The rice gene OsOLP1 (GenBank accession number: Os01g0839900) was used in the present study. The gene has no intron in its coding region. To generate OsOLP1 knockout mutant, a gene editing vector for OsOLP1 was constructed using the CRISPR/Cas9 method (Ma et al., 2015). To improve the targeting efficiency of CRISPR, two sites in the coding region of OsOLP1, T1 (ACCTTGGTCCGGCCGCAT’TGG) and T2 (AGTTGCCGGCCCTCGTGC’GGG) were selected and ligated to OsU6a and OsU6b promoters by overlapping PCR, and sgRNA was ligated to the end of the target site to construct the “OsU6a-T1-sgRNA-OsU6b-T2- sgRNA” expression cassette, and assembled into the binary pYLCRISPR/Cas9 Pubi-H vectors (digested with Bsa I) to generate transgenic rice plants. The target sites and primers ( Supplementary Table 1 ) required for vector construction were selected using CRISPR-GE(http://skl.scau.edu.cn/).

To generate OsOLP1 overexpression line, the full-length coding sequence of OsOLP1 was amplified from Oryza sativa (japonica Nipponbare cultivar) using the All-in-one First-Strand cDNA synthesis SuperMix and TransStart^® FastPfu Fly DNA polymerase (TransGene Biotech, Beijing, China) and inserted into the pRHV-cHA (He et al., 2018) vector carrying the Ubi promoter and C-terminal HA tag. Supplementary Table 1 shows the primer pairs required for construction of the vector.

The Agrobacterium tumefaciens EHA105 transformation was performed as described previously (Hiei et al., 1994; Clough and Bent, 1998). All constructed vectors were transferred to A. tumefaciens EHA105 via chemical transformation. 1 μg of recombinant plasmid DNA was added to EHA105 competent cells, placed on ice for 5 min, in liquid nitrogen for 5 min, at 37°C for 5 min, and on ice for 5 min, 700 mL of antibiotic-free LB medium was added and incubated for 2-3 h with shaking, followed by centrifugation to collect the cells, and 100 µL of the suspension was spread on LB plates with 50 μg/mL kanamycin and 25 μg/mL rifampicin. A. tumefaciens was co-incubated with callus tissues induced from mature rice seeds, and the desired rice lines were obtained by hygromycin resistance screening and differentiation.

For the identification of CR-OsOLP1 lines, DNA was extracted from five-week-old leaves, and the coding region of OsOLP1 was amplified using the Cas-OsOLP1 primer pairs ( Supplementary Table 1 ). The PCR products were then sequenced and aligned using DNAMAN. RNA was also extracted from the selected lines (CR-OsOLP1#20 and #27), and the expression levels of OsOLP1 were analyzed by qRT-PCR.

OsOLP1 overexpression lines (OE-OsOLP1) were determined by immunobloting and qRT-PCR. Total protein was extracted from five-week-old leaves in the OE-OsOLP1 lines using a plant protein extraction kit (Solarbio life science, Beijing, China) according to the manufacturer’s instructions. Rice leaves were extracted with extraction buffer to obtain total protein and then subjected to SDS-PAGE analysis. Samples were incubated with anti-HA antibodies (1:1000) and anti-rabbit (1:5000) and then exposed and photographed with Tanon Chemiluminescence Imaging System. Wild-type seedlings were used as the negative controls. Staining of total proteins with Ponceau S was used as the control. The leaves of the selected lines (OE-OsOLP1#28 and #38) were sampled for qRT-PCR analysis. Primer pairs used for the PCR are listed in Supplementary Table 1 .

Sterilization of experimental rice seeds consisting of OsOLP1 overexpressing and OsOLP1 knockout lines was accomplished by immersing them in 75% ethanol for 15 minutes, and residual ethanol was removed by rinsing seven times. Rice seeds were spread out on a Petri dish containing moistened filter paper and germinated in a growth chamber at 28°C under 16 h light and 8 h in darkness. Sprouted seedlings were transplanted and grown under the same conditions for five weeks. Each pot contains equal amounts of breathable vermiculite and nutrient soil. Samples were quick-frozen in liquid nitrogen and stored temporally at -80°C until analysis.

For the treatment of drought tolerance, five-week-old rice seedlings were continuously deprived of water for 5 days to test for their drought tolerance rate. The seedlings were rewatered for 7 days until the plants recovered, and plant survival was determined by the color and appearance of new leaves. The survival rate was determined by calculating the ratio of total survived seedlings to the total number of seedlings subjected to drought stress. Three replicates of each experiment with 20 seedlings per replicate were performed. To calculate the soil water content (SWC), a 1 ml pipette tip was inserted into the culture soil and vertical sampling of the soil was performed at different (0, 1, 2, 3, 4, 5) days of the drought treatment and SWC was calculated using the formula below. To calculate the relative water content (RWC) of leaves, the full-expanded leaves of OsOLP1 knockout and overexpression line, and WT were sampled at 0 and 5 days, and RWC was analyzed using the formula below.

where Sf represents soil fresh weight, Sd represents soil dry weight, Wf represents the fresh weight of rice leaves, Wd represents the dry weight, and Ws represents the saturated weight of leaves after immersion in sterile water for 6 h.For the treatment of NaCl, irrigated water was discarded from five-week-old rice seedlings cultured normally, and then the seedlings were watered with water containing 150 mM NaCl. The rice tissues were sampled at different time points to determine the level of gene transcription.

Triplicate samples of OsOLP1 overexpression lines, OsOLP1 knockout lines, and WT were obtained. Total RNA was extracted from related plant tissues (roots, stems, leaves, nodes, and leaf sheaths for OsOLP1 expression pattern analysis; leaves for CR-OsOLP1 and OE-OsOLP1 identification, transcript analysis of ABA and proline synthesis genes) using the EasyPure^® Plant RNA Kit (TransGen Biotech, Beijing, China). 1 μg of total RNA was reverse-transcribed into first-strand cDNA using All-in-One First-Strand cDNA Synthesis SuperMix (TransGen Biotech, Beijing, China).

qRT-PCR primers pairs ( Supplementary Table 1 ) were designed using the primer premier 5 and the 2×RealStar Fast SYBR green qRT-PCR Mix purchased from GenStar, Beijing, China. 20 microliters (µl) of the qRT–PCR mix was prepared following the manufacturer’s description, and analysis was carried out using the 96-well desktop QuantStudio 5 real-time PCR system. Osactin1 (LOC_Os03g50885) was used as the internal control for normalization. The relative expression analysis was calculated using the 2^–ΔΔCt method as previously performed (Livak and Schmittgen, 2001).

ABA concentration analysis was performed following Ding et al. approach (Ding et al., 2011). About 0.2 g of 3-week-old rice leaves were ground with liquid nitrogen to obtain a fine powder. 500 µL of extraction buffer (90% (v/v) methanol containing 200 mg/L diethydithiocarbamic acid sodium salt) was added and mixed them thoroughly. The extracts were transferred to glass tubes and incubated at 4°C overnight, then centrifuged for 10 min at 8000 g. The supernatant was evaporated to dryness in a new centrifuge tube, and the residue was dissolved with 500 μL of methanol tris buffer. Each experiment included 3 replicates. ABA content was determined using an ABA ELISA Kit following the manufacturer’s instructions (Mlbio, Shanghai, China).

Five-week-old rice leaves (before and 2 h after drought treatment) were fixed in a fixative buffer containing 2.5% glutaraldehyde overnight at 4°C. The fixative solution was discarded, and samples were washed three times with PBS buffer. After removing excess PBS buffer, samples were desiccated in a serially graded ethanol of 30%, 40%, 50%, 70%, 90%, and 100% for 15 min each. The leaf samples were dried in a vacuum desiccator, mounted on a metal stub with two-sided tape, and sprayed with gold. The stomata were examined using Zeiss Gemini 500 field emission scanning microscope (Gemini 500 FESEM). X500 magnification was used to count the number of stomata per square millimeter, and X1500 magnification to count the percentage of stomata opening and closing from six plants for each line.

The proline concentration was determined using Lee et al. method (Lee et al., 2013). 0.2 g of five-week-old leaf tissue was placed in a 2 ml tube and pulverized using a tissue lyser after quickly freezing the samples in liquid nitrogen. 3% sulfosalicylic acid was added to the extract. The samples were boiled at 100°C for 10 minutes and centrifuged at 10,000 g for 10 minutes at room temperature. The supernatant was pipetted into a fresh centrifuge tube, and glacial acetic acid (1 ml) and acidic ninhydrin solutions (1 ml) were added after cooling and after boiling for 1 h. The wavelength of the sample was determined at 520 nm using a spectrophotometer. Each experiment included three replicates. Proline content was calculated using the L-proline concentration standard curve.

Plant growth and preparation were the same as above. The thioglycolic acid (TGA) of lignin content determination method was carried out as previously described (Suzuki et al., 2009). Samples pulverization was the same as performed for proline content analysis. The powdered samples were oven-dried at 60°C for 1 h. 1.8 ml of methanol was added and extracted at 60°C for 20 min. The extract was centrifuged at 16100 g for 10 min at room temperature. The supernatant was discarded, and the extraction step was repeated with methanol. 1 ml of 3 N HCl and 0.1 ml of thioglycolic acid were added, and the samples were heated at 80°C for 3 h and centrifuged at 16100 g for 10 min. The supernatant was carefully removed and discarded, and 1 ml of distilled water was added to the pellet and vortexed for 30 s to mix. The samples were centrifuged as above, and the pellet was resuspended in 1 ml of 1 N NaOH and then vertically rocked for 16 h at 80 rpm. After centrifugation at 16100 g for 10 min at room temperature, the supernatant (about 1 ml) was transferred into a new 1.5 ml tube and acidified with 0.2 ml concentrated HCl. After freezing at 4°C for 4 h, the samples were centrifuged at 16,100 g for 10 min at room temperature. The supernatant was removed, and the pellet was dissolved in 1 ml of 1 N NaOH. After a 50-fold dilution with 1 N NaOH, the absorbance of the solution at 280 nm was measured using the spectrophotometer. Each experiment included four replicates. The lignin concentration was computed using the lignin standard curve as a reference.

Cross-sections of rice roots were stained as previously described (Pradhan Mitra and Loque, 2014). The cross-sections were obtained from the roots of five-week-old rice seedlings and incubated in 50% alcohol for 2 h. The sections of root tissues were embedded in an embedding medium, then cut into 100 mm sections with a cryostat, and stained with phloroglucinol-HCl solution (one volume of 37 N HCl to two volumes of 3% phloroglucinol in ethanol). The prepared sections were observed and photographed using the OLYMPUS BX61 microscope at 10X magnifications.

The amino acid sequence, signal peptide, and disulfide bond analysis of the protein were performed by phytozome (https://phytozome-next.jgi.doe.gov/) and Uniprot (https://www.uniprot.org/); amino acid sequence alignment by DNAMAN 9; gene expression profiling was performed by RiceXPro (https://ricexpro.dna.affrc.go.jp/) and Rice Expression Database (http://expression.ic4r.org/index); Analysis of the spider plot is performed by Hiplot (https://hiplot.cn/).

Data sets from the experiments were subjected to analysis of variance and graphs plotted on the GraphPad Prism, version 9.0.0. Data significance analysis was performed using one-way ANOVA and Tukey’s multiple comparison test. Each bar represents the means and standard deviations from each observation determined at a 95% significance level.

To ascertain that OsOLP1 has the characteristics of osmotins and OLPs, we performed cluster analysis of 40 osmotins and OLPs from rice. The result confirmed that OsOLP1 formed a cluster with multiple rice osmotins and OLPs ( Figure 1A ). By Uniprot and phytozome analysis, OsOLP1 was identified as a secreted protein with an N-terminal signal peptide containing 22 amino acids. Without signal peptide, the molecular weight of the mature protein was 24 kDa. The amino acid sequence alignment indicated that the protein has 16 cysteines and is capable of forming eight disulfide bonds ( Supplementary Figure 1 ), whose position could be obtained by Uniprot analysis. Furthermore, relative quantification of OsOLP1 revealed transcript abundance in various rice tissues, especially in the leaf nodes ( Figure 1B ). RiceXpro study revealed that OsOLP1 expression was present throughout the rice life cycle and was highest during the blooming stage. Rice Expression Database exhibited changes in OsOLP1 under various stresses, and environment Ontology indicated that OsOLP1 was associated with desiccation/dehydration stress response. ( Supplementary Figure 2 ). Therefore, Drought and NaCl were used to treat rice to explore whether the transcript levels of OsOLP1 respond to osmotic stress. After rice seedlings were exposed to drought and NaCl (150 mM) stress for 12 and 24 hours, the expression level of OsOLP1 was high in the roots, stems, and leaves ( Figures 1C, D ). These results suggest that OsOLP1 is widely expressed in tissues and responds to different osmotic stresses.

The genetic regulatory activities of OsOLP1 in drought-stressed rice seedlings were further ascertained by constructing OsOLP1 knockout plants by the CRISPR/Cas9 system and overexpressing plants. The OsOLP1 knockout mutant was generated using two targets in the coding sequence (CDS) regions of OsOLP1 driven by OsU6a and OsU6b promoters were ligated to sgRNA to generate sgRNA expression cassettes. Then the cassettes were inserted into the pYLCRISPR/Cas9Pubi-H vector for gene editing in wild-type rice, and 48 transgenic lines were obtained. DNA from all the lines was extracted, and the region OsOLP1 located was amplified and sequenced to determine knockout efficiency. However, only specific target fragments from three lines, #20, #27, and #31, were amplified. Sequencing analysis showed that all three lines were homozygous with a deletion of 206 base pairs between the two sites ( Figure 2A ). The transcript level of OsOLP1 determined by qRT-PCR was found to be lower in the OsOLP1 gene editing lines (CR-OsOLP1#20 and #27) ( Figure 2B ); therefore, these two lines were selected for further evaluation of drought stress.

Meanwhile, the coding sequence of OsOLP1 was fused with a 4HA tag driven by the Ubi promoter to generate overexpressing rice plants ( Figure 2C ). The expression of OsOLP1 in overexpression lines OE-OsOLP1#28 and #38 was successfully detected by immunoblot analysis using HA antibodies ( Figure 2D ). OsOLP1 expression levels of OE-OsOLP1#28 and #38 were 7-10-fold higher than that of wild-type ( Figure 2E ). The generation of CR-OsOLP1 and OE-OsOLP1 helped us to evaluate further the effect of OsOLP1 on drought tolerance in rice.

Although OsOLP1 also conforms to the structural characteristics of osmotin-like proteins, its role in drought conditions still needs to be verified. Five-week-old seedlings from the CR-OsOLP1 and OE-OsOLP1 lines were removed from the irrigation water and subsequently subjected to a 5-day drought without watering. After 5 days, the soil water content was reduced to about 30%, and the plants were in a severe state of water loss ( Figure 3B ). The survival rates of rice were calculated after recovery of 7 days of watering. After five days of drought treatment, the wild type showed survival rates of 28%. Compared to wild type, CR-OsOLP1 knockout lines showed lower average survival rates of 8.2%, exhibiting more sensitive to water loss, significant drought-induced symptoms, such as leaf curling and wilting, as well as significantly lower recovery rates after rehydration. However, OE-OsOLP1 lines grew better and maintained an average high survival rate of 53.1% after rehydration ( Figures 3A, D ). In addition, the relative water content of OE-OsOLP1 (~65%) was higher than that of WT (51.4%), whereas that of CR-OsOLP1 was lower (~ 25%) ( Figures 3C ). In a nutshell, OsOLP1 overexpression can avoid rapid water loss and improve rice drought tolerance.

ABA can respond to the osmotic stress faced by plants and utilize water rationally. The dehydration signal leads to ABA accumulation and activation of downstream stress-resistance signaling pathways. Therefore, variation in ABA content can be used as the barometer for plant drought tolerance assessment. The ABA content in CR-OsOLP1, OE-OsOLP1, and WT was analyzed. Under normal condition, no difference in ABA concentration among OE-OsOLP1, CR-OsOLP1 and WT. However, when subjected to a one-day drought, the ABA level in the CR-OsOLP1 knockout lines decreased by 50%, while that in the OE-OsOLP1 lines increased by 50% ( Figure 4A ).

However, the question of how OsOLP1 promotes endogenous ABA synthesis was raised. Zeaxanthin epoxidase (ZEP) and 9-cis-epoxycarotenoid dioxygenase (NCED) are crucial to ABA synthesis. ZEP1 is the initial enzyme in the ABA synthesis pathway, and NCED3 regulates ABA synthesis under water deficit, and their expression contributes to ABA-mediated drought tolerance. The transcript levels of these two enzymes were quantified by qRT-PCR. The results showed consistency with the ABA content above. The relative transcript abundance of OsZEP1 and OsNCED3 in OE-OsOLP1 lines were significantly up-regulated under drought conditions, while vice versa was observed in CR-OsOLP1 lines ( Figures 4B, C ). Thus, OsZEP1 and OsNCED3 mediated the OsOLP1-enhanced ABA deposition under drought.

The stomata are enclosed by specialized pairs of guard and subsidiary cells for mechanical support. The stomata regulate water utilization in higher plants through transpiration, and the density and closure of stomata are critical factors for plant drought tolerance. Therefore, we observed the closure and density of epidermal stomata in rice leaves using the scanning electron microscopy technique. As shown in Figures 5A, B , there was no significant difference in stomatal density per square millimeter between CR-OsOLP1, OE-OsOLP1, and WT. Then, after 2 hours of drought exposure, we calculated the percentage of stomatal closure per square millimeter using Figure 5C as the standard. The percentage of stomatal closure in CR-OsOLP1, OE-OsOLP1, and WT lines was not significantly different before drought treatment. However, after drought exposure, the percentage of stomatal closure was 85.6% in the wild type and ~80% in knockout CR-OsOLP1, indicating that CR-OsOLP1 had less stomatal closure than the wild type, while overexpression plants OE-OsOLP1 showed more stomatal closure numbers of ~96% ( Figure 5D ), suggesting that OsOLP1 can reduce water loss in rice by regulating stomatal apparatus to maintain water balance.

When plants suffer from unfavorable external conditions, they usually adapt by accumulating protective components. Among these protective components, proline accumulation is characteristic for plant stress defense. It was determined that the proline level was low under normal conditions. However, compared with WT, its levels rapidly increased by 2.7-6.1-fold in the OsOLP1 overexpression lines and decreased by 1.9-4.4-fold in the OsOLP1 knockout lines under water deficit conditions ( Figure 6A ). Consistent with this result, the transcript levels of genes encoding delta-1-pyrroline-5-carboxylate synthase (OsP5CS1 and OsP5CS2) involved in the proline synthesis were significantly increased by about 1 to 10 folds in OE-OsOLP1 lines ( Figures 6B, C ). In addition, the transcription level of the proline-degrading protein proline dehydrogenase (OsPDH1) was significantly increased in CR-OsOLP1 lines compared to WT ( Figure 6D ). Proline accumulation levels in our report are partly attributable to these regulatory genes’ expression.

Deposition of lignin leads to the lignification of plant tissues, which often strengthens plants against abiotic stresses. The degree of lignification is proportional to the lignin content. We measured the lignin content of CR-OsOLP1, OE-OsOLP1, and WT lines using the Thioglycolic acid (TGA) method. The results revealed that OsOLP1 deletion adversely affects lignin content in rice. However, the lignin concentration of the OsOLP1-overexpression lines was significantly 1.3-1.6-fold higher than that of WT ( Figure 7A ). Lignin-specific (Phloroglucinol-HCl) staining of rice roots also visually showed the degree of lignification of CR-OsOLP1 and OE-OsOLP1 lines. The sclerenchyma tissues and vasculature in OE-OsOLP1 lines were highly lignified, whereas WT and CR-OsOLP1 lines were less lignified ( Figure 7B ). Taken together, OsOLP1 may have mediated the synthesis of lignin.

Finally, a spider plot was used to normalize the effects of OsOLP1 on rice survival rate, relative water content, ABA and proline contents, stomatal closure, and lignin accumulation under drought conditions ( Figure 7C ). The evenly distributed spokes on the wheel showed that CR-OsOLP1#20 and CR-OsOLP1#27 were closer toward the center (0%), and therefore weakened their tolerance to drought stress. However, OE-OsOLP1#28# and OE-OsOLP1#38 showed strong drought tolerance as most of their spokes were far from the center on the wheel.