Background

Rice (Oryza sativa L.) is the most important staple crop worldwide. Improving rice yield is an urgent need to solve the crisis of food shortage. Rice yield mainly depends on growth environment, plant height, tillering and other important agronomic traits [1].

Under optimal environmental conditions, rice yield is mainly controlled by tiller numbers per plant, grain yield per plant and thousand-grain weight [1]. Starting with shoot branching, rice tiller experience two stages, axillary meristem formation and tiller bud outgrowth [26]. Subclass I SnRK2s have been identified as playing important roles in early processes of drought stress [27,28,29,30]. For example, SnRK2.4 and SnRK2.10 are activated by ABI1 and PP2CA under osmotic stress in A. thaliana [27, 28]. OsSAPK6 can phosphorylate and activate OsbZIP10 and OsbZIP46 in the ABA signaling pathway [29, 30]. For subclass II SnRK2s, SnRK2.8 positively regulates drought resistance response and participates in root growth [31]. A recent report showed that SnRK2.7 and SnRK2.8 regulated drought response genes and transcription factors bound to ABA response elements in A. thaliana [32]. Recent report indicated that SAPK2 is involved in the osmotic stress response by phosphorylating OsbZIP23 and OsbZIP46 and promoting the transcription of stress response genes [33, 34]. These data indicate that SnRK2s are crucial for abiotic stress responses in rice. Therefore, it is necessary to clarify the special function of SnRK2s.

Up till now, the function of OsSAPK3 has not been reported yet. In this study, sapk3 mutants (sapk3-1 and sapk3-2) generated by CRISPR/Cas9 system were used to characterize the role of OsSAPK3 in responding to drought stress and regulating rice yield. These results provide new insight for functional analysis of SnRK2s and engineering of drought resistant rice.

Results

Cloning and sequence analysis of OsSAPK3

The open reading frame of OsSAPK3 consists of 1,005 bp, which encoded 334 amino acids (Annotation identified in Introduction to the Rice Genome Annotation Project. database:LOC_Os10g41490 http://rice.plantbiology.msu.edu/index.shtml). SnRK2s contain a Ser/Thr enzyme activity conservative domain DFGYSKSSVLHSQPKSTVGTPAYIAPE and an ATP binding site GXGXXGX. The C-terminal sequence has high polymorphism, mainly including response to osmotic and ABA stress [16]. OsSAPK1, OsSAPK2 and OsSAPK3 are all rice SnRK2 II subfamily members. Similar to SnRK2 III members that play an important role in ABA pathway, OsSAPK3 contained several conservative domains, such as an ATP binding domain (ATP) (marked with black), the activation loop (marked with red lines) and SnRK2s conserved motif (marked with blue lines) (Fig. 1 A, Supplementary Table 1). But there is a significant difference in the ABA-activated motif (ABA box) (marked with yellow lines) comparing with ABA-pathway SnRK2s (Fig. 1 A, Supplementary Table 1).

Fig. 1
figure 1

Sequence analysis and phylogenetic relationship of SnRK2II and SnRK2III subfamily genes. A Amino acids comparison and B Phylogenetic relationship of selected SnRK2 members in Arabidopsis thaliana (At), Oryza sativa (Os) and Zea mays (Zm). The ATP-binding domain (ATP) and the activation loop are marked with black and red lines respectively. The SnRK2s conserved motif and the ABA-activated SnRK2s motif (ABA box) are marked with blue and yellow lines respectively

Full size image

To analyze the closeness of OsSAPK3 to ABA dependent SnRK2 III members, a phylogenetic tree was constructed based on their amino acids sequence of selected SnRK2 II and SnRK2 III members from A. thaliana (At), Oryza sativa (Os) and Zea mays (Zm) (Supplementary Table 1). These results showed that OsSAPK3 was similar to ZmSAPK3 (Fig. 1 B). In addition, it is highly similar to AtSnRK2.8 in A. thaliana and OsSAPK2 in rice (Fig. 1 B). These results indicated that OsSAPK3 might have functions different from subclass III SnRK2s but similar to OsSAPK2 and AtSnRK2.8.

Expression profile and localization of OsSAPK3

To analyze the responsiveness of OsSAPK3 under different abiotic stresses in rice, OsSAPK3 expression profiles were conducted via quantitative RT-PCR (qRT-PCR). Under drought, NaCl and PEG treatments, the expression of OsSAPK3 remarkably increased (Fig. 2 A). In ABA-treated plants, OsSAPK3 expression was weakly induced (Fig. 2 A).These results indicated that OsSAPK3 played a role in osmotic stress response, but had moderate role in ABA tolerance.

Fig. 2
figure 2

The expression pattern of OsSAPK3 A Transcription analyses of OsSAPK3 under various stress, such as 150 mM NaCl, 20% PEG6000, exposed to the air or 100 μM ABA. B Transcription analyses of OsSAPK3 in different rice tissues by qRT-PCR. C GUS staining analyses of ProSAPK3-GUS expression in different rice tissues. a, root; b, stem; c, leaf. (D) Subcellular localization of OsSAPK3 in rice protoplasts. Scale bar = 10 μm. Data represent means ± SD of three biological replicates (5 plants for each replicate)

Full size image

To investigate the biological role of OsSAPK3, the spatial and temporal expression of OsSAPK3 in different tissues were analyzed. The qRT-PCR results displayed that OsSAPK3 expression level was highest in roots, followed by leaves (Fig. 2 B). To further verify this result, ProOsSAPK3-Gus plants were used for detecting tissue expression patterns. The GUS activity was relatively higher in roots than in leaves (Fig. 2 C), which was consistent with qRT-PCR analyses. The subcellular localization of OsSAPK3 was assessed through transient expression of the OsSAPK3-GFP fusion protein in rice protoplasts. Fluorescence microscopy analyses showed that OsSAPK3 was localized in the nucleus and cytoplasm (Fig. 2 D).

OsSAPK3 mutated plants show decreased ABA sensitivity

To explore the function of OsSAPK3 in responses to ABA stress, OsSAPK3 mutated plants were generated using CRISPR/Cas9 gene editing technology. Two homozygous lines were chosen for ABA tolerance evaluation, which were named sapk3-1 and sapk3-2.

The sapk3-1 plants contained a 7-bp deletion in the first exon of OsSAPK3, while sapk3-2 plants carried a 14-bp deletion in second exon (Fig. 3 A, Supplementary Table 2). Amino acid sequence analyses showed that OsSAPK3 protein from sapk3-1 and sapk3-2 mutant plants had only 5 or 48 amino acids due to premature termination of translation. (Fig. 3 B, Supplementary Table 2). In addition, OsSAPK3 expression level was much lower in sapk3-1 and sapk3-2 plants than in wild-type plants (Fig. 3 C). These results indicated that sapk3-1 and sapk3-2 mutant lines were OsSAPK3 loss-of-function mutants.

Fig. 3
figure 3

OsSAPK3 mutated plants show decreased ABA sensitivity. A Sequence comparison of wild-type plants and two sapk3 mutant lines. B Protein structures of wild-type plants and two sapk3 mutant lines. C Transcript accumulation of OsSAPK3 in wild-type plants and two sapk3 mutant lines. D Phenotypes of wild-type plants and two sapk3 mutant lines before and after 100 μM ABA treatment for10 days. Scale bars = 2 cm. E Shoot and F Root length corresponding to D. Data represent means ± SD of three biological replicates (16 plants for each replicate). **P < 0.01 (Student’s t-test)

Full size image

In order to determine whether OsSAPK3 participates in ABA responses, shoot and root length of the wild-typeplants and two sapk3 mutant lines were examined after treatment with 100 μM exogenous ABA for 10 days. Compared with wild-type plants, the shoot and root growth inhibition significantly reduced in two sapk3 mutant lines after ABA treatment (Fig. 3 D-F), which indicated that sapk3-1 and sapk3-2 plants were not sensitive to ABA. These results demonstrated that OsSAPK3 played a role in ABA dependent signaling pathway.

OsSAPK3 mutated plants show reduced drought stress tolerance

To confirm the function of OsSAPK3 in drought tolerance in rice, the performance of wild-type plants and two sapk3 mutant lines under drought stress was examined by withholding water for 7 days. The survival rate of wild-type plants (46.4%) was significantly higher than that in two sapk3 mutant lines (i.e., sapk3-1 and sapk3-2 plants survival rates were only 20.4% and 22.5%, respectively) (Fig. 4 A and B). This result demonstrated that OsSAPK3 played a positive role in drought stress response.

Fig. 4
figure 4

OsSAPK3 mutated plants show reduced drought stress tolerance A Phenotype of wild-type plants and two sapk3 mutant lines before and after drought stress treament for 7 days. Scale bars = 2 cm. B Survival rates corresponding to A. C Relative water loss rate of leaves from 2-week-old wild-type plants and two sapk3 mutant lines under drought stress. D Stomatal density and E Stomatal conductance from wild-type plants and two sapk3 mutant lines before and after drought stress. F Transcript level of stomata genes OsSLAC1 and G OsSLAC7 in wild-type plants and two sapk3 mutant lines before and after drought stress. Values represent the means ± SD of three biological replicates (5 plants for each replicate). **P < 0.01 (Student’s t-test)

Full size image

Since the drought tolerance is closely related to relative water loss rate, the relative water loss rate was observed. As shown in Fig. 4, the relative water loss rate of wild-type leaves was higher than that in two sapk3 mutant lines (Fig. 4 C). Stomatal status is important for drought tolerance, then the stomatal density and stomatal conductance at 6-leaves stage were examined. There was no obvious difference for stomatal density between the wild-type plants and two sapk3 mutant lines (Fig. 4 D). Under normal conditions, the stomatal conductance of all plants had no significant difference between wild-type plants and sapk3 mutants. However, stomatal conductance of wild-type plants was significantly lower than that in two sapk3 mutant lines under drought conditions (Fig. 4 E). Based on these results, OsSAPK3 may change the stomatal conductance by participating in ABA dependent stomatal regulation. To test this assumption, the expression levels of slow anion channel-associated 1 (OsSLAC1) and slow anion channel-associated 7 (OsSLAC7) were verified. The expression levels of these two genes in two sapk3 mutant lines were significantly lower than that in wild-type plants under drought conditions (Fig. 4 F and D).

These results indicated that OsSAPK3 improved the tolerance to drought stress by regulating stomatal conductance through regulating the expression of OsSLAC1 and OsSLAC7.

Metabolic regulation of sapk3 mutants in response to drought stress

Plants adapt to drought stress through physiological and biochemical changes, such as the accumulation of proline and soluble sugar [23]. Thus, the content of proline and soluble sugar was examined. Under normal conditions, there is no significant difference in proline and soluble sugar content among all plants (Fig. 5 A and B). After drought treatment, the proline and soluble sugar content in two sapk3 mutant lines were significantly lower than that in wild-type plants (Fig. 5 A and B). Proline biosynthesis is catalyzed by 1-pyrroline-5- carboxylate synthetase (OsP5CS) [35]. According to qRT-PCR data, the OsP5CS expression level was much lower in sapk3 mutants than in wild-type plants under drought conditions (Fig. 5 E), which was consistent with proline content. These results suggested that OsSAPK3 might positively regulate the resistance to drought stress via changing the accumulation of osmolyte, such as proline and soluble sugars.

Fig. 5
figure 5

Metabolic regulation of sapk3 mutants in response to drought stress. A Proline B Soluble sugar C MDA contents in the seedlings of wild-type plants and two sapk3 mutant lines before and after drought stress treament for 7 days. D Relative ion leakage in wild-type plants and two sapk3 mutant lines leaves before and after drought stress treament for 7 days. E Relative expression levels of OsP5CS in the seedlings of wild-type plants and two sapk3 mutant lines before and after drought stress treament for 7 days. F AsA content in the seedlings of wild-type plants and two sapk3 mutant lines before and after drought stress treament for 7 days. G CAT H SOD I POD activities in the seedlings of wild-type plants and two sapk3 mutant lines before and after drought stress treament for 7 days. Values represent the means ± SD of three biological replicates (5 plants for each replicate). **P < 0.01 (Student’s t-test). FW: Fresh weight

Full size image

OsSAPK3 affects ROS detoxification via regulating expression of antioxidant genes under drought conditions

Plants under abiotic stress generate various kinds of ROS molecules [36]. ROS concentrations is frequently quantified by measuring the amount of Malondialdehyde (MDA) and relative electrolytic leakage [37]. MDA content and relative electrolytic leakage of the leaves from wild-type plants and two sapk3 mutant lines were compared. After drought treatment, two sapk3 mutant lines accumulated more MDA than the wild-type plants (Fig. 5 C). Moreover, relative electrolytic leakage of two sapk3 mutant lines was higher than that in wild-type plants (Fig. 5 D). Given the increased ROS-induced damage of sapk3 mutants, the activities of catalase (CAT), superoxide dismutase (SOD), peroxidase (POD) and the content ascorbic acid (AsA) from wild-type plants and two sapk3 mutant lines were also measured. Compared with wild-type plants, two sapk3 mutant lines showed lower CAT, SOD, POD activities and AsA content than wild-type plants under drought conditions (Fig. 5 F-I).

To determine whether OsSAPK3 is involved in ROS-scavenging, the expression levels of several antioxidant genes including OsCAT, OsSOD1, OsSOD2 and ascorbic acid peroxidase 2 (OsAPX2) were assessed. The expression levels of these genes were significantly lower in two sapk3 mutant lines than that in wild-type plants (Fig. 6A-D). These results demonstrated that OsSAPK3 was also important for improving ROS tolerance by regulating ROS detoxification.

Fig. 6
figure 6

The expression of abiotic stress-responsive genes. A-I Relative expression levels of OsCAT, OsSOD1, OsSOD2, OsAPX2, OsTRAB1, OsbZIP23, OsbZIP46, OsABI5 and OsDREB2A from the seedings of 2-week-old wild-type plants and two sapk3 mutant lines before and after drought stress. Values represent the means ± SD of three biological replicates (5 plants for each replicate). **P < 0.01 (Student’s t-test)

Full size image

OsSAPK3 involved in drought stress response in ABA dependent and independent manner

In order to explore the possible mechanism of OsSAPK3 in the regulation of drought stress, the expression of 5 well-characterized drought resistance-related genes (i.e., OsTRAB1 [38], OsABI5 [39], OsbZIP23 [34], OsbZIP46 [33] and OsDREB2A [40]) were analyzed. Among these genes, OsTRAB1, OsABI5, OsbZIP23 and OsbZIP46 are involved in ABA dependent signaling, OsDREB2A is involved in ABA independent regulatory systems. All their expression levels were significantly lower in two sapk3 mutant lines than that in wild-type plants after drought treatment (F ig. 6E-I). These results suggested that OsSAPK3 upregulated the expression of some stress-responsive genes both in ABA-dependent and ABA-independent manner under drought stress.

OsSAPK3 improved rice yield by altering tiller numbers and grain size

In order to assess the effects of OsSAPK3 on rice growth and productivity, the wild-type plants and two sapk3 mutant lines were grown in the field. At maturity, compared to wild-type plants, plant height of two sapk3 mutant lines decreased significantly (Fig. 7 A and C). In addition to dwarf phenotype, two sapk3 mutant lines showed significantly fewer tiller numbers than wild-type plants (Fig. 7 E). However, effective tiller numbers of sapk3 mutant plants were slightly fewer than wild-type plants (Fig. 7 F). The effective tiller rate of sapk3 mutants was significantly higher than wild-type plants (Fig. 7 G). These results indicated that OsSAPK3 participated in the regulation of rice tillering.

Fig. 7
figure 7

OsSAPK3 affected growth phenotypes in rice A Growth phenotypes B Grain phenotypes C Plant height D Grain length E Tiller numbers F Effective tiller numbers G Effective tiller rate H Grain yield per plant of wild-type plants and two sapk3 mutant lines.Values represent the means ± SD of three biological replicates (25 plants for each replicate). *P < 0.05 and **P < 0.01 (Student’s t-test)

Full size image

The tiller number is one key factor determining rice grain yield. Efficient utilization of nitrogen (N) is one important factor affecting rice tillering [41]. So far, many nitrate transporter genes, such as nitrate and di/tripeptide transporters (NPFs), have been found in rice [42]. OsNPF7.1, OsNPF7.2 and OsNPF7.4 was reported to positively regulate the tiller number [41, 42]. Additionally, OsSAPK2, which is homologous to OsSAPK3, could enhance grain production by regulating nitrogen utilization efficiency [43]. The function of OsSAPK3 on rice tillering was investigated by analysing tiller buds development of wild-type plants and two sapk3 mutant lines. Tiller buds grew slower in two sapk3 mutant lines than that in wild-type plants, and this phenomenon was observed continuously for 34 days after germination (DAG) (Fig. 8 A and B). Based on these results, it could be concluded that OsSAPK3 promoted rice tiller bud outgrowth between 20 and 34 DAG. To investigate how OsSAPK3 altered the tiller buds, the expression levels of OsNPF7.1, OsNPF7.2 and OsNPF7.4 were measured. These results showed that the expression level of OsNPF7.2 in two sapk3 mutant lines was significantly lower than that in wild-type plants (Fig. 8 D). However, expression levels of OsNPF7.1 and OsNPF7.4 exhibited no significant difference between wild-type plants and two sapk3 mutant lines (Fig. 8 C and E). These results suggested that OsSAPK3 positively regulated rice tillering by promoting rice tiller bud outgrowth through regulating expression of nitrate transporter genes.

Fig. 8
figure 8

OsSAPK3 altered tiller bud outgrowth in rice A Comparison of tiller buds at 20 days after germination among wild-type plants and two sapk3 mutant lines. bar = 1 cm. B Statistical analysis at different days after germination among wild-type plants and two sapk3 mutant lines. Values represent the means ± SD of 3 biological replicates (25 plants for each replicate). C Expression levels of OsNPF7.1 D OsNPF7.2 E OsNPF7.4 F OsMIS2, G OsGSN1 and H OsGIF1 among wild-type plants and two sapk3 mutant lines. Values represent the means ± SD of three biological replicates (5 plants for each replicate). **P < 0.01 (Student’s t-test)

Full size image

On the other hand, rice yield is also estimated via panicle length, grain number, seed setting rate per panicle, grain size and grain yield per plant [1]. Analyses of panicle presented that there was no significant difference in panicle length, grain number and seed setting rate per panicle between two sapk3 mutant lines and wild-type plants (Fig. S1 A and C-E). This result suggested that OsSAPK3 might not participate in the regulation of pollen fertility. In addition, there was no significant difference in grain width and 1000-seed weight (Fig. S1 B and F). Further analyses of grain size and yield revealed that grain length and grain yield per plant of two sapk3 mutant lines were significantly lower than that of wild-type (Fig. 7 B, D and H).

Several genes related to grain size in rice have been identified, such as GRAIN SIZE MINI SEED 2 (MIS2) [44], AND NUMBER1(GSN1) [45] and GRAIN INCOMPLETE FILLING 1 (GIF1) [46]. In order to investigate how OsSAPK3 changed grain length, the expression levels of OsMIS2, OsGSN1 and OsGIF1 were measured. These results demonstrated that the expression levels of OsMIS and OsGIF1 in two sapk3 mutant lines were significantly lower than that in wild-type plants (Fig. 8 F and H). But the expression level of OsGSN1 showed no significant difference between two sapk3 mutant lines and wild-type plants (Fig. 8 G). These results suggested that OsSAPK3 regulated grain length by promoting expression levels of seed size related genes.

Discussion

OsSAPK3 confers drought tolerance in an ABA-dependent manner in rice

Plant-specific kinases, SnRK2s are considered to be major regulators in the plant responses to challenging environmental conditions [17]. OsSAPK2, member of subclass II SnRK2s, is involved in the osmotic stress response by regulating the expression of ABA dependent genes [47, 48]. These reports suggest that subclass II SnRK2s play important roles in osmotic stress signaling and promotes us to explore the potential role of OsSAPK3. However, direct experimental evidence of the role of OsSAPK3 in abiotic stress is largely lacking.

In this study, sequence alignment showed that OsSAPK3 differed dramatically from subclass III SnRK2s in terms of ABA-activated motif (marked with yellow lines) (Fig. 1 A). But it was highly similar to AtSnRK2.8 in A. thaliana and OsSAPK2 in rice (Fig. 1 B). SnRK2.8 is strongly activated by salt and mannitol, slightly activated by ABA [31]. OsSAPK2 is strongly induced by drought, NaCl and PEG, but not by ABA in rice [49]. It is found that OsSAPK3 was significantly induced by drought, NaCl and PEG stress, and weakly induced by ABA (Fig. 2 A). In addition, OsSAPK3 was highly expressed in roots and weakly in leaves (Fig. 2 B and C) which was consistent with the expression patterns of AtSnRK2.8 [31], but different from OsSAPK2 [49]. These findings suggested that OsSAPK3 might acquire functions different from subclass III SnRK2s but similar to OsSAPK2 and AtSnRK2.8 during long-term evolution.

Mizoguchi et al. report that AtSnRK2.8 play important roles in drought stress signaling in A. thaliana [32]. Here, the results demonstrated that OsSAPK3 was localized in the nucleus and cytoplasm (Fig. 2 D), which was quite similar to AtSnRK2.8 [32]. In addition, the hypersensitivity to exogenous ABA reduced in sapk3 mutants lines (Fig. 3 D-F). Further research showed that sapk3 mutants exhibited decreased drought tolerance (Fig. 4 A), as charactered by lower survival rate (Fig. 4 B) and higher water loss rate under drought conditions (Fig. 4 C). Under environmental stresses, ABA induces stomatal closure through the activation and inactivation of ion channels, such as SLAC1 and SLAC7 in A. thaliana [50, 51]. In rice, SAPK8 has been found to activate the function of OsSLAC1 [52]. SAPK2 [29] and SAPK9 [53] positively regulated the stomatal closure by up regulating the expression of OsSLAC1 and OsSLAC7. Under drought conditions, stomatal conductance in sapk3 mutant plants leaves was remarkably increased (Fig. 4E), and expression patterns of SLAC1 and SLAC7 in sapk3 mutant plants leaves were remarkably decreased compared with wide type plants (Fig. 4 F and G). These results suggested that OsSAPK3 functioned positively in drought stress response by ABA-dependent manner.

OsSAPK3 affects physiological and metabolic regulation in response to drought stress

In rice, understanding the physiological adaptation of drought resistance is very useful for develo** high-yield varieties under drought conditions. Previous studies have shown that plant growth is severely restricted by abiotic stresses, mainly due to protein denaturation, lipid peroxidation, cell homeostasis destruction, cell oxidative damage and DNA mutations caused by the accumulation of ROS [54, 55]. Lipid peroxidation induced by ROS accumulation can damage cell membranes [55]. Therefore, relative ion leakage is a good indicator to evaluate membrane stability under cold, drought or salt stress conditions in rice and wheat [56, 57]. The results in this study revealed that MDA content and relative ion leakage in sapk3 mutants were significantly higher than that in wild-type plants (Fig. 5 C and D), suggesting that OsSAPK3 played a crucial role in alleviating oxidative damage in rice.

According to previous reports, ROS detoxification plays a key role in drought and heat stress response [58, 59]. The antioxidant system that protects plants against adverse effects of ROS contains enzymatic antioxidants and non enzymatic molecules. Enzymatic antioxidants include SOD, CAT and APX [60]. AsA serves as non enzymatic antioxidant [60]. Therefore, CAT, SOD, POD activities and AsA content can be used as important indicators for drought tolerance in plants. In this study, activities of CAT, SOD, POD and AsA content of two sapk3 mutant lines were significantly lower than that in wild-type plants (Fig. 5 F-I). To understand the molecular mechanism of decreased ROS levels and higher antioxidant activities in two sapk3 mutant lines, the expression levels of OsCAT, OsSOD1, OsSOD2 and OsAPX2 were tested. In this study, the expression levels of OsCAT, OsSOD1, OsSOD2 and OsAPX2 exhibited significant downregulation in two sapk3 mutant lines (Fig. 6 A-D), which was consistent with lower CAT, SOD, POD activities and AsA content in two sapk3 mutant lines (Fig. 5 F-I). These results supported that OsSAPK3 regulated the accumulation of ROS through altering the expression of ROS-scavenging genes under drought stress. ABA not-activated SnRK2s, such as SnRK2.4 and SnRK2.10, take part in ROS production/accumulation under salt or high osmotic stress conditions in A. thaliana [61]. In rice, OsSAPK2 improved ROS detoxification by promoting the generation of ROS scavengers under salt or drought stress [33, 42]. SAPK9 decreased cellular oxidative damage by reducing ROS accumulation [53]. These results suggested that OsSAPK3 could be involved in the SnRK2-dependent ROS production and scavenging.

Early studies showed that under dehydration conditions, the accumulation of compatible osmoregulation substances such as proline and soluble sugar could certify the normal metabolic pathway by providing osmotic balance [54]. Under drought stress, plants can repair damage by accumulating proline to increase antioxidant activity [34]. Therefore, proline content can be used as an important reference for drought resistance. In transgenic rice, overexpression of OsP5CS showed a great improvement for drought tolerance [55]. SAPK2 [29] and SAPK9 [53] positively regulates drought or salt stress tolerance by increasing the osmotic adjustment and stomatal closure in rice. To test if OsSAPK3 plays a role in proline and soluble sugar accumulation under drought stress, proline and soluble sugar contents were examined. Under drought condition, contents of proline and soluble sugar in two sapk3 mutant lines were significantly reduced compared with wild-type plants (Fig. 5 A and B). Along with this, relative expression of OsP5CS was significantly reduced in two sapk3 mutant lines (Fig. 5 E). These results implied that OsSAPK3 functioned positively by influencing accumulation of compatible osmolytes in drought stress response.

Subclass III SnRK2s phosphorylate and activate several transcription factors, such as ABA-responsive element binding factor (ABF/AREB) transcription factors OsTRAB1 (OsbZIP66) [38], OsABI5 [39], OsbZIP23 [34] and OsbZIP46 [33]. Dehydration responsive element-binding protein 2A (OsDREB2A) mediates transcriptional changes to acquire stress resistance in ABA-independent pathways [40]. In this study, the transcript levels of OsTRAB1, OsABI5, OsbZIP23 and OsbZIP46 were all significantly lower in two sapk3 mutant lines than in wild-type plants under drought conditions (Fig. 6 E–H). In addition, the analyses of OsDREB2A expression level showed that there was also significantly decreased in two sapk3 mutant lines than in wild-type plants under drought conditions (Fig. 6I). As ABA-responsive bZIP transcription factors, OsbZIP23 and OsbZIP46 can be activated by ABA-dependent SnRKs including SAPK2 [29], SAPK6 [29] and SAPK9 [53]. These results suggested that OsSAPK3 had several substrates in both ABA-dependent and ABA-independent signaling.

In total, OsSAPK3 might play an important role in both ABA-dependent and ABA-independent drought stress signaling pathways.

OsSAPK3 improved rice yield by altering tiller numbers and grain length

It has reported that overexpression of SRK2C/SnRK2.8 enhances plant growth in A. thaliana, which may due to the phosphorylation of enzymes involved in metabolic processes [31]. SAPK9 increases rice yield by influencing panicle weight and spikelet fertility [53]. Under drought stress in reproductive period, sapk2 mutant showed dwarf phenotype, with decreased grain number per panicle and grain yield per plant [40]. To gain more comprehensively understanding of OsSAPK3, plants height between wild-type plants and two sapk3 mutant lines was evaluated. The plant height of two sapk3 mutant lines decreased significantly (Fig. 7 A and C). These results implied that OsSAPK3 had a similar function to AtSnRK2.8 or OsSAPK2 in regulating plant growth.

Previous studies have reported that effective tillers per plant is an essential determinant of rice yield, which is affected by genetic and environmental factors [62]. Concisely, the first and second exons of SAPK2 were selected for guide RNA design. Double-strand DNA generated by annealing the oligo pairs, and then was cloned into the pYLCRISPR/Cas9Pubi-H vector. Wild-type rice plants (Oryza sativa L.) was used for transformation. Then transgenetic seedlings were kept in growth chamber at 28◦C under long-day conditions (14 h light/10 h dark cycles). For mutation detection, genomic DNA extracted from mutant seedlings (all plant) were used for PCR. Then PCR products (sequence is in Supplementary Table 3) were identified by comparing the 19-bp gRNA target sequences (gtttcgagggggccaatgg, gattgatgaaaatgtgcag) to the rice reference genome (sequence is in Supplementary Table 3).

Based on mutation detection results, we selected two independent homozygous mutant lines in the T1 generation, which are named sapk3-1 and sapk3-2. The primers used for CRISPR/Cas9 (U6a-SAPK3-F, U6a-SAPK3-R, U3-SAPK3-F, U3-SAPK3-R) and mutation detection (Cas9-SAPK3-F and Cas9-SAPK3-R) were listed in the Supplementary Table 3. Nucleotide sequence of SAPK3 CDS and predicted amino acid sequences in different genotypes were listed in the Supplementary Table 2.

Phylogenetic tree and alignment

Sequences of selected SnRK2II and SnRK2III members in A. thaliana (At), Oryza sativa (Os) and Zea mays (Zm) were aligned with CLUSTALX (Sequences were listed in the Supplementary Table 1) [16, 63]. The phylogenic tree was constructed by MEGA 5.0. The reliability for the internal branch was evaluated by the bootstrap with 1000 bootstrap replicates and marked above the nodes.

RNA extraction and qRT-PCR analyses

To detect the transcript level of target genes under different stresses, wild-type plants were kept in pot with same soil in growth chamber at 28◦C under long-day conditions (14 h light/10 h dark cycles) for 2 weeks. Two weeks old seedlings were treated with 100 μM ABA, 175 mM NaCl or 20% PEG 6000 (m/v) for 24 h. For drought stress, two weeks old seedlings were exposed to the air without water supply for 24 h. Leaves were sampled every 3 h.

For the qRT-PCR analysis, total RNA was isolated from rice seedlings using the TriZol reagent (Invitrogen). The cDNAs were obtained by using Superscript II in accordance with manufacturer’s instructions (Invitrogen). The qRT-PCR analysis was performed using SYBR Premix Ex Taq kit (Takara). The primers used in qRT-PCR analysis were listed in Supplementary Table 3.

GUS Staining of OsSAPK3

For GUS reporter analysis, OsSAPK3 promoter was amplified using primers Promoter-SAPK3-F and Promoter-SAPK3-R (Supplementary Table 3). The fused ProSAPK3-GUS was cloned into the P1300 vector. For the GUS staining analysis, leaves, stems and roots were destained with pure ethanol and then examined. The GUS activity was detected using the GUS blue kit (Huayueyang, Bei**g, China) according to the manufacturer’s manual.

Subcellular localization

For the transient expression assay in rice protoplasts, the full-length OsSAPK3 CDS sequences were inserted into the 35S: GFP (PAN580) empty vector to generate the green fluorescent protein (GFP) recombinant vector 35S-SAPK3-GFP (pAN580). The 35S-GFP empty vector and the 35S-SAPK3-GFP recombinant vector were inserted into separate rice protoplasts as previously described [64]. Fluorescent signals were detected using the Zeiss LSM 710 laser scanning confocal microscope.

Determination of metabolites

Water loss rate, relative ion leakage, the content of proline, soluble sugar and malondialdehyde (MDA) were checked following the method as previously described [49].

For water loss rate, flag leaves were detached and left on the laboratory bench at room temperature. Then weighed every 20 min. Water loss rate was expressed as a percentage of initial fresh weight.

For relative ion leakage, About 1 g flag leaves separated from different lines were cut into 5 mm length and placed in test tubes containing 10 ml deionized water. The tubes were covered with plastic caps and placed in a water bath maintained at the constant temperature of 22 °C for 2 h. The conductance of H2O was measured by conductivity meter (HORIBA TWIN COND B-173).

For the content of proline, about 0.5 g dried leaf segments from different lines were ground into powder with liquid nitrogen, and then homogenized with 10 ml of 3% sulphosalicylic acid in tube. Collected the supernatant after centrifuging for 20 min (3000 × g). 2 ml of supernatant was reacted with 2 ml acid ninhydrin and 2 ml glacialacetic acid in a test tube at 100◦C for 1 h. Then cooled on ice. The absorbance at 520 nm was measured by spectrophotometer.

Total soluble sugar content in leaves was determined using anthrone reagent. Approximately 0.5 g dried leaf segments from different lines were ground into powder with liquid nitrogen, and then homogenized with 2 ml 80% ethanol in shaker at 200 rpm for 1 h. Following centrifuge at 6,000 × g for 10 min, and then collected as much supernatant as possible. Added equal volume of chloroform, completely mix, and then centrifuged at 12,000 × g for 10 min. The aqueous part was transferred to a new tube, 50 μl of each was mixed with 4.95 ml anthrone reagent and then boiled for 15 min. Measured the optical density of glucose standards at 620 nm by spectrophotometer.

For the content of MDA, about 1 g leaf segments from different lines were homogenized in 10 ml of 10% trichloroacetic (v/v) and centrifuged at 5,000 × g for 10 min. Following 2 ml of supernatant was reacted with 2 ml thiobarbituric acid in a test tube at 100◦C for 15 min, quickly cooled on ice, and the absorbance at 532 nm was measured by spectrophotometer.

The activities of SOD, POD, CAT and AsA content were detected using copper-zinc superoxide dismutase (CuZn-SOD) assay kit (A001-4–1), peroxidase assay kit (A084-3–1) catalase (CAT) assay kit (A007-1–1) and Vitamin C assay kit (A009-1–1) (http://www.njjcbio.com/, Nan**g Jiancheng Bioengineering Institute) according to the manufacturer’s manual.

Two-week-old seedlings of two sapk3 mutant lines and wide type plants withheld water for 7 days. According to the manual, leaf segments (0.5 g) from control and dehydration lines were homogenized using a chilled mortar.

For total SOD activity, 0.5 g leaf segments were ground into homogenate with 0.2 ml phosphate buffer solution (0.1 mol/L pH7-7.4), and then homogenized with 0.2 ml reagent No. 7. Centrifuged at 4,000 × g for 15 min. Then collected supernatant for Cu/Zn SOD determination. 2 ml of supernatant was reacted with 0.1 ml reagent No. 1, No. 2, No. 3 and No. 4 in a test tube at 37◦C for 40 min. Then reacted with 2 ml chromogenic agent for 10 min at room temperature. Measured the absorbance at 550 nm.

For POD activity, 0.5 g leaf segments were ground into homogenate with 0.2 ml phosphate buffer solution (0.1 mol/L pH7-7.4). Then collected the supernatant after centrifuging for 10 min (3500 × g). 0.1 ml of supernatant was reacted with 2.4 ml reagent No. 1, 0.3 ml reagent No. 2 and 0.2 ml reagent No. 3 in a test tube at 37◦C for 30 min. Then reacted with reagent No. 4 for 10 min at 37◦C. Collected the supernatant after centrifuging for 10 min (3500 × g). Measure the absorbance at 420 nm.

For CAT activity, 0.5 g leaf segments were ground into homogenate with 4.5 ml phosphate buffer solution (0.1 mol/L pH7-7.4). Then collected the supernatant after centrifuging for 10 min (2000 × g). 0.05 ml supernatant was reacted with 1 ml reagent No. 1 and 0.1 ml reagent No. 2 in a test tube at 37◦C for 1 min. Then mixed with 1 ml reagent No. 3 and 0.1 ml reagent No. 4. Measured the absorbance at 405 nm.

For the content of ascorbic acid (AsA), 0.5 g leaf segments were ground into homogenate with 5 mL of 5% (w/v) m-phosphoric acid. Following centrifugation at 10,000 × g for 15 min at 4 °C, the supernatant was used for the determination of AsA. Following 2 ml of supernatant was reacted with specified dosage of R1, R2, R3 and R4 (According to the manual) in a test tube at 22 ◦C for 30 min, quickly cooled on ice, and the absorbance at 536 nm was measured.

Stomatal number and stomatal conductance analysis

Leaves from two weeks wild-type plants and two sapk3 mutant lines were fixed with 2.5% glutaraldehyde and observed by scanning electron microscopy (S-3400 N, Hitachi, Japan). The second leaves of two weeks rice plants were selected for the stomatal conductance analysis, which was performed using the LI-6400XT Portable Photosynthesis System (LI-COR, USA) before and after 7 days exposure to drought stress.

Statistical analyses

Three independent experiments were conducted for drought stress analysis and gene expression analyses. Three biological replicates were performed for each experiment. For agronomic traits analyses, 25 plants of wild-type plants or two sapk3 mutant lines were randomly chosen to detect agronomic traits. Excel 2010 was used for making charts. All results are presented as means ± standard derivation (SD) of three biological replicates. Statistically significant difference analysis was conducted by “Student’s t-test” using SPSS statistics software.