Construction of the AtSTZ1 overexpression vector was undertaken following previously established methods (Yu et al., 2016). Briefly, the complete AtSTZ1 cDNA sequence was inserted into the binary vector pCB3000, containing the CaMV 35S promoter. Subsequently, the recombinant plasmid ( Supplementary Figure S1A ) was introduced into hypocotyl explants of upland cotton (Gossypium hirsutum) cultivars Zhongmian 44 and R15 (Qin et al., 2017). Positive transgenic plants from T₀ - T₃ generations were selected and determined according to Qin et al. (2017). The primers utilized for vector construction can be found in Supplementary Table S1 .

Extraction and purification of total RNA from the transgenic cotton leaves and synthesis of first-strand cDNA are referred to Li et al. (2005). As mentioned earlier, quantitative real-time PCR (qRT-PCR) analysis was utilized to detect the transcriptional level of AtSTZ1 and additional genes that are associated with drought stress. This analysis was conducted in accordance with the methodology described by Li et al. (2005). The GhHis3 gene (AF024716) served as the internal reference for gene expression normalization. The specific primers utilized for quantitative reverse transcription polymerase chain reaction (qRT-PCR) are detailed in Supplementary Table S1 .

The seed germination trials involved sowing seeds from homozygous AtSTZ1-overexpressing transgenic lines (OE-4, -6, generated from R15 as explants, and OE-7 and -9 generated from Zhongmian 44 as explants) and the wild type on 10% PEG6000-saturated filter papers. These seeds were germinated in an artificial light incubator with 16-hour light/8-hour dark cycles at 28°C. The germination rate was determined referring to Qin et al. (2014).

Drought tolerance tests were conducted on seedlings from both the wild-type and transgenic cotton in a greenhouse according to Yu et al. (2016). The same author’s method was used to assess the drought resistance and agronomic performance of transgenic cotton plants in the field. These field experiments were performed in Yuncheng City, Shanxi Province, China and Shihezi City, Xinjiang Province, China. Images were taken when plants exhibited noticeable drought stress phenotypes during seedling development, the reproductive period and the mature stage (Yu et al., 2016).

The growth status of the seedlings was recorded on day 15, and a lateral root count was calculated. After 25 days of drought stress, the primary root length of early plants was measured. The roots were cut from AtSTZ1-overexpressing transgenic cotton plants treated with drought for 15 days and 25 days, respectively. Samples of leaves and primary roots were collected from plants that had undergone a 25-day drought treatment. The root and leaf surface imprint methods, referring to Yu et al. (2008), were used. The number of stomata and root cells were calculated, and their sizes were measured referring to the methods of Yu et al. (2016).

All experiments were independently replicated three times.

The water use efficiency (WUE), transpiration rate, and photosynthetic rate in the leaves from 25-day-old transgenic cotton plants were determinated according to Yu et al. (2013). The WUE, transpiration rate and photosynthetic rate were assessed utilizing a photosynthesis measuring instrument (CIRAS-3, Portable photosynthesis System, Lufthansa Scientific Instruments Co., Ltd.).

Leaves from plants subjected to drought for 25 days or from normal control plants were utilized for the analysis of MDA, free proline, soluble sugar, and chlorophyll contents and electrolyte leakage. Proline and chlorophyll levels were quantified following the methodology outlined by Qin et al. (2014), while soluble sugars were determinated as per the protocol by Dubois et al. (1956). As previously mentioned, the MDA content was detected according to the procedure of the thiobarbituric acid (TBA) test (Cao et al., 2009), and electrolyte leakage was assessed referring to Bajji et al. (2002).

The assays of POD, SOD, and CAT activities were performed following the protocol outlined by Ruan et al. (2011). Total protein content was determined by utilizing a Bradford Protein Assay Kit (Solarbio Science &Technology Co., Ltd, Beijing, China).

Fourth true leaf tissues from both transgenic cotton lines and the wild-type controls following drought treatment for 25 days were collected, and extraction and purification of total RNA from the fourth true leaf were conducted, following the methodology outlined by Qin et al. (2017). Subsequently, RNA-seq libraries were constructed and sequenced by Nanjing Personal Gene Technology Co., Ltd. (Nanjing, Jiangsu, China, http://www.personalbio.cn). Details regarding the sequencing location/facility, read length, and type have been previously described by Zhou et al. (2020). Gene function was predicted and the distribution frequency of functional categories was calculated by gene ontology (GO) analysis.

Edger software was used for the analysis of differential gene expression derived from transcriptome sequencing data. The determination of differential gene expression was founded on fold change (FC ≥ 2 or FC ≤ -2) and P value (P ≤ 0.01). The Cutadapt (v2.7) software and parameters utilized for GO and heat map analyses refer to Zhang et al. (2021). The statistical analysis incorporated a twofold alteration in the average of biological replicates, coupled with a stringent false discovery rate threshold of below 0.05 (P < 0.05), to ensure rigor and accuracy.

To explore the role of AtSTZ1 in cotton yield, we obtained 21 transgenic cotton lines overexpressing AtSTZ1 in the R15 and Zhongmian 44 wild-type backgrounds and verified them via kanamycin resistance and PCR analyses ( Supplementary Figure S1 ). To investigate the function of AtSTZ1 in response to drought, we selected four overexpressing homozygous lines with higher AtSTZ1 expression levels ( Supplementary Figure S1B , OE-4 and -6, generated from R15 as explants; OE-7 and -9, generated from Zhongmian 44 as explants) for subsequent analyses.

To evaluate the transgenic lines’ performance in the field, considering both irrigated and natural drought conditions, we constructed field trials at two different field locations (Shihezi, Xinjiang and Yuncheng, Shanxi). A major field experiment was performed in Shihezi, Xinjiang Province, China, from April to September, a period characterized by scarce rainfall throughout the growing season. As illustrated in Figures 1A, B, D, E , compared to the wild type, noticeably better growth and improved drought resilience is observed in four transgenic lines (-4, -6, 7, and -9), and their agronomic trait data were collected along with those of the wild-type controls.

A comparison of fiber yield and agronomic traits between AtSTZ1 transgenic lines and the wild- type counterparts was conducted under normal irrigation and natural non-irrigation conditions in the field. As indicated in Table 1 , under adequate water conditions, relative to that of the wild-type controls (R15 and Zhongmian 44), the fruit branch number (P ≤ 0.02), boll number (P ≤ 0.03), plant height (P ≤ 0.03), and boll fresh weight (P ≤ 0.03) of the transgenic lines 4, 6, 7 and 9 significantly increased, and the yield of seed cotton per plant rose by 18.20%, 11.22%, 18.25%, and 14.24%, respectively (P ≤ 0.03), with a significant decrease in boll shedding (P ≤ 0.04). Notably, under natural drought stress, the AtSTZ1 transgenic plants displayed better agronomic traits than did the wild-type plants ( Table 1 ). The yield of cotton seed fiber from transgenic lines 4, 6, 7 and 9 increased markedly, by 21.88%, 20.82%, 25.81% and 23.04%, respectively (P ≤ 0.02), relative to that of the wild type controls. Besides, the fruit branch number (P ≤ 0.02), boll number (P < 0.01), boll fresh weight (P ≤ 0.02), and plant height (P < 0.01) have an obvious increase, while boll shedding per plant (P ≤ 0.03) decreased in the four transgenic lines in contrast to the wild type controls ( Table 1 ). Importantly, the relative cotton fiber yield of the transgenic lines demonstrated a more significantly increase under natural arid environment than under normal, well-watered conditions ( Table 1 ). Under natural drought condition, the field performance (including cotton fiber yield, boll fresh weight, boll number, fruit branch number, and plant height) of AtSTZ1 transgenic cotton lines was positively correlated with the expression level of AtSTZ1 ( Table 1 ). Our findings conclusively demonstrated that the overexpression of AtSTZ1 in cotton not only enhanced drought resistance and simultaneously improved fiber yield in the field.

To evaluate whether AtSTZ1-overexpressing cotton plants exhibit enhanced drought tolerance, we conducted drought resistance tests during seed germination, seedling development, and the reproductive phase, respectively. Seeds from wild-type and AtSTZ1-overexpressing transgenic lines 4, 6, 7 and 9 were sown on 10% PEG6000-saturated filter papers ( Figure 2A ). Figure 2B shows the germination rates of the wild-type and AtSTZ1 overexpression seeds in the condition containing 10% PEG6000. No significant variation was observed in the germination of AtSTZ1-overexpressing lines and wild type counterparts on sterilized water-saturated filter paper, with nearly complete germination in all seeds after 6 days (P ≥ 0.36, Figures 2A, B ). Whereas, seeds from AtSTZ1-overexpressing plants germinated earlier and more rapidly than did from the wild-type controls ( Figures 2A, B ) under PEG6000 treatment. After 6 days, approximately 50-60% of the transgenic seeds had germinated, compared to only about 20% of the wild-type seeds (P ≤ 8.11E-10, Figure 2B ). The above findings suggest that the AtSTZ1 overexpressing transgenic cotton significantly enhanced drought tolerance during seed germination.

To assess the effect of AtSTZ1 overexpression on drought tolerance, transgenic cotton plants with an overexpression of AtSTZ1 were cultivated in a greenhouse under drought conditions. For evaluating drought resistance during the early seedling stage, seeds from four homozygous T₃ generation lines (-4, -6, -7 and -9) of AtSTZ1-overexpressing transgenic cotton were planted in nutrition pots, and watering ceased when the cotton plants developed two cotyledons. As depicted in Figure 3A , no discernible difference was observed between the transgenic cotton lines and the wild-type plants during seedling development on the 15th day without watering. However, as drought stress persisted on days 25 to 28, cotyledons of the wild-type control were entirely shed, whereas AtSTZ1 transgenic plants exhibited a significant delay in leaf-wilting phenotypes and reduced wilting percentages in contrast to those of wild type plants. Specifically, over 90% of the transgenic plants remained healthy ( Figures 3A, B ), and the wilting rate of the transgenic lines was only 7-8% (P ≤ 9.15E-5) after 25 days of water deprivation, while the wilting rate of the wild type controls was 20%. Following drought treatment for 28 days, the wilting rate (30-39%, P ≤ 8.74E-7) of the transgenic plants was notably lower than the wild-type (over 90%). After 28 days of drought stress and a subsequent two-day recovery period, the leaves of most of the transgenic plants returned to normal and still grew healthily, whereas almost all of the wild type plants fully wilted and even died ( Figures 3A, B ). In addition, we observed a significant increase (0.56-0.68 fold) in the dry biomass of the AtSTZ1 overexpressing transgenic lines under drought stress in comparison to the wild-type counterparts (P ≤ 3.83E-12, Figure 3C ).

To assess drought resistance during the reproductive phase, thirty-five days old greenhouse-grown plants with well-watering went through a 40-day period without watering. As shown in Figure 3D , the wild-type control exhibited a pronounced drought-stress phenotype, with leaves almost completely wilted. In contrast, all four transgenic lines maintained a relatively normal state of growth. These findings indicate that AtSTZ1 can dramatically improve the drought tolerance of cotton plants in different growth and development periods. On the whole, our results indicate that overexpression of AtSTZ1 improves drought resistance in cotton.

We observed that four AtSTZ1-overexpressing transgenic cotton lines (-4, -6, 7, and -9) exhibited stronger growth than the wild-type controls in the field in Yuncheng, Shanxi Province, China ( Figures 1C, F ; Supplementary Figure S2 ). The AtSTZ1-overexpressing transgenic cotton plants not only produced significantly larger bolls ( Figures 1C, F ; Supplementary Figure S2 ), but also developed an expanded root system ( Figures 1C, F ; Figures 4A, B , the roots were cut from AtSTZ1-overexpressing transgenic cotton plants treated with drought for 15 days and 25 days, respectively) characterized by increased length of primary roots ( Figures 4B, E ; P ≤ 5.81E-8) and increased numbers of lateral roots ( Figure 4C , P ≤ 2.8E-10), significantly augment of root dry biomass ( Figure 4F , P ≤ 1.61E-7), and an increase of roots cell size in length ( Figures 4D, H ; P ≤ 4.92E-51). Therefore, the average number of root cells per apparent area in transgenic lines 4 and 6 was obviously less than the wild type control ( Figure 4G , P ≤ 2.25E-11). This enhanced root system could potentially benefit cotton plant growth and drought tolerance.

We measured stomatal size and number in the leaves of AtSTZ1-overexpressing transgenic plants and the wild type controls. Both the length (P ≤ 1.66E-35) and width (P ≤ 3.26E-26) of the stomatal guard cells were larger in two transgenic cotton lines than the wild-type control ( Figures 5A, B ). The average stomatal densities of the transgenic lines 4, 6, 7 and 9 were respectively 26.04%, 22.26%, 28.45% and 27.13% less than in the wild type counterparts, respectively ( Figures 5A, C ; P ≤ 1.91E-16). The observation that epidermal cells in the transgenic lines overexpressing AtSTZ1 was larger compared to wild type (P ≤ 2.49E-9), which resulted in a notable decrease in stomatal density of transgenic lines ( Figure 5D ). Consequently, the AtSTZ1-overexpressing plants exhibited greater water-use efficiency (WUE, Figure 5E , P ≤ 4.72E-6), greater photosynthetic rates ( Figure 5F , P ≤ 4.17E-9), lower stomatal conductance ( Figure 5G , P ≤ 2.36E-5), and lower transpiration rates ( Figure 5H , P ≤ 3.98E-5) than did the wild-type controls. Accordingly, those data revealed that the increase in WUE of the transgenic cotton lines was attributable to a decrease in stomatal density and conductance.

We compared several physiological parameters between AtSTZ1 transgenic lines and the wild type plants following drought treatment. Under normal well-watered conditions, malondialdehyde (MDA), proline, chlorophyll, and soluble sugar contents; electrolyte leakage; and peroxidase (POD), catalase (CAT), and superoxide dismutase (SOD) activities were similar in both the transgenic lines and the wild-type control ( Figure 6 ). However, following drought stress, the MDA content ( Figure 6A , P ≤ 1.83E-7) and electrolyte leakage ( Figure 6B , P ≤ 1.97E-5) in the wild-type cotton plants were greater than those in the transgenic lines, suggesting that there was less oxidative damage in the transgenic cotton plants than in the wild-type control plants. Moreover, the contents of soluble sugars ( Figure 6C , P ≤ 0.0002), proline ( Figure 6D , P ≤ 0.0001) and chlorophyll ( Figure 6E , P ≤ 2.68E-6) in the AtSTZ1-overexpressing cotton lines increased significantly relative to the wild-type control after drought treatment. Notably, the activities of reactive oxygen species (ROS)-scavenging enzymes, such as SOD ( Figure 6F , P ≤ 1.51E-5), POD ( Figure 6G , P ≤ 2.08E-5) and CAT ( Figure 6H , P ≤ 4.26E-6), were markedly greater in the transgenic lines compared to those of in the wild type plants following drought treatment. Hence, our data display that the transgenic cotton lines overexpressing AtSTZ1 exhibited an enhanced drought resistance in comparison to the wild-type plants, likely because it better protects against oxidative damage during drought stress.

In order to deeper explore the regulatory function of AtSTZ1 in cotton in responding to drought stress, we performed transcriptomic analysis to identify differentially expressed genes (DEGs) under both irrigated and drought conditions. A Venn diagram ( Figure 7A ) displays the DEGs in transgenic lines 4, 7 and 9 and in the wild-type plants (R15 and Zhongmian 44) (control vs drought). A total of 5860 DEGs (including 2906 downregulated and 2954 upregulated genes) and 5805 DEGs (comprising 2824 downregulated and 2981 upregulated genes) were identified in the wild-type controls (Zhongmian 44 and R15) following drought treatment, respectively ( Supplementary Table S2 ). In AtSTZ1-overexpressing lines 4, 7 and 9, 392 DEGs, including 188 downregulated genes and 204 upregulated genes, were identified under drought stress ( Supplementary Table S2 ; Supplementary Appendix S1 ; Figure 7A ). 94 DEGs (comprising 42 downregulated and 52 upregulated genes) were identified in three AtSTZ1-overexpressing lines and the wild type plants following non-stressful growth environments. In contrast to those in the wild type controls, 106 DEGs (including 56 downregulated and 60 upregulated genes) were identified in those three AtSTZ1-overexpressing lines under drought treatment ( Supplementary Appendixes S1 - 2 ; Supplementary Table S2 ; Figure 7A ). The differentially regulated transcripts of these 106 DEGs are engaged in a variety of biological processes, cellular compartments, and molecular functions according to the functional classification via GO enrichment ( Figure 7B ). The expression levels of these 106 DEGs in AtSTZ1-overexpressing lines 4, 7 and 9 and the wild-type controls (Zhongmian 44 and R15) following non-irrigated conditions are depicted in heatmaps ( Figure 7C ).

Among the 106 DEGs, 60 were upregulated ( Figure 7D ; Supplementary Table S3 ), encompassing major functional gene classes. These included an E3 ubiquitin-protein ligase (GH_D09G2516, associated with stomatal conductance) (Tong et al., 2021), a calmodulin protein (GH_D12G2948), a calcium-binding protein (GH_A05G2319), and a peroxidase (GH_D12G2674, associated with root antioxidation) (Luo et al., 2016). Two aquaporins (GH_D03G0641 and GH_D13G2562) related to the regulation of stomatal response and root growth have been recognized (Kapilan et al., 2018; Guo et al., 2022). The expression of six kinases, including a PR5-like receptor kinase (GH_A01G0439, associated with stomatal conductance) (Baek et al., 2019), a proline-rich protein kinase (GH_D02G0727), a receptor protein kinase (GH_A01G2189, linked to regulating root hair growth) (Wei and Li, 2018), and three Ser/Thr protein kinases (GH_A04G0041, GH_A12G2843 and GH_A02G1960), were also upregulated. Seven transcription factors, comprising an AP2/ERF dehydration-responsive element-binding protein (GH_D06G1018, associated with root development) (Lee et al., 2016; El-Esawi and Alayafi, 2019), a zinc finger protein (GH_A07G1346, linked to regulating the function of stomata) (Han et al., 2021), a BLH/BELL transcription factor (GH_D13G0313, regulating secondary cell wall biosynthesis) (Ma et al., 2019), a bHLH (GH_D03G1091, associated with stomatal development) (Yang et al., 2023), a NAC (GH_A01G0644, linked to root development) (Liu et al., 2014), and two MYBs (GH_D08G0405 and GH_D12G2894, controlling stomatal size and opening) (Butt et al., 2017; Chen et al., 2015) were identified. Additionally, 10 heat shock proteins (GH_A09G1779, GH_A02G0345, GH_A07G0271, GH_A09G2360, GH_D08G1971, etc.) linked to ROS scavenging (Hou et al., 2018; Zhang and Huang, 2020), were dramatically upregulated in AtSTZ1-overexpressing lines following drought conditions. Our RNA-seq results are in alignment with the observed phenotypic traits exhibited by the transgenic lines.

To verify the accuracy of drought-response relative gene expression patterns of the DEGs, the expression levels of 20 representative upregulated DEGs from the AtSTZ1-overexpressing plants were determinated by qRT-PCR ( Figure 7E ). The qRT-PCR validation of the transcriptional levels of the drought-related genes screened exhibited congruence with the RNA-seq transcript abundance data, demonstrating the accuracy and reliability of the transcriptomic data.