Overexpression of a natural chloroplast-encoded antisense RNA in tobacco destabilizes 5S rRNA and retards plant growth

Background The roles of non-coding RNAs in regulating gene expression have been extensively studied in both prokaryotes and eukaryotes, however few reports exist as to their roles in organellar gene regulation. Evidence for accumulation of natural antisense RNAs (asRNAs) in chloroplasts comes from the expressed sequence tag database and cDNA libraries, while functional data have been largely obtained from artificial asRNAs. In this study, we used Nicotiana tabacum to investigate the effect on sense strand transcripts of overexpressing a natural chloroplast asRNA, AS5, which is complementary to the region which encodes the 5S rRNA and tRNAArg. Results AS5-overexpressing (AS5ox) plants obtained by chloroplast transformation exhibited slower growth and slightly pale green leaves. Analysis of AS5 transcripts revealed four distinct species in wild-type (WT) and AS5ox plants, and additional AS5ox-specific products. Of the corresponding sense strand transcripts, tRNAArg overaccumulated several-fold in transgenic plants whereas 5S rRNA was unaffected. However, run-on transcription showed that the 5S-trnR region was transcribed four-fold more in the AS5ox plants compared to WT, indicating that overexpression of AS5 was associated with decreased stability of 5S rRNA. In addition, polysome analysis of the transformants showed less 5S rRNA and rbcL mRNA associated with ribosomes. Conclusions Our results suggest that AS5 can modulate 5S rRNA levels, giving it the potential to affect Chloroplast translation and plant growth. More globally, overexpression of asRNAs via chloroplast transformation may be a useful strategy for defining their functions.


Background
Chloroplasts originated around 1.5 billion years ago via an endosymbiotic event where a primitive eukaryote engulfed an ancestor of modern-day cyanobacteria. Subsequently, massive gene transfer to the nucleus occurred, resulting in a highly reduced plastid genome of 120-160 kb that pos-sesses~120 genes. These remaining genes are mostly organized into clusters, and their expression requires a combination of prokaryotic and eukaryotic-like posttranscriptional events including maturation of polycistronic transcripts, splicing, RNA editing, and 5' and 3' end trimming [reviewed in [1]]. These processes are catalyzed by nucleus-encoded proteins, many of which were originally encoded by the chloroplast ancestor [2].
Post-transcriptional regulation in both prokaryotes and eukaryotes is also exerted by non-coding RNAs (ncRNAs), which include antisense RNAs (asRNAs) and transcribed intergenic sequences. The asRNAs can act in cis, on the cognizant sense strand transcript, or in trans, targeting one or more distantly-located genes through specific base pairing. Antisense RNAs regulate many steps in gene expression including translation initiation, mRNA stability, alternative splicing, RNA editing, and transcription termination [3,4]. Clear evidence exists that asRNAs are prevalent in cyanobacteria, the prokaryote most closely related to chloroplasts. In Prochlorococcus MED4 and Synechocystis sp. PCC 6803, asRNAs were found using microarrays, and some were verified by RNA blot and shown to be differentially regulated under altered stress and nutrient conditions [5,6]. Functions have been ascribed to two cyanobacterial asRNAs, α-furA and IsrR, which act in cis to occlude a ribosomal binding site and regulate sense strand transcript accumulation, respectively [7,8]. Thus, the chloroplast progenitor likely used asRNAs for gene regulation, suggesting that at least part of this capacity might have been retained in the present-day organelle.
The occurrence of organellar ncRNAs has been established experimentally from sequencing of cDNA libraries [9,10]. In one study, ncRNAs were abundantly detected in plant mitochondria deficient for polynucleotide phosphorylase (PNPase), a 3' to 5' exoribonuclease involved in RNA maturation and decay, some of which were also found in wild-type (WT) plants [1,9]. In chloroplasts, sequencing of small RNAs identified a number of ncRNAs [10,11]. Although their functions were not directly tested, at least some of them appear to be footprints of sequence-specific RNA-binding proteins involved in mRNA 3' end formation [12]. More recently, an RNA antisense to the chloroplast ndhB gene in Arabidopsis, Nicotiana tabacum and poplar was identified, which was hypothesized to regulate RNA maturation or stability [13]. Direct evidence for chloroplast asRNA function comes primarily from two studies, one in which an ectopically expressed asRNA resulted in reduced C-to-U editing in tobacco [14], and a second where a chloroplast genome rearrangement in Chlamydomonas led to expression of an asRNA that stabilized the corresponding sense strand transcript [15].
The present study focuses on AS5, a chloroplastencoded asRNA that is represented by several ESTs (see below), and was found to be highly abundant in the Arabidopsis rnr1-3 mutant (Sharwood, Hotto, Bollenbach and Stern, unpublished results). This mutant lacks ribonuclease R (RNR1), a 3' to 5' exoribonuclease, which is dually targeted to mitochondria and chloroplasts [16,17], and whose prokaryotic orthologue is involved in rRNA maturation and the degradation of structured RNAs [18,19]. The rnr1-3 mutant accumulates precursors for several transcripts encoded in the chloroplast ribosomal operon, but is particularly deficient in the accumulation of both precursor and mature forms of 5S rRNA, which is encoded by the rrn5 gene that lies just upstream of trnR, which encodes tRNA Arg . Because AS5 is antisense to the 5S-trnR region, and because rnr1-3 overaccumulates AS5 and underaccumulates the complementary 5S rRNA, we postulated that AS5 might regulate 5S rRNA processing or stability. Here we have overexpressed AS5 in tobacco (Nicotiana tabacum) chloroplasts to determine its in vivo function in the absence of any pleiotropic effects caused by RNR1 deficiency. The results show that overepxressing AS5 leads to a slow-growth phenotype and decreased 5S rRNA stability, suggesting a possible function for AS5 in WT chloroplasts.

Creation of tobacco chloroplast transformants
Transplastomic tobacco plants were obtained using the vector pAS5OX, which resulted in the overexpression of AS5 ( Figure 1A). This vector targets the trnI-trnA region in the chloroplast ribosomal (rrn) operon located in the inverted repeat region through homologous recombination, and relies on readthrough from the upstream P rrn promoter for AS5 transcription. This insertion site is well-characterized and has not been reported to affect plant growth and development [reviewed in [20]]. The selectable marker aadA possesses its own promoter and is transcribed in tandem with the rrn operon. As shown below, this leads to enhanced transcription of genes downstream of aadA. The AS5 sequence expressed from the transformation vector is the antisense strand extending from just downstream of the trnR gene, to the 3' end of rrn5. Endogenous AS5 is also expressed from this operon, albeit on the opposite strand.
Chloroplast transformants were identified by growth on selective medium and PCR (Experimental Procedures). After 3-4 rounds of regeneration, the transgenic lines were further validated by DNA gel blot after restriction digest with XhoI and HindIII (Figures 1A &1B). The XhoI site is within the trnI intron, while HindIII cleaves in the middle of the AS5 transgene and just downstream of trnA. Two independent AS5 overexpression (AS5 ox ) lines were confirmed (AS5 ox -1 and AS5 ox -2), with expected bands of 2.3 and 0.9 kb ( Figure 1B). Surprisingly, both lines retained a weak band at 1.3 kb, corresponding to a WT-like size. This is likely due to a nuclear-encoded fragment, rather than a plastome sequence, and has been seen in other chloroplast transformants targeting the rrn operon and other regions [21,22]. Furthermore, an identical pattern was observed after the plants had been self-crossed, which normally resolves any residual heteroplasmy. To verify that increased AS5 expression was occurring in the AS5 ox lines, we used quantitative RT-PCR (qRT-PCR; Figure 1C). After normalization to two nuclear control mRNAs, we found that the AS5 ox lines accumulated >500-fold more AS5, confirming the expected transgene expression.

Plant growth is slowed in the AS5 ox lines
Growth was compared between AS5 ox -1, AS5 ox -2 and WT tobacco grown in the greenhouse. The AS5 ox lines exhibited slower growth beginning 19 days after germination, which resulted in delayed flowering time (Table 1; Figure 2). Forty days after germination, we observed that internode 3 was not significantly reduced in length, while internode 9 was shorter. In addition, the stem circumference at both internodes was significantly reduced (~7% decreased at internode 3, and~16% decreased at internode 9). The differences in internode length and width resulted in shorter plants with lighter shoots after 40 days. Shoot fresh weight at maturity was reduced by about 10% compared to the WT, which can likely be ascribed to reduced stem diameter, as the plant height at maturity did not differ (data not shown). Leaf 3 and leaf 9 dimensions and weight on a per area basis were measured 40 days after germination. Leaf 3, near the base of the plant, was not significantly altered in length, width or fresh weight, while the dry mass was slightly increased in the AS5 ox plants. However, leaf 9 was significantly smaller in both length and width, while unit area weight was unaffected. Lastly, total chlorophyll was reduced by~10% in the AS5 ox lines at maturity, corresponding to a slightly pale-green leaf phenotype. Taken together, these data indicate that the overall vigor of the AS5 ox lines is diminished, especially with respect to stem elongation.

AS5 is present as multiple transcripts
EST data from tobacco and its solanaceous relative tomato suggested that AS5 is present in multiple overlapping forms ( Figure 3A). These ESTs, however, represent partial sequences from size-selected cDNAs, and are derived from polyadenylated transcripts, which are inherently unstable in chloroplasts [23]. To assess which discrete species might accumulate in WT or transformed plants, a series of overlapping strand-specific probes spanning the 5S-trnR region were used on polyacrylamide gel blots of total leaf RNA ( Figure 3B).
The longest probe, P1, contained the 5S-trnR intergenic region and the trnR coding region. The resulting banding pattern revealed four AS5 transcripts of 70-400 nt in the WT (WT1-4), and an additional 80 nt species in the AS5 ox lines. The multiple species may result from endo-and/or exonucleolytic maturation of an AS5 primary transcript, and/or from specific cleavages of AS5 bound to a sense RNA target. Each of the four AS5 bands common to all samples overaccumulated in the AS5 ox lines, suggesting that at least a portion of the transgene transcript was functioning in a manner similar to the endogenous one. The AS5 ox lines also accumulated larger transcripts not resolved in the polyacrylamide gel (see below).
Additional probes were used to further characterize the AS5 species. Probe P2 contained only the 5S-trnR intergenic region and identified all four WT bands, indicating that the WT1-4 bands must contain part of the intergenic region. However, the WT4 band was most strongly detected. This increased hybridization to the WT4 band, as compared to results with P1, may reflect secondary structure differences between probes P1 and P2. The AS5 ox -specific band was not detected with P2, demonstrating that this band contains sequences antisense to the trnR coding region, and possibly additional sequence derived from the transgene.
Probe P3 spans the proximal part of the 5S-trnR intergenic region. This probe hybridized more weakly than P1 and P2, which may be due to less sequence overlap with the targets, and/or its low (27%) G+C content compared to the full intergenic region (41%). The strongest hybridization was to WT1 and WT4, although WT2 was faintly visible. This suggests that WT1 and WT4 are fully derived from the region covered by P3. WT2 appears to have some overlap with P3, but is likely to be mostly derived from more distal portions of the intergenic sequence, whereas WT3 appears to be derived entirely from the distal part of the intergenic region. It is possible that WT1 serves as a precursor for each of the smaller transcripts, but this cannot be directly inferred. In summary, the AS5 ox lines overproduce the native AS5 species and at least one novel small AS5 molecule.

AS5 ox lines accumulate processing intermediates of rrn transcripts
Because the transgene was inserted between trnI and trnA, rrn operon transcripts were analyzed by RNA gel blot. The operon encodes four rRNAs and three tRNAs that are transcribed as an~7.4 kb polycistronic precursor from a bacterial-like promoter upstream of rrn16 [ Figure 4A; [24]]. A series of endo-and exonucleolytic steps, along with splicing of the trnI and trnA introns, yield the mature RNAs. The probes shown along the bottom of Figure 4A were used to detect any perturbations due to AS5 transgene insertion or regulation.
When blots were probed for rrn16, mature 16S rRNA and a precursor were detected ( Figure 4B). This precursor is created by endonucleolytic cleavage in the rrn16-trnI intergenic region, leaving a 3' extension that is processed exonucleolytically [16,25]. Abundance of the mature and precursor rrn16 transcripts was slightly reduced in the AS5 ox lines.
The trnI probe detected mature tRNA Ile (72 nt), a large polycistronic intermediate (~3.3 kb), and the~0.8 kb unspliced trnI monocistron. In the AS5 ox lines, two additional polycistronic precursors accumulated, which appear to contain transgene sequences because they are not present in the WT and comigrate with AS5 ox -specific Results represent an average ± the standard error of measurements taken 40 days after germination (WT -n = 18; AS5 ox -1 and AS5 ox -2 -n = 17) or at maturity (n = 10). Internode and leaf numbers were counted starting from the base of the plant. *Traits with a significant difference (p < 0.05) between both AS5 ox lines and WT.
AS5 bands. Unspliced trnI was invariant, and accumulation of mature tRNA Ile decreased by~50% in the AS5 ox lines. Decreased efficiency of endonucleolytic cleavage at the 3' end of trnI, near the transgene insertion site, may be contributing to the reduced formation of mature tRNA Ile and increased accumulation of polycistronic intermediates. The AS5 probe identified several transcripts. The largest polycistronic transcript appears to comigrate with the WT trnI polycistron, in addition to the two AS5 ox -specific trnI bands mentioned above. Other less-abundant bands were apparent, the smallest of which migrated near tRNA Ile and represents the abundant AS5 ox -specific band detected with probe P1 in Figure 3B. We next probed for the aadA selectable marker transcript. The aadA cassette generates a monocistronic mRNA of 1.2 kb, which was by far the most abundant species. However, increased exposure revealed both longer bands and one shorter band. The former appear to be cotranscripts with trnI and the AS5 transgene, whereas the shorter band appears to be a cotranscript with only AS5. Thus, AS5 ox plants accumulate a variety of AS5-containing transcripts, some of which are the same size as endogenous species and others that are transgene-specific products. The implications of these findings for possible AS5 biological activity are considered in the Discussion.
The lower part of Figure 4B shows probes for the rrn operon downstream of aadA. Like trnI, the trnA probe revealed accumulation of additional precursors in the AS5 ox lines, however no effect on tRNA Ala accumulation was observed. The two precursor transcripts also found in WT were a doublet at 3.2 kb and a 0.8 kb species, the latter corresponding to unspliced trnA. Each of these RNAs overaccumulated compared to the WT. At least five additional bands were detected in the AS5 ox lines, which appear to be cotranscripts with aadA.
Analysis of rrn23 transcripts revealed seven major species. This complex pattern is due to accumulation of rrn23-rrn4.5 processing intermediates and hidden breaks within the rrn23 transcript [26]. There were no size  differences between the WT and AS5 ox lines and only minor quantitative differences, suggesting that there was no accumulation of aadA cotranscripts. Similarly, accumulation of mature 4.5S rRNA was not affected.
Finally, strand-specific probes were used to assess processing and accumulation of 5S rRNA and tRNA Arg . Maturation of these species involves the formation of a transient 5S-trnR intermediate that is cleaved by RNase P to form the 5' end of tRNA Arg , followed by a second endonuclease cleavage in the 5S-trnR intergenic region near the mature 5S 3' end [27]. The pre-5S rRNA is further matured by the 3' to 5' exonucleolytic activity of RNR1 and possibly PNPase [16]. When probed for 5S rRNA, accumulation was similar between the WT and AS5 ox lines. In contrast, tRNA Arg overaccumulated in the AS5 ox lines. As discussed below, this results from increased transcription from the aadA promoter. No precursors were observed for either rrn5 or trnR, including when analyzed following high-resolution polyacrylamide gel electrophoresis (data not shown), indicating that maturation of their 5' and 3' ends was largely unaffected.
Because the rrn5 and trnR transcripts were the most likely targets of regulation by the overexpressed asRNA, qRT-PCR was used to quantify their accumulation relative to two nuclear transcripts ( Figure 5). As suggested by the gel blots ( Figure 4B), there was only a small increase in rrn5 accumulation in the AS5 ox lines, while the trnR transcript increased three-fold. Overall, the results in Figures 4, 5 show that some rrn operon transcripts accumulated differently in the AS5 ox lines. This may be due to differences in transcription rates, RNA processing, and/or RNA stability.

Increased transcription of 5S-trnR in the AS5 ox plants
We had hypothesized that AS5 regulates rrn5 processing and/or stability, and we observed that in the AS5 ox lines 5S rRNA accumulation was unaffected whereas tRNA Arg was three-fold more abundant. One explanation to account for increased tRNA Arg accumulation was enhanced transcription due to the upstream aadA promoter. This would a priori affect rrn23, rrn4.5, rrn5 and trnR, but not genes upstream of aadA, such as rrn16.
To determine relative transcription rates in the WT and AS5 ox lines, a run-on assay was performed. Purified chloroplasts were labeled with α-32 P-UTP for 5 min, after which labeled RNAs were used to probe a slot blot containing 250 and 500 ng of PCR products from rRNAs, the Photosystem II D1 protein-coding gene (psbA), and the chloroamphenicol acetyltransferase gene (cat) as a negative control ( Figure 6). Because the signals did not increase in the 500 ng slots, we concluded that the target DNA was in excess over the probe. We also noted a weak signal from the cat gene, which was considered background noise and was subtracted from subsequent calculations. Signal intensities were normalized to psbA, whose transcription should be unaffected in AS5 ox lines.
The analysis showed that rrn16 transcription varied less than 10%, with the slight decrease mirroring the RNA gel blot result ( Figure 4B). As rrn16 is proximal to the transgene promoter, a different transcription rate was not expected. When rrn23 was examined, the transcription rate was almost doubled in the AS5 ox lines, and 5S-trnR transcription increased four-fold. The disparity between rrn23 and 5S-trnR results may be attributed to the low signal to noise ratio for the latter in the WT, making the measurement less accurate. In any event, these results show that while transcription of rrn5 and trnR are increased in the AS5 ox lines, only tRNA Arg increases in abundance, indicating that 5S rRNA is less stable in the AS5 ox lines. The unstable species could either be the mature form or a precursor with a 3' extension following trnR excision. In summary, increased transcription of 5S rRNA masks the instability imparted by overexpression of AS5.

5S rRNA polysome association decreases in AS5 ox lines
The AS5 ox lines have a slow-growth phenotype and decreased 5S rRNA stability. If AS5 were destabilizing mature or precursor 5S rRNA, it might also interfere with its incorporation into mature ribosomes. Polysome analysis was therefore conducted to assess ribosomal association of 5S rRNA and two representative mRNAs. As shown in Figure 7, polysome fractions were analyzed for rrn5, rbcL, psbA and AS5. The rrn5 probe revealed that a large portion of mature 5S rRNA is not associated with polysomes in either WT or AS5 ox lines. This was expected, since 5S rRNA exists in a dynamic relationship with the 50S ribosomal subunit, an association thought to influence ribosome performance [28]. In the AS5 ox lines, however, there was a small but distinctive and reproducible shift of 5S rRNA towards less dense fractions (lanes 1-6), with about one quarter being present in the polysome fractions (lanes 7-12). This may be an indirect influence of AS5 overaccumulation (see Discussion).
Polysomal loading of the rbcL and psbA mRNAs were investigated to identify any effect on chloroplast translational activity. As was the case for 5S rRNA, the rbcL mRNA shifted modestly towards the nonpolysomal fractions in the AS5 ox lines, suggesting that its translation rate had decreased, in keeping with their slower growth phenotype. The polysomal loading of psbA mRNA was not appreciably altered between WT and AS5 ox lines. However, this transcript is mostly nonpolysomal even in WT plants, except under conditions where PSII repair is induced [e.g. [29]].
Finally, the AS5 transcript itself was examined. Considering the data shown in Figure 4B for total RNA, we expected to observe approximately six bands between 80 nt and 3.3 kb, with the smaller species being only visible when polyacrylamide gels were used (Figure 3). Figure 7 shows that the vast majority of AS5 RNAs in the AS5 ox lines migrated in the nonpolysomal fractions. This result was largely expected for an ncRNA, however we noted that a portion of the longer AS5 transcripts were found in heavier fractions. Intriguingly, their accumulation pattern roughly parallels that of 5S rRNA. This raises the possibility that AS5 may bind to rrn5 that is incorporated into polysomes, although the hybridization results do not demonstrate that the two RNAs are actually associated. The 80 nt AS5 band found only in the AS5 ox lines was in fraction 1, at the top of the gradient. The other small AS5 bands are not visible in this gel system. Faint higher molecular weight bands visible on the WT blot were judged to be rRNA artifacts based on their migrations.

Discussion
NcRNAs have been detected in both mitochondria and chloroplasts, yet there is little knowledge of their functional significance. In this study, the chloroplast asRNA AS5 was investigated for possible cis-acting regulation of rrn5 and trnR. We used the approach of chloroplast transformation to overexpress AS5, a strategy not previously used for a natural chloroplast asRNA. However, overexpression of nuclear-encoded miRNA is routinely used to confirm miRNA targets [reviewed in [30,31]]. While transgenes can create pleiotropic effects, we were able to correlate increased AS5 expression with decreased 5S rRNA stability and slightly reduced polysome association. While the underlying mechanism remains to be elucidated, we speculate that AS5 may Figure 6 Run-on transcription in WT and AS5 ox chloroplasts. 250 or 500 ng of PCR product from chloroplast-encoded genes (right) were fixed to nylon membranes, with the chloroamphenicol acetyltransferase gene (cat) included as a negative control. Run-on transcription was conducted as described in Experimental Procedures. The ratio of AS5 ox to WT transcription (left) was calculated after background subtraction of the cat gene and normalization to psbA. The average of three experiments is shown. restrict 5S rRNA ribosomal incorporation and/or pair with its precursor, targeting the duplex for degradation via a double-stranded endoribonuclease.
RNA processing factors for the rrn operon may be limited in AS5 ox lines When AS5 was expressed from within the rrn operon, several probes showed varied accumulation of precursor transcripts or processing intermediates compared to the WT. Some of these disparities may have arisen because RNA processing capacity had been exceeded, a phenomenon that has previously been observed. Transgenes inserted into the chloroplast rrn operon have been shown to accumulate >100-fold more RNA than an analogous nuclear (e.g. aadA) transformant due to plastid genome polyploidy [32,33]. Excess RNA produced by a chloroplast transgene would unlikely stimulate any increase in the nuclear-encoded and chloroplast-targeted proteins necessary for its maturation. For example, when the clpP 5' UTR was overexpressed in tobacco chloroplasts a reduction in endogenous mature clpP mRNA was observed due to the limited availability of a clpPspecific mRNA maturation factor [34].
In this study, we observed a several-fold increase in transcription rates distal to the aadA insertion, which is substantial given that the rrn promoter is already the Figure 7 Polysome analysis of WT and AS5 ox samples as described in Methods. Radiolabeled probes used for rrn5, rbcL and psbA detection were dsDNA, while an RNA probe was used for AS5 detection. Estimated band sizes for AS5 are shown to the left. The ethidium bromide-stained gel represents equal proportional loading of the 12 gradient fractions. EDTA and puromycin controls (not shown) were used to determine the free ribosomal fractions (< 80S, lanes 1-6) and polysomal fractions (lanes 7-12). The percent of total RNA from polysomal and non-polysomal fractions are shown below the respective blots, as determined from averaging four (WT) or eight (AS5 ox -1 and -2) experiments ± the standard error.
strongest in the chloroplast [35]. The most strongly affected gene in this region was trnA, most likely due to its proximity to the aadA promoter. Extra precursor accumulation from the rrn operon, however, probably does not contribute to the slow-growth phenotype of AS5 ox lines. Previous studies where transgenes were targeted to the rrn operon also resulted in substantial accumulation of polycistronic transcripts, but little effect on plant development and chloroplast protein expression was noted, other than the intended effects of the transgenes, such as increasing tolerance to abiotic stress [32,36,37]. However, in one case plant development was delayed [38].
Mature tRNA Ile accumulation is decreased somewhat in the AS5 ox lines, while tRNA Ala was not affected. We considered whether tRNA Ile deficiency could explain the polysome shift ( Figure 7) and slow growth phenotype. We concluded this is unlikely because the chloroplast encodes a second tRNA Ile isoacceptor (CAU) within the inverted repeat region that could compensate for the decrease in the affected tRNA Ile -GAU using wobble base pairing, which is common in chloroplasts [39]. The accumulation of tRNA precursors can probably be ascribed to transgene sequences that interfere with RNA structures recognized by processing enzymes.
Multiple AS5-containing precursor transcripts also accumulated in the transgenic lines, while a family of smaller AS5 transcripts (< 400 nt) were detected by RNA gel blot for both WT and AS5 ox plants. The precursor transcripts are polycistronic intermediates, while the smaller transcripts may be a result of AS5 maturation or degradation. The AS5-containing polycistronic transcripts appear to be poor substrates for chloroplast RNA processing enzymes, although the smaller species exhibited a 3-to 10-fold greater accumulation in the AS5 ox lines than in the WT (Figure 4). This is far less than the qRT-PCR results, which showed a >500-fold overaccumulation of AS5 in the transgenic lines due to quantification of the very abundant polycistronic precursors in addition to the smaller AS5 species. These longer RNAs could have less biological activity than the small molecules in terms of 5S rRNA binding and/or destabilization, although a small proportion of them was found in the heavier fractions of polysome gradients (Figure 7).

5S rRNA stability is decreased in transgenic plants
Accumulation of mature rRNA transcripts differed in the AS5 ox lines. It was expected and observed that transcription of rrn23 and rrn5 would increase due to the position of the transgene promoter, and that rrn16 would be unaffected. However, there was only a minor (~1.5-fold) increase in mature 23S rRNA, no difference in 4.5S and 5S rRNAs, and a small decrease in 16S transcripts. The mechanisms surrounding rRNA degradation and ribosome copy control in plant organelles are poorly understood, although in bacteria it has been established that rRNAs accumulate in a molar ratio with their ribosomal protein counterparts, and if one or the other is in excess they are targeted for degradation [reviewed in [40]]. Bacterial rRNA turnover also occurs as a way to regulate translation during slow growth and under certain stress conditions, and to prevent the accumulation of misassembled ribosomes. The variability in mature rRNA accumulation in the AS5 ox lines is likely due to post-transcriptional mechanisms that equalize rRNA stoichiometry by degrading excess transcripts according to the bacterial paradigm. To this end, we speculate that AS5 may be regulating 5S accumulation, leading to degradation of excess 23S and 4.5S rRNAs. Crosstalk between chloroplast rRNAs was previously observed in the tomato dcl mutant, whose primary deficiency is in 4.5S rRNA, but also exhbited decreased polysomal loading of 5S rRNA [41].

Native AS5 may be processed in vivo
Multiple AS5 species were detected in both WT and AS5 ox lines. Among ncRNAs, the existence of multiple transcripts is relatively common for eukaryotic miRNAs and their precursors [42]. The transcript profiles of organellar ncRNAs, however, have been characterized in only a few cases. In the case of chloroplast ncRNAs, both very small [22-55 nt; 10] and longer [400-650 nt; 43] transcripts have been verified by gel blots. In the latter case, three transcripts accumulated from the ndhB antisense strand. AS5 could be produced by transcriptional read-through from trnN, and/or from a promoter between trnN and AS5. Either would create longer precursors that could give rise to the observed species.
Further evidence that AS5 is processed comes from the EST database, which shows variable 5' and 3' ends from tomato AS5 transcripts ( Figure 3A), something that is mirrored when AS5 ESTs are extracted for Arabidopsis (data not shown). However, ESTs are often derived from protocols that exclude smaller transcripts (< 200 nt), so AS5 transcripts such as the 70 nt species detected by RNA gel blot ( Figure 3B) would not be found even if it were polyadenylated.
The 70 nt AS5 species is derived from the 5S-trnR intergenic region near the mature 5S rRNA 3' end, a region previously shown to be a site for endonuclease cleavage during 5S rRNA maturation [27]. This raises the possibility that some AS5 bands could be derived from cleavage of a longer AS5 species bound to its putative 5S rRNA target. Both 5S rRNA and the 5S-trnR intergenic region are predicted to form multiple stemloops [44,45], which would interfere with full-length sense-antisense duplex formation and leave singlestranded regions available for ribonucleolytic cleavage, or promote cleavage by an RNase III-like activity. Partial duplexing between exposed loops or bulges is common for prokaryotic ncRNAs and their targets [46,47]. One example is binding of DsrA to its mRNA target, rpoS. This creates an RNase III processing site, which, following cleavage, produces stable transcripts of each [48].
Some AS5 transcripts may be degradation intermediates unrelated to 5S rRNA interactions. In prokaryotes, both bound and unbound ncRNAs are regulated by multiple ribonucleases, including RNase E, PNPase, RNase II and RNase III. For example, the bacterial ncRNA RNAI degradation intermediate is visible by RNA blot after RNase E cleavage near the 5' end. Following this, the transcript is polyadenylated and degraded by RNase II and PNPase [49]. Similarly, multiple transcripts are observed for the Salmonella typhimurium ncRNAs CsrA, CsrB, MicA, and SraL, which were determined to be degradation intermediates [50].

AS5 may regulate 5S rRNA
As discussed above, our results lead to the hypothesis that at least a portion of overexpessed AS5 is binding to pre-and/or mature 5S rRNA to inhibit its maturation or incorporation into ribosomes, ultimately targeting it for degradation via an RNase E or RNase III-dependent pathway. RNase E is present in chloroplasts [51,52], and one RNase III-like protein has been detected to date [53]. Additional members of the prokaryotic RNase III family have also been predicted [54].
Based on bacterial data, 5S rRNA is thought to stabilize the 50S subunit, enhance peptidyl transferase activity, and assist in tRNA binding [45]. In E. coli, sequential deletion of 5S rRNA genes resulted in a slower growth rate [55], which parallels the slower growth of the AS5 ox lines where 5S rRNA availability may be limited by AS5. Mature 5S rRNA binds to ribosomal proteins L5 and L18, and this complex is incorporated into the pre-50S subunit, a process that may be facilitated by the RNAbinding proteins CSP41a and CSP41b [56,57]. Excess AS5 may be a competitive inhibitor of L5 and L18, preventing ribosomal incorporation.
Our working hypothesis is that AS5 is part of a quality/ quantity control mechanism for 5S rRNA. While 5S rRNA has not been proposed as a regulatory target in chloroplasts, other examples do exist. In the case of nuclearencoded 5S rRNA in Xenopus oocytes, the Ro protein associates specifically with 5S sequences that have extra 3' nucleotides and/or point mutations. These faulty 5S sequences are eventually degraded, and it is thought that binding of Ro acts as a chaperone for target recognition by ribonucleases [58]. A second regulatory mechanism occurs in Arabidopsis via a heterochromatic siRNA, siR1003. In this case, aberrant 5S transcripts extending into an intergenic spacer region are cleaved into siRNAs. These siRNAs regulate the methylation status of the 5S rDNA to prevent further accumulation of long 5S transcripts [59]. While these particular modes of 5S regulation could not occur in chloroplasts, they do suggest that organisms have evolved ways of detecting excess and/or faulty 5S rRNA.

Conclusion
AS5 is one of the first natural chloroplast-encoded asR-NAs to be investigated for a possible cis regulatory role in gene expression. This function is commonly observed for both prokaryotic and eukaryotic ncRNAs. Overexpression of AS5 led to slower growth, which we interpret as a pleiotropic effect of its interaction with 5S rRNA. AS5 may interact dynamically with 5S rRNA as influenced by the accumulation of other ribosomal transcripts and proteins. We hypothesize that the role of AS5 is to prevent the accumulation of misprocessed 5S rRNA, as well as to control its stoichiometry. Further studies are needed to determine the enzymes involved in the maturation and turnover of AS5, and other chloroplast-encoded asRNAs.

Nucleic Acid Manipulation
The transformation vector diagrammed in Figure 1A was obtained by modification of the plasmid ptrnI-RT [60]. Briefly, the T7G10 5' UTR was removed by digestion with AscI and NdeI. The 450 bp AS5 sequence was PCR-amplified from total tobacco DNA with the primers AS5 ox 5' and 3' (Additional File 1), incorporating NheI and ClaI restriction sites, respectively, then inserted into pCR2.1. The resulting plasmid was NheI-NotI digested, and ligated to the modified ptrnI-RT vector to yield the final pAS5OX plasmid.

Chloroplast Transformation
Chloroplast transformants were obtained by bombardment of two week-old Nicotiana tabacum cv. petite Havana seedlings using 0.6 μm gold particles coated with pAS5OX [22,61]. Transformants were selected on RMOP-spectinomycin (500 mg L -1 ), then subjected to 3-4 rounds of regeneration on RMOP-spectinomycin/ streptomycin (500 mg L -1 ea.). AS5 ox transformants were monitored by PCR and DNA gel blot. Two homoplasmic lines were obtained (AS5 ox -1 and -2), and transferred to MS-spectinomycin (500 mg L -1 ) for rooting. The transformants were grown on soil at 25°C with a 16-h light/dark photoperiod and the seeds collected. All subsequent analyses utilized homoplasmic AS5 ox plants germinated on soil.

DNA Gel Blots
Leaf or callus tissue was ground in liquid nitrogen with a mortar and pestle. DNA was extracted using the DNeasy Plant Mini Kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. Total DNA was digested with XhoI and HindIII and hybridized using the method of Church and Gilbert [62]. Primers trnI-trnA F and R (Additional File 1) were used to amplify a 250 bp region between the trnI 3' exon and trnA intron, which was used as a template for probe synthesis as described below. The hybridized membrane was visualized using a Storm Scanner (Molecular Dynamics, Sunnyvale, CA).

Phenotypic Analysis
Plants were grown in flats for two weeks on soil supplemented with osmocote fertilizer under a 16-h light/dark cycle, then transferred to two gallon pots with the same soil until flowering, when total chlorophyll and shoot height were determined (Table 1). For chlorophyll analysis, duplicate 0.1 cm 2 leaf discs were frozen in liquid nitrogen and ground in 1 mL of methanol using a Wheaton homogenizer. Cell debris was pelleted by centrifugation at 13,000 rpm for 3 min at 4°C, after which total chlorophyll was determined as previously described [63]. In a second analysis, plants were germinated and grown under the same conditions, but phenotypic measurements were taken 40 days after germination. Shoot height was measured from the top of the soil to the apical meristem. Internodes and leaves were counted starting at the base of the plant to the apical meristem. Internode circumference was measured midway between surrounding leaves. Leaf length/width were measured at the longest/widest points, and leaf weight was determined on a per area basis.

RNA Isolation and qRT-PCR
Mature leaf tissue was ground in liquid nitrogen, and total RNA extracted using TRI reagent (Molecular Research Center, Cincinnati, OH) with minor modifications to the manufacturer's instructions. RNA was precipitated overnight with isopropanol at -20°C, and the pellet was washed with 75% ethanol and dissolved in water. For strand-specific cDNA synthesis, 1 μg of DNase-treated RNA was reverse transcribed with Super-Script III (Invitrogen, Carlsbad, CA) using the 3' qPCR gene-specific primers (Additional File 1). The qPCR reaction contained 1× Fast SYBR Green Master Mix (Applied Biosystems, Carlsbad, CA), 2 ng cDNA, and 300 nM of each primer, except for 18S rRNA (200 nM each primer), in a 20 μL volume. Amplification was done in a Bio-Rad CFX96 real-time PCR detection system (Hercules, CA) using the following two-step cycling conditions: initial denaturation at 95°C for 3 min, followed by 40 cycles at 95°C for 10S and 59°C for 30 s (+plate read), and a final incubation at 95°C for 10 s, after which a melt curve analysis was completed (59-95°C in 0.5°C steps) to ensure amplification specificity. Quantification and primer efficiencies were determined by comparison to four-step standard curves (0.04-10.00 ng cDNA in five-fold increments). Relative quantification compared to WT samples (given a reference value of 1) was achieved after normalization to 18S rRNA and GAPDH mRNA by the Bio-Rad CFX Manager software, taking into account differences in primer efficiencies. The final data is an average of three biological and at least two technical replicates.

RNA Gel Blots
One microgram of total RNA per sample was separated in 1.2% agarose/formaldehyde gels, which were probed with either double-stranded DNA or single-stranded RNA probes as indicated in the Figure Legends. DNA probes were synthesized from 100 ng of PCR template, and hybridized to the RNA gel blot at 65°C according to Church and Gilbert [62]. RNA probes were made from 100 ng of PCR template containing a T7 promoter using T7 polymerase and 40 μCi α-32 P-UTP, and then gel purified. For RNA probes, membranes were pre-hybridized for 6 hrs in 50% formamide, 20× SSC, 2% bovine serum albumin, 0.6% SDS and 200 μg mL -1 denatured salmon sperm DNA at 65°C, after which the RNA probe was denatured and allowed to hybridize overnight at 65°C. Membranes were then washed at 65°C twice for 5 min in 1× SSC and 0.6% SDS, followed by two 20 min washes in 0.3× SSC and 0.6% SDS. For polysome analysis, 200 mg of leaf tissue was extracted and fractionated through a 15-55% sucrose density step gradient [64], and the RNA extracted and analyzed as described above. To analyze small RNA fragments, 5 μg of total RNA was separated in a 10% polyacrylamide gel, electroblotted onto Hybond-N+ nylon membrane (GE Healthcare, Piscataway, NJ) in 1× TBE buffer using a TE 77 semi-dry transfer apparatus (GE Healthcare), UV-crosslinked, and probed as described above.

Chloroplast Isolation and Run-on Transcription
Chloroplasts from WT and AS5 ox lines were isolated from 40 g of leaf tissue as previously described [65], with minor modifications. Intact chloroplasts were collected from the interface of a 40-80% sucrose density gradient. The final chloroplast pellet was washed with two volumes of IC buffer (50 mM HEPES, pH 8.0 and 0.33 M sorbitol) and resuspended in 1 mL IC buffer. Plastid number was determined with a hemacytometer. Transcription was initiated by the addition of 2.5 × 10 6 plastids per 25 μL reaction containing 4 μCi μL -1 α-32 P-UTP that was equilibrated at 25°C [66]. The reaction was terminated after 5 min by the addition of an equal volume of stop solution (5% SDS, 50 mM Tris-HCl pH 8.0, and 25 mM EDTA), and then extracted with phenol:chloroform. The supernatant was treated with DNase, followed by a second phenol:chloroform extraction and ethanol precipitation.
The pellet was resuspended in 50 μL of TE, and unincorporated nucleotides were removed on a Sephadex G-25 column. The final RNA sample was denatured at 65°C for 15 min, and then used as a probe for DNA slot blots [67]. PCR products representing psbA, rrn16, rrn23, rrn5-trnR, aadA, and cat (negative control) were denatured and spotted onto Hybond-N+ membrane by vacuum filtration using a Hybri-Slot Manifold (Bethesda Research Laboratories, Carlsbad, CA). The membrane was UVcrosslinked, and then pre-hybridized for 12 hrs at 65°C in hybridization solution (6× SSC, 5× Denhardt's solution, 0.5% SDS, and 40 μg mL -1 denatured salmon sperm DNA). The probe was allowed to hybridize for 24 hrs, after which the membrane was washed twice for 30 min in wash buffer 1 (2× SSC, 0.1% SDS) and once for 30 min in wash buffer 2 (0.5× SSC, 0.1% SDS) at 65°C [68]. The membrane was visualized and quantified using a Storm scanner (GE Healthcare).

Additional material
Additional file 1: PCR Primers.