(Received for publication, February 11, 1997, and in revised form, March 19, 1997)
From the Laboratoire de Génétique Moléculaire des Plantes, Université Joseph Fourier and CNRS, B. P. 53, F-38041 Grenoble, France
The spinach rrn operon is used as a model system to study transcriptional regulation in higher plant photosynthetic and non-photosynthetic plastids. We performed capping experiments to determine whether P1, PC, or P2 promoters are employed for rrn transcription start sites in cotyledon and root tissues. By using a new method of analysis of capped RNA we demonstrate for the first time that 1) in both organs the rrn operon is expressed in a constitutive manner by cotranscription with the preceding tRNA(GAC)Val gene, and 2) the PC transcription start site is used only in cotyledons and leaves, i.e. we demonstrate the organ-specific usage of a plastid promoter. Both start sites, PC and that of the tRNA(GAC)Val cotranscript, lack Escherichia coli-like consensus sequences. The cotranscript is initiated 457 base pairs upstream of the tRNA(GAC)Val gene. The PC-specific DNA-binding factor, CDF2, is not detectable in root tissues confirming its regulatory role in PC-initiated rrn expression and the organ specificity of PC expression. Furthermore, our results show that rrn operon expression patterns differ in spinach and tobacco indicating species-specific transcriptional regulation of plant plastid gene expression.
In accordance with the hypothesis that plastids are of endosymbiotic origin most plastid genes are organized into polycistronic transcription units reminiscent of bacterial operons. Plastid rRNA operons show the typical procaryotic gene order of 16 S, 23 S, and 5 S rDNA. These genes are transcribed as large precursor RNAs that are subsequently processed into the various mature rRNA species (1, 2).
The promoter regions of plastid rrn operons harbor Escherichia coli-like "-10" and "-35" consensus sequences, like most of the plastid transcription units. These E. coli-like consensus sequences serve as promoter structures (3-5) or as regulatory elements (6, 7). However, the interpretation of results on studies of transcriptional regulation in plastids is complicated by the existence of different types of RNA polymerases. One is nuclear encoded (8-11) and the other one is encoded on the plastid genome (12-15). The plastid-encoded enzyme can be considered as "E. coli-like" with respect to its subunit composition (16, 17) and promoter usage (6, 18-20). The composition of the nuclear-encoded RNA polymerase and the promoter structures that are used by this enzyme are not yet clear although several potential transcription start sites for this enzyme have been mapped (5, 21, 22).
In spinach, the rrn operon upstream region contains three different promoter elements (P1, PC, P2), and transcription is thought to be regulated by the transcription factor CDF2 (23). CDF2 acts as a repressor of rRNA transcription by the E. coli-like plastid RNA polymerase and probably as an activator of rRNA transcription by the nuclear-encoded RNA polymerase (6), i.e. rRNA transcription could be regulated exclusively by CDF2. On the other hand, up to now correct initiation at the putative PC start site could not be demonstrated in vitro raising the question whether PC is indeed an initiation site. In addition, two transcription start sites that are different from the three of spinach have recently been reported for the tobacco rrn operon (5). They are differentially used in leaf chloroplasts or in amyloplasts of cultured cells. Considering the high sequence homology of the tobacco and the spinach rrn operon promoter regions, it is very surprising to find differences in rrn expression patterns between two closely related plants. Therefore, we analyzed the rrn expression patterns in both plants in parallel under the same experimental conditions. Also, we analyzed rrn expression of different plastid types in intact plants, because gene expression in cultured cells (in particular in BY2 cells, which are not competent for regeneration) might differ from that in intact tissues of plants.
In general, the rate of ribosome formation (including rRNA synthesis) is adapted to the individual cellular requirement for protein biosynthesis. Correspondingly, overall transcription rates of rrn operons change drastically in accordance with changes of cellular metabolic activities. For instance, in E. coli the rRNA transcription is subject to stringent and growth-rate control (for review see Ref. 24). In higher plants, overall transcription rates of the plastid genome change drastically in a developmental and organ-specific manner (25-27). The transcription rate is about 30 times higher in leaf chloroplasts than in root amyloplasts (27). This difference in gene expression is related to the photosynthetic activities of chloroplasts and correlates with the number of plastid ribosomes. Consequently, rRNA expression should be regulated on the transcriptional level to respond to changes in plastid metabolic activities.
To learn more about the regulation of rrn transcription in different plastid types of intact plants, we have analyzed rrn expression in spinach root and cotyledon tissues by capping and primer extension experiments to determine and compare transcription start sites and transcript quantities. We also analyzed protein/rrn promoter interactions using different protein extracts. Results are interpreted with respect to the presence of multiple RNA polymerases in plastids (for recent review see Ref. 28), and the rrn transcription patterns obtained from plastids of intact plants are compared with transcription patterns observed using cell cultures.
Spinach (Spinacia oleracea L. var. Geant d'hiver) was grown either in vermiculite at 15 and 22 °C (night/day) in a 10-h light/14-h dark cycle or in darkness or kept on moistened filter paper in darkness for 7 days and subsequently illuminated for 2-8 h. After 1 week, cotyledons and roots were collected, washed extensively (three times with sterile water), and homogenized for plastid purification. For nucleic acid analysis, the plant material was immediately frozen in aliquots of 2 g in liquid nitrogen. Tobacco leaves were taken from 45-day-old plants.
Nucleic Acid IsolationFrozen tissue (2 g) was ground in
liquid nitrogen until a very fine powder was obtained. The powder was
poured immediately into an ice-cold mixture of phenol/chloroform/buffer
(1:1:2; 10 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, and 1% SDS) for nucleic acid extraction. DNA was
removed by two successive LiCl precipitations. RNA was dissolved in
sterile water to a final concentration of 2 µg/µl and stored at
20 °C in 15-µg or 1-µg aliquots until usage.
Oligonucleotides
were 5-end-labeled with [
-32P]ATP and T4
polynucleotide kinase. The primer (105 cpm, 1-10 ng)
was annealed to 1 or 15 µg of the total RNA, and cDNA synthesis
was performed using 50 units of Moloney murine leukemia virus reverse
transcriptase (Boehringer Mannheim). The reaction products were
analyzed on 8% polyacrylamide, 7 M urea gels.
Dideoxy-sequencing reactions were performed on double-stranded plasmid
DNA using
-35S-ATP and the T7 sequencing kit from
Pharmacia Biotech Inc. Each primer extension series was performed in
parallel using different primer/RNA ratios to verify linearity of
signal responses.
For S1 nuclease mapping of capped RNA, 300 ng of single-stranded DNA
were hybridized to 15 µg of capped RNA and digested with 100 units of
S1 nuclease at 20 °C for 2 h if not otherwise indicated. Capping reactions were performed in 30-µl aliquots containing 15 µg
of total RNA, 150 µCi of [-32P]GTP, and 25 units of
guanylyltransferase (Life Technologies, Inc.) in a solution containing
50 mM Tris-HCl, pH 7.9, 1.25 mM MgCl2, 6 mM KCl, 1 mM
dithiothreitol at 37 °C for 30 min.
The
831-bp1 BglII-PvuII
plastid DNA fragment (30) was recloned into Bluescript KS+ after
excision by EcoRI from the PHp 34 plasmid. Clones were
selected containing the 16 S rRNA 5-end oriented in the opposite
direction of the
-galactosidase gene. The 831-bp EcoRI
fragment was further cleaved by HinPI and inserted into
Bluescript KS+ cleaved by EcoRI and AccI.
Plastid DNA was amplified using primers e and f
(primer e, 5-AAGATTTGGCTCGGCATG-3
; primer
f, 5
-CCATAGGTACAGCGTTTG-3
), and the corresponding
fragment was cloned into pCRTMII (Invitrogen) according to
the manufacturer's protocol.
50 ng of primer were annealed to 3 µg of capped RNA in a final volume of 10 µl (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM MgCl2, 10 mM dithiothreitol) for 10 min at the melting temperature for the primer. 1 µl of RNase H (5 units) was added, and the reaction was incubated at 37 °C for 20 min. The reaction was stopped by adding 10 µl of loading buffer and was directly run on a denaturing polyacrylamide gel. Control reactions are incubated for the same time without primer or without RNase H.
Gel Retardation AssaysPurification of root plastids was done according to Deng and Gruissem (27). Isolation of chloroplasts, preparation of protein extracts, and gel shift assays were performed as described (6).
Primer
extension assays were optimized to give quantitative results. Fig.
1A shows the analysis of cotyledon and root
RNA using a primer that anneals within the mature 16 S rRNA (primer a, 5-GGGCAGGTTCTTACGCGT-3
, for primer localization see
scheme of Fig. 2A). Radioactivity
incorporated into the cDNA that corresponds to mature 16 S rRNA was
counted. The obtained values indicated a 77-fold difference in rRNA
content in root and cotyledon tissues, which corresponds well to
previously reported results (27), and thus confirms that our assay
conditions are quantitative. In addition to the mature rRNA, two other
RNAs are revealed in cotyledon tissues (Fig. 1A, lane
C). The band named Pro2 corresponds in size to the processing
intermediate, which was previously characterized in maize and tobacco
chloroplasts (3, 29), suggesting that this processing site is universal
in plastids of different plant species. The band named PC corresponds
to the previously characterized putative transcription start site of
the spinach rrn operon (23). Interestingly, PC is not
detected in root tissues (lane R). Pro2 appears as a very
faint band. The intermediate sized transcript that is marked with an
asterisk is probably an artifact since it is not revealed with another
primer (compare with Fig. 1B, lane R). The
cDNA of about 900 bases indicates the presence of very long
precursor transcripts in root tissues. Such a long transcript should
include the tRNA(GAC)Val that is encoded upstream of the
rrn operon. Therefore, we named this site PtRNA. The band
corresponding to PtRNA is also detectable in cotyledon tissues after
longer exposure (not shown).
The quantitative difference of the PC transcripts in cotyledon and root
tissues was re-analyzed with the same RNA preparations using primer
g (5-TTCATAGTTGCATTACT-3
). The sequence of primer g is complementary to the rrn precursor region
immediately upstream of mature 16 S rRNA. Therefore, the majority of
the radiolabeled primer is not trapped by mature rRNA, and the
sensitivity of the method is amplified by more than 100 times. Also
under this condition, cDNA corresponding to PC initiation is not
detectable in root tissues (Fig. 1B, lane R). RNA
corresponding to Pro2 cannot be detected since the primer is positioned
too close to it. No other intermediate sized RNA is revealed, including
that corresponding to the product from the promoter for rrn
transcription in cultured tobacco cells (Ptobacco, see Ref.
5).
Pro2-cleaved precursor transcripts are abundant in cotyledon/leaf
tissues (Fig. 1A) and are present in newly assembled plastid ribosomes (29). To analyze the fate of the supposed
tRNA(GAC)Val-containing precursor transcript (PtRNA, Fig.
1A) after Pro2 cleavage, we used a primer that is located
upstream of processing site Pro2 for primer extension (Fig.
1C; primer c, 5-TCTTTCATTCCAAGGCATAACTTGT-3
). Again, PC is observed only in cotyledon tissues but not in root tissues
(lanes L, D, and R). In addition, a higher
molecular weight cDNA is revealed in roots and cotyledons (Pro1).
Fine mapping localizes the 5
-end of this transcript 125 nucleotides
upstream of the tRNA(GAC)Val gene (Fig. 1, D and
E, for sequence see Ref. 30). The amount of these
transcripts does not differ considerably in cotyledon tissues of
light-grown or dark-grown plantlets (Fig. 1C, lanes L and D) and in root tissues (lane R).
Next we addressed three questions. 1) Is the rrn operon indeed cotranscribed with the tRNAVal(GAC) gene, and is PtRNA the transcription start site for this large transcript? 2) If so, is PC a cotyledon/leaf-specific transcription start site or a cotyledon/leaf-specific processing site of the large transcript that starts at PtRNA? 3) Is Pro1 a processing site of the PtRNA-initiated precursor RNA or another transcription initiation start site that produces a transcript that is terminated upstream of Pro2?
Cotranscription of the tRNAVal(GAC) Gene with the rrn OperonTo ensure that PtRNA corresponds to cotranscription of the
tRNA(GAC)Val gene with the rrn operon we
analyzed root and cotyledon RNA by RT-PCR (Fig. 2B). The
cDNA was primed within the sequence of mature 16 S rRNA (primer
a), and for amplification a second primer was used that
anneals within the tRNA(GAC)Val gene (primer b,
5-GGTATAACTCAGCGGTAG-3
). RNA corresponding to cotranscription of the
tRNA(GAC)Val gene, and the rrn operon is
detected in roots and in cotyledons (Fig. 2, lanes 1 and
4). Controls were made omitting reverse transcription prior
to PCR amplification (lanes 2 and 5) or omitting
primer b during amplification (lanes 3 and
6). The corresponding 417-bp product was cloned and sequenced. The
5
and 3
parts of the sequence are shown in Fig. 2B on the
right-hand side.
RT-PCR was also performed with primer pairs b/c and
d/c (Fig. 2C; primer d,
5-GGGGTTGATCCGTATCAT-3
). Transcripts covering the sequences between
primers c and b are more abundant than
transcripts extending between primers c and d
(Fig. 2C) or a and b (Fig.
2B). This difference could be explained by a processing
event that takes place at Pro1. Pro2 has already been determined as a
processing site (3, 29).
Having confirmed the existence of transcripts extending from tRNA(GAC)Val downstream into the rrn precursor region, it is necessary to distinguish between transcription initiation and RNA-processing sites.
PC Is a Transcription Initiation Site and Pro1 Is a Processing SiteWith the method of primer extension, it is not possible to
distinguish whether the revealed RNA results from transcription initiation or from processing (see above and also Ref. 23). Transcription initiation can only be shown by in vitro
capping of RNA that contains a 5-triphosphate followed by positioning of the capped 5
-end of the RNA by S1 nuclease mapping. Therefore, we
analyzed cotyledon and root RNA after in vitro capping.
Complementary single-stranded DNA was produced from a cloned DNA
fragment that comprises 140 bp of the mature 16 S rRNA gene and 691 bp
of the upstream region (831-base probe in Fig. 2A; for
sequence see Ref. 30). We observe only one transcript in cotyledons
that contains a 5
-triphosphate (Fig. 3A, lane
1). This transcript corresponds from its size to initiation at PC.
In roots, we could not detect any transcript under the same
experimental conditions (lane 2).
To analyze the Pro1-corresponding transcript we cloned a 603-bp DNA fragment that comprises the Pro1 site and the tRNA(GAC)Val gene but does not extend to the Pro2-processing site (see Fig. 2A). The analysis of capped root (lane 1) and cotyledon (lane 2) RNA is shown in Fig. 3B. The two bands that resist S1 nuclease digestion (lane 2) correspond to the A and G nucleotides located 3 and 4 bases upstream of the primer extension signal (23). This difference of 3 and 4 bases can be explained by the addition of GTP to the analyzed RNA in the capping reaction and to the fact that cDNA (primer extension) and RNA (capping) migrate differently in polyacrylamide gels (31). Lane 3 shows the profile of the capped RNA without hybrid selection.
These results show that PC is a transcription start site of the 16 S rRNA in cotyledon/leaf plastids of spinach plantlets. In root amyloplasts, the 16 S rRNA is cotranscribed with the tRNA(GAC)Val gene. In cotyledon/leaf tissues, both types of transcription exist (cotranscription with tRNA(GAC)Val and transcription initiation at PC). The transcript corresponding to Pro1 seems to result from processing since no capping could be demonstrated (Fig. 3B). The start of the primary rrn transcript, which includes the tRNA(GAC)Val, could not be determined by capping and S1 nuclease mapping because it starts upstream of the DNA fragment that was used for hybridization, i.e. more than 691 bp upstream from the sequence coding for mature 16 S rRNA.
The Cotranscript tRNA(GAC)Val-rrn Starts 755 bp Upstream of the Sequence Coding for Mature 16 S rRNATo localize
the transcription start site of the tRNA(GAC)Val gene and
to ensure that this transcript in fact extends into the mature 16 S
rRNA we had to develop a new method. If the large transcript stops
before the mature 16 S rRNA-coding region, S1 nuclease mapping would
give wrong results as we cannot determine the 3-end of the hybrid
since the labeling by capping is at the 5
-end of the transcript.
(Usually the 3
-end of the single-stranded DNA is labeled.) Therefore,
we developed a method that is based on the enzymatic property of RNase
H to digest RNA in DNA-RNA hybrids. After the capping reaction, one
part of the RNA is annealed to primer a (which anneals
within mature 16 S rRNA) and digested with RNase H. The ribosomal
precursor RNA should be cleaved at the position of the primer. The
product(s) of this reaction is (are) analyzed concomitantly with two
control reactions, one which contains only the primer and another
one that is treated with RNase H but in the absence of primer. What we
are looking for now is the appearance of an additional band within the
pattern of capped total RNAs that is not seen in the two control
reactions. The size of this band corresponds to the distance of the
transcription start site(s) upstream from the position of the
primer.
First of all, we analyzed whether capping reactions are reasonably
reproducible. This is essential to identify newly appearing bands in
total capping patterns. Fig. 4A shows results
with total capped RNAs (lanes 1-4). Spinach seeds were
germinated in darkness for 7 days (lane 1), and seedlings
were subsequently illuminated for 2 (lane 2), 4 (lane
3), or 8 (lane 4) h. Total RNA was isolated from
cotyledons and labeled by capping. The most strongly labeled band of
about 120 bases was isolated and sequenced. It corresponds to
cytoplasmic 5 S RNA (not shown). To avoid the high background of this
band the following gels were run long enough to elute it from the gel.
RNase H mapping of capped cotyledon RNA using primer a produces two different supplementary bands of about 250 and 900 bases (marked by arrows in Fig. 4B, lane 2). The 250-base RNA corresponds to PC. The 900-base transcript confirms the existence of a very long rrn precursor transcript that comprises the tRNA(GAC)Val and corresponds well in size to the large transcript revealed by primer extension using the same primer (see Fig. 1). The asterisk (Fig. 4B) indicates the position predicted for the second rrn promoter that was observed in tobacco plastids (5, 21).
For fine mapping of the 900-base RNA, we hybrid-selected capped RNA
using complementary single-stranded DNA with the 5-end localized about
200 nucleotides downstream of PtRNA. The S1 nuclease-protected RNA is
shown in Fig. 4C. It localizes the transcription start site
(PtRNA) 457 bp upstream of the tRNA(GAC)Val gene. This
coincides well with the 870-base RNA found by cleavage with primer
a in the RNase H mapping experiment. PtRNA is not preceded
by E. coli-like consensus sequences (Fig. 4D)
suggesting transcription by the nuclear-encoded plastid RNA polymerase.
If we compare PtRNA with the main rrn transcription start
site in rpoB-depleted tobacco plastids (21) we notice that 7 of 8 bases are identical (Fig. 4D, bold
letters).
A
comparison of spinach and tobacco rrn promoter regions is
shown in Fig. 5. Differences in base composition are
marked by open boxes, transcription start sites are
indicated by bent arrows. The fact that PC is obviously not
used in spinach root tissues and that transcription start sites in this
region are different in spinach (PC) and tobacco (P1 and P2) plastids
prompted us to analyze and compare DNA/protein interactions by gel
retardation. The rrn promoter fragment (WT) and the promoter
mutation (M1) on which CDF2 does not bind (6) were analyzed in the
presence of plastid extracts obtained from spinach cotyledon (Fig.
6B, lanes 1-6) or root tissues
(Fig. 6B, lanes 7-12). Taking into account that
rrn transcription is 50 to 80 times lower in root than in
leaf tissues we used much higher protein concentrations to analyze root
extracts compared with leaf extracts. The formation of the
CDF2-promoter complex is detectable using leaf extracts (lanes
1-6). No protein/promoter interaction is detected with root
plastid extracts even if the protein concentration was scaled up to 52 times that of cotyledon extracts (lanes 9 and
12).
Fig. 6C shows DNA/protein interactions among the spinach rrn promoter DNA fragment, three mutated fragments (see Fig. 6A), and tobacco leaf plastid extract. The complex that is formed with the wild-type promoter fragment (lanes 1-3) is highly sensitive to M2 and M3, i.e. to modifications in the "-35" and "-10" sequence elements of the E. coli-like spinach P2 promoter. Using M1, the mutation of the CDF2-binding site, DNA/protein interactions are only moderately reduced (lanes 11 and 12) in contrast to results obtained with spinach plastid extracts (compare with Fig. 6B). This experiment suggests that the differences in spinach and tobacco chloroplast rrn initiation (Fig. 5) are due to differences in protein/promoter interactions in the two plant species.
If we compare root and leaf RNA from spinach and tobacco directly by
primer extension analysis using primer c (Fig.
7A), we find that cotranscription of the
rrn operon with the tRNA(GAC)Val gene seems also
to exist in tobacco plastids (lane C Spinach and lanes
C and R Tobacco, Pro1). The differential
localization of the transcription start sites (PC in spinach, P1 in
tobacco) is evident (compare lane C Spinach to lane C
Tobacco). Analysis of rRNA from tobacco cell cultures (BY2) shows
the two reported transcription start sites P1 and P2 (Fig.
7B, lane 2). In contrast, in plastids of spinach
cell cultures we determine only transcripts corresponding to PC
(lane 1).
Regulation of transcription in plastids is not yet well understood, and results on the expression of higher plant rDNA transcription seem to be contradictory. (i) The spinach plastid 16 S rDNA upstream region contains two tandem E. coli-like promoters, P1 and P2, which are efficiently used in vitro by E. coli RNA polymerase (Ref. 32 and Fig. 5). However, in vivo, we could not find products that correspond to these two initiation sites in photosynthetically active organs (7, 23). On the other hand, tobacco plastid rDNA transcription starts mainly at an E. coli-like promoter (P1), which corresponds to the spinach P2 promoter (Refs. 3 and 5 and Fig. 5). (ii) The higher plant plastid genome is transcribed by at least two different types of RNA polymerase (for recent review see Ref. 28). In spinach, plastid rDNA is probably transcribed by the nuclear-encoded enzyme as suggested by in vitro transcription studies (6) and the results presented here. On the other hand, transcription at the P1 promoter of tobacco plastid rDNA is likely to be initiated by the E. coli-like plastid-encoded RNA polymerase (21). (iii) A second minor transcription start site that is not preceded by E. coli-like promoter elements has been demonstrated recently in tobacco plastids. The steady-state level of the corresponding transcript is higher in BY2 plastids than in leaf chloroplasts (5). When the plastid rpoB gene is deleted by plastid transformation only this non-E. coli-like promoter is active suggesting that the minor transcription start site is used by the nuclear-encoded plastid RNA polymerase (21). In spinach leaf chloroplasts, this start site has not been reported so far.
These contradictions prompted us to analyze rrn expression in spinach plastids. We found that transcription of the rrn operon in intact spinach plants is regulated by usage of two promoters. In photosynthetic and non-photosynthetic organs (cotyledons and roots), the rrn operon is cotranscribed with the preceding tRNA(GAC)Val gene (Figs. 1 and 2). The transcription start site of the tRNA(GAC)Val-including precursor RNA (PtRNA) does not employ E. coli-like consensus sequences (Fig. 4). Instead, there is some upstream sequence homology to the recently reported rrn promoter that is active in rpoB-depleted plants (Ref. 21, P2 in Fig. 5, and see also Fig. 4C). This suggests that PtRNA is transcribed by the nuclear-encoded RNA polymerase. To compare our results with tobacco we performed primer extension analysis also with tobacco leaf and root RNA. The results indicate that cotranscription of the rrn operon with tRNA(GAC)Val exists also in tobacco (Fig. 7A).
The enhancement of rrn transcription in photosynthetically active cotyledon/leaf tissues is due to the organ-specific activation of the PC promoter in spinach (Figs. 1 and 3). This transcription start site is different from the main transcription start site of tobacco leaf plastids (Fig. 7A, compare also spinach PC and tobacco P1 in Fig. 5). Also, the analysis of RNA isolated from spinach cell cultures in parallel with tobacco BY2 cells shows differences in rrn expression patterns. The tobacco P2 transcript is only detectable in tobacco BY2 cells but not in spinach cell cultures (Fig. 7B). The reasons for these differences remain unclear. It seems, however, that the spinach PC and the tobacco P1 transcription start sites are recognized by different RNA polymerases and/or transcription factors (6, 21) (Fig. 6). The evolutionary transfer of genes coding for components of the plastid transcriptional machinery to the nucleus and the existence of multiple RNA polymerases in plastids provide a large background for species-specific rearrangements of the plastid transcriptional apparatus. Nevertheless, the difference in transcription initiation between tobacco and spinach remains surprising. Full understanding of these differences will require the purification of transcriptional components and in vitro reconstitution of initiation events in both systems.
The tRNA(GAC)Val-including precursor RNA is probably processed at the Pro1 site. This can be concluded from the negative capping result (Figs. 4 and 6) and from the quantitative differences of RT-PCR products which start upstream or downstream from Pro1 (Fig. 2C). The hypothetical secondary structure of the Pro1-processing site, as shown in Fig. 1E, locates the cleavage site within the loop of a hairpin structure. It was demonstrated that chloroplasts contain an endonuclease activity that cleaves within loop structures of inverted repeats (33). An enzyme of this type might be implicated in the processing mechanism occurring at Pro1.
Altogether, our results show species-specific assembly of the plastid
transcriptional complexes and the existence of novel mechanisms of gene
expression in plastids, i.e. the organ-specific usage of
different promoter structures. Fig. 8 summarizes the results obtained for spinach plastid rrn transcription.
We thank M. Rocipon for photographical work, H. Pesey for technical assistance and cell culture maintenance, and Dr. J.-G. Valay for critical reading of the manuscript.