Using a sterile plastic spatula and 10 mL of sterile distilled water, both Streptomyces spp. DEF17 and DEF19 spores were collected from the surface of three-week-old cultures cultured on Czapek’s yeast extract agar (CZY) plates at 25 °C (Mattei et al., 2022; Colombo et al., 2019). After soaking tomato seeds (Solanum lycopersicum “Moneymaker”) in 1% sodium hypochlorite for three minutes, the seeds were rinsed with sterile distilled water until disinfectant elimination. Following surface sterilization, the seeds were separated into three groups: a control group that was prepared by soaking seeds in sterile deionized water, and two treatment groups, which were prepared by soaking the seeds in either DEF17 or DEF19 spore solutions (1 × 10⁷ CFU/mL). After treatment, seeds were left under laminar hood flow until completely dried. A total of 20 seeds per treatment were prepared.

To avoid cross-contamination, each seed was seeded separately in 50 mL Falcon tubes filled with 5 mL of ½ Murashige and Skoog (MS) medium containing 0.2% agar and grown in growth chamber under controlled environment with a photoperiod of 16 h light/8 h dark, with a photosynthetically active radiation (PAR) of approximately 130 μmol m^-2 s^-1 and temperatures of 24 °C/18 °C. Seedlings were harvested, and the roots were cut after 21 days of sterile growth. The residual medium was centrifuged for 10 minutes at 4 °C at 10,000 g. The supernatant, containing root exudates, was then recovered and used for chemotropism assays and LC-MS/MS analysis.

The root exudates were analyzed without further sample preparation. The roots were dried under vacuum and weighed, and 3 inox marbles were added for grinding the roots in a bead grinder for 2 minutes. To the ground samples extraction solvent (70% ethanol/MQ) was added in a ratio of 1/100 w/v, and the samples were sonicated for 5 minutes at room temperature and subsequently incubated for 1 hour at 25 °C in a thermo block with 1400 g agitation. After centrifugation (4 °C, 20000g for 20 minutes) 100 µL was recovered and dried under vacuum. The thus dried extract was resolubilized in 100 µL 20% MeOH/MQ.

All samples were filtered through a PTFE syringe filter (0.22 µm, Millex-LG, Merck KGaA, Germany) and analyzed with LC-MS/MS, as previously described (Halime et al., 2025). Ten microliters of the sample were injected and separated with an Acquity UPLC I-Class system equipped with a diode array detector using a reversed-phase Acquity UPLC BEH C18 column (2.1 × 100 mm, 1.7 μm; Waters, USA). The column was maintained at 50 °C, and a flow rate of 0.5 mL/min was used. The mobile phase consisted of 0.1% (v/v) formic acid in water (A) and 0.1% (v/v) formic acid in acetonitrile (B), with the following gradient: 0 min, 1% B; 4 min, 1% B; 16 min, 5% B; 35 min, 40% B; 45 min, 100% B; 50 min, 100% B; 54 min, 1% B; and 60 min, 1% B.

The UPLC was coupled to a TripleTOF 6600+ mass spectrometer (SCIEX, USA) in positive and negative ionization modes. Electrospray ionization was performed using the following parameters: source temperature 650 °C; ion spray voltage of 4.5 and -4.5 kV for positive and negative mode. The curtain gas (nitrogen) was at 30 psi, and nebulizer and turbine gas (air) at 55 and 50 psi. The declustering potential was set at 60 V in positive and -60 V in negative mode. Survey scans of 175 ms were acquired for information-dependent acquisition. The ten highest single-charged MS ions with an intensity higher than 100 counts/sec were selected for fragmentation. MS/MS spectra were collected in high sensitivity mode with an accumulation time of 200 ms. A sweeping collision energy of 15 eV was applied to all precursor ions for collision-induced dissociation. The dynamic exclusion was set for 2 s after three occurrences before the precursor could be fragmented again.

Progenesis QI (v2.3, Waters) was used for identification and relative quantification; raw data files were aligned, normalized, and relative quantitative analysis was performed based on treatments. The two sample types were analyzed in separate experiments. Features without MS/MS data or known contaminants were omitted for the identification stage. Initial identification relied on the use of an in-house database containing MS, MS/MS and metadata on all identifications obtained in the group. For compounds not present in this in-house database, identification was achieved through MS/MS matching of experimental spectra with spectra found in databases such as GNPS (https://gnps.ucsd.edu/ProteoSAFe/libraries.jsp), MZCloud™ (https://beta.mzcloud.org/), LipidMaps (https://www.lipidmaps.org/), and PubChem (https://pubchem.ncbi.nlm.nih.gov) or available in literature. All accepted identifications were manually validated, and accepted identifications were incorporated into the in-house database for future dereplication. The reported identifications are level 2 identifications as defined by the Metabolomics Standards Initiative.

Fusarium oxysporum f. sp. lycopersici, sequenced strain FOL 4287 (NRRL 34936/CBS 123668/FGSC 9935) was used in all experiments. Fol was cultured on potato dextrose agar (PDA; Difco, USA) and incubated at 24 °C for 7 days. For conidia production, a carboxymethyl cellulose (CMC) medium was prepared based on a protocol adapted from Colombo et al. (2019). The CMC medium consisted of 15 g/L carboxymethyl-cellulose (Sigma-Aldrich, USA), 1 g/L NH₄NO₃, 1 g/L KH₂PO₄, 0.5 g/L MgSO₄·7H₂O (Carlo Erba Reagents, Italy), and 1 g/L yeast extract (Difco Laboratories, USA), adjusted to pH 6.5. After autoclaving, 100 mL of sterile CMC medium was poured into sterilized 250 mL Erlenmeyer flasks and inoculated with six 0.6 cm-diameter plugs of Fol mycelium previously grown on PDA. Cultures were incubated at 24 °C on a rotary shaker for 5 days. After incubation, cultures were filtered through a single layer of Miracloth (Calbiochem, USA) into sterile 50 mL Falcon tubes. Conidia were harvested by centrifugation at 9,500 g for 10 min at 4 °C. The pellet was washed three times with sterile MQ and finally resuspended in 0.01% (v/v) Tween 20 solution. Conidia were counted using a Bürker chamber and adjusted to a final concentration of 5 × 10⁶ conidia/mL for further experiments.

A dual-culture method was used to test DEF17 and DEF19 antifungal activity against Fol growth following the protocol from Kunova et al. (2016). Briefly, strain DEF17 and DEF19 were symmetrically streaked on PDA plates, 2.5 cm from the center of the Fol mycelial plugs (d = 5 mm). As controls, PDA plates that had only been inoculated with fungal plugs were employed. Three replications of each treatment were carried out. For five days, all plates were incubated at 26 °C. Each Fol colony’s diameter was measured, and the inhibition rate (IR) was computed using the formula below reported in Zhao et al., 2022:

where R1 = Diameter of pathogenic fungus in the control plate. R2 = Diameter of the pathogenic fungus interacting with the antagonist. Dual culture experiments were repeated in triplicate.

The same exudates used to analyze the tomato metabolomic responses were also used to assess the chemotropic response of Fol conidia to root exudates following the method described by Turrà et al. (2015) with modifications. A 0.5% water agar (WA0.5%) solution was prepared and maintained at 40 °C. Conidial suspensions (5 × 10⁶ conidia/mL) were embedded in 5 mL of WA0.5% to a final concentration of 5 × 10⁴ conidia/mL and poured into standard Petri dishes. After solidification, two wells spaced 1 cm apart were created in the agar using a sterile pipette tip. One well was filled with 50 µL of root exudates from Streptomyces-treated plants (DEF17 or DEF19), and the opposing well with 50 µL of control exudates from untreated plants. Plates were incubated at 24°C for 16 hours in the dark. Microscopic evaluation was performed using an optical microscope (200× magnification) to assess the germ tube emergence direction. For each condition, at least 100 hyphal tips were scored. Chemotropic responses were expressed as the proportion of hyphae orienting toward the treatment well (% toward treatment well) versus the control well. Values >50% indicate chemoattraction, while<50% indicate repulsion. Seven independent batches of cells (n = 100 hyphal tips per batch) were scored for each sample. Finally, the pH of the exudates was measured with a pH meter (XS instruments, Italy).

To assess the biocontrol activity of both DEF17 and DEF19 in planta, a Fol infection assay was done by sowing a total of 60 tomato seeds. Twenty plants per treatment were prepared (DEF17, DEF19 and Control). Seed treatment was done by soaking the tomato seeds in DEF17 or DEF19 spore solution (1x10⁷ spores/ml) or deionized water (control). Tomato plants were grown in a growth chamber, with a 16-h light (∼PPFD of 600 μmol of photons/(m₂ s₋₁)) and an 8-h dark photoperiod. During the experiment, the recorded average temperatures during the light and dark periods were 28 °C and 22 °C, respectively. The relative humidity (HR) percentage averaged between 74.3% and 51.7% during the dark and light periods, respectively. Solanum lycopersicum ‘Moneymaker’, which is susceptible to FOL, was grown in pots of 7 cm × 7 cm x 10 cm, sowing one seed per pot. Plants were grown in a blend (1:1 ratio) of Irish and enriched-black peat-based growth substrate (SER CA-V7 and SER V10-14P, Vigorplant, Italy), previously sterilized for 3 weeks at 55 °C. Pots were randomly distributed and watered every two days with tap water. After 10 days, half of the pots (n=30) were soil drenched with 8 ml of Fol conidia (10⁶ conidia/mL). Disease severity was visually assessed every week for 30 days, starting two weeks after inoculation and following Marlatt (1996) protocol. Briefly, each plant was given a final disease rating on a 1–5 scale: 1 = no visible symptoms; 2 = mild chlorosis with slight wilting or stunting; 3 = moderate chlorosis with noticeable wilting or stunting; 4 = severe chlorosis with strong wilting or stunting; and 5 = plant death. Final assessment was carried out 30 days after inoculation.

One-way ANOVA was used on the data, and the Tukey post-hoc test (P<0.05) was employed to evaluate the mean differences. GraphPad Prism version 10 for macOS (GraphPad Software, La Jolla, California, USA, www.graphpad.com) was used to conduct the analyses. The captions of each figure and table provide more details.

Moreover, Metaboanalyst v.6.0 (https://www.metaboanalyst.ca) was used to perform multivariate analysis of metabolomics data. Before the analysis, the data underwent sum-normalization, square root transformation, and Pareto scaling. The hierarchical clustering was obtained with the following options: Euclidean distance measure and Ward clustering algorithm. Using a variable importance in projection (VIP) plot, which was produced by a partial least squares discriminant analysis (PLS-DA), the most significant variables (metabolites) linked to the variations between clusters were chosen. VIP > 2 variables are indicated, and a one-way ANOVA was used to determine their significance (p< 0.05).

All experiments were performed using at least three biological replicates unless otherwise stated.

In dual culture assays, DEF19 limited the fungal growth by 50% compared to the control, whereas DEF17 did not show any inhibition (Figure 1; Supplementary Figure S1).

However, Fol conidia showed different chemotaxis behaviors when exposed to the root exudates obtained from different treatments (Figure 2). Exposure of Fol to root exudates of tomato plants grown after inoculation with DEF17 resulted in a significant change in the orientation of Fol conodia hyphal tips. In this regard, 46% of Fol germination tubes grew oriented towards exudates of tomato incubated with DEF17 on average across different replicas compared to both control (Figure 2a) and DEF19 (Figure 2c). In addition, a shift in the orientation of the conidia grew germination tube was also observed when Fol conidia were exposed to control and DEF19 plant exudates. In fact, 43% and 57% were oriented towards control and DEF19-treated plant exudates, respectively (Figure 2b). Furthermore, DEF17 seed-treated plant exudates showed the lowest pH (pH 4.2) compared to both control (pH 4.5) and DEF19-treated plant exudates (pH 4.6), respectively. In this regard, these results highlight the complexity of screening biocontrol agents (BCAs) efficacy using in vitro assays.

Untargeted LC-MS/MS analysis was performed on root tissues and root exudates of tomato seedlings treated with DEF17, DEF19, and an untreated control. A total of 4272 LC-MS features were detected in root tissues, of which 172 were annotated. Similarly, 4296 LC-MS features were detected in root exudates, of which 104 were annotated. However, after polishing datasets from repeated metabolite hits and selecting only significant metabolites (p value > 0.05), a total of 123 and 81 metabolites were identified for root tissues and exudates, respectively (Supplementary Table S1). Representative base peak chromatograms of control, DEF17- and DEF19-treated tomato exudates are provided in Supplementary Figure S3.

Principal component analyses were conducted separately for root tissues and root exudates, reflecting their distinct metabolomic compartments. While DEF17-treated plants clustered closely with control plants at the root tissue level (Figure 3a), they formed a distinct cluster in root exudates (Figure 3b), indicating a preferential modulation of metabolite secretion rather than bulk root metabolism. This implies that DEF17 has a more limited but still unique effect on the roots and exudates, whilst DEF19 causes a more pronounced and widespread alteration of the studied metabolomes.

The unique impact of inoculation with the different streptomycetes on the metabolomes is confirmed by hierarchical clustering. The samples from the inoculated seeds with either strain are separated from the controls, but the height of branches containing samples inoculated with DEF19 is higher than that of samples inoculated with DEF17 (Figure 4a). On the other hand, DEF17-treated plants exhibited a distinct and reproducible chemical fingerprint in root exudates, as revealed by the heatmap analysis (Figure 4b), characterized by treatment-specific patterns of relative metabolite abundance, suggesting a significant impact on secretion dynamics. As for root tissues, DEF19 induced the strongest response in tomato root exudates.

In root tissues (Figure 5a), the top 30 discriminant features (VIP > 4.4) showed distinct accumulation patterns across treatments. In this analysis, most metabolites were more abundant in streptomycetes-treated samples than in control samples, but strain-specific effects are obvious, as seen already in Figure 4. The view in root exudates is different (Figure 5b); the majority of the high-VIP compounds are more abundant in exudates from DEF19-treated plants. These abundances are lowest in control-exudates, with DEF17-treated plants taking an intermediate abundance. Notably, tomatine showed higher abundance in control samples and was strongly reduced in DEF19. Since this decrease is much less pronounced in DEF17-exudates, these data show that there is a strain-specific alteration in the exudation of this defense-related metabolite.

Overall, these results are in line with those observed in the heatmap, indicating that Streptomyces spp. treatments, particularly DEF19, induced a different reprogramming in tomato root and root exudate metabolomes.

Among the identified metabolites that change significantly depending on the treatment, a total of 22 metabolites were detected in both roots and exudates (Supplementary Table S2) showing different abundances depending on the sample type and treatment.

For instance, γ-glutamyl leucine (Figure 6a) exhibited clear compartment-specific patterns. In root tissues, DEF19-treated plants exhibited the highest concentrations, followed by DEF17-treated plants and the control. On the other hand, the highest concentrations were found in the root exudates of DEF17-treated plants. The abundance profile of γ-glutamyl methionine (Figure 6b) has the same patterns, being more abundant in roots of DEF19-treated plants and significantly increased in exudates of DEF17-treated plants compared to both DEF19 and control plants. Instead, phenylacetic acid dihexoside (PAA-dihex) (Figure 6c) showed similar patterns in both root tissues and root exudates with DEF17-treated plants showing significantly higher abundances in roots tissues and root exudates. Overall, these results show that the abundance of metabolites found in root tissues and root exudates is differently modulated by the two streptomycete strains in a compartment-specific manner.

In addition to these compartmentalization patterns, some finer modulation of the metabolome is observed. DEF19-treated plants showed a significantly higher signal for 2,4-di-tert-butylphenol hexoside (2,4-DTBP Hex), while the control group presented higher levels of a related derivative, 2,4-di-tert-butylphenol hex-hex (2,4-DTBP HexHex) (Figures 7a, b), with DEF17 showing in both cases at an intermediate abundance.

In planta assays revealed that DEF17 seed-treated plants showed significantly reduced disease severity compared to both control and DEF19 seed-treated plants, at 30 days post-inoculation (Figure 8; Supplementary Figure S2). On the other hand, control plants showed the higher severity scores, consistent with the known susceptibility of the ‘Moneymaker’ cultivar to Fol. In contrast, DEF19- seed-treated plants did not differ significantly from the infected control plants, indicating that this strain did not reduce significantly disease severity in vivo.