Bacteria were previously identified by Rebollar et al. (2019). The bacteria C26G and C32I were recently described as new species of the Acinetobacter genus (Cevallos et al., 2022), while C23F was identified at the family level as Enterobacteraceae through 16S rRNA sequencing (Rebollar et al., 2018). All bacteria isolates were initially screened for their antifungal activity against the pathogen B. cinerea by the direct confrontation assay. For this, each bacteria was cultured on Luria-Bertani medium [LB] at 30°C for 24 h until an optical density of 0.6 [O.D_600nm]. The spore suspension of B. cinerea strain B05.10 [provided by Brigitte Mauch-Mani] was prepared as previously described (L’haridon et al., 2011). Dishes with potato dextrose agar [PDA, Sigma-Aldrich], were inoculated with 10 µl of each bacterial suspension at one end of the plate and 6 µl of the spore suspension of B. cinerea at the other end and incubated in the dark at 24°C for 7 days. The test was done in triplicate [n=3]. The fungal growth inhibition was evaluated by measuring the area of mycelium growth with ImageJ2 software [version 2].

Bacteria were cultivated in LB liquid medium, at 30 °C in the dark until an optical density of 0.6 [O.D_600nm]. Each bacterial liquid culture was centrifuged at 12,500 rpm for 15 min and the supernatant was filtered through a 0.22 μm Millipore membrane. For the inhibitory assay, Petri dishes [60 mm] containing PDA were supplemented with 50, 60, 70 and 80% [v/v] of bacterial filtrates [BFs]. As a control, dishes with PDA were supplemented with LB medium. A 5 mm sterile filter paper disk with 6 μl of suspension of B. cinerea [5 × 10⁴ spores ml⁻¹] spores was placed at the center of the dish and incubated at 24°C for 7 days in the dark. The test was done in triplicate [n=3]. Inhibition was evaluated by measuring the diameter of the mycelium on the dish. Images of growing B. cinerea mycelium were analyzed using the ImageJ2 analysis software [version 2].

Three drops of a spore suspension of the fungus [5×10⁴ spores ml⁻¹], including each bacteria filtrate at 80%, were placed on slides in a growth chamber at 24°C with >95% relative humidity for 24 h. Subsequently, the germinated spores were quantified in a Neubauer chamber. Staining of spores was performed (Romero-Contreras et al., 2019), and were observed under fluorescence microscope [Zeiss Axioskop 2, 40x].

To detect the inhibitory activity of bacterial volatile organic compounds [VOCs], 20 μl of a cell suspension of C23F, C26G, and C32I bacteria [0.6 O.D_600nm] was inoculated in the center of LB plates. Simultaneously, a suspension of B. cinerea spores [5×10⁴ spores ml^-1) was inoculated on PDA medium plates. Subsequently, the bottom of each plate [fungus and bacteria] were placed face to face and immediately sealed with parafilm. The plates were then incubated at 26°C for 7 days, together with a control, which consisted of PDA plates inoculated with the pathogen and uninoculated LB plates. The experiment was conducted in duplicate [n=2]. Inhibition was assessed by measuring the diameter of the mycelium on the plates. The growth halo of B. cinerea was measured using the ImageJ2 software [version 2].

Seeds of the ecotype Columbia-0 [Col-0, Nottingham Arabidopsis Stock Centre, Nottingham, UK] as wild type and the mutants of eds5-1 (Nawrath and Métraux, 1999), jar1 (Chassot et al., 2007) and ein3 (Roman et al., 1995), were germinated on 0.2X Murashige-Skoog [MS] medium supplemented with sucrose [0.5% w/v], and 1% [w/v] agar (Murashige and Skoog, 1962) for 7 days. The plants were then transplanted into germination trays containing a mixture of soil and vermiculite [3:1] and kept in a greenhouse at 22 ± 2°C and 60% humidity with 16 h/8 h light/dark cycles for one month with constant irrigation of non-sterile water. Then, 1 ml of each bacterial cell suspensions or bacterial filtrates was added [adjusted to 0.6 O.D₆₀₀] every five days for four weeks. For the pathogenicity assay, leaves of water-treated control or treated A. thaliana were infected with 6 µL of a suspension of B. cinerea spores [5×10⁴ spores ml⁻¹]. The inoculated plants were covered with plastic lids to maintain >95% relative humidity and transferred to a growth chamber at 22 ± 2°C and a 24 h dark cycle. After 72 hours post-infection [hpi], symptoms were evaluated. To determine the percentage of plant survival, we quantified the number of infected plants [disease incidence]. The lesion sizes were quantified using the ImageJ2 software [version 2]. The experiment was performed with a randomized design with 45 plants [three leaves per plant] per treatment.

Arabidopsis thaliana Col-0 plants were root-inoculated with Acinetobacter sp. C32I and foliar infected with the B. cinerea B05.10. After 6 hours post-infection [hpi], ten leaves per plant were collected and preserved in liquid nitrogen. Conditions were set for both dual and tripartite interactions of the plant with each microorganism. Total RNA extraction was performed from two independent experiments using Trizol, following the supplier instructions [Invitrogen]. RNA concentration and purity were measured with a NanoDrop spectrophotometer [Implen NP80, Thermo Fisher Scientific, USA]. Afterwards, 1 μg of total RNA was treated with DNAse [Thermo Scientific™, USA] to remove the genomic DNA, according to manufacturer of instructions. Library construction and sequencing were carried out by the Beijing Genomics Institute [BGI] Americas 2 using DNBSeq TM technology. The sequences are publicly available in the NCBI with number project PRJNA986187 and PRJNA1044848. To identify differentially expressed genes [DEGs], DESeq2 software in the Integrated Differential Expression Analysis MultiEXperiment [IDEAMEX] was used, with a FoldChange ≥ 2 or ≤ -2, and an adjusted p-value ≤ 0.05 (Jiménez-Jacinto et al., 2019). The DEG analysis was grouped through Gene Ontology [GO] analysis using PANTHER [v17.0]. GO term enrichment for the plant analysis was performed using the TAIR web tool [The Arabidopsis Information Resource [TAIR], https://www.arabidopsis.org/tools/go_term_enrichment.jsp, on www.arabidopsis.org, Feb 24, 2022], analyzed with the Fisher test and FDR correction. Non-redundant enriched terms were obtained by using REVIGO software (Supek et al., 2011). The graphs were created using the ggplot2 library with RStudio and heatmap libraries [v4.2.1].

To analyze the expression of genes, a third independent biological replica was performed. Leaves from root-inoculated plants with Acinetobacter sp. C32I and foliar infected with the B. cinerea B05.10 were collected. RNA isolation using TRI Reagent^® [Sigma-Aldrich, United States], were performed following the instructions of the manufacturer. RNA integrity was evaluated by using denaturing gel electrophoresis and measured using NanoDrop spectrophotometer [Implen NP80, Thermo Fisher Scientific, USA]. Afterwards, 1 μg of total RNA was treated with DNase I [Thermo Fisher Scientific, Inc., Waltham, MA, USA]. For cDNA synthesis, we used 1 μg of the total RNA, and then processed with a RevertAid H Minus RevertAid First Strand cDNA Synthesis kit [Thermo Fisher Scientific, Inc., Waltham, MA, USA], according to manufacturer’s instructions. Primers for the RT-qPCR gene expression analysis were: AOS [S_5’-ATCCAAAGATCTCCCGATCC-3’, AS_5’-GTGGATTCTCGGCGATAAAA-3’], ACS6 [S_5’-TGGTTGGTTAAAGGCCAAAG-3’, AS_5’-TGGTCCATATTCGCAAAACA-3’], PR1 [S_5’-TTCTTCCCTCGAAAGCTCAA-3’, AS_5’-AAGGCCCACCAGAGTGTATG-3’] and ZAT12 [S_5’-ATCAAGTCGACGGTGGATGT-3’, AS_5’-ACAAAGCGTCGTTGTTAGGC-3’]. To normalized transcript abundance two housekeeping genes were used. ACTIN [S_5’-TGCTTTGCCACATGCTATCC-3’, AS_5’-GACTTCAGGGCATCGGAAAC-3’] and CF150 [S_5’-CCGACAAGGAGAAGCTTAACAAGTT-3’, AS_5’-CGGCAGATTTGGATGGACCAGCAAG-3’]. The conditions of the reactions were performed according to Aragón et al. (2021).

For in vivo infection assays, blueberries at commercial maturity were harvested from organic fields that had not received any preharvest treatment with synthetic pesticides. Subsequently, the fruits were disinfected with a 2% NaClO solution for 2 minutes, followed by washing with tap water, and then left to dry at room temperature. Bacterial cultures [C23F, C26G, and C32I] were grown to a density of 0.6 [O.D_600nm]; subsequently, centrifuged at 12,500 rpm for 15 min. The supernatants were recovered and transferred through a 0.22 μm Millipore filter to obtain the bacterial filtrates [BF], which were then resuspended in physiological solution [NaCl 9 g/L], to achieve a concentration of 80% [v/v]. The cell pellets were dissolved in 3 ml of physiological solution. Blueberries were punctured to induce a lesion and were then independently sprayed with 3 ml of the bacterial solution and the BF. Subsequently, they were incubated at 25 ± 1°C and 95% relative humidity for 24 h. Three independent biological replicates, each one with 10 blueberries, were used for each treatment. Physiological solution served as the control. Following bacterial inoculation, blueberries were sprayed with a solution of B. cinerea ISIB-MMA/F-Bc01-S spores [1x10⁶ conidia/mL] and kept at 25 ± 1°C for 7 dpi. Infection was assessed by counting healthy and infected fruits (Chacón et al., 2022).

Solanum lycopersicum L. seeds were washed with a 3% solution of sodium hypochlorite [3 times] and absolute ethanol [1 time]. The seeds were germinated in hydrated vermiculite. After 7 days of growth, they were transferred to pots with a mixture of soil and vermiculite [3:1] and maintained in a greenhouse at 22 ± 2°C with 60% relative humidity under a 16 h/8 h light/dark cycle. Over five weeks, inoculations of the soil every three days were done with C32F, C26G, and C32I bacteria, grown in 0.2x MS liquid medium. Leaves were collected from plants pretreated with each of the bacteria and then infected with B. cinerea spores [5×10⁴ spores ml^-1]. The plants were maintained under >95% relative humidity at 24°C for three days in a plant growth chamber with a 16 h/8 h light/dark cycle. Lesion evaluation was conducted with ImageJ2 software [version 2]. Three biological replicates were performed, each with 10 leaves per treatment.

The data obtained from each test are presented in the figures as mean values [ ± SEM]. Statistical analyses were conducted using a One-way ANOVA followed by Tukey post-hoc test [p < 0.05] to identify significant differences for each experiment. The graphs were generated using R [https://posit.co/]. These analyses were based on results from three independent experiments.

Previously, we had demonstrated the inhibitory activity of bacteria isolated from amphibian skin against B. cinerea (Cevallos et al., 2022). Based on these results, two more promising strains belonging to the genus Acinetobacter C26G and C32I were selected, and the reproducibility and consistency of their antagonistic behavior was corroborated. In addition, a third bacterium, C23F, previously classified at the Enterobacteraceae family level, was included (Rebollar et al., 2018). The results revealed that all three bacteria exerted an antagonistic effect on the growth of B. cinerea ( Figure 1A ). In particular, the measurement of mycelial growth of the fungus showed that the three bacteria were able to significantly inhibit the growth of B. cinerea, compared to the uninoculated control, where the fungus invaded the entire plate ( Figure 1B ). These results indicated that bacteria from frog skin inhibited B. cinerea growth under in vitro assays.

Several BCAs have been shown to have the ability to secrete various compounds with antifungal activity that contribute to counteract the effect of pathogenic fungi (Ferreira-Saab et al., 2018). To investigate whether C23F, C26G, and C32I bacteria secrete compounds that interfere in the germination and growth process of B. cinerea, we confronted the bacterial filtrates [BF] with the pathogen. Our results revealed that the BF of each bacteria inhibited the growth of B. cinerea in a dose-dependent manner ( Figure 2A ). We observed that when the pathogen was grown on PDA medium supplemented with 50% and 60% BF, approximately 10% inhibition occurred for the three bacterial strains. However, at concentrations of 70 and 80% BF, a direct correlation in the inhibition of B. cinerea growth was observed ( Figure 2A ). To determine the effect of BFs on spore germination, we co-cultured a suspension of B. cinerea spores, along with 80% of BFs for 24 h ( Figure 2B ). The data showed that strain C23F influenced the germination rate of spores of 8.5%, followed by C26G and C32I with 11.33% and 13.76%, respectively, compared to the unfiltered control, where 100% germination was observed ( Figure 2C ). Our results indicate that BFs significantly inhibits the development and growth of B. cinerea, which highlights their potential use as BCAs.

To test the antagonistic activity of volatile organic compounds [VOCs] in inhibiting B. cinerea, a double-sided assay was performed. The data revealed an inhibitory effect of the VOCs produced by two frog-associated bacteria [C23F and C32I] on the growth of the pathogenic fungus, in contrast to the control without bacteria, where mycelial growth was observed throughout the entire surface of the plate ( Figure 3A ). A quantification of the growth area demonstrated that the bacteria C26G had the least inhibitory effect against B. cinerea, exhibiting similar growth as the untreated control and covering a total area of approximately 15 cm². In contrast, VOCs from bacteria C23F and C32I displayed inhibitory effects on the fungus, resulting in growth areas of 6.58 cm² and 3.17 cm², respectively ( Figure 3B ). Interestingly, our findings revealed that, in addition to the release of diffusible organic compounds, volatile organic compounds are also produced, which might play a role in restraining B. cinerea.

Several biocontrol agents induce plant defense systems, which confer a protection system against pathogenic fungi (Wang et al., 2021). To analyze the effect of bacteria C23F, C26 and C32I directly on the plant, we inoculated the soil, in which A. thaliana plants were growing, with a suspension of cells or BFs and evaluated their effect against B. cinerea ( Figure 4 ). Upon infecting A. thaliana leaves with B. cinerea, we observed that the lesions caused by the pathogen on the plants pre-treated with the bacterial cells or BFs were significantly smaller compared to the non-inoculated control after three days post-infection [3 dpi], ( Figure 4A, B ). Next, we measured the incidence of infection on the leaves and observed that plants inoculated with C23F cells and filtrates showed a disease incidence of 46% and 71% respectively, and plants treated with C26G showed 48% and 66% incidence, followed of plant inoculated with C32I with 44% and 65%, disease incidence, compared with the controls that showed a 98% incidence ( Figure 4C ). These results indicate that both the bacteria and the BF induce defense responses in the plant when present in the soil, protecting leaves against B. cinerea infection.

To analyze the effect of bacteria to protect commercial crops such as blueberries during post-harvest, we performed an in vivo test using either cells or BFs of C23F, C26G and C32I, and inoculating with B. cinerea ISIB-MMA/F-Bc01-S (Chacón et al., 2022). We observed that blueberries fruits, treated with bacteria and infected with the fungus, showed a lower lesion grade compared to the non-inoculated control ( Figure 5A ). In untreated condition, 95% of the fruits developed the disease, while the fruits treated with cells bacteria, showed incidences of 58%, 49%, and 26% for bacteria C23F, C26G, and C32I, respectively ( Figure 5B ). For the treatment of blueberries inoculated with BF, the pathogen incidence was similar to the treatment with bacterial cells. The control treatment showed an incidence of 99%, while the BF treatments showed incidences of less than 80% ( Figure 5C ). Additionally, we analyzed the protective effects of each bacterium against B. cinerea using leaves from tomato plants inoculated with the respective bacterium. Our results showed a reduction in lesions in plants treated with each bacterium, compared to the control group where lesions were significantly higher ( Supplementary Figure 1 ). Overall, the results indicate that C23F, C26G, and C32I and their filtrates [BF] protect plants and fruits against B. cinerea, suggesting a potential for biological control applications.

To investigate the transcriptional changes in the plant upon inoculation with the C32I bacterium and the pathogen B. cinerea, we conducted both bipartite and tripartite interactions among the organisms under greenhouse conditions. We observed a difference in the number of up-regulated and down-regulated genes ( Figure 6 ; Supplementary Data 1 ). We identified that 6,246 and 31 DEGs up-regulated, while 34,297, and 43 DEGs were down-regulated for each treatments, respectively ( Figure 6A ; Supplementary Data 2 ). Additionally, we explored whether these DEGs were shared among the treatments: we did not find shared DEGs for all the treatments and only few were shared when two treatments were compared ( Figure 6B ).

To assess the potential functions of the identified DEGs, these genes were assigned to significant annotations by Gene Ontology [GO] term enrichment analysis and were classified with respect to biological processes ( Supplementary Data 3 ). All DEGs were classified into 22 biological processes, including the biological processes involved in interspecies interaction between organisms, response to biotic stimulus, response jasmonic acid, response to fungus, among others ( Figure 6C ). The results suggests that plants in the presence of C32I and B. cinerea, triggers the expression of various functional groups that differ with respect to dual interaction.

To comprehend changes in the expression levels of genes associated with hormonal pathways, we performed a clustering of genes categorized into salicylic acid [SA]-, jasmonic acid [JA]-, ethylene [ET]-response to pathogen and others related to transcription factors. We analyzed the accumulation of transcripts in A. thaliana during the dual and tripartite interactions between the Acinetobacter sp. C32I and B. cinerea [6 hpi] ( Figure 7 ). In the initial analysis, we compared the control condition [non-inoculated] with the plant exposed only to B. cinerea. In the ethylene [ET] pathway, we detected a down-regulation of genes, that participate in the resistance to B. cinerea. However, in the dual plant-bacteria interaction, we observed an induction of ERF4, ACS6, CORI3 and PR4 ET-related genes and down-regulation of the receptors EIN2 and EIN3 (Wójcikowska and Gaj, 2017; Jiang et al., 2023; Shin et al., 2023; Tiwari et al., 2023). Regarding tripartite interaction, the genes showed the same up-regulation, except for PR4 gene. Concerning jasmonic acid [JA] pathway, we observed a down-regulation of the responsive genes in presence of fungus. On the other hand, in the interaction of the plant with C32I, is observed up-regulation of the PDF1.2, TAT3, EP1, AOS, AOC3, AOC2, BBD1, and OC4 genes (Stenzel et al., 2003; You et al., 2010; Omidvar et al., 2021; Mathan et al., 2022), and these expression levels were maintained in the tripartite interaction. Furthermore, in the salicylic acid [SA] pathway, PR1 gene was induced during the infection with fungus and during the tripartite interaction.

Additionally, we analyzed a group of genes that have been described to respond to the defense against pathogens. Notably, during plant-bacteria interaction condition, there was an up-regulation of the genes DHAR1 [DEHYDROASCORBATE REDUCTASE] and PME17 and PME41 [PECTIN METHYLESTERASE], the latter recently identified in defense against B. cinerea (Del Corpo et al., 2020; Aragón et al., 2021). The expression levels were similar during the tripartite interaction to genes, including UPOX [UPREGULATED BY OXIDATIVE] and ZAT12 [ZINC FINGER OF ARABIDOPSIS] that participate in oxidative stress (Davletova et al., 2005; Vaahtera et al., 2014). Taken together, our results demonstrate that in the presence of the C32I bacterium, the transcriptional response of the plant is modified, indicating an early activation of the plant defense system, that is maintained in during the tripartite interaction, suggesting that the bacteria induce the plant immunity against B. cinerea. Finally, we confirmed the expression patterns found with RNA-seq with the expression of four genes selected from each hormone pathway: ACS6, AOS, PR1, and ZAT12 through qRT-PCR ( Figure 7B ).

To elucidate the signal transduction pathways involved in the defense response induced by Acinetobacter sp. C32I against B. cinerea, the wild-type plant and mutants involved in the JA [jar1], ET [ein3], and SA [eds5-1] signaling pathways were inoculated with the bacterium C32I and infected with B. cinerea. The data demonstrated variation in the lesion size for each inoculated and non-inoculated treatment was detected ( Figure 8A ). An analysis of the lesion area in the Col-0 [WT] and eds5-1 [SA] mutant showed a reduction in the lesion area caused by B. cinerea at 3dpi occurred when the plant was inoculated with the bacteria. However, the defense response in the jar1 [JA] and ein3 [ET] mutants showed the same lesion area, in both conditions ( Figure 8B ). Overall, the results suggested resistance of C32I-inoculated A. thaliana against B. cinerea, dependent on JA and ET signaling, rather than SA signaling.

We have observed that bacterium C32I induces transcriptional changes in the plant, manifesting differentially expressed genes [DEGs], both up-regulated and down-regulated, and some of the are related to defense systems. When comparing our results with other transcriptional reports of previously characterized biocontrol agents, we noticed a difference in the expressed genes in the presence of C32I, compared to Bacillus valezensis FBZ42, Pseudomonas fluorescences SS01, and Burkholderia sp. SSG ( Figure 9 ) (van de Mortel et al., 2012; Tzipilevich et al., 2021; Kong et al., 2022). When we compared the transcriptional changes induced by C32I only few genes are shared with the other BCAs, suggesting that, even though all the analyzed BCAs have a biocontrol activity, transcriptional regulation by C32I triggers defense induction mechanisms distinct from other microorganisms.