Variations in efficiency of plastidial RNA editing within ndh transcripts of perennial ryegrass (Lolium perenne) are not linked to differences in drought tolerance

Projected climate change is likely to subject key temperate grassland species, such as perennial ryegrass (Lolium perenne) to drought stress. Previous studies have shown that the NADH dehydrogenase complex (NDH) is involved with countering oxidative stress during environmental stresses like drought. We studied RNA editing within plastidial transcripts of the NDH complex in relation to the drought response of several accessions of perennial ryegrass. We found dramatic and reproducible differences in RNA editing efficiency between accessions, but efficiency was not influenced by imposition of drought stress, and a direct relationship between editing behaviour and drought response was not detected.


Introduction
Maintenance of healthy grasslands is essential for efficient livestock production, yet projected climate change is likely to place a heavy drought stress burden on key grassland species, such as perennial ryegrass (Lolium perenne). Perennial ryegrass-dominated pasture is the basis for livestock production in many temperate regions. However, as a consequence of estimated climate change over the next 100 years, the viability of the production of forage grasses will be threatened due to changes in temperature and rainfall (Holden and Brereton 2002). By improving the drought tolerance of cultivars of perennial ryegrass, the impact of climate change can be countered. This could be accomplished by either traditional breeding or genetic engineering, both of which will benefit from a deeper understanding of the underlying basis of drought stress responses in these crops.
The finding (Shinozaki et al. 1986) that the chloroplast genome includes close homologues to the genes encoding the subunits of the mitochondrial NADH:ubiquinone oxidoreductase (NDH) posed questions on the role of this complex in chloroplasts, which were resolved only when the availability of chloroplast transformation procedures allowed the characterization of gene knockouts (Burrows et al. 1998;Horvá th et al. 2000;Joët et al. 2001). These confirmed that the complex is fully functional but is nonessential under normal growth conditions (Burrows et al. 1998), and suggested an adaptive role in relation to drought stress (Horvá th et al. 2000). Specifically, when the ndhB gene was inactivated in transplastomic plants, the dark reduction in the plastoquinone (PQ) pool was impaired, and enhanced growth retardation was observed under humidity stress conditions (Horvá th et al. 2000). Support for this role comes from the finding that NDH complex activity in thylakoids increases under a combination of drought, high light and temperature stress (Ibanez et al. 2010). Sequencing of the perennial ryegrass plastome revealed that genes encoding NDH proteins appear to be a hot spot for RNA editing sites, with more than half the detected editing sites being located in these genes (Diekmann et al. 2009). RNA editing alters the nucleotide sequence of an RNA molecule so that it deviates from the sequence of its DNA template and provides a novel post-transcriptional regulatory system in organelles (Tillich et al. 2006). Different RNA editing systems exist and each is thought to have evolved independently (Gray 2012). For example, individual pentatricopeptide repeat (PPR) proteins have recently been implicated in editing at specific sites in the ndh complexes of both mitochondria (Yuan and Liu 2012) and chloroplasts (Okuda et al. 2010). RNA editing in chloroplasts belongs to the conversion system, where exclusively C-to-U substitutions occur, with the exception of U-to-C substitutions in the bryophyte Anthoceros formosae (Kugita et al. 2003). mRNA editing usually results in the restoration of codons for conserved amino acids (Bock et al. 1994). It is plausible that the functionality of the NDH complex could be impaired by the lack of RNA editing, thereby decreasing the tolerance to oxidative stress caused by water deficit. This would imply that the RNA editing machinery could act as a 'switch' to turn on defences when exposed to stress conditions. To evaluate the potential of RNA editing efficiency as a marker for stress tolerance, or as a target for genetic modification, the current investigation sought to establish whether there is a correlation between the editing efficiency of ndh genes and drought tolerance, in a number of L. perenne accessions.

Plant culture conditions and drought treatment
At least 12 plants of each accession were allowed to establish in a hydroponic system supplemented with 4.4 g L 21 Gamborg B5 medium + vitamins (Gamborg et al. 1968). The system was aerated by an aquatic pump to supply oxygen in the solution. Two separate systems were set up. After 1 week, the solution was refreshed to prevent depletion of nutrients. In the second week after experimental set-up, the drought stress experiment was initiated by replacing the solutions in both systems. In the first system the solution was replaced with 4.4 g L 21 MS (Murashige and Skoog 1962) salts; this system acted as the control. In the second system, the solution was replaced with a solution comprising 4.4 g L 21 MS salts supplemented with 20 % PEG-6000 (Duchefa cat. no. P0805) for the induction of drought stress. This concentration of PEG results in a water potential of about 20.45 MPa (Michel and Kaufmann 1973). The experiment was performed in a controlled glass house at Teagasc, Oak Park, Carlow, Ireland, with a mean daily temperature of 22 8C and supplemented with lighting (photosynthetically active radiation ¼ 650 mE m 22 s 21 ) for 16 h. After 2 weeks in these conditions, samples were taken for analysis.

Phenotypic growth assessment using pixel detection
Photographs were taken of the plants after the completion of the experiment and analysed using Adobe Photoshop 5.5. Pixel detection of leaf tissue and root tissue was performed ( Fig. 1), revealing the number of pixels in each photograph consisting of either leaf (Fig. 1C) or root tissue (Fig. 1D), thereby indicating the amount of tissue present after each treatment. Data from individual photographs were converted using a reference area in each photograph to detect the number of pixels therein (Fig. 1B), and the ratio of pixels between each photograph for this reference area represented the difference in the size of the photograph. A measure of relative biomass growth is represented by the ratio of pixels between treatments/accessions. The number of pixels within leaf tissue of individual plants was predicted by taking the total number of pixels consisting of leaf tissue divided by the number of plants in the photograph. For pixel detection of root tissue, individual plants could be assessed, since root tissue did not overlap between separate clones.

Relative water content measurements
Relative water content (RWC) was measured as described previously (Barrs and Weatherley 1962). It was calculated for each plant by taking a 2-cm leaf piece from the middle of the plant and weighing the fresh weight (FW), turgor weight (TW) and dry weight (DW) of these leaf tissues. The tissues were submerged in distilled water for 3 h, after which the seedlings were blotted dry. For these seedlings, the TW was measured on a fine scale. Subsequently, the seedlings were placed in an oven to dry at 70 8C, after which the DW was measured.
RWC is calculated using the formula:

Total dry biomass measurements
The total roots from each plant were harvested, wrapped in tin foil and dried in an oven at 70 8C for 3 days. The total root DW was recorded afterwards for each separate plant. The leaf DW could not be recorded, as the total leaf tissue was required to extract total RNA for RNA editing analyses.

cDNA preparations
Total RNA from three plants per treatment and per accession was extracted using the RNeasy w Plant Mini Kit from Qiagen (cat. no. 74903). mRNA extraction was performed using the manufacturer's instructions. During the RNA extraction, DNase was added. cDNA was synthesized from the RNA template using Superscript III reverse transcriptase following the manufacturer's instructions (Invitrogen cat. no. 18080-400).

Trace-file method to identify the editing efficiency
All PCR products were sent to a commercial sequencing company for PCR purification and sequencing. The returned trace files were analysed with the program 'Chromas Lite' for RNA editing sites and the corresponding efficiency. To analyse the efficiency of editing, the heights of the peaks at an editing location within the trace files AoB PLANTS www.aobplants.oxfordjournals.org were measured and compared with one another as illustrated in Fig. 2. The editing efficiency was calculated using the formula (Nakae et al. 2008): Editing efficiency = height edited peak height edited peak + height unedited peak × 100 % Verification of trace-file methodology by comparison with a colony screen Polymerase chain reaction products of the ndhB and one ndhF fragment were sequenced. Peaks in the resulting trace files were analysed using the formula stated above. The same PCR products were cloned into the cloning vector pCR2.1-TOPO, and subsequently introduced into Escherichia coli strain TOP10. Agar stab cultures were made in 96-well plates, each well containing a separate clone, derived from E. coli with pCR2.1-ndhB and E. coli with pCR2.1-ndhF. The clones containing the ndhB fragment were sequenced in both directions, whereas the clones containing the ndhF fragment were sequenced only in one direction. The editing efficiency was calculated as the percentage of clones containing the edited site, in comparison with the total number of clones. Colony screen results were compared with those of the trace-file method for quality assurance purposes. Peptide alignments of ndhB and ndhF were made in CLC Sequence Viewer.

Statistical analyses
The arcsine transformation and t-tests were conducted in Microsoft Excel, whereas the variance tests were performed in the program Minitab Solutions 15. All data sets were analysed for equal variance using the Levene test to determine whether t-tests were performed with equal or unequal variance. For the RWC analyses, arcsine transformation of the values was necessary to obtain a normal distribution to carry out statistical analyses, as percentage values cannot be used directly for comparison studies. These values were subsequently tested for statistical differences using the t-test with a one-tailed distribution and equal variance. A one-tailed distribution was chosen, as the hypothesis was that stressed plants would exhibit lower values compared with plants in control conditions.
For the total DW analyses, the results were tested for a normal distribution and subsequently tested for statistical differences using the t-test with a two-tailed distribution and unequal variance. A two-tailed distribution was chosen, based on the hypothesis that the root biomass could be either higher or lower for the stressed conditions compared with the control conditions.
For editing efficiency, the values were transformed into arcsine values. These values were subsequently tested for statistical differences using the t-test with a two-tailed distribution and unequal variance. A two-tailed distribution was chosen, based on the hypothesis that efficiency could be either lower or higher for plants under stressed conditions compared with plants in control conditions.

Characterization of plant growth, evaluated after a drought stress period
The average leaf biomass (Fig. 3) of accessions PI418701 and PI462336 was slightly decreased in leaf development under stress conditions, indicating a drought-tolerant response for clones within these accessions. All the other accessions had a clear decrease in biomass production under stress conditions, indicating a drought-susceptible response (Foito et al. 2009).
There was a statistically significant difference (P , 0.05) in RWC between stressed and non-stressed plants for Cv. 'Cashel', and accessions PI462336, PI231565 and PI632553 (Fig. 4), indicating that plants from these accessions were susceptible to drought stress.
The increase in root biomass under drought stress reflects an adaptive response involving an increase in root length to reach water deeper in the soil (van den Berg and Zeng 2006). Accessions PI611044, PI632553, PI223178, PI201187 and PI231565 showed a significant decrease in root development after exposure to drought stress compared with control conditions, indicating that these accessions are drought susceptible (Fig. 5). Accession PI418701 exhibited a slight increase in root growth during drought stress compared with controll conditions, suggesting a drought-tolerant response. Accessions PI462336 and PI610825 and Cv. 'Cashel' showed an apparent reduction in root development during stress, but this was not statistically significant. There was, however, a statistical difference in root dry biomass between treatments for accessions PI632553 and PI223178 (Fig. 6), further indicating that clones of these accessions were more susceptible to drought than the other accessions.

Overall response to drought stress and control conditions
The results from the preceding section were combined to determine how each accession was affected by drought stress. The responses are ranked in Table 1. Accession PI418701 was not subject to negative effects under drought stress, so the clones of this accession could be considered drought tolerant, while PI462336 was only mildly affected by drought and PI610825 had an intermediate negative response under drought stress. The following cultivars and accessions had an increasing drought-susceptible response, Cv. 'Cashel', PI231565, PI201187, PI611044, PI223178 and PI632553.

RNA editing evaluation
cDNA samples were randomly selected for analyses of RNA editing. A colony screen is considered a highly reliable method to determine the differences in RNA editing (Roberson and Rosenthal 2006). Results obtained from the colony screen were compared with those obtained from the trace-file method (results not shown). The  . Relative water content after 2 weeks of exposure to drought stress compared with control conditions. Error bars represent the standard deviation of the mean. **Statistical difference between stress and non-stress treatments according to a t-test (one-tailed distribution, equal variance on arcsine transformed values) at P , 0.05.
AoB PLANTS www.aobplants.oxfordjournals.org & The Authors 2013 highest difference observed between methods was a 10.8 % difference in editing, whereas the lowest difference observed was 0.8 %. This confirmed our confidence in the trace-file method for determining editing efficiency.
All observed editing events were C-to-U changes with a polar amino acid serine to a hydrophobic leucine conversion in ndhF and serine-to-leucine, proline-toleucine, histidine-to-tyrosine and serine-to-phenylalanine changes in ndhB ( Fig. 7A and B). In addition to the serineto-leucine change in ndhB, the serine-to-phenylalanine and histidine-to-tyrosine changes were changes from polar amino acids to hydrophobic amino acids. The most significant change is probably the proline-to-leucine amino acid change leading to a greater structural change and the function of the protein.
No statistically significant differences in RNA editing of any ndhB or ndhF transcripts were detected within accessions, between stressed and non-stressed clones, for any of the analysed editing sites (data not shown). However, there were significant and reproducible Figure 5. Mean root biomass per plant based on pixel detection. Statistical differences in the number of pixels between stress and non-stress treatments were calculated according to the t-test (two-tailed distribution, unequal variance). Error bars represent the standard deviation of the mean. Statistical differences at **P , 0.01, and *P , 0.05.  Table 1. Overall review of the results for the in vivo drought stress experiment. +, 'tolerant', i.e. no difference in response between stressed and non-stressed conditions; I, 'intermediate', i.e. modest difference in response between stressed and non-stressed conditions; -, 'sensitive', i.e. strong difference in response between stressed and non-stressed conditions. Overall response was scored by taking the average of +s, Is and -s.  AoB PLANTS www.aobplants.oxfordjournals.org PI632553 (data for the drought-stressed plants are shown in Fig. 8). The observed differences were dramatic. Some accessions showed almost complete editing, while other accessions almost completely lacked editing at these sites. The known editing site at genome position 103675 within the ndhF transcript showed a similar difference between accessions as was evident for the ndhB editing sites; however, the editing efficiency of accessions PI223178 and PI610825 was not statistically different (P , 0.05) from that of accession PI418701 (data for the droughtstressed plants are shown in Fig. 9).
Relationship between drought stress response and RNA editing efficiency within the ndhB and ndhF transcripts There was no correlation between drought tolerance and editing efficiencies for editing sites within the ndhB and the ndhF genes (see Table 2). Forexample, the two most droughttolerant accessions, PI462336 and PI418701, showed very different editing efficiencies, while the more susceptible accessions could be either efficiently or inefficiently edited.

Discussion
Editing efficiencies for the editing sites within ndhB and ndhF were evaluated within the accessions tested for drought tolerance, and subsequently compared with the respective drought tolerance of these clones. These results showed dramatic differences in the editing efficiency of these transcripts between accessions, but no alteration in efficiency in response to stress or any correlation between drought tolerance and editing efficiency within these ndh genes. Previous reports (Casano et al. 2000) have demonstrated that expression levels of plastid NDH complex genes were up-regulated during drought stress (a situation which causes photo-oxidative stress) and play a role in reducing PQ in conjunction with superoxide dismutase and hydroquinone peroxidase (Casano et al. 2000;Abdeen et al. 2010). The current finding does not contradict the involvement of the NDH complex in circumventing oxidative stress, but indicates that the RNA editing of the involved transcripts is not a major determining factor for regulation of this complex. Nevertheless, very low editing efficiencies (as low as 5 %) were observed within certain accessions. This may indicate that these genes are highly up-regulated, to the extent that, despite inefficient editing, enough functional transcripts are produced to allow for correct assembly of sufficient NDH complex to counter oxidative stress.
An alternative explanation could be that another pathway is more prominently involved in countering oxidative stress. This pathway could be the PGR5/PGRL1dependent route, also known as the non-NDH pathway (Rumeau et al. 2007;Suorsa et al. 2009). The involvement of this pathway with cyclic electron transfer was shown to be important under near-optimal conditions in Arabidopsis (Munekage et al. 2004). A recent publication showed that components of the PGR5/PGRL1 route were up-regulated during drought stress, whereas those of the NDH complex ndhH were not affected (Lehtimäki et al. 2006). The observed differences in RNA editing efficiencies may be due to different expression levels of the proteins involved in the editing of these specific editing sites. These may be genotype specific and unrelated to environmental stimuli. Another possibility could be the difference in the amount of transcripts of ndhB and ndhF available for editing. If there are fewer transcripts available, then the editing efficiency might increase. Both these explanations could contribute to the observed effects. Other studies have identified certain trans-factors that are essential for editing of certain sites; however, this does not exclude the possibility that other proteins may be involved in the editing machinery, as is implied by Chateigner-Boutin et al. (2008). The editing machinery can be limited by the least available protein within that editing complex. This was demonstrated when chimeric RNA was expressed containing the editing site of psbL in tobacco chloroplasts, and this led to a significant decrease in the editing efficiency of the endogenous psbL RNA. This competitive effect of the transgene was specific to the psbL gene, with other editing sites being properly edited, indicating depletion of the psbL-specific transacting factor (Chaudhuri et al. 1995).
Some proteins that bind to specific cis-factors surrounding the editing sites have been identified (Okuda et al. 2010). These belong to the PPR protein family. This large family of proteins is believed to be involved in RNA maturation in plastids and mitochondria (Shikanai 2006). In Arabidopsis mitochondria, gene knockout of a PPR did result in an altered drought stress response, in this case an enhancement, but this occurred in plants severely retarded in growth as a result of impaired mitochondrial NDH activity (Yuan and Liu 2012). It is difficult to draw parallels between gene knockouts impairing NDH activity in the mitochondria, where the complex has an essential bioenergetic function, and in the chloroplasts where it has a purely adaptive role, and can be abolished without phenotypic consequence in the absence of stress (Horváth et al. 2000).
While PPR proteins appear to have a role specific to individual editing sites, other proteins within the editing complex might have a more general function, and if knocked out, could impair the whole editing machinery. For example, when the CP31 protein in tobacco was  AoB PLANTS www.aobplants.oxfordjournals.org knocked out, editing within the psbL transcript was completely absent, while editing in the ndhB gene was partially impaired. Further elucidation of the regulation of RNA editing in perennial ryegrass plastids, and its consequences for the production of functional NDH complex, await further identification of all the trans-factors involved in the ndhB and ndhF genes in this species.

Conclusions
This study shows that different varieties or accessions of crop plants can differ markedly in the extent to which plastidial transcripts are edited, with some sites, in certain accessions, being edited to very low levels. However, in the case of genes in the NDH complex, associated with the oxidative stress response, there is no evidence that RNA editing makes a significant contribution to regulation. Up-regulation of the complex, associated with drought stress, is primarily mediated through other processes, which merit further investigation.

Sources of Funding
The work was funded by Teagasc (Ireland) through Walsh postgraduate Fellowships to Rob van den Bekerom and Kerstin Diekmann.