2.1 AtCIPK20 is responsive to dehydration and encodes a cMT-localised protein

To investigate the responsiveness of AtCIPK20 to drought stress, we utilised qRT-PCR to monitor the expression changes of AtCIPK20 throughout the dehydration process of detached leaves (Figure 1a). The result revealed a significant upregulation of AtCIPK20 transcripts at 4, 6, and 8 h of dehydration compared to the control (0 h) (Figure 1a). At the 6-h mark of dehydration, the expression level of AtCIPK20 peaked, displaying a 10-fold increase relative to the control (Figure 1a). Given the pivotal role of ABA signals in orchestrating plant responses to drought stress (Raghavendra et al., 2010; J. K. Zhu, 2016), we further explored the expression pattern of AtCIPK20 in response to exogenous ABA treatment in seedlings (Figure 1b). Our results demonstrated a significant induction of AtCIPK20 expression at 2, 4, and 8 h post ABA treatment, followed by a decrease at 12 and 24 h post ABA treatment (Figure 1b). These results indicated that AtCIPK20 was responsive to drought stress and was downstream of ABA signalling pathway.

To gain insights into the biological role of AtCIPK20, we performed a subcellular localisation analysis of its encoded protein. We expressed the AtCIPK20-eYFP fusion protein in Arabidopsis mesophyll protoplasts and AtCIPK20-GFP in Nicotiana benthamiana leaves, and observed the fluorescence signals using confocal microscopy. Interestingly, contrast to the diffuse distribution throughout the entire cell of previously reported CIPK members (Batistič et al., 2010; Cui et al., 2018; X. Deng, Zhou, et al., 2013; Ma et al., 2019, 2020), AtCIPK20-eYFP and -GFP both exhibited a filamentous distribution that resembled the cytoskeleton within cells (Figure S1). To confirm this, we co-expressed the C-terminal domain of a cMT-localised protein ABS6 (Li et al., 2021) fused with RFP (ABS6_C-RFP) to label the cMT, or an 17-amino-acid peptide Lifeact (Riedl et al., 2008) fused with RFP (Lifeact-RFP) to lable the actin filaments (F-actin), respectively with the AtCIPK20-eYFP in protoplasts (Figure 1c). Using confocal microscopy and Z-stack imaging, we found a co-localisation between AtCIPK20-eYFP and ABS6_C-RFP, but not with RFP and Lifeact-RFP, indicating the cMT-association of AtCIPK20 (Figure 1c). To further verify the cMT localisation of AtCIPK20, we treated protoplasts expressing AtCIPK20-eYFP with the cMT depolymerising agent Oryzalin. Untreated cells showed filamentous distributions of both AtCIPK20-eYFP and ABS6_C-RFP, with co-localisation (Figure 1d). Conversely, in cells treated with Oryzalin, the characteristic filamentous arrangement was disrupted (Figure 1d).

To investigate whether AtCIPK20 protein possesses MT-binding activity, an MT binding spin-down assay was conducted (Figure 1e–f). The results revealed that, following ultracentrifugation, MBP was exclusively detected in the supernatant and did not co-precipitate with in vitro assembled MTs (Figure 1e). In contrast, a fraction of MBP-AtCIPK20 co-precipitated with the assembled MTs (Figure 1f). This finding, along with subcellular localisation analysis (Figure 1c), imply that AtCIPK20 encodes a cMT-associated protein.

The online tool of NCBI Conserved Domain Database (CDD) predicted that AtCIPK20 comprises two distinct domains, namely the N-terminal kinase domain (KD) and the C-terminal regulatory domain (RD) (Figure 1g and Figure S2). This structural arrangement, featuring both domains, aligns with the pattern observed in previously characterised members of the CIPK family (Batistič et al., 2010; Cui et al., 2018; X. Deng, Zhou, et al., 2013; Ma et al., 2019, 2020). To determine the cMT localisation characteristics conferred by specific domains of AtCIPK20, we fused the N-terminal segment containing the kinase domain (AtCIPK20_N-eYFP) and the C-terminal region containing the regulatory domain (eYFP-AtCIPK20_C) with eYFP for expression in protoplasts (Figure 1g). The results revealed that AtCIPK20-eYFP, eYFP-AtCIPK20, and eYFP-AtCIPK20 all displayed filamentous distributions within the cells and showed significant colocalization with the ABS6_C-RFP signal (Figure 1h). In contrast, AtCIPK20_N-eYFP exhibited a diffuse distribution, with correlation analysis indicating a lower degree of colocalization with ABS6_C-RFP compared to AtCIPK20-eYFP, eYFP-AtCIPK20, and eYFP-AtCIPK20 (Figure 1h). These results demonstrate that AtCIPK20 localises to cMT and largely relies on its C-terminal Regulatory domain.

2.2 AtCIPK20 plays a positive role in Arabidopsis response to drought

To investigate the involvement of AtCIPK20 in Arabidopsis drought response, we employed a T-DNA insertion mutant (SALK_040637C) (atcipk20, Figure S3). The T-DNA insertion occurs at the junction between the AtCIPK20 promoter and the 5′UTR (Figure S3A). The RT-PCR and qRT-PCR results indicate that the expression of AtCIPK20 is significantly reduced in atcipk20 compared to wild-type (Col-0), suggesting that this mutant is a knock-down variant (Figure S3B). Phenotypic analysis of the drought response revealed that atcipk20 leaves exhibited significantly exacerbated wilting after 14 days of drought treatment compared to those of Col-0 (Figure 2a). Upon rehydration, the survival rate of atcipk20 mutant was strikingly low at 32.1%, significantly lower than that of Col-0 (83.9%) (Figure 2b). Subsequently, genetic complementation was performed by introducing the AtCIPK20 coding sequence into the atcipk20 background, generating two independent complementation lines (Com#1 and Com#2). The expression of AtCIPK20 in the complementation lines was confirmed using RT-PCR and qRT-PCR (Figure S3B). Under drought conditions, and in terms of survival upon rehydration, the reintroduction of AtCIPK20 effectively restored the drought sensitivity of the atcipk20 mutant to levels comparable to those of Col-0 (Figure 2a, b). Furthermore, two overexpression lines of AtCIPK20 driven by the CaMV35S promoter were established in the Col-0 background (OEAtCIPK20#1 and OEAtCIPK20#2). RT-PCR and qRT-PCR results confirmed significantly elevated expression levels of AtCIPK20 in both overexpression lines compared to Col-0 (Figure S4). Drought phenotype assessment demonstrated that following an 18-day drought treatment, the majority of leaves of the OEAtCIPK20 plants retained their green colour, in contrast to the yellowing leaves of Col-0 (Figure 2c). The survival rate of the OEAtCIPK20 lines after rehydration exceeded 75%, which contrasted starkly with the 18.5% survival rate observed in Col-0 (Figure 2d). These results provide strong evidence that AtCIPK20 plays a vital role in the Arabidopsis drought response, exerting a positive regulatory influence on the plant ability to withstand drought conditions.

2.3 AtCIPK20 promotes the ABA-mediated stomatal closure

Detached leaf water loss rates were measured in the atcipk20 mutant and two OEAtCIPK20 overexpression lines. The results revealed that water loss from atcipk20 rosette leaves was significantly faster than that from the wild-type Col-0 (Figure 3a). Conversely, leaves from both OEAtCIPK20#1 and OEAtCIPK20#2 lines exhibited significantly slower water loss rates compared to those of Col-0 (Figure 3a). Stomatal aperture plays a pivotal role in controlling leaf water loss (Bauer et al., 2013). Under normal growth conditions, Col-0, atcipk20, OEAtCIPK20#1, and OEAtCIPK20#2 maintained opening stomatal apertures, with no notable differences among the different genotypes (Figure 3b, c). After dehydration for 2 h, the stomatal aperture of Col-0 significantly decreased compared to normal conditions (Figure 3b, c). Although dehydration led to some stomatal closure in atcipk20 leaves, their aperture remained significantly wider than that of wild type Col-0 leaves (Figure 3b, c). In contrast, AtCIPK20 over-expressor exhibited greater degrees of stomatal closure than wild type Col-0 after dehydration (Figure 3b, c). These results indicated AtCIPK20 positively regulated the stomatal closure and involved in drought response.

Given the subcellular localisation assay suggests that AtCIPK20 may function on cMT (Figure 1), and previous studies have reported that ABA positively regulates cMT depolymerization in guard cells during drought-induced stomatal closure (Assmann, 2003; Dou et al., 2021), we hypothesised that AtCIPK20 might operate downstream of ABA signalling in the drought response. Since sensitivity to exogenous ABA can reflect the seedling's response to drought (Assmann, 2003), we investigated the sensitivity of atcipk20 and OEAtCIPK20 genotypes to ABA. The results revealed a notably higher germination rate for atcipk20 compared to Col-0 on ABA-containing medium (Figure 3d, e). In contrast, no significant difference between the germinations of Col-0 and atcipk20 was observed on ABA-free medium (Figure 3d, e). Further analysis was conducted on seedlings with fully developed green cotyledons (SGC) to measure primary root lengths. In the absence of ABA, primary root lengths of atcipk20 seedlings were comparable to those of Col-0. However, in the presence of ABA, atcipk20 seedlings exhibited a significantly reduced primary root length compared to that of Col-0 (Figure 3f, g). On the other hand, the germination rate of OEAtCIPK20 lines showed a substantial decrease relative to Col-0 on ABA-containing medium (Figure 3h, i). Moreover, the inhibitory impact of ABA on primary root growth was significantly more pronounced in the OEAtCIPK20 lines compared to that in Col-0 (Figure 3j, k). At an ABA concentration of 200 nM, the primary root growth of both OEAtCIPK20 lines was almost entirely arrested (Figure 3j, k).

Given the role of AtCIPK20 in positively regulating plant drought resistance and simultaneously negatively influencing leaf water loss, stomatal aperture during dehydration, and ABA sensitivity (Figures 2 and 3a–k), the investigation proceeded to explore the stomatal aperture of Col-0, atcipk20, and two OEAtCIPK20 lines in response to exogenous ABA. After treatment with a stomatal opening solution, the stomata of all genotypes fully opened, displaying no difference among the various genotypes (Figure 3l, m). Subsequently, rosette leaves, pretreated with the stomatal opening solution, underwent ABA exposure to induce stomatal closure. After ABA application, the atcipk20 mutant showed a wider stomatal aperture compared to Col-0, whereas the OEAtCIPK20 lines exhibited a pronounced reduction in stomatal aperture (Figure 3l, m). These collective findings suggest that AtCIPK20 plays a regulatory role in the ABA-mediated stomatal closure process under drought stress.

2.4 AtCIPK20 positively regulates the ABA-mediated cMT depolymerization in guard cells

AtCIPK20 protein associated to cMT (Figure 1), and recent reports suggested that cMT depolymerization processes play a crucial role in regulating stomatal closure (Dou et al., 2021). To test whether AtCIPK20 is involved in cMT depolymerization within guard cells, we crossed TUB6-GFP line with cipk20 mutant and AtCIPK20 overexpressing line (OEAtCIPK20#2) to observe cMT stability. Confocal microscopy revealed that under normal growth conditions, cMT in guard cells from wild-type (Col-0), atcipk20 mutant, and OEAtCIPK20#2 plants exhibited well-organised radial filaments (Figure 4a). However, after 2 h of dehydration treatment at room temperature, stomatal cMT in the wild type, atcipk20, and OEAtCIPK20#2 lines showed varying degrees of depolymerization (Figure 4a, b), with the cipk20 mutant displaying significantly less depolymerization than the wild type, while the OEAtCIPK20#2 line exhibited significantly greater depolymerization than the wild type (Figure 4a, b). These results indicate that AtCIPK20 controls cMT depolymerization in guard cells induced by drought stress.

Since AtCIPK20 regulates ABA-mediated stomatal closure (Figure 3), we hypothesised that it might also be involved in ABA-mediated cMT depolymerization in guard cells. As expected, stomatal cMT in the untreated wild-type and atcipk20 mutant plants displayed apparent radical alignment, with no significant difference in cMT density between the two genotypes (Figure 4c, d). The ABA treatment (20 μM for 2 h) disrupted the cMT filamentous arrangement in wild type, while the effect on cMT in atcipk20 guard cells was limited, compared to the wild type (Figure 4c, d). Conversely, ABA treatment (10 μM for 1 h) led to significantly greater cMT depolymerization in guard cells of the OEAtCIPK20 line compared to the wild type (Figure 4f, g). Furthermore, exogenously cMT depolymerising agent oryzalin could compensate for the ABA insensitivity of cMT in guard cells of atcipk20 mutant and its ABA-induced stomatal closure defect (Figure 4c–e). It has been reported that CaCl₂, downstream of the ABA signal, can induce cMT depolymerization and stomatal closure (Yu et al., 2020). We found that cMT in guard cells of atcipk20 mutant exhibited higher stability than the wild type under exogenous CaCl₂ treatment, while the stomatal closure in the atcipk20 mutant was significantly lower than that in the wild type (Figure 4c–e). These results indicate that AtCIPK20 controls ABA-mediated cMT depolymerization in stomata, positively regulating stomatal closure and thereby limiting transpiration to enhance drought resistance in Arabidopsis (Figure 4h).

2.5 AtCBL9 antagonises the cMT-associated attributes of AtCIPK20, exerting a negative regulation in the ABA-induced depolymerization of cMT in guard cells

Numerous studies have reported that the abiotic stress resistance function of CIPK relies on its physical interaction with CBL, occurring at the C-terminal regulatory domain of CIPK members (Ma et al., 2020). Our data indicated that the cMT attachment of AtCIPK20 is dependent on its C-terminal regulatory domain (Figure 1c, d), prompting us to speculate whether the cMT targeting of AtCIPK20 is associated with its interaction with a specific Arabidopsis CBL partner. Yeast two-hybrid assays identified two AtCBL family members, AtCBL1 and AtCBL9, as interactors of AtCIPK20 (Figure 5a). Notably, the β-GAL reporter assay indicated a stronger interaction between AtCBL9 and AtCIPK20 compared to interaction between AtCBL1 and AtCIPK20 (Figure 5a). Further investigation using Bimolecular Fluorescence Complementation (BiFC) revealed that the interactions between AtCIPK20 and AtCBL1/9 occurred specifically at the plasma membrane (PM) (Figure 5b and Figure S5A). Moreover, overexpression of AtCBL9-RFP or AtCBL1-RFP altered the filamentous distribution of AtCIPK20-eYFP, leading to its aggregation at the PM (Figure 5c and Figure S5B). This suggests that the interaction with AtCBL9 or AtCBL1 impairs the cMT localisation of AtCIPK20.

The qRT-PCR analysis showed that AtCBL1 and AtCBL9 respond to dehydration and ABA treatments (SFigure S6). The expression of AtCBL1 and AtCBL9 peaks at 6 h following dehydration treatment, similar to AtCIPK20 (Figure S6A). However, unlike the rapid response of AtCIPK20 to ABA (Figure 1b), AtCBL1 and AtCBL9 are initially suppressed by ABA and only significantly upregulated at the later stage of treatment (24 h) (Figure S6B).

Since AtCBL9 shows a stronger interaction with AtCIPK20 than AtCBL1 in yeast (Figure 5a), we have focused further research on AtCBL9. Subsequently, we employed a knockdown mutant variant of the AtCBL9 gene (atcbl9, SALK_142774C) (Figure S7) to investigate its biological function under drought stress conditions. Following an 18-day drought period, the atcbl9 mutant demonstrated a less severe wilting phenotype in leaves compared to the wild type (Figure 5d). Upon rehydration, atcbl9 plants exhibited superior recovery compared to the wild type, with a survival rate of 88.3%, significantly surpassing the wild type's 6.8% (Figure 5d, e). Additionally, atcbl9 mutants exhibited a slower rate of detached leaf water loss compared to the wild type (Figure 5f). The extent of ABA-induced stomatal closure in the atcbl9 mutant was significantly higher than that observed in the wild type (Figure 5g, h), and atcbl9 seedlings exhibited greater ABA sensitivity compared to the wild type (Figure S8). Under normal conditions, the cMT in guard cells of atcbl9 displayed a distinct radial arrangement similar to the wild type (Figure 5i). However, exogenous ABA application led to higher depolymerization cMT in atcbl9 guard cells, resulting in a lower cMT density compared to the wild type (Figure 5i, j). These findings suggest that AtCBL9, as an interacting partner of AtCIPK20, antagonises AtCIPK20′s cMT localisation, and negatively regulating ABA-mediated cMT depolymerization in guard cells, stomatal closure, and drought resistance in Arabidopsis.

2.6 The cMT localisation pattern of AtCIPK20 is not unique in plant CIPK family

To elucidate whether the cMT localisation of AtCIPK20 is uniquely specific or more widespread among other CIPK family members of the Arabidopsis, subcellular localisation analysis was conducted on all 26 Arabidopsis CIPKs (AtCIPK1-26). Results showed that, aside from AtCIPK20, eYFP-fused proteins of AtCIPK5, AtCIPK7, and AtCIPK9 exhibited filamentous patterns and colocalized with ABS6_C-RFP (Figure 6a). Additionally, AtCIPK4, AtCIPK17 and AtCIPK25 exhibited partial co-localisation with cMT (Figure 6a). Interestingly, AtCIPK15 displayed a punctate distribution and appears to be adhering to the cMT network (Figure 6a). Other Arabidopsis CIPK members did not exhibit cMT localisation (Figure S9). Additionally, domain truncation experiments were carried out for AtCIPK5 and AtCIPK7, and their fusion expression with eYFP was conducted in a manner similar to that of AtCIPK20 (Figure 1g). Correlation analysis of eYFP and RFP fluorescence signal intensities reveals that, similar to AtCIPK20, the C-terminal regulatory domain, rather than the N-terminal kinase domain, plays a predominant role in the microtubule localisation of AtCIPK5 and AtCIPK7 (Figure S10). To identify conserved protein structural features among CIPK members with cMT localisation, we conducted evolutionary analyses using the full-length protein sequences, the C-terminal Regulatory domain sequences, and the N-terminal Kinase domain sequences of 26 Arabidopsis CIPKs, respectively. However, the results showed that CIPK members with cMT localisation did not cluster together in any of the three phylogenetic trees (Supplemental Figure 11). Given that AtCIPK20 can directly bind to MTs in vitro (Figure 1e–f), we utilised AlphaFold 3 to predict the protein structure of AtCIPK20 and performed molecular docking with known Tubulin dimers (PBD accession: 6WVR) (Abramson et al., 2024; Debs et al., 2020; Singh et al., 2024). The results suggested that certain amino acid residues in the C-terminal Regulatory domain might be involved in the binding of AtCIPK20 to MTs (Figure S12). Unfortunately, these amino acids were not conserved among CIPK members with cMT localisation characteristics (Figure S2).

The prevalence of Arabidopsis CIPK cMT localisation leads to the hypothesis that the associations between CIPKs and cMT might extend across a broader spectrum within the plant kingdom. To confirm this hypothesis, the AtCIPK20 protein sequences were blasted in Maize (B73 inbred line) and Rice (nipponbare) genome database. In the genomes of maize and rice, there are 46 and 39 CIPK gene members, respectively, designated as ZmCIPK1-46 and OsCIPK1-39 (Figure S13). We selected the top 8 homologs with the highest amino acid sequence similarity to AtCIPK20, which are ZmCIPK35, ZmCIPK2, ZmCIPK22, ZmCIPK5 for maize, and OsCIPK16, OsCIPK31, OsCIPK20, OsCIPK8 for rice (Figure S13). Expression vectors with eYFP fusion were constructed for these CIPK members, and they were transiently expressed in the maize protoplasts or rice protoplasts. The ZmCIPK5, which shares the highest homology with AtCIPK20 in maize, displayed cMT association characteristics similar to AtCIPK20, with exhibiting co-localisation with ABS6_C-RFP when transiently expressed in maize protoplasts (Figure 6b). Additionally, ZmCIPK2-eYFP fusion protein also partially exhibited cMT localisation features (Figure 6b). However, ZmCIPK35 and ZmCIPK22 did not possess cMT localisation properties (Figure S14A). Among the four AtCIPK homologous proteins in rice, only OsCIPK20 displayed filamentous distribution similar to AtCIPK20 in rice protoplasts, colocalizing with ABS6_C-RFP (Figure 6c). OsCIPK31, OsCIPK16, and OsCIPK8, on the other hand, exhibited a diffuse distribution throughout the entire cell (Figure S14B). These results imply that there are certain CIPK members with cMT localisation characteristics in various plant species.

4.1 Plasmid constructions

For generating the AtCIPK20 over-expressing vector, the CDS of AtCIPK20 was amplified from cDNA library of Col-0 leaves. The AtCIPK20 CDS was inserted between SpeI and BstEII sites of pCAMBIA1302 vector for CaMV35S promoter driven expression (35S:AtCIPK20).

For cMT subcellular co-localisation assays, the C-terminal segment coding sequence of a reported cMT-association protein ABS6 (Li et al., 2021) was cloned from cDNA library of Col-0 rosette leaves and inserted into HindIII of pRFPLT vector for RFP fusion expression. For labelling the Filaments actin (F-actin), a 17 amino acids length peptide Lifeact (Riedl et al., 2008) which has F-actin bound capacity was fused with RFP based on the pRFPLT vector (XhoI/HindIII).

For purification of MBP-AtCIPK20 refusion protein, the full length of AtCIPK20 CDS was amplified and inserted into site between BamHI and SalI of pMAL-C2x vector.

For subcellular localisation of CIPKs, the coding regions of Arabidopsis CIPK members (AtCIPK1-26) were amplified from cDNA library of Col-0 rosette leaves by RT-PCR, and ligated into HindIII/MluI sites of pEYFPLT vector. The full-length coding sequences of AtCIPK5, AtCIPK7, and AtCIPK20, or segments containing the Regulatory domain in their C-termini, were fused to the EcoRI site of the pEYFPLT vector to be fused to the C-terminus of eYFP. The segments containing the kinase domain in the N-termini of AtCIPK5, AtCIPK7, and AtCIPK20 were connected into the HindIII/MluI sites of the pEYFPLT vector to be fused to the N-terminus of eYFP. The orthologs of AtCIPK20, ZmCIPK5 (Zm00001d008901), ZmCIPK22 (Zm00001d038048), ZmCIPK2 (Zm00001d022450) in zea mays, and OsCIPK8 (LOC_Os01g10890), OsCIPK31 (LOC_Os07g48100), OsCIPK20 (LOC_Os05g26820) and OsCIPK26 (LOC_Os03g22050) in Oryza sativa, whose CDS were amplified from the B73 inbred line and nipponbare genomic DNA by PCR, respectively, due to no intron in the coding regions of these genes. Subsequently, these coding sequences were ligated into HindIII/MluI sites of pEYFPLT vector for CaMV35S promoter driven eYFP-fused expression.

For Yeast two hybrids, the AtCIPK20 coding sequence was cloned onto pGBK-T7 vector (EcoRI/BamHI) to construct the GAL4 BD-AtCIPK20 fusion expression vector. The AtCBL1-10 coding regions were amplified from cDNA library of Col-0 rosette leaves, and ligated into pGAD-T7 vector (XhoI/HindIII) to construct the GAL4 AD-AtCBLs fusion expression vectors. For co-localisation assays of AtCIPK20 with its AtCBL partners, the AtCBL1 and AtCBL9 CDS regions were cloned into HindIII/MluI sites of pRFPLT vector.

The primers sequences used in vector constructs in this study were listed in Table S1.

4.2 Plant materials

For generation of AtCIPK20 over-expressing transgenic lines (OEAtCIPK20), the 35S:AtCIPK20 vector was transformed into Agrobacterium tumefaciens strain GV3101. The transformations into A. thaliana ecotype Columbia (Col-0) were performed according to the floral dip method (Clough & Bent, 1998). The transgenic plants were selected by growth on a 1/2 MS plates containing 25 mg/L of hygromycin; The atcipk20 (SALK_040637C) and atcbl9 (SALK_142774C) T-DNA insertion mutants were obtained from AraShare (https://www.arashare.cn/index/). The TUB6-GFP line was offered by Prof. Xiayan Liu of College of Life Sciences, Northwest A&F University. The atcipk20 TUB6-GFP, OEAtCIPK20 TUB6-GFP and atcbl9 TUB6-GFP lines were obtained by crossing. The Arabidopsis plants were grown under a light/dark cycle of 16 h/8 h at 150 μmol m⁻² s⁻¹.

4.3 DNA extraction and PCR identification of atcipk20 mutant

Genomic DNA isolated from the leaves of Arabidopsis, maize and rice using the CTAB method (Porebski et al., 1997). The genotype of atcipk20 mutant was identified by PCR using AtCIPK20 gene-specific and T-DNA specific primers (LP, RP and BP, Figure S3 and Table S1).

4.4 RNA extraction and quantitative RT-PCR

Total RNA extraction was carried out from 2-week-old plant rosette leaves Trizol reagent (Takara, Japan). The obtained RNA samples were subjected to the Transcriptor 1st-Strand cDNA Synthesis kit (Roche, Switzerland) as per the manufacturer's protocol to generate the cDNA library. The resulting cDNA was diluted 30-fold using ddH₂O and utilised as the template for conducting quantitative RT-PCR (qRT-PCR). For normalisation of expression levels, AtACTIN2 was employed as the internal reference. The experimental procedure was replicated thrice using independent biological samples. Amplification of AtACTIN2, AtCIPK20, AtCBL1 and AtCBL9 was achieved using the primer pairs AtACTIN2-250F × AtACTIN2-250R, AtCIPK20-RTF × AtCIPK20-RTR, CBL1-RT-F × CBL1-RT-R, and CBL9-RT-F × CBL9-RT-R (Table S1), respectively.

4.5 Drought treatment

The seeds of different genotypes were germinated on 1/2 MS medium containing 1% sucrose for 7 days under a light/dark cycle of 16 h/8 h at 150 μmol m⁻² s⁻¹. The seedlings were transplanted into pots (7 cm × 7 cm × 7.5 cm) filled with 110 g of nutrient soil for further growth. There were six pots for each genotype, and each pot contained nine seedlings. During the 2-week growth period, the water supply was strictly controlled to ensure consistent weight for each pot. Then, water was withheld as a drought treatment. After 14 days of drought (for Col-0 and atcipk20) and 18 days of drought (for Col-0, OEAtCIPK20 and atcbl9), watering was resumed to allow plants to recover and the number of surviving plants was recorded 3 days later.

4.6 Determination of water loss

For the leaf water loss assay, the rosette leaves from 2-week-old plants of different genotypes were collected (0.5 g). The detached leaf tissue was weighed immediately and then placed on filter paper at room temperature (25°C). The leaves tissue was subsequently weighed at specified intervals (15, 45, 75, 105, 135, 165, and 245 min). The percentage of water loss was calculated using the formula [(initial weight - final weight)/(initial weight)] × 100.

4.7 Stomatal aperture assay

For the stomatal aperture measurements, rosette leaves from 2-week-old plants of different genotypes were collected. Nail polish was applied to the abaxial surface of both freshly harvested leaves and leaves dehydrated for 2 h to preserve the stomatal status. After the nail polish solidified, it was peeled off and placed onto glass slides for stomatal density and aperture analysis. Stomatal apertures were observed for 80 stomata in 8 fields per genotype using a microscope (Nikon, Japan). For measurement of stomatal aperture response to ABA, the rosette leaves from different genotypes plants (2-weeks-old) were pretreated with a stomatal opening buffer (10 mM MES, 50 KCl, 0.1 mM Ca₂Cl and 1% sucrose, pH = 6.15) 2 h, and consequently treated with same solution containing 5 μM ABA for 30 min. The ratio of width/length of the inner edges of the guard cells was quantified as an indicator of stomatal aperture using ImageJ software.

4.8 ABA sensitivity assay

The seeds of different genotype lines were germinated on 1/2 MS medium containing 1% sucrose and different concentrations of ABA (0, 150, and 300 nM for Col-0 and atcipk20; 0, 100 and 200 nM for Col-0, atcbl9 and AtCIPK20 over-expressing lines). After germination for 7 days under a light/dark cycle of 16 h/8 h at 150 μmol m⁻² s⁻¹, the seedlings with green cotyledons (SGC) were counted. The experiments were performed with three biological replicates, each replicate consisting of the 72 seeds of each genotype. The SGC were sequentially measured for its primary root length using the imageJ software.

4.9 Cloning of AtCIPK20 orthologs in maize and rice

For identification of orthologs of AtCIPK20 in maize and rice, the AtCIPK20 amino acid sequence was blasted in maize inbred line B73 and nipponbare reference Genomic database (https://www.maizegdb.org/ and http://rice.uga.edu/). The four genes with the highest homology to AtCIPK20 in maize or rice were cloned for analysis of cMT co-localisation.

4.10 Subcellular localisation

Fluorescent protein fusion expression vectors were extracted and purified, and their concentrations were adjusted to 2 mg/ml. Plasmid purification was performed using the PEG-8000 method (K. Zhu et al., 2005). The recombinant vectors were transformed into Arabidopsis, maize and rice Mesophyll protoplasts for transient expression. The Arabidopsis, maize and rice protoplasts preparation and transformation were performed according to the published methods (Armstrong et al., 1990; F. He et al., 2016; Yoo et al., 2007). After 12 h of culture, eYFP and RFP fluorescence signals were observed by Z-axis scanning with a laser confocal microscope (Nikon A1R, Japan). Co-localisation analysis was performed using ImageJ software to quantify the relative fluorescence density.

4.11 MT binding protein spin-down assays

The expression vectors for MBP and MBP-AtCIPK20 fusion were transformed into Escherichia coli strain BL21 (DE3). Protein expression was induced by 0.05 μM IPTG. The bacterial cells were collected by centrifugation and lysed mechanically to obtain crude protein extracts for purification. Protein purification was performed using MBPSep Dextrin Agarose Resin 6FF (Cat #: 20515ES08, Yeasen Biotechnology (Shanghai) Co., Ltd.) following the manufacturer's instructions.

For the MT binding spin-down assays, we utilised a Microtubule Binding Protein Spin-down Assay kit (Cat. #: BK029, Cytoskeleton, Inc) to assemble MTs in vitro according to the provided protocol. The purified MBP and MBP-AtCIPK20 proteins were initially centrifuged at 100,000 × g for 30 min to remove insoluble material. Subsequently, 7.5 μg of MBP and MBP-AtCIPK20 were incubated with 2×10¹¹ MT/ml (prepared as per the kit instructions) in a 50 μL reaction volume at room temperature for 30 min. After incubation, the mixtures were subjected to ultracentrifugation at 100,000 × g for 40 min at room temperature to separate the supernatant and pellet. The pellet was resuspended in 50 μL of 1× Laemmli buffer. The samples analysed by SDS-PAGE and immunoblotting. The proteins in SDS-PAGE gels were detected by silver staining and immunoblotting. The immunoblotting was performed using commercial MBP antibody (Cat #: AE016, ABclonal Technology Co., Ltd.) and β-Tubulin antibody (Cat #: ABP0128, Abbkine Scientific Co., Ltd).

4.12 Yeast two hybrids

The GAL4 BD-AtCIPK20 and GAL4 AD-AtCBLs yeast expression vectors were co-transformed into the yeast strain AH109. The transformants were selected on solid SD medium plates lacking leucine and tryptophan (SD-LW). The resulting colonies were then amplified in liquid SD-LW medium (for 24 h, 28°C, in the dark). The yeast cells were subsequently collected by centrifugation (2500 × g, 4°C for 3 min) and resuspended in sterile ddH₂O to OD_600nm = 0.15. The 3 μL suspension of the yeast cells was spotted onto both SD-LW plates and SD medium plates lacking leucine, tryptophan, adenine, and histidine (SD-LWHA). β-galactosidase filter assays were conducted according to the protocol outlined in the Yeast Protocols Handbook (Clontech).

4.13 cMT measurement in guard cells

The cMT in guard cells of Col-0 TUB6-GFP, atcipk20 TUB6-GFP, OEAtCIPK20 TUB6-GFP, and atcbl9 TUB6-GFP lines were visualised using a confocal laser scanning microscope (Nikon A1R, Japan). The GFP fluorescence was excited at 488 nm, and GFP signals in the lower epidermal stomata of the leaves were observed using a 100 × oil-immersion objective and recorded. For the preprocessing of quantitative cMT density assessments, maximum-intensity projections were generated from consecutive optical sections. The resulting images were then transformed into binary format using thresholding techniques and subjected to skeletonization using the 'Process-Binary-Skeletonize' function in ImageJ. The cMT density was determined by measuring the occupancy of GFP signal within guard cells as previously reported (Dou et al., 2021). All analyses were performed using the original 8-bit raw scanning images.