Defective chloroplast development inhibits maintenance of normal levels of abscisic acid in a mutant of the Arabidopsis RH3 DEAD-box protein during early post-germination growth


For correspondence (e-mail


The plastid has its own translation system, and its ribosomes are assembled through a complex process in which rRNA precursors are processed and ribosomal proteins are inserted into the rRNA backbone. DEAD-box proteins have been shown to play roles in multiple steps in ribosome biogenesis. To investigate the cellular and physiological roles of an Arabidopsis DEAD-box protein, RH3, we examined its expression and localization and the phenotypes of rh3–4, a T–DNA insertion mutant allele of RH3. The promoter activity of RH3 is strongest in the greening tissues of 3-day and 1-week-old seedlings but reduced afterwards. Cotyledons were pale and seedling growth was retarded in the mutant. The most obvious abnormality in the mutant chloroplasts was their lack of normal ribosomes. Electron tomography analysis indicated that ribosome density in the 3-day-old mutant chloroplasts is only 20% that of wild-type chloroplasts, and the ribosomes in the mutant are smaller. These chloroplast defects in rh3–4 were alleviated in 2-week-old cotyledons and true leaves. Interestingly, rh3–4 seedlings have lower amounts of abscisic acid prior to recovery of their chloroplasts, and were more sensitive to abiotic stresses. Transcriptomic analysis indicated that nuclear genes for chloroplast proteins are down-regulated, and proteins mediating chloroplast-localized steps of abscisic acid biosynthesis are expressed to a lower extent in 1-week-old rh3–4 seedlings. Taken together, these results suggest that conversion of eoplasts into chloroplasts in young seedlings is critical for the seedlings to start carbon fixation as well as for maintenance of abscisic acid levels for responding to environmental challenges.


Germination is the onset of development of the plant embryo from the mature seed when environmental and internal conditions are appropriate. Storage reserves in the seed are mobilized to support growth of a newly germinated seedling. However, this nutrient reserve is exhausted rapidly, and the development of chloroplasts is essential for the new seedling to become an autotroph (Rajjou et al., 2012). Dicotyledonous embryos such as those of Arabidopsis have normal chloroplasts, but these de-differentiate into eoplasts when the embryo cells become dormant (Ruppel et al., 2011). The eoplasts in the cotyledons are converted back into chloroplasts, and the new chloroplasts proliferate when conditions allow germination to proceed. Plastids account for less than 1% of the cotyledon cell volume in the Arabidopsis seed, but they enlarge, divide and differentiate into chloroplasts within 48 h after imbibition, increasing their volume in the cell by 70-fold (Mansfield and Briarty, 1992, 1996).

Integrated activity of the chloroplast genome and the nuclear genome is required for normal development of chloroplasts. Although the chloroplast/plastid is a semi-autonomous organelle with its own genome and protein synthesis machinery, more than 90% of plastid proteins are encoded by the nuclear genome and have to be imported into the chloroplast from the cytosol (Lee et al., 2009; Andres et al., 2010). Multi-protein complexes in the chloroplast, such as ribosomes and Ribulose-1,5-bisphosphate carboxylase oxygenase (RuBisCo), consist of subunits encoded by nuclear and plastid genomes, and it has been shown that plastid defects are associated with the reduced expression of nuclear-encoded plastid proteins (Inaba and Schnell, 2008; Bang et al., 2012). Furthermore, plant cells have metabolic pathways comprising reactions that take place in the cytosol and the plastid, and such coupling of metabolic reactions necessitates the expression of plastid enzymes in coordination with cytosolic enzymes of the pathway. Not surprisingly, plant cells have evolved mechanisms to control nuclear gene expression in accordance with the developmental and metabolic status of the plastid to ensure appropriate assembly of such chloroplast macromolecular complexes and the efficient regulation of metabolic pathways that cross the chloroplast membrane (Pogson et al., 2008).

Abscisic acid (ABA) is a plant hormone that is required for several aspects of plant development. ABA synthesis is most active in embryos entering dormancy and in plants under adverse environmental conditions (Xiong and Zhu, 2003). However, low levels of ABA are maintained in vegetative tissues under non-stressful conditions, and ABA-deficient mutant plants display reduced growth phenotypes, suggesting that ABA is required for normal plant growth (Finkelstein and Rock, 2002; Lee et al., 2006). The first ABA synthesis-specific reaction is mediated by ABA1, and this step and subsequent steps until xanthoxin production take place in the chloroplast (Seo and Koshiba, 2002; Nambara and Marion-Poll, 2005). Enzymes catalyzing ABA synthetic reactions in the chloroplast but encoded by nuclear genes such as ABA1 must be imported into the chloroplast. ABA biosynthesis illustrates the importance of collaboration between the nuclear genome and the plastid genome for hormone homeostasis in the plant cell.

Plastids have ribosomes that are related those of prokaryotes, in agreement with the plastid's cyanobacterial origin (Harris et al., 1994). The original transcripts from ribosomal RNA (rRNA) genes are large precursors that are cleaved and spliced to form mature rRNA molecules. A group of RNA helicases, namely DEAD-box RNA helicases, play important roles for ribosomal RNA processing (Rocak and Linder, 2004). The Arabidopsis genome contains 58 DEAD-box genes (RH1–RH58; 56 intact genes and two pseudogenes) (Mingam et al., 2004), and three of them have been implicated in chloroplast ribosome biogenesis. RH39 binds to 23S chloroplast rRNA and is required for cleavage of the rRNA into its mature form. rh39 mutant plants contain a low level of RuBisCO because inactivation of RH39 causes wholesale inhibition of translation in the chloroplast (Nishimura et al., 2010). RH22, another chloroplast-localized DEAD-box protein, directly interacts with the 23S rRNA assembly intermediate as well as the RPL24 ribosomal protein, and disruption of RH22 causes aberrant chloroplast development (Chi et al., 2011). Recently, it was shown that RH3 is implicated in the splicing of group II introns in 23S rRNA and in the splicing of RNAs for ribosomal proteins (Asakura et al., 2012). These observations suggest disruption in ribosome biogenesis in the chloroplast in the rh3 mutant.

Here, we examined the defective chloroplast development and stress response of an RH3 mutant line, rh3–4. RH3 is highly expressed in greening tissues of germinating seedlings, and chloroplasts in the rh3–4 mutant have fewer ribosomes, lower amounts of chlorophyll, and poorly developed thylakoids. rh3–4 seedlings displayed ABA-deficient phenotypes. ABA levels were lower in the mutant seedlings, and supplementation of ABA in the growth medium partially rescued the phenotypes of the mutant. Transcriptomic analysis indicated that transcripts for plastid-localized enzymes involved in ABA biosynthesis are suppressed in the mutant seedling.


RH3 is a chloroplast DEAD-box protein that is highly expressed in germinating seedlings

At5g26742 encodes a DEAD-box protein consisting of 748 amino acids with a chloroplast targeting sequence at its N–terminus (Figure 1a). We call the gene RH3, according to the nomenclature proposed by Boudet et al. (2001) who systematically compiled the Arabidopsis DEAD-box genes. The SMART domain search program (Schultz et al., 2000) identified the HELICc domain and the GUCT domain in the RH3 amino acid sequence. These domain sequences often accompany the DEAD-box domain.

Figure 1.

GUS promoter activity assay for RH3, and subcellular localization of RH3. (a) Domain architecture of the RH3 protein. The chloroplast targeting sequence (CTS, green), the DEAD-box domain (purple), the helicase superfamily C–terminal (HELICc) domain (dark blue) and the zinc binding motif (light blue) are indicated. (b–f) Expression patterns of RH3pro:GUS fusion in young tissues. Arrows indicate higher expression regions throughout the cotyledon in (d) and in the rosette margin of 2-week-old seedlings in (f). (g,h) Immunoelectron microscopy localization of RH3–FLAG. A 1-week-old rh3–4 seeding expressing the RH3–FLAG fusion protein was labeled with FLAG-specific immunogold particles. The gold particles are seen in the chloroplast stroma. Ch, chloroplast; Cy, cytosol; St, starch. Scale bars = 500 nm.

To examine the tissue-specific expression patterns of RH3, we generated transgenic plants expressing a transcriptional fusion construct with the RH3 promoter (2.0 kb) and the β–glucuronidase (GUS) reporter gene. T2 progeny from three independent transgenic RH3p-GUS lines were used for GUS activity analysis (Figure 1b–f). GUS staining was strongest in the cotyledons and hypocotyls of germinating seedlings (Figure 1b). The GUS activity became weak in the cotyledons and hypocotyls in 1-week-old seedlings, and decreased further in true leaves in older seedlings (Figure 1c–e). Interestingly, GUS expression was detected in the apical zone of the first pair of true leaves (Figure 1(f)). GUS activity was also detected in various tissues of mature plants, including the inflorescence stem, stigmata, anthers and petals (Figure S1).

To determine the subcellular localization of RH3, we performed immunoelectron microscopy with transgenic Arabidopsis plants expressing an RH3::FLAG fusion protein under the control of its native promoter. FLAG-specific immunogold particles were observed mostly in the chloroplast stroma of the transgenic cotyledon cells (Figure 1g,h). The RH3::FLAG construct complemented the phenotypic defects of rh3–4 mutant seedlings (Figures 2 and 3). These results indicate that RH3 is a chloroplast-targeted DEAD-box protein.

Figure 2.

Phenotypes of the rh3–4 T–DNA insertion mutant. (a) The exon–intron structure of the RH3 gene showing the T–DNA insertion site (triangle). Black boxes represent exons, and white boxes at the 5′ and 3′ ends represent untranslated regions. The arrows indicate the positions of primers used for RT–PCR in (b). LB, T–DNA left border sequence. (b) RT–PCR analysis of the RH3 transcript in the homozygous rh3–4 mutant seedlings. Note that the region spanning the T–DNA insertion site is not amplified. The TUBULIN8 gene was used as a control. (c) Pale cotyledons of rh3–4 mutant seedlings. Cotyledons of 3-day-old and 1-week-old mutant seedlings are yellow (first and second columns). Cotyledons are indicated by arrows. (d) Reduced chlorophyll contents in rh3–4 seedlings. Relative chlorophyll contents in the mutant seedlings are shown after setting the chlorophyll contents in wild-type seedlings to 1. (e,f) Lesions in the first pair of true leaves of rh3–4 seedlings. Trypan staining of the leaves shows that the apical zone has dead cells (arrowheads) and regions in which dead cells have been cleared away during staining and washing [arrowheads in (e‘)]. Abnormal cell death is seen only in the first pair of true leaves [arrows in (e“)]. Wild-type seedlings do not display such abnormalities.

Figure 3.

Complementation of rh3–4 with the RH3pro:RH3–FLAG transgene. (a) Wild-type (WT), homozygous rh3–4 and homozygous rh3–4 transformed with the RH3pro:RH3–FLAG construct. Three transformed lines were chosen and further characterized. (b) PCR genotype analysis of wild-type (WT), homozygous and complemented rh3–4 seedlings using RH3 gene-specific primers (first row, RH3), T–DNA-specific primers (second row, rh3–4 T–DNA), and transgene-specific primers (third row, transgene). PCR amplification of genomic DNA from the three complemented lines confirmed that they are homozygous for the T–DNA insertion in RH3 and that they contain the RH3–FLAG construct. (c) Immunoblot analysis of chloroplast proteins. Protein extracts from 1-week-old wild-type (WT), rh3–4 and complemented rh3–4 seedlings were probed with antibodies against FLAG, RBCL (RuBisCO large subunit), LHCP (light-harvesting chlorophyll a/b protein), CRD1 (chloroplast-localized di-iron protein), OE23 (23-kDa oxygen evolving protein), and HSP70 (heat shock protein 70). PBA (proteasome β subunit A) and tubulin (TUB) were used as controls.

Disruption of RH3 causes defective chloroplast development and abnormal cell death in young seedling leaves

To understand the developmental and cellular roles of RH3, we obtained a T–DNA insertion mutant line of RH3 from the Arabidopsis Biological Resource Center (SALK_emb1138–1, CS16011). This line is a weak mutant allele of RH3, and was named rh3–4 by Asakura et al. (2012). The T–DNA is inserted into intron 9 of RH3 (Figure 2a), with T–DNA left border sequence on both sides. RT–PCR analysis indicated that full-length RH3 mRNA is not synthesized in homozygous rh3–4 mutant plants (Figure 2b).

Cotyledons and young leaves are pale in homozygous rh3–4 mutant seedlings (Figure 2c). The cotyledons and upper hypocotyl of 3-day-old rh3–4 homozygous mutant seedlings were albino, while those of 3-day-old wild-type seedlings were green. Cotyledons remained pale green in 1-week-old mutant seedlings, and the seedlings grew more slowly than wild-type seedlings. Although the albino phenotype of the cotyledon alleviated in 2-week-old seedlings, the first pair of true leaves were stunted and lighter green in color. They had short petioles and pale peripheries with irregular margins (Figure 2c). The apical region of the mutant leaf was yellow but was speckled with green spots (Figure 2d). Trypan blue staining of the leaf revealed dead cells in the apical periphery as well as empty patches in the upper half where dead cells had been cleared away (Figure 2e). This abnormal cell-death phenotype was not observed in true leaves of wild-type seedlings. The albino phenotypes and the irregular leaf morphology were not present in the second pair of true leaves of 2- and 3-week-old rh3–4 seedlings, although the leaves continued to be smaller and their petioles were shorter in the mutant (Figure 2c,e). The chlorophyll contents of the rh3–4 mutant seedlings at 1, 2 and 3 weeks after growth commenced were approximately 10, 35 and 70% of the chlorophyll contents of wild-type seedlings, respectively, in agreement with the gradual recovery in older leaves (Figure 2f). rh3–4 mutant plants grew normally after the seedling stage and produced viable homozygous mutant seeds. We did not notice any abnormality in root development of rh3–4 plants except that their primary root elongation was slower from approximately 1 week after growth commenced (Figure S2).

To confirm that the phenotypes observed in the rh3–4 seedlings are due to the disruption of RH3, we transformed rh3–4 mutant plants with a construct containing the RH3 cDNA fused to its 2 kb promoter region at the 5′ end and the FLAG tag at its 3′ end. PCR genotyping results verified that transgenic rh3–4 seedlings that did not exhibit the mutant phenotypes carried the complementation construct (Figure 3a,b). All the rescued rh3–4 seedlings were positive for expression of the RH3–FLAG fusion protein when verified by immunoblot analysis (Figure 3c). Immunoblot analysis of 1-week-old seedlings indicated that the levels of five chloroplast proteins involved in photosynthesis, chlorophyll biosynthesis, chloroplast protein import and thylakoid membrane assembly were reduced. Together with the lower chlorophyll contents, the reduced levels of chloroplast proteins indicate that chloroplasts are severely affected in rh3–4 mutant seedlings.

Microscopic analysis of rh3–4 mutant chloroplasts

We compared microscopic features of chloroplasts in 1-week-old cotyledons of the rh3–4 mutant and those of wild-type seedlings using confocal laser scanning microscopy. Wild-type chloroplasts were round, and their sizes ranged between 5.0 and 9.1 μm (mean 6.7 μm, standard deviation 1.0 μm) (Figure 4a,c). Chloroplasts in rh3–4 mutant cotyledons were smaller, with diameters ranging between 2.0 and 5.2 μm (mean 3.4 μm, standard deviation 0.86 μm), and were irregularly shaped (Figure 4(b,d)). In high-magnification micrographs, the stroma of mutant chloroplasts appeared coarse, and some chloroplasts were curved (arrows in Figure 4(d)). Autofluorescence from the mutant chloroplasts was dimmer, in agreement with their lower chlorophyll levels (Figure 2b,d).

Figure 4.

Abnormal chloroplasts in the rh3–4 cotyledon. (a–d) Chloroplasts of WT and rh3–4 cotyledon mesophyll cells (1-week-old) visualized by chlorophyll autofluorescence. The photomultiplier gain was set 50% higher for the micrograph in (d) than for the micrograph in (c) in order to visualize rh3–4 chloroplasts, because autofluorescence in the mutant chloroplast was lower. (e,f) TEM analysis of rh3–4 chloroplasts. (e) Wild-type chloroplast in a 1-week-old cotyledon. (f) rh3–4 chloroplast in a 1-week-old cotyledon. (g,h) Chloroplasts from 2-week-old rh3-4 mutant cotyledons. Thylakoid stacks are marked with brackets. (i,j) Wild-type chloroplast in a 3-day-old cotyledon (i), and rh3–4 chloroplast in a 3-day-old cotyledon (j). Starch granules and primitive thylakoids are seen in the mutant chloroplast and the wild-type chloroplast. (k,l) High-magnification electron micrographs of chloroplasts in 3-day-old wild-type (k) and rh3–4 (l) cotyledons. Four ribosome in the wild-type chloroplast stroma are marked with red circles in (k). The brackets in (k) indicate where the thylakoid membrane begins to stack. Very few ribosomes are observed, and stacking of the thylakoid membrane is not seen in the mutant chloroplast [arrowhead in (l)]. Scale bars = 50 μm (a), 5 μm (c,d,g), 1 μm (e,f,h) and 300 nm (i–l). Ch, chloroplast; St, starch.

We then examined the ultrastructure of the mutant chloroplasts by transmission electron microscopy (TEM). The chloroplasts of 1-week-old wild-type cotyledons had a smooth ellipsoid profile, and their internal membrane system was well developed (Figure 4e). Chloroplasts in 1-week-old mutant cotyledons were small and had tubules enclosing cytosol in their peripheries (arrows in Figure 4f). Thylakoid development is stunted in the mutant. The thylakoids were convoluted and failed to form normal grana consisting of compact layers of membrane Starch granules were absent in the 1-week-old mutant chloroplasts. In 2-week-old seedlings, chloroplasts in the mutant cotyledon increased in size (mean 4.6 μm, standard deviation 0.81 μm), but were still significantly smaller than chloroplasts in wild-type cotyledons (Figure 4g). Electron microscopy imaging revealed that normal-looking grana start to form in the mutant chloroplasts, and they are oriented in parallel with the longer axis of the chloroplast ellipsoid as in the wild-type chloroplast (Figure 4h).

In 3-day-old seedlings, both wild-type and rh3–4 mutant chloroplasts had starch granules and tubular thylakoid membranes in the stroma (Figure 4i,j). The thylakoids were thicker in wild-type chloroplasts because they contain stacks of 2–3 membrane tubules (Figures 4i,k) and 5a) In contrast, the thylakoids in mutant chloroplasts consist of simple tubules (Figures 4i and 5b). In higher-magnification electron micrographs from 3-day-old cotyledon cells, the stroma of wild-type chloroplasts was filled with ribosomes, identified as small black dots with diameters of approximately 15 nm (Figure 4k). Interestingly, ribosomes were rarely identified in the stroma of rh3–4 mutant chloroplasts (Figure 4l). Mitochondria and endomembrane compartments appeared normal in rh3–4 mutant cotyledon samples prepared from the three developmental stages.

Figure 5.

Quantitative analysis of ribosome density and ribosome size using electron tomography. (a–f) Electron tomographic slices showing chloroplasts of 3-day-old (a,b) and 1-week-old (c,d) cotyledons. In each panel, low-magnification (left) and high-magnification (right, boxed area in the left image) tomographic slices are shown. Groups of ribosomes are enclosed by circles in the high-magnification images. Thylakoid stacking is delayed in the mutant [brackets in (a), (c) and (d)]. A tubular thylakoid in the mutant chloroplast is marked with an arrowhead in (b). Ribosomes are denoted with arrows in (e) and (f). (g-j) Histograms showing ribosome densities (g and h) and ribosome sizes (i and j). In each histogram, measurements from wild-type (WT) and rh3-4 mutant samples at three developmental stages are compared. Ch, chloroplast; L, lipid body; M, mitochondrion.

Ribosomes are fewer and smaller in the rh3–4 mutant chloroplast than in the wild-type chloroplast

We used electron tomography to better visualize ribosomes in the chloroplasts. Individual ribosomes are not clearly resolved in conventional micrographs because ribosome diameters are smaller than the thickness of regular TEM sections. One ribosome may lie on top of another ribosome, blocking the view of the ribosome located below. This problem may be overcome by means of electron tomography methods that generate 2 nm slices from TEM sections. Individual ribosomes are unambiguously observed in 6–7 sequential 2 nm thick tomographic slices (Kang and Staehelin, 2008).

Three samples from 3-day-old, 1-week-old and 2-week-old cotyledon parenchyma cells were analyzed by electron tomography (Figure 5a–f). The stromal ribosome densities in the mutant cells were only 17% (3179 per μm3 versus 18 511per μm3) and 41% (7770 per μm3 versus 18 808 per μm3) of those in wild-type cells after 3 days and 1 week of growth, respectively (Figure 5g). Ribosome densities in the cytosol were comparable in the wild-type and mutant at all stages (Figure 5h).

We also noticed that ribosomes in the mutant chloroplasts were smaller than ribosomes in the wild-type chloroplasts (Figure 5e,f). Ribosomes in the cytosol were approximately 14 nm in diameter in both mutant and wild-type cells. In contrast, ribosomes in the chloroplasts of 3-day-old mutant seedlings have diameters of approximately 10 nm (Figure 5e,i). The smaller size of the ribosomes may be due to failure in the assembly of 70S ribosome particles or failure in processing intermediates of ribosome subunits. However, in the cotyledon cells of 1-week-old seedlings, most chloroplast ribosomes were as large as wild-type ribosomes, although the ribosome density in the mutant chloroplasts was still less than half that in the wild-type chloroplasts (Figure 5g,i). Thus, electron tomography analysis provides evidence that RH3 is involved in chloroplast ribosome biogenesis of germinating seedlings and indicates that the defective chloroplast ribosome biogenesis of rh3–4 starts to be alleviated in 1-week-old seedlings.

To further examine chloroplast ribosome biogenesis in the rh3–4 mutant, we performed RNA blot analysis using probes that recognize chloroplast rRNA molecules (Figure 6a). Consistent with the electron tomography imaging results (Figure 5), which showed a decrease in chloroplast ribosome density, the amounts of chloroplast rRNA were significantly reduced. We did not detect unusual forms of chloroplast rRNA in total RNA samples from the 3-day-old rh3–4 mutant when equal amounts of samples were analyzed (Figure 6b).

Figure 6.

RNA blot analysis of the chloroplast rRNA operon. (a) Chloroplast rRNA operon of Arabidopsis and rRNA probes used in (b). Exons and introns are indicated by black boxes and black lines, respectively. The lines above the exons indicate the regions from which probes were prepared. (b) RNA blot analyses of chloroplast rRNA. Cytosolic 25S rRNA amounts are shown below as loading controls. To better visualize chloroplast rRNA from mutant samples, twice the amount of total RNA from the mutant was blotted compared to the amount used for the wild-type.

rh3–4 mutant seedlings display ABA-deficient phenotypes

rh3–4 seedlings were more sensitive to dehydration stress than wild-type seedlings (Figure S3). Their growth retardation and yellowing of leaves are reminiscent of the phenotypes of ABA-deficient mutants (Lee et al., 2006). We quantified ABA in 1-week-old wild-type and rh3–4 seedlings using an immunoassay method. Levels of ABA in rh3–4 seedlings were 56.6% of those in wild-type seedlings (Figure 7a, left). In 2-week-old mutant seedlings, in which the defective chloroplasts have recovered, the ABA level approaches that of wild-type seedlings (Figure 7a, right).

Figure 7.

Stress- and ABA-related phenotypes of the rh3–4 mutant. (a) Reduced ABA contents in rh3–4 seedlings. One-week-old seedlings of wild-type and the rh3–4 mutant line were analyzed for ABA quantification. Mean amounts of ABA per mutant seedling (= 4) are shown after setting the mean ABA amount per wild-type seedling (= 4) to 1.0. Error bars show standard deviations of each measurement. (b) Sensitivity of rh3–4 plants to salt stress. Three-week-old wild-type plants were grown on soil without (top left) or with (top right) NaCl stress for 2 weeks. Three-week-old rh3–4 plants were grown on soil without (bottom left) or with (bottom right) NaCl stress for 2 weeks. (c) Resumption of growth after removing NaCl stress. Most wild-type plants resumed growth after the salt stress was released (top), but half of the mutant plants died (bottom). (d) Rescue of rh3–4 phenotypes by exogenous ABA treatment. The first true leaves of wild-type, rh3–4, rh3–4 grown with 0.5 μm ABA, and rh3–4 grown with 1.0 μm ABA (from left to right) are shown. Trypan blue staining was performed to visualize dead cells and lesions in the leaves (bottom panels). Note that leaves are larger and free of dead cells when grown with 1.0 μm ABA in comparison with mutant leaves in the second column. Scale bars = 5 mm (b) and 2 mm (d). Images of wild-type and rh3–4 seedlings were taken at the same magnifications.

To further assess ABA deficiency in the mutant, we exposed 1-week-old rh3–4 seedlings in soil to 200 mm NaCl for 2 weeks, and then grew the stressed seedlings without salt for 1 week (Figure 7b). During the stress period, the shoot apical meristem of wild-type seedlings produced several pairs of leaves, but leaves produced by mutant seedlings were severely stunted. Most of the wild-type seedlings (32 of 36 seedlings) survived the stress and resumed growth when the salt stress was removed. In contrast, 50% of rh3–4 seedlings (18 of 36 seedlings) failed to resume growth and subsequently died, indicating that they were irreversibly damaged by the salt stress (Figure 7c).

We grew rh3–4 seedlings on plates supplemented with ABA to examine whether exogenous application of ABA influenced the rh3–4 phenotypes. The extra ABA promoted leaf expansion in mutant seedlings, partially rescuing the growth retardation phenotype. In addition, the abnormal cell death in the apical zone of the first pair of true leaves was alleviated (Figure 7d). The effects of ABA supplementation increased in a concentration-dependent manner up to 1.0 μm.

Expression of nuclear genes encoding chloroplast proteins is suppressed in the rh3–4 mutant

To investigate the effect of chloroplast disruption on gene expression in germinating seedlings, we performed RNA-seq analysis of 1-week-old rh3–4 and wild-type seedlings using Ion Torrent sequencing technology (Life Technologies, Two independent experiments were performed, and, in each experiment, we obtained 100 000–200 000 high-quality reads that mapped to the five annotated Arabidopsis chromosomes (Accession numbers: NC_003070, NC_003071, NC_003074, NC_003075, and NC_003076). (Table S1). To remove background noise of the experiment, nuclear genes that have RPKM (reads per kilobase pair exon model per million mapped reads) values larger than 20 in either wild-type or mutant data were chosen for further analysis (6038 genes from experiment 1 and 2315 genes from experiment 2). Differential expression of the selected genes in the mutant and wild-type was assessed using the mapman pathway viewer (Thimm et al., 2004). Nuclear genes encoding proteins that are involved in processes in the chloroplast such as the photosynthesis light reaction, the Calvin cycle, photorespiration and tetrapyrrole synthesis are down-regulated in the mutant seedlings (Figure 8a, Figure S3 and Table S3).

Figure 8.

Genome-wide RNA-seq analysis of nuclear mRNA. (a) Heat map of metabolic pathways showing a general suppression of genes involved in chloroplast-specific pathways including the photosynthesis light reaction, Calvin cycle, photorespiration and tetrapyrrole synthesis in rh3–4 compared with wild-type 1-week-old seedling samples. The diagram was obtained using the mapman Bin analysis tool (Thimm et al. 2004). Red and green squares indicate genes with reduced and increased transcript accumulation, respectively. The heat map scale is on the logarithmic scale (base 2), and is shown in the lower right corner. (b) ABA synthesis pathway showing suppression of ABA1 (sixfold) and NCED4 (2.5-fold). Identification numbers and expression values of the genes shown in the pathways are listed in Table S2. The fold changes in rh3-4 are listed in Table S3.

We also examined hormone biosynthesis pathways in the mapman viewer to determine whether the abscisic acid (ABA) biosynthesis pathway is suppressed in the mutant (Figure 8b and Figure S4). Expression of ABA1, the first enzyme of the pathway, and NCED4 (9–cis–epoxycarotenoid dehydrogenase 4), a member of the dehydrogenase family that cleaves 9–cis–carotenoids to xanthoxin, was suppressed. Genes in other hormone biosynthesis pathways had RPMK values smaller than 20 or did not display consistent suppression/up-regulation in the two RNA-seq experiments. Semi-quantitative RT–PCR analysis of several genes that displayed differential expression levels between the mutant and wild-type validated our RNA-seq results (Figure S5).


Plastids are dynamic organelles that can grow, divide and inter-convert into various forms (Sakamoto et al., 2008; Pogson and Albrecht, 2011). The remodeling of dormant eoplasts in the dry seed into active chloroplasts in green seedlings requires massive expression of chloroplast proteins from both the nuclear genome and the plastid genome. Ribosome production is inhibited in the rh3-4 mutant chloroplast in the first two weeks after germination. This leads to delay in chloroplast development and suppression of nuclear genes encoding chloroplast proteins during the period. One of the consequences of the defective chloroplast ribosome biogenesis is that ABA biosynthesis is impaired in the mutant. Our analytical study results for the rh3–4 mutant suggest that formation of chloroplasts in young seedlings is required not only for photosynthetic competency but also for ABA synthesis, both of which are required for normal post-germination growth.

Three Arabidopsis DEAD-box proteins, RH39, RH22 and RH3, have been implicated in ribosome production in the chloroplast. In all three mutants, precursor forms of 23S rRNA accumulate because processing of the rRNA is inhibited by inactivation of the DEAD-box genes (Nishimura et al., 2010; Chi et al., 2011; Asakura et al., 2012). The three proteins interact directly with 23S rRNA and/or ribosomal proteins, and are thought to facilitate assembly of the 50S ribosome subunit. We have shown that RH3 is required for ribosome biogenesis in the cotyledon during the first 2 weeks of seedling growth using electron tomography and RNA blot analysis. Asakura et al. (2012) examined chloroplast ribosome biogenesis in rh3–4 plants at stages later than we did, and successfully demonstrated that RH3 is involved in splicing of group II introns. We were not able to detect accumulation of precursor forms of rRNA in 1-week-old rh3–4 mutant seedlings because of a severe decrease in the amounts of all types of chloroplast rRNA molecules (Figure 6(b)). Among rh3 mutant alleles, rh3–4 is a weak allele in which expression of RH3 is not completely inactivated (Asakura et al., 2012). The weak RH3 activity appeared to be sufficient for producing ribosomes in the chloroplast of rosette leaves, even though the inefficient ribosome processing leads to accumulation of intermediate forms of rRNA to a degree that can be detected by RNA blot analysis. A shortage of ribosomes in the chloroplasts of young cotyledons and the gradual replenishment of ribosomes in later stages of growth suggest that strong RH3 activity is required to sustain the rapid transformation of eoplasts into chloroplasts in germinating seedlings. The profile of the RH3 promoter activity from 3 days to 3 weeks of growth observed by GUS histochemical staining agrees with this possibility (Figure 1b–f).

It is well known that expression of nuclear genes and plastid genes is coordinated, and the two organelles exchange information about their status to facilitate collaboration between the two genomes (Kleine et al., 2009). Our transcriptomic analysis confirms the coordination between the two organelles in germinating Arabidopsis seedlings. As a result of the scarcity of ribosomes in the rh3–4 chloroplast, translation in the stroma is blocked, causing serious problems for normal chloroplast development. At the same time, nuclear genes for photosynthesis, chlorophyll synthesis and photorespiration are strongly down-regulated (Figure 8(a)). In addition to the chloroplast-specific pathways, the ABA biosynthetic pathway, whose reaction steps extend from the plastid to the cytosol, is also affected. The mutation in RH3 leads to suppression of plastid-localized enzymes of the pathway, including ABA1 and NCED4. The two enzymes are thought to be important for regulating ABA biosynthesis as their expression levels correlate with levels of ABA (Seo and Koshiba, 2002; Wasilewska et al., 2008).

ABA is an isoprenoid phytohormone derived from carotenoids. The plastidic terpenoid pathway provides carotenoid precursors for ABA as well as other phytohormones such as brassinosteroids, cytokinins and gibberellins (Crozier et al., 2000). Therefore, we cannot rule out the possibility that metabolism of multiple hormones may be affected in the mutant. However, analysis of differentially expressed genes in rh3–4 seedlings using the mapman pathway analysis program indicates that key genes of ABA biosynthesis display high expression levels in wild-type and are consistently suppressed in the mutant. Transcriptomic analysis of a mutant termed scabra3 (sca3) supports the association of the ABA synthesis defect with aberrant chloroplast formation in the Arabidopsis seedling. SCA3 is an Arabidopsis nuclear gene encoding a plastid RNA polymerase, and is highly expressed in germinating seedlings (Hricova et al., 2006). Inactivation of the gene causes phenotypes similar to those of rh3–4, and microarray analysis of the mutant identified ABA1 and NCED4 (among other genes involved in isoprenoid biosynthesis) as down-regulated in sca3 seedlings.

Seed germination is suppressed by ABA, but once seed-to-seedling transition has been completed, ABA contributes to adaptation of young seedlings to environmental stresses (Chen et al., 2008). Furthermore, failure to maintaining a normal level of ABA leads to growth retardation (Xiong and Zhu, 2003; Sreenivasulu et al., 2012), and leaf cell death has been associated with ABA deficiency (Dong et al., 2007). In this study, we showed that a delay in chloroplast ribosome biogenesis in germinating seedlings of rh3–4 makes the young seedling more vulnerable to dehydration and high salt stresses. rh3–4 is a virescent mutant that develops normally once past the seedling stage if grown under the laboratory conditions. The inability of rh3–4 to adapt to adverse growth conditions suggests that failure to sustain normal levels of ABA may mean that young seedlings are unable to survive in the natural environment.

Experimental procedures

Plant growth conditions

An RH3 T-DNA insertion mutant line obtained from the Arabidopsis Biological Resource Center (rh3–4, SALK_emb1138-1; ecotype Col–0) was back-crossed three times to wild-type Col–0 plants. Because homozygous rh3–4 plants are fertile, it is possible to harvest homozygous rh3–4 seeds from self-fertilized homozygous plants. The primers used for genotyping the rh3–4 mutant were P9 (T–DNA left border), P10 and P11 (listed in Table S4). Wild-type and rh3–4 seedlings were grown on B5 medium (Sigma-Aldrich) with 1% sucrose, 0.5 g l−1 MES (pH 5.7) and 0.75% agar. Seedlings for soil assays were sown directly in soil (Fafard germination mix, and grown for 3 weeks at 22°C.

Complementation of rh3–4

The promoter region, full-length RH3 cDNA and the FLAG tag were fused by PCR using primers P14, 15, 16 and 17 (Table S4). The recombinant DNA was inserted in to the pCAMBIA3301 binary vector ( and the construct was introduced into homozygous rh3–4 seedlings by the floral-dip method using Agrobacterium tumefaciens (Clough and Bent, 1998). Normal-looking seedlings that were resistant to BASTA (BioWorld, were selected and were genotyped by PCR to identify rh3–4 homozygous mutants.

GUS promoter activity assay

The native promoter of RH3 (2 kb) was amplified using primers P12 and P13 and cloned in front of the GUS open reading frame in the pCAMBIA 3301 vector. Transformation of wild-type Col–0 plants and GUS staining were performed as described by Lee et al. (2012). T2 seedlings and plants from 15 T1 lines were examined.

Trypan blue assay

The staining solution was prepared by diluting the trypan blue stock solution (10 ml glycerol, 10 ml lactic acid, 10 ml water, 10 g phenol and 0.02 g trypan blue) twofold using 95% ethanol. Wild-type or rh3–4 samples were stained by boiling them in the staining solution for 1 min on a heating block. The samples were cooled down in the staining solution and de-stained by incubating in chloral hydrate solution (250 g chloral hydrate in 100 ml distilled water).

Immunoblot analysis and RNA blot analysis

Preparation of total protein extracts from wild-type, rh3–4 and transgenic plants expressing RH3:FLAG and immunoblot analysis were performed as described by Lee et al. (2012). Anti-PBA1 antibody and anti-RBCL (RuBisCO large subunit) antibody were purchased from Abcam ( Antibodies against LHCP (light-harvesting chlorophyll a/b protein), CRD1 (chloroplast-localized di-iron protein), OE23 (23-kDa oxygen evolving protein), and HSP70 (heat shock protein 70) were as described by Payan and Cline (1991), Tottey et al. (2003), Henry et al. (1997) and Yuan et al. (1993), respectively. Anti-FLAG-tag antibody (catalog number F3165) and anti-tubulin antibody (catalog number T6199) were purchased from Sigma-Aldrich ( For RNA blot analysis, total RNA samples were isolated from aerial tissues of 1-week-old wild-type and rh3–4 seedlings using the TRIzol Plus RNA purification system (Life Technologies, Total RNA samples (1 μg) were analyzed using the chloroplast rRNA probes as described by Kang et al. (2003).

Stress treatment, ABA treatment, and ABA ELISA assay

The sensitivity of rh3–4 seedlings to salt stress was assessed by watering the plants with 200 mm NaCl twice a week for 2 weeks. After the salt stress, the seedlings were grown under normal conditions for 1 week. Treatment of rh3–4 plants with various concentrations of ABA was performed by transferring the 3-day-old seedlings to ABA-containing plates and then growing tem for 2 weeks. ABA measurement was performed using an ABA ELISA kit from CUSABIO (catalog number CSB-E09159PI, according to the manufacturer's instructions.

Confocal laser scanning microscopy and electron microscopy

Cotyledons from 1 and 2-week-old cotyledons were imaged using a Zeiss Pascal LSM5 confocal laser scanning microscope ( The company setting for rhodamine imaging was used for visualizing chloroplasts by their autofluorescence. The lengths of the long axis of chloroplast ellipsoid profiles were measured for 33–35 chloroplasts in each sample using the imagej software package (

Cotyledons were excised from 3-day-old, 1-week-old and 2-week-old seedlings, and rapidly frozen in an HPM100 high-pressure freezer (Leica Microsystems, The frozen samples were freeze-substituted at −80°C in anhydrous acetone containing 2% OsO4 for 2 days, and extra OsO4 was removed by washing with pre-cooled anhydrous acetone at −80°C. The samples were slowly warmed to room temperature and embedded in EMBED 812 resin (Product number 18109, TedPella, Sample processing for immunogold labeling, ultramicrotomy and imaging were performed as described by Kang (2010).

Electron tomography analysis

Tomograms of chloroplasts were generated as described by Kang et al. (2011). To measure ribosome densities, stromal and cytosolic volumes of 8–15 μm3 were analyzed in tomograms of 3-day-old, 1-week-old and 2-week-old cotyledon cells from wild-type and rh3–4 seedlings. The analyzed volumes do not include membrane elements such as endoplasmic reticulum and thylakoids that may influence density measurement. After modeling all the ribosomes in the volume, the ribosome density in each volume was calculated using the IMOD program. ( Ribosome densities were calculated in three volumes of 0.1∼0.5 cubic micrometer in the wild-type and the rh3-4 mutant samples. Ribosome diameters were determined by averaging the diameters of 60–70 ribosomes for each tomogram.

Ion Torrent library preparation, sequencing, and sequence analysis

Total RNA samples were isolated from cotyledons of 1-week-old wild-type and rh3–4 seedlings using a Norgen Biotek total RNA purification kit (catalog number: 17200, After degrading ribosomal RNA using a Ribo-Zero rRNA removal kit for plant leaves (catalog number: MRZPL116, Epicentre,, cDNA libraries were prepared using a Ion Total RNA-Seq kit (Life Technologies). The resulting cDNA libraries were purified using AMPure beads (Beckman Coulter,, and their concentrations and sizes were determined using an Agilent BioAnalyzer DNA high-sensitivity LabChip (Agilent Technologies Emulsion PCR and enrichment of cDNA conjugated particles were performed using an Ion Xpress template kit (Life Technologies) according to the manufacturer's instructions. The final particles were loaded on an Ion 314 chip and sequenced using a personal genome machine (Life Technologies).

Raw data from the personal genome machine were pre-processed using the torrent suite version 2.0.1 to remove poor quality reads and trim adapter sequences. The processed fastaq files were imported into the CLC Genomics Workbench version 4.9 (CLC bio,, and reads were mapped to TAIR10 Arabidopsis thaliana chromosome sequences downloaded from using the RNA-seq toolkit of the CLC Genomics Workbench software package with default parameters. Differential expression between wild-type and the rh3–4 mutant was analyzed using the expression analysis tool of the CLC Genomics Workbench software package, with RPKM as the expression value. Genes with an RPKM larger than 20 in either the wild-type or the rh3–4 cDNA were chosen, and their RNA-seq data were analyzed using mapman version 3.5.1, which displays expression data from a large number of genes on diagrams of metabolic pathways and biological pathways (Thimm et al., 2004). For fold change ratio calculation, genes with no read in either the wild-type samples or in the mutant samples were set as having one read to avoid the problem of division by zero. The Arabidopsis Genome Initiative TAIR9 annotation (January 2010) was used to map genes in the pathways.


We are grateful to Tony Romeo and Christopher Vakulskas (Microbiology and Cell Science Department, University of Florida, Gainesville, FL, USA) for performing Escherichia coli complementation experiments with RH3 cDNA even though the results are not included in the paper. We thank Kenneth C. Cline (Horticultural Sciences Department, University of Florida, Gainesville, FL, USA) for sharing antibodies used for immunoblot analysis. We also thank Andreas Hoenger and members of the Boulder Laboratory for 3D Electron Microscopy of Cells for allowing us to use their intermediate voltage electron microscopes. This research was supported by US National Science Foundation grants MCB-0958107 and IOS-1025976 to B.–H.K., and by the world class university project (R31-2008- 000-10105) of the National Research Foundation, Ministry of Education, Science and Technology (Korea) to I.H.