Introduction

Endophytes are microorganisms that colonize internal plant tissues for all or part of their lifetime (Hardoim et al. 2015). These communities likely reside in all plants and are strongly influenced by plant characteristics, such as host species, age and health status, and environmental conditions. Endophytic communities and their hosts strongly interact with one another and also respond to biotic and abiotic stresses, and as a result, endophytes are involved with most aspects of plant fitness (Rodriguez et al. 2009), including resistance to pathogens (Raghavendra and Newcombe 2013). Non-pathogenic endophytes can be exploited to antagonize pathogens via hyperparasitism, competition or antibiosis, by counteracting virulence factors, or by producing specific elicitors capable of triggering plant defences and improving resistance to disease (Compant et al. 2010; Hardoim et al. 2015). For grapevine, the composition and ecology of the endophytic communities have been extensively investigated (Bulgari et al. 2011; Compant et al. 2011). Endophytic organisms in grapevine have been isolated with both antibacterial and antifungal activities (Andreolli et al. 2016), suggesting the possibility of exploiting these microorganisms to control grapevine diseases. Bacillus subtilis and Pseudomonas spp. increase the resistance to Botrytis cinerea in grapevine (Trotel-Aziz et al. 2008), and Streptomyces atratus significantly controls downy mildew (Liang et al. 2016). In addition to bacteria, several beneficial root-colonizing fungi are reportedly effective biological control agents (BCA) against grapevine diseases (Legler et al. 2016). Trichoderma spp. are actual commercial biocontrol agents and their effects were investigated against downy mildew (Palmieri et al. 2012) and esca disease (Pertot et al. 2016). However, not all Trichoderma species or strains confer plant protection, and the defence responses can be affected by plant genotype and influenced by environmental and stress conditions (Babalola 2010; Roatti et al. 2013). Agrobacterium vitis is the primary bacterial pathogen responsible for grapevine crown gall, and to date the control of the disease has relied primarily on using healthy propagative material and resistant rootstock. In this study, we show that selected grapevine endophytic species can limit crown gall symptoms (CGS) in high-value susceptible grapevine. Additionally, commercial BCAs were evaluated with the selected grapevine endophytes to compare their effects as biocontrol agents against A. vitis.

Materials and methods

Isolation of grapevine endophytes

Ten asymptomatic grapevine plants (Vitis vinifera cv. Corvina and cv. Corvinone; over ten years old) growing in vineyards showing CGS located in the Valpolicella area (Veneto, Italy) were selected for endophyte isolation. With the aim to obtain a representative variety of endophytes colonizing the entire plant, the different tissues of canes, shoots, petioles and leaves (Burruano et al. 2008; approximately five samples for each type of tissue and plant) were collected in autumn 2012. Grapevine canes were disinfected by treatment with 15% H2O2 for 6 min, absolute ethanol for 1 min and washing three times with sterile water for 1 min. Grapevine shoots, petioles and leaves were disinfected according to the method described by Musetti et al. (2006). Small pieces of surface-disinfected tissues were plated in nutrient yeast dextrose agar (NYDA) and incubated at 30 °C to recover the bacterial endophytes in the plant tissue. Colonies grown in these Petri plates were collected from the plates, and then pure cultures were obtained by repeated streaking onto NYDA medium. To assess the presence of Agrobacterium vitis in these collected tissues, selective RS medium was used (Roy and Sasser 1983). To isolate endophytic fungi, plant tissues were treated as above, plated on potato dextrose agar (PDA) supplemented with streptomycin sulphate (50 mg l−1) and incubated at 25 °C for five days in the dark. Fungal colonies grown on Petri plates were collected and purified until axenic fungal cultures were obtained. Before taxonomic identification, a visual inspection prevented the selection of identical isolates arising from the same tissues. The commercial biocontrol agents Trichoderma asperellum T1 (Trifender®; AGRAnova UG, Bonn, Germany), T. harzianum T22 (Trianum® G; Koppert, Lansingerland, the Netherlands), T. harzianum TH01 and Bacillus subtilis SR63 (both isolated by Micosat® F; CCS AOSTA, Aosta, Italy) were also tested as potential biocontrol agents against crown gall disease. These BCAs were isolated in purity from their respective commercial products to avoid the interference with additives and biological contaminants.

Characterization of grapevine endophytes

Endophytic bacterial isolates were identified based on 16S rDNA sequences. The DNA of the endophytic samples was extracted using an alkaline lysis solution (Baldan et al. 2014). One microliter of DNA from a 1:100 dilution was used for amplification using universal primers targeting the 16S rDNA gene (63F-1387R primers) (Marchesi et al. 1998). PCR amplifications were conducted in a volume of 25 μl, with 1 × PCR buffer (100 mM Tris–HCl, pH 9.0, 15 mM MgCl2 and 500 mM KCl), 0.2 mM dNTPs, 0.2 μM each primer and 0.5 U of Hot Start Taq DNA polymerase (PRIMM; Milan, Italy). The PCR-derived fragments were resolved in 2% agarose/TAE gels and visualized under UV light via GelRed™ Nucleic Acid Stain (Biotium, Fremont, USA). All amplification products were subjected to ExoSAP (Thermo Fischer Scientific, Waltham, USA) treatment, sequenced and then compared with the 16S rDNA available in the NCBI GenBank database (https://www.ncbi.nlm.nih.gov/genbank/) using BLASTn to identify the corresponding bacteria. The 16S nucleotide sequences obtained from Curtobacterium spp. isolates, subsequently characterized by in vitro screening as potential biocontrol agents, were compared with sequences of endophytic species of Curtobacterium isolated from grapevine (C. flaccumfaciens and C. herbarum) or other plant species (C. luteum isolated from rice and sugarcane). Multiple sequence alignment was performed using MEGA 7.0 software (Kumar et al. 2016), and a homology tree based on pairwise nucleotide sequence divergences was constructed using the neighbour-joining method with 1000 bootstraps. The 16S nucleotide sequence from a sister genus Frigoribacterium was used as the out-group (Chase et al. 2016).

The isolated fungal species were identified according to morphological features (Barnett and Hunter 1998; Watanabe 2011). One endophytic isolate was identified in the genus Acremonium (hereafter named AVR67) and was molecularly characterized using the primer pairs ITS4/ITS5, NS1/NS4 and NL1/NL4 (White et al. 1990; O’Donnell 1993) targeting the ITS, 18S and 28S rRNA regions, respectively, followed by a homology search using BLAST to identify the species. The obtained 16S and ITS rRNA gene sequences of Curtobacterium isolates and AVR67, respectively, were used to design specific primers for both endophytes as follow: Cf16S-F 5′ GGTGGTGGAAAGATTTTTTG 3′ and Cf16S-R 5′ CTAATCCTGTTCGCTCCCCAT 3′ for Curtobacterium, and avr67ITS-F 5′ GTATTTTCTGAGTGGCATAAGCA 3′ and ITS4 primer for Acremonium.

In vitro screening for antagonistic activity against A. vitis

Three tumorigenic strains of A. vitis (provided by the Centre for Research in Viticulture, CREA, Conegliano, Italy; named 3NT4, 3NT11 and 3NT12) were used in trial experiments. Potential BCA selection was based on the ability to inhibit the in vitro growth of A. vitis on agar plates according to Bechard et al. (1998). Bacterial endophytes, B. subtilis and A. vitis, were cultured in a nutrient yeast dextrose broth (NYDB) at 30 °C for 24 h. For each bacterium, a cell suspension was prepared and adjusted to a density of 108 CFU ml−1. One hundred microliters of each potential BCA bacterial suspension was placed in a well (6 mm in diameter) positioned at the centre of 94 mm Petri dishes containing NYDA medium and left to dry in a laminar flow cabinet. The plates were incubated in an upright position at 30 °C overnight, and subsequently a cell suspension (250 μl) of each strain of A. vitis was spread on these plates and incubated as previously described. A boiled bacterial suspension for each BCA was used as a control. Five replicates were used for each combination of BCA isolate and A. vitis. After incubation, the areas of the growth inhibition zones around the central well were measured, and the calculated radius, compared with that of the control, was used to estimate the inhibitory activity.

For fungal endophytes, to evaluate antagonism and competition for the substrate, a suspension of A. vitis and a plug of fresh mycelium (5 mm in diameter) of each fungal isolate were simultaneously inoculated at the opposite positions on PDA in Petri plates. The plates were incubated at 30 °C in the dark and observed daily until the organisms were in contact or showed inhibition. The index of antagonism (IA) between these microorganisms and the bacterial pathogen was assigned and calculated according to Wicklow et al. (1980) and modified as follows: the reaction type in which the fungus contacted and stopped the growth of the bacterial colony was assigned a numerical value of 1, whereas the reaction in which the fungus grew over the bacterial colony was assigned a numerical value of 2.

In planta biocontrol assays

Three bacterial isolates showing in vitro antagonistic activity (two Curtobacterium sp. named CVR5 and CVR38 and B. subtilis SR63) and four fungal isolates (AVR67, T. asperellum T1, and T. harzianum TH01 and T22) were used for in planta biocontrol assays. A bacterial suspension for each BCA was prepared as described above and adjusted to 108 CFU ml−1 with 10 mM phosphate-buffered saline (PBS) pH 7.0 and 0.03% Tween-20. For fungal inoculum, the conidia suspension was obtained by fungal isolate growth on PDA for ten days at 25 °C in the dark. Collected conidia were resuspended in PBS at the density of 107 CFU ml−1.

Biocontrol assays were performed on Vitis vinifera cv. Corvinone grafted onto V. berlandieri × V. rupestris 110 Richter (110R), a rootstock well known for susceptibility to crown gall (Burr et al. 1998). Each trial consisted of fifteen plants that were analysed by PCR to verify the absence of Agrobacterium vitis. The analysis was conducted with primer pairs VIRD59F26/VIRD59R122 and VIRD62F23/VIRD62R135 according to Bini et al. (2008). Plants were transplanted into large pots containing soil (70% sandy loam and 30% peat; pH 7.2) and grown in a vineyard for four weeks before the BCA trials were conducted. During the first year of study (2013), trials were performed with each biocontrol agent. In the second year (2014), the effective BCAs were tested on a new set of plants. Specifically, T. harzianum strains were not assayed because of the low efficacy observed in the 2013 trials. The endophytic bacterial isolates and B. subtilis were inoculated by mechanical drilling into grapevine rootstocks (with a drill bit of 1 mm) at 5 cm from the ground. Twenty-five microliters of the culture suspension or sterile PBS (control) was inoculated into each lesion, which was then wrapped with Parafilm®. Because Acremonium is an endophyte, the AVR67 fungal isolate was inoculated into the rootstock as an endophytic bacterium. Because the recommended application of commercial products based on Trichoderma spp. requires soil inoculation, the conidial suspension was inoculated and mixed into pots to obtain a final concentration in soil of 106 CFU g−1. To verify the root colonization by Trichoderma spp., root fragments (100 per plant, 0.5 cm in length) from three grapevine plants for each treatment (T. asperellum T1, T. harzianum T22 and TH01) were analysed as previously described (Ferrigo et al. 2014). Root fragments from untreated plants were examined as controls.

Four weeks after the BCA treatments, grapevine rootstocks were inoculated with 25 μl of a mixed cell suspension of the three A. vitis strains at a final concentration of 108 CFU ml−1. Inoculation was performed 3 cm above the BCA treatment to avoid interference among BCA, pathogen and wound. Inoculation of sterile PBS was used as the control. The new lesions were wrapped in Parafilm®, and the plants were left for one week in a greenhouse and then transferred into a nursery until gall formation (16 weeks). The symptom incidence was estimated for each treatment as the number of symptomatic plants among 15 plants. Gall size was recorded as the average of the diameter perpendicular to the stem.

Quantification of Agrobacterium vitis by quantitative PCR (qPCR)

To determine the amount of A. vitis in grapevine treated with CVR5, CVR38, AVR67, T. asperellum T1 and B. subtilis SR63, qPCR was performed on DNA samples obtained from three symptomatic and asymptomatic grapevine rootstocks analysed during the first year of symptom evaluation. The analysis was also conducted on untreated grapevine inoculated with A. vitis as a control. Rootstock tissues collected 3 cm above the site of A. vitis inoculation were removed with a scalpel and thoroughly washed before DNA extraction, according to Doyle (1990). qPCR was performed using a 7500 real-time PCR detection system (Applied Biosystems, Foster City, USA). Fifty nanograms of total DNA was amplified in a 25 μl reaction containing SYBR® Green PCR Master Mix (Applied Biosystems; Foster City, USA) with two primer pairs, VIRD59F26/VIRD59R and VIRD62F23/VIRD62R, at a final concentration of 200 nM each in a multiplex qPCR (Bini et al. 2008). All reactions were conducted in triplicate. To calculate the reaction efficiency of the PCR assays, standard curves were generated by plotting threshold cycle values (Ct values) obtained from amplification of five-fold dilutions against the logarithm of the DNA quantities. Slope (M), linear correlation coefficient of standard curve (R2) and amplification efficiency (E) for each primer pair were as follow: VIRD59F26/VIRD59R (M = −3.1867; R2 = 0.99; E = 105.9%) and VIRD62F23/VIRD62R135 (M = −3.2147; R2 = 0.96; E = 104.7%).

Quantification of grapevine endophytes and B. subtilis by qPCR

qPCR analysis was performed to determine the amount of inoculated antagonists after 16 weeks. To quantify the grapevine endophytes and B. subtilis after treatment, three symptomatic and asymptomatic grapevine rootstocks randomly selected from those treated with CVR5, CVR38, AVR67 and B. subtilis SR63 were analysed during the first year of symptom evaluation. As a control, the same analysis was conducted on untreated grapevine inoculated with A. vitis. For qPCR analysis, 50 ng of total DNA was amplified in a 25 μl reaction containing SYBR® Green PCR Master Mix supplemented with 200 nM primers Cf16S-F/Cf16S-R for Curtobacterium sp. (annealing at 52 °C), avr67ITS-F/ITS4 for Acremonium sp. (annealing at 55 °C), and ytcP-F/ytcP-R for B. subtilis (annealing at 50 °C; Kwon et al. 2009). The amplification conditions consisted of 40 cycles as follow: 95 °C for 15 s, annealing at the primer pair-specific temperature for 30 s, and extension at 72 °C for 35 s, followed by a final extension step of 4 min at 72 °C. All reactions were performed in triplicate. Details of standard curves analysed for primer efficiency were as follow: Cf16S-F/Cf16S-R (M = −3.3645; R2 = 0.93; E = 98.3%), avr67ITS-F/ITS4 (M = −3.4701; R2 = 0.97; E = 94.2%), and ytcP-F/ytcP-R (M = −3.2901; R2 = 0.99; E = 101.4%). Collected data were analysed using the Applied Biosystem 7500 1.4 software.

Statistical analyses

Statistical analyses were performed using χ2 tests for comparisons of disease incidence and root colonization by Trichoderma spp. between treated and untreated plants. Because of the unequal numbers of galls among treated plants, Welch’s unequal variances t test was used for gall size comparisons. The differences in in vitro antagonism were analysed with two-way ANOVA followed by Tukey’s HSD post-hoc test examining the influence of different BCAs on pathogen inhibition compared with the control. The differences in microorganism DNA content (Curtobacterium sp., B. subtilis, Acremonium sp. and A. vitis) between non-treated and treated plants were analysed with Kruskal–Wallis one-way ANOVA followed by Dunn’s test for nonparametric multiple comparisons. Correlations between DNA level of BCAs and that of A. vitis were analysed by Spearman’s rank correlation. Analyses were performed with the XlStat 2016 software package (Addinsoft, NY, USA).

Results

Isolation and characterization of grapevine endophytes

Endophytic bacteria were isolated from asymptomatic grapevines in vineyards with a high incidence of crown gall and identified using 16S rDNA sequencing. The most prevalent bacteria were the genera Bacillus and Pantoea, in addition to other less abundant genera, such as Curtobacterium, Pseudomonas, Serratia, Sphingomonas and Stenotrophomonas. No Agrobacterium vitis colonies were isolated from these plants. The 16S nucleotide sequences of Curtobacterium spp. were deposited in GenBank under the accession numbers KY458624 and KY458630 for isolates CVR5 and CVR38, respectively. The phylogenetic tree constructed with the 16S ribosomal sequences of endophytic Curtobacterium isolates identified at the species level clustered the two Curtobacterium isolates CVR5 and CVR38 into a clade including C. flaccumfaciens (Fig. 1), indicating a high identity at the sequence level with this species.

Fig. 1
figure 1

Neighbour-joining tree constructed using pairwise nucleotide divergences in 16S rRNA gene sequences of Curtobacterium sp. VR5 and VR38 isolated in this study (bold) and sequences of endophytic Curtobacterium species from GenBank. The percentage of replicate trees in which the associated strains clustered together in the bootstrap test (1000 replicates) is shown next to the branches. The distances are in the units of the number of base substitutions per site. The GenBank accessions are presented

Full size image

Moreover, several fungi were recovered and morphologically identified as different genera, including Acremonium, Alternaria, Botrytis, Cladosporium, Epicoccum, Fusarium, Geotrichum, Hansfordia, Nigrospora, Penicillium, Phialophora, Pithomyces, Sordaria and Ulocladium. Among these fungi, the genus Alternaria was the most prevalent, with an amount that was approximately 30% of all isolates. The endophytic Acremonium isolate was further characterized on the basis of ITS, 18S and 28S rRNA gene sequences (Accession Numbers KY488147, KY488148 and KY864401, respectively), and this analysis revealed 99% identity with sequences deposited in GenBank for A. alternatum and A. sclerotigenum.

Antagonistic activity of potential biocontrol agents

Bacterial and fungal grapevine endophytes were assayed for their antagonistic activity against Agrobacterium vitis. Regarding endophytic bacteria, only two isolates in the genus Curtobacterium, named CVR5 and CVR38, inhibited the in vitro growth of A. vitis. CVR5 was more effective against the three strains of A. vitis (3T4, 3T11 and 3T12) than CVR38. Similar antagonistic activity against A. vitis was also demonstrated with B. subtilis strain BS63 isolated from the commercial product Micosat F (Table 1). Boiled bacterial suspensions did not display any inhibitory activity (data not shown). Significant differences in effect of different BCAs were observed (F3,48 = 1146.22, p < 0.0001), also revealing the influence of A. vitis strains (F2,48 = 105.98, p < 0.0001) and their interaction (F6,48 = 15.09, p < 0.0001).

Table 1 Inhibitory activity of Curtobacterium sp. and Bacillus subtilis against Agrobacterium vitis strains
Full size table

By contrast, none of the fungal endophytes consistently inhibited the in vitro growth of A. vitis (data not shown). However, because of the potential of Acremonium species as biocontrol agents with a different mode of action (Jäschke et al. 2010; Lo Piccolo et al. 2015), AVR67 was evaluated for ability to antagonize crown gall disease. The potential antagonistic activity against A. vitis of Trichoderma spp. isolated from commercial products was also compared. Trichoderma asperellum T1 (AGRAnova), T. harzianum TH01 (Micosat F) and T. harzianum T22 (Trianum G) competed with A. vitis, hindering pathogen growth. In particular, T. asperellum T1 had the highest IA value, showing the best performance as competitor against A. vitis (Table 2).

Table 2 In vitro interactions of AVR67 and Trichoderma spp. against Agrobacterium vitis strains
Full size table

Quantification of A. vitis in grapevines after BCA treatments

To verify that reductions in symptoms were related to different levels of pathogen colonization, the A. vitis content was monitored in inoculated grapevines treated with different BCAs and compared with untreated ones. Of the 105 grapevine plants inoculated with A. vitis and treated with BCAs, only 28 plants showed symptoms (Table 3). However, all symptomatic and asymptomatic plants were positive for A. vitis when analysed with PCR (data not shown). In the first year of trials, three symptomatic rootstocks randomly chosen for each BCA treatment were analysed for the amount of A. vitis DNA by qPCR and compared with untreated controls. Symptomatic BCA-treated plants were less infected when compared with the symptomatic untreated plants (χ 25  = 16.339, p = 0.012), with the lowest amount of pathogen DNA obtained with CVR38 and AVR67 inoculation (Fig. 2). Regarding asymptomatic plants, a similar amount of pathogen was detected independent of treatment (χ 25  = 7.427, p = 0.191; data not shown). When comparing DNA content of A. vitis and that of CVR, AVR67 or B. subtilis, a significant negative correlation was found for some microorganisms. Specifically, an increase in the antagonist population corresponded to a decrease in the amount of A. vitis for CVR5 (r = 0.861, p = 0.017) and B. subtilis S63 (r = 0.697, p = 0.049), whereas no significant correlation was found in the interaction between A. vitis and CVR38 (r = 0.595, p = 0.103) or AVR67 (r = 0.618, p = 0.165).

Table 3 Symptom incidence and gall size in two years of in planta trials
Full size table
Fig. 2
figure 2

Levels of Agrobacterium vitis DNA in treated grapevine plants. Data were normalized with respect to the levels detected in not treated symptomatic plants (NT, grey bar) set at 100%. White bars represent level of A. vitis DNA (+ SE) in symptomatic treated samples. Asterisks indicate significant differences (*p ≤ 0.05; **p ≤ 0.01) compared with the untreated controls according to Kruskal–Wallis and Dunn’s tests. (NT not treated; Bs Bacillus subtilis; Ta Trichoderma asperellum)

Full size image

Quantification of biocontrol agents in grapevine after BCA treatments

To verify that the disease reduction was correlated with different levels of BCA colonization, the amount of Curtobacterium sp., B. subtilis and Acremonium sp. was compared in infected grapevine tissues between treated and untreated plants (Fig. 3). The analysis was conducted during the first year of trials in grapevine plants showing or not showing CGS. The level of colonization by the potential BCAs was compared with the background level of each species commonly found as usual inhabitants of grapevine tissues. The quantification of Curtobacterium sp. in CVR5-treated plants \(\left( {\chi_{2}^{2} = 7.231,\;\;p = 0.0 2 7} \right)\) showed no differences in symptomatic grapevines compared with the control (p = 0.18) but a significant increase up to 190% in asymptomatic ones (p = 0.007). By contrast, no significant variation was found in the level of Curtobacterium sp. after the CVR38 treatment (χ 22  = 4.863, p = 0.088) (Fig. 3). For B. subtilis, the quantification of bacteria in treated plants (χ 22  = 7.2, p = 0.027) showed an increase of 115% in asymptomatic plants (p = 0.007) compared with untreated controls, whereas no difference was observed among symptomatic ones (p = 0.18). Regarding Acremonium, the content was always higher in AVR67-treated plants than that in control grapevines \(\left( {\chi_{2}^{2} = 7.448,\;\;p = 0.024} \right).\) However, no significant difference was found between symptomatic and asymptomatic plants (p = 0.172) (Fig. 3). For the Trichoderma strains, root colonization increased during the first-year trial by 15 ± 1, 19 ± 1, and 24 ± 2% for TH01, T22, and T1, respectively, compared with the controls, and the differences were significant for T22 (χ 21  = 9.800, p = 0.002) and T1 (χ 21  = 20:836, p = 0.0001)

Fig. 3
figure 3

Levels of biological control agent DNA in treated grapevine plants. Data were analysed separately for each antagonistic species and were normalized with respect to the levels detected in not treated symptomatic plants set at 100%. Bars represent DNA level (+ SE) of Curtobacteriun sp. (a), Bacillus subtilis (b) and Acremonium sp. (c) in different samples normalized with respect to symptomatic not treated plants (NT, grey bar). Dark grey and white bars represent DNA levels found in symptomatic (S) and asymptomatic (A) treated plants, respectively. Asterisks indicate significant differences (p ≤ 0.05) compared with the untreated controls according to Kruskal–Wallis and Dunn’s tests. (NT not treated; Bs Bacillus subtilis)

Full size image

Reduction of crown gall symptoms in grapevine after BCA treatments

The inoculation of grapevine rootstock with CVR5 showed significant two-year average reductions of 75 and 76.5% in symptom incidence and gall size, respectively, compared with the control plants (Fig. 3), and the results were comparable for inoculation with B. subtilis SR63 (compared with controls, 81% lower incidence and 62% smaller gall size). By contrast, CVR38 showed a significant reduction in incidence of 89% (2014), whereas an average reduction of 63.5% in gall size was observed during both years. For the fungal strain AVR67, the efficacy of the reduction of CGS was 75.0 and 63.5% on average for gall incidence and size, respectively. Other known biocontrol agents, such as T. asperellum and T. harzianum, were assayed for their ability to reduce symptoms, and T. asperellum T1 was the most effective (75% lower incidence; 53.5% smaller gall size). By contrast, the T. harzianum strains (T22 and TH01) assayed during the first-year trial reduced disease incidence but did not reduce gall size (Table 3).

Discussion

Endophytic organisms can greatly influence the health of their host and recent evidence demonstrates their potential biocontrol activity against pathogens (Busby et al. 2016; Lòpez-Fernàndez et al. 2016). Specific microorganisms associated with a specific plant species provide the most effective biocontrol activity, likely due to adaptation to the specific host environment (Adhikari et al. 2001). With the aim to identify the best-adapted microorganisms to control crown gall in grapevine, endophytes were isolated from asymptomatic plants grown in vineyards highly symptomatic for crown gall. These endophytes and bacterial and fungal isolates from commercial products known for their antagonistic activity were assayed against A. vitis. Among the isolated bacterial endophytes, we identified some species that showed high homology with bacterial populations isolated in grape tissues by other authors (West et al. 2010; Andreolli et al. 2016). The relative abundance of some genera (e.g., Bacillus, Pantoea) found on ten-year-old grapevines analysed in this study is consistent with the bacterial populations found on 15-year-old plants (Andreolli et al. 2016). Note that the type of tissue analysed (Compant et al. 2011; Martins et al. 2013), cultivar (Pancher et al. 2012), plant age (Andreolli et al. 2016), grapevine physiological state (Bulgari et al. 2011), date of sample collection (Bulgari et al. 2014) and pest management (Campisano et al. 2014) can affect the overall endophytic community. Moreover, the isolation of endophytes can also be influenced by culturing techniques (Pei et al. 2017). Thus, the community isolated in the present work cannot be considered fully representative of the entire endophytic community in the analysed grapevines.

The two isolates of Curtobacterium sp. that showed in vitro antagonism against A. vitis were tested for their ability to reduce CGS in grapevine. Data were collected during trials conducted for two years, and a significant reduction in gall development and disease incidence was observed for CVR5, whereas the reduction in disease incidence in CVR38 plants was significant only in the second-year trial. To explain the divergent behaviours between the strains, environmental factors could have had different effects on the ability of these strains to colonize plant tissues and develop galls. Curtobacterium species are well-known endophytes of grapevine (Baldan et al. 2014), and our isolates presented high homology with C. flaccumfaciens, suggesting a likely affinity with this species. Curtobacterium flaccumfaciens is more abundant in grapevines without crown galls than in symptomatic ones (Faist et al. 2016), suggesting a correlation between CGS amelioration and C. flaccumfaciens population density, which is consistent with our results. Moreover, C. flaccumfaciens can reduce symptoms caused by Xylella fastidiosa (Lacava et al. 2007) with the induction of systemic resistance or alternatively, by direct antagonism and interactions between the bacteria. Additionally, C. flaccumfaciens exhibits in vitro inhibitory activity against the pathogen Agrobacterium tumefaciens and reduces crown gall disease in rose, kalanchoe and squash fruits, likely because of bioactive substances (Tolba and Soliman 2013). Antibiosis could be an important component in the reduction of CGS, as suggested by the in vitro experiments in which the two strains CVR5 and CVR38 inhibited the growth of the three A. vitis strains in this study. Furthermore, these bacteria could provide additional protection through niche competition, because an increase of Curtobacterium levels was observed in asymptomatic CVR5-treated plants, whereas an increase was not observed for those treated with CVR38, which did not show a significant decrease of disease incidence based on the data collected during 2013.

A commercial bacterial isolate, Bacillus subtilis SR63, was also compared as a biocontrol agent. Bacillus subtilis is one of the most studied bacterial BCAs, showing several valuable traits (Ongena and Jacques 2008), and most strains produce a multitude of broad-spectrum antibiotic compounds that contribute to antagonistic activity (Stein 2005). Bacillus subtilis SR63 inhibited the in vitro growth of A. vitis and reduced CGS. This result supports previous findings in which B. subtilis strains prevent gall formation induced by Rhizobium vitis on Nicotiana glauca (Eastwell et al. 2006) and by A. tumefaciens on tomato (Hammami et al. 2009). Further studies are required to identify the metabolites involved in the A. vitis-BCA interactions.

Among the fungal endophytes obtained, we identified several genera that are described in previous works (Casieri et al. 2009; Hofstetter et al. 2012). Alternaria spp. were highly prevalent in our isolations, as previously described by other authors (González and Tello 2011), which indicates that this genus is among the most abundant fungal endophytes in grapevine. Of the fungal isolates obtained from asymptomatic grapevine, we focused attention on Acremonium sp. VR67 that showed high nucleotide similarity to A. alternatum and A. sclerotigenum. Alternaria alternatum limits infection by Plasmodiophora brassicae on Arabidopsis thaliana (Jäschke et al. 2010), whereas A. sclerotigenum has been suggested to have potential as a biocontrol agent against Plasmopara viticola based on the inhibitory activity shown in vitro against sporangia germination (Lo Piccolo et al. 2015). Some Acremonium sp. strains show antagonistic activity in plants directly with production of antimicrobial compounds such as acremonidins, pyrrocidines and acremines (Isaka et al. 2009) or indirectly by eliciting plant defence responses (Jäschke et al. 2010). Other endophytic species with known potential BCA activity such as Epicoccum were isolated, but the effectiveness of these isolates from grapevine has been primarily reported against fungal pathogens (Compant and Mathieu 2016). In the present work, Acremonium sp. strain VR67 decreased CGS in plants without exhibiting in vitro inhibition of A. vitis, which indicated that antibiosis might not be involved in the control of this pathogen. In fact, although gall incidence in AVR67-treated plants was lower than that of the control, no significant differences were observed between symptomatic and asymptomatic treated plants, suggesting the influence of factors beyond competition. The potential involvement of plant defence responses as part of the mode of action of AVR67, as suggested for A. alternatum (Jäschke et al. 2010), should be investigated. Trichoderma spp. isolates were also tested for their ability to limit CGS. Concerning the use of Trichoderma in grapevine, T. harzianum T39 and T. atroviride SC1 and UST1 have been previously evaluated for their potential activity as biocontrol agents. T39 induces systemic resistance to Plasmopara viticola (Perazzolli et al. 2011), and the strains SC1 and UST1 counteract Phaeoacremonium aleophilum and Phaeomoniella chlamydospora infections during grafting (Pertot et al. 2016) and pruning (Mutawila et al. 2015). Regarding the Trichoderma strains used in the present work, no antibiotic activity was revealed in in vitro assays. However, during the dual in vitro culture of Trichoderma spp. and A. vitis, interference in pathogen growth was found. Although the tested Trichoderma strains colonized grapevine roots at a low level, these treatments were effective, even with differences in their efficacy, in controlling disease caused by A. vitis. The defensive mechanisms triggered by these isolates were not investigated, and further experiments are required to elucidate the mode of action that, as proposed for AVR67, could involve the elicitation of host plant defence responses.

Finally, the level of A. vitis found in symptomatic BCA-treated plants was lower than that detected in the untreated control. The increasing contents of inoculated endophytes in treated rootstock demonstrated both the ability of these microorganisms to colonize plant tissues and the possibilities to artificially increase their abundance inside the plant. The data presented in this work demonstrate the efficacy and the high potential of endophytic isolates to hamper disease caused by A. vitis in grapevine and the possibility to draw on reserves of commercial BCAs to identify potential agents that can control crown gall through different strategies. Moreover, the combination of different protective strategies appears to be a promising approach to control crown gall disease in vineyards.