Seeds from the wild-type (WT) and the homozygous transgenic plants T2 generation (T₃-26, T₃-45 and T₃-48) were collected and grown hydroponically in a growth room with 65-70% relative humidity, 28 ± 2°C daytime and 18 ± 2°C nighttime temperatures, a photoperiod of 14 h light/10 h darkness at a light intensity of approximately 4000 Lx. Seeds were first germinated in an I-shaped square seedling sponge with Hoagland’s solution. The germinated seeds were then transferred to vermiculite for growth. Tobacco plant seedlings with three true leaves were transplanted into hydroponic plastic pots for the experiments.

For poly-ethylene glycol (PEG-6000) treatments, 10-week-old seedlings were transferred to Hoagland’s nutrient solution supplemented with 0% PEG-6000 (Control) and 20% PEG-6000 (Drought) for cultivation. And seedlings were sampled separately after PEG-6000 treatment for 6h, and plant phenotypes were recorded. These experiments were performed in three biological replicates. Harvested samples were immediately frozen in liquid nitrogen and then stored at −80°C for subsequent experiments.

For heat treatments, 10-week-old seedlings were maintained at 45°C for heat stress (65% relative humidity) in growth chambers, and seedlings were sampled separately after heat treatment (0, 48h), and plant phenotypes were recorded. And fresh PVY-infected or TMV-infected tobacco leaves were frozen in liquid nitrogen and thoroughly crushed, then diluted 1:100 with water (Ren et al., 2021). The transgenic and WT plants were inoculated with PVY or TMV by rubbing, and susceptibility symptoms were assessed after 5 or 7 days. These experiments were performed in ten biological replicates. Harvested samples were immediately frozen in liquid nitrogen and then stored at −80°C for RNA isolation.

The gene knockdown model of NtCPS2 was constructed using CRISPR/Cas9 gene editing technology in tobacco (Nicotiana tabacum cv. 8306) to inhibit the function of NtCPS2 in vitro (Zhang et al., 2020). Based on the provided mRNA sequences and corresponding genomic sequence information, 2 CRISPR target sites ( Table S1 ) were designed to improve the gene targeting efficiency. The target loci PCR amplification primers were designed ( Table S2 ), and the primers were sent to Tianyi Huiyuan Biotechnology Co. for synthesis. After primer synthesis, the fragment containing the target site was amplified by overlap extension PCR. PCR fragment was cloned into the final CRISPR expression vector using recombinase from Nanjing Novozymes Biotechnology Co. The constructed CRISPR vector was electrotransferred into E. coli and positive clones were screened based on colony PCR for transformation of Agrobacterium and genetic transformation of tobacco strains.

Mutation detection of target gene target site sequences in edited plants

Tobacco genomic DNA was extracted by a modified CTAB method. samples T1-26, T1-45, and T1-48 were ground using a grinder, and then genomic DNA was extracted using a CTAB extraction buffer. DNA samples were air-dried and added to ddH2O for 100 μL.

The primers 17 KN48-df (5’→3’): ATCATAGCGGAATTGTTTGTCTC and 17 KN48-dr (5’→3’): TCCGTATAGATACCTAAGCGATCTG were designed to amplify the target bands and purified. The sequencing results were compared with the template sequences using DANMAN 6.0 software. Amplification reaction system: 2×PCR Mix10μL each primer 0.3μL, ddH₂O 8. 9μl, tobacco genome dna0.5μl. reaction procedure: 94°C pre-denaturation 5 minutes, 32 cycles (94°C30s, 56°C30s, 72°C40s), 72°C5 minutes, 25°C1 minute. The above PCR amplification products were detected by electrophoresis using 1% lipose gel.

The relative expression of Nicotiana tabacum ribosomal protein L25 (L25, L18908) was stable across treatments, tissues, and periods and was suitable for use as an internal reference. Schmidt et al. evaluated the expression stability of eight commonly used tobacco internal reference genes based on 22 samples of K326, which were in descending order of L25 > EF-Ia > Ntubc2 > PP2A > 18SrRNA > Actin > B-Tubulin > a-Tubulin (Schmidt and Delaney, 2010).

Four promoter sequences (pGhU6.1, pGhU6.4, pGhU6.7, and pGhU6.9) were amplified from the tobacco genome and verified with Sanger sequencing. Promoter pGhU6.9 was finally chosen for vector modification because of its high identity with that of AtU6-26. Its Bsa I restriction site was mutated to avoid multiple Bsa I sites in the final vector. The pRGEB32 plasmid, a gift from Xie et al. (2015), was linearized with Hind III and Sbf I double digestion, resulting in deletion of the gRNA terminator fragment. The promoter pGhU6.9 was assembled with the gRNA-terminator segment using an overlapping PCR method. The assembled fragment was inserted to the linearized pRGEB32 using ClonExpressII One Step Cloning Kit (Vazyme, Nanjing, China), thus generating the pRGEB32-GhU6.9 vector. It has a hpt selection marker and was used to target a reporter gene DsRed2, which was firstly transformed into cotton with a vector carrying NPT II (Neomycin phosphotransferase II). For endogenous gene targeting, we changed the selection marker hpt with NPT II between two restriction sites, Pspx I and Xmal I, using the abovementioned one-step cloning strategy. This vector, defined as pRGEB32-GhU6.9-NPT II, was used for stable genetic transformation to knock out NtCPS2 in tobacco ( Figure S1 ).

To verify the subcellular localization of the CPS2 protein, the coding sequence of CPS2 was cloned into the GFP vector (pHBT-GFP-NOS) driven by the 35S promoter, to generate the CPS2-GFP fusion protein ( Figure S2 ). The resulting vector was then transformed into protoplasts of Nicotiana benthamiana following a previously published protocol (Liu et al., 2017; Xu et al., 2020). Afterward, the fluorescence signals of the protoplasts were observed using a confocal laser scanning microscope (Lecia sp8, Germany).

The moistened and sterilized absorbent cotton and filter paper were soaked with deionized water in a clean petri dish, then laid on the petri dish. 100 disinfected seeds were evenly ordered on the surface filter paper and cultured in an artificial climate incubator (MGC-350BP-2L, Yiheng, China). The light intensity was set at 4950 LX, the light duration was 12 h/d, the relative humidity was 60%, and the temperature was 26°C.Timed observations were made every day, and deionized water was added in time. The number of germinated seeds was counted on 7d (germination potential) and 14d (germination rate) after sowing, and the germinating standard was the radicle exceeding 1/2 of the seed length (Ye et al., 2017). Plant biomass measurements included fresh weight. The leaf fresh weight of the plant was measured 70d after seeding.

For water loss rate studies, the leaves of 10-week-old transgenic and WT plants were detached, placed on dry philtre paper at room temperature, and weighed at time points (0,0.5, 1, 1.5, 2, 2.5 and 3 h). The water loss rates were determined on the basis of the water content of the leaves. Water loss rates were calculated based on the initial fresh weight of the plants (Zhao et al., 2018).

To understand whether leaf water loss rate mediated by NtCPS2 is related to stomata movement, leaves were detached from 10-week-old transgenic and WT plants, placed on dry philter paper at room temperature, and stomata were observed with a metalloscope (ML10, Mingmei, China) and stomata rate calculated at time points 0 and 3 h. The stomata size was determined by the stomata movement. The size of the stomata that were smaller than 0.5 µm was considered as closure.

Leaf samples of WT and T₃-26, T₃-45, and T₃-48 were sampled immediately after PEG-6000 treatment for 6h. Leaf samples were measured for relative electrolytic leakage (Zhang et al., 2019), soluble protein (sPRO), hydrogen peroxide (H₂O₂) and activities of major antioxidant enzymes, including catalase (CAT), and peroxidase (POD), as described previously. H₂O₂ accumulation was detected by 3,3’-diaminobenzidine (DAB) staining as described by Sun et al. (2019). The antioxidant enzymes sPRO, H₂O₂, MDA, CAT and POD content or activities were measured using the corresponding detection kits (PC0010, BC3595, BC0025, BC0205, and BC0095, Solarbio, China) following the manufacturer’s protocols. The experiment was repeated at least three times.

ABA and GA in tobacco leaves were determined at different time points (0, 6 h) during PEG-6000 treatment using an ELISA kit provided by China Agricultural University and included three independent biological replicates as described by Chen et al. (2021). The detection limit of quantitative GA in the assay kit is 2–313 ng·mL−1, and that of ABA is 2–311 ng·mL−1. All samples were rapidly frozen in liquid nitrogen and stored at −80 °C prior to analysis.

The extraction, purification and determination of endogenous levels of GA3, ABA and iP + iPA by an indirect ELISA technique were performed as described by He (1993). The samples were homogenised in liquid nitrogen and extracted in cold 80% (v/v) methanol with butylated hydroxytoluene (1 mmol·L−1) overnight at 4°C. The extracts were collected after centrifugation at 10000 × g (4°C) for 20 min, the extracts were passed through a C18 Sep-Pak catridge (Waters, Milford, MA) and dried in N₂. The residues were dissolved in PBS (0.01 mol·L−1, p H 7.4) in order to determine the levels of GA3, ABA and iP + iPA. Microtitration plates (Nunc) were coated with synthetic GA3, iP + iPA, or ABA-ovalbumin conjugates in NaHCO3 buffer (50 mmol·L−1, pH 9.6) and left overnight at 37°C. Ovalbumin solution (10 mg/ml) was added to each well in order to block nonspecific binding. After incubation for 30 min at 37°C, standard GA3, ABA, iP + iPA, samples and antibodies were added and incubated for a further 45 min at 37°C. The antibodies against GA3, ABA and iP + iPA were obtained as described by Weiler et al. (1981). Then horseradish peroxidase-labelled goat antirabbit immunoglobulin was added to each well and incubated for 1 h a t 37°C. Finally, the buffered enzyme substrate (orthophenylenediamino) was added, and the enzyme reaction was carried out in the dark at 37°C for 15 min, then terminated using 3 mol·L−1 H₂SO₄. The absorbance was recorded at 490 nm. Calculations of the enzyme-immunoassay data were performed as described by Weiler et al. (1981). In this study, the percentage recovery of each hormone was calculated by adding known amounts of standard hormone to a split extract. Percentage recoveries were all above 90%, and all sample extract dilution curves paralleled the standard curves, indicating the absence of nonspecific inhibitors in the extracts.

Leaf transpiration rate (E) and stomatal conductance (GS) were measured in the culture chamber using a CIRAS-3 portable photosynthetic system (PP-Systems, UK). Environmental conditions were tightly controlled, with light intensity (PAR) set at 1000μmol/(m²·s), molar content of CO₂ at 390μmol/mol, and temperature at 25°C. The average value of 10 consecutive cycles was taken as the test value. Five representative plants were selected for each variety, and each plant was repeated for three times to get its average value.

Extraction of total RNAs and synthesis of cDNAs were performed according to the method described by Xia et al. (2018). The quantitative real-time PCR(RT-qPCR) was checked with gene-specific primers to investigate the transcription levels of some genes for cis-abiol, gibberellins and abscisic acid biosynthetic enzymes. The RT-qPCR assay was performed using SYBR Green PCR Master Mix (Tiangen Biotech, China) for 20 µL of the reaction mixture on an IQ5 Light Cycler System (Bio-Rad, Hercules, CA, USA). Relative transcript levels were calculated as described by Zhang et al. (2019), and using the L25 as the reference gene. Primers used for RT-qPCR are listed in Table S3 .

Microsoft Excel (Microsoft Corporation, USA) was used for data collection, and SPSS (version 20.0, SPSS Inc., Chicago, IL, USA) for statistical analysis of data, and RStudio (1.4.1106 RStudio, Boston USA) was used for mapping. Data were reported as mean values ± standard error (SE). Data were analyzed using one-way analysis ANOVA, and means were compared using Duncan’s multiple range test at a significance level of p< 0.05.

To investigate the gene expression patterns of NtCPS2 under different abiotic and biotic stresses, we examined the transcript levels of NtCPS2 in 10-week-old tobacco seedlings exposed to drought, heat, PVY, and TMV. As shown in Figure 1 , NtCPS2 gene expression increased significantly within 6 hours after drought treatment, and transgenic plants wilted more than WT. After hot treatment, the expression level of the NtCPS2 gene increased by 50% on average after 48 h, and the leaves of transgenic plants were browned earlier than WT ( Figure 1B ). After inoculation with PVY and TMV virus, the transgenic plants developed disease on days 5 and 7, respectively, but WT did not develop the disease. Meanwhile, the expression level of the NtCPS2 gene was significantly increased. The results confirmed that the NtCPS2 gene expression could be induced by abiotic stress and biological stress.

To monitor the subcellular location of NtCPS2, we fused the NtCPS2 gene (using its CDS sequence) to the N-terminus of a GFP reporter gene, and the resulting expression cassette was transiently expressed in tobacco leaf epidermal cells. As shown in Figure 2 , the fluorescence of the NtCPS2-GFP fusion protein was observed in the cell membrane.

Under optimal growth conditions, the seed germination and seedling stage of the three transgenic lines and WT differed on the Petri dish ( Figure 3A ), the transgenic plants were larger than the WT plants ( Figure 3A ). Then, we counted the germination potential and germination rate of the tobacco seedlings on petri dish, and found that the transgenic plants had higher germination potential and germination rate than the WT plants ( Figures 3B, C ). Similarly, the transgenic lines exhibited higher seedling fresh weight than the plants from WT ( Figure 3D ).

According to the biosynthetic pathways of diterpenoids, the same substrate, geranylgeranyl pyrophosphate (GGPP), is used for the synthesis of cis-abienol and GA. NtCPS2 knockdown positively affects GA synthesis. The GA3 contents in transgenic plants were 18.94%, 23.20% and 28.68% higher, respectively, under control treatment compared with that of the WT ( Figure 4A ). The GA3 content was not significantly changed under drought treatment from that of the control treatment. As shown in Figure 4B, the ABA contents in transgenic plants were 71.82%, 72.61% and 53.36% lower, respectively, under control treatment compared to WT. In addition, the content of ABA increased sharply under drought treatment with that of the control treatment. Meanwhile, the GA3/ABA in transgenic plants increased sharply compared with that of the WT, indicating that NtCPS2 knockdown resulted in increased gibberellin content, and decreased abscisic acid content, and then increased the ratio of gibberellin content to abscisic acid content (Figure 4C). Furthermore, the changes in GA and ABA contents in transgenic plants and WT plants under drought treatment were consistent with those under control treatment.

To further prove that NtCPS2-knockdown promotes GA synthesis by regulating the expression of gibberellin synthesis pathway genes, and the antagonistic effect of gibberellin and abscisic acid on transcription level, L25 was used as a housekeeping gene to analyse the expression of 6 key genes by qRT-PCR. These genes included four gibberellin genes (ent- acid synthase gene KS, ent- acid oxidase gene KO, and ent-acid oxidase gene KAO, GA2-oxidase gene GA2ox) and two abscisic acid genes (zeaxanthin epoxidase gene ZEP, and 9-cis epoxycarotenoid dioxygenase gene NCED), which have been shown to play a key role in gibberellin and abscisic acid synthesis. As indicated in Figures 5A-D , the relative expression levels of KS, KO and KAO genes promoting gibberellic acid synthesis in these transgenic lines were higher than those in the WT under control and drought conditions. And the relative expression level of the GA2ox gene which inhibited gibberellic acid synthesis was lower in transgenic than in WT. Meanwhile, the relative expression levels of ZEP and NCED, which promote ABA synthesis, in transgenic lines were lower than those in the WT, especially the relative expression level of NCED in transgenic lines was 59.58% lower than that in the wild type under drought conditions (Figures 5, F). These data indicate that NtCPS2 knockdown regulates the expression of gibberellin and abscisic acid-related genes, and significantly affects the expression of abscisic acid-related genes under drought stress.

Next, we investigated the functions of NtCPS2 in drought stress tolerance using transgenic tobacco. Water loss rate and stomatal aperture are often used as indicators of drought tolerance in plants. The stomatal openings of the NtCPS2 transgenic NtCPS2 plants were significantly larger under drought stress at room temperature than those of the WT ( Figure 6A ). And after 3 hours of dehydration stress, the stomatal closure rate of NtCPS2 transgenic plants was significantly smaller than that of WT ( Figure 6B ). Similarly, water loss from the detached leaves was recorded every half hour in both NtCPS2 transgenic lines and the plants from WT. As shown in Figure 6C , the three transgenic plants had higher water loss rates than the plants from WT.

To further detect stomatal closure and water loss of leaves, we tested stomatal conductance and transpiration rate of leaves, and found that after drought treatment for 6 hours, stomatal conductance and transpiration rate of the NtCPS2 transgenic plants were significantly higher than those of WT ( Figures 6D, E ). These tests showed that the decreased expression of NtCPS2 gene was not good for plant leaf water retention, resulting in the lower drought resistance of NtCPS2 transgenic plants than WT.

To uncover the possible physiological mechanisms underlying the enhanced drought tolerance of NtCPS2-knockdown plants, we measured the levels of hydrogen peroxide (H₂O₂) and the activities of several antioxidant enzymes in NtCPS2-knockdown plants and WT plants grown under optimal conditions and drought.

To confirm the ability of the transgenic tobacco plants to scavenge ROS, the accumulation of ROS was determined under optimal growth conditions and under drought stress. There was a difference in H₂O₂ accumulation between WT and transgenic seedlings ( Figure 7A ). After drought treatment for 6 hours, H₂O₂ accumulation in transgenic lines was higher than WT, observed as brown (DAB staining) pigments ( Figure 7A ). To quantify this difference, H₂O₂ content was measured in whole tobacco seedlings with or without drought stress. Under optimal growth conditions, H₂O₂ content was slightly higher in the transgenic line than in WT ( Figure 7B ). After 6 hours of drought, three transgenic lines contained significantly more H₂O₂ than WT, especially T₃-45 and T₃-48 ( Figure 7B ).

Next, we evaluated the activities of two antioxidant enzymes, guaiacol peroxidase (POD) and catalase (CAT) in the transgenic and WT plants ( Figures 7C, D ). The results showed that POD and CAT activities significantly increased after drought stress, however, the increase in transgenic plants was less than that in WT plants ( Figures 7C, D ).

These results suggest that NtCPS2-knockdown increases the accumulation of ROS by reducing the activities of several antioxidant enzymes.

Furthermore, to further confirm the status of the membrane system in transgenic plants, we examined the relative electrolytic leakage and soluble protein content ( Figures 8A, B ). The relative electrolytic leakage in these transgenic lines was about 25% and 42% higher than those in WT plants under suitable and drought conditions, respectively. ( Figure 8A ). Furthermore, the soluble protein content was significantly lower in the transgenic lines compared to WT, especially after treatment with drought stress ( Figure 8B ). These results evidenced that the increased membrane permeability caused by NtCPS2 knockdown is due to the decreased sPRO content, which reduces the tolerance to drought stress.