*2.6. OsCAF1 through C-terminal Interacts with OsCRS2*

In maize, CRS2 could interact with CAF1 and form a CRS2–CAF1 complex in chloroplasts [22]. Therefore, we investigated whether OsCAF1 interacted with OsCRS2 and formed an OsCRS2–OsCAF1 complex in rice. We obtained rice *OsCRS2* (*Os01g0132800*) by homologous alignment with *ZmCRS2* and observed that *OsCRS2* was also expressed highly in green tissue (Supplementary Figure S1A). The results of the yeast two-hybrid (Y2H) experiment revealed that OsCAF1 interacted with OsCRS2 and formed an OsCRS2–OsCAF1 complex in rice (Figure 7A). To determine which OsCAF1 section was essential for interactions with OsCRS2, we first generated three truncated forms of OsCAF1, including OsCAF1-N, OsCAF1-M, which contains two CRM domains, and OsCAF1-C (Figure 7B). The results showed that only OsCAF1-C could interact with OsCRS2, while OsCAF1-N and OsCAF1-M failed to interact with OsCRS2 (Figure 7C). The result indicated that the C-terminal of OsCAF1 was essential for interactions with OsCRS2. To further explore the function of the of OsCAF1 C-terminal, we analyzed the alignment of the C-terminals of OsCAF1 residues 387–701 aa in *Oryza sativa* with different species, including *Z. mays*, *A. thaliana*, *Panicum hallii*, *Brachypodium distachyon*, *Sorghum bicolor*, *Setaria italica*, *Triticum aestivum*, *Elaeis guineensis*, and *Phoenix dactylifera*. The sequence alignment results showed that the OsCAF1 C-terminal residues of 585–598 aa and 685–701 aa were highly conserved (Supplementary Figure S2). To test the functional regions of the OsCAF1 in C-terminal, we divided the C-terminal containing OsCAF1 residues C1 (387–584 aa), C2 (585–598 aa), C3 (599–684 aa), and C4 (685–701 aa) (Figure 7B) for further study. The results of the Y2H experiment showed that OsCAF1-C1 could bind OsCRS2 (Figure 7C). The results indicated that the C1-terminal of OsCAF1 was required for interactions with OsCRS2 in rice; however, the two highly conserved regions of the C-terminal of OsCAF1 could not interact with OsCRS2.

**Figure 7.** C-terminal of OsCAF1 interacts with OsCRS2. (**A**) OsCAF1 interacts with OsCRS2. The full-length CDS sequences of OsCAF1 and OsCRS2 were cloned into PGBKT7 (BD) and PGADT7 (AD) vector, respectively, to generate BD-OsCAF1 and AD-OsCRS2 constructs. The successfully cloned two constructs were then co-transformed into yeast AH109 cells. AD+ BD-OsCAF1 and BD+AD-OsCRS2 constructs were used as control. (**B**) The protein structure of OsCAF1. Diagram of OsCAF1 protein structure and different structural domains of OsCAF1, OsCAF1-N (1-184), OsCAF1-M (185-386), OsCAF1-C (387-701), OsCAF1-C1 (387-584), OsCAF1-C2 (585-598), OsCAF1-C3 (599-684), and OsCAF1-C4 (685-701). (**C**) Yeast two-hybrid results of the interaction of OsCRS2 and OsCAF1. Truncated versions of OsCAF1 were cloning into BD vector and to generate BD-OsCAF1-N, BD-OsCAF1-M, BD-OsCAF1-C, BD-OsCAF1-C1, BD-OsCAF1-C2, BD-OsCAF1-C3 and BD-OsCAF1-C4 vectors. These vectors were co-transferred to AH109 yeast cells with AD-OsCRS2, respectively. Yeast cells were selected on synthetic dextrose medium lacking Leu and Trp (SD/-Trp-Leu) and the medium lacking Leu, Trp, His, and Ade (SD/-Trp-Leu-His-Ade) with different dilution series (1, 10<sup>−</sup>1, and 10<sup>−</sup>2).

The *oscaf1* mutant exhibited albinism and survived for only approximately three weeks. The chloroplast numbers in *oscaf1* mutants decreased, accompanied by abnormal chloroplast development (Supplementary Figure S3) and damaged thylakoid membranes compared with the WT, which led to a decrease in chlorophyll contents in *oscaf1* mutants. Similar phenotypes of CRM domain proteins have been reported in rice, such as *oscfm3* and *oscrs1* [19,23]. OsCAF1 might play an important role in chloroplast development. Among the altered chloroplast and photosynthesis-related gene expressions in *oscaf1* mutants, PEP-dependent genes, including *PsaA*, *PsbA*, and *RbcL*, and encoding photosynthesis-associated proteins significantly decreased in *oscaf1* mutants, although NEP-dependent genes, *RpoC1* and *RpoC2*, markedly increased in *oscaf1* mutants. In addition, the expression levels of *AtpB* and *AtpE*, transcribed by both NEP and PEP, significantly decreased in *oscaf1* mutants. Therefore, OsCAF1 might regulate chloroplast development by influencing PEP activities.

The ROS contents such as H2O2 increased in the three leaves stage leaves of *oscaf1* mutants, compared with the WT. In addition, the expression levels of ROS-scavenging related genes decreased in *oscaf1* mutant leaves. Considering the results of a previous study [19], the functional obstruction of OsCAF1 could lead to abnormal chloroplast development, ROS over-accumulation in leaves, and eventually the death of rice seedlings after the three leaves stage.

Numerous leaf-color associated mutants have been identified and cloned in different plants [25–27]. Abnormal splicing of chloroplast gene introns could cause chlorophyll deficiency, which then leads to visible losses in green color in leaves or death of plants [25–27]. In rice, *wsl4* exhibit bleached appearances before the four leaves stage, and after the five leaves stage, all leaves turned to a normal color similar to WT under normal growth environments. The bleached appearance phenomenon in *wsl4* was caused by defective intron splicing of chloroplast genes, including *atpF*, *rpl2*, *ndhA*, and *rps12* [26]. The results of such studies indicate that the abnormal splicing of the four introns does not cause

lethal phenotypes at seedling stages. In the present study, the splicing of *ndhB* and *ycf3* introns was also affected. The ndhB protein was an NADPH dehydrogenase and bound to the thylakoid membrane, influenced NDH activity, and regulated cyclic electron transport around photosystem I (PS I) and respiratory electron transport [28]. In tobacco, the *ndhB* mutant exhibited abnormal intron splicing of only chloroplast *ndhB*, and the *ndhB* mutant could grow normally under normal conditions [28]. In addition, both *crr2-1* and *crr2-2*, with impaired intron splicing of *ndhB* in *Arabidopsis*, could grow normally under normal conditions [29]. In the present study, NDH activity or electron transfer efficiency of PS I may be affected in *oscaf1* mutants; however, the impairing of the *ndhB* intron was not the key factor influencing seedling death after the three leaves stage. Another intron splicing impaired gene, *ycf3*, the chloroplast-encoded Ycf3 protein, was responsible for the stability of the PS I protein complex [16]. Previous studies have shown that numerous mutants, such as *otp51, osppr6,* and *tha8*, have albino seedling lethal phenotypes similar to the one in the *oscaf1* mutant, and all the mutants have similar genetic defects including reduced intron splicing efficiency of *ycf3,* similar to the results in the present study [14,16,27]. Therefore, the Ycf3 protein could be required for early chloroplast development. In *oscaf1* mutants, the reduced splicing efficiency of *ycf3* was responsible for mutant deaths after the three leaves stages.

In *Arabidopsis*, AtCAF1 is required for the splicing of group IIB introns, including *ndhA*, *petD*, *rpl16*, *rps16*, *trnG*, *ycf3*, *rpoC1*, and *ClpP* [21]. ZmCAF1 in maize is also required for the splicing of group IIB introns, almost similar to *A. thaliana*, excluding *rpoC1* and *ClpP* [11]. Previous results have shown that CAF1 is not required for group IIA intron splicing in maize and *A. thaliana* [11,21]. Notably, our results indicated that the efficiency of intron splicing of group IIA introns, including *atpF, rpl2*, and *rps12*, and group IIB introns, including *ndhB*, *ycf3*, and *ndhA*, were affected. Particularly, the group IIA introns were not spliced in *oscaf1* mutants completely. The results indicated that OsCAF1 could play a key role in the regulation of intron splicing in group IIA and group IIB introns in rice, which was different from maize and *A. thaliana*. Previous study showed that OsCRS1 not only participated in the splicing of *atpF* introns but also promoted the splicing of other introns, including *trnL, rpl2, ndhA, ndhB, petD*, and *ycf3* [22]. Therefore, the functions of OsCRS1 and OsCAF1 could partially overlap but are not redundant in rice. In addition, the intron splicing of subgroup IIB introns, including *petD, rpl16, rps16*, and *trG* does not require OsCAF1 in rice, while the splicing of the subgroup IIB introns requires ZmCAF1 and AtCAF1 in maize and *A. thaliana* [11,21]. Moreover, in rice, there are three additional proteins containing two CRM domains, and the functions of the proteins may be different from OsCAF1. Intron splicing of subgroup IIA and subgroup IIB introns was regulated by OsCAF1, which, in turn, influenced chloroplast development in rice.

Nuclear-encoded protein ZmCRS2 could interact with ZmCAF1 to form the CRS2–CAF1 complex, which regulates the splicing of group IIB introns in maize chloroplasts [22]. In addition, our study demonstrated that OsCAF1 could interact with OsCRS2 through the C-terminal region, suggesting that the interaction mechanism of CAF1 and CRS2 might be conserved in maize and rice. The homologous alignment analysis results of the C-terminal region of CAF1 among various plants suggested high conservation at C-terminal positions 585–598 aa and 685–701 aa. However, the results of our study showed that the C1 region of OsCAF1, 387–584 aa, was essential for interactions with OsCRS2 rather than the conserved regions of the C2 and C4 regions. The C2 and C4 conserved regions of OsCAF1 could interact with other proteins, which requires further investigations. In addition, the *OsCRS2* expression levels decreased significantly in *oscaf1* (Supplementary Figure S2B), suggesting the mutation of OsCAF1 also influenced *OsCRS2* expression. Based on the above findings, the C1 region of OsCAF1 interacts with OsCRS2, and the generated complex, OsCRS2–OsCAF1, could participate in the splicing of subgroup IIA and subgroup IIB introns in rice chloroplast.

OsCAF1, which has two CRM domains, is an RNA intron-splicing factor in rice. *OsCAF1* influences the expression of chloroplast-associated genes, affects the accumulation of H2O2 in rice leaves, and plays a key role in early chloroplast development in rice. We demonstrated that OsCAF1 is potentially involved in the splicing of both group IIA and group IIB introns of chloroplast transcripts in rice. In addition, the C1-terminal region of OsCAF1 could interact with OsCRS2 and the OsCRS2–OsCAF1 complex could participate in the splicing of group II introns.
