Triggering Receptor Expressed on Myeloid Cells-1 (TREM-1) Contributes to Bordetella pertussis Inflammatory Pathology

ABSTRACT

Whooping cough (pertussis) is a severe pulmonary infectious disease caused by the bacteria Bordetella pertussis. Pertussis infects an estimated 24 million people annually, resulting in >150,000 deaths. The NIH placed pertussis on the list of emerging pathogens in 2015. Antibiotics are ineffective unless administered before the onset of the disease characteristic cough. Therefore, there is an urgent need for novel pertussis therapeutics. We have shown that sphingosine-1-phosphate receptor (S1PR) agonists reduce pertussis inflammation without increasing bacterial burden. Transcriptomic studies were performed to identify this mechanism and allow for the development of pertussis therapeutics that specifically target problematic inflammation without sacrificing bacterial control. These data suggested a role for triggering receptor expressed on myeloid cells-1 (TREM-1). TREM-1 cell surface receptor functions as an amplifier of inflammatory responses. Expression of TREM-1 is increased in response to bacterial infection of mucosal surfaces. In mice, B. pertussis infection results in Toll-like receptor 9 (TLR9)-dependent increased expression of TREM-1 and its associated cytokines. Interestingly, S1PR agonists dampen pulmonary inflammation and TREM-1 expression. Mice challenged intranasally with B. pertussis and treated with ligand-dependent (LP17) and ligand-independent (GF9) TREM-1 inhibitors showed no differences in bacterial burden and significantly reduced tumor necrosis factor-α (TNF-α) and C-C motif chemokine ligand 2 (CCL-2) expression compared to controls. Mice receiving TREM-1 inhibitors showed reduced pulmonary inflammation compared to controls, indicating that TREM-1 promotes inflammatory pathology, but not bacterial control, during pertussis infection. This implicates TREM-1 as a potential therapeutic target for the treatment of pertussis.

INTRODUCTION

Bordetella pertussis is the etiologic agent of whooping cough. Despite widespread vaccination, pertussis is the only vaccine-preventable disease to have been consistently on the rise in the United States since 1976 (1). Antibiotics are ineffective unless administered early during disease (2, 3). Therefore, there is an urgent need for new therapeutics to treat B. pertussis (4). Previously, we demonstrated the potential of sphingosine-1-phosphate receptor-1 (S1PR) agonists to dramatically reduce inflammation in the lungs of B. pertussis-challenged mice (5, 6). Importantly, despite reducing inflammatory pathology and cytokine expression, S1PR agonists did not impair the ability of mice to control and clear infection (5). However, recent work has shown that chronic administration of S1PR agonists is associated with a reduced ability to clear infection (7). S1PR is present on a wide variety of cell types, regulating a multitude of biological responses (8–11). Understanding the mechanisms by which S1PR agonists protect against disease may inform the development of improved host-directed therapies that promote resolution of inflammation and bacterial clearance.

Transcriptomic studies designed to explore the mechanisms of action of S1PR agonists in reducing pertussis inflammation identified differential regulation of multiple genes and pathways. Previously, we demonstrated the ability of peptidoglycan recognition protein 4 (PGLYRP4), a host protein expressed following S1PR agonist treatment, to limit inflammatory responses to tracheal cytotoxin (peptidoglycan [PGN] fragment) following B. pertussis infection (12).

Here, we describe the increased expression of triggering receptor expressed on myeloid cells-1 (TREM-1) observed in response to B. pertussis infection. Further, we demonstrate the ability of S1PR agonists to decrease TREM-1 expression and the consequences of targeted TREM-1 inhibition during B. pertussis infection. TREM-1 is a member of the Ig superfamily expressed on monocytes and neutrophils in response to pattern recognition receptor (PRR) activation (e.g., Toll-like receptors [TLRs]) (13). Here, we report on a possible role for TLR9 in the potentiation of TREM-1 responses to B. pertussis. TLR9 is located within endosomes (14) and recognizes unmethylated cytosine- and guanine-rich nucleic acid motifs typical of bacterial DNA (15). TLR9 activation has been associated with enhanced vaccine-mediated protection against B. pertussis (16). However, no reports exist detailing its role in B. pertussis pathogenesis.

The engagement of TREM-1 by its ligands amplifies the expression of inflammatory factors, such as tumor necrosis factor-α (TNF-α) and C-C motif chemokine ligand 2 (CCL2) (13). Whether TREM-1 expression is protective in infectious diseases is unclear. TREM-1 activity is protective in models of Streptococcus suis infection and bacterial peritonitis (17, 18) and against the dissemination of Klebsiella pneumoniae (19). TREM-1-deficient animals, however, were unaffected in their ability to control or clear Legionella pneumophila or Leishmania major growth and were protected against excessive inflammatory responses to infection (20). These data suggest that TREM-1 blockade may be a therapeutic possibility in certain inflammation-producing bacterial infections while potentially deleterious in others.

Here, we hypothesize that TREM-1 activation serves as a checkpoint between efficient and deleterious inflammatory responses to B. pertussis. Additionally, we test the hypothesis that TREM-1 inhibition will reduce pulmonary inflammation without sacrificing bacterial control. The obtained data provide, for the first time, evidence that TREM-1 contributes to the excessive inflammatory response to B. pertussis infection and highlight the potential of TREM-1 inhibition for the treatment of pertussis and other hyperinflammatory diseases.

RESULTS

TREM-1 expression is increased in response to B. pertussis infection and is reduced by S1PR agonism.

Transcriptomic studies comparing RNA isolated from the lungs of animals challenged with B. pertussis or a “sham” inoculum of phosphate-buffered saline (PBS) identified TREM-1 and its associated cytokines (TNF-α, CCL-2, C-X-C motif chemokine ligand 3 [CXCL3], and interleukin-1β [IL-1β]) as significantly upregulated during B. pertussis infection (Fig. 1A). Additionally, canonical pathway analysis using Ingenuity Pathway Analysis (IPA) software identified the TREM-1 signaling pathway as having the 7th most significant overlap with our data set out of all associated with B. pertussis infection (Fig. 1B). Expression of TREM-1-associated genes was confirmed by reverse transcription-quantitative PCR (qRT-PCR) (Fig. 1C). Our previous studies identified S1PR agonists as potential treatments for pertussis disease (5). Interestingly, S1PR agonists reduced the expression of TREM-1 and TREM-1-associated genes in B. pertussis-infected mice compared to treatment with a vehicle control (Fig. 1A and B). This is the first time TREM-1 expression has been associated with either B. pertussis infection or the S1PR system.

FIG 1.

Transcriptomic analysis of TREM-1-associated genes following B. pertussis infection. (A) Differential expression of TREM-1-associated genes identified by whole-lung transcriptomics. Gene expression was compared using RNA that had been isolated from whole-lung tissue of C57BL/6 mice 4 days postinfection with B. pertussis or PBS sham infection (black bars). White bars represent gene fold induction in infected animals treated with vehicle or S1PR agonist. (B) Canonical pathway analysis performed with the Ingenuity Pathway Analysis software identified TREM-1 signaling as being significantly associated with genes whose expression is increased by B. pertussis infection compared to PBS (black bars) and in infected, water control-treated versus S1PR agonist (AAL-R, 0.5 mg/kg)-treated mice (white bars). The expression of TREM-1-associated genes at 4 days postinfection was confirmed by qRT-PCR (C). Fold induction is calculated by normalization to PBS-challenged mice. Each data point represents the mean of triplicate technical replicates from one of four individual mice (n = 4).

Increased expression of TREM-1 in response to B. pertussis infection is TLR9 dependent.

TREM-1 is expressed on the surface of macrophages and neutrophils following engagement of TLRs (13). In models of B. pertussis pathogenesis, TLR4 has been shown to mediate neutrophil recruitment, bacterial clearance, and cytokine production following B. pertussis infection and is required for vaccine-induced protective immunity (21, 22). Our transcriptomic screen identified the expression of DNA-sensing receptor TLR9 as being highly upregulated in the lungs following B. pertussis infection. These data were confirmed by qRT-PCR (Fig. 1C). To determine if TLR4 or TLR9 contributes to the induction of TREM-1 expression, we challenged knockout (KO) mice (TLR4 KO, TLR9 KO) with B. pertussis and assessed TREM-1 expression by qRT-PCR at 7 days postinfection (dpi). B. pertussis infection induces similar TREM-1 expression in wild-type C57BL/6 and TLR4-deficient mice (Fig. 2). This suggests that TLR9 contributes significantly to the induction of TREM-1 expression, and that TLR4 has minimal impact on the expression of TREM-1.

FIG 2.

B. pertussis-induced expression of TREM-1 is TLR9 dependent. qRT-PCR analysis of lung tissue harvested from C57BL/6, TLR9 KO, or TLR4 KO mice at 7 days postinfection with B. pertussis. Data are normalized to PBS-challenged animals of the same genetic background. Each datapoint represents the mean value obtained from triplicate samples from one of four individual mice (n = 4); ***, P < 0.005; NS, P > 0.05.

However, the robust upregulation of TREM-1 observed in wild-type and TLR4 KO mice was not evident in TLR9 KO mice (Fig. 2). This result suggests that TLR9, which mediates the recognition of CpG-rich DNA, promotes the expression of TREM-1 following B. pertussis infection.

TREM-1 does not contribute to bacterial control in B. pertussis infection.

Deficiencies in TREM-1 have been linked to increased bacterial burden in animal models of Streptococcus suis, Klebsiella pneumoniae, and Streptococcus pneumoniae infections (19, 23, 24). In other infectious models, however, TREM-1 promotes inflammation without contributing to pathogen control or clearance (e.g., L. pneumophila, L. major) (20). Additionally, inhibiting TREM-1 in combination with antibiotic treatment allowed for the efficient killing of S. suis and reduced inflammatory responses (17). To determine if TREM-1 contributes to the control of B. pertussis infection, infected mice were treated daily with one of two TREM-1 peptide inhibitors LP17 (5 mg/kg) or GF9 (20 mg/kg) or vehicle, and pulmonary bacterial burden was assessed at 4 and 7 dpi. These inhibitors employ a ligand-dependent (LP17) or ligand-independent (GF9) mechanism of TREM-1 inhibition (Fig. 3A). LP17 is a peptide with sequence homology to the extracellular region of TREM-1 (25). LP17, therefore, functions as a decoy receptor preventing interactions between TREM-1 and its ligands. By contrast, GF9 blocks interactions between the TREM-1 receptor and its signaling partner DNAX-activating protein of 12 kDa (DAP12) in the cell membrane (26). These two different inhibitors were used in our studies to address the uncertain nature of TREM-1 ligands. No differences in bacterial burden were observed between mice treated with vehicle or either TREM-1 inhibitor (Fig. 3B and C) at 4 or 7 dpi. This suggests that TREM-1 upregulation and signaling does not contribute to the control of B. pertussis bacterial burden. Additionally, this suggests that TREM-1-targeting therapeutics should not impact the host’s ability to control infection.

FIG 3.

TREM-1 does not contribute to bacterial control in B. pertussis. We used ligand-dependent (LP17) and ligand-independent (GF9) mechanisms to inhibit TREM-1 activity. LP17 binds potential TREM-1 ligands, preventing receptor binding. GF9 prevents transmembrane interactions between TREM-1 and its binding partner DAP12 (A). Bacterial burden was assessed at 4 and 7 dpi by plating the lungs of B. pertussis-challenged C57BL/6 mice receiving TREM-1 inhibitors (gray bars) LP17 (5 mg/kg) (B) or GF9 (20 mg/kg) (C) or vehicle control (black bars) on Bordet-Gengou agar plates. Vehicle controls consisted of 5 mg/kg scrambled peptide (LP17 studies) or PBS (GF9 studies). Each data point represents the average of three technical replicates obtained from one of four individual mice (n = 4). Data are shown as mean with standard deviation. P values were determined by two-way ANOVA using Sidak’s multiple-comparison test; NS, P > 0.05.

TREM-1 promotes inflammatory pathology during B. pertussis infection.

Acute inflammation can limit bacterial growth but does so at the risk of damaging the host. For this reason, anti-inflammatories have been widely applied in diseases of sterile inflammation, but less so in infectious diseases. Delineating protective and beneficial elements of the inflammatory response to infectious diseases will benefit the design of host-directed therapeutics. To determine if TREM-1 contributes to the inflammation observed during murine B. pertussis infection, mice were challenged with B. pertussis before treatment with LP17 (5 mg/kg), GF9 (20 mg/kg), or vehicle control. Lungs were harvested at 4 dpi, and expression of CCL-2, CXCL3, and TNF-α was measured by qRT-PCR. Expression of each analyte was found to be increased upon infection with B. pertussis, and that increase was significantly lower following TREM-1 inhibition than following treatment with vehicle control (Fig. 4A to C). TNF-α protein production was assessed by enzyme-linked immunosorbent assay (ELISA) at 4 and 7 dpi. At both time points, TNF-α levels were increased in response to B. pertussis infection and were significantly decreased by GF9-mediated TREM-1 inhibition (Fig. 4D). These data suggest that TREM-1 contributes to the expression of inflammatory cytokines observed following B. pertussis infection of C57BL/6 mice.

FIG 4.

TREM-1 promotes inflammatory responses to B. pertussis. Lungs were isolated from B. pertussis-challenged mice treated with TREM-1 inhibitors LP17 (5 mg/kg) or GF9 (20 mg/kg) at 4 dpi to assess expression of CCL2 (A), CXCL3 (B), and TNF-α (C) by qRT-PCR. qRT-PCR data are normalized to uninfected animals. Black bars represent B. pertussis-challenged, vehicle-treated animals. Gray bars represent B. pertussis-challenged animals treated with TREM-1 inhibitors LP17 (5 mg/kg) or GF9 (20 mg/kg). TNF-α protein levels were increased in response to B. pertussis infection (black bars) and decreased in response to treatment with GF9 (gray bars) (D). Data represent mean and standard deviation values obtained from triplicate technical replicates and four biological replicates (n = 4). P values were determined by two-way ANOVA using Sidak’s multiple-comparison test; *, P < 0.05; **, P < 0.01; ***, P < 0.005.

Pulmonary histopathology was assessed at 7 dpi to determine the contribution of TREM-1 to lung inflammation. Lungs taken at 7 dpi from B. pertussis-infected mice receiving LP17 (5 mg/kg) or GF9 (20 mg/kg) demonstrated reduced bronchovascular inflammation and consolidation of alveolar spaces compared to lungs from vehicle controls (Fig. 5), although only GF9 treatment reached statistical significance. Further studies are needed to determine if this effect is dose related or associated with a ligand-independent mechanism of TREM-1 inhibition. In summary, these data demonstrate that TREM-1 promotes inflammation and lung pathology in B. pertussis infection without inhibiting the antibacterial response. They also highlight the therapeutic potential of inhibiting TREM-1 for the treatment of pertussis.

FIG 5.

TREM-1 pathogenesis in B. pertussis infection. Lungs were isolated from B. pertussis-challenged mice treated with TREM-1 inhibitors LP17 (5 mg/kg, gray dots) or GF9 (20 mg/kg, gray dots) or vehicle (black dots; 5 mg/kg scrambled peptide for LP17 studies and PBS for GF9 studies) at 7 dpi. Lung tissues were stained with hematoxylin and eosin for assessment of histopathology. Three blinded investigators scored slides on a scale of 1 to 9 using our established semiquantitative system (A). Representative images are shown for infected C57BL/6 mice treated with vehicle (B), LP17 (C), or GF9 (D). Each data point represents one biological replicate (n = 4); *, P < 0.05.

DISCUSSION

The development of next-generation therapeutics for pertussis that improve clinical disease, reduce long-term lung damage, and retain the ability of antimicrobials to reduce dissemination is essential. B. pertussis infection elicits potent inflammatory responses, resulting in severe lung damage (27–29). Further, the infection is associated with long-term pulmonary sequelae (30). We hypothesize that reducing the inflammatory response to B. pertussis will improve the long-term outcomes of mild to severe disease and potentially rescue fatal infection. Previously, our group identified that short-term intranasal treatment with an S1PR agonist reduced inflammatory responses without impacting bacterial control in adult mice and prolonged survival in a neonatal model of lethal disease (5). However, long-term treatment of B. pertussis-challenged mice with oral S1PR agonists has been associated with a reduced ability to clear the infection (7). Transcriptomic studies attempting to discern the beneficial components of S1PR agonist treatment identified TREM-1 expression as increased during B. pertussis infection and decreased by treatment with S1PR agonists (Fig. 1).

TREM-1 is known to contribute to the inflammatory response to many pathogens. However, its role in bacterial control appears to be pathogen specific. TREM-1 inhibition using recombinant extracellular TREM-1 during S. suis infection reduced bacterial clearance despite promoting the resolution of inflammatory responses in mice (17). However, when combined with antibiotic treatment, TREM-1 inhibition not only dampened inflammatory responses but also promoted clearance of S. suis (17). Future studies will assess the potential of combining antibiotics with TREM-1 inhibitor treatment. In mouse models of bacterial (L. pneumophila), fungal (L. major), and viral (influenza) infections, the deletion of murine TREM-1 reduced inflammation without impacting pathogen control (20). Studies using TREM-1 inhibitors demonstrate its importance in sepsis, cancer, and collagen-induced arthritis (18, 26, 31).

To determine if TREM-1 could be a candidate for host-directed therapies in B. pertussis, we challenged mice with B. pertussis before treatment with one of two TREM-1 inhibitors, LP17 or GF9, or vehicle control. Mice treated with TREM-1 inhibitors showed no signs of altered bacterial control (Fig. 3). However, TREM-1 inhibitor-treated mice demonstrated reduced expression of inflammatory markers (Fig. 4). Further, TREM-1 inhibition was associated with reduced inflammatory pathology in the lungs of inhibitor-treated mice compared to in vehicle-treated control mice (Fig. 5). These data demonstrate for the first time a role for TREM-1 in pertussis pathogenesis. The ability to reduce inflammation without impacting bacterial control highlights the potential of TREM-1 inhibition as a therapy for pertussis.

Due to the uncertain nature of TREM-1 ligand(s), two classes of TREM-1 inhibitors were used in these studies: ligand-dependent and ligand-independent inhibitors LP17 and GF9, respectively. LP17 functions as a decoy receptor and competes with TREM-1 recognition of its endogenous ligand(s) (32) (Fig. 3A). By contrast, GF9 functions within the cell membrane by interfering with the interaction between TREM-1 and its signaling partner DAP12, thereby allowing receptor-ligand interactions but preventing downstream signaling (26) (Fig. 3A). In these studies, both GF9 and LP17 significantly reduced B. pertussis-induced expression of TREM-1 and its associated cytokines. Further, histopathology revealed that TREM-1 inhibition reduced inflammation in the lungs of infected mice (Fig. 5). The ability to reduce inflammation without inhibiting bacterial killing makes TREM-1 inhibitors attractive candidates for host-directed therapeutics for pertussis. TREM-1 may function as an inflammatory checkpoint in the response to infectious diseases. We hypothesize that initial macrophage activation results in sufficient antibacterial responses to resolve infection, but repeated or secondary activation via TREM-1 promotes a hyperinflammatory response that may be deleterious to the host. This is supported by the significant, but not complete, disruption to inflammatory responses observed in TREM-1 inhibitor-treated mice.

TREM-1 is expressed primarily on monocytes and neutrophils. GF9 formulated into macrophage-specific high-density lipoprotein (HDL)-mimicking particles (GF9-HDL) targets monocyte TREM-1 (26). In this situation, GF9 released by GF9-HDL in the cell reaches its intramembrane site of action from inside the cell. Our previous data demonstrated an essential role for monocyte S1P1 in S1PR agonist-mediated disease attenuation (5). Therefore, we hypothesize that the efficacy of TREM-1 inhibitors for the treatment of B. pertussis can be further improved by formulation with HDL-mimicking particles.

TREM-1 expression is induced by the action of pathogen-associated molecular patterns (PAMPs) on PRRs (13). Our transcriptomic studies identified TLR9 as highly expressed in response to B. pertussis infection. Studies in TLR9 KO mice demonstrated that DNA sensing is a major driver of TREM-1 expression in response to B. pertussis infection (Fig. 2). Future studies will seek to determine the potential of TLR9-targeting therapies for the treatment of pertussis disease.

TREM-1 ligands include peptidoglycan complexed with host peptidoglycan recognition protein-1 (PGLYRP1) (33). It is possible that B. pertussis PGN, tracheal cytotoxin (TCT), may be contributing to TREM-1 activation. In addition to dampening TREM-1 expression, S1PR agonists induce expression of PGLYRP4 but not PGLYRP1 (12). PGN, when bound to PGLYRP4, fails to induce the inflammatory responses induced by PGLYRP1-bound PGN (34). Future work will explore the hypothesis that PGLYRP4 competes with PGLYRP1 for PGN, influencing TREM-1 activation.

An impressive aspect of S1PR-mediated attenuation of pertussis disease is the ability to delay treatment without affecting outcomes (5). It will be important to assess the impact of delaying TREM-1 inhibitor administration to more therapeutically relevant time points on anti-inflammatory capability. The data presented here demonstrate a role for TREM-1 in the inflammatory response to B. pertussis despite not contributing to bacterial control. These data suggest that TREM-1 may be a future target for host-directed therapies in pertussis.

MATERIALS AND METHODS

Bacterial strains.

A streptomycin- and nalidixic acid-resistant derivative of the B. pertussis Tohama I strain, developed by Carbonetti et al. (35), was used for all experiments described. B. pertussis was grown on Bordet-Gengou (BG) agar plates supplemented with 10% defibrinated sheep blood and 200 μg/ml streptomycin.

Mouse infections.

Six-week-old C57BL/6 mice (Charles River) and TLR4 and TLR9 knockout mice (Jackson Laboratory) were used in accordance with the University of Maryland, Baltimore (UMB), Institutional Animal Care and Use Committee. For infections, bacterial inocula were prepared in sterile phosphate-buffered saline (PBS) following a 48-h incubation on BG agar. Anesthetized mice (isoflurane) received 2 × 10⁶ CFU B. pertussis intranasally in a total volume of 50 μl.

TREM-1 inhibitor studies.

Ligand-dependent TREM-1 inhibitory peptide LP17 (LQVTDSGLYRCVIYHPP, C-terminal amidated; Thermo Fisher) is based on the complementarity-determining region 3 (CDR3) region of TREM-1 (32). Based on previous publications (32), a solution of a scrambled peptide (TDSRCVIGLYHPPLQVY, C-terminal amidated; Thermo Fisher) in sterile PBS was used as a vehicle in LP17 studies. Ligand-independent TREM-1 inhibitory peptide GF9 (GFLSKSLVF; SignaBlok) is based on the transmembrane domain of TREM-1 (26). LP17 (5 mg/kg), GF9 (20 mg/kg), or vehicle controls (scrambled peptide and PBS, respectively) were administered daily by intraperitoneal injection beginning on day 0 based on previous studies (26, 32). Lungs were harvested at 4 and 7 days postinfection (dpi) for assessment of bacterial burden (superior, middle, and inferior lobes), inflammatory cytokine expression (postcaval lobe), and pulmonary histopathology (left lobe).

RNA isolation and processing.

Lung tissue was harvested directly into RNAlater (Sigma-Aldrich) and stored overnight at 4°C before RNA extraction using RNA Stat60 (TelTest, Inc.) as per the manufacturer’s instructions. In brief, samples were homogenized using an Omni TH mixer (Omni, Inc.), phase separated with the addition of chloroform, and precipitated with isopropanol. RNA was quantified, and 1 μg was reverse transcribed using a reverse transcription system (Promega). Quantitative real-time PCR was performed in a BIORAD CFX96 real-time PCR instrument. The hypoxanthine phosphoribosyltransferase (HPRT) gene was used as an internal housekeeping control gene, with all genes normalized to the HPRT gene and expression calculated as fold change compared with PBS-inoculated control animals (calculated by the threshold cycle 2^−ΔΔCT method).

Pathology.

Lungs were perfused with PBS before removal into 10% (wt/vol) neutral buffered formalin. Hematoxylin-eosin staining was performed by the Pathology, Electron Microscopy and Histology Laboratory at the University of Maryland. A semiquantitative scoring system was employed based on the degree of infiltrate in the bronchovascular region, and the degree of tissue consolidation was observed. Lung sections were given a score between 0 and 9 based on the degree of bronchovascular inflammation (0 to 3), the percentage of bronchovascular bundles involved (0 to 3), and the amount of consolidation or infiltration of cells in the alveolar spaces (0 to 3).

RNA sequencing.

RNA sequencing analysis was performed by the Informatics Resource Center, Institute for Genome Sciences, University of Maryland School of Medicine (UMSOM). Paired-end Illumina libraries were mapped to the mouse reference, Ensembl release GRCm38.74, using TopHat v1.4.0 with the default mismatch parameters. Read counts for each annotated gene were calculated using HTSeq. The DESeq Bioconductor package (v1.5.24) was used to estimate dispersion, normalize read counts by library size to generate the counts per million for each gene, and determine differentially expressed genes between two conditions. Differentially expressed transcripts with a false discovery rate of <0.05 and a log₂ fold change in expression were used for downstream analysis. The list of differentially expressed genes was used to compute the enrichment of biological pathways using Ingenuity Pathway Analysis (IPA). IPA was used to identify canonical pathways that were most significantly associated with the data sets analyzed.

Enzyme-linked immunosorbent assay.

Protein levels of TNF-α were assessed by enzyme-linked immunosorbent assay (ELISA) (BioLegend) using whole-lung tissue homogenate at 4 and 7 dpi from mice challenged with B. pertussis and treated with GF9 (20 mg/kg) or PBS. Lungs were harvested into 2 ml of PBS containing 20 μl of proteinase inhibitor cocktail (Sigma). Lungs were then homogenized as described above and centrifuged at 300 × g for 10 min at 4°C, and the supernatant was harvested for protein detection.

Statistical analysis.

Data were analyzed and graphs prepared in GraphPad Prism statistical software. Graphs represent mean values with standard deviation. Student’s t tests were used to determine significance between two data sets, and two-way ANOVA was used to determine significance between multiple groups. For canonical pathway analysis, the Fisher exact test was used to calculate a P value, determining the probability that the association between the genes in the data set and the canonical pathway was significant.

Data availability.

Data discussed in the manuscript have been deposited to the NCBI Sequence Read Archive under BioProject PRJNA493118 (https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA493118).