ABSTRACT
Incidence of whooping cough (pertussis), a bacterial infection of the respiratory tract caused by the bacterium Bordetella pertussis, has reached levels not seen since the 1950s. Antibiotics fail to improve the course of disease unless administered early in infection. Therefore, there is an urgent need for the development of antipertussis therapeutics. Sphingosine-1-phosphate receptor (S1PR) agonists have been shown to reduce pulmonary inflammation during Bordetella pertussis infection in mouse models. However, the mechanisms by which S1PR agonists attenuate pertussis disease are unknown. We report the results of a transcriptome sequencing study examining pulmonary transcriptional responses in B. pertussis-infected mice treated with S1PR agonist AAL-R or vehicle control. This study identified peptidoglycan recognition protein 4 (PGLYRP4) as one of the most highly upregulated genes in the lungs of infected mice following S1PR agonism. PGLYRP4, a secreted, innate mediator of host defenses, was found to limit early inflammatory pathology in knockout mouse studies. Further, S1PR agonist AAL-R failed to attenuate pertussis disease in PGLYRP4 knockout (KO) mice. B. pertussis virulence factor tracheal cytotoxin (TCT), a secreted peptidoglycan breakdown product, induces host tissue damage. TCT-oversecreting strains were found to drive an early inflammatory response similar to that observed in PGLYRP4 KO mice. Further, TCT-oversecreting strains induced significantly greater pathology in PGLYRP4-deficient animals than their wild-type counterparts. Together, these data indicate that S1PR agonist-mediated protection against pertussis disease is PGLYRP4 dependent. Our data suggest PGLYRP4 functions, in part, by preventing TCT-induced airway damage.
INTRODUCTION
Whooping cough (pertussis) is a severe respiratory infection caused by the bacterium Bordetella pertussis. Pertussis is characterized by a paroxysmal cough lasting several weeks, leading to an estimated 24 million cases and 160,000 deaths annually (1, 2). Pertussis is the only vaccine-preventable disease to be consistently on the rise in the United States since 1976 (3). It has been suggested that the reason for increased incidence of disease is the rapidly waning immunity elicited by the currently used acellular pertussis vaccine (4–6). Antibiotic treatment of pertussis is only effective if diagnosed before onset of paroxysmal cough and is therefore rarely helpful (7, 8). Waning immunity, increased disease incidence and ineffective postexposure therapeutics have resulted in an urgent need for novel drug targets for the treatment of pertussis.
Murine models of pertussis have been well characterized. Since mice do not produce an audible cough, pulmonary histopathology is an important readout of disease progression. Pertussis toxin (PT) and tracheal cytotoxin (TCT) are two virulence factors produced by B. pertussis that may contribute to airway histopathology. PT is secreted by the bacteria via a type IV secretion system, promotes colonization of the respiratory tract and exacerbates airway pathology (9–12). TCT is released by B. pertussis as a by-product of cell wall turnover. Cell wall turnover was first discovered in Bacillus megaterium (13, 14) and is now known to occur in a variety of Gram-negative bacteria (15, 16). Escherichia coli releases 5 to 8% of its peptidoglycan (PGN) into culture media per generation. In E. coli this recycling occurs via the permease AmpG (17). B. pertussis PGN recycling via AmpG is inefficient, resulting in accumulation of extracellular TCT, a monomeric peptidoglycan subunit. Replacing B. pertussis AmpG with that of E. coli resulted in a strain with 99% reduction of released TCT (18). TCT is involved in damage and extrusion of mammalian ciliated epithelial cells, hindering the ability of the large airways to clear infecting microorganisms (19–21).
Sphingosine-1-phosphate (S1P) can signal via five G-protein-coupled receptors (S1P1 to S1P5). Agonists of these receptors have been used to treat multiple sclerosis and are being investigated for the treatment of psoriasis, Crohn’s disease, ulcerative colitis, and more (22, 23). Previous work from our group demonstrated the potential of host-directed therapies for treatment of pertussis with S1P receptor agonists in a mouse model. Single doses of S1P receptor (S1PR) agonists (specifically S1PR1) at prophylactic (1 h postinfection) and postexposure (day 4 [D4] postinfection) time points resulted in reduced inflammatory cytokine expression and ablated pulmonary pathology (24, 25). Further, S1PR agonism resulted in improved survival rates in a lethal model of neonate disease (24). Although these compounds demonstrated significant benefit in the murine model of disease, the S1PR is present on a wide variety of cell types throughout the body, influencing multiple physiological processes (26). Understanding the mechanism(s) of protection may allow us to develop more specific therapeutics.
To further our understanding of the mechanisms behind S1PR agonist-mediated pertussis disease attenuation, we performed transcriptome sequencing (RNA-seq) studies on B. pertussis-infected mice after drug administration. These studies identified peptidoglycan recognition protein 4 (PGLYRP4) as one of the most abundantly transcribed genes in response to treatment. PGLYRPs are secreted antimicrobial proteins produced at mucosal surfaces in response to infection (27). PGLYRPs elicit bacteriostatic and bactericidal effects through multiple mechanisms. For example, mammalian PGLYRP2 has been shown to exert amidase activity upon peptidoglycan (28). PGLYRP4 can bind to peptidoglycan on Gram-positive and Gram-negative bacteria through a zinc-dependent mechanism (29). Here, we demonstrate a novel role for PGLYRP4 in both B. pertussis pathogenesis and S1PR agonist-mediated disease attenuation. In addition, we demonstrate that PGLYRP4 limits early inflammatory responses to B. pertussis. Further, we identify the B. pertussis virulence factor TCT as a driver of lung inflammation during the early stages of B. pertussis infection. Finally, we identify TCT as a putative target of PGLYRP4 activity. Based on these data, we hypothesize that the S1PR agonist AAL-R induces expression of PGLYRP4 from an unknown cell type, which provides protection against the actions of B. pertussis virulence factor TCT.
(This study was presented in part at the 2018 Gordon Research Conference on Biology of Acute Respiratory Infection, Ventura, CA, and the 2018 Bordetella Research Day, Baltimore, MD.)
RESULTS
Transcriptional responses to S1PR agonism in B. pertussis-infected mice.Previously, we showed that treatment of B. pertussis-infected animals with S1PR agonist AAL-R resulted in reduced pulmonary inflammatory pathology (25). To elucidate the mechanisms by which AAL-R reduced inflammation, we used RNA-seq to compare lung transcriptional responses to AAL-R treatment versus vehicle control treatment in B. pertussis-infected C57BL/6 adult mice at 4 days postinoculation (dpi). A total of 1,687 genes were >2-fold differentially regulated in infected AAL-R-treated animals compared to controls. Of these 1,687 genes, 573 were upregulated (and 1,114 were downregulated) in lungs receiving S1PR1 agonist (AAL-R) versus vehicle (data not shown). A representation of the most significantly up- and downregulated transcripts after S1PR agonism is presented in Fig. 1A. Transcripts enriched to a greater degree in vehicle- than S1PR agonist-receiving mice (that is, genes downregulated by S1PR agonism) include (i) ubiquitin-protein transferase Trim30c, a regulator of autophagy (30); (ii) NF-κB and MyD88 activator IL-17A; (iii) aconitate decarboxylase (Acod1), which is involved in innate responses to microbes and limiting Toll-like receptor responses (31); and (iv) sodium-phosphate ion channel Slc17A2 . Together, these data demonstrate a reduction in proinflammatory gene expression following S1PR agonism. Further, the genes most highly upregulated in agonist receiving animals can be found in Fig. 1A. Included in this list are genes encoding C-type lectins (Glycam1 and CD209b), zinc-finger proteins (Vezf1 and Zcchc5), genes associated with myogenic function (Myod1 and Myf5), collagen-associated genes (Lrrc15 and Col9a2), a neuronal growth gene (Lsamp), aryl-hydrocarbon interacting protein (Aipl1), Nup62cl (nucleoporin), Cartpt (neuropeptide), food satiation-associated gene Ucn2, and peptidoglycan recognition protein 4 (PGLYRP4). This is a diverse collection of genes, distributed across multiple biological classes and processes. Peptidoglycan recognition proteins have been associated with the control of infectious and noninfectious inflammatory responses (32, 33). Based on this property, PGLYRP4 was selected as a candidate for further investigation and forms the basis for the focus of the present study.
Transcriptional analysis following S1PR agonism. Infected mice were treated with either S1PR agonist AAL-R or an H2O vehicle control. RNA was isolated from the lungs of four mice per group, and transcriptional analysis identified the differentially expressed genes (A), predicted upstream regulators (B), and canonical pathways predicted to be altered by S1PR agonism (C). Gene expression is expressed as the log2-fold change. PA positive fold change indicates greater expression in S1PR agonist-treated animals than in vehicle-treated animals. The activation z-score was calculated using IPA software. P values were determined by a Fisher exact test.
Ingenuity Pathway Analysis (IPA; Ingenuity Systems) of differentially expressed genes was used for prediction of upstream regulators of these data sets (Fig. 1B). This analysis predicts that, upstream of the time point analyzed, the levels of the proinflammatory cytokines tumor necrosis factor alpha (TNF-α), gamma interferon (IFN-γ), and interleukin-1β (IL-1β) were increased in animals receiving vehicle compared to levels in those receiving S1PR agonist. Further, vehicle-treated animals are predicted to have more NF-κB and MyD88 expression. In addition, IPA was used to make predictions on pathways that are activated or inhibited in our data sets (Fig. 1C). S1PR agonist treatment was associated with reduced expression of genes associated with adhesion and movement of leukocytes, phenomena regulated at multiple levels (barrier function, adhesion, and homing) by S1PR (34, 35). IPA also predicted that S1PR agonist-receiving animals have altered dendritic cell maturation and differences in signaling via the triggering receptor expressed on myeloid cells 1 (TREM-1). TREM-1 expression has been associated with increased expression of CCL2, TNF-α, IL-1α, IL-1β, and IL-6, all of which are reduced following S1PR agonism. These data indicate a less inflammatory environment in S1PR agonist-receiving mouse lungs.
PGLYRP4 limits early inflammatory responses to B. pertussis.After the identification of PGLYRP4 as one of the most highly enriched transcripts following S1PR agonism in B. pertussis-infected mice, we sought to determine its contribution to host responses to B. pertussis infection. To this end, we obtained PGLYRP4-deficient mice, which are on a BALB/c background (36). Real-time quantitative PCR confirmed the minimal differential regulation of PGLYRP genes by B. pertussis infection and significant upregulation of PGLYRP4 in BALB/c mice receiving S1PR agonist after infection (Fig. 2A), consistent with our RNA-seq data. S1PR agonist treatment of uninfected mice resulted in modest upregulation of PGLYRP4 expression. However, S1PR agonism in the presence of B. pertussis infection induced significantly greater PGLYRP4 upregulation than in the absence of infection (≈200-fold versus 4-fold) (Fig. 2A). Bacterial burdens in the lungs of these mice were not significantly different between groups (data not shown).
PGLYRP4 protects against B. pertussis-induced inflammatory pathology. PGLYRP4 expression was confirmed by quantitative reverse transcription-PCR (qRT-PCR). (A) RNA was isolated from the lungs of BALB/c mice receiving a sham inoculum (PBS) or B. pertussis infection (wild type), S1PR agonist AAL-R-treated uninfected mice (AAL-R), or AAL-R-treated infected mice (WT + AAL-R) at 4 dpi. (B) The pulmonary bacterial burden was determined by plating lung homogenate from BALB/c (gray circle) or PGLYRP4 KO (open circle) mice at days 4 (D4), 7 (D7), and 14 (D14) postinfection on Bordet-Gengou agar plates. (C) The expression of the proinflammatory cytokines IFN-γ and IL-6 4 days after infection of BALB/c (gray) or PGLYRP4 KO (open) animals was measured by qRT-PCR. (D) Pulmonary inflammation was semiquantitatively scored and compared between BALB/c (gray) and PGLYRP4 KO (open) animals at 4, 7, and 14 days postinfection. P values were determined by using a Student t test (A and C) and two-way ANOVA (B and D). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.
PGLYRP4-deficient animals had a slight reduction in bacterial burden compared to BALB/c mice at 4 dpi, but this was not found to be significant by two-way analysis of variance (ANOVA). Similarly no differences in CFU were observed at 7 or 14 dpi (Fig. 2B). B. pertussis-infected PGLYRP4-deficient mice demonstrated increased early inflammatory responses compared to their wild-type counterparts, with knockout animals expressing significantly greater levels of the inflammatory cytokines IFN-γ (P = 0.001) and IL-6 (P = 0.003) (Fig. 2C) at 4 dpi. B. pertussis-infected PGLYRP4 knockout (KO) animals displayed significantly increased pulmonary inflammation as determined by histopathology at 4 dpi (P < 0.01) (Fig. 2D). Interestingly, this increase was not seen at later time points, with PGLYRP4 KO mice showing slight but significant reductions in inflammation compared to infected wild-type animals (P = 0.02) at 7 dpi. These data demonstrate a role for PGLYRP4 in controlling early inflammatory responses to B. pertussis infection.
PGLYRP4 is required for S1PR agonist-mediated attenuation of pertussis disease.The contribution of PGLYRP4 to S1PR agonist-mediated attenuation of disease was determined by comparing host responses in wild-type and PGLYRP4-deficient animals receiving AAL-R or vehicle after infection. We previously showed that S1PR agonism reduces expression of B. pertussis-induced IL-6, IFN-γ, and TNF-α in C57BL/6 mouse lungs (25). B. pertussis-infected PGLYRP4 KO animals showed significantly stronger IL-6 and IFN-γ responses than wild-type animals at 4 dpi (Fig. 2). B. pertussis-infected BALB/c mice showed reduced IL-6 (P = 0.05), TNF-α (P = 0.05). and IFN-γ (not significant [NS], P = 0.2) gene expression in response to S1PR agonism versus vehicle treatment (Fig. 3A to C), similar to our previous findings in C57BL/6 mice. PGLYRP4 KO mice, however, failed to dampen these inflammatory cytokines in response to S1PR agonism (Fig. 3A to C). S1PR agonist-receiving PGLYRP4 KO mice actually show increased expression of IFN-γ and TNF-α (Fig. 3B and C). The reason for this increase is unclear and will be the focus of future studies, but further underscores the anti-inflammatory activity of PGLYRP4 induced by S1PR agonism. Further, S1PR agonism mediated a significant reduction in pulmonary histopathology scores at 7 dpi in infected BALB/c mice (P = 0.002) but not in PGLYRP4 KO mice (P = 0.76) (Fig. 3D). This demonstrates an essential role for PGLYRP4 in S1PR agonist-mediated attenuation of inflammation in B. pertussis-infected mice.
PGLYRP4 is required for S1PR agonist-mediated disease attenuation. Four days after infection and treatment with either vehicle (circles) or S1PR agonist (AAL-R) (squares), BALB/c (gray shapes) or PGLYRP4 KO (open shapes) mice were analyzed for expression of inflammatory cytokines IL-6 (A), IFN-γ (B), and TNF-α (C). (D) Pulmonary pathology was assessed at days 4 and 7 after infection of BALB/c or PGLYRP4 KO mice receiving vehicle or AAL-R using a previously published semiquantitative scoring system. P values were calculated by two-way ANOVA (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).
TCT drives early inflammatory responses in the absence of PGLYRP4.PGLYRP family members bind and exert activity upon both lipopolysaccharide (LPS) and PGN (27, 37, 38). B. pertussis virulence factor TCT is structurally equivalent to a monomeric subunit of PGN and released from the bacterial cell during cell wall turnover. To investigate the role of TCT in pulmonary inflammatory pathology, murine challenge studies were carried out using an engineered B. pertussis strain that overproduces TCT (24-fold more than the wild type; W. E. Goldman, unpublished data) or the parental wild-type B. pertussis strain (BC36). Animals infected with the TCT-hyperproducing strain showed enhanced pulmonary inflammatory pathology (Fig. 4A) compared to those challenged with parental B. pertussis strain BC36. Importantly, infection with TCT-hyperproducing B. pertussis resulted in significantly greater pulmonary inflammatory pathology in PGLYRP4-deficient animals than in wild-type BALB/c mice at 4 dpi (Fig. 4A). S1PR agonist treatment attenuated early (4 dpi) inflammatory pathology, as seen following infection with TCT-hypersecreting strains, in BALB/c but not in PGLYRP4 KO mice (Fig. 4A). Infection with TCT-hypersecreting strains also resulted in increased expression of IL-6 (P < 0.001) at early time points (day 4) compared to challenge with parental B. pertussis BC36 despite similar bacterial burdens (Fig. 4B and C). This result demonstrates a role for TCT in induction of early inflammatory pathology and mirrors that observed infection of PGLYRP4-deficient animals (Fig. 2D), consistent with the idea that PGLYRP4-dependent control of early inflammatory responses is mediated via its actions on TCT. Interestingly, mice infected with the TCT-hyperproducing strain induced significantly greater expression of the PGLYRP4 gene at 4 dpi than mice infected with parental B. pertussis BC36 (Fig. 4D), but it should be noted that this increase (2-fold) is small and paled in comparison to the increase induced by AAL-R/S1PR agonist treatment (>200-fold; Fig. 2A) and was clearly insufficient to inhibit pathology. Together, these results support the hypothesis of a TCT-PGLYRP4 axis in B. pertussis pathology, where PGLYRP4 inhibits TCT-induced pathology.
TCT drives early inflammatory responses and PGLYRP4 expression. Pulmonary pathology was assessed after hematoxylin-eosin staining of B. pertussis-infected BALB/c (gray) or PGLYRP4 KO (open) mice. (A) TCT-hyperproducing B. pertussis (TCT+, inverted triangle) induces early pathology in BALB/c mice comparable to that observed after wild-type (WT, circle) infection of PGLYRP4 KO mice. (B) IL-6 expression was measured by qRT-PCR after infection of BALB/c mice with wild-type (circle) or TCT-hyperproducing (triangle) B. pertussis. (C) Exacerbated pathology and inflammatory cytokine production were noted between B. pertussis strains despite no differences being detected in bacterial burden. (D) qRT-PCR analysis of PGLYRP4 expression from murine lung samples revealed increased PGLYRP4 expression in lungs challenged with TCT-hyperproducing B. pertussis compared to wild-type mice. P values were determined by two-way ANOVA (A) or Student t test (B to D). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.
DISCUSSION
Pertussis disease has reached incidence levels not seen since the 1950s. When outbreaks occur we lack effective postexposure therapeutics to treat the disease or complications arising from the paroxysmal coughing associated with disease (vomiting, apnea, and broken ribs). Therefore, pertussis is a reemerging, predictable, public health crisis for which we are not prepared, necessitating the development of effective treatments. Macrolide antibiotic treatment represents the current standard of postexposure pertussis treatment (7). However, unless administration of antibiotics begins during the catarrhal stage of disease, when symptoms are nonspecific and diagnoses therefore difficult, their use is not associated with improved symptoms (8).
Here, we follow on from our recent reports demonstrating the potential of S1PR agonists as potent suppressors of pertussis-mediated lung inflammation (24, 25). Manipulation of S1P signaling has been associated with improved outcomes in a variety of infectious and noninfectious diseases (39–41). Therefore, understanding the mechanisms of protection is important for development of improved therapeutics. To identify key mediators of S1PR agonist-mediated disease attenuation we performed RNA-seq studies comparing the transcriptional profiles of (i) uninfected and B. pertussis-infected animals and (ii) B. pertussis-infected animals with or without S1PR agonism. Transcripts coding for secreted antimicrobial protein PGLYRP4 were highly enriched in infected animals receiving S1PR agonist versus vehicle. PGLYRP4 targets peptidoglycan on dividing Gram-positive and Gram-negative bacteria (29) but has yet to be described in response to S1PR agonism or B. pertussis infection and may represent an interesting, novel mechanism by which S1PR agonism contributes to protection from disease. It will be interesting to determine whether PGLYRP4 is induced in other lung diseases for which S1PR agonism has been shown to be protective, such as tuberculosis and non-peptidoglycan-producing influenza (40, 41).
Studies performed in knockout mouse models allowed us to identify a role for PGLYRP4 in suppressing early inflammatory responses to pertussis disease. However, animals lacking PGLYRP4 experienced a slight but significant reduction in inflammatory pathology at 7 dpi. This may be related to the slight reduction in bacterial burden noted at day 4 in deficient animals, but these differences are small and only present before day 7. This may also imply that promoting an efficient early response to peptidoglycan or lipopolysaccharide may exacerbate disease in the short term but could ultimately be beneficial over the course of disease.
Importantly, S1PR agonist-mediated attenuation of pertussis disease was dependent on the presence of PGLYRP4. S1P signaling has been associated with the improvement or exacerbation of several infectious and noninfectious diseases. Induction of PGLYRP family members represents a novel mechanism by which S1PR agonist disease attenuation is mediated. Understanding the mechanism(s) by which PGLYRP4 contributes to S1PR agonist-mediated disease attenuation may allow for powerful next-generation therapeutics. The mechanism of PGLYRP4 induction by S1PR agonists will be investigated in future studies.
One of the mammalian PGLYRP proteins, PGLYRP2, has enzymatic activity allowing its binding and alteration of PGN and LPS. Specifically, murine PGLYRP2 showed N-acetyl-muramoyl-l-alanine amidase (NAMLAA) activity on PGN and B. pertussis TCT resulting in its degradation (28, 42). PGLYRP4, however, lacks the essential zinc-binding cysteine moiety, preventing it from acting as an amidase (29). PGLYRP4 has been shown to elicit bactericidal responses through induction of two-component stress responses, oxidative stress, and thiol and metal stresses and generating reactive oxygen species (43, 44). It has further been predicted based on structure to disrupt bacterial cell maturation via steric inhibition of biosynthetic enzymes and prevention of peptide cross-linking (45). This is interesting as, in our model, loss of PGLYRP4 had minimal effect on bacterial burden or growth. Indeed, the greatest impact of PGLYRP4 on infection came in the context of infection with TCT-overproducing strains. TCT, a contributor to tissue damage during infection, is a released monomeric DAP-type peptidoglycan, whose structure is linked to its ability to induce inflammatory responses (46). Our data indicate that loss of PGLYRP4 or increases in bacterial TCT expression result in excess early inflammation. This is the first report of a phenotype for a TCT-overexpressing strain. Further, loss of PGLYRP4 results in increased susceptibility to TCT-induced inflammation (Fig. 4D). Understanding the nature of any PGLYRP4-TCT interaction is important and will require protein-level studies. We predict that PGLYRP4 binds to TCT and physically inhibits its proinflammatory activity. The results presented here raise interesting questions about B. pertussis pathogenesis and treatments. If our hypothesis, i.e., that TCT contributes to pathogenesis, is correct: would mounting robust anti-TCT responses reduce the severity of disease in following infection? Could B. pertussis be simultaneously gaining advantage from the release of an immune target (TCT) while directing responses away from the bacterium? What is the contribution of TCT to pathogenesis relative to other B. pertussis virulence factors and is targeting released TCT sufficient to protect against pathogenesis? Deacetylation of PGN can prevent the mounting of an efficient/inflammasome-mediated response (47). B. pertussis produces two putative enzymes for deacetylation of its peptidoglycan, LpxC and BpsB. Do these enzymes contribute to the prevention of an effective antipertussis response? It will also be important to determine the impact of any immunomodulatory treatments on future immunity against the pathogen. If infants awaiting vaccination, or midvaccination schedule, are particularly susceptible, it may be necessary to alter the vaccination schedule of treated patients.
In addition to their direct effect on PGN, PGLYRP family members have been shown to bind bacterial cells and induce two-component stress responses (48). B. pertussis stress responses are poorly understood. Recent work showed altered virulence factor secretion following induction of the two-component stress response operon Rpo/Rse, with increased secretion of ACT and decreased production of PT (49). Our future work will seek to determine the nature of the PGLYRP4-TCT interaction, whether PGLYRP4 also mediates its anti-inflammatory activity through effects on other virulence factors, whether PGLYRP4 induces B. pertussis stress responses or direct bacterial killing, and the impact of these responses on infection.
MATERIALS AND METHODS
Bacterial strains.In this study, our wild-type streptomycin-resistant strain of B. pertussis Tohama I (wild-type) was used unless otherwise stated (9). B. pertussis was grown on Bordet-Gengou agar supplemented with 200 μg/ml streptomycin and 10% defibrinated sheep blood (Lampire Biological Laboratories) for 48 h at 37°C. To investigate the role of tracheal cytotoxin a mutant strain was engineered to overproduce TCT. Briefly, allelic-exchange vector pBP10 was used to insert a kanamycin resistance gene into B. pertussis BC36 to disrupt the ampG gene. Southern blot analysis demonstrated that the wild-type allele was interrupted with the kanamycin resistance gene. The supernatant of in vitro log-phase cultures of this ΔampG mutant accumulated 24 times more TCT than its wild-type counterpart, as measured by high-pressure liquid chromatography.
Mouse infections.C57BL/6, BALB/c (Charles River), or PGLYRP4–/– (kindly provided by R. Dziarski) animals were used in accordance with Institutional Animal Care and Use Committee protocol 0417005 (University of Maryland, Baltimore, MD). Bacterial inoculum was prepared following 48 h of growth on Bordet-Gengou agar in a phosphate-buffered saline (PBS) suspension and administered intranasally. For S1PR agonist treatment, AAL-R (a generous gift from H. Rosen) was prepared in sterile water at a concentration of 0.5 mg/kg and delivered intranasally. Lungs were removed for analysis of bacterial burden, histopathology and transcriptional analysis.
RNA isolation and processing.Immediately following harvest, tissues were snap-frozen using a dry-ice/isopropanol bath. RNA was extracted using the RNeasy microarray tissue kit (Qiagen) according to the manufacturer’s instructions and DNase treated on-column to prevent DNA contamination. RNA was checked on the 2100 Bioanalyzer (Agilent) for quality control. Quantitative real-time PCR was performed with Maxima SYBR green/ROX quantitative PCR master mix in an Applied Biosystems 7500 Fast real-time PCR system. As an internal housekeeping control gene, hypoxanthine phosphoribosyltransferase was used. Expression was calculated as the fold change compared to PBS “sham”-infected animals using the 2−ΔΔCT method.
RNA sequencing.RNA-seq analysis was carried out by the Informatics Resource Center, Institute for Genome Sciences, UMDSOM. 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 log2-fold change were used for downstream analysis. The list of differentially expressed genes was used to compute the enrichment of biological pathways using IPA.
IPA (Ingenuity Systems) was also used to perform upstream regulator analysis, predicting the regulators of each data set and their activation state. This is based on the IPA Ingenuity Pathway Knowledge Base. Predicted upstream regulators are assigned a z-score related to their activation state. These predictions are based on previously experimentally observed transcription events from the literature. A z-score of >2 indicates the regulator is predicted to be activated; a z-score of <2 predicts the regulator to be inhibited. IPA software was used to identify canonical pathways that were most significant to the data set. 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.
Pathology.Before removal into 10% (wt/vol) buffered formalin, lungs were perfused with PBS. Hematoxylin-eosin staining was performed by the Pathology, EM, and Histology Laboratory at University of Maryland. Histopathological findings were scored on a scale of 0 to 9. Each slide was scored 0 to 3 across three categories (for a maximum score of 9): the severity of bronchovascular bundle inflammation, the percentage of bronchovascular bundles involved, and the degree of tissue consolidation observed.
Statistical analysis.Graphs were plotted and data analyzed using statistics software GraphPad Prism 6.0. Fold changes, for real-time PCR, were calculated per mouse compared to PBS-inoculated groups. All plots represent mean values with standard deviations. For RNA-seq data, IPA software was used to calculate a z-score for upstream regulators, and a Fisher exact test was used to assess the probability that the associations of genes in our data set and canonical pathways were significant. A Student t test was used when assessing significance between two means.
Accession number(s).Reads were deposited in the NCBI Sequence Read Archive under BioProject PRJNA493118.
ACKNOWLEDGMENTS
This study was supported by Public Health Service grants AI-101055 and AI-119566 and AI-135465 from the National Institute of Allergy and Infectious Diseases.
We declare there are no potential conflicts of interest.
FOOTNOTES
- Received 2 August 2018.
- Returned for modification 24 August 2018.
- Accepted 26 November 2018.
- Accepted manuscript posted online 3 December 2018.
- Copyright © 2019 American Society for Microbiology.
