Seeds of ‘TM-1’ and ‘CRI50’ were used in current investigation. TM-1 is widely used as a genetic standard (Zhang et al., 2015), and CRI50 is a commercial Chinese cotton cultivar with high yield and stress tolerance (Li et al., 2021). Cotton seeds were delinted with H₂SO₄ (98%) and rinsed in water, then soaked in distilled water for 1 day and followed by germination on wet gauzes for another day at 25°C. Nicotiana benthamiana RA-4 accession was used for subcellular location experiment. Germinant cotton seeds and Nicotiana benthamiana were transferred to pots containing vermiculite in the greenhouse at 28/20°C under 16-h light/8-h dark photoperiod (60% relative humidity). Arabidopsis thaliana ecotype Columbia-0 (Col-0) was used for transgenic analysis. The Arabidopsis thaliana and all transgenic plants were planted in the growth chamber at a temperature regime of 23/20°C with 16-h light/8-h dark cycle (50% relative humidity).

For drought treatment, ‘TM-1’ were used to water withholding at the trefoil stage of seedlings for two weeks, and rewatering for three days. A no treatment control was always included. Transcript levels were detected in cotton seedling. Hormone treatment were performed at the trefoil stage of the cotton seedlings with solutions of various substances. Cotton seedlings were sprayed evenly on the leaves with ABA solution (100 µM), MeJA solution (100 µM), IAA solution (100 µM), SA solution (100 µM), or GA solution (100 µM) for up to 24 h. The leaf samples were collected at 0, 6, 12, 24 h after treatment to assess candidate gene expression response to various hormone. These samples were frozen in liquid nitrogen immediately and stored at -80°C for RNA isolation. Three biological repeats in each treatment were performed.

34 AtCDPKs sequences were downloaded from the Arabidopsis Information resource (http://www.arabidopsis.org/), and 30 OsCDPKs sequences have been downloaded from either China Rice Data Center (https://www.ricedata.cn/gene/) or GenBank (http://ncbi.nlm.nih.gov/genbank). The G.hirsutum genome sequence was downloaded from the sequenced genome in Cottongen (https://www.cottongen.org). These CDPK proteins in model plants were set as the query in a BLASTp (https://blast.ncbi.nlm.nih.gov/Blast.cgi) search to identify the CDPK proteins in G.hirsutum. Meanwhile, to identify the GhCDPK family members more accurately, the profile hidden Markov model of the HMMER3.0 program was further applied to search all of the hits with the default parameters by utilizing EF hand domains (PF13499) and kinases (PF00069) (Finn et al., 2011). Candidate genes were obtained by combining two results and removing the duplicates ( Supplementary Table S1 ). SMART (http://smart.embl-heidelberg.de/) and PFAM (http://pfam.janelia.org/) were used to verify the presence of protein kinase domains and EF hand domains in all candidates. Motif prediction and visualization of GhCDPKs was done by MEME website (http://meme-suit/org/) and MEME Suit Wrapper of TB tools, respectively. The molecular mass of each GhCDPKs were predicted by the online tools ProtParam (http://web.expasy.org/protparam/).

Based on physical location data provided in Cottongen, the chromosomal locations of GhCDPKs were predicted and subsequently visualized using Gene Location Visualize from GTF/GFF tool of TBtools software. Multiple amino acid sequence alignment was performed using ClustalW. The phylogenetic tree was constructed by MEGA7.0 using the neighbor-joining method with 1000 bootstrap replications (Kumar et al., 2016) and then displayed with the online iTOL tool (https://itol.embl.de/tree).

The expression profile in G. hirsutum TM-1 at different tissues, including root, stem, leaf, bract, torus, pental, pistil, filament, anther, ovules and fibers, were obtained from published RNA-seq dataset reported previously (Zhang et al., 2015). The expression data were gene-wise normalized, and the expression patterns were illustrated using the MultiExperiment Viewer (MeV) software.

Total RNA was extracted from the frozen samples with three replicates using the RNA Prep Pure kit (TIANGEN Biotech) following the manufacture’s recommendations. The first-strand cDNA was synthesized from 2 μg total RNA based on a PrimeScript Reverse Transcriptase Kit (Takara). qRT-PCR was performed in a LightCycler 480 II PCR System (Mannheim, Germany) using the SYBR Premix Ex Taq (Takara) with GhACT4 (Artico et al., 2010) and AtUBQ7 (Zhang et al., 2021b) as reference genes in cotton and Arabidopsis, respectively. Relative quantification in gene expression levels were calculated based on three biological replicates referring the 2^-ΔΔCt method (Livak and Schmittgen, 2001). Specific primers used for qRT-PCR are listed in Supplementary Table S2 .

The physicochemical property analysis of GhCDPK60 were predicted by the online tools CBS (http://www.cbs.dtu.dk/services/). Cis-regulatory elements in the promoter of GhCDPK60 were predicted through online website plantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/). Subcellular location prediction of GhCDPK60 was carried out by using the WoLF PSORT website (https://wolfpsort.hgc.jp/). In order to detect the subcellular localization of GhCDPK60 in vivo, we used tobacco for transient expression. The full-length coding region of GhCDPK60 was amplified from the G. hirsutum variety TM-1 and cloned into the pCAMBIA1305-GFP vector driven by CaMY35S promoter to generate GFP-fusion protein. Primer pairs were listed in Supplementary Table S2 . Subsequently, the plasmids were introduced into Agrobacterium tumefaciens strain GV3101 and then infiltrated 3-4-week-old Nicotiana benthamiana leaves epidermis. The signal of GFP was observed after 48 h using a laser confocal scanning microscope (LSM 880, Zeiss, Germany). The 35S-OsAlaAT1-mCherry (Yang et al., 2015; Fang et al., 2022) and 35S-mCherry-SYP132 (Xia et al., 2019) were used as plant cytoplasm and plasma membrane marker for the colocalization experiments, respectively. Excitation wavelength used in 488 nm for GFP, and the wavelength range of captured light at 515-555 nm. The excitation wavelength and gain wavelength of mCherry were 555 nm and 580-630 nm, respectively.

The coding sequence of GhCDPK60 was amplified and cloned into a pBI121 vector driven by CaMY35S promoter. Primer pairs were listed in Supplementary Table S2 . The recombinant plasmid was transformed into Arabidopsis thaliana (Columbia-0, Col-0) through the floral dipping method (Clough and Bent, 1998), and positive transgenic plants were selected on the 1/2 Murashige and Skoog (MS) medium containing 50 mg/mL kanamycin. Gene-specific primers were used to isolate homozygous plants and confirm transcription status. T₃ transgenic pure lines were subjected to gene expression analysis and stress tolerance evaluation.

To observe the effects of ABA and mannitol on seed germination and phenotypic differences between wild type and OE-GhCDPK60 plants, three independent GhCDPK60-overexpressing lines (T₃) and Col-0 plants were tested according to method mentioned previously (Wang et al., 2016). Col-0 and GhCDPK60-overexpressing seeds were sown on 1/2 MS medium solid plates with 0.25 μM ABA, 200 mM or 400 mM mannitol, stratified at 4°C for 3 days and then transferred to long-day growth conditions (16-h light/8-h dark cycle at 25°C) in the growth chamber.

To identify the stress tolerance of GhCDPK60 overexpressed Arabidopsis (OE4, OE5 and OE12) and Col-0, seeds cultured on MS medium for 7 days were transplanted into soil for about 3 weeks with sufficient watering followed by a 15-days drought stress (withholding irrigation). Normally watered plants were used as the control. And then the relative water content, malondialdehyde (MDA), proline, catalase (CAT) activities, superoxide dismutase (SOD) activities, and peroxidase (POD) activities were examined by leaf samplings. Each sample represented three replicates (each replicate had 4-6 seedlings). These experiments were performed at least three times independently with similar results.

An about 360 bp fragment of GhCDPK60 was amplified and then ligated into Virus Induced Gene Silencing (VIGS) vector pTRV2 (Reyes et al., 2017). Constructed vectors were transformed into Agrobacterium GV3101 competent cells by a heat-shock method. The GV3101 lines contained pTRV-GhCDPK60, pTRV-GhPDS, and pTRV (pYL156, empty vector as control) were mixed with an equal volume of Agrobacterium containing pYL192 (helper vector), respectively. The mixed solutions were used to infiltrate unfolded cotton cotyledons of TM-1 and CRI50, respectively (Hayward et al., 2011; Gu et al., 2014). The VIGS experiments were repeated at least three times with more than three individual plants were included. The quantitative real-time PCR (qRT-PCR) was performed to further confirm that candidate genes had been silenced in VIGS experiments. The primers used in the VIGS experiments and qRT-PCR analysis are listed in Supplementary Table S2 .

For drought tolerance assay, when pTRV-GhPDS plants showed phenotype, pTRV-GhCDPK60 and pTRV plants were subjected to water withholding at the trefoil stage of seedlings. The phenotypes were observed after drought treatment for two weeks, and the leaf samples were collected to evaluate relative water content, contents of MDA, proline, activities of CAT, SOD, and POD. The treatments were repeated at least twice.

In order to observe the accumulation of H₂O₂, fresh leaves from control and VIGS-cotton after two-weeks drought stress treatment were taken and incubated completely in 1mg/mL, pH 3.8 3-3’-diaminobenzidine (DAB) solution for 6 h under 70% humidity conditions till brown precipitates are observed, and then decolorized in 96% ethanol at 40°C in order to remove chlorophyll (Jambunathan, 2010). For the O 2 − detection, detached leaves were immersed in 100 mL staining solution containing 0.1% (w/v) nitroblue tetrazolium (NBT), 10 mM sodium azide, 50 mM potassium phosphate, pH 6.4 for 15 min. After stopping the reaction with 95% ethanol, the samples were decolorized in 96% ethanol under heating at 40°C (Jambunathan, 2010). Superoxide ions react with NBT and appear as blue. These stained leaves can be photographed by light microscope.

To check the physiological indices changes, the gene-silenced cottons leaves or overexpressing Arabidopsis seedlings were used to identify relative water content (RWC), contents of MDA, proline, activities of CAT, SOD, or POD after drought treatments, whereas some plants were kept untreated as controls. Relative water content was measured according to (Hu et al., 2013). The dehydrated leaves were soaked in distilled water for 4 h and turgid weight (TW) was recorded. Leaves were finally dried for 48 h at 80°C to obtain total dry weight (DW). Relative water content was calculated as follows: RWC (%) = [(desiccated weight – DW)/(TW– DW)] × 100. The MDA concentration was determined through thiobarbituric acid method described by (Schmedes and Hølmer, 1989), and proline content measured via a ninhydrine method (Irigoyen et al., 1992). The CAT activity, SOD activity, and POD activities were determined according to the previous studies (Alici and Arabaci, 2016; Ullah et al., 2018). The absorbance was measured using a UV-2550 UV-vis spectrophotometer (SHIMADZU). Three biological replications were performed.

To determine the GhCDPK genes in G. hirsutum, we performed a genome-wide prediction of GhCDPK genes by BLAST analysis of Arabidopsis and rice CDPKs against the publicly available G. hirsutum genome (https://www.cottongen.org). Meanwhile, we conducted a hidden Markov model profile-based search for putative CDPK amino acid sequences and then identified probable GhCDPKs using SMART and PFAM. A total of 80 GhCDPKs were identified and numbered according to their genome location ( Supplementary Table S1 , Supplementary Figure 1 ). All putative GhCDPKs exhibited the typical protein structures of the CDPK family, which consist of a variable N-terminal domain, a protein kinase domain, a junction domain, a calmodulin-like Ca²⁺ binding domain, and a C-terminal domain ( Supplementary Figure 2 ). The number of amino acid residues, molecular mass, and isoelectric point of the GhCDPKs are listed in Supplementary Table S1 . The numbers of GhCDPKs amino acids ranged from 368 (GhCDPK51) to 648 residues (GhCDPK11), the predicted molecular mass varied from 41.626 (GhCDPK51) to 63.564 kDa (GhCDPK11) and the predicted isoelectric point ranged between 4.452 (GhCDPK49) and 9.481 (GhCDPK72), which were comparable with CDPK genes from other plant species (Hrabak et al., 2003; Ray et al., 2007; Ma et al., 2013). To investigate the evolutionary relationships of the GhCDPKs proteins, sequences of the 80 GhCDPKs, Arabidopsis AtCDPKs proteins (Hrabak et al., 2003), and rice OsCDPKs proteins (Ray et al., 2007) were used to construct a neighbor-joining (NJ) phylogenetic tree ( Figure 1 ). According to evolutionary relationships, all the CDPKs can be classified into four groups (Group I-IV). CDPKs from monocots and dicots populated evenly in all four groups, which indicated that the CDPKs have already diverged from the common ancestor of these clades. CDPKs were reported to play essential roles in tolerance to abiotic stresses or seed development in plants (Coca and San Segundo, 2010; Asano et al., 2012b; Wei et al., 2014; Liu et al., 2017; Chen et al., 2017). We used public available high-throughput sequencing data from G. hirsutum acc. TM-1 to investigate the expression profiles of CDPK family genes in various tissues, including roots, stems, leaves, -3 days post anthesis (dpa), 0 dpa and 3 dpa ovules and 5 dpa, 10 dpa, 20 dpa and 25 dpa fibers among others (Zhang et al., 2015). The CDPK genes exhibited various expression patterns and functional divergence in vegetative and reproductive organs ( Supplementary Figure 3 ). In our previous genome-wide analysis of the CDPK family in Gossypium raimondii, we identified some CDPK genes as positive regulators in the response of cotton to drought stress, including GhCDPK60 (Zhu et al., 2013; Li et al., 2015b). GhCDPK60 shares a high sequence similarity with Arabidopsis AtCDPK8 (79.1% identity) (Zou et al., 2015) and both belongs to group II of the CDPK family ( Figure 1 ), suggesting their functional similarity; thus, GhCDPK60 was further used for functional study.

According to previous study, AtCDPK8 functions in ABA-mediated stomata regulation in response to drought stress through phosphorylate AtCAT3 (Zou et al., 2015). In view of high similarity of protein sequence between GhCDPK60 and AtCDPK8, we deduced that GhCDPK60 maybe participate in drought stress tolerance. To verify the hypothesis, we carried out drought treatment and found that transcription level of GhCDPK60 was significantly induced after water withdrawing but recovered to comparably lower after rewatering treatment ( Figure 2A ). Phytohormones play crucial roles in mitigating and minimizing drought stress-related detrimental effects and improving plant growth and survival under different environmental stresses (Jogawat et al., 2021). In order to investigate whether GhCDPK60 respond to hormonal signals, we evaluated the transcript profiles of GhCDPK60 according to different hormonal treatments for trefoil stage of the cotton seedlings. Results showed that GhCDPK60 was upregulated after four types of hormone treatments ( Figure 2 ). During the ABA treatment, the expression level of GhCDPK60 was at least two-fold higher than that of the control within 12 h ( Figure 2B ). The expression level of GhCDPK60 was approximately three-fold higher than that of the control after 24 h of MeJA treatment ( Figure 2C ). In addition, the expression level of GhCDPK60 was approximately two-fold higher than that of the control after 12 h of treatment with SA ( Figure 2E ). The distinct expression patterns of GhCDPK60 suggested that it could respond to drought stress and might function differently during different drought processes.

The GhCDPK60 (Ghir_D07G013310.1) has an open reading frame of 1590 bp encoding a 59.438 kDa protein, and possesses a structure typical of the CDPK family. In addition, there were several cis-regulatory elements in the promoter of GhCDPK60 through online website analysis (plantCARE), such as ABRE, Box 4, CAAT-box, CGTCA-motif, G-box, TATA-box, and MBS among others ( Supplementary Table S3 ), which are the binding sites for abiotic stress factors, indicating that this gene is essential to stress responses. CDPKs were predicted to localize on the plasma membrane, cytoplasm, endoplasmic reticulum, as well as in the nucleus (Asano et al., 2012a). To confirm subcellular localization of GhCDPK60, the fusion protein of GhCDPK60 with GFP under the constitutive CaMV35S promoter was successfully expressed in the epidermal cells of tobacco, and the free GFP vector was used as a positive control ( Supplementary Figure 4 ). Transient expression in tobacco leaves was performed by agroinfiltration method. The empty GFP protein was localized in the membrane, cytoplasm, and nucleus, while the signal of GhCDPK60-GFP was perceived in the plasma membrane ( Supplementary Figure 4 ). To further confirm the plasma membrane localization of GhCDPK60, GhCDPK60-GFP fusion protein was transiently co-expressed with cytoplasm ( Figure 3A ) and plasma membrane ( Figure 3B ) markers in leaf epidermal cells of N. benthamiana, respectively. The results showed that the colocalization signal of GhCDPK60-GFP with plasma membrane marker, SYP132 (Xia et al., 2019), instead of cytoplasm marker were captured ( Figure 3 ), indicating that GhCDPK60 is a plasma membrane associated protein kinase.

To confirm the function of GhCDPK60 in response to drought stress, we firstly overexpressed it in wild-type Arabidopsis and selected three independent homozygous lines of the T₃ generation (OE4, OE5 and OE12) via qRT-PCR ( Figure 4A ), which were used for the subsequent physiological experiment. In order to examine whether GhCDPK60 affects drought tolerance in transgenic Arabidopsis, we assessed the germination of Col-0 and GhCDPK60-overexpressing lines after 200 mM or 400 mM mannitol treatments. The results showed that GhCDPK60-overexpressing lines germination was obviously higher than that of Col after mannitol treatment ( Figure 4B ). The phytohormone ABA plays important roles in the adaptation of plants to abiotic stresses, such as high salinity and drought (Cutler et al., 2010). Drought stress induces ABA accumulation and triggers ABA-dependent signaling pathways (Zhu, 2002). Thus, we investigated the GhCDPK60 response to ABA and found that transgenic Arabidopsis exhibited more sensitivity in medium with 0.25 μM ABA ( Figure 4B ), suggesting that GhCDPK60 may be involved in ABA-regulated physiological processes and drought stress tolerance.

To further investigate the drought stress tolerance of GhCDPK60-OE, 3-week-old Arabidopsis seedlings were subjected to a drought treatment. After 5, 10 or 15 days of drought treatment, the growth of Col was inhibited compared with that of GhCDPK60-OE lines ( Figure 4C ). Plants with high capacity for water retention can better survive drought or dehydration stress. After a dehydration treatment, GhCDPK60-OE lines retained a significant high RWC compared with the Col ( Figure 4D ). In plants, osmotic adjustment and stomatal closure are the main physiological mechanisms for reducing water loss under dehydration or drought. To elucidate the physiological mechanism by which GhCDPK60 confers tolerance to drought and dehydration stresses and improves the ability of plant to retain water, we quantified the osmolyte proline (Per et al., 2017) in GhCDPK60-OE lines. Under normal growth conditions, there were no significant differences between controls and transgenic lines in terms of their proline content, but GhCDPK60-OE lines accumulated larger amounts of proline under drought conditions ( Figure 4E ). MDA is the product of the peroxidation reaction, which used as a drought indicator to evaluate the degree of plasma membrane damage and the strength of the stress reaction (Zhang et al., 2021a). To assess whether GhCDPK60 is involved in oxidative damage, we tested the MDA concentration. Although there were no significant differences in malondialdehyde (MDA) contents between controls and GhCDPK60-OE lines under normal growth conditions, clear differences were observed between control and GhCDPK60-OE lines after drought treatment ( Figure 4F ). These results indicated that GhCDPK60 played a positive role in drought stress tolerance.

Under drought stress, the transcript levels of stress-responsive genes AtCAT3, AtNXH1, AtRD29A/B, and AtDREB2A were obviously higher in GhCDPK60-OE lines than in Col ( Supplementary Figure 5 ). In previous studies, higher transcript levels of AtCAT3, AtNXH1, AtRD29A/B, and AtDREB2A enhanced tolerance to abiotic stresses (Sakuma et al., 2006; Asif et al., 2011; Jia et al., 2012; Zou et al., 2015). These results demonstrated that GhCDPK60 was involved in increasing transcription of stress-associated genes in Arabidopsis, thereby improving tolerance to drought stress.

VIGS is a fast and simple method for transient silencing of genes that is widely used in cotton research (Hayward et al., 2011). To functionally characterize GhCDPK60 in cotton drought tolerance, GhCDPK60 was successfully silenced in two different backgrounds using a VIGS strategy ( Figure 5A ). The pTRV : GhPDS-inoculated plant leaves showed a bleached phenotype ( Figure 5A ), which was an affirmative indication that the knockdown vector was effective. According to qRT-PCR validation, the expression level of GhCDPK60 was significantly reduced in the VIGS plants than that in control (plants transformed in empty vector) ( Figure 5B ). To further evaluate the phenotype of the gene-silenced plants under drought stress, the plants were subjected to a two-week drought treatment. We found that the leaves of GhCDPK60-silenced plants showed more wilted and shrunken than in control plants, especially in the TM-1 genetic background ( Figure 5C ). The RWC from detached leaves was much lower for the GhCDPK60-silenced plants than that in control. In addition, the RWC of GhCDPK60-silenced plants in the cv. CRI50 background was slightly higher than that in the TM-1 background ( Figure 5D ), consistent with the dehydration phenotypes. After two-week drought treatment, GhCDPK60-silenced plants showed a lower proline content and higher MDA contents, compared with those of vector control ( Figure 5E, F ). To obtain a deeper understanding of the function of GhCDPK60 under drought stress, we analyzed the transcript levels of some stress-inducible genes and found that they were significant lower in GhCDPK60-silenced plants than in control under drought stress ( Supplementary Figure 6 ). These results indicated that the silencing of GhCDPK60 decreased the drought tolerance in cotton.

Increasing evidence suggests that drought induces the production of active oxygen species (Fu and Huang, 2001). ROS production triggers defense mechanisms associated with Ca²⁺ fluxes and ABA signaling under drought stress, but over-accumulation of ROS leads to cell death via progressive oxidative damage (Mahmood et al., 2019). Considering the lower MDA content in GhCDPK60-OE lines and higher MDA content in GhCDPK60-silenced plants under drought stress ( Figure 4F , 5F ), we hypothesized that GhCDPK60 might be involved in ROS production. In order to determine whether GhCDPK60 functions on the production of ROS, GhCDPK60-silenced plants after drought treatment were stained with 3-3’-diaminobenzidine (DAB) and nitroblue tetrazolium (NBT), respectively, and the results demonstrated that the ROS levels were comparatively higher in GhCDPK60-silenced plants in the TM-1 background than that of control ( Figure 6A ), consistent with the phenotypes after drought stress. Plants have developed the scavenging mechanisms to maintain homeostasis of ROS redox reactions and have been protected against the detrimental effects of active oxygen and biotic/abiotic stress by non-enzymatic antioxidants and enzymatic components (Van Breusegem et al., 1998), such as catalase (CAT), superoxide dismutase (SOD), and peroxidase (POD) (Bowler et al., 1992). Subsequently, we determined the CAT, SOD, and POD activities both in transgenic Arabidopsis and silenced cotton under drought stress. We found that CAT, SOD, and POD activities in the GhCDPK60-overexpressing Arabidopsis lines were significantly higher than in the Col, while the GhCDPK60-silenced cotton displayed decreased trends compared to the control ( Figure 6B-G ). These results indicated that the GhCDPK60 was involved in the elimination of ROS in the defense response to drought stress.

Plant CDPKs stand for a multigene family of numerous calcium-dependent protein kinases which are vital for several physiological processes, including biotic and abiotic stress tolerance, by participating in signal transduction pathways and inducing downstream effects, such as altered protein phosphorylation and gene expression patterns (Harmon et al., 2001). However, to the best of our knowledge, studies focused on the roles of CDPKs in response to abiotic stress in upland cotton were limited. The release of plant genome sequencing data has enabled the identification of CDPKs in Arabidopsis (Cheng et al., 2002; Hrabak et al., 2003), Oryza sativa (Asano et al., 2005), Triticum aestivum L. (Li et al., 2008), Populus trichocarpa (Zuo et al., 2013), Fragaria x ananassa (Crizel et al., 2020), banana (Li et al., 2020), and Solanum habrochaites (Li et al., 2022), which provides models for the characterization of the CDPKs gene family in upland cotton. Benefiting from the advent and availability of the published genome sequence in G. hirsutum TM-1 (Zhang et al., 2015; Wang et al., 2019; Chen et al., 2020), we identified and analyzed CDPKs gene and proteins by extracting and aligning their sequences in TM-1 (HAU_v1) (Wang et al., 2019). Totally, 80 putative GhCDPKs were manually reannotated and confirmed in upland cotton, most of which existed as gene pairs in At and Dt subgenome, while only a few members existed in one of the subgenomes, such as GhCDPK5, GhCDPK6, and GhCDPK22, suggesting that the GhCDPKs maybe experience genome replication event. Similar to CDPKs in rice and Arabidopsis, all the members of GhCDPKs have typical conserved EF-hand motifs, which contribute to binding Ca²⁺. The phylogenetic analysis on GhCDPKs proteins showed that most of them were greatly conserved during evolution.

Drought stress are the major abiotic threats to plants that result in alters in transpiration rate, development and composition of photosynthetic apparatus, excess ROS production, and biochemical composition changes, and further affect plant growth and decrease crop yield. Plants have developed a network of signal transduction pathways to regulate metabolism to adapt to environments, including CDPKs mediated Ca²⁺ pathway (Asano et al., 2012a). Cotton is an essential economic fiber crop with high quality oil and protein seeds, frequently undergoes drought stress, which severely damages the cotton yield. In this research, we found that the GhCDPK60 was significantly induced by drought and hormones treatments, suggesting its potential function on regulating drought stress tolerance. GhCDPK60 was confirmed to be a plasma membrane associated protein. The plasma membrane localization of GhCDPK60 could facilitate activation of several enzyme by localized calcium signals generated in response to abiotic or biotic stimuli (Kudla et al., 2010). To assess the role of GhCDPK60 under drought conditions, we employed overexpression assay in Arabidopsis and VIGS assay in cotton. Our results demonstrated that GhCDPK60-overexpressing plants were stronger than the Col and retained more water content, while GhCDPK60-silence plants were more vulnerable to drought stress compared with the control, reflected in the withered leaves and lower leaf relative water content, suggesting that GhCDPK60 is a positive regulator of the response to drought stress. These results are consistent with those of previous studies on some other CDPKs that positively regulate drought stress tolerance (Wei et al., 2014; Zou et al., 2015; Huang et al., 2018). It was noted that phenotypes of GhCDPK60-silence plants in the TM-1 background were more severe than that in the CRI50 background under drought stress, suggesting that there existed other genes to help resist to drought stress in CRI50.

To gain a deeper understanding of the molecular mechanism of GhCDPK60 under drought stress, we analyzed the transcription level of several stress-inducible genes. In previous studies, the atcat3 mutant displayed a drought stress-sensitive phenotype (Zou et al., 2015). Overexpression of AtNHX1 gene improved drought tolerance in transgenic groundnut (Asif et al., 2011). The RD29 (Responsive to Desiccation) genes RD29A and RD29B were induced by desiccation stress (Jia et al., 2012). AtDREB2A interacts with a cis-acting DRE sequence to activate the expression of downstream genes that are involved in drought stress response in Arabidopsis (Sakuma et al., 2006). Under drought stress, the transcription levels of drought-responsive genes, AtCAT3, AtNXH1, AtRD29A/B, and AtDREB2A were higher in GhCDPK60-overexpressing Arabidopsis than in Col. By contrast, our results showed that the transcription levels of drought-related genes GhCDPK1, GhZEP, GhAAO3, and GhABA2 were lower in GhCDPK60-silence cotton than control. GhCDPK1, as a positive regulator, was induced by drought stress (Tian et al., 2016). AtZEP, a homologous gene of GhZEP, played important roles in response to osmotic stress (Park et al., 2008). Two ABA biosynthetic-related genes GhAAO3 and GhABA2 were homologous with AtAAO3 and AtABA2, respectively, which were required for drought tolerance (Khan et al., 2019). Considering the result that GhCDPK60 responded to ABA signal, we speculated that GhCDPK60 may play role in increasing expression of stress-associated genes and ABA-regulated drought stress tolerance.

The ability to reserve enough water is crucial for plants to overcome drought stress. Our results showed that GhCDPK60 played a positive role in improving the ability of the plant to retain water under dehydration conditions. Subsequently, we explored the physiological mechanism by which GhCDPK60 enables the plants to retain water. Plants are inclined to accumulate compatible osmolytes such as proline to reduce the cellular osmotic potential under limited water conditions (Per et al., 2017). Our results showed that there were increased proline content in GhCDPK60-overexpressing Arabidopsis, but decreased content in GhCDPK60-silence cotton. Hence, we deduced that GhCDPK60 functions in osmotic adjustment and improvement of ability to retain water during drought in plants. Generally, plants tend to generate ROS under abiotic or biotic stress conditions, thereby impairing the production of biomolecules and increasing the MDA concentration as well as the permeability of the plasma membrane. MDA is produced by polyunsaturated fatty acid peroxidation in the cells, which is a robust diagnostic indicator for determining the injury degree to a stressed plant (Morales and Munné-Bosch, 2019). In our research, GhCDPK60-overexpressing Arabidopsis produced less MDA than Col under drought stress, whereas GhCDPK60-silence cotton produced more compared with control, suggesting that expression of GhCDPK60 might alleviate the damages induced by oxidative stress. Plants have developed a complicated ROS scavenging system to minimize and/or prevent oxidative damage in cells (Van Breusegem et al., 1998). Indeed, GhCDPK60-silence cotton exhibited obvious reduced activities of antioxidant enzymes under drought stress, such as CAT, SOD, and POD, on the contrary, GhCDPK60-overexpressing Arabidopsis showed enhanced enzymes activities. These results are consistent with that GhCDPK60-overexpressing Arabidopsis displayed increased drought stress tolerance, which possibly as a result of reduced ROS accumulation. Overall, GhCDPK60 could enhance the ability of plants to resist drought stress by accumulating more osmotic adjustment substances, and enhancing the activity of the antioxidant system to scavenge ROS.