The plasmids and bacterial strains used in this study are listed in Table 1. Strain MS82 was isolated from the rhizosphere of a soybean plant grown at the R. R.

Km^r, kanamycin resistance; Tc^r, tetracycline resistance; Ap^r, ampicillin resistance; Cm^r, chloramphenicol resistance.

Foil Plant Science Research Farm, Mississippi State University. Assessment of antifungal activity of P. fluorescens MS82 was conducted using plates of potato dextrose agar (PDA) (Difco, BD Biosciences, Detroit, MI). Strain MS82 was grown on Pseudomonas Agar F (MP Biomedicals, Aurora, OH) for detection of fluorescence. Strain MS82 grew at 28˚C on nutrient broth-yeast extract medium (NBY) [21] with kanamycin (50 μg・mL⁻¹) and tetracycline (25 μg・mL⁻¹).

Bacterial genomic DNA was extracted using the Wizard® Genomic DNA Purification Kit (Promega Corporation. Madison, WI). Multilocus sequence typing (MLST) analysis was conducted as described by Jolley [22] . The seven housekeeping genes (glnS, gyrB, ileS, nuoD, recA, rpoB and rpoD) were amplified from the DNA of strain MS82 using the primers listed in the website http://pubmlst.org/pfluorescens (Table 2). The cycling program included a 2

min initial denaturation step at 95˚C, followed by 30 cycles of 95˚C for 1 min, 56˚C for 30 s, 72˚C for 1 min, and final elongation at 72˚C for 10 min. Sequencing reads were assembled using the Lasergene version software package (DNAStar, Inc. Madison, WI). A phylogenetic tree based on the concatenated sequences (3,453 bp) of seven housekeeping gene fragments [glnS (501 bp), gyrB (492 bp), ileS (552 bp), nuoD (516 bp), recA (435 bp), rpoB (477 bp) and rpoD (480 bp)] of MS82 was produced using MEGA6 software.

The antifungal activities of Pseudomonas fluorescens MS82 and its mutants in this study were analyzed using PDA plate bioassays as described previously [23] . Briefly, strain MS82 was grown overnight in NBY liquid medium, after which bacterial cells were collected by centrifugation and washed once with sterile distilled water. The cells were suspended in sterile distilled water to an optical density of 0.3 at 420 nm wavelength (approximately 2 × 10⁸ CFU/mL), a drop of bacterial suspension (5 μL) was spotted on the center of each PDA plate, and plates were incubated 3 d at 28˚C. Each plate was then oversprayed with a standardized (OD₄₂₀ = 0.3) conidial suspension of one of the three fungal indicator species: M. perniciosa JS01 (collection of Jindi Song), Trichoderma viride TX12 (Collection of Shien Lu) and G. candidum F-260 (a gift of Dennis Gross). The oversprayed plates were further incubated in the dark at 28˚C for 2 d. To evaluate the antifungal activity of MS82 against the mushroom fungus A. bisporus (Strain MS46, isolated from fresh mushroom), two PDA discs (5 mm in diameter) containing the mycelium of A. bisporus were placed 4 cm apart onto each of 9-cm PDA plates. A 5-μL aliquot of bacterial cell suspension (OD₄₂₀ = 0.3) was immediately spotted onto the same plate equidistant from the edge of each fungal disc. The inoculated plates were incubated in the dark for three weeks at 25˚C. The inhibitory zones of MS82 and its mutants against the indicator fungi were recorded and compared with each other. Three replicates of the plate bioassays were performed independently.

In order to find the genes associated with antifungal activity, EZ-Tn5 Tnp Transposome Kit (Epicentre Biotechnologies, Madison, WI) was used as described by the manufacturer to randomly mutate strain MS82. Strain MS82 was grown in LB broth with shaking (220 rpm) at 28˚C until early log phase (OD ₆₀₀ = 0.8 - 1.0). The cells were harvested by centrifugation at 5000 rpm for 6 min at 4˚C, washed twice with the same volume and once with 0.5 volume of 300 mM sucrose, re-centrifuged, and finally re-suspended in 0.01 volumes of 300 mM sucrose [24] . Before electroporation, the bacterial suspension was chilled on ice for 30 min. Aliquots of 80-μL cell suspension were mixed with one μL of EZ-Tn5 < R6Kγori/KAN-2 > Tnp Transposome and transferred to chilled 0.1 cm gap cuvettes. Electroporation was conducted using a Gene Pulser II (Bio-Rad Laboratories, Hercules, CA) with the settings of 25 μFD, 200 Ω and 1.8 kV. Immediately after electroporation, the cells were transferred into 3 ml LB broth and shaken (220 rpm) at 28˚C for 2h. The colonies growing on NBY plates supplemented with 50 μg・mL ⁻¹ kanamycin were picked up as potential mutants for antifungal plate bioassays. The mutant MS82MT19 that had no visible antifungal activity against M. perniciosa was cultured on NBY plates with 50 μg・mL ⁻¹ kanamycin for further study. The Tn5 transposon fragment was amplified by PCR with the primers EzTn5F (5’-TTA CGA AAC ACG GAA ACC-3’) and EzTn5R (5’-TCT AAT ACC TGG AAT GCT GTT-3’), provided by the EZ-Tn5 Tnp Transposome Kit manufacturer, to confirm the presence of the transposon in the mutant.

Bacterial genomic DNA of mutant MS82MT19 was extracted as described above. The bacterial genomic DNA was digested by restriction enzymes Pst I and Xba I, self-ligated, and then transformed into electrocompetent E. coli strain TransforMax EC100D pir⁺ (Epicentre Biotechnologies, Madison, WI) as instructed by the kit to generate plasmids pMT19E1 and pMT19E2 (Table 1). Sequencing reactions were performed using the primers KAN-2FP-1 and KAN-2RP-1 supplied in the transposon kit to the flanking genomic DNA fragments of the Tn5 transposon in mutant MS82MT19.

The entire gene (i.e., the PafR gene) targeted by transposon in mutant MS82MT19 was sequenced by the Sanger primer walking method [25] . In brief, five pairs of primers were designed and used to sequence the DNA flanking the Tn5 transposon in the plasmid pMT19E2. A consensus sequence of the entire gene region was generated using the Lasergene package. The gene functions were predicted through BLAST analysis against the GenBank database. The web-based software BPROM in the Softberry package was used for promoter prediction [26] .

Efforts to complement the mutant MS82MT19 with a full-length targeted gene in trans were made through introduction of the plasmid pUCP26 harboring the full-length gene [18] . However, the antifungal activity of the mutant could not be restored to the wild type level at all. Therefore, we used insertion of the nptII gene into the PafR gene to generate a nonpolar site-directed mutation. A 1kb fragment containing the SmaI site of the PafR gene of the wild-type strain MS82, for which PCR primers containing restriction sites of AatII and PstI were designed (FP-AatII-1: 5’-GAC AGA CGT CTA CCG AGG TCA ACC AGG TTG-3’; RP-PstI-1: 5’-CGA ACT GCA GCG AGG TCG AGT GCT CGA AAA C-3’), was cloned into the pGEM-T Easy vector to produce the plasmid pMS1. The 1 kb fragment was digested by AatII and PstI and then subcloned into pBR325 [19] , generating plasmid pMS2 containing tetracycline resistance. A 1.1-kb SmaI fragment of plasmid pBSL15 [20] , which carries the nptII gene without a transcriptional terminator, was inserted into pMS2 at the SmaI site to generate plasmid pMS3 containing kanamycin and tetracycline resistance. Plasmid pMS3 was electroporated into strain MS82 for marker exchange mutagenesis as previously described [27] , resulting in the generation of the mutant MS82SD19 which has resistance to kanamycin but sensitive to tetracycline. PCR was used to verify the double-crossover marker exchange in the mutant MS82SD19. The antifungal activity of the mutant MS82SD19 against M. perniciosa was evaluated with plate bioassays as described above.

Seven housekeeping gene fragments (glnS, gyrB, ileS, nuoD, recA, rpoB and rpoD) were obtained by PCR amplification. These DNA sequences were uploaded into GenBank with the accession numbers: KU510319 (glnS), KU382113 (gyrB), KU510314 (ileS), KU510315 (nuoD), KU510316 (recA), KU510317 (rpoB) and KU510318 (rpoD). The concatenated sequence (3453 bp) of the seven gene fragments [glnS (501 bp), gyrB (492 bp), ileS (552 bp), nuoD (516 bp), recA (435 bp), rpoB (477 bp) and rpoD (480 bp)] was generated in silico. A phylogenetic tree based on the concatenated sequences of MS82 was constructed using MEGA6 (Figure 1). Strain MS82 was tightly clustered with the P. fluorescens strain Pf0-1 with the bootstrap value of 100%. These data suggest that strain MS82 belongs to the bacterial species P. fluorescens.

Strain MS82 suppressed the mycelial growth of the fungal pathogen M. perniciosa with an inhibition zone of 34.17 ± 0.44 mm in the plate bioassay (Figure 2(B)). Strain MS82 also inhibited the growth of T. viride, and G. candidum (Figure 3). Interestingly, strain MS82 did not affect the mycelial growth of the mushroom fungus A. bisporus (Figure 2(A)). The antifungal activities of strain Pf0-1, the closest strain to MS82 based on phylogenetic analyses, were tested using the same methods as MS82. The results showed that strain Pf0-1 did not inhibit the growth of M. perniciosa, T. viride, or G. candidum (Figure 3). These results suggest that strain MS82 has an antifungal spectrum different from P. fluorescens strain Pf-01 and that its ability to inhibit a variety of fungi may be exploited for biocontrol and development of antibiotics for use in mushroom production.

Strain MS82MT19 was obtained from random mutagenesis that completely lost the antifungal activity and was chosen for further genetic analyses (Figure 2(C)). The 16S rRNA sequencing confirmed that t MS82MT19 was a mutant of strain MS82. Two plasmids, pMT19E1 and pMT19E2 (Table 1) carrying the re-

gions flanking the EZ-Tn5 transposon insertion, were produced from MS82MT19 genomic DNA using the plasmid rescue method. The plasmids were digested by Pst I and Xba I, demonstrating that they carried different lengths of genomic DNA insertions that include the Tn5 transposon. Preliminary BLAST analyses of the flanking regions of the transposon in the two plasmids revealed that the Tn5 transposon carried by the mutant MS82MT19 was inserted into a gene that has a high identity (~98%) at nucleotide level to the putative gene Pfl01_4876 that was predicted to code for a sensory box GGDEF/EAL domain protein of P. fluorescens Pf0-1 (GenBank Accession No. CP000094.2).

Sequence analysis showed that the putative proteins encoded by the gene disrupted by the transposon in MS82MT19 shares high identities (99-100%) with sensory box GGDEF/EAL domain proteins of Pseudomonas. A 4.5kb flanking DNA region of the transposon, which was inserted in the genome of the mutant MS82MT19, was sequenced from the plasmids pMTE1 and pMTE2 (Table 1) by using primer walking and Sanger DNA sequencing. Further analysis revealed that only one open-reading frame of 3849 nt encoding the sensory box GGDEF/ EAL domain protein is disrupted by the transposon in MS82MT19. This gene was named as the PafR gene (Pseudomonas antifungal regulatory gene) and its sequence was deposited into GenBank with the accession number KU510320. The promoter region (−35 box: ATGCTA and −10 box: GCTTATAGT) was identified 38 nucleotides upstream of the putative start codon of the predicted protein PafR. A typical sequence-dependent terminator was not found from the untranslational 3’ end of the gene. The BLAST results revealed that the putative protein of this gene shares 99% amino acid identity to the sensory box GGDEF/ EAL domain protein of P. fluorescens Pf0-1 (GenBank accession: ABA76613). In addition, the results showed that the transposon in the MS82MT19 genome was inserted between the nucleotides 1393 and 1394 of the sensory box GGDEF/ EAL domain gene.

The role of sensory box GGDEF/EAL domain PafR gene targeted in MS82MT19 in production of antifungal activity was further determined using the site-directed mutagenesis. After 16S RNA DNA sequencing confirmed the identity of the mutant MS82SD19, the insertion of the terminitorless nptII cassette in the PafR gene was determined by PCR analysis with the primers FP-AatII-1 and RP-PstI-1. As expected, the PCR products were 1.1 kb in size from genomic DNA of the wild type strain MS82 and 2.2 kb containing the nptII cassette from both MS82SD19 genomic DNA and plasmid pMS3 (Figure 4), respectively. The

plate bioassay results revealed that the mutant MS82SD19 containing nptII gene did not produce the antifungal activity against M. perniciosa (Figure 2(D)), which is the same as the phenotype of the mutant MS82MT19. Collectively, the destruction of the PafR gene of MS82 strain either by random mutagenesis or by site-directed insertional mutagenesis resulted in the loss of antifungal activity against M. perniciosa. These results demonstrate that the PafR gene is the key gene in the antifungal activity of MS82 production.