In our attempt to screen rice UGT genes responsive to abiotic stress, UGT85E1 (LOC_Os02g51900) drew our attention because of its response to drought stress. We examined the responsiveness of UGT85E1 to mannitol and exogenous ABA via real-time quantitative reverse transcription PCR (qRT-PCR). The expression level of UGT85E1 under 200 mM mannitol treatment gradually increased from 6 to 24 h, decreasing slightly at 48 h. The induction of UGT85E1 was also observed from 6 to 48 h under 100 μM ABA treatment (Figure 1A). These two treatments significantly induced the expression of UGT85E1, suggesting that this gene may be involved in drought stress. To analyze the function of UGT85E1, we created the overexpression lines and mutant lines of the UGT85E1 gene. Two overexpression lines (OE-3 and OE-5) and two CRISPR/cas9 mutant lines (ko-8 and ko-9) were subjected to drought stress for further analysis (Figure 1B, Supplementary Figure 1).

Under normal conditions, 2-week-old UGT85E1 overexpression lines and mutants showed normal and similar growth phenotypes with wild-type plants (WT). After 150 mM mannitol treatment, most mutants appeared more withered than the WT, however, the two overexpression lines performed better growth than WT (Figures 2A,C). Survival rates showed consistency with the phenotype observation (Figures 2B,D). We also conducted drought treatments using 4-week-old mutants and overexpression lines. No growth difference of mutants and overexpression lines when compared with WT can be observed under soil culture and normal irrigation conditions. After deprival of irrigation for about 1.5 weeks, some mutant plants wilted. After re-irrigating for 2 days, the mutants were more withered and showed a lower survival rate than that of WT (Figures 2E,F). However, the overexpression lines grew better than that of WT under drought conditions, showing more green, upright petiole, and a much higher survival rate (Figures 2G,H). These experiments proved that UGT85E1 is involved in drought stress response and its overexpression can significantly enhance the drought tolerance of rice.

Upon exposure to drought stress, water loss of plant leaves mainly occurred through the stomatal opening. Thus, we investigated the stomatal aperture of 4-week-old different transgenic lines under dehydration conditions. Here, the stomatal opening was classified into three types, i.e., fully open, partially open, and completely closed as shown in Figure 3A. Before dehydration, the ratio of three types of stomatal opening is roughly the same between WT, mutants, and overexpression lines. After 2 h of dehydration of detached leaves, however, the proportion of stomatal closure increased and the proportion of partially open stomata significantly decreased in overexpression line OE-3 compared with WT (Figure 3B). The mutant line ko-8 was on the contrary with less closed and more open stomata compared to WT. For example, the percentage of completely closed stomata in WT, ko-8, and OE-3 lines was 30.1, 17.3, and 70%, respectively (Figure 3B). This result showed that the drought tolerance of UGT85E1 transgenic rice was due to the reduced stomatal opening.

Then, we detected the rate of water loss from detached leaves of each UGT85E1 transgenic line within 120 min. The result indicated that the two overexpression lines exhibited lower water loss than that of WT, but the mutants showed more water loss (Figure 3C).

It is well-known that ABA can induce stomatal closure. We thus examined the endogenous ABA levels in UGT85E1 transgenic lines and WT under normal and drought stress conditions. Although ABA levels were similar for different genotypic plants under control conditions, overexpression lines showed much higher ABA levels than WT under drought stress, whereas the mutants showed lower ABA levels (Figure 3D). This result prompted us to hypothesize that UGT85E1 may be involved in regulating ABA biosynthesis in drought stress. To confirm this, the transcript levels of ABA biosynthesis and responsive genes were analyzed. As shown in Figure 4A, the expression levels of ABA biosynthesis genes OsABA1, OsABA2, OsABA4, and OsNCED1 and ABA-responsive genes OsABI5 and OsbZIP23 were highly upregulated by drought stress in the UGT85E1 overexpression lines as compared with WT, while in the mutants, they were down-regulated. The upregulation or downregulation of ABA biosynthesis and responsive genes would change the sensitivity of seed germination to exogenously applied ABA. As shown in Figures 4B,C, our investigation indicated that the root length of overexpression lines was shorter than WT seedling under 5–10 μM of ABA treatment, whereas the mutants showed the longer root. These results suggest that the role of UGT85E1 in enhancing drought tolerance may be due to the enhanced ABA response in rice.

Stresses often lead to the accumulation of hydrogen peroxide (H₂O₂) in plant cells, and the ability of cells to remove H₂O₂ reflects the level of resistance to stresses (Li et al., 2004). It was reported that ABA could significantly promote the activities of ROS scavenging enzymes such as SOD (Superoxide dismutase), APX (Ascorbate peroxidase), and CAT (Catalase) under stress conditions (Jiang and Zhang, 2002; Xu et al., 2018). Given that UGT85E1 is involved in ABA response as mentioned above, we thus examined ROS accumulation and the relative expression of several genes encoding ROS scavenging enzymes in UGT85E1 overexpression lines and mutants. 3,3′-Diaminobenzidine (DAB) staining indicated that H₂O₂ accumulated more in ugt85e1 mutants than WT, the overexpression lines showed the opposite changes under mannitol treatment (Figure 5A). Transcript levels of several genes encoding ROS scavenging enzymes were significantly upregulated in overexpression lines, but down-regulated in ugt85e1 mutant lines compared with WT after 250 mM mannitol treatment (Figure 5B). These results indicated that UGT85E1 overexpression promotes the scavenging of ROS.

As osmoprotective molecules, proline and sugar are recognized to protect cells against osmotic stress. Numerous studies have demonstrated that the proline and sugar accumulation is enhanced in response to different abiotic stresses and also by the induction of ABA (Verslues et al., 2007; Szabados and Savoure, 2010; Vishwakarma et al., 2017; Živanović et al., 2020). Here, we measured the contents of proline and soluble sugar in the UGT85E1 overexpression lines and mutants subjected to drought treatment. Our results showed that much more proline and soluble sugar were accumulated in OE-3 and OE-5 plants, but less in ko-8 and ko-9 compared with WT (Figures 5C,D).

We also investigated the expression changes of four stress-related genes (OsRAB16, OsRAB21, OsDREB2B, and OsDREB1A) after 250 mM mannitol treatment. The results showed that the expression levels of four genes in overexpression lines were significantly higher than that in WT, their expression levels in mutant lines were lower than that in WT (Figure 5E). The above-mentioned results suggest that the UGT85E1 gene exerts its important influence on these stress-related genes, directly or indirectly, and plays a broad role when facing stress conditions.

To further investigate the role of UGT85E1 in scavenging ROS, we also tested the detached leaves of the overexpression lines and mutant lines with the exogenously applied H₂O₂ and herbicide methyl viologen (MV), which can generate ROS locally in chloroplasts. As shown in Figure 6A, under the control condition, there was no difference in the green color of the detached leaves from each of the lines soaked in distilled water for 1 week. When treated with 2% H₂O₂ or 10 μM MV, we can see that most overexpression samples were greener than WT, and mutant samples were browner than WT. The determination of chlorophyll contents also confirmed this observation (Figure 6B). This result again demonstrated that the overexpression of UGT85E1 can enhance the ability to scavenge H₂O₂ and to resist oxidative stress.

To further verify the effect of UGT85E1 in enhancing drought stress tolerance, we overexpressed UGT85E1 in Arabidopsis and investigated the drought tolerance using two overexpression lines OE-4 and OE-7 (Figure 7A). Dehydration assay indicated that Arabidopsis OE-4 and OE-7 had a lower water loss rate than WT within 120 min (Figure 7B). When 3-week-old soil-grown seedlings were subjected to drought stress for 1 week and followed by re-irrigation for 24 h, we can see that the overexpression lines grew better than WT. Not only the leaves of OE-4 and OE-7 were fully extended but almost all survive. However, about half of the WT wilted (Figures 7B,C,D). These results suggest that UGT85E1 overexpression in Arabidopsis also contributes to drought tolerance.

We also investigated the ROS levels in Arabidopsis overexpression lines and WT plants under stress conditions. The DAB and nitroblue tetrazolium test (NBT) staining indicated that ROS accumulated much less in UGT85E1 transgenic Arabidopsis lines than in WT (Figures 8A,B). We further analyzed the expression of the abiotic stress-related genes and found that the transcript levels of AtRD29A, AtRD29B, AtDREB1A, and AtDREB1B in transgenic Arabidopsis lines were higher than in WT (Figure 8C). These ectopic expression data provide more powerful evidence for the involvement of UGT85E1 in enhancing drought stress tolerance.

The WT rice plant and also the genetic background of the transgenic rice plants were ZH11 (Oryza sativa L. subsp. japonica cv. Zhonghua No.11). Rice seedlings were hydroponically grown or soil-grown in a growth chamber under standard growth conditions. For Arabidopsis thaliana, the ecotype Columbia 0 (Col-0) was used as wild type and transformed by the constructed plasmids. Seeds were sown on Murashige and Skoog (MS) medium, cold-treated for 3 days at 4°C, and transferred to controlled growth chamber under 16 h light/ 8 h dark condition with a fluency rate of 100 μmol/s/m² of white light (produced by cool-white fluorescent lamps) at 22°C.

To construct the rice overexpression vector, the full-length cDNA (1,461 bp) of UGT85E1 were amplified and cloned into pUN1301 binary vector under the control of maize ubiquitin promoter. For the ectopic expression of UGT85E1 in Arabidopsis, the full-length cDNA of UGT85E1 were cloned into a pBI121 binary vector under the control of CaMV 35S promoter. For the knocking-out of UGT85E1 in rice, the CRISPR/Cas9 BGK01 kit was purchased from Biogle Biotechnology Company (Hanghzhou, China) (http://www.biogle.cn) and the knocking-out vector was constructed according to the instructions. Rice transformation was carried out by the Agrobacterium-mediated method (Hiei et al., 1994), and the flower dipping method was used for Arabidopsis transformation (Clough and Bent, 1998).

For the hydroponic treatment, rice seedlings grown for 3 days were treated with 150 mM mannitol for 1 week. Non-treated seedlings were used as the control. For the drought stress treatment of soil-grown seedlings, uniformly growing seedlings were transplanted to the soil and grown for 4 weeks under normal irrigation conditions. Then, irrigation was stopped for about 10 days and re-irrigation was provided. After the recovery for 2 days, pictures of growth phenotypes were taken and the survival ratio (the number of surviving plants over the total number of treated plants in the pot) of each line was calculated. For oxidative stress tolerance assays, leaf disks from 3-week-old seedlings were soaked into H₂O₂ or MV according to the indicated concentrations for 5 days, those leaf discs soaked in water were used as control.

To test the ABA sensitivity at the germination stage, seeds of the UGT85E1 transgenic lines were germinated on 6-well plates containing 5 μM or 10μM ABA, the deionized water was used as the control. These seeds were germinated under 10 h light/14 h dark condition at 28°C in a growth chamber for 5 days, and then the root length of each seedling was measured.

For the drought treatment of UGT85E1 transgenic Arabidopsis, 3-week-old soil-grown seedlings were kept without irrigation for 6 days and then again provided with watering. After recovery for 24 h, the survival rate was calculated and growth performance was photographed.

The detached leaves of 3-week-old rice were placed in the air for 2 h. Water is naturally lost to promote stomatal closure. Firstly, the leaves were fixed with 2.5% glutaraldehyde fixator at 4°C for 12–16 h, and then washed with 1×phosphate buffered saline (PBS) for 5 times; Secondly, gradient ethanol dehydration, and then drying the samples in the dryer, spraying gold in ion sputtering device, finally a scanning electron microscope (FEI Quanta250 FEG, Oregon, USA) was used to observe and take photos.

Leaves of 2-week-old UGT85E1 transgenic rice and WT plants were detached and kept in the air for 2 h. Approximately 50 mg of leaf tissues were collected and rapidly frozen with liquid nitrogen, followed by homogenizing the sample in PBS (PH7.4). Samples were centrifuged for 20 min, and the supernatant was collected carefully for determination of ABA. The ABA concentration was assayed according to the instructions provided by the plant hormone abscisic acid ELISA Kit (Shanghai Fusheng Industrial Co., Ltd, Catalog number: A112641-96T, Shanghai, China).

For calculation of water loss rate, the detached leaves of the 3-week-old rice plants were firstly weighed to record the fresh weight (FW), and then the samples were dehydrated in the air. The samples were weighed every 20 min to record the dry weight (DW; Xiong et al., 2014). Ten plants of each line were used in each replicate with three replicates. The water loss rate is defined as the (FW-DW)/FW×100%. For the determination of H₂O₂ accumulation under drought stress, 3-week-old rice plants or 4-week-old Arabidopsis plants were treated with 250 mM mannitol for 24 h. The leaves were subjected to 0.1% DAB staining for 16 h (pH = 5.8), followed by incubation in the de-staining buffer (ethanol/lactic acid/glycerol=3:1:1; Thordal-Christensen et al., 1997). Similarly, NBT-staining for superoxide detection was conducted as described by Wohlgemuth et al. (2002).

To determine the content of proline and soluble sugar, the detached leaves of 4-week-old rice plants were dehydrated at room temperature for 1 h, the non-treated plants were served as control 0.1 g of leaves were homogenized with 2 ml 3% sulfosalicylic acid and followed by centrifugation. 1 ml supernatant was taken and an equal volume of glacial acetic acid and ninhydrin were added, followed by inoculation in boiling water bath for 60 min. After cooling, 3 ml toluene was added and the content of proline was determined at 520 nm (Nakazawa et al., 1982). Similarly, to measure the content of soluble sugar, 0.1 g of the harvested leaves were homogenized in 2 ml distilled water and boiled for 30 min, followed by centrifugation. About 1 ml supernatant was taken, and 2 ml anthrone and 2 ml sulfuric acid were added, followed by a full mix and inoculation in boiling water bath for 10 min, then soluble sugar was measured at 620 nm (Morris, 1948).

For analyzing the induced expression of UGT85E1, 2-week-old hydroponically grown rice seedlings were subjected to treatments with 200 mM mannitol or 100 μM ABA, and the leaf tissues were harvested at 0, 6, 12, 24, and 48 h time points. For detecting expression levels of the ABA biosynthesis and signal genes, ROS scavenging enzyme genes, and stress-related genes under stress conditions, 2-week-old hydroponically grown rice seedlings or Arabidopsis seedlings were treated with 250 mM mannitol for 12 h, non-treated materials were used as the control. All these harvested leaf samples were then rapidly frozen in liquid nitrogen and stored at 80°C for further expression analysis.

Total RNA was extracted from plant samples using the Trizol reagent (TaKaRa, Japan). Reverse transcription was performed using a Prime Script RT reagent kit (TaKaRa). qRT-PCR was performed with a CFX Connect Thermal Cycling System (Bio-Rad, USA) using a SYBR Green PCR Master Mix kit (TaKaRa). The gene expression levels were normalized with the reference gene OsActin1 in rice and the reference gene AtActin2 in Arabidopsis. The fold-change in the expression of each target gene was calculated by using the 2^−ΔΔCt method. Primer information for the qRT-PCR assay is included in Supplementary Table 1.

All data in this work were obtained from at least three biological replicates with three technical replicates each. Statistically significant differences were determined by one-way ANOVA, followed by Tukey's test (P < 0.05).