General DNA manipulations, including gene cloning and plasmid isolation were performed using standard protocols (Sambrook et al., 1989). All restriction enzymes and modified enzymes were purchased from FastDigest, Thermo Scientic Korea Ltd (Seoul, Korea). The genome of P. chlororaphis O6 (Loper et al., 2012) contains both the fit gene cluster harboring fitD and the HCN biosynthesis operon, hcnABC. Specific primer sets were designed to amplify target genes. The fitD forward primer bearing a EcoR1 recognition site, 5′-CGGAATTCGGTTCTTGTCGGCAAACCAC-3′ and the reverse primer harboring a HindIII site, 5′-CCCAAGCTTCCAATCACCTGGTGTCGGAA-3′ encoded the amino acid sequences NHLVSE and WFADKNR within the FitD protein. The 4.3 kb PCR product contained a partial sequence of the fitD gene was cloned into the EcoRI- and HindIII-digested pEX18Tc marker exchange vector (Hoang et al., 1998).

To disrupt HCN production a forward primer was tagged with a NsiI site, 5′-CCAATGCATGCCTCTATGCCCTTCTGACC-3′ and the reverse primer was tagged with a SacI site, 5′-TACGAGCTCCAAGGCGTTTGCGGAGTATG-3′. These primer sequences were designed to target sequences within the gene encoding a hypothetical protein located upstream and the gene for glutathione S-transferase located downstream of the hcnABC operon (Fig. 1). The 3.6 kb PCR product containing the hcn operon was digested with NsiI and SacI and cloned into the SacI- and PstI-digested pEX18Tc vector.

The pEX18Tc vectors containing the PCR products of the partial fitD and hcn operon genes were used for in vitro mutagenesis using EZ::Tn5 <KAN2> transposon (Epicentre Biotechnologies, WI, USA). This system randomly inserts a kanamycin resistance gene in the target DNA. The potential fitD or hcn mutants cloned into the pEX18Tc plasmid possessing tetracycline resistance were selected on LB agar plates containing tetracycline and kanamycin. The recombinant plasmids were prepared and used to determine the flanking sequence of the EZ-Tn5 transposon using the KAN-2 FP-1 forward primer and RP-1 reverse primers provided in the EZ::Tn5 <KAN2> transposon kit from Epicentre.

Two pEX18Tc recombinant clones, containing the EZ::Tn5 inserts within the structural genes of fitD and hcnA (Fig. 1), were selected and transferred into P. chlororaphis O6 through tri-parental mating. The marker-exchange mutant of each gene was selected on LB agar medium (Becton Dickinson GmbH, Heidelberg, Germany) supplemented with 5% sucrose and kanamycin in LB agar, as described previously (Miller et al., 1997). To confirm the mutation of both genes, PCR products were amplified from both mutants using gene specific primers, and the PCR products were sequenced to determine flanking sequences of EZ-Tn5 insertions in the mutants.

To complement the hcn mutant, the 3.6-kb PCR product containing the full-length hcnABC operon was cloned into pCRII vector, digested with NsiI, and transferred into the PstI-digested pCPP54 vector, which has a broad host range (Park et al., 2011). We could not complement the fitD mutant because this gene was embedded in the very large (about 12 kb) fit operon.

The wild-type strain of P. chlororaphis O6, mutant strains defective in gacS, hcnA, and fitD, and the complemented gacS and hcnA mutants (Kang et al., 2007; Spencer et al., 2003) were used in the studies. Cells were grown to stationary phase for 36 h at 22°C with shaking at 200 rpm in liquid King’s medium B (KB) (King et al., 1954), which was amended with glycine (4.5 g/l) to enhance HCN production. Glycine is the precursor of HCN (Zdor, 2015). Escherichia coli DH5α was cultivated for 48 h at 37°C with shaking at 250 rpm in the same media to provide a control bacterium not producing HCN or FitD.

P. chlororaphis O6 strains and E. coli DH5α, as a negative control, were grown in KB or KB plus glycine for 48 h at 22°C with agitation at 200 rpm in a shaking incubator. The cultures were centrifuged at 10,000 g to pellet the cells and the supernatants were filtered through 0.2 μm filters, and assayed to measure HCN (Guibault and Kramer, 1966). Dilutions (0.05 to 100 μm) of a KCN stock solution were used as the calibration standard.

To detect HCN production in the rhizosphere, tomato seeds (TENTEN, Koregon, Anseong, Korea) were surface sterilized with 70% ethanol for 5 min, treated with 1% sodium hypochlorite for 1 min and rinsed three times in sterile distilled water. Seeds were transferred to a sterile nursery soil mixture of Bio-Santo and Vermiculite (7:3, vol/vol, Seminis Korea, Seoul, Korea) contained in Magenta-boxes (7.2 × 7.2 × 10 cm, Sigma-Aldrich, St. Louis, MO, USA) and incubated for 16 h in light and 8 h in dark at 26°C. Two weeks after seeding, the roots were drenched with 10 ml of bacterial cells suspended in sterile water (1 × 10⁸ cfu/ml) or with KB broth medium as a negative control. Three days after root drenching, HCN production was determined qualitatively using Cyantesmo paper (Machery-Nagel GmbH & Co., Duren, Germany).

One bioassay for nematicidal activity was performed with the second juvenile stage (J2) of the root-knot nematode, M. hapla. The nematode juveniles were isolated from naturally infested soil at depths of 0-30 cm from the Fruit Experimental Station, in Haenam, Jeonnam, Korea (Lee et al., 2011). P. chlororaphis O6 bacterial strains were grown for 36 h in KB or KB plus glycine broth and adjusted to 1 × 10⁸ cfu/ml (OD_600nm = 1.0) with sterile water, before preparing 10-fold dilutions in sterile water. The diluted cultures were mixed 1:1 v/v with a suspension of J2-stage juveniles (100 nematodes). After the mixtures were incubated for 1 h at room temperature, J2 mortality was evaluated under a low-power microscope (Carl Zeiss Discovery V12, Gottingen, Germany) by touching the nematodes with a sharp tip. Nematodes that did not respond with movement were considered dead. Assessment of each treatment involved three replicates examining responses with approximately 100 nematodes and three independent studies were conducted. Data represent the average of five replicates and the assay was repeated three times.

A second in planta nematicidal evaluation involved examining the effects on root-knot symptom formation on tomato following the methods modified by those of Lee et al (2011). Tomato seeds were surface-sterilized by soaking in 70% ethanol for 30 s and then 1% sodium hypochlorite for 10 min. After extensive washing with water, each tomato seed (Betatiny, PPS Seed, Yongin, Korea) was planted into sterile nursery soil mixture of Bio-Santo and Vermiculite housed in sterile Magenta boxes. Seedlings were established in closed Magenta boxes with lids. The lids were removed after development of the second true leaves of tomato plants, and 20 ml of sterile water was added as a drench every 2 days. The boxes were incubated in chambers under a regime of 16 h light (2,000 lux, 80 μmol photons m⁻² s⁻¹) and 8 h darkness for 5 weeks.

The preventive and curative potential of the diluted bacterial cultures on the root knot symptoms were assessed. Cultures were grown on KB plus glycine as described above, and adjusted to 1 × 10⁸ cfu/ml with sterile water. Each treatment box was drenched with 10 ml of the 10-fold diluted culture of the P. chlororaphis O6 strains. The nematicide, Fosthiazate 30% SL (FarmHannong, Seoul, Korea), was added to boxes as a positive treatment at the manufacturer’s recommended dose (250 μl/l). Negative controls involved adding 10 ml of 10-fold diluted noninoculated KB plus glycine broth to the boxes. To assess the effects on nematode infection, the bacterial cultures were applied 1 week before nematode inoculation. To determine assess any curative effects, the bacterial cultures were added 1 week after nematode infection. Approximately 200 M. hapla J2 juveniles in 10 ml of sterile water were applied to each box. As a negative control, the boxes were drenched with the same volume of sterile water. The boxes were returned to the growth chamber under the same conditions as described above. After 2 weeks, the roots of plants were collected, and the numbers of root-knot galls were counted, and root fresh weights were recorded. The study was conducted with three replicates per treatment with three tomato plants per treatment.

Data were analyzed by ANOVA using SPSS 21.0 K for Windows software (SPSS institute, NC, USA). The significance of the effect of bacterial treatment was evaluated by Duncan’s multiple range test (P < 0.05). The significance of the effect of glycine amendment in the KB medium of P. chlororaphis O6 strains was evaluated by Student’s T-comparison (P < 0.05).

Sequence analysis of PCR products obtained using DNA from the mutants confirmed the insertion of the EZ-transposon into the bacterial chromosome. The EZ-transposon disrupted the hcnA gene between the 16th amino acid, alanine, and the 17th amino acid, aspartic acid, relative to the start methionine. The selected fitD mutant contained an insertion between the 834th base pair in a codon encoding arginine, and the 835th base pair in a codon encoding tyrosine (Fig. 1). The altered genomes of the mutants were confirmed by PCR using the primer sets for each gene; an increase of about 1 kb in product size was observed as anticipated due to insertion of the kanamycin-resistance gene (Data not shown). The fitD mutant was additionally confirmed by transcriptional analysis between the wild-type and the fitD mutant (Supplementary Fig. 1). The fitD transcript was induced in the late-log phase and stationary phase in P. chlororaphis O6 wild-type, but no fitD transcripts were detected in the fitD mutant.

Growth of the strains on KB broth with and without glycine amendments confirmed HCN production in the culture fluid at similar levels for the wild-type strain, the fitD mutant, and the complemented hcn and gacS mutants (Fig. 2). These strains all presented increased HCN production when the medium was supplemented with the HCN-precursor, glycine (Fig. 2). However, no HCN production was observed following mutation of the hcnA gene in the HCN synthase operon and following mutation of the gacS gene. As anticipated, the control E. coli strain produced no HCN.

Juvenile death was observed following application of intact cultures of the wild-type strain at 1 × 10⁸, and to a lesser degree at 1 × 10⁷ cfu/ml (Fig. 3). Higher mortality was observed when the wild-type strain was grown on KB plus glycine versus KB, supporting the involvement of HCN production in juvenile death. Similar levels of juvenile mortality were observed when the cultures were treated with the complemented hcnA mutant and the fitD mutant compared with the wild-type. Cultures with the hcnA mutant exhibited approximately 50% mortality, similar to the level (53 to 60% mortality) observed with the gacS mutant, which does not produce HCN. No nematicidal activity of E. coli DH5α and KB medium was observed during the initial inoculation. However, after 1 h of treatment with E. coli DH5α cultures presented 20-30% loss in viability; these values are similar to that observed with non-inoculated KB medium.

In this study, all bacterial cultures were grown on KB plus glycine broth to optimize HCN formation. No galls were observed on tomato roots grown in the absence of M. hapla infection (Fig. 4). The plant infective J2 juveniles were used as the inoculum and galls were assessed in 14 day-old plants. Under the preventative regime, when nematodes and bacterial preparations were applied at the same time, and application of the nematicide, Fosthiazate, eliminated gall formation. Gall numbers also decreased from about 100 galls/plant in the nematode control study to about 40 galls/plant with treatments of the wild-type cultures or the preparations from the complemented hcn mutant and the fitD mutant. There was no significant difference in the gall numbers between the no-treatment control and treatment of the hcn mutant bacterial preparations.

To examine a curative effect, nematode inoculation was performed 1 week prior to the addition of bacterial preparations or the commercial nematicide. The gall numbers were lower with the 7 day-delayed applications compared with the preventive regime for the control (Fig. 4). No curative effects were observed for the treatments with the wild-type strain, the hcn mutant, and Fosthiazate. However, there was a trend for lower gall numbers in treatments with the complemented hcn mutant and the fitD mutant.

To establish whether the loss of protection against gall formation by the hcn mutant was due to a lack of root colonization, colonization of this mutant was compared with that of the wild-type strain. The same number of cells was recovered from the tomato roots in the wild-type and mutant when assayed at 3, 5 and 7 days after inoculation (Supplementary Fig. 2). The use of Cyantesmo paper in the air space of the plant growth boxes confirmed HCN production, as indicated by blue coloration of the indicator paper, only when the tomato roots were colonized by the wild-type cells. As anticipated, no coloration was detected when roots were colonized by the hcnA or gacS mutants (Supplementary Fig. 3).