Bacterial isolate RS-25 from jujube fruit, MG-4 from strawberry fruit, Z-14 from wheat rhizosphere soil and Pnf-4 from wheat plant significantly inhibited the growth of various phytopathogenic fungi and were maintained at 4°C on streak-inoculating nutritive agar (NA) slants (Zhang et al., 2017a, 2017b). The strains were cultured overnight at 37°C on NA slants, transferred to Erlenmeyer flasks containing 50 ml of the seed culture medium, and shaken at 220 rpm for 48 h at 37°C. The vegetative cells and supernatant were obtained after centrifugation at 10,000 ×g for 15 min. The cells were suspended in sterile distilled water (SDW) and adjusted to the appropriate concentration (1 × 10⁸ cells/ml), and the supernatant was filtered through a 0.22-μm syringe to obtain the culture filtrate.

Using the dual-culture in vitro assay described by Kilani-Feki et al. (2016), the antifungal activities of the four isolates against B. cinerea were evaluated. A single 5-mm-diameter mycelial disc of the test fungus was placed at the center of potato dextrose agar (PDA) plates. The samples were treated with the bacterial isolates at four symmetrical sites, a distance of 3 cm from the center of the plate, and then allowed to grow for 5 days at 27°C in darkness. The antifungal activity of the culture filtrate was measured using a diffusion plate assay modified from that described by Pretorius et al. (2015). Four evenly spaced wells (7 mm in diameter) were made 2.5 cm from the center of PDA plates containing 40 μg/ml of streptomycin sulfate. A plug of B. cinerea was placed at the center of the plate, and an aliquot (50 μl) of the filtrate or sterile water (control) was added to each well. The plates were incubated at 27°C, and the inhibition zones were measured after clear halos became visible.

HCl (6 M) was used to lower the pH of the culture filtrates of the four antagonistic bacteria to 2. The filtrates were maintained overnight at 4°C and centrifuged to isolate the precipitate, which was then washed twice using diluted HCl (pH 2) and extracted twice using methanol (Zhang et al., 2017b). The extract was dried using a rotary vacuum evaporator and resuspended in an equivalent volume of sterile water to detect the biocontrol effects of the extract in controlling the gray mold caused by B. cinerea on postharvest fruit.

To evaluate the efficiency of the four antagonist bacteria, RS-25, MG-4, Z-14, and Pnf-4, and their culture filtrates and extracts in controlling the gray mold on the postharvest fruit of tomato, strawberry, and grape, the in vivo assay reported by Panebianco et al. (2014) and Parafati et al. (2015) was used with slight modifications. Tomato (Lycopersicon esculentum Mill. ‘Qianxi’), strawberry (Fragaria × ananassa Duch. ‘Shuangliu’), and grape (Vitis vinifera L. ‘Red Globe’) fruit were obtained from commercial orchards in Baoding, China. In this study, the tomato, strawberry, and grapefruit did not receive any preharvest fungicide treatment, and only healthy and homogenous fruit were selected. Before treatment and inoculation, the fruits were washed with tap water, superficially disinfected for 1 min in 0.1% (v/v) sodium hypochlorite, rinsed with SDW, and air-dried at room temperature.

A sterilized needle was used to make artificial wounds (0.3 cm in depth × 0.5 cm in width) at equatorial areas of the fruit (2 wounds on each fruit). A drop (10 μl; 1 × 10⁶ conidia/ml) of B. cinerea was inoculated in each wound and air-dried for 2 h; then, another drop (10 μl; 1 × 10⁸ cells/ml) of each bacterium (culture filtrate or extract) was added to each wound. The fruits were placed in plastic packaging cartons (5 fruit/carton) and incubated at 25°C and 95% relative humidity for 5 days. Fruits treated with SDW and infected with the pathogen were used as the negative control. Three replicates of 5 fruit each were used per treatment (15 fruit/treatment), and the experiment was repeated thrice. The disease index (DI) and disease reduction (DR) of the gray mold were calculated according to the formulas published by Gong et al. (2017). The degree of disease severity due to inoculation with B. cinerea was calculated as follows: 0 = no lesion; 1 = a few scattered lesions covering < 2% of the fruit surface; 2 = extensive lesions covering > 2% but < 5% of the fruit surface; 3 = extensive lesions covering > 5% but < 25% of the fruit surface, and 4 = extensive lesions covering > 25% of the fruit surface. The DI was calculated using the following formula:

Where i is the disease severity (0-4), and n_i is the number of fruit with a severity of i. The DR was calculated using the following formula: DR (%) = 100 × (DI-Bc − DI-test)/DI-Bc. Where Bc is B. cinerea.

A dual-culture method was used to assess the efficacy of volatile organic compounds (VOCs) produced by the antagonists in decreasing the growth of B. cinerea (Gao et al., 2017; Grzegorczyk et al., 2017). Twenty-microliter aliquots of the bacterial suspensions (10⁸ cells/ml) were seeded on NA plates and incubated at 37°C for 24 h. Then, discs (5-mm circular plugs) of actively growing fungal mycelia of B. cinerea were placed at the center of Petri dishes with PDA. Each pathogen plate was covered with a plate containing a bacterial culture incubated for 24 h. Three plates were used for each antagonist, and unseeded PDA plates were used as the negative control. The two plates were superimposed with Parafilm around the edges to prevent the leakage of air and incubated at 25°C. The mycelial diameter was measured after 5 days, and the rate of inhibition of radial mycelial growth was calculated using the equation published by Gao et al. (2018). Three replicates were used for each treatment, and the experiment was repeated thrice. VOCs from the four antagonists were analyzed using solid-phase micro extraction-gas chromatography-mass spectrometry (SPME-GC-MS). For collecting the VOCs, the antagonists were inoculated in NA medium with 25 ml of headspace for 7 days at 28°C. The NA medium without the antagonistic bacteria was used as the control. DVB/CAR/PDMS fiber (Supelco, Inc., Bellafonte, PA, USA) was applied for headspace SPME of the VOCs at 55°C for 40 min. The SPME fiber was immediately inserted into the injection port of the gas chromatography-mass spectrometry (GC-MS; Agilent 7820A-5977E, Santa Clara, CA, USA) for thermal desorption at 250°C for 5 min with a DB-5MS column (30 m × 0.25 mm i.d. × 0.25 μm film thickness; Agilent J&W Scientific, Folsom, CA, USA). The initial oven temperature was 60°C for 2 min, and it was increased to 100°C at a rate of 10°C/min, then increased to 180°C at a rate of 5°C/min, and further increased to 240°C at a rate of 10°C/min and held for 5 min. The mass spectrometer was operated in the electron ionization mode (70 eV). The ion source, quadruple, and GC-MS interface temperatures were set at 230°C, 150°C, and 250°C, respectively. Mass spectrometry (MS) was performed in the full-scan mode with the m/z ranging from 50 to 400 for the analysis of VOCs. The mass spectra data of the volatile compounds were compared with those in the NIST/EPA/NIH Mass Spectrometry Library and Mainlib and/or Replib databases (Gao et al., 2017).

The protease activity of the antagonists, indicated by casein degradation, was measured by inoculation on skimmed milk agar with the cross-line method, and the width of each halo was measured as an indicator of the level of protease activity (Yang et al., 2008). The cellulase activity of the antagonists was determined by observing the widths of the clear zones on carboxymethyl cellulose (CMC) agar plates (Ghose, 1987). The chitinase activity of the antagonists was determined by measuring the widths of the clear zones on chitin agar (CA) plates (Roberts and Selitrennikoff, 1988). To measure the activity of the metabolites secreted by the antagonists, four evenly spaced wells (7 mm in diameter) were formed 2.2 cm from the center of the plate by excision, and a 30-μl aliquot of the culture filtrate was added to each well.

The chrome azurol S (CAS) assay was performed to evaluate the effectiveness of the Bacillus antagonists in siderophore production (Kurabachew and Wydra, 2013; Schwyn and Neilands, 1987). After 5 days at 28°C, the siderophore-producing isolates generated orange zones, and the diameter of the clear zone of these colonies was visually rated. Siderophore production in the culture filtrate was measured as described above.

The antagonists were cultured in nutritive broth and shaken to midexponential growth; then, 10 μl of the cultured medium was transferred to 1 ml of minimal salts glycerol glutamate (MSgg) medium (Branda et al., 2001) in 24-well polyvinylchloride microtiter plates (Thermo Fisher Scientific, Waltham, MA, USA) and incubated under stationary conditions at 37°C for 4 days (Dietel et al., 2013).

The ability of the antagonists to colonize the wounds on the grape, strawberry, and tomato fruit was tested using previously described and partially modified procedures (Cirvilleri et al., 2005; Parafati et al., 2015). A sterile needle was used to make artificial wounds (4 wounds/fruit) on superficially sanitized fruit, and 10 μl aliquots of the bacterial suspensions (1 × 10⁶ cells/ml) were individually inoculated into the wounds; the fruit were incubated for 5 days under the conditions mentioned previously (6 fruit/box). Tissue samples containing the whole wound were cut using a sterile knife after inoculation for 0, 24, 48, 72, 96, and 120 h. The samples were weighed, crushed with a sterilized mortar and pestle, serially diluted, and plated on NA. The bacterial colonies were counted after overnight culture at 37°C to calculate the mean colony forming units (log₁₀ cfu) on the basis of fresh weight. Three replicates of each fruit type were used per treatment (6 fruit per treatment), and the experiment was repeated twice.

The methanol extracts of the culture filtrates of the four antagonistic bacteria were analyzed using the 5800 matrix-assisted laser desorption ionization-time-of-flight (MALDI-TOF) mass spectrometer (AB SCIEX, Redwood, CA, USA) operated in the positive reflectron mode; α-cyano-4-hydroxycinnamic acid was used as the matrix (prepared by dissolving 5 mg in 1 ml of 50:50 acetonitrile: water containing 0.1% trifluoroacetic acid) (Velho et al., 2011).

Genomes extracted from the antagonists isolates were analyzed for the 16S rRNA gene sequence by using universal primers, 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1495R (5′-CTACGGCTACCTTGTTACGA-3′), that amplified a 1,400-bp region of the 16S rRNA gene. Phylogenetic analysis was conducted using maximum likelihood in MEGA 5.10 (Tamura et al., 2011). The topology of the phylogenetic tree was evaluated by 1,000 resamplings.

The data from the replicates were expressed as mean ± SD values. SPSS version 17.0 software package (SPSS Inc., Chicago, IL, USA) was used to perform the calculations and comparisons of the treatment means for each experiment. Duncan’s test (P ≤ 0.05) and one-way analysis of variance were used to determine whether the means differed significantly.

The four isolates, RS-25, MG-4, Z-14, and Pnf-4, and their culture filtrates all demonstrated significant antifungal activities against B. cinerea. The culture filtrate of RS-25 demonstrated the strongest antifungal activity, whereas that of Pnf-4 was the weakest (Fig. 1). The efficacies of RS-25, MG-4, Z-14, and Pnf-4 and their culture filtrates and extracts against the gray mold caused by B. cinerea on the postharvest tomato, strawberry, and grapefruit are reported in Table 1 and Fig. 2. Generally, Z-14, MG-4, and their culture filtrates and extracts were more active against B. cinerea. Pnf-4, RS-25, and their culture filtrates and extracts were relatively ineffective in reducing the DI of the strawberry fruit. The DR of B. cinerea was observed to be the highest (82.14) when the strawberry fruits were treated with the cells of Z-14, whereas the culture filtrate of RS-25 was the most ineffective and yielded the lowest DR value (55.08). The DR values of the tomato fruit treated with Z-14, RS-25, and the culture filtrate and extract of Z-14 were higher (82.62%, 80.46%, 79.25%, and 75.23%, respectively) than those of the tomato fruit treated with Pnf-4, MG-4, and the culture filtrate of RS-25 (68.65%, 67.65%, and 67.29%, respectively). The DR values of the tomato fruit treated with the culture filtrates of Pnf-4 and MG-4 were the lowest (55.27% and 54.03%, respectively). The DR values of the grapefruit treated with MG-4 and RS-25 and culture filtrates of Pnf-4, MG-4, and RS-25 were the highest (84.69%, 86.66%, 83.24%, 86.28%, and 86.81%, respectively), whereas the culture filtrate of Z-14 showed the lowest efficiency in controlling the gray mold (12.20%) on grapefruit.

The variable inhibition efficiency of the VOCs produced by RS-25, MG-4, Z-14, and Pnf-4 against the mycelial growth of B. cinerea is shown in Fig. 3. The highest inhibition values were detected for Pnf-4 against B. cinerea (70.71 ± 3.94%), followed by RS-25 (62.57 ± 3.26%) and MG-4 (53.44 ± 2.71%). In contrast, the VOCs of Z-14 were the most ineffective in reducing the mycelial growth of B. cinerea (24.21 ± 1.85%). Based on the headspace SPME-GC-MS, 109 potential VOCs produced by RS-25, 93 VOCs by MG-4, 92 by Z-14, and 91 by Pnf-4 were identified. The main components of the VOCs were ketone, alcohol, aldehyde, olefin, ester, alkane, and phenol, and esters accounted for the highest percentage in the VOCs produced by RS-25, MG-4, Z-14, and Pnf-4 (36.80%, 29.58%, 30.78%, and 36.26%, respectively) (Table 2).

All the tested Bacillus strains and their culture filtrates were able to hydrolyze CMC, and they produced obvious clear zones in the CMC agar plates. The culture filtrate of Z-14 exhibited the strongest cellulase activity; the diameter of the clear zone was 3.4 cm, whereas those of RS-25, Pnf-4, and MG-4 were 2.4 cm, 2.3 cm, and 2.8 cm, respectively (Fig. 4A and B). The four antagonists and their culture filtrates all exhibited potent protease activity, and the hydrolysis zones of RS-25, Z-14, and MG-4 were relatively bigger than that of Pnf-4; this is similar to the protease activity results of the culture filtrates of the antagonists (Fig. 4C and D). The four antagonists and their culture filtrates did not produce clear zones in CA plates, which demonstrated they did not synthesize chitinase. These results are consistent with the findings of Ren et al. (2013), who suggested that Bacillus pumilus JK-SX001 does not secrete chitinase but instead secretes protease and cellulase.

All four antagonists exhibited the ability to form biofilms on the inert surface of liquid MSgg medium and evolved into whitish aerial structures that appeared to be dry and had a cotton-like texture (Fig. 4G). The population dynamics of RS-25, MG-4, Z-14, and Pnf-4 strains on artificially wounded tomato, strawberry, and grapefruit are shown in Fig. 5. The population trends of the antagonists in the wound tissue of tomato fruit, starting from a similar concentration (log₁₀ cfu/g of tissue = 3.50), remained relatively stable for a period of 24 h; increased for MG-4, Z-14, and Pnf-4 and a few drops of RS-25 up to 72 h; and were similar (log₁₀ cfu/g of tissue = 4.30) after 120 h of incubation (Fig. 5A). The population trend of the antagonists in the wound tissue of strawberry fruit showed fluctuations to a certain extent during 120 h after inoculation; RS-25 showed the highest concentration (log₁₀ cfu/g of tissue = 3.76), whereas MG-4 showed the lowest concentration (log₁₀ cfu/g of tissue = 3.06) (Fig. 5B). The population trend of the antagonists in the wound tissue of grapefruit, starting from similar concentrations, ranged from 3.11 (log₁₀ cfu/g of tissue) for MG-4 to 3.37 (log₁₀ cfu/g of tissue) for Pnf-4; growth appeared to be quite stationary up to 24 h, but it showed variations after that period. The population densities of Z-14 and Pnf-4 increased significantly (log₁₀ cfu/g of tissue = 4.07 and 4.05, respectively), whereas the population density of MG-4 remained relatively stable (log₁₀ cfu/g of tissue = 3.10) and that of RS-25 showed a notable decline (log₁₀ cfu/g of tissue = 2.41) (Fig. 5C). The results prove that the population densities of the four antagonists in the three types of fruit showed different colonization capacities. The colonization abilities of the antagonists were similar in the wounds of the tomato fruit but different in those of the strawberry and grapefruit. The colonization ability of RS-25 was the strongest in the wounds of the strawberry fruit, but weakest in those of the grapefruit.

MALDI-TOF mass spectrometry was used to analyze the lipopeptide antibiotics produced by the antagonists. The mass spectra of the crude extract from RS-25 revealed two major clusters with peaks at m/z 1,016.69-1,074.70 Da, which correspond to the isoforms of surfactin B, and peaks at m/z 1,433.88-1,529.88 Da, which correspond to the isoforms of fengycin (Fig. 6A). The mass spectra of the crude extract from Pnf-4 revealed three major clusters with peaks at m/z 1,030.70-1,074.72 Da, which correspond to surfactin B, peaks at m/z 1,079.62-1,125.57 Da, which correspond to iturin A, and peaks at m/z 1,435.87-1,543.88 Da, which correspond to fengycin (Fig. 6B). Surfactin B, iturin A, and fengycin were detected from Z-14 (Fig. 6C), and iturin A and fengycin, from MG-4 (Fig. 6D).

When compared with the GenBank database, the 16S rRNA sequences of the strains were matched to those of Bacillus spp. The constructed unrooted phylogenetic tree revealed that antagonist Pnf-12 was most similar to B. subtilis, according to the alignment results. Antagonist RS-25 was identified as Bacillus amyloliquefaciens; MG-4, Bacillus licheniformis; and Z-14, B. subtilis (Fig. 7).