Rice seeds were soaked in 10% sodium hypochlorite for 5 min under dark conditions. The seeds were then washed thoroughly, soaked in deionised water, and placed in an incubator at 37°C for germination. Uniformly emerging rice seedlings were evenly placed on a floating net and cultured in 0.5 mol L⁻¹ CaCl₂ nutrient solution in the dark for two days to induce rooting. After CaCl₂ culture, rice seedlings were cultured with kimura B nutrient solution containing 0.18 mmoL⁻¹ KH₂PO₄, 0.36 mmoL⁻¹ (NH₄)₂SO₄, 0.54 mmoL⁻¹ MgSO₄·7H₂O, 0.18 mmoL⁻¹ KNO₃, 0.36 mmoL⁻¹ Ca(NO₃)₂·4H₂O, 46.25 μmoL⁻¹ H₃BO₃, 0.32 μmoL⁻¹ CuSO₄·5H₂O, 0.76 μmoL⁻¹ ZnSO₄·7H₂O, 9.15 μmoL⁻¹ MnCl₂·4H₂O, 0.11μmoL⁻¹ H₂MoO₄·H₂O, 20 μmoL⁻¹ EDTA−FeSO₄ (pH, 5.6).The nutrient solution was replaced every 2 days.

Two-week-old rice plants were cultured for 5 days under normal conditions, 10 μmol L⁻¹ CuSO₄ treatment, or 25 μmol L⁻¹ CdCl₂ treatment. A normal kimura B nutrient solution was used as the control. Four replicates were set, and each replicate was comprised of 5 rice seedlings. All rice seedlings were grown in a greenhouse under long-day conditions (14 h light/10 h dark) at 28°C/24°C. Oryza sativa cv ‘Dongjin’ was used as the wild type (WT) in this study.

The CRISPR/Cas9 system was used to construct the rice mutants. Two sequences of 20 bp in exons of GLP8-2 were selected as gRNAs. Primers were designed based on these sequences, and the annealed product was fused with the pRGEB31 vector. The coding sequences of OsGLP8-2 were amplified from the cDNA of the WT (Dongjin, DJ). Two specific primers (CriOsGLP8-2F and CriOsGLP8-11R) were used to identify mutants, with 10 genes (OsGLP8-2 to OsGLP8-11) knocked out. If a clear band was observed after 1% agarose gel electrophoresis ( Supplementary Figure S1 ), this indicated that the multi-gene knockout mutant was successfully constructed. The coding region sequence of OsGLP8-2 was fused with the pOx vector to form a recombinant plasmid (GLP8-2OE) driven by the 35S promoter. The primers used for vector construction are listed in Supplementary Table S1 .

The recombinant plasmids were sequenced and introduced into Agrobacterium tumefaciens EHA105. The A. tumefaciens-mediated genetic transformation system was used to construct transgenic rice.

Four-day-old WT rice seedlings were treated with 3 μmol L⁻¹ CuSO₄ for 12 h. About 50 mg of root tips were collected and snap-frozen in liquid nitrogen to extract total RNA for transcriptome sequencing (GENE DENOVO Biotechnology Co., Ltd, Guangzhou, China). Bioinformatic analysis of the data was performed using the Omicsmart online real-time interactive platform. Each material was repeated three times. Fold change ≥ 2, and FDR ≤ 0.05.

The roots of the rice were washed with 20 mmol L⁻¹ Na₂EDTA for 30 min to remove heavy metal ions attached to the surface of the roots. They were then placed in an oven at 80°C until a constant weight, and the dry weight was recorded. Dry plant samples or cell wall materials (0.2 g) were digested with 5.0 mL of guaranteed HNO₃:HClO_{4 =} 87:13 (v:v) mixed acid. Cd and Cu concentrations were determined using an inductively coupled plasma optical emission spectrometer (ICP-OES, PerkinElmer, Optima 8000, America). A plant standard [GBW10043 (GSB-21)] was purchased from the National Research Centre for Standards of China and used to ensure reliable results during the digestion and analysis processes.

Extraction of crude cell walls was according to the methods of Yang et al. (2011) and Zhu et al. (2020). About 0.5 g of fresh samples of rice shoots and roots were ground with 10 times the volume of 95% ethanol to homogenise them. The mixture was centrifuged at 8,000×g for 5 min, and the supernatant was discarded. The pellet was washed 3 times with 95% ethanol. Finally, the pellet was washed twice with ethanol-hexane solution (v:v=1:2) and dried at room temperature to obtain the crude cell wall. The determination of heavy metals in the cell wall was performed as previously described in Determination of Cu and Cd concentrations section.

The stems of the rice plants were stained with Safranin O-Fast Green staining and paraffin sectioned by embedding technology to determine lignin deposition in the cell walls (Wang et al., 2016; Han et al., 2021). The roots were placed on a Petri dish, and a few drops of 1% phloroglucinol ethanol solution were added. A drop of 35% HCl was added, and the roots were covered with cover glass. After volatilisation and colour development, the roots were placed on a type microscope and magnified four times for observation, and photos were taken.

Six milligrams of cell wall residue were transferred to a glass test tube, and 2.5 mL of 25% bromoacetyl-acetic acid solution (v/v=1:3) and 0.1 mL of 70% perchloric acid were added. The tube was covered, sealed, and placed in a water bath at 70°C for 30 min. The test tube was shaken every 10 min. After cooling, 10 mL of 2 mol L⁻¹ NaOH solution was added, and the reaction mixture was diluted to 25 mL with glacial acetic acid. The mixture was centrifuged at 1000×g for 5 min. The absorbance of the supernatant was measured at 280 nm, with the reaction solution containing no sample as a blank control.

Total RNA from rice seedlings was extracted using an RNA Extraction Kit (TaKaRa, 9697, China). The cDNA was then obtained after inversion was used as a template, and the SYBR Green fluorescence quantitative kit (TaKaRa, RR420A, China) was used for fluorescence quantitative PCR amplification. The expression of the target gene was calculated using the 2^-ΔΔCt method (Gaonkar et al., 2018). The housekeeping gene ACTIN1 (LOC_Os03g50885) was used as the internal control. Three biological replicates were used for qRT-PCR, and three technical replicates were set for each biological replicate. The primers used are listed in Supplementary Table S2 .

The Diaminobenzidine (DAB) staining method was used for the quantitative detection of H₂O₂. DAB powder was dissolved in 50 mmol L⁻¹ Tris-HCl (pH 6.0) to prepare a 1 mg mL⁻¹ dye solution. Rice leaves (3–4 cm) were immersed in DAB dye solution, vacuumed for 2 hours until the leaves sank to the bottom of the tube, and placed in the dark for 12 hours. The dyed rice leaves were boiled in 95% ethanol for decolourisation. After bleaching was complete, the leaves were immersed in 70% ethanol, and pictures were taken using a stereo microscope (Nikon, SMZ1000, Japan).

Leaf samples (0.1 g) were ground with 1.5 mL of trichloroacetic acid (TCA) on ice and centrifuged at 12,000×g for 15 min at 4°C. Then, 500 μL of the supernatant was transferred to a clean 2 mL centrifuge tube, and 1.5 mL of 0.5% thiobarbituric acid (TBA) was added and mixed well. The tubes were placed in a water bath at 90°C for 20 min. After cooling, the mixture was centrifuged at 10,000 rpm for 5 min. The absorbance of the supernatant was determined at 450, 532, and 600 nm. C (μmol L^-1)=6.45×(A₅₃₂-A₆₀₀)-0.56×A₄₅₀

Rice root tips (1 cm) were placed in 3 μg L⁻¹ propidium iodide (PI) solution and soaked in the dark for 15 min. The root tips were removed from the dye solution and rinsed repeatedly with deionised water. Staining was observed with a fluorescence microscope (Zeiss, Axio Imager A1, Germany).

The data were analysed using Excel and SPSS25.0 for analysis of variance and LSD multiple comparison testing (P ≤ 0.05). GraphPad Prism 6 was used to graph the data after processing. The values in the graph are the mean ± SD (n = 3). Different letters indicate the differences between several rice lines.

Thirty-two members of the OsGLP family were tandem repeat genes and were divided into 8 gene clusters located on chromosomes 1, 2, 3, 8, 9, and 12. Among them, the tandem repeat gene cluster on chromosome 8 was the largest, containing 11 OsGLP genes (OsGLP8-1 to OsGLP8-11) (Li et al., 2016). Four-day-old WT seedlings were treated with 3 μmol L⁻¹ CuSO₄ for 12 h. Total RNA was isolated from rice roots and used for transcriptome sequencing. The heat map showed that the expression of multiple OsGLP family genes, such as OsGLP8-2, OsGLP8-5, OsGLP8-6, OsGLP8-7, OsGLP8-9, OsGLP8-10, and OsGLP8-11, increased significantly after Cu treatment ( Figure 1 ). It was inferred that some OsGLPs are Cu-responsive proteins.

To understand the contribution of OsGLPs to heavy metal tolerance, we designed the primers for OsGLP8-2 and OsGLP8-11 to knock out multiple genes at the same time, and obtained the glp8-2 and glp8-(2-11) mutants, respectively ( Figure 2A ; Supplementary Figure S2 ). Furthermore, we identified 6 overexpression lines and selected GLP8-2OE1 (hereinafter referred to as GLP8-2OE) with the highest expression of OsGLP8-2 for subsequent experiments ( Figure 2B ).

The WT, two mutants glp8-2 and glp8-(2-11), and GLP8-2OE seedlings were treated with 10 μmol L⁻¹ CuSO₄ or 25 μmol L⁻¹ CdCl₂ for 5 days. Both the glp8-2 and glp8-(2-11) mutants showed hypersensitivity to Cu and Cd toxicity compared with the WT seedlings ( Figure 2C ; Supplementary Figure S3 ). Excess Cu and Cd had a significant inhibitory effect on the shoot height and root elongation of glp8-2 and glp8-(2-11) mutants, while OsGLP8-2 overexpression increased heavy metal tolerance in rice ( Figures 2D, E ). Quantitative analysis further confirmed that the dry weights of glp8-2 and glp8-(2-11) seedlings were significantly lower than those of WT and GLP8-2OE seedlings ( Figures 2F, G ). Overall, OsGLP knockout led to a decrease in chlorophyll content ( Supplementary Figure S2 ). These results suggest that OsGLPs play an important role in regulating heavy metal tolerance in rice.

To further investigate the mechanism of OsGLPs regulating heavy metal tolerance in rice, we measured the Cd and Cu concentrations in the shoots and roots, and in those of their cell walls. As shown in Figure 3 , the Cd and Cu concentrations in both shoots and roots increased significantly with elevated heavy metal levels. At 3 and 10 μmol L⁻¹, glp8-2 and glp8-(2-11) seedlings had higher Cu concentrations than the WT, while that in GLP8-2OE was lower. This was especially obvious when the rice seedlings were treated with 10 μmol L⁻¹ CuSO₄ ( Figures 3A, B ). Cd accumulation in different rice seedlings at high levels of Cd (25 μmol L⁻¹) displayed a trend similar to Cu toxicity ( Figures 3C, D ). Cd and Cu concentrations in the roots were higher than those in the shoots of the different rice seedlings at the same level of heavy metals.

Cell walls are the main compartments that accumulate heavy metals (Krzeslowska, 2011; Wang et al., 2020; Yan et al., 2022). We determined Cd and Cd concentrations in the cell walls of different rice seedlings to investigate whether OsGLP changes the distribution of heavy metals. In contrast to the results of Cu concentrations in the shoots and roots, the glp8-2 and glp8-(2-11) mutants accumulated less Cu in the cell wall than those of the WT and GLP8-2OE when exposed to 3 and 10 μmol L⁻¹ CuSO₄ ( Figures 4A, B ). Similar to Cu concentration in the cell wall, the loss of OsGLP8-2 resulted in lower Cd retention in the cell wall than the WT and overexpressing rice seedlings at high levels of CdCl₂ (25 μmol L⁻¹) ( Figures 4C, D ). However, there was no obvious difference in heavy metal concentrations of the cell wall between the glp8-2 and glp8-(2-11) rice seedlings.

Our previous studies reported that lignin may play a vital role in Cu and Cd stress (Liu et al., 2015; Xia et al., 2018; Su et al., 2020). To investigate the relationships among loss of OsGLPs, lignin synthesis, and heavy metal accumulation, we comparatively analysed lignin synthesis in different rice seedlings treated with elevated Cd and Cu levels. The Safranin O-Fast Green staining and phloroglucinol-HCl staining in stems and roots showed that the lignin content in the glp8-2 and glp8-(2-11) mutants was lower than that of the WT and GLP8-2OE rice ( Figures 5A, B ). To further confirm this, we quantitatively determined the lignin content in different rice seedlings treated with Cu and Cd using the acetyl bromide-soluble method (Van Acker et al., 2013; Jang and Lee, 2020). As expected, there was a decrease in the lignin content of the shoots and roots from the two glp8-2 and glp8-(2-11) mutants under Cu and Cd stress, and the lignin content was about 4.6% lower than the WT ( Figures 5C, D ). The lignin content in the roots of OsGLP8-2OE induced by Cu and Cd was higher than that of the WT. A significant negative correlation between the heavy metal concentrations and lignin content was observed in rice roots (p< 0.0001 for Cu; p = 0.0024 for Cd) ( Figures 5E, F ).

To determine whether the alteration of OsGLPs expression levels affects the lignin synthesis pathway, we tested the changes in the expression levels of lignin-related genes, including phenylalanine ammonia-lyase (PAL), 4-coumarate CoA ligase (4CL), caffeoyl-CoA-O-methyltransferase (CCoAoMT), cinnamate-4-hydroxylase (C4H), and cinnamoyl-CoA reductase (CCR). Cu and Cd treatments significantly induced the expression of five lignin biosynthetic enzyme genes (PAL, 4CL, CCoAOMT, C4H, and CCR) ( Figure 6 ). When treated with Cu and Cd, the expression levels of these five genes were reduced in the glp8-2 and glp8-(2-11) mutants, especially in the OsGLP8-(2-11) mutants, but elevated significantly in the OsGLP8-2OE seedlings. The expression pattern of five lignin-related genes under Cu and Cd toxicity showed the same trend as lignin content in the roots and heavy metal accumulation in the cell wall.

To explore the role of OsGLPs in oxidative stress caused by heavy metal stress, we compared DAB staining in WT and transgenic rice lines. There was no significant difference in the leaf colour of the four lines in the absence of excess Cu and Cd. When exposed to Cu and Cd, the leaf colour was darker than that of the control and GLP8-2OE. Compared with the WT, the glp8-2 and glp8-(2-11) mutants were darker, especially the OsGLP8-(2-11) mutant ( Figure 7A ). This indicates that OsGLPs participate in the elimination of active oxygen in rice cells and can reduce the accumulation of active oxygen caused by heavy metal stress, thereby alleviating the oxidative damage of rice.

Malondialdehyde (MDA) content indicates the degree of peroxidation of the cell membrane and is an important indicator of plant stress resistance (Yang et al., 2019). As shown in Figure 7B , Cu and Cd stress aggravated the peroxidation of membrane lipids and increased the MDA content. Loss of the function of OsGLPs led to increased MDA content, which was the opposite of the GLP8-2OE rice seedlings.

Propidium iodide (PI) is a nuclear fluorescent dye that indicates the integrity of the plasma membrane (Li et al., 2021). When the root tip cells are damaged, the permeability of the plasma membrane increases, PI can enter the cell and bind to DNA, and red fluorescence can be observed with a fluorescence microscope. In this study, PI was used to determine the integrity of the plasma membrane in the root tip. Cu and Cd treatments damaged the integrity of the plasma membrane in rice roots. By comparing the intensity of red fluorescence in different rice, the red fluorescence intensity of the glp8-2 and glp8-(2-11) mutants were higher than that of the WT and OsGLP8-2OE ( Figure 7C ) when subjected to Cu and Cd stress, which was consistent with the H₂O₂ histochemical localisation and MDA content. This indicates that OsGLP genes play a certain role in maintaining the integrity of the plasma membrane.

The knockout mutant of 10 genes (glp8-(2-11)) and the mutant of OsGLP8-2 (glp8-2) showed the same phenotypes, such as heavy metal tolerance and accumulation, lignin deposition and gene expression levels, and antioxidant defence abilities. We speculated that OsGLP8-2 may display the main contribution in the tandem repeat gene clusters on chromosome 8 in responding to heavy metal stress. The time course for expression levels of OsGLP8-2, OsGLP8-3, OsGLP8-5, OsGLP8-7, and OsGLP8-11 genes on chromosome 8 were detected under Cu and Cd treatment. The expression of these five genes was significantly upregulated and reached a peak under Cu exposure for 3 h and Cd exposure for 12 h. Among these genes, Cd and Cu treatments upregulated the expression of OsGLP8-2, with the highest fold change. Its highest level was 71 times higher than that of the control under Cu stress and 11.3 times higher under Cd stress ( Figure 8 ). These data demonstrate that OsGLP8-2 is more sensitive to Cu and Cd and is upregulated more than other tandem genes.