GhCTSF1, a short PPR protein with a conserved role in chloroplast development and photosynthesis, participates in intron splicing of rpoC1 and ycf3-2 transcripts in cotton

Abstract

Cotton is one of the most important textile fibers worldwide. As crucial agronomic traits, leaves play an essential role in the growth, disease resistance, fiber quality, and yield of cotton plants. Pentatricopeptide repeat (PPR) proteins are a large family of nuclear-encoded proteins involved in organellar or nuclear RNA metabolism. Using a virus-induced gene silencing assay, we found that cotton plants displayed variegated yellow leaf phenotypes with decreased chlorophyll content when expression of the PPR gene GhCTSF1 was silenced. GhCTSF1 encodes a chloroplast-localized protein that contains only two PPR motifs. Disruption of GhCTSF1 substantially reduces the splicing efficiency of rpoC1 intron 1 and ycf3 intron 2. Loss of function of the GhCTSF1 ortholog EMB1417 causes splicing defects in rpoC1 and ycf3-2, leading to impaired chloroplast structure and decreased photosynthetic rates in Arabidopsis. We also found that GhCTSF1 interacts with two splicing factors, GhCRS2 and GhWTF1. Defects in GhCRS2 and GhWTF1 severely affect intron splicing of rpoC1 and ycf3-2 in cotton, leading to defects in chloroplast development and a reduction in photosynthesis. Our results suggest that GhCTSF1 is specifically required for splicing rpoC1 and ycf3-2 in cooperation with GhCRS2 and GhWTF1.

This study reports that GhCTSF1 encodes a chloroplast-localized protein with two PPR motifs that is required for splicing of rpoC1 and ycf3-2 through interaction with GhCRS2 and GhWTF1. Silencing of CTSF1 will decrease chlorophyll content and impair chloroplast development and photosynthesis in Arabidopsis and cotton.

Introduction

Chloroplasts are the organelles responsible for photosynthesis in higher plants and play essential roles in other aspects of plant physiology and development, such as amino acid and fatty acid biosynthesis, environmental stimulus sensing, cell signaling, and plant immunity (Hölzl and Dörmann, 2019; Teardo et al., 2019; Kachroo et al., 2021; Li and Kim, 2022). Chloroplasts evolved from a cyanobacterial ancestor and established an endosymbiotic relationship with eukaryotic cells. During endosymbiosis, some cyanobacterial genes were lost from the genome or relocated to the host nucleus, whereas a few genes were retained in the plastid genome (Hadariova et al., 2018). Most of the remaining plastid genes encode components of the photosynthetic apparatus. These include the major photosystem I (PSI) and PSII subunits, the cytochrome b6f complex, the NAD (P)H dehydrogenase-like (NDH) complex, and some ATP synthase subunits (Zhang et al., 2023b). The plastid genome also encodes some of the plastid translation and transcription machinery and a few key chloroplast biogenesis genes such as ycf1, ycf2, clpP1, accD, and matK. Angiosperm plastid genomes have highly conserved structures and gene content. However, despite the lower gene content of higher plant chloroplasts, their RNA processing is more complex than that of bacteria (Zoschke and Bock, 2018; Small et al., 2023).

Several chloroplast genes that encode proteins or structural RNAs are disrupted by introns in land plants. The plastid genomes of land plants contain approximately 20 group II introns and a single group I intron (within the pre-trnL-UAA region) (Zhang et al., 2021). A variety of nuclear-encoded splicing factors have been reported to participate in intron splicing in higher plant organelles; these include chloroplast RNA splicing and ribosome maturation (CRM) proteins, pentatricopeptide repeat (PPR) proteins, plant organellar RNA recognition (PORR)-domain proteins, and members of the accumulation of photosystem one (APO) domain family (Wang et al., 2022). PORR-domain (formerly called DUF860) proteins are present in angiosperms and contain an RNA-binding domain. Members of the PORR family, including WTF1, WTF9, LEFKOTHEA, and RPD1, are involved in splicing of organellar group II introns (Kroeger et al., 2009; Francs-Small et al., 2012; Daras et al., 2019; Edris et al., 2023). The plant-specific domain of unknown function DUF794 is an RNA-binding domain that harbors zinc-binding motifs, and the DUF794 family member APO1 is required for splicing of several group II introns in chloroplasts (Watkins et al., 2011). Studies have shown that several CRM-domain-containing proteins, including CFM2, CFM3, chloroplast RNA splicing 1 (CRS1), CRS2-associated factor 1 (CAF1), and CAF2, are critical for chloroplast RNA splicing (Wang et al., 2022). ZmCFM2 and AtCFM2 have four CRM domains and participate in splicing ndhA and ycf3-1 subgroup IIB and group I introns (Asakura and Barkan, 2007). CFM3 and CRS1 contain three CRM domains, and AtCRS1 and ZmCRS1 are associated with splicing of atpF (Ostersetzer et al., 2005). CAF1 and CAF2 contain two CRM domains required for the splicing of ndhA, ndhB, petB, petD, rpl16, rps16, trnG, and ycf3-1 in maize and A. thaliana (Wang et al., 2020). CRS2 encodes a protein that is homologous to peptidyl-tRNA hydrolases (PTHs) but has lost PTH activity (Jenkins and Barkan, 2001). In maize, CRS2 can interact with CAF1 and CAF2 to form CRS2–CAF1 and CRS2–CAF2 complexes, which participate in splicing of different subsets of chloroplast group II introns and in the regulation of chloroplast development (Asakura et al., 2008).

In addition to the proteins mentioned above, PPR proteins also function in post-transcriptional processing in chloroplasts. PPR proteins are characterized by tandem copies of a degenerate 35-amino-acid repeat and are encoded by one of the most prominent gene families, which contains more than 450 members in angiosperms (Schmitz-Linneweber and Small, 2008; Gutmann et al., 2020). PPR proteins are divided into the P and PLS subfamilies. The PLS subfamily can be further classified into E, E+, and DYW groups on the basis of their specific C-terminal motifs. Most PPR proteins are targeted to the mitochondria or plastids and play essential roles in a wide range of organelle-related physiological and developmental functions such as photosynthesis, respiration, cytoplasmic male sterility, and early embryogenesis (Barkan and Small, 2014). Studies have revealed that most P-subclass PPR proteins are involved in RNA stabilization, translation initiation, RNA cleavage, RNA maturation, and intron splicing (Cao et al., 2022; Barkan and Small, 2014).

The cotton (Gossypium hirsutum) genome contains 1059 PPR proteins, but only a few studies have investigated their functions (He et al., 2019). In the present study, we characterized a nuclear-encoded P-type PPR protein with only two PPR motifs, GhCTSF1, which is highly expressed in leaves and subcellularly localized in chloroplasts. Silencing of GhCTSF1 decreases chlorophyll content and impairs chloroplast development and photosynthesis. In A. thaliana, loss-of-function mutations in GhCTSF1 orthologs result in similar phenotypes. Further investigation revealed that GhCTSF1 participates in the cis-splicing of ycf3 intron 2 and rpoC1 by interacting with the chloroplast-targeted PORR-domain-containing protein GhWTF1 and the splicing facilitator GhCRS2. Disruption of GhWTF1 and GhCRS2 by CRISPR–Cas9 severely affected the intron splicing of rpoC1 and ycf3-2, impairing chloroplast structure and reducing photosynthesis in cotton. Our data revealed that GhCTSF1 was required for splicing of rpoC1 and ycf3-2 through its interaction with GhCRS2 and GhWTF1.

Results

The GhCTSF1 gene plays an important role in chloroplast development and photosynthesis

The leaf is the primary photosynthetic organ in most plants, and photosynthesis is performed mainly by chloroplasts in the leaf mesophyll cells. Previous studies have shown that PPR proteins are involved in leaf development (Liu et al., 2010; Huang et al., 2018). There are 1059 PPR proteins in the widely cultivated allotetraploid cotton G. hirsutum (He et al., 2019), 10 in algae (Schmitz-Linneweber and Small, 2008), and nearly 100 in Physcomitrella patens (O'Toole et al., 2008), suggesting plant-specific functions. We analyzed transcriptome data from different G. hirsutum tissues to identify the PPR proteins involved in cotton leaf development. Twenty PPR genes were highly expressed in the leaves (FPKM > 10; Supplemental Table 1) and considered essential for leaf development. To assess the physiological functions of the candidates, gene-specific fragments (approximately 300 bp) were cloned and inserted into the pCLCrV-A vector for virus-induced gene silencing (VIGS) to suppress the expression of these genes in cotton. After inoculation with the gene-silencing vector, we found that CotAD_24272-silenced cotton plants displayed a yellow-variegated leaf phenotype. Three specific fragments of CotAD_24272 (GhCTSF1) cDNA were selected for silencing to confirm the role of GhCTSF1. As shown in Figure 1A, the variegated leaf phenotypes were observed in all three VIGS lines. Expression of GhCTSF1 in GhCTSF1-silenced leaves was assessed by quantitative reverse-transcription polymerase chain reaction (RT–qPCR). The transcript levels of GhCTSF1 in the three silenced plants were reduced to 40%, 29%, and 57% of those in the empty-vector (CLCrV-A) plants (Figure 1B), indicating that GhCTSF1 was effectively silenced in the VIGS plants. Consistent with the yellow-variegated color of the leaves, chlorophyll a and b contents were significantly lower in the GhCTSF1-silenced leaves compared with those of empty-vector plants (Figure 1C).

Figure 1.

Phenotypic characterization of GhCTSF1-silenced plants.

(A) The leaf phenotypes of 2-month-old empty-vector control plants (CLCrV-A) and GhCTSF1-silenced plants (GhCTSF1-RNAi). Scale bars, 1 cm.

(B) Quantification of the expression level of GhCTSF1 in CLCrV-A and GhCTSF1-RNAi plants. Error bars represent mean ± SD. Asterisks indicate significant differences (∗∗∗P < 0.001) by Student’s t-test.

(C) Chlorophyll a and b contents of leaves from CLCrV-A and GhCTSF1-RNAi plants. Error bars indicate mean ± SD (n = 3), and asterisks indicate statistically significant differences by Student’s t-test: ∗∗P < 0.01 and ∗∗∗P < 0.001.

(D) Chloroplast ultrastructure in true leaves from 2-month-old plants. Scale bars, 1 μm. The chloroplasts of GhCTSF1-RNAi plants were from the albino sector of the leaf. SG, starch grain; Thy, thylakoid; Gr, granum; PG, plastoglobule.

(E–G) The quantum yield of PSI (E), electron transport rate of PSI (F), and non-photochemical quenching (G) of CLCrV-A and GhCTSF1-RNAi-2 plants. Data in (E)–(G) are shown as means ± SD of three individual replicates.

Chloroplast ultrastructure was analyzed using a transmission electron microscope (TEM) to investigate the physiological role of GhCTSF1 in chloroplast biogenesis. Chloroplasts in the leaves of empty-vector plants showed a well-structured system of stromal and grana thylakoids, whereas chloroplast formation was altered in the yellow-variegated leaves of GhCTSF1-silenced plants. No thylakoids or plastoglobules were observed in the GhCTSF1-silenced leaves (Figure 1D). In addition, the quantum yield of PSI, electron transport rate of PSI, and non-photochemical quenching were significantly lower in GhCTSF1-silenced plants than in the empty-vector plants (Figures 1E–1G). These results indicated that the GhCTSF1 gene product is important for chloroplast development and photosynthesis.

GhCTSF1 is a chloroplast-localized P-subfamily PPR protein

Sequence analysis revealed that GhCTSF1 encodes a protein containing 312 amino acids with only two PPR repeats (Figure 2A). We performed RT–qPCR with gene-specific primers to document GhCTSF1 expression patterns in major cotton tissues. GhCTSF1 was highly expressed in the leaves (Figure 2B), consistent with the leaf phenotype of the GhCTSF1-silenced plants.

Figure 2.

Domain structure and subcellular localization of GhCTSF1.

(A) Schematic diagram of the functional domains of GhCTSF1. cTP, chloroplast transit peptide predicted by TargetP.

(B) Quantitative RT–PCR (RT–qPCR) analysis of relative GhCTSF1 transcript levels in various tissues and fibers at different stages. The cotton UBQ7 gene was used as an internal control. Values are means ± SD (n = 3).

(C) Subcellular localization of the GhCTSF1–GFP fusion protein in N. benthamiana leaf epidermal cells. GhCTSF1–GFP exhibits a chloroplast localization pattern, whereas free GFP is cytosolic and nuclear. GFP, green fluorescent protein; Chlorophyll, chlorophyll autofluorescence.

(D) The GhCTSF1–mCherry construct co-localized with WTF1–GFP. Protoplasts were obtained from N. benthamiana leaves expressing GhCTSF1–mCherry and WTF1–GFP/PEND–GFP/PGL34–GFP/PIC1–GFP fusion proteins. Scale bar, 10 μm.

Using TargetP, GhCTSF1 was predicted to be localized in plastids (Emanuelsson et al., 2000) (Figure 2A). To experimentally determine the subcellular localization of GhCTSF1, a 35S::GhCTSF1–GFP (green fluorescent protein) construct was transiently expressed in tobacco leaf epidermal cells. The GFP signal of GhCTSF1–GFP was present mainly in the chloroplasts and merged with the chlorophyll autofluorescence (Figure 2C). By contrast, the GFP signal of the blank control GFP (35S::GFP) was visible in the cytosol and nucleus (Figure 2C). These results indicated that GhCTSF1 is localized in the chloroplast. To further investigate its precise localization, specific proteins targeting different chloroplast structures, including plastoglobules (PGL34) (Vidi et al., 2007), stroma (WTF1) (Kroeger et al., 2009), nucleoid (PEND) (Terasawa and Sato, 2009), and inner envelope membrane (PIC1) (Duy et al., 2007), were fused with a GFP tag. The resulting fusion proteins were transiently expressed in N. benthamiana leaf protoplasts together with the GhCTSF1–mCherry fusion protein. The GhCTSF1–mCherry fusion protein co-localized with WTF1–GFP (Figure 2D). We also investigated the subcellular localization of GhCTSF1 using GhCTSF1–Myc transgenic A. thaliana plants. Immunoblot analysis showed that GhCTSF1 was localized to the chloroplast stroma and thylakoid fraction (Supplemental Figure 1). All these results indicated that GhCTSF1 may play roles in the chloroplast.

GhCTSF1 is required for RNA splicing of plastid rpoC1 and ycf3-2

PPR proteins participate in various organellar RNA metabolic pathways. Strand-specific RNA sequencing (RNA-seq) was performed to gain a global understanding of the role of GhCTSF1 in chloroplast RNA metabolism. We identified 3013 differentially expressed genes between GhCTSF1-silenced plants and CLCrV-A plants, using a fold change cutoff of 2 and P < 0.05 (Supplemental Figure 2A). Among these, 2118 genes were downregulated and 895 genes were upregulated in GhCTSF1-silenced plants compared with CLCrV-A plants (Supplemental Figure 2B). KEGG analysis of the differentially expressed genes revealed that ribosomes were the main enriched pathway (Supplemental Figure 2C). Ribosome enrichment in GhCTSF1-silenced plants reflects crosstalk between the nucleus and chloroplast. In addition, photosynthetic damage signals that originate from the plastids influence nuclear gene expression (Kendrick et al., 2022; Zhang et al., 2023a).

Because GhCTSF1 was a chloroplast-localized PPR protein, we first quantified the RNA-editing levels and RNA expression patterns of all known plastid genes. On the basis of the RNA-seq data, there were no differences in plastid RNA editing between GhCTSF1-silenced and CLCrV-A plants, a conclusion that was further confirmed by Sanger sequencing (Supplemental Figure 3; Supplemental Table 2). We investigated the RNA expression profiles of all known plastid genes and found that 22 transcripts were downregulated in all three GhCTSF1-silenced plants (Figures 3A and 3B). Among these, ycf3 and rpoC1 were the most abundant (Figures 3B and 3C). To test whether GhCTSF1 was involved in the regulation of mRNA stability, leaves of GhCTSF1-silenced and CLCrV-A plants were treated with actinomycin D. RT–qPCR result showed no significant difference in mRNA stability between the two plant materials (Supplemental Figure 4). RNA splicing is required for the complete maturation of chloroplast transcripts, and intron-splicing efficiency is closely related to the abundance of mature mRNA. We observed that the transcripts of rpoC1 and ycf3 contained one and two cis-introns, respectively (Supplemental Figure 5). To confirm that the absence of mature transcripts was due to defects in intron splicing, we designed intron-spanning primers to detect the splicing of the two transcripts. RT–PCR analysis of intron splicing showed that the cis-spliced fragments of rpoC1 and ycf3-2 were attenuated, and signals from the unspliced RNA precursors of rpoC1 and ycf3-2 were more substantial in GhCTSF1-silenced plants than in CLCrV-A plants (Figure 3D). By contrast, ycf3 intron 1 was correctly and efficiently spliced in GhCTSF1-silenced plants (Supplemental Figure 6). We also assayed the splicing events of other plastid genes with group II introns via RT–PCR, and the results showed no difference between GhCTSF1-RNAi and CLCrV-A plants (Supplemental Figure 6A). We next analyzed splicing efficiency using RT–qPCR (Supplemental Figure 6B). A significant decrease in splicing efficiency was observed for ycf3 intron 2 and rpoC1 intron 1 in GhCTSF1-silenced plants (Figure 3E). These results suggested that GhCTSF1 participates in the cis-splicing of plastid rpoC1 and ycf3-2 transcripts. We detected the protein levels of RpoC1 and Ycf3 and found that RpoC1 and Ycf3 failed to accumulate in GhCTSF1-silenced plants (Supplemental Figure 7A). The plastid rpoC1 gene encodes a subunit of plastid-encoded RNA polymerase (PEP) that is essential for chloroplast development, as it regulates PEP-dependent plastid gene transcription (Hernandez-Verdeja and Strand, 2018). We therefore analyzed the transcript levels of PEP-dependent plastid genes and found that most PEP-dependent transcripts were downregulated in GhCTSF1-silenced plants (Figure 3F). Ycf3 encodes an essential assembly factor for PSI, and we therefore investigated the abundance of photosynthetic complexes by blue native polyacrylamide gel electrophoresis (BN–PAGE). We found that the accumulation of PSII and PSI complexes was significantly reduced in all three GhCTSF1-silenced plants compared with that in CLCrV-A plants (Supplemental Figure 7B). Immunoblot analyses of representative proteins of the thylakoid protein complexes showed that PsaA, PsaB, D1, and D2 failed to accumulate in GhCTSF1-silenced plants, whereas ATPase complex subunits were still present in CLCrV-A plants (Supplemental Figure 7C).

Figure 3.

Silencing of GhCTSF1 expression results in impaired intron splicing of ycf3-2 and rpoC1.

(A) Log2 average fold change of read counts across the complete plastid genome.

(B) Chloroplast transcript levels in GhCTSF1-silenced plants assessed by organelle RNA-seq. The values are presented as log2 ratios of chloroplast gene expression in GhCTSF1-silenced plants (GhCTSF1-RNAi) relative to that in empty-vector control plants (CLCrV-A). Three independent biological replicates were analyzed. Error bars represent means ± SD.

(C) RT–qPCR was performed to analyze transcript levels of ycf3 and rpoC1. Ubiquitin was used as an internal control. Representative results from three biological replicates are shown. Error bars indicate mean ± SD. Asterisks indicate statistically significant differences by Student’s t-test: ∗∗∗P < 0.001.

(D) RT–PCR analysis of the intron splicing of ycf3-2 and rpoC1 in GhCTSF1-RNAi and CLCrV-A plants. U, unspliced transcripts; S, spliced transcripts. All PCR products were confirmed by sequencing.

(E) RT–qPCR analysis of splicing efficiency of ycf3-2 and rpoC1. Primers spanned adjacent exons and introns to measure differences in splicing efficiency. U, unspliced transcripts; S, spliced transcripts. Error bars indicate mean ± SD.

(F) Expression levels of PEP-dependent chloroplast genes. The values are presented as log2 ratios of gene expression in GhCTSF1-silenced plants (GhCTSF1-RNAi) relative to that in empty-vector control plants (CLCrV-A). Data are shown as means ± SD.

The GhCTSF1 homolog EMB1417 also participates in intron splicing of rpoC1 and ycf3 in Arabidopsis

Phylogenetic analysis of the amino acid sequences revealed that GhCTSF1 is a homolog of Arabidopsis embryo defective 1417 (EMB1417) (Figure 4A). EMB1417 encodes a PPR protein of 307 amino acids that was predicted to be a P-type PPR protein with two PPR motifs (Supplemental Figure 8A). RT–qPCR analysis showed that EMB1417 was highly expressed in true leaves, cotyledons, and stems (Supplemental Figure 8B). We determined the subcellular localization of EMB1417 and found that it was localized in the chloroplast (Supplemental Figure 8C). To investigate the role of EMB1417, we obtained the knockout (KO) line SALK_007827, which had a T-DNA insertion in the 5′ UTR. However, we found no noticeable phenotypic differences between the SALK_007827 line and the wild type (WT) (Supplemental Figure 9). We generated null mutants of EMB1417 (EMB1417-KO1, EMB1417-KO2, and EMB1417-KO3) using CRISPR–Cas9-based genome editing (Figure 4B). Analysis of the genomic DNA sequence revealed that EMB1417-KO1 and EMB1417-KO2 contained a 1-base insertion and a 2-base deletion in EMB1417, respectively. Both mutations resulted in a frameshift. EMB1417-KO3 had a 582-nt deletion that removed 194 amino acids (Figure 4C; Supplemental Figure 10A). In the T2 generation, the homozygous EMB1417-KO3 mutant produced by CRISPR–Cas9 displayed a yellow-variegated leaf color with significantly reduced chlorophyll content (Figures 4D and 4E). EMB1417 was predicted to be an EMB gene, and we therefore examined embryonic development in the EMB1417-KO3 mutant. However, no noticeable differences in seed phenotype were observed between WT plants and the EMB1417-KO3 mutant under normal growth conditions (Supplemental Figure 10B).

Figure 4.

Mutation of EMB1417 results in disrupted chloroplast development and impaired RNA splicing of ycf3-2 and rpoC1.

(A) A. thaliana EMB1417 is a homolog of GhCTSF1 in G. hirsutum. The phylogenetic tree was constructed with MEGA v.5.2 (https://www.megasoftware.net). Numbers are percentage bootstrap values for 1000 replicates.

(B) The cotyledon phenotypes of three EMB1417 mutant lines. Pictures were scored using 5-day-old seedlings grown on half-strength MS plates at 22°C. Scale bar, 1 mm.

(C) Sequencing results for the three EMB1417 mutant lines.

(D) Phenotype of the EMB1417-KO3 mutant. Pictures were scored at 10, 20, and 40 days. Scale bar, 2.5 mm. CD, cotyledon; FL, first leaf; TL, fourth leaf.

(E) Chlorophyll a and b contents of EMB1417-KO3 plants. Error bars indicate mean ± SD (n = 3), and asterisks indicate statistically significant differences by Student’s t-test: ∗∗∗P < 0.001.

(F) RT–PCR analysis of the intron splicing of ycf3-2 and rpoC1 in EMB1417-KO3 plants. U, unspliced transcripts; S, spliced transcripts. clpP-1 was used as a control. All PCR products were confirmed by sequencing.

(G) Analysis of the splicing efficiency of ycf3-2 and rpoC1 in the EMB1417-KO3 mutant assessed by RT–qPCR. Error bars represent ± SD of three biological replicates.

(H) Expression levels of PEP-dependent chloroplast genes. The values are presented as log2 ratios of gene expression in EMB1417-KO3 plants relative to that in wild-type (WT) plants. Data are shown as means ± SD.

(I–K) The quantum yield of PSI (I), electron transport rate of PSI (J), and non-photochemical quenching (K) of EMB1417-KO3 plants.

Data in (F)–(H) are shown as means ± SD of three individual replicates.

As EMB1417 is a homolog of GhCTSF1, we investigated whether EMB1417 functions in plastid RNA splicing. As shown in Figure 4F, unspliced ycf3-2 and rpoC1 transcripts were more abundant in the EMB1417-KO3 mutant than in the WT, whereas levels of spliced ycf3-2 and rpoC1 transcripts were dramatically reduced. RT–qPCR confirmed that the intron-splicing efficiencies of ycf3-2 and rpoC1 were dramatically reduced in the EMB1417-KO3 mutant (Figure 4G). However, intron splicing of other plastid mRNAs (e.g., clpP-1) was not significantly affected in the mutant (Figures 4F and 4G; Supplemental Figures 11A and 11B). Plastid RNA editing was examined using strand-specific RNA-seq, and editing events were not significantly changed in EMB1417-KO3 plants (Supplemental Table 3). We also investigated the transcript levels of all plastid chloroplast genes and found that most PEP-dependent genes were downregulated in the EMB1417-KO3 mutant (Supplemental Table 4). The decreased expression of PEP-dependent plastid genes was confirmed by RT–qPCR (Figure 4H). TEM observations revealed destroyed chloroplasts in a section of the yellow-variegated area of the mutant leaf, whereas well-developed chloroplasts were observed in the leaves of WT plants (Supplemental Figure 10C). We next analyzed photosynthetic parameters and found that the quantum yield of PSI, the electron transport rate of PSI, and non-photochemical quenching were significantly lower in EMB1417-KO3 than in the WT (Figures 4I–4K). These results indicate that EMB1417, a homolog of GhCTSF1, influences chloroplast development and photosynthesis, likely by affecting the splicing of plastid rpoC1 and ycf3-2 transcripts in Arabidopsis.

GhCTSF1 and GhCRS2 or GhWTF1 proteins physically interact

Several factors, including the CAF, CRS, APO, CRM, PPR, and PORR proteins, can form a splicing complex that participates in group II intron splicing in chloroplasts (Asakura et al., 2008; Majeran et al., 2012; Jin et al., 2016). Our results indicated that splicing of rpoC1 and ycf3-2 was impaired in GhCTSF1-silenced plants and emb1417 mutant plants (Figures 3D and 4G). We hypothesized that GhCTSF1 interacts with these splicing factors to promote the splicing of introns from pre-mRNAs in chloroplasts. We explored potential interactions between GhCTSF1 and these splicing factors using yeast two-hybrid assays. The results showed that GhCTSF1 interacted with GhCAF2, GhCRS1, GhCRS2, GhCFM2, GhWTF1, GhOTP51, and GhCTSF1 (Figure 5A). To confirm these interactions, we performed bimolecular fluorescence complementation (BiFC) assays using 35S-promoter-driven expression of the yellow fluorescent protein (YFP) C terminus fused to GhCTSF1 (GhCTSF1–cYFP). This was co-expressed in tobacco leaves with N-terminal fusions of GhCAF2, GhCRS1, GhCRS2, GhCFM2, GhWTF1, GhOTP51, and GhCTSF1. Strong YFP fluorescence signals were observed in chloroplasts when GhCTSF1–cYFP was co-expressed with GhCRS2, GhWTF1, or GhCTSF1 (Figure 5B; Supplemental Figure 12). We tested these interactions using a co-immunoprecipitation (Co-IP) assay. GhCTSF1 fused with a Myc tag (GhCTSF1–Myc) and GhCRS2 or GhWTF1 fused with HA were transiently co-expressed in tobacco leaves. GhCTSF1–Myc was co-immunoprecipitated when GhCRS2 or GhWTF1 was pulled down from leaf extracts using an anti-HA antibody (Figure 5C). The interaction of GhCTSF1 with GhCRS2 or GhWTF1 was further confirmed by in vitro pull-down assays (Figure 5D). These results suggest that GhCTSF1 physically interacts with GhCRS2 and GhWTF1 both in vitro and in vivo. Because EMB1417 is a homolog of GhCTSF1, we explored whether EMB1417 and Arabidopsis CRS2 or WTF1 could also form a complex. Yeast two-hybrid assays showed that EMB1417 could interact with CRS2, WTF1, or itself in Arabidopsis (Supplemental Figure 13). These results suggested that GhCTSF1 and EMB1417 are structurally and functionally conserved in cotton and Arabidopsis.

Figure 5.

GhCTSF1 interacts with GhCRS2 and GhWTF1.

(A) Yeast two-hybrid assay showing that GhCTSF1 interacts with GhCAF2, GhCRS1, GhCRS2, GhCFM2, GhWTF1, GhOTP51, and GhCTSF1 itself. Full-length GhCTSF1 was fused with the DNA-binding domain in pGBKT7, and the potential interactors were separately fused with the activation domain (AD) in pGADT7. Transformed yeast was grown on SD/-Trp-Leu (SD-TL) and SD/-Trp-Leu-His-Ade (SD-TLHA) dropout plates to test protein interactions. The empty pGADT7 vector was co-transformed with GhCTSF1 as a negative control.

(B) Bimolecular fluorescence complementation assay showing that GhCTSF1 interacts with GhWTF1, GhCRS2, and GhOTP51 in N. benthamiana leaf epidermal cells. GhCTSF1 was fused to the C-terminal fragment of YFP (cYFP), and GhWTF1, GhCRS2, or GhOTP51 was fused to the N-terminal fragment of YFP (nYFP). Chlorophyll autofluorescence (chlorophyll) was used to reveal the chloroplasts. Bright, bright-field image under transmitted light; Merge, merged image of YFP, chlorophyll, and Bright. Excitation/emission wavelengths were 510/530 nm for YFP and 488/670 nm for chlorophyll autofluorescence. Scale bar, 50 μm.

(C) Co-IP analysis shows the interaction between GhCTSF1 and GhWTF1/GhCRS2 using a transient expression system in tobacco leaves. IP was performed using an anti-Myc matrix, and the co-immunoprecipitated proteins were detected with anti-HA antibodies. This assay was repeated three times with similar results.

(D) MBP-pull-down assay demonstrating the interaction between GhCTSF1 proteins and GhWTF1/GhCRS2.

CRS2 and WTF1 are required for rpoC1 and ycf3-2 splicing

WTF1 is a PORR-domain-containing protein, and CRS2 contains a PTH domain (Supplemental Figures 14 and 15). Both factors are involved in RNA splicing of chloroplast-encoded introns (Kroeger et al., 2009), and we next sought to determine the role of GhCRS2 and GhWTF1 in RNA splicing of chloroplast transcripts in cotton. The two genes were individually silenced in cotton seedlings using VIGS, and GhCRS2-RNAi and GhWTF1-RNAi plants displayed a yellow-variegated leaf phenotype and decreased chlorophyll content (Supplemental Figures 16 and 17). We also performed CRISPR–Cas9-mediated KO of GhCRS2 and GhWTF1 in the cotton cultivar Jin668. After pedigree selection assisted by kanamycin and PCR amplification, 14 and 9 independent T₀ transgenic lines were generated for GhWTF1-KO and GhCRS2-KO, respectively. GhWTF1-KO2 and GhCRS2-KO5 lines were selected for detailed analysis. After self-pollination of the T₀ parents, GhWTF1-KO and GhCRS2-KO lines were obtained, and the seedlings were evaluated by PCR and Sanger sequencing. The results showed that GhWTF1-KO2 and GhCRS2-KO5 seedlings contained a 1-base insertion mutation and an 11-base insertion at their target sites, respectively (Figures 6A and 6B; Supplemental Figures 18A and 18B). GhWTF1-KO2 and GhCRS2-KO5 plants displayed a yellow-variegated phenotype in the cotyledons and leaves and lower total chlorophyll content (Figures 6C–6E). The developmental status of chloroplasts in the GhWTF1-KO2 and GhCRS2-KO5 mutants was examined using transmission electron microscopy. In WT plants, the chloroplasts were crescent shaped and contained well-developed thylakoid membranes with large grana stacks. However, the albino leaves of GhWTF1-KO2 and GhCRS2-KO5 plants contained numerous abnormal chloroplasts with fewer thylakoid membranes (Figure 6F). We observed a significant decrease in the quantum yield of PSI, the electron transport rate, and non-photochemical quenching in GhWTF1-KO2 and GhCRS2-KO5 plants (Figures 6G–6I). These results indicated that GhCRS2 and GhWTF1 play important roles in chloroplast development and photosynthesis in cotton.

Figure 6.

Loss of GhWTF1 and GhCRS2 will greatly delay chloroplast development and impair photosynthesis in cotton.

(A and B) KO of GhWTF1 and GhCRS2 was verified by gold-standard Sanger sequencing.

(C and D) Cotyledon and leaf phenotypes of GhWTF1 and GhCRS2 KO plants. Pictures of cotyledons and leaves were scored at 10 and 60 days, respectively. Scale bars, 2.5 cm in (C) and 5 cm in (D).

(E) Chlorophyll a and b contents of leaves from GhWTF1 and GhCRS2 KO plants. Error bars indicate mean ± SD (n = 3), and asterisks indicate statistically significant differences by Student’s t-test: ∗∗∗P < 0.001.

(F) Chloroplast ultrastructure in true leaves from 60-day-old plants. Chloroplasts of GhWTF1 and GhCRS2 KO plants were obtained from the albino sector of the leaf. SG, starch grain; Thy, thylakoid; Gr, granum; PG, plastoglobule.

(G–I) The quantum yield of PSI (G), electron transport rate of PSI (H), and non-photochemical quenching (I) of GhWTF1 and GhCRS2 KO plants. Data in (G)–(I) are shown as means ± SD of three individual replicates.

(J) RT–qPCR analysis of splicing efficiency of ycf3-2 and rpoC1 in GhWTF1 and GhCRS2 KO plants. The values are presented as log2 ratios of gene expression in GhWTF1 and GhCRS2 KO plants relative to that in WT plants. Data are shown as means ± SD.

(K) Expression levels of PEP-dependent chloroplast genes.

RNA splicing of rpoC1 and ycf3-2 was also assessed in GhWTF1-KO2 and GhCRS2-KO5 plants. The results showed that mature transcripts of rpoC1 and ycf3-2 were dramatically reduced in GhWTF1-KO2 and GhCRS2-KO5 plants, as was the splicing efficiency of rpoC1 and ycf3-2 (Figure 6J; Supplemental Figure 18C). We also evaluated plastid RNA editing by RNA-seq and found that editing events were not significantly affected in GhWTF1-KO2 and GhCRS2-KO5 plants (Supplemental Table 5). In addition, we investigated the transcript levels of PEP-dependent plastid genes and found that they were decreased in GhWTF1-KO2 and GhCRS2-KO5 plants (Figure 6K; Supplemental Tables 6 and 7). These results indicated that GhWTF1 and GhCRS2 participate in cis-splicing of plastid rpoC1 and ycf3-2 transcripts in cotton.

GhWTF1 binds to rpoC1 and ycf3 transcripts

Because GhCTSF1 interacts with GhWTF1 and GhCRS2, all three participate in RNA splicing of rpoC1 and ycf3-2, and GhWTF1 contains an organellar RNA recognition (PORR) domain, we investigated whether GhWTF1 binds to the pre-mRNAs of rpoC1 and ycf3-2. RNA IP coupled with qPCR (RIP–qPCR) was used to assess potential interactions between GhWTF1-rpoC1 and GhWTF1-ycf3. GhWTF1 protein with an HA tag (GhWTF1–HA) was transiently expressed in cotton cotyledons by agroinfiltration. After 72 h of incubation, high levels of GhWTF1–HA expression were detected in the leaves (Figure 7A). RIP–qPCR demonstrated that the negative control, clpP, was barely detectable in precipitates from GFP-HA plants, confirming the low background noise in the control samples. As expected, clpP mRNA was not enriched after IP in WTF1–HA plants, indicating that GhWTF1 could not bind to pre-mRNAs of clpP (Figure 7B).

Figure 7.

GhWTF1 associates with ycf3 and rpoC1 transcripts.

(A) Immunoblots showing the WTF1–HA protein in cotton cotyledons.

(B–D) RIP–qPCR assay to analyze the region of WTF1 binding to clpP, rpoC1, and ycf3. Top, diagram depicting the structures of clpP, ycf3, and rpoC1 transcripts. The locations of regions used for RIP–qPCR analysis are indicated by black lines. Bottom, relative enrichment of each region in clpP, rpoC1, and ycf3. GFP-HA was used as the control. Data are means ± SD, n = 3; ∗∗∗P < 0.001.

(E) Gel-mobility shift assays (EMSAs) demonstrating the binding activity of recombinant GhWTF1 to rpoC1 and ycf3-2.

(F) A proposed working model for the involvement of CTSF1-mediated RNA splicing in chloroplast development and photosynthesis. The nuclear-encoded PPR protein CTSF1 is imported into plastids and interacts with CRS2 and WTF1 to form part of the plastid RNA splicing complex. The splicing complex then associates with the pre-mRNA and promotes the intron splicing of rpoC1 and ycf3-2. As a result, the spliced plastid rpoC1 transcripts maintain high activity of PEP, which plays an essential role in chloroplast development by activating expression of photosynthesis-associated plastid-encoded genes (PhAPGs). Meanwhile, the spliced ycf3 transcript promotes assembly of the PSI complex, which in turn functions in photosynthesis.

To better understand the mechanism by which GhWTF1 binds to mRNAs, we designed five and six pairs of primers for different rpoC1 and ycf3 regions, covering all introns and exons (Figures 7C and 7D). As shown in Figure 7C, significant GhWTF1 enrichment was detected in the P2 and P3 regions of rpoC1, with the highest level detected at P3. No significant enrichment of GhWTF1 was observed in the P1, P4, or P5 region of rpoC1. P1 and P5 are located in the first and second exons of the rpoC1 transcript, respectively, and we speculated that WTF1 binds to the intron region of rpoC1. Similarly, the P3 region of ycf3 was significantly enriched in the GhWTF1–HA group, whereas the P1, P2, P4, P5, and P6 regions were not, indicating that GhWTF1 could bind to intron 2 of ycf3 (Figure 7D). We verified the binding between the GhWTF1 protein and pre-mRNAs of rpoC1 and ycf3-2 using RNA electrophoresis mobility shift assays (RNA-EMSAs). The result showed that GhWTF1 bound to rpoC1 and ycf3-2 probes, whereas MBP did not (Figure 7E). We next performed RIP–qPCR using GhCTSF1 and found that CTSF1 was also associated with the rpoC1 and ycf3-2 introns (Supplemental Figure 19A). However, no binding affinity of GhCTSF1 to rpoC1 and ycf3-2 introns was observed in RNA–EMSAs (Supplemental Figure 19B). These results imply that GhWTF1 can directly bind to rpoC1 and ycf3 mRNAs, contributing to the regulation of rpoC1 and ycf3 mRNA splicing.

Discussion

GhCTSF1 participates in RNA splicing of rpoC1 and ycf3-2, which is essential for plastid development and photosynthesis in cotton

Chloroplasts are the sites of photosynthesis, and their development requires tight coordination between gene expression systems in the nucleus and organelles (Yoo et al., 2019; Richter et al., 2023; Wang et al., 2023). Plastid-encoded genes are regulated mainly by post-transcriptional processes such as RNA processing, stability, editing, and intron splicing, most of which depend on PPR proteins (Barkan and Small, 2014). In this study, we defined the function of a novel short P-type PPR protein, GhCTSF1, which is required for the RNA splicing of rpoC1 and ycf3-2 transcripts in cotton chloroplasts. Silencing of GhCTSF1 expression decreased the splicing efficiency of ycf3 intron 2 and rpoC1 intron 1, leading to impairments of chloroplast structure and photosynthetic rate, both of which are crucial for the growth and development of cotton leaves.

Ycf3 is an essential assembly factor for PSI complex biogenesis and is encoded by the chloroplast ycf3 gene in the chloroplast genomes of algae and higher plants. Studies have shown that Ycf3 and pre-existing Y3IP1 form the Ycf3–Y3IP1 module, which assists in the initial assembly of PsaA and PsaB into the PSI RC subcomplex. This subcomplex, combined with Ycf4, forms the main component of the assembly apparatus that mediates assembly of the PSI and light-harvesting complex I (LHCI) subunits into a mature PSI–LHCI supercomplex (Nellaepalli et al., 2018). In Chlamydomonas reinhardtii and Nicotiana tabacum, KO of ycf3 completely blocked accumulation of the PSI complex, and the mutants could not grow photoautotrophically (Albus et al., 2010). Similarly, a striking decrease in the transcript level of ycf3 in Arabidopsis otp51 and pbf2 mutants blocked the assembly of the PSI complex, which in turn led to a lack of PSI activity (de Longevialle et al., 2008; Wang et al., 2020). In this study, we observed that RNA splicing of ycf3-2 was abolished, and abundance of the mature ycf3 transcript was dramatically reduced in GhCTSF1-silenced plants and emb1417 mutants. Immunoblot analyses and BN–PAGE showed that accumulation of PSII and PSI complexes was significantly reduced. The quantum yield of PSI, the electron transport rate of PSI, and non-photochemical quenching were significantly lower in GhCTSF1-silenced plants and emb1417 mutants, suggesting that these plants were unable to accumulate the PSI complex, resulting in lower photosynthetic activity.

Plastid transcription in plants depends on the nucleus-encoded RNA polymerase (NEP) and PEP. According to RNA-polymerase dependence, the transcription of chloroplast genes can be divided into three classes. Class I is PEP dependent, class II is both NEP and PEP dependent, and class III is NEP dependent. PEP drives high-level transcription of photosynthesis-related genes, most tRNA genes, and rRNA synthesis. Here, we also found that levels of many other mRNAs were reduced in GhCTSF1-silenced plants and emb1417 mutants. A possible explanation may be that the plastid rpoC1 encodes a core subunit of PEP and that deficiency in RNA splicing of rpoC1 causes a reduction in functional mature rpoC1 transcripts and a low level of PEP activity, resulting in a substantial decrease in PEP-dependent transcripts (e.g., rbcL, petB, petD, psaC, and psbA). PEP plays a pivotal role in chloroplast development by governing the transcription of chloroplast genes. Mutants lacking any PEP subunit will exhibit chloroplast dysfunction and abnormal development, with albino or yellowish phenotypes (Yang et al., 2019). We observed that GhCTSF1-silenced plants and the EMB1417-KO3 mutant both displayed a yellow-variegated leaf phenotype. Similarly, disruption of Arabidopsis OTP70 leads to a substantial defect in splicing of the plastid transcript rpoC1 and a reduction in PEP activity. The otp70 seedlings show a marked growth delay and exhibit short, rounded, and distinctly paler leaves (Chateigner-Boutin et al., 2011). In rice, KO of OsPPR16 leads to impaired accumulation of RpoB and reduces the expression of PEP-dependent genes, and the osppr16 mutant exhibited abnormal chloroplast structure and pale leaves (Huang et al., 2020).

In addition to ycf3-2 and rpoC1, ycf3-1, rpl16, rps12-2, and petD also showed slightly lower splicing efficiency in the EMB1417-KO3 mutant. This may have resulted from the suppression of matK, which encodes a maturase required for RNA splicing. The expression of matK is also dependent on PEP activity (Small et al., 2023). In GhCTSF1-silenced plants, RNA editing at psbJ-59, ndhD-2, ndhB-149, ndhB-746, and psaI-83 was altered, implying an additional indirect effect of GhCTSF1 on RNA editing. In addition, we cannot exclude the possibility that CTSF1 may have other functions in the chloroplast, as the protein co-localizes with WTF1 not only in the stroma but also in the thylakoids.

GhCTSF1 interacts with WTF1 and CRS2 to facilitate RNA splicing in cotton

In land plants, chloroplast genomes have 17–20 group II introns and only one group I intron in trnL (UAA). Both rpoC1 and ycf3-2 belong to group IIB of the group II introns (Wang et al., 2022). Group II intron splicing requires additional proteins from terrestrial plants. GhCTSF1 is a novel, minor P-class PPR protein and participates in the process of chloroplast RNA splicing, as demonstrated by analysis of RNA processing events. In vitro and in vivo assays demonstrated that GhCTSF1 physically interacts with the nucleus-encoded proteins GhWTF1 and GhCRS2. RIP–qPCR and RNA–EMSA showed that GhWTF1 binds to rpoC1 and ycf3-2 introns. Thus, GhCTSF1 may assist WTF1 in facilitating the splicing of ycf3 intron 2 and rpoC1.

Ycf3 intron 2 is exceptional among all subgroup IIB introns requiring CRS2 and CAFs. A previous study showed that ycf3 intron 2 splices independent of CRS2 were also independent of CAF1 and CAF2 (Ostheimer et al., 2003). However, this intron requires several other splicing factors to assist in splicing. In maize, slot-blot hybridization and RIP–chromatin IP assays have demonstrated that THA8 is responsible for splicing of ycf3 intron 2 and the tRNA intron, and its molecular functions are conserved in Arabidopsis (Khrouchtchova et al., 2012). THA8 may also cooperate with OTP51 and APO1 to facilitate splicing of ycf3 intron 2. EMSAs have shown that maize recombinant APO1 binds 40–240 nt in the intron domain of ycf3 intron 2 with high affinity, similar to results in Arabidopsis (Watkins et al., 2011). WTF1, another splicing factor identified in CAF1 and CAF2, co-immunoprecipitates through mass spectrometry analysis (Kroeger et al., 2009). WTF1 and RNC1 have been shown to form a heterodimer to promote the splicing of most chloroplast group II introns. In addition, a co-precipitation assay showed that THA8 is associated with WTF1 and RNC1 (Khrouchtchova et al., 2012). It is likely that THA8 interacts with the WTF1/RNC1 heterodimer to facilitate splicing of their shared intron target, ycf3 intron 2. Thus, WTF1/RNC1, OTP51, and APO1 may be recruited by THA8 to participate in ycf3 intron 2 splicing via protein–protein interactions. Considering that GhCTSF1 interacts with WTF1, we speculated that GhCTSF1 might be involved in ycf3 intron 2 splicing using two models. First, GhCTSF1 may act as an important component by sharing common introns with the splicing complexes to promote ycf3 intron 2 splicing. Second, GhCTSF1 may cooperate with the WTF1 protein and directly bind to other domains of ycf3 intron 2 to facilitate splicing. Evidence showed that WTF1 binds to the 535–585 nt region of ycf3 intron 2 and APO1 binds to the 40–240 nt region of ycf3 intron 2. We suppose that a second splicing model is more likely to exist. APO1 and WTF1 may simultaneously bind to different positions of ycf3 intron 2 to facilitate splicing. Although intron 2 of ycf3 did not co-immunoprecipitate with WTF1 in maize, we speculate that WTF1 may require CTSF1 to increase its affinity for the target sequence of ycf3 intron 2. However, this hypothesis needs to be tested in future research.

Our study shows that GhCTSF1 interacts with GhCRS2 in cotton. CRS2 is a general splicing factor involved in splicing several group II introns. CRS2 often associates with CAF1 and CAF2 to assist in RNA splicing by forming CRS2–CAF1 or CRS2–CAF2 complexes. The CRS2–CAF1 complex is necessary for splicing petD, trnG, rps16, rpl16, ndhA, and ycf3 intron 1 in maize. CAF1 is also involved in intron splicing of rpoC1 and clpP intron 1 in Arabidopsis. The CRS2–CAF2 complex is required for the splicing of five group IIB introns: ndhB, petB, ndhA, rps12 intron 1, and ycf3 intron 1 (Ostheimer et al., 2003; Asakura and Barkan, 2006). CRS2–CAF1/2 complexes require additional proteins to target different introns (Zhang et al., 2023b). For example, the P-class PPR protein PBF2 is required for splicing of ycf3 intron 1; it interacts with CAF1 and CAF2 and may contribute to the specificity of CRS2–CAF1 and CRS2–CAF2 complexes toward ycf3 intron 1 (Wang et al., 2020). The P-class OsPPR11 may contribute to forming specific splicing complexes that regulate ndhA and ycf3-1 intron splicing through interactions with the CRM family protein OsCAF2 (Zhang et al., 2023a). The PPR protein EMB1270 is required for the splicing of clpP1 intron 2, ycf3 intron 1, ndhA, and ndhB through its specific interaction with the splicing factor CFM2 (Zhang et al., 2021). GhCTSF1 is a P-class PPR protein that interacts with the general splicing factors CRS2 and WTF1. In recent years, some small PPR proteins (SPRs) have been discovered that may mediate RNA splicing by specific intermolecular interactions in vivo. The PPR protein THA8, which has only four PPR motifs, is closely related to RNA splicing in chloroplasts through its interaction with WTF1 and RNC1 (Khrouchtchova et al., 2012). SPR2, which contains only four PPR repeats, is required for intron splicing by interacting with other splicing factors in maize mitochondria (Cao et al., 2022). GhCTSF1 may have a similar mode of action, forming splicing complexes through interactions with specific factors to promote the splicing of rpoC1 and ycf3 intron 2. Of course, we cannot rule out the possibility that other splicing factors or PPR proteins interact with GhCTSF1 to facilitate splicing of different introns.

GhCTSF1 is functionally conserved as a splicing factor in higher land plants

Cotton is an ancient dicotyledonous plant with nearly 1059 PPR proteins, making the PPRs one of the largest protein families in land plants. Some researchers have found that PPR families are surprisingly similar by analyzing monocot and dicot genome and transcriptome sequences (O'Toole et al., 2008). Much of the expansion and diversification of the PPR family occurred between the origin of the first terrestrial plants (∼475 mya) and the first appearance of modern angiosperms (∼140 mya) (Gutmann et al., 2020). The ancestors of GhCTSF1 and EMB1417 may have arisen after the separation of the land plant lineage from green algae and were maintained and stabilized in seed plants, making it likely that their roles are conserved across land plants.

Chloroplasts predominantly contain group II introns that diverged from the canonical intron structure during evolution and whose splicing mechanism is highly conserved (Hausner et al., 2006). The P-class PPR proteins mainly play a role in RNA stabilization and intron splicing. More than 60 (mostly P-class) PPR proteins are involved in splicing introns in different organelles, and these factors are genetically selected (Small et al., 2023). The retained P-class PPR factors may be necessary for splicing. We found that a GhCTSF1 homolog is present in A. thaliana, G. hirsutum, G. max, Z. mays, O. sativa, C. taitungensis, G. biloba, S. moellendorffii, and P. patens. This implies that the short P-class PPR proteins may have conserved functions in intron splicing of rpoC1 and ycf3-2 in plants (Supplemental Figure 20).

Methods

Plant material and growth conditions

G. hirsutum (Xuzhou 142) seeds were planted in pots containing soil and grown in the greenhouse at 30°C under a 14-h light/10-h dark photoperiod. A. thaliana seeds were surface sterilized with 70% ethanol and 20% bleach and sown on Murashige and Skoog (MS) medium. Seedlings were then transferred to soil after 1 week and grown under long-day conditions (21°C, 16-h light/8-h dark cycle). For VIGS assays, 2-week-old cotton plants with two fully expanded cotyledons were used for Agrobacterium infiltration. After inoculation, plants were transferred into a growth chamber set to a 16-h light/8-h dark photoperiod at 23°C. Ten seedlings were used for each treatment, and three biological triplicates were performed per assay.

Chlorophyll content, chloroplast ultrastructure, and photosynthetic rate

Leaf chlorophyll content was measured as described previously (Borsuk and Brodersen, 2019). In brief, 0.2 g fresh leaf was immersed in 10 ml ethanol in the dark. After 48 h, the supernatants were analyzed by spectrophotometric scanning at 470, 649, and 665 nm. The microstructure of chloroplasts was observed using a TEM. Young leaves were fixed in 4% glutaraldehyde, dehydrated using different concentrations of ethanol, embedded and aggregated with ethoxyline resin, and cut into sections. After being stained with uranyl acetate and alkaline lead citrate for 15 min, the sample was observed using a Model H-7650 TEM (HITACHI). Net photosynthetic rates of flag leaves were measured with an LI-6400 portable photosynthesis system (LI-COR Biosciences, Lincoln, NE, USA). For each material, Chl content and net photosynthetic rate were measured using at least three biological replicates.

Subcellular localization of GhCTSF1 protein

To investigate the subcellular localization of GhCTSF1, the open reading frame of GhCTSF1 was cloned into the pTF486-35S-GFP expression vector to generate a GhCTSF1–GFP fusion protein. The fusion constructs and the empty vector were transiently transformed into Arabidopsis protoplasts as described previously (Ryu et al., 2019). The transformed protoplasts were incubated at 23°C for 18 h, and florescence signals were visualized using a confocal laser scanning microscope (TCS SP8). GFP was visualized with excitation at 488 nm and emission at 505–530 nm. Chlorophyll fluorescence was visualized with excitation at 488 nm and emission at 650–710 nm. Detailed methods were described in our previous study (He et al., 2018).

RNA-seq analysis

Total RNA was extracted using an RNA Isolation Kit (Invitrogen) and digested with RNase-Free DNase (Qiagen) according to the manufacturer’s instructions. The rRNA-depleted RNA was used to prepare RNA-seq libraries with the TruSeq Stranded mRNA Library Prep Kit (Illumina). The library was sequenced using the Illumina HiSeq 2500 platform. RNA-seq reads were mapped to the chloroplast genome of G. hirsutum using the SOAPaligner/SOAP2 program in strand-specific mode (Li et al., 2008). Chloroplast RNA editing and RNA splicing were analyzed as described previously (Sun et al., 2018).

RNA isolation, RT–PCR, and RT–qPCR analysis

Total RNA was extracted using an RNA isolation kit, digested with RNase-Free DNase (Invitrogen), and used to synthesize cDNA. RT–PCR was used to analyze the splicing of chloroplast introns. Chloroplast genes were amplified with specific primer pairs. For analysis of RNA splicing efficiency by RT–qPCR, specific primers located in intron-exon (as unspliced forms) and exon-exon (as spliced forms) junctions were designed for each gene. The primer pairs used for RNA editing/splicing analysis and RT–qPCR are listed in Supplemental Table 8.

BN–PAGE and immunoblotting

Chloroplasts were prepared and BN–PAGE was performed as described previously (He et al., 2019). The isolated thylakoid pellets were suspended in resuspension buffer (25 mM Bis–Tris–HCl [pH 7.0], 1% n-dodecyl b-D-maltoside, and 20% glycerol [w/v]) at 1.0 mg chlorophyll ml⁻¹. After incubation at 4°C for 5 min and centrifugation at 12 000 × g for 10 min, the supernatant was added to one-tenth volume of loading buffer (100 mM Bis–Tris–HCl [pH 7.0], 0.5 M 6-amino-n-caproic acid, 5% Serva blue G, and 30% [w/v] glycerol) and applied to 1-mm-thick 4%–13% acrylamide gradient gels. For immunoblot analysis, proteins were solubilized and separated on 12% SDS–polyacrylamide gels. After electrophoresis, the proteins were transferred to polyvinylidene difluoride membranes (Millipore) and detected using antibodies specific for the PSII subunits (D1, D2), PSI subunits (PsaA, PsaB), and ATP synthase β-subunit (AtpB, AtpF). All the primary antibodies against chloroplast-related proteins used in this study were obtained from Agrisera (https://www.agrisera.com) or PhytoAB (https://www.phytoab.com/).

Yeast two-hybrid assay

The full-length cDNA of GhCTSF1 was cloned into the prey vector pGADT7, and the full-length or truncated cDNAs of splicing factors were cloned into pGBKT7. The prey and bait plasmids were co-transformed into AH109 and plated on SD/-Leu-Trp medium for 3 days at 30°C. Interactions between bait and prey were further tested on SD/-Trp-His-Leu-Ade medium.

BiFC assay

For BiFC, the coding sequence of GhCTSF1 (without stop codon) was cloned into the pSP-cYFP vector to form a C-terminal, in-frame fusion with cYFP. The coding regions of splicing factors (without stop codons) were cloned into the pSP-nYFP vector to produce N-terminal, in-frame fusions with nYFP. The BiFC constructs were transferred into Agrobacterium tumefaciens strain GV3101 and transiently expressed in tobacco leaves. YFP fluorescence was detected with a Leica TCS SP8 confocal laser scanning microscope. The experiments were repeated three times. The empty pSP-nYFP vector was used as a negative control.

Co-IP assays

For Co-IP assays, 4-week-old N. benthamiana leaves were infiltrated with A. tumefaciens. After 72 h, the infected leaves were frozen in liquid nitrogen and homogenized with Co-IP buffer (50 mM Tris–HCl [pH 7.5], 150 mM NaCl, 0.1% Triton X-100, 0.2% Nonidet P-40, and 0.6 mM phenylmethylsulfonyl fluoride with protease inhibitor cocktail). After protein extraction, anti-MYC beads (Thermo Scientific) were added and incubated for 3 h at 4°C. The beads were washed three times with 25 column volumes of washing buffer (25 mM Tris–HCl, 500 mM NaCl, 1 mM EDTA, 10% glycerol). The proteins were eluted from the beads and detected using the corresponding antibody.

In vitro pull-down assay

For pull-down assays, recombinant proteins were expressed in Escherichia coli Rosetta (DE3). The coding sequence of GhCTSF1 was cloned into pGEX-6P-1, and those of GhWTF1 and GhCRS2 were cloned into pMAL-c5x. For in vitro binding assays, GhCTSF1–GST and GhWTF1–MBP or GhCRS2–MBP proteins were incubated with MBP Sepharose dextrin agarose resin at 4°C for 4 h. After incubation, the resin was washed three times and centrifuged to remove the supernatant. Retained proteins were released by adding loading buffer and boiled for 5 min, then resolved by SDS–PAGE and detected with anti-MBP and anti-GST antibodies.

Creation of the CRISPR–Cas9 KO mutant

CRISPR–Cas9-mediated gene editing of GhWTF1 and GhCRS2 was performed as described previously (Wang et al., 2018). In brief, the two target sites of each gene were selected according to the CRISPR-P web tool (http://cbi.hzau.edu.cn/crispr/). Each pair of single guide RNAs (sgRNAs; sgRNA1 and sgRNA2) was integrated in a single vector: fragments containing tRNA–sgRNA1 and tRNA–sgRNA2 fusions were obtained using pGTR as a template, and the two fragments were then fused by overlapping PCR. The assembled fragment was ligated into the pRGEB32-GhU6.9-NPT II expression vector, which was transformed into WT cotton (Jin668) using the Agrobacterium-mediated hypocotyl method as described previously (Tian et al., 2015).

To generate the Arabidopsis EMB1417 KO mutant, a 20-nt gRNA was designed using CRISPR-P (http://cbi.hzau.edu.cn/crispr/) to target exon 2 of EMB1417. The gRNA was synthesized as a pair of complementary single-stranded oligonucleotides with adaptors at the 5′ ends. The two oligonucleotides were annealed, phosphorylated by T4 Polynucleotide Kinase (NEB), and inserted into the pRGEB32 vector for plant transformation (Xie et al., 2015). The binary construct was introduced into A. tumefaciens strain GV3101 for transformation of WT plants through the floral dip method. For mutant characterization, T1 transgenic seeds were selected on half-strength MS medium supplemented with 40 mg/l hygromycin. The specific mutations in EMB1417 were examined by PCR amplification using gene-specific primers followed by sequencing.

RIP–qPCR analysis

RIP–qPCR was performed as described previously with minor modifications (Wang et al., 2021). In brief, RIP was performed using cotton cotyledons transiently expressing GhWTF1–HA protein. Fresh cotyledon samples were ground into fine powder and suspended. The homogenate was centrifuged for 30 min at 12 500 × g and 4°C, and 200 μl of cell lysate was saved as the input sample. The remaining sample was divided into two equal parts (IP sample and mock sample) and incubated with anti-HA antibody (CST, C29F4) or anti-immunoglobulin G1 antibody (Thermo Scientific, MA1-10406) together with protein A+G magnetic beads (Merck Millipore, 16-663) at 4°C overnight with rotation. After incubation, the bead–protein–RNA complexes were washed and treated with 30 μg proteinase K at 37°C for 30 min to release the RNP complexes, and RNA was then extracted with the TRIzol reagent. IP and input RNA samples were used for qPCR analysis.

RNA–EMSA

RNA–EMSA was carried out as described previously (Xiao et al., 2018). In brief, the corresponding cDNA fragments of GhWTF1 and GhCTSF1 were cloned into the pMAL-c5x vector to generate recombinant GhWTF1–MBP and GhCTSF1–MBP. Biotin-labeled RNA probes for rpoC1 and ycf3-2 were synthesized by GenScript (Nanjing, China). For RNA–EMSAs, the recombinant protein was incubated with an RNA probe in a 10-μl reaction mixture. The mixture was incubated at 25°C for 20 min, then separated by 4% native PAGE in 0.5× TBE buffer and transferred to a nylon membrane (Roche).

Funding

This work is supported by the National Natural Science Foundation of China (32170367 and 32000146), the Shaanxi Fundamental Science Research Project for Chemistry & Biology (22JHZ007), the Department of Science and Technology Innovation Team Project of Shaanxi Provincial (2024RS-CXTD-72), the Fundamental Research Funds for the Central Universities (2020TS053), and the Excellent Graduate Training Program of Shaanxi Normal University (LHRCCX23185).

Author contributions

J.Y. and P.H. designed the research and wrote the paper. Y.H., M.C., and M.T. performed most of the experiments and analyzed most of the data. M.Z., X.Y., Y.Z., and T.Z. performed subcellular localization and western blotting assays.

Published: March 5, 2024