RpoN (sigma factor 54) contributes to bacterial fitness during tracheal colonization of Bordetella bronchiseptica

The Gram‐negative pathogenic bacterium Bordetella bronchiseptica is a respiratory pathogen closely related to Bordetella pertussis, the causative agent of whooping cough. Despite sharing homologous virulence factors, B. bronchiseptica infects a broad range of mammalian hosts, including some experimental animals, whereas B. pertussis is strictly adapted to humans. Therefore, B. bronchiseptica is often used as a representative model to explore the pathogenicity of Bordetella in infection experiments with laboratory animals. Although Bordetella virulence factors, including toxins and adhesins have been studied well, our recent study implied that unknown virulence factors are involved in tracheal colonization and infection. Here, we investigated bacterial genes contributing to tracheal colonization by high‐throughput transposon sequencing (Tn‐seq). After the screening, we picked up 151 candidate genes of various functions and found that a rpoN‐deficient mutant strain was defective in tracheal colonization when co‐inoculated with the wild‐type strain. rpoN encodes σ54, a sigma factor that regulates the transcription of various genes, implying its contribution to various bacterial activities. In fact, we found RpoN of B. bronchiseptica is involved in bacterial motility and initial biofilm formation. From these results, we propose that RpoN supports bacterial colonization by regulating various bacteriological functions.


INTRODUCTION
Bordetella pertussis, B. parapertussis, and B. bronchiseptica, which are closely related to each other, cause respiratory infections in mammals, including humans.These bacteria share the Bordetella virulence gene (Bvg) system, which is activated at 37°C and leads the bacteria to the virulent Bvg + phase, which produces virulence factors.At ambient temperatures (lower than 26°C) or in the presence of modulators, such as nicotinic acid and MgSO 4 , the Bvg system is inactivated and converts the bacteria into the avirulent Bvg -phase, which represses the expression of the virulence factors.For decades, previous studies have revealed that a number of homologous Bordetella virulence factors, including toxins (adenylate cyclase toxin and dermonecrotic toxin), adhesins (filamentous hemagglutinin and fimbriae), effectors of the type III secretion system, and autotransporters (pertactin and Vag8), are regulated by the Bvg system. 1,2Therefore, the pathogenicity of the Bordetella species is considered to be exerted in a similar manner.
However, the detailed mechanism for the Bordetella species to establish infection is not fully understood.For example, we recently reported that B. bronchiseptica mutant strains deficient in adenylate cyclase toxin, dermonecrotic toxin, or the type III secretion system colonized rat tracheas and lungs similar to a wild-type strain, 3 implying the existence of unknown additional virulence factors or redundant roles of known virulence factors.Several studies also pointed out that the bacteria express Bvg -phase-dependent factors in host cells and animals, [4][5][6] but the roles of these factors in the infection process are not elaborated.Therefore, in this study, to seek novel factors involved in the Bordetella pathogenicity, we conducted a genome-wide screening for genes that support colonization/infection of B. bronchiseptica using a transposon sequencing technique (Tn-seq), which combines large-scale transposon mutagenesis with next-generation sequencing. 7Since B. pertussis is strictly a human pathogen, we infected rats with B. bronchiseptica as a model of the natural infection process because the rat is a natural host for this organism.Consequently, 151 genes of various functions were picked up from the screening.Among them, rpoN, which encodes sigma factor σ 54 (RpoN) was found to contribute to the bacterial fitness for tracheal colonization.We further characterized the function of RpoN in B. bronchiseptica to explore the possible roles of RpoN in tracheal colonization/infection of the bacteria.

Bacterial strains and plasmids
The bacterial strains, plasmids, and primers used in this study are listed in Supporting Information: Tables S1  and S2.B. bronchiseptica RB50 was provided by P.A. Cotter (University of North Carolina).The B. bronchiseptica strains were cultured at 37°C with Stainer-Scholte (SS) medium or Bordet-Gengou (BG) agar plates (Becton Dickinson) containing 20% defibrinated horse blood, 1% hipolypeptone (Nihon Pharmaceutical), 1% glycerol, and 10 µg/mL ceftibuten.Bacteria grown on BG plates were suspended in SS medium to set an initial OD 650 value of 0.02 and incubated at 37°C with shaking for 15-16 h.The amount of B. bronchiseptica was determined by OD 650 values according to the following formula: 1 OD 650 = 3.3 × 10 9 CFU/mL.Escherichia coli strains were grown at 37°C on LB agar plates or LB broth.Antibiotics were added at the following concentrations when necessary: gentamicin (Gm), 20 µg/mL; ceftibuten, 10 µg/mL; ampicillin, 50 µg/mL; kanamycin, 50 µg/mL; and tetracycline (Tet), 10 µg/mL.Bacterial strains with gene deletion were constructed as described previously with slight modifications. 8For example, a rpoN-deficient mutant (ΔrpoN) was constructed as follows.The fragments of the 5′-and 3′-flanking regions of the rpoN gene (named BB-rpoN-U and BB-rpoN-D, respectively) were amplified by PCR and ligated to each other by overlap PCR.The ligated fragment was inserted into the SmaI site 14 bp upstream of the lac operator in a suicide plasmid, pABB-CRS2-Gm, 6 using an In-Fusion HD Cloning Kit (Clontech).The resultant plasmid, pABB-CRS2-Gm-ΔrpoN, was introduced into B. bronchiseptica RB50 through biparental conjugation using E. coli S17-1 λpir.For the biparental conjugation, RB50 and E.coli S17-1 λpir were mixed at a ratio of 4:1.After conjugation, host RB50 was cultivated in the presence of 10% sucrose for counter selection, and RB50 ΔrpoN mutants generated via two-step homologous recombination were recovered from the grown colonies.The Bvg + phaselocked and Bvg − phase-locked mutants of RB50 were constructed as previously described. 8Strains with antibioticresistant genes were constructed through the broad-host-range mini-Tn7 system as described previously. 9A plasmid, pUC18T-mini-Tn7T-Tet, was generated by replacing the Gm-resistant gene (aacC1, hereafter called Gm r ) of pUC18Tmini-Tn7T-Gm with the Tet-resistant gene (tet, hereafter called Tet r ) of pBBR1MCS3. 10These plasmids were used to introduce Gm r or Tet r to the bacterial chromosome at the attTn7 site.

Animal experiments
Three-week-old female Wistar rats (Japan SLC) were anesthetized with isoflurane and intranasally inoculated with 10 µL of SS medium containing approximately 10 4 CFU of B. bronchiseptica RB50.Aliquots of the inoculum were serially diluted with Dulbecco's phosphatebuffered saline (PBS) and cultivated on BG plates for 2 days, and the number of colonies grown on the plates (CFU) was counted to confirm bacterial load.After 5 days of infection, rats were euthanized using pentobarbital sodium, and their tracheas were removed and homogenized in PBS with BioMasher I (Nippi, 49118-52).Tracheal homogenates were serially diluted with PBS and subjected to CFU counting.

Gene screening with the transposon library
A Himar1-based transposon vector, pMariG, 11 was introduced into B. bronchiseptica RB50 via biparental conjugation through E.coli S17-1 λpir.After a 6-h incubation, the bacteria in the mixture were collected and cultured on BG plates supplemented with Gm for 48 hr at 37°C.The bacteria were recovered from all colonies grown on the plates, resuspended in fresh SS medium, and cultured for 6 hr at 37°C with shaking.The resultant bacteria in the culture were diluted to give an OD 650 value of 1.0, and the diluted samples were divided into 1-mL aliquots and stored at −80°C until use as the transposon library.
For in vivo screening, 1-mL aliquots of the transposon library were thawed, washed with fresh SS medium, and diluted to 10 6 CFU/mL.Ten rats per group were intranasally infected with the diluted samples (10 µL/rat).The remaining inoculum was used as input samples.The screening procedures were conducted with three independent groups of rats.After a 5-day infection, tracheal homogenates were prepared and plated on BG plates as mentioned above.After a 2-day incubation at 37°C, bacteria were collected from all grown colonies, suspended in SS medium, and incubated at 37°C for 4 h with shaking.The cultured bacteria were used as the output samples.Genomic DNA from three input samples and three output samples was extracted using the DNeasy Blood and Tissue Kit (QIAGEN) following the manufacturer's instruction and subjected to Illumina sequencing as previously described with slight modifications. 12Six-micrograms of genomic DNA from the three input and three output samples was digested with MmeI (New England Biolabs) for 2.5 h.After a 1-h treatment with rAPid alkaline phosphatase (Roche), the genomic DNA was re-extracted using phenol:chloroform:isoamyl alcohol (25:24:1) (Nacalai Tesque) and precipitated by ethanol.The MmeI-digested fragments were ligated with adapters for Illumina sequencing and amplified with PCR using a pair of primers: Linker and a Tn-index (Supporting Information: Table S2), which are complementary to the adapter and transposon junctional region, respectively.After PCR, the amplified approximately 120 bp fragments, which contain an approximately 20-bp bacterial genome sequence with transposon junctional region, the adapter sequence, and a unique barcode (bold sequences in Supporting Information: Table S2) derived from the Tn-index to distinguish the three input and three output samples, were recovered, and purified using a QIAquick Gel Extraction Kit (QIAGEN, 28706).All PCR products were submitted to the Genome Information Research Center at the Research Institute for Microbial Diseases, Osaka University, for quality control analysis and sequencing using a NovaSeq 6000.The obtained sequence data was analyzed as follows.In brief, all reads obtained from the sequencing were separated based on the unique barcode for the three input and three output samples and trimmed to remove all nongenomic sequences using cutadapt.Trimmed reads were mapped to the sequence of the reference genome (Accession number: NC_002923.7)using Bowtie2, and the transposon insertion sites were identified.Finally, the number of each insertion site over the whole genome was counted.The counts of each insertion site were compared between the input and output samples using the resampling test, and the conditional essentiality of genes was determined.The resampling test of the TRANSIT suite, cutadapt, and Bowtie2 are available on the Galaxy Project's public server. 13

Competition assay
For the in vivo competition assay, three-week-old rats were intranasally inoculated with 10 4 CFU of the bacterial mixture in 10 µL SS medium, which contained wild-type (WT) carrying Gm r and gene-deficient mutants (Mut) carrying Tet r in equal amounts.After a 5-day infection, tracheal homogenates were prepared and plated on BG plates containing ceftibuten plus Gm or ceftibuten plus Tet, and the CFUs of WT::Gm r and Mut::Tet r were determined as mentioned above.If necessary, the experiment was repeated using WT::Tet r and Mut::Gm r .
For the in vitro competition assay, overnight cultures of WT::Gm r , WT::Tet r , ΔrpoN::Gm r , and ΔrpoN::Tet r were diluted to an OD 650 value of 0.2.Equal amounts of WT and ΔrpoN carrying the different antibiotic-resistant genes were mixed and cultured in SS medium at 37°C for 15 and 24 h.Aliquots of the bacterial culture were taken at the indicated times, serially diluted, and cultivated on BG plates containing proper antibiotics to determine CFU. The

Other analysis methods
The bacterial motility assay was performed as described previously with slight modification. 14Overnight cultures of B. bronchiseptica in SS medium were diluted to give an OD 650 value of 0.8, and 2 μL of the dilution was stabbed into SS soft agar plates [0.4% (w/v) agar] that were prepared immediately before use.After a 48-h incubation at 37°C, the diameters of the grown colonies with spreading areas were measured at the indicated times.For the western blotting of flagellin, WT, ΔrpoN, and ΔflaA were recovered from the soft agar plates and from 48-h static culture with SS medium at 37°C.The recovered bacteria were diluted to an OD 650 value of 1.0 in 50 µL of PBS, mixed with 10 µL of sixfold concentrated SDS sample loading buffer, boiled for 5 min, and subjected to SDS-PAGE followed by electrotransfer to a polyvinylidene difluoride membrane (Millipore Billerica).The western blotting was performed with rabbit anti-FtsZ 15 (1:5000) or mouse anti-FliC serum prepared by immunization with recombinant FliC prepared previously 16 (1:5000) as the first antibody and goat anti-rabbit IgG-HRP (Jackson ImmunoResearch, 111-035-144) or goat anti-mouse IgG-HRP (Jackson ImmunoResearch, 115-035-003) as the second antibody.The target proteins were visualized by enhanced chemiluminescence using an Immobilon Western (Merck Millipore, WBKLS0500) and LAS-4000 mini Luminescent Image Analyzer (GE Healthcare).
Real-time PCR to determine the relative expression of target mRNAs was carried out as follows.Total RNA was extracted using NucleoSpin RNA (TaKaRa) from test bacteria cultured with and without shaking in Bvg + phase and Bvg − phase conditions and treated with Recombinant DNase I RNase-free (TaKaRa).The resultant DNA-free RNA was reverse-transcribed into complementary DNA using a PrimeScript RT Reagent Kit (TaKaRa), followed by the real-time PCR with Fast SYBR Green Master Mix (Thermo Fisher Scientific) and the appropriate primers using the StepOnePlus Real-Time PCR System (Applied Biosystems).The expression level of target mRNAs was determined by the ΔΔCt method using recA as the reference gene.
The level of biofilm formation was evaluated as described previously with slight modifications. 15,17In brief, test bacteria were suspended in SS medium at an OD 650 value of 0.05.Then, 100 µL of the suspensions were transferred into one well of a 96-well polyvinylchloride (PVC) microtiter plate, and 500 µL of the suspensions were transferred into a 5-mL polystyrene tube.After incubation at 37°C for the indicated times, the media and planktonic cells were discarded, and loosely adherent cells were removed by three washes with tap water.The tubes and 96-well plates were air-dried, and biofilm formed on the plastic was stained with 0.1% w/v crystal violet for 30 min at room temperature.After thorough washings with tap water, the tubes were air-dried and photographed, or the remaining stain in the well of the 96-well plate was extracted in 200 µL of 33% acetic acid and quantitated by its OD 600 value.
RpoN-binding sites on the B. bronchiseptica RB50 genome were predicted using the position-specific scoring matrices (PSSMs) that were generated from the 186 characterized RpoN-dependent −24/−12 promoter sequences from 44 different bacterial species. 18The promoter regions of all genes on the B. bronchiseptica RB50 genome (300 bp upstream of the start codon) were scanned by the matrix-scan program of the Regulatory Sequence Analysis Tools (RSAT) with the PSSMs of the RpoN-binding sites (Markov order: 1, weight score ≥ 1, p ≤ 0.001). 19

Accession number
The raw sequence reads of this study are available in NCBI's short read archive (SRA) under BioProject number PRJNA973810.

Screening for genes contributing to bacterial colonization of rat tracheas
The transposon library generated in this study was composed of approximately 80,000 clones of transposoninserted mutants, which is 16 times the total number of genes of B. bronchiseptica RB50 (5073 genes, according to NCBI: txid257310).The transposon library was divided into independent pools of 1 mL with an OD 650 value of 1.0.Using three randomly selected pools, we mapped the transposon-inserted sites on the bacterial genome by highthroughput sequencing and estimated the levels of saturation to be 77.9%,66.4%, and 84.4% of the possible insertion sites, respectively (Table 1 and Supporting Information: Figure S2A).In total, the transposons were found to be inserted into 99.1% of genes on the genome of B. bronchiseptica RB50.We then determined the proper inoculum dose to avoid population contraction of the inoculated bacteria.When the bacteria were intranasally inoculated at approximately 10 4 CFU/rat and 10 7 CFU/rat, similar numbers (approximately 10 5 CFU) of the bacteria were recovered from the tracheas 5 days postinoculation (Supporting Information: Figure S1).Therefore, we considered 10 4 CFU sufficient to establish tracheal colonization and bacterial expansion.For each screening, 10 rats were intranasally inoculated with the transposon library, the bacteria were recovered from rat tracheas after 5 days, and transposon-inserted sites on the genome were localized.After three independent screenings with three independent pools of the transposon library, we picked up 151 genes that exhibited a log 2 fold change (log 2 FC) ≤ −3 in the sequence counts in the output (postinfection) samples relative to the input (preinfection) samples (Supporting Information: Table S3).When categorized based on the database of Clusters of Orthologous Genes (COGs), about onethird of the selected genes were presumed to be involved in metabolism, including energy production and conversion.Twenty-four genes were related to transcription or translation.Eight genes were involved in cell wall biogenesis.Other genes were related to posttranslational modification, DNA replication and repair, transportation, motility, cell cycle control, signal transduction mechanisms, and so on (Supporting Information: Figure S2B).No genes related to known virulence factors, such as fimbriae (fim), filamentous hemagglutinin (fhaB), dermonecrotic toxin (dnt), and adenylate cyclase toxin (cyaA), were included in the candidates after our screening.The transposon library we used for Tn-seq consists of various mutant clones.Therefore, the deficiency of a mutant clone might be rescued by another mutant in the library, particularly in the case of secreted factors like DNT and CyaA.Alternatively, or additionally, the protocol for infection with a larger inoculation size of bacteria and a shorter infection period may influence the obtained results because virulence factors may function at specific times of infection and may be functionally redundant as mentioned above.Additionally, BvgA and BvgS, which compose the Bvg system to regulate the expression of major virulence factors, showed a low fold-change after screening (bvgS: 0.7825; bvgA: 0.6828).Twenty genes were characterized as Bvg-activated genes, including the gene encoding Bordetella autotransporter B (BatB), which is required to resist inflammatory clearance during infection. 20Most of the Bvg-activated genes were hypothetical proteins, including a putative toxin BB3242, which shares homology with a binding domain of aerolysin and pertussis toxin.3][24][25] The other 90 genes are reported to be independent of the Bvg system. 21Of the 151 candidates, only 13 genes are specific to B. bronchiseptica, and 117 and 127 genes are found in B. pertussis (Tohama strain) and B. parapertussis (12822 strain), respectively (Supporting Information: Table S3 and Figure S2C).One hundred and six genes are conserved in the three species.

RpoN contributes to bacterial fitness in rats
Among the 151 genes, we focused on three genes: rpoN (σ 54 ), BB1983 (an orthologue of the response regulator belonging to the BasR/S two-component system of E.coli), and BB3242 (a putative toxin), and evaluated their contributions to B. bronchiseptica RB50 infection, because the first two are involved in the global regulation of gene expressions and the third encodes a protein showing 32% and 29% identity to binding regions of pertussis toxin and aerolysin, respectively.In the single infection experiments, we did not find statistically significant differences in the colonization abilities of ΔrpoN, ΔBB1983, and ΔBB3242 (Figure 1a), although ΔrpoN was not recovered from three of five tested rats.In principle, the Tn-seq that we adopted is designed to screen for genes that contribute to bacterial fitness in the host environments because varied Tn mutants were simultaneously co-inoculated for the experiments.Therefore, we further evaluated the involvement of rpoN, BB1983, and BB3242 in bacterial colonization using the competitive infection assay.As shown in Figure 1b,c, ΔrpoN::Tet r was totally outcompeted by WT::Gm r after a 5-day co-infection.Similar results were obtained in the coinfection experiments using ΔrpoN::Gm r and WT::Tet r , excluding the possibility that the drug-resistant genes affected the fitness of the bacteria to colonize (Figure 1d,e).The competitive advantage in the co-infection of WT and ΔBB1983 or ΔBB3242 was not apparent.Therefore, we considered that RpoN may contribute to fitness during colonization in rats.To examine whether ΔrpoN was outcompeted by WT due to growth retardation, we assessed the growth rate of ΔrpoN and conducted an in vitro competition assay against WT.As shown in Figure 1f,g, all tested strains similarly grew, and no competition was observed among them.

RpoN is independent of the Bvg system
To understand the relationship between rpoN and the Bvg system, we examined whether rpoN expression is dependent on the Bvg phase.The expression level of rpoN was not markedly different among WT and the Bvg − and Bvg + phase-locked mutants, unlike the expression level of Bvg - specific flaA and Bvg + specific fhaB (Figure 2a).We next compared the expression patterns of Bvg + specific genes (fhaB, cyaA, dnt, and bscN) and Bvg -specific genes (flaA and brtA 15 ) between WT and ΔrpoN in the absence or presence of MgSO 4 (Bvg + phase-or Bvg − phaseoriented conditions).The Bvg phase-dependency of the expression of these genes was unchanged between WT and ΔrpoN (Figure 2b).These results indicate that the expression of rpoN is not regulated by the Bvg system, and the regulation of gene expression by the Bvg system is unlikely mediated by RpoN.

Motility
Because RpoN has been reported to regulate motility and flagella expression in several bacterial species, 26,[32][33][34] we investigated whether this is the case in B. bronchiseptica, which is known to be motile in the Bvg -phase. 35When cultivated in the presence of MgSO 4 , which leads the bacteria to the Bvg -phase, WT, and ΔrpoN exhibited motility; however, ΔrpoN was less motile than WT (Figure 3a right and 3b right).In the absence of MgSO 4 , WT was nonmotile within 24 h of incubation but displayed obvious motility at the extended incubation periods, as reported previously, 35 while ΔrpoN did not show such motility (Figure 3a left and 3b left).These results indicate that RpoN positively regulates the motility of B. bronchiseptica in Bvg + phase and Bvg − phase conditions.As shown in Figure 2b, the expression of flaA was positively regulated by RpoN in the Bvg + phase condition (WT vs. ΔrpoN) but not the Bvg − phase condition (WT + MgSO 4 vs.ΔrpoN + MgSO 4 ).Therefore, we further evaluated the expression of flagellin (FliC), which is encoded by flaA, in bacteria growing on soft agar or in SS medium without shaking after a 48-h incubation.Real-time PCR and western blotting confirmed the expression of flagellin was much less in ΔrpoN compared to WT (Figure 3c,d).These results demonstrate that flagellin is expressed during static incubation even in the Bvg + phase condition, and RpoN positively regulates bacterial motility, probably by inhibiting the expression of flagellin in the Bvg + phase condition.However, why ΔrpoN was less motile in the Bvg -phase condition regardless of the comparable flaA expression remains unknown (Figures 2b, and (d and e) In vivo competitive assays for ΔrpoN against WT.Rats were intranasally inoculated with a 1:1 mixture of WT::Tet r and ΔrpoN::Gm r .Bacteria were collected from tracheas after 5 days of infection, the bacterial load was determined, and the CI values were calculated as described in Section 2. Horizontal bars represent geometric means ± SD (d) and medians ± 95% confidence intervals (e) (n = 5).(f) Growth curves of WT, WT::Gm r , WT::Tet r , ΔrpoN::Gm r , and ΔrpoN::Tet r .All strains were precultured for 15 h to reach the mid-log phase.After adjusting the initial OD 650 to 0.2, the suspension of each strain was cultured with shaking at 37°C.The OD 650 of each culture was measured at the indicated times.Horizontal bars represent means (n = 3).(g) In vitro competitive assays for ΔrpoN against WT.A 1:1 mixture of WT and ΔrpoN was incubated for the indicated times, and the number of each strain was enumerated as described in Section 2. Horizontal bars represent medians ±95% confidence intervals (n = 3).The dashed lines indicate the detection limit (b and d) and the reference CI value of 1.0 (c, e, and g).Data were statistically analyzed by an unpaired t test (a), paired t test (b, d), nonparametric Mann-Whitney U test (compared to the hypothetical CI of 1) (c, e, and g), and two-way ANOVA with Šídák's multiple-comparison test (f).*p < 0.05; **p < 0.01; ns, no significant (p > 0.05).
through the control of flagellin expression, we carried out an in vivo competition assay with WT and ΔflaA; unexpectedly, these bacterial strains equally colonized trachea 5 days after co-inoculation (Figure 3e,f), indicating that the motility regulated by RpoN unlikely contributes to bacterial fitness in tracheal colonization.

Biofilm formation
B. bronchiseptica establishes persistent infection through biofilm formation, 36 a phenomenon that is regulated by RpoN in many bacterial species. 27,34,37Therefore, we examined the involvement of RpoN in the biofilm formation of B. bronchiseptica.The biofilm formation of ΔrpoN was poorer than of WT after 8 hr of incubation but similar after 24 h (Figure 4a,b).These results indicate that RpoN is involved in the early stage of biofilm formation.The deletion of flaA did not affect the biofilm formation, suggesting that RpoN is involved in the early biofilm formation in a flagellin/motility-independent manner.

Prediction of RpoN-regulated genes of B. bronchiseptica
In various bacterial species, RpoN initiates the transcription of target genes by interacting with promoter sequences located −24 bp and −12 bp upstream of start codons, 38 where the consensus sequence GGN 10 GC exists. 18Therefore, we screened possible RpoN-binding sites over the B. bronchiseptica RB50 genome using the matrix-scan program of RSAT with the weight matrix generated from 186 RpoNbinding sites of varied bacterial species. 18,19We found 173 genes of B. bronchiseptica that were presumably regulated by RpoN (Supporting Information: Table S4).Among them, two nitrogen utilization-related genes, glnA and BB2035 (which showed 70% identity to glnK of E. coli), 31,39 whose expressions are known to be regulated by RpoN in other bacteria, are included.Predicted target genes were divided into 19 categories according to the functional classification based on the database of COGs (Supporting Information: Table S4 and Figure S3).1][42][43][44] Five genes (BB1818, dxs, BB2035, ccoS, and BB4145) were also picked up by the Tn-seq screening.

DISCUSSION
Previously, we developed the in vivo expressed-tag immunoprecipitation (IVET-IP) technique to explore the gene expression profile of B. bronchiseptica colonizing rat tracheas. 6IVET-IP identified 173 bacterial genes that were persistently expressed in rats up to 30 days post-inoculation (the number of genes identified by IVET-IP equals the number of predicted RpoN-regulated genes in our study; however, most of the identified genes are different).Among these genes, which we named in vivo-induced (ivi) genes, we subsequently characterized a Bvg --specific gene, brtA, which encodes a Ca 2+ -dependent adhesin involved in biofilm formation. 15These results proved the utility of IVET-IP for a comprehensive analysis of the expression of genes involved in bacterial pathogenicity.However, IVET-IP includes an immunoprecipitation procedure, which is a bottleneck for the recovery of bacteria from host tissues, resulting in a lower detection limit.In addition, among ivi genes, only brtA has been characterized because we could not pick up other candidate genes.Thus, in this study, we sought B. bronchiseptica genes required for colonization/infection in rats using Tn-seq and obtained a list of candidates, which includes few ivi genes.Unlike IVET-IP, which identifies genes that are highly expressed during infection, Tn-seq identifies necessary genes for infection regardless of their expression levels in host animals.We, therefore, expected to find additional genes involved in bacterial infection/colonization by using Tn-seq.
Eventually, we implicated RpoN in bacterial fitness during the colonization of B. bronchiseptica in rat tracheas.RpoN, also known as sigma factor 54 (σ 54 ), initiates the transcription of genes associating with RNA polymerases. 45he functions of RpoN have been explored in other bacterial species, such as E. coli and Pseudomonas aeruginosa. 44,4628][29][30]44 In the present study, we first described Bordetella RpoN, which is implicated in bacterial fitness for tracheal colonization.RpoN of B. bronchiseptica RB50 strain shares 43% homology with E. coli RpoN with the conserved domain structure, implying functional similarity.Indeed, we confirmed the involvement of B. bronchiseptica RpoN in bacterial motility and biofilm formation, similar to that of other bacterial species.We also examined the involvement of RpoN in oxidative stress resistance by using ΔrpoN mutant strain; however, we could not obtain reproducible results, implying a complicated relationship between them.Although further analyses are required to understand the role of RpoN in bacterial colonization, our results suggest that early biofilm formation may contribute to the fitness for the bacteria to colonize.8][49] In contrast, our results demonstrated that RpoN regulated the gene expressions independently of the Bvg system, and its own expression of RpoN was not regulated by the Bvg system.RpoN requires an activator to energize the RNA polymerase-RpoN complex for transcription initiation. 38any of these activators (referred to as bacterial enhancerbinding proteins) are response regulators of the bacterial two-component systems. 38These response regulators transduce signals from sensor kinases, which sense alterations in the extracellular conditions.These findings suggest that RpoN regulates gene expressions in response to environmental alterations through two-component systems other than the Bvg system.
1][52] Consistently, we predicted 173 candidate genes, some of which were also found in other bacterial species, indicating the wide impact of RpoN on gene expressions in B. bronchiseptica.In particular, we note that BB2035 encoding an orthologue of E. coli glnK was among the candidate genes.E. coli glnK is a RpoN-regulated gene implicated in ammonium assimilation upon nitrogen limitation. 31,53Considering that RpoN was originally identified as a regulator of genes involved in nitrogen metabolism and assimilation, and host environments are under nitrogen limitation, it may be worth exploring whether RpoN is involved in bacterial fitness by compensating for the depletion of nitrogen sources within hosts.
Taken together, we identified RpoN of B. bronchiseptica as a factor conferring the fitness to colonize rat tracheas.Given that B. pertussis and B. parapertussis share RpoN with high homology (99.6% and 99.8% vs. B. pertussis and B. parapertussis, respectively) and RpoN regulates varied biological functions in various bacterial species, it is conceivable that RpoN has similar roles in the bacterial fitness of B. pertussis and B. parapertussis.RpoN of B. bronchiseptica was involved in early biofilm formation and motility.RpoN was also reported to regulate nitrogen metabolism. 31,53Furthermore, RpoN likely controls the expression of more than a hundred downstream genes.Previously, many comprehensive studies using genomewide screenings, including microarray, RNA-seq, IVET-IP, and Tn-seq techniques, have been performed to understand the pathogenicity of B. bronchiseptica and B. pertussis 4,6,[54][55][56] ; however, our understanding of influential genes and the network of gene expressions during the course of infection is still limited.We propose that RpoN coordinates the gene expressions to make bacteria adapt to the host environments.
3a right, and 3b right).To verify whether RpoN is involved in bacterial fitness F I G U R E 1 Tracheal colonization of ΔrpoN, ΔBB1983, and ΔBB3242.(a) Tracheal colonization of Bordetella bronchiseptica RB50 WT and mutants deficient of rpoN, BB1983, and BB3242.WT and mutants were intranasally inoculated into rats.The number of bacteria recovered from the tracheas after 5 days were enumerated as described in Section 2. Horizontal bars represent geometric means ± SD (n = 5).(b and c) In vivo competitive assays for mutants against WT.Rats were intranasally inoculated with a 1:1 mixture of WT::Gm r and ΔrpoN::Tet r , ΔBB1983::Tet r , or ΔBB3242::Tet r .The number of bacteria recovered from the tracheas after 5 days was enumerated, and the competition index (CI) values were calculated as described in Section 2. Horizontal bars represent geometric means ± SD (b) and medians ± 95% confidence interval (n = 5).

F I G U R E 2
RpoN is independent of the Bvg system.(a) Real-time PCR for rpoN, flaA, and fhaB mRNA.Total RNA was extracted from WT, Bvg + phase-, and Bvg − phase-locked mutants after a 15-h incubation in the Bvg + phase condition with shaking in Stainer-Scholte (SS) medium.Data represent fold changes (mean ± SD) in rpoN, flaA, and fhaB expression in each mutant against WT (n = 6).(b) fhaB, dnt, cyaA, bscN, flaA, and brtA expression in Bordetella bronchiseptica RB50 WT and mutants deficient of rpoN in the Bvg + phase and Bvg − phase conditions.Total RNA was extracted from the bacteria after a 15-h incubation with shaking in SS medium with or without 40 mM MgSO 4 .Data represent fold changes (mean ± SD) in target mRNAs expression against WT in the absence of MgSO 4 (n = 6).Data were statistically analyzed by one-way ANOVA with Tukey's multiple comparison (a) and t test (b).*p < 0.05, **p < 0.01 and ****p < 0.0001; ns, not significant (p > 0.05).

F
I G U R E 3 RpoN regulates motility and flagellar synthesis.(a, b) Motility assays for WT, ΔrpoN, ΔflaA, and the Bvg − phase-locked mutant.(a) Images of colonies grown on SS (Stainer-Scholte) soft agar plates.The Bvg -phase-locked mutant of Bordetella bronchiseptica, which is known to be constitutively motile, was used as a positive control.The mutant deficient of flaA (encoding flagellin), which is supposedly non-motile, was used as a negative control.(b) Numerical data representing means of colony diameters ± SD (n = 4).The dashed line indicates the initial diameter of the inoculum on SS soft agar plates.(c) Western blotting for flagellin (FliC) expression.WT, ΔrpoN, and ΔflaA were recovered from the edges of the colonies on SS soft agar plates (left panels) and the static culture in SS medium (right panels) after a 48-h incubation at 37°C (without MgSO 4 ) and subjected to SDS-PAGE followed by western blotting.FtsZ in the bacterial lysates was simultaneously detected as an internal control.The blot images are representatives of two independent experiments.(d) Real-time PCR for flaA mRNA.Total RNA was extracted from WT and ΔrpoN before and after 24-h and 48-h static incubations in SS medium.Data represent fold changes (mean ± SD) in flaA expression against WT at 0 h of the incubation (n = 6).(e, f) In vivo competitive assays for ΔflaA against WT.Rats were intranasally inoculated with 1:1 mixture of WT and ΔflaA carrying different antibiotic-resistant genes.The number of bacteria recovered from the tracheas after 5 days was enumerated, and the competition index (CI) values were calculated as described in Section 2. The dashed lines indicate the detection limit (e) and the reference CI value of 1.0 (f).Horizontal bars represent geometric means ± SD (e) and medians ± 95% confidence intervals (f) (n = 5).Data were statistically analyzed by an unpaired t test (b, d), paired t test (e), and nonparametric Mann-Whitney U test (compared to the hypothetical CI of 1) (f).*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; ns, not significant (p > 0.05).

F I G U R E 4
RpoN is involved in the early stage of biofilm formation.(a) Biofilm formation of WT, ΔrpoN, and ΔflaA in 5-mL polystyrene tubes.(b) Qualitative evaluation of biofilm formation in 96-well plates.Stained biofilm was solubilized with 33% acetic acid and quantitated by OD 600 values.Plotted data represent means ± SD (n = 4).Data were statistically analyzed by a one-way ANOVA with Šídák's multiple-comparison test.**p < 0.01; ns, not significant (p > 0.05).
Overview of Tn-seq results and mapping data.
T A B L E 1 RPON CONTRIBUTES TO BACTERIAL FITNESS IN VIVO