Surprising features of plastid ndhD transcripts: addition of non-encod
http://www.100md.com
《核酸研究医学期刊》
Westf?lische Wilhelms-Universit?t Münster, Institut für Biochemie und Biotechnologie der Pflanzen, Hindenburgplatz 55, D-48143 Münster, Germany
*To whom correspondence should be addressed. Tel: +49 251 83 24790/91; Fax: +49 251 83 28371; Email: rbock@uni-muenster.de
Present address:
Aitor Zandueta-Criado, Instituto de Agrobiotecnología y Recursos Naturales, Universidad Pública de Navarra/Consejo Superior de Investigaciones Científicas, Ctra. de Mutilva s/n, Mutilva Baja 31192 Navarra, Spain
ABSTRACT
RNA editing in higher plant plastids is a post- transcriptional RNA maturation process changing single cytidine nucleotides into uridine. In the ndhD transcript of tobacco and several other plant species, editing of an ACG codon to a standard AUG initiator codon is believed to be a prerequisite for translation. In order to test this assumption experimentally, we have analyzed the editing status of ndhD mRNA species in the process of translation. We show that unedited ndhD transcripts are also associated with polysomes in vivo, suggesting that they are translated. This surprising finding challenges the view that ACG to AUG editing is strictly required to make the ndhD message translatable and raises the possibility that ACG can be utilized as an initiator codon in chloroplasts. In addition, we have mapped the termini of the ndhD transcript and discovered a novel form of RNA processing. Unexpectedly, we find that highly specific sequences are added to the 3' end of the ndhD mRNA at high frequency. We propose a model in which these sequences are added by the successive action of a CCA-adding enzyme (tRNA nucleotidyltransferase) and an RNA-dependent RNA polymerase (RdRp) activity. The presence of an RdRp activity may have general implications also for other steps in plastid gene expression.
Introduction
RNA editing alters the identity of single nucleotides in organellar transcripts of higher plants, mainly by cytidine to uridine conversion . In most instances, these changes are functionally significant in that codons for conserved amino acid residues are restored, and lack of RNA editing would result in impaired protein function (9,10).
Editing in plant cell organelles is a post-transcriptional process and largely independent of other RNA processing steps, such as intron splicing and cutting of polycistronic precursor transcripts into oligocistronic or monocistronic mRNAs (11–15). In contrast, the relationship between RNA editing and translation is much less clear. There seems to be no uniform way in which the organellar translational apparatus deals with unedited or partially edited mRNAs. Using specific antibodies against polypeptides reflecting edited or unedited mRNAs, plant mitochondria were assayed for the presence of proteins resulting from the translation of incompletely edited transcripts. Interestingly, translation products made from both edited and unedited mRNAs accumulated for a ribosomal protein, Rps12 (16,17). However, only the ribosomal protein S12 resulting from fully edited rps12 transcripts appears to become incorporated into ribosomes, and whether or not the other S12 species serve some function remains unclear. In contrast, for the ATPase subunit Atp6 (18) and another ribosomal protein, Rps13 (19), proteins made from unedited transcripts could not be detected. This is most probably due to their rapid degradation rather than exclusion of unedited transcripts from translation. In accordance with this assumption, both fully edited and partially edited mRNAs were found associated with mitochondrial polysomes (20). Similarly, the plastid translational apparatus does not seem to discriminate against unedited transcripts (21). In contrast to mRNAs with internal editing sites, it has been generally assumed that the situation was much clearer with editing events creating initiation codons (by changing a genomically encoded ACG triplet into a standard AUG start codon). Here, it seemed conceivable that unedited transcripts would be excluded from translation. This assumption was supported, at least for two cases, by transgenic experiments (22) and in vitro translation data (23). However, recent data have cast considerable doubt on the strict requirement of an AUG codon for translation to initiate in both plant mitochondria and plastids. When the AUG start codon was mutated in a chloroplast mRNA of the unicellular green alga Chlamydomonas reinhardtii, translation, though being significantly reduced, was not abolished and still initiated at the correct site (24). In addition, the radish (Raphanus sativus) mitochondrial cox2 gene was recently found to contain an ACG codon at the predicted translation initiation site which does not undergo editing into a standard AUG initiator codon (25). Similarly, the start codon of the ndhD message in Nicotiana tabacum and Nicotiana sylvestris is created by ACG to AUG editing, whereas, in Nicotiana tomentosiformis, the ACG remains unedited (26). However, in both plastids and mitochondria, experimental evidence for translation of mRNAs with unedited ACG start codons is lacking.
In this study, we have analyzed the relationship between RNA editing and translation for the plastid ndhD mRNA. The ndhD gene specifies a subunit of a plastid NADH dehydrogenase and is unique in that a significant proportion of its transcripts carries an unedited initiation codon . In fact, only between 30 and 60% of the ndhD transcripts carry the edited AUG codon which has been suggested to be required for initiation of ndhD translation (23,27). Here we show that also the unedited ACG-containing ndhD transcripts are heavily loaded with ribosomes, suggesting that they are actively translated. In addition, by analyzing the 5' and 3' termini of mature ndhD mRNAs, we have discovered that the 3' ends frequently carry stretches of highly specific non-encoded nucleotides which presumably are added by an RNA-dependent RNA polymerase (RdRp) activity.
Materials and Methods
Oligonucleotides
The oligonucleotides used were as follows: ndhD3', 5'-CAA GGTCGAAGTTATTCTATC-3'; ndhD5', 5'-TGGTCCAA GTGTATCTTGTC-3'; ndhDi, 5'-AGAAAAACAGAAC CCCCC-3'; and ndhDs, 5'-GAAAATTAAGGAACCCGC-3'. Their location and orientation are depicted in Figure 1.
Figure 1. Experimental strategy towards mapping the 5' and 3' ends and analyzing the RNA editing status of plastid ndhD transcripts. In the upper panel, the location and orientation of primers for cDNA synthesis and PCR are shown relative to the ndhD coding region. Relevant restriction sites for cloning are indicated. Transcripts are self-ligated with T4 RNA ligase, thereby fusing their 5' and 3' ends to produce circularized mRNA molecules. After cDNA synthesis primed with an ndhD-specific oligonucleotide, the region containing the 5' UTR, 3' UTR and the RNA editing site within the ndhD start codon is amplified by PCR. Products are then cloned and individual clones are sequenced to determine the termini of the mRNAs and the editing status of the start codon.
Isolation of polysome fractions and polysome-associated RNAs
Polysomes and polysomal RNAs were purified as described previously (28) with the following modifications: Tobacco leaves (2 g) were ground in liquid nitrogen and treated with 8 ml of ice-cold polysome extraction buffer. Fractions of 400 μl were collected from the analytical sucrose gradients. Prior to RNA isolation, all fractions were diluted with 280 μl of distilled water. Isopropanol precipitation was carried out without the addition of glycogen. Subsequently, RNA was washed with 70% ethanol and pellets were resuspended in 20 μl of sterile distilled water.
Northern blot and hybridization procedures
Aliquots of 4 μl of RNA extracted from polysome fractions were electrophoresed on formaldehyde-containing 1% agarose gels and transferred onto Hybond-XL membranes (Amersham). For hybridization, an dCTP-labeled probe was generated by random priming (Amersham) following the manufacturer’s instructions. An ndhD-specific probe was prepared by PCR amplification with primers ndh5' and ndhDi using total plant DNA as a template.
RNA circularization, cDNA synthesis and PCR
A 2 μl aliquot of RNA from fractions P1 and P6 (EDTA-free polysome gradients) and 0.5 μl of RNA from fraction C10 (EDTA-containing polysome gradient) were ligated at 37°C for 1 h with 20 U of T4 RNA ligase (New England Biolabs) in a final volume of 100 μl. cDNA synthesis was performed as follows. A 1 μg aliquot of RNA (precipitated with ethanol after ligation and redissolved in water) and 38 ng of oligonucleotide ndhDi were denatured for 5 min at 70°C in a total volume of 14 μl and then cooled on ice. Subsequently, reaction buffer and dNTPs (final concentration 2 mM each) were added and the volume was adjusted to 24 μl. The reverse transcription reaction was initiated by addition of 16 U of M-MLV reverse transcriptase (Promega) and was allowed to proceed at room temperature for 10 min followed by incubation at 45°C for 1 h. The resulting cDNA templates were amplified according to standard PCR protocols (30 s at 93°C, 1 min 30 s at 52°C, 1 min 30 s at 72°C; 30 cycles). Primers ndhDs and ndhD3' were used for specific amplification of the head-to-tail ligated 5' and 3' untranslated regions (UTRs). Amplification products were separated by agarose gel electrophoresis and the 300–400 bp band was excised and purified.
Cloning and sequencing
Purified amplification products were digested with StyI and XmnI and ligated into a pBluescript(SK+)-derived plasmid (carrying a cloned DNA fragment providing a CCTTGG StyI site) cut with StyI and EcoRV. Individual clones were sequenced with the M13 forward primer using the dideoxy chain termination method.
RESULTS AND DISCUSSION
A strategy for simultaneous mapping of transcript termini and determining the editing status of ndhD mRNAs
A variety of methods is available for the precise mapping of transcript termini, including primer extension (for 5' mapping), RNase protection assays, RNA ligation with anchor oligonucleotides and RNA circularization producing head-to-tail ligated transcripts (30). To determine both the 5' and the 3' ends of mature ndhD transcripts and to simultaneously analyze the RNA editing site within the start codon in individual mRNA molecules, we chose the experimental strategy illustrated in Figure 1. In order to exclude artifacts that could be experimentally introduced at the mRNA termini by the possible non-specific action of reverse transcriptases or DNA polymerases, we decided to circularize the RNA molecules prior to any treatment with nucleotide polymerizing enzymes (Fig. 1). Head-to-tail ligated mRNAs were then reverse transcribed using a gene-specific primer followed by amplification of the region containing the fused UTRs and the editing site-containing start codon (Fig. 1). As we were interested in characterizing mature translated mRNAs, we purified polysomes from tobacco leaves and fractionated polysomal RNAs in sucrose gradients. This polysomal RNA was then circularized and subsequently treated as shown in Figure 1.
Polysome association of plastid ndhD mRNAs
Polysome fractionation on sucrose gradients separates mRNAs according to the number of ribosomes bound to individual RNA molecules: mRNAs carrying large numbers of ribosomes migrate to the bottom of the gradient, whereas mRNAs displaying no or poor loading with ribosomes stay in the top fractions of the gradient (Fig. 2). This fractionation pattern is nicely seen with polysomes from tobacco leaves (Fig. 2; lower left panel). In contrast, control gradients centrifuged in the presence of EDTA (which causes release of ribosomes from the mRNAs) displayed no RNA migration into the bottom fractions of the gradient (Fig. 2; upper left panel), confirming successful mRNA fractionation according to ribosome number.
Figure 2. Detection of ndhD transcripts in polysome preparations from tobacco leaves. Polysomes were fractionated in sucrose gradients in either the presence (upper panel; control) or absence (lower panel) of EDTA. RNA samples from individual gradient fractions were analyzed on ethidium bromide-stained gels (left) and assayed for the presence of ndhD mRNAs by northern blotting and hybridization to an ndhD-specific probe (right). The presence of ndhD transcripts only in the upper fractions (9–12) of the EDTA gradient indicates release of ribosomes from mRNAs as caused by EDTA addition. In contrast, the EDTA-free gradient shows migration of large amounts of ndhD mRNAs into the bottom fractions. The prominent band at 2.5 kb represents dicistronic psaC–ndhD transcripts, whereas the 1.7 kb band consists of monocistronic ndhD message (31,32).
In order to detect ndhD mRNA species that are actively translated, RNAs extracted from sucrose gradient fractions were assayed for the presence of ndhD transcripts by northern blot analysis. The ndhD gene is part of an operon with the functionally unrelated psaC cistron (encoding a photosystem I subunit) located immediately upstream. Compared with psaC, ndhD mRNAs accumulate to rather low levels (31), consistent with the presumably much lower demand for NADH dehydrogenase subunits than for photosystem I proteins. It has been observed previously that the transcript pattern of ndhD is highly complex, with both dicistronic psaC–ndhD and monocistronic ndhD mRNAs accumulating in addition to numerous larger polycistronic precursor transcripts (31,32). When assaying polysome-associated mRNAs, we detected a similarly complex ndhD transcript pattern with a prominent band for the psaC–ndhD dicistronic mRNA (2.5 kb) in addition to somewhat less abundant transcript species representing monocistronic (1.7 kb) and several polycistronic mRNAs (Fig. 2). The dicistronic and polycistronic species are of limited interest for the purpose of this study since their ribosome association could be attributed to translation of the adjacent cistrons and does not necessarily suggest active translation of ndhD. We therefore focused our further studies on monocistronic ndhD transcripts whose isolation from polysomal fractions suggests their translation.
Multiple 5' and 3' ends of translated ndhD mRNAs
In order to precisely map the 5' and 3' termini of translated ndhD mRNAs, RNA samples from three different polysome gradient fractions were selected: fractions 1 and 6 from the EDTA-free polysome gradient (subsequently referred to as P1 and P6) and fraction 10 from the EDTA-containing control gradient (subsequently referred to as C10; Fig. 2). Using the strategy outlined in Figure 1, PCR with the ndhD-specific primers ndhDs and ndhD3' resulted in successful amplification of the head-to-tail ligated 3' and 5' UTRs of monocistronic ndhD transcripts. Since in PCRs, the amplification of small molecules is strongly favored over the amplification of larger molecules, only products corresponding to monocistronic message were obtained, in spite of the higher abundance of polycistronic transcript species (Fig. 2 and data not shown). Following cloning into a plasmid vector (Fig. 1), 16–22 individual clones were sequenced for each of the three samples (P1, P6 and C10).
Previous mapping of the 5' ends of ndhD transcripts by primer extension and RNase protection analyses had revealed the presence of multiple 5' ends (31,32). Our sequence analyses of individual head-to-tail ligated ndhD-derived cDNAs confirmed the presence of multiple 5' termini which all mapped inside the intercistronic spacer region in between ndhD and the upstream psaC (Fig. 3). As our RNA circularization strategy allowed the simultaneous mapping of 3' ends, which had not been determined previously for ndhD, we were interested in identifying processing sites within the 3' UTR in relation to possible secondary structure elements. Most 3' UTRs of plastid transcripts harbor stem–loop-type inverted repeats which act as RNA processing and stabilizing elements but do not terminate transcription (33,34). Indeed, we could identify such a putative stem–loop-type RNA secondary structure element also within the ndhD 3' UTR (Fig. 3). Most ndhD transcripts from polysomal fractions were found to terminate 2–6 nt downstream of the stem–loop (Fig. 3), strongly supporting that this secondary structure is relevant in vivo. The termini of the remaining clones were found to map inside the stem (Fig. 3), which could indicate that the corresponding RNA molecules are condemned to degradation (since they lack the putative transcript-stabilizing secondary structure). Decay of plastid transcripts frequently initiates from an endonucleolytic cleavage within the 3' UTR removing the protecting stem–loop structure . As psaC and ndhD are co-transcribed from an operon, but yet psaC accumulates to much higher levels than ndhD (31), a reasonable assumption could be that ndhD transcripts are turned-over more rapidly. Hence, the detection of RNA degradation intermediates may not be surprising. The majority of the 3'-truncated clones carried an unedited start codon (fraction P6 in Fig. 3) which could be indicative of ACG-containing transcripts being more susceptible to degradation than AUG-containing mRNAs.
Figure 3. Mapped ndhD mRNA termini and editing status of ndhD mRNAs. RNAs from fraction 10 of the EDTA-containing control gradient (C10; see Fig. 2) and fractions 1 and 6 of the EDTA-free gradient were circularized with RNA ligase and analyzed as displayed in Figure 1. Arrows indicate mapped termini, with the number of clones given for each terminus. Asterisks indicate ambiguity with respect to the terminal nucleotide (due to the presence of identical nucleotides at the putative 5' and 3' ends). The stem–loop indicates a potential secondary structure within the 3' UTR which is typical of plastid mRNAs. The number of unedited versus edited clones is given in parentheses above the initiation codon.
Translation of ndhD mRNAs with an unedited start codon
Having confirmed that we had successfully cloned head-to-tail ligated monocistronic ndhD transcript-derived cDNAs, we were most interested in analyzing the editing status of the start codon. If translation was indeed strictly dependent on editing of the ACG into a standard AUG initiator codon, then all polysome-loaded monocistronic ndhD mRNAs should harbor the edited AUG codon. Sequencing of 59 individual clones altogether from the three different gradient fractions (P1, P6 and C10; Fig. 2) revealed a large number of clones carrying the unedited ACG codon (38 clones equaling 64%), suggesting that unedited ndhD mRNAs are also translated. Moreover, the percentage of edited clones (36%) derived from polysomal mRNA is in the range of the previously determined efficiency of ndhD start codon editing , indicating that the chloroplast translational apparatus does not strongly discriminate against ACG-containing transcripts. However, the amount of edited clones was significantly higher in the bottom fraction (P1) of the polysome gradient (with 13 out of 21 clones being edited) than in the middle fraction (P6) of the gradient (with only two out of 22 clones being edited), indicating that AUG-containing mRNAs are, on average, loaded with more ribosomes. This finding could suggest that re-initiation of translation is more efficient at the AUG codon than at the ACG codon, and raises the possibility that editing is employed to regulate translation of the ndhD message. Consistent with such a translational regulation through RNA editing, it has been reported that, in N.tabacum, the editing efficiency at the ndhD start codon is strongly influenced by plastid differentiation, tissue type, plant development and the environmental factor light (23).
Translation of ndhD mRNA with an unedited ACG codon could also explain why this editing site remains completely unedited in N.tomentosiformis (26) and raises the possibility that, at least in the ndhD sequence context, ACG can be used as an initiator codon for plastid translation. However, at present, we cannot entirely rule out the possibility that translation of unedited ndhD mRNAs initiates at an alternative site and not from the unedited ACG codon. At least three different codons can be utilized as initiator codons in chloroplasts (AUG, GUG and UUG), and any such codon upstream or downstream of the ACG could, in theory, also serve as initiator codon.
Presence of non-encoded sequences in mature ndhD transcripts
Upon inspection of the sequences at the ligation site of the head-to-tail joined 5' and 3' UTRs of ndhD, we surprisingly discovered long stretches of non-encoded sequence (Figs 4 and 5). Significantly, these stretches were found in the majority of the analyzed clones (i.e. in 39 out of 59 clones). These non-encoded extra sequences, although differing in length, were highly homologous to each other (Fig. 5), strongly suggesting that they share a common mechanism of addition. Although the head-to-tail ligation approach does not allow us to distinguish between addition to the 5' UTR and addition to the 3' UTR, the 5' to 3' working direction of nucleotide polymerases strongly suggests that the extra sequences are added to the 3' end of the mRNA. (This assumption gained further support by our identification of the probable origin of the non-encoded sequence; see below and Figs 4–6.)
Figure 4. Example of a clone containing a stretch of non-encoded sequence at the 3' end. The complementary sequence motif upstream of the stem–loop structure is marked. The mapped 5' and 3' ends, the sequence of the stem–loop structure, the edited initiation codon and the StyI site used for cloning (see Fig. 1) are also indicated.
Figure 5. Presence of non-encoded sequences at the 3' end of plastid ndhD transcripts. The non-encoded sequences are aligned with a stretch of complementary sequence present upstream of the stem–loop structure. Arrows indicate mapped 3' termini and the corresponding number of clones. Nucleotides in parentheses were not found in all clones.
Figure 6. Model for the addition of 3'-non-encoded sequences to plastid ndhD transcripts. The frequent presence of CCA-like sequences suggests initial involvement of the CCA-adding enzyme (tRNA nucleotidyltransferase) which may recognize the stem–loop because of structural similarity with its regular tRNA substrate. Complementarity of the free CCA end with the sequence immediately upstream of the stem–loop structure may prime RNA-dependent RNA polymerization using the mRNA as a template. The RNA-dependent RNA polymerase (RdRp) activity does not necessarily have to come from a specific RdRp enzyme, but also could be a hitherto undetected side activity of normally DNA-dependent RNA polymerases, such as the chloroplast transcription enzymes PEP and NEP (48,52,53) or the primase for plastid DNA replication .
The presence of the non-encoded nucleotides cannot be attributed to an experimentally induced artifact since the mRNAs were circularized prior to any contact with nucleotide-polymerizing activities and (d)NTPs (Fig. 1).
To date, two types of non-encoded nucleotide stretches have been found at the 3' end of organellar RNAs: (i) homopolymeric stretches of mostly A nucleotides resulting in polyadenylated 3' UTRs; and (ii) short CCA-like additions pointing to an involvement of CCA-adding enzyme (tRNA nucleotidyltransferase). The former are found in both plastids (35,37,38) and mitochondria (39–41), whereas the latter so far have been reported only for plant mitochondria (42). Polyadenylation is generally believed to mark RNA molecules for degradation, which is supported by polyadenylated plastid and mitochondrial RNAs being highly unstable (38,43–45). In contrast, the 3' addition of relatively long, highly specific and non-homopolymeric sequence stretches to an mRNA, as seen in the majority of our ndhD clones, has not been observed before.
To us, it seemed unlikely that such long specific sequences are synthesized without using a template. We therefore searched for homologies of the non-encoded sequence stretch with other sequences in the tobacco chloroplast genome. Interestingly, we found a perfectly complementary sequence motif immediately upstream of the stem–loop structure within the ndhD 3' UTR (Figs 4 and 5). This finding strongly suggests that the sequence upstream of the stem–loop is used as a template to polymerize the extra sequence found at the transcript terminus. Such an RNA-dependent RNA polymerization has not been observed before in organellar RNA processing and, to our knowledge, no candidate enzymatic activity has been identified to date in either chloroplasts or plant mitochondria.
In some clones, the first 1–3 extra nucleotides were not complementary to the putative template sequence upstream of the stem–loop (Fig. 5). Interestingly, these nucleotides are always CCA (or CCA-like; Fig. 5). Moreover, the sequence at the junction between the 3' end and the non-encoded extra sequence is CCA-like in all clones (Fig. 5). This might indicate that this first part of the non-encoded sequence is added by an enzymatic activity similar or identical to the CCA-adding enzyme normally involved in tRNA maturation. Recently, non-encoded CCA-like sequences have been found at the 3' end of maize mitochondrial mRNAs, and it has been hypothesized that they are added by a tRNA nucleotidyltransferase-like activity (42).
A model that could explain the origin of the 3' non-encoded sequences found at the 3' end of ndhD mRNAs is presented in Figure 6. Briefly, addition of a CCA sequence by the CCA-adding enzyme (tRNA nucleotidyltransferase) would create a sequence complementarity between the 3' end and a UGG motif upstream of the stem–loop. In this way, RNA-dependent RNA polymerization could be primed, which then is likely to proceed by a different enzymatic activity (i.e. an RdRp). It seems unnecessary to postulate the existence of a separate RdRp enzyme in plastids, as DNA-dependent RNA polymerases can have some RdRp-like side activity. This has been demonstrated, for example, for T7 RNA polymerase (46). It seems noteworthy in this respect that chloroplasts possess, in addition to a plastid-encoded Escherichia coli-like RNA polymerase, also a nuclear-encoded T7-like RNA polymerase (47–49).
At present, we can only speculate about the possible functional significance of these extra sequences added to the 3' end of the ndhD message. As >80% of the stem–loop-containing ndhD transcripts carry such extensions, nucleotide addition appears to be a fairly efficient post-transcriptional process. Whether or not this nucleotide addition, for example, influences or even regulates transcript lifetime (e.g. by providing an extended stabilizing double-stranded RNA structure) remains to be investigated. Moreover, the occurrence of RNA template-directed RNA synthesis in plastids, as suggested by our data, may also have implications for other steps in plastid gene expression. Synthesis of antisense strands by RdRp and the resulting formation of stable double-stranded RNAs could, for example, affect translatability of plastid mRNAs, influence RNA degradation, control 3' processing or intercistronic RNA processing (RNA cutting), or even trigger RNA amplification. It is of note in this regard that antisense RNAs can be important regulators of gene expression in prokaryotes (50,51).
It will be interesting to see how widespread 3' processing by addition of non-encoded sequence stretches is in chloroplast (and perhaps also mitochondrial) mRNAs. Preliminary data suggest that not all chloroplast transcripts are processed in this manner. This may be unsurprising since not all 3' UTRs of chloroplast mRNAs harbor a UGG motif upstream of the stem—loop which, according to our proposed model, would facilitate priming of RNA-dependent polymerization (Figs 5 and 6). One intensely studied example is the petD mRNA 3' end where no such sequence additions have been detected .
Taken together, during the course of this work, we have provided evidence for translation of ndhD mRNAs carrying an unedited start codon and detected a novel form of RNA processing by addition of specific sequences to the 3' end, presumably through CCA addition followed by RNA- dependent RNA polymerization.
ACKNOWLEDGEMENTS
We thank Mrs Daniela Ahlert for DNA sequencing, Dr Javier Pozueta-Romero for valuable discussion and support, and Dr J?rg Kudla for critical reading of the manuscript. This work was financed by grants from the Deutsche Forschungsgemeinschaft to R.B. A.Z.C. was supported by a fellowship from the Ministerio de Educación Cultura y Deporte, Spain.
REFERENCES
Gualberto,J.M., Lamattina,L., Bonnard,G., Weil,J.-H. and Grienenberger,J.-M. (1989) RNA editing in wheat mitochondria results in the conservation of protein sequences. Nature, 341, 660–662.
Covello,P.S. and Gray,M.W. (1989) RNA editing in plant mitochondria. Nature, 341, 662–666.
Hiesel,R., Wissinger,B., Schuster,W. and Brennicke,A. (1989) RNA editing in plant mitochondria. Science, 246, 1632–1634.
Hoch,B., Maier,R.M., Appel,K., Igloi,G.L. and K?ssel,H. (1991) Editing of a chloroplast mRNA by creation of an initiation codon. Nature, 353, 178–180.
Kudla,J., Igloi,G.L., Metzlaff,M., Hagemann,R. and K?ssel,H. (1992) RNA editing in tobacco chloroplasts leads to the formation of a translatable psbL mRNA by a C to U substitution within the initiation codon. EMBO J., 11, 1099–1103.
Mulligan,R.M., Williams,M.A. and Shanahan,M.T. (1999) RNA editing site recognition in higher plant mitochondria. J. Hered., 90, 338–344.
Bock,R. (2000) Sense from nonsense: how the genetic information of chloroplasts is altered by RNA editing. Biochimie, 82, 549–557.
Bock,R. (2001) RNA editing in plant mitochondria and chloroplasts. In Bass,B. (ed.), Frontiers in Molecular Biology: RNA Editing. Oxford University Press, pp. 38–60.
Bock,R., K?ssel,H. and Maliga,P. (1994) Introduction of a heterologous editing site into the tobacco plastid genome: the lack of RNA editing leads to a mutant phenotype. EMBO J., 13, 4623–4628.
Zito,F., Kuras,R., Choquet,Y., K?ssel,H. and Wollman,F.-A. (1997) Mutations of cytochrome b6 in Chlamydomonas reinhardtii disclose the functional significance for a proline to leucine conversion by petB editing in maize and tobacco. Plant Mol. Biol., 33, 79–86.
Sutton,C.A., Conklin,P.L., Pruitt,K.D. and Hanson,M.R. (1991) Editing of pre-mRNAs can occur before cis- and trans-splicing in Petunia mitochondria. Mol. Cell. Biol., 11, 4274–4277.
Yang,A.J. and Mulligan,R.M. (1991) RNA editing intermediates of cox2 transcripts in maize mitochondria. Mol. Cell. Biol., 11, 4278–4281.
Gualberto,J.M., Bonnard,G., Lamattina,L. and Grienenberger,J.M. (1991) Expression of the wheat mitochondrial nad3–rps12 transcription unit: correlation between editing and mRNA maturation. Plant Cell, 3, 1109–1120.
Ruf,S., Zeltz,P. and K?ssel,H. (1994) Complete RNA editing of unspliced and dicistronic transcripts of the intron-containing reading frame IRF170 from maize chloroplasts. Proc. Natl Acad. Sci. USA, 91, 2295–2299.
Freyer,R., Hoch,B., Neckermann,K., Maier,R.M. and K?ssel,H. (1993) RNA editing in maize chloroplasts is a processing step independent of splicing and cleavage to monocistronic mRNAs. Plant J., 4, 621–629.
Phreaner,C.G., Williams,M.A. and Mulligan,R.M. (1996) Incomplete editing of rps12 transcripts results in the synthesis of polymorphic polypeptides in plant mitochondria. Plant Cell, 8, 107–117.
Lu,B., Wilson,R.K., Phreaner,C.G., Mulligan,R.M. and Hanson,M.R. (1996) Protein polymorphism generated by differential RNA editing of a plant mitochondrial rps12 gene. Mol. Cell. Biol., 16, 1543–1549.
Lu,B. and Hanson,M.R. (1994) A single homogeneous form of ATP6 protein accumulates in petunia mitochondria despite the presence of differentially edited atp6 transcripts. Plant Cell, 6, 1955–1968.
Williams,M.A., Tallakson,W.A., Phreaner,C.G. and Mulligan,R.M. (1998) Editing and translation of ribosomal protein S13 transcripts: unedited translation products are not detectable in maize mitochondria. Curr. Genet., 34, 221–226.
Lu,B. and Hanson,M.R. (1996) Fully edited and partially edited nad9 transcripts differ in size and both are associated with polysomes in potato mitochondria. Nucleic Acids Res., 24, 1369–1374.
Hirose,T., Kusumegi,T. and Sugiura,M. (1998) Translation of tobacco chloroplast rps14 mRNA depends on a Shine–Dalgarno-like sequence in the 5'-untranslated region but not on internal RNA editing in the coding region. FEBS Lett., 430, 257–260.
Chaudhuri,S. and Maliga,P. (1996) Sequences directing C to U editing of the plastid psbL mRNA are located within a 22 nucleotide segment spanning the editing site. EMBO J., 15, 5958–5964.
Hirose,T. and Sugiura,M. (1997) Both RNA editing and RNA cleavage are required for translation of tobacco chloroplast ndhD mRNA: a possible regulatory mechanism for the expression of a chloroplast operon consisting of functionally unrelated genes. EMBO J., 16, 6804–6811.
Chen,X., Kindle,K.L. and Stern,D.B. (1995) The initiation codon determines the efficiency but not the site of translation initiation in Chlamydomonas chloroplasts. Plant Cell, 7, 1295–1305.
Dong,F.G., Wilson,K.G. and Makaroff,C.A. (1998) The radish (Raphanus sativus L.) mitochondrial cox2 gene contains an ACG at the predicted translation initiation site. Curr. Genet., 34, 79–87.
Sasaki,T., Yukawa,Y., Miyamoto,T., Obokata,J. and Sugiura,M. (2003) Identification of RNA editing sites in chloroplast transcripts from the maternal and paternal progenitors of tobacco (Nicotiana tabacum): comparative analysis shows the inolvement of distinct trans-factors for ndhB editing. Mol. Biol. Evol., 20, 1028–1035.
Neckermann,K., Zeltz,P., Igloi,G.L., K?ssel,H. and Maier,R.M. (1994) The role of RNA editing in conservation of start codons in chloroplast genomes. Gene, 146, 177–182.
Ruf,S., K?ssel,H. and Bock,R. (1997) Targeted inactivation of a tobacco intron-containing open reading frame reveals a novel chloroplast-encoded photosystem I-related gene. J. Cell Biol., 139, 95–102.
Liu,X. and Gorovsky,M.A. (1993) Mapping the 5' and 3' ends of Tetrahymena thermophila mRNAs using RNA ligase mediated amplification of cDNA ends (RLM-RACE). Nucleic Acids Res., 21, 4954–4960.
Kuhn,J. and Binder,S. (2002) RT–PCR analysis of 5' to 3'-end-ligated mRNAs identifies the extremities of cox2 transcripts in pea mitochondria. Nucleic Acids Res., 30, 439–446.
delCampo,E.M., Sabater,B. and Martin,M. (2002) Post-transcriptional control of chloroplast gene expression. J. Biol. Chem., 277, 36457–36464.
Matsubayashi,T., Wakasugi,T., Shinozaki,K., Yamaguchi-Shinozaki,K., Zaita,N., Hidaka,T., Meng,B.Y., Ohto,C., Tanaka,M., Kato,A., Maruyama,T. and Sugiura,M. (1987) Six chloroplast genes (ndhA–F) homologous to human mitochondrial genes encoding components of the respiratory chain NADH dehydrogenase are actively expressed: determination of the splice site in ndhA and ndhB pre-mRNAs. Mol. Gen. Genet., 210, 385–393.
Stern,D.B. and Gruissem,W. (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.
Hayes,R., Kudla,J., Schuster,G., Gabay,L., Maliga,P. and Gruissem,W. (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.
Hayes,R., Kudla,J. and Gruissem,W. (1999) Degrading chloroplast mRNA: the role of polyadenylation. Trends Biochem. Sci., 24, 199–202.
Schuster,G. and Bock,R. (2001) Editing, polyadenylation and degradation of mRNA in the chloroplast. In Aro,E.-M. and Andersson,B. (eds), Advances in Photosynthesis and Respiration. Regulation of Photosynthesis. Kluwer Academic Press, Dordrecht, The Netherlands, Vol. 11, pp. 121–136.
Lisitsky,I., Klaff,P. and Schuster,G. (1996) Addition of destabilizing poly(A)-rich sequences to endonuclease cleavage sites during the degradation of chloroplast mRNA. Proc. Natl Acad. Sci. USA, 93, 13398–13403.
Kudla,J., Hayes,R. and Gruissem,W. (1996) Polyadenylation accelerates degradation of chloroplast RNA. EMBO J., 15, 7137–7146.
Gagliardi,D. and Leaver,C.J. (1999) Polyadenylation accelerates the degradation of the mitochondrial mRNA associated with cytoplasmic male sterility in sunflower. EMBO J., 18, 3757–3766.
Lupold,D.S., Caoile,A.G.F.S. and Stern,D.B. (1999) Polyadenylation occurs at multiple sites in maize mitochondrial cox2 mRNA and is independent of editing status. Plant Cell, 11, 1565–1577.
Kuhn,J., Tengler,U. and Binder,S. (2001) Transcript lifetime is balanced between stabilizing stem–loop structures and degradation-promoting polyadenylation in plant mitochondria. Mol. Cell. Biol., 21, 731–742.
Williams,M.A., Johzuka,Y. and Mulligan,R.M. (2000) Addition of non-genomically encoded nucleotides to the 3'-terminus of maize mitochondrial mRNAs: truncated rps12 mRNAs frequently terminate with CCA. Nucleic Acids Res., 28, 4444–4451.
Lisitsky,I., Kotler,A. and Schuster,G. (1997) The mechanism of preferential degradation of polyadenylated RNA in the chloroplast. J. Biol. Chem., 272, 17648–17653.
Lisitsky,I., Klaff,P. and Schuster,G. (1997) Blocking polyadenylation of mRNA in the chloroplast inhibits its degradation. Plant J., 12, 1173–1178.
Gagliardi,D., Perrin,R., Maréchal-Drouard,L., Grienenberger,J.-M. and Leaver,C.J. (2001) Plant mitochondrial polyadenylated mRNAs are degraded by a 3'- to 5'-exoribonuclease activity, which proceeds unimpeded by stable secondary structures. J. Biol. Chem., 276, 43541–43547.
Cazenave,C. and Uhlenbeck,O.C. (1994) RNA template-directed RNA synthesis by T7 RNA polymerase. Proc. Natl Acad. Sci. USA, 91, 6972–6976.
Lerbs-Mache,S. (1993) The 110-kDa polypeptide of spinach plastid DNA-dependent RNA polymerase: single-subunit enzyme or catalytic core of multimeric enzyme complexes? Proc. Natl Acad. Sci. USA, 90, 5509–5513.
Hedtke,B., B?rner,T. and Weihe,A. (1997) Mitochondrial and chloroplast phage-type RNA polymerases in Arabidopsis. Science, 277, 809–811.
Hess,W.R. and B?rner,T. (1999) Organellar RNA polymerases of higher plants. Int. Rev. Cytol., 190, 1–59.
Wassarman,K.M., Repoila,F., Rosenow,C., Storz,G. and Gottesman,S. (2001) Identification of novel small RNAs using comparative genomics and microarrays. Genes Dev., 15, 1637–1651.
Massé,E. and Gottesmann,S. (2002) A small RNA regulates the expression of genes involved in iron metabolism in Escherichia coli. Proc. Natl Acad. Sci. USA, 99, 4620–4625.
Igloi,G.L. and K?ssel,H. (1992) The transcriptional apparatus of chloroplasts. Crit. Rev. Plant Sci., 10, 525–558.
Hajdukiewicz,P.T.J., Allison,L.A. and Maliga,P. (1997) The two RNA polymerases encoded by the nuclear and the plastid compartments transcribe distinct groups of genes in tobacco plastids. EMBO J., 16, 4041–4048.
Chang,C.-H. and Wu,M. (2000) The effects of transcription and RNA processing on the initiation of chloroplast DNA replication in Chlamydomonas reinhardtii. Mol. Gen. Genet., 263, 320–327.(Aitor Zandueta-Criado and Ralph Bock*)
*To whom correspondence should be addressed. Tel: +49 251 83 24790/91; Fax: +49 251 83 28371; Email: rbock@uni-muenster.de
Present address:
Aitor Zandueta-Criado, Instituto de Agrobiotecnología y Recursos Naturales, Universidad Pública de Navarra/Consejo Superior de Investigaciones Científicas, Ctra. de Mutilva s/n, Mutilva Baja 31192 Navarra, Spain
ABSTRACT
RNA editing in higher plant plastids is a post- transcriptional RNA maturation process changing single cytidine nucleotides into uridine. In the ndhD transcript of tobacco and several other plant species, editing of an ACG codon to a standard AUG initiator codon is believed to be a prerequisite for translation. In order to test this assumption experimentally, we have analyzed the editing status of ndhD mRNA species in the process of translation. We show that unedited ndhD transcripts are also associated with polysomes in vivo, suggesting that they are translated. This surprising finding challenges the view that ACG to AUG editing is strictly required to make the ndhD message translatable and raises the possibility that ACG can be utilized as an initiator codon in chloroplasts. In addition, we have mapped the termini of the ndhD transcript and discovered a novel form of RNA processing. Unexpectedly, we find that highly specific sequences are added to the 3' end of the ndhD mRNA at high frequency. We propose a model in which these sequences are added by the successive action of a CCA-adding enzyme (tRNA nucleotidyltransferase) and an RNA-dependent RNA polymerase (RdRp) activity. The presence of an RdRp activity may have general implications also for other steps in plastid gene expression.
Introduction
RNA editing alters the identity of single nucleotides in organellar transcripts of higher plants, mainly by cytidine to uridine conversion . In most instances, these changes are functionally significant in that codons for conserved amino acid residues are restored, and lack of RNA editing would result in impaired protein function (9,10).
Editing in plant cell organelles is a post-transcriptional process and largely independent of other RNA processing steps, such as intron splicing and cutting of polycistronic precursor transcripts into oligocistronic or monocistronic mRNAs (11–15). In contrast, the relationship between RNA editing and translation is much less clear. There seems to be no uniform way in which the organellar translational apparatus deals with unedited or partially edited mRNAs. Using specific antibodies against polypeptides reflecting edited or unedited mRNAs, plant mitochondria were assayed for the presence of proteins resulting from the translation of incompletely edited transcripts. Interestingly, translation products made from both edited and unedited mRNAs accumulated for a ribosomal protein, Rps12 (16,17). However, only the ribosomal protein S12 resulting from fully edited rps12 transcripts appears to become incorporated into ribosomes, and whether or not the other S12 species serve some function remains unclear. In contrast, for the ATPase subunit Atp6 (18) and another ribosomal protein, Rps13 (19), proteins made from unedited transcripts could not be detected. This is most probably due to their rapid degradation rather than exclusion of unedited transcripts from translation. In accordance with this assumption, both fully edited and partially edited mRNAs were found associated with mitochondrial polysomes (20). Similarly, the plastid translational apparatus does not seem to discriminate against unedited transcripts (21). In contrast to mRNAs with internal editing sites, it has been generally assumed that the situation was much clearer with editing events creating initiation codons (by changing a genomically encoded ACG triplet into a standard AUG start codon). Here, it seemed conceivable that unedited transcripts would be excluded from translation. This assumption was supported, at least for two cases, by transgenic experiments (22) and in vitro translation data (23). However, recent data have cast considerable doubt on the strict requirement of an AUG codon for translation to initiate in both plant mitochondria and plastids. When the AUG start codon was mutated in a chloroplast mRNA of the unicellular green alga Chlamydomonas reinhardtii, translation, though being significantly reduced, was not abolished and still initiated at the correct site (24). In addition, the radish (Raphanus sativus) mitochondrial cox2 gene was recently found to contain an ACG codon at the predicted translation initiation site which does not undergo editing into a standard AUG initiator codon (25). Similarly, the start codon of the ndhD message in Nicotiana tabacum and Nicotiana sylvestris is created by ACG to AUG editing, whereas, in Nicotiana tomentosiformis, the ACG remains unedited (26). However, in both plastids and mitochondria, experimental evidence for translation of mRNAs with unedited ACG start codons is lacking.
In this study, we have analyzed the relationship between RNA editing and translation for the plastid ndhD mRNA. The ndhD gene specifies a subunit of a plastid NADH dehydrogenase and is unique in that a significant proportion of its transcripts carries an unedited initiation codon . In fact, only between 30 and 60% of the ndhD transcripts carry the edited AUG codon which has been suggested to be required for initiation of ndhD translation (23,27). Here we show that also the unedited ACG-containing ndhD transcripts are heavily loaded with ribosomes, suggesting that they are actively translated. In addition, by analyzing the 5' and 3' termini of mature ndhD mRNAs, we have discovered that the 3' ends frequently carry stretches of highly specific non-encoded nucleotides which presumably are added by an RNA-dependent RNA polymerase (RdRp) activity.
Materials and Methods
Oligonucleotides
The oligonucleotides used were as follows: ndhD3', 5'-CAA GGTCGAAGTTATTCTATC-3'; ndhD5', 5'-TGGTCCAA GTGTATCTTGTC-3'; ndhDi, 5'-AGAAAAACAGAAC CCCCC-3'; and ndhDs, 5'-GAAAATTAAGGAACCCGC-3'. Their location and orientation are depicted in Figure 1.
Figure 1. Experimental strategy towards mapping the 5' and 3' ends and analyzing the RNA editing status of plastid ndhD transcripts. In the upper panel, the location and orientation of primers for cDNA synthesis and PCR are shown relative to the ndhD coding region. Relevant restriction sites for cloning are indicated. Transcripts are self-ligated with T4 RNA ligase, thereby fusing their 5' and 3' ends to produce circularized mRNA molecules. After cDNA synthesis primed with an ndhD-specific oligonucleotide, the region containing the 5' UTR, 3' UTR and the RNA editing site within the ndhD start codon is amplified by PCR. Products are then cloned and individual clones are sequenced to determine the termini of the mRNAs and the editing status of the start codon.
Isolation of polysome fractions and polysome-associated RNAs
Polysomes and polysomal RNAs were purified as described previously (28) with the following modifications: Tobacco leaves (2 g) were ground in liquid nitrogen and treated with 8 ml of ice-cold polysome extraction buffer. Fractions of 400 μl were collected from the analytical sucrose gradients. Prior to RNA isolation, all fractions were diluted with 280 μl of distilled water. Isopropanol precipitation was carried out without the addition of glycogen. Subsequently, RNA was washed with 70% ethanol and pellets were resuspended in 20 μl of sterile distilled water.
Northern blot and hybridization procedures
Aliquots of 4 μl of RNA extracted from polysome fractions were electrophoresed on formaldehyde-containing 1% agarose gels and transferred onto Hybond-XL membranes (Amersham). For hybridization, an dCTP-labeled probe was generated by random priming (Amersham) following the manufacturer’s instructions. An ndhD-specific probe was prepared by PCR amplification with primers ndh5' and ndhDi using total plant DNA as a template.
RNA circularization, cDNA synthesis and PCR
A 2 μl aliquot of RNA from fractions P1 and P6 (EDTA-free polysome gradients) and 0.5 μl of RNA from fraction C10 (EDTA-containing polysome gradient) were ligated at 37°C for 1 h with 20 U of T4 RNA ligase (New England Biolabs) in a final volume of 100 μl. cDNA synthesis was performed as follows. A 1 μg aliquot of RNA (precipitated with ethanol after ligation and redissolved in water) and 38 ng of oligonucleotide ndhDi were denatured for 5 min at 70°C in a total volume of 14 μl and then cooled on ice. Subsequently, reaction buffer and dNTPs (final concentration 2 mM each) were added and the volume was adjusted to 24 μl. The reverse transcription reaction was initiated by addition of 16 U of M-MLV reverse transcriptase (Promega) and was allowed to proceed at room temperature for 10 min followed by incubation at 45°C for 1 h. The resulting cDNA templates were amplified according to standard PCR protocols (30 s at 93°C, 1 min 30 s at 52°C, 1 min 30 s at 72°C; 30 cycles). Primers ndhDs and ndhD3' were used for specific amplification of the head-to-tail ligated 5' and 3' untranslated regions (UTRs). Amplification products were separated by agarose gel electrophoresis and the 300–400 bp band was excised and purified.
Cloning and sequencing
Purified amplification products were digested with StyI and XmnI and ligated into a pBluescript(SK+)-derived plasmid (carrying a cloned DNA fragment providing a CCTTGG StyI site) cut with StyI and EcoRV. Individual clones were sequenced with the M13 forward primer using the dideoxy chain termination method.
RESULTS AND DISCUSSION
A strategy for simultaneous mapping of transcript termini and determining the editing status of ndhD mRNAs
A variety of methods is available for the precise mapping of transcript termini, including primer extension (for 5' mapping), RNase protection assays, RNA ligation with anchor oligonucleotides and RNA circularization producing head-to-tail ligated transcripts (30). To determine both the 5' and the 3' ends of mature ndhD transcripts and to simultaneously analyze the RNA editing site within the start codon in individual mRNA molecules, we chose the experimental strategy illustrated in Figure 1. In order to exclude artifacts that could be experimentally introduced at the mRNA termini by the possible non-specific action of reverse transcriptases or DNA polymerases, we decided to circularize the RNA molecules prior to any treatment with nucleotide polymerizing enzymes (Fig. 1). Head-to-tail ligated mRNAs were then reverse transcribed using a gene-specific primer followed by amplification of the region containing the fused UTRs and the editing site-containing start codon (Fig. 1). As we were interested in characterizing mature translated mRNAs, we purified polysomes from tobacco leaves and fractionated polysomal RNAs in sucrose gradients. This polysomal RNA was then circularized and subsequently treated as shown in Figure 1.
Polysome association of plastid ndhD mRNAs
Polysome fractionation on sucrose gradients separates mRNAs according to the number of ribosomes bound to individual RNA molecules: mRNAs carrying large numbers of ribosomes migrate to the bottom of the gradient, whereas mRNAs displaying no or poor loading with ribosomes stay in the top fractions of the gradient (Fig. 2). This fractionation pattern is nicely seen with polysomes from tobacco leaves (Fig. 2; lower left panel). In contrast, control gradients centrifuged in the presence of EDTA (which causes release of ribosomes from the mRNAs) displayed no RNA migration into the bottom fractions of the gradient (Fig. 2; upper left panel), confirming successful mRNA fractionation according to ribosome number.
Figure 2. Detection of ndhD transcripts in polysome preparations from tobacco leaves. Polysomes were fractionated in sucrose gradients in either the presence (upper panel; control) or absence (lower panel) of EDTA. RNA samples from individual gradient fractions were analyzed on ethidium bromide-stained gels (left) and assayed for the presence of ndhD mRNAs by northern blotting and hybridization to an ndhD-specific probe (right). The presence of ndhD transcripts only in the upper fractions (9–12) of the EDTA gradient indicates release of ribosomes from mRNAs as caused by EDTA addition. In contrast, the EDTA-free gradient shows migration of large amounts of ndhD mRNAs into the bottom fractions. The prominent band at 2.5 kb represents dicistronic psaC–ndhD transcripts, whereas the 1.7 kb band consists of monocistronic ndhD message (31,32).
In order to detect ndhD mRNA species that are actively translated, RNAs extracted from sucrose gradient fractions were assayed for the presence of ndhD transcripts by northern blot analysis. The ndhD gene is part of an operon with the functionally unrelated psaC cistron (encoding a photosystem I subunit) located immediately upstream. Compared with psaC, ndhD mRNAs accumulate to rather low levels (31), consistent with the presumably much lower demand for NADH dehydrogenase subunits than for photosystem I proteins. It has been observed previously that the transcript pattern of ndhD is highly complex, with both dicistronic psaC–ndhD and monocistronic ndhD mRNAs accumulating in addition to numerous larger polycistronic precursor transcripts (31,32). When assaying polysome-associated mRNAs, we detected a similarly complex ndhD transcript pattern with a prominent band for the psaC–ndhD dicistronic mRNA (2.5 kb) in addition to somewhat less abundant transcript species representing monocistronic (1.7 kb) and several polycistronic mRNAs (Fig. 2). The dicistronic and polycistronic species are of limited interest for the purpose of this study since their ribosome association could be attributed to translation of the adjacent cistrons and does not necessarily suggest active translation of ndhD. We therefore focused our further studies on monocistronic ndhD transcripts whose isolation from polysomal fractions suggests their translation.
Multiple 5' and 3' ends of translated ndhD mRNAs
In order to precisely map the 5' and 3' termini of translated ndhD mRNAs, RNA samples from three different polysome gradient fractions were selected: fractions 1 and 6 from the EDTA-free polysome gradient (subsequently referred to as P1 and P6) and fraction 10 from the EDTA-containing control gradient (subsequently referred to as C10; Fig. 2). Using the strategy outlined in Figure 1, PCR with the ndhD-specific primers ndhDs and ndhD3' resulted in successful amplification of the head-to-tail ligated 3' and 5' UTRs of monocistronic ndhD transcripts. Since in PCRs, the amplification of small molecules is strongly favored over the amplification of larger molecules, only products corresponding to monocistronic message were obtained, in spite of the higher abundance of polycistronic transcript species (Fig. 2 and data not shown). Following cloning into a plasmid vector (Fig. 1), 16–22 individual clones were sequenced for each of the three samples (P1, P6 and C10).
Previous mapping of the 5' ends of ndhD transcripts by primer extension and RNase protection analyses had revealed the presence of multiple 5' ends (31,32). Our sequence analyses of individual head-to-tail ligated ndhD-derived cDNAs confirmed the presence of multiple 5' termini which all mapped inside the intercistronic spacer region in between ndhD and the upstream psaC (Fig. 3). As our RNA circularization strategy allowed the simultaneous mapping of 3' ends, which had not been determined previously for ndhD, we were interested in identifying processing sites within the 3' UTR in relation to possible secondary structure elements. Most 3' UTRs of plastid transcripts harbor stem–loop-type inverted repeats which act as RNA processing and stabilizing elements but do not terminate transcription (33,34). Indeed, we could identify such a putative stem–loop-type RNA secondary structure element also within the ndhD 3' UTR (Fig. 3). Most ndhD transcripts from polysomal fractions were found to terminate 2–6 nt downstream of the stem–loop (Fig. 3), strongly supporting that this secondary structure is relevant in vivo. The termini of the remaining clones were found to map inside the stem (Fig. 3), which could indicate that the corresponding RNA molecules are condemned to degradation (since they lack the putative transcript-stabilizing secondary structure). Decay of plastid transcripts frequently initiates from an endonucleolytic cleavage within the 3' UTR removing the protecting stem–loop structure . As psaC and ndhD are co-transcribed from an operon, but yet psaC accumulates to much higher levels than ndhD (31), a reasonable assumption could be that ndhD transcripts are turned-over more rapidly. Hence, the detection of RNA degradation intermediates may not be surprising. The majority of the 3'-truncated clones carried an unedited start codon (fraction P6 in Fig. 3) which could be indicative of ACG-containing transcripts being more susceptible to degradation than AUG-containing mRNAs.
Figure 3. Mapped ndhD mRNA termini and editing status of ndhD mRNAs. RNAs from fraction 10 of the EDTA-containing control gradient (C10; see Fig. 2) and fractions 1 and 6 of the EDTA-free gradient were circularized with RNA ligase and analyzed as displayed in Figure 1. Arrows indicate mapped termini, with the number of clones given for each terminus. Asterisks indicate ambiguity with respect to the terminal nucleotide (due to the presence of identical nucleotides at the putative 5' and 3' ends). The stem–loop indicates a potential secondary structure within the 3' UTR which is typical of plastid mRNAs. The number of unedited versus edited clones is given in parentheses above the initiation codon.
Translation of ndhD mRNAs with an unedited start codon
Having confirmed that we had successfully cloned head-to-tail ligated monocistronic ndhD transcript-derived cDNAs, we were most interested in analyzing the editing status of the start codon. If translation was indeed strictly dependent on editing of the ACG into a standard AUG initiator codon, then all polysome-loaded monocistronic ndhD mRNAs should harbor the edited AUG codon. Sequencing of 59 individual clones altogether from the three different gradient fractions (P1, P6 and C10; Fig. 2) revealed a large number of clones carrying the unedited ACG codon (38 clones equaling 64%), suggesting that unedited ndhD mRNAs are also translated. Moreover, the percentage of edited clones (36%) derived from polysomal mRNA is in the range of the previously determined efficiency of ndhD start codon editing , indicating that the chloroplast translational apparatus does not strongly discriminate against ACG-containing transcripts. However, the amount of edited clones was significantly higher in the bottom fraction (P1) of the polysome gradient (with 13 out of 21 clones being edited) than in the middle fraction (P6) of the gradient (with only two out of 22 clones being edited), indicating that AUG-containing mRNAs are, on average, loaded with more ribosomes. This finding could suggest that re-initiation of translation is more efficient at the AUG codon than at the ACG codon, and raises the possibility that editing is employed to regulate translation of the ndhD message. Consistent with such a translational regulation through RNA editing, it has been reported that, in N.tabacum, the editing efficiency at the ndhD start codon is strongly influenced by plastid differentiation, tissue type, plant development and the environmental factor light (23).
Translation of ndhD mRNA with an unedited ACG codon could also explain why this editing site remains completely unedited in N.tomentosiformis (26) and raises the possibility that, at least in the ndhD sequence context, ACG can be used as an initiator codon for plastid translation. However, at present, we cannot entirely rule out the possibility that translation of unedited ndhD mRNAs initiates at an alternative site and not from the unedited ACG codon. At least three different codons can be utilized as initiator codons in chloroplasts (AUG, GUG and UUG), and any such codon upstream or downstream of the ACG could, in theory, also serve as initiator codon.
Presence of non-encoded sequences in mature ndhD transcripts
Upon inspection of the sequences at the ligation site of the head-to-tail joined 5' and 3' UTRs of ndhD, we surprisingly discovered long stretches of non-encoded sequence (Figs 4 and 5). Significantly, these stretches were found in the majority of the analyzed clones (i.e. in 39 out of 59 clones). These non-encoded extra sequences, although differing in length, were highly homologous to each other (Fig. 5), strongly suggesting that they share a common mechanism of addition. Although the head-to-tail ligation approach does not allow us to distinguish between addition to the 5' UTR and addition to the 3' UTR, the 5' to 3' working direction of nucleotide polymerases strongly suggests that the extra sequences are added to the 3' end of the mRNA. (This assumption gained further support by our identification of the probable origin of the non-encoded sequence; see below and Figs 4–6.)
Figure 4. Example of a clone containing a stretch of non-encoded sequence at the 3' end. The complementary sequence motif upstream of the stem–loop structure is marked. The mapped 5' and 3' ends, the sequence of the stem–loop structure, the edited initiation codon and the StyI site used for cloning (see Fig. 1) are also indicated.
Figure 5. Presence of non-encoded sequences at the 3' end of plastid ndhD transcripts. The non-encoded sequences are aligned with a stretch of complementary sequence present upstream of the stem–loop structure. Arrows indicate mapped 3' termini and the corresponding number of clones. Nucleotides in parentheses were not found in all clones.
Figure 6. Model for the addition of 3'-non-encoded sequences to plastid ndhD transcripts. The frequent presence of CCA-like sequences suggests initial involvement of the CCA-adding enzyme (tRNA nucleotidyltransferase) which may recognize the stem–loop because of structural similarity with its regular tRNA substrate. Complementarity of the free CCA end with the sequence immediately upstream of the stem–loop structure may prime RNA-dependent RNA polymerization using the mRNA as a template. The RNA-dependent RNA polymerase (RdRp) activity does not necessarily have to come from a specific RdRp enzyme, but also could be a hitherto undetected side activity of normally DNA-dependent RNA polymerases, such as the chloroplast transcription enzymes PEP and NEP (48,52,53) or the primase for plastid DNA replication .
The presence of the non-encoded nucleotides cannot be attributed to an experimentally induced artifact since the mRNAs were circularized prior to any contact with nucleotide-polymerizing activities and (d)NTPs (Fig. 1).
To date, two types of non-encoded nucleotide stretches have been found at the 3' end of organellar RNAs: (i) homopolymeric stretches of mostly A nucleotides resulting in polyadenylated 3' UTRs; and (ii) short CCA-like additions pointing to an involvement of CCA-adding enzyme (tRNA nucleotidyltransferase). The former are found in both plastids (35,37,38) and mitochondria (39–41), whereas the latter so far have been reported only for plant mitochondria (42). Polyadenylation is generally believed to mark RNA molecules for degradation, which is supported by polyadenylated plastid and mitochondrial RNAs being highly unstable (38,43–45). In contrast, the 3' addition of relatively long, highly specific and non-homopolymeric sequence stretches to an mRNA, as seen in the majority of our ndhD clones, has not been observed before.
To us, it seemed unlikely that such long specific sequences are synthesized without using a template. We therefore searched for homologies of the non-encoded sequence stretch with other sequences in the tobacco chloroplast genome. Interestingly, we found a perfectly complementary sequence motif immediately upstream of the stem–loop structure within the ndhD 3' UTR (Figs 4 and 5). This finding strongly suggests that the sequence upstream of the stem–loop is used as a template to polymerize the extra sequence found at the transcript terminus. Such an RNA-dependent RNA polymerization has not been observed before in organellar RNA processing and, to our knowledge, no candidate enzymatic activity has been identified to date in either chloroplasts or plant mitochondria.
In some clones, the first 1–3 extra nucleotides were not complementary to the putative template sequence upstream of the stem–loop (Fig. 5). Interestingly, these nucleotides are always CCA (or CCA-like; Fig. 5). Moreover, the sequence at the junction between the 3' end and the non-encoded extra sequence is CCA-like in all clones (Fig. 5). This might indicate that this first part of the non-encoded sequence is added by an enzymatic activity similar or identical to the CCA-adding enzyme normally involved in tRNA maturation. Recently, non-encoded CCA-like sequences have been found at the 3' end of maize mitochondrial mRNAs, and it has been hypothesized that they are added by a tRNA nucleotidyltransferase-like activity (42).
A model that could explain the origin of the 3' non-encoded sequences found at the 3' end of ndhD mRNAs is presented in Figure 6. Briefly, addition of a CCA sequence by the CCA-adding enzyme (tRNA nucleotidyltransferase) would create a sequence complementarity between the 3' end and a UGG motif upstream of the stem–loop. In this way, RNA-dependent RNA polymerization could be primed, which then is likely to proceed by a different enzymatic activity (i.e. an RdRp). It seems unnecessary to postulate the existence of a separate RdRp enzyme in plastids, as DNA-dependent RNA polymerases can have some RdRp-like side activity. This has been demonstrated, for example, for T7 RNA polymerase (46). It seems noteworthy in this respect that chloroplasts possess, in addition to a plastid-encoded Escherichia coli-like RNA polymerase, also a nuclear-encoded T7-like RNA polymerase (47–49).
At present, we can only speculate about the possible functional significance of these extra sequences added to the 3' end of the ndhD message. As >80% of the stem–loop-containing ndhD transcripts carry such extensions, nucleotide addition appears to be a fairly efficient post-transcriptional process. Whether or not this nucleotide addition, for example, influences or even regulates transcript lifetime (e.g. by providing an extended stabilizing double-stranded RNA structure) remains to be investigated. Moreover, the occurrence of RNA template-directed RNA synthesis in plastids, as suggested by our data, may also have implications for other steps in plastid gene expression. Synthesis of antisense strands by RdRp and the resulting formation of stable double-stranded RNAs could, for example, affect translatability of plastid mRNAs, influence RNA degradation, control 3' processing or intercistronic RNA processing (RNA cutting), or even trigger RNA amplification. It is of note in this regard that antisense RNAs can be important regulators of gene expression in prokaryotes (50,51).
It will be interesting to see how widespread 3' processing by addition of non-encoded sequence stretches is in chloroplast (and perhaps also mitochondrial) mRNAs. Preliminary data suggest that not all chloroplast transcripts are processed in this manner. This may be unsurprising since not all 3' UTRs of chloroplast mRNAs harbor a UGG motif upstream of the stem—loop which, according to our proposed model, would facilitate priming of RNA-dependent polymerization (Figs 5 and 6). One intensely studied example is the petD mRNA 3' end where no such sequence additions have been detected .
Taken together, during the course of this work, we have provided evidence for translation of ndhD mRNAs carrying an unedited start codon and detected a novel form of RNA processing by addition of specific sequences to the 3' end, presumably through CCA addition followed by RNA- dependent RNA polymerization.
ACKNOWLEDGEMENTS
We thank Mrs Daniela Ahlert for DNA sequencing, Dr Javier Pozueta-Romero for valuable discussion and support, and Dr J?rg Kudla for critical reading of the manuscript. This work was financed by grants from the Deutsche Forschungsgemeinschaft to R.B. A.Z.C. was supported by a fellowship from the Ministerio de Educación Cultura y Deporte, Spain.
REFERENCES
Gualberto,J.M., Lamattina,L., Bonnard,G., Weil,J.-H. and Grienenberger,J.-M. (1989) RNA editing in wheat mitochondria results in the conservation of protein sequences. Nature, 341, 660–662.
Covello,P.S. and Gray,M.W. (1989) RNA editing in plant mitochondria. Nature, 341, 662–666.
Hiesel,R., Wissinger,B., Schuster,W. and Brennicke,A. (1989) RNA editing in plant mitochondria. Science, 246, 1632–1634.
Hoch,B., Maier,R.M., Appel,K., Igloi,G.L. and K?ssel,H. (1991) Editing of a chloroplast mRNA by creation of an initiation codon. Nature, 353, 178–180.
Kudla,J., Igloi,G.L., Metzlaff,M., Hagemann,R. and K?ssel,H. (1992) RNA editing in tobacco chloroplasts leads to the formation of a translatable psbL mRNA by a C to U substitution within the initiation codon. EMBO J., 11, 1099–1103.
Mulligan,R.M., Williams,M.A. and Shanahan,M.T. (1999) RNA editing site recognition in higher plant mitochondria. J. Hered., 90, 338–344.
Bock,R. (2000) Sense from nonsense: how the genetic information of chloroplasts is altered by RNA editing. Biochimie, 82, 549–557.
Bock,R. (2001) RNA editing in plant mitochondria and chloroplasts. In Bass,B. (ed.), Frontiers in Molecular Biology: RNA Editing. Oxford University Press, pp. 38–60.
Bock,R., K?ssel,H. and Maliga,P. (1994) Introduction of a heterologous editing site into the tobacco plastid genome: the lack of RNA editing leads to a mutant phenotype. EMBO J., 13, 4623–4628.
Zito,F., Kuras,R., Choquet,Y., K?ssel,H. and Wollman,F.-A. (1997) Mutations of cytochrome b6 in Chlamydomonas reinhardtii disclose the functional significance for a proline to leucine conversion by petB editing in maize and tobacco. Plant Mol. Biol., 33, 79–86.
Sutton,C.A., Conklin,P.L., Pruitt,K.D. and Hanson,M.R. (1991) Editing of pre-mRNAs can occur before cis- and trans-splicing in Petunia mitochondria. Mol. Cell. Biol., 11, 4274–4277.
Yang,A.J. and Mulligan,R.M. (1991) RNA editing intermediates of cox2 transcripts in maize mitochondria. Mol. Cell. Biol., 11, 4278–4281.
Gualberto,J.M., Bonnard,G., Lamattina,L. and Grienenberger,J.M. (1991) Expression of the wheat mitochondrial nad3–rps12 transcription unit: correlation between editing and mRNA maturation. Plant Cell, 3, 1109–1120.
Ruf,S., Zeltz,P. and K?ssel,H. (1994) Complete RNA editing of unspliced and dicistronic transcripts of the intron-containing reading frame IRF170 from maize chloroplasts. Proc. Natl Acad. Sci. USA, 91, 2295–2299.
Freyer,R., Hoch,B., Neckermann,K., Maier,R.M. and K?ssel,H. (1993) RNA editing in maize chloroplasts is a processing step independent of splicing and cleavage to monocistronic mRNAs. Plant J., 4, 621–629.
Phreaner,C.G., Williams,M.A. and Mulligan,R.M. (1996) Incomplete editing of rps12 transcripts results in the synthesis of polymorphic polypeptides in plant mitochondria. Plant Cell, 8, 107–117.
Lu,B., Wilson,R.K., Phreaner,C.G., Mulligan,R.M. and Hanson,M.R. (1996) Protein polymorphism generated by differential RNA editing of a plant mitochondrial rps12 gene. Mol. Cell. Biol., 16, 1543–1549.
Lu,B. and Hanson,M.R. (1994) A single homogeneous form of ATP6 protein accumulates in petunia mitochondria despite the presence of differentially edited atp6 transcripts. Plant Cell, 6, 1955–1968.
Williams,M.A., Tallakson,W.A., Phreaner,C.G. and Mulligan,R.M. (1998) Editing and translation of ribosomal protein S13 transcripts: unedited translation products are not detectable in maize mitochondria. Curr. Genet., 34, 221–226.
Lu,B. and Hanson,M.R. (1996) Fully edited and partially edited nad9 transcripts differ in size and both are associated with polysomes in potato mitochondria. Nucleic Acids Res., 24, 1369–1374.
Hirose,T., Kusumegi,T. and Sugiura,M. (1998) Translation of tobacco chloroplast rps14 mRNA depends on a Shine–Dalgarno-like sequence in the 5'-untranslated region but not on internal RNA editing in the coding region. FEBS Lett., 430, 257–260.
Chaudhuri,S. and Maliga,P. (1996) Sequences directing C to U editing of the plastid psbL mRNA are located within a 22 nucleotide segment spanning the editing site. EMBO J., 15, 5958–5964.
Hirose,T. and Sugiura,M. (1997) Both RNA editing and RNA cleavage are required for translation of tobacco chloroplast ndhD mRNA: a possible regulatory mechanism for the expression of a chloroplast operon consisting of functionally unrelated genes. EMBO J., 16, 6804–6811.
Chen,X., Kindle,K.L. and Stern,D.B. (1995) The initiation codon determines the efficiency but not the site of translation initiation in Chlamydomonas chloroplasts. Plant Cell, 7, 1295–1305.
Dong,F.G., Wilson,K.G. and Makaroff,C.A. (1998) The radish (Raphanus sativus L.) mitochondrial cox2 gene contains an ACG at the predicted translation initiation site. Curr. Genet., 34, 79–87.
Sasaki,T., Yukawa,Y., Miyamoto,T., Obokata,J. and Sugiura,M. (2003) Identification of RNA editing sites in chloroplast transcripts from the maternal and paternal progenitors of tobacco (Nicotiana tabacum): comparative analysis shows the inolvement of distinct trans-factors for ndhB editing. Mol. Biol. Evol., 20, 1028–1035.
Neckermann,K., Zeltz,P., Igloi,G.L., K?ssel,H. and Maier,R.M. (1994) The role of RNA editing in conservation of start codons in chloroplast genomes. Gene, 146, 177–182.
Ruf,S., K?ssel,H. and Bock,R. (1997) Targeted inactivation of a tobacco intron-containing open reading frame reveals a novel chloroplast-encoded photosystem I-related gene. J. Cell Biol., 139, 95–102.
Liu,X. and Gorovsky,M.A. (1993) Mapping the 5' and 3' ends of Tetrahymena thermophila mRNAs using RNA ligase mediated amplification of cDNA ends (RLM-RACE). Nucleic Acids Res., 21, 4954–4960.
Kuhn,J. and Binder,S. (2002) RT–PCR analysis of 5' to 3'-end-ligated mRNAs identifies the extremities of cox2 transcripts in pea mitochondria. Nucleic Acids Res., 30, 439–446.
delCampo,E.M., Sabater,B. and Martin,M. (2002) Post-transcriptional control of chloroplast gene expression. J. Biol. Chem., 277, 36457–36464.
Matsubayashi,T., Wakasugi,T., Shinozaki,K., Yamaguchi-Shinozaki,K., Zaita,N., Hidaka,T., Meng,B.Y., Ohto,C., Tanaka,M., Kato,A., Maruyama,T. and Sugiura,M. (1987) Six chloroplast genes (ndhA–F) homologous to human mitochondrial genes encoding components of the respiratory chain NADH dehydrogenase are actively expressed: determination of the splice site in ndhA and ndhB pre-mRNAs. Mol. Gen. Genet., 210, 385–393.
Stern,D.B. and Gruissem,W. (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.
Hayes,R., Kudla,J., Schuster,G., Gabay,L., Maliga,P. and Gruissem,W. (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.
Hayes,R., Kudla,J. and Gruissem,W. (1999) Degrading chloroplast mRNA: the role of polyadenylation. Trends Biochem. Sci., 24, 199–202.
Schuster,G. and Bock,R. (2001) Editing, polyadenylation and degradation of mRNA in the chloroplast. In Aro,E.-M. and Andersson,B. (eds), Advances in Photosynthesis and Respiration. Regulation of Photosynthesis. Kluwer Academic Press, Dordrecht, The Netherlands, Vol. 11, pp. 121–136.
Lisitsky,I., Klaff,P. and Schuster,G. (1996) Addition of destabilizing poly(A)-rich sequences to endonuclease cleavage sites during the degradation of chloroplast mRNA. Proc. Natl Acad. Sci. USA, 93, 13398–13403.
Kudla,J., Hayes,R. and Gruissem,W. (1996) Polyadenylation accelerates degradation of chloroplast RNA. EMBO J., 15, 7137–7146.
Gagliardi,D. and Leaver,C.J. (1999) Polyadenylation accelerates the degradation of the mitochondrial mRNA associated with cytoplasmic male sterility in sunflower. EMBO J., 18, 3757–3766.
Lupold,D.S., Caoile,A.G.F.S. and Stern,D.B. (1999) Polyadenylation occurs at multiple sites in maize mitochondrial cox2 mRNA and is independent of editing status. Plant Cell, 11, 1565–1577.
Kuhn,J., Tengler,U. and Binder,S. (2001) Transcript lifetime is balanced between stabilizing stem–loop structures and degradation-promoting polyadenylation in plant mitochondria. Mol. Cell. Biol., 21, 731–742.
Williams,M.A., Johzuka,Y. and Mulligan,R.M. (2000) Addition of non-genomically encoded nucleotides to the 3'-terminus of maize mitochondrial mRNAs: truncated rps12 mRNAs frequently terminate with CCA. Nucleic Acids Res., 28, 4444–4451.
Lisitsky,I., Kotler,A. and Schuster,G. (1997) The mechanism of preferential degradation of polyadenylated RNA in the chloroplast. J. Biol. Chem., 272, 17648–17653.
Lisitsky,I., Klaff,P. and Schuster,G. (1997) Blocking polyadenylation of mRNA in the chloroplast inhibits its degradation. Plant J., 12, 1173–1178.
Gagliardi,D., Perrin,R., Maréchal-Drouard,L., Grienenberger,J.-M. and Leaver,C.J. (2001) Plant mitochondrial polyadenylated mRNAs are degraded by a 3'- to 5'-exoribonuclease activity, which proceeds unimpeded by stable secondary structures. J. Biol. Chem., 276, 43541–43547.
Cazenave,C. and Uhlenbeck,O.C. (1994) RNA template-directed RNA synthesis by T7 RNA polymerase. Proc. Natl Acad. Sci. USA, 91, 6972–6976.
Lerbs-Mache,S. (1993) The 110-kDa polypeptide of spinach plastid DNA-dependent RNA polymerase: single-subunit enzyme or catalytic core of multimeric enzyme complexes? Proc. Natl Acad. Sci. USA, 90, 5509–5513.
Hedtke,B., B?rner,T. and Weihe,A. (1997) Mitochondrial and chloroplast phage-type RNA polymerases in Arabidopsis. Science, 277, 809–811.
Hess,W.R. and B?rner,T. (1999) Organellar RNA polymerases of higher plants. Int. Rev. Cytol., 190, 1–59.
Wassarman,K.M., Repoila,F., Rosenow,C., Storz,G. and Gottesman,S. (2001) Identification of novel small RNAs using comparative genomics and microarrays. Genes Dev., 15, 1637–1651.
Massé,E. and Gottesmann,S. (2002) A small RNA regulates the expression of genes involved in iron metabolism in Escherichia coli. Proc. Natl Acad. Sci. USA, 99, 4620–4625.
Igloi,G.L. and K?ssel,H. (1992) The transcriptional apparatus of chloroplasts. Crit. Rev. Plant Sci., 10, 525–558.
Hajdukiewicz,P.T.J., Allison,L.A. and Maliga,P. (1997) The two RNA polymerases encoded by the nuclear and the plastid compartments transcribe distinct groups of genes in tobacco plastids. EMBO J., 16, 4041–4048.
Chang,C.-H. and Wu,M. (2000) The effects of transcription and RNA processing on the initiation of chloroplast DNA replication in Chlamydomonas reinhardtii. Mol. Gen. Genet., 263, 320–327.(Aitor Zandueta-Criado and Ralph Bock*)