Bordetella bronchiseptica strain RB50, B. pseudohinzii 8-296-03, B. hinzii L60, B. petrii DSM12804, B. avium 197N and B. trematum H044680328 were grown and maintained on BG agar (Difco) supplemented with 10% defibrinated sheep’s blood (Hema Resources). Liquid cultures were grown overnight at 37°C to mid-log phase (OD ∼0.6) in Stainer Scholte (SS) liquid broth (Stainer and Scholte, 1970). Klebsiella aerogenes was grown and maintained on Luria-Bertani (LB) agar (Difco) and liquid cultures were grown at 37°C to mid-log phase in LB broth (Difco).

RAW 264.7 macrophages cells were grown to 80% confluency (∼1 × 10⁵ CFU/well) in Dulbecco’s Modified Eagle Media (DMEM) supplemented with 10% FBS, glucose and glutamine in 48-well tissue-culture plates at 37°C. Bacteria were added in 10 μl PBS containing 10⁷ CFU (MOI of 100), 10⁶ CFU (MOI of 10) or 10⁵ CFU (MOI of 1) as indicated. Plates were centrifuged at 250 g for 5 min at room temperature and incubated at 37°C for 1 h, after which gentamicin solution (Sigma-Aldrich) was added to a final concentration of 300 μg/ml. Plates were incubated at 37°C for an additional 1, 3, 7, or 23 h and subsequently washed with PBS. 0.1% Triton-X solution was administered, followed by 5 min incubation and vigorous pipetting to lyse the macrophages. The samples were serially diluted and plated on BG agar plates to quantify total bacteria numbers.

RAW 264.7 macrophages were seeded in 6-well tissue-culture plates at a density of 1.5 × 10⁵ cells/ml, inoculated with B. bronchiseptica RB50 at a MOI of 10:1 and centrifuged for 5 min at 250 g. Following 1 h incubation at 37°C, the macrophages were washed with PBS, and DMEM media containing 300 μg/ml gentamicin was added. After 1 h, the macrophages were washed with PBS and suspended in a final volume of 300 μl of PBS. The macrophages were then collected by centrifugation and fixed with fresh 2% glutaraldehyde for Transmission Electron Microscopy at the University of Georgia Electron Microscopy Core Facility.

Green fluorescent protein (GFP)-expressing B. bronchiseptica strain RB50 (Taylor-Mulneix et al., 2017a) was exposed to RAW 264.7 macrophages at a MOI of 100:1 for 1 h, followed by gentamycin treatment for 1 h. Live cell fluorescence microscopy was performed using a Zeiss Axio Obsever.Z1/7 microscope. Imaging was performed at 488 nm for GFP (green), and transmitted light for DIC II (white) at a magnification of 40x using an LD Plan-Neoflaur 40x/0.4 Korr M27 objective.

RAW 264.7 cells were seeded in 6-well tissue-culture-treated plates with coverslips in the bottom at a density of 1.5 × 10⁵ cells/ml in 3 ml and inoculated with B. bronchiseptica RB50 at a MOI of 10 (2 × 10⁶ CFU) 12 h later. To synchronize the bacterial exposure to macrophages the plates were centrifuged at 300 g for 10 min. After 45 min incubation at 37°C, bacteria in the supernatant were removed by washing the macrophages three times with 1X PBS. The plates were incubated with DMEM medium containing 300 μg/ml gentamicin for 2 h to kill the remaining extracellular bacteria, and then washed 3 times with 3 ml of 1X PBS. Cells were fixed in 4% paraformaldehyde for 10 min at room temperature. The cells were then washed three times with 1X PBS and subsequently permeabilized with 0.1% Triton X-100 in 1X PBS for 20 min at room temperature. Primary antibodies derived from sera of B. bronchiseptica-infected mice were added after dilution in 1X PBS containing 2% BSA. After incubation at room temperature for 1 h, the cells were washed 3 times in 1X PBS. Then, the preparation was incubated with secondary donkey anti-mouse antibodies conjugated to FITC and with phalloidin for actin staining for 1 h. After 3 washes in 1X PBS, the coverslips were removed from the plates and fixed on glass slides with mounting medium containing DAPI. The images were taken with a Zeiss LSM 710 Confocal Laser Microscope for Z-stack imaging at 0.5 μm intervals.

RAW 264.7 macrophages were seeded in 6-well tissue-culture plates and inoculated with B. bronchiseptica RB50 at a MOI of 100:1. A subset of the bacteria was cultured in DMEM medium without macrophages as the negative control. Following 1 h incubation at 37°C, the remaining bacteria in the supernatant were removed by washing the macrophages with PBS and followed by addition of DMEM medium containing 300 μg/ml gentamicin. After 1 h the DMEM was removed, and the macrophages were washed with PBS. The samples were suspended in 1 ml of TRIzol for RNA extraction.

RNA was extracted from RB50 lysates using TRIzol (Ambion) and the Bacterial RNA isolation Kit (Max Bacterial Enhancement Reagent, Ambion) with implemented PureLink DNase treatment (Invitrogen) following the manufacturer’s instructions. RNA quality was assessed using the NanoDrop 2000 (Thermo Scientific) and BioAnalyzer (Agilent). Samples were submitted for Illumina sequencing at the Molecular Research Laboratory in Shallowater, TX, United States. Ribosomal RNA was depleted from each biological replicate (n = 3) during preparation of the Illumina sequencing library.

Quality control of raw reads was performed using FASTQC and TRIMMOMATIC for filtering of low quality reads and trimming of Illumina library adapters. Filtered reads were mapped to Bordetella bronchiseptica RB50 genome assembly NC_002927.3 using “Bowtie2.” The resulting output files were used to evaluate differential gene expression between three biological replicates of intracellular B. bronchiseptica (n = 3) and controls (n = 3) using the “EdgeR” package for the statistical environment R distributed within the Bioconductor project.

Total protein sequences were extracted from the NCBI archive for: B. bronchiseptica RB50 (RefSeq assembly accession: GCF_000195675.1), B. parapertussis 12822 (GCF_000195695.1), B. pertussis Tohama I (CF_000195715.1), B. hinzii L60 (GCF_000657715.1), B. pseudohinzii 8-296-03 (GCF_000657795.2), B. avium 197N (GCF_000070465.1), B. petrii DSM12804 (GCF_000067205.1), and B. trematum H044680328 (GCF_900078695.1). Similarities between B. bronchiseptica proteins and proteins of the non-classical species were calculated in mGenomeSubtractor (Shao et al., 2010) as the H value for each protein, defined as H = i x (l_m/l_q). H is the highest BLASTp identity score (i), multiplied by the ratio of the matching sequence length (l_m) and the query length (l_q). Based on our previous work (Linz et al., 2018), proteins with an H value < 0.5 were considered absent. Pairwise tBLASTx genome comparisons in the Artemis Comparison Tool (Carver et al., 2008) validated proteins with values of H > 0.5 as true orthologs.

Real-time PCR analyses were performed on a QuantStudio (Applied Biosystems) using Power SYBR Green PCR Master Mix (Applied Biosystems). Complementary DNA (cDNA) transcript libraries were prepared from biological triplicates of the control and of bacteria incubated with macrophages in DMEM + 10% FBS. Samples were processed for RNA extraction using TRIzol Reagent (Ambion by Life Technologies) and treated with PureLink DNase (Invitrogen). Primers were manually designed and purchased from IDT (Supplementary Table S1). The cycling parameters were as follows: 5-min preincubation at 95°C followed by 40 cycles of a 2-step PCR at 95°C and 60°C. Gene expression was calculated using the ΔΔCt method with expression of the 16S rRNA used as reference. Data were analyzed using DataAssist version 3.0 (Applied Biosystems).

The allelic exchange vector pSS4245 (Inatsuka et al., 2010) was used for the generation of deletion mutants. Briefly, ∼1 kb of DNA flanking each end of the target gene was PCR amplified using primers provided in Supplementary Table S2, joined and inserted into the vector by PIPE cloning (Klock and Lesley, 2009). The construct was verified by sequencing, transformed into E. coli SM10λpir, and transferred into the parental B. bronchiseptica strain RB50 by mating. Colonies containing the integrated plasmid were selected and incubated on BG agar to stimulate allelic exchange by homologous recombination. Emerging colonies were screened by PCR for replacement of the wildtype by the mutant allele and confirmed by Sanger sequencing. In vitro growth curves showed that none of the deletion mutants had growth defects compared to the wildtype strain RB50 (data not shown). For complementation, the target gene was cloned into plasmid pBBR1 (Antoine and Locht, 1992).

The mean ± standard error (error bars in figures) was determined for all appropriate data. Two-tailed, unpaired student’s t-tests were used to determine the statistical significance between two normally distributed populations. GraphPad Prism version 6.04 was used to conduct these statistical tests and to generate figures.

We had earlier observed using gentamicin protection assays that the prototype B. bronchiseptica strain RB50 (Bb) can enter and survive within murine macrophage-like cell line RAW 264.7 in vitro (Bendor et al., 2015). To determine the number and proportion of bacteria entering these macrophages, we performed an assay of macrophages infected with Bb at multiplicities of infection (MOI) of 100, 10 and 1 for 1 h. The percentage of recovery ranged from 0.7 to 1% of the original inoculum at all three MOIs, indicating that a relatively constant fraction of bacteria entered and resisted digestion by macrophages (Figure 1A and Supplementary Figure S2). The observation that the ratio of bacteria to macrophage did not affect survival rate suggested that this is not simply macrophages being overwhelmed or overcome by bacterial numbers. Electron microscopy (Figure 1C), confocal microscopy (Supplementary Figure S3A) and z-stack images (Figure 1B and Supplementary Figure S3B) taken after 2 h incubation confirmed the presence of bacteria within phagocytic vacuoles. Once inside the RAW 264.7 cells, bacterial numbers remained relatively stable and decreased only slowly over time. Bacterial CFUs recovered at 4 and 8 h showed no significant change in numbers for any of the MOIs used, and even at 24 and 48 h intracellular Bb were recovered in substantial numbers (Figure 1A). In contrast to Bb, Klebsiella aerogenes (Figure 1A) failed to persist in RAW 264.7 cells and was recovered at numbers over two orders of magnitude lower.

We hypothesized that to survive within professional phagocytic cells, Bb would require distinct groups of genes to be transcriptionally modulated once the bacteria reached the intracellular niche. To examine this transcriptional response, we analyzed the RNA profile of intracellular Bb at 2 h post inoculation and compared it to that of bacteria grown in vitro. Total RNA was isolated from samples collected after antibiotic treatment and sequenced on an Illumina MiSeq (RNA-Seq). On average, 8.8 × 10⁵ reads of intracellular Bb (n = 3) and 5.6 × 10⁶ reads of the planktonic bacterial control (n = 3) mapped to non-rRNA regions of the Bb reference genome (NC_002927.3). Those reads were used to evaluate the differential gene expression of Bb inside macrophages in comparison to that of bacteria grown in vitro.

A Principal Component Analysis (PCA) of the normalized read distribution revealed a clear difference in the global gene expression between intracellular and in vitro grown Bb (Supplementary Figure S4). The PCA plot showed clustering of the replicates and separation of the two groups along the first principal component (PC1), indicating a distinct transcriptional response to the intracellular environment. Differentially expressed genes that displayed a log2-fold change of either ≥ 1.5 (upregulated genes) or ≤−1.5 (downregulated genes) with a p-value < 0.05 were selected for further analysis, which resulted in a list of 318 upregulated and 243 downregulated genes. To validate our RNA-seq dataset we performed a quantitative real-time (qRT) PCR to assess the transcriptional changes in five highly upregulated and four strongly downregulated genes (Supplementary Figure S5), which confirmed the RNA-Seq data.

Since the human-restricted pathogens B. pertussis and B. parapertussis arose from B. bronchiseptica-like ancestors and can persist inside human macrophages, we evaluated whether the 318 upregulated genes were conserved among the three classical Bordetella species. Two hundred and seventy two intact genes (86%) were identified in the genome of B. pertussis strain Tohama I, the other genes were missing or truncated by frameshifts or premature stop codons. Similarly, 301 of the 318 genes (95%) were present in the genome of B. parapertussis strain 12822, showing conservation of most genes (Figures 2A,C).

While many non-classical bordetellae are also human and animal pathogens, their ability to persist inside phagocytic cells has not been evaluated. Therefore, we tested the presence or absence of the upregulated genes among the non-classical Bordetella species, including the bird pathogens B. hinzii (Vandamme et al., 1995) and B. avium (Kersters et al., 1984), the mouse pathogen B. pseudohinzii (Ivanov et al., 2016), the human opportunistic pathogen B. trematum (Vandamme et al., 1996), and the environmental species B. petrii (von Wintzingerode et al., 2001). We calculated the protein similarity (H value) of the 318 genes upregulated in B. bronchiseptica inside macrophages and their corresponding homologs in the non-classical bordetellae, with a gene considered to be present with a protein similarity value of H ≥ 0.5. An average of 77–81% of the 318 upregulated genes were present in the non-classical species (Figures 2A,C) with 95 (30%) of the genes displaying similarity values of H ≥ 0.9 (Figure 2E). In contrast, only 46–55% of the total of 4,981 evaluated B. bronchiseptica genes were identified in the genomes of the non-classical species (P < 0.0001), where only 448 (9%) of the genes reached protein similarity scores of H ≥ 0.9 (Figures 2B,D,E).

This high evolutionary conservation of genes that are upregulated in B. bronchiseptica during intracellular survival in phagocytic cells suggests that the non-classical bordetellae may be able to persist inside macrophages. To test this hypothesis, the non-classical species were assessed for intracellular survival in RAW 264.7 macrophages for 2 and 4 h. All examined Bordetella species were recovered, with the exception of B. avium (Figure 3). The inoculated B. pseudohinzii, B. hinzii, B. trematum and B. petrii bacteria survived internalization by macrophages at similar rates to B. bronchiseptica. The genomes of these species share 222 out of the 318 (70%) transcriptionally upregulated genes, which implies a critical function for intracellular persistence in mammalian phagocytic cells.

In contrast to the other analyzed Bordetella species, B. avium was severely impaired in its ability to persist inside macrophages. Only 0.001% of the inoculum was recovered after 2 h and no viable bacteria were detected after 4 h. Therefore, we performed a comparative genome analysis to identify transcriptionally upregulated genes that were only missing in B. avium, which resulted in the identification of six genes (Table 3). Deletion of two of these genes (BB0096 and BB1908) resulted in a significant reduction in intracellular survival (Figure 4 and Supplementary Figure S6). Complementation of these knock-out mutants with plasmid-borne gene copies restored the wildtype phenotype in both mutants (Figure 4), confirming that loss of malate synthase transcriptional regulator glcC (BB0096) or the tripartite tricarboxylate transporter BB1908 negatively impacts intracellular persistence in macrophages. In addition, previous studies showed important roles of transcriptionally upregulated (Table 1) risA and hfq genes in intracellular persistence of Bb and B. pertussis (Zimna et al., 2001; Bibova et al., 2013). We also assessed Bb knock-out mutants of several other transcriptionally upregulated genes (Supplementary Table S3), however, intracellular survival of none of the tested mutants was significantly different from the RB50 wildtype strain.