An Arabidopsis Pentatricopeptide Repeat Protein, SUPPRESSOR OF VARIEGATION7, Is Required for FtsH-Mediated Chloroplast Biogenesis 1[W][OA]

The Arabidopsis ( Arabidopsis thaliana ) yellow variegated2 ( var2 ) mutant has green- and white-sectored leaves due to loss of VAR2, a subunit of the chloroplast FtsH protease/chaperone complex. Suppressor screens are a valuable tool to gain insight into VAR2 function and the mechanism of var2 variegation. Here, we report the molecular characterization of 004-003 , a line in which var2 variegation is suppressed. We found that the suppression phenotype in this line is caused by lack of a chloroplast pentatricopeptide repeat (PPR) protein that we named SUPPRESSOR OF VARIEGATION7 (SVR7). PPR proteins contain tandemly repeated PPR motifs that bind speciﬁc RNAs, and they are thought to be central regulators of chloroplast and mitochondrial nucleic acid metabolism in plants. The svr7 mutant has defects in chloroplast ribosomal RNA (rRNA) processing that are different from those in other svr mutants, and these defects are correlated with reductions in the accumulation of some chloroplast proteins, directly or indirectly. We also found that whereas var2 displays a leaf variegation phenotype at 22 (cid:1) C, it has a pronounced chlorosis phenotype at 8 (cid:1) C that is correlated with defects in chloroplast rRNA processing and a drastic reduction in chloroplast protein accumulation. Surprisingly, the cold-induced phenotype of var2 cannot be suppressed by svr7. Our results strengthen the previously established linkage between var2 variegation and chloroplast rRNA processing/chloroplast translation, and they also point toward the possibility that VAR2 mediates different activities in chloroplast biogenesis at normal and chilling temperatures. 2008). Genes are driven by the cauliﬂower mosaic virus 35S promoter in this construct. Primers 16390F and 16390GFPR were used to amplify the SVR7 sequence using pfu Turbo DNA polymerase (Stratagene). The PCR product was digested with Bam HI and Nco I and cloned in-frame into pTF486. The resulting construct (designated P35S : SVR7 : GFP ) was sequenced to verify that the open reading frame of eGFP was correctly fused to the 3 # end of SVR7 . A control construct lacked the SVR7 sequence (designated P35S : GFP ). Methods for transient expression of wild-type Arabidopsis leaf protoplasts have been described (Yoo et al., 2007; Yu et al., 2008). Confocal microscopy (Leica TCS NT) with a ﬂuorescein isothiocyanate-tetramethyl rhodamine isothiocy-anate ﬁlter combination was used to capture ﬂuorescence signals of GFP and chlorophyll autoﬂuorescence.

Variegation mutants are powerful tools to gain insight into mechanisms of chloroplast biogenesis (Sakamoto et al., 2003; for review, see Rodermel, 2002;Yu et al., 2007). The Arabidopsis (Arabidopsis thaliana) yellow var-iegated2 (var2) mutant has green cotyledons but its true leaves are green and white sectored; sectoring is produced in homozygous recessive plants. The green sectors contain normal-appearing chloroplasts, while the white sectors contain abnormal plastids that lack pigments and developed lamellar structures (Chen et al., 1999). Heteroplastidic cells that have some normalappearing chloroplasts are also found in var2 white tissues, indicating that individual plastids are affected differently by the nuclear mutation (i.e. var2 acts in a plastid-autonomous manner; Chen et al., 1999). Because the tissues of var2 have a uniform genetic constitution, a major question is why some chloroplasts are able to bypass the requirement for VAR2 during chloroplast biogenesis. Why isn't the mutant all white?
We have proposed a threshold model to explain the mechanism of var2 variegation (Yu et al., 2004(Yu et al., , 2005(Yu et al., , 2007. According to this model, chloroplast biogenesis requires a threshold of FtsH activity, and the threshold varies from plastid to plastid because of intrinsic differences in the metabolic profiles of the 100+ developing plastids in a typical Arabidopsis mesophyll cell (e.g. due to gradients, sometimes steep, of CO 2 , light, chlorophyll, and photosynthesis across the thickness of a leaf; Smith et al., 1997). The plastids in a cell might also have variable amounts of the various FtsH proteins. It is proposed that subthreshold FtsH activities result in an arrest of chloroplast development; consequently, white plastids are formed that generate white cells and white sectors in the developing leaf. Threshold activities, on the other hand, result in the formation of chloroplasts and green sectors. Our working hypothesis is that thresholds in var2 are achieved by FtsH activities per se and/or by the activities of factor(s)/process(es) that are able to compensate for a lack of VAR2.
To gain insight into the compensating mechanisms that allow green sector formation in var2, we and others have carried out second-site suppressor screens Miura et al., 2007;Yu et al., 2008). These screens have led to the identification of a number of recessive suppressor genes. These include ClpC2, the gene for a chloroplast ClpC/Hsp100 chaperone ; SUPPRESSOR OF VARIEGATION1 (SVR1), which codes for a chloroplast pseudouridine synthase (Yu et al., 2008); SVR2, also designated ClpR1 (Koussevitzky et al., 2007b), the gene for a subunit of the chloroplast ClpP/R protease complex (Yu et al., 2008); FUG1, which encodes the chloroplast translation factor IF-2 (Miura et al., 2007); and SCO1, which codes for the chloroplast translation factor EF-G (Miura et al., 2007). Interestingly, most of these suppressors are involved in the coupled processes of chloroplast ribosomal RNA (rRNA) processing and translation. This has given rise to the hypothesis that reductions in chloroplast translation are able to compensate for a lack of VAR2 by lowering the threshold demand for FtsH in the population of developing chloroplasts. Another mechanism for suppression has recently come to light with the demonstration that ectopic expression of the heterotrimeric G protein a-subunit (GPA1) suppresses var2 variegation, perhaps by regulating the expression of nuclear genes for FtsH proteins (Zhang et al., 2009).
In this report, we characterize another var2 suppressor (designated 004-003) and demonstrate that suppression of variegation in this line is due to down-regulated expression of SVR7, encoding a chloroplast-localized pentatricopeptide repeat (PPR) protein. We show that svr7 mediates defects in chloroplast rRNA processing that are different from those reported for svr1 and svr2 (Yu et al., 2008) and that processing proceeds by a hierarchy of events, as highlighted by double mutant analyses. The defects in rRNA processing are accom-panied by decreases in the accumulation of some chloroplast proteins, consistent with the idea that decreased protein synthesis is able to suppress FtsHmediated defects in chloroplast biogenesis. Interestingly, examination of 004-003 revealed that there is a unique genetic interaction between svr7 and var2 at chilling temperatures (8°C): at this temperature, var2 displays a pronounced chlorosis in newly emerging tissues that cannot be bypassed by the loss of SVR7 activity. Taken together, the data presented here reinforce our earlier findings of a strong genetic interaction between chloroplast rRNA processing/translation and VAR2 function and raise the intriguing possibility that VAR2 mediates chloroplast biogenesis via independent processes that act at normal and chilling temperatures.

Phenotype of the var2 Suppressor Line, 004-003
We have previously reported the isolation of var2 suppressors using ethyl methanesulfonate mutagenesis  and T-DNA activation tagging (Yu et al., 2008). One of the suppressors in our collection of activation-tagged mutants (designated 004-003) is the topic of this paper. Backcrosses between 004-003 and wild-type Columbia showed that the suppressor gene in 004-003 behaves in a recessive manner. We named the suppressor gene locus SVR7, and the mutant allele in 004-003 was designated svr7-1. svr7-1 single mutants were isolated in the F2 progeny of the backcross using derived cleaved-amplified polymorphic sequence primers (Neff et al., 1998) to genotype the VAR2 locus . Figure  1A shows that the 004-003 double mutant (var2-5 svr7-1) resembles the svr7-1 single mutant and that both are smaller and more pale green than the wild type and var2-5. The reductions in chlorophyll content in 004-003 and svr7 are significant (approximately 25%; Fig. 1B) but do not reflect an underlying perturbation in photosystem structure or stoichiometry, since chlorophyll a/b ratios are not appreciably altered (Fig. 1C).
The var2-5 allele is caused by a missense mutation and is leaky, inasmuch as the mutant plants accumulate VAR2 mRNAs and low amounts of the mutant protein (Chen et al., 2000). Its variegation phenotype is less severe than that of var2-4, which has a defect in mRNA splicing and approximates the null phenotype (Yu et al., 2004). To test whether the suppression of variegation in 004-003 is allele specific, we generated svr7-1 var2-4 double mutants. Figure 1D shows that these plants are not variegated and that they resemble 004-003. Taken together, the data in Figure 1 indicate that svr7-1 is epistatic to var2 and that suppression of var2 variegation by svr7-1 is not allele specific.
to a T-DNA insert but, rather, is caused by a recessive mutation at another site (data not shown). Hence, we turned to methods of positional cloning to identify the SVR7 locus (see "Materials and Methods"). In brief, this method involved first mapping the gene to an approximately 76-kb interval between FCA5#3 and FCA6#2 and then determining the genomic DNA sequences of candidate nuclear genes for chloroplast proteins in this interval in the mutant svr7-1 plants (Fig. 2). One of these genes (At4g16390) was found to contain a series of short (1-5 bp) deletions and a few point mutations in a discrete region of the gene. The At4g16390 gene model contains a single exon and codes for a protein of 702 amino acids (approximately 78.2 kD), and the mutations would be predicted to generate a premature stop codon and a truncated translation product (317 amino acids) lacking important functional domains of the protein (described below). Hence, we tentatively designated At4g16390 as SVR7.
Complementation of 004-003 by SVR7 If suppression of variegation in 004-003 is due to a loss of functional SVR7, it should be possible to restore the variegation phenotype by overexpression of SVR7 in 004-003. To test this hypothesis, a full-length SVR7 genomic DNA sequence was cloned into a binary T-DNA vector, and the construct was transformed into 004-003. Figure 3A shows that the transgenic plants (004-003 P35S:SVR7) have a variegated phenotype, similar to var2-5. Figure 3B shows that the overexpression plants are indeed transformed in that they have dramatically enhanced SVR7 mRNA levels when compared with the 004-003 plants. Taken together with the prediction that a translation product lacking the C-terminal half of SVR7 would be generated in svr7-1 (Fig. 2), the extremely low levels of SVR7 transcripts in 004-003 suggest that svr7-1 is a molecular null, or at least that it approximates the null phenotype. We conclude from the data in Figure 3 that disruption of At4g16390 is responsible for the suppression of var2 variegation in 004-003.
The C-terminal motifs are thought to define sites for binding of effectors (such as proteins or nucleic acids; Lurin et al., 2004;Delannoy et al., 2007). Figure 4A shows that SVR7 contains 10 PPR motifs, nine of which occur in a tandem array, spanning amino acid residues 176 to 491. SVR7 also contains an SMR (for small MutS related) domain near its C terminus. The SMR domain was first identified in the C-terminal region of the Synechocystis sp. MutS2 protein (Moreira and Philippe, 1999), and it is also present in the E. coli SMR protein (Fig. 4A). MutS2 might be involved in the inhibition of homologous recombination (Pinto et al., 2005), but the function of the E. coli SMR protein is unclear.
In Arabidopsis, at least eight proteins have an architecture similar to SVR7 (i.e. tandem arrays of PPR motifs + an SMR domain; Fig. 4A). These include GUN1, which is a key regulator of chloroplast-tonucleus retrograde signaling (Koussevitzky et al., 2007a), and pTAC2, which copurifies with plastid transcriptionally active chromosomes (pTACs; Pfalz et al., 2006). Figure 4B shows that among the eight proteins, SVR7 is most closely related to At5g46580 (38% amino acid identity), while GUN1 is most similar to pTAC2.

Chloroplast Localization of SVR7
Organelle-targeting algorithms predict that SVR7 has an N-terminal chloroplast transit peptide of 53 amino acids (Emanuelsson et al., 2007). To confirm that SVR7 resides in the chloroplast, an SVR7/GFP fusion (P35S:SVR7:GFP) was generated and transiently ex-pressed in wild-type Arabidopsis leaf protoplasts; confocal microscopy was used to monitor the fluorescence signals. Figure 5, A to C, reveals that the green fluorescence signals from GFP are present in the cytoplasm of cells transformed with a control construct lacking SVR7 (P35S:GFP); these signals do not merge with the red autofluorescence signals from chloroplasts. By contrast, when P35S:SVR7:GFP is expressed ( Fig. 5, D-F), the green fluorescence signals colocalize with chloroplasts, some as discrete foci within the organelle (as indicated by merging of the red and green fluorescence signals). These data indicate that SVR7 is a plastid protein.

Disruption of SVR7 Results in Chloroplast rRNA Processing Defects
Chloroplast rRNA genes (23S, 16S, 4.5S, and 5S) constitute an operon (rrn operon) in the chloroplast genome (Fig. 6A). The maturation of chloroplast rRNAs involves a series of endonuclease-cleavage and exonuclease-trimming events (Bollenbach et al., 2005). Casual inspection of an ethidium bromidestained denaturing gel of RNAs from 004-003 ( Fig.  3B) reveals that the stoichiometry of rRNA bands is disrupted in this line. To examine this in greater detail, RNA gel-blot analyses were performed on RNAs from 004-003 and svr7-1 using rRNA gene-specific probes (Fig. 6B). Blots of total leaf Arabidopsis RNAs probed with 23S rDNA contain seven readily detectable bands (Fig. 6B, lane WT): the 3.2-kb band is the 23S-4.5S dicistronic precursor; the 2.9-, 2.4-, and 1.7-kb bands Map-based cloning of SVR7. SVR7 was mapped by positional cloning as described in "Materials and Methods." The gene was fine-mapped between FCA5#3 and FCA6#2 using markers described in Supplemental Table S1 (indicated by vertical lines). The number of recombinants is shown under each marker; 665 F2 plants (1,330 chromosomes) were examined. Genomic sequencing of candidate genes between FCA5#3 and FCA6#2 led to the identification of SVR7 (At4g16390). The position of SVR7 is indicated by an asterisk. There is a single exon in the SVR7/At4g16390 gene model (indicated by the box). The gray parts of the box in the gene model represent the 5# and 3# untranslated regions. The deletions (1-5 bp) in svr7-1 are shown as dashed lines, and the point mutations are indicated in boldface. are processing intermediates; and the 1.2-, 1.0-, and 0.5-kb bands are the mature forms of 23S rRNA (Bollenbach et al., 2005;Yu et al., 2008). Figure 6B shows that var2-5 and the wild type have similar banding patterns, but that in 004-003 and svr7-1, the levels of the 2.9-and 2.4-kb rRNAs are significantly increased while those of the 1.7-, 1.2-, and 1.0-kb rRNAs are significantly decreased. The metabolism of chloroplast 16S rRNA and 4.5S rRNA is also perturbed in 004-003 and svr7-1, inasmuch as the 16S rRNA precursor and the 23-4.5S dicistronic precursor accumulate in these lines (Fig. 6B). Figure 6C shows that the 23S rRNA profile is normalized in the 004-003 complementation line (004-003 P35S:SVR7), indicating that the defect in chloroplast rRNA processing in 004-003 is caused, either directly or indirectly, by a lack of SVR7.

Suppressors of var2
One question that arises is whether the defects in chloroplast rRNA processing are unique to svr7-1. To address this question, we examined 23S rRNA accumulation patterns in three other suppressor mutants: svr1-1 (Yu et al., 2008), svr3-1, and svr8-1. The suppressor genes in the latter two lines have not yet been cloned, but fine-mapping studies indicate that they do not map to other svr loci or to genes for PPR proteins. Figure 6D shows that 23S rRNA processing is defective in all four svr lines and that each line has a unique rRNA profile. Although we cannot rule out that these defects are secondary effects of other, more general perturbations in chloroplast function, the data suggest that each mutant defines a distinct step in chloroplast rRNA processing, either directly or indirectly.
We next carried out western immunoblotting analysis to investigate the impact of svr7 on chloroplast protein accumulation ( Fig. 6E; Supplemental Fig.  S2A). On a fresh weight basis, the levels of three representative chloroplast DNA-encoded proteins we examined (the D1 protein of PSII, the large subunit of Rubisco, and the a-subunit of the ATP synthase) are reduced to approximately 50% of those of the wild type in 004-003 and svr7-1 (Fig. 6E). By contrast, the levels of most nucleus-encoded chloroplast proteins (Lhcb2, PsbP protein of PSII, and the Rieske Fe-S protein of the cytochrome b 6 f complex) are not reduced in 004-003 and svr7-1, with the exception of the PSI PsaF protein, which appears to be reduced to approximately 50% of the wild-type level (Fig. 6E). As anticipated, sharp reductions in VAR2 in var2-5 and 004-003 are accompanied by decreased amounts of the AtFtsH1 protein, likely because of turnover of unassembled subunits of the AtFtsH complex (Sakamoto et al., 2003;Yu et al., 2004). Considered together, the data in Figure 6E indicate that the accumulation of some chloroplast proteins, especially those of chloroplast origin, is impaired in the svr7 background. However, at present, it is not clear whether SVR7 mutation directly, or indirectly, leads to this impairment.

Chilling Sensitivity of var2
Chloroplast rRNA processing and translation are coupled events, and analyses of various var2 suppressor lines have indicated that there is a genetic interaction between VAR2 function and these processes (Miura et al., 2007;Yu et al., 2008). A critical aspect of chloroplast translation is its importance under chilling temperatures (Tokuhisa et al., 1998;Rogalski et al., 2008). Therefore, we felt that it might be instructive to examine the responses of var2, svr7-1, and the 004-003 double mutant to chilling stress. For these experiments, we grew the plants at 22°C for 3 weeks and then transferred them to 8°C for another 4 weeks. At both temperatures, the plants were maintained under continuous light (100 mmol m 22 s 21 ). Figure 7A shows that leaves that develop under chilling stress are green in wild-type plants but uniformly chlorotic in var2-5. The same is true of the severe allele, var2-4. These data indicate that the phenotype of var2 is temperature dependent: variegation at normal growth temperatures and a chilling-induced chlorosis.
We also examined the chilling response of var1, which lacks the AtFtsH5 subunit of the AtFtsH complex (Sakamoto et al., 2002). Figure 7B shows that the basal parts of var1 leaves that develop at 8°C are yellow but that, as the leaves expand, their tips eventually turn green. This suggests that chloroplast bio- genesis in var1 is delayed at 8°C. The less severe phenotype of var1 versus var2 at 8°C is in agreement with the lower abundance of VAR1 (AtFtsH5) than VAR2 (AtFtsH2) in AtFtsH complexes and with the relatively weak phenotype of var1 versus var2 under normal growth conditions (Sakamoto et al., 2002).
We next tested the response of svr7-1 and 004-003 (var2-5 svr7-1) to chilling stress. Figure 7C reveals that the leaves of 004-003 are chlorotic at 8°C, similar to var2-5, while the leaves of svr7-1 are all green at this temperature. In fact, chlorophyll concentrations at 8°C are reduced approximately 25% in svr7-1 versus the wild type at 8°C (similar to the reductions observed at 22°C; Fig. 1B) and sharply reduced to less than 5% of normal in var2-5 and 004-003 (Fig. 7D). Accompanying these changes are drastic alterations in chloroplast rRNA profiles: whereas temperature does not alter these profiles in svr7-1 and the wild type (the profiles in Figs. 6B and 7E are similar), the overall levels of 23S and 16S rRNAs are greatly reduced in both Figure 4. Protein sequence analysis of SVR7. A, Eight Arabidopsis PPR proteins have an architecture similar to SVR7. Full-length Synechocystis sp. MutS2 (P73625) and E. coli SMR (P0A8B2) protein sequences were obtained from the UniProt protein database (www.uniprot.org). Protein sequences of Arabidopsis genes were obtained from The Arabidopsis Information Resource (www. arabidopsis.org). InterProScan (www.ebi.ac.uk/ Tools/InterProScan/; Quevillon et al., 2005) was used to identify conserved domains. B, Phylogenetic relatedness of Arabidopsis PPR proteins containing an SMR domain. Full-length protein sequences were obtained as in A. The phylogenetic tree was constructed using MEGA4 software (Tamura et al., 2007). var2-5 and 004-003, with precursor forms accumulating at the expense of mature forms (Fig. 7E).
The dramatic changes in chlorophyll accumulation and chloroplast rRNA processing prompted us to assess the patterns of chloroplast protein accumulation in cold-developed tissues ( Fig. 7F; Supplemental Fig.  S2B). Figure 7F shows that some proteins, when compared on a fresh weight basis, are modestly reduced in amount in svr7-1, including the plastid DNA-encoded D1, LS, and ATPa proteins and the nuclear DNAencoded PsaF protein. These findings are similar to those found at 22°C (Fig. 6E), indicating that chloroplast protein synthesis in not affected by temperature in svr7-1. In contrast, the levels of most chloroplast proteins are drastically reduced in the chlorotic tissues of var2-5 and 004-003 (many are present at lower than 25% of wild-type levels). The only exception appears to be AtFtsH1, which is accumulated at around the wild-type level in 004-003 and is more abundant in cold-developed svr7-1 single mutant tissues. Taken together, Figure 7F suggests a much greater impact of chilling temperature on chloroplast protein accumulation in var2-5 and 004-003, although we are not clear if this is directly or indirectly the consequence of VAR2 mutation.
In summary, the data in Figures 3 and 6 show that at 22°C, 004-003 and svr7-1 have similar visual and molecular phenotypes, as monitored by chlorophyll concentrations, chloroplast rRNA profiles, and chloroplast protein accumulation profiles. This suggests that svr7-1 is epistatic to var2 at this temperature (i.e. a loss of SVR7 is able to bypass the requirement for VAR2 during chloroplast biogenesis). The data in Figure 7, on the other hand, show that this is conditional, inasmuch as svr7 does not suppress the chillinginduced chlorosis phenotype of var2, nor does it suppress the var2-mediated defects in rRNA processing and protein accumulation at the low temperature. This suggests that var2-5 is epistatic to svr7-1 at 8°C (i.e. a loss of SVR7 does not compensate for a loss of VAR2 in chloroplast development).

Genetic Interactions between svr7 and svr2
In addition to svr1 (Fig. 6D), we have previously shown that svr2 has a chloroplast rRNA processing defect (Yu et al., 2008). SVR2 is a subunit of the chloroplast ClpP/R protease, a stromal enzyme that is thought to be responsible for most protein turnover in the plastid (Adam and Clarke, 2002; Adam et al., Figure 6. Chloroplast rRNA processing defects and protein accumulation in 004-003 and svr7-1. A, Structure of the chloroplast rrn operon in Arabidopsis. Thick solid lines under each rRNA gene represent the probes used for the RNA gel-blot analyses in B to D. Equal amounts of total leaf RNA (2 mg) from 3-week-old seedlings were loaded on the gel. See Supplemental Figure S1 for photographs of the ethidium bromidestained gels in B to D. B, Accumulation of chloroplast 23S, 16S, and 4.5S rRNAs in the wild type (WT), var2-5, 004-003, and svr7-1. C, Accumulation of 23S rRNAs in the wild type, var2-5, 004-003, and 004-003 P35S:SVR7 (the 004-003 complementation line; see Fig. 3). D, Accumulation of 23S rRNAs in the wild type and four var2 suppressor lines (svr1-1, svr7-1, svr3-1, and svr8-1). SVR1 codes for a chloroplast pseudouridine synthase (Yu et al., 2008), while molecular characterization of svr3-1 and svr8-1 has not yet been reported (F. Yu, X. Liu, and S. Rodermel, unpublished data). E, Accumulation of chloroplast proteins in wild-type, var2-5, 004-003, and svr7-1 plants. Total leaf protein from 3-week-old seedlings was loaded onto the gels based on an equal fresh weight basis. See text for description of proteins; see Supplemental Figure S2A for the quantification of immunoblots.  Figure S1 for a photograph of the ethidium bromide-stained gel. F, Accumulation of chloroplast proteins in leaves of wild-type, var2-5, 004-003, and svr7-1 plants developed at 8°C. Total leaf protein was extracted and loaded on a fresh weight basis for immunoblotting. See Supplemental Figure S2B for the quantification of immunoblots. 2006). As a first step to understand how SVR proteins are involved in chloroplast rRNA processing, we generated svr2 svr7 double mutants. These were identified using a PCR-based assay (data not shown). Figure 8A shows that the svr2 svr7 double mutant and svr2-1 single mutants have similar growth characteristics and morphologies and that these are markedly different in svr7-1. Figure 8B reveals that the 23S rRNA profiles of svr2-1 and svr7-1, although similar, have different precursor stoichiometries and that the profile of the double mutant resembles that of svr2-1, not svr7-1. Taken together, these data indicate that svr2 is epistatic to svr7. One interpretation consistent with these data is that SVR2 and SVR7 function in the same process or pathway of chloroplast rRNA processing and that SVR2 is involved in a more upstream event than SVR7.

Function of SVR7
Our analyses have shown that SVR7 is not essential for chloroplast development but that loss of SVR7 is able to suppress var2-mediated defects in chloroplast biogenesis in a temperature-dependent manner. To gain insight into these observations, perhaps the first question concerns the function of SVR7. PPR proteins are central regulators of RNA metabolism in organelles, and they bind RNA via the PPR motif (Delannoy et al., 2007;Small, 2008, Stern et al., 2010). Because PPR motifs do not have known catalytic sites, PPR proteins are thought to act as adaptors to facilitate interaction between an effector molecule and a specific RNA substrate (Delannoy et al., 2007). Consistent with this idea, a number of PPR proteins contain C-terminal sequences that confer specific functions. An increasing number of substrate RNAs and effectors of PPR proteins have been identified either biochemically or genetically Small, 2008, Stern et al., 2010).
Our studies have shown that SVR7 has canonical PPR motifs and a conserved C-terminal SMR domain. Support for the idea that SVR7 binds RNA comes from studies of the radish (Raphanus sativus) P67 protein (Echeverria and Lahmy, 1995;Lahmy et al., 2000). The sequence of P67 is very similar to SVR7, and it was identified in radish extracts as a nuclear protein with in vitro binding activity to a fragment of nuclear pre-rRNA, designated rBP (Echeverria and Lahmy, 1995). However, subsequent studies showed that P67 resides in the plastid rather than the nucleus-cytoplasm, suggesting that rBP is not the physiological binding partner of P67 (Lahmy et al., 2000). Because the nucleotide sequence of rBP does not share homology with any chloroplast RNA species, Lahmy et al. (2000) proposed that the interaction between P67 and its RNA binding partner might be directed by the secondary or tertiary structure of the RNA fragment. This RNA has yet to be defined biochemically.
SVR7 is one of at least eight Arabidopsis PPR proteins that have an SMR domain in their C terminus (Fig. 4A). Two of these, pTAC2 and GUN1, have been partially characterized. pTAC2 is a component of pTACs that have been isolated from Arabidopsis and mustard (Brassica juncea; Pfalz et al., 2006), while GUN1 has received much attention as a putative central integrator of retrograde signaling pathways (Susek et al., 1993;Koussevitzky et al., 2007a). Because GUN1 colocalizes with pTAC2 in discrete foci within chloroplasts (reminiscent of the punctuate GFP signals of SVR7; Fig.  5), it has been suggested that GUN1, like pTAC2, is associated with sites of transcriptionally active plastid chromosomes (Koussevitzky et al., 2007a). In support of this hypothesis, the SMR domain of GUN1 is able to bind DNA (Koussevitzky et al., 2007a). This is similar to the SMR domains of the human B3BP/Nedd4-binding protein 2 and the eubacterial Thermus thermophilus MutS2, both of which have nicking and endonuclease activities in addition to DNA binding activity (Fukui et al., 2007;Diercks et al., 2008).
Given the demonstrated biochemical functions of PPR and SMR domains, perhaps the simplest interpretation of our data is that SVR7 is directly involved Figure 8. Genetic interaction between svr7 and svr2. A, Phenotypes of representative 3-week-old wild-type (WT), svr7-1, svr2-1, and svr7-1 svr2-1 plants growing under normal conditions. B, RNA gel-blot analysis of chloroplast 23S rRNA accumulation in the plants in A. Equal amounts of RNA (2 mg) extracted from 3-week old seedlings were loaded on the gel. See Supplemental Figure S1 for a photograph of the ethidium bromide-stained gel.
in chloroplast rRNA processing. For example, the PPR domain could bind an rRNA species and the SMR domain could bind a nuclease that acts on the rRNA. However, until the biochemistry is completed, we cannot rule out the possibility that the chloroplast rRNA processing defects in svr7 are secondary effects of a more general impairment in chloroplast function. Against this notion are our observations that many of the var2 suppressors isolated to date are impaired in chloroplast rRNA processing and that the processing defects in these lines are unique. These findings suggest that each of the different SVR proteins mediates a distinct step of rRNA metabolism, either directly or indirectly. In support of this idea is our finding from double mutant analyses that svr2 is epistatic to svr7 (Fig. 8), which suggests that the defects in rRNA processing mediated by SVR7 are downstream of those mediated by SVR2. This further implies that chloroplast rRNA processing proceeds by a hierarchy of events.
In summary, our data indicate that characterization of var2 suppressors is a powerful and facile tool to identify factors that are involved in the poorly understood process of chloroplast rRNA processing. This is underscored by the fact that very few proteins have been identified with a defined role in this process. These include one endoribonuclease, CSP41, which is involved in 23S rRNA processing (Yang et al., 1996;Yang and Stern, 1997;Beligni and Mayfield, 2008) and two 3#-5# exoribonucleases, PNPase and RNR1 (Walter et al., 2002;Kishine et al., 2004;Bollenbach et al., 2005;Sauret-Gü eto et al., 2006).

Model of Suppression of Variegation by svr7
We have proposed that a threshold of VAR2 activity is required early in chloroplast biogenesis in the developing leaf and that plastids respond differently to a lack of VAR2, depending on such factors as substrate availability and the presence of compensating factors (Yu et al., 2004(Yu et al., , 2007. As was the case for svr1 and svr2 (Yu et al., 2008), svr7 is able to suppress variegation in both var2-5 and the null allele, var2-4. Therefore, suppression is not likely the consequence of a direct interaction between SVR7 and VAR2. Rather, we propose that loss of SVR7 changes the functional state of the chloroplast in such a manner that the threshold of VAR2 activity needed to ensure proper chloroplast biogenesis is lowered.
Evidence for how this might occur is provided by several var2 suppressors that have pointed to a fundamental linkage between the suppression of variegation and an impairment in chloroplast rRNA processing and/or chloroplast translation (Miura et al., 2007;Yu et al., 2008). rRNA processing, ribosome assembly, and translation are coupled events in chloroplasts, and defects in one are accompanied by defects in the others (Shen and Bremer, 1977;Keus et al., 1984;Barkan, 1993;Leal-Klevezas et al., 2000;Walter et al., 2002;Bellaoui et al., 2003;Williams and Barkan, 2003;Bollenbach et al., 2005;Sauret-Gü eto et al., 2006;Schmitz-Linneweber et al., 2006;Watkins et al., 2007;Yu et al., 2008). Our findings with svr7, as well as with a number of other svr lines that include svr3 and svr8 (Fig. 6), thus strengthen the linkage between chloroplast rRNA processing/ translation and suppression of variegation. This linkage is highlighted by the observation that variegation is suppressed in var2 seedlings that are grown in the presence of inhibitors of chloroplast translation, such as chloramphenicol and spectinomycin (Yu et al., 2008).
What is the relationship between VAR2 and chloroplast rRNA processing/translation? One possibility is that the interaction is direct, as in yeast mitochondria, where the inner membrane m-AAA complex (composed of FtsH homologs) directly regulates mitochondrial translation (Nolden et al., 2005). VAR2 might also exert its effect on chloroplast translation in a more indirect manner (e.g. by influencing the activities of proteins that are involved in rRNA processing, ribosome assembly, or translation). Because FtsH proteins act as proteases/chaperones and are important in protein quality control (for review, see Adam and Clarke, 2002;Sakamoto, 2006;Koppen and Langer, 2007), another possibility is that translation is influenced by VAR2 control of protein translocation across the membrane, as found with the i-AAA (Yme1) FtsH homolog in yeast mitochondria (Rainey et al., 2006). This could result, for example, in reduced accumulation of imported proteins of nuclear origin necessary for chloroplast gene expression, resulting in reduced translation. A lack of VAR2 might also result in an accumulation of misfolded polypeptides and protein aggregates, with deleterious effects on organelle development. Such effects are thought to explain mitochondrial abnormalities in mutants of human and yeast that lack mitochondrial FtsH proteins (Nolden et al., 2005). Suppressors such as svr1, svr2, and svr7 might thus reduce this load by decreasing the amount of protein produced, thus lowering the threshold of VAR2 activity required for folding/degradation. A related hypothesis is that decreased chloroplast protein synthesis in the suppressors causes a reduction in photosynthetic protein accumulation, which, in turn, might decrease ROS production at PSII and hence the amount of VAR2 needed for D1 turnover and photoprotection.
Other mechanisms of var2 suppression based on an impairment of chloroplast translation have been discussed in detail (Yu et al., 2008). These mechanisms will not be recapitulated here but, in brief, proceed from the premise that translation affects the duration of chloroplast biogenesis, the propagation of a retrograde (plastid-to-nucleus) signal, or the synthesis of a chloroplast DNA-encoded substrate of VAR2 that regulates chloroplast development. Given the large number of possibilities, and the likelihood that there is not a single mechanism of suppression applicable to all svr mutants, a future challenge will be to describe suppression mechanisms at the molecular level.

VAR2 Mediates Chloroplast Biogenesis in a Temperature-Dependent Manner
One of the central findings of this paper is that var2 displays a leaf variegation phenotype at 22°C but a chlorotic phenotype at 8°C (Fig. 7). var1 has similar temperature-dependent phenotypes (Fig. 7). Our suppressor studies further revealed that the svr7 mutation is able to suppress the leaf variegation phenotype but not the chilling-induced chlorosis phenotype. These observations are consistent with molecular data (chloroplast rRNA profiling and chloroplast protein accumulation) showing that var2-5 is epistatic to svr7-1 at 8°C, while the converse is true at 22°C (i.e. svr7-1 is epistatic to var2-1; Figs. 6 and 7).
One hypothesis consistent with our data is that VAR2 mediates a single process during chloroplast biogenesis and that this process is only modestly impaired at 22°C, resulting in a variegation phenotype, but severely impaired at 8°C, resulting in chlorosis. According to this scenario, a lack of SVR7 activity would be sufficient to compensate for the mild VAR2 defect, perhaps by decreasing chloroplast protein synthesis, but not sufficient to compensate for the severe VAR2 defect. A second hypothesis is that VAR2 mediates two different processes, one that occurs at normal growth temperatures and one that occurs at chilling temperatures. In this model, a lack of SVR7 activity would be able to compensate for the defective var2mediated process that is required at 22°C but not for the defective process that is required at 8°C.
Chilling-induced chlorosis is a frequent symptom of chilling injury and can be caused by different mechanisms (Tokuhisa et al., 1998). While it is premature to speculate on the role of VAR2 in chloroplast development at low temperature, it is interesting that chillinginduced chlorosis has frequently been associated with a loss of chloroplast translation. For example, early studies showed that several maize (Zea mays) mutants that display chilling-induced chlorosis, including M-11 (Millerd et al., 1969), hcf7 (Barkan, 1993), and v16 (Hopkins and Elfman, 1984), have defects in chloroplast ribosome accumulation and/or translation that are more pronounced at low temperature. A chillingsensitive Arabidopsis mutant (chs1) has reduced chloroplast protein accumulation at low temperatures (Schneider et al., 1995), and chilling-induced chlorosis of the Arabidopsis paleface mutant, which defines the gene for a homolog of the yeast 18S rRNA methylase (DIM1), is caused by a chilling-sensitive step in chloroplast rRNA processing that inhibits ribosome assembly (Tokuhisa et al., 1998). More recently, it has been demonstrated that a tobacco (Nicotiana tabacum) transplastomic line that is devoid of chloroplast ribosomal protein L33 is sensitive to cold stress (Rogalski et al., 2008).
In addition to chilling-induced effects on chloroplast translation, it has been known for some time that chilling stress results in a decrease in the de novo synthesis and degradation of D1 (Salonen et al., 1998), another process that involves VAR2 activity. This decrease has recently been correlated with ribosome pausing and a delayed translation elongation of psbA mRNAs that results in decreased D1 synthesis (Grennan and Ort, 2007).
Given that rRNA processing and chloroplast protein accumulation are drastically impaired in var2 (Fig. 7) and that D1 turnover is also perturbed in var2 (Kato et al., 2009), it is tempting to speculate that a loss of VAR2 accentuates a cold-induced impairment in one or more of these processes. The latter would be consistent with the suggestion that VAR2 plays an important role in chloroplast biogenesis beyond its extensively studied role in D1 turnover (Chen et al., 2000;Zaltsman et al., 2005a).

Plant Material and Growth Conditions
All Arabidopsis (Arabidopsis thaliana) mutants used in this study are in the Columbia ecotype background. Plants were grown at either 8°C or 22°C under continuous illumination with a light intensity of approximately 100 mmol m 22 s 21 . The svr1-1 single mutant (derived from the TAG-FN suppressor line) has been described (Yu et al., 2008). Other single mutants used in this study include svr7-1, svr3-1, and svr8-1, which were derived from the suppressor lines 004-003 (this work), TAG-11, and 023-005, respectively (F. Yu, X. Liu, and S. Rodermel, unpublished data).

Chlorophyll Determination
Two-week-old seedlings were harvested, weighed, and finely ground in liquid N 2 . Total chlorophyll was extracted with 95% ethanol, and chlorophyll concentrations and chlorophyll a/b ratios were calculated according to Lichtenthaler (1987).

Positional Cloning
Procedures for positional cloning have been described in detail . In brief, an F2 mapping population was generated by crossing svr7-1 with Landsberg erecta. Bulked segregant analysis (Lukowitz et al., 2000) was then performed using a pool of DNAs from 96 F2 plants having an svr7-1 phenotype, and the mutant gene was mapped between simple sequence length polymorphism markers ciw6 and ciw7 on chromosome IV (Bell and Ecker, 1994). The gene was fine-mapped using markers that were designed based on insertion/deletion or single nucleotide polymorphisms between Landsberg erecta and Columbia ecotypes (Jander et al., 2002;Supplemental Table S1). A total of 665 F2 plants (1,330 chromosomes) were used in fine-mapping the suppressor gene.

Complementation of 004-003
A genomic DNA fragment encompassing the full-length At4g16390 (SVR7) sequence was amplified by pfu Turbo DNA polymerase (Stratagene) using primers 16390F and 16390R, and the amplified fragment was cloned into the BamHI site of pBluescript KS+. Following verification of the SVR7 sequence, the BamHI fragment was subcloned into a modified pBI121 vector (Yu et al., 2004); sequences in this vector are driven by the cauliflower mosaic virus 35S promoter. The resulting construct was transformed into Agrobacterium tumefaciens by electroporation, and the floral dip method was used to transform 004-003 plants (Clough and Bent, 1998). After transformation, T1 seeds were collected and grown on plates in a medium containing 13 Murashige and Skoog salts, 1% Suc, and 0.8% agar (pH 5.7) supplemented with kanamycin (50 mg L 21 ). The phenotype of the complementation lines was confirmed in the T2 generation.

Confocal Microscopy of SVR7 Expression
A construct was made containing the full-length SVR7 gene fused at its C terminus to the open reading frame of eGFP in the vector pTF486 (Yu et al., 2008). Genes are driven by the cauliflower mosaic virus 35S promoter in this construct. Primers 16390F and 16390GFPR were used to amplify the SVR7 sequence using pfu Turbo DNA polymerase (Stratagene). The PCR product was digested with BamHI and NcoI and cloned in-frame into pTF486. The resulting construct (designated P35S:SVR7:GFP) was sequenced to verify that the open reading frame of eGFP was correctly fused to the 3# end of SVR7. A control construct lacked the SVR7 sequence (designated P35S:GFP). Methods for transient expression of wild-type Arabidopsis leaf protoplasts have been described (Yoo et al., 2007;Yu et al., 2008). Confocal microscopy (Leica TCS NT) with a fluorescein isothiocyanate-tetramethyl rhodamine isothiocyanate filter combination was used to capture fluorescence signals of GFP and chlorophyll autofluorescence.

DNA, RNA, and Protein Manipulations
Leaf DNAs were isolated using the cetyl-trimethyl-ammonium bromide method (Wetzel et al., 1994), and total leaf RNAs were purified using the Trizol RNA reagent (Invitrogen). Radiolabeled probes were generated according to Yu et al. (2008). Northern-blot analyses were performed as described by Wetzel et al. (1994). Total leaf proteins from 2-week-old seedlings were extracted, and western immunoblot analyses were carried out using a series of antibodies to representative proteins, as described previously (Yu et al., 2008). The quantification of western-blot signals was carried out using QuantityOne software (Bio-Rad).

Generation of svr2 svr7 Double Mutants
The svr7-1 single mutant was crossed with svr2-1 (Yu et al., 2008), and genomic DNAs were isolated from the F2 progeny. The genotype of the F2 plants at these two loci was determined by PCR-based assays: PCR with primers 004-003F and 004-003R was used to determine the genotype of the SVR7 locus, whereas PCR with primers 49970F2 and 49970R2 was used to determine the genotype at the SVR2 locus (Yu et al., 2008).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Ethidium bromide staining of RNA gels.
Supplemental Figure S2. Quantification of immunoblot signal intensities.
Supplemental Table S1. Primers used in this study.