Soil derived metabolic profiling and their impact on the root growth in peanuts (Arachis hypogaea L.)

Abstract

See More

Plant growth is intricately regulated by soil ecosystems, where dynamic interactions between plants and soil metabolites shape root development. As critical mediators of these interactions, soil metabolites not only reflect biogeochemical cycling but also directly modulate root morphogenesis by eliciting stimulatory or inhibitory responses. To decode the mechanisms driving peanut (Arachis hypogaea L.) root system development, utilizing UPLC-HRMS we profiled 702 soil specific metabolites across soil samples collected from five different regions. Further 118 differentially expressed metabolites were identified in collected soil samples, and 10 metabolites were selected to validate their function associated with peanut root length phenotype. Through systematic screening, four root promoting metabolites (nicotinamide, carbendazim, vanillic acid, and raffinose) and four phytotoxic compounds (phthalic acid, myristic acid, formononetin, and syringic acid) were identified. Our results showed that the seedlings treated with nicotinamide, carbendazim, vanillic acid, and raffinose promotes root elongation by up to 28.3% as compared to untreated seeds. Whereas, seedlings treated with phthalic acid, myristic acid, formononetin, and syringic acid, suppressed root growth by 56.6%, demonstrating a bimodal inhibition pattern. Dose response assays revealed hierarchical efficacy among these metabolites, with carbendazim and formononetin representing the most potent enhancer and suppressor, respectively. Current findings reveal a causal link between soil metabolite composition and peanut root development, providing a biochemical basis for harnessing soil specific metabolites in precision agriculture.

Graphical Abstract

Introduction

Peanut (Arachis hypogaea L.), native to tropical South America, serves as a vital source of edible oil and protein in developing countries and is globally recognized as one of the four major oilseed crops [56]. This crop combines nutritional, economic, and ecological values, making it an indispensable agricultural commodity worldwide [3, 31]. As a leguminous plant (Fabaceae family), peanut is unique for its geocarpic fruiting habit. During its growth and development, the root system serves as a critical organ for nutrient uptake and biosynthesis. Notably, peanut roots form symbiotic associations with rhizobia, enabling biological nitrogen fixation (BNF) through root nodules. Consequently, root traits profoundly influence peanut growth vigor and yield potential, with root architecture directly determining crop productivity [21, 43].

Root growth is intrinsically linked to soil, which serves not only as a physical substrate for root anchorage but also as a dynamic medium for environmental sensing and response. Soil harbors diverse metabolites that play pivotal roles in regulating root system development [61]. Soil metabolites refer to compounds generated through diverse biological and biochemical processes, which accumulate in soils and modulate their physical, chemical, and biological properties. The composition and concentration of these metabolites vary significantly across soil types, driven by differences in pedogenesis, parent material, climatic conditions, and vegetation cover [8]. The soil metabolome exhibits remarkable complexity and diverse origins, encompassing root exudates, microbial metabolites, and decomposition products of plants, microorganisms, and soil organic matter. Characteristic metabolite classes include sugars, amino acids, organic acids, phenolics, and alkaloids, which collectively shape the biochemical milieu to regulate nutrient availability, microbial symbiosis, and root architecture [27, 58]. In leguminous crops such as peanut, symbiotic nitrogen fixation (SNF) represents a critical yet highly sensitive process that requires precise coordination of soil biochemical conditions and metabolite networks [17]. Despite variations in edaphic factors, the extent to which specific soil metabolites mechanistically regulate root morphogenesis remains an unresolved question in plant-soil interactions.

Metabolomics has emerged as an increasingly crucial tool in related research, offering distinct advantages of high sensitivity and high-throughput capacity. Leveraging advanced detection and data processing technologies, it enables comprehensive and accurate profiling of biological metabolic signatures, thereby effectively reflecting metabolic abundance in biological samples [55]. In recent years, metabolomics has played an important role in soil metabolite research. Through technological innovations and multidimensional applications, it has evolved from single metabolite detection to a systemic tool for deciphering soil ecological function [48, 49]. Emerging studies have revealed that numerous soil metabolites play fundamental roles in ecological regulation and biological interactions, nutrient cycling and energy flux, as well as biosynthetic potential within soil ecosystems [5, 26, 60]. Notably, benzoxazinoids (BXs), as key secondary metabolites, not only mediate plant–microbe interactions but also enhance plant stress adaptation by reshaping soil fungal communities (e.g., suppressing pathogen abundance). Meanwhile, soil metabolites such as amino acids and organic acids directly participate in microbially driven nitrogen and phosphorus cycling, thereby regulating soil fertility [16]. Furthermore, organic acids modulate soil bacterial composition, influencing the degradation of organic pollutants [36]. However, given the vast diversity of soil metabolites, further research is needed to elucidate their specific roles in plant growth regulation.

In this study, we aimed to systematically decode the region-specific chemical profiles of agriculturally diverse soils and determine their influence on peanut root system development. Using ultra-performance liquid chromatography coupled with high-resolution mass spectrometry (UPLC-HRMS), we performed untargeted soil metabolomics across five distinct soil types. By integrating these metabolic profiles with phenotypic root architecture screening, we identified soil-type-specific metabolite signatures that exert direct regulatory effects on root growth. Specifically, we uncovered a set of metabolites, including four organic acids (e.g., carbendazim) that promoted root development and four phytotoxic compounds (e.g., formononetin) that inhibited root growth. These findings provide mechanistic insight into the role of rhizosphere chemical cues in shaping root architecture. The primary objective of this research was to establish a functional link between soil metabolite composition and root phenotypic responses, thereby laying the groundwork for metabolite-informed agronomic strategies to enhance peanut resilience and productivity across variable edaphic environments.

Materials and methods

Sampling locations and collection of soil samples

This experiment was conducted at the Crops Research Institute of Guangdong Academy of Agricultural Sciences in Guangzhou, Guangdong. On August 12, 2024, soil samples were collected from five peanut production areas in Guangdong Province, including Liucheng Town, Dongyuan County, Heyuan City (HY) (115°7′27.948″E, 24°1′49.264″N); Lintou Town, Dianbai District, Maoming City (MM) (111°3′32.544″E, 21°39′47.160″N); Chengcun Town, Yangxi County, Yangjiang City (YJ) (111°42′55.476″E, 21°49′8.580″N); Zhongluotan Town, Baiyun District, Guangzhou City (GZ) (113°24′16.416″E, 23°23′7.368″N); and Dongshi Town, Pingyuan County, Meizhou City (MZ) (115°57′13.104″E, 24°40′22.728″N). Current sampling sites were selected to comprehensively reflect the overall soil types and regional variations within Guangdong Province. At each sampling site, a five-point sampling method was employed, where five topsoil samples (0–20 cm depth) were randomly collected in an “S” pattern. The five subsamples from the same location were mixed to form one composite sample. The samples were labeled as MZ, YJ, HY, GZ, and MM, corresponding to their respective locations, and stored in sample bags with detailed information, including sampling location, date, and collector. The soil sample were collected and flash frozen in liquid nitrogen, and stored in the ultra-low temperature freezer (−80 °C). The soil samples were finely ground, sieved through a 5 mm mesh, and stored at −80 °C. Samples were further sent to Guangzhou Hezhong Biotechnology Co., Ltd. for analysis of soil physicochemical properties.

Soil color spectrometry

Following standardized preparation (air-drying at 25 ± 2 °C, grinding through 2 mm stainless steel sieve, and organic debris removal), homogenized soil samples were stored in light-protected polyethylene bags (UV-blocking grade, 0.1 mm thickness) [34]. Further, for colorimetric measurements an MSP-800 spectrophotometer (MSP; X-Rite Pantheon, USA) was utilized. A randomized 10-point sampling strategy per sample ensured spatial representativeness, with three technical replicates averaged to minimize positional bias [35]. Soil color was quantified using the CIE (Commission Internationale de l’Éclairage (International Commission on Illumination)) L*, a*, b* color space. The L* value represents lightness on a scale from 0 (black) to 100 (white), with higher values indicating brighter soils. The a* value represents the red-green axis, where positive values denote increasing redness and negative values indicate greenness. The b* value represents the yellow-blue axis, with positive values corresponding to yellowness and negative values to blueness.

Soil metabolite extraction

For metabolite extraction, soil aliquots (50 mg) were weighed into 2.0 mL microcentrifuge tubes, combined with 500 μL of 80% methanol (v/v, ice-cold), and homogenized with stainless steel beads using a ball mill. The homogenates were placed in a −20 °C refrigerator for 30 min for protein precipitation. The mixture was then centrifuged at 13,000 g for 15 min, and 400 μL of the supernatant was transferred to new microcentrifuge tubes. After that the supernatant was freeze-dried and reconstituted by adding 100 μL of 50% ice methanol solution. Centrifugation was performed at 13,000 g for another 15 min, and then the supernatant was transferred to the injection bottle for detection via UPLC-HRMS. An aliquot of 10–20 μL of each sample was taken to form a quality control (QC) sample for UPLC-HRMS detection. The whole extraction process was operated on ice [11].

Detection of soil metabolites via UPLC-HRMS

In the present study, untargeted metabolomic profiling of soil samples was performed using ultra-high-performance liquid chromatography coupled with quadrupole time-of-flight mass spectrometry (UPLC-QTOF-MS/MS). Chromatographic separation was carried out on an ACQUITY UPLC I-Class system (Waters, USA), equipped with an ACQUITY UPLC HSS T3 column (100 mm × 2.1 mm, 1.8 μm; Waters, UK). The column was maintained at 40 °C throughout the analysis. The mobile phase consisted of solvent A (water containing 5 mmol/L ammonium acetate and 5 mmol/L acetic acid) and solvent B (LC–MS grade acetonitrile). A gradient elution protocol was applied at a flow rate of 0.3 mL/min. High-resolution mass spectrometric detection was conducted on a TripleTOF 6600 system (SCIEX, Framingham, MA, USA). Each sample was analyzed in both positive and negative ionization modes to ensure comprehensive metabolite coverage. The ion source conditions were optimized as follows: curtain gas pressure at 30 PSI, nebulizer gas (Gas 1) and auxiliary heater gas (Gas 2) both set at 60 PSI, and source temperature at 500 °C. The ion spray voltage was set to + 5000 V for positive mode and −4500 V for negative mode.

Data acquisition was conducted in information-dependent acquisition (IDA) mode. For each cycle, full MS scans were recorded in the range of 60–1200 Da with an accumulation time of 150 ms. The top 12 precursor ions (signal intensity > 100 counts) were selected for MS/MS fragmentation using dynamic exclusion set at 4 s to minimize repeated scanning of abundant ions. To maintain data quality and consistency, the instrument was calibrated after every 20 sample injections. Quality control (QC) samples were injected at regular intervals (every 10 samples) throughout the run. Systematic drift in mass accuracy or retention time was corrected post-acquisition by aligning the observed deviations in QC sample profiles.

Data pre-processing and analysis

The obtained raw mass spectrometry data were converted to the mzXML format using the MSConvert tool within ProteoWizard. Subsequent data preprocessing, including peak detection, peak grouping, retention time correction, a second round of peak grouping, and the annotation of isotopes and adducts, was performed using the XCMS software package. The processing was conducted within the R environment, utilizing the CAMERA and metaX toolboxes for comprehensive metabolomic analysis. XCMS processing employed the centWave method with the following parameters: minfrac = 0.5, snthr = 6, ppm = 30, peakwidth = c (5, 25), bw2 = 5, mzwid = 0.015, and profStep = 0.01. Each ion was characterized by its specific retention time and m/z value. Peak intensities were recorded to construct a three-dimensional matrix consisting of peak indices (defined as RT-m/z pairs), sample names, and ion intensity values.

Peak intensity data were further processed using metaX. Initial metabolite identification was performed by matching accurate mass against the KEGG and HMDB databases with a mass tolerance of 10 ppm, considering common adducts in positive mode ([M + H]⁺, [M + Na]⁺, [M + K]⁺, [M + NH₄]⁺) and negative mode ([M-H]⁻, [M + NH₄-2H]⁻, [M + 2Cl]²⁻, [2 M-3H]³⁻). Metabolic features detected in fewer than 50% of quality control (QC) samples or 80% of biological replicates were discarded. Missing values in the remaining data were imputed using the k-nearest neighbor algorithm. The dataset was then normalized using the Probabilistic Quotient Normalization method. A quality control-based robust LOESS signal correction was applied to the QC samples to compensate for instrumental drift over the acquisition sequence. Furthermore, metabolic features with a relative standard deviation (RSD) greater than 50% across all QC samples were removed to ensure data quality.

Reproducibility was assessed through principal component analysis (PCA) and Spearman correlation analysis of intra-group and QC samples. Differences in metabolite abundance between two phenotypic groups were evaluated using Student’s t-tests. Resulting p-values were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate (FDR) method, with metabolites retaining significance at an adjusted p-value < 0.05. Supervised multivariate analysis was performed using partial least squares-discriminant analysis (PLS-DA) implemented in metaX to discriminate between groups. Features with a variable importance in projection (VIP) score greater than 1.0 were selected as important. Orthogonal PLS-DA (OPLS-DA) was conducted using the ropls R package. Final differentially abundant metabolites (DAMs) were identified by integrating VIP values from the OPLS-DA model (VIP > 1), fold change (FC > 1), and statistical significance (p value < 0.05).

Metabolite annotation was conducted by matching the accurate mass (m/z) of samples to entries in the online KEGG and HMDB databases within a mass error of 10 ppm. Proposed molecular formulas were subsequently validated by assessing their isotopic distribution patterns. For the identification of secondary metabolites, a more stringent workflow was applied using a mass tolerance of 0.01 Da for MS1 and 0.05 Da for MS2 data, with an identification score cutoff of 70%. Identifications were further confirmed using a combined fragment spectral database (Lipidblast v37, Massbank, HMDB v4.0, and an in-house spectral library). All identified compounds were classified, and their associated pathway information was retrieved from the KEGG database. Finally, differentially abundant metabolites were functionally annotated within KEGG to elucidate their potential physicochemical properties and biological roles.

Chemical reagents preparation

All reagents and solvents used were of analytical grade. Ten metabolites were selected from the differentially abundant metabolites profile in distinct soil samples for further functional validation, including the syringic acid (SA), nicotinamide (NAM), carbendazim (CBZ), vanillic acid (VA), raffinose (Raf), phthalic acid (PA), paclobutrazol (PBZ) (a commonly used synthetic plant growth regulator), betaine (Bet), myristic acid (MA), and formononetin (FNT). All compounds were sourced from the CASmart chemical platform (https://physics.casmart.com.cn/).

For compounds with poor water solubility (SA, CBZ, VA, PA, PBZ, MA, and FNT), 0.5 g of each was accurately weighed and dissolved in 100 mL of HPLC-grade methanol. The solutions were gently heated to 40–50 °C to facilitate dissolution and subsequently prepared as 5 g/L stock solutions. These stock solutions were aliquoted into amber vials and stored at 4 °C to prevent photodegradation. Working solutions at a concentration of 0.5 g/L were prepared by a tenfold dilution of the stock using methanol immediately prior to use.

For water-soluble metabolites (NAM, Raf, and Bet), 0.05 g of each compound was dissolved in 100 mL of ultrapure water, yielding stock solutions at 0.5 g/L. These aqueous stocks were also stored at 4 °C and used directly for bioassay treatments without further dilution. This standardized preparation protocol ensured consistency and stability across all metabolite treatments in subsequent peanut root growth experiments (Table 1).

Table 1 The concentration of ten soil metabolites used for the root germination tests

Peanut seed treatment

Seed germination

Uniform sized seeds of the peanut variety Yueyou1823 were selected and surface-sterilized by soaking in 75% ethanol for 8 min, then rinse three times with distilled water. Petri dishes were sterilized with 75% ethanol and rinsed with distilled water. Three layers of filter paper were placed in each petri dish, and an equal volume of distilled water was added. Thirty seeds were evenly distributed per petri dish (330 seeds total), and then petri dishes were placed in a light incubator under controlled conditions: 26 °C temperature, 6600 lx light intensity (400–750 nm wavelength), and a 12 h light/12 h dark photoperiod. Petri dishes were observed for approximately 12 h post incubation and germination was defined as successful upon radicle emergence.

Planting and root measurement

For planting, a cultivation substrate of vermiculite and perlite (1:2 ratio) was prepared and germinated peanut seeds were planted in the substrate and treated with 0.5 g/L of the tested organic reagents. Root lengths were recorded 7 days post-planting. The SPSS 23.0 software (https://www.ibm.com/products/spss) was used for significance analysis of observed root lengths.

Results

Soil type and color difference analysis in different peanut growing areas of Guangdong

Soil color holds significant diagnostic value for assessing physical properties, as it provides preliminary judgments regarding soil formation processes and fertility levels, thereby providing foundational data for subsequent soil type identification. Initial observations revealed distinct color variations among the five soil samples (Fig. 1A, B). Soil samples collected from site YJ exhibited a light gray color, whereas those from site MM displayed a dark brown coloration. The L* values of the five soil samples ranged between 42.65 and 55.27, with MM showing the lowest L* value (displayed dark brown coloration) and HY showing the color highest L* value (displayed bright yellow coloration). Analysis of a* and b* values of collected soil samples showed that a* values ranged from 3.68 to 16.55, and b* values ranged from 13.17 to 24.05. The soil samples collected from YJ had the lowest a* and b* values (3.68 and 13.17, respectively), corresponding to a light gray soil color (Fig. 1C).

Fig. 1

Sampling of soil samples from different peanut growing areas and color difference analysis. A Collection points for soils from different peanut growing areas. B Photographs of soil samples. C Soil sample color difference analysis data. The L* value represents the brightness of the color, ranging from 0 (black) to 100 (white). The a* value indicates the red-green tone of the color, with positive values denoting red and negative values denoting green. The b* value represents the yellow-blue tone of the color, with positive values indicating yellow and negative values indicating blue.

Global metabolomic profiling and statistical characterization of soil samples

A comprehensive mass spectrometry analysis was conducted using both positive and negative ionization modes, resulting in a large dataset of raw spectral data. Post-processing of this data revealed 17,573 metabolic features in the positive mode (Table S1) and 11,707 features in the negative mode (Table S2). The distribution of total ion signals was examined across two dimensions: mass-to-charge ratio (m/z) and retention time (RT), capturing the global profile of metabolite hydrophobicity/hydrophilicity and molecular weight (Figs. S1–S2). Total Ion Chromatograms (TIC) were generated by summing the chromatographic intensities across all detected metabolites in each sample (Fig. 2A), providing an overview of the global signal during separation. Across the 15 soil samples, retention time patterns were consistent. Ion intensities sharply increased and then decreased to baseline within the 1-5 min range. Between 5 and 10 min, ion intensities increased progressively. In this interval, YG group samples displayed significantly lower ion intensities (both positive and negative modes) compared to other groups. In contrast, the GZ group exhibited markedly higher intensities in negative ion mode, and the MZ group showed higher intensities in positive ion mode.

Fig. 2

Quality control analysis of soil metabolomics data. A Total ion flow diagram of all metabolite intensities superimposed in soil samples. B The taxonomic classification of secondary-identified metabolites in soil samples. C Cumulative plot of MS/MS spectral matching and retention time versus mass-to-charge ratio. D Soil metabolite principal component analysis (PCA). E Correlation analysis between samples. F Heat map of clustering metabolite intensities for soil secondary identification.

Metabolite annotation and compound classification

Initial attempts to annotate metabolites were limited by isomeric ambiguities in standard spectral databases, leading to non-specific m/z-metabolite matches. To improve specificity, we employed orthogonal validation by comparing experimental MS/MS spectra with those in authenticated libraries. This enabled high-confidence classification of metabolic features across 15 compound categories (Fig. 2B), including alkaloid derivatives, benzenoids, hydrocarbons, lignans/neolignans, and lipids. Among them, lipids accounted for the highest number of features (n = 237), whereas organohalogen compounds had the fewest (n = 1). Phenylpropanoids and organic acids had comparable feature counts (45 and 44, respectively). To further enhance metabolite identification accuracy, we integrated retention time information into our statistical analysis (Fig. 2C). This revealed clear chromatographic elution order (retention time) was contributed to understand the features in the soil compound composition. Organic oxygen compounds showed an abrupt decline in relative abundance during the 0-2 min window, accompanied by a significant increase in organoheterocyclic compounds. From 2 to 8 min, organoheterocyclic proportions declined steadily, while lipid features increased. After 8 min, lipids represented over 50% of detected features, underscoring their central role in soil metabolite composition. Rigorous quality control protocols were applied to ensure analytical reliability. Principal Component Analysis (PCA) was conducted on 18 experimental samples, including QC replicates, for dimensionality reduction and assessment of sample variability (Fig. 2D). The first two principal components accounted for 42.76% of total variance (PC1: 23.66%, PC2: 19.10%). Notable separation patterns were observed: GZ and MM samples differed along PC1, while YJ and HY groups diverged along PC2, suggesting clear compositional heterogeneity.

Pairwise correlation analysis (Fig. 2E) revealed the strongest inter-sample similarity between GZ and MZ groups (r = 0.9125) and the weakest between HY and YJ (r = 0.7838). Based on these findings, comprehensive metabolite quantification was performed across all samples, followed by hierarchical clustering of the 702 high-confidence metabolites identified via MS/MS (Table S3, Fig. 2F). The results displayed distinct group-specific intensity patterns, with statistically significant inter-group differences. The average intensity distribution of MS/MS-verified metabolites was visualized to assess overall abundance profiles across groups (Fig. S3). These results reinforced the robustness and reproducibility of our data, confirming the reliability of the profiling workflow. Collectively, these validated procedures offer a strong technical framework for downstream applications such as metabolite screening, structural elucidation, and functional annotation.

Screening of specific differential metabolites among soil sample groups

Five groups of soil samples were systematically compared through 10 pairwise combinations to identify specific differential metabolites. Differential screening was conducted using a hybrid approach combining univariate analysis with T-statistics, enabling detection of significantly altered positive and negative ions for each group comparison. To interpret biologically meaningful differences, high-confidence secondary metabolites verified by MS/MS were subjected to differential abundance analysis and pathway mapping (Fig. 3). Volcano plots were generated to visualize fold change and statistical significance for each comparison (Fig. 3A), with red and green dots denoting upregulated and downregulated metabolites, respectively. Among all comparisons, the MM/GZ group showed the largest number of differentially abundant metabolites (n = 306; 56 upregulated, 250 downregulated), while the MZ/GZ comparison yielded the fewest (n = 183; 79 upregulated, 104 downregulated). To further delineate the biological categories of group-specific differential metabolites, their classifications were quantified across all pairwise comparisons (Fig. 3B). Lipid-related metabolites consistently represented the largest proportion of differential compounds, suggesting that lipid metabolism is a key distinguishing feature among the soil groups. The MM/YJ group had the highest proportion of lipid metabolites (35%), emphasizing the metabolic diversity across samples.

Fig. 3

Specific differential metabolite analysis of distinct soil samples. A Differential ion volcano plots for each comparison group. B Statistical pie chart of the number of difference ions in each comparison group. C Venn diagram illustrating the unique and shared ion species among five soil types. D Top 5 marker metabolites ranked by intensity across five soil types.

Analysis of differential metabolite overlaps across the 10 comparisons (Fig. 3C) revealed no common metabolites in the GZ/HY, GZ/YJ, GZ/MZ, and GZ/MM comparisons. Conversely, 30 metabolites were shared among the MZ/HY, MZ/YJ, MZ/GZ, and MZ/MM groups. This distribution highlights unique biochemical signatures and reinforces the distinct metabolic identities of each soil group, as also shown in the significantly different metabolite profiles. The top five (Fig. S4) most abundant metabolites in each group were identified as potential region-specific soil biomarkers. These metabolites are chemically informative indicators of biological activity, organic 5-(2-Furanyl)-3,4-dihydro-2H-pyrrole (M194T369), Mono-2-ethylhexyl phthalate (M277T439), Dansyl acid (M252T269), Dimethenamid OXA (M270T229), and Herbacetin-3,8-diglucopyranoside (M625T314). The GZ region was characterized by 7C-aglycone (M297T298), Nb-p-Coumaroyltryptamine (M307T334), Caffeine (M195T203), N-Methylnicotinamide (M137T178), and 13-Oxo-9,11-tridecadienoic acid (M242T274). The HY region featured (6β,7α,12β,13β)-7-Hydroxy-11,16-dioxo-8,14-apianadien-22,6-olide (M383T384), Macrocarpal E (M471T359), Corosolic acid (M471T412), Eucalyptone (M485T358), and Macrocarpal H (M471T395). MZ samples contained 7-Aminonitrazepam (M252T393), Metolachlor (M284T393), Acetochlor (M270T393_1), Triamterene (M254T393), and Retinyl ester (M301T540). Finally, the YJ region was distinguished by Maltitol (M343T50), Pyraclostrobin (M388T419), 4-Methylumbelliferone (M194T419), all-trans-4-Ketoretinoic acid (M313T392), and Thiamethoxam (M292T228) (Fig. 3D). These metabolites underscore the compositional diversity and ecological specificity of each soil region.

Identification of the differentially abundant metabolites (DAMs) in different soil samples

To rigorously analyze the significantly different metabolites among the five soil groups, a combination of ANOVA (one-way Analysis of Variance) and PLS-DA (Partial Least Squares Discriminant Analysis) was employed. This preliminary screening yielded 13,477 primary differential metabolites. From this pool, 702 high-confidence secondary metabolites were selected based on MS/MS verification for further investigation and quantification. To explore expression trends among the 702 confirmed differential metabolites, Mfuzz clustering analysis was applied, organizing metabolites with similar dynamic profiles into 16 distinct clusters (Fig. 4A). Cluster-wise analysis enabled visual identification of metabolite category distributions across regions. Lipids and organoheterocyclic compounds emerged as dominant categories in most clusters. Notably, Cluster 8 contained the highest number of lipid metabolites (n = 29), whereas alkaloids, organonitrogen, and organosulfur compounds were less represented. A refined subset of 118 differentially abundant metabolites (DAMs) was selected for detailed intensity analysis and visualization across samples (Table S4, Fig. 4B). This analysis revealed minimal intra-group variation, contrasted by substantial inter-group differences. The distribution of metabolite categories across clusters mirrored earlier trends (Fig. 4A). For example, HY samples showed elevated expression in Clusters 14 and 16, while MZ samples displayed strong differential signals in Clusters 4 and 5, reinforcing the presence of region-specific metabolic fingerprints.

Fig. 4

Identification and metabolic pathway analysis of differentially expressed metabolites (DEMs) in different soil samples. A Mfuzz cluster analysis plot of differential expressed metabolites. B Heat map of clustering information for differential expressed metabolites. C Metabolic pathway map of differential expressed metabolites.

Further to elucidate the biological pathways associated with differential metabolites, KEGG enrichment analysis was performed (Fig. 4C). The analysis revealed 11 relevant metabolic pathways, encompassing amino acid metabolism, biosynthesis of other secondary metabolites, carbohydrate metabolism, and chemical structural transformation pathways. Lipid-related pathways were the most enriched, encompassing 11 sub-pathways including fatty acid biosynthesis, elongation, and degradation. Glycerophospholipid metabolism was the most populated sub-pathway, involving 9 identified compounds. In contrast, the “metabolism of cofactors and vitamins” pathway showed the least enrichment, with only three minor sub-pathways: beta-alanine metabolism, phosphonate and phosphinate metabolism, and cyanoamino acid metabolism each represented by a single compound (n = 1). These results underscore that lipid metabolism is the dominant metabolic process among the differentially abundant metabolites in all five soil groups.

Effects of differential soil metabolites on peanut root growth

We initially screened significantly expressed metabolites from the differentially abundant metabolites (DAMs) and selected compounds with known effects on plant growth. Therefore, from the 118 differentially abundant metabolites (DAMs), ten bioactive compounds were selected based on their known or predicted phytological activity. These included syringic acid, nicotinamide, carbendazim, vanillic acid, raffinose, phthalic acid, paclobutrazol, betaine, formononetin, and myristic acid. Syringic acid (SA) and vanillic acid (VA) are widely distributed in plants, these metabolites influence growth and development in a concentration dependent manner [32, 41]. Formononetin (FNT), betaine (Bet), and raffinose (Raf) are known regulators of plant growth and stress responses [12, 39, 54]. Paclobutrazol (PBZ) and carbendazim (CBZ) are synthetic compounds which serve as plant growth regulator and fungicide respectively, and their effects on peanut root growth may provide insights for soil ecological management [62]. Nicotinamide (NAM) is a precursor of NAD, it plays crucial roles in regulating NAD metabolism, redox balance, and immune signaling networks during plant growth and metabolic integration [23]. Phthalic acid (PA) and myristic acid (MA) are significantly expressed organic acids demonstrate important functions. PA, is a major metabolite of phthalate esters (PAEs) in soil, which is widely distributed in agricultural fields and affects crop yield and quality upon plant uptake and metabolism [45, 53]. MA serves as a key functional metabolite involved in metabolic regulation, microbial interactions, and environmental remediation through metabolic pathway enrichment and synergistic effects with other metabolites [59].

Functional validation experiments assessed their influence on peanut root growth, revealing three distinct activity profiles: four compounds significantly promoted root elongation (Fig. 5), four exhibited inhibitory effects (Fig. 6), and two metabolites (paclobutrazol, betaine) showed no statistically significant impact (Fig. S5).

Fig. 5

Effects of beneficial metabolites on peanut root growth. A Molecular and structural formulae of beneficial metabolites. B Root length of peanut treated with beneficial metabolites. C Significance analysis of peanut root length in treatment and control groups. Asterisks indicate statistical significance: *p < 0.05, **p < 0.01. D Significance analysis of peanut root length in different treatment groups. Bars with different lowercase letters indicate statistically significant differences (p < 0.05) as determined by one-way ANOVA followed by Tukey’s honestly significant difference (HSD) post hoc test.

Fig. 6

Inhibitory effects of phytotoxic metabolites on peanut root development. A Molecular and structural formulae of harmful metabolites. B Root length of peanut treated with harmful metabolites. C Significance analysis of peanut root length in treatment and control groups. Asterisks indicate statistical significance: *p < 0.05, **p < 0.01. D Significance analysis of peanut root length in different treatment groups. Bars with different lowercase letters indicate statistically significant differences (p < 0.05) as determined by one-way ANOVA followed by Tukey’s honestly significant difference (HSD) post hoc test.

Soil metabolites enhancing peanut root growth

Among the ten tested compounds, nicotinamide, carbendazim, vanillic acid, and raffinose significantly enhanced root development (Fig. 5A). Dose-dependent analysis revealed a clear efficacy gradient. Carbendazim demonstrated the most pronounced root-promoting activity (average root length = 6.8 cm; 28.3% increase over control), followed by nicotinamide (6.3 cm), vanillic acid (5.8 cm), and raffinose (5.7 cm) (Fig. 5B). The near-equivalent effect of vanillic acid and raffinose (Fig. 5C, D ) suggests potential functional redundancy or synergy in modulating early-stage root growth. The dose-responsiveness observed across all four compounds confirms their role in promoting peanut root elongation, identifying carbendazim as a leading candidate for potential agronomic application.

Soil metabolites inhibiting peanut root elongation

Conversely, four metabolites, phthalic acid, myristic acid, formononetin, and syringic acid exhibited inhibitory effects on root growth (Fig. 6A). Formononetin exerted the strongest suppression (root length = 2.3 cm; 56.6% decrease vs. control), followed by myristic acid (2.4 cm), syringic acid (4.2 cm), and phthalic acid (4.8 cm) (Fig. 6B). This represents a 2.4-fold difference in potency between the most and least inhibitory agents. Based on severity, the compounds clustered into two groups: formononetin and myristic acid elicited severe inhibition (< 2.5 cm), while syringic acid and phthalic acid caused moderate effects (Fig. 6C, D). These findings suggest that specific soil-derived metabolites can significantly modulate peanut root development through diverse inhibitory mechanisms.

Discussion

Regional variability in soil metabolites and their agronomic implications

Plant growth is inextricably linked to soil, which together with plants forms a dynamic and multidimensional symbiotic network [6, 52]. This plant–soil interaction operates as a bidirectional flow of materials and signals, with soil metabolites functioning as central chemical mediators. These metabolites not only reflect the status of soil biogeochemical cycles but also play an active role in shaping the rhizospheric microenvironment, directly impacting root development [33].

In this study, soils from five peanut-growing regions in Guangdong Province Guangzhou (GZ), Meizhou (MZ), Heyuan (HY), Yangjiang (YJ), and Maoming (MM) were analyzed. Substantial differences in key nutrients such as organic matter, nitrogen, and phosphorus were observed. For instance, HY exhibited elevated organic matter levels, whereas GZ had significantly higher available nitrogen. Coupling these findings with untargeted metabolomic profiling revealed distinct metabolic signatures across regions, likely driven by the underlying nutrient heterogeneity. A total of 118 differentially abundant soil metabolites were identified, with lipids constituting the dominant class. Lipid metabolites are known to act as signaling agents within microbial communities, modulating both interspecies communication and structural organization [10]. Their accumulation also enhances soil resilience by reducing contaminant bioavailability [46]. These multifunctional roles underscore the ecological and agricultural importance of lipids in soil metabolic networks, highlighting their potential as biomarkers for soil health and crop productivity.

Insights from LC–MS/MS-based untargeted soil metabolomics

Soil metabolomes are composed of a complex array of substances, including root exudates, microbial secretions, and the degradation products of organic matter. These include organic acids, alkaloids, terpenoids, and phenolic compounds [24]. Traditional research often focused on a narrow set of well-known compounds due to technological limitations. However, advancements in high-throughput metabolomics now enable broad-spectrum profiling of soil metabolomes at an unprecedented scale [63]. LC–MS/MS has emerged as a gold-standard technique for such analyses due to its superior sensitivity, specificity, and resolution [37]. In this study, methanol was used for the extraction of soil metabolites. While methanol efficiently extracts many polar and semi-polar metabolites, it provides limited coverage for certain metabolite classes, particularly lipids and highly polar compounds. A methanol–acetonitrile mixture has been reported to improve metabolite recovery and may represent a more suitable approach for achieving broader coverage. Moreover, metabolite stability in soil differs between chemical classes: polar metabolites (e.g., sugars, amino acids) generally degrade faster due to preferential microbial utilization, whereas nonpolar metabolites such as lipids tend to persist longer. Consistent with this, our study identified a higher proportion of lipid related metabolites, which can originate from soil microorganisms, plant residues, and soil fauna following crop cultivation. Thus, the relative identification of polar versus nonpolar metabolites in soil reflects both crop derived inputs and environmental influences. To enhance metabolite profiling accuracy, future analyses should employ multi solvent extraction strategies, allowing for a more comprehensive representation of soil metabolomes.

In our study, LC–MS/MS-based untargeted metabolomics effectively delineated the metabolic architecture of soils from diverse regions, exposing both inter- and intra-regional metabolic diversity. Previous reports support that soil metabolite distributions are heavily influenced by edaphic factors, including nutrient profiles and physicochemical characteristics [51]. These differences shape localized ecological environments, thus influencing plant growth potential and microbial community structures.

Beneficial soil metabolites promote peanut root development

Soil metabolites serve as crucial biochemical interfaces in the plant–soil continuum, mediating nutrient exchange, signaling, and adaptation. Beneficial metabolites can enhance root growth by modulating nutrient uptake, hormonal pathways, and stress responses. Notably, microbial derived metabolites have been shown to improve phosphorus uptake and influence carbohydrate metabolism in roots, often through the action of arbuscular mycorrhizal fungi (AMF) and their interactions with plant carbon pools [29, 30].

In this study, the application of select soil metabolites to peanut seeds revealed that raffinose, vanillic acid, and nicotinamide significantly stimulated root elongation. Raffinose, a non-reducing trisaccharide, plays an essential role in seed development and germination by providing a stable carbon and energy source [4]. It also enhances seed desiccation tolerance, leading to improved seed vigor [42]. Our experimental data confirm that raffinose treatment resulted in enhanced root growth, possibly due to improved germination kinetics. Vanillic acid, a phenolic compound, was also found to positively affect root development, likely by increasing biomass and enhancing water and nutrient uptake efficiency. Similarly, nicotinamide, a precursor in nicotinamide adenine dinucleotide (NAD) biosynthesis has been associated with stress mitigation and growth promotion. These findings underscore the diverse molecular mechanisms through which soil metabolites can directly support early stage root growth in crops.

Inhibitory effects of soil-derived metabolites on root elongation

While many soil metabolites are beneficial, others exhibit allelopathic or toxic properties that impair plant growth. These detrimental effects are often mediated through hormonal interference, oxidative stress induction, or disruption of rhizospheric microbial balance. Our exogenous application tests identified four metabolites, phthalic acid, syringic acid, formononetin, and myristic acid as inhibitory to peanut root elongation.

Phthalic acid and its derivatives (phthalate esters) are recognized environmental pollutants with endocrine-disrupting properties, known to persist in soil and negatively affect both soil microbiota and plant physiology [48, 49]. Formononetin, an isoflavone primarily found in leguminous plants like Astragalus (Astragalus membranaceus Bunge) and Pueraria (Pueraria montana (Lour.), has pharmacological relevance due to its estrogenic activity but demonstrated strong inhibitory effects on root growth in our study [7]. Myristic acid, though beneficial as a fungal carbon source at appropriate levels, can disrupt rhizosphere dynamics and hinder root development when present in excessive concentrations [20]. Syringic acid is widely found in plant kingdom, a common phenolic acid, modulates hormonal signaling particularly auxin and ethylene pathways and can alter microbial composition via root exudate modification, leading to reduced root expansion [29, 30]. Moreover, the reduced form of syringic acid serves as a precursor in plant lignin biosynthesis, while syringic acid itself is readily degraded by soil bacteria. Given the critical role of root development in crop yield, especially for underground crops like peanuts, identifying such inhibitory metabolites is essential. Metabolomics enables precise characterization of these compounds, providing valuable data for root development regulation and soil quality assessment.

Paclobutrazol, phthalic acid, syringic acid, and myristic acid represent chemically diverse compounds with distinct biological relevance, yet all showed inhibitory effects on peanut root growth under our in vitro assay conditions. Paclobutrazol is a widely used plant growth regulator that modulates gibberellin biosynthesis and is applied in field crop production at concentrations higher than the 0.5 g/L used in this study (≈ 0.0017 mol/L). At this lower concentration, paclobutrazol did not exert a pronounced inhibitory effect on root development, consistent with its reported threshold-dependent activity in regulating root-shoot balance rather than inducing acute toxicity [15, 40, 47, 50].

In contrast, myristic acid, a saturated fatty acid naturally present in plants, animals, and fungi, displayed significant repression of root elongation at 0.5 g/L. Myristic acid is generally regarded as a structural lipid component but also exhibits antimicrobial properties [28]. Although common in the plant kingdom, our data suggest that supra-physiological concentrations can negatively impact root growth, a phenomenon rarely reported previously. Similarly, syringic acid and formononetin phenolic metabolites commonly associated with lignin biosynthesis and secondary metabolism also suppressed peanut root elongation under the same conditions. Phenolic acids are known to interfere with auxin signaling and cell wall dynamics when present at elevated levels [2, 22], supporting our observation that over-accumulation of these compounds in vitro may repress root growth, even though they are ubiquitous in natural soils and plant tissues.

Phthalic acid and its derivatives (phthalates) deserve special attention, as they are recognized anthropogenic pollutants in agricultural soils arising from plastic debris, fertilizers, and industrial inputs [1, 19]. Phthalates can induce phytotoxicity by inhibiting seed germination, root elongation, and altering antioxidant metabolism [14, 25, 44]. Plants typically detoxify these xenobiotics via conjugation processes such as esterification or glycosylation, reducing their toxicity and facilitating vacuolar sequestration [9, 13, 57]. In our study, we applied analytical grade phthalic acid in a controlled system to evaluate its direct biological effects, independent of soil co-contaminants. While we did not assess its metabolic fate in peanut seedlings, our findings confirm a strong inhibitory effect on root elongation, consistent with its classification as an environmental contaminant. Future studies incorporating targeted metabolomics and transcriptomics will be essential to clarify how peanut roots metabolize phthalic acid in vivo and whether detoxification pathways, such as esterification, confer tolerance.

Together, these results highlight that both endogenous plant metabolites (e.g., phenolic acids, fatty acids) and anthropogenic soil contaminants (e.g., phthalates) can influence root development in concentration-dependent manners. Moderate levels of such metabolites may be compatible with, or even beneficial for, plant growth, whereas excessive accumulation suppresses phytohormone-mediated developmental pathways. Additionally, microbial degradation profoundly shapes the chemical environment of soils [18, 38]. Integration of soil metabolomics with metagenomic sequencing could therefore provide critical insights into the biotransformation of these compounds by microbial communities and their ultimate impacts on crop root development.

Conclusion

This study provides a comprehensive characterization of soil metabolomes across five major peanut-growing regions in Guangdong Province and demonstrates a causal link between soil-specific metabolites and peanut root development. Untargeted LC–MS/MS profiling revealed strong regional heterogeneity, with lipids emerging as dominant and ecologically significant constituents. Functional validation identified both root-promoting metabolites (raffinose, vanillic acid, nicotinamide, carbendazim) and root-inhibitory compounds (phthalic acid, myristic acid, syringic acid, formononetin), underscoring the dual role of soil metabolites as stimulants or constraints in root system establishment. The phytotoxicity of phthalic acid highlights the agronomic relevance of anthropogenic pollutants, while the inhibitory effects of naturally occurring metabolites such as myristic and syringic acids demonstrate the concentration dependent nature of soil plant chemical interactions. These findings broaden our understanding of the biochemical determinants of peanut root morphogenesis and establish a foundation for metabolite informed soil management. Future work integrating soil metabolomics with metagenomics and transcriptomics will be essential to clarify microbial transformations, plant detoxification pathways, and the environmental fate of both beneficial and inhibitory compounds. Such integrative approaches may enable the development of predictive soil metabolic biomarkers for crop performance and guide targeted interventions to optimize root growth. From an applied perspective, metabolite informed strategies could support peanut farmers by identifying soils rich in growth promoting compounds, mitigating risks from allelopathic or pollutant derived metabolites, and designing amendments or microbial inoculants to rebalance soil chemistry. Ultimately, leveraging soil metabolite knowledge offers a promising route for precision agriculture, enabling improved peanut productivity and resilience while ensuring sustainable soil health.