Soil preparation
The soil used in the pot experiment were collected from the topsoil (0-20 cm) in Shennong Garden of the Jiangxi University of Chinese Medicine in Jiangxi, China (E 115°44′24′′, N 28°41′24′′). The soil was air-dried, sieved to remove gravels through a 4 mm mesh sieve, and then mixed thoroughly. The basic physicochemical properties (total nitrogen (N), available N, total phosphorus (P), available P, total potassium (K), available K and total Cd concentration, organic matter content, and pH value) (Table 1) of the soil were measured following the methods of Wen et al. [52]. The total Cd concentration was 0.95 mg kg-1, less than the critical value (1.0 mg kg-1) of the soil to ensure normal plant growth in agriculture and forestry production.
The appropriate contents of cadmium chloride hemi (pentahydrate) (CdCl2·2.5H2O, Sinopharm Chemical Reagent Co., Ltd., China) were thoroughly blended with soil to obtain final concentrations of 25, 50 and 100 mg Cd per kilogram of soil, respectively. The selected Cd concentrations were consistent with that of previous studies of Wang et al. [53] and Zhang et al. [25]. Then the soils with or without Cd treatment were transplanted into plastic pots (16 cm in diameter × 17 cm in height) and watered every 5 d with pure water. Before use, they were equilibrated for 30 d.
Plant material and Cd treatment
S. miltiorrhiza seedlings were purchased from S. miltiorrhiza planting areas (Shangdong, China). The formal identification of the samples was carried out by Assoc Prof. Xiaoyun Wang. Cultivated in Shennong garden for 30 d, the seedlings were transferred into appropriate pots containing 2.5 kg soil. Each pot contained two seedlings. A pot was taken as one replicate, and three replicates were performed for each treatment. In the experiment, seedlings without Cd treatment were used as the control group (CK), and seedlings with 25, 50, 100 mg kg-1 Cd treatment were named TR, TS, and TT, respectively. Each seedling presented uniform growth. After being watered once per week for 4 weeks, plant samples were harvested for further study.
In order to perform LC-MS and physiological analyses, root samples from each pot were selected to make a composite root sample, simultaneously frozen in dry ice and stored at -80 °C. Leaves and remaining roots from each pot were collected to make a composite leaf sample and root sample, respectively, to perform Cd concentration analysis. Then both samples were dried in the oven at 60 °C for 4 d. Voucher specimens (NO. DS-001) were deposited in a public herbarium in Research Center for Traditional Chinese Medicine Resources and Ethnic Minority Medicine of Jiangxi University of Chinese Medicine.
Determination of Cd concentration
The oven-dried samples were ground to fine powder, passed through a 2-mm sieve, and then digested with a mixture of nitric acid (HNO3) and hydrogen peroxide (H2O2) (3:1, v/v) in the Teflon tanks using an electric heating board at 160 ℃ thoroughly. The Cd concentration was determined by inductively coupled plasma mass spectrometry (ICP-MS, Thermo Scientific, USA).
Physiological analysis
The ninhydrin colorimetry was used to determine the proline content [54, 55]. A total of 0.05 g fresh root samples were pooled in the centrifugal tube with 5 mL 3% sulfosalicylic acid, extracted in the boiling water bath for 10 min, and shaken frequently. The extraction was filtered into a volumetric flask. Then the volume was constant to 25 mL with distilled water. A total of 2 mL extraction solution mixed with 2 mL glacial acetic acid and 2 mL acidic ninhydrin was added in a centrifuge tube. The mixture was treated with the boiling water bath for 30 min. After cooling down to room temperature, 4 mL toluene was added to the mixture and shaken thoroughly. When they were stratified during standing, the absorbance of the upper solution was measured at 520 nm with an ultraviolet spectrophotometer (UV-8000, Metash, China).
The content of MDA and activities of POD, SOD and CAT were detected by assay kits (Suzhou Keming, China). A total of 0.1 g fresh samples and 1 mL solutions (1:10, v/v) were grounded in a water bath and centrifuged at 8000 × g and 4 ℃ for 10 min. According to the manufacturer’s instructions, supernatant absorbances were measured at 532 nm and 600 nm to assess MDA content and at 470 nm, 240 nm and 560 nm to evaluate the activities of POD, CAT and SOD, respectively.
Metabolite analysis
The mothed of metabolites extraction was referred to Wang et al. [56]. A total of 25 mg fresh samples were placed into an EP tube with 500 μL extract solution (methanol:water = 3:1 (v/v), with the isotopically-labeled internal standard mixture), homogenized at 35 Hz for 4 min and sonicated for 5 min with an ice-water bath. The homogenization and sonication cycle were conducted 3 times. The samples were incubated for 1 h at -40 ℃ and centrifuged at 12000 rpm and 4 ℃ for 15 min. Subsequently, the resulting supernatant was transplanted into a fresh glass vial for further analysis. Besides, the quality control (QC) sample was formed by mixing an equal aliquot of the supernatants from all samples [57].
Combining a UHPLC system (Vanquish, Thermo Fisher Scientific, US) and Q Exactive HFX mass spectrometer (Orbitrap MS, Thermo, US), LC-MS/MS analyses were performed to analyze sample metabolites using a Waters ACQUITY UPLC HSS T3 (2.1 mm × 100 mm, 1.8 μm; Waters, USA). Mobile phase A was water containing 5 mmol L-1 ammonium acetate and 5 mmol L-1 ammonia hydroxide, while mobile phase B was acetonitrile. The mobile phase elution procedure was set as follows: the concentration of mobile phase A was 5% in 0-5 min, 35% at 7 min, increased to 60% holding for 1 min, then decreased to 5% holding for 3 min; the mobile phase B was 95% in 0-5 min, 75% at 5 min, decreased to 20% holding for 1 min, then increased until 95% holding for 3 min. The auto-sampler temperature was 4 ℃, the injection volume was 3 μL, and the flow rate was 0.5 mL min-1. The QE HFX mass spectrometer was conducted to acquire MS/MS spectra based on information-dependent acquisition mode in controlling the acquisition software (Xcalibur, Thermo, USA). In the mode, the acquisition software continuously evaluated the full scan MS spectrum. The ESI source conditions were conducted as follows: The flow rates of sheath gas and Aux gas were 30 Arb and 10 Arb, respectively; the capillary temperature was 350 ℃; the full MS resolution was 60000; the MS/MS resolution was 7500; the collision energy was 10/30/60 in NCE mod; the spray voltage was 4.0 kV (positive) or -3.8 kV (negative).
After the original data was converted into mzXML format using the software ProteoWizard (https://proteowizard.sourceforge.io/), the R package XCMS (version 3.2) was used for the peak recognition, extraction, alignment, and integration. The preprocesses of the original data included the following: 1) Data filtering. The filtering standard was to remove the data with no definite substance name or no spectrum comparison similarity. 2) Missing values processing. Substances of more than 50% missing in comparisons were filtered directly, and substances of less than 50% missing were performed the imputation of missing values using the k-nearest neighbor (KNN) algorithm. 3) Normalization. The internal standard (IS) or total ion current (TIC) of each sample was used for the normalization. A total of 1111 and 305 peaks of the original data were retained for positive and negative ion modes, respectively. The excel sheets, including the name of peak and sample, and the standard data of normalized peak area were obtained for further data analysis.
Statistical analysis
Firstly, the transfer factor (TF) and bio-concentration factor (BCF) of Cd were calculated as follows [26-28]:
TF = Cd concentration of aboveground parts (mg kg-1) / Cd concentration of roots (mg kg-1)
BCF = Cd concentration in the tissues / Cd concentration in soils
Secondly, the multivariate ordination principal component analysis (PCA) was conducted to reveal the overall distribution of samples. The supervised partial least squares discrimination analysis (PLS-DA) was utilized to assess the differences in roots between the control and Cd treatment groups. The PLS-DA model was validated by permutation tests (200) or ANOVA of the cross-validated residuals (CV-ANOVA). A Q2 value of above 0.5 and an R2 value of above 0.7 for the permutation test or p value of below 0.05 for the cross-validated residuals (CV-ANOVA) denoted the highly significant model [58]. The variable importance in projection (VIP) was crucial for explaining the data of the PLS-DA model. Metabolites with a p value of below 0.05 and a VIP of above 1.0 were filtered out as differential metabolites, playing larger roles in distinguishing roots between the control and Cd addition groups.
Thirdly, the pathway analysis of all the metabolites was conducted by the software MetaboAnalyst 4.0 (http://www. metaboanalyst.ca/faces/ModuleView.xhtml). Based on the pathway impact value of above 0.1, the potential metabolic target pathways were obtained [59]. Thirdly, the Venn diagram was obtained using the software Venny 2.1 (http://bioinfogp.cnb.csic.es/tools/venny/) to present the overlap of all the named metabolites in the control and Cd-treatment groups. Subsequently, t-test and correlation analyses were conducted to evaluate the variability of identical metabolites in samples and relationships between Cd concentration in tissues (leaves and roots) and Cd concentration in soils with SPSS 18.0 (SPSS Inc., USA), respectively. The PCA and PLS-DA models were performed by SIMCA-P version 14.1 (Umetrics, Sweden). All data were log10-transformed before analysis.
Availability of data
All analyzed data are included in this article and its Supplementary Files.