Biosynthesis of the antimicrobial cyclic lipopeptides nunamycin and nunapeptin by Pseudomonas fluorescens strain In5 is regulated by the LuxR‐type transcriptional regulator NunF

Abstract Nunamycin and nunapeptin are two antimicrobial cyclic lipopeptides (CLPs) produced by Pseudomonas fluorescens In5 and synthesized by nonribosomal synthetases (NRPS) located on two gene clusters designated the nun–nup regulon. Organization of the regulon is similar to clusters found in other CLP‐producing pseudomonads except for the border regions where putative LuxR‐type regulators are located. This study focuses on understanding the regulatory role of the LuxR‐type‐encoding gene nunF in CLP production of P. fluorescens In5. Functional analysis of nunF coupled with liquid chromatography–high‐resolution mass spectrometry (LC‐HRMS) showed that CLP biosynthesis is regulated by nunF. Quantitative real‐time PCR analysis indicated that transcription of the NRPS genes catalyzing CLP production is strongly reduced when nunF is mutated indicating that nunF is part of the nun–nup regulon. Swarming and biofilm formation was reduced in a nunF knockout mutant suggesting that these CLPs may also play a role in these phenomena as observed in other pseudomonads. Fusion of the nunF promoter region to mCherry showed that nunF is strongly upregulated in response to carbon sources indicating the presence of a fungus suggesting that environmental elicitors may also influence nunF expression which upon activation regulates nunamycin and nunapeptin production required for the growth inhibition of phytopathogens.


| INTRODUCTION
Pseudomonas fluorescens strain In5 originally isolated from an agricultural suppressive soil in southern Greenland is a promising biocontrol agent capable of suppressing Rhizoctonia solani infection of tomato seedlings and inhibiting the growth of diverse phytopathogens (Michelsen, Watrous, Glaring, Kersten, Koyama, et al., 2015).
In order to begin unraveling regulation within the nun-nup gene clusters of P. fluorescens In5, the LuxR-type regulator-encoding gene nunF located downstream of the nunamycin biosynthesis genes was selected for characterization. Using a combination of insertional mutagenesis, gene expression and secondary metabolite profiling, the role of nunF in the production of the CLPs nunamycin and nunapeptin in P.
fluorescens In5 was investigated. In addition, we investigated whether nunF expression is induced in response to either specific fungalassociated carbon sources or root-associated carbon sources present in the rhizosphere that can potentially indicate to the bacterium that it is in close vicinity to a fungal hyphae or a plant root. If the bacterium is specialized in using nunF-regulated genes to induce hyphal leakages, it is hypothesized that the fungal-associated carbon sources should result in a higher upregulation of nunF.

| Strains, plasmids, and growth conditions
The strains and plasmids used in this work are listed in Table 1. All enzymes used in this study were from New England Biolabs (NEB) supplied by BioNordika, Herlev, Denmark. Escherichia coli DH5α ™ (NEB, BioNordika) was used as the host strain for cloning procedures and broad-host range plasmids were transferred to P. fluorescens strain In5 by electroporation as described previously by Michelsen, Watrous, Glaring, Kersten, Koyama, et al. (2015).

| Insertional mutagenesis of nunF by homologous recombination
Gene knockout by homologous recombination of nunF gene was carried out using the gene replacement vector, pEX100T (1).

| Complementation of nunF knockout strain
Plasmid pHN1270 harboring the apramycin selectable marker was used as a complementation vector. The nunF gene was amplified from strain In5 genomic DNA by PCR using Phusion High Fidelity Polymerase (Fisher Scientific) and forward (5′-GGAATTAACCATGC AGTGGTGGTGGTGGTGGTGCTCGAGAGGAGGACCGACCATGAAT CG-3′) and reverse (5′-AATCTGTATCAGGCTGAAAATCTTCTCTCAT CCGCCAAAACTAGTTTACGCCCCGATCATCCATTTG-3′) primers to yield a 931-bp fragment which was fused by Gibson Assembly ® (NEB) into pHN1270 linearized by NcoI and SpeI and transformed into E.
coli DH5α™ (NEB). Fusion of the amplicon and plasmid was confirmed by restriction digest of plasmid DNA followed by Sanger sequencing (GATC-Biotech, Konstanz, Germany) to confirm integrity of the DNA sequence. The resultant construct was then transformed into strain In5 by electroporation described earlier. Complementation was tested using the antifungal activity assay described below with the following strains: P. fluorescens In5 WT, ΔnunF with the control empty vector pHN1270 with or without IPTG (2 mmol/L) induction, and ΔnunF with the complementation plasmid pHN1270::nunF with or without IPTG (2 mmol/L) induction. Complementation was performed with biological triplicates and repeated twice.

| Phenotypic analysis
Antifungal activity was assayed as described previously by Michelsen, Watrous, Glaring, Kersten, Koyama, et al. (2015). Briefly, a plug of R. solani Ag3 was placed in the center of small Petri dish (60Ø×15 mm) (Greiner Bio One, Frickenhausen, Germany) with a fifth potato dextrose agar (PDA; Difco, Lawrence, KS) and 10 μl of overnight culture of bacterial strains was spotted 2.5 cm away from the fungal plug. Where appropriate, 50 μg ml −1 and/or 10 μg ml −1 apramycin and gentamicin, respectively, were used and either 0 or 2 mmol/L IPTG (Fisher Scientific) was added. Plates were incubated at 20°C for 72 hr. Phenotypic analysis was conducted in triplicate (three biological replicates) and the experiment was repeated twice. Antifungal activity was determined as the percentage inhibition of radial growth (PIRG) as previously described by Michelsen, Watrous, Glaring, Kersten, Koyama, et al. (2015).

| Peptide extraction
Strains were grown as described previously for phenotypic analysis.
The agar and biomass from a single plate were transferred to a Falcon tube and 10 mL 2-butanol containing 1% formic acid was added. The tubes were shaken and then placed in an ultrasonic bath for 30 min before being subjected to centrifugation at 4,000g for 10 min. The organic phase was removed and then dried under a constant stream of nitrogen. The residue was resuspended in 300 μl of methanol, subjected to centrifugation at 4,000g for 10 min, and then 250 μl of the solution was collected for LC-HRMS analysis.

| Liquid chromatography-high-resolution mass spectrometry analysis
Ultrahigh-performance liquid chromatography-diode array detectionquadruple time of flight mass spectrometry (UHPLC-DAD-QTOFMS) was performed on an Agilent Infinity 1290 UHPLC system (Agilent Technologies, Santa Clara, CA) equipped with a diode array detector.
Separation was obtained on an Agilent Poroshell 120 phenyl-hexyl column (2.1 × 250 mm, 2.7 μm) with a linear gradient consisting of T A B L E 2 Primers used in this study for qRT-PCR

| Construction of a mCherry-based reporter of nunF gene expression
The reporter strain P. fluorescens In5 harboring the nunF gene promoter region cloned in front of mCherry was previously constructed (Hennessy, Stougaard, & Olsson, 2017). Briefly, a fragment containing the predicted promoter region upstream of the nunF start codon was amplified using LongAmp™ Taq DNA polymerase (NEB) from strain In5 genomic DNA; the PCR product was then cloned using Gibson Assembly ® (NEB) into pSEVA237R (RK2-Km R -mCherry) upstream of an mCherry-expressing cargo. The resultant construct was then transformed into strain In5 by electroporation as described earlier.

| Analysis of nunF gene expression in vitro
For monitoring nunF gene expression in vitro, strain In5 carrying nunF-mCherry on pSEVA237R was grown overnight in a minimal media (DFM) supplemented with 0.5% wv −1 glucose and 25 μg ml −1 kanamycin with shaking 200 rpm at 28°C. Cells were washed twice with 0.9% wv −1 NaCl and resuspended to an OD 600 nm = 0.1 and 20 μl was added to a 96-well microtiter plate, together with 180 μl of DFM supplemented with 0.05% wv −1 of different carbon sources.
Analysis of nunF gene expression was performed using biological triplicates. Increase in mCherry responses faster than glucose was seen as a preferential nunF regulation in response to that carbon source.

| Genomic context of nunF in P. fluorescens In5
In previous studies, whole genome sequencing and genome walk-
T A B L E 3 Antifungal and anti-Pythium activity of Pseudomonas fluorescens In5 and mutants defective in CLP production and nunapeptin synthesis ( Table 3). The nunF mutation did not affect growth of strain In5 as the nunF knockout mutant showed comparable growth to that of the WT as did mutants M2D1 and 5F5 ( Figure   S2). The role of CLPs in biofilm formation has been widely reported in pseudomonads (Bonnichsen et al., 2015;de Bruijn & Raaijmakers, 2009a). To investigate whether a nunF mutant and mutants defective in peptide production (M2D1 and 5F5) played a role in biofilm formation, the crystal violet method in microtiter plates was used with P. fluorescens SS101 as a reference strain (de Bruijn & Raaijmakers, 2009b) ( Figure S3). Compared to WT, all three mutants showed a significant reduction in biofilm formation with the nunF mutant the lowest biofilm-forming strain at 4 hr (p < .001) ( Figure S3A). There was no significant difference between planktonic cells among strains (p > .05) ( Figure S3B). Swarming motility was also tested (Table S2). Compared to the reference strain P. fluorescens SS101, In5 showed a featureless swarming phenotype and not dendritic. Swarming was observed only on 0.25% agar where the nunF mutant showed a significant reduction in swarming ability compared to WT and mutant strains (p < .05).

Previous observations have shown that nunamycin and nunapeptin
are key components of the antimicrobial activity of strain In5. To confirm the role of nunF in antimicrobial activity, complementation of the nunF mutant of strain In5 was performed using the low-copy, broadhost range vector pHN1270::nunF which restored antifungal activity when tested against R. solani, whereas the empty vector control pHN1270 had no effect (Figure 3).

| Transcriptional analysis of the nunF mutant of P. fluorescens In5
NRPs are synthesized by nonribosomal peptide synthetases ( (Figure 4b). Interestingly, the expression of nupA and nupB compared to the WT were 19% and 47%, respectively, suggesting that a more complex regulation is involved in nunapeptin synthesis that may not solely rely on nunF. In order to investigate the effect of a nunF knockout on transcription of additional genes putatively encoding regulators located adjacent to NRPs biosynthesis genes, qRT-PCR analysis was also used to analyze the transcript levels of nupP, nupR1, and nupR2 in strain In5 WT compared to the nunF single-gene knockout mutant ( Figure 5). A significant reduction in the expression of nupP, nupR1, and nupR2 was observed in the nunF mutant strain compared to the WT suggesting that nunF interacts with the two putative regulators located on the NRPS genomic island downstream of the nunapeptin biosynthetic genes and possibly also nupP located upstream of the nup genes.

| NunF is required for CLP biosynthesis in P. fluorescens In5
LC-HRMS confirmed that the nunF mutant did not produce nunamycin or nunapeptin ( Figure 6) in detectable quantities. Analysis of extracts from strains grown on fifth strength PDA required to induce CLP production, detected both nunamycin and nunapeptin in the WT and complemented nunF knockout extracts (ΔnunF pHN1270::nunF), and neither peptide was produced in the nunF knockout empty vector (ΔnunF pHN1270) control ( Figure 6). Based on these results, nunF is required for both nunamycin and nunapeptin biosynthesis in strain In5. As nunF is putatively encoding a LuxR-type regulator, the presence of Nacyl-homoserine lactones (AHLs) was also investigated. The presence of Nhexanoyl-homoserine lactone was confirmed by accurate mass (observed m/z: 200.1279 for [M + H] + C 10 H 17 NO 3 + , mass deviation: 1.1 ppm) as well as retention time and tandem MS spectrum, particularly the diagnostic fragment at m/z 102.0549 (Kildgaard et al., 2014). This AHL was observed in all extracts, regardless of the growth medium or the presence/absence of the cyclic peptides.

| Dynamics of nunF gene expression in P. fluorescens In5
In order to investigate whether environmental factors play a role in regulating nunF expression, a reporter strain harboring the nunF promoter fused to mCherry was assayed in microtiter plates for growth and mCherry expression recorded on carbon sources indicating the presence of a plant rhizosphere or indicating the presence of a fungus and compared to glucose controls (Figure 7 and S4). The mCherry expression with the "fungal carbon sources" glycerol and trehalose was shown to be higher than with "plant carbon sources" like arabinose and cellobiose (Figure 7, Table S3 and Figure S4).

| DISCUSSION
The genomic island of P. fluorescens In5 harboring two large gene clusters required for synthesis of two CLPs nunamycin and nunapeptin, encodes three genes encoding putative LuxR-type transcriptional regulators, of which nunF was characterized in this study. The present findings demonstrate that the LuxR-like protein NunF is involved in the antimicrobial activity and regulation of the two CLPs nunamycin and nunapeptin in strain In5.
In Pseudomonas spp., the closest characterized homolog of NunF is the SyrF protein from P. syringae pv. syringae strain B301D and the recently characterized SyrF protein from the causal agent of brown spot disease on bean P. syringae pv. syringae B728a (Vaughn & Gross, 2016).
All three proteins have no defined N-terminal regulatory domain, but in contrast have a highly conserved C-terminal domain marked by a HTH DNA binding domain (Aravind, Anantharaman, Balaji, Babu, & Iyer, 2005;Vaughn & Gross, 2016). Phylogenetic analysis showed that NunF together with SyrF in addition to SalA and SyrG from P. syringae pv. syringae strains could be classified into a new subfamily of LuxR proteins as previously proposed (de Bruijn & Raaijmakers, 2009a;Jacobs et al., 2003). This family would also include the viscosin and massetolide LuxR regulators described for P. fluorescens strains SS101 and SBW25 (de Bruijn & Raaijmakers, 2009a).
F I G U R E 5 Effect of a nunF single-gene deletion on transcription of the nupP, nupR1, and nupR2 genes putatively encoding transcriptional regulators located adjacent to the nunapeptin biosynthesis genes. qRT-PCR was conducted on RNA samples harvested at 24 hr from strains incubated on fifth strength PDA at 20°C. mRNA accumulation of genes putatively encoding the transcriptional regulators nupR1 and nupR2 in addition to nupP located upstream of the nunapeptin biosynthetic genes was quantified relative to the housekeeping gene encoding DNA gyrase subunit B (gyrB) (HM070426) and the percentage mRNA accumulation of the nunF mutant relative to the WT representing 100% was calculated. Quantitative RT-PCR (qRT-PCR) was conducted twice with biological triplicates. Error bars represent the standard error of the means. Pseudomonas fluorescens In5 nunF single-gene deletion strain (ΔnunF) nupP, nupR1, and nupR2 mRNA accumulation significantly different from WT strain is highlighted with an asterisk (level of significance: *≤.05, **≤.01, ***≤.001) F I G U R E 7 Analysis of the temporal response of the nunF promoter to rhizosphere-associated carbon sources using a nunF promoter mCherry reporter strain. A nunF promoter fusion to mCherry was constructed and used as a reporter system in P. fluorescens In5 for monitoring nunF gene expression in vitro. Pseudomonas fluorescens In5 reporter strain (pSEVA237R::PnunF::mCherry) and the control strain (pSEVA237R::mCherry) were screened in a 96-well microtiter plates with varying rhizosphere-associated carbon sources. proposing nunapeptin to be potent against Fusarium spp., and Pythium in contrast to nunamycin which is more effective in the growth inhibition of R. solani (Michelsen, Watrous, Glaring, Kersten, Koyama, et al., 2015). Interestingly, interactions between the In5 strains and N. crassa produced phenotypes similar to those observed for R. solani and not for F. graminearum as would have been expected as both fungi belong to the Ascomycota phylum. It is important to note that the M2D1 and 5F5 were generated by random transposon insertion which may have polar effects on the expression of neighboring genes (Jacobs et al., 2003). The nunF mutant generated by insertional-directed mutagenesis when screened in a dual-culture assay showed a complete loss of antimicrobial activity against all pathogens tested. This was expected based on the phenotypic analysis of M2D1 and 5F5, but surprising when compared to previous studies in P. syringae pv. syringae where a mutant in syrF showed a 61% reduction in virulence in P. syringae pv. syringae B728a and 83% in P. syringae pv. syringae B310D, respectively (Lu et al., 2002;Vaughn & Gross, 2016). This loss of antagonism toward fungi and Pythium can be attributed to the loss in production of both nunamycin and nunapeptin as validated by LC-HRMS analysis. The nunF mutant did not produce detectable levels of nunamycin or its derivatives and also did not produce nunapeptin or its derivatives. In contrast to strain In5, where a functional copy of nunF is required for production of both peptides, syrF appears to be critical for syringomycin production in P. syringae pv. syringae B301D and to a lesser extent syringopeptin production (Lu et al., 2002). Nunamycin and nunapeptin production were restored when a functional copy of nunF was expressed on a broad-host range RK2 ori complementation plasmid. Levels were not comparable to the WT correlating with the dual-culture assay where the complemented nunF mutant showed a restoration of antifungal activity of 79%. This could be the result of using a synthetic RBS and spacer region and low copy complementation vector. Partial restoration of nunamycin and nunapeptin is in accordance with complementation studies described in P. syringae pv. syringae B728a where partial restoration of syringomycin was reported following expression of syrF in trans (Vaughn & Gross, 2016 (Lu et al., 2002(Lu et al., , 2005Wang et al., 2006). As discussed earlier, the organization of the strain In5 antifungal genomic island is similar to that encoding the P. syringae phytotoxins. However, one difference is the absence of SalA or SyrG homologs on the strain In5 antifungal genomic island. An important difference between strain In5 and P. syringae pv. syringae is that In5 is a fungal pathogen, while the latter is a plant pathogenic isolate of Pseudomonas, thus we would expect that regulation of antifungal compounds is different from phytotoxins.
An interesting observation is the fact that In5 produces two different lipopeptides which is unusual for pseudomonads with the exception of pathogenic isolates (Raaijmakers et al., 2006). It has not yet been determined whether purified nunamycin, nunapeptin, or a combination of both peptides is phytotoxic. In P. syringae pv.
syringae, syrB required for syringomycin production is activated in response to plant molecules (Mo & Gross, 1991). Therefore, we conducted mCherry reporter assays to determine whether nunF transcription is affected by environmental factors, more specifically by carbon sources indicating the presence of fungi rather than plant-related carbon sources. Among the fungal-associated carbon sources tested was laminarin which is a linear beta-1,3 glucan similar to the outer parts of fungal cell walls for which we have previously observed a strong mCherry signal when In5 is grown on high concentrations of the substrate (data not shown) (Brown & Gordon, 2003;Fesel & Zuccaro, 2016;Klarzynski et al., 2000;Trouvelot et al., 2014). Oxalate and citrate are common organic acids fungi used to solubilize minerals from the environment and bacteria growing on fungal hyphae are often oxalotrophic (Bravo et al., 2013;Deveau et al., 2010;Gadd, 2007;Scheublin et al., 2010). Oxalate and citrate did not stimulate mCherry formation which may be due to the buffering capacity of the acids. The strongest signals were recorded in response to trehalose and glycerol which are typically accumulated in hyphae during stress and in particular during drought stress (Bhaganna et al., 2010;Hallsworth, 1997;Wyatt et al., 2015). In addition, glycerol has been found to be supporting bacterial growth in the fungal hyphosphere, where bacteria can ensure survival in the environment by accessing nutrients from fungal exudates (Boer, Folman, Summerbell, & Boddy, 2005;Boersma, Otten, Warmink, Nazir, & van Elsas, 2010;Nazir, Warmink, Voordes, van de Bovenkamp, & van Elsas, 2013). A strong signal was also observed for glucose which is readily mineralized by both bacteria and fungi in the rhizosphere. The stimulation of nunF expression in response to carbon sources indicating the presence of a fungus suggests that In5 is specialized for growth in the fungal hyphosphere. Conversely, this has been demonstrated in P. syringae pv. syringae which is specialized in plant pathogenesis, and therefore responds to plant-associated compounds (Mo & Gross, 1991).
The ability of fungal-associated carbohydrates to elicit an mCherry response by the In5 reporter strain indicates that they also play an important role in fungal-bacterial interactions.
Understanding the mode of action underpinning the biocontrol activity of In5 is essential for the application of this agent for the biological management of plant diseases. This study is a first step toward unraveling the regulatory network of the nun-nup regulon in P. fluorescens In5 required for nunamycin and nunapeptin synthesis which are key factors underpinning the biocontrol activity of this isolate. The NunF regulator was shown to be part of the nun-nup regulon and is essential for the production of both nunamycin and nunapeptin, and therefore, is also a key factor involved in the antimicrobial activity of In5. In future studies, it will be important to determine the function of nunF in the biocontrol activity of In5 in soil systems, and also to investigate in further detail the interplay between the additional LuxR encoding genes nupR1 and nupR2 and whether a more complex regulatory network involving additional genes is at play and mediating the antimicrobial activity of strain In5 through the synthesis of nunamycin and nunapeptin.

ACKNOWLEDGMENTS
This work was supported by the Villum Foundation grant VKR7310 (Microbial Communication-A Key to the Development of Novel Sustainable Agri-and Aquaculture Practices Using Biological Control Bacteria). We would like to thank Dr Nakashima (Tokyo Institute of Technology, Japan) for kindly providing the plasmid pHN1270. We are also thankful to Susanne Iversen for skillful technical assistance in the laboratory. Agilent Technologies is acknowledged for the Thought Leader Donation of the 6545 UHPLC-QTOF.