Community Structures and Antifungal Activity of Root-Associated Endophytic Actinobacteria of Healthy and Diseased Soybean

The present study was conducted to examine the influence of a pathogen Sclerotinia sclerotiorum (Lib.) de Bary on the actinobacterial community associated with the soybean roots. A total of 70 endophytic actinobacteria were isolated from the surface-sterilized roots of either healthy or diseased soybeans, and they were distributed under 14 genera. Some rare genera, including Rhodococcus, Kribbella, Glycomyces, Saccharothrix, Streptosporangium and Cellulosimicrobium, were endemic to the diseased samples, and the actinobacterial community was more diverse in the diseased samples compared with that in the heathy samples. Culture-independent analysis of root-associated actinobacterial community using the high-throughput sequencing approach also showed similar results. Four Streptomyces strains that were significantly abundant in the diseased samples exhibited strong antagonistic activity with the inhibition percentage of 54.1–87.6%. A bioactivity-guided approach was then employed to isolate and determine the chemical identity of antifungal constituents derived from the four strains. One new maremycin analogue, together with eight known compounds, were detected. All compounds showed significantly antifungal activity against S. sclerotiorum with the 50% inhibition (EC50) values of 49.14–0.21 mg/L. The higher actinobacterial diversity and more antifungal strains associated with roots of diseased plants indicate a possible role of the root-associated actinobacteria in natural defense against phytopathogens. Furthermore, these results also suggest that the root of diseased plant may be a potential reservoir of actinobacteria producing new agroactive compounds.


Introduction
Sclerotinia stem rot (SSR) caused by a fungus Sclerotinia sclerotiorum (Lib.) de Bary is a highly destructive disease leading to serious economic losses to crops throughout the world. This fungus can infect over 400 plant species, including many economically important crops and vegetables [1][2][3]. Generally, the development of resistant cultivars is a long-term approach for controlling the disease [2,4]. However, the disease has yet been difficult to control because of the limited resource of the resistant genes. Therefore, fungicides have been used as the auxiliary method for controlling SSR in practice [5]. The benzimidazole and dicarboximide fungicides were the most efficient fungicides in controlling

Plant Materials
Root samples were collected from soybean (cultivar: Hefeng-50) plants identified as SSR symptomatic (diseased) or asymptomatic (healthy) based on typical symptoms in a heavily infected soybean field in Suihua, Heilongjiang province, North China (46.63 • N 126.98 • E). The diseased samples showed lesions encircling up to 1/3 of stem diameter referring to a severity scale of three [32]. Each healthy plant and diseased plant located as close neighbors were determined as one group. The samples were brought to the lab in a cooler with ice in July 2017 and were processed immediately.

Isolation of Endophytic Actinobacteria
Three groups of root samples were used for isolation of endophytic actinobacteria. The root samples were air dried for 24 h at room temperature and then washed in water with an ultrasonic step (160 W, 15 min) (KH-160TDV, Hechuang, China) to remove the surface soils and adherent epiphytes completely. After drying, the sample was cut into pieces of 5-10 mm in length and then subjected to a seven-step surface sterilization procedure: A 60-sec wash in sterile tap water containing cycloheximide (100 mg/L) and nalidixic acid (20 mg/L), followed by a wash in sterile water, a 5 min wash in 5% (v/v) NaOCl, a 10 min wash in 2.5% (w/v) Na 2 S 2 O 3 , a 5 min wash in 75% (v/v) ethanol, a wash in sterile water and a final rinse in 10% (w/v) NaHCO 3 for 10 min. After being thoroughly dried under sterile conditions, the surface-sterilized samples were subjected to continuous drying at 100 • C for 15 min. The sample was then cut up in a commercial blender and ground with a mortar and pestle, employing 1 mL of 0.5 M potassium phosphate buffer (pH 7.0) per 100 mg tissue. Tissue particles were allowed to settle down at 4 • C for 20-30 min, and an aliquot of 200 µL supernatants were spread on a series of isolation media and incubated at 28 • C for 2-3 weeks. Each isolation medium was supplemented with nalidixic acid (20 mg/L) and cycloheximide (50 mg/L) to inhibit the growth of Gram-negative bacteria and fungi. Five isolation media: Humic acid-vitamin (HV) agar [33], Gause's synthetic agar no. 1 [34], dulcitol-proline agar (DPA) [35], cellulose-proline agar [36], and amino acid agar (serine 0.05%, threonine 0.05%, alanine 0.05%, arginine 0.05%, agar powder 2%, pH 7.2-7.4) were selected for the isolation. After 14 days of aerobic incubation at 28 • C, the actinobacterial colonies were transferred onto oatmeal agar (International Streptomyces Project medium 3, ISP3) [37] and repeatedly re-cultured until pure cultures were obtained, and maintained as glycerol suspensions (20%, v/v) at −80 • C.

Phenotypic and Molecular Characterization of Actinobacterial Isolates
The purified colonies were cultivated on ISP 3 at 28 • C for two weeks, and then grouped according to their phenotypic characteristics, including the characteristics of colonies on plates, color of aerial and substrate mycelium, spore mass color, spore chain morphology, and production of diffusible pigment. Those colonies with the same characteristics were classified as one species. The number of species was counted to compare the diversity of root-associated endophytic actinobacteria from healthy and diseased soybean.
Different phenotypic isolates were further subjected to 16S rRNA gene sequence analysis for the genus and species identification. The total DNA was extracted using the lysozyme-sodium dodecyl sulfate-phenol/chloroform method [38]. The primers and procedure for PCR amplification were carried out as described by Kim et al. [39]. The PCR products were purified and ligated into the vector pMD19-T (Takara Biomedical Technology, Beijing, China) and sequenced by an Applied Biosystems DNA sequencer (model 3730XL). The almost full-length 16S rRNA gene sequences (~1500 bp) were obtained and aligned with multiple sequences obtained from the GenBank/EMBL/DDBJ databases using CLUSTAL X 1.83 software. Phylogenetic tree was constructed with neighbor-joining method [40] using Molecular Evolutionary Genetics Analysis (MEGA) software version 7.0 [41]. The bootstrap method with 1000 repetitions was using to assess the topology of the phylogenetic tree [42]. Phylogenetic distances were calculated according to the Kimura two-parameter model [43]. The 16S rRNA gene sequence similarities were determined using the EzBiocloud server [44]. The obtained gene sequences have been deposited in the GenBank database.

Screening for Antagonistic Actinobacteria
The phytopathogenic S. sclerotiorum strain used in this study was kindly provided by the Soybean Research Institute of Northeast Agricultural University (Harbin, China). Antagonistic activity of isolates were evaluated through the dual culture plate assay [45]. The isolates were point-inoculated at the margin of potato dextrose agar (PDA) [46] plates and incubated for three days at 28 • C, after which a fresh mycelial PDA agar plug of the fungus was transferred to the opposite margin of the corresponding plate. After additional days of incubation at 20 • C for seven days, inhibition of hyphal growth of the fungus was scored. The inhibition rates were calculated as follows: where Wi is the width of inhibition and W is the width between the pathogen and actinobacteria. Each test was repeated three times and the average was calculated.

Isolation and Characterization of Antifungal Compounds
The antifungal compounds were isolated using an in vitro antifungal activity-guided method [47]. The active isolates were inoculated into 250 mL flask containing 50 mL of tryptone soy broth (TSB) [48] and cultivated for two days at 28 • C with shaking at 200 rpm. Then, 12.5 mL of the seed culture was transferred into 1 L Erlenmeyer flask containing 250 mL of the fermentation medium (soluble starch 1%, dextrose 2%, tryptone 0.5%, yeast extract 0.5%, NaCl 0.4%, K 2 HPO 4 3H 2 O 0.05%, MgSO 4 7H 2 O 0.05%, CaCO 3 0.5%, pH 7. 2-7.4) and incubated at 28 • C for seven days with shaking at 200 rpm. The fermentation broth (20 L) was centrifuged (4000 rev/min, 20 min), and the supernatant and bacterial biomass were extracted with ethylacetate and methanol, respectively. Both extracts were concentrated by a rotary evaporator under reduced pressure until dry and mixed after dissolving their dried residues with methanol.
Structural determination of the active compounds were made according to spectroscopic analysis. NMR spectra were measured with a Bruker Avance III-600 spectrometer in CDCl 3 or CD 3 OD using TMS as internal standard. The ESI-MS spectrum was taken on a Waters Xevo TQ-S ultrahigh pressure liquid chromatography triple quadrupole mass spectrometer. The HR-ESI-MS spectrum was acquired with an Agilent G6230 Q-TOF mass instrument. The UV spectrum was recorded in chloroform using a Shimadzu UV-2401PC UV-VIS spectrophotometer. The IR spectrum was obtained using a Bruker Tensor 27 FTIR. Optical rotation was determined in chloroform using a JASCO P-1020 polarimeter. The ECD spectrum was measured on a Chirascan circular dichroism spectrometer (Applied Photophysics Corporation Limited, Leatherhead, UK).

Antifungal Assay of Elucidated Bioactive Compounds
The active compounds were dissolved in methanol and diluted to different concentrations, which were then added to PDA medium. A fresh fungal plug of the fungus (5 mm in diameter) was placed in the center of the agar plate and incubated at 20 • C. Experiments were performed in triplicate, and the plate with the same amount of methanol was used as control. When the control plate was covered completely with fungal mass, the percentage of inhibition was calculated with the formula as follows: where D is average diameter of the treatment and D c is average diameter of the control. Data were subjected to linear regression analysis, and the effective concentrations required for 50% inhibition (EC 50 ) were calculated.

Culture-Independent Community Analysis
Total community DNA was extracted from three groups of surface-sterilized root samples using FastDNA®SPIN for soil kit (MP Biomedicals, Solon, CA, USA) according to the manufacturers' instructions. The purity and concentration of DNA were detected using NanoPhotometer spectrophotometer (Implen, München, Germany) and Qubit 2.0 Flurometer (Life Technologies, Carlsbad, CA, USA). Bacterial DNA pyrosequencing was based on~460 bp amplicons generated by the PCR primers: 341F (5 -CCTACGGGNGGCWGCAG-3 ) and 805R (5 -GACTACHVGGGTATCTAATCC-3 ) with the barcode spanning the hypervariable regions V3-V4 of the 16S rRNA gene The PCR reaction was carried out in 30 µL reactions with 15 µL of Phusion High-Fidelity PCR Master Mix (New England Biolabs, Ipswich, MA, USA), 0.2 µM of forward and reverse primers, and about 10 ng template DNA. Thermal cycling conditions were 95 • C for 3 min, followed by 25 cycles of 95 • C for 30 s, 55 • C for 30 s, and 72 • C for 30 s, with a final extension at 72 • C for 10 min. PCR products was mixed in equidensity ratios. Then, mixture PCR products was purified with GeneJET Gel Extraction Kit (Thermo Scientific, Fermentas, Germany).
Sequencing libraries were generated using NEB Next Ultra DNA Library Prep Kit for Illumina (New England Biolabs, Ipswich, Massachusetts, USA) following the manufacturer's recommendations and index codes were added. The library quality was assessed on the Qubit@ 2.0 Fluorometer and Agilent Bioanalyzer 2100 system (Agilent Technologies, Palo Alto, CA, USA). At last, the library was sequenced on an Illumina MiSeq platform and 250 bp paired-end reads.
When the sequencing was finished, we needed to filter the raw data to secure the quality, which mainly included the following steps: (1) Cut the polluted adapter, (2) remove low quality reads, specifically reads with average quality less than 19, based on the Phred algorithm, and (3) remove the reads with N base exceeding 5%.
According to overlap of the clean data, we spliced the paired reads by using the PEAR software [49] to merged sequences. The sequences were then removed Chimeras and clustered into operational taxonomic units (OTUs) by UCLUST [50] based on 97% pairwise identity. Taxonomic classification of the representative sequence for each OTU was done using a RDP classifier or QIIME's closed reference strategy against the 16S rRNA gene database [51].

Statistical Analysis
Statistical analyses were performed via SPSS software, version 18.0. Means were compared via analysis of variance one-way ANOVA using the least significant differences test (LSD, p < 0.05). The data were reported as means ± standard deviation.

Isolation and Distribution of Endophytic Actinobacteria
A total of 1574 endophytic actinobacterial colonies were successfully isolated. Based on their phenotypic characteristics, the colonies were preliminary classified into 70 species. Among the 70 species, 15 species were from healthy soybean roots, 27 species were from diseased soybean roots, and 28 species were shared by healthy and diseased soybean roots. The diversity of actinobacteria from diseased soybean root were greater than those from healthy soybean root. 16S rRNA gene sequence analysis of the 70 isolates revealed that they were distributed under 14 genera: Streptomyces, Micromonospora, Actinomadura, Nonomuraea, Microbacterium, Rhodococcus, Promicromonospora, Microbispora, Kribbella, Mycolicibacterium, Glycomyces, Saccharothrix, Streptosporangium and Cellulosimicrobium within the class Actinobacteria. Streptomyces was the most frequently isolated genus (58%, 41 isolates), followed by Micromonospora (5 isolates) and Nonomuraea (4 isolates). Some rare genera, including Rhodococcus, Kribbella, Glycomyces, Saccharothrix, Streptosporangium and Cellulosimicrobium, were endemic to the diseased samples ( Figure 1). The 16S rRNA gene sequences were deposited in GenBank with accession numbers: MH919371-MH919374 and MN058215-MH058280.
Compound 9 was obtained as a yellow, amorphous powder. Its molecular formula C 22 H 27 N 3 O 4 S was determined by high resolution electrospray ionization mass spectrometry (HRESIMS) data (m/z 468.1555, [M + Na] + , calculated for 468.1564), corresponding to 11 degrees of unsaturation. The IR spectrum indicated the presence of hydroxy (3424 cm −1 ) and carbonyl (1720, 1682 cm −1 ) groups. The 1 H NMR data ( Table 2)   Compound 9 was obtained as a yellow, amorphous powder. Its molecular formula C22H27N3O4S was determined by high resolution electrospray ionization mass spectrometry (HRESIMS) data (m/z 468.1555, [M + Na] + , calculated for 468.1564), corresponding to 11 degrees of unsaturation. The IR spectrum indicated the presence of hydroxy (3424 cm −1 ) and carbonyl (1720, 1682 cm −1 ) groups. The 1 H NMR data ( Table 2)   Comparison the NMR data of 9 with FR900452 [58], an indole diketopiperazine motif linked with a cyclopentenone moiety, which was isolated from the fermentation broth of Streptomyces sp. B9173, implied that 9 was identified as a reduced form of FR900452 in which the cyclopentenone moiety is hydrogenated to cyclopentanone. As shown in Figure 4, the accurate assignments of all protons and carbons for compound 9 were preformed through the correlations in 2D-NMR spectra ( 1 H-1 H COSY, HSQC and HMBC, Figure S2  configuration of was identified as 3S, 10R, 11R, 14R, resulting from the same trends of cotton effects (CEs) in the experimental ECD spectra of 9 and N-demethylmaremycin B. The in vitro antifungal activity of compounds 1-9 against S. sclerotiorum was determined at various concentrations. All compounds showed significantly antifungal activity against S. sclerotiorum with the EC50 values ranging from 49.14 mg/L to 0.21 mg/L (Table 3). Thus, it further confirmed that these compounds were the main antifungal constituents produced by the four active strains. Values are the means ± SE (n = 9). Data within the same column followed by different letters are significantly different.

Culture-Independent Communities
A total of 4116 OTUs containing 745708 high-quality reads were detected in the soybean root microbiome. The raw sequencing reads for this project were submitted to the National Center for The relative configuration of compound 9 was determined by interpretation of its ROESY NMR spectrum. The correlations of H-14/Me-24, and H-10/H-11, indicated that H-14 and Me-24 were α-oriented, whereas H-10 and H-11 were β-oriented (Figure 4). Based on the close skeleton, the comparison of ECD spectra between 9 and N-demethylmaremycin B [57], and the largely consistent data supported that the configurations of 3-OH was α-oriented. Ultimately, the absolute configuration of was identified as 3S, 10R, 11R, 14R, resulting from the same trends of cotton effects (CEs) in the experimental ECD spectra of 9 and N-demethylmaremycin B.
The in vitro antifungal activity of compounds 1-9 against S. sclerotiorum was determined at various concentrations. All compounds showed significantly antifungal activity against S. sclerotiorum with the EC 50 values ranging from 49.14 mg/L to 0.21 mg/L (Table 3). Thus, it further confirmed that these compounds were the main antifungal constituents produced by the four active strains.

Culture-Independent Communities
A total of 4116 OTUs containing 745708 high-quality reads were detected in the soybean root microbiome. The raw sequencing reads for this project were submitted to the National Center for Biotechnology Information Short Read Archive under accession numbers SRR8056376-SRR8056381. The predominant bacterial phyla were Proteobacteria, Bacteroidetes and Actinobacteria in the soybean roots. To compare the microbial communities obtained in healthy and diseased root samples from each group, the relative abundance of order Rhizobiales that can improve rhizobial nodulation and nitrogen fixation was significantly greater in the healthy samples, whereas the order Actinobacteria were more abundant in the diseased samples ( Figure 5).

Discussion
The multifaceted approach adopted in this study, linking culture-independent and culture-dependent analysis, showed that actinobacteria were more abundant or diverse in the diseased soybean roots. This finding was in agreement with the previous study that the phylum Actinobacteria was higher in 'Candidatus Liberibacter asiaticus'-infected citrus samples compared with that in healthy samples [31]. Another similar study also showed that potato plants infected with Erwinia carotovora subsp. atroseptica increased bacterial diversity [59]. The higher diversity of endophytic actinobacteria in diseased but healthy plants suggests that they may be involved in pathogen defense [60]. Indeed, extensive research has shown that endophytic actinobacteria has the capacity to control plant pathogens [22]. The in vitro antagonism assays demonstrated that four

Discussion
The multifaceted approach adopted in this study, linking culture-independent and culture-dependent analysis, showed that actinobacteria were more abundant or diverse in the diseased soybean roots. This finding was in agreement with the previous study that the phylum Actinobacteria was higher in 'Candidatus Liberibacter asiaticus'-infected citrus samples compared with that in healthy samples [31]. Another similar study also showed that potato plants infected with Erwinia carotovora subsp. atroseptica increased bacterial diversity [59]. The higher diversity of endophytic actinobacteria in diseased but healthy plants suggests that they may be involved in pathogen defense [60]. Indeed, extensive research has shown that endophytic actinobacteria has the capacity to control plant pathogens [22]. The in vitro antagonism assays demonstrated that four strains showed strong antifungal activity against S. sclerotiorum. Among the four antagonistic strains, all colonies of three strains were absolutely from diseased soybean roots, and another strain was also significantly enriched in diseased soybean roots compared to healthy soybean roots. A similar study has also shown that the rhizosphere soil of diseased tomato plant harbored a high percentage of antagonists [61]. Studies over the past few years have provided important information that plants possess a sophisticated defense mechanism by actively recruiting root-associated microbes from soil upon pathogen attack [18,62]. By adjusting the quantity and composition of its root secretion, plants can determine the composition of the root microbiome by affecting microbial diversity, density, and activity [63,64]. Our results seem to be consistent with previous observations. However, those strains with antagonistic activity in vitro may not be simply translated into biocontrol bacteria. Their biocontrol effects are influenced by various factors. For example, the antagonistic strains should reach a certain amount inside the plants to demonstrate a significant biocontrol effect [65,66]. Moreover, their secondary metabolite producing ability inside the plants may be influenced by plant physiological environment. Therefore, further work is required to assess the biocontrol efficiency in vivo and root-colonizing capacity of antagonistic strains by pathogen infection.
To learn more about the chemical nature of the antifungal activity, nine active compounds including six macrolides, two diketopiperazines and one 2-oxonanonoids, were finally obtained. Out of which, bafilomycin B 1 (3) showed strongest inhibitory activity against S. sclerotiorum. Bafilomycin B 1 has been reported to be produced by several Streptomyces strains and to show inhibitory activity against various fungi in vitro, such as Rhizoctonia solani, Aspergillus fumigatus, Botrytis cinerea, Penicillium roqueforti, and so on [67,68]. The antifungal activity of this compound against S. sclerotiorum was first reported in this paper. Azalomycins F 4a (1) and F 5a (2) were first isolated from the broth of Streptomyces hygroscopicus var. azalomyceticus [69]. Azalomycins F complex, including azalomycins F 3a , F 4a and F 5a showed remarkable antifungal activity against asparagus (Asparagus officinalis) pathogens Fusarium moliniforme and Fusarium oxysporumas as well as powdery mildew pathogen Botrytis spp. [70]. The antifungal activity of the pure compound was first demonstrated in our research. Azalomycins possess broad-spectrum antibacterial and antifungal activities, and almost all of them were produced by Streptomyces, which were isolated repeatedly from soil and plant roots in the field by our laboratory (data no shown). This emphasizes the possible importance of Streptomyces producing azalomycins to protect plants against phytopathogens. A mixture of dinactin (7), trinactin and the major component tetranactin (6) is known as commercial pesticides polynactin (liuyangmycin in China), which can effectively control spider mites under wet conditions [71]. In addition, tetranactin (6) also exhibited significant antifungal activity against plant pathogen Botrytis cinerea with a minimum inhibitory concentration (MIC) of 24 µg·mL -1 [72]. Besides dinactin (7) and tetranactin (6), the monomer (4) and dimer (5) of polynactin were also isolated in this study, all of which were active against S. sclerotiorum. The antifungal activities of actinolactomycin (4), dimeric dinactin (5) and dinactin (7) have not been reported as yet. The findings reported here shed new light on the application of polynactin. Natural indole diketopiperazines exhibited a wide range of biological activities including antitumor [73], antibacterial [74], antifungal [75] and antiviral activities [76]. FR900452 is sulfur-containing indole diketopiperazines that showed specific and potent inhibitory activity against the platelet aggregation induced by platelet-activating factor [77]. Maremycin G (8) and compound 9, structurally related to FR900452, showed significant antifungal activity against S. sclerotiorum. To our knowledge, this is the first report of the antifungal property of maremycins. Further research is needed to confirm the efficacy of in vivo disease control provided by the nine active compounds under laboratory and field conditions.

Conclusions
In summary, we report that soybean infected by S. sclerotiorum (Lib.) de Bary had a higher populations of actinobacteria and enhanced root colonization of antagonistic populations. In addition, eight known compounds and one new compound that exerted significant antifungal activity against S. sclerotiorum were obtained. These findings suggest that diseased plant samples could be a potential source for screening novel agroactive compounds, which contribute to a better understanding of plant-microbe interactions and provide new strategies for the development of agricultural antibiotics.

Conflicts of Interest:
The authors declare that they have no conflicts of interest.