The accession ZM24 of upland cotton was used as the research material. Seedlings of ZM24 were grown hydroponically in half-strength Hoagland nutrient solution until the third true leaf expanded, and then they were exposed to 200 mM NaCl and 6% polyethylene glycol 6000 (PEG6000) solution between 0 and 24 h. The roots were harvested at 0, 3, 6, 12 and 24 h. Seeds of A. thaliana were surface-sterilized with 50% sodium hypochlorite for 4 min and then rinsed three times using sterile water. The seeds were sown on solidified Murashige and Skoog (MS) medium and then vernalized at 4°C for two days in the dark before being transferred to grow at 22°C.

Total RNAs were isolated from Arabidopsis or cotton seedlings using RNAprep Pure Plant Kit (Tiangen, Beijing, China). Total RNA (1 μg) was reverse transcribed into cDNA using the MonScript™ RTIII kit (Monad). Gene-specific primers are listed in Supplementary Table S1. The housekeeping genes were G. hirsutum histone3 (GhHis3, Gh_D03G0370) and A. thaliana actin2 (AtActin2, At3G18780). Quantitative real-time PCR (qRT-PCR) was performed using MonAmp™ SYBR® Green qPCR Mix and reactions were run in a Monad Selected q225 PCR system. Relative expression levels of genes were calculated by the 2^−ΔΔCt method.

For the subcellular localization analysis, GhSCL4 coding sequences (CDS) were ligated into the EcoRV site of the pEarlyGate101 vector to construct the 35S::GhSCL4-YFP vector [36]. The positive construct was then transformed into tobacco (Nicotiana benthamiana) cells. After 48 h of inoculation, the epidermal cells of the infiltrated tobacco leaves were visualized using a confocal microscope (OLYMPUS FV1200). For transactivation assay, the full-length CDS of GhSCL4 was inserted between the EcoR I and Sal I sites of the pGBKT7 vector. The recombined construct and inset-less vector pGBKT7 were both transformed into yeast strain AH109. Transformants were selected by growing on SD\-Trp medium at 30°C for two days and the surviving colonies were then transferred to both SD\-Trp and SD\-Trp\-His\-Ade medium to grow for three days. The primers used in these two assays are listed in Supplementary Table S1.

The full-length coding sequence of GhSCL4 (Gh_D12G0611) was retrieved from the CottonGen database (https://www.cottongen.org/data/download) [37]. Subsequently, it was amplified using specific primers p2300-GhSCL4-F and p2300-GhSCL4-R (see Supplementary Table S1) and cloned between the Xba I and Kpn I sites of the pCAMBIA2300 vector to generate the 35S::GhSCL4 construct. The construct was transformed into wild-type Arabidopsis (Columbia ecotype) (for gene overexpression) by the Agrobacterium tumefaciens-mediated floral dip method. The salt and osmotic tolerance of GhSCL4 transgenic Arabidopsis during the germination period was tested on MS agar plates saturated with 150 mM NaCl and 400 mM mannitol, respectively. Plant growth was monitored based on cotyledon greening rates on the tenth day post-germination. To evaluate the salt tolerance of adult seedlings grown in soil, 3-week-old Arabidopsis plants were irrigated with 200 mM NaCl for two weeks. To observe the primary root length of plants, Arabidopsis seeds were grown on 1/2 MS medium supplemented with NaCl or mannitol for 10 d and then the root length was measured.

The tobacco rattle virus (TRV) system was used for virus induced gene silencing (VIGS) as previously reported [38]. A nearly 200-bp fragment of each of the candidate genes GhSCL4 and GhCaM7 was inserted into TRV-based inset-less vector pYL156 to produce TRV::GhSCL4 and TRV::GhCaM7, respectively, for VIGS in cotton plants. Plants injected with the pYL156 vector (TRV::00) were used as controls. Plasmids containing the positive recombinant VIGS vectors were transformed into A. tumefaciens strain GV3101. The cotyledons of 7-day-old ZM24 cotton seedlings were inoculated with equal amounts of the Agrobacterium expressing the TRV vectors as previously described [39]. qRT-PCR was performed to detect the interference efficiency of GhSCL4/GhCaM7 approximately two weeks after infiltration, and empty vector agro-infiltrated cotton plantlets were the control. For the salt tolerance assay, control and GhSCL4/GhCaM7-silenced plants were irrigated with 300 mM NaCl solution when plants produced the third true leaf.

To further detect the antioxidant enzyme activities in cotton plants exposed to salt treatment for 10 days, approximately 200 mg of fresh leaf materials were collected and ground using 2 ml of pre-chilled 50 mM phosphate buffer. Activities of superoxide dismutase (SOD), peroxidase (POD), and catalase (CAT) were performed as described previously [40].

The CDS of GhSCL4 was ligated into the pXY104 (cYFP) vector. The CDS of GhCaM7 was cloned into the pXY106 (nYFP) vector [41]. The primers used in this assay are listed in Supplementary Table S1. The recombinant constructs were introduced into Agrobacterium strain GV3101. Agrobacterium cells were collected and resuspended in infiltration medium to an OD₆₀₀ of 1.2. Agrobacterium cells containing nYFP and cYFP fusion proteins were then mixed at a ratio of 1:1 and infiltrated into tobacco leaves. After 48 h of inoculation, the infiltrated leaves were imaged using a confocal microscope (OLYMPUS FV1200).

The CDS of GhSCL4 and GhCaM7 were ligated into the pCAMBIA1300-nLuc and pCAMBIA1300-cLuc vectors, respectively [42]. Agrobacterium cells containing the indicated constructs were collected and resuspended to a final concentration of OD600 = 0.5. Bacterial suspensions were infiltrated into leaves of N. benthamiana plants and then placed at 23°C for 48 h in the dark. Subsequently, plants were incubated in an artificial growth chamber (25°C, 16 h light/8 h dark photoperiod) before luciferase (LUC) activity was measured.

To select the GRAS family genes involved in salt stress, the raw data of RNA-seq was download from NCBI Sequence Read Archive (SRA: PRJNA248163). A salt-induced SCL4/7 subgroup gene in cotton, namely GhSCL4 (Gh_D12G0611), was identified from the RNA-seq data of salt-treated cotton (Fig. 1A). Subsequently, the expression level of this gene was confirmed by qRT-PCR, and the result showed that the amounts of transcripts of GhSCL4 were higher than that in H₂O-treated controls, especially after 3 and 6 h of salt treatment (Fig. 1B). In addition, the transient expression assay using N. benthamiana cells showed that the GhSCL4-YFP fusion protein accumulated mainly in the nucleus (Fig. 1C). Furthermore, GhSCL4 was cloned into the pGBKT7 vector and expressed in the AH109 yeast cells to confirm its transactivation activity. As shown in Fig. 1D, only transformants expressing the pGBKT7-GhSCL4 fusion protein could grow normally on both SD\-Trp and SD\-Trp\-Ade\-His media, whereas those transformed with the empty pGBKT7 vector did not survive on the SD\-Trp\-Ade\-His plates.

To investigate the function of GhSCL4 in plants, GhSCL4 was ectopically overexpressed in wild-type Arabidopsis plants and three homozygous overexpression lines (OE1, OE2 and OE3) were obtained. To determine whether GhSCL4 is involved in plant responses to salt, the growth of OE and WT plants grown on solidified medium was compared. When grown on MS medium, the growth of OE plants was similar to that of WT plants (data not shown). However, in the presence of 150 mM NaCl, the GhSCL4-overexpression plants exhibited higher tolerance to NaCl as indicated by the higher cotyledon greening rates of the OE plants than that of the control plants. An estimated 31.41% ± 4.24% of WT seedlings had green leaves, while 92.59% ± 4.24%, 78.7% ± 6.99% and 88.89% ± 5.56% of the OE1, OE2 and OE3 lines had green cotyledons, respectively (Figs. 2A, 2B).

Considering that root architecture has an important role in the response of plants to abiotic stress, root lengths of transgenic and wild-type plants under salt stress were examined. Regardless of whether NaCl was present, lengths of the primary roots of GhSCL4-overexpressing plants were generally longer than roots of WT. Roots of OE lines were also longer than that of WT lines in the NaCl treatments at concentrations of 100 and 120 mM (Figs. 2C, 2D). Thus GhSCL4-overexpressing plants were able to maintain higher root activity under either normal or salt stress conditions, indicating enhanced salt tolerance in overexpressing plants.

Salt tolerance assays of the seedlings grown in soil were also performed, and the result showed that the growth of GhSCL4-overexpressing plants and wild type plants were both severely inhibited, though the wild type plants were more sensitive to salt stress (Fig. 2E). After salt treatment, the leaf withering rate of OE lines was approximately 15%, which is significantly lower than that (>55%) of the WT plants (Fig. 2F).

The tolerance of GhSCL4-overexpressed lines and control plants to mannitol-simulated osmotic stress was determined. After 10 days of growth on mannitol medium, the cotyledon greening rates of three overexpression lines (OE1, OE2 and OE3) were all significantly higher than that of WT (36.11% ± 2.77%) (Figs. 3A, 3B). Additionally, the effect of osmotic stress inhibited elongation of primary roots. The root lengths from both overexpression and WT lines gradually decreased as the mannitol concentration increased from 200 to 250 mM, though the root lengths of overexpression lines were significantly longer than that of wild-type plants in the presence of various concentrations of mannitol (Figs. 3C, 3D). Thus, over-expression of GhSCL4 could enhance the osmotic tolerance of plants.

To study the molecular mechanism of GhSCL4 in response to salt stress, the transcript levels of salt-responsive genes were determined under both normal conditions and NaCl treatments by real-time PCR analysis. The result showed that the transcript levels of salt-induced genes, including AtNHX1 and AtSOS1, were stress inducible in both OE lines and WT, but the transcript levels were substantially higher in the OE lines than in WT plants after salt treatment (Fig. 4). Salt stress can lead to accumulation of ROS and result in oxidative damage, the transcript levels of oxidation-related genes were calculated as well. In transgenic Arabidopsis over-expressing GhSCL4, the oxidation-related genes AtAPX3 and AtCSD2 had more transcripts under salt-stressed conditions than those in WT plants (Fig. 4). Overall, the results showed that GhSCL4 may enhance tolerance to salt in part by promoting the expression of salt stress-induced and oxidation-related genes under salt condition.

To further confirm the role of GhSCL4 in modulating abiotic stress in cotton, TRV-based VIGS was conducted to interfere with the expression of GhSCL4 and obtain GhSCL4-silenced cotton. Cotton lines injected with the empty vector were taken as the control. The interference efficiency was assessed using qRT-PCR, and the result showed that the transcript level of GhSCL4 in GhSCL4-silenced plants was significantly lower than that of the control (Fig. 5A). Under normal growth condition, no obvious difference was detected between the control and GhSCL4-silenced plants. However, following irrigation with 300 mM NaCl solution for three weeks, the leaves of GhSCL4-silenced plants yellowed, shrunk, rolled or wilted, while the control plants grew normally (Fig. 5B).

To investigate whether GhSCL4 could participate in regulating H₂O₂ content under salt stress, we detected the accumulation of H₂O₂ in GhSCL4-silenced and control plants using DAB staining. Under normal conditions (no salt stress), very little H₂O₂ accumulated in both GhSCL4-silenced and control plants and there were no significant differences between them. However, when they were exposed to NaCl, H₂O₂ was highly accumulated in the leaves of NaCl-treated plants, as indicated by the extensive staining by DAB (Fig. 5C). However, the GhSCL4-silenced plants accumulated more H₂O₂ than control plants under salt treatment (Fig. 5C). In addition, changes in the activities of antioxidant enzyme (CAT, SOD, and POD) in GhSCL4-silenced plants and control plants were investigated. The results showed that under normal conditions, there were no significant differences in the antioxidant enzyme activities between GhSCL4-silenced and control plants, while GhSCL4-silenced plants exhibited lower CAT and SOD activities than those of control plants under salt treatment (Fig. 5D). Overall, these results suggest that GhSCL4 enhances salt tolerance of plants by detoxifying ROS.

In order to provide insight into the regulation network of GhSCL4 in response to abiotic stress, we predicted the GhSCL4-interacting proteins based on the database of co-expression networks for cotton (http://structuralbiology.cau.edu.cn/gossypium/cytoscape/network.php) and identified 14 genes that co-expressed with GhSCL4. Gene Ontogeny analysis was conducted to detect the pathways of these co-expression genes. The genes classified into nine pathways listed in Fig. 6A, including calcium ion binding, MAP kinase activity, ATP binding, chromatin binding, protein tyrosine kinase activity. Among the pathways, the number of co-expression genes belonging to the calcium ion binding pathway was the highest, and annotation of these genes revealed that they showed high similarity to calmodulin proteins (CaM7 and CaM5) and CaM-like proteins (CML11 and CML8) in Arabidopsis, respectively (Supplementary Table S2). To further confirm the interaction of GhSCL4 and these calcium ion binding proteins, a LCI assay was conducted to confirm the interaction between GhSCL4 and GhCaM7. As shown in Fig. 6B, only the coexpression of CLuc-GhCaM7 and NLuc-GhSCL4 in tobacco epidermal cells complemented the LUC activity. Additionally, BiFC assay also showed that GhCaM7 (Gh_A05G3893) could interact with GhSCL4 in the nucleus (Fig. 6C). Additionally, VIGS assay was conducted to obtain GhCaM7 silenced plants and investigate the role of GhCaM7 in response to salt stress in cotton. qRT-PCR analysis confirmed that the transcript level in GhCaM7-silenced plants was significantly lower than that in control plants (Fig. 6D). When all plants were treated with 300 mM NaCl for nearly four weeks, results showed that the leaves of GhCaM7-silenced plants were more yellow and wilted, indicating GhCaM7 is a positive regulator in response to salt stress (Fig. 6E).

With the growing conflict of land use to grow grain crops or cotton, cotton has been gradually displaced to areas with saline-alkali soils [3]. Therefore, mining salt tolerance genes is important for breeding salt-tolerant cotton cultivars. Several SCL4/7 subgroup genes in the GRAS family of genes have been reported to participate in plant development and abiotic stress [43–45]. For example, transgenic Arabidopsis overexpressing the poplar GRAS gene PeSCL7 exhibited significantly increased salt and drought tolerance at the seedling stage [43]. In this study, transgenic Arabidopsis overexpressing GhSCL4 showed enhanced resistance to salt and osmotic stress. Silencing of GhSCL4 resulted in reduced salt and drought tolerance in cotton. Salt stress can led to osmotic stress, which in turn can result in turgor loss in plant cells [45]. Thus, the osmotic stress tolerance conferred by GhSCL4 may contribute to the increased salt resistance in plants. Recently, several GRAS proteins, such as LISCL and NSP1/NSP2, have been confirmed as transcription factors with transcription activation activity [46]. Our transient expression assay showed that the GhSCL4 protein could localize to the cell nucleus, and our transactivation activity assay in the AH109 yeast cells showed that GhSCL4 could activate the reporter genes Ade and His of the yeast (Figs. 1B, 1C). Thus, GhSCL4 may function as a transcriptional regulator.

When a plant is subjected to abiotic stress in the soil, root tips are the plant parts that are first exposed to stress signals, and consequently the adverse conditions will inhibit primary root growth by decreasing cell production and size [7,47]. Plants that can produce longer and more extensive roots can explore farther soil regions in search of available water, thus some studies have demonstrated that root length is positively associated with salt stress in several crop species such as cotton and soybean [40,48]. Here, our comparison of the primary root lengths of GhSCL4 transgenic and wild-type Arabidopsis showed that regardless of whether roots were exposed to NaCl or mannitol, the primary root length of GhSCL4 transgenic plants was significantly longer than that of control plans (Figs. 2C, 3C). Thus, GhSCL4 functions as a positive regulator of primary root elongation, and the longer primary root length of GhSCL4 transgenic seedlings likely enhanced its ability to reach and absorb more water in deeper soils to provide partial relief from the low availability of water from the high osmotic conditions. Although the exact molecular mechanism underlying the role of GhSCL4 in primary root elongation requires further investigation, we report that root architecture may be a key determinant in plant response to soil abiotic stresses.

At present, there are only a few studies that report the mechanisms of GRAS family genes in plant abiotic stress response. A poplar SCL gene, PeSCL7, alleviated salt toxicity in transgenic plants by elevating enzyme activity of the ROS-scavenging SOD [43]. Salt stress induces the accumulation of ROS, and excessive accumulation of ROS causes oxidative damage to membrane lipids [19]. Thus, the increased activity of the SOD enzyme induced by overexpression of PeSCL7 likely controlled the accumulation of ROS to more tolerable levels under the abiotic stress. Here, we demonstrated that GhSCL4 may regulate ROS levels in response to salt stress. The major species of ROS, H₂O₂, was detected at higher amounts in GhSCL4-silenced plants than in control plants under salt treatment, and GhSCL4-silenced plants exhibited lower CAT and SOD activities than those of control plants under salt treatment (Fig. 5D). Furthermore, after salt treatment, the transcripts of oxidation-related genes (AtAPX3 and AtCSD2) of GhSCL4 over-expression Arabidopsis were significantly higher than those in WT plants (Fig. 4). Overall, these results suggest that GhSCL4 enhances salt tolerance of plants by detoxifying ROS.

The transient expression assay using both LCI and BiFC confirmed the interaction between GhSCL4 and GhCaM7 (Figs. 6B, 6C). Similarly, AtSCL4, the homologous protein of GhSCL4, was also identified as an AtCaM7-interacting protein via protein microarrays [49]. Calmodulin is one of the prominent calcium sensors in plants and are reportedly involved in plant response to abiotic stress [13,15,16]. Here, silenced GhCaM7 in cotton decreased its level of salt tolerance, indicating that GhCaM7 is a positive regulator in cotton response to salt stress. Intracellular CaM may also play a critical role in maintaining ROS homeostasis. In Arabidopsis, CaM interacted with the H₂O₂-scavenging enzyme CAT3 to control ROS levels in a Ca²⁺ dependent manner [50], and CaM also participated indirectly in regulating ROS levels through CaM-regulated λ-aminobutyrate (GABA) synthesis and the GABA shunt metabolic pathway [51]. Thus, the enhanced ROS-scavenging capacity conferred by GhSCL4 might be associated with Ca²⁺/CaM7-mediated signaling. Thus, we surmised that the interaction between GhSCL4 and GhCaM7 may be occurring at the protein level, so further exploration using biochemical and molecular experiments are needed to elucidate the complex signalling mechanism.

Overall, we identified a new SCL gene GhSCL4 that conferred enhanced tolerance to salt treatments in cotton plants. The interaction between GhSCL4 and GhCAM7 indicated that GhSCL4 might be involved in Ca²⁺/CaM7-mediated control of ROS levels when plants are under salt stress. Gene function analysis of GhSCL4 provides beneficial information in developing bio-engineering strategies to improve the stress tolerance of cotton.