Rice (Oryza sativa ssp. japonica) “Zhonghua 11” (ZH11) and Arabidopsis thaliana (Ecotype Col-0) served as the study’s wild type plants. In a greenhouse with a 14 h/10 h light-dark cycle, 28°C, and 50%–60% humidity, rice seedlings were cultivated with 1/2 Hoagland’s solution (HS) and stroma (nutritive soil: vermiculite = 3:1), respectively. Arabidopsis thaliana seedlings were cultivated with MS medium and stroma (nutritive soil: vermiculite = 3:1) in a controlled environment with a 16 h/8 h light-dark cycle, 100 µmol/s^-1/m^-2 of light intensity, 22°C, and 40% - 60% humidity.

The Hangzhou Biogle Vector Construction Kit (http://www.biogle.cn) was used to generate the CRISPR/Cas9-mediated mutation construct of OsDUGT1. To construct the rice overexpression plasmid, the full-length CDS of OsDUGT1 (LOC_Os01g41430) was amplified from rice cDNA using ApexHF HS DNA Polymerase CL (AG Company, China), and then cloned into the pUN1301 vector under the control of the maize ubiquitin promoter. The Agrobacterium-mediated genetic transformation of rice was completed by BioRun Biotechnology Co., Ltd (Wuhan, China). In addition, the full-length cDNA of OsDUGT1 were cloned into a pBI121 binary vector under the control of CaMV 35S promoter, which was subsequently introduced into Arabidopsis via the flower-dipping technique in order to achieve ectopic expression of OsDUGT1. The primers used to construct the plasmids are listed in Supplementary Table S1 .

The high temperature stress experiments of rice and Arabidopsis were carried out in the same incubator and same conditions to keep environmental consistency (including temperature, light intensity, light-dark cycle, humidity etc.)

DAB and NBT staining are performed in accordance with the procedure outlined by literature (Dolui et al., 2024), with slight modifications. Rice leaves were collected following a 24 h of heat stress treatment and submerged in 0.5 mg/mL NBT solution in 10 mM phosphate buffer (pH7.6) before being vacuum-infiltrated for ten minutes. Samples were then dyed for 12 h at room temperature in the absence of light. After discarding the dyeing solution, samples were cooked for 10 minutes in the fixed solution (ethanol: lactic acid: glycerin = 3:1:1) until all of the chlorophyll was gone, at which point they were photographed.

Rice leaves with same heat treatment were gathered and submerged in a 1 mg/mL dimethylbenzidine (DAB) solution in 50 mM Tris-HCl (pH3.8) for 15 minutes before being vacuum-infiltrated. Samples were then dyed for 16 h in the dark at room temperature. Following a 15 minutes of decolonization in a water bath at 70°C with 90% ethanol, the pigmented samples were photographed.

Chlorophyll fluorescence parameters were recorded by a PAM fluorometer (Walz; PAM 2500) after dark adaptation (30 min) (Ghorbani et al., 2021).

The Plant RNA Kit (Vazyme Biotech Co., Nanjing, China) was used to extract total RNA from Arabidopsis and rice seedlings. The PrimeScript RT reagent kit with gDNA Eraser (Perfect Real Time) (Accurate Biology, Hunan, China) was used for reverse transcription and cDNA synthesis. The SYBR Green Supermix (Transgen, Beijing, China) was used for qRT-PCR. Pre-denaturation was done at 95°C for 3 minutes, and then there were 45 cycles (5 s at 95°C, 30 s at 60°C, and 30 s at 72°C). Data was taken at 72°C for each cycle. OsACTIN1 and OsUBIQ1 were used as the internal reference genes for rice gene analysis. AtACTIN2 and AtTUB2 were used as the internal reference genes for Arabidopsis gene analysis. Gene expression levels were normalized to these reference genes and the fold change in the expression of each target gene was calculated by using the 2^−ΔΔCt method. Supplementary Table S1 contained the primers for the qRT-PCR study.

Two-weeks-old OsDUGT1 transgenic and wild type rice seedlings were heat-treated for 24 h at 45°C, whereas untreated seedlings served as the control. According to the manufacturer’s instructions (Shanghai Sangon Biotech, Shanghai, China), the hydrogen peroxide (H₂O₂) content, catalase (CAT, EC 1.11.1.6), superoxide dismutase (SOD, EC 1.15.1.1), peroxidase (POD, EC 1.11.1.7), and malondialdehyde (MDA) contents were measured using the H₂O₂ Assay Kit (D799774-0100), CAT Test Kit (D799598-0100), SOD Assay Kit (D799594-0100), POD Assay Kit (D799592-0100), and MDA Assay Kit (D799762-0100), respectively. As previously mentioned, the relative ion leakage rate was measured (Fang et al., 2015).

The pGEX-4T vector (Takara, Japan) was used to clone the whole OsDUGT1 cDNA sequence. The E.coli strain BL21 was inoculated with the constructs, and the transformed single colony was cultivated for approximately 12 h at 37°C in LB medium containing 100 mg/mL ampicillin. The culture solution was then transferred to 75 mL 2×YT media containing 50 mg/mL ampicillin at 37°C until the OD₆₀₀ reached 0.6~0.9. After adding 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) and incubating for 24 h at 16°C, the OsDUGT1 fusion protein was expressed. Centrifugation at 5000 × rpm for 5 minutes at 4°C was used to harvest the cells, and they were then re-suspended in 25 mL of PBS solution containing 50 µL of phenylmethylsulfonyl fluoride (PMSF, 100 mM). After being lysed using an ultrasonic cell crusher, the supernatant was got by centrifuging suspension at 7000 × rpm for 25 minutes at 4°C. 100 µL of GST agarose beads were added and the tube was incubated for one hour at room temperature. The purified protein was then eluted using elution buffer.

A 100 µL aliquot of reaction mixture including 0.1 mM substrate, 2.5 mM MgSO4, 10 mM KCl, 100 mM Tris-HCl (pH8.0), 14 mM 2-mercaptoethanol, 5 mM UDP-glucose, and 1 μg of OsDUGT1 fusion protein was used for the enzyme activity experiment. After 30 minutes of incubation at 30°C, 100 µL of methanol was added to the reaction mixtures to quench the reaction. The mixtures of enzymatic reactions were centrifuged at 4°C for 10 minutes at 12,000 × rpm. Liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis was subsequently performed on the supernatant.

The secondary metabolomics experiment was carried out by Metware Co. (Wuhan, China). First, the sample was freeze-dried with Scientz-100F, and then powdered with MM 400, Retsch (30 Hz, 1.5 min). Next, 50 mg of the sample was weighed and added to pre-cooled 70% methanol for extraction. Following centrifugation (rotation speed 12000 rpm, 3 minutes), the sample was filtered using a microporous filter membrane (0.22 μm pore size), the supernatant was absorbed for UPLC-MS/MS analysis. The following are the primary liquid phase conditions: 1 Column: Agilent SB-C18 1.8 µm, 2.1 mm, 100 mm; (2) The mobile phase consists of acetonitrile (with 0.1% formic acid added) and ultra-pure water (phase A); (3) elution gradient: A/B phase: 95:5 V/V at 0 min, 5:95 V/V at 9.0 min, 5:95 V/V at 10.0 min, 95:5 V/V at 11.1 min, 95:5 V/V at 14.0 min; (4) flow rate: 0.35mL/min;temperature:40°C; injection volume: 2 µL. Next, we attached an ESI-triple quadrupole-linear ion trap (QTRAP)-MS to the effluent. Triple quadrupole (QQQ) scanning and liquid chromatographic analysis were performed using the Q TRAP mass spectrometer and API 6500 Q TRAP LC/MS/MS system. This system ran in positive ion mode and had an ESI Turbo Ion-Spray interface. The electrospray ionization (ESI) temperature of 550°C and the ion spray voltage (IS) of 5500 V (positive ion mode) or -4500 V (negative ion mode) are the primary mass spectrum conditions. The collision-induced ionization parameters are set to high, and the ion source gas I, gas II, and curtain gas were set to 50, 60, and 25 psi, respectively. QQQ scans obtained with nitrogen as collision gas for MRM experiments. Triple quadrupole mass spectrometry with multiple reaction monitoring (MRM) was used to quantify metabolites. The program Analyst 1.6.3 was used to process the mass spectrum data.

Using the Trizol Reagent, total RNA was extracted from rice seedlings. Illumina NovaSeq (Wuhan, China) was used to implement the RNA-seq libraries. The fastqc software (version: 0.11.5) was used to analyze the raw data. The hisat2 software (version: 2.0.1-beta) was used to compare the clean read data to the reference genome (https://rapdb.dna.affrc.go.jp). All RNA-seq libraries’ quantitative expression analyses were compiled using FeatureCounts (version: v1.6.0). Edge R package was used to evaluate the differentially expressed genes (DEGs) compared to wild type. Log2 FC ≥1 and p value ≤ 0.05 were regarded as significant differential genes. GO and KEGG enrichment analyses were performed on these DEGs.

At least three biological replicates and three technical replicates each provided the data present in this study. The mean ± SD value was used to report the results. One-way ANOVA followed by Tukey’s test was performed using SPSS software.

Using Genevestigator data platform (https://genevestigator.com/gv), we searched the UGT genes in rice genome to find some members that were up-regulated or down-regulated by high temperature. It was noticed that OsDUGT1 gene (LOC_Os01g41430.1) was annotated as high temperature inducible in this database. To confirm this, we treated wild type rice (ZH11) with high temperature and detected OsDUGT1 expression during treatment by qRT-PCR. We found that OsDUGT1 transcription was significantly increased under heat stress ( Figure 1A ; Supplementary Figure S1A ), suggesting it is high temperature inducible gene. To further explore its function and mechanism, we created its knockout mutants using CRISPR/Cas9 strategy, and obtained its homozygous mutant lines Osdugt1-ko18 and Osdugt1-ko75 ( Figure 1B ). Besides, we also created and identified several overexpression lines of OsDUGT1 (OE19, OE21, etc.) ( Figure 1C ; Supplementary Figure S1B ). These lines were used for subsequent studies.

Using the two mutant lines (ko18, ko75) and two overexpression lines (OE19, OE21), we investigated the potential function of OsDUGT1 in response to high temperature for the hydroponically grown rice and soil-grown rice. We discovered that mutants ko18 and ko75 consistently displayed a more susceptible phenotype to the heat stress with reduced survival rates in comparison to wild type ZH11, regardless of whether the rice was cultivated hydroponically or in soil. In contrast to the wild type, the overexpression lines OE19 and OE21 consistently exhibited the improved resistance to heat stress and provided greater survival rates ( Figure 2 ). These findings showed that OsDUGT1 confers heat tolerance on rice and has a beneficial regulatory role in the response to high temperatures.

When plants are consistently exposed to severe heat stress, intracellular ROS levels will increase significantly, thus breaking the ROS homeostasis and causing oxidative damage (Zhao et al., 2018b). To find out how OsDUGT1’s physiological mechanism affect rice’s ability to withstand heat, we measured H₂O₂ content and discovered that while there were no appreciable differences between the mutant lines, wild type lines and overexpression lines under the control condition, however, the H₂O₂ content in Osdugt1 knockout mutant lines after high temperature stress was significantly higher compared to ZH11 and overexpression lines, increased by 0.2 and 0.5 folds, respectively ( Figure 3A ). These differences were statistically significant. This suggests that at high temperatures, the loss of OsDUGT1 gene function causes a large amount of ROS accumulation. Besides, using diaminobenzidine (DAB) and nitroblue tetrazolium (NBT) staining techniques, we further explored the impact of OsDUGT1 in improving antioxidant capacity. According to the staining results, the rice leaves of the ko18 and ko75 mutant seedlings stained more deeply than those of ZH11 when exposed to high temperature stress, while the leaves of the OE19 and OE21 transgenic seedlings stained lighter than those of ZH11; however, there was no discernible difference between the transgenic and ZH11 seedlings’ leaves under normal circumstances ( Supplementary Figure S2 ). To cope with oxidative stress, plants often rely on enzymatic or non-enzymatic antioxidant defenses. Antioxidant enzymes including catalase (CAT), superoxide dismutase (SOD), and peroxidase (POD) are thought to be the most effective kinds of defense. After heat stress, our measurements showed that the overexpression lines’ CAT, SOD, and POD enzyme activities were significantly higher than ZH11, whereas those enzyme activities in mutant lines were lower than ZH11 ( Figures 3B–D ). These results suggested that OsDUGT1 responds to heat stress by eliminating reactive oxygen species to a greater degree, thus alleviating plant damage.

Usually, high-temperature increases membrane fluidity, causes changes in cell membrane permeability, and results in a significant loss of cellular electrolytes, which inhibits cellular function and reduces heat resistance. On the other hand, membrane lipid peroxidation in plants is detected by measuring malondialdehyde (MDA). MDA is a widely used marker of oxidative lipid injury caused by environmental stress. Because MDA concentration and ion leakage rate represent the degree of cell membrane damage (Prerostova et al., 2022), we additionally assessed their changes. Following 45°C treatment, it was observed that the Osdugt1 mutant lines’ MDA content and ion leakage rate were much higher than those of the wild type, whereas the overexpression lines’ values were significantly lower ( Figures 3E, F ). These results suggested that OsDUGT1 confers heat tolerance on rice at least through the increased ROS scavenging and the cellular membrane protection.

To further verify the function of OsDUGT1 in high temperature response, we heterogeneously expressed this gene in Arabidopsis thaliana, and acquired several overexpression lines (OE1, OE2, OE8, OE14) ( Figure 4A ; Supplementary Figure S1C ). The hypocotyl length of OsDUGT1 overexpression lines (OE8, OE14) were first measured after heat stress treatment at 45°C. We discovered that after heat stress treatment, the hypocotyl length of overexpression lines was noticeably longer than that of wild-type plants ( Figures 4B, C ). Then, we assessed the transgenic Arabidopsis’s root elongation after heat treatment. Similar with the hypocotyl growth, the roots of OsDUGT1 overexpression lines were significantly longer than those of the wild type, and with much more lateral roots ( Figures 4D, E ). To further investigate whether OsDUGT1 overexpressing plants can enhance the heat tolerance of Arabidopsis thaliana, OsDUGT1 overexpressing Arabidopsis plants grown on media and grown on soil were subjected to heat stress treatment. The results indicated that the growth phenotype and survival rate of plants overexpressing OsDUGT1 were obviously superior to those of the wild type ( Figure 5 ). These results further demonstrated the function of OsDUGT1 gene in enhancing plant tolerance to heat stress.

In order to know the molecular mechanism of OsDUGT1 in enhancing plant tolerance to heat stress, we analyzed the glycosyltransferase activity of OsDUGT1 by screening its substrates in vitro. The OsDUGT1 fusion protein with GST tag was first expressed in E.coli and purified ( Figure 6A ). Using the purified OsDUGT1-GST fusion protein as recombinant enzyme and UDP-glucose as sugar donor, a variety of candidate substrates were detected, including flavonoids, phenolic acids, plant hormones and other compounds. We found that OsDUGT1 had a clear glycosyltransferase activity toward ten flavonoids ( Figures 6B, C and Supplementary Figure S3 ), but no catalytic activity toward other compounds tested here was detected. Thus, these results suggested that the catalytic activity of OsDUGT1 was likely broad spectrum but specifically towards flavonoid group. In order to verify the glycosylation activity of OsDUGT1, the catalytic products of two typical flavonoids, quercetin and kaempferol, were subjected to LC-MS/MS analysis. Experimental results showed that the reaction products of quercetin gave the dominant ion peaks m/z 465.1024 [quercetin+Glc+H⁺] and the reaction products of kaempferol gave the dominant ion peaks m/z 449.1078 [kaempferol+Glc+H⁺] ( Figures 6D, E ). These results are in good agreement with the anticipated kaempferol glucoside and quercetin glucoside, respectively. Therefore, we concluded that OsDUGT1 has glycosylation activity on flavonoids.

To further explore the potential in vivo effects and metabolic adjustments resulted by OsDUGT1, we adopted metabolomics analysis to find out the metabolite differences between Osdugt1 mutants (ko) and wild type ZH11 under high temperature stress. 1491 metabolites were detected and quantified successfully by LC-MS/MS, among which 166 were significantly different, including 144 decreased and 22 increased metabolites in mutants ( Figure 7A ). By cluster analysis, the metabolites with known differences were separated into eight major groups (flavonoids, terpenes, alkaloids, phenolic acids, lignans and coumarins, quinines, tannins, and various others) ( Figures 7B, C ). It is worth pointing out that the most significant changes and the largest portion in differential metabolites were flavonoids, and their contents in knockout mutants were obviously less than that in wild type ( Figures 7B, C ; Supplementary Table S2 ). The KEGG enrichment of all differential metabolites in ko vs ZH11 indicated that the differential metabolites were focused on biosynthesis of secondary metabolites, flavonoid biosynthesis, flavones and flavonol biosynthesis, etc. ( Figure 7D ). It is well-known that flavonoids in planta are mainly in the form of glycosides. Our metabolome analysis indicated that OsDUGT1 can affect the flavonoid metabolism in rice, leading to a notable reduction in the accumulation levels of flavonoid glycosides and total flavonoids in OsDUGT1 loss-of-function mutants under high temperature stress. Clearly, these metabolic changes and adjustments were highly relevant to the enzyme activity of OsDUGT1 identified here.

To know the subsequent effects following the metabolic changes and adjustments by OsDUGT1 in high temperature response of rice, we further performed transcriptomic analysis using RNA-seq technique for the Osdugt1 mutants and wild type under heat treatment condition. Totally, 5664 differentially expressed genes (DEGs) were found (3106 downregulated and 2558 upregulated) ( Figure 8A ; Supplementary Figure S4A ).

According to gene ontology (GO) analysis, DEGs were shown to be enriched in molecular function, cellular component, and biological process ( Figure 8B ; Supplementary Figure S4B ). In detail, those down-regulated DEGs for biological process were enriched in response to heat (GO:0009266), response to temperature stimulus (GO:0009266), response to stress (GO:0006950). As is well known, the genes for the HSF and HSP are essential for the plant’s reaction to heat stress. In Osdugt1 knockout mutants, a higher proportion of heat stress-responsive genes, such as heat shock proteins (Hsp20, Hsp70 and Hsp90) and transcription factors related to heat stress, were down-regulated ( Figure 8C ), which well matched up with the heat sensitive phenotype of mutant lines. Intriguingly, 14 ROS scavenging-related genes associated with response to oxidative stress (GO:0006979) were also significantly down-regulated in Osdugt1 knockout mutants ( Figure 8D ), which was well in line with the excessive ROS accumulation observed in mutant lines. We also observed those DEGs associated with chloroplast structure and function.

According to KEGG enrichment analysis of DEGs, we obtained 20 highly enriched terms in Osdugt1 knockout mutants ( Supplementary Figure S4C ). We found that several genes related to flavonoid pathway and anthocyanin pathway were downregulated in knockout mutants, such as OsCHI, OsF3’5’H, OsNOMT, and OsDFR ( Figure 8E ). However, several genes associated with lignin synthesis in the phenylalanine pathway (OsPAL, OsC4H, OsCOMT, OsCAD) were upregulated in Osdugt1 knockout mutants ( Figure 8E ). These results suggested that knockout of OsDUGT1 might lead to a redirection of metabolic flux from flavonoid metabolism. This may be a reason why the content of flavonoids was reduced in the mutant rice.