 1998 In vivo evidence for 5′3′ exoribonuclease degradation of an unstable chloroplast mRNA. Plant J., 13, 85 96. , , , ,
- 1992 Evidence for nuclear control of the expression of the atpA and atpB chloroplast genes in Chlamydomonas. Plant Cell, 4, 283 295. , ,
- 1993 Nonlinear ribosome migration on cauliflower mosaic virus 35S RNA. Cell, 73, 789 802. , ,
- 1998 Deadenylation-dependent and -independent decay pathways for α1-tubulin mRNA in Chlamydomonas reinhardtii. Mol. Cell. Biol., 18, 1498 1505. &
- 1992 A nuclear mutation affects the synthesis of the chloroplast psbA gene product in Chlamydomonas reinhardtii. Curr. Genet., 22, 47 52. , ,
- 1991 Transgenic expression of aminoglycoside adenine transferase in the chloroplast: a selectable marker for site-directed transformation of Chlamydomonas. Nucl. Acids Res., 19, 4083 4090.
- 1995 Nuclear mutants of Chlamydomonas reinhardtii defective in the biogenesis of the cytochrome b6f complex. Plant Mol. Biol., 29, 921 932. , , , , ,
- 1989. The Chlamydomonas Sourcebook: a Comprehensive Guide to Biology and Laboratory Use. San Diego. Academic Press.
- 1996 Chloroplast mRNA 3′-end processing by a high molecular weight protein complex is regulated by nuclear encoded RNA binding proteins. EMBO J., 15, 1132 1141. , , , , ,
- 1994 Oat phytochrome A mRNA degradation appears to occur via two distinct pathways. Plant Cell, 6, 1007 1019. &
- 1998 Inversions in the Chlamydomonas chloroplast genome suppress a petD 5′ untranslated region deletion by creating functional chimeric mRNAs. Plant J., 14, 663 671. , , , ,
- 1996 Transcription of tufA and other chloroplast-encoded genes is controlled by a circadian clock in Chlamydomonas. Proc. Natl Acad. Sci. USA, 93, 996 1000. , ,
- 1997 Rat1p and Xrn1p are functionally interchangeable exoribonucleases that are restricted to and required in the nucleus and cytoplasm, respectively. Mol. Cell. Biol., 17, 6122 6130.
- 1993 Direct evidence for selective modulation of psbA, rpoA, rbcL, and 16S RNA stability during barley chloroplast development. Plant Mol. Biol., 22, 447 463. , ,
- 1991 Changes in chloroplast mRNA stability during leaf development. Plant Cell, 3, 517 530. &
- 1989 Mutation at the Chlamydomonas nuclear NAC2 locus specifically affects stability of the chloroplast psbD transcript encoding polypeptide D2 and PS II. Cell, 58, 869 876. , , ,
- 1985 Rapid and efficient site-specific mutagenesis without phenotypic selection. Proc. Natl Acad. Sci. USA, 82, 488 492.
- 1996 Function of 3′ non-coding sequences and stop codon usage in expression of the chloroplast psaB gene in Chlamydomonas reinhardtii. Plant Mol. Biol., 31, 337 354. , ,
- 1997 A nuclear mutation that affects the 3′ processing of several mRNAs in Chlamydomonas chloroplasts. Plant Cell, 9, 825 836. , ,
- 1999 A nuclear gene in maize required for the translation of the chloroplast atpB/E mRNA. Plant Cell, in press. &
- 1998 Ribonuclease E is a 5′-end-dependent endonuclease. Nature, 395, 720 723.
- 1996 RNA-binding proteins of 37/38 kDa bind specifically to the barley chloroplast psbA 3′-end untranslated RNA. Plant Mol. Biol., 30, 1195 1205. , ,
- 1992 Accumulation of chloroplast psbB RNA requires a nuclear factor in Chlamydomonas reinhardtii. Mol. Gen. Genet., 231, 449 459. , ,
- 1994 Premature translational termination triggers mRNA decapping. Nature, 370, 578 581. &
- 1994 Deadenylation of the unstable mRNA encoded by the yeast MFA2 gene leads to decapping followed by 5′3′ digestion of the transcript. Genes Dev., 8, 855 866. , ,
- 1994 Determinants for stability of the chloroplast psbD RNA are located within its short leader region in Chlamydomonas reinhardtii. EMBO J., 13, 3182 3191. , , ,
- 1999 Identification of cis-acting RNA leader elements required for chloroplast psbD gene expression in Chlamydomonas. Plant Cell, 11, 957 970. , , , ,
- 1996 Post-transcriptional regulation of chloroplast gene expression in Chlamydomonas. Plant Mol. Biol., 32, 327 341.
- 1993 In vivo analysis of Chlamydomonas chloroplast petD gene expression using stable transformation of β-glucuronidase translational fusions. Proc. Natl Acad. Sci. USA, 90, 497 501. , ,
- 1994a Function of the Chlamydomonas reinhardtii petD 5′ untranslated region in regulating the accumulation of subunit IV of the cytochrome b6/f complex. Plant J., 6, 503 512. , , ,
- 1994b petD mRNA maturation in Chlamydomonas reinhardtii chloroplasts: the role of 5′ endonucleolytic processing. Mol. Cell. Biol., 14, 6180 6186. , , ,
- 1993 Light-regulated and endogenous fluctuations of chloroplast transcript levels in Chlamydomonas. Regulation by transcription and RNA degradation. Plant J., 3, 213 219. , ,
- 1995 Characterization of cDNA encoding mouse homolog of fission yeast dhp1+ gene: structural and functional conservation. Nucl. Acids Res., 23, 357 361. , , ,
- 1997 A nuclear-encoded function essential for translation of the chloroplast psaB mRNA in Chlamydomonas reinhardtii. Plant Cell, 9, 773 782. , , , , ,
- 1987 Control of plastid gene expression: 3′ inverted repeats act as mRNA processing and stabilizing elements, but do not terminate transcription. Cell, 51, 1145 1157. &
- 1991 A 3′ stem/loop structure of the Chlamydomonas chloroplast atpB gene regulates mRNA accumulation in vivo. Plant Cell, 3, 285 297. , ,
- 1997 Transcriptional and translational activities of chloroplasts. Trends Plant Sci., 2, 308 315. , ,
- 1995 5′-exonuclease-2 of Saccharomyces cerevisiae. Purification and features of ribonuclease activity with comparison to 5′-exonuclease-1. J. Biol. Chem., 270, 16063 16069. &
- 1994 The petD gene is transcribed by functionally redundant promoters in Chlamydomonas reinhardtii chloroplasts. Mol. Cell. Biol., 14, 6171 6179. , , , , , ,
- 1996 Regulation of gene expression in chloroplasts of higher plants. Plant Mol. Biol., 32, 315 326. &
- 1993 Conducting the G-quartet. Curr. Biol., 3, 893 895.
- 1992 The rate-limiting step in yeast PGK1 mRNA degradation is an endonucleolytic cleavage in the 3′-terminal part of the coding region. Mol. Cell. Biol., 12, 2986 2996. &
- 1996 CSP41, a sequence-specific chloroplast mRNA binding protein, is an endoribonuclease. Plant Cell, 8, 1409 1420. , ,
- 1996 Selective translation initiation by ribosome jumping in adenovirus-infected and heat-shocked cells. Genes Dev., 10, 1557 1567. &
GenBank accession number AI536027.