Multi-omics reveals rhizosphere soil metabolismte-microbiota interactions in different varieties of sorghum under nutrient-deficient stress

1 Introduction

As the fifth-largest cereal crop worldwide, sorghum (Sorghum bicolor (L.) Moench) can persist across diverse soil environments, including arid, saline–alkali, and infertile soils, reflecting its strong adaptability and stress tolerance (1). Sorghum is an important minor grain crop in China. Beyond its application as a food crop, it is a promising raw material for brewing, animal husbandry, functional food production, and bioenergy (2).

Microbial communities that colonize plant roots in soil are regarded as the second genome of plants and play essential roles in plant growth and development (3, 4). These root-associated communities are bacterial, and their beneficial members can promote nutrient transformation, increase plant nutrient uptake, and enhance resistance to both abiotic and biotic stresses (5). Since the 1970s, nitrogen-fixing microorganisms from the sorghum rhizoplane have been isolated and cultivated using Biolog microplate assays and plate dilution methods. With the development of nucleic acid-based molecular techniques, such as denaturing gradient gel electrophoresis (DGGE), more plant growth-promoting bacteria with nitrogen-fixing capacity and biocontrol strains have been identified, including Rhizobium, Bradyrhizobium, and Paenibacillus, along with nifH gene sequences (6, 7). Following the broad application of 16S rDNA sequencing, sorghum rhizosphere bacterial communities have been reported to consist predominantly of Proteobacteria, Actinobacteria, Acidobacteria, Verrucomicrobia, Bacteroidetes, and Ascomycota (8). Their composition is shaped by cropping systems, including continuous cropping and crop rotation, and by fertilization practices. In continuous sorghum cropping soil, Acidobacteria and Gemmatimonadetes are strongly enriched (9). In soybean–sorghum and maize–sorghum rotation soils, Pseudomonadota predominated (10). Under nitrogen fertilizer application, Proteobacteria, Actinobacteria, and Bacteroidetes significantly decreased in sorghum rhizosphere soil (11). Furthermore, rhizosphere microbial diversity in sorghum has been suggested to be higher during the grain filling and ripening stages than during the heading and jointing stages. During reproductive growth, the sorghum rhizosphere microbiome has been shown to exert significant effects on grain yield and protein content (12, 13).

Soil metabolites originate principally from root exudates, soil microbial metabolism, and the metabolism of animals inhabiting soil ecosystems. As signaling molecules, root exudates can influence the soil microbiome through promotion, inhibition, and exclusion (14). Among sorghum root exudates, biological nitrification inhibitors (BNIs) (e.g., sorghumone) can suppress nitrification by inhibiting ammonia monooxygenase and hydroxylamine oxidoreductase activities in nitrobacteria (15, 16). Similarly, 3-(4-hydroxyphenyl) propionate has been shown to inhibit ammonia-oxidizing microorganisms, thereby reducing nitrogen loss and enhancing nitrogen fertilizer utilization (17, 18).

Recently, plant–soil–microbe interactions have increasingly been used as a strategy to achieve synergistic improvements in the productivity and ecological functions of saline–alkali and infertile lands. The rhizosphere microbial community composition is generated through complex interactions and feedback among plant roots, microorganisms, and the soil physicochemical environment (19). In sorghum-cultivated soils, distinct bacterial environments have been observed across different salt stress levels. Under high-salinity conditions, Proteobacteria and Bacteroidetes increased significantly, whereas root-derived phenolic compounds decreased, with these shifts being linked to rhizosphere bacterial diversity (20). In peanut–sorghum intercropping systems under salt stress, Shao et al. suggested that sorghum regulated the rhizosphere microbial structure through the soil sugar metabolism pathway. Soil metabolites (e.g., d-arabinose 2 and rhamnose) were significantly negatively correlated with Bacteroidota, Verrucomicrobia, and Gemmatimonadetes (21). Another study reported that sorghum root exudates significantly inhibited ammonia-oxidizing bacteria (AOB) (22). Sorgoleone, isolated from sorghum-associated soils, has been identified as the sole carbon and energy source for Acinetobacter, Burkholderia, and Pseudomonas (15). Despite the fact that there are many researches about soil microbial communities and metabolites of sorghum rhizosphere (15, 23), the natural variability of the microbiome and metabolites in the rhizosphere soil of sorghum varieties with different tolerance to nutrient-deficient stress is still unclear. To address this gap, this study adopted 16S rDNA and LC-MS to characterize the bacterial community structures and soil metabolite compositions in rhizosphere and bulk soils under nutrient-deficient stress. The analysis focused on the composition of microorganisms and metabolites in the tolerant sorghum rhizosphere under nutrient-deficient stress and the dynamic adjustment of rhizosphere metabolite components by sorghum to recruit specific microorganisms, optimize nutrient cycling, and enhance stress resistance. This study provides theoretical support for improving sorghum tolerance to nutrient-deficient conditions, developing microbial inoculants, and increasing the productivity of low-yield farmlands in China.

2 Materials and methods

2.1 Experimental site

The field experiment was conducted at the Dongbai Experimental Site (37°40′ N, 112°44′ E), Sorghum Research Institute, Shanxi Agricultural University. During the 2023 sorghum growing season, the site received 288.3 mm of precipitation and had a mean temperature of 20.05°C. The field was characterized as nutrient-poor land, with no fertilizer input for 11 years and continuous sorghum cultivation. The soil contained 4.08 g/kg soil organic matter (SOM), 37.03 g/kg alkali-hydrolyzed nitrogen (AN), 5.55 mg/kg available phosphorus (AP), and 110.78 mg/kg available potassium (AK).

2.2 Experimental design

The experimental materials were Jinnuo 101 and Jinnuo 102, two sorghum cultivars bred at the Sorghum Research Institute of Shanxi Agricultural University. The experiment was arranged with three biological replicates, two sorghum cultivars were used, resulting in a total of 6 plots. Each plot measured 2.5 m × 8 m (20 m²). In each plot, dry weight of the root were measured at the jointing stage (JS), heading stage (HS), anthesis stage (AS), and filling stage (FS), as well as superoxide dismutase (SOD) activity, catalase (CAT) activity, and MDA content.

Four treatments were established: rhizosphere soil of Jinnuo 101 (R1); non-rhizosphere soil of Jinnuo 101 (Non R1); rhizosphere soil of Jinnuo 102 (R2); non-rhizosphere soil of Jinnuo 102 (Non R2). During the filling stage of sorghum, five sorghum plants with uniform growth in each plot were selected using a five-point sampling scheme. From these plants, the rhizosphere soil and bulk soil were collected together from the 0–20 cm layer. Each whole plant was excavated with its root system, and large soil clods were removed by tapping the roots. Rhizosphere soil was collected by brushing the root-adherent soil, whereas non-rhizosphere soil was obtained through mild agitation of the roots. Five plant-derived samples were pooled and separated into three portions. All samples were placed in foam boxes with ice packs. After sieving through a 2 mm mesh, one portion was air-dried and used to determine the physicochemical properties, with three replicates. The remaining portions were stored at −80 °C and used for soil bacterial communities and metabolomic profiling analysis, with six replicates.

2.3 Analysis of root antioxidant enzyme activity, MDA content, and soil physicochemical properties

SOD (U/g FW), and CAT (U/g/min) were determined using ultraviolet absorption colorimetry, and nitroblue tetrazolium (NBT) photochemical reduction method, respectively. MDA content (nmol/g FW) was measured by the thiobarbituric acid (TBA) method.

Soil pH was measured using a pH meter (FE28, China) at a soil-to-water ratio of 1:2.5. SOM, AN, AP, and AK were quantified using dichromate oxidation, potassium persulfate oxidation, molybdenum blue colorimetry, and flame photometry, respectively. Moreover, total nitrogen (TN), total phosphorus (TP), and total potassium (TK) were measured separately using Kjeldahl digestion and distillation, HClO₄–H₂SO₄ acid digestion, and alkaline fusion.

2.4 Soil DNA extraction and sequencing

For each of the six biological replicate soil samples, total DNA was extracted from 1 g of soil using the CTAB method (detailed information in Supplementary Materials 1.1). The bacterial 16S rRNA gene V3–V4 hypervariable regions were amplified by PCR using the 341F/805R primer pair. Each 25 µL amplification reaction consisted of 12.5 µL Premix Taq (Phusion, New England Biolabs, USA), 2.5 µL of each primer, 50 ng template DNA, and PCR-grade water to volume. PCR products were purified using AMPure XT beads (Beckman Coulter Genomics, Danvers, MA, USA) and quantified using a Qubit fluorometer (Invitrogen, USA). The library size was assessed using an Agilent 2100 Bioanalyzer (Agilent, USA), and the library concentration was determined using a Library Quantification Kit for Illumina (Kapa Biosciences, Woburn, MA, USA). Sequencing was performed using a NovaSeq PE250 platform (Illumina, USA). The average amount of sequencing data per sample was 75985 reads, the average Q20 is 97.10%, and the average Q30 is 92.10%.

FLASH was used to merge paired-end reads, and fqtrim (v0.94) was used to quality-filter raw reads. After chimeric sequences were removed using Vsearch (v2.3.4), DADA2-based dereplication generated an amplicon sequence variant (ASV) table and representative sequences. Following normalization by random subsampling, α- and β-diversity indices were calculated using QIIME2. Feature abundances were normalized based on relative abundance using the SILVA classifier (release 138), and sequence alignment was performed against SILVA using BLAST. Graphs and additional analyses were produced using R packages (v3.5.2).

2.5 Metabolite extraction and LC-MS analysis

For metabolite extraction, the collected samples were thawed on ice, and 100 mg of each sample from six biological replicates was weighed and ground in liquid nitrogen. Metabolites were extracted with 120 µL of precooled 50% methanol. The extraction mixture was vortexed for 1 min, incubated for 10 min at room temperature, and stored overnight at -20 °C. After centrifugation at 4, 000 × g for 20 min, the supernatant was transferred into 96-well plates and stored at -80 °C until LC-MS analysis. A pooled quality control (QC) sample was generated by combining 10 μL of each extraction mixture.

Both ionization modes were used to analyze samples using a Triple TOF 5600 Plus high-resolution MS/MS system (SCIEX, UK). Chromatographic separation was achieved using an ultra-performance liquid chromatography (UPLC) system (SCIEX, UK) with an ACQUITY UPLC T3 column (100 mm × 2.1 mm, 1.8 µm; Waters) maintained at 35 °C. Water (A) and acetonitrile (B), each containing 0.1% formic acid, were used as mobile phases. At a flow rate of 0.4 mL/min, the gradient was set as follows: 5% B, 0–0.5 min; 5–100% B, 0.5–7 min; 100% B, 7–8 min; 100–5% B, 8–8.1 min; and 5% B, 8.1–10 min. Mass spectrometry data were acquired in IDA mode under the following source conditions: curtain gas, 30 psi; GS1/GS2, 60 psi; source temperature, 650 °C; and ion spray voltages, ± 5 kV/-4.5 kV. TOF MS scanning was performed over 60–1200 Da at 150 ms/scan with a 0.56 s cycle time.

LC-MS raw data were pre-processed using XCMS in the R environment, with metabolite annotation performed against the Kyoto Encyclopedia of Genes and Genomes (KEGG) and HMDB databases (mass tolerance 10 ppm) (detailed information in Supplementary Materials 1.2). Identifications were based on database matching with MS/MS spectra and retention time (confidence level 2). After log transformation and normalization, principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were applied to identify differentially expressed metabolites (DEMs) based on variable importance in projection (VIP) ≥ 1, p < 0.05, and |log₂(fold change)| ≥ 2. Correlations between DEMs and soil microorganisms were assessed by Spearman’s heatmap analysis. All analyses were conducted using the OmicStudio platform.

3 Results

3.1 Antioxidant enzyme activity and MDA content of sorghum root

Under nutrient-deficient stress, the dry root weight of Jinnuo 102 was significantly 50.4% higher than that of Jinnuo 101 at the grain-filling stage. The SOD activity of Jinnuo 102 reached the highest level at the anthesis stage, which was significantly higher than that of Jinnuo 101. The CAT activity of Jinnuo 102 showed significantly higher than Jinnuo 101 at the filling stage. The MDA content gradually increased with the growth stage, and was significantly lower than that of Jinnuo 101 (Table 1). Thus, the nutrient-deficient stress had different effects on the roots of the two varieties of sorghum. Jinnuo 102 exhibited a more obvious antioxidant enzyme activity.

Table 1. Root dry weight, antioxidant enzyme activities, and malondialdehyde content in roots of two sorghum under nutrient-deficient stress.

RDW, root dry weight; SOD, superoxide dismutase; CAT, catalase; MDA, malondialdehyde; JS, jointing stage; HS, heading stage; AS, anthesis stage; FS, filling stage. Values are presented as mean ± standard error (n = 3). Different lowercase letters within the same column indicate significant differences among treatments (P < 0.05).

3.2 Physicochemical properties of sorghum soil

Under low-nutrient stress, sorghum rhizosphere and non-rhizosphere soils were both weakly alkaline. AK and TK contents were moderate, whereas SOM, AN, AP, and TP remained at extremely low levels. As shown by the comparative analysis of soil physicochemical properties (Table 2), pH differed significantly among treatments (P < 0.05). In particular, Non R1 and Non R2 had significantly higher pH values than R1 and R2. For AP, R2 exhibited a significantly higher content than Non R2. Highly significant treatment differences were detected in AK and SOM contents (P < 0.01), and the highest SOM content occurred in R1. Overall, the rhizosphere soils of both sorghum cultivars compared with their corresponding non-rhizosphere soils were characterized by significantly lower pH and significantly higher AP, AK, and SOM contents, whereas the remaining parameters showed no significant differences.

Table 2. Physicochemical properties of rhizosphere and non-rhizosphere soils of sorghum under low-nutrient stress.

SOM, soil organic matter; AN, alkali-hydrolyzed nitrogen; AP, available phosphorus; AK, available potassium. R1, rhizosphere soil of Jinnuo 101; Non R1, non-rhizosphere soil of Jinnuo 101; R2, rhizosphere soil of Jinnuo 102; Non R2, non-rhizosphere soil of Jinnuo 102. Values (mean ± standard error, n = 6) are presented, with distinct lowercase letters in the same column signifying significant inter-treatment differences (P < 0.05).

3.3 Soil bacterial communities

3.3.1 Alpha diversity

The treatment-related patterns of the alpha diversity of soil bacterial communities under nutrient-limited stress are shown in Table 3. No significant differences were detected in the Shannon and Good’s coverage, indicating that the sequencing coverage was high, and community evenness was comparable across the four sample groups. In contrast, both the Observed species and Chao1 indices were significantly increased in sorghum rhizosphere soil than in non-rhizosphere soil. Additionally, these two indices were significantly higher in Jinnuo 102 rhizosphere soil than in Jinnuo 101 rhizosphere soil, indicating that Jinnuo 102 had the highest rhizosphere species diversity. Overall, species diversity was higher in sorghum rhizosphere soil than in non-rhizosphere soil, and cultivar-dependent variation was evident.

Table 3. Alpha diversity of bacterial communities in rhizosphere and non-rhizosphere soils of sorghum under low-nutrient stress.

R1, rhizosphere soil of Jinnuo 101; Non R1, non-rhizosphere soil of Jinnuo 101; R2, rhizosphere soil of Jinnuo 102; Non R2, non-rhizosphere soil of Jinnuo 102. Values (mean ± standard error, n = 6) are presented, with distinct lowercase letters in the same column signifying significant inter-treatment differences (P < 0.05).

3.3.2 Beta diversity

To visualize intergroup dissimilarities in the bacterial community structure, principal coordinate analysis (PCoA) was performed. PCoA1 and PCoA2 explained 17.43% and 6.67% of the variation, respectively, with a total explanatory power of 18.10% (R² = 0.230, P = 0.002) (Figure 1). This ordination pattern demonstrated that the bacterial community structures of sorghum rhizosphere soils were significantly separated from those of non-rhizosphere soils (Non R1 and Non R2). Differences in bacterial community structures were observed among the rhizosphere soils of different sorghum cultivars.

3.3.3 Analysis of bacterial community structure

At the phylum level, the bacterial community composition in the rhizosphere soils of different sorghum cultivars is shown in Figure 2. Across the different groups, the soil bacterial 16S rDNA sequences were classified into 40 phyla, 255 families, and 439 genera. Proteobacteria, Actinobacteria, Acidobacteria, Planctomycetes, and Chloroflexi were the dominant phyla in sorghum rhizosphere and bulk soils. Notably, Actinobacteria, Acidobacteria, and Bacillobacteria were significantly more abundant in sorghum rhizosphere soil than in bulk soil (Figures 2A, B). Within the rhizosphere, Jinnuo 102 showed a significantly higher abundance of Actinobacteria, whereas Jinnuo 101 showed a significant enrichment of Cyanobacteria (P < 0.05).

At the genus level, Ralstonia, Gemmataceae unclassified, Methyloversatilis, Sphingomonas, Vicinamibacteraceae unclassified, and Gemmatimonadaceae unclassified were the main genera detected across treatments. When dominant bacterial genera were defined as those with an average relative abundance >1%, 18 and 16 genera were detected in the rhizosphere soils of Jinnuo 102 and Jinnuo 101, respectively, whereas only 13 and 12 genera were detected in the corresponding non-rhizosphere soils. This pattern indicates that the bacterial genus-level diversity was higher in sorghum rhizosphere soil. In addition, Burkholderiales unclassified, Streptomyces, and Subgroup7 unclassified had significantly higher relative abundances among dominant rhizosphere bacterial genera than in non-rhizosphere soil (Figures 2C, D). Within rhizosphere soils, the two cultivars demonstrated distinct genus-level enrichment patterns. Jinnuo 101 was characterized by the significantly higher relative abundances of Burkholderiales unclassified, Chloroplast unclassified, and Ohtaekwangia, whereas Jinnuo 102 was characterized by the significantly higher relative abundances of Actinobacteriota unclassified, Solirubrobacter, and Nocardioides.

In addition to phylum- and genus-level differences in bacterial relative abundance, additional variation was detected at the family, class, order, and species levels. Across these taxonomic levels, bacterial enrichment was reflected by 9 biomarkers in Non R1, 11 biomarkers in R1, 6 biomarkers in R2, and 2 biomarkers in Non R2, all identified with LDA (log10) values >3.0. These biomarkers represented species that differed among treatments (Figures 3A, B).

3.4 Soil metabolome

A total of 518 metabolites were identified using HPLC-MS/MS (Supplementary Figure 1). Lipids and lipid-like molecules represented the largest metabolite class, with 234 compounds accounting for 45.17% of the total. The remaining metabolites included 77 organic heterocyclic compounds (14.86%), 51 benzoates (9.85%), 47 phenylpropane and polyketone compounds (9.07%), 47 organic acids and derivatives (9.07%), 26 organic oxygen compounds (5.02%), 12 organic nitrogen compounds (2.31%), and other metabolites (4.63%) (Figure 4A).

In the PCA, PC1 and PC2 explained 38.85% of the variation (Figure 4B). The QC samples confirmed the stability of the analytical method and data quality. Within each treatment, the samples clustered closely, whereas evident separation was observed among different treatments. The two non-rhizosphere soil groups exhibited little divergence in their metabolic profiles, whereas the other treatment groups were distinctly separated.

3.4.1 Screening of differentially expressed metabolites

The significantly regulated metabolites between groups were identified using VIP ≥ 1, p-value < 0.05, and |log2FC| ≥ 2. KEGG was used for metabolite annotation. Based on these criteria, 216 DEMs were detected in R1 vs. Non R1, comprising 158 upregulated and 58 downregulated metabolites (Supplementary Figure 2A); 252 DEMs were detected in R2 vs. Non R2, comprising 176 upregulated and 76 downregulated metabolites (Supplementary Figure 2B); and 79 DEMs were detected in R1 vs. R2, comprising 49 upregulated and 30 downregulated metabolites (Supplementary Figure 2C). DEM-associated pathways were ranked by ascending P-value, and the top 20 pathways were used to construct bubble plots. In both R1 vs. Non R1 and R2 vs. Non R2 comparisons (Figures 5A, B), DEMs were significantly enriched in “Biosynthesis of phenylpropanoids”, “Glycerophospholipid metabolism”, “Tryptophan metabolism”, “Linoleic acid metabolism”, and “ABC transporters”. Within “Biosynthesis of phenylpropanoids”, the enriched metabolites were pos-M269T246 (formononetin), pos-M359T216_2 (matairesinol), pos-M417T212 (daidzin), and neg-M459T200 (podophyllotoxin). Within “Tryptophan metabolism”, pos-M190T58 (kynurenic acid), pos-M157T242 (3-indoleacetonitrile), and pos-M164T235 (3-methyldioxyindole) were significantly enriched. Within “Glycerophospholipid metabolism”, pos-M190T399 (glycerol 3-phosphate), pos-M520T334 (LysoPC 18:2), and neg-M739T455 (PE 36:4) were significantly enriched. Within “ABC transporters”, neg-M341T57 (D-(+)-trehalose), pos-M190T399 (glycerol 3-phosphate), and pos-M175T58 (D-xylitol) were significantly enriched. These metabolites were detected in both R1 vs. Non R1 and R2 vs. Non R2 comparisons and were upregulated in their respective pathways. In contrast, the downregulated neg-M293T392 (9-oxoODE), neg-M313T286 (12, 13-DHOME), and pos-M337T289 (12, 13-dihydroxy-9Z-octadecenoic acid) were significantly enriched in “Linoleic acid metabolism” (Supplementary Tables 1, 2). In the R1 vs. R2 comparison (Figure 5C), DEMs were significantly enriched in “Biosynthesis of phenylpropanoids”, “Biosynthesis of plant secondary metabolites”, “Plant hormone signal transduction”, and “ABC transporters”. In R2, neg-M137T211 (salicylic acid) and pos-M179T292 (aesculetin) were significantly upregulated in the “Biosynthesis of phenylpropanoids” pathway (Supplementary Table 3).

3.4.2 Characteristic metabolites in sorghum rhizosphere soil

In sorghum rhizosphere soil, most significantly upregulated metabolites were classified as lipids, glycosides, or amino acids. Among metabolites with significantly high relative abundances, 12 potential plant-derived secondary metabolites were selected (Supplementary Table 4). These consisted of three terpenoids, five fatty amides or fatty acids, two phenolic acids, one carbohydrate, and one flavonoid. Their relative abundances differed highly significantly between R1 and Non R1 and between R2 and Non R2 (Figure 6). Compared with the other treatments, R1 contained higher levels of pos-M256T465 (palmitamide), and pos-M296T361 (punicic acid), whereas R2 contained higher level of neg-M137T211 (salicylic acid) (P < 0.05). The remaining metabolites exhibited no significant differences.

3.4.3 Correlation analysis of microbes and metabolites

To clarify crop root–environment interactions, the relationships between soil microorganisms and major metabolites were investigated. Therefore, correlation analyses were conducted among the top six dominant bacterial phyla, top eight dominant bacterial genera, and differentially abundant metabolites with high expression levels (Figure 7). Both significant (P < 0.05) and highly significant (P < 0.01) correlations were detected. Neg-M137T211 (salicylic acid) was highly significantly and positively correlated with Actinobacteriota, significantly and positively correlated with Acidobacteriota, and significantly and negatively correlated with Proteobacteria. Moreover, Pos-M163T54 (4-hydroxycinnamic acid) showed a highly significant positive correlation with Acidobacteriota and a significant positive correlation with Actinobacteriota. At the genus level, salicylic acid was highly significantly and negatively correlated with Ralstonia, whereas Burkholderiales unclassified and Streptomyces were highly significantly and positively correlated with multiple differentially abundant metabolites.

4 Discussion

4.1 The root phenotypes and antioxidant enzyme activity of two types of sorghum

Plant cells maintain a homeostatic balance between reactive oxygen species (ROS) synthesis and degradation under regular physiological conditions. However, under abiotic stress, the overproduction of ROS can perturb protein conformation and disrupt the cell membrane system, thereby adversely affecting plant growth and development. The antioxidant system reduces cellular damage by eliminating ROS accumulated in the plant body (24). MDA is the product of lipid peroxidation in plant cell membranes. Its content can reflect the extent of membrane lipid oxidation in plants and the strength of their response to adverse conditions. Nitrogen-efficient rice have higher SOD and CAT activity and lower MDA content than nitrogen inefficient under low-nitrogen treatment (25). Overall, the antioxidant enzyme activity of Jinnuo 102 was significantly higher compared to Jinnuo 102. During the entire growth period, Jinnuo 102 maintained a relatively stable and low oxidative damage state. The results demonstrated that the two types of sorghum exhibited different responses to poor soil conditions.

4.2 Microbial composition in the rhizosphere of sorghum under low-nutrient stress

Rhizosphere microbial diversity and community structure are closely linked to normal plant growth and development (26). Under low-nutrient stress, bacterial species diversity significantly increased in the sorghum rhizosphere. Across the rhizosphere soils of both sorghum cultivars, Actinobacteriota, Proteobacteria, and Acidobacteriota were the dominant bacterial phyla, consistent with previous reports (9, 20, 27). One possible explanation for this dominance is that sorghum roots release carbohydrates, fatty acids, and amino acids to chemotactically attract Actinobacteriota and Acidobacteriota into the rhizosphere (28, 29). These copiotrophic bacteria may support root acquisition of potassium, phosphorus, and other trace elements (30). Actinobacteria, which are predominant in infertile soils, can degrade recalcitrant organic matter such as chitin and produce antibiotics that suppress pathogen growth (23). Proteobacteria, including Pseudomonas, have been reported to exhibit plant growth-promoting functions such as phosphorus solubilization and siderophore production. In saline or arid soils, they are abundant and can account for 30–50% of the bacterial community under salt stress (31). Dawn et al. (32) detected a predominant Pseudomonas population in early season sorghum rhizosphere soil, whereas this population was scarcely detected in bulk soil. In addition, Lourenço et al. reported that the addition of phosphorus fertilizer increased the Acidobacteria abundance in sorghum soil (33). In the present study, Cyanobacteria were notably more abundant in the rhizosphere soil of Jinnuo 101. Because this phylum can fix carbon dioxide, produce organic matter, and participate in carbon and nitrogen cycling (21), its enrichment is consistent with the highest organic matter content observed in the rhizosphere soil of Jinnuo 101. In contrast, the Jinnuo 102 rhizosphere soil contained a higher abundance of Actinobacteriota, most genera of which have nitrogen-fixing and pathogen-inhibiting capacities (34). Nevertheless, AN and TN contents did not differ significantly in the rhizosphere soil of Jinnuo 102. This may be because nitrogen fixation requires substantial energy input, and its efficiency may be constrained by the low organic matter content of soil.

In sorghum rhizosphere soil, the relative abundances of Burkholderia and Streptomyces were significantly higher than those in bulk soil. Through nitrogen fixation and phosphate solubilization, Burkholderia can promote nitrogen and phosphorus uptake by secreting indoleacetic acid and antimicrobial substances (35, 36). Streptomyces have been reported to alleviate salt, drought, and heavy metal stresses in plants (37). Under different nitrogen levels, Lopes et al. observed the enrichment of Streptomycetaceae and Streptomyces in the rhizosphere of nitrogen-stress-tolerant sweet sorghum, although the mechanisms underlying this enrichment remain unresolved (23). Moreover, cultivar-specific enrichment was observed. Jinnuo 101 rhizosphere soil contained the higher relative abundances of Burkholderiales unclassified, Chloroplast unclassified, and Ohtaekwangia, while Jinnuo 102 rhizosphere soil contained the significantly higher abundances of Actinobacteriota unclassified, Solirubrobacter, and Nocardioides. Among these taxa, Nocardioides can solubilize phosphate, fix nitrogen, and produce siderophores (38).

4.3 Soil metabolites in the rhizosphere of sorghum under low-nutrient stress

Soil metabolites can alleviate abiotic stress by improving nutrient use efficiency. For example, abscisic acid and organic osmolytes, including proline, betaine, and trehalose, contribute to drought and salt stress mitigation (39). These metabolites can also shape surrounding microbial communities by recruiting beneficial bacteria or suppressing pathogens (40). In sorghum rhizosphere soil, DEMs showed significant upregulation in “Biosynthesis of phenylpropanoids”, “Glycerophospholipid metabolism”, “Tryptophan metabolism”, and “ABC transporters” but significant downregulation in “Linoleic acid metabolism”. The “Biosynthesis of phenylpropanoids” pathway produces lignin and flavonoids. Lignin strengthens cell walls and contributes to resistance against pathogenic bacteria. Flavonoids, including formononetin, daidzin, and aesculetin, were upregulated. As shown in previous studies, flavonoids can function as signaling compounds for rhizosphere microorganisms and recruit beneficial soil microbial communities to enhance plant resistance to adverse stresses (41). Rice roots can secrete flavonoids and lipid substances that stimulate the plant defense system against Cd stress (42). Lipid substances, including glycerol 3-phosphate and LysoPC (18:2), were significantly upregulated. As chemical signals, lipids can promote rhizosphere interactions and regulate plant defense responses during exposure to beneficial or pathogenic microorganisms. Furthermore, they maintain plasma membrane integrity, promote abiotic stress responses, support intracellular and intercellular communication, and mediate nutrient exchange during interactions (43). As a precursor of many important secondary metabolites, tryptophan can be utilized by root microorganisms, especially beneficial bacteria, to synthesize IAA and promote plant growth. ABC transporters that transport small molecules (e.g., amino acids) assist in nutrient acquisition by rhizosphere microorganisms. Plant roots produce large amounts of linoleic acid to adapt to abiotic stress, and the decrease in unsaturated fatty acids in sorghum rhizosphere soil may reflect their extensive decomposition by surrounding microorganisms as carbon sources.

More common DEMs were shared between the R1 vs. Non R1 and R2 vs. Non R2 comparisons. This pattern indicates that the two sorghum cultivars exerted similar effects on rhizosphere soil metabolite composition. These shared DEMs may represent the key substances defining the common metabolic characteristics of sorghum rhizosphere soil under nutrient stress. In the R1 vs. R2 comparison, DEMs were enriched in “Biosynthesis of plant secondary metabolites” and “Biosynthesis of phenylpropanoids”. This result indicates that sorghum rhizosphere soil metabolites were affected by cultivar. Biosynthesis of phenylpropanoids was significantly up regulated in rhizosphere soil of Jinnuo 102, which has more abundance of phenols. Some secondary metabolites derived from the phenylpropanoid metabolic pathway, such as flavonoids, terpenoids, and phenols, are excreted by plant roots into the surrounding soil. They modulate the rhizosphere microbial community and subsequently influence plant growth and stress resistance (44).

Soil microorganisms and metabolites are closely interconnected. Soil microorganisms and metabolites are tightly interconnected. Although soil microorganisms rapidly utilize most metabolites secreted by plant roots, some metabolites remain in the rhizosphere and act as important drivers of root–soil–microbe interactions. This can shape the microbial community structure and function (45, 46). In the present study, shifts in sorghum rhizosphere metabolite profiles were significantly correlated with changes in the relative abundance of bacterial communities. This suggests that rhizosphere metabolites may promote or inhibit microbial growth, whereas soil microorganisms may influence metabolite synthesis. Previous studies have demonstrated that triterpene and diterpene biosynthesis affects the root-associated microbial composition of Arabidopsis thaliana. Purified triterpenoid compounds stimulate Sphingomonas proliferation and inhibit Arthrobacter growth (47). Flavonoids function as chemical inducers that attract rhizobia, enhance nod gene expression, and promote nodulation (48). In addition, p-hydroxybenzaldehyde secreted by Sorghum halepense roots suppresses Proteobacteria abundance while increasing the relative abundances of Acidobacteriota, Verrucomicrobiota, and Cyanobacteria (49). Neg-M357T426_2 was sorgoleone analogue. Sorgoleone analogues and oleamide were highly significant positive correlation with Burkholderiales unclassified. These metabolites may provide carbon and energy sources for this genus. This genus can fix nitrogen and solubilize phosphate in soil, which may promote nutrient absorption by sorghum roots under low-nutrient stress. Consistently, Burkholderia, Acinetobacter, and Pseudomonas isolated from sorghum-associated soils have been shown to grow with sorgoleone as their sole carbon and energy sources (15). The amount of shikimic acid in sorghum roots, which serves as a substrate for salicylic acid synthesis, was correlated to rhizosphere Pseudomonas under N stress (50). In Jinnuo 102, the enriched salicylic acid was significantly correlated with Actinobacteriota and may activate plant systemic defense while alleviating the oxidative damage caused by nutrient-poor stress (51). Collectively, we hypothesize that these two sorghum cultivars regulate the colonization and activity of functional microbial communities by different key metabolites. However, because the soil metabolic composition can be complex and sensitive to environmental factors, further experimental validation is needed to identify the specific sorghum rhizosphere metabolites that promote or inhibit the growth of specific bacterial species.

5 Conclusion

This study systematically examined the bacterial community composition and metabolomic profiles of rhizosphere and bulk soils of Jinnuo 101 and Jinnuo 102 under low-nutrient stress. The bacterial structure in rhizosphere soil of sorghum was restructured by increasing the relative abundance of beneficial bacteria. Moreover, the enrichment of metabolites such as terpenoids, fatty acids, sugars, and phenols in the sorghum rhizosphere provides functional support and energy for the sorghum root system and microorganisms to resist low-nutrient stress. Under low-nutrient stress, Jinnuo102 exhibits higher antioxidant enzyme activity and lower MDA content compared with Jinnuo101. The rhizosphere soil of Jinnuo102 harbored higher bacterial diversity and enriched more beneficial bacteria, such as Actinobacteriota unclassified, Solirubrobacter, and Nocardioides. Metabolic pathway analysis demonstrated the significant upregulation of “Biosynthesis of phenylpropanoids” in Jinnuo102 rhizosphere soil, coinciding with an increased abundance of phenolic metabolites. Significant correlations were detected between these microbial taxa and phenolic metabolites, indicating that alterations in the metabolite composition affected the community structure and functionality of the associated microorganisms. This finding reveals that tolerant sorghum enhances low-nutrient resistance by coordinately upregulating phenylpropanoid pathways and enriching beneficial rhizosphere bacteria. Together, these results from this work could guide future mechanistic studies in regulating the colonization and activity of functional microbial communities, forming a metabolite–microbe interaction network that supports sorghum growth under low-nutrient stress.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fsoil.2026.1893513/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Bahar S. Razavi, University of Kiel, Germany

Reviewed by: Xiaolong Shi, Shenyang Agricultural University, China

Jasmine Gill, Medical University of South Carolina, United States