Description of the experimental design
An inoculation experiment was conducted within a long-term field trial (LTE) to investigate the effects of a consortium of beneficial microorganisms (BMc; Pseudomonas sp. RU47, Bacillus atrophaeus ABi03, and Trichoderma harzianum OMG16) on maize performance (cv. Benedictio). The experimental plots employed conservation tillage and received either the recommended use of nitrogen (N) and pesticides for maize (intensive, Int) or reduced N use and no fungicide application (extensive, Ext). Maize plants were drenched twice with the BMc at growth stages EC 12 and 14–16 and harvested at maturity (EC 53–63). During the experiment duration, spring precipitation was exceptionally low compared to the average spring rainfall of the past 30 years, with almost no rainfall in April and less in May and July (Fig. S1). To elucidate the effects of the BMc on the maize-microbiome interaction in the rhizosphere, we analyzed various parameters, including root traits (total root length, root hair length), shoot dry mass (SDM), physiological stress indicators, phytohormones, nutrient status, root and rhizosphere metabolites, and bacterial/archaeal and fungal community compositions, as well as soil metagenome profiles.
BMc strains colonized the maize rhizosphere and increased shoot dry mass and iron uptake
The inoculated strains effectively colonized the rhizosphere, as determined by CFU counts through selective plating (Fig. 1A). Among the BMc strains, ABi03 exhibited the most efficient colonization with a density two orders of magnitude higher than RU47 and OMG (Two-Way ANOVA, F = 1889.68, p < 0.05, n = 8; Fig. 1A). N-fertilization intensity influenced the ABi03 rhizo-competence (Two-Way ANOVA, F = 16.45, p < 0.05, n = 8; Fig. 1A), but had no significant effect on RU47 and OMG16 (Two-Way ANOVA, F = 1.2–0.8, p > 0.05, n = 8; Fig. 1A). Maize SDM was significantly higher by approximately 30% in BMc inoculated plants compared to non-inoculated controls (Two-Way ANOVA, F = 149.7, p < 0.05, n = 8; Fig. 1B). The increase in SDM was accompanied by higher concentrations of sulfur, iron, zinc and manganese in maize leaves of inoculated plants (Two-Way ANOVA, F = 0.3-17.14, p < 0.05, n = 8; Table S1 and Table 1). However, only iron concentration showed an increase regardless of N-fertilization intensity (Two-Way ANOVA, F = 17.14, Tukey’s HSD, p < 0.05, Table 1). Moreover, none of the nutrients (except N in extensive fertilized control plants) showed concentrations below the deficiency limit (Table 1). In contrast, we did not observe any significant effects of the experimental variables on most of the soil measured chemical parameters (Table S2). Furthermore, BMc inoculation did not significantly affect root growth (total root and hair length) or arbuscular mycorrhizal fungi (AMF) colonization, processes crucial for the spatial acquisition of mineral nutrients (Fig. S2). However, AMF root colonization was significantly reduced under intensive N-fertilization, supporting the notion that high N-fertilization and pesticide use have detrimental effects on mycorrhizal symbiosis38,39.
Table 1
Macro- and micro-nutrient concentrations measured in maize shoots (cv. Benedictio) grown under extensive and intensive N-fertilization intensities, without (Ctrl; control plants) and with BMc (beneficial microorganisms consortium) inoculation. DT indicates deficiency threshold based on Campbell (2000). Data represent mean values ± standard deviation. Different letters indicate significant differences (p < 0.05, Tukey’s HSD test, Benjamini-Hochberg correction, n = 4).
| | Extensive | | Intensive |
| DT | Ctrl | | BMc | | Ctrl | | BMc |
Macro-nutrients [g kg− 1] |
C | | 453 | ± 0 | | | 454 | ± 3.00 | | | 453 | ± 1.00 | | | 454 | ± 3.00 | |
N | 30 | 29.7 | ± 0.30 | | | 31.0 | ± 0.80 | | | 31.1 | ± 1.40 | | | 31.8 | ± 1.20 | |
P | 2.5 | 2.72 | ± 0.16 | | | 2.83 | ± 0.14 | | | 2.57 | ± 0.10 | | | 2.59 | ± 0.13 | |
K | 20 | 22.1 | ± 0.80 | | | 22.6 | ± 1.50 | | | 22.0 | ± 0.90 | | | 21.1 | ± 1.00 | |
Mg | 2.5 | 1.84 | ± 0.07 | a | | 1.85 | ± 0.05 | a | | 1.72 | ± 0.07 | ab | | 1.63 | ± 0.09 | b |
Ca | 4 | 5.00 | ± 0.36 | | | 5.03 | ± 0.32 | | | 4.83 | ± 0.20 | | | 5.05 | ± 0.72 | |
S | 1.2 | 1.95 | ± 0.06 | b | | 2.13 | ± 0.10 | ab | 2.14 | ± 0.08 | ab | | 2.23 | ± 0.07 | a |
Micro-nutrients [mg kg− 1] |
Fe | 15 | 85.0 | ± 5.0 | bc | 96.2 | ± 4.70 | a | | 80.2 | ± 5.70 | c | | 91.7 | ± 6.30 | ab |
Cu | 5 | 8.78 | ± 0.32 | | | 8.41 | ± 0.48 | | | 10.07 | ± 1.01 | | | 9.29 | ± 0.77 | |
Zn | 15 | 37.5 | 3.3 | b | | 40.8 | ± 4.50 | ab | 40.9 | ± 3.90 | ab | | 47.0 | ± 3.60 | a |
Mn | 15 | 59.1 | ± 5.0 | b | | 61.7 | ± 6.20 | ab | 62.4 | ± 4.60 | ab | | 77.3 | ± 16.4 | a |
Campbell CR (ed). Reference Sufficiency Ranges for Plant Analysis in the Southern Region of the United States. Southern Cooperative Series Bulletin No. 2000, 394. |
In response to BMc inoculation, a substantial number of stress-related maize genes, in terms of relative transcript abundances, tended to be repressed in the leaves regardless of N-fertilization level (Fig. 1C). In contrast, stress-related genes encoding for iron-dependent enzymes (Table S3), involved in detoxification of reactive oxygen species (ROS), such as superoxide dismutase (SOD) and ascorbate peroxidase (APX) activity, and genes associated with nutrient acquisition pathways, exhibited distinct upregulation in response to BMc inoculation, particularly those involved in iron uptake. Specifically, NAS3, a gene associated with the synthesis of the iron/zinc chelator nicotianamine (Table S3) likely contributing to enhanced iron concentrations in leaves, was consistently upregulated (t-test, p < 0.05, n = 4; Fig. 1C). The iron transporter IRTa showed a contrasting pattern, albeit statistically significant only in the context of extensive N-fertilization samples (t-test, p < 0.05, n = 4).
BMc inoculation exerted systemic effects on maize hormonal status and improved stress resilience independent of N-fertilization intensity
BMc inoculation significantly increased the concentration of growth-promoting phytohormones (indole acetic acid, gibberellic acid, and zeatin) in the shoots by approximately twofold, while decreasing the concentrations of stress-related hormones (abscisic acid (ABA) and jasmonic acid (JA)) (Two-Way ANOVA, F = 28.20-279.34, p < 0.05, Fig. 2). Salicylic acid (SA) concentrations in shoots remained unaffected by BMc inoculation (Fig. 2). In the roots, a similar response of plant-growth hormones to BMc inoculation was observed, with significantly higher concentrations of indole acetic acid (IAA) and zeatin (cytokinin) in the inoculated plants compared to the control (Fig. S3). Interestingly, BMc inoculation led to a significant increase in the concentration of stress-related hormones involved in plant defense in the roots (JA and SA, t-test, p < 0.05, n = 4, Fig. S3), which could potentially be associated with the plant response to the presence of a high microbial load or the drastic change in microbial community composition in the maize rhizosphere.
Since BMc inoculation enhanced the gene expression and shoot activities of iron-dependent ROS detoxification (APX and SOD, ANOVA, F = 45.09, p < 0.05, Fig. 3), consequently it halved the accumulation of hydrogen peroxide (H2O2) in the leaf tissue (ANOVA, F = 285.5, p < 0.05, Fig. 3), and additionally increased leaf concentrations of total antioxidant and glycine betaine (ANOVA, F = 12.54, p < 0.05, Fig. 3), which have functions in non-enzymatic ROS detoxification and osmotic adjustment. The observed decrease in stress-resilience markers within leaf tissues of BMc-inoculated maize plants aligns with the downregulation of stress-related genes in this tissue, indicating a plausible involvement of BMc inoculation in regulating plant stress responses.
The patterns of rhizosphere metabolites are shaped by BMc inoculation
To evaluate the impact of BMc inoculation on rhizosphere metabolites, we established experimental plots with root observation windows that enabled the sampling and profiling of low molecular weight metabolites across distinct maize root zones (Fig. S4). As expected, metabolite concentrations exhibited significant variation across the sampled root zones, reflecting the distinct physiological and functional characteristics of different root types (Fig. S5, Table S4). Given the higher detection of secondary metabolites in the apical root zone, likely exerting a significant effect on microbiota assembly around maize roots in contrast to other root types, our investigation specifically targeted the analysis of rhizosphere metabolites from the apical root zone (Fig. S5). In line with the improved growth performance exhibited by BMc-inoculated plants, a majority of the identified metabolites showed significant increases following BMc inoculation, irrespective of N-fertilization intensity (p < 0.05, Tukey’s HSD test, Fig. 4). These enhancements encompassed organic acid anions (predominantly malate), sugars (such as glucose and trehalose), specific phenolic acids and flavonoids (like caffeic acid, quercetin/naringenin), as well as benzoxazinoid metabolites with antibiotic, allelopathic, and iron-mobilizing properties (e.g., MBOA) (Fig. 4).
BMc inoculation increased Myxococcota, Bacteroidota, and Proteobacteria in the maize rhizosphere
BMc inoculation significantly impacted bacterial β-diversity within the maize rhizosphere, suggesting its potential role as a determinant factor shaping bacterial community assembly as revealed by 16S rRNA gene amplicon sequencing (PERMANOVA, BMc: R2 = 0.12, p = 1x10− 4, n = 8; Fig. 5A). Actinobacteriota, Proteobacteria, and Firmicutes were the most abundant phyla, while Acidobacteria, Bacteroidota, Gemmatimonadota, Chloroflexi, Myxococcota, and Verrucomicrobiota occurred in relative abundances below 5% (Fig. 6A). BMc inoculation specifically decreased the relative abundance of Actinobacteriota in both N-fertilization intensities, while increasing the relative abundance of Acidobacteria, Bacteroidota, Chloroflexi, and Myxococcota under intensive fertilization and Proteobacteria under extensive fertilization (logistic regression, p < 0.05, n = 4; Fig. 6A). Additionally, BMc inoculation reduced the relative abundance of Gemmatimonadota in the rhizosphere of maize grown under intensive fertilization (logistic regression, p < 0.05, n = 4; Fig. 6A).
BMc inoculation significantly increased the relative abundance of 55 and 48 ASVs in intensive and extensive fertilization, respectively, out of a total of 6595 bacterial ASVs (logistic regression, p < 0.05, Benjamini-Hochberg correction, n = 4; Table S5). Concurrently, the relative abundance of 48 and 33 bacterial ASVs decreased due to BMc inoculation under intensive or extensive N-fertilization, respectively (logistic regression, p < 0.05, Benjamini-Hochberg correction, n = 4, Table S5). Out of the 15 most abundant bacterial ASVs identified in the maize rhizosphere, the relative abundance of five of these ASVs, classified as four genera belonging to the Actinobacteriota phylum (Agromyces: ASV3, Pseudarthrobacter: ASV2 and ASV5, unclassified Micrococcaceae ASV24, and Streptomyces: ASV27), decreased in response to BMc inoculation under extensive N-fertilization intensity (logistic regression, p < 0.05, Benjamini-Hochberg correction, n = 4, Fig. 6B). Moreover, the relative abundance of three out of 15 ASVs classified as Actinobacteriota (Agromyces: ASV3, Streptomyces: ASV25 and ASV27), decreased in response to BMc inoculation under extensive N-fertilization intensity (logistic regression, p < 0.05, Benjamini-Hochberg correction, n = 4, Fig. 6B). In contrast, one ASV (ASV20), classified as Bacillus sp., increased due to BMc treatment in samples under extensive N-fertilization (logistic regression, p < 0.05, Benjamini-Hochberg correction, n = 4; Fig. 6B). Notably, ASV20 exhibited high sequence identity with Bacillus atrophaeus and the ABi03 inoculated strain based on pairwise sequence alignment of the amplified region of the 16S rRNA gene (Fig. S6A). Furthermore, ASV20 was consistently among the most abundant ASVs detected in the rhizosphere across all samples. Given its high abundance in control samples (Fig. 6B) and similarity to other B. atrophaeus strains (Fig. S6A), we retained ASV20 in the dataset used to assess BMc-driven modulation of the microbiome. Unlike ABi03, we did not observe ASVs affiliated with Pseudomonas (especially RU47) among the dominant ASVs in the maize rhizosphere, a result that was expected based on our culture-dependent data (Fig. 1A). However, bacterial ASV1083 was phylogenetically associated with RU47, and its relative abundance tended to increase with BMc inoculation (logistic regression, p > 0.05, Benjamini-Hochberg correction; n = 4, data not shown).
Fungal community assembly in the rhizosphere was remarkably affected by BMc inoculation
The fungal community composition in the maize rhizosphere was significantly affected by BMc inoculation (PERMANOVA, R2 = 0.33, p = 1x10− 4, n = 8, Fig. 5). Fertilization intensity also influenced fungal community composition, albeit to a lesser extent (PERMANOVA, R2 = 0.09, p = 0.028, n = 8, Fig. 5B). Accordingly, both experimental variables, particularly BMc inoculation, significantly affected the composition of the fungal community across all taxonomic levels in the rhizosphere (Fig. 6C and 6D). For example, at the phylum level, the relative abundance of Ascomycota, the most dominant phylum, increased from 68.2–80.7% in extensively fertilized samples due to BMc inoculation (logistic regression, p < 0.05, n = 4; Fig. 6B). Contrarily, Mortierellomycota and unclassified fungi (Fungi_phy_Incertae) significantly decreased in relative abundance following BMc inoculation under extensive N-fertilization (logistic regression; p < 0.05, n = 4; Fig. 6C).
Among the 1605 fungal ASVs identified, 121 and 124 exhibited increased relative abundance in response to BMc inoculation under intensive and extensive N-fertilization, respectively, while 49 and 110 ASVs displayed decreased relative abundance under intensive and extensive N-fertilization, respectively (Table S6). Notably, eight dominant ASVs responded to BMc inoculation in the extensive treatments, while only three ASVs responded in the intensive N-fertilization context. Specifically, four fungal ASVs, affiliated with the genera Trichoderma, Marquandomyces, Talaromyces, and Penicillium, were significantly enriched under extensive fertilization (logistic regression, p < 0.05, n = 4; Fig. 6D), while Podila, Mortierella, Stachybotrys and Fungi_gen_incertae_sedis showed an opposite trend (logistic regression, p < 0.05, n = 4; Fig. 6D). Interestingly, BMc inoculation also led to increased abundance of Trichoderma and Talaromyces in intensive fertilized treatments (Fig. 6D). Moreover, the shifts in abundance of the highly-abundant fungal taxa under intensive N-fertilization were restricted to ASV1530 (Pseudogymnoascus), which significantly decreased from the explicitly high abundance of 40.23 ± 19.89% to 8.51 ± 3.74% (logistic regression, p < 0.05, n = 4; Fig. 6D). However, none of the enriched Trichoderma ASVs in response to BMc were closely phylogenetically associated with OMG16, based on the pairwise alignment of the amplified ITS region (Fig. S6).
BMc inoculation enhanced relative abundance of microbial lipopeptide and siderophore biosynthesis genes in the maize rhizosphere
Analysis of the metagenomic dataset revealed numerous sequences annotated to OMG16 ABi03, and RU47, corroborating their establishment in the rhizosphere. BMc inoculation significantly increased the relative abundance of reads assigned to B. atrophaeus ABi03 (ANOVA, F = 96.0, p = 4x10− 8, n = 4; Fig. 7A) and Pseudomonas sp. RU47 (ANOVA, F = 50.6, p = 12x10− 4, n = 4; Fig. 7A), while no such increase was detected for Trichoderma harzianum OMG16. To investigate the impact of BMc inoculation on microbial functions in the rhizosphere, metagenomic shotgun sequencing reads were mapped against a customized database targeting potential plant-beneficial microbial functions (Table S7). BMc inoculation significantly influenced the functional composition of bacterial communities in the rhizosphere (PERMANOVA, R2 = 0.15, p = 0.0014, Fig. 7B) and led to a significant increase in the abundance of genes associated with chemotaxis, quorum sensing, the degradation of aromatic compounds, as well as synthesis of auxin, siderophores, lipopeptides (e.g., surfactin, iturin, athrofactin) and spermidine (potA) (edgeR, p < 0.05, n = 4. Figure 7C). Genes such as mycobactin synthetase (mbtE), involved in the synthesis of iron-chelating siderophores, were highly enriched due to BMc inoculation, particularly under extensive N-fertilization (Fig. 7C). In contrast, lvH, dppA, rbsB, potD and 4,5-diphosphate decarboxylase (involved in type-3-secretion system, chemotaxis, spermidine production and aromatic compound degradation, respectively) decreased with BMc inoculation (edgeR, p < 0.05, n = 4; Fig. 7C).
In order to link microbial taxa to their respective functions, we taxonomically annotated sequences assigned to differentially abundant functions. Approximately 66% of sequences assigned to specific functions met the criteria for reliable taxonomic classification. Eight genes associated with the production of lipopeptides were linked to Bacillus spp., as discerned by sequence similarity. The majority of these genes exhibited 100% identity and 100% alignment with ABi03 (reads from each gene annotated as ABi03: fenD 63%, ituA 75%, ituB 67%, ituC 61%, srfAA 79%, srfAB 54%, srfAC 74%, and srfAD 93%; Fig. 7C). Furthermore, five BMc-enriched genes associated with siderophore production (dhbf 64%, entE 66%, entF 66%, mbtE 63%, and mbtF 59% reads, Table 2) were mapped to bacterial taxa affiliated with the phylum Actinobacteriota, with the highest proportion of attributed sequences annotated as Streptomyces spp. (Fig. S7). Notably, only a limited percentage of these genes was annotated to RU47 and ABi03 genomes, and these instances were exclusively restricted to the BMc samples (Table 2).
Table 2
Annotated siderophore genes in the shotgun sequencing data and their association with Bacillus atrophaeus ABi03 and Pseudomonas sp. RU47. In addition, the association of these genes to ABi03 and RU47 genomes was evaluated by DIAMOND (e = 10− 5, identity = 100%).
Siderophore genes | Total counts | Number of counts annotated to genera | Number of genera annotated per gene | Read counts annotated as Pseudomonas | Read counts annotated as RU47 (100% identity) | Read counts annotated as Bacillus | Read counts annotated as ABi03 (100% identity) |
2,3-dihydroxybenzoate-AMP-ligase, entE | 647 | 428 | 185 | 4 | 0 | 48 | 27 |
L-serine ligase, entF | 3389 | 2242 | 59 | 167 | 0 | 30 | 26 |
Glycine ligase, dhbF | 4889 | 3138 | 176 | 257 | 1 | 191 | 141 |
Mycobactin peptide synthetase, mbtE | 1146 | 721 | 98 | 78 | 2 | 4 | 2 |
Mycobactin peptide synthetase, mbtF | 579 | 344 | 56 | 5 | 0 | 8 | 6 |
Improved plant micronutrient uptake, rhizosphere metabolite patterns and microbiome modulation are linked to BMc inoculation
To delve deeper into the interconnected shifts within the soil-plant system resulting from BMc application, we conducted an integrated network analysis. Only highly significant correlations (Pearson ρ ≥ |0.8|, p < 0.05, Benjamini-Hochberg correction, n = 16) and distance-based clustering were used (Fig. 8, Table S8). This analysis unveiled two primary clusters, Module 1 and Module 2, incorporating all variables positively or negatively correlated with maize growth, respectively (Fig. 8). Module 1 comprised ASV20, taxonomically linked to Bacillus ABi03 from the applied BMc (Fig. S6A), alongside various fungal ASVs taxonomically categorized as Trichoderma spp., although none were phylogenetically associated with Trichoderma OMG16 (Fig. S6B). All of these were positively correlated with growth-promoting variables (Fig. 8, Table S8). Interestingly, the concentrations of iron, zinc, and manganese, along with the functional genes encoding bacterial siderophore synthetases (mycobactin synthetase MbtE, Table S8) displayed a positive correlation with maize growth and the expression of NAS3, a plant gene involved in iron/zinc translocation (Fig. 1C). These positive correlations suggest that BMc inoculation improves iron uptake by facilitating alterations in bacterial siderophore release, and by stimulating root exudation of iron-mobilizing phenolics. This resulted in an improved iron-nutritional status, promoting ROS detoxification with protective effects under drought stress.