Effects of nanoscale zinc oxide treatment on growth, rhizosphere microbiota, and metabolism of Aconitum carmichaelii

Sampling investigation of soil trace elements’ correlation with yield and monoester-type alkaloids in A. carmichaelii

The sampling area, Butuo County, is located in Xichang City and covers an area of 1685 km², situated between the north latitudes of 27°16′ to 27°56′ and east longitudes of 102°43′ to 103°04′ in southwestern Sichuan Province, China. The main herb and economic crop in this region is A. carmichaelii, which is exported to East Asian countries including Japan, Korea, Mongolia, and India (Yu et al., 2016).The maps of the sample locations were constructed using ArcMap Desktop (version 10.8) with the Spatial Analyst module (Liu, Shao & Wang, 2013; Mahapatra et al., 2020).

Soil and plant samples were collected from 22 locations in six different villages within Butuo (Fig. S1). The sample collection process involved employing a five-point sampling method at each location. A total of 10 plants were sampled from each plot. Soil samples were collected from the cultivated layer within a 20 cm depth surrounding the roots of the plants at each sampling point, resulting in approximately 0.5 kg of mixed soil sample. All soil samples underwent a natural air-drying process, followed by grinding and sieving through a two mm sieve. These processed soil samples were then utilized for soil parameter analysis in the laboratory.

Soil trace element analysis

Available zinc (AZn) was estimated by triplicate soil extractions with 0.02 M Na₂-EDTA + 0.5 M NH₄Ac at pH 4.65, using 5 mL solution and shaking for 30 min (Campillo-Cora et al., 2019). Available manganese (AMn), available boron (AB), available iron (AFe) and available copper (ACu)were determined by shaking 5 g of soil with 10 mL diethylene triamine penta acetic acid (DTPA) extractant (pH 7.3) for 2 h (Lindsay, 1978). Total trace element assays used microwave-assisted digestion with some modifications. The included elements were total zinc (TZn), total manganese (TMn), total boron (TB), total iron (TFe) and total copper (TCu) (Bettinelli et al., 2000). For trace element assessment, 0.3 g of each soil sample was digested with a mixture of 5 mL nitric acid (67% w/v), 3 mL hydrochloric acid (37% w/v), and 1 mL hydrofluoric acid (40% w/v) in a microwave chemical reactor (MDS-6G, SINEO, Shanghai, China) with a temperature program of 160 °C for 10 min, then increased to 210 °C for 30 min. The resulting clear brownish solution was treated with an acid scavenger (TK12, SINEO, Shanghai, China) for 30 min and diluted to 50 ml with 2% nitric acid. All samples were centrifuged at 8,586 × g for 10 min and the supernatant was filtered through a 0.45 µm membrane filter. The resulting samples were analyzed for heavy metals using an atomic absorption spectrophotometer (AA-7020, EWAI, Beijing, China). Three replicates were prepared for each analytical sample.

Analysis of yield and monoester-type alkaloids

The yield was obtained by measuring the fresh weight of lateral roots from 10 plants at each sampling point, after removing the soil. To analyze the content of monoester-type alkaloids, lateral root powder was sieved through a sieve with a diameter of 0.15 mm. Next, 1 g of the powder was added to 10 mL of 1% hydrochloric acid, and the resulting mixture was thoroughly shaken and sonicated for 30 min at 150 W and 40 kHz. After allowing the mixture to stand at room temperature for 30 min, it was centrifuged at 8,586 × g for 10 min. The supernatant was collected and filtered using a 0.45 µm membrane filter. Subsequently, 20 µL of the filtered solution was taken for further analysis using high performance liquid chromatography(HPLC). In the HPLC method, a C18 column (250 × 4.5 mm, 5 µm) was used at 25 °C. A 20 µL sample was injected and monitored at 235 nm. A mobile phase of acetonitrile-tetrahydrofuran (9:1):0.1 mol/L ammonium acetate solution (70:30 v/v) with a flow rate of 0.8 mL min −1 was utilized. The analyses were performed using a Shimadzu LC-20A liquid chromatograph with a diode array detector (CBM20A, Shimadzu, Tokyo, Japan).

ZnO NPs-treated pot experiments

The test was conducted in a greenhouse located at Chengdu Normal University (30°40′47″N, 103°49′16″E) in Chengdu, Sichuan Province, China. Pots were prepared and filled with 6 kg of soil for the plant growth experiment. The soil used in the study was obtained from a farm near the university, and natural drying was used to remove inclusions such as rocks and plant matter. The dried soil samples were ground to pass through a two mm sieve and mixed thoroughly and homogeneously. The soil had a pH of 6.4, electrical conductivity (EC) of 143.1 ms/m, soil organic matter (SOM) of 161.7 mg/kg, available nitrogen (AN) of 75.62 mg/kg, available phosphorus (AP) of 26.53 mg/kg, available potassium(AK) of 110.32 mg/kg, and available zinc (AZn) of 0.35 mg/kg.

The A. carmichaelii plants used in this study were artificially cultivated in Butuo County. ZnO NPs (with a diameter of 50 ± 10 nm and 99.9% purity) were purchased from Shanghai Macklin Biochemical Co. Ltd and used as test NPs in this study. In each pot, we applied 100 mL of a solution consisting of either distilled water (CK), bulk ZnO at a concentration of 30 mg/L, or ZnO NPs at the same concentration. To avoid particle aggregation, both bulk ZnO and ZnO NPs were suspended directly in deionized water, followed by ultrasonic vibration (200 W, 40 kHz) for 30 min prior to use. These treatments were applied three times to each pot.

Metabolite extraction

The metabolites of A. carmichaelii lateral roots were analyzed soon after harvest. The samples were freeze-dried, ground into powder, and dissolved in one mL of extract solution (methanol: water = 7: 3, v/v) that had been pre-cooled to −20 °C before being ground again. The mixtures were left overnight and then centrifuged at 13,000 × g for 10 min at 4 °C. Following centrifugation, 800 µL of supernatant was collected and filtered through a 0.22 µm membrane filter before being transferred to a sample bottle for LC-MS analysis.

LC-MS/MS technology was used for conducting a broad-spectrum targeted metabolomic analysis to quantitatively measure plant metabolites. The QTRAP 6500 Plus high sensitivity mass spectrometer (SCIEX, Framingham, MA, USA) was employed for quantitative detection of MRM (Multiple reaction monitoring) in the samples. The Waters ACQUITY UPLC I-Class (Waters, Milford, MA, USA), tandem QTRAP6500 Plus high sensitivity mass spectrometer (SCIEX, Framingham, MA, USA) was utilized for effective separation and accurate quantitative detection of metabolites. An ACQUITY UPLC HSS T3 column (100 × 2.1 mm, 1.8 µm, Waters) was used as the chromatographic column. The mobile phase consisted of liquid A, an aqueous solution containing 0.1% formic acid, and liquid B, 100% acetonitrile containing 0.1% formic acid. The elution process was carried out in four steps: 0–2 min, 5% B solution; 2–22 min, 5%–95% B liquid; 22–27 min, 95% B liquid; 27–30 min, 5% B solution. The flow rate used was 0.3 mL/min at a column temperature of 40 °C. For the QTRAP 6500 Plus equipped with the EST Turbo ion spray interface, the ion source parameters were set as follows: Ion source temperature: 500 °C; Ion spray voltage (IS): 4500 V (positive mode) or −4500 V (negative mode); The ion source gases I (GS1), II (2), and curtain gas (CUR) were set at 40 psi each.

Metabolite data analysis

To identify and quantitatively analyze the metabolites, Skyline (MacCoss Lab, Seattle, WA, USA) and BGI-Wide Target-Library (BGI Co., Ltd., Shenzen, China) were used (Wen et al., 2017). For data statistical analysis, metabolite classification annotation, and functional annotation, MetaboAnalyst 5.0 was employed (Pang et al., 2022).

The variable importance in projection (VIP) of the orthogonal partial least-squares discriminant analysis (OPLS-DA) model was used to screen out differential metabolites between the control group and the experimental group (Jia et al., 2021; Zhou et al., 2021). Metabolites were considered differential if they had VIP ≥1, absolute Log2FC (fold change) ≥1, and fold change (FC) ≥2 or FC ≤0.5 for group discrimination (Cao et al., 2019). To confirm the OPLS-DA model’s screening accuracy, principle component analysis (PCA) was used to assess the overall clustering pattern of the two treatment groups. The correlation between samples was calculated using Pearson correlation.

Rhizosphere microbial analysis

DNA samples for the soil microbiome were collected from the rhizosphere area of the plants at harvest. All high-throughput sequencing was completed by Biomarker Technologies Co., Ltd (Beijing, China). DNA purification was performed using the Soil DNA Kit (MN NucleoSpin 96 Soil). The method adhered to the manufacturer’s instructions for the extraction procedure. Initially, the soil was weighed and 700 µL of Buffer SL1 and Buffer SL2 was added. The MN Bead Tube was then filled with fresh lysis buffer up to the 1.5 mL mark, followed by the addition of 150 µL of Enhancer SX. To prevent leakage, the cap was tightly closed. Subsequently, the sample was lysed by horizontally attaching the MN Bead Tube to a vortexer and vortexing at full speed and room temperature (25 °C) for 5 min. Afterward, the tube underwent centrifugation at 11,000 × g for 2 min to eliminate any foam generated by the detergent. Following this step, 150 µL of Buffer SL3 was added, and the mixture was vortexed for 5 s. An incubation period of 5 min at 0−4 °C ensued. The tube was then centrifuged at 11,000 × g for 1 min to separate the lysate. To remove any impurities, the lysate was filtered through an appropriate filter. The DNA was bound to a silica membrane according to the manufacturer’s instructions. Subsequently, the silica membrane was washed and dried following the kit’s protocol to eliminate any contaminants. To ensure optimal DNA recovery, the silica membrane was thoroughly dried. Finally, the DNA was eluted from the silica membrane using the recommended elution buffer or water, as specified in the kit’s instructions. The V3-V4 hypervariable region of the 16S rRNA gene was amplified for each sample using barcoded universal primers 338 F (5′-ACTCCTACGGGAGGCAGCA-3′) and 806 R (5′- GGACTACHVGGGTWTCTAAT-3′) (Zhang et al., 2023). For amplicon sequencing, the Illumina NovaSeq 6000 PE250 platform (Illumina, San Diego, CA, USA) was utilized. The reads were merged and filtered by size and quality, and then clustered into operational taxonomic units (OTUs) using an open reference strategy based on 97% identity with UPARSE (version 7.0). The bacterial 16S rRNA data were analyzed on the BMKCloud (http://www.biocloud.net) platform. Alpha-diversity was evaluated using the Shannon and Simpson index, while Beta-diversity was assessed using Principal Coordinates Analysis (PCoA) based on Bray-Curtis distance.

Correlation between soil trace elements, yield and monoester-type alkaloids content in A. carmichaelii

The relationship between soil trace elements, yield and monoester-type alkaloids was analyzed using Pearson correlation (Fig. 1, p < 0.05). The results s indicate a moderate positive correlation between the content of AMn in the soil and yield (0.3 ≤r ≤ 0.5). Additionally, AZn is strongly positively correlated with yield and content of monoester alkaloids (0.5 <r < 1). On the other hand, the content of AFe and TFe exhibit a moderate negative correlation with yield (−0.5 ≤ r ≤−0.3), while TZn content is significantly negatively correlated with the content of monoester alkaloids (−0.5 ≤ r ≤−0.3). No significant correlation was found between the other elements and the plant. In terms of the relationship between different trace elements in soil, it was found that AZn is positively correlated with ACu (0.5 <r <1) but strongly negatively correlated with AFe (−1 < r <  − 0.5) and moderate negatively correlated with TFe (−0.5 ≤ r ≤−0.3). Moreover, ACu exhibits a moderate negative correlation with AFe(−0.5 ≤ r ≤−0.3), while no significant correlation was found between the other elements.

Response of plants to ZnO NPs

To investigate the impact of ZnO NPs on the growth and pharmacodynamic components of A. carmichaelii, we analyzed the plant height, fresh weight of the lateral root, and content of monoester-type alkaloid. The phenotypic changes of the plant were illustrated in Figs. 2A & 2B, while the lateral roots were shown in Fig. 2C. The bar graph (Figs. 2D–2F) clearly demonstrates that the application of ZnO NPs significantly promotes both plant growth and the content of monoester-type alkaloids (*p < 0.05, **p < 0.01). Specifically, our observations revealed an increase in plant height by 23.31% when compared to the distilled water treatment, and a remarkable 63.77% increase when compared to bulk ZnO treatment (Fig. 2D). The fresh weight of the lateral roots also witnessed significant improvement with an increase of 52.69% compared to the distilled water treatment and a staggering 88.58% improvement when compared to bulk ZnO treatment (Fig. 2E). Furthermore, we found an increase in the content of monoester-type alkaloids by 20.32% when compared to the distilled water treatment and an impressive 30.48% increase when compared to bulk ZnO treatment (Fig. 2F). These findings underscore the tremendous potential of ZnO NPs in enhancing plant growth and improving pharmacodynamic components in the field.

Impact of ZnO NPs on lateral root metabolites of A. carmichaelii

Further metabolomics analysis was performed on the lateral roots of A. carmichaelii that were treated with ZnO NPs and bulk ZnO to investigate the impact of ZnO NPs on the plant’s metabolites. The PCA score plot showed PC1 could explain 50.7% of the total metabolic variation, while PC2 could explain 32% (Fig. 3A). Moreover, PC1 clearly distinguished the bulk ZnO and ZnO NPs treatments, revealing that the content of the selected metabolites was significantly altered between two treatments. The Volcano plot (Fig. 3B) clearly indicates a significant difference between the bulk ZnO and ZnO NPs treatments, as evidenced by the separation of the data points into two distinct clusters. The significant 38 metabolites were selected based on VIP ≥1, absolute Log2FC (fold change) ≥1, FC ≥2 or FC ≤0.5, and P < 0.05, with 23 metabolites upregulated and 15 metabolites downregulated (Table S1 & Fig. 3B). The differential metabolites between the two treatments encompassed a wide array of compound classes, including carbohydrates, amino acids, guanidines, amines and derivatives, polyketides, terpenoids, flavonoids, branched unsaturated hydrocarbons, hydroxyindoles, alkaloids, and more. The ZnO NPs treatment upregulated certain metabolites involved carbohydrate and amino acids related to plant growth or stress resistance, such as D-Mannosamine and D-Proline, while downregulating certain glycosides and terpenes, including Rehmannioside C. Notably, the alkaloid 14-Benzoylaconine, a clinically significant component of A. carmichaelii, showed an increase. The correlation of these 38 metabolites was presented in a correlation heatmap (Fig. 3C).

To gain a comprehensive understanding of how plants respond to ZnO NPs, a metabolic pathway analysis was conducted using MetaboAnalyst 5.0. This study assessed 53 pathways in the lateral roots of A. carmichaelii and identified eight pathways where differential metabolites are involved. The metabolic pathways were arginine and proline metabolism (1), aminoacyl-tRNA biosynthesis (2), flavone and flavonol biosynthesis (3), glutathione metabolism (4), alkaloid biosynthesis (5), ascorbate and aldarate metabolism (6), tryptophan metabolism (7), and amino sugar and nucleotide sugar metabolism (8). Among these eight metabolic pathways, five of them were significantly upregulated by ZnO, excluding flavone and flavonol biosynthesis (3), ascorbate and aldarate metabolism (6), and amino sugar and nucleotide sugar metabolism (8), as shown in Fig. 3D.

Diversity, composition, and functional prediction of rhizosphere bacteria

Similarly, rhizospheric bacteria of A. carmichaelii treated with ZnO NPs and bulk ZnO were assessed to investigate the impact of ZnO NPs on the plant’s rhizosphere. The analysis of the bacterial 16S rRNA data was performed on the BMKCloud (http://www.biocloud.net) platform. The P values of alpha-diversity index (Shannon and Simpson) in rhizosphere bacteria were all smaller than 0.05, suggesting that there was a statistically significant difference in species richness and relative abundance between bulk ZnO and ZnO NPs treatments in rhizosphere bacteria (Figs. 4A & 4B). Beta diversity was analyzed with Principal Coordinates Analysis (PCoA), using Bray-Curtis dissimilarity. The PCoA plot revealed distinct differences between the rhizosphere soil bacterial communities from the ZnO NPs treatment and those associated with the bulk ZnO treatment. The first PCoA axis accounted for 60.44% of the total variability, as shown in Fig. 4C.

To demonstrate the compositional differences of rhizosphere soil bacterial communities between bulk ZnO and ZnO NPs treatment, a family-level analysis was conducted. Specifically, the treatment with ZnO NPs resulted in a decrease in the relative abundance of Gemmatimonadaceae from 10.12% to 8.2%, Sphingomonadaceae from 9.51% to 7.22%, SC I 84 from 3.76% to 2.56%, Haliangiaceae from 3.52% to 2.44%, and Unclassified_Elsterales from 2.89% to 2.3%, among others. On the other hand, there was an increase in the levels of Streptomycetaceae from 0.29% to 3.06%, Oxalobacteraceae from 1.49% to 2.29%, Rhizobiaceae from 1.97% to 2.86%, Chitinophagaceae from 3.18% to 4.17%, and Solibacteraceae from 4.92% to 6.34% (Fig. 4D). Specifically, the treatment with ZnO NPs led to a reduction in the relative abundance of SC I 84 showed a decrease from 3.76% to 2.56%, accounting for a reduction of 31.89%. Haliangiaceae decreased from 3.52% to 2.44%, resulting in a decrease of 30.42%. Sphingomonadaceae decreased from 9.51% to 7.22%, indicating a decrease of 24.07%. Unclassified_Elsterales decreased from 2.89% to 2.3%, representing a decrease of 20.21%. Additionally, Gemmatimonadaceae from 10.12% to 8.2%, showing a decrease of 18.92%. In contrast, there was an increase in the levels of Streptomycetaceae from 0.29% to 3.06%, representing an increase of 941.58%. Oxalobacteraceae increased from 1.49% to 2.29%, indicating an increase of 54.42%. Rhizobiaceae increased from 1.97% to 2.86%, showing an increase of 44.71%. Moreover, Chitinophagaceae increased from 3.18% to 4.17%, accounting for an increase of 30.9%, while Solibacteraceae increased from 4.92% to 6.34%, representing an increase of 28.82% (Fig. 4D). To identify the specific bacteria associated with the ZnO NPs treatment and distinguish the predominant taxa in the rhizosphere soil, LEfSe was utilized to generate a cladogram (all LDA scores (log10) >3.5, P < 0.05). The circles, representing the classification levels from phylum to species, were ordered from inside to outside. Each small filled circle indicated a classification at that level, with the size proportional to its relative abundance. The relative abundances of Bacteroidota and Actinobacteriota in the ZnO NPs treatment were highly significantly higher compared to the bulk ZnO treatment. Specifically, the relative abundance of Bacteroidota increased by 47.55%, while Actinobacteriota showed a 41.62% increase in the ZnO NPs treatment, as demonstrated in Fig. S2 (P < 0.01). These augmented levels of Bacteroidota and Actinobacteriota were considered as the biomarkers of the ZnO NPs treatment (Figs. 5A & 5C).

In order to explore the functional changes of rhizosphere bacteria under different treatments in this study, PICRUSt2 (Hierarchy level 2) was used to analyze the changes of the rhizosphere bacterial community of A. carmichaelii. Results based on KEGG database (Kyoto encyclopedia of genes and genomes) were indicated in Fig. 5B and Table S2. PICRUSt2 analysis revealed significant differences in the functional abundance of bacteria from bulk ZnO and ZnO NP treatments, including changes in 22 metabolic pathways, such as carbohydrate metabolism, energy metabolism, cofactor and vitamin metabolism, membrane transport, and nucleotide metabolism.

Covariation between bacterial and metabolites of the lateral roots

To investigate the relationship between 38 upregulated/downregulated metabolites (Table S1 & Fig. 3B) and the top 20 most abundant bacteria at the family level, we conducted a correlation analysis. We obtained the covariation between rhizosphere soil bacteria and metabolites in the lateral roots of A. carmichaelii using Spearman’s rank correlation with two-tailed nominal p values (Fig. 6). Only metabolites and bacteria displaying significant correlations were included in the figure.

The results presented in Fig. 6 showed significant differences between the bulk ZnO and ZnO NP treatments. The lateral roots of A. carmichaelii exhibited up-regulated expression of L-Alanyl-L-glutamine, 14-Benzoylaconine, Serotonin, D-Mannosamine, Metformin, and DL-Mevalonolactone in response to ZnO NPs treatment. What is intriguing is that their expression levels displayed positive correlations with specific microbial communities and reduced negative correlations. For example, the abundance of Solibacteraceae and Chitinophagaceae, which showed positive correlation with D-Mannosamine expression, significantly increased, while Gemmatimonadaceae and unclassified_Elsterales, which were negatively correlated with D-Mannosamine expression, decreased. Similarly, the expression levels of L-(+)-Selenomethionine, Ribitol, and (R)-(+)-Limonene were downregulated in the lateral roots of A. carmichaelii treated with ZnO NPs. Interestingly, bacteria that negatively correlated with these metabolites were more abundant while those that positively correlated with them were less abundant.