The PPR-SMR Protein ATP4 Is Required for Editing the Chloroplast rps8 mRNA in Rice and Maize 1[OPEN]

Chloroplast gene expression involves the participation of hundreds of pentatricopeptide repeat (PPR) RNA binding proteins, and proteins in the PLS subfamily typically specify sites of RNA editing, whereas those in the P-subfamily typically stabilize RNA, activate translation, or promote intron splicing. Several P-type PPR proteins include a small MutS-related (SMR) domain, but the biochemical contribution of the SMR domain remains enigmatic. Here, we describe a rice ( Oryza sativa ) mutant, osatp4 , lacking the ortholog of ATP4, a PPR-SMR protein in maize ( Zea mays ). osatp4 mutants were chlorotic and had a plastid-ribosome de ﬁ ciency when grown in the cold. Like maize ATP4, OsATP4 was required for the accumulation of dicistronic rpl16 - rpl14 transcripts. Surprisingly, OsATP4 was also required for the editing of a speci ﬁ c nucleotide in the ribosomal protein S8 transcripts, rps8 , and this function was conserved in maize. By contrast, rps8 RNA was edited normally in the maize PROTON gradient regulation3 mutant, pgr3 , which also lacks rpl16 - rpl14 transcripts, indicating that the editing defect in atp4 mutants is not a secondary effect of altered rpl16 - rpl14 RNA metabolism. Expression of the edited rps8 isoform in transgenic osatp4 mutants complemented the cold-sensitive phenotype, indicating that a rps8 expression defect accounts for the cold-sensitivity. We suggest that ATP4 stimulates rps8 editing by facilitating access of a previously characterized PLS-type RNA editing factor to its cognate cis-element upstream of the edited nucleotide. Total isolated using TRIzol Reagent (Invitrogen). Approximately 20 m g of RNA was separated on a 1.5% (v/v) formaldehyde/morpholinopro-panesulphonic acid gel by electrophoresis and capillary-transferred onto the positively charged nylon membrane. Hybridizations were performed with DIG-labeled probes at 50°C for 12 h. Probes were produced by labeling PCR products with the PCR DIG Probe Synthesis Kit (Roche). Hybridization signals were detected using the CDP-Star reagent (Sigma-Aldrich). Prehybridization, hybridization, washing, and detection were performed following Roche ’ s DIG Manual. rps8 rpl16

Chloroplasts originated from a cyanobacterial ancestor (Timmis et al., 2004;Keeling, 2010). The core gene expression machinery in chloroplasts is similar to that in bacteria, but the expression of chloroplast genes also involves more recently acquired steps, including RNA editing, protein-mediated RNA stabilization, and the processing of polycistronic transcripts to smaller isoforms (Stern et al., 2010;Barkan, 2011). Several families of nuclear-encoded RNA binding proteins that mediate these steps emerged through a process of nuclearorganellar coevolution (Barkan, 2011). This is exemplified by the pentatricopeptide repeat (PPR) proteins, a large family of RNA binding proteins involved in various steps in RNA metabolism in chloroplasts and mitochondria (Barkan and Small, 2014). PPR proteins are characterized by tandem repeats of a degenerate 35-amino acid motif (Small and Peeters, 2000). Consecutive repeats stack to form a superhelix, which binds single-stranded RNA, often in a sequencespecific manner (Barkan and Small, 2014). In land plants, the family has dramatically expanded, typically including more than 400 members (Lurin et al., 2004;O'Toole et al., 2008). The PPR family is subdivided into P and PLS subfamilies depending on the presence of long (L) or short (S) variant PPR motifs (Lurin et al., 2004). P-type PPR proteins are usually involved in RNA splicing, stabilization, and translational activation, whereas PLS-type PPR proteins are mostly dedicated to RNA editing (Barkan and Small, 2014).
Some P-type PPR proteins carry a small MutSrelated domain (SMR) at the C terminus (Liu et al., 2013). The SMR domain was originally identified as the C-terminal domain of bacterial MutS2 proteins, which are involved in DNA recombination and repair (Moreira and Philippe, 1999). The SMR domain in MutS2 proteins harbors endonuclease activity and binds branched nucleic acid structures such as those at Holliday junctions (Fukui and Kuramitsu, 2011). The PPR-SMR proteins that have been characterized in plants are implicated primarily in RNA transactions. GENOMES UNCOUPLED1 (GUN1) is required for multiple stressrelated retrograde signaling pathways (Koussevitzky et al., 2007), and affects the efficiency of editing for multiple sites in plastid RNAs during retrograde signaling under norflurazon treatment (Zhao et al., 2019). pTAC2 is necessary for the activity of the plastidencoded RNA polymerase, but its biochemical role is unknown (Pfalz et al., 2006;Williams-Carrier et al., 2014;Wang et al., 2016). PPR-SMR1 is involved in mitochondrial RNA splicing in maize (Zea mays; Chen et al., 2019). PPR53 in maize and its ortholog SOT1 in Arabidopsis (Arabidopsis thaliana) function in processing the 23S and 4.5S ribosomal RNAs (rRNA) and in stabilization of ndhA mRNA (Wu et al., 2016;Zoschke et al., 2016). ATP4, which was so-named because it was the fourth mutant recovered from the maize Photosynthetic Mutant Library (Belcher et al., 2015), is required specifically for accumulation of the chloroplast ATP synthase and increases the translational efficiency of the chloroplast atpB/E mRNA. In addition, ATP4 stabilizes psaJ and dicistronic rpl16-rpl14 RNAs (Zoschke et al., 2012(Zoschke et al., , 2013b, whereas its ortholog in Arabidopsis, SVR7, enhances chloroplast rRNA processing (Liu et al., 2010) and chloroplast ATP synthase levels (Zoschke et al., 2013a). The biochemical role of the SMR domain in PPR-SMR proteins is unclear. The SMR domain of SOT1 has RNA endonuclease activity in vitro (Zhou et al., 2017), but it is unclear how endonuclease activity could account for the phenotypes reported for mutations in genes encoding PPR-SMR proteins.
In this study, we describe a transfer DNA (T-DNA) insertion allele of the rice (Oryza sativa) ortholog of maize atp4. The molecular defects are very similar to those reported previously in maize. In addition, however, we show that the phenotype is cold-sensitive, demonstrate an unanticipated role for ATP4 in rice and maize in editing the rps8 transcript, and provide evidence that a defect in rps8 expression underlies the cold-sensitive phenotype.

Identification and Characterization of the osatp4 T-DNA Mutant
In a previous report, mutants generated by the CRISPR/Cas9 system in the rice atp4 ortholog, denoted The T-DNA insertion in osatp4 (PFG_4A-50341) was positioned at the ninth PPR motif. The gene-specific primers P1 and P2 and the T-DNA border primer P3 used for genotyping are shown. q-F and q-R denote the primer positions for RT-qPCR; cr-osatp4 refers to the CRISRP/Cas9 knockout mutant with the target sites indicated as T1 and T2. C, Genotyping by PCR analysis and Southern blot of the T-DNA insertion mutant. The mutant osatp4,5 was homozygous for the insertion in osatp4 and lacked other T-DNA insertions. This line was used for further analysis. Approximately 10 mg of genomic DNA was digested with the restriction endonuclease HindIII. The probe used for Southern blot hybridizations was a part of the selection marker gene (hygromycin phosphotransferase gene, hpt), indicated by the black line in B. D, Detection of OsATP4 transcripts in wild type and osatp4 by RT-qPCR analysis. Primer positions are indicated in B (q-F and q-R). Actin was used as the reference gene. Three biological replicates were used; technical replicates were n 5 3. Error bars indicate 6 SE.
osppr676 (Liu et al., 2017), caused defects in plant growth, pollen development, and a reduction in the AtpB subunit of the ATP synthase. However, the molecular defects were not characterized to the extent of those in the orthologous mutants in maize and Arabidopsis. To further study the molecular functions of OsATP4 in rice, we identified a T-DNA insertion mutant (PFG_4A-50341), osatp4, whose seedling blades exhibited an albino phenotype when grown in the cold (Fig. 1A). Thermal asymmetric interlaced PCR of the sequences flanking the T-DNA insertion site showed that the T-DNA inserted itself in the coding region, 1,294 bp downstream of the start codon of rice gene Os03g11670 (Fig. 1B). The predicted OsATP4 amino acid sequence includes a predicted chloroplast transit peptide at the N terminus, followed by 10 PPR motifs and an SMR domain at its C terminus (Fig. 1B). In the osatp4 mutant, the T-DNA inserted at codon 432, resulting in the disruption of the ninth PPR motif (Fig. 1B). Southern blot and PCR analyses (Fig. 1C) identified one mutant line with a single T-DNA insertion (line 5 in Fig. 1C), and this line was used for further analysis. OsATP4 transcripts were undetectable by reverse transcription quantitative PCR (RT-qPCR) in osatp4 (Fig. 1D). Mutant plants expressing a transgene encoding OsATP4 were restored to the wild-type phenotype (Fig. 1A, complemented [COM]), indicating that disruption of OsATP4 is responsible for the phenotype.
To further study the function of OsATP4, especially that of the SMR domain, we generated osatp4 mutants by the CRISPR/Cas9 system. Two target sequences (t1 and t2) were chosen as guide RNAs to construct the CRISPR/Cas9 vector (Fig. 1B), which was then transformed into rice callus and used to regenerate plants. Sequence analysis revealed six mutant lines with different deletions or insertions at the target sites (Supplemental Fig. S1A), all of which exhibited the same cold-sensitive phenotype as osatp4 (Supplemental Fig. S1B).
To confirm the subcellular localization of OsATP4, we expressed a fusion protein consisting of full-length OsATP4 fused to the yellow fluorescent protein (YFP) in rice protoplasts. YFP colocalized with red chlorophyll autofluorescence (Supplemental Fig. S2), confirming that OsATP4 is a chloroplast-localized protein (Liu et al., 2017), consistent with the location of the maize (Zoschke et al., 2012) and Arabidopsis orthologs (Liu et al., 2010).

Phenotypic Characterization of osatp4 in Normal and Cold Conditions
When the osatp4 mutant was grown under natural field conditions, alterations in plant height ( Fig. 2A; Table 1) and tiller number ( Fig. 2B; Table 1) were observed. The mutants had 13 tillers on average, which was less than that of the wild type (16 tillers; Table 1). Tillering is an important agronomic trait that affects grain production. Rice tillering occurs in a two-stage process: the formation of an axillary bud, and bud outgrowth (Li et al., 2003). Examination under the stereomicroscope showed that the osatp4 mutation caused Figure 2. Phenotype of osatp4, cr-osatp4, and COM (P OSATP4 : OSATP4) under natural field conditions. A, Phenotypes of mature wild-type (WT), osatp4, cr-osatp4, and COM plants when grown in the field in Wuhan, China (30.4°N,114.2°E) in 2018 (picture taken in October 2018). cr-osatp4 contains a deletion generated by CRISPR/Cas9 using the targeting sequences shown in Figure 1B. Plants were moved to pots for photography. B, Phenotypes of wild type and osatp4 at 3 weeks after transplanting under natural field conditions. C, Tiller buds of five-leaf stage seedlings. The numbers indicate tiller-bud numbers in the wild type and osatp4. .33 6 0.65** no effect on the initiation of axillary buds but inhibited their outgrowth (Fig. 2C). In addition, the mutant exhibited decreases in plant height, panicle length, average grain weight, and yield per plant (Table 1). The mutant also exhibited reduced leaf width and chlorophyll under natural field conditions (Fig. 3A). Similarly, a decrease in chlorophyll content was observed in osatp4 when grown at 30°C (Fig. 3B), however, F v /F m , a measure of maximum PSII quantum yield, was unaffected ( Fig. 3D; Table 2). When grown at 20°C, which confers cold stress on rice (Hasanuzzaman et al., 2018), the chlorophyll content ( Fig. 3C) and the F v /F m value significantly decreased in osatp4 ( Fig. 3E; Table 2).

OsATP4 Is Required for ATP Synthase Accumulation and Exhibits a Global Loss of Photosynthetic Complexes at Reduced Growth Temperatures
Mutations in SVR7 and ATP4, the orthologs of OsATP4 in Arabidopsis and maize, respectively, cause a decrease in ATP synthase levels and have little effect on other photosynthetic complexes (Zoschke et al., 2012(Zoschke et al., , 2013a. Immunoblot analysis of osatp4 mutants grown at 30°C gave similar results ( Fig. 4A): the AtpB and AtpF subunits of the ATP synthase were reduced roughly 4-fold whereas the D1 protein of PSII (PsbA), subunit of cytochrome b 6 f complex (PetB), and Rubisco large subunit accumulated to normal levels. Interestingly, however, when plants were grown at 20°C, all of Chlorophyll content of two-leaf seedlings of wild type and osatp4 grown at 30°C (B) and at 20°C (C). FW, Fresh weight. In B and C, the data represent the mean of five independent experiments 6 SE. Asterisks indicate statistically significant differences according to Student's t test (**P , 0.01). Representative false-color image of F v /F m in two-leaf seedlings grown at 30°C (D) and at 20°C (E). F v /F m values are represented between blue (0, lowest) and red (0.9, highest) extremes of the false-color scale. Table 2. Maximum fluorescence yield F v /F m after 30-min dark adaptation of two-leaf stage seedlings of wild type, osatp4, and COM osatp4 Data represent the means 6 SE (n 5 6) from independent plants. Asterisks indicate statistically significant differences between osatp4 and wild-type plants according to Student's t test (**P , 0.01).

Condition
Maximum  these proteins accumulated to reduced levels (Fig. 4B). Protein reductions of this type can result from global defects in chloroplast gene expression. In fact, it was proposed previously that svr7 mutants have a global reduction in chloroplast translation (Liu et al., 2010). Our results suggest that this function is conserved in rice and that it is particularly important under cold stress.
ATP4 Affects the C-to-U Editing of rps8 RNA in Rice and Maize Recently, the PPR-SMR protein GUN1 was shown to affect chloroplast RNA editing specifically in norflurazon-treated plants, and this effect was linked to an interaction between GUN1 and the accessory editing factor MORF2 (Zhao et al., 2019). To address whether OsATP4 influences RNA editing, we sequenced the transcripts containing 23 chloroplast editing sites (Tsudzuki et al., 2001) in osatp4 and wild-type plants grown at 30°C and at 20°C (Supplemental Table S1). To our surprise, the editing of nucleotide 182 in the rps8 gene (rps8-182) was strongly reduced in osatp4 plants at both growth temperatures. The editing of rps8-182 generates a Ser-to-Leu amino acid change in the RPS8 protein. The editing efficiency was 87% to 98% in wild-type plants, whereas editing was not detectable in osatp4 ( Fig. 5A; Supplemental Table S1) in either our cr-osatp4 CRISPR lines (Fig. 5B) or the independent allele osppr676-1 identified previously (Supplemental Fig.  S3). In the complementation lines expressing the OsATP4 transgene, editing was restored (Fig. 5A). Transcripts from rps8 accumulated normally in the osatp4 mutant (Fig. 5C), indicating that this editing defect is not a secondary effect of a defect in rps8 transcript processing.
To investigate whether the function is conserved in maize, we examined the editing efficiency of rps8-182 in the maize atp4 mutant. The results showed that ZmATP4 is also required for the editing of this nucleotide (Fig. 5D). This function had been missed in previous analyses of this mutant (Zoschke et al., 2012(Zoschke et al., , 2013b.
The Defect in rpl16-rpl14 RNA Accumulation Does Not Cause the rps8 Editing Defect in atp4 Mutants In the orthologous maize atp4 mutant, dicistronic rpl16-rpl14 transcripts fail to accumulate due to a defect in transcript stabilization (Zoschke et al., 2013b). RNA gel-blot analysis of RNA prepared from osatp4 showed that this function is conserved in rice (Fig. 6).
Because rpl16 and rpl14 are adjacent to and cotranscribed with rps8, it seemed possible that the rps8 editing Sequence chromatograms are shown for wild-type (WT), osatp4, and osatp4complemented lines (COM1, COM2, and COM3). The editing site of rps8-182 is shaded in gray. The relative expression level of OsATP4 in these lines is shown to the left. Actin was used as the reference gene. Three technical replicates were performed. Error bars indicate 6 SE. B, The editing efficiency of rps-182 in six CRISPR/CAS9 mutant lines of OsATP4 (cr-osatp4-1, cr-osatp4-2, cr-osatp4-3, cr-osatp4-4, cr-osatp4-5, and cr-osatp4-6). C, RNA gel-blot analysis showing size and abundance of rps8 mRNA in wild type and osatp4. The positions of RNA size markers are shown on the right side. The probe used for northern blotting is indicated by the dashed line in Figure 6. The ethidium-bromide-stained gels are shown below to indicate equal sample loading. D, The editing efficiency of rps8-182 in maize atp4 and pgr3 mutants.
defect is a secondary effect of the rpl16-rpl14 RNA defect. To address that possibility, we took advantage of the fact that the maize pgr3 mutant lacks the same rpl16-rpl14 transcripts as atp4 (Rojas et al., 2018). To determine whether PGR3 affects rps8 editing, we sequenced complementary DNA (cDNA) from rps8 in the maize pgr3 mutant. We found that rps8-182 is edited normally in the maize pgr3 mutant (Fig. 5D), demonstrating that the editing defect does not result from the loss of rpl16-rpl14 dicistronic transcripts.

Expression of the Edited Isoform of RPS8 in osatp4
Mutants Complemented the Cold-Sensitive Phenotype By surveying the editing efficiency of rps8-182 in different cultivated rice accessions (Zhao et al., 2015), we observed that two accessions, W104 and C015, demonstrated 0% editing at this site (Supplemental Fig.  S4, A and B). Both accessions exhibit an albino phenotype when grown at 20°C (Supplemental Fig. S4C), similar to the osatp4 mutant. Interestingly, both accessions have an 8-bp deletion in the DUA1 gene (Supplemental Fig. S4, A and B), which encodes a PPR-PLS protein that was shown previously to affect numerous editing sites, including rps8-182 (Cui et al., 2019). C015 and W104 have the same 8-bp deletion as reported in the original DUA1 study (Cui et al., 2019). We observed the rps8 editing defect in green plants from these accessions grown under standard conditions, as well as in cold-stressed plants. These results suggest a possible correlation between rps8 editing and cold-sensitivity.
To test whether the rps8 expression defect accounts for the cold-sensitivity of osatp4 mutants, we transformed osatp4 mutants with a nuclear transgene encoding the edited isoform of RPS8 (rps8-182U). The chloroplast transit peptide from the rbcS gene was used to target RPS8 into chloroplasts, and transcription was driven by the ubiquitin promoter, which is a strong constitutive promoter. Seeds of transgenic T 0 plants (RPS8U) were cultivated in a growth chamber under 12-h light/12-h dark at a constant temperature of 20°C. Expression of the nuclear RPS8U transgene restored much of the chlorophyll to osatp4 mutants grown in the cold (Fig. 7). These results suggest that the rps8 editing defect underlies the cold-sensitive phenotype of osatp4. However, we cannot eliminate the possibility that elevated expression of RPS8, independent of the editing defect, accounts for the complementation we observed.

DISCUSSION
In this work, we describe phenotypes of T-DNA insertion and CRISPR-induced mutant alleles of the rice ortholog of maize ATP4. Our results showed that the molecular defects in osatp4 are very similar to those reported previously in maize, including defects in accumulation of the ATP synthase and in the stability of dicistronic rpl14-rpl16 transcripts (Figs. 4 and 6). Intriguingly, our results also revealed that the editing of rps8-182 is absent in both osatp4 and zmatp4 mutants (Fig. 5, A, B, and D). Other characterized PPR-SMR proteins have roles in transcription, transcript processing, RNA stabilization, and translation (reviewed by Zhang and Lu, 2019). Recently, GUN1 was shown to impact RNA editing specifically in norflurazon-treated plants (Zhao et al., 2019). However, a role for PPR-SMR proteins in RNA editing under nonstress growth conditions had not been observed previously. The editing of rps8 is normal in the maize pgr3 mutant, which also lacks dicistronic rpl16-rpl14 transcripts (Rojas et al., 2018). Conversely, the native rice accessions C015 and W104 showed a defect in rps8-182 editing (Supplemental Fig. S4, A and B), but accumulated rpl16-rpl14 transcripts normally ( Fig. 6; Supplemental Fig. S5). These results provided strong evidence that the editing of rps8-182 Figure 6. RNA gel-blot hybridization demonstrating loss of rpl16 transcripts in osatp4 mutants. The diagram displays genes surrounding rpl16 in the rice chloroplast genome with probes used for RNA gel-blot hybridizations indicated by dotted lines. Total leaf RNA (20 mg per lane) of wild type (WT), osatp4, and W104 was analyzed by RNA gel blot using a rpl16 exon 2 probe. The positions of RNA size markers are shown on the right side. The symbols ; and * indicate the unspliced and spliced forms of processed rpl16-rpl14 transcripts, respectively, according to previous results of maize atp4 (Zoschke et al., 2013b). The ethidium-bromide-stained gels are shown below to indicate equal sample loading. and stabilization of rpl16-rpl14 occur independently of one another. P-type PPR proteins are usually involved in RNA splicing, RNA stabilization, and translational activation, but two P-type PPRs affect editing (Doniwa et al., 2010;Leu et al., 2016). Although ATP4, with its SMR domain, is not a typical P-type protein, these results suggest that P-type PPR proteins may affect RNA editing more often than previously considered, and it would be interesting to analyze RNA editing status in previously reported P-class PPR mutants.
Sites of C-to-U RNA editing in chloroplasts are generally specified by PLS-type PPR proteins, which bind a short distance upstream of the edited cytidine (Barkan and Small, 2014). The PLS-type PPR subfamily is divided into three subgroups: E, E1, and DYW, according to the characteristics of the C-terminal motifs. The DYW domain harbors the deaminase activity required for C-to-U conversion (Oldenkott et al., 2019;Hayes and Santibanez, 2020;Small et al., 2020). It is unlikely that ATP4 functions directly in editing because it lacks both a DYW domain and a PLS-type PPR repeat tract. Instead, we favor the possibility that ATP4 facilitates access of a classic PLS-type editing factor to the rps8 site. This possibility is in line with previous evidence that ATP4 and a different PPR-SMR protein, PPR53, support the action of other PPR proteins that bind and stabilize specific transcript termini (Zoschke et al., 2016;Rojas et al., 2018). It is intriguing in this regard that the RNA immediately upstream of the rps8-182 site is predicted by the tool Mfold (Zuker, 2003) to form an RNA hairpin with a substantial double-stranded stem (Fig. 8A). DUA1, a PLS-DYW type PPR protein, is required for rps8 editing, and its binding site has been mapped to the downstream side of the stem (blue nucleotides in Fig. 8, A and C; Cui et al., 2019). Secondary structures of this nature inhibit binding by PPR proteins in vitro Figure 7. Expression of the edited isoform of RPS8 from the nuclear genome rescues the cold-sensitive phenotype of osatp4. A, The phenotypes of wild-type (WT), osatp4, and RPS8U at the two-leaf stage in cold (20°C). Plants labeled RPS8U contain a nuclear transgene encoding the edited isoform of RPS8. Three independent lines are shown. B, Chlorophyll content of two-leaf seedlings of wild-type, osatp4, and RPS8U in cold (20°C), FW, Fresh weight. An average of eight biological replicates is shown. Error bars represent SE.  (Zuker, 2003); dg 5 28.9 kcal mol 21 . The deep blue C at the base of the stem is the site of editing (rps8-182). Nucleotides in the predicted binding site of the editing factor DUA1 (Cui et al., 2019) are shaded in light blue. Nucleotides shaded in red match the predicted binding site of PPR motifs in OsATP4 Yan et al., 2019). B, The amino acid residues at positions 5 and 35 of the 10 PPR motifs present in OsATP4 and the predicted binding nucleic acids based on the PPR code. The rps8 RNA sequence is shown underneath. Highly correlated matches are shaded in red. C, The amino acid residues at positions 5 and 35 of the 10 PPR motifs present in DUA1 and its binding site in rps8 RNA (Cui et al., 2019). Highly correlated matches are shaded in light blue. D, Model for the role of OsATP4 in rps8 editing. A potential ATP4 binding site maps to the left side of the stem-loop. Binding of ATP4 to this site may disrupt the stem-loop, and thereby facilitate binding of DUA1. (McDermott et al., 2018) and in vivo (Kindgren et al., 2015). The PPR code Yan et al., 2019) suggests a possible binding site for ATP4 on the 59 side of the predicted stem (red nucleotides in Fig. 8,  A and B). We suggest that binding of ATP4 to this site may reduce the fraction of the time that the DUA1 binding site is sequestered in this RNA hairpin (Fig. 8D). An analogous mechanism has been proposed for PPR translational activators such as PPR10 (Prikryl et al., 2011). P-type PPR proteins that stabilize RNA termini typically have long PPR tracts and bind with high affinity to RNA. PLS-type PPR proteins involved in RNA editing likely bind with lower affinity, as these bind in open reading frames and stable binding might interfere with translation (Barkan and Small, 2014). In this case, we proposed that ATP4 transiently binds to rps8 mRNA to facilitate RNA editing. The proposed binding of ATP4's PPR tract to discontinuous RNA segments (Fig. 8B) seems plausible in light of the diversity of RNA ligands discovered for several other PPR proteins (Rojas et al., 2018), and the recent discovery of an alternative mode of RNA interaction for the PPR domain in protein-only RNAse P (Teramoto et al., 2020). Furthermore, transient binding is consistent with the finding that maize ATP4 did not detectably coimmunoprecipitate rps8 transcripts in RNA co-immunoprecipitation microarray analysis (Zoschke et al., 2012). Alternatively, ATP4 might interact directly with DUA1 and affect its RNA binding or editing activities. How the SMR domain contributes to this activity is unclear, but the fact that SMR domains in other contexts bind branched nucleic acid structures (Fukui and Kuramitsu, 2011) may be relevant. For example, the SMR domain might bind to a branched structure at the base of an RNA hairpin and facilitate hairpin unwinding. This possibility could be explored in the future with in vitro assays and recombinant proteins.
Low temperature is a major abiotic stress that adversely affects plant growth, productivity, and quality. Rice evolved in tropical, subtropical, and temperate areas, and the optimal growth temperature is between 27°C and 32°C (Hasanuzzaman et al., 2018). Therefore, constant growth under 20°C is considered cold-stress for rice (Hasanuzzaman et al., 2018). Understanding cold-adaptability is critical to the improvement of coldtolerance in rice. Previous studies have shown that RPL33 and RPS15 are required for maintaining sufficiently high chloroplast translation capacity under cold-stress in tobacco (Nicotiana tabacum; Rogalski et al., 2008;Fleischmann et al., 2011). Rps8 in Escherichia coli participates in the assembly of 30S ribosomal subunits (Mizushima and Nomura, 1970;Sashital et al., 2014) and is essential for survival of the cell (Shoji et al., 2011). The RNA editing of rps8-182 generates a Ser-to-Leu amino acid change in the RPS8 protein. This is a nonconservative change, as Ser is a polar amino acid whereas Leu is a nonpolar aliphatic amino acid. Impaired RNA editing of rps8-182 likely disrupts RPS8 function, and along with the defect of rpl16-rpl14 RNA stabilization, could compromise chloroplast translation capacity (Supplemental Fig. S6) and lead to a coldsensitive phenotype (Fig. 1A). In our study, expression of the edited isoform of rps8 in the osatp4 background complemented the cold-sensitive phenotype (Fig. 7). In the future, expression of the unedited isoform of rps8 in the mutant background would be interesting to distinguish whether elevated expression of RPS8 or the edited RPS8 accounts for the complementation we observed. Nevertheless, our results provided a strong evidence that optimal RPS8 expression contributes to cold-adaptability in rice.

Plant Materials and Growth Conditions
Rice (Oryza sativa) T-DNA insertion lines for PPR genes were obtained by searching the OrgGenesDB (https://orygenesdb.cirad.fr/). OsATP4 corresponded to the T-DNA insertion line 4A-50341. The wild-type background of this line is the cv Dongjin. P1, P2, and P3 of genotyping primers are listed in Supplemental Table S2. C015 and W104 are two rice accessions belonging to the aus subpopulation (Zhao et al., 2015). Agronomic trait analyses were performed by growing wild-type and mutant plants under natural field conditions at the experimental station of Huazhong Agricultural University, Wuhan,China (30.4°N,114.2°E). To assess the phenotypes under cold-stress, seeds were germinated at 37°C for 3 d in the dark. Uniformly germinated seedlings were then selected and grown in a growth chamber under 12-h light/12-h dark at constant temperatures of 20°C and 30°C (as control condition). Maize (Zea mays) atp4 and pgr3 plants were sown in soil and grown in diurnal cycles (16-h light/ 8-h dark) at a temperature of 28°C and 26°C for the light and dark periods, respectively.

Complementation of osatp4 and CRISPR/Cas9 Knockout of OsATP4
For transgene complementation, a 5,617-bp wild-type genomic DNA fragment containing the entire OsATP4 coding region with a 3,156-bp upstream region and a 430-bp downstream sequence was amplified using the primers Pcom-F and Pcom-R (Supplemental Table S2). The primers incorporated a KpnI site at the N-terminal end and the C-terminal end. PCR products were cloned into the pCAMBIA2300 vector. The pCAMBIA2300-OsATP4 plasmid was introduced into Agrobacterium tumefaciens EHA105 by electroporation and then used to infect the calli induced from mature seeds of osatp4 mutants following a published method (Hiei et al., 1994), and the resistant calli were selected with 15 mg L 21 G418.
To express the edited RPS8 isoform in osatp4 mutants, plasmid pRPS8U (Ubi:ctp:rps8-182U:35S polyA) was constructed, in which the edited isoform of RPS8 (the 182th nucleotide of rps8 is T) was preceded by the chloroplast transit peptide from the rbcS gene and the ubiquitin promoter, with Isoptericola variabilis-EPSPS as selection marker, as described in Cui et al. (2016). The plasmid was transformed into osatp4 calli, and transformants were selected with 200 mg L 21 glyphosate.
To generate CRISPR-mediated alterations in the SMR domain of OsATP4, the targeting sequences shown in Supplemental Table S2 were designed. The target-site sequences were cloned into the single guide RNA expression cassette pYLgRNA and then into the destination vector pYLCRISPR/Cas9-MH. The CRISPR/Cas9 vector system for multiplex targeting of gene sites in plants was provided by Yaoguang Liu (South China Agriculture University) and applied as described in Ma et al. (2015). The CRISPR vectors were introduced into the rice 'Dongjin' by A. tumefaciens-mediated transformation following a previous study (Hiei et al., 1994). Nucleotide sequences were analyzed with Sequencher software (Gene Codes Corporation).

Measurement of Chlorophyll
Chlorophyll content was measured using the method described by Arnon (1949). The leaves of two-leaf seedlings were harvested, weighed, and finely ground in liquid N 2 . Total chlorophyll was extracted with a mixture of ethanol, acetone, and water (4.5:4.5:1, v/v/v). Absorption values at 663 and 645 nm were recorded to calculate the chlorophyll content.
For the chlorophyll fluorescence measurement, the seedlings were incubated in the dark for 30 min and subsequently placed in a kinetic imaging fluorometer (model no. 800MF; FluorCam). The F v /F m was measured in accordance with the manufacturer's instructions.

Confocal Microscopy of OsATP4 Expression
The full-length OsATP4 gene was fused at its C terminus to the open reading frame of YFP in the vector pXCSG-YFP (Feys et al., 2005). The full-length cDNA of OsATP4 was amplified using the primers listed in Supplemental Table S1. The plasmid was transformed into protoplasts obtained from green rice tissue as described in Zhang et al. (2011). Protoplasts were incubated overnight at room temperature in the dark, and the fluorescence signal was observed with a confocal microscope (Leica).

RT-qPCR
Total RNA was extracted with Trizol reagent (TransGen Biotech) from the osatp4 and wild-type cv Dongjin seedlings. The RNA was reverse-transcribed using M-MLV reverse transcriptase (Invitrogen) and random primers to obtain cDNA in accordance with the manufacturer's instructions. Quantitative PCR was performed in a total volume of 20 mL containing 1 mL of cDNA products, 10-mL FastStart Universal SYBR Green Master (Roche), and 30 mM of genespecific primers (Supplemental Table S2). A model no. 7500 Real-Time PCR System (Applied Biosystems) was used to perform the RT-qPCR with the following program: 95°C for 10 min, followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. Actin was used as the reference gene. Changes in transcription were quantified using the ΔΔ CT method (Livak and Schmittgen, 2001).

Analysis of RNA Editing
To assay RNA editing, Multiplex RT-PCR Mass Spectrometry (Suizhi Biotech; Germain et al., 2015) was used with customized primers (Supplemental Table S2). The editing efficiency of rps8, ndhB-467, ndhB-586, ndhB-611, ndhB-737, ndhB-830, rps14-80, and rpoB-560 was determined by sequencing of RT-PCR products. The cDNA above was used as the template. RT-PCR was performed with a mixture containing 1 mL of cDNA, 2 mL of 103 PCR buffer (Mg 21 plus), 0.4 mL of dNTPs (10 mM), 0.2 mL of primer-F (10 mM), 0.2 mL of primer-R 10 (10 mM; Primers are presented in Supplemental Table S2), and 1 U of Taq DNA polymerase in a total volume of 20 mL. The amplified PCR products were directly sequenced by Sangon Biotech, and the results were compared with the corresponding DNA sequences for each transcript. Nucleotide sequences were analyzed with Sequencher software (Gene Codes Corporation).

Southern Blot Analysis
Genomic DNA was isolated by the cetyl trimethyl ammonium bromide method (Murray and Thompson, 1980). Southern blot analysis was performed with the probe labeled by digoxigenin (DIG). Approximately 10 mg of genomic DNA digested with restriction endonuclease HindIII was separated on a 0.8% (w/v) agarose gel by electrophoresis and then capillary-transferred onto the positively charged nylon membrane. The DIG-labeled probe was prepared with PCR by supplementing 0.01 mM of DIG-deoxyuridine triphosphate into the reaction mixture. Prehybridization, hybridization, and chemiluminescent detection were performed following the DIG application manual provided by Roche Diagnostics.

Northern Blot
Total RNA was isolated using TRIzol Reagent (Invitrogen). Approximately 20 mg of RNA was separated on a 1.5% (v/v) formaldehyde/morpholinopropanesulphonic acid gel by electrophoresis and capillary-transferred onto the positively charged nylon membrane. Hybridizations were performed with DIG-labeled probes at 50°C for 12 h. Probes were produced by labeling PCR products with the PCR DIG Probe Synthesis Kit (Roche). Hybridization signals were detected using the CDP-Star reagent (Sigma-Aldrich). Prehybridization, hybridization, washing, and detection were performed following Roche's DIG Application Manual. Supplemental Table S2 lists the primers of rps8 and rpl16 primers used to generate PCR products for DIG labeling.

Immunoblot Analysis
Total proteins of osatp4 and wild-type cv Dongjin seedlings were isolated with an extraction buffer containing 50 mM of Tris-HCl at pH 8.0, 150 mM of NaCl, 1% (v/v) NP40, and 23 protease inhibitor (COMPLETE; Roche Diagnostics). The protein concentration was determined by the Bradford method (Bradford, 1976). Bovine serum albumin was used as the standard. The proteins were separated by SDS-PAGE and then transferred onto a polyvinylidene fluoride membrane by semidry transfer. The subsequent immunodetection was performed following a standard protocol for the primary antibody. The secondary antibody HRP-conjugated goat antirabbit IgG (ABclonal) was diluted at a ratio of 1:10,000. The AtpB and PetB probings used 10 mg of leaf proteins and antisera from Agrisera (AS05085 and AS03034, respectively) diluted at a ratio of 1:10,000. The AtpF, PsbA, PetC, and PsaD probings used 30-mg leaf proteins and antisera (AS101604, AS05084, AS07230, and AS09461, respectively; Agrisera) diluted at a ratio of 1:1,000.

rRNA Ratio Analysis
Total RNA was isolated using an miRNeasy kit (Qiagen; https://www. qiagen.com/). rRNAs were analyzed and quantified by analyzing total RNA of the seedlings with the Experion Automated Electrophoresis System (Bio-Rad). rRNA ratios were calculated by comparing the area of the peaks corresponding to the rRNAs.

Supplemental Data
The following supplemental data are available.
Supplemental Figure S2. Subcellular localization of OsATP4-YFP when transiently expressed in rice protoplasts.
Supplemental Figure S4. rps8 editing defects in the W104 and C015 rice accessions.
Supplemental Figure S5. RNA gel blot hybridization of rpl16 transcripts in C015.
Supplemental Figure S6. Accumulation of rRNAs in wild type and osatp4.
Supplemental Table S1. The editing efficiency of wild type and osatp4 under both 30°C and 20°C conditions.
Supplemental Table S2. List of oligonucleotides used in this study.