DNA Polymorphisms and Biocontrol of Bacillus Antagonistic to Citrus Bacterial Canker with Indication of the Interference of Phyllosphere Biofilms

Citrus bacterial canker caused by Xanthomonas axonopodis pv. citri is a devastating disease resulting in significant crop losses in various citrus cultivars worldwide. A biocontrol agent has not been recommended for this disease. To explore the potential of bacilli native to Taiwan to control this disease, Bacillus species with a broad spectrum of antagonistic activity against various phytopathogens were isolated from plant potting mixes, organic compost and the rhizosphere soil. Seven strains TKS1-1, OF3-16, SP4-17, HSP1, WG6-14, TLB7-7, and WP8-12 showing superior antagonistic activity were chosen for biopesticide development. The genetic identity based on 16S rDNA sequences indicated that all seven native strains were close relatives of the B. subtilis group and appeared to be discrete from the B. cereus group. DNA polymorphisms in strains WG6-14, SP4-17, TKS1-1, and WP8-12, as revealed by repetitive sequence-based PCR with the BOXA1R primers were similar to each other, but different from those of the respective Bacillus type strains. However, molecular typing of the strains using either tDNA-intergenic spacer regions or 16S–23S intergenic transcribed spacer regions was unable to differentiate the strains at the species level. Strains TKS1-1 and WG6-14 attenuated symptom development of citrus bacterial canker, which was found to be correlated with a reduction in colonization and biofilm formation by X. axonopodis pv. citri on leaf surfaces. The application of a Bacillus strain TKS1-1 endospore formulation to the leaf surfaces of citrus reduced the incidence of citrus bacterial canker and could prevent development of the disease.

Bacillus species are genotypically diverse organisms. The comparison of small-subunit ribosomal RNA sequences reveals the presence of five genetically distinct groups in the genus [23]. Those Bacillus strains that are known to have the potential to protect plants from pathogens or pests or stimulate plant growth are attributed to two groups, the B. cereus group and the B. subtilis group. The B. cereus group includes B. anthracis, B. cereus, B. thuringiensis, B. mycoides, B. pseudomycoides, and B. weihenstephanesis; the B. subtilis group includes B. subtilis, B. pumilus, B. atrophaeus, B. licheniformis and B. amyloliquefaciens [23]. Many Bacillus species are generally considered harmless, and B. subtilis has even been granted GRAS status by the United States Food and Drug Administration (US FDA). However, B. anthracis can cause anthrax in humans and cattle, and B. cereus is known to produce enterotoxins that cause food poisoning [21]. Molecular techniques, including 16S rRNA gene sequencing, DNA polymorphism analyses by tDNA-PCR for the tDNA-intergenic spacer region, ITS-PCR for the 16S-23S intergenic transcribed spacer region, and repetitive element sequence-based PCR (rep-PCR) using the ERIC2, BOXA1R and (GTG) 5 primers [24,25,26], have been developed for rapid species identification of the Bacillus genus.
Citrus fruits are of economic importance worldwide [27,28]. The major bacterial disease of citrus, citrus bacterial canker, is caused by X. axonopodis pv. citri [29], for which the currently published nomenclature is X. citri subsp. citri [30]. To control this disease, copper salts and antibiotics are suggested [31]; however, several Xanthomonas strains have been found to both of these methods [32]. Thus, the development of alternative control strategies for this disease is necessary.
Microbial communities attached to a surface are referred to as biofilms [33]. The synergistic or antagonistic interactions between biofilm organisms and their respective hosts can contribute to the successful establishment of symbiotic or pathogenic relationships [34]. Consequently, interfering with bacterial biofilm formation has been suggested as a novel strategy for disease control [35,36,37]. It has been shown that biofilm formation was necessary for epiphytic fitness and canker development by the phytopathogen X. axonopodis pv. citri [38]. For the beneficial antagonist, root colonization plays a key role in the interaction of B. subtilis with Arabidopsis and the pathogen Pseudomonas syringae [18]. Our previous study indicated that antagonistic B. amyloliquefaciens WG6-14 was a potential biopesticide for controlling citrus bacterial canker (unpublished data), and an endospore formulation of this antagonist has been officially recommended for controlling bakanae disease of rice in Taiwan. However, the interaction of X. axonopodis pv. citri and antagonistic Bacillus species in the phyllosphere of citrus has not been investigated.
In this study, native bacilli isolated from potting mixes, organic compost, and soil in Taiwan were assessed for antagonistic activity against citrus canker bacteria. The genetic identities determined by rDNA sequences of bacilli from Taiwan, their respective type strains, and other industrial strains were compared. DNA polymorphisms were determined by molecular typing of the 16S-23S intergenic transcribed spacer region, tDNA intergenic spacer length analysis and repetitive element sequence-based PCR. In addition, the efficacy of reducing disease incidence by application of Bacillus species and the interaction between the antagonist and the pathogen in the phyllosphere of citrus were investigated.

Sequence and phylogenetic analyses of 16S rRNA genes in native Bacillus species
To identify the Bacillus strains, each strain was subjected to physiological and biochemical characterization using the methods described in Bergey's Manual of Systematic Bacteriology [39] and was identified using the Biolog system (Biolog Inc., CA, USA). The physiological and biochemical tests included Gram staining, endospore staining, starch hydrolysis, Voges-Proskauer test, the oxidase-fermentation test, gelatin hydrolysis, citrate utilization, nitrate reduction, arginine dihydrolase activity, growth in 7% sodium chloride and growth at 50uC. Strains HPS-1, OF3-16, SP4-17, TKS1-1, WP8-12, and WG6-14 all showed positive reactions for these tests were classified as B. licheniformis (data not shown). Strain TLB7-7 did not hydrolyze starch or reduce nitrate and was classified as B. pumilus (data not shown). The results of the Biolog analysis indicated that strain HSP-1 was B. licheniformis, strain SP4-17 was B. megaterium, strains TKS1-1 and WP8-12 were B. subtilis, strain TLB7-7 was B. pumilus and strain WG6-14 was B. amyloliquefaciens; strain OF3-16 could not be classified using the Biolog system (Table 1). Thus, the species attributes for most of the strains were designated based on Biolog analysis except for strain OF3-16, which was based on the physiological and biochemical characteristics described in Bergey's Manual.
ITS-PCR, tDNA-PCR, and rep-PCR fingerprint and cluster analysis of Bacillus species ITS-PCR, tDNA-PCR and rep-PCR fingerprinting have been used to differentiate isolates among a wide range of bacterial and fungal genera and species as well as to study genomic diversity [24,40,41,42]. To evaluate the DNA polymorphisms of Bacillus species native to Taiwan and their respective type stains, ITS-PCR using the primers L1 and G1 to amplify the 16S-23S intergenic transcribed spacer region, tDNA-PCR using the primers T5A and T3B to amplify the tDNA-intergenic spacer region, and rep-PCR analyses using the primers ERIC2, BOXA1R and (GTG) 5 as described by Freitas et al. [24] were performed. DNA polymorphisms were assessed four times with reproducible results. ITS-PCR fingerprinting and unweighted pair group method with arithmetic mean (UPGMA) cluster analysis classified all tested strains into 4 distinct groups. Bacillus strains SP4-17, WP8-12, WG6-14, and TKS1-1, which showed the greatest antagonistic activity, were in a cluster with the B. amyloliquefaciens type strain BCRC11601 (Fig. 3). However, the reference strains B. subtilis subsp. subtilis BCRC10255, B. licheniformis BCRC11702, B. subtilis subsp. spizizenii BCRC80045, B. pumilus BCRC11706, and B. cereus UW85; the strains NT-A1, NT-B1 and NT-2 isolated from Japanese natto and the native strains HSP1, OF3-16, and TLB7-7 all showed the same ITS-PCR fingerprint patterns. These results suggest that ITS-PCR fingerprint analysis was not able to differentiate Bacillus isolates at the species level and discriminate B. subtilis from B. cereus.
Using tDNA-PCR fingerprinting, the tested strains showed nine pattern types (Fig. 4). Strains WG6-14, WP8-12, SP4-17 and TKS1-1 were homologous and showed the same DNA banding pattern as B. amyloliquefaciens BCRC11601; strains NT-2, NT-B1, NT-A1 and HSP-1 were homologous and showed the same DNA banding pattern as the B. subtilis type strains BCRC80045 and BCRC10255. Strains TLB7-7 and OF3-16 were designated as B. pumilus and B. licheniformis, respectively, according to Biolog analysis, 16S rRNA sequence analysis and physiological and biochemical characterization. These strains showed tDNA-PCR fingerprints that were distinct from their respective type strains.
Three sets of primers, ERIC2, (GTG) 5 and BOXA1R [40], were used for rep-PCR fingerprint analysis. Based on BOXA1R-PCR fingerprint analysis, ten banding patterns were observed (Fig. 5). Strain HSP-1 showed the same pattern as B. subtilis subsp. subtilis BCRC10255. Strains NT-B1, NT-2 and NT-A1 isolated from natto formed a cluster that was different from that of their close relatives, strains B. subtilis subsp. subtilis BCRC10255 and B. subtilis subsp. spizizenii BCRC80045. Strains SP4-17, TKS1-1, WG6-14 and WP8-12 were homologous and showed a unique banding pattern. Negative results were observed with BOXA1R- Figure 1. Antagonistic activity of Bacillus species against X. axonopodis pv. citri XW19. A 20 ml aliquot of X. axonopodis pv. citri XW19 suspension (OD 620 = 0.3) was spread on an SYB agar plate. Following overnight incubation at 30uC, 20 ml of Bacillus suspension (OD 620 = 0.3) was spotted (A) inside the stainless steel ring or (B) on a paper disc. The plates were incubated at 30uC, for 5 days. CK, 20 ml of sterile Milli-Q water was used as a control. The results represent the means and standard deviations (error bars) of a representative experiment. Different lowercase letters indicate significant differences (p,0.05) according to Tukey's HSD test. doi:10.1371/journal.pone.0042124.g001 PCR fingerprinting for strains B. cereus UW85 and B. pumilus BCRC11706. Both ERIC2-PCR and (GTG) 5 -PCR amplification were negative for most of the tested strains (data not shown). Of the three primer sets, BOXA1R-PCR showed unique patterns that could differentiate strains native to Taiwan from the reference strains. Moreover, strains SP4-17, TKS1-1, WG6-14 and WP8-12, which showed superior antagonistic activity against X. axonopodis pv. citri ( Fig. 1), had the same BOXA1R-PCR fingerprint, which was distinct from those of all reference strains.

Attenuated symptom development of citrus bacterial canker by treatment with B. subtilis and B. amyloliquefaciens
Our previous results indicated that the application of B. amyloliquefaciens WG6-14 endospores one day prior to inoculation with citrus canker bacteria reduced disease incidence from 97.7% to 3.03% (unpublished data). To assess the effect of B. subtilis TKS1-1 and B. amyloliquefaciens WG6-14 on the disease severity of citrus bacterial canker, Bacillus suspensions (overnight cultures diluted to an OD 620 of 0.3, ca. 10 8 CFU/ml) were sprayed on the leaves of Mexican lime 1 day prior to inoculation with X. axonopodis pv. citri TPH2 (overnight cultures diluted to an OD 620 of 0.3, ca. 10 8 CFU/ml), and the number of cankers per cm 2 on each leaf with and without Bacillus treatment was determined. Less severe canker symptoms or no symptom were observed on the Bacillustreated leaves compared to the water control ( Fig. 6 A). The number of cankers per cm 2 for the untreated control was 6.462.5, compared to 0.360.3 and 0.660.5 for the B. subtilis TKS1-1 and B. amyloliquefaciens WG6-14 treatments, respectively (Fig. 6 B). The number of cankers per cm 2 developing following with the application of Bacillus suspensions was significantly reduced by up to 6-fold (p,0.05). The effect of Bacillus on colonization and biofilm formation by citrus canker bacteria on leaf surfaces According to Rigano et al. [38] and our previous findings (unpublished data), biofilm formation is important for epiphytic survival and the development of canker disease. Colonization of the leaf surfaces of Mexican lime by X. axonopodis pv. citri strain TPH2 harboring a green fluorescent protein expressing plasmid, pGTKan (Table 1), was examined by confocal laser scanning microscopy. Individual cells attached to the surfaces of leaves submerged in bacterial suspension (overnight cultures diluted to an OD 620 of 0.05 in trypticase soy broth) were observed 1 day postinoculation, and microcolony and biofilm development were observed after 2 days (data not shown). Biofilms consisting of multicellular aggregates similar to those observed by Rigano et al. [38] were observed 1 day post-inoculation with X. axonopodis pv.
citri strain TPH2 harboring pGTKan on the leaf surfaces of Mexican lime grown in the greenhouse (Fig. 7 A and E). Bacterial aggregates could be observed surrounding and inside the stomata (Fig. 7 A). Treatment with B. subtilis strain TKS1-1 or B. amyloliquefaciens strain WG6-14 resulted in fewer X. axonopodis pv. citri cells attaching to the leaf surface compared to no treatment, and the cells were dispersed (Fig. 7 B and F, respectively). B. subtilis strain TKS1-1 and B. amyloliquefaciens strain WG6-14 cells were stained with acridine orange and showed red fluorescence (Fig. 7  C and G, respectively). The combined green and red fluorescent images indicated that small aggregates of Bacillus cells (red) were scattered around the X. axonopodis pv. citri cells (green) (Fig. 7 D and  H). These results suggest that by spraying antagonistic Bacillus 1 day prior to inoculation with the pathogen, colonization and Bacillus endospore formulations are effective in reducing the development of canker symptoms and the incidence of citrus bacterial canker disease B. subtilis TKS1-1 endospore formulations were applied to the leaves of navel orange trees grown in the greenhouse to assess disease control efficacy for citrus bacterial canker. The results indicated that the spray-application of an endospore formulation diluted 100-fold (final concentration ca. 10 9 spores/ml) was effective in reducing symptom development and disease incidence of citrus bacterial canker compared to no treatment (Fig. 8). The efficacy of treatments applied 24 h prior to pathogen inoculation and treatments applied post-inoculation on reducing disease incidence was similar, and was not significantly affected by the frequency of application.

Discussion
No known biocontrol agents have been developed for the disease management of citrus bacterial canker. To explore the potential of bacilli native to Taiwan to control this disease, Bacillus species with a broad spectrum of antagonistic activity against various phytopathogens were isolated from potting mixes, organic compost and rhizosphere soils. By dual culture assay, seven strains TKS1-1, OF3-16, SP4-17, HSP1, WG6-14, TLB7-7, and WP8-12 showing superior antagonistic activity were chosen for biopesticide development and for further investigation. Using established and patented methods, we mass-produced strain TKS1-1 endospores, and showed them to be effective in reducing the severity and incidence of citrus bacterial canker. In addition, an endospore formulation of strain WG6-14 reduced bacterial black spot of mango and bacterial leaf blight of rice (unpublished data). Endospore formulations of Bacillus strain WG6-14 have been commercialized and registered as biocontrol agents for rice  bakanae disease in Taiwan. As part of the safety requirements for biopesticide development, GRAS organisms are preferred as biocontrol agents. Our results, based on physiological and biochemical characteristics, 16S rDNA sequences and tDNA-PCR analyses, indicate that all seven native strains with antagonistic activity against X. axonopodis pv. citri and that demonstrated high efficacy in suppressing citrus bacterial canker disease were in the same clades as the type strains of the B. subtilis group that are listed as GRAS bacteria by the US FDA and that are distinct from strains of the B. cereus group [23].
ITS-PCR, tDNA-PCR and rep-PCR analyses have been successfully used to investigate the species and intraspecific variability of Bacillus species [24,26,41,43]. Of these molecular typing techniques, all of which were used in this study, rep-PCR analysis using the BOXA1R primer displayed the best resolving power for discriminating between native strains exhibiting superior antagonistic activity against X. axonopodis pv. citri and the reference strains. ITS-PCR analysis was not sufficient to distinguish strains of the B. subtilis group from B. cereus strain UW85 [23]. This result suggests that ITS-PCR analysis was not adequate for discriminating between Bacillus strains at the species level as was demonstrated by Freitas et al. [24]. In contrast, Wunschel et al. showed that the banding patterns generated by PCR analysis of the rRNA spacer region could distinguish B. subtilis from species in the B. cereus group but could not differentiate between species within the B. cereus group [25]. On the basis of cell wall constituents and DNA-DNA relatedness data, B. subtilis strains were reclassified into two subspecies: B. subtilis subsp. subtilis and B. subtilis subsp. spizizenii [44]. Our data indicate that these two subspecies were grouped into one cluster by tDNA-PCR analysis and two clusters by BOXA1R-PCR analysis. In addition, DNA polymorphisms in strains WG6-14, SP4-17, TKS1-1, and WP8-12, as revealed by rep-PCR using the BOXA1R primer, were similar to each other, but different from their respective type strains. These four strains were associated with the greatest antagonistic activity. Our results suggest that the DNA fingerprint generated with BOXA1R-PCR could be valuable not only for patenting or commercializing these Bacillus strains, but also for creating markers for the selection of antagonists against X. axonopodis pv. citri.
Epiphytic and root colonization are considered as the process of biofilm formation [45]. Bacterial biofilm formation has been shown to be necessary for epiphytic fitness, pathogenesis, antagonism and symbiosis with the host organism [34]. Thus, microbial infection control strategies could be developed based on interfering with biofilm formation [35,36,37]. Rigano et al. [38], as well as our unpublished data, demonstrated that biofilm formation by X. axonopodis pv. citri on the leaf surfaces of citrus was associated with the occurrence of citrus canker symptoms. Here, we  investigated the efficacy of using antagonistic bacilli to interfere with this process. Our data indicate that biofilm development by X. axonopodis pv. citri on the leaf surfaces of Mexican lime was reduced following spray inoculation of Bacillus strains WG6-14 and TKS-1 1 day prior to pathogen inoculation when compared to no treatment. Application of these two Bacillus strains to citrus leaves resulted in reduced symptom development, suggesting that these antagonistic bacilli are potential biocontrol agents for citrus bacterial canker disease. Taken together, these results suggested that control may be associated with the interference of colonization and biofilm formation by X. axonopodis pv. citri in the phyllosphere, which is the site of initial colonization and infection. The biocontrol efficacy of citrus canker disease by Bacillus strain TKS1-1 was further demonstrated by spray application of endospore formulations in the greenhouse. Rhizosphere-coloniz-ing B. subtilis 6051 forms a stable biofilm and secretes surfactin, which work together to protect Arabidopsis plants from infection by pathogenic P. syringae [18]. We did not exclude the possibility that Bacillus strains TKS1-1 and WG6-14 also may secrete surfactin [18], bacteriocins such as xanthobacidin [46], or other cyclic lipopeptides [47] because both strains inhibited the growth of X. axonopodis pv. ciri XW19. In addition, cyclic lipopeptides are reportedly involved in biofilm formation by Bacillus species [18,45,48]. Alternatively, some Bacillus species including B. subtilis, B. amyloliquefaciens, B. pumilus, B. mycoides, B. pasteurii, B. thuringiensis, and B. cereus apparently induce plant resistance [44]. Determinants for elicitating plant resistance responses have been demonstrated and include surfactins and fengycins [49] and volatile organic compounds such as 2,3-butanediol [50]. Our preliminary results also indicate that B. amyloliquefaciens strain WG6-14 produces  butanediol derivatives and that these volatile metabolites induce the expression of plant disease resistance genes such as those encoding phenylalanine ammonia lyase and pathogenesis related protein PR-1 in the leaves of rice plants (unpublished data). As another example, the control of Cercospora leaf spot on sugar beet by a phyllosphere-colonizing B. mycoides was attributed to its ability to induce systemic resistance [51].
In conclusion, our results demonstrate that Bacillus strains native to Taiwan, particularly strains WG6-14 and TKS1-1, can attenuate the symptoms and decrease the incidence of citrus bacterial canker disease. Because members of the B. subtilis group are GRAS bacteria, it would be safe to use these strains in the environment and maintain sustainability of the agricultural ecosystem. Biofilm formation as well as epiphytic colonization and survival are important for canker development in X. axonopodis pv. citri. The biocontrol efficacy of applying antagonistic bacilli to citrus leaves may be associated with their ability to interfere with colonization and biofilm formation by X. axonopodis pv. citri. Additionally, information obtained from molecular typing using the BOXA1R-PCR assay would provide DNA fingerprints valuable for patenting or commercializing these Bacillus strains.

Strains and growth conditions
The Bacillus and Xanthomonas strains and plasmids used in this study are listed in Table 1. Bacillus strains were routinely cultured on Difco TM potato dextrose agar (PDA, Becton Dickinson, Sparks, MD, USA) at 30uC. Xanthomonas strains were cultured on Difco TM Nutrient agar (NA; Becton Dickinson) or in trypticase soy broth (TSB; Becton Dickinson) at 27uC unless otherwise stated. When required, gentamicin (Gm; Sigma) was added to the medium at a concentration of 25 mg/ml. For the isolation of Bacillus strains, 1 gram of soil from the root rhizospheres of different plants, organic compost, potting mixes or natto was suspended in 5 ml of distilled water, heated at 80uC for 10 min, spread-plated on PDA and then incubated at 30uC for 1 day. Bacillus-like colonies were selected and tested for antagonistic activity against various phytopathogens including Pythium aphanidermatum, Rhizoctonia solani AG4, Xanthomonas axonopodis pv. vesicatoria XVT12, and X. axonopodis pv. citri XW19. Seven (HSP1, TKS1-1, OF3-16, SP4-17, WG6-14, TLB7-7 and WP8-12) of the 326 tested strains showed higher antagonistic activity than the remaining strains and were used for further study. To identify the Bacillus species, strains were subjected to physiological and biochemical characterization according to the methods described in Bergey's Manual [39]; they were identified using the Biolog system (Biolog Inc., CA, USA). For pathogenicity assays, X. axonopodis pv. citri strain TPH2 was generated by electroporating pBBR1MCS5 into X. axonopodis pv. citri strain XW19. For confocal laser scanning microscopy, pGTKan was electroporated into X. axonopodis pv. citri strain TPH2. Electroporation (12.5 kV/cm, 25 mF, 400 V) was performed using standard procedures [52].
Sequence and phylogenetic analysis of Bacillus 16S rRNA Genomic DNA was extracted from the Bacillus isolates using the WizardH genomic DNA purification kit (Promega, Madison, WI, USA) according to the manufacturer's instructions. The 16S rRNA genes were amplified by PCR using the primers 8F and 907R according to the conditions described by Freitas et al. [24], except that 26GoTag Master Mix (Promega) was used. The PCR products were then sequenced at the Automated DNA Sequencing Service Laboratory, National Chung Chung-Hsing University, Taiwan.
The 16S rRNA sequences were aligned using the Pileup program, SeqWeb version 3.1.2 (GCG Wisconsin Package, Accelrys Inc., San Diego, CA, USA). A distance matrix was generated by the Kimura 2-parameter method with the Dnadist program, Phylip version 3.6 (University of Washington, Seattle WA, USA). Phylogenetic trees were constructed using the neighbor-joining method (Neighbor program; Phylip version 3.6). The Seqboot program (Phylip version 3.6) was used to generate 1000 bootstrapped data sets. All sequences were compared with their respective type strains using the BLASTN program in the GenBank nucleotide database (http://www.ncbi. nlm.nih.gov/BLAST).

DNA fingerprint and cluster analysis of Bacillus species
The genomic diversity of native and Bacillus type strains was assayed using molecular typing of the 16S-23S intergenic transcribed spacer region (ITS-PCR), tDNA-intergenic spacer polymorphism (tDNA-PCR) analysis, and repetitive element sequence-based PCR (rep-PCR) with the primers BOXA1R, ERIC, and (GTG) 5 as described by Freitas et al. [24]. The primer sequences and amplification conditions were as previously described, except that 26GoTag Master Mix (Promega) was used.
For cluster analysis, the similarity matrix was generated based on Jaccard's coefficient and was used to build a tree with the unweighted pair group arithmetic mean method (UPGMA) available as part of the UVP Vision Works LS 6.5 software (UVP, Cambridge, UK).
Pathogenicity assay X. axonopodis pv. citri strain TPH2 was cultured in TSB supplemented with 50 mg/ml gentamicin at 27uC, 100 rpm for 2 days; B. subtilis TKS1-1, and B. amyloliquefaciens WG6-14 were cultured in TSB at 27uC, 100 rpm for 1 day. The culture suspensions were adjusted to an OD 620 of 0.3, and then sprayed on the leaves of Mexican lime in the greenhouse; the Bacillus suspensions were sprayed to the point of runoff 1 day prior to inoculation with X. axonopodis pv. citri strain TPH2. Milli-Q water was used as a control. The development of symptoms was recorded weekly for 1 month. The disease severity of citrus bacterial canker disease with and without Bacillus treatment was expressed as the number of cankers per cm 2 .

Confocal laser scanning microscopy
Bacillus and Xanthomonas strains were cultured and inoculated onto leaves under conditions similar to those used for pathogenicity assays, except that X. axonopodis pv. citri strain TPH2 harboring pGTKan was used. Cells colonized on leaf surfaces were stained with acridine orange (0.025% in 0.026 M citric acid buffer [pH 6.6]; Sigma) and then examined with an Olympus Fluoview FV1000 confocal microscope (Olympus Optical Co. Ltd., Tokyo, Japan) equipped with an argon laser. Excitation and emission wavelengths were 510 nm and 488 nm, respectively.
Efficacy of Bacillus endospore formulations on the reduction of the disease incidence of citrus bacterial canker Bacillus endospore formulations were produced using established and patented methods [53] with a 750 liter liquid fermentor. Briefly, a single colony of B. subtilis TKS1-1 was inoculated into SYB broth and incubated at 30uC, 125 rpm overnight; this culture was used as seed inoculum for a large-scale preparation of endospores using liquid fermentation. Stepwise scaled-up fermentation was conducted in SYB in a series of fermentors at 30uC for 5 days with agitation at greater than 150 rpm and aeration rate greater than 1 air volume/culture volume/min. To assess its effect on the incidence of citrus bacterial canker disease, the endospore formulation was diluted 100-fold to achieve a final concentration of 10 9 endospores/ml and applied 24 h pre-or post-inoculation with X. axonopodis pv. citri XW19. The leaves of navel orange trees were wounded with 20 pinpoint needle pricks per leaf. An X. axonopodis pv. citri XW19 suspension (OD 620 of 0.3) was diluted 10fold, then sprayed to the point of runoff on the wounded leaves in the greenhouse. Four treatments were conducted to evaluate the effect of the application time and frequency of B. subtilis strain TKS1-1 application on symptom development and disease incidence of citrus bacterial canker: (i) treatment T-B-T/1WK: the strain TKS1-1 endospore formulation was applied 24 h prior to X. axonopodis pv. citri XW19 inoculation, and then every week post-pathogen inoculation for 4 weeks; (ii) treatment T-B-D/ 1WK: the endospore formulation was applied 24 h prior to X. axonopodis pv. citri XW19 inoculation, followed by weekly spraying of Milli-Q water for 4 weeks; (iii) treatment D-B-T/1WK: the leaves were sprayed with Milli-Q water 24 h prior to strain XW19 inoculation, and then the strain TKS1-1 endospore formulation was spayed every week post-pathogen inoculation for 4 weeks; and (iv) treatment D-B-D/1WK: Milli-Q water was applied 24 h prior to XW19 inoculation and then every week post-inoculation for 4 weeks. The leaves treated with water were used as the control. Disease incidence (DI) at 4 weeks post-inoculation was calculated using the formula: DI(%) = (number of pinpoints with canker symptoms)/206100.

Statistical analysis
All experiments were performed at least three times. Data represent the means and standard deviations from at least four replicates of a representative experiment. The significant difference among treatments was analyzed by one-way ANOVA and Tukey's honestly significant difference (HSD) test using SPSS 15.0 software (SPSS Inc., Chicago, IL, USA).