The microbiota protects against respiratory infection via GM-CSF signaling

Abstract

The microbiota promotes resistance to respiratory infection, but the mechanistic basis for this is poorly defined. Here, we identify members of the microbiota that protect against respiratory infection by the major human pathogens Streptococcus pneumoniae and Klebsiella pneumoniae. We show that the microbiota enhances respiratory defenses via granulocyte-macrophage colony-stimulating factor (GM-CSF) signaling, which stimulates pathogen killing and clearance by alveolar macrophages through extracellular signal-regulated kinase signaling. Increased pulmonary GM-CSF production in response to infection is primed by the microbiota through interleukin-17A. By combining models of commensal colonization in antibiotic-treated and germ-free mice, using cultured commensals from the Actinobacteria, Bacteroidetes, Firmicutes, and Proteobacteria phyla, we found that potent Nod-like receptor-stimulating bacteria in the upper airway (Staphylococcus aureus and Staphylococcus epidermidis) and intestinal microbiota (Lactobacillus reuteri, Enterococcus faecalis, Lactobacillus crispatus and Clostridium orbiscindens) promote resistance to lung infection through Nod2 and GM-CSF. Our data reveal the identity, location, and properties of bacteria within the microbiota that regulate lung immunity, and delineate the host signaling axis they activate to protect against respiratory infection.

Fig. 1

The microbiota protects against Gram-negative and Gram-positive lung infection via GM-CSF signaling. a, b S. pneumoniae and c, d K. pneumoniae burden 6 h (a, c) and 12 h (b, d) post-intranasal inoculation. e–j Cytokine levels in lung tissue 12 h post-inoculation with S. pneumoniae (e–g) and K. pneumoniae (h–j). k, l Cytokine levels in lung tissue 12 h post-intranasal inoculation with K. pneumoniae LPS (1 μg/mouse) (k) or lipoteichoic acid (LTA) (50 μg/mouse) (l). For (e–l) data are from n = 4–5 mice/group and error bars are s.d. Statistical comparisons were by Mann–Whitney (a–f and h–l) and Student’s t-test (g), *P < 0.05, **P < 0.01. m Survival after intranasal challenge with S. pneumoniae (n = 12–13 mice/group), groups were compared using the log-rank (Mantel–Cox) test, *P < 0.05. n S. pneumoniae and o K. pneumoniae burden in the lung 12 h post-intranasal inoculation. Indicated groups were intranasally administered a GM-CSF-neutralizing antibody, or isotype control (10 μg/mouse) (n, o), rGM-CSF, or vehicle control (5 μg/mouse) (p). q S. pneumoniae burden in the lung 12 h post-intranasal inoculation. Indicated groups were intranasally administered a GM-CSF- neutralizing antibody, or isotype control (10 μg/mouse) concomitant with bacterial challenge in the lung. For microbiota transfer, mice were intranasally inoculated with 10 μL of upper respiratory tract lavage fluid and orally inoculated with 200 μL of fecal suspension 72 h prior to S. pneumoniae infection. Each point represents a single mouse and horizontal lines indicate median values (a–d and n–q). Statistical comparisons were made by Student’s t-test with a post-hoc Sidak–Bonferroni correction for multiple comparison (n), and one-way ANOVA with post-hoc Sidak’s test (o–q) *P < 0.05, **P < 0.01, and NS, not significant. r Survival after intranasal challenge with S. pneumoniae ± rGM-CSF (5 μg/mouse daily) (n = 10 mice/group) groups were compared using the log-rank (Mantel–Cox) test, *P < 0.05

Fig. 2

The microbiota primes GM-CSF production during infection through IL-17A. a–c IL-17A levels in lung tissue 12 h post-inoculation with K. pneumoniae (a) and S. pneumoniae (b, c). For microbiota transfer, mice were intranasally inoculated with 10 μL of upper respiratory tract lavage fluid and orally inoculated with 200 μL of fecal suspension 72 h prior to S. pneumoniae infection. d–g GM-CSF levels in lung tissue 12 h post-inoculation with K. pneumoniae (d, f) and S. pneumoniae (e, g). Indicated groups were intranasally administered an IL-17A-neutralizing antibody (100 μg/mouse via intraperitoneal injection), or isotype control 72 h prior to, and concomitant with, intranasal infection. h, i GM-CSF levels in lung tissue 12 h post-inoculation with K. pneumoniae (h) and S. pneumoniae (i). Indicated groups of mice were treated intranasally with rIL-17A, or vehicle control (5 μg/mouse) concomitant with bacterial challenge. Data (a–i) are from n = 5–6 mice/group and error bars are s.d. Statistical comparisons were by Mann–Whitney (a, b, d–g), or by one-way ANOVA with post-hoc Sidak’s or Dunnett’s test (c, h and j), *P < 0.05, **P < 0.01, ***P < 0.001, and NS, not significant

Fig. 3

The microbiota promotes alveolar macrophage-mediated clearance of bacteria from the lung and the antibacterial activity of these cells is promoted by GM-CSF signaling via the ERK pathway. a, b Myeloperoxidase (MPO) levels in the lung 12 h post-inoculation with S. pneumoniae (a) and K. pneumoniae (b). Statistical comparisons were by Mann–Whitney, data are from n = 5 mice/group and error bars are s.d. (a, b). c S. pneumoniae and d K. pneumoniae burden in the lung 12 h post-intranasal inoculation. Indicated groups were administered liposome clodronate, or empty liposomes, 48 h prior to lung infection. rGM-CSF (5 μg/mouse) was given concomitant with intranasal inoculation with bacteria. Each data point represents a single mouse and horizontal lines indicate median value. e, f Alveolar macrophage killing of S. pneumoniae (e) or K. pneumoniae (f). Indicated groups of alveolar macrophages were treated with rGM-CSF (100 ng/mL) for 1 h and then either U0126 for 1 h, SB203580 for 1 h, SP600125 for 1 h, DPI for 30 min, or inhibitor vehicle control (DMSO, denoted by “−”) for 1 h prior to incubation with bacteria. Inhibitor concentration is displayed below each bar. Values are from n = 4 mice/group and error bars are s.e.m. Statistical comparisons were by one-way ANOVA with post-hoc Sidak’s or Dunnett’s test as appropriate (c, e and f) and Kruskal–Wallis test with Dunn’s multiple comparison test (d), *P < 0.05, **P < 0.01, ***P < 0.001, and NS, not significant

Fig. 4

Potent NLR-stimulating bacteria in the upper airway and intestine regulate lung immunity through GM-CSF signaling. a S. pneumoniae lung burden 12 h post-intranasal inoculation. Indicated groups were orally inoculated with TLR ligands (LPS 50 μg and P3C 50 μg) or NLR ligands (MDP 50 μg and MurNAcTriDAP 50 μ μg) 48 and 24 h prior to lung infection. Antibody treatment was as described in Fig. 1m. Statisitcal comparisons were by Kruskal–Wallis test with Dunn’s test (a). b, c NOD2-dependent SEAP production by HEK293 cells 30 h post-stimulation with individual members of the microbiota (b), or bacterial consortia at an MOI of 1:10 (c). Values are from 7 to 16 biological replicates and error bars are s.d. (b, c). d, e S. pneumoniae (d) and K. pneumoniae (e) lung burden 12 h post-intranasal inoculation. Mice were orally inoculated with indicated bacterial consortia (~5 × 10⁸ CFU) 48 and 24 h prior to intranasal infection. f GM-CSF levels in lung tissue 12 h post-inoculation with K. pneumoniae, data are from n = 5–8 mice/group and error bars are s.d. g 16 s rRNA gene copies in feces 3 days after oral inoculation with indicated consortia. h S. pneumoniae lung burden 12 h post-inoculation in outbred Swiss germ-free mice, and Swiss mice with a microbiota. Indicated groups were orally inoculated with bacterial consortia (10⁸ CFU) 72 h prior to S. pneumoniae infection. i S. pneumoniae lung burden 12 h post-intranasal inoculation. Indicated groups of mice received the complete microbiota as described in Fig. 1p or the “High Nod2-stimulating” consortia as described in Fig. 4d. j S. pneumoniae lung burden 12 h post-intranasal inoculation. Indicated groups of mice were intranasally administered 10⁶ CFU of indicated bacteria into the upper airway 48 h prior to inoculation with S. pneumoniae into the lung. Antibody treatment was as described in Fig. 1m. k GM-CSF levels in lung tissue 12 h post-inoculation with S. pneumoniae. l GM-CSF levels in lung tissue of wild-type and Nod2 ^−/− mice 12 h post-intranasal inoculation with K. pneumoniae LPS (1 μg/mouse), data are from n = 5–8 and error bars are s.d. Indicated mice were orally inoculated with indicated bacterial consortia as described in Fig. 4d. For upper airway colonization, indicated mice were intranasally inoculated with 10⁶ CFU “High Nod2-stimulating” S. aureus and “Low Nod2-stimulating” S. gallinarum. Indicated groups were treated intranasally with rIL-17A, or vehicle control (5 μg/mouse) concomitant with K. pneumoniae LPS inoculation. Statistical comparisons were by Student’s t-test, with a post-hoc Sidak–Bonferroni correction for multiple comparison where appropriate (d–k), and one-way ANOVA with post-hoc Sidak’s test (l), *P< 0.05, **P < 0.01, ***P < 0.001, and NS, not significant. Each point represents a single mouse and horizontal lines indicate median values (a, d, e, g–j)