All bacterial strains and plasmids used in this study are presented in Table 1. P. fluorescens PF07 and its mutants were grown in Luria-Bertani (LB) or tryptone broth (10 g/L tryptone) medium at 28°C.

The deletion mutants ΔrpoN, ΔfapC, and ΔbrfA of P. fluorescens PF07 were constructed by double-crossover allelic exchange, according to the method by Liu et al. (2018). All primers used in the mutant construction are listed in Supplementary Table 1. Briefly, two fragments flanking rpoN were amplified by PCR with 07rpoN-MF1/MR1 and 07rpoN-MF2/MR2 primers, respectively. The sequences of the two primers 07rpoN-MR1 and 07rpoN-MF2 were reverse and complementary. The two fragments were purified and fused in subsequent PCR reactions using 07rpoN-MF1/MR2 primers. In the same way, two fragments flanking the fapC or brfA were amplified and fused using the corresponding primers of each gene. The fused segment of the rpoN gene with a deletion of 1,258 bp in its 1,494-bp open reading frame (ORF), the fused segment of the fapC gene with a deletion of 676 bp in its 756-bp ORF, and the fused segment of the brfA gene with a deletion of 1,134 bp in its 1,326-bp ORF, were cloned into the suicide vector pLP12Cm using a ClonExpress II One Step Cloning Kit (Vazyme, China), respectively. The recombinant plasmids pLP12Cm-rpoN, pLP12Cm-fapC, and pLP12Cm-brfA were transferred from E. coli β2163 (Demarre et al., 2005) into P. fluorescens PF07 by conjugation. The single crossover transconjugants were selected on LB agar medium containing 20 μg/mL chloramphenicol and 0.3% D-glucose at 28°C. L-arabinose can induce the expression of the virulence gene vmi480 on the suicide vector; thus, the second crossover mutants were screened on LB agar with 0.4% L-arabinose. The PF07 ΔrpoN, ΔfapC, and ΔbrfA mutants were confirmed via PCR, using external primer pairs rpoN-TF/TR, fapC-TF/TR, and brfA-TF/TR, respectively. The deletion mutants were further confirmed by sequencing.

The overnight cultures of PF07 and the rpoN mutant (5 μL each) were spotted on tryptone plates (1% tryptone, 1% agar). After 1 day of incubation at 28°C, 24 colonies of PF07 were gently scraped from six tryptone plates and mixed in 10 mL of 0.9% NaCl. The sample was harvested by centrifugation at 7,000 × g for 10 min and frozen in liquid nitrogen prior to RNA isolation. After 5 days of incubation, macrocolonies of PF07 and the rpoN mutant were collected using the same method, respectively. Two biological replicates were prepared for each treatment. The obtained samples were sent to Novogene Co., Ltd. (Beijing, China) for RNA-seq analysis. Total RNA was isolated from the frozen samples using an RNAprep Pure Cell/Bacteria Kit and a miRcute miRNA Isolation Kit (Tiangen, China). The concentration and quality of total RNA were validated using a NanoDrop spectrophotometer (Thermo, United States) and Agilent 2100 Bioanalyzer (Agilent, United States), and the qualified RNA samples were used as input materials. mRNA was purified from total RNA using probes to remove rRNA. Then, strand-specific RNA sequencing libraries were constructed using NEBNext Ultra Directional RNA Library Prep Kit for Illumina (NEB, United States) according to the manufacturer’s introductions. Following validation with an Agilent Bioanalyzer 2100 system (Agilent, United States), the libraries were sequenced on an Illumina Hiseq platform and paired-end reads were generated. The raw reads were filtered to remove adapter-containing reads, high N rate reads (N rates > 10%), and low-quality reads (50% bases with Q-score ≤ 5) using in-house Perl scripts. The genome sequence of P. fluorescens PF08, an isolate from refrigerated turbot, was used as the reference genome,¹ and the obtained clean reads were mapped to the reference genome using Bowtie2 v2.2.3 (Langmead and Salzberg, 2012). The novel gene and sRNA were identified by Rockhopper (McClure et al., 2013). The gene expression levels were quantified, and differential expression analysis was performed using DESeq2 (Love et al., 2014). Genes with an adjusted P-value < 0.05 and | log₂ fold change| ≥ 1 were considered to be differentially expressed (Wang et al., 2018a; Liu et al., 2019).

qRT-PCR analyses were performed with RNA samples under the same culture conditions used for the RNA-seq experiments. Total RNA was isolated with a TRIzol Plus RNA Purification Kit (Invitrogen, United States) and treated with RNase-Free DNase Set (Qiagen, Germany) to remove the contaminating DNA. cDNA was synthesized using the SuperScript™ III First-Strand Synthesis SuperMix (Invitrogen, United States), and qRT-PCR analyses were conducted using a CFX384 Touch Real-Time PCR Detection System (Bio-Rad, United States) with the Power SYBR1 Green PCR Master Mix (Applied Biosystems, United States). Primers for qRT-PCR are listed in Supplementary Table 2. The 16S rRNA gene was used as a reference control for sample normalization. The relative expression of the target genes was calculated using the 2^–ΔΔCt method (Livak and Schmittgen, 2001). Three biological replicates were prepared, and samples were run in triplicate.

To identify the 5′-end of a gene, rapid amplification of 5′-cDNA ends (5′-RACE) was performed according to a method described previously (Liu et al., 2018). Briefly, the Vaccinia Capping System (NEB, United States) and GeneRacer Kit (Invitrogen, United States) were used according to the manufacturers’ instructions. The 5′-ends of the transcripts were capped with the Vaccinia Capping System, and the GeneRacer Kit ensured the amplification of only full-length transcripts via the elimination of truncated messages from the amplification process. The fapA specific primers used for 5′-RACE are listed in Supplementary Table 1. The 5′ RACE products were directly sequenced to identify the 5′-end of fapA.

Macrocolony biofilms were observed by Congo red assay and transmission electron microscopy (TEM), using a previously reported method (Liu et al., 2019). Congo red plates (1% tryptone, 1% agar, 40 μg/mL Congo red, and 10 μg/mL Coomassie brilliant blue G250) were used to assess macrocolony biofilm color and morphology. Cells of PF07 and the mutants were grown overnight in liquid LB medium with shaking at 28°C. A total of 5 μL of the overnight cultures was spotted on Congo red plates. The plates were incubated at 28°C for 1, 3, 5, and 7 days to observe the macrocolony morphology. For TEM analysis, the macrocolonies on day 5 were gently scraped from a tryptone plate (1% tryptone, 1% agar) and fixed in 2.5% glutaraldehyde for more than 4 h at room temperature. Then, the macrocolony biofilms were postfixed, dehydrated, embedded in TAAB resin (TAAB Laboratories, United Kingdom), ultrathin-sectioned, stained, and observed via TEM (Hitachi H-600, Japan).

Overnight cultures of PF07 and the mutants (6 μL each) were added into 6 mL of fresh sterile tryptone broth. The dilution culture was statically grown in a glass tube at 28°C for 3 days. Pellicles were assayed by visual inspection of the air–liquid interface of the standing culture.

Quantification of biofilms grown in microplates was performed as described by Liu et al. (2021). Tryptone broth was used as the growth medium, and cultures were incubated at 28°C for 6, 12, 24, and 48 h with shaking (150 rpm). After crystal violet staining, the absorbance at 595 nm was measured using a microplate reader (Infinite 200, Tecan, Switzerland). Eight individual samples were assessed for each strain at each time point, and the experiment was repeated at least twice.

Overnight cultures were inoculated into 12-well dishes containing tryptone broth at a 1:1,000 ratio. Glass coverslips (20 mm × 20 mm) were partially submerged in the medium and grown at 28°C with shaking (150 rpm). After incubation for 24 h, the glass slips were washed two times with 0.9% NaCl. The biofilm on the glass slips was successively stained with 150 ng/mL 4′,6-diamidino-2-phenylindole (DAPI) for 20 min and 2 mg/mL Congo red for 20 min in the dark at room temperature (Larsen et al., 2007). The coverslips were washed and dried, and then a drop of liquid Antifade mounting medium (Beyotime, China) was placed on the coverslips to prevent fluorescence quenching. All the coverslips with stained biofilms were placed on slides and imaged with a Nikon Eclipse Ti2 confocal laser scanning microscopy (CLSM) (Nikon, Japan) and a 63 × oil objective. Fluorescence signals were recorded in the blue channel (excitation 405 nm, emission 425–475 nm) for DAPI and in the red channel (excitation 561 nm, emission 570–620 nm) for Congo red, where at least 10 fields of view were captured. NIS-Elements AR confocal software was used to analyze the biofilm images, allowing for collection of z-stacks.

Overnight cultures were diluted 1:100 into 150 mL fresh tryptone broth medium, and cultivated at 28°C with constant shaking (220 rpm) for 24 h. The optical absorbance at 600 nm was measured at 4-h intervals. Three independent experiments were performed, and the mean optical absorbance values at each time point were used to determine the growth rate.

P. fluorescens PF07 is a potent biofilm producer, and it formed a red, wrinkled macrocolony on a Congo red plate after at least 3 days of growth. The red and wrinkled macrocolony of bacteria often results from the production of large amounts of extracellular matrices (Friedman and Kolter, 2004). In the present work, RNA-seq-dependent transcriptomic analysis was utilized to identify genes involved in biofilm matrix formation and regulation. The macrocolony biofilms of PF07 grown on tryptone plates for 5 days (PF07_5d) were compared with those grown for 1 day (PF07_1d). RpoN is an important sigma factor involved in biofilm regulation of bacteria (Xu et al., 2019; Liu et al., 2021); therefore, the rpoN mutant of PF07 was constructed to determine the regulatory role of RpoN in biofilm matrix formation (Supplementary Figure 1). The raw sequencing data of RNA-seq were deposited in the Sequence Read Archive,² with an accession number PRJNA800221. After the raw reads were filtered, the clean reads were mapped to the reference genome sequence of P. fluorescens PF08, with a total mapped rate of at least 96.31% for all samples. The high mapped rates suggest that PF07 strain is very closely related to the reference strain PF08, an isolate from refrigerated turbot (Xu et al., 2021). The obtained read data and mapping data are shown in Supplementary Tables 3, 4. According to the screening criteria for differentially expressed genes (DEGs) (| log2 fold change| ≥ 1, padj ≤ 0.05), a total of 1,403 DEGs were identified, including 641 significantly upregulated genes and 762 significantly downregulated genes in PF07_5d, compared with PF07_1d (Figure 1A). In addition, 159 DEGs were identified by comparing the macrocolony biofilms of rpoN mutant and PF07 on day 5, including 29 significantly upregulated genes and 130 significantly downregulated genes (Figure 1B). The detailed information on the two groups of DEGs is summarized in Supplementary Tables 5, 6. The relationship between the two groups of DEGs was further analyzed, and a total of 80 DEGs were obtained in the overlap of the two groups (Figure 1C).

Based on the functional annotation, chromosomal location, and expression pattern, the representative DEG clusters or DEGs are listed in Table 2, and the results were validated by qRT-PCR. The qRT-PCR results agreed with the expression patterns from the RNA-seq analyses, except for pgaABCD, a gene cluster related to the biosynthesis of the extracellular polysaccharide PNAG. The main components of the biofilm matrix are proteins or extracellular polysaccharides, and the genes coding for these components are usually present in the form of gene clusters (Blanco-Romero et al., 2020). The proteinaceous components of the biofilm matrix mainly include secreted extracellular proteins, cell surface adhesins and protein subunits of cell appendages, such as flagella and pili (Fong and Yildiz, 2015). As shown in Table 2, the main DEGs related to matrix proteins are fapABCDE, flp/tad, and flagellar genes. Interestingly, the gene cluster fapA-E, encoding functional amyloids, was significantly upregulated in PF07_5d compared with PF07_1d, and significantly downregulated in ΔrpoN compared with PF07 (Table 2). This gene cluster is likely to be involved in the formation of biofilm matrix and is positively regulated by RpoN. However, the flp/tad gene cluster, encoding a special type IVb pilus, was significantly downregulated in PF07_5d compared with PF07_1d. The expression of genes associated with flagella was also significantly downregulated in PF07_5d, compared with PF07_1d, as well as in ΔrpoN compared with PF07 (Table 2). In addition to the matrix protein genes, the representative DEGs related to polysaccharide metabolism are pgaABCD and glycogen metabolism genes. According to the results of RNA-seq, the pga gene cluster was upregulated in PF07_5d compared with PF07_1d, but the qRT-PCR results showed downregulation. The discrepancy suggests that the expression of pga genes in biofilms may lack repeatability as the qRT-PCR tests were performed using independently extracted RNAs. The expression of the genes associated with glycogen metabolism was also downregulated in PF07_5d compared with PF07_1d. Moreover, the gene cluster rhlABRI was significantly downregulated in PF07_5d compared with PF07_1d, but upregulated in the rpoN mutant compared with PF07_1d. The gene cluster encodes enzymes (RhlAB) for rhamnolipid synthesis and the quorum sensing system (RhlRI). Finally, the bacterial secretion system is also likely to be associated with biofilm formation. As shown in Table 2, the type II secretion system was upregulated, while another possible secretion system was downregulated at the transcription level in PF07_5d, compared with PF07_1d.

Some genes that may be involved in the regulation of biofilm matrix synthesis were also identified (Table 2). A novel RpoN-dependent transcription factor D7M10_RS14495, which we named BrfA, was significantly upregulated in PF07_5d compared with PF07_1d. Some genes encoding diguanylate cyclase and phosphodiesterase (D7M10_RS04965, D7M10_RS06580, and D7M10_RS22095) were significantly altered during the formation of the biofilm. The second messenger c-di-GMP regulates the expression of biofilm matrix components in most bacteria, and the level of intracellular c-di-GMP concentration is controlled by diguanylate cyclase and phosphodiesterase (Fazli et al., 2014). In addition, the transcript levels of several novel sRNAs were also altered during the formation of biofilms (Supplementary Table 5). However, these novel sRNAs require further confirmation.

With regard to the usage of carbon and nitrogen sources, we found that genes related to galactose catabolism (D7M10_RS00150-65) were significantly upregulated in the mature biofilm, but genes related to amino acid utilization (D7M10_RS13750-85, arcDABC) were significantly downregulated during the biofilm formation. In addition, a gene cluster with unknown function (D7M10_RS18830-15) was highly upregulated in PF07_5d compared with PF07_1d, and significantly downregulated in ΔrpoN compared with PF07 (Table 2).

The RNA-seq results indicated the gene cluster fapA-E was positively regulated by RpoN. RpoN recognizes promoter sequences with conserved GG and GC elements located –24 and –12 nucleotides upstream from the transcriptional start site (Barrios et al., 1999). According to chromosomal location and expression pattern, fapA-E are arrayed in an operon, and fapA is the first gene of the operon. The 5′-RACE method was performed to identify the transcriptional start site of fapA. As shown in Figure 2, a conserved –24/–12 element was present upstream from the transcriptional start site. This result suggests that fapA-E genes are directly regulated by RpoN.

The fap cluster contains six ORFs fapABCDEF. According to the results of the RNA-seq, the transcription of fapA-E was significantly upregulated during biofilm formation, but the expression of fapF was not upregulated (Table 2). The time-course analysis of fap gene expression during biofilm formation by qRT-PCR further supported the RNA-seq results (Figure 3A). These results suggest differences in the expression patterns of fapF and fapA-E. FapC is the main component of the Fap fibers (Bleem et al., 2018); thus, the fapC mutant was constructed to observe the role of the Fap amyloids during biofilm formation (Supplementary Figure 1). Biofilm morphology assays on Congo red plates were performed (Figure 3B). After 1 day of growth, the colonies of the fapC mutant were the same as the wild-type colonies and characterized as pale and smooth. After 3 days, the colonies of the wild type started to become red and wrinkled, but those of the fapC mutant remained pale and smooth. On days 5 and 7, the wild-type strain formed mature macrocolony biofilms, which were redder and more wrinkled than the fapC mutant. Additionally, we used TEM to visualize the structure of the macrocolony biofilms. After 5 days, the wild-type cells were encased in a basket-like matrix, whereas the mutant cells generated much less extracellular matrix (Figure 3B). These results suggest that PF07 can develop wrinkled macrocolony biofilms through the high production of the extracellular matrix, and that Fap amyloid is the main component of PF07 biofilm matrices.

Similar to the fapC mutant, the rpoN mutant also failed to form red and wrinkle macrocolonies, and it failed to generate an extracellular matrix in the biofilm (Figure 3B). The RNA-seq results show the cells in the macrocolony biofilm of the rpoN mutant significantly downregulated the transcription of fapA-E genes (Table 2). These results indicate that RpoN regulates the production of the biofilm matrix via regulation of the fap gene cluster.

Downstream of the fapA-F gene sequence of PF07, there is an adjacent reverse gene, brfA (D7M10_RS14495), which encodes an RpoN-dependent Fis family transcriptional regulator and is upregulated in mature biofilms (Table 2). The deletion mutant of brfA was constructed (Supplementary Figure 1), and ΔbrfA showed macrocolony biofilms similar to ΔfapC and ΔrpoN (Figure 3B). In addition, the transcription of fapA and fapC was also significantly down-regulated in ΔbrfA as in ΔrpoN (Figure 3C). These results suggest BrfA is likely to act in conjunction with RpoN to regulate the transcription of fap genes.

P. fluorescens PF07 formed robust pellicles when grown in standing culture (1% tryptone) after 3 days of growth. However, the mutants ΔfapC, ΔrpoN, and ΔbrfA grew as turbid cultures and failed to form robust pellicles, especially the rpoN mutant (Figure 4A). Thus, the three genes are required for pellicle formation by P. fluorescens PF07.

The ability of the bacteria to form SSA biofilms in plastic microplates was analyzed by crystal violet staining. As shown in Figure 4B, after 6 h of growth, ΔrpoN showed a slight defect in the initial adhesion. After 12 h, the biofilm biomass of ΔrpoN and ΔbrfA was significantly reduced compared with the wild type. After 24 and 48 h, the three mutants produced significantly less biofilm than the wild type, by at least 50%. To determine whether the three mutants had growth defects, their growth curves were plotted. As shown in Figure 4C, all the strains started to enter the stationary phase after 12 h of growth. During the logarithmic growth phase (4–12 h), the growth of ΔrpoN was slower than the other strains. After 16 h, the optical density of the wild type was lower than the mutants. These results suggest that fapC, rpoN, and brfA genes are required for SSA biofilm maturation in P. fluorescens PF07. However, FapC and BrfA may not be involved in the initial adhesion process of biofilm formation.

The biofilms formed on glass coverslips were stained with both Congo red and DAPI to show the structure of the biofilm by CLSM. Congo red is frequently used for amyloid and polysaccharide staining in biofilm matrices, and the DNA stain DAPI (blue) is used to visualize the bacterial cells (Larsen et al., 2007). As shown in Figure 5, PF07 formed thick and rough biofilms covering most of the surface after 24 h of incubation. The cells with blue fluorescence were wrapped and covered by extracellular matrices with red fluorescence. Compared to PF07, ΔfapC, ΔrpoN, and ΔbrfA formed notably thinner biofilms that sparsely covered the surface, and the red fluorescent signal was significantly reduced. The CLSM results are consistent with the results of the crystal violet assay in plastic microplates. These results further confirm that Fap amyloid is the main component of the SSA biofilm matrices, and both RpoN and BrfA regulate the production of biofilm matrices by controlling the transcription of the fap genes.