Potential of Pantoea dispersa as an effective biocontrol agent for black rot in sweet potato

Biocontrol offers a promising alternative to synthetic fungicides for the control of a variety of pre- and post-harvest diseases of crops. Black rot, which is caused by the pathogenic fungus Ceratocytis fimbriata, is the most destructive post-harvest disease of sweet potato, but little is currently known about potential biocontrol agents for this fungus. Here, we isolated several microorganisms from the tuberous roots and shoots of field-grown sweet potato plants, and analyzed their ribosomal RNA gene sequences. The microorganisms belonging to the genus Pantoea made up a major portion of the microbes residing within the sweet potato plants, and fluorescence microscopy showed these microbes colonized the intercellular spaces of the vascular tissue in the sweet potato stems. Four P. dispersa strains strongly inhibited C. fimbriata mycelium growth and spore germination, and altered the morphology of the fungal hyphae. The detection of dead C. fimbriata cells using Evans blue staining suggested that these P. dispersa strains have fungicidal rather than fungistatic activity. Furthermore, P. dispersa strains significantly inhibited C. fimbriata growth on the leaves and tuberous roots of a susceptible sweet potato cultivar (“Yulmi”). These findings suggest that P. dispersa strains could inhibit black rot in sweet potato plants, highlighting their potential as biocontrol agents.


Molecular identification and phylogenetic analysis of the microbial communities in sweet potato plants.
We collected a total of 75 species of microorganisms from the tuberous roots (RO) and shoots (SH) of field-grown sweet potato plants and the bulk soil (SO). These isolates could be identified to the genus and species level with a sequence similarity of 97-100% based on 16S or 5.8S rRNA gene sequences in the NCBI database.

Colonization of Pantoea in sweet potato plant tissues.
To visualize the location of Pantoea in sweet potato plants by live-cell imaging, we inoculated the roots of the sweet potato cultivar "Yulmi" with a suspension of GFP-labeled P. dispersa . Individual colonies of GFP-expressing RO-21 were observed in the roots 18 h after inoculation ( Fig. 2A,B; Supplementary Fig. S3). After 48 h, bacterial aggregates were observed in horizontal sections of stems ( Fig. 2C-F), and cells were effectively observed residing in the intercellular spaces of the outer cortex and there was extensive colonization in the cellular pits of the xylem tracheids in the stems (Fig. 2C-F). After 7 d, GFP-expressing macro-colonies of RO-21 was found in the leaf petioles ( Supplementary Fig. S4), and it was able to recover by selective medium (Supplementary Fig. S5). This result suggests that P. dispersa effectively colonizes sweet potato tissues as an endophytic bacterium.

Evaluation of in vitro antagonism.
To determine the potential of endophytic Pantoea to control C. fimbriata, Pantoea isolates were screened for their inhibitory ability using an in vitro dual culture assay on PDA media. Mycelial plugs of C. fimbriata (5-mm diameter) were placed at the center of PDA plates and bacterial colonies were streaked around the borders of the plates ( Fig. 3A; Supplementary Fig. S6). The antagonistic activity of the Pantoea strains was then estimated by comparing the fungal growth with C. fimbriata alone. Since P. agglomerans was previously identified from sweet potato 31 , we introduced P. agglomerans KCTC2564 (=ATCC27155) as additional Pantoea strain for this test. Fungal growth was monitored by measuring the diameter of the mycelium until day 16 at 28 °C ( Fig. 3A; Supplementary Fig. S6A and S6B). This approach allowed the relative growth of C. fimbriata mycelia to be quantitatively assessed (Fig. 3B).
Pantoea strains RO-18, RO-20, RO-21, and SO-13, which were identified as P. dispersa, had the highest inhibition rates, reaching >63% at 16 d after co-incubation (Fig. 3B). Among these, RO-21 and SO-13 showed the best inhibition activity, decreasing mycelial growth by up to 72% and significantly reducing the diameter of the fungal disks from 7.3 cm with the control treatment to 2.4-2.6 cm ( Supplementary Fig. S6B). All four of these strains exhibited a clear inhibition halo (Fig. 3A). Some of the other bacterial strains also showed weak inhibition activity on C. fimbriata growth (19-34%), including Escherichia coli DH5 α, P. agglomerans KCTC2564, and P. www.nature.com/scientificreports www.nature.com/scientificreports/ ananatis strains SH-5, SH-9, SH-1, RO-22, SH-3, and RO-1 (Fig. 3A,B; Supplementary Fig. S6A); however, there was no statistically significant reduction in the diameter of the fungal disks with these treatments and the negative control (P ≥ 0.05) (Fig. 3B). These results indicate that P. dispersa strains RO-18, RO-20, RO-21, and SO-13 are good candidates as potential biocontrol agents of C. fimbriata.  www.nature.com/scientificreports www.nature.com/scientificreports/ Effect of cell-free culture supernatant of Pantoea on spore germination and hyphal morphology. To determine whether the extracellular compounds that are produced by P. dispersa have antifungal activity, we observed the spore germination rates and morphologies of C. fimbriata following co-culture of the spores (10 5 CFU mL −1 ) with 900 μL of cell-free culture supernatant of Pantoea strains SH-9, RO-18, RO-20, RO-21, and SO-13. Light microscopy was then used to evaluate the effect of extracellular metabolites in the culture filtrates on the spore germination rates. Incubation with the cell-free culture supernatant of SH-9 did not significantly affect www.nature.com/scientificreports www.nature.com/scientificreports/ the C. fimbriata germination rates compared with the control after 20 h [germination rates = Control (84.0%) and SH-9 (65.1%), respectively]. However, the germination rates of the spores were inhibited by incubation with the cell-free supernatant of RO-18 (35.1%), RO-20 (36.0%), RO-21 (33.1%), and SO-13 (33.4%) after 20 h (Fig. 4A,B). Staining with FITC-WGA further showed that extracellular metabolites in the culture filtrates of P. dispersa strains RO-21 and SO-13 caused cellular changes in the hyphal morphology of C. fimbriata, including hyphal swelling, distortion, and cytoplasmic aggregation, while incubation with the cell-free culture supernatant of P. ananatis strain SH-9 did not cause any changes in hyphal morphology (Fig. 4C). Together, these findings suggest that the extracellular metabolites of P. dispersa have antifungal activity against C. fimbriata.

In vitro interaction between Pantoea isolates and C. fimbriata.
To better understand the mode of action of the antifungal activity of P. dispersa strains, C. fimbriata was grown alongside Pantoea strains and the cell viability was visualized by staining with Evans blue, which stains dead cells blue, and Neutral red, which stains viable cells red. P. dispersa RO-21 and SO-13 appeared to cause the dramatic breakage of C. fimbriata hyphae in the contact zone, with the mycelia that were exposed to P. dispersa staining blue, while those in the control zone on the other side only exhibited faint blue staining (Fig. 5A,B). Similar results were obtained from RO-20 and RO-18 (Supplementary Fig. S7A and S7B). By contrast, in the presence of P. ananatis SH-9 and SH-3, faint blue staining and red staining were observed in both the contact and control zones ( Fig. 5C; Supplementary Fig. 7C). These findings indicate that P. dispersa has fungicidal activity rather than fungistatic activity against C. fimbriata. www.nature.com/scientificreports www.nature.com/scientificreports/

Inhibition of C. fimbriata infection in sweet potato leaves and tuberous roots.
To assess the biocontrol efficiency of P. dispersa in the leaves and tuberous roots of the sweet potato cultivar "Yulmi", which is susceptible to C. fimbrata infection 23 , samples were pre-treated with RO-21 or water (as a control) and then inoculated at the same site with droplets containing C. fimbriata spores. The disease incidence on the leaves and  www.nature.com/scientificreports www.nature.com/scientificreports/ tuberous roots was then determined. Leaves that had been pre-treated with water showed disease symptoms at the inoculation site, with dark brown circular patches of 4-6 mm diameter appearing on the upper surface of the leaves and the surrounding areas turning yellow after 36 h. By contrast, the disease incidence was reduced in leaves that had been pre-treated with RO-21 (Fig. 6A). Disease symptoms were also dramatically reduced in tuberous roots that had been pre-treated with RO-21 cells for 1 d before C. fimbriata inoculation compared with www.nature.com/scientificreports www.nature.com/scientificreports/ those that had been pre-treated with water as a control (Fig. 6B,C). Furthermore, pre-treatment with RO-21 for only 10 min also reduced the size of the black lesions around the inoculation site compared with the water pre-treatment (Fig. 6C). Together, these findings indicate that P. dispersa could prevent the infection of sweet potato leaves and tuberous roots by the pathogenic fungus C. fimbriata.

Discussion
The composition of any given plant-associated microbial community is likely to be determined by multiple factors, including the soil type, phytopathogen population 12 , plant age 35 , and plant genotype, as well as stochastic sampling factors 36 . In the present study, we identified 75 culturable microbes that were associated with field-grown sweet potato plants based on their colony morphologies, colors, and patterns of growth. These included members of the genera Bacillus, Pantoea, Enterobacter, Serratia, and Microbacterium (Supplementary Tables S2 and S3), indicating that these plants are associated with diverse bacterial communities. However, only a few species dominated these communities, with the genera Bacillus and Pantoea predominating in the shoots and tuberous roots (Supplementary Tables S2 and S3).
Bacillus species are considered attractive biocontrol agents due to their ability to produce hard, resistant endospores and antibiotics that control a broad range of plant fungal pathogens 37 . However, it has previously been found that B. subtilis does not effectively inhibit the growth of C. fimbriata 38 , which is consistent with the findings for the Bacillus strains isolated in the present study (data not shown). The bacterial genus Pantoea comprises many versatile species that have been isolated from a multitude of environments, such as mammals, agricultural areas, water, and particularly soil 39 . El Amraoui 40 found that species showed stronger antifungal activity against Candida albicans CIP 48.72, Candida albicans CIP 884.65, and Cryptococcus neoformans CIP 960 than Bacillus sp., while Town 41 showed that Pantoea sp. inhibits the potato pathogen Phytophthora infestans. In the present study, Pantoea strains RO-18, RO-20, RO-21, and SO-13, which were isolated from sweet potato, had a high homology with the type strain P. dispersa LGM 2603 T and exhibited remarkable antifungal activity against C. fimbriata both in vitro and in vivo, with the in vitro dual culture assay showing that they inhibited C. fimbriata mycelium growth (Fig. 3A) and spore germination (Fig. 4A,B), appeared to cause dramatic breakage of the fungal hyphae (Fig. 4C), and killed the fungal hyphae, acting as a fungicidal agent (Fig. 5A,B). Therefore, genus Pantoea has a greater potential to control C. fimbriata than genus Bacillus. www.nature.com/scientificreports www.nature.com/scientificreports/ Root-inoculated GFP-labeled RO-21 cells were initially visualized in the root surface and central vascular system of primary root in sweet potato plants within 18 h, following which a proliferation of cells was observed in the intercellular spaces between adjacent xylem tracheid cells in the caulosphere at 48 h after inoculation. The rapid spread of this strain from the root to the aerial tissues suggests that it uses the vascular system as a route for systemic colonization. Other endophytic bacteria are also located in the vascular system, with large fluorescent colonies having been clearly observed in the pits of the xylem cells in the plant cell wall 42,43 .
The mode of action of P. dispersa is similar to that of B. subtilis against Eutypa lata in grapes, whereby irregularities occur in the hyphal morphology, such as tip narrowing and vesicle formation 44 . B. subtilis has been shown to cause abnormalities in the hyphae of Fusarium oxysporum, causing cell wall lysis, breakage, granulation, and vacuolization 45 . Similar phenomenon was observed by P. dispersa isolates (Figs 4 and 5). Moreover, the leaves and tuberous roots of sweet potato plants had a significantly decreased incidence of disease symptoms when cultured in the presence of RO-21 (Fig. 6A,B), indicating that P. dispersa can inhibit the development of black rot disease in sweet potato. Moreover, strains RO-18, RO-20, RO-21, and SO-13 were initially isolated from the tuberous roots of agricultural sweet potato and soil, which can be explained by the fact that both P. dispersa and C. fimbriata are soil-borne microorganisms that primarily compete for the tuberous roots 5 .
Although P. ananatis strains SH-1, SH-3, SH-5, SH-9, SH-10, SH-13, RO-1, and RO-22 did not effectively inhibit the growth of C. fimbriata in this study, it has previously been shown that P. ananatis CPA-3 has strong antifungal activity against Penicillium expansum 46 . Similarly, P. ananatis 125NP12 have been found to protect tomato fruit against the grey mold fungus, Botrytis cinerea, by producing antifungal compounds 47 . In this respect, P. ananatis strains SH-1, SH-3, SH-5, SH-9, SH-10, SH-13, RO-1, and RO-22 might control other plant pathogens in sweet potato plants. In addition, the potential of P. ananatis SCB4789F-1 to promote plant growth has been demonstrated by its ability to solubilize phosphorus and zinc, produce siderophores, and synthesize IAA. P. ananatis B1-9 strain isolated from the rhizosphere of onions showed potential for promoting plant growth in peppers, cucumbers, and melons 48 . Therefore, further detailed studies are needed on the plant growth-promoting effect of P. ananatis isolates.
Several mechanisms have been reported for the biocontrol of plant pathogens, including competition for nutrients 49 , the induction of host resistance 50,51 , and the production of killer toxins 52 , degradable enzymes such as chitinase 6 , and antifungal metabolites [53][54][55] . Many Pantoea strains are strong environmental competitors that produce a variety of natural products with antibiotic activity, such as pantocins, microcins, and phenazines [56][57][58][59][60][61][62][63][64] . More recently, Pantoea Natural Product 1 (PNP-1), which has been isolated from P. ananatis BRT175, has been shown to have inhibitory activity against Erwinia amylovora and is likely to be similar in action to 4-formylaminooxyvinylglycine (FVG) 65,66 . In the present study, cell-free supernatant derived from P. dispersa cultures was found to have inhibitory effects on C. fimbriata spore germination and hyphae growth when added simultaneously or after some time, indicating that the antifungal activity of these strains may occur via similar mechanisms. Hence, further investigation is required to evaluate the potential bioactive compounds that may be of biocontrol importance in the metabolites of these isolates.
Together, our results indicate that P. dispersa strains represent useful biocontrol agents for protecting sweet potato plants from post-harvest infection by C. fimbriata and for decreasing the use of agricultural chemical.

Methods
Isolation of microorganisms. Tuberous roots and shoots were collected from healthy sweet potato (I. batatas) plants growing in a cultivation area in Jeongeup, Jeollabuk-do, Republic of Korea (35°30′29.8″N, 126° 50′13.9″E). To acquire plant extracts, the plant materials were extensively washed with sterile distilled water, and aliquot (100 μL) of the final rinse water was plated onto Laurie-Bertani (LB) and Potato Dextrose agar (PDA) media (Difco) to check the disinfection process. The plant tissues were then crushed with 50 mL of sterile phosphate-buffered saline (PBS: 100 mM Tris-HCl, pH 8.0; 150 mM NaCl) and filtered through four layers of sterile cheesecloth. The bacterial cells in the filtrate were serially diluted with PBS buffer, and 100 μL of each diluted sample was spread onto an LB and PDA media plates, and then incubated at 28 °C for 5-14 d.
The colony types in each sample were categorized according to their growth rate and phenotypic characteristics, such as sizes, colors, and colony morphology. Each type was then counted and a representative was taken for purification and identification. The following voucher specimens were deposited at the Korean Collection for Species identification and phylogenetic analysis using 16S and 5.8S rRNA gene sequences.
The 16S, 5.8S rRNA and housekeeping genes (gyrB, and rpoB) 67 of each species were amplified using the primers listed in Supplementary Table S1. 16S and 5.8S rRNA sequencing was carried out at BioFact Inc. (Daejeon, Korea), and sequence alignment was performed with the BLAST search (https://blast.ncbi.nlm.nih.gov/Blast. cgi). Phylogenetic trees were constructed using the neighbor-joining (NJ), minimum-evolution (ME), and maximum-likelihood (ML) method based on 16S, gyrB, and rpoB genes in MEGA (ver.7) 68 , and a bootstrap analysis of 1,000 replications was performed to evaluate the stability of the nodes.
Monitoring bacterial colonization on sweet potato plants. Sweet potato plants that had been separated from their tuberous roots were placed in a continuous-floating hydroponic system for 5 d until their roots had reached a length of 4-5 cm. The roots of the plants were then inoculated with a suspension of GFP-expressing P. dispersa RO-21 [10 8 colony-forming units (CFU) mL −1 ]. After 15 min, the plant roots were washed three times with distilled water and then kept in the floating hydroponic system. GFP-expressing bacteria were detected in the root 18 h after inoculation. At 4 d and 7 d after inoculation, the stems of each sweet potato plants were cut into sections that were as thin as possible using a razor blade from 5 cm above the roots. Stem cross-sections and selected roots from each plant were then placed on a glass slide with distilled water, and the GFP-expressing bacteria were observed using an epifluorescence microscope with a GFP filter (Moticam Pro 205 A; Motic) at 20× or 40× magnification. To confirm the GFP signals were not due to the auto-fluorescence of the plant tissue itself, re-isolation of the Pantoea strain RO-21 were performed. Leaf petioles from the strain RO-21 inoculated plants for 7 d were collected, then surface-sterilized by using 1.05% sodium hypochlorite for 10 min and several repeats rinse in sterile distilled water. After homogenized with PBS buffer, samples were spread on LB agar plates supplementary with ampicillin (100 μg mL −1 ).

Dual culture assay.
To test the activity of the bacterial strains against C. fimbriata, mycelial plugs of C. fimbriata (5-mm diameter) were placed in the center of PDA plates (pH 5.7) and the various bacterial strains were streaked in a square shape around each agar disk at a distance of 3 cm from the mycelial plug. In addition, mycelial plugs were placed on PDA plates without any bacterial strains as a negative control. Fungal growth was assessed by measuring the diameter (in centimeters) of the colony until 16 d at 25 °C. Each bacterial strain was tested in triplicate and the experiment was carried out twice. The relative inhibition (RI) was then calculated using the formula RI (%) = [(radial growth in control − radial growth in samples)/radial growth in control] × 100, as previously described 69 . Data were analyzed by one-way analysis of variance (ANOVA) using GraphPad Prism 4. Significant differences (P ≤ 0.05) between the means were determined by unpaired t-test and Tukey's multiple comparison test.

Microscopic analysis.
A fungal spore germination assay was performed using cell-free supernatant of the Pantoea strains. Bacterial cells were harvested from 6-d-old liquid cultures of SH-9, RO-18, RO-20, RO-21, and SO-13 by centrifugation for 20 min at 10,000×g. The resulting supernatant was then filtered through a 0.2 μm filter (Whatman TM ) to remove the cells. C. fimbriata spores were collected from a 2-week-old culture of fungi growing on PDA media and placed in 10 mL of sterile distilled water, following which they were passed through 25 μm sterile Miracloth TM (Millipore) to remove the hyphae and subsequently diluted to 10 5 spore mL −1 for the bioassay. Then, 10 4 spores were co-incubated with 900 μL of each of the cell-free supernatant of SH-9, RO-18, RO-20, RO-21, and SO-13 for 13, 15, 17, and 20 h in triplicate. Spore germination (%) was evaluated in each sample by placing the spores on a microscope slide and counting at least 300 spores per sample. Once the fungal spores had been left to germinate in the cell-free supernatant of the Pantoea strains for 36 h, the hyphae were stained with fluorescein isothiocyanate-labeled wheat germ agglutinin (FITC-WGA, 1 μg mL −1 ) for 30 min to investigate whether the cell-free supernatant of P. dispersa altered the fungal cell wall.
To evaluate the viability of this pathogenic fungus in the presence of Pantoea strains, C. fimbriata was co-cultivated alongside RO-18, RO-20, RO-21, SH-3 or SH-9 on PDA media at 25 °C for 10 d. The C. fimbriata mycelia that were adjacent to the bacteria, were then stained with the vitality stains Neutral red (0.1 mg mL −1 ; Dae Jung, Cat # 5603-4125) or Evans blue (0.5 mg mL −1 ; Alfa Aesar, Cat # A16774) by placing 10 μL of the solution on a slide, incubating the cells for 3-5 min at room temperature, and then washing 3-4 times with deionized water. Mycelia growing on the opposite sides from the bacteria were treated as a negative control. There were 3-4 replicates per treatment and photographs were taken under a light microscope (Nikon Eclipse Ci).

Disease incidence and symptoms.
To evaluate the biocontrol effect of P. dispersa on C. fimbriata, the leaves (5-6 weeks old) and tuberous roots of plants of the sweet potato cultivar "Yulmi", which is susceptible to C. fimbrata, were analyzed. The leaf surfaces were punched with a needle on the upper side, and then the punched sites were inoculated with 10 μL of P. dispersa cell suspension (10 7 CFU mL −1 ) pre-treatment for 1 d. The C. fimbriata spores after incubation for 7 d at 25 °C were scraped from well-grown PDA plates and filtered through 25 μm Miracloth in order to obtain spore suspensions. 5 μL of C. fimbriata suspension (10 6 CFU mL −1 ) were inoculated at the same site as used for pre-treatment. The inoculated leaves were covered with plastic wrap for 36 h, the yellow to black area of the punched leaves were recorded. All experiments were performed in triplicate, repeated at least twice. Surface of the tuberous roots was punched with a needle, then same treatment as described with leaves, unless inoculated with 10 μL of P. dispersa cell suspension (10 7 CFU mL −1 ) pre-treatment for 1 d and 10 min, then inoculated with 5 μL of C. fimbriata suspension at the same site as was used for pre-treatment, under moist conditions in the plastic box at 25 °C for 10 d. The disease incidence was recorded by observing the formation of the black spots of C. fimbriata and softening of the tissue in the collar region. All measurements were made in triplicate and repeated twice with similar results.