Priming tomato with AgNPs coated with Pseudomonas N5.12 metabolites enhances plant resilience to drought stress

Under current climate change conditions, agronomic yield is compromised by limited water availability and sudden drought events throughout the plant cycle. Among the tools to overcome drought stress, beneficial bacteria have been used in agriculture as biofertilizers, but the potential of bacterial metabolites has been increasingly underscored. Furthermore, bacterial metabolites can be formulated into nanoparticles (AgNPs) to further improve their activity. Tomato plants were treated with biogenic silver nanoparticles or bacterial metabolites derived from Pseudomonas N5.12 and exposed to drought stress (via foliar spray or soil drench). Physiological and biochemical parameters, including photosynthetic performance, pigment content, oxidative stress markers, and antioxidant enzyme activities, were evaluated. Transcriptome analysis was also performed to assess the whole-plant response to drought under the different treatments. Drought stress significantly reduced tomato shoot length (−14.4%), fresh weight (−30.8%), and dry weight (−29.5%); increased H 2 O 2 content (18.2%); and altered antioxidant enzyme activities. Pretreatment with AgNP (NPs coated with Pseudomonas N5.12 metabolites, 60 ppm) improved shoot length (11.5%), fresh weight (49.3%), and dry weight (50.1%) under drought conditions. AgNPs increased transpiration without affecting net photosynthesis and induced oxidative signaling, as indicated by increased H 2 O 2 (18.5%) and MDA (14.1%) levels, along with enhanced SOD activity (66.5%). Transcriptomic analysis revealed a distinct AgNP-induced gene expression profile. Approximately 50% of the differentially expressed genes were encoded in chloroplasts and mitochondria and were primarily related to protein synthesis, photosynthesis, and energy metabolism. AgNPs coated with Pseudomonas N5.12 metabolites enhance drought tolerance in tomato by improving growth, modulating redox homeostasis, and inducing extensive organelle-associated gene reprogramming. These findings demonstrate the advantages of nanoparticle-based formulation of bacterial metabolites. Overall, AgNPs show strong potential as a tool to support crop growth under drought conditions.

In recent years, the increasing challenges of climate change have significantly impacted global agricultural productivity. Drought impairs plant growth by reducing leaf expansion, stem extension, and root proliferation while disrupting water relations and photosynthetic pigment synthesis. This leads to decreased light harvesting, reduced carbon fixation, and ultimately lowers plant productivity and crop yield (Wach and Skowron, 2022; Haghpanah et al., 2024). Drought stress remains one of the most limiting factors for global agricultural productivity, and as such, researchers are actively exploring novel approaches to mitigate its effects on crops (Iqbal et al., 2020; Kapoor et al., 2020). Among these approaches, the use of plant growth-promoting rhizobacteria (PGPR) or their metabolites to improve plant adaptation to harsh conditions (Galicia-Campos et al., 2023; Lucas et al., 2023) has led to significant growth in the biofertilizer market (https://www.grandviewresearch.com/industry-analysis/biofertilizers-industry). However, the performance of biofertilizers is not always consistent since PGPR need to successfully colonize plant roots while overcoming root microbiome constraints and adverse soil conditions (Biswas et al., 2024). Despite the effectiveness of bacterial metabolites (Alabdallah and Hasan, 2021; Dhaka et al., 2023), delivery systems for agronomic applications are to be developed. On the other hand, recent efforts have focused on metal nanoparticles, typically 1–100 nm in size, which are endowed with high penetration capacity and are known for their unique properties and various applications, including in the agriculture sector (Alabdallah and Hasan, 2021). Among the nanoparticles, silver nanoparticles (AgNPs) have garnered attention for their potential to enhance plant growth and plant resilience under abiotic stress conditions, including drought (Alabdallah et al., 2021).

AgNPs can be synthesized through a variety of methods (chemical, physical, and biological), with biological synthesis consolidating as an eco-friendly and cost-effective approach (Dhaka et al., 2023). This method not only reduces the use of toxic chemicals but also ensures the production of NPs with unique properties, constituting an attractive alternative for sustainable nanomaterial production (Siddiqi et al., 2018). Biological synthesis uses organic matter as a bioreductant, producing NP coated with organic matter, either from plants, fungi, or other sources (Siddiqi et al., 2018). The biological synthesis of AgNPs using bacterial strains or metabolites, endowed with strong antibacterial and antifungal properties, has been previously demonstrated (Siddiqi et al., 2018; Ansari and Rehman, 2021; Plokhovska et al., 2025a; 2025b). Despite the number of AgNP synthetized with biological methods, using metabolites from specific PGPR would result in an NP coated with active bacterial molecules and, therefore, endowed not only with the advantages of the NP size but also with the active coating (Nie et al., 2023). In line with this reasoning, the formulation of bacterial metabolites on AgNP has resulted in the activation of plant secondary metabolism, while the non-formulated metabolites are not effective (Gutierrez-Albanchez et al., 2025).

AgNPs can influence key plant processes, with effects depending on concentration, size, shape, and species. At optimal doses, they can enhance growth, seed germination, photosynthesis, and nutrient uptake (Sadak, 2019; Kulikova et al., 2020; Khan I. et al., 2021; Lahuta et al., 2022) and are also applied as nano-fertilizers and nano-pesticides to improve crop productivity and stress management (Kamil et al., 2017; Manimegalai et al., 2020). AgNPs enhance plant tolerance to abiotic stresses such as salinity, drought, and heavy metals (Khan M. et al., 2021; Nejatzadeh, 2021). They reduce oxidative damage, improve photosynthesis and chlorophyll content, maintain ionic balance, and activate antioxidant enzymes (Khan I. et al., 2021; Mansour et al., 2021). At the molecular level, AgNPs modulate stress-responsive genes and enhance nutrient uptake and osmolyte accumulation, supporting crop resilience (Kamil et al., 2017; Gupta et al., 2018).

Tomato (Solanum lycopersicum) is an important vegetable crop worldwide, valued for its nutritional and economic significance. However, tomato plants are highly sensitive to drought stress, which can significantly reduce growth and fruit quality (Prabha and Deepak, 2025). Due to their well-characterized physiological and stress responses, tomatoes are an ideal model for studying the effects of AgNPs on plant adaptation and resilience under water-limited conditions.

Our recent studies reported the synthesis of AgNPs coated with metabolites from the beneficial strain Pseudomonas N5.12 endowed with antimicrobial activity (Plokhovska et al., 2025b). These AgNPs were also active on plants, enhancing tomato metabolism and resilience at 60 ppm (NP60); this dose was based on the optimization performed in a previous study (Plokhovska et al., 2025c). Furthermore, despite the in vitro antimicrobial potential (Plokhovska et al., 2025b), no significant negative effects on soil microbial diversity are detected under controlled conditions, therefore appearing as an environmentally safe alternative (Plokhovska et al., 2025c). We hypothesize that metabolites from Pseudomonas N5.12, when formulated as AgNPs, enhance plant drought tolerance more effectively than the bacterial strain or its metabolites alone (Lucas et al., 2023). To differentiate the effects of formulated metabolites (AgNPs) from those of bacterial metabolites alone, the treatments included AgNPs (NP60), non-formulated bacterial metabolites (MLs), and a non-treated control. Key markers included growth, photosynthetic pigments, ROS (reactive oxygen species), ROS scavenging enzymes, antioxidants, osmolytes, and photosynthetic efficiency (fluorescence and net photosynthesis); genetic reprogramming in all three treatments was analyzed by transcriptomic analyses.

Biosynthesis AgNPs from Pseudomonas N5.12

AgNPs were synthesized as described in Plokhovska et al. (2025b) and were directly used without further synthesis, modification, or functionalization. In brief, a 24 h culture of Pseudomonas N5.12 was centrifuged and filtered through a 0.25 µm filter to remove cells. The bacterial filtrate was mixed with 1 mM AgNO 3 (2:4, v/v) and incubated at 37 °C for 24 h. The product was characterized by UV–VIS (430 nm) and TEM, showing spherical particles (20.71 ± 0.43 nm) with an organic crown of bacterial metabolites. X-ray diffraction (XRD) analysis confirmed the crystalline nature of the AgNPs. The XRD spectrum showed characteristic Bragg reflection peaks corresponding to a face-centered cubic structure of AgCl crystals. Fourier-transform infrared spectroscopy (FT-IR) revealed organic functional groups (O–H, C–H, C=O, and C–O), indicating successful coating of AgNPs by bacterial metabolites. Zeta potential was −26.07 ± 0.8. Additionally, these AgNPs demonstrated strong antibacterial activity against both human and plant pathogens, confirming their bioactive potential. The physicochemical characterization of AgNPs was previously reported in detail in our earlier study (Plokhovska et al., 2025b), and the same characterized nanoparticles were used in the present work. AgNPs were lyophilized, resuspended in Milli-Q water with a stock concentration of 6,000 μg/mL, and a working solution of 60 μg/mL (NP60) was used (Plokhovska et al., 2025c); the same batch used for characterization by Plokhovska et al. (2025b) was used in this experiment (Batch: 231205BIO-N512). The bacterial filtrate was also used as treatment (ML).

Experimental setup and design

Experiments were conducted on tomato plants S. lycopersicum F1-Hybrid (Razymo RZ, Netherlands). A schematic diagram of the experimental procedure is shown in Figure 1. The experiment was conducted following a completely randomized design (CRD). Seeds were sown in trays (12) of 8 alveoli (6 cm × 6 cm × 6 cm = 216 cm 3), each tray being a replicate with 8 plants; a peat: sand mixture (1/1; v/v) was used as substrate. Each tray was watered with 1 L of tap water every 2 days; once a week, plants were supplied with the same volume of half strength Hoagland solution (Phytotechlab H353, Spain). Plants were grown in the greenhouse, under natural photoperiod, with a 10 °C–30 °C day/night temperature range. One month after sowing, 24 seedlings (three trays) were inoculated with bacterial filtrate (ML), and another 24 seedlings (three trays) were inoculated with AgNPs (NP60); two other groups (3 trays each) were left untreated as controls, with the aim of having an irrigated control (Control 100%) and a drought control. Bacterial MLs were delivered to plant roots by soil drench (5 mL/plant), while AgNPs (60 μg/mL) were sprayed directly onto tomato leaves (1 mL per plant); these doses and concentrations were selected based on optimization in our previous study (Plokhovska et al., 2025c), which was carried out with the same batch (231205BIO-N512) as by Plokhovska et al. (2025b). The treatment was performed three times, as shown in Figure 1. Three days after the last application, plants were subjected to drought stress for 6 days, leaving 3 trays as non-stressed controls (control 100%); therefore, four groups (three trays each) were in the trial: control 100, control (C), bacterial MLs, and nanoparticles (NP60). After 6 days of non-watering and just before the point of no return for wilting (Supplementary Figure S1), leaf samples were collected for transcriptomic analysis, and watering was restored, allowing the plants to recover. This period without watering was shown to reliably induce drought stress under similar conditions (Lucas et al., 2023). Photosynthetic efficiency measurements were performed 6 days post watering in three plants, and the aerial parts of plants from each treatment (n = 3) were harvested for biometric analysis (height, fresh weight, and dry weight), and the remaining plants in each replicate were pooled, powdered in liquid nitrogen, and stored at −80 °C for further analysis.

Chlorophyll fluorescence

Photosynthetic efficiency was assessed based on the chlorophyll fluorescence emitted by photosystem II. Chlorophyll fluorescence was measured using a pulse amplitude-modulated (PAM) fluorometer (Hansatech FM2, Hansatech Instruments Ltd., United Kingdom) on leaves that were dark-adapted for 1 h. The variable fluorescence (Fv) was calculated as the difference between the maximum fluorescence (Fm) and the minimum fluorescence (Fo). The maximum photosynthetic efficiency of photosystem II (PSII) was calculated as Fv/Fm. Immediately, the leaf was continuously irradiated with red–blue actinic beams and equilibrated to record a steady-state fluorescence signal (Fs). The effective PSII (φPSII) and non-photochemical quenching (NPQ) coefficients were calculated as follows: φPSII = (Fm’-Fs)/Fm’) and NPQ = (Fm-Fm’)/Fm’). All measurements were carried out on five plants per treatment.

Photosynthetic pigments

One hundred milligrams of leaf powder were dissolved in 1 mL of acetone 80% (v/v), incubated overnight at 4 °C, and then centrifuged 5 min at 10,000 rpm in a Hermle Z233 M-2 centrifuge. Immediately after, absorbance at 647, 663, and 470 nm was measured to quantify chlorophyll a, chlorophyll b, and carotenoids, respectively, using a BioMate 5 spectrophotometer. The concentrations of chlorophylls (a and b) were calculated using the equation proposed by Porra et al. (1989), and carotenoids with the one proposed by Lichtenthaler (1987).

Net photosynthetic and transpiration measurements

Net photosynthetic rate (Pn) and transpiration rate (E) were measured simultaneously in nine plants using a CI-340 portable photosynthesis system (CID Bio-Science, Inc., Camas, WA, United States). Water use efficiency (WUE) was calculated as WUE = Pn/E. The CI-340 system was equipped with a light control module (CI-301 LA) to adjust light intensity to 500 μmol m -2 ·s -1. The net photosynthetic rate was determined by measuring CO 2 concentration before and after air entered the leaf chamber, allowing calculation of the CO 2 assimilation rate per unit leaf area (μmol·m -2 ·s -1). Transpiration rate was determined by measuring water vapor concentration before and after air entered the leaf chamber, allowing calculation of the water vapor flux per one-sided leaf area (mmol·m -2 ·s -1).

Oxidative stress markers

The quantification of H 2 O 2 was performed following the method described by Shukla et al. (2022). Samples (0.2 g of leaf powder) were suspended and vortexed with 2.0 mL of 0.1% (w/v) trichloroacetic acid (TCA) in an ice bath. The homogenate was centrifuged at 10,000 rpm for 20 min. The supernatant (0.5 mL) was mixed with 0.5 mL of 10 mM of potassium phosphate buffer (pH 7.0) and 1 mL of 1 M potassium iodide. After 5 min, absorbance was measured spectrophotometrically at 390 nm. The amount of H 2 O 2 formed was estimated from a standard curve and expressed as nmol/g H 2 O 2 of FW. The H 2 O 2 standard curve (0.1–0.9 nmol) was prepared in the same extraction buffer, with absorbance measured at 390 nm.

Malondialdehyde (MDA) content was determined as described by Hu et al. (2016). In brief, 0.1 g of leaf powder was added to 2 mL of 10% (v/v) trichloroacetic acid (TCA) and centrifuged at 15,000 rpm for 30 min at 4 °C. The supernatant (1 mL) was mixed with 4 mL of 20% TCA and 0.5% thiobarbituric acid (TBA). The reaction was incubated at 95 °C for 30 min and cooled to room temperature. Then, it was centrifuged at 10,000 rpm for 10 min, and the absorbance at 532 and 600 nm of the supernatant was measured. The concentration of MDA was determined as follows: where ε is the molar extinction coefficient (155 mM -1 /cm).

Enzyme activities

For enzyme activity measurements, 10 mg of leaf powder was suspended in 1 mL of 0.1 M potassium phosphate buffer (pH 7) containing 2 mM phenylmethylsulfonyl fluoride at 4 °C. The mixture was sonicated for 10 min and centrifuged at 12,000 rpm for 10 min to obtain the soluble protein fraction. Total protein was determined by mixing 50 μL of supernatant with 250 μL of Bradford reagent in 96-well plates (Thermo Fisher Scientific, United States) and measuring absorbance at 595 nm after 30 min of incubation. Protein concentration was calculated from a BSA calibration curve (0.05–2 mg/mL). The supernatant was then used to assess ROS-scavenging enzyme activities spectrophotometrically.

Ascorbate peroxidase (APX) was measured using the method described by Garcı́a-Limones et al. (2002). The enzyme extract (100 μL) was mixed with 860 μL of potassium phosphate buffer (50 mM pH 7.0) and 120 μL of 2.5 mM sodium ascorbate. The reaction was started by adding 120 μL of 50 mM H 2 O 2. A decrease in absorbance at 290 nm was indicative of ascorbate oxidation. APX activity was calculated using the following formula: where ΔA290/min is the change in absorbance per minute at 290 nm (A 0 − A 20 s × 3 to convert to per minute); ε is the molar extinction coefficient of ascorbate (2.8 mM −1 /cm); L is the path length of the cuvette (1 cm); Vt is the total reaction volume (μL); Vs. is the volume of enzyme extract used (μL); and [protein] is the protein concentration in the sample (mg/mL). Activity was expressed as μmol of ascorbate oxidized per minute per mg of protein.

Catalase (CAT) was measured by the method described by Garcí-Limones et al. (2002). The enzyme extract (50 μL) was mixed with 980 μL of potassium phosphate buffer (50 mM pH 7.0) and 120 μL of 200 mM H 2 O 2. The absorption at 240 nm was measured immediately and after 20 s; a decrease indicates H 2 O 2 breakdown. CAT activity was calculated using the following formula: where ΔA240/min is the change in absorbance per minute at 240 nm (A 0 − A 20 s × 3 to convert to per minute); ε is the molar extinction coefficient of H 2 O 2 (36 mM −1 /cm); L is the path length of the cuvette (1 cm); Vt is the total reaction volume (μL); Vs. is the volume of enzyme extract used (μL); and [protein] is the protein concentration in the sample (mg/mL). Activity is expressed as μmol of H 2 O 2 decomposed per minute per mg of protein.

SOD activity was determined following the specifications of the SOD activity detection kit (SOD Assay Kit-WST, Sigma-Aldrich, Darmstadt, Germany). With this method, xanthine was converted to superoxide radical ions, uric acid, and hydrogen peroxide by xanthine oxidase (XO). Superoxide reacted with WST1 (water-soluble tetrazolium 1) to generate a water-soluble formazan product that absorbed light at 440 nm. SOD prevented the reduction of WST1 to WST-1formazan, thereby reducing the absorption at 440 nm, which was proportional to SOD activity; the rate of WST1 reduction with O 2 was linearly related to xanthine oxidase (XO) activity. The activity of SOD was expressed as the percentage of inhibition of WST-1 reduction, calculated as follows: where Abs(control)​ is the absorbance of the reaction mixture without enzyme and Abs(sample)​ is the absorbance of the reaction mixture with the enzyme extract.

RNA-seq analysis

RNA-seq analysis was carried out by AllGenetics and Biology SL (www.allgenetics.eu). Three individual plants, one from each biological replicate, were pooled to constitute one analytical replicate. From each analytical replicate, three independent RNA extractions were performed and subsequently pooled to generate a single RNA library per replicate. Each sample was mechanically homogenized in a TissueLyser II (QIAGEN) using a pre-chilled tube holder. The resulting powder was used as input for RNA extraction. RNA was isolated from each sample using the NZY Total RNA Isolation Kit (NZYTech), strictly following the manufacturer’s instructions. The RNA extracts were quantified and quality-checked in an Agilent 2100 Bioanalyzer using the Agilent RNA 6000 Nano Kit.

For library preparation, the NEBNext® UltraTM II Directional RNA Library Prep Kit (New England BioLabs) was used. In brief, each RNA extract was enriched in mRNA by selecting molecules with a poly-A tail at their 3′ end with the NEBNext Poly(A) mRNA Magnetic Isolation Module. The captured mRNA molecules were then converted into cDNA, and sequencing adapters were added to their ends. The molecules within each library were then indexed with a unique oligonucleotide tag so that the libraries could be pooled for sequencing and demultiplexed afterward. The fragment size distribution and concentration of the libraries were assessed using an Agilent 2100 Bioanalyzer with the Agilent HS DNA Kit. The finished libraries were pooled in equimolar amounts according to the results of a Qubit dsDNA HS Assay (Thermo Fisher Scientific) quantification. The resulting pool was sequenced in a fraction of a NovaSeq PE150 flow cell (Illumina), aiming for a total output of 60 gigabases.

The quality of the FASTQ files was assessed using FastQC software (Andrews, 2010), and the output was summarized with MultiQC (Ewels et al., 2016). Trimmomatic v0.39 (Bolger et al., 2014) was used to remove the adapters (ILLUMINACLIP option), low-quality regions (AVGQUAL:26), and remaining reads shorter than 80 bp (MINLEN:80). STAR 2.7.8a (Dobin et al., 2013) was used to align the retained reads to the S. lycopersicum genome assembly SL3.1. Secondary alignments were removed, and uniquely aligned reads were used for further analysis. Read counts were computed using htseq-counts v0.13.5 (Anders et al., 2015). Read counts were normalized and filtered following the Trimmed Mean of M-values (TMM) method (Robinson and Oshlack, 2010). Low-quality raw reads and adaptor sequences were removed, yielding clean reads that ranged from 54.05 to 71.27 million. By comparing the reads to the tomato genome, the genomic alignment of each sample was obtained, with an alignment rate of approximately 92%. Differential gene expression was calculated for three pairwise comparisons: C vs. ML, C vs. NP60, and ML vs. NP60 (Ewels et al., 2016).

To compare the three treatments, genes showing significant differential expression (log 2 FC > 2 and log 2 FC < −2) in each pairwise comparison were selected and merged into a single table, in which raw expression values were presented, assigning a value of zero to genes for which no expression was detected in those treatments (Supplementary Table S2). Then, a second selection was carried out, identifying genes with a log 2 FC value between 0.2 and −0.2, focusing on those genes simultaneously triggered by ML and NP60 with zero expression in the controls, indicative of common targets of the metabolites, irrespective of whether they were formulated or not (Supplementary Table S3).

Statistical analysis

The statistical data analysis was performed using IBM SPSS Statistics (Version 29.0.2.0). Values were recorded as the mean ± standard error (n = 3). A student’s t-test between control (100) and the drought control was run (p < 0.05), with significant differences marked by an asterisk. Prior to ANOVA, homoscedasticity and normality of variance were assessed using Statgraphics Plus 5.1 for Windows, and the data met the requirements for analysis. An ANOVA test was conducted under a one-way completely randomized design (CRD) to analyze differences between treatments and the control under drought stress. When significant differences appeared (p < 0.05), an LSD test was used to reveal differences between them; different letters (a, b, and c) indicate significant differences between treatments (C, ML, and NP60).

A principal component analysis (PCA, Canoco™ for Windows v.4.5 software) was performed with expression values in the three treatments of differentially expressed genes (genes that showed a significant differential expression, log 2 FC > 2, and log 2 FC < −2 in each pairwise comparison) (Supplementary Table S3). In the ordination provided by the PCA, the genes that clustered closest to each treatment were selected and classified into functional categories (manually curated).

Basal plant response to drought: Drought significantly reduced shoot length by 14.4% (p = 0.001) and caused a notable decrease in shoot fresh weight by 30.8% (p < 0.05) and dry weight by 29.5% (p < 0.05), as shown in Supplementary Table S1. Drought increased photosynthetic pigment concentration, but these changes were not statistically significant compared to non-stressed controls. The onset of drought was evidenced by a significant decrease in transpiration compared to non-stressed plants; however, it did not affect net carbon fixation (Pn), water use efficiency (WUE), or photosynthetic efficiency (Supplementary Table S1). Drought significantly increased H 2 O 2 content by 18.2% (p < 0.01), while no significant differences in MDA content (p > 0.05) were found. Notably, drought reduced APX activity by 29.1% (p > 0.05), while causing a significant increase in CAT activity by 142.3% (p < 0.01; Supplementary Table S1).

Once that drought response was characterized by comparing control 100 vs. drought control, the physiological response of plants under drought to ML and NP60 was characterized.

Plant growth and physiological responses

Given these observations, we evaluated whether pretreatment with bacterial MLs or AgNPs at 60 ppm (NP60) could alleviate the negative effects of drought. Accordingly, pretreatment of seedlings with ML or NP60 enhanced plant growth and biomass under drought stress (Figure 2a). Specifically, ML induced an increase in shoot length by 14.8% (p < 0.001), similar to the NP60 treatment, which showed an increase of 11.5% (p < 0.001), as shown in Figure 2b. Additionally, both treatments resulted in an increase in shoot fresh weight by 57.8% with ML (p < 0.001) and 49.3% with NP60 (p < 0.01) (Figure 2c) and shoot dry weight by 46.2% with ML (p < 0.001) and 50.1% with NP60 (p < 0.01) (Figure 2d).

Photosynthetic performance and pigment content

To better understand the physiological mechanisms underlying the observed changes in growth and biomass, the effects of drought and pretreatment with ML or NP60 on photosynthetic performance and pigment content were evaluated.

Transpiration (E) and net photosynthesis (Pn) are shown in Figure 3. NP60 significantly increased transpiration by 22.9% (p < 0.001) (Figure 3a); however, net carbon fixation for both treatments showed similar values to the control (Figure 3b).

Photosynthetic efficiency (through chlorophyll fluorescence) in tomato plants after 6 days of drought, prior to rewatering, showed changes only in the minimal fluorescence (Fo), which significantly increased in ML- and NP60-treated plants by 17.3% (p < 0.01) and 13.4% (p < 0.05), respectively. Maximum photosynthetic efficiency (Fv/Fm), effective quantum yield (φPSII), and non-photochemical quenching (NPQ) were not affected by any treatment (Supplementary Figure S2). In summary, pretreatment with ML and NP60 modulate specific photosynthetic traits, including transpiration and minimal fluorescence, while preserving overall photosynthetic efficiency.

Oxidative stress markers and antioxidant enzyme activity

NP60 treatment did not affect pigments’ concentration compared to drought control (Figure 4). Plants treated with ML exhibited a 20.9% increase in chlorophyll a content (p < 0.01; Figure 4a), an 18.5% increase in chlorophyll b content (p < 0.01; Figure 4b), and a 15.8% increase in carotenoids compared to non-stressed plants (p < 0.05; Figure 4c).

Oxidative stress markers (H 2 O 2 and MDA) are shown in Figure 5. NP60 increased H 2 O 2 and MDA contents by 18.5% (p < 0.01) and 14.1% (p < 0.05), respectively; however, ML treatment did not affect these parameters.

Regarding ROS scavenging enzymes, no significant changes in APX activity were detected in response to either treatment compared to the control, with a trend toward decrease in ML and the opposite trend in NP60; however, differences between ML and NP60 were significant (p < 0.05; Figure 6a). A significant decrease in CAT activity was observed in ML- and NP60-treated plants, with reductions by 73.5% (p < 0.001) and 38.7% (p < 0.05), respectively (Figure 6b). In contrast, NP60 resulted in a significant increase in SOD activity of 66.5% (p < 0.01) compared to the drought control (Figure 6c).

The significant changes observed in plant growth and redox status under each treatment suggested a dramatically different response involving more changes than those detected by the specific markers analyzed above, prompting us to perform transcriptome profiling using RNA-seq to explore these mechanisms.

Transcriptomic analyses

Differential gene expression is presented in three pairwise comparisons: C vs. ML, C vs. NP60, and ML vs. NP60 (https://doi.org/10.5281/zenodo.15089625) and only includes genes expressed in the two elements being compared; therefore, when comparing control with ML, a gene that was expressed only in ML would not appear in the comparison, similarly in control to NP and ML to NP. However, a gene that was not expressed in the control but was expressed in the ML and NP will show up in this latter comparison, leading to different genes and numbers of genes in each pairwise comparison. In order to compare the three treatments, all genes were first merged into a single table in which raw expression values and differential expression in each pairwise comparison were included; a zero value was assigned for genes in those treatments in which no expression was detected (Supplementary Table S2). A total of 17,369 genes were analyzed. In order to identify overexpressed genes in each treatment, a selection of differentially expressed genes with expression values log 2 FC > 2 and log 2 FC < −2 was performed (Supplementary Table S3). A total of 755 genes were found.

Additionally, in order to identify genes exclusively triggered by ML and NP60, several consecutive selections were carried out. First, genes showing a log 2 FC expression value between 0.2 and −0.2 in the ML vs. NP pairwise comparison were selected; these represent common targets responsive to ML and NP60. Next, genes showing log 2 FC expression values over 2 were selected in order to rule out the possibility of any of these genes being overexpressed in the control in each pairwise comparison. Finally, log 2 FC values under −2 were considered indicative of high expression either in ML or NP compared to the control. A total of 690 genes indicative of ML and NP60 common targets were identified, showing zero expression in the controls. A heatmap with the expression values of these 690 genes in the three treatments is shown in Supplementary Table S4a. Moreover, the genes with raw expression value over 100 in NP60 were classified in functional categories (manually curated) (Figure 11; Supplementary Table S4b).

In order to characterize the gene expression pattern under each treatment, a PCA (Figure 7) was performed with absolute expression values of all genes with a log 2 FC value higher than 2 or lower than −2 in any of the three pairwise comparisons (C vs. NP60, C vs. ML, and NP60 vs. ML); 755 genes were differentially expressed, among which 93 genes were grouped toward the control, 112 toward ML and 182 toward NP60 (Supplementary Table S3a–c, respectively).

To characterize the biological functions of the differentially expressed genes under the three treatments, a functional classification analysis was conducted and location of the genes was determined (nuclear, mitochondrial, or chloroplastic) within the three groups of genes (Supplementary Table S3a–c, control, LM, and NP60, respectively). The subcellular localization prediction showed that most genes upregulated under drought stress were encoded in the nucleus for controls (Figure 8) and ML (Figure 9); however, under NP60 treatment, 42% were encoded in the chloroplast and 8% in the mitochondrion (Figure 10).

The functional classification of the overexpressed genes in drought controls in the PCA is presented in Supplementary Table S3a. In particular, the 93 overexpressed genes encoded for proteins involved in metabolism and synthesis of cellular components (21%), transcription factors (8%), stress response (10%), cell wall synthesis (5%), defense against pathogens and herbivores (3%), and oxidative stress (2%) (Figure 8).

The main functional categories associated with ML comprised 112 genes (Supplementary Table S3b). These included defense against pathogens and herbivores (20%), metabolism and synthesis of cellular components (17%), stress response (10%), cell wall synthesis (4%), and oxidative stress (3%) (Figure 9). Importantly, 17% of the overexpressed transcription factors were involved in the regulation of several processes’ responses: stress response (8%), hormone response (4%), development (3%), and defense (2%). Therefore, a marked activation of stress response takes place under ML treatment. Interestingly, there were abundant transcripts indicating activation of hormone-mediated signal transduction pathways, with those related to ethylene being the most abundant.

The 182 genes (Supplementary Table S3c) specific to NP60-treated samples revealed a striking activation of chloroplast- (42%) and mitochondrion-encoded genes (8%), as shown in Figure 10. Among the chloroplast encoded genes, three groups appeared: one for protein synthesis (ribosome synthesis, tRNA, and transcription and translation genes), which accounted for 40%; another one for photosynthesis (non-photosystem photophosphorylation-related genes and structural photosystem genes), which accounted for 50%; and transport and signaling genes accounted for the remaining (Figure 10). Interestingly, numerous structural genes related to photosystems and ribosomes were detected, along with genes involved in photophosphorylation processes occurring independently of the photosystems. Thus, NP treatment stimulated metabolism, energy production, protein synthesis, photosynthesis, and growth-related signaling in plant cells.

The 690 genes that were overexpressed by NP60 and ML but were not expressed in the control (Supplementary Table S4a) were organized from the highest expression value (902.56) in the NP60 treatment to the lowest expression value (3.04). A functional classification was performed on the 64 genes that showed expression values over 100.00 in the NP60 treatment (Figure 11). These genes were involved in key physiological processes, among which those related to transcription and translation of proteins, especially those involved in the synthesis of structural components of ribosomes (33%), were prominent, along with genes of unknown function (23%) (Figure 11).

Among the strategies to tackle drought’s impact on crop yields, beneficial bacteria PGPR have been reported to enhance plant yield and fitness under adverse conditions by activating the plant’s adaptive mechanisms (Galicia-Campos et al., 2022; Gupta et al., 2022; Lucas et al., 2023). Not only bacterial strains have shown these effects, but bacterial metabolites also trigger plant metabolism (Gutierrez Albanchez et al., 2018; Martin-Rivilla et al., 2020). Interestingly, if bioactive bacterial metabolites reduce Ag + ions during the synthesis of AgNPs, the resulting NPs acquire a unique organic coating composed of these metabolites, among which are active compounds (elicitors). This coating determines NP size and contributes to a more efficient delivery system and the concomitant plant response (Khan et al., 2019; Rani et al., 2023). In this context, the application method is relevant as root drench could affect the rhizosphere, whereas foliar spraying allows direct NP uptake and rapid systemic translocation without impacting soil microbiota, making it a practical and consistent delivery approach (Plokhovska et al., 2025c).

Drought impact in non-treated plants resulted in H 2 O 2 accumulation and CAT activation, indicating a typical oxidative stress response (Parkash and Singh, 2020). In ML-treated plants, growth benefits were associated with higher photosynthetic pigments and photosynthesis-related nuclear-encoded genes, indicating activation of light energy absorption. Despite this, oxidative stress was managed as H 2 O 2 and MDA levels were not elevated, and ROS-scavenging enzymes remained unchanged under the influence of ML. Furthermore, overexpressed genes triggered by ML reveal a hormone-mediated response involving mainly ABA, ethylene, and auxins (Supplementary Table S3b). The relaxed status of these plants is consistent with the depletion phase of the faster and more active response of primed plants to drought stress during the post-challenge primed state (Martinez-Medina et al., 2016; Mauch-Mani et al., 2017; Gutierrez Albanchez et al., 2018).

However, formulated bacterial metabolites (NP60) impact drought response, shifting toward controlled ROS signaling rather than oxidative damage. Although NP60 increased H 2 O 2 and MDA levels (Yang et al., 2021), plants showed improved physiological performance, suggesting activation of adaptive redox signaling pathways (Kanojia and Dijkwel, 2018; Waszczak et al., 2018). Enhanced penetration due to NP size and efficient foliar delivery likely contribute to the stronger metabolic and transcriptomic reprogramming observed. The enhanced drought tolerance induced by AgNPs likely results from the combined effect of NP size and metabolite coating. Small NPs improve tissue interaction and uptake, while the coating provides bioactive signals that modulate stress responses, hormone regulation, antioxidants, and osmolytes. These molecular adjustments under NP60 treatment directly translate into higher shoot length, biomass, and photosynthetic efficiency after drought, demonstrating the link between gene expression reprogramming and phenotypic recovery.

Transcriptomic analysis evidences a markedly different response to drought in the three groups of plants (PCA), associated with gene expression reprogramming, supporting physiological and metabolic results. The natural response to drought stress identified in controls involves 93 genes, of which only 4% are related to stress response. Interestingly, bacterial MLs induce a more active response to drought stress, activating a significant overexpression of more genes (112), with a relevant abundance of those involved in stress response. When these metabolites are formulated in NPs, the response is even more marked (183 genes), doubling the number of overexpressed genes in controls. In addition to increasing the number of differentially expressed genes, their nature is also different. While ML activates transcription of nuclear-encoded genes involved in a hormone-mediated response, AgNPs trigger defense and photosynthesis, activating chloroplast-encoded genes, increasing respiratory metabolism by targeting mitochondrial-encoded genes, as well as nuclear-encoded genes involved in defense (Supplementary Table S3c). Overexpression of more genes in ML- and NP-treated plants at the peak of the stress period, right before the rewatering, is consistent with a post-challenge-priming response as primed plants have activated a faster and more intense response upon stress (Martinez-Medina et al., 2016; Mauch-Mani et al., 2017; Gutierrez Albanchez et al., 2018); this priming effect is consistent with the reported priming effect induced by AgNPs (Yadav et al., 2024). The higher number of DEGs and their high expression (log 2 FC > 2) confirms that the sampling time right at the end of the stress period was a good choice as the more intense response has been fully evidenced for both treatments.

It is worth mentioning that a subset of genes is activated only by ML and NP60 (Figure 11); this indicates that specific bacterial metabolites are present in the NP corona, confirming their uniqueness in activating specific target genes (Plokhovska et al., 2025b). The increased expression of genes involved in transcription, translation, and DNA repair suggests that NP60 treatment enhances protein synthesis and triggers protective cellular mechanisms that directly enable recovery of stressed plants by maintaining metabolic activity and restoring growth after rewatering. Among the highest DEGs are iron superoxide dismutase (Supplementary Table S4a), a key enzyme in reactive oxygen species detoxification, and a plastidic ATP/ADP-transporter, which is essential for energy balance, exchanging ATP for ADP from chloroplasts to the cytosol. Ribosomal proteins such as 60S ribosomal protein L7-4 and 40S ribosomal protein S6 also show strong upregulation, indicating activation of the protein synthesis machinery. The expression of stress-related genes such as temperature-induced lipocalin and transcription factors such as CONSTANS interacting protein 2b suggests a broad cellular response aimed at improving stress tolerance and signaling.

A striking finding is that 50% of NP60-triggered genes are encoded in the chloroplast’s (42%) and the mitochondrion’s (8%) genomes, consistent with the reported penetration capacity of NP due to nano-size (Gutierrez Albanchez et al., 2018), reinforcing the notion that the unique response of NP is due to both the organic crown and its size, which allows it to reach double-membrane organelle DNA. This suggests that NPs may directly interact with organelles, enhancing metabolic and photosynthetic activity, while also possibly inducing systemic signaling cascades. These organelle-targeted responses contribute to maintaining photosynthetic efficiency and energy production, which are directly linked to the faster recovery and biomass accumulation observed after drought. In contrast, ML mainly affected nuclear-encoded stress and defense genes, highlighting the unique effect of formulating metabolites into NPs. While activation of photosynthesis by NPs has been reported and related to overexpression of nuclear encoded genes that improve photosynthetic performance (Tombuloglu et al., 2020), the activation of mitochondrion- and chloroplast-encoded genes by NPs has rarely been reported before and only in Arabidopsis with magnetic NPs (Pagano et al., 2022). Hence, the strong activation of chloroplast- and mitochondrion-encoded genes by AgNPs is reported here for the first time and highlights the strong potential of bacterial elicitor formulation.

These findings suggest that AgNPs coated with elicitors play a key role in stimulating the defense mechanisms of tomatoes. They enhance drought tolerance and help maintain normal plant functioning under water-stress conditions. Together, the priming effect, the transcriptional reprogramming, and the organelle-targeted gene activation explain the improved phenotypic resilience. This is evidenced by faster recovery, higher biomass, and restored photosynthetic performance after rewatering. The exploratory nature of the transcriptomic analysis to unveil differences between formulated and non-formulated elicitors has been carried out with a sufficiently high log 2 FC value to be significant and reliable (Everaert et al., 2017). Nevertheless, further analyses by RT-qPCR of specific subsets of genes should be carried out in the future to validate the proposed mechanisms. This is of special interest as the environmental safety for soil microbial biodiversity of these AgNPs has been evidenced (Plokhovska et al., 2025c). Based on the above, proposed mechanisms summarizing the effects of AgNPs on tomato adaptive responses to drought are shown in Figure 12.

In summary, AgNPs coated with metabolites from Pseudomonas N5.12 induce a faster and stronger response to drought than non-formulated metabolites, generating a distinct priming pattern and confirming our starting hypothesis. This response involves extensive gene reprogramming, improved ROS homeostasis, and activation of defensive and adaptive metabolic pathways, likely driven by both the bacterial elicitors and the nanoscale properties of AgNPs. Formulating bacterial metabolites in NPs thus represents a promising strategy to enhance plant drought resilience, promoting higher survival and faster recovery after rewatering. These findings highlight the potential of AgNPs for sustainable crop management and provide a basis for future research on NP-mediated modulation of plant stress responses.