Background

Biocontrol bacteria provide an alternative strategy to synthetic chemicals for crop protection against disease. Many microorganisms have been shown to be effective microbial biological control agents (mBCAs) against soil-borne pathogens and in particular, fluorescent pseudomonads have been the focus of much research [12, 16, 38, 39]. Among these studies, the production of secondary metabolites and especially the role of cyclic lipopeptides in the biological control of plant pathogens has been widely documented [18, 27, 28, 30, 31].

Pseudomonads with potential applications in biocontrol have been widely studied using whole-genome sequencing in combination with functional analysis using either mutants derived by random or site-directed mutagenesis as tools for the identification of key traits underpinning biocontrol activity [4, 5, 8, 9, 24, 25, 36, 37]. However, knowledge of the overall response of biocontrol bacteria to specific pathogens is limited and therefore transcriptional profiling by RNA-seq of dual cultures is a valuable tool to elucidate potential mode of actions underpinning biocontrol bacterial antagonism toward pathogens. Despite the advances in the application of high-throughput RNA sequencing (RNA-seq) as a tool for transcriptomics, only few studies to date report using the method to study bacterial-fungal interactions [11, 26]. A recent study by [26] investigated the transcriptional response of Serratia plymuthica against R. solani in dual-culture and found that antibiosis appeared to be a key mode of action utilised by the bacterium against the pathogen.

P. fluorescens In5 is a potential biocontrol agent previously isolated from a disease suppressive soil in southern Greenland showing antimicrobial activity against a broad range of phytopathogens [24, 25]. Using a combination of whole-genome sequencing, mutant generation and characterisation coupled with microbial metabolomics analysis, key biocontrol traits of this isolate have recently been identified [13, 25]. In order to build upon research conducted to date and to contribute toward our knowledge of bacterial-fungal/oomycete interactions, the aim of the present study was to investigate the transcriptional changes in P. fluorescens In5 during dual-culture with two phytopathogens. Genome-wide RNA-seq was used as a method to investigate the transcriptional response of P. fluorescens In5 in dual-culture with either the basidiomycete R. solani or the oomycete P. aphanidermatum compared to a single culture of P. fluorescens In5 in the absence of both pathogens.

Methods

Dual culture assay

Nunc™ OmniTray™ (Fisher Scientific, Roskilde, Denmark) sterile plates were filled with 35 ml of fifth strength potato dextrose agar (PDA; Difco Lawrence, KS) and ten 5 mm plugs of either R. solani or P. aphanidermatum were place in the center of the plate (see Additional file 1 for plate layout) and incubated 24 h at 20 °C. P. fluorescens In5 was grown in 10 ml Luria–Bertani broth overnight at 28 °C and subsequently streaked 3 cm away from the fungal or oomycete plugs using a sterile inoculation loop. Plates were incubated at 20 °C for 48 h.

RNA isolation

At 48 h, bacterial cells representing biological triplicates were scraped into 500 µl of RNAlater® (Thermo Fisher), micropipetted and briefly vortexed before being stored at 4 °C overnight. Before proceeding to RNA extraction, 1 ml of cold phosphate-buffered saline (PBS) was added to each tube containing 500 µl RNAlater® and centrifuged 5 min at 14,000 g and supernatant removed before proceeding to RNA extraction. Total RNA was isolated using the ZR Fungal/Bacterial RNA MiniPrep™ kit according to the manufacturer’s instructions (Zymo Research, Nordic Biosite, Copenhagen, Denmark).

RNA sequencing

RNA sequencing (RNA seq) libraries were constructed and sequenced following a default Illumina stranded RNA protocol including rRNA depletion (BGI tech, Hong Kong, China). The short-insert library was sequenced on an Illumina HiSeq system by 2 × 100 bp paired-end sequencing producing approximately 10 million reads per sample. Trimming and quality filtering of sequences and transcriptomics analysis were performed using CLC Genomics Workbench (CLC bio, Qiagen, Aarhus, Denmark). The trimmed sequences were mapped to the annotated In5 genome (GenBank accession no. LIRD01000000) using default options for prokaryotes, except that the minimum length and similarity fraction of a matched read was set to 0.8 and 0.9, respectively. Only reads where both ends of a paired-end read could be mapped were counted, all other reads were discarded. Expression values for individual coding sequences (CDS) were calculated as Reads Per Kilobase of transcript per Million mapped reads (RPKM). The average expression value from the three biological replicates was used to calculate fold change differential expression of all annotated CDS in response to R. solani or P. aphanidermatum compared to the control (P. fluorescens In5 single culture). For calculating fold-change, only genes with a minimum expression value (RPKM) >5 in both control and interaction was included.

Significance of the transcriptomic dataset was calculated as follows. For the three replicate RPKM values (control, Rs, Pa) the standard error (SE) was calculated for each gene and for the three treatments (control, Rs, Pa). The standard error of the means (SEM) was subsequently calculated for the ratio of the two fungal treatments (Rs, Pa) and the control treatment (C). From the resulting values, the margin of error (MOE) was calculating containing 95% of the distribution (MOE = 1.96*SEM). Finally, for significance testing the following was determined: if treatment/control ratio − MOE is larger than 1 then fungal treatment is significantly larger (P < 0.05) than control. If the ratio + MOE is smaller than 1 then the fungal treatment is significantly smaller (P < 0.05) than control.

Results and discussion

Phenotypic analysis and targeted response against specific pathogens by P. fluorescens In5

In order to gain insight into the P. fluorescens In5 mechanism of action against phytopathogens, a dual-culture assay was established to study the bacterial transcriptome in response to R. solani and P. aphanidermatum. Transcriptional profiling of P. fluorescens In5 using RNA-seq was conducted on single bacterial cultures as a control, or in dual-culture with either R. solani or P. aphanidermatum. Approximately 10 million paired-end sequence reads were generated from each of the three biological replicates for each setup and gene expression levels were determined by comparison to the available genome sequence [13].

The most striking feature of this comparative transcriptomics study was the organism-specific response of P. fluorescens In5. Overall, the majority of genes were unchanged in expression relative to the control (Fig. 1; Additional file 2). Importantly, only two transcripts were up-or downregulated at least twofold in response to both R. solani and P. aphanidermatum (Table 1; Additional file 3: Table S1; Table 2; Additional file 4: Table S2), an upregulated gene encoding an Mbth-like protein (AL066_04890) and a downregulated gene encoding a putative AlpA transcriptional regulator (AL066_26360). Interestingly, no transcripts were upregulated more than twofold in R. solani and downregulated more than twofold in P. aphanidermatum or vice versa (Additional file 2).

Fig. 1
figure 1

Gene expression 2log plot of P. fluorescens In5 during interactions with R. solani and P. aphanidermatum. Graph represents 2log of expression means for each treatment (Rs, R. solani or Pa, P. aphanidermatum) compared to the control (C)

Full size image
Table 1 Differential expression of P. fluorescens In5 genes during an interaction with R. solani compared to P. aphanidermatum
Full size table
Table 2 Differential expression of P. fluorescens In5 genes during an interaction with P. aphanidermatum compared to R. solani
Full size table

During dual cultivation of the bacterium with R. solani, a zone of inhibition was established between P. fluorescens In5 and the fungus which persisted for over three weeks. Analysis of dual-cultures of P. fluorescens In5 and R. solani identified a total of 37 genes, including 12 hypothetical genes, with at least a threefold change in expression level compared to the control (Table 1). Of the 37 differentially expressed transcripts, 30 were shown to be significant (Additional file 3: Table S1). In contrast, only four genes showed significant differential gene expression by more than threefold change during the interaction with P. aphanidermatum (Table 2; Additional file 4: Table S2). Among the 37 differentially expressed genes during the interaction with R. solani, 29 genes were upregulated and eight genes were downregulated.

Upregulation of genes involved in secondary metabolism and detoxification in response to R. solani

Many of the genes of known function whose expression was highly upregulated in response to R. solani encode enzymes associated with aromatic compound metabolism and detoxification (Table 1; Additional file 3: Table S1). Two genes, both putatively coding for quercetin 2,3-dioxygenases (AL066_14420, AL066_31575) were upregulated over sixfold. In the rhizosphere, microbes are continually exposed to secondary metabolites including aromatic compounds and flavonoids such as quercetin, which has been shown to possess antibacterial activity by inhibiting DNA gyrase [17, 29]. Quercetin 2, 3-dioxygenase is required for quercetin degradation and functions by opening the C-ring forming a depside and releasing a carbon monoxide. In addition to the upregulation of quercetinases, two hydrolases (AL066_06105, AL066_09895) were the most highly upregulated genes in response to R. solani. The first hydrolase (AL066_06105) shows 83% identity to a characterised esterase from P. putida, while the second hydrolase (AL066_09895) is distantly related by 31% to a phosphoesterase from E. coli. The final step in quercetin metabolism is hydrolysis of the depside formed from the dioxygenase activity by an esterase to yield 2-protocate-chuoyl-phloroglucinol carboxylic acid and protocatechuic acid [15]. In relation to the rhizosphere, it has been reported that the ability of the biocontrol strain P. fluorescens WCS365 to utilise organic acids is the nutritional basis for tomato root colonisation by the bacterium [23]. These results suggest that P. fluorescens In5 detoxifies fungal-derived aromatic compounds potentially produced by R. solani in response to the bacterial interaction and possibly also subsequently metabolises the degradation products. To the best of our knowledge, R. solani or other fungi have not yet been shown to naturally produce quercetin with the exception of a medicinal plant endophyte [22]. However, quercetin 2,3-dioxygenases have been shown to act on different flavonoids indicating the potential production of phenolic-like compounds by R. solani, which has been documented for Rhizoctonia spp. [6, 21, 33]. Additional significantly upregulated genes relating to aromatic compound metabolism were two FMN-dependent NADH-azoreductases (AL066_18305, AL066_ 10590), an aromatic ring-opening dioxygenase (AL066_04230) and a glutathionyl-hydroquinone reductase (AL066_05530).

Analysis of the quercetinases discussed above found that both proteins belong to the pirin family with one of the enzymes (AL066_31575) possessing the pirin c-terminal cupin domain. In addition to these two enzymes, a pirin-like protein encoding gene (AL066_10105) was also significantly upregulated (>40-fold change) in response to R. solani. In P. stutzeri, a pirin-like protein has been shown to possess quercetinase activity, although the biological function of pirin remains largely unknown [1]. In eukaryotes, pirin has been proposed to be involved in transcriptional activation and cell apoptosis, while in prokaryotes, for example in cyanobacteria, it has been shown to be stress induced [1, 14, 34, 40].

Interaction with R. solani induces the expression of small hypothetical proteins

A notable feature of the transcriptional response of P. fluorescens In5 to R. solani was the upregulation of several genes encoding hypothetical proteins (Table 1; Additional file 3: Table S1). The genes were found to encode small proteins all predicted to be secreted with the exception of AL066_12150 (Additional file 5). Antimicrobial peptides (AMPs) are small molecules produced and secreted by diverse organisms and can be referred to as cationic host defence peptides, anionic or cationic peptides or alpha-helical antimicrobial peptides among others [2]. Antifungal peptides can be diverse in structure and typically target fungal cell walls or membranes; for example they can bind the major structural component of fungal cells walls chitin, or disrupt fungal membranes increasing permeability or directly form pores [2]. These hypothetical proteins could be involved in the response of the bacterium to R. solani or antagonism toward the fungus. However, further studies providing proteomic or immunological evidence is required to establish whether these proteins are present extracellularly.

In contrast to R. solani, there was no significant inhibition zone observed for the interaction between P. fluorescens In5 and P. aphanidermatum at 48 h, though the oomycete did not grow past the bacterium. The anti-Pythium activity did not however persist beyond 72 h dual cultivation. In response to P. aphanidermatum, there were no genes showing threefold or greater differential expression relative to the control (Table 2). Of the genes recorded to have fold-change values above two, nine transcripts were upregulated and 16 downregulated. Among the upregulated genes, five encode hypothetical proteins and the remaining genes putatively encode three transporters (AL066_27625, AL066_07870, AL066_7605), a malonate decarboxylase delta subunit (AL066_27960) and an antibiotic synthesis mbth protein (AL066_04890). These results indicate that antibiosis may be a mode of action by P. fluorescens In5 during the interaction with P. aphanidermatum similarly to the mechanism reported to underpin the interaction between S. plymuthica and R. solani [26]. Transcripts downregulated by a fold-change greater than two included two hypothetical genes (AL066_11985, AL066_24330) in addition to a putative AlpA transcriptional regulator (AL066_26360) which as mentioned previously, was also downregulated during the R. solani interaction.

Interaction with P. aphanidermatum induces an Mbth-like protein and ABC-transporter like encoding genes

As discussed earlier, differential expression analysis showed few to no changes in response to the oomycete P. aphanidermatum. Among the Pythium-induced genes, two ABC-transporter like genes were strongly upregulated (AL066_27625, AL066_07605). Biocontrol agents must be able to tolerate antibiotics produced by plant pathogens in addition to their own compounds and thus upregulation of these genes could indicate a potential role in cell detoxification during the biocontrol interaction [32]. Interestingly, a beta-lactamase gene (AL066_07745) typically required for antibiotic resistance against beta-lactams was downregulated. This could point to Pythium not producing beta-lactams in response to In5 or that In5 responds specifically to fungi rather than oomycetes, which more closely resemble plants. The latter is in accordance with the transcriptomic data here presented whereby the Pythium interaction when compared to the control showed few to no up-or-downregulated transcripts.

Another upregulated transcript was an MbtH-like protein (AL066_04890) located downstream of an NRPS encoding gene (AL066_04845). MbtH-like proteins are small proteins of unknown function although a potential role in NPRS biosynthesis has been proposed [3, 10]. It has also been reported that co-production of such proteins with NPRS components can enhance protein production [10]. Interestingly, MbtH-like proteins have been proposed as useful target genes for the identification of novel secondary metabolite gene clusters by genome mining [3]. Genome analysis of In5 identified a second MbtH-like protein also located downstream of an NRPS gene cluster. Of note, disruption of this cluster renders the mutant unable to inhibit Pythium and switches off nunapeptin production (data not shown).

Identification of co-expressed gene clusters during bacterial-phytopathogen interactions

The transcriptomic study conducted by [26] reported the upregulation of genes in clusters. Clustering of differentially expressed genes was also observed in this study for a number of transcripts in response to both phytopathogens (Fig. 2). For example, four genes in a cluster were upregulated in response to R. solani or both R. solani and P. aphanidermatum (Fig. 2a). Interestingly, within this cluster, two putative LysR transcriptional regulators (AL066_06080, AL06110) were identified. Although not differentially expressed at 48 h, these regulator-like proteins may play an important role in antifungal activity at earlier or later time points. In P. chlororaphis, a knockout mutant of ptrA encoding a LysR-type regulator was shown to be defective in antifungal activity against Scelorotinia sclerotiorum, indicating a functional role relating to the biocontrol activity of the bacterium [20]. Four of the downregulated genes in P. fluorescens In5 in response to R. solani were clustered and putatively encode proteins involved in sulphur oxidation (Fig. 2b). Expression of these genes might be more relevant in response to the microbial production of sulfur-containing compounds or environments rich in organic C such as the rhizosphere [7, 19, 35]. At present, the role of genes located up- or downstream of differentially expressed transcripts remains to be determined.

Fig. 2
figure 2

Example of the clustering of genes differentially regulated in response to both R. solani and P. aphanidermatum (a) or R. solani (b). Genes upregulated in response to R. solani are represented by green arrows, genes upregulated in response to both R. solani and P. aphanidermatum are represented by blue arrows, and yellow arrows indicate genes downregulated in response to R. solani

Full size image

In a previous study, the non-ribosomal peptides (NRPs) nunamycin and nunapeptin were characterised as key components of the biocontrol activity of P. fluorescens In5 [25]. None of the genes with the exception of pseC, located on the NRPs-encoding genomic island were differentially regulated in response to R. solani or P. aphanidermatum at the time point tested. As production of the peptides decreases after 48 h (data not shown), it is likely that the biosynthesis genes are upregulated in the early stages of the interaction and once the non-ribosomal peptides synthetase (NRPS) genes are expressed, the mega-enzyme complex is formed and directs peptide biosynthesis independently from ribosomes and mRNA. An important limitation of this study was the technical issue of the quantity of RNA that could be extracted at early time-points and consequently 48 h was selected for analysis similarly to the study described by [26]. Overall, the results from this study investigating the transcriptional response of P. fluorescens In5 during interactions with two phytopathogens, point towards a general mechanism of antibiotic biosynthesis and transport and most notably an organism-specific mode of action which in response to R. solani is secondary metabolite detoxification, degradation and metabolism.

Conclusions

Transcriptomic profiling of the P. fluorescens In5 responses toward two significant phytopathogens has directed our genomic analysis of biocontrol traits to reveal previously uncharacterised genes for future functional genomics studies. In addition to genes of known function, many genes encoding proteins of unknown function were differentially expressed indicating that multiple genes and processes appear to be involved in both the antagonism of P. fluorescens In5 toward phytopathogens and the bacterium’s response. It has been widely documented that the primary mechanism of pathogen inhibition by biocontrol bacteria is the production of antimicrobial secondary metabolites. Based on the results presented here, transcripts required for secondary metabolite degradation were highly upregulated indicating that secondary metabolite detoxification may be a key defence mode of action during the interplay between P. fluorescens In5 and plant pathogens. This preliminary transcriptome analysis also demonstrates that P. fluorescens In5 is specialised in the antagonism of fungal pathogens and, while having demonstrated anti-Pythium activity, the response of the bacterium toward the oomycete is not as defined. Going forward it will be important to conduct similar transcriptomic-based studies. Ultimately, understanding how biocontrol agents respond to different pathogens is critical if such microbes are to play a role in the management of plant diseases.