Antifungal Effects of Volatiles Produced by Bacillus subtilis Against Alternaria solani in Potato

Antifungal activities of plant-beneficial Bacillus have been widely studied in recent years. Numerous studies have studied the antifungal mechanisms of soluble non-volatile bioactive compounds such as lipopeptides and proteins produced by Bacillus against soil-borne diseases. However, the antagonistic mechanisms of volatile organic compounds (VOCs) from Bacillus against airborne phytopathogens are still largely unknown, and the function of Alternaria solani pathogenic genes has not been well identified. Here, we first isolated a Bacillus strain with strong antifungal activity and finally identified it as B. subtilis ZD01. Then, the antagonistic mechanisms of VOCs produced by strain ZD01, against A. solani, an airborne fungal pathogen that can cause early blight diseases of potato, were studied. We showed that VOCs produced by strain ZD01 can reduce the colony size and mycelial penetration and can cause serious morphological changes of A. solani. Scanning electron microscope (SEM) observation showed that VOCs released by ZD01 could cause more flaccid and gapped hyphae of A. solani. Also, we found that VOCs produced by ZD01 can inhibit the conidia germination and reduce the lesion areas and number of A. solani in vivo significantly. Meanwhile, based on gas chromatography/mass spectrometry (GC/MS) analysis, 29 volatile compounds produced by strain ZD01 were identified. Out of 29 identified VOCs, 9 VOCs showed complete growth inhibition activities against A. solani. Moreover, we identified two virulence-associated genes (slt2 and sod) in A. solani. slt2 is a key gene that regulates the mycelial growth, penetration, sporulation, and virulence in vivo in A. solani. In addition, sod plays a significant role in the SOD synthetic pathway in A. solani. Results from qRT-PCR showed that the transcriptional expression of these two genes was down-regulated after being treated by VOCs produced by ZD01. These results are useful for a better understanding of the biocontrol mechanism of Bacillus and offer a potential method for potato early blight disease control.


INTRODUCTION
Alternaria solani is a kind of fungal pathogen that can cause early blight disease of tomato, potato, tobacco, and many other vegetables and crops, and lead to huge losses in agricultural production. Potato is one of the most important crops in the world. Early blight disease on potato can cause up to 80% of annual yield losses in some regions of the world (Morgan et al., 2002;Pasche et al., 2004;Peters et al., 2008). Nowadays, chemical fungicides are the most effective agents to control early blight. However, the indiscriminate using and even abuse of chemical fungicides have already caused the appearance of resistant pathogens, which will further threaten the food safety and human health. Due to this, there is a rising interest to seek for alternative antifungal agents for plant disease control.
Bacillus species are known for their capacity to produce a great variety of antifungal compounds to suppress or kill fungal pathogens (Chaurasia et al., 2005). Among them, nonribosomal cyclic lipopeptides (Nakano et al., 1988;Munimbazi and Bullerman, 1998;Kim et al., 2004;Timmusk et al., 2005) are the most well-studied ones. However, these non-volatile antibiotics cannot spread over long distances. In recent years, volatile organic compounds (VOCs) produced by Bacillus have been evaluated as a new approach for plant fungal disease control. Due to their capability of diffusing between the soil particles and spreading into the atmosphere over very large distances from their original application point, VOCs can exert their inhibitory activity without direct or physical contact between the VOC-producing microorganisms and the target pathogens (Minerdi et al., 2009;Heydari and Pessarakli, 2010). Their strong antifungal activities, along with the characteristic of harmless to both environment and human beings, make VOCs a promising and sustainable agent to replace fungicides for plant pathogen control in the future (Fialho et al., 2010;Parafati et al., 2017;Tilocca et al., 2020).
However, limited knowledge is known about bacterial VOCs controlling airborne plant fungal pathogens, especially A. solani.
In several recent studies, VOCs produced by Bacillus species have been identified to inhibit the airborne pathogens, mainly as Botrytis cinerea, the causal agent of tomato gray mold. Two Bacillus velezensis strains 5YN8 and DSN012 could suppress the growth and spore formation of B. cinerea by releasing numbers of VOCs (Jiang et al., 2018). The VOCs of B. velezensis ZSY-1 strain exhibit significant antifungal activity against B. cinerea (Gao et al., 2017). The mixed volatiles produced by B. atrophaeus CAB-1 strains, mainly composed of hexadecane, 2,3-dimethoxybenzamide, and oanisaldehyde, resulted in an effective inhibition of B. cinerea (Zhang et al., 2013).
In this study, B. subtilis ZD01, an effective antifungal strain, was isolated and identified from potato rhizosphere, and the antifungal effects of VOCs produced by it on A. solani were investigated by both in vivo and in vitro experiments. We demonstrated that VOCs of B. subtilis have great potential to be used as a biologically synthesized fungicide in the agricultural field.

RESULTS
B. subtilis ZD01 Showed Strong Inhibitory Activity Against A. solani A total of 103 isolated Bacillus strains were obtained from rhizosphere of potato in Shandong and Hebei Province in China. Among them, 34 isolates showed a certain inhibition effect on A. solani (Supplementary Table S1). Notably, strain ZD01 showed the strongest antagonistic activity against A. solani (Supplementary Table S1). Then, ZD01 were tested for inhibitory activity mediated by volatile against A. solani using face-to-face Petri dish method (Gao et al., 2017). The result showed that the growth rates of A. solani mycelia were reduced by 50.1 ± 2.1% in the presence of volatiles released by ZD01 ( Figure 1A). Figures 1C,D showed that a broad range of pathogens were also inhibited by the volatiles released from strain ZD01.
Strain ZD01 was then identified by 16s rRNA and complete genome sequencing. We performed BLAST search using the DNA sequence of the16S rRNA gene of ZD01 as a query against the NCBI GenBank database. Furthermore, we sequenced the complete genome of ZD01 by Pacbio RSII. After that, a single circularized chromosome of 4,015,360 bp in length with a GC content of 43.71% from ZD01 was obtained ( Figure 1B, CP046448). According to the BLAST and complete genome sequencing results, we found that strain ZD01 showed extremely high similarity (identity 99%, E-value = 0) to B. subtilis. We finally classified the strain ZD01 as B. subtilis.

VOCs Produced by ZD01 Changed the Mycelia Morphology and Inhibited the Conidia Germination of A. solani
Inhibition of fungal mycelia growth and spore germination is conventional mechanism of biocontrol agents for disease controlling. So, the suppression of VOCs produced by ZD01 on mycelia growth and conidia germination were evaluated. The result showed that the inhibition rate of A. solani growth by VOCs was 50.1 ± 2.1% compared with the control after 6 days, suggesting that the VOCs from ZD01 showed a strong inhibitory activity on fungal mycelia growth. Moreover, with the treatment of VOCs produced by B. subtilis ZD01, the mycelia of A. solani became denser and thicker, and the color of the colony turned into white ( Figure 1A). In addition, strong decrease in mycelial penetration was observed after treatment. Mycelia in control groups were able to penetrate cellophane sheets during the cellophane penetration assay with the colony diameter of 5.5 ± 1.6 cm. However, the A. solani colony co-cultured with VOCs produced by ZD01 did not grow after the cellophane was removed, which indicated that the penetration ability of mycelia lost completely (Figure 2A).
A scanning electron microscope (SEM) was used to investigate detailed morphological changes of A. solani hyphae caused by VOCs. Morphological changes in A. solani cells are shown in Figure 2B. Regular length and intact cell walls with uniform composition and structure were present in the hyphae of A. solani in the control group ( Figure 2B). However, hyphae treated with VOCs from ZD01 exhibited substantial structural destruction ( Figure 2B). In detail, some of the hyphae became expanded, and the formation of empty segments was presented (red arrows, Figure 2B). Treated mycelium appeared with a more flaccid hyphae, and the surface of the cell walls became uneven (yellow arrows, Figure 2B). Also, thin or gapped structures presenting a retracted protoplasm were seen in Figure 2B (green arrows).
The broken structures might lead to the leakage of cytoplasmic components. Thus, volatiles produced by ZD01 could change the morphology of A. solani seriously, and decompose the cell walls and membrane.
Spore germination plays a crucial role during the infection of fungal pathogens, so the conidia germination suppression capacity of VOCs produced by B. subtilis ZD01 was evaluated. As shown in Figures 2C,D, the B. subtilis ZD01 could inhibit the germination of A. solani conidia significantly (P < 0.05) by releasing some VOCs compared with the control group. In the control group, conidia began to germinate and germ tube formed. Then, the germ tubes formed regular hyphae (black arrow, Figure 2C), which can directly penetrate the host epidermal cell junctions and formed infection hyphae on leaf. However, some conidia treated with VOCs from ZD01 could not germinate completely (blue arrow, Figure 2C), and some conidia formed irregular germ tubes, which were much shorter than those in the control group (red arrows, Figure 2C). The inhibition rate of VOCs from ZD01 on conidia germination was 19.2 ± 2.1% ( Figure 2D).

VOCs Produced by B. subtilis ZD01 Reduced the Symptoms of Early Blight Disease on Potato Leaf
The development and expansion of disease symptom induced by A. solani was inhibited effectively by VOCs produced by strain ZD01 in in vivo leaf test. As shown in Figure 3A, the lesion areas on potato leaves (cultivar Helan 15) inoculated by A. solani were significantly reduced (P < 0.05) after co-cultivation with ZD01 volatiles by divided plate method. For leaves in the control group with no treatment of ZD01 volatiles, the lesion areas extended to 5.2 ± 1.7 cm 2 after 5-7 days incubation at 25 • C, whereas for the leaves exposed to volatiles from ZD01, the lesion areas were limited to 0.8 ± 0.3 cm 2 ( Figure 3B). These results corresponded to relative pathogen copy numbers per milligram leaf of 3.98 ± 0.67 × 10 8 and 1.08 ± 0.22 × 10 8 for control and treatment, respectively ( Figure 3C).
slt2 and sod Are Virulence-Associated Genes in A. solani Identification of virulence-associated genes is important to reveal the pathogenic mechanisms and biological control approaches of fungal pathogens. The complete genome sequence of A. solani HWC-168 has been sequenced and analyzed in our previous study (Zhang et al., 2018). We compared the whole genome sequence of HWC-168 with its wellstudied Saccharomyces cerevisiae (EF058927.1) and close relative A. alternate (GQ414510.1). After that, two typical pathogenic genes (slt2 and sod) were found in the genome of HWC-168. Then, the functions of these two genes were determined through gene knockout and phenotype verification. To determine whether slt2 and sod affects the pathogenicity of A. solani, we compared the deletion mutants with the wildtype strain and complementation strains in mycelia growth, sporulation, and virulence by in vivo potato leaf tests. Compared with the wild-type strain and complementation strains, the mycelium of slt2 mutant was significantly denser, and the color of colony was gray-white without pigments ( Figure 5A). The colony diameter of slt2 deletion mutants was 2.79 ± 0.52 cm, while that of wild-type strain and complementation strains were 6.18 ± 0.14 and 6.18 ± 0.08 cm, respectively ( Figure 5E). No obvious changes of the colony morphology and diameter of sod deletion mutant were observed (Figures 5A,E), while the colony diameter of sod and its complementation strain ( sod-C) were 6.04 ± 0.14 and 6.11 ± 0.27 cm, respectively. These results suggested that slt2 is the key gene involved in the regulation of mycelium growth and development. Also, we found that the slt2 gene was critical for the penetration ability of A. solani. As shown in Figure 5B, the wild-type and complemented strains of slt2 but not the mutants were able to penetrate cellophane sheets in the cellophane penetration assay.
We next tested whether the slt2 and sod genes are also responsible for the sporulation of A. solani. The results showed that slt2 mutant lost the capacity of conidia production and the yield of conidia of sod deletion mutant per area was 514 ± 149/cm 2 , while the wild-type strain was 2131 ± 301/cm 2 ( Figure 5D). Compared with the wild-type strain, the sporulation yield of slt2 and sod deletion mutants per area decreased significantly (P < 0.05). Through the complementation strain, the yield of sporulation increased to 2073 ± 415/cm 2 and 1709 ± 171/cm 2 , respectively. The sporulation ability of the slt2 deletion mutant was completely lost, and the sporulation ability of the sod deletion mutants was weakened compared to that of the wild-type strain, suggesting that slt2 is a key gene regulating the sporulation in A. solani. Meanwhile, a reduction of sod mutant in superoxide dismutase (SOD) content was detected. The content of SOD in mutant was 86.8 ± 1.4 and 89.6 ± 3.2 ng/L, respectively, after 1 and 2 days incubation as compared to the wild-type HWC-168, which was 127.7 ± 9.2 and 134.2 ± 12.0 ng/L, respectively ( Figure 5G). In addition, SOD contents in complementation strains were similar to those of the wild type (124.0 ± 3.2 ng/L, 133.5 ± 3.9 ng/L). These results suggested that sod has a significant role in the SOD synthetic pathway in A. solani.
In in vivo tests, potato leaves inoculated with wild-type HWC-168, and complementation strains of slt2-C and sod-C showed obvious lesions and yellow halos ( Figure 5C). The lesion diameter extended to 0.99 ± 0.16, 1.01 ± 0.02, and 1.03 ± 0.09 cm after 7 days incubation at 25 • C, whereas for the leaves inoculated with slt2 and sod mutants, the lesion diameters were limited to 0.00 ± 0.00 cm and 0.73 ± 0.07 cm, respectively ( Figure 5F). The result showed that the deletion of slt2 and sod can significantly reduce the pathogenicity of A. solani (P < 0.05).

B. subtilis ZD01 Volatiles Down-Regulated the Expression of Virulence-Associated Genes in A. solani
The transcriptional expression profiles of slt2 and sod under condition of co-culture with ZD01 volatiles were investigated by real-time RT-PCR. The results showed that after A. solani strain HWC-168 was exposed to volatiles emitted by ZD01 for 3 days, the transcriptional expression of sod was strongly induced (∼2.45-fold) compared with the control group and then repressed (∼0.61-fold) after 4 days ( Figure 5H). The expression of slt2 was strongly repressed (∼0.83-and 0.40-fold) in the presence of VOCs after 3-and 4-day co-culture ( Figure 5H). The down-regulated expression of slt2 was consistent with the virulence reduction in A. solani.

DISCUSSION
Many plant-beneficial Bacillus species exhibit their biocontrol capacity to plant pathogens through non-volatile antibiotic production, nutrients, and niche competitions, as well as induction of plant systemic resistance (Pliego et al., 2008;Lemfack et al., 2014;Farag et al., 2017;Xie et al., 2018). However, limited knowledge is known about the antifungal mechanisms of volatiles produced by Bacillus strains. Most studies just focus on mycelia morphology and penetration and spore germination at a cellular level. However, the molecular mechanisms of antifungal activity have not been revealed. For example, transmission electron microscopy observation of fumigated and untreated B. cinerea showed excessive vesication or thickened cell walls in exposed conidia and increased strong retraction of plasma membrane in exposed hyphae (Li et al., 2012). The VOCs of B. velezensis 5YN8 can suppress the mycelium growth and conidia formation of B. cinerea BC1301 (Jiang et al., 2018). In this study, the insights into the mechanisms of B. subtilis ZD01 volatiles against A. solani showed similar biocontrol strategies. The hyphae penetration, conidia germination, and virulence of A. solani were significantly reduced when treated with VOCs produced by ZD01 (P < 0.05). Scanning electron microscopy showed thin, inward or gapped structures and altered surface morphology in the majority of A. solani cells after co-culture with strain ZD01 volatiles. Meanwhile, A. solani cells exposed to VOCs produced by strain ZD01 formed swollen part of hyphae with defective ability, leading to aborted invasion to the plant barrier. These results were consistent with the reduced pathogenicity in vivo. All of these findings indicated that the mode of action of volatiles can be explained, at least in part, by their activities that lead to the functional degradation through the damage of mycelium structure, killing A. solani cells partially, and the reduction of spore germination (Figure 6). All of the active volatiles produced by bacteria so far can be grouped into alcohols, ketones, aldehydes, alkenes, alkynes, benzenes, esters, terpenoids, heterocycles, and sulfurcontaining compounds (Fernando et al., 2005;Corcuff et al., 2011;Lemfack et al., 2014). Among nine identified chemicals with strong antifungal effects, the antifungal effects of 6methyl-2-heptanone, acetophenone, 2-pentylfuran, 2,5-dimethyl pyrazine, and benzothiazole have also been analyzed. 6-Methyl-2-heptanone produced by B. subtilis completely inhibited mycelial growth of F. oxysporum f. sp. lactucae (Liu et al., 2018). Acetophenone released by Streptomyces globisporus and Paenibacillus polymyxa can inhibit Penicillium italicum and F. oxysporum growth, respectively Raza et al., 2015). The multiple functions of 2-pentylfuran have also been identified. 2-Pentylfuran isolated from the volatile products of bacterial strains presented strong inhibition on both mycelial growth and conidia germination of F. oxysporum (Wu et al., 2015;Liu et al., 2018). 2-Pentylfuran fumigated from Bacillus megaterium strain XTBG34 could significantly promote plant growth of Arabidopsis thaliana (Zou et al., 2010). Moreover, 2,5dimethylpyrazine can significantly inhibit growth of several plant pathogens such as Magnaporthe oryzae, Phytophthora capsici, and A. solani (Munjal et al., 2016;Gao et al., 2017), and it can also be used as a food additive at low concentration (Müller and Rappert, 2010). Benzothiazole produced by Bacillus species can inhibit growth of F. oxysporum , and it can also be used to produce riluzole and pramipexole (Powell, 2015;Faridbod et al., 2016).
In this study, benzenes and ketones are the most abundant volatiles produced by strain ZD01. The aniline and acetophenone, which belonged to benzene compounds, could completely inhibit the growth of A. solani at certain concentrations and also exhibited high yield, as indicated by the peak areas of the GCs, so it was likely that they are main VOCs that play key roles during the antifungal process. For the VOCs that belong to ketones, 2nonanone, 5-methyl-2-hexone, and 2-heptanone exhibited 100% inhibition of A. solani under certain concentrations when we did in vitro inhibition test, but their yields were relatively low among VOCs produced by ZD01, as indicated by GC-MS. However, 6-methyl-2-heptanone exhibited large peak areas on the GCs, and it did show strong antifungal effects against A. solani with 100% inhibition. Considering both contents and antifungal effects, aniline, acetophenone, and 6-methyl-2-heptanone might be considered active antifungal compounds.  6 | A model for the mode of action of VOCs produced by ZD01 against A. solani. Volatiles produced by ZD01 mediate the conidia germination and mycelium penetration of A. solani. Acetophenone, 6-methyl-2-heptanone, and aniline biosynthesized by ZD01 is the major antifungal compounds against A. solani. These volatiles produced by ZD01 target the conidia and mycelium, which subsequently leads to suppression of fungal growth, mycelium penetration, conidia germination, and pathogenicity of A. solani as well as virulence gene expression.
Based on whole genome sequences and annotation of A. solani strain HWC-168 (Zhang et al., 2018), two virulencerelated genes slt2 and sod were predicted. In some fungal pathogens, like Colletotrichum lagenarium, B. cinerea, and Mycosphaerella graminicola, the role of virulence-related gene slt2 has also been well studied, involved in the maintenance of cell-wall integrity and various aspects of saprotrophic and pathogenic growth (Kojima et al., 2002;Mehrabi et al., 2006;Rui and Hahn, 2007). In B. cinerea, a mutant defective in the slt2 homolog mp3 was defective in penetration and nonpathogenic infection (Rui and Hahn, 2007). Meanwhile, the biological role of sod gene has been extensively investigated in the model organism S. cerevisiae. However, the roles of these two genes in A. solani are still unclear. Here, we present the functions of slt2 and sod in A. solani identified by knockout mutant construction and phenotypical characterization. We found that slt2 is a key gene involved in mycelial growth, hyphae penetration, and sporulation of A. solani, which could further affect its pathogenicity. Meanwhile, we also found that sod is responsible for the synthesis of SOD in A. solani and its deletion mutant can reduce virulence of A. solani to some extent. Furthermore, we found that VOCs produced by B. subtilis ZD01 can decrease the transcriptional expression of these two genes. The down-regulated expression of slt2 was consistent with the reduced virulence in A. solani. However, the expression of sod was induced co-cultured with VOCs produced by strain ZD01 after 3 days. A. solani responds to and resists the biocontrol agent, B. subtilis VOCs, resulting to produce the high level of reactive oxygen species (ROS). SOD, encoded by sod, is involved in scavenging the high level of ROS into molecular oxygen and hydrogen peroxide. Then, with the accumulation of VOCs emitted by strain ZD01, the pathogenicity of sod decreased along with the increasing VOCs. In addition, VOC composition of a given species is highly dynamic over time, resulting in a changing composition of the produced VOCs depending on the age of the VOC-producing species . That would be the reason for the induced sod response after 3 days and repression after 4 days. These results reveal the molecular mechanism by which VOCs antagonize A. solani.
This study shed light on the interaction mechanisms between A. solani and VOCs produced by B. subtilis, and a potential biocontrol method for potato early blight disease caused by A. solani.

Strains and Culture Conditions
Soil samples were collected from four different fields (Chengde, Hebei; Zhang Jiakou, Hebei; Qinhuangdao, Hebei; Tengzhou, Shandong) which had serious potato early blight disease by a fivespot sampling method (Chen et al., 2018) in China. Briefly, five soil samples that were 0-20 cm away from infected potato plants (Helan 15) were collected and then pooled as one sample. After that, the dilution plating method with miner modification was used for spore-forming bacteria isolation (López-Berges et al., 2010). In brief, soil sample was homogenized and placed in a sterilized beaker. Macerated samples were heated in a water bath for 15 min at 80 • C to kill non-spore-forming bacteria and vegetative cells of spore-forming bacteria. The resulting supernatants were serially diluted in sterile 0.85% NaCl solution and plated onto LB agar plates. Plates were incubated at 37 • C for 1 day.
Collected strains were tested for antagonistic activity. Briefly, bacterial antagonistic activities toward A. solani were tested by dual-culture assay on PDA (Hu et al., 2014), which support the vegetative growth of either bacteria or fungi. The bacterial strains (5 µl, 1 × 10 8 CFU/ml) were inoculated 2 cm away from A. solani. Four strains were spotted in each dish, and the blank control without bacterial strains was set. All isolates were tested in triplicate, and their inhibition zones were measured after 7 days of dual culture at 25 • C.
Isolated bacterial strains were cultured in LB medium at 37 • C. Strain ZD01 was then identified by 16S rRNA gene analysis and whole genome sequencing (CP046448). The genome of ZD01 was sequenced using the PacBio RSII sequencing platform (Pacific Biomarkers, CA, United States). All the strains used in this study are listed in Supplementary Table S2.

Antagonistic Assay of VOCs Against Fungal Pathogens
By following the methods of Gong et al. (2015), two Petri dishes were placed face to face. The bottom Petri dish contained LB agar (1.5%, wt/vol), which was inoculated with ZD01 (200 µl, 1 × 10 8 CFU/ml), or pure identified VOCs (100 µl; 1 g/ml) were added. All the commercial solid VOCs were dissolved in dimethyl sulfoxide (DMSO). Purchase information of all the used reagents is listed in Supplementary Table S3. The top Petri dish contained 1.5% of potato dextrose agar (PDA). For the growth inhibition test, plugs (5 mm in diameter) of plant pathogenic fungi were placed onto the center of PDA plates. All the used fungal pathogens are listed in Supplementary  Table S2. For the conidia germination inhibition test, 100 µl of conidia solution (10 5 CFU/ml) was added. The top and bottom Petri dishes were sealed together with parafilm and incubated at 25 • C. After 4-7 days, the diameters of the fungal colonies were measured. Inhibition of A. solani conidia germination by VOCs was determined after 8 h of incubation. LB plates without bacterial inoculum were used as control. Inhibition rate of mycelium growth and conidia germination was calculated by the following formulas:

Penetrability Assays
The penetration ability of each A. solani strain was examined on grown PDA plates covered with cellophane membranes, as described previously (López-Berges et al., 2010). Briefly, each strain was grown on PDA plates covered with a cellophane membrane. For the VOC treatment groups, plugs (5 mm in diameter) of wild-type A. solani were spotted onto the center of PDA plate with a cellophane. Then, the wild-type strains were co-cultured with VOCs produced by strain ZD01 in LB plate (200 µl, 10 8 CFU/ml) for 3 days at 25 • C using the face-toface Petri dishes. For another 3 days, the wild-type strain was grown without VOC treatment and the cellophane membrane on the PDA plate was removed. Then, the penetrability halo was examined. LB plates without bacterial inoculum were used as control. For mutants and complementation strains, plugs (5 mm in diameter) of wild type, mutants of slt2 and sod, and complementation strains of A. solani were spotted onto the center of the PDA plate covered with the cellophane, respectively. After 3 days of incubation at 25 • C, the cellophane membrane with the colony was removed from each plate. Each strain was grown for another 3 days at 25 • C. Then, penetrated mycelial growth on each plate was examined after incubation. The experiment was repeated three times.

Scanning Electron Microscopy
The mycelia morphologies of A. solani in control groups or those treated with VOCs released by ZD01 were visualized by SEM. To observe structural changes on A. solani, the wild-type strains were co-cultured with VOCs produced by strain ZD01 for 6 days at 25 • C using the face-to-face Petri dishes. Mycelia of each group were harvested and fixed in 2% glutaraldehyde at 4 • C and then dehydrated with gradient ethanol solutions (30, 50, 80, 90, and 100%). After that, ethanol was replaced by 100% tertiary butyl ethanol. Cells were then freeze-dried, coated with gold, and imaged using a Hitachi S-3500N field emission SEM (Hitachi, Tokyo, Japan). The experiment was repeated three times.

In vivo Antagonistic Activity of VOCs Produced by ZD01
Strain ZD01 was inoculated in 2 ml of LB broth and grown overnight. 1% of overnight culture was re-inoculated into 50 ml of fresh LB broth and incubated at 37 • C under the shaking condition of 200 rpm for 12 h. Subsequently, 200 µl of cellfree supernatant was transferred into one compartment of the divided Petri dish with 1.5% of LB agar and spread out. One piece of fresh potato leaf (Helan 15) was placed onto the other compartment with 0.5% water agar containing 10 µg/ml tetracycline hydrochloride and 20 µl of A. solani conidia suspension (10 5 CFU/ml) was inoculated onto the center of the leaf. LB plates without ZD01 were used as control. For the pathogenicity of different A. solani strains, 5-mm plugs of the wild type, slt2 and sod mutants, and complementation strains were inoculated onto the center of one piece of fresh potato leaf. After 5 days of growth with 12 h of light and 12 h of darkness alternately at 25 • C, the lesion areas were measured.

SOD Concentration Measurement of A. solani
The concentrations of SOD in wild-type, sod mutant, and complementation strains of A. solani were determined by double antibody sandwich method using a commercial kit. One gram of ground mycelium was resuspended in 9 ml of PBS buffer (pH 7.2-7.4) and then centrifuged at 3,000 rpm for 20 min. The supernatant was then collected for SOD concentration measurement using the Enzyme Immunoassay Kit for Superoxide Dismutase (Omega Bio-Tek, Norcross, GA, United States) according to the manufacturer's instructions.

Collection of VOCs With SPME and Analysis by GC-MS
For VOC extraction, the Bacillus strain was inoculated into 6 ml of LB medium in a 20-ml vial. After incubation at 37 • C for 4 days, samples were used for analysis. To provide a repeatability of the experiment, four samples were prepared and the LB medium without the antagonistic bacteria was set up as a control.
Volatile organic compounds were analyzed using solid phase microextraction (SPME) coupled with gas-chromatography tandem mass spectrometry (GC-MS) analysis. The SPME fiber (2 cm, 50/30 µm divinylbenzene/carboxen/polydimethylsiloxane fiber, DVB/CAR/PDMS) was inserted into the headspace of the vial and then placed at 50 • C for 40 min. Compounds were then desorbed for 10 min in the injection port of the gas chromatograph at 220 • C with the purge valve off (splitless mode).
An HP-5 capillary column (30.0 m × 0.25 mm × 0.25 µm, Thermo) and helium as the carrier gas were used for GC-MS. A Thermo Trace 1300-ISQ MS was used for peak separation and detection. Each run was performed for 45 min. The initial oven temperature of 40 • C was held for 4 min, ramped up at a rate of 5 • C/min to 150 • C holding for 1 min, further ramped up at a rate of 10 • C/min to 280 • C, and held for 5 min. The mass spectrometer was operated in the electron ionization mode at 70 eV with a source temperature of 280 • C, and a continuous scan from 35 to 400 m/z was used. The analysis was performed in full-scan mode. Mass spectral data of the volatile compounds were compared with data in the National Institute of Standards and Technology (NIST) Mass Spectrum Library.

Construction of Fungal Deletion and Complementation Strains
Gene deletion vector construction and transformation of A. solani were generated by the double-joint PCR method with minor modification (Yu et al., 2004). The primers used for flanking sequences amplification for each gene are listed in Supplementary Table S4. Open reading frames (ORFs) of slt2 and sod were replaced with a hygromycin resistance cassette (hyg) and the constructed fragment was inserted into the pEASY-T1 cloning vector (Supplementary Table S2). After transforming the constructed plasmid into HWC-168, the subsequent deletion mutants were verified by PCR with slt2-F/R and sod-F/R (Supplementary Table S4). For complementation, the respective ORFs were fused to a neomycin selection marker (neo) and introduced into the corresponding deletion mutants. Specific primers slt2-F/R and sod-F/R and marker gene primers neo-F/R (Supplementary Table S4) were used for verification.

Quantitative Real-Time PCR
Total RNAs of A. solani cells co-cultured with volatiles produced by ZD01 after 3 and 4 days were extracted by using the Bacterial RNA Kit (Omega Bio-Tek, Norcross, GA, United States) according to the manufacturer's instructions. First-strand cDNA was obtained using reverse transcriptase (TransGen Biotech, Beijing, China) with random hexamer primers. Real-time PCR was performed with SYBR Premix Ex Taq TM (TransGen Biotech, Beijing, China). ITS gene was used as an internal reference gene. The specific primers used are listed in Supplementary Table S4. The relative expression of specific genes was calculated by using the 2 − CT method (Livak and Schmittgen, 2001).