The soils used for a previous study (Li X. et al., 2022) were sieved and mixed to obtain homogeneous composite soil. The Cd concentration gradients of 0 (Cd0), 5 (Cd5), 20 (Cd20), and 45 (Cd45) mg Cd kg⁻¹ dry soil, which were chosen according to several previous studies (Wu et al., 2018; Dhaliwal et al., 2020), were set to understand whether D. pinnata can be defined as a Cd hyperaccumulator. The weighed Cd (via CdCl₂•2.5H₂O) was dissolved in an appropriate amount of deionized water, and the solutions were adequately mixed with the Cd-free soils to obtain soils with targeted Cd concentrations. Actual Cd concentrations (Cd5: 6.84 mg kg⁻¹, Cd20: 20.10 mg kg⁻¹, and Cd45: 45.40 mg kg⁻¹) were measured prior to experiments. The prepared soils were equally loaded into uniform flowerpots (Li X. et al., 2022) with impermeable plastic trays to catch leachates. The soils were equilibrated for 1 month in a glass greenhouse (light: approximately 82% natural light, 12–14 h, 23–25°C; darkness: 10–12 h, 18–20°C; humidity: 40–60%) at Kunming Institute of Botany, Chinese Academy of Sciences.

Mature D. pinnata seeds purchased from a horticultural company (Wuhan, China) were surface-sterilized, cleaned, and sown (three seeds per pot) in the aforementioned flowerpots. One seedling was left in each pot after sprouting. Plants were watered regularly, and pots were position-changed to control for microenvironment variation. After 3 months of growth, plant roots, stems, and leaves, as well as rhizospheric soils, were collected following published methods (Wu et al., 2018; Li X. et al., 2022). Cd²⁺ adsorbed on the root surface was removed using Na₂EDTA solution (15 mM, 20 min) (Wu et al., 2018). Three biological replicates were prepared.

To explore the molecular mechanisms of the Cd response without interference from soil, D. pinnata was grown hydroponically. Seeds were surface-sterilized (Li et al., 2021b) and germinated in culture dishes (25 ± 2°C, 14 h light/10 h dark). Two weeks later, seedlings in similar size were individually cultured in 50 mL plastic tubes containing 1/2 modified Hoagland nutrient solution (Zhou et al., 2016) for 15 d in the above-mentioned greenhouse. The culture medium was replaced every 3 d. Subsequently, the plants were transferred into two hydroponic boxes (l × w × h: 37.5 cm × 25.0 cm × 13.5 cm; 12 seedlings per box) with 10 L 1/2 modified Hoagland nutrient solution (renewed every 3 d). One month later, seedlings in one box were treated with 50 μM Cd²⁺ (via CdCl₂•2.5H₂O), an appropriate Cd concentration that can effectively trigger changes in plant gene expression in a short time (Chen et al., 2021; Wang et al., 2022), whereas those in the other remained untreated as controls.

After 48 h of treatment, two to three top leaves and all roots of each plant were harvested separately. For transcriptomic analysis, samples were immediately frozen in liquid nitrogen and stored at −80°C. For Cd analysis, samples were cleaned for Cd²⁺ removal using Na₂EDTA solution (15 mM, 20 min) (Wu et al., 2018) and oven-dried at 80°C for 48 h. Three biological replicates were prepared for each measurement.

Raw reads from our high-throughput sequencing of the rhizospheric bacterial community were deposited in the NCBI Sequence Read Archive (SRA) (accession number: PRJNA874226). The samples generated 117,148–134,460 raw paired-end sequencing reads and 2,129–2,427 bacterial OTUs ( Supplementary Table S3 ).

The PCA showed that Cd0- and Cd45-treated samples formed distinct clusters ( Figure 2A ). However, bacterial communities did not differ significantly in four alpha indices (Shannon, Simpson, Chao1, and Ace) between Cd0 and Cd45 soils ( Supplementary Table S4 ). The two soils shared 1,457 core bacterial OTUs ( Figure 2B ), accounting for 66.0% and 64.2% of the total OTUs in the Cd0 (2,206) and Cd45 (2,270) soils, respectively. We identified 21 bacterial phyla and 233 genera, with most of them (20 phyla and 188 genera) common across the Cd0 and Cd45 soils ( Figures 2C, D ). Additionally, both soils had at least one sample that contained 16 bacterial phyla ( Supplementary Table S5 ) and 75 bacterial genera ( Supplementary Table S6 ) with a relative abundance of > 0.1%.

The Cd0 and Cd45 soils differed significantly in microbial taxa composition (P < 0.05, Welch’s t-test). The abundance of the phyla Patescibacteria, Cyanobacteria, and Elusimicrobia increased in the Cd45 soil compared with that in the Cd0 soil, whereas the abundance of the phyla Bacteroidetes and Armatimonadetes decreased ( Figure 2E ). Additionally, 12 bacterial genera differed in abundance between Cd0 and Cd45 soils ( Figure 2F ).

After 50 mM Cd treatment for 48 h under hydroponic conditions ( Figure 3A ), root and leaf Cd concentrations reached 11,900 and 227 mg kg⁻¹, respectively ( Supplementary Figure S3 ).

Approximately 67,988,424–68,000,000 raw RNA reads, which were deposited in the NCBI SRA (accession number: PRJNA811758), were generated for 12 samples, with clean reads accounting for 99% ( Supplementary Table S7 ). High Q20 (> 97%) and Q30 (> 92%) values ( Supplementary Table S7 ) indicated good sequencing quality. Of the 219,444 unigenes obtained, 91,692 (41.8%) were annotated in different databases ( Supplementary Table S8 ). We quantified 40,679 ( Supplementary Table S9 ) and 29,304 unigenes ( Supplementary Table S10 ) in D. pinnata roots and leaves, respectively.

The PCA results ( Figure 3B ) indicated that Cd stress had a greater influence on gene expression in roots than in leaves. Under Cd stress, D. pinnata roots and leaves had 13,556 (4,028 upregulated and 9,528 downregulated) and 4,904 DEGs (2,260 upregulated and 2,644 downregulated), respectively ( Figures 3C, D ; Supplementary Tables S9 and S10 ). We observed tissue-specific expression for 88.8% (roots) and 68.9% (leaves) of DEGs ( Figure 3E ). The remaining 1,524 DEGs were identified in both roots and leaves, and 90.4% exhibited the same change patterns ( Figure 3E ).

Significant correlations between the RNA sequencing and qRT‐PCR data in both roots (R² = 0.7199, P < 0.01) and leaves (R² = 0.6165, P < 0.01) ( Figure 3F ) were observed, indicating reliable RNA sequencing results in this study.

In both roots and leaves, DEGs were enriched in similar GO terms under the biological process, molecular function, and cellular component categories ( Supplementary Figures S4 and S5 ). Nearly all enriched terms contained upregulated and downregulated genes ( Supplementary Figures S4 and S5 ), suggesting that Cd stress exerts multiple physiological effects on D. pinnata.

The results of enrichment analysis on upregulated genes found that they were significantly enriched (P < 0.001) in 15 KEGG pathways in roots and four in leaves ( Figure 4 ; Supplementary Table S11 ). The pathways were primarily associated with substance metabolism, signal transduction, and substance transport ( Figure 4 ). Only one pathway (phenylpropanoid biosynthesis) was enriched in both roots and leaves ( Figure 4 ).

We selected several pathways for further analysis based on the number of enriched genes ( Figure 4 ). For roots, the pathways were related to phenylpropanoid biosynthesis, mitogen-activated protein kinase (MAPK) signaling, plant hormone signal transduction, and plant‐pathogen interaction; for leaves, pathways were involved in phenylpropanoid biosynthesis and cysteine/methionine metabolism. Many upregulated genes were involved in key reactions of these pathways ( Supplementary Figures S6–S11 ), providing insight into the mechanisms of Cd tolerance in D. pinnata.

The downstream products of the phenylpropanoid metabolism pathway, such as lignins, flavonoids, and procyanidins, improve plant tolerance to heavy metals through multiple mechanisms (Ge et al., 2022). In this study, upregulated genes were highly enriched in the synthesis process of lignins in the phenylpropanoid metabolism pathway in both the roots and leaves of D. pinnata ( Supplementary Figures S6 and S10 ), which may lead to an increase in the production of various lignins (e.g., syringyl lignin, 5-hydroxyguaiacyl lignin, and guaiacyl lignin). Lignins, as the main components of the secondary wall of plant cells, can fix heavy metal ions in the cell wall through their functional groups, such as carboxyl, phenolic, and aldehyde groups, to inhibit the entry of heavy metals into the cytoplasm (Li Y. et al., 2018; Su et al., 2020; Yu et al., 2023). Therefore, the results suggest that lignin-mediated cell wall compartmentalization ( Figure 5 ) may be an important Cd detoxification mechanism in both the roots and leaves of D. pinnata.

The ability of plants to perceive, transmit, and translate stress signals into an appropriate physiological response determines their ability to tolerate heavy metals (Islam et al., 2015). Plant responses are typically modulated via crosstalk between signaling molecules such as Ca²⁺, reactive oxygen species (ROS), nitric oxide (NO), phytohormones, and hydrogen sulfide (Luo et al., 2016). In this study, many upregulated genes in D. pinnata roots were enriched in three signaling pathways that are likely critical to Cd tolerance: MAPK signaling ( Supplementary Figure S7 ), hormone signal transduction ( Supplementary Figure S8 ), and plant‐pathogen interaction ( Supplementary Figure S9 ).

As a second messenger, Ca²⁺ plays an important role in intracellular signaling pathways of acclimation to heavy metals and other stressors. Ca²⁺ enters the cytoplasm through cyclic nucleotide-gated calcium channels (CNGCs) and interacts with calcium-binding proteins, including calcium-dependent protein kinase (CDPK), calmodulin (CaM), and calmodulin-binding protein kinase (Li Y. Q. et al., 2022). Of these, CDPKs are involved in phosphorylating NADPH oxidase to produce ROS (Li Y. Q. et al., 2022), whereas the Ca²⁺/CaMs complex promotes NO synthase activity to produce NO (He et al., 2015). Here, we observed that Cd induced CNGC activity, along with Ca²⁺-related ROS and NO production, in D. pinnata roots ( Supplementary Figure S9 ). These results strengthen the hypothesis that intracellular Ca²⁺, ROS, and NO are pivotal in plant acclimation to heavy-metal stress (Luo et al., 2016).

Phytohormones such as ethylene (ETH), abscisic acid (ABA), auxin, jasmonic acid, and salicylic acid often act as signals that trigger a plant’s heavy-metal stress response (Saini et al., 2021). Unsurprisingly, Cd treatment upregulated signaling pathways in D. pinnata roots that involve auxin, ABA, and ETH ( Supplementary Figure S9 ), supporting the importance of phytohormone signal transduction in the Cd stress response.

Heavy metal exposure also initiates MAPK cascades in plants as part of signal transduction (Mondal, 2022). Consistent with previous research (Li S. C. et al., 2022; Mondal, 2022), ETH, ABA, and H₂O₂ were primarily responsible for initiating MAPK cascades in D. pinnata roots under Cd stress ( Supplementary Figure S7 ). Our results support the idea of crosstalk between MAPK cascades and other signaling molecules during the coordination of D. pinnata responses to Cd stress.

In summary, Cd stress induced a complex signal transduction network involving Ca²⁺, ROS, NO, phytohormones, and MAPK cascades in D. pinnata roots ( Figure 5 ). This network likely triggers detoxification processes that ensure D. pinnata tolerance to Cd.

As a source of sulfur, methionine is required for the biosynthesis of glutathione and phytochelatins, which contribute to heavy-metal detoxification in plants (Thakur et al., 2022). Methionine is also involved in metal uptake and transport in plants (Mousavi et al., 2021), as well as being the immediate precursor of S-adenosylmethionine, itself a precursor to ETH and polyamine biosynthesis. In this study, Cd stress upregulated genes encoding enzymes key to methionine synthesis of ETH in D. pinnata leaves ( Supplementary Figure S11 ). Recent evidence suggests that ETH mediates Cd resistance in plants by positively regulating flavonoid biosynthesis and antioxidant activity (Chen et al., 2022). Moreover, in the Cd hyperaccumulator Sedum alfredii, high ETH concentrations postponed apoplastic-barrier formation and thus promoted Cd accumulation in root apoplasts (Liu et al., 2021). Taken together, the data strongly imply that ETH signaling based on methionine metabolism ( Figure 5 ) is critical to regulating Cd stress responses in D. pinnata leaves.