Integrated multi-omics profiling reveals effects of exogenous plant growth regulators on tobacco rhizosphere under Vulpia myuros rotation

Abstract

Background

Soil degradation poses a significant threat to agricultural sustainability. The integration of plant growth regulators (PGRs) with cover cropping systems represents a promising strategy for enhancing soil health and plant resilience. However, the synergistic effects of PGRs and cover crops on soil microbial communities and plant metabolism remain poorly understood.

Results

Exogenous PGRs under Vulpia myuros rotation were linked to distinct restructuring of tobacco rhizosphere microbiomes and metabolic networks. High-throughput sequencing of 16 S rRNA and ITS regions revealed that NaHS was associated with enhanced fungal diversity, promotion of beneficial saprotrophic taxa, and activation of phenylpropanoid and lignin metabolism. VB₁-based treatments modulated bacterial communities, enriched cooperative Firmicutes–Proteobacteria clusters, and drove sphingolipid, coenzyme, and amino acid pathways, while multivitamins promoted synergistic microbial clusters and stimulated phenolic antioxidant and secondary metabolite production. Glutamate maintained overall microbial diversity but induced rare taxa, reshaping lipid and ceramide metabolism and activating neurotransmitter and amino acid pathways. Integrated metabolomics–microbiome analysis revealed treatment-specific microbial–metabolite interactions, including lipid remodeling, pollutant degradation, and phenolic activation, linking microbial shifts with metabolic reprogramming. Network analyses demonstrated that PGRs differentially modulate microbial co-occurrence, enhancing cooperation, centrality, or competitive partitioning, collectively optimizing rhizosphere ecological functions and plant stress adaptation.

Conclusions

Our findings demonstrate that exogenous PGRs differentially reshape rhizosphere microbial communities and metabolic networks, enhancing plant stress adaptation in a cover crop system. These results provide a mechanistic basis for optimizing PGR-cover crop synergies to improve soil health and crop resilience.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-026-08312-4.

Background

Soil degradation poses a major threat to the sustainability of agricultural systems, driving the urgent need for environmentally sound remediation strategies that enhance soil health while minimizing ecological disturbance [1]. Intensive agricultural practices, such as monoculture cropping, excessive tillage, and overuse of chemical fertilizers, have led to significant declines in soil organic matter, reduced biodiversity, and compromised soil structure, thereby diminishing soil fertility and resilience [2]. These changes not only impair plant growth but also exacerbate environmental issues like erosion, nutrient leaching, and greenhouse gas emissions [3]. Consequently, restoring soil health has become a critical focus for sustainable agriculture [4].

Within this context, plant growth regulators (PGRs) have emerged as promising tools for modulating plant physiological processes, including nutrient assimilation, stress tolerance, and root system architecture [5]. For instance, B-vitamin complexes (VB₁, VB₂, and VB₆) act as enzymatic cofactors that optimize carbohydrate metabolism and antioxidant defenses [6], whereas indole butyric acid (IBA) functions as an auxin analog to stimulate root proliferation [7]. Similarly, sodium hydrosulfide (NaHS) serves as a hydrogen sulfide (H₂S) donor, mitigating oxidative stress [8], and glutamate contributes to nitrogen metabolism and the biosynthesis of signaling molecules [9]. Although the individual benefits of these PGRs are well documented, their synergistic interactions with soil amendment practices, such as cover cropping, remain poorly understood, particularly regarding the coupled metabolic responses of soil microbiota and plants. Nevertheless, it remains unclear how PGRs interact with cover cropping systems to modulate soil microbial functions and, in turn, influence plant metabolic pathways and overall soil health.

In this study, we focused on the Vulpia myuros cover crop system, which has been recognized as a model for sustainable soil management [10]. Rotating tobacco with V. myuros, followed by tillage and incorporation of plant residues, has been shown to enhance soil organic matter, improve structure, and recruit beneficial microbial communities [11]. Building on this framework, we conducted a field experiment integrating distinct PGR treatments (T1: VB₁ + IBA + Zn²⁺; T2: NaHS; T3: VB₁ + VB₂ + VB₆; T4: glutamate) with the V. myuros rotation system [12]. We hypothesized that these PGR formulations would differentially modulate soil microbial communities and elicit specific metabolic responses in tobacco. By combining 16 S rRNA and ITS sequencing with non-targeted metabolomics, we aimed to link PGR-driven microbial shifts to plant metabolic reprogramming and assess the potential of PGR-cover crop synergies for improving soil health. This multi-omics approach provides mechanistic insights into PGR efficacy within soil-plant-microbe systems and informs sustainable strategies to enhance resilience in degraded agricultural landscapes.

Materials and methods

Plant materials, growing conditions and treatments

Field experiments were conducted in Xijiadian County, Danjiangkou City, Hubei Province, China (111.18354°E, 32.748336°N). The region experiences a northern subtropical monsoon climate, significantly moderated by the vast Danjiangou Reservoir. During the core tobacco growing season (April–August), the area maintains a mean monthly temperature of 23.6 °C and a relative humidity of 70.7%, with mild, often diffuse sunlight. The experimental site featured yellow-brown earth soil with a neutral pH (averaging 7.5), moderate organic matter content (8.4 g/kg), and was notably rich in available potassium (averaging 161.7 mg/kg). Tobacco cultivar Nicotiana tabacum cv. Yunyan 87 was provided by the Wuhan Tobacco Research Institute. Seeds of Vulpia myuros were obtained from Suqian Lvpeng Seed Industry Technology Company (Suqian, China). All plant collection and experiments complied with institutional, national, and international guidelines.

Rhizosphere soil was collected using a five-point sampling method. For each experimental group, six individual plants were selected to ensure representative sampling. Samples were transported on ice and stored at − 80 °C until analysis. In September 2021, V. myuros was sown (300 g per plot). The following March, fields were plowed and fertilized (1.0 kg compound fertilizer per row), and in June the dried biomass of V. myuros was incorporated into ridges to promote soil mounding. Annual rotation of tobacco with V. myuros was maintained throughout the experiment [13]. Five treatments were applied, each with 200 mL solution per plant: CK (control check, untreated group; water control); T1, 20 mg L⁻¹ VB₁ + 1 mg L⁻¹ indolebutyric acid + 10 mg L⁻¹ ZnSO₄ [14]; T2, 600 mg L⁻¹ NaHS [15]; T3, 20 mg L⁻¹ VB₁ + 20 mg L⁻¹ VB₂ + 20 mg L⁻¹ VB₆ [16]; T4, 60 mg L⁻¹ glutamic acid [17]. Target root exudation (VB₁/IBA), sulfur metabolism (NaHS), vitamin signaling, and amino acid flux (glutamic acid) were pathways probably linked to rhizosphere microbiome modulation in prior work [1]. t.T11.D1: t represents the green manure experiment, T1 represents the T1 treatment, 1 represents the first replicate sample of the T1 treatment, D1 represents the first batch of samples, and t.T11.D1, t.T12.D1, t.T13.D1 represent the three replicates of the first batch of samples treated with T1 in the experiment. CK.1; T1.1; T2.1; T3.1; T4.1: Represent the five treatments (each treatment has three replicates) of the first batch of experiments. CK.2; T1.2; T2.2; T3.2; T4.2: Representing the five treatments of the second batch of experiments (each treatment has three replicates). CK.3; T1.3; T2.3; T3.3; T4.3: Representing the five treatments of the third batch of experiments (each treatment has three replicates).

Untargeted metabolomics

Tobacco tissues (100 mg) were ground in liquid nitrogen and extracted with 80% methanol. The QC sample is prepared by mixing equal volumes of experimental samples, and is tested on the instrument before, during, and after the LC-MS/MS injection of the experimental samples. The first three QCs before sample injection are used to monitor the instrument status and balance the chromatography-mass spectrometry system before injection. The following three QCs are used for segmented scanning, and the secondary spectra obtained together with the experimental samples are used for metabolite identification. The QCs inserted in the middle of sample detection are used to evaluate the system stability throughout the entire experimental process and perform data quality control analysis. After centrifugation and dilution, supernatants were analyzed using a Vanquish UHPLC system coupled with an Orbitrap Q Exactive™ HF-X mass spectrometer (Thermo Fisher, Germany) [18]. Separation was performed on a Hypersil Gold column (100 × 2.1 mm, 1.9 μm) under both positive and negative ion modes. Raw data were processed in Compound Discoverer 3.1 (Thermo Fisher) for peak alignment, detection, and quantification, with annotation against mzCloud, mzVault, and MassList databases. Metabolites with CV > 30% in QC samples were excluded, and remaining peaks were normalized to total spectral intensity. Annotation was further supported by KEGG, HMDB, and LIPID Maps databases. Statistical analyses were performed using R (3.4.3), Python (2.7.6), and CentOS (6.6). Multivariate methods (PCA, PLS-DA in metaX [19]); and univariate t-tests were used to identify differential metabolites (VIP > 1, p < 0.05, fold change ≥ 2 or ≤ 0.5). Visualization included volcano plots (ggplot2), clustering heatmaps (pheatmap), and correlation matrices (corrplot). KEGG enrichment analysis was conducted to evaluate metabolic functions and pathways (p < 0.05).

Sequencing and data analysis

Total genomic DNA was extracted using the CTAB/SDS method, and quality was checked on 1% agarose gels before diluting to 1 ng/µL. Target regions of 16 S rRNA (V4/V3/V3-V4/V4-V5), 18 S rRNA (V4/V9), ITS (ITS1/ITS2), and Arc V4 were amplified with barcoded primers (e.g., 16 S V4: 515 F–806R). PCR was performed with Phusion^® High-Fidelity Master Mix (New England Biolabs) under standard cycling conditions. Amplicons were verified by 2% agarose gel electrophoresis, pooled in equal amounts, and purified using a Qiagen Gel Extraction Kit. Sequencing libraries were prepared with the NEBNext^® Ultra™ II DNA Library Prep Kit (E7645) and evaluated with Qubit 2.0 and Agilent Bioanalyzer 2100. Libraries were sequenced on an Illumina NovaSeq platform to produce 250 bp paired-end reads [20]. Raw reads were processed by trimming barcodes and primers, merging with FLASH (v1.2.11) [21], and quality-filtering with fastp (v0.20.0). Chimera removal was conducted against the Silva (16 S/18S) and Unite (ITS) databases using Vsearch (v2.15.0), yielding Effective Tags. Denoising with DADA2/deblur in QIIME2 (2020.6) generated Amplicon Sequence Variants (ASVs), with low-abundance ASVs (< 5 reads) removed. Taxonomic annotation was performed in QIIME2 (202202), Silva (138.1) for 16 S/18S and Unite (v9.0) for ITS. ASV tables were rarefied to the minimum sequencing depth, and community structure was evaluated by alpha diversity (Observed OTUs, Chao1, Shannon, Simpson, Good’s coverage, Pielou_e) and beta diversity (weighted/unweighted UniFrac) [22]. The absolute abundance of ASVs was normalized using a standard of sequence number corresponding to the sample with the least sequences. Subsequent analysis of alpha diversity and beta diversity were all performed based on the output normalized data. The basic statistical results of data quality control for 16 S and ITS amplicon sequencing are presented in Table S1. The primer sequences used for each region were listed in Supplementary Table S2.

Combined transcriptome and metabolome analysis

To integrate microbial community dynamics with host metabolic responses, we combined metabolomic profiling and metagenomic sequencing [23], followed by correlation analysis between differential metabolites and microbial taxa [24]. The Spearman correlation coefficient varies between − 1 and 1, with negative and positive values indicating inverse or direct associations, respectively, and values closer to ± 1 reflecting stronger relationships. Correlation analyses were performed in R using the cor function, and statistical significance (p-values) was assessed with the Student method in the WGCNA package.

Results

Effects of exogenous PGRs on rhizosphere bacterial and fungal diversity

To assess the impact of PGRs on microbial community structure, ANOSIM based on 16 S and ITS sequencing showed that all R values between the control check (CK) and PGR treatments were greater than 0 (Tables 1 and 2), indicating potential dissimilarity between groups. However, all P-values exceeded 0.05, suggesting that these differences were not statistically significant. For instance, the CK-T2 comparison had the highest R value (0.12071 for 16 S), yet the P-value was 0.08, underscoring the need for caution in interpretation. Despite the lack of significance, the R values and patterns (e.g., negative R in T2-T4 ITS comparisons) hint at potential directional trends in community structure under PGR treatments. Therefore, we refrain from claiming significant effects and instead note that these observations may warrant further investigation with larger sample sizes or complementary tests (e.g., PERMANOVA).

Table 1.

Anosim intergroup difference analysis of 16 S seq

Group	R-value	P-value
CK-T1	0.02092	0.273
CK-T2	0.12071	0.08
CK-T3	0.09962	0.112
CK-T4	0.06139	0.151
T1-T2	-0.07064	0.847
T1-T3	-0.04184	0.614
T1-T4	0.03841	0.205
T2-T3	-0.04612	0.648
T2-T4	0.03841	0.204
T3-T4	0.01749	0.285

Table 2.

Anosim intergroup difference analysis of ITS seq

Group	R-value	P-value
CK-T1	0.02709	0.267
CK-T2	0.11403	0.095
CK-T3	0.03824	0.186
CK-T4	0.05676	0.166
T1-T2	0.0751	0.159
T1-T3	0.08196	0.132
T1-T4	0.0415	0.223
T2-T3	0.11883	0.105
T2-T4	-0.00857	0.389
T3-T4	0.09088	0.122

To further explore which taxa were affected, LEfSe analysis identified ammonia-oxidizing archaea and bacteria, including f-Nitrosotaleaceae, o-Nitrosotaleales, and c-Nitrososphaeria, as biomarkers in CK (Fig. 1). No significant bacterial biomarkers were detected in T1–T4, indicating that exogenous PGRs suppressed nitrifying guilds. ITS-based LEfSe revealed treatment-specific fungal biomarkers: T1 enriched c_Agaricomycetes, T2 enriched f_Phaeosphaeriaceae, o_Pleosporales, and c_Dothideomycetes. T3 enriched f_Mortierellaceae, o_Mortierellales, while T4 showed no significant biomarkers. These results collectively highlight selective promotion of beneficial or antagonistic fungi by different PGRs.

Fig. 1.

LEfSe analysis of microbial biomarkers under different fertilization regimes. A LDA scores of bacterial biomarkers enriched in T1 (red), T2 (green), and T3 (blue). B Fungal biomarkers showing CK-specific enrichment. C Cladogram (phylum to family level) highlighting taxa associated with T1 (red), T2 (green), and T3 (blue); node size indicates relative abundance. D Phylogenetic clustering of CK-linked bacterial lineages (red gradient) and rare taxa (gray). Solid and dashed lines represent mean and median relative abundance, respectively; absent bars indicate undetectable biomarkers (LDA > 2, Kruskal-Wallis p < 0.05). T1, VB1 + IBA + ZnSO₄; T2, NaHS; T3, VB₁ + VB₂+ VB₆; T4, glutamate

In addition, alpha diversity analyses confirmed these differential responses, revealing distinct patterns among bacterial and fungal communities (Fig. 2, Table S3). For bacteria, Shannon indices decreased under T1 (8.65) versus CK (8.95), were maintained under T2 (8.92), and increased under T3 (9.15) and T4 (9.25) (Fig. 2A). Chao1 indices showed similar trends (Figurer 2A). For fungi, Shannon indices decreased under T1 (4.5) and T3 (4.2), increased under T2 (5.8), and remained stable under T4 (5.0) (Fig. 2B). Chao1 indicated increases in T2 (600) and decreases in T3 (500), with T1 and T4 largely unchanged (Fig. 2B). Overall, these patterns suggest NaHS enhances fungal diversity and heterogeneity, whereas VB₁+IBA + Zn²⁺ and multivitamin combinations can suppress diversity.

Fig. 2.

Alpha diversity indices of bacterial and fungal communities across treatments. A Shannon and Chao1 of bacterial based on 16 S rRNA gene sequencing. B Shannon and Chao1 of bacterial based on ITS sequencing. T1, VB1 + IBA + ZnSO₄; T2, NaHS; T3, VB₁ + VB₂ + VB₆; T4, glutamate.

To characterize shifts in microbial composition, taxonomic profiling revealed pronounced microbial shifts (Fig. 3, Table S4). CK bacterial communities were dominated by Acidobacteriota, Proteobacteria, and Chloroflexi. T1 increased Firmicutes and reduced Acidobacteriota; T2 enriched Acidobacteriota and decreased Proteobacteria; T3 elevated Actinobacteriota and decreased Chloroflexi; T4 enhanced Proteobacteria and Bacteroidota. At the genus and species level, T1 promoted Blautia, T2 enriched Candidatus Nitrosotalea and Roseburia intestinalis, T3 increased Klebsiella and Dialister sp., while T4 enhanced Sphingomonas and Ruminococcus bicirculans. These observations indicate that exogenous PGRs selectively restructure bacterial communities to optimize ecological functions.

Fig. 3.

Relative abundance of bacterial and fungal communities at the phylum, genus, and species levels under PGR treatments. Relative abundance of bacterial (A–C) and fungal (D–F) communities at the phylum (A, D), genus (B, E), and species (C, F) levels across treatments (CK, T1, T2, T3, T4). Bar charts display the top 10 taxa at each taxonomic level, with remaining taxa grouped as “Others.” Colors represent different microbial taxa, the x-axis indicates sample names, and the y-axis shows relative abundance (%). Low-abundance taxa beyond the top 10 are summed and represented as “Others”. T1, VB1 + IBA + ZnSO₄; T2, NaHS; T3, VB1 + VB2 + VB6; T4, glutamate

Similarly, fungal communities exhibited distinct compositional shifts (Fig. 3, Table S3). At the phylum level, T1 increased Ascomycota but reduced Basidiomycota; T2 enriched Chytridiomycota and Mucoromycota; T3 elevated both Ascomycota and Mortierellomycota; T4 enhanced rare taxa such as Rozellomycota. At finer taxonomic resolution, T1 promoted antagonistic Clonostachys rosea while suppressing pathogenic Fusarium oxysporum; T2 enriched beneficial Trechispora; T3 increased saprophytic Saitozyma podzolica; and T4 induced rare taxa such as Humicola nigrescens and Leptobacillium. Collectively, these results demonstrate that PGRs under V. myuros rotation selectively modulate fungal communities, enhancing rhizosphere ecological function.

In conclusion, these findings indicate that distinct exogenous PGRs differentially shape bacterial and fungal diversity, richness, and community composition. NaHS was primarily associated with enhanced fungal complexity and beneficial taxa, vitamin-auxin combinations favor specific bacterial groups but may suppress fungal diversity, multivitamins promote saprophytic Basidiomycota, and glutamate preserves baseline diversity while inducing rare taxa. Together, this selective modulation of rhizosphere microecology under V. myuros rotation suggests that tailored PGR applications can optimize soil functional potential and plant-microbe interactions.

Exogenous PGRs differentially reprogram tobacco rhizosphere metabolism

Principal component analysis (PCA) revealed distinct clustering patterns across PGR treatments (Figure S1), confirming treatment-specific metabolic reprogramming under V. myuros rotation. Principal component analysis of the non-targeted metabolomics data revealed distinct separation patterns between groups under both negative and positive ionization modes. In negative ion mode, the first two principal components accounted for 20.89% and 11.77% of the total variance, respectively. The CK group clustered separately from all treatment groups, with T3 exhibiting the greatest separation along PC1, indicating substantial metabolic alterations induced by the combined vitamin B treatment. T1 and T4 groups showed moderate separation from CK, while T2 demonstrated minimal displacement. Conversely, positive ion mode analysis explained 17.11% and 11.62% of variance through PC1 and PC2, respectively, with QC samples tightly clustered confirming data quality. Here, T2 and T4 displayed pronounced separation from CK along PC1, suggesting NaHS and glutamate treatments induced distinct metabolic profiles detectable primarily in positive mode. T3 maintained significant separation, though less pronounced than in negative mode, highlighting the complementary nature of both ionization techniques. Overall, each treatment elicited specific metabolic responses distinguishable from control, with ionization mode influencing the detection of these variations.

Based on KEGG pathway annotation analysis, the functional distribution of metabolites showed significant differences between positive and negative ion modes (Fig. 4, Table S5). The Metabolism category was the most significantly enriched, with lipid metabolism (82 in negative ion mode, 22 in positive ion mode) and amino acid metabolism (27 in negative ion mode, 32 in positive ion mode) being particularly prominent. Global overview maps (75 in positive ion mode), carbohydrate metabolism, and energy metabolism were also well represented. In contrast, non-metabolic categories such as environmental information processing, genetic information processing, and cellular processes contained very few metabolites (only 1–3 each). Specific metabolites associated with drug development, including those related to nuclear receptors and structure-based classification, were also identified. These results indicate that the biological system under study is in an active metabolic state, prioritizing biosynthesis and energy generation, while non-metabolic functions are likely regulated at the transcriptional, translational, or protein activity levels. The presence of drug-related metabolites suggests potential bioactivity or therapeutic value.

Fig. 4.

Annotation of metabolites in KEGG pathways under negative and positive ion modes. A Negative ion mode metabolites. B Positive ion mode metabolites. x-axis, the number of metabolites. y-axis, KEGG pathways at the secondary classification level. Each bar corresponds to a specific pathway, with bar height indicating the number of metabolites annotated

PGRs differentially restructure bacterial and fungal co-occurrence networks

To further investigate how PGRs influence microbial interactions, microbial co-occurrence network analysis revealed treatment-specific topological restructuring (Fig. 5, Table S6). For bacteria, the CK network was relatively simple, centered on Proteobacteria and Acidobacteriota, with sparse positive correlations and an average node degree of 2.1. T1 increased network complexity, forming dense Firmicutes–Actinobacteriota clusters with strong positive correlations and an average node degree of 3.5, indicating enhanced cooperative interactions. T2 featured a single-core network dominated by Acidobacteriota, with intensified positive correlations to Chloroflexi but increased negative edges, suggesting strengthened centrality accompanied by heightened competition. T3 exhibited the largest network scale, with Proteobacteria and Bacteroidota forming multi-modular structures, reflecting expanded synergistic interactions. T4 induced topological restructuring, showing strong negative correlations between Nitrospirota and Crenarchaeota and maximal node degree heterogeneity, indicating competitive niche partitioning.

Fig. 5.

Microbial interaction networks based on Spearman correlations for 16 S and ITS sequencing across treatments. A 16 S bacterial communities. B ITS fungal communities. Nodes represent genera (size indicates average relative abundance; color denotes phylum), and edges indicate interactions (thickness shows absolute correlation; red, positive; blue, negative). Correlations were filtered to include only coefficients ≥ 0.6, exclude self-connections, and remove nodes with abundance < 0.005%

Similarly, fungal co-occurrence networks displayed distinct structural patterns (Fig. 5, Table S6). CK displayed a dense network dominated by Ascomycota and Basidiomycota, with prevalent positive correlations, an average node degree of 3.2, and a single negative correlation between Mortierellomycota and Kickxellomycota. T1 reduced network scale, decreased Ascomycota abundance and positive interactions, lowered the average node degree to 2.5, and introduced novel negative correlations between Blastocladiomycota and Chytridiomycota, suggesting weakened fungal cooperation. T2 intensified Ascomycota dominance with thicker positive edges and elevated average node degree to 3.8, while eliminating Mortierellomycota nodes, reflecting strengthened centrality but reduced functional diversity. T3 induced topological restructuring, forming synergistic clusters between Glomeromycota and Mucoromycota, with negative correlations increasing to 20%, indicative of novel cooperative interactions. T4 exhibited maximal node heterogeneity, featuring strong negative correlations between Basidiobolomycota and Chytridiomycota and an uneven node degree distribution (1.5–4.0), reflecting competitive partitioning.

To sum up, these results demonstrate that exogenous PGRs under V. myuros rotation selectively reshape microbial interaction networks: T1 promotes bacterial cooperation and weakens fungal Ascomycota interactions; T2 strengthens bacterial and fungal centrality but reduces fungal diversity; T3 expands bacterial–fungal synergistic clusters; and T4 induces competitive niche partitioning. These findings highlight the differential modulation of ecological functionality by distinct PGR treatments.

LIPID MAPS-based metabolomics reveal distinct lipidomic alterations under PGR treatments

To investigate the functional roles of lipids in rhizosphere ecology, lipidomic annotation against the LIPID MAPS database revealed key ecological mediators under negative ion mode (Fig. 6, Table S7). The significant detection of fatty acids and conjugates, including fatty acids and conjugates, eicosanoids, and fatty esters, points to their role in energy transfer and as precursors for signaling molecules that regulate microbial behavior and plant-microbe interactions. Glycerophospholipids such as Glycerophosphoglycerols, Glycerophosphoethanolamines, and Glycerophosphocholines were prominently observed, indicating their critical function in maintaining membrane integrity under fluctuating soil conditions and supporting biofilm formation and microbial colonization. The presence of polyketides like flavonoids and linear tetracyclines, along with Isoprenoids, suggests active secondary metabolism tied to chemical defense and interspecies communication. Additionally, sterol lipids including steroids, steroid conjugates, and bile acids and derivatives imply eukaryotic involvement in modulating ecological niche adaptation and stress response.

Fig. 6.

LIPID MAPS annotation of lipid metabolites under negative and positive ion modes. A Negative ion mode. B Positive ion mode. x-axis, the number of metabolites. y-axis, lipid classes at the main level. Bar length reflects metabolite counts for each class

Complementary insights were obtained under positive ion mode, where lipid profiles further expanded the ecological functional landscape. Fatty acids and related compounds remain highly prevalent, with subclasses such as fatty alcohols and fatty amides indicating additional dimensions of lipid signaling and microbial antagonism. The strong representation of Glycerophosphocholines emphasizes their role in membrane dynamics under biotic and abiotic stresses. Consistent detection of flavonoids and isoprenoids reinforces their ecological relevance as antimicrobial or antioxidant agents mediating soil health and plant defense. Moreover, the broad coverage of sterol lipids, including starols, steroids, steroid conjugates, and bile acids and derivatives, highlights their contribution to eukaryotic membrane stability, hormonal signaling, and community structuring in heterogeneous soil environments.

Altogether, the lipidomic profiles from both ionization modes reveal a sophisticated and interconnected ecological network within the soil microenvironment. The co-occurrence of diverse lipid classes reflects multifaceted mechanisms underpinning energy flux, chemical defense, structural adaptation, and cross-kingdom interactions. This integrated lipid landscape underscores the functional resilience and ecological complexity of soil communities, driven by coordinated metabolic strategies that sustain microbial and plant fitness in dynamic terrestrial ecosystems.

Exogenous PGRs are associated with treatment-specific KEGG pathway reprogramming

To elucidate how exogenous PGRs influence rhizosphere metabolism, untargeted metabolomics KEGG enrichment revealed distinct, treatment-specific metabolic reprogramming under complementary negative and positive ionization modes (Fig. 7, Table S8). For T1, negative ion mode highlighted activation of ferroptosis and moderate enrichment of terpenoid backbone biosynthesis and steroid hormone biosynthesis, whereas positive ion mode emphasized arginine/proline metabolism and flavonoid/zeatin biosynthesis (Fig. 7A). Together, these alterations coordinate lipid peroxidation defense, amino acid reprogramming, and secondary metabolism. For T2, negative ion mode enriched secondary bile acid biosynthesis and phenylpropanoid biosynthesis, while positive ion mode highlighted tryptophan metabolism and the sulfur relay system, indicating dual regulation of lipid remodeling, phenolic accumulation, and sulfur-dependent amino acid metabolism to enhance stress resilience (Fig. 7B). For T3, negative ion mode broadly suppressed lipid-associated pathways, whereas positive ion mode strongly activated phenylpropanoid, phenylalanine, and shikimate-derived alkaloid biosynthesis, reflecting VB₁/₂/₆-driven secondary metabolite induction (Fig. 7C). For T4, negative ion mode enhanced fatty acid biosynthesis, steroid biosynthesis, and antibiotic pathways, while positive ion mode enriched tryptophan metabolism, serotonergic synapse, plant hormone biosynthesis, and sulfur amino acid metabolism, demonstrating coordinated lipid remodeling and provision of neurotransmitter/hormone precursors (Fig. 7D). Overall, these results reveal that exogenous PGRs induce pathway-specific metabolic reprogramming: T1 reinforces ferroptosis and secondary metabolism; T2 integrates bile acid, phenylpropanoid, tryptophan, and sulfur metabolism; T3 suppresses lipid pathways while promoting phenolics and alkaloids; and T4 orchestrates lipid and nitrogen/sulfur metabolism for optimized stress adaptation.

Fig. 7.

Top 20 KEGG pathways enriched with differential metabolites. A T1 vs. CK; (B) T2 vs. CK; (C) T3 vs. CK; (D) T4 vs. CK. x-axis, the ratio of differential metabolites to total metabolites in each pathway. y-axis, pathway terms. Point color indicates the P-value of the hypergeometric test (smaller values indicate higher significance), and point size represents the number of differential metabolites. T1, VB₁ + IBA + Zn²⁺; T2, NaHS; T3, VB₁ + VB₂ + VB₆; T4, glutamate

Metabolic clustering highlights PGR-specific lipid and secondary metabolite patterns

To explore treatment-specific metabolic reprogramming, hierarchical clustering revealed distinct PGR-dependent metabolic patterns (Figure S2, Table S9). In negative ion mode, CK and T4 clustered closely with low-expression metabolites, whereas T2 formed an independent branch with strong red signals indicating NaHS-associated glycerophospholipid and sterol accumulation. T1 positioned between CK and T2, marked by localized sphingolipid enrichment, while T3 exhibited long branches and mottled patterns, reflecting heterogeneous lipid reprogramming under multivitamins (Fig. S2A). In positive ion mode, T4 separated as a distinct cluster enriched in ceramides, T2 partially overlapped with CK but showed flavonoid hotspots, T1 displayed amino acid–derived metabolites, and T3 concentrated in flavonoid and phenolic regions (Fig. S2B). Taken together, these patterns demonstrate that: VB₁+Zn²⁺ remodels sphingolipid and amino acid metabolism, NaHS drives lipid and secondary metabolite homeostasis, multivitamins enhance phenolic synthesis, and glutamate reprograms networks through ceramide-centered lipid signaling.

PCA and FunGuild analyses reveal treatment-specific microbial restructuring

To clarify microbial-metabolite interactions under different treatments, we analyzed 16 S rRNA and ITS sequencing data alongside metabolomics. For bacterial communities (16 S), CK showed distinct clustering in PC1 negative region, while VB₁ + Zn + auxin treatment (T1) shifted Proteobacteria enrichment to PC1 positive. NaHS (T2) induced Acidobacteriota perturbation in PC1 negative, multivitamins (T3) caused Actinobacteriota heterogeneity at maximal PC1 dispersion, and glutamate (T4) stimulated Proteobacteria proliferation at PC1 positive. Fungal communities (ITS) demonstrated PC2 migration: VB₁ + Zn (T1) suppressed Ascomycota while promoting Basidiomycota, NaHS (T2) stabilized Ascomycota, multivitamins (T3) activated ectomycorrhizal symbiosis, and glutamate (T4) specifically enhanced saprophytic fungi like Leptobacillium (Fig. 8, Table S10).

Fig. 8.

PCA of predicted functional profiles from 16 S (A) and ITS (B) sequencing. Points represent samples, colored by group (T1: VB₁ + IBA + ZnSO₄; T2: NaHS; T3: VB₁ + VB₂ + VB₆; T4: glutamate). PC1 and PC2 indicate the first and second principal components, with percentages showing their contribution to sample variation. Closer points reflect more similar functional composition

Key functional shifts emerged through microbial-metabolite correlations. T1 exhibited Sphingomonas-lysophosphatidylcholine association and suppressed Pseudomonas pollutant metabolism. T2 showed Halolactibacillus-phenyllactic acid interaction and stabilized nitrifier-polyphenol synergy. T3 revealed multivitamin-driven amino acid remodeling and fungal phenolic defense activation. T4 highlighted glutamate-promoted aromatic hydrocarbon degradation and antioxidant synthesis. These patterns demonstrate VB₁ + Zn enhances rhizosphere balance through bacterial-lipid interactions and fungal suppression, NaHS maintains heterogeneity with bacterial perturbation and fungal stabilization, multivitamins strengthen mycorrhizal symbiosis and pathogen defense, while glutamate improves stress adaptation via pollutant degradation and antioxidant activation (Fig. 9, Table S10).

Fig. 9.

Functional prediction of fungal communities based on ITS sequencing using FUNGuild. Heatmaps show the top 35 functions by abundance across experimental groups. A Level 1 database results. B Mode database results. Values are Z-scores calculated from standardized relative abundances. T1, VB₁ + IBA + ZnSO₄; T2, NaHS; T3, VB₁ + VB₂ + VB₆; T4, glutamate

Metabolomics-microbiome correlations reveal treatment-specific microbial-metabolite networks

To further dissect the functional links between microbial shifts and metabolic reprogramming, untargeted metabolomics coupled with microbiome analysis revealed distinct microbial-metabolite correlations between the CK and T1-T4 treatments (Table S11). For T1, 16 S showed a strong negative correlation between Subgroup_2 and the antioxidant N-acetylcysteine, and Sphingomonas positively correlated with lysophosphatidylcholine LPC 18:1, relationships not observed in CK; ITS showed Verticillium shifted from positive to strong negative correlation with phenyl derivatives (3-(methoxycarbonyl)amino-2,2,3-trimethylbutanoic acid), Talaromyces newly correlated positively with glyoxylate, and Aspergillus showed intensified negative correlation with DL-myelin esters, with 24 new or enhanced thin-ellipse correlations in T1 versus CK (Figure S3A, Figure S4A). For T2, 16 S exhibited a strong positive correlation between Halolactibacillus and phenyllactic acid, Candidatus Nitrosotalea shifted from negative to strong positive correlation with 4-hydroxyphenylacetic acid, and Candidatus Solibacter showed enhanced negative correlation with sucrose, with 19 new significant correlations in T2 versus CK; ITS showed Cladosporium newly correlated positively with phenylpropanoid derivatives and intensified positive correlations with lignin monomers, Xanthophyllomyces showed significant negative correlation with bile acids (glycodeoxycholic acid), with no significant associations in CK (Figure S3B, Figure S4B). For T3, 16 S showed Subgroup_2 positively correlated with dimethylglycine (absent in CK), Candidatus Solibacter shifted from weak positive to strong negative correlation with branched-chain fatty acid FAHFA (2:0/18:1), Halolactibacillus lost positive correlation with scopoletin (observed in CK) and newly correlated negatively with royal jelly acid, and ITS showed Verticillium reversed from positive to strong negative correlation with deoxycholic acid, Talaromyces newly correlated positively with scopoletin, Cladosporium intensified positive correlation with resveratrol, with 22 new significant correlations in T3 versus CK (Figure S3C, Figure S4C). For T4, 16 S showed Candidatus Solibacter positively correlated with polycyclic aromatic hydrocarbon derivative 2-(1-benzothiophen-3-ylmethylene)-3,4-dihydronaphthalen-1(2 H)-one, Halolactibacillus shifted from weak positive to strong negative correlation with stearic acid, Subgroup_2 newly showed negative correlation with antioxidant N-formylkynurenine, and ITS showed Cladosporium displayed extremely strong positive correlation with polyphenol resveratrol and intensified positive correlation with lignin monomer 2,5-dihydroxybenzaldehyde, Paecilomyces showed significant negative correlation with bile acid glycochenodeoxycholic acid-2-cyanoacrylate, with 19 new significant correlations in T4 versus CK (Figure S3D, Figure S4D). These results validate that VB₁ + Zn under V. myuros rotation employs dual mechanisms: bacterially activating Sphingomonas-lysophospholipid interactions while suppressing Pseudomonas pollutant metabolism, and fungally inhibiting Verticillium secondary metabolism while activating Talaromyces oxalate metabolism, synergistically optimizing rhizosphere microecological balance; NaHS under V. myuros rotation remodels microecological networks through dual mechanisms: bacterially activating Halolactibacillus-phenolic acid interactions for organic acid accumulation and nitrifier-polyphenol synergy for nitrogen utilization, and fungally driving Cladosporium-mediated lignin degradation via phenylpropanoid pathways while inhibiting sterol metabolism, synergistically enhancing soil organic matter transformation efficiency and plant stress resilience; multivitamins optimize microecology under V. myuros rotation through dual mechanisms: bacterially restructuring amino acid metabolism via Subgroup_2-dimethylglycine synergy while inhibiting cellulolytic-lipid interactions, and fungally reversing Verticillium sterol metabolism and synergistically activating Talaromyces/Cladosporium-mediated phenolic synthesis such as scopoletin/resveratrol correlations, collectively enhancing plant stress resilience; glutamate under V. myuros rotation employs dual mechanisms: bacterially promoting Candidatus Solibacter-aromatic hydrocarbon synergy for accelerated pollutant degradation while blocking Halolactibacillus-mediated fatty acid accumulation, and fungally enhancing Cladosporium-driven antioxidant synthesis and suppressing saprophytic sterol metabolism, collectively improving plant stress adaptation.

Discussion

In this study, the integration of multi-omics approaches provides a comprehensive perspective on how exogenous PGRs reshape the tobacco rhizosphere under a V. myuros cover crop system. Our results demonstrate that distinct PGR formulations induce specific structural and functional responses in both microbial communities and plant metabolism, highlighting their potential as targeted soil amendment strategies.

Each PGR treatment was correlated with distinct community assembly, suggesting it may act as a selective force. The most pronounced shift occurred under NaHS (T2), which enriched Acidobacteriota and the beneficial fungal phylum Mortierellomycota. Acidobacteria are typically associated with oligotrophic conditions and play key roles in soil organic matter degradation and acid stress tolerance [25], while Mortierella species contribute to phosphorus solubilization, pathogen suppression, and the production of hormonelike compounds and polyunsaturated fatty acids that enhance plant stress resilience [26]. In contrast, the VB₁+IBA + Zn²⁺ treatment favored Firmicutes, including copiotrophic and potentially probiotic members such as Blautia, which can enhance host nutrient uptake [27]. Notably, all PGR treatments suppressed ammonia-oxidizing archaea and bacteria, suggesting that while PGRs promote plant-beneficial taxa, they may temporarily decouple nitrification processes, highlighting a trade-off that should be considered in soil management strategies [28].

Untargeted metabolomics revealed sophisticated metabolic reprogramming underpinning plant adaptation. NaHS was associated with activation of both the sulfur relay system and eicosanoid metabolism, confirming H₂S as a signaling molecule that primes defense responses beyond mere oxidative stress protection [29]. Glutamate specifically enriched fatty acid amides and tryptophan/serotonin pathways, indicating an intersection between microbial modulation and neural-like signaling in plants, which likely influences root architecture and stress perception [30]. The multivitamin treatment (T3) suppressed general lipid metabolism while activating phenylpropanoid and flavonoid biosynthesis, reflecting a “defense-over-growth” strategy where carbon resources are diverted to antioxidants and antimicrobial compounds [31]. Coordinated shifts in microbial taxa (e.g., Cladosporium and Talaromyces) and specific metabolites (e.g., resveratrol and scopoletin) suggest a co-metabolic interaction between plants and their rhizosphere microbiome, mediated by PGRs.

Network analyses further illuminated the ecological impacts of PGRs. The T1 bacterial network exhibited increased complexity and stronger positive correlations, indicative of a more cooperative and potentially resilient microbial community [32]. Conversely, NaHS enhanced Acidobacteriota centrality but increased negative correlations, suggesting intensified competition and niche partitioning. Fungal networks under T3 and T4 formed novel synergistic clusters (e.g., Glomeromycota-Mucoromycota) and more negative correlations, reflecting reorganization of community interactions. Functional profiling with FunGuild revealed that PGRs diversified fungal trophic modes, increasing pathotroph and symbiotroph representation. In particular, T3 and T4 enhanced mycorrhizal guilds, potentially improving nutrient acquisition and systemic resistance [33]. The V. myuros cover crop established a foundational soil microbiome and organic matter base. We hypothesize that without this cover crop, the soil would be less receptive, and PGR effects would be weaker or more variable. The PGRs then acted as targeted modulators upon this preconditioned system, fine-tuning the community. This synergy suggests cover crops may be a prerequisite for consistent PGR benefits in degraded soils [34].

Beyond mechanistic insights, the practical application and scalability of these PGR treatments are critical for agricultural implementation. The application rates used in this study are feasible for field-scale operations, as they align with standard foliar or soil drench practices in precision agriculture [35]. The vitamin-based and glutamate treatments similarly involve low-cost, readily available compounds. The integration with the V. myuros cover crop system enhances feasibility, as the cover crop itself contributes to soil health, potentially reducing the long-term dependency and required frequency of PGR applications. Future work should focus on large-scale, multi-season trials to validate the economic viability and optimize application schedules for different soil types and cropping systems.”

It is important to note that this study provides a snapshot of the rhizosphere system at a key developmental stage under PGR treatments. While this design effectively captures treatment-specific differences at this critical point, it does not allow us to infer the temporal dynamics of microbial succession or metabolic reprogramming. The community structures and metabolic profiles we report are likely part of an ongoing succession process. Future longitudinal studies tracking the rhizosphere weekly or biweekly from cover crop incorporation through tobacco maturation are needed to elucidate the trajectory of these PGR-induced changes and identify optimal intervention timings.

Conclusion

Our multi-omics analysis demonstrates that exogenous PGRs, in combination with a V. myuros cover crop, selectively shape rhizosphere microbial communities and plant metabolic pathways. This mechanistic understanding underscores the potential of integrating chemical and biological interventions for sustainable soil management and offers a framework for optimizing plant-microbe interactions to enhance resilience in degraded agricultural systems.

Data Availability

16 S-seq and ITS-seq raw data have been deposited in the NCBI sequence read archive (SRA) under project accession PRJNA1314933 and PRJNA1314939. The raw data for the metabolome of this study have been uploaded to MetaboLights (Number: MTBLS12916).

Authors’ contributions

Conceptualization, X.L. and P.L.; methodology, X.L.; investigation, X.L., Y.Q., Q.Z., X.C., Y.X., and P.L.; resources, C.Y., Y.Y., J.Y., and P.L.; writing-original draft preparation, X.L.; writing-review and editing, P.L.; supervision, P.L.; funding acquisition, C.L. and J.Y. All authors reviewed the manuscript.

Funding

This work was supported by the Key Science and Technology Project of China National Tobacco Corporation [grants 110202201040 (XJ-11) and 110202101009 (XJ-01)] and the Open Project of the School of Life Sciences at Hubei University (grant SMKXXB2026004). The funding bodies played no role in the design of the study, collection, analysis, interpretation of data, or writing of the manuscript.

Data availability

16 S-seq and ITS-seq raw data have been deposited in the NCBI sequence read archive (SRA) under project accession PRJNA1314933 and PRJNA1314939. The raw data for the metabolome of this study have been uploaded to MetaboLights (Number: MTBLS12916).

Declarations

Ethics approval and consent to participate

Not applicable.

Competing interests

The authors declare no competing interests.

Consent for publication

Not applicable.