Isolated Bacillus subtilis strain 330-2 and its antagonistic genes identified by the removing PCR

Plant growth-promoting bacteria (PGPB) may trigger tolerance against biotic/abiotic stresses and growth enhancement in plants. In this study, an endophytic bacterial strain from rapeseed was isolated to assess its role in enhancing plant growth and tolerance to abiotic stresses, as well as banded leaf and sheath blight disease in maize. Based on 16S rDNA and BIOLOG test analysis, the 330-2 strain was identified as Bacillus subtilis. The strain produced indole-3-acetic acid, siderophores, lytic enzymes and solubilized different sources of organic/inorganic phosphates and zinc. Furthermore, the strain strongly suppressed the in vitro growth of Rhizoctonia solani AG1-IA, Botrytis cinerea, Fusarium oxysporum, Alternaria alternata, Cochliobolus heterostrophus, and Nigrospora oryzae. The strain also significantly increased the seedling growth (ranging 14–37%) of rice and maize. Removing PCR analysis indicated that 114 genes were differentially expressed, among which 10%, 32% and 10% were involved in antibiotic production (e.g., srfAA, bae, fen, mln, and dfnI), metabolism (e.g., gltA, pabA, and ggt) and transportation of nutrients (e.g., fhu, glpT, and gltT), respectively. In summary, these results clearly indicate the effectiveness and mechanisms of B. subtilis strain 330-2 in enhancing plant growth, as well as tolerance to biotic/abiotic stresses, which suggests that the strain has great potential for commercialization as a vital biological control agent.


Results
Identification and properties of the strain. The isolated strain was catalase-positive, gram-positive, rodshaped, aerobic, and motile and had distinct fermentation profiles for different carbon sources (Supplementary Table 1). Based on the observed phenotypic characteristics, the isolated strain was grouped into the genus Bacillus. Furthermore, it exhibited high levels of similarity (99%) to the closest known species in the database. Phylogenetic analysis of the 16S rDNA gene sequence revealed that the isolated 330-2 strain was Bacillus subtilis. In addition to B. subtilis, another species of the genus Bacillus, B. amyloliquefaciens, showed (99%) similarity with the strain during the BLAST search. The phylogenetic tree analysis using Mega5 version 5.2 showed that strain 330-2 was more closely related to B. subtilis and B. amyloliquefaciens (Fig. 1). To further confirm the strain, a BIOLOG test was performed.
BIOLOG identification of the bacterial strain. Phenotypic analyses was done using 96-well BIOLOG GENIII MicroPlates assay. The list of carbon substrates used by B. subtilis sp. is presented ( Fig. 2 and Supplementary Table 1). The results of the 16S rDNA sequencing and BIOLOG identification indicated that the strain used in this study was B. subtilis.

Production of hydrolytic enzymes and metabolites by bacterial antagonists.
Progression of the pink color in the presence or absence of tryptophan in the culture broth showed the production of IAA. Maximum IAA production was recorded as 24.13 μg ml −1 in the B. subtilis strain 330-2, followed by 2.10 μg ml −1 in Bacillus subtilis strain 168 broth culture (B. subtilis strain 168 was taken as control). The B. subtilis strain 330-2 also produced the siderophore through the development of orange halo zones around their spot of inoculation on Chrome-azurol S (CAS) medium, while B. subtilis strain 168 did not produce a siderophore.
The B. subtilis strain 330-2 showed a positive result for phosphate solubilization by spot inoculation in the solid state of Pikovskaya's media, but did not solubilize the Di-calcium phosphate (DCP) by spot inoculation on Pikovskaya's medium. On the plates, B. subtilis strain 330-2 P-solubilizing zones around the bacterial colony were 4.2 and 6.3 mm using Tri-calcium phosphate (TCP) and zinc phosphate (ZP), respectively. However, B. subtilis strain 168 could not solubilize the phosphate in Pikovskaya's solid medium. Both strains were tested for phytase production, whereas B. subtilis strain 330-2 solubilized calcium and sodium phytate by formation of halo zones around its spot of inoculation (12.7 and 9.6 mm for calcium and sodium phytate, respectively), suggesting the release of free P. The B. subtilis strain 168 did not solubilize the sodium phytate, except on calcium phytate-amended medium through the formation of a 6.6 mm halo zone around the bacterial colony.
The B. subtilis strain 330-2 solubilized zinc oxide and zinc phosphate, which led to the formation of clear halo zones of 12.26 and 13.5 mm around the bacterial spot on the respective medium compared to B. subtilis strain 168, which did not solubilize zinc oxide and zinc phosphate. Both strains showed positive protease activity by producing clear halo zones of 29.5 mm and 17.8 mm around the spot inoculation. Both strains grew on laminrain azure and cellulose-amended minimal media, except strain B. subtilis strain 168, which did not grow on cellulose-amended medium (Table 1 and Fig. 3).
In vitro antagonistic activity assays. The B. subtilis strain 330-2 illustrated the highest antifungal activity against all the phytopathogens listed (Fig. 4). The data on mycelium growth was recorded based on the respective control treatment (4-7 days) and revealed considerable reductions in the growth of fungal mycelia under the influence of B. subtilis strain 330-2. The maximum inhibition (%) in the radial growth by B. subtilis strain 330-2 was 49.53% for C. heterostrophus, followed by 47.96% for A. alternata, 44.57% for B. cinerea, 42.25% for F.   Generation of differentially expressed clones by R-PCR. The sequence identities between the tester and driver were found to be <87% as determined under diversifying differentiation with a specified selection 33 . Therefore, 87% identity was used as one of the evaluating criteria to determine R-PCR cloned genes. In the recent study, the R-PCR efficiency was estimated to be 92.10% (Supplementary Table 3). To underpin the removal mechanisms of R-PCR at the genomic DNA sequence level, both the tester and driver sequences were compared using MEGA 5 version 5.2 and it was found that, in the three main cases, the driver could not eliminate the tester ( Fig. 6 and Supplementary Fig. 3). The sequence identities of the R-PCR cloned genes between tester and driver are presented (Supplementary Table 4).
Five genes involved in abiotic stress tolerance, e.g., pabA, the precursor of tryptophan (indole-3-acetic acid synthesis); bioA, responsible for biotin biosynthesis; narI, nitrate reductase production; pckA was shown to be involved in nitrogen fixation, ispH is responsible for isoprenoid biosynthesis, were found to be differentially expressed in B. subtilis strain 330-2 were found to be differentially expressed in B. subtilis strain 330-2. All these processes are critical for the normal growth and development of plants.  Table 2).

Discussion
The present study was conducted to isolate an antagonistic B. subtilis strain 330-2 that can efficiently control a wide range of phytopathogens. The biochemical, physiological and phylogenetic analyses of the 16S rDNA gene sequences and BIOLOG test confirmed that strain 330-2 was B. subtilis. The BIOLOG micro plate assay has mainly been used for bacterial biocontrol agents in unraveling their specific carbon sources. The isolation of endophytic Bacillus species from various crops has been extensively studied 7,34 . B. subtilis strain 330-2 produced significant amounts of IAA (24.13 µg ml −1 ), which might play an important role in seedling growth and development 35 . Eighty percent of the soil rhizosphere bacteria can produce IAA, whereas almost all (98%) of the PGPR strains isolated from the plant rhizosphere were able to produce IAA 36 . Significant amounts of in vitro IAA production by Bacillus spp. have also been documented by Singh et al. 37 . The production of siderophores by B. subtilis strain 330-2 demonstrated its antagonistic nature against various pathogens along with plant growth regulation. In the present study, B. subtilis strain 330-2 was found to produce phytase and was a good solubilizer of phosphate and zinc 38 . Phytase has been identified and characterized previously from several gram-positive and gram-negative soil bacteria, such as B. subtilis 39 , B. laevolacticus, Pseudomonas sp. 40,41 , and Enterobacter sp. 42 . Extracellular phytase from B. amyloliquefaciens FZB45 enhanced the growth of maize seedlings under in vitro conditions and, in another study, phytase producing bacterial isolates promoted the plant growth of Indian mustard 43,44 .
Because of the restricted radial growth of the selected phytopathogens, formation of inhibition zones by B. subtilis strain 330-2 might be linked to the production of antimicrobial compounds such as antibiotics, cellulase, laminarinase, protease, siderophore production, as well as nutrient competition 37 . Notably, B. subtilis strain 330-2 secreted a complex of hydrolytic enzymes, including β-1,3-glucanase, β-1,4-glucanase and proteases, which possibly degrade the contents of the fungal cell wall, such as β-1,3-glucan and glucosidic bonds. However, rare literature is available on the production of hydrolytic enzymes by microorganisms 45,46 . Therefore, it is assumed that the cell wall lysis of the pathogenic fungi is due to the coordinated action of hydrolytic enzymes, such as laminarase, cellulase and protease 47 . In the present study, B. subtilis strain 330-2 was quite effective against all six of the pathogens tested, indicating its potential efficacy as a biocontrol agent and suggesting its use as a bioinoculant for banded leaf sheath in both rice and maize.
The efficiency of the isolated strain in enhancing plant growth was further assessed in a screen-house experiment. B. subtilis strain 330-2 was found to produce a considerable amount of IAA and significantly enhance the growth of plants compared with the control. For instance, compared with the control, the B. subtilis strain 330-2-treated rice/maize seedlings showed 12-34% enhancements in seedling elongation and biomass accumulation (Fig. 5a-f). It has been stated that the endophytic Bacillus (B. megaterium strain HNSQJYH124, B. subtilis  strain HNSQJYH135, and B. atrophaeus strain HNSQJYH170) significantly enhanced root length, shoot/root fresh and dry weight and chlorophyll content in wheat seedlings when inoculated with Bacillus 7 . B. subtilis and P. aeruginosa increased the plant growth of African spinach, okra and tomato in terms of shoot length and dry biomass 48 . More recently, it was also found that the inoculation of plants with PGPB enhanced the root colonization and plant growth 4,49 . The mechanisms underlying the promotion of plant growth are diverse, including synthesis of phytohormones and signaling molecules, improvement of the plant mineral nutrition and production of siderophores and volatile compounds 50 . It has been reported that Bacillus spp. can elicit induced systemic resistance (ISR), which may result in the promotion of plant growth 51 . It was clear from the results that B. subtilis strain 330-2 is safe for the health of plants without any adverse effects. It is suggested that the B. subtilis strain 330-2 used in this study could be used as a PGPB 52 .
To assess the differential expression of genes, the most advanced and authentic R-PCR method was used to compare the B. subtilis strain 330-2 with B. subtilis 168. In total, 114 single clones were differentially expressed by R-PCR, suggesting the involvement of many genes in biotic and abiotic stress tolerance, as well as growth enhancement. Approximately 10% of fragments, including polyketide synthases (PKS) and non-ribosomal peptide synthetases (NRPSs), were involved in the synthesis of secondary metabolites and defense mechanisms for the PGPB-induced suppression of plant pathogens (Supplementary Table 2). Biosynthesis of bioactive lipopeptides plays an important role in the interaction of Bacillus species and plants either by suppressing the soil-borne pathogens in the rhizosphere or/and enhancing the plant defense mechanisms. Biosynthesis of bioactive lipopeptides occurs due to various non-ribosomal peptide synthetases (NRPSs) that have antagonistic activities 53 . Activation of several PKS/NRPS genes related to the production of antibiotics (e.g., srfAA) plays a key role in root colonization by emulsifying, foaming and decreasing the surface tension of solids to assure proper root colonization and a suitable environment that facilitates proliferation 54 . In a previous study, it has been reported that srfAA, macrolactin polyketide synthase, difficidin synthase and fengycin synthetase were induced in response to the root exudates and were directly linked to the biological control activity of B. amyloliquefaciens FZB42 55 . Bacteria in the rhizosphere sense the root exudates that are released by the plants via methyl-accepting proteins, which activates motility-related genes (e.g., hag, mcpA) and then move to the root surface for attachment 56 . In the present study, several genes associated with cell motility and biofilm formation (e.g., epsB, wapA, spo0A) were differentially expressed in B. subtilis strain 330-2. They are suspected to enhance the ability of B. subtilis strain 330-2 to exploit various plant-derived polysaccharides in the rhizosphere. These differentially expressed genes might favor host plant sensing and defense against pathogens in plants treated with B. subtilis strain 330-2. Iron has been reported to play a key role in bacterial biofilm formation 57 . In this study, the ferrichrome transport system permease protein (fhuG) gene was differentially expressed by strain 330-2 (Supplementary Table 2), which is known to be involved in scavenging iron from environmental sources 58 .
In the natural environment, B. subtilis is frequently faced with different stress factors, including salinity, oxidative stress, heat stress, and deprivation of nutrients 59 . These stress factors negatively influence the growth and productivity of living organisms, including plants and bacteria, leading to significant economic losses globally. Thus, it is useful to develop stress-tolerant species and to understand the underlying mechanisms 60 . Nine genes Figure 8. Venn diagram indicating the common genes shared by salinity, oxidative, heat, cold and drought stresses. In total, 12 common differentially expressed genes were identified in all the five abiotic stresses. associated with abiotic stress tolerance (e. g., yuaI, gpmI, glpT, coxA, glcD, yceD, yugJ, ggt, metK) were differentially expressed in B. subtilis strain 330-2. Specifically, approximately 7, 5, 3, 2, and 1 genes related to tolerance against salinity, oxidative, heat, drought cold and oxidative stress, respectively, were differentially expressed in B. subtilis strain 330-2 (Fig. 8). These differentially expressed genes belonged to diverse functional classes, such as photosynthesis, lipid metabolism (G3P), stress/defense carbohydrate metabolism, competition and membrane transport (Supplementary Table 2). In the past, it has been reported that inoculation with PGPR enhanced the drought tolerance via inducing transcription of drought response genes 61 , affecting the phytohormonal balance 62 and sugar accumulation 63 . Glycerol-3-phosphate (glpT), an important component of carbohydrates as well as lipid metabolic processes, enhanced the basal defense against Colletotrichum higginsianum 64 . Likewise, the involvement of yuaI, and gpmI against salinity stress is evident 65,66 . In the present study we found that the sporulation cortex protein (coxA) was differentially expressed in the strain 330-2, which is involved in low temperature stress 67 .
The genes differentially expressed by B. subtilis strain330-2 are involved in oxidative stress (Supplementary Table 2) responses and the activation of different cellular defense mechanisms. These genes primarily include glycolate oxidase (glcD), tellurium resistant (yceD), which is important for tellurite resistance 68 , putative NADPH-dependent butanol dehydrogenase (yugJ), which acts in direct detoxification of hydrogen peroxide 69 , and gamma-glutamyltranspeptidase (ggt), for resistance to organic peroxides 70 . Furthermore, we observed that metK, the precursor of ethylene and polyamine biosynthesis 71 , was differentially expressed by B. subtilis strain 330-2. This gene has been known to trigger tolerance against salinity 72 and drought stresses 73 . In conclusion, we have presented strong evidence that using the R-PCR approach allowed the identification of a substantial number of resistant genes in biotic/abiotic stresses.

Methods
Strains and growth conditions. The Bacillus subtilis strain 330-2 used in this study was isolated from the rapeseed in Wuhan, China and maintained at −80 °C. The Bacillus subtilis strain 168 and phytopathogenic fungi, Rhizoctonia solani AG1-IA, Fusarium oxysporum, Botrytis cinerea, Alternaria alternata, Cochliobolus heterostrophus, and Nigrospora oryzae, were obtained from the Department of Plant Pathology, Huazhong Agricultural University, Wuhan, 430070, P.R. China. The bacterial strain B. subtilis strain 168 was initially streaked from −80 °C glycerol stocks onto LB medium plates and then was cultured from a fresh single colony into LB broth. The fungal cultures were maintained on potato dextrose agar (PDA) by sub-culturing at regular interval and incubated at 28 °C.
Plant and soil sample collection. Samples of maize (Zea mays L.), wheat (Triticum astevam L.), rice (Oryza sativa L.), wild rapeseed (Brassica napus) and rhizosphere soils were collected from the field of Huazhong Agricultural University, Wuhan, 430070, P.R. China. Briefly, for endophytic isolation plant tissue samples were washed with running tap water to remove adherent soil particles and microbes. Samples were cut into 1-2 cm small pieces aseptically and surface sterilized with 70% ethanol for 1 min, 1.2% NaClO solution for 5 min, followed by 70% ethanol for 1 min. Then, samples were washed three times with sterilized distilled water and surface dried on sterilized filter paper. Surface sterilized samples were ground to a slurry with a sterilized mortar and pestle containing sterilized quartz sand. One gram each of rhizopheric soil, crushed roots, crushed leaves and crushed stem were homogenized separately in 9 ml saline (0.85% NaCl) in 20 ml test tube. The suspensions were homogenized, serial diluted up to 10 −5 . For each of these dilutions, 0.1 ml was spread on different Luria Bertani (LB) medium plates and then incubated at 28 °C for 4 days. The water from the last rinse was used as a control. Then, the colonies with different morphology were selected and purified by sub-culturing using maximum possible antiseptic condition 74 . Pure cultures were examined for growth studies.
Isolation and identification of the bacterial strain. The bacterial B. subtilis strain 330-2 used in this experiment was isolated from wild rapeseed. B. subtilis strain 330-2 indicated the highest inhibitory effect in vitro against R. solani and other fungi. Phenotypic profiling of the B. subtilis strain 330-2 was tested using BIOLOG GenIII MicroPlates (Model EL311, BioTek Instruments, USA), containing 71 carbon source utilization assays and 23 chemical sensitivity assays according to the manufacturer's protocol. The 10 μl microbial suspension was inoculated into each well of the GENIII MicroPlates. The plates were incubated at 30 °C for 24 to 48 h and observed for color development at 12 h intervals. Data were recorded at a 590 nm wavelength and the color development in the plates was read and compared in the BIOLOG database (Microlog SystemTm, Release 4.0) to identify the strain to species level 75 . The 16S rDNA gene of the strain was amplified using universal primer, both forward 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and reverse 1492R (5′-GGTTACCTTGTTACGACTT-3′) 76 . These primers were synthesized by T-Singke, Wuhan, China. The total volume (50 μl) of the PCR reaction mixture contained 5 μl of 10 × Taq buffer, 1.5 μl of MgCl 2 (25 mmol L −1 ), 4 μl of each dNTPs (10 mmol L −1 ), 2 μl of each primer (30 pmol), 1 μl of DNA template, 0.25 μl of Taq DNA polymerase, and 35.75 μl of ddH 2 O. The PCR reaction was performed using MyCycler Thermal Cycler (Bio-RAD, Foster, California, USA) with the following thermal program: 95 °C for 5 min, followed by 94 °C for 30 s, 50 °C for 30 s, and 72 °C for 1 min 30 s. These steps were repeated for 30 cycles, and a final 10 min extension at 72 °C. The molecular size of the resulting PCR product was analyzed on a 1.0% agarose gel to confirm that the fragment was 1.5 kb, and the amplified 16S rDNA fragment was sequenced by T-Singke, Wuhan, China. The 16S rDNA sequence results were analyzed and compared on the website, http://www.ncbi.nlm.nih.gov 77 . of Salkowski reagent (50 ml, 35% of perchloric acid, 1 ml 0.5 M FeCl 3 solution). The appearance of a pink color in the supernatant confirmed the production of IAA. After 30 min, the optical density (OD) of the mixture was read at an absorbance of 530 nm in a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific Inc., Wilmington, USA). The amount of IAA was estimated using a standard curve.

Indole
Siderophore production detection. Siderophore production of the strains was determined on blue agar plates containing Chrome-azurol S (CAS) medium following Schwyn and Neilands' method 79 . The bacterial strains (24 h-old culture) were spotted on CAS medium and incubated at 28 °C for 4 days. Formation of an orange halo around the colony was the indicator of siderophore production.
Phosphate solubilization. Phosphate solubilization of the isolated strain 330-2 and B. subtilis strain 168 was detected by spotting these strains separately on Pikovskaya's (PVK) agar plates using zinc phosphate (ZP), di-calcium phosphate (DCP), and tri-calcium phosphate (TCP) as sources of insoluble inorganic phosphate. Plates were incubated at 28 °C for 7 days 80 . The halo zone and colony diameter were measured and data were recorded. The phosphate solubilization index (S.I.) was calculated using the equation: S.I. = (colony diameter + halo zone diameter)/colony diameter 81 .
Phytase production detection. Phytase production of the strains was evaluated by spot inoculation of log phase culture on phytase screening media using sodium and calcium phytate as sources of insoluble organic phosphate. The plates were incubated at 28 °C for 7 days. The halo zone around the colonies was observed on the plates.

Zinc solubilization.
To evaluate zinc solubilization abilities, the log phase culture of the bacterial strains was spotted on Tris-minimal medium plates containing zinc phosphate and zinc oxide as sources of insoluble inorganic zinc. The inoculated plates were then incubated at 28 °C for 7 days. Clearing zones around the colonies were observed for zinc solubilization.
Cell wall degrading enzymes. β-1,3-glucanase and β-1,4-glucanase activities were studied using minimal medium (MM) containing laminarin azure (Sigma-Aldrich Co., USA) and cellulose powder as sole sources of carbon, respectively. Laminarin azure and cellulose amended minimal medium indicate the production of β-1,3-glucanase and β-1,4-glucanase, respectively. The medium plates were incubated at 28 °C for 7 days. Protease production was assessed by spotting each of the bacterial strains on Skim Milk Agar (SMA) medium plates. The bacterial strains that produced protease were identified by halo zones around the bacterial colonies 82 . In vitro antagonistic activity assays. The strain was tested for its ability to inhibit the growth of R. solani AG1-IA, B. cinerea, F. oxysporum, A. alternata, C. heterostrophus, and N. oryzae in vitro. The B. subtilis strain 330-2 was dabbed at two corners 3 cm from the center by stabbing the plate with sterilized tooth pick. After 24 h, fresh mycelium of test fungi (0.8-cm diameter) was placed in the center of the PDA plates. All treatments were conducted in triplicate. Plates were incubated at 28 °C. The inhibition of fungal growth was monitored by recording the diameter of the inhibition zone (mm).The percentage of inhibition was calculated using the following formula 83 . where "C" is the mycelium diameter (cm) of the fungus growing on the control plates (without bacteria) and "T" is the mycelium diameter (cm) of the fungus growing in the bacterial-treated plates. The experiment was repeated thrice with three biological replicates each time.
Plant growth assay. B. subtilis strain 330-2 was grown in 100 ml LB medium in a rotary shaker (170 rpm) at 37 °C, until they reached an optical density at 600 nm (OD600) of 0.8. The bacterial cells were harvested by centrifugation at 6000 × g for 5 min at 4 °C and resuspended in 100 ml of distilled water and adjusted to (10 10 CFU ml −1 ). Seeds of rice (Oryza sativa, L.) (variety: Huanghuazhan) and corn (Zea mays L.) (variety: B73) were obtained from the College of Plant Science and Technology of Huazhong Agricultural University, Wuhan, China. The seeds were surface sterilized with 70% alcohol for 1 min, followed by 0.5% NaClO solution for 1.30 min and 70% alcohol for 1 min. The seeds were then washed three times with distilled water. The sterilized seeds were bio-primed with bacterial culture for 4 h (control seeds were treated with distilled water). Seeds were then placed in the sterile petri dishes and dried back to their original moisture contents at room temperature. After bio-priming, seed germination tests were carried out. Ten rice seeds and three maize seeds for each treatment were placed in plastic pots of 7 cm diameter and 15 cm diameter, respectively and were filled with sterilized soil. Each treatment was replicated three times, respectively. Then, pots were incubated in a growth chamber at 28 °C with a photoperiod of 16 h light and 8 h dark. Data were calculated for shoot length, root length, fresh weight and dry weight.
B. subtilis strain 330-2 effects in tobacco plants. The addition of the PGPB, which aimed to trigger plant growth, should not possess any detrimental effect on other plant species. B. subtilis strain 330-2 was tested in tobacco plants as a well-studied model and widely cultivated crop species 84 . Tobacco seeds were surface sterilized and grown in autoclaved soil in plastic pots (7 cm diameter). When the seedling shoots reached 1 cm, they were inoculated with a bacterial suspension (10 8 -10 9 CFU ml −1 ). Four seedlings per condition were sampled on 7, 14, and 21 days, and data were recorded on plant height, fresh weight, and dry weight post inoculation. M9 buffer without bacterial inoculum was used as a negative control.
Generation of differentially expressed clones by R-PCR. The removing PCR (R-PCR) is a restriction enzyme-based method, which eliminates unwanted genes from the gene population very efficiently. Using this method, the undesirable genes can be removed by means of removing drivers. This recently developed R-PCR technique was used to isolate the differentially expressed antagonism-related genes by comparing genomic DNA populations of the B. subtilis strain 330-2 and B. subtilis strain 168. To identify specific genes in B. subtilis strain 330-2, the B. subtilis strain 330-2 and B. subtilis strain 168 were taken as a tester and driver, respectively. The B. subtilis strain 168 is a well-known model strain without roles in enhancing plant growth and tolerance to biotic/ abiotic stresses. The R-PCR was performed between the 'driver' and 'tester' according to the previous method of Huan et al. 33 . All PCR steps were performed on MyCycler Thermal Cycler (Bio-RAD, Foster, California, USA). Bands of the second R-PCR were excised from the gel, and PCR products were purified using the QIAquick gel purification kit (QIAGEN, Germany). The purified PCR products were cloned in pmD18-T Easy vector following manufacturer's protocol (Promega) and transformed into DH5α cells. Plates were incubated at 37 °C until small colonies were visible. Sequencing of the positive clones were done by T-Singke, Wuhan, China using primers both forward M13F (−47) (5′-CGCCAGGGTTTTCCCAGTCACGAC-3′) and reverse M13R (−48) (5′-AGCGGATAACAATTTCACACAGGA-3′).

Sequences analysis.
A total of 180 white-clones were picked from the library. The inserted fragments were amplified, and fragments of the differentially expressed clones were sequenced using the primer pair of M13F and M13R. Sequences were trimmed of vector using DNAMAN software. Homology searches of all sequences were queried in the GenBank database with nucleotide blasts available at the NCBI (http://www.ncbi.nlm.nih. gov/BLAST).
Statistical analysis. Data were analyzed by one way analysis of variance (ANOVA) using STATISTIX software (8.1) (Analytical Software, Tallahassee, FL, USA). The mean variance was analyzed using a least significant difference (LSD) test at 0.05 probability level.