GeD lines were generated using Ilmi as a background. The generated GeD, and Ilmi were cultivated at 28/26°C during a 16/8-h light/dark photoperiod. Three sgRNAs (sgRNA 1–3) were designed using the CRISPR RGEN Tools¹ program to target the first exon of OsSAP (Oryza sativa Senescence-associated protein, Accession no. AB734097) and edit the genome (Park et al., 2015); these sgRNAs were integrated into the pRGEB32 vector, which is capable of expressing Cas9 in plants (Mikami et al., 2015). The obtained binary vector was transformed through calli derived from Ilmi using Agrobacterium tumefaciens EHA105 and the rice plant was regenerated as previously described (Yukoh and Toshihiko, 2008). We also used OsSAP-T₆ lines for compared GeD lines related to drought tolerance were acquired from Prof. Ubaidillah (Ubaidillah et al., 2013).

Genomic DNA was extracted from rice leaves using a DNeasy Plant Mini Kit (QIAGEN, Cat. 69104, Hilden, Germany) according to the manufacturer’s instructions. To amplify the target genome region of CRISPR/Cas9, PCR analysis was performed with primers using these region sequences. The amplified PCR product was purified and mutants for which the target gene was genome-edited were detected through Ilmi sequences and alignment via sequencing chromatogram analysis. For all regeneration GeD plants, each line of the T₀ generation was analyzed for the target region of OsSAP.

Roots, stems, and leaves were sampled at the seedling stage of rice, whereas the roots, stems, leaves, and panicles at the heading stage were sampled. Total RNA was extracted from the sampled organs using the RNeasy Micro Kit (QIAGEN, Cat. 74004, Hilden, Germany) according to the manufacturer’s instructions. Subsequently, cDNA was synthesized using 1 μg of total RNA from each sample with the qPCRBIO cDNA Synthesis Kit (PCRBIOSYSTEMS, United States), used according to the manufacturer’s instructions. The synthesized cDNA was diluted 1:10 (v/v) using deionized water and then used for qPCR analysis together with the housekeeping gene OsActin. qPCR analysis was performed using an Eco™ Real-Time PCR System (Illumina, Cat. EC-900-1001, CA, United States) with 2X Multi-Star OneStep qRT-PCR Master Mix (BIOFACT, Cat. RQ351-50h, Seoul, Korea); relative expression levels were analyzed according to the manufacturer’s instructions. Cycle threshold values for each sample were normalized using OsActin as a standard (Ariani et al., 2004). Samples were each analyzed in five independent biological replicates. The list of primer sets used for the analysis is listed in Supplementary Table S1.

GeD lines, Ilmi, and OsSAP-T₆ lines were grown by filling a Wagner pot with soil. Each pot used in the experiment was filled with the same weight of soil, and water was supplied equally to each pot at a height of 5 cm from the soil. Drought conditions were created by not watering plants when they reached the fourth leaf stage; the water supply was stopped until the soil moisture content reached 20%. When the soil moisture content reached 20%, the plants were watered for 7 days and the survival rate was analyzed (Li et al., 2013). To analyze growth, shoot length, root length, chlorophyll content, number of tillers, and 1,000-grain weight, data were collected when the soil moisture content was 20%.

To detect H₂O₂ after drought treatment, the 3,3′-diaminobenzidine (DAB) staining method of Liu et al. (2014) was modified and used. After drought treatment, leaves were cut to a length of 5 cm and then submerged in 1-mg/ml DAB (pH 3.8) to be dyed in darkness at 25°C for 16 h. To prevent the penetration of DAB solution from being disturbed by foreign substances on the leaves, they were cut and their surface was washed with deionized water. After decolorization using 100% ethanol, the reaction of H₂O₂ on each leaf was analyzed through newly created brown spots. DAB staining was analyzed on leaves of five different plants for each line and representative images were collected. In addition, to analyze the content of H₂O₂ contained in leaves, H₂O₂ was extracted using the method proposed by Rao et al. (2000). The content of H₂O₂ contained in plant leaves was determined via the Quantitative Peroxide Assay Kit (Thermo Fisher, Cat. 23280, Seoul, Korea), used according to the manufacturer’s instructions. Leaf extracts were measured at OD₅₆₀ on a spectrometer. The absorbance values measured for each leaf were calibrated with the standard curve generated with from a known H₂O₂ concentration.

To evaluate the generation of O₂⁻ in the leaves of plants after drought stress, 0.1% nitro blue tetrazolium (NBT) solution was added to 10-mM potassium phosphate buffer (pH 7.8), and then leaf samples were cut to a length of 5 cm and precipitated (Fitzgerald et al., 2004). They were incubated in darkness at 25°C for 16 h. To analyze the blue spots generated by the reaction of NBT and O₂⁻, the chlorophyll contained in the leaves was removed. To remove chlorophyll, the leaves were fixed in alcohol lactophenol (2:1:1, 95% ethanol: lactic acid: phenol) at 65°C for 30 min before they were rinsed with 50% ethanol. Afterward, NBT was found to react with superoxide on the surface of the leaf to form blue spots.

Proline content was determined using the method from Bates et al. (1973). After drought treatment on GeD, and OsSAP-T₆ lines, 0.5 g of leaf was ground, and homogenized with 3% sulfosalicylic acid (w/v) at 100°C for 10 min. This was then centrifuged at 4,500 RPM for 30 min after 10 min. Subsequently, 2 ml acidninhydrin and 2-ml glacial acetic acid were added to 2 ml of the supernatant, boiled at 100°C for 40 min, and then cooled using ice. Proline levels in the drought-treated samples were determined at OD₅₂₀. MDA content was analyzed via the method outlined by Heath and Packer (1968). A 1-g leaf sample was placed in 10 ml of 10% trichloroacetic (v/v) reagent and was homogenized, after which it was centrifuged at 4,500 RPM for 10 min. Next, 2 ml of the supernatant was reacted with 2 ml of thiobarbituric acid at 100°C for 15 min and then cooled with ice before it was analyzed at OD₄₅₀, OD₅₃₂, and OD₆₀₀ to determine MDA concentration. Finally, the MDA content was determined by applying the formula (μmol L⁻¹) = 6.45 (OD₅₃₂ − OD₆₀₀) − 0.56 OD₄₅₀, which was proposed by Shi et al. (2012). Catalase (CAT) activity was analyzed by modifying the method of Beers and Sizer (1952). The 3.0 ml reaction mixture contained 2 ml of sodium phosphate buffer (50 mM at pH 7.0), 0.5 ml of H₂O₂ (40 mM), and 0.5 ml of enzyme extract. The decomposition of H₂O₂ at OD₂₄₀ was analyzed. The activity of superoxide dismutase (SOD), an antioxidant enzyme, under drought stress was analyzed according to the protocol of Duan et al. (2012). The unit of SOD activity was determined as the minimum amount of enzyme required to inhibit the photochemical reduction of NBT chloride by 50% at OD₅₆₀. Peroxidase (POD) activity was analyzed by the guaiacol oxidation method proposed by Weydert and Cullen (2010). The POD reaction mixture (3 ml) contained 50 mM sodium acetate buffer (pH 5.0), 20-mM guaiacol, 40-mM H₂O₂, and 0.1-ml enzyme extract. The prepared POD reaction mixture was analyzed at OD₂₄₀ at 1 min intervals; data were recorded before 3 min had passed after the start of the reaction.

Oryza sativa Senescence-associated protein genome-edited plants with CRISPR/Cas9 were analyzed through genomic sequencing. Regenerated plants from callus were named T₀ plant, and seeds harvested from T₀ plant were named T₁ seed. Also, when T₁ seeds are sowing and transplanted into the field, they are named T₁ plants, and the seeds harvested from T₁ plants become T₂ seeds. Among genome-edited plants, homozygous types were selected from T₀. An accession number was assigned to each tiller of the T₀ plants, and T₁ seeds were harvested, and each was assigned an accession number. T₁ seeds were transplanted in one row for each line in the field, and 25 plants were transplanted for each row. Twenty-five plants were planted in each line, and the plants that died after growth were replaced with purple rice and were not used for subsequent analysis. In T₁ plants, T₂ seeds were harvested in bulk. In each generation, the expression level of Cas9 and the target gene, OsSAP, was analyzed using OsActin as a control.

All data were statistically analyzed using one-way ANOVA and Duncan’s multiple range tests via the SPSS program (IBM SPSS Statistics, version 22, NC, United States). Value of p < 0.05 was considered statistically significant.

To evaluate the role of OsSAP in drought conditions, we developed GeD₀ (Genome-editing Drought gene OsSAP 0 generation) lines through genome editing of OsSAP using CRISPR/Cas9. In the CRISPR/Cas9 vector, which can be expressed in rice, the OsSAP guide RNA is regulated by the U3 promoter (Figure 1A). The guide RNA is inserted into the BsaI site of the pRGEB32 vector (Figure 1B). Three guide RNAs were designed, and the GC contents were set at 50–70% (Figure 1C). Afterward, the CRISPR/Cas9 vector, which the sgRNA was inserted into, was transformed into the callus by Agrobacterium medium (Figures 2A–F). For callus inoculation, callus cultured for 21 days were selected (Figure 2A). After co-culture with Agrobacterium for 3 days, green spots were investigated after 7 days in the regeneration medium (Figure 2B). After 45 days on the regeneration plate, shoot growth was almost completed (Figure 2C), and culture was performed for rooting on a new regeneration plate for 30 days (Figure 2D). After 3 days of acclimatization (Figure 2E), T₀ plants were transplanted into pots for T₁ seed harvesting (Figure 2F). A total of 237 regenerated plants were developed, but only 104 regenerated plants survived when transplanted into soil after acclimatization. Among 104 regenerated plants, there were 45 regenerated plants with genome editing via OsSAP-sgRNA1, and T-DNA insertions were analyzed in 23 of them (Figure 2G). There were 38 regenerated OsSAP-sgRNA2 plants, and T-DNA insertion was analyzed in 15 of them. When OsSAP-sgRNA3 was used, only 15 out of 21 regenerated plants had T-DNA insertion. Around 1,000 Ilmi seeds were used for callus culture, and 237.0 ± 12.8 regenerated plants were developed. Among them, T-DNA was successfully inserted in 53 regenerated plants (Figure 2H). The editing type of OsSAP was analyzed with each sgRNA (Figure 3). When OsSAP-sgRNA1 was used, deletion occurred in 12 of 23 regenerated plants. There were also two insertions, and nine did not respond to genome editing (Figure 3A). Among the regenerated plants, the editing was homozygous in three plants (Figure 3B). In OsSAP-sgRNA2 plants, there were eight deletions, two insertions, and six did not respond to editing. In one of these plants, the homozygous edited type was analyzed. In OsSAP-sgRNA3 plants, there were four deletions, one insertion, and 10 were not edited. Among editing plants, two plants were edited homozygous. For continuous research, three plants (designated as GeD₀ 1–1, GeD₀ 3–1, and GeD₀ 3–2) were selected for the generation of GeD lines.

To evaluate the response of OsSAP to drought conditions, the soil moisture content of the GeD, and OsSAP-T₆ lines was reduced to 20%. At this level, the survival rate, shoot length, root length, chlorophyll content, number of tillers, and 1,000-grain weight of each line were investigated (Supplementary Figure S1). These variables showed no significant differences in all lines of GeD, and OsSAP-T₆. The survival rates of GeD lines (GeD₀ 1–1, GeD₀ 3–1, and GeD₀ 3–2) under drought stress were lower than that of OsSAP-T₆ (43.5% vs. 17.4, 16.8, and 15.6%, respectively). The overall shoot length showed a trend under drought stress: GeD lines (GeD₀ 1–1, GeD₀ 3–1, and GeD₀ 3–2) had shoot lengths of 6.4, 5.8, and 6.1 cm, respectively, which were lower than that of OsSAP-T₆ (15.1 cm). Under drought stress, the root lengths (2.8 cm, 3.5 cm, 4.9 cm, respectively) of GeD lines (GeD₀ 1–1, GeD₀ 3–1, and GeD₀ 3–2) were shorter than or equal to that of OsSAP-T₆ lines (9.1 cm). The chlorophyll content of OsSAP-T₆ plants under drought conditions was 42.1 SPAD; that of the GeD lines (GeD₀ 1–1, GeD₀ 3–1, and GeD₀ 3–2) was lower than that of the OsSAP-T₆ Lines. Drought stress treatment reduced the number of tillers; the tiller number of OsSAP-T₆, 12.3, was larger than that of the GeD lines at (GeD₀ 1–1, GeD₀ 3–1, and GeD₀ 3–2), 2.3, 2.3, and 2.0, respectively. Finally, when 1,000-grain weights were measured in each line after drought stress, the weight of OsSAP-T₆ lines was 20.3 g, whereas the weights of the GeD lines (GeD₀ 1–1, GeD₀ 3–1, and GeD₀ 3–2) were lower at 9.5, 7.8, and 7.1 g, respectively.

Superoxide dismutase and CAT are important enzymes that protect plant cells from oxidative damage caused by ROS: SOD acts as an antioxidant defense mechanism in almost all living cells by catalyzing the conversion of superoxide (O₂⁻) to O₂ and endogenous H₂O₂; CAT then catalyzes the conversion of H₂O₂ to H₂O and O₂⁻. To analyze the ROS scavenging capacity of OsSAP under drought stress, GeD, and OsSAP-T₆ lines were subjected to drought stress until the soil moisture content reached 20%. Under control conditions, the H₂O₂ contents of GeD, and OsSAP-T₆ lines were the same (Figure 4). However, under the drought stress conditions, the amount of H₂O₂ increased and the H₂O₂ increase rate differed in each line. In OsSAP-T₆, H₂O₂ increased by 31.6 ± 0.9% due to drought stress. However, in the GeD, H₂O₂ increased by an average of 71.6 ± 4.5%. Under drought stress, the GeD lines generated more H₂O₂ than OsSAP-T₆. All lines of GeD, and OsSAP-T₆ had similar levels of malondialdehyde (MDA) content under control conditions. However, when drought stress was applied, MDA content increased in all analyzed lines relative to the control conditions. However, the OsSAP-T₆ lines increased by only 6.5 ± 0.1% on average, whereas the GeD lines showed an average increase of 12.6 ± 3.2%. Proline production by drought treatment was also different in the GeD lines, compared to the OsSAP-T₆ lines. Under control conditions, almost no proline was generated in all analyzed lines. However, when drought stress was applied, proline content increased overall in all analyzed lines: the average increase in the OsSAP-T₆ line was 494.7 ± 9.1%, which represented an increase of 493.7% ± 9.1%. The average proline content of GeD lines was 283.9 ± 43.9%, which was an increase of 282.9 ± 43.9% when compared with the control. Thus, proline content increased more in the OsSAP-T₆ lines following drought stress application. These findings suggest that the expression of OsSAP reduces cell damage in leaves when a rice plant in under drought stress conditions. Under normal conditions, the level of SOD activity was consistent in all lines. However, when drought stress was applied, SOD activity increased in all lines: the SOD activity was highest in the OsSAP-T₆ lines and lowest in the GeD lines. POD activity was also the same in all analyzed lines under normal conditions; however, when drought stress was applied, POD activity was highest in the OsSAP-T₆ lines and lowest in the GeD lines. CAT showed a similar trend to that of SOD: it was highest in OsSAP-T₆ lines and lowest in GeD lines when drought stress was applied. These results suggest the possibility that OsSAP increases drought resistance by acting on the oxidative stress response.

When the rice reached the four-leaf stage, the GeD and OsSAP-T₆ lines were subjected to drought stress by stopping the water supply until the soil moisture content reached 20%. Before drought stress treatment, there was no difference in the phenotype of all lines and, the growth rate was similar (Figure 4). However, when water was supplied after drought stress, the recovery of the GeD lines was slow, whereas the recovery of the OsSAP-T₆ lines was rapid. To determine whether ROS production was related to OsSAP-mediated resistance to drought stress (with moisture content at 20%, as described earlier), the H₂O₂ content of GeD, and OsSAP-T₆ lines after drought stress treatment were analyzed using DAB staining. In the controls, all analyzed lines showed very little staining. Under drought stress, OsSAP-T₆ lines were barely stained but the GeD lines produces many dark and brown spots were on their leaves; indeed, most of their leaves were covered with brown spots. In addition, NBT staining was analyzed to evaluate the relationship between OsSAP and O₂⁻ production under drought conditions. Under control conditions, almost no blue spots were observed in all lines; however, after drought stress, the OsSAP-T₆ lines still showed a similar pattern to the control, whereas the incidence of blue spots increased in GeD lines compared to both. NBT and DAB staining indicated that drought stress generated H₂O₂ and O₂⁻ at increased levels in the GeD lines relative to the levels in the OsSAP-T₆ lines, suggesting that OsSAP could control overproduction and regulation of H₂O₂ under drought stress. Thus, the expression of OsSAP could reduce peroxidation and ROS accumulation by enhancing the activity of major antioxidant enzymes under drought stress conditions.

To understand the molecular mechanism of OsSAP in drought stress, the gene expression level of transcription factors that respond to drought stresses was analyzed. Specifically, after drought stress on GeD, and OsSAP-T₆ lines, relative expression levels of the stress-related transcription factors OsABI5, OsRAB16A, OsLEA3, OsLIP9, OsCATA, OsDREB2A, OsNAC5, and OsNAC6 were examined using qRT-PCR (Figure 5). All analyzed stress-related transcription factors were maintained at low levels or were significantly downregulated in all analyzed lines before drought stress treatment. However, after drought stress, all expression levels were upregulated. Expression levels of these stress-related transcription factors showed contrasting trends in OsSAP-T₆ and GeD lines; significantly lower expression levels were observed in GeD compared to OsSAP-T₆ lines. These results suggest that overexpression of OsSAP supports an increase in drought stress tolerance and enhances the expression of stress-related genes.

Expression levels of OsSAP and Cas9 were analyzed in GeD₀ and GeD₁ plants (Figure 6). In all GeD₀ plants, Cas9 was expressed at a level similar to OsActin, but OsSAP was not expressed. In all GeD₁ plants, OsSAP was not expressed; however, Cas9 was expressed in various concentrations throughout the GeD₁ lines. Cas9 was very weakly expressed in GeD₀ 1–2, GeD₀ 1–3, and GeD₀ 1–4, and GeD₀ 1–1 and GeD₀ 1–5. OsSAP was not expressed in all OsSAP-G-2 lines; however, Cas9 expression of each GeD₀ 2 line was different. In GeD₀ 2–1, Cas9 was expressed very weakly compared to OsActin, and in GeD₀ 2–3, Cas9 was expressed at a level similar to that of OsActin. In addition, GeD₀ 2–2 showed weaker Cas9 expression than OsActin, but stronger Cas9 expression than GeD₀ 2–1. OsSAP was not expressed in GeD₀ 3 lines, and the Cas9 expression level of all GeD₀ 3 lines was weaker than that of OsActin. The expression levels of OsSAP and Cas9 in GeD₂ seeds were also analyzed. OsSAP was not expressed in all lines of GeD₂ seeds. In the GeD₀ 1-1-1, GeD₀ 1-2-1, and GeD₀ 1-4-1 lines, Cas9 expression was weaker than that of OsActin. Cas9 was strongly expressed in GeD₀ 1-4-1, and Cas9 expression of GeD₀ 1-3-1 and GeD₀ 1-5-1 is similar to that of OsActin. In all the GeD₀ 2 lines, the Cas9 expression level was lower than that of OsActin. In particular, Cas9 was not expressed in GeD₀ 2-3-1. In GeD₀ 3-1-1, the Cas9 expression level was lower than that of OsActin, but the Cas9 expression level of GeD₀ 3-3-1 and GeD₀ 3-6-1 was similar to that of OsActin. Agricultural traits of T₀ plants, T₁ plants, T₁ seeds, and T₂ seeds were investigated. The coefficient of variation (CV) of the heading date, plant height, and culm length of T₀ plant were 2.5, 12.9, and 11.7%, respectively (Supplementary Figure S2). The CVs of the heading date, plant height, and culm length of T₁ plant were 3.1, 2.2, and 2.6%, respectively. The CVs of grain length, grain width, and 1,000-grain weight of T₁ seed were 4.5, 5.6, and 17.1%, respectively, and the CV of grain length, grain width, and 1,000-grain weight of T₂ seed were 1.9, 3.4, and 4.1%, respectively.