Fusarium oxysporum KB-3 as a mycorrhizal fungus of Bletilla striata (Jiang et al., 2019) was used for the detection of EHB. It was inoculated on potato dextrose agar (PDA) and cultured at 28°C for 5 days to collect the mycelia. The genomic DNA was extracted by CTAB method (Stenglein and Balatti, 2006). To detect endohyphal bacterial DNA in the fungal culture, 16S ribosomal DNA (16S rDNA) region was used for PCR amplification with primer pair of ER10 (5′-GGCGGACGGGTGAGTAA-3′) and ER11 (5′-ACTGCTGCCTCCCGTAG-3′) (Widjojoatmodjo et al., 1994). The PCR consisted of 2 μl gDNA, 10 μm forward primer, 10 μm reverse primer, 2 × PCR Master Mix (Takara, Japan). It was programmed on GeneAmpR PCR System 2700 (Applied Biosystems, USA) under the following condition: 94°C for 2 min, 35 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 15 s, extension at 72°C for 30 s, and final extension at 72°C for 10 min. To further determine the existence of bacteria in the mycelium, fungal hyphae stained with SYTO-9 green fluorescent nucleic acid solution (Obasa et al., 2017) were observed under a fluorescence Nikon ECLIPSE Ni-U microscope system (Nikon, Japan). The hyphae were collected from the edge of 3-day-old PDA colony, stained with 500 μl SYTO-9 solution (10 μL SYTO-9 with 490 μl Mcilvaine buffer) for 2 min, and rinsed three times with 1 ml Mcilvaine buffer for observation.

To isolate EHB from Fusarium oxysporum KB-3, the method was referred to the study of Obasa et al. (2017). The fungus was inoculated in potato dextrose broth (PDB) shaking incubated with 200 rpm at 28°C for 72 h. The hyphae were collected, disinfected with 70% ethanol for 2 min, and washed with sterile distilled water for 5 times. The fifth washing liquid was spread on Luria-Bertani (LB) agar medium grown at 30°C for 48 h to examine possible bacterial contamination. Meanwhile, the remaining fifth solution was centrifuged at 12,000 g for 10 min to concentrate the possible bacteria for the PCR amplification of 16S rDNA (primers: ER10 and ER11) region as previously mentioned. The surface sterilized hyphae were transferred to a sterile mortar supplemented with silica sand and distilled sterile water and grounded thoroughly with a sterile stick. The resulted solution was grown on LB agar at 28°C to obtain bacterial colonies. The upcoming colonies were further purified and deposited in glycerol solution at −80°C.

The genomic DNA of obtained bacterium was extracted according the procedure of E.Z.N.A.^® Bacterial DNA Kit (Omega BIO-TEK, USA). A number of three gene regions of 16S rDNA (Galkiewicz and Kellogg, 2008), DNA gyrase subunit B (gyrB) (Deletoile et al., 2009), and RNA polymerase subunit B (rpoB) (Deletoile et al., 2009) were amplified to identify the EHB using primer pairs of 27F/1492R, gyrB3/gyrB4, and Vic3/Vic2, respectively. The PCR amplification was conducted in a 40 μl mixture using the regents as previously reported under the referenced conditions (Galkiewicz and Kellogg, 2008; Deletoile et al., 2009). Successful products were purified and sequenced by BGI (Wuhan, China) using two specific primers for each gene region. The resulting sequences were determined by BLAST searches in GenBank and also deposited with accession numbers. The three gene sequences of the present bacterium and its relevant species were edited manually and concatenated in MEGA v.7.0.26 (Kumar et al., 2016). A phylogenetic analysis was conducted by Bayesian inference (BI) using MrBayes v.3.2.1 (Ronquist et al., 2012) based on a Markov Chain Monte Carlo (MCMC) algorithm. The best-fit nucleotide substitution model was specified by jModelTest 2.1.7 (Darriba et al., 2012) under default settings followed by Akaike information criterion (AIC). Trees were sampled every 100 generations from 5,000,000 generations resulting in total 50,000 trees. The first 12,500 trees (first 25% samples by default in the software) were discarded representing the burn-in phase and the remaining 37,500 trees were for calculating posterior probability (PP) values in the majority rule consensus tree.

The wild strain of Fusarium oxysporum KB-3 was inoculated in PDB medium containing 50 μg/ml tetracycline, 70 μg/ml gentamicin, 0.1 M CaCl₂, and 1% DMSO, incubated at 28°C for 72 h. Every three times sub-cultured fungus was determined by PCR amplification of 16S rDNA region (primers: ER10 and ER11) to detect the EHB. If the 16S rDNA amplicon was still present, antibiotic treatment would be continued until it disappeared. In the meantime, the resulted bacteria-free hyphae were stained with SYTO-9 solution and examined under the fluorescence microscope as previously mentioned to examine the bacterial existence. The successful cured strain (KB-3⁻) was confirmed by both methods and deposited in glycerol solution preserved at −80°C for future use.

To visualize Klebsiella aerogenes KE-1 in the strain KB-3, the bacterium was labeled with green fluorescent protein (GFP) and restored to the cured strain KB-3⁻. First, the bacterium was transformed by electroporation with plasmid pLac-EGFP-Chl-signal-Hyg. In the meantime, the protoplasts of cured strain KB-3⁻ were prepared as previously described by Lakshman et al. (2012). Subsequently, the GFP-labeled bacteria and protoplasts were separately rinsed with precooled shock buffer (0.6 mol/L mannitol, 0.1 g/L CaCl₂) for three times, and then, both were mixed and placed in ice bath for 10 min. The mixture was converted by electroporation at 200 Ω, 25 μF in 2-mm cuvettes (voltage range up to 5 kV cm⁻¹). The mixture (100 μl) was inoculated in 0.5 ml 1/4 PDB medium incubated for 12 h at 28°C. Afterward, the mixture was spread on 1/4 PDA and then covered a layer of 1/4 PDA medium to form a double-layer plate. Once hyphae grew through the upper medium, the hyphal tip was transferred to new medium. It was repeated for several times to ensure no bacteria attached the hyphae. The hyphae were observed under the fluorescence microscope as previously mentioned. The successfully restored strains (KB-3⁺) were deposited and also determined by PCR amplification of 16S rDNA using primers of ER10 and ER11.

To detect IAA production from the cultural broth of Klebsiella aerogenes KE-1, wild strain KB-3, cured strain KB-3⁻ and restored strain KB-3⁺, it was examined using high-performance liquid chromatography (HPLC) compared with IAA standard. A number of three mycelium blocks (9 mm in diam.) for fungal strain grown on PDA for 5 days were transferred into 200 ml LB liquid comprising 10 mg/ml L-tryptophan, cultured at 200 rpm and 28°C. Meanwhile, 100 μl LB cultural solution of the bacterium KE-1 (OD₆₀₀ = 0.6) was inoculated as previous treatment for each fungal strain. After incubated for 60 h, each filtrate (100 ml) collected by filtration with lens wiping paper and centrifugation was extracted twice with 100 ml ethyl acetate (EtOAc). The EtOAc layers were condensed to dry by rotary evaporation vacuum (45°C). The residues were dissolved in methanol to 10 ml and filtered by a 0.22-μm Millipore filter (Merck Millipore, Germany) for HPLC analysis. IAA (Sigma, USA) was dissolved in methanol as a standard sample. Besides, ten-time-diluted KE-1 extraction was used for HPLC analysis. Venusil XBP-C18 column (4.6 × 250 mm, 5 μm) was used to analyze the EtOAc extract in Wooking HPLC system (Wooking K2025, China). The mobile phase was acetonitrile and 2% acetic acid aqueous solution at the flow rate of 1 ml/min, and the injection volume was 10 μl. The results were analyzed by Wookinglab software v00.02.20.00 (Wooking, China).

To evaluate the IAA production of Klebsiella aerogenes KE-1, wild strain KB-3, cured strain KB-3⁻, and restored strain KB-3⁺, it was conducted using Salkowski colorimetric technique (Ehmann, 1977; Glickmann and Dessaux, 1995; Huddedar et al., 2002; Chaiharn and Lumyong, 2011). After every 12 h of incubation using previous culturing method for IAA examination (3 replications), 2 ml culture broth (n = 3) was taken and centrifuged, followed by 1 ml supernatant treated with 2 ml of Salkowski reagent (1 m, 0.5 mm FeCl₃, and 50 ml, 35% HClO₄) (Chaiharn and Lumyong, 2011). The samples were incubated at 25°C for 30 min in the dark and evaluated optical density (OD) values at 530 nm by spectrophotometer (BIO-RAD, USA). LB medium containing 10 mg/ml L-tryptophan was used as control. The experiment was repeated for two times. IAA concentration was calculated using IAA standard curve (y = 0.0238; x = 0.0395) determined by different concentrations IAA (Sigma, USA) solutions.

The LB containing 10 mg/ml L-tryptophan without NaCl cultural supernatants of wild strain KB-3, cured strain KB-3⁻, and restored strain KB-3⁺ was prepared as previous treatment for IAA examination. After incubated for 60 h, each supernatant was modified to pH 7.0 ± 0.1 with 0.5 M NaOH if needed and filtered two times by the 0.22-μm Millipore filter (Merck Millipore, Germany) to obtain the 20-ml filtrate. The filtrates were used for seedling growth promotion assay. Tomato seeds were surface-disinfected with 75% ethanol for 2 min and 4% sodium hypochlorite for 5 min and then rinsed with distilled sterile water for 5 times. The surface sterilized seeds were transferred to 90 mm Petri dishes with moistened sterile filter paper to accelerate seed germination. Healthy similar sprouted seeds (n = 10) were randomly transferred to crisper containing filter papers saturated with distilled water (5 ml). Per seedling was covered with 50 μl each of previous LB cultural supernatants and LB liquid containing 10 mg/ml L-tryptophan without NaCl (control) (Hoffman et al., 2013). Each treatment had three replicates (i.e., 30 seedlings/treatment). The crispers were tilted placing in an illumination incubator at 25°C for 5 days (12-h light/dark period). Stem and root lengths of tomato seedlings were measured and the experiment was repeated for two times.

Data obtained from the IAA production and plant growth-promoting assay were subjected to the analysis of variance (ANOVA) with SPSS 17.0 software (SPSS Inc., Chicago, IL, USA). The results were presented as average means and standard deviation. Duncan's multiple range test was used to compare the mean value at the 5% (p < 0.05) level of significance.

The PCR amplification of 16S rDNA for either the wild strain KB-3 and restored strain KB-3⁺ resulted a band of 255-bp fragment; however, the cured strain KB-3⁻ and the final rinse solution of wild strain for endohyphal bacterial isolation remained empty (Figure 1). The resulted showed that the cured strain KB-3⁻ was successfully obtained by continuously cultured in modified PDB. Similarly, hyphae of the wild strain KB-3 (Figure 1B) contained the fungal nucleus and also bacterial DNA under fluorescence microscopy, while the cured strain KB-3⁻ (Figure 1C) only comprised fungal nucleus.

One EHB (KE-1) was obtained from surface disinfected and grounded hyphae of Fusarium oxysporum KB-3 on LB agar. The fifth washing liquid obtained from hyphal surface sterilization was no bacterial colonies observed on LB agar plates, which came out that the hyphae were completely disinfected. To identify the bacterium KE-1, the 16S rDNA (1,440 bp), gyrB (506 bp), and rpoB (512 bp) gene sequences were amplified and deposited in GenBank database with accession numbers of MZ144165, MZ150560, and MZ150561, respectively. BLAST searches showed that each of the three gene sequences shared 99–100% identity with that of Klebsiella aerogenes (Syn. Enterobacter aerogenes) strains, for example, the type strain KCTC 2190 (Taxonomy ID: 1028307). Bayesian tree (Figure 2) based on the combined 16S rDNA, gyrB and rpoB gene sequence dataset was constructed with Klebsiella spp.. The results clearly showed that the EHB KE-1 was K. aerogenes and it was clustered together with the type strain KCTC 2190 and K. aerogenes AR_0161 (NCBI genome accessions CP028951) supported with high PP values (1.0), from which the present bacterium formed a sub-clade with strain KCTC 2190 (PP = 1.0).

The cured strain KB-3⁻ was gained after continuously culturing wild Fusarium oxysporum KB-3 in modified PDB broth (Figures 1A,C). Meanwhile, the GFP-labeled Klebsiella aerogenes KE-1 was obtained by the electroporation and it was also successfully restored in the hyphae of the cured strain KB-3⁻ (KB-3⁺) (Figure 1A) observed by fluorescence microscopy (Figure 3). The GFP-labeled K. aerogenes KE-1 bacteria could be precisely exhibited in the hyphae of the restored strain KB-3⁺.

In HPLC analysis, IAA production was found in the cultural broth of Klebsiella aerogenes KE-1, wild strain KB-3, cured strain KB-3⁻, and restored strain KB-3⁺ (Figure 4). It was evaluated every 12 h by the Salkowski colorimetric technique within 84 h of inoculation. The IAA production of all the tested samples was gradually increased and then decreased after the highest (Figure 5). However, the EHB could secrete the highest amount of IAA (76.9 μg/ml) after 48 h inoculation (Figure 5A) and the three fungal strains secreted the highest IAA production at 60 h post-inoculation. At the highest IAA production, the wild stain KB-3 produced the highest amount of IAA (av. 31.4 μg/ml), followed by the restored strain KB-3⁺ (av. 19.4 μg/ml), and then the cured strain KB-3⁻ (av. 9.6 μg/ml) (Figure 5B). The IAA production was significantly different (p < 0.05) among the three fungal strains. In addition, it was noteworthy that the IAA production of restored strain KB-3⁺ did not reach to the same level of wild strain KB-3 (around three-fifths of the wild one), but significantly higher than the cured strain KB-3⁻ produced.

The wild Fusarium oxysporum strain KB-3, its cured strain KB-3⁻, and its restored strain KB-3⁺ exhibited plant growth-promoting effects on tomato using their fermentation broth compared with the controls (Figure 6). The results were consistent with their IAA production abilities. The root length and stem length of tomato seedlings treated with the wild strain KB-3 (around two times of those generated from controls) were the longest, followed by the restored strain KB-3⁺ and then the cured strain KB-3⁻. Similarly, the restored strain KB-3⁺ induced shorter length of both tomato root and stem than those induced by the wild strain but significantly higher than the cured strain KB-3⁻. The root length and stem length of tomato seedlings under the four treatments were significantly different from each other (p < 0.05). Besides, lateral root development of tomato seedlings was observed only in the treatment of the wild strain KB-3 (Figure 6).