Muc5b Is Required for Airway Defense

Abstract

Respiratory surfaces are exposed to billions of particulates and pathogens daily. A protective mucus barrier traps and eliminates them via mucociliary clearance (MCC)^1,2. However, excessive mucus contributes to transient respiratory infections and to the pathogenesis of numerous respiratory diseases¹. MUC5AC and MUC5B are evolutionarily conserved genes that encode structurally related mucin glycoproteins, the principal macromolecules in airway mucus^1,3. Genetic variants are linked to diverse lung diseases^4-6, but specific roles for MUC5AC and MUC5B in MCC, and the lasting effects of their inhibition, are unknown. Here we show that Muc5b (but not Muc5ac) is required for MCC, for controlling infections in the airways and middle ear, and for maintaining immune homeostasis in the lungs. Muc5b deficiency caused materials to accumulate in upper and lower airways. This defect led to chronic infection by multiple bacterial species, including Staphylococcus aureus, and to inflammation that failed to resolve normally⁷. Apoptotic macrophages accumulated, phagocytosis was impaired, and IL-23 production was reduced inMuc5b^−/− mice. By contrast, in Muc5b transgenic (Tg) mice, macrophage functions improved. Existing dogma defines mucous phenotypes in asthma and chronic obstructive pulmonary disease (COPD) as driven by increased MUC5AC, with MUC5B levels either unaffected or increased in expectorated sputum^1,8. However, in many patients, MUC5B production at airway surfaces decreases by as much as 90%^9-11. By distinguishing a specific role for Muc5b in MCC, and by determining its impact on bacterial infections and inflammation in mice, our results provide a refined framework for designing targeted therapies to control mucin secretion and restore MCC.

Mucosal surfaces are central interfaces between organisms and their external environments. Mucus-coated barriers defend against pathogens¹² and re-distributed commensal organisms¹³. Gastrointestinal (GI) mucins prevent Helicobacter pylori growth¹², colitis¹⁴, and colorectal carcinogenesis¹⁵. To test whether secreted airway mucins serve correspondingly significant roles, we examined MCC and responses to bacterial infections in Muc5ac^−/−16,Muc5b^−/−, and Muc5b^Tg mice (Extended Data Fig. 1a-d). We identified unique mechanisms by which Muc5b mediates effective respiratory mucosal defense (Extended Data Fig. 2). In Muc5b^−/− upper airways, olfactory gland glycoconjugates were absent, but nasopharyngeal surfaces were unaffected (Fig. 1a, b). Muc5ac and Muc5b were lacking in respective knockout airways, but tracheobronchial glycoconjugates increased inMuc5b^−/− mice due to induced Muc5ac (Fig. 1c, d, Extended Data Figs. 1f and 3). Despite retaining mucous phenotypes in many airway tissues, growth and survival were impaired in Muc5b^−/− animals (Fig. 1e, f) while Muc5ac^−/− and Muc5b^Tg mice survived normally (Fig. 1f). Acute MCC was normal in Muc5ac^−/− and Muc5b^Tg mice, but severely reduced in Muc5b^−/− mice (Fig. 1g, h and Extended Data Fig. 1e), even though functional ciliated cells were present (Fig 1d, i). Mucus transport was impaired in Muc5b^−/− tracheal epithelial cells in vitro, confirming that defective clearance reflected altered mucociliary interactions specifically (Fig. 1j-l and Extended Data Videos 1, 2). Collectively, these data identify non-redundant protective requirements for Muc5b in survival and MCC.

Figure. 1. Muc5b is required for survival and particle clearance.

a-d, Muc5b^+/+, Muc5b^−/−, and Muc5ac^−/− olfactory glands (a), nasopharynxes (b), tracheal glands (c), and bronchial surface epithelial cells (d) stained with AB-PAS or with anti-Muc5b (c,d bottom). Solid arrowheads, mucous cells. Open arrowheads, ciliated cells. e-i, Effects of mucin expression on growth, survival, MCC, and ciliary function. j-l, Microsphere movement, transport velocity, and transport index in primary tracheal epithelial cells in vitro. Scale bars, 20 μm (a, j), 10 μm (b-d). Error bars, sem. ^‘*’ significant differences (p<0.05). Numbers in parentheses, ‘n’ mice.

Impaired MCC in Muc5b^−/− mice was accompanied by abnormal breathing (Fig. 2a, b) and hypoxemia (Fig. 2c). We assessed lung function in mechanically ventilated mice. Baseline airflow and responses to the bronchoconstricting agent methacholine were normal in the lower airways (Fig. 2d-f). Circumventing the upper airways restored ventilation in spontaneously breathing tracheostomized animals (Fig. 2g, Extended Data Videos 3, 4). Thus, upper respiratory obstruction impeded airflow in Muc5b^−/− mice. Micro-CT confirmed this with radiological evidence of upper airway obstruction (Fig 2h), and middle ear effusion consistent with otitis media (OM) (Extended Data Fig. 4a, b). The latter was unexpected given associations between increased MUC5B and human OM¹⁷. In Muc5b^−/− mice, but not Muc5ac^−/− mice (data not shown), hair fragments encased in mucus-like material were consistently found in posterior nasopharynxes (Fig. 2i, j, Extended Data Fig. 4c) and middle ears (Fig. 3a and Extended Data Fig. 4d). Bacteria and inflammation in Muc5b^−/− middle ear lavage samples confirmed infectious OM (Fig. 3b, c).

Figure. 2. Muc5b deficiency causes severe upper airway obstruction.

a-b, Airflow and ventilation in Muc5b^+/+ and Muc5b^−/− mice. c, Oxygen saturation (SaO₂) and rescue with 100% O₂ supplementation (FiO₂ 1.0). d-f, Lung (R_L) and lower airway (R_AW) resistance and hyperresponsiveness to methacholine (MCh) in mechanically ventilated mice. g, Upper airway by-pass by tracheostomy under normoxia (FiO₂ 0.21). h, Upper airway obstruction in Muc5b^−/− mice confirmed by micro-CT (red arrows, nasopharynxes). i,j, Hair encased in mucus-like plugs visible during necropsy. Scale bars, 1 mm. Error bars, sem. ^‘*’, significant difference (p<0.05). Numbers in parentheses, ‘n’ mice.

Figure. 3. Infection is the cause of death in Muc5b^−/− mice.

a-c, Hyperemic blood vessels (red arrows), hair (white arrows), bacteria, and leukocytes (>95% neutrophils) in Muc5b^−/− middle ears. d,e, Hair fragments and leukocytes in Muc5b^−/− lungs. f,g, Chronically accumulating bacteria in lungs, leading to moribundity and septicemia in Muc5b^−/− mice. h-j, Mortality, infection, and ventilation with antibiotic treatment. k, Bacterial genera, and staphylococcal species in Muc5b^−/− mice. l, Survival following intranasal S. aureus inoculation (10⁷ CFU/mouse). Scale bars, 20 μm (d,e). Error bars, sem ^‘*’ statistical significance (p<0.05). Numbers in parentheses, ‘n’ mice.

In Muc5b^−/− lower airways, aspirated materials and inflammatory infiltrates were also common (Fig. 3d, e). Culturable bacteria in the lungs increased 2.9-21.6 fold over time, and 7.6-75 fold further in spontaneously moribund mice who also had increased bacteria in spleen cultures (Fig. 3f, g), suggesting that disseminated infections contributed to mortality. To test this hypothesis, mice were placed on antibiotic-supplemented diets. Antibiotics reversed spontaneous lethality and reduced lung bacterial burden, but did not restore normal ventilation in Muc5b^−/− mice (Fig. 3h-j). Thus, the cause of death in Muc5b^−/− mice was infectious, and not directly due to airflow limitation. These data identify a natural course in which Muc5b deficiency causes upper airway obstruction, spontaneous infection of connecting auditory tubes and the lungs, and fatal bacteremia.

Because of the importance of bacterial infection in Muc5b^−/− mice, we identified organisms whose accumulation in the lungs was normally prevented by Muc5b-rich mucus. 16S rRNA phylotyping revealed 19 genera and >25 species in the lungs. Most were oral, gut, and skin microbiota. At 3 mos age, populations were heterogeneous and indistinguishable (Fig. 3k and Extended Data Fig. 5). However, in moribund Muc5b^−/− lungs, streptococci and staphylococci increased significantly (Fig. 3k). One streptococcal species, (S. acidominimus) increased in abundance but was also frequently found in controls (Extended Data Fig. 5f-h). By contrast, Staphylococcus aureus, an important pneumonia-causing pathogen, selectively increased in spontaneously moribund Muc5b^−/− mice (150-389 fold; Fig. 3k). Its incidence was 78% in moribund mice compared to 25% in 3 mo old Muc5b^−/− mice, and 8% in Muc5b^+/+ mice (χ²p=0.002). Moribund mouse spleen cultures contained bacteria also found in the lungs, including S. aureus (Extended Data Fig. 6). Based on these associations, we tested the effects of mucin expression on pathogenesis following airway inoculation with S. aureus USA300¹⁸. Though non-lethal in Muc5b^+/+, Muc5b^Tg, and Muc5ac^−/− animals, only 40% of Muc5b^−/− mice survived (Fig. 3l, Extended Data Fig. 7a-e). Thus, Muc5b is selectively required for preventing acquisition and enrichment of virulent bacteria such as S. aureus in the airways.

The airways are exposed to bacteria from birth onward. MCC is considered the major route for their elimination. However, although Muc5b^−/− mice showed severely impaired MCC, most survived several months with increased lung bacterial burden (Figs. 1f and 3f, h), but were ultimately unable to control S. aureus infection (Fig. 3k, l). Therefore, we investigated additional potential mechanisms forMuc5b-dependent airway defense. Purified MUC6 (but not MUC5AC) inhibits H. pylori growth¹², and purified MUC5AC (but not MUC2) inhibits Trichuris muris survival¹⁶. However, neither MUC5AC nor MUC5B inhibited S. aureus growth (Extended Data Fig. 7f), suggesting that Muc5b-mediated defense was not mediated by direct inhibitory interactions with S. aureus.

We then investigated whether Muc5b deficiency affected inflammatory defenses. Leukocytes in Muc5b^−/− lung lavage fluid were significantly altered. In the steady state, neutrophils and eosinophils accumulated, but lymphocytes were absent (Extended Data Fig. 8a). There were also striking changes in macrophages in Muc5b^−/− lungs. Macrophages accumulated over time (Fig. 4a). They frequently contained undigested cytoplasmic inclusions (Fig. 4b), and had impaired phagocytic functions (Extended Data Fig. 9a). By age 12 mos., IL-23, a mediator of anti-microbial inflammatory responses normally produced by macrophages and dendritic cells, was reduced by 93% in Muc5b^−/− mice (Fig. 4c, Extended Data Table 1). As Muc5b^−/− mice became moribund over this timecourse, dysfunctional inflammatory phenotypes progressed (Fig. 4d, Extended Data Fig. 8a). Macrophages that accumulated in airspaces were increasingly apoptotic (Fig. 4e, f). IL-23 remained significantly lower in moribund Muc5b^−/− mice than controls (Fig. 4g), despite their severely infected states (Fig. 3f, k). We found no evidence that macrophages produce Muc5b (Extended Data Fig. 9b), so it is unlikely that these pathologies were cell autonomous. Rather, the ability of Muc5b^−/− lungs to resolve inflammation following microbial exposure was significantly inhibited⁷.

Figure. 4. Muc5b maintains functioning lung macrophage populations.

a-c, Temporal changes in lung lavage macrophage numbers and morphologies and in IL-23. d-g, Macrophage numbers and apoptosis, and IL-23, in spontaneously moribund and control lungs. h-j, Macrophage numbers and apoptosis, and IL-23, in young (3 mos) and old (6-7 mos) lungs 7d after intratracheal USA300 infection (10⁸ CFU/mouse). j-l, IL-23 and macrophage MHC-II in infected Muc5b^Tg mice. Muc5b^Tg lavage USA300 DNA was >98% reduced; incidence was 25% vs. 67% (Muc5b^+/+) and 100% (Muc5b^−/−), χ²p=0.03. Error bars, sem. ^‘*’, statistical significance (p<0.05). Representative data in k. Numbers in parentheses, ‘n’ mice.

Apoptotic macrophage accumulation in Muc5b^−/− mice suggests that in addition to acute microbial elimination, MCC is required for chronically maintaining phagocyte clearance and immune functions in vivo. We thus tested the impacts of age and Muc5b expression on acute and resolving lung inflammation following S. aureus infection. In young animals, acute neutrophilic responses were similar irrespective of genotype (Extended Data Fig. 8b). By contrast, aged Muc5b^−/− animals were exquisitely sensitive, exhibiting 100% mortality within the first few hours of infection, while Muc5b^+/+ and Muc5b^Tg mice survived (Fig. 4h). Thus, the combined absence of MCC and the chronic effects of underlying spontaneous infections enhanced the deleterious consequences of S. aureus infection. Young mice survived to day 7 post-infection, a peak timepoint of resolving inflammation⁷. Infected Muc5b^−/− lungs exhibited increases in lung macrophages and mixed granulocytes (Fig. 4h and Extended Data Fig. 8c). Accumulated macrophage pools were predominated by apoptotic cells (Fig. 4i) and linked to significantly reduced lung IL-23 (Fig. 4j), similar to spontaneously moribund Muc5b^−/− mice. Conversely, in Muc5b^Tg mice, IL-23 production, macrophage activation, and S. aureus elimination were enhanced (Fig. 4j-l).

Collectively, these data show that Muc5b expression affects inflammation and macrophage-mediated responses in the lungs. In its absence, mice failed to mount effective anti-bacterial responses. In its sustained presence, these were enhanced. Apoptotic cells, including neutrophils, accumulated in MCC deficient Muc5b^−/− airspaces (Fig. 4e, Extended Data Fig. 9c). Efferocytosis of neutrophils is reported to down-regulate IL-23¹⁹, which may explain its decline in the lungs. It will be of great interest to test the degrees to which Muc5b-dependent changes in macrophages are mediated by altered elimination due to impaired MCC, and by other processes such as opsonization or pharmacologic interactions²⁰. Nonetheless, these studies functionally link two fundamental first line host defense components – mucus and phagocytes. To the best of our knowledge, Muc5b is the only bronchial secretory protein identified to date demonstrating such singularly strong biological necessity. Mice survive normally in the absence of Muc5ac or anti-microbial peptides such as βdefensin-1²¹, calgranulin B²², lipocalin-2²³, or lactoferrin²⁴. In addition to controlling MCC and macrophage readiness, it is conceivable that Muc5b is a scaffold for these and other defensive molecules.

The novel relationships between Muc5b, MCC, and inflammation reported here have the potential to impact therapies for airway diseases. Cystic fibrosis, COPD, and asthma are associated with mucus hypersecretion, impaired MCC, and increased risk for respiratory infections^25-29. MUC5AC and MUC5B expression varies among individuals, but the overall result is excessive mucin production^1,8-11. Such pathological associations have overshadowed the benefits of airway mucus and the potential functions of individual mucins in infections^{1,12,13,15,16,30}. Human MUC5B is highly polymorphic. A recently identified promoter variant found in ~20% of the general population increases its expression 37.4-fold in healthy lungs⁵. Our findings in Muc5b deficient and overexpressing mice suggest that MUC5B variants may regulate airway homeostasis, disease pathogenesis, and mucosal immune function in humans broadly. While controlling mucin hypersecretion is an attractive therapeutic goal for transient and chronic lung diseases¹, a therapy that completely disrupts mucus and inhibits MUC5B for extended periods may not be advisable. Rather, promoting adequate expression and enhancing MCC and airway defense in controlled manners are better strategies.

Methods Summary

Mice were used with Institutional Animal Care and Use Committee approvals. Muc5ac^−/− mice were generated previously¹⁶. Muc5b^−/− and Muc5b^Tg mice were generated here. Muc5b protein was assessed immunohistochemically using rabbit polyclonal antisera. Ciliary beat, MCC, and transport were assessed as described previously. Lung function was measured using a head-out plethysmograph and a flexiVent (Scireq, Montreal, Quebec, Canada), and blood oxygen was assessed using a pulse oximeter. Otitis media was assessed by visual otoscopy and middle ear lavage (MEL). Pulmonary inflammation was assessed by histology and lung lavage. Lavaged leukocytes were identified by light microscopy and flow cytometry. Neutrophils, macrophages, MHC-II, and apoptotic cells, were detected using commercially available Ab’S and reagents. S. aureus was administered by 10 μl intranasal or 50 μl intratracheal inocula at 10⁷-10⁸ CFU/animal. Bacteria and bacterial DNA were isolated from MEL, lung homogenates, and lung lavage pellets. Isolated colonies were phylotyped by 16S rRNA and mecA sequencing. Kaplan-Meier (1f and 3h, l), regression (1e and 2f), one-sided t-test (1g-i, k, l; 2b-e, g; 3b, c, f, g, j, k, and 4c, d, f, g, i, j), and one-way ANOVA (3i and 4a, h, j, l) with appropriate corrections for multiple comparisons, unequal variances, and non-Gaussian distribution were carried out using GraphPad Prism v5.04 (GraphPad Software, Inc., La Jolla, CA). Full methods are found in Supplementary Information.

Methods

Mouse husbandry

Male and female specific pathogen free, 6-8 week old C57BL/6J mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Mice were housed in accordance with the Institutional Animal Care and Use Committees of the MD Anderson Cancer Center, the University of Colorado School of Medicine, and the University of North Carolina-Chapel Hill. Mice were kept in specific pathogen free housing areas that were maintained by institutional animal care staff. Mice were fed ad libitum either a normal diet of water and irradiated chow (Harlan Teklad, Indianapolis, IN), or in specified experiments, the same diet supplemented with trimethoprim and sulfadiazine (275 and 1,365 mg/kg chow, respectively) and enrofloxacin (2.7 mg/ml drinking water). Mice used for survival analysis were assigned prior to genotype identification and were studied up to 12 months age. Moribund mice were identified by observing changes in body weight and behaviors. In accordance with the veterinary care procedures at MD Anderson and UC Denver, a loss of >15% of body weight without recovery, hunching, fur ruffling, and lethargy were used as criteria for determining moribundity. For all other studies, mice were assigned non-descriptive numbers at weaning. Aside from knowledge of a particular strain background, genotypes were unknown during experiments and post-experimental analyses. These criteria held during inflammation, microscopy, and bacterial species identification studies, which were performed by different investigators. Male and female mice were used in all studies. Animals were euthanized by exsanguination under anesthesia or asphyxiation by CO₂ inhalation.

Genetically engineered mice

Muc5ac knockout mice were generated previously¹⁶. Muc5b knockout mice were generated using a targeting allele with LoxP sites flanking exon 1 (fig S1). The targeting allele was linearized with NotI and electroporated into CJ7 embryonic stem cells at the MD Anderson Cancer Center Genetically Engineered Mouse Facility. Homologous insertion of the targeting allele was confirmed by long-range PCR, restriction digest, and DNA sequencing (fig S1). One targeting allele-positive cell line was expanded and inserted into albino C57BL/6J blastocysts to generate chimeric founders. CJ7 cells are derived from a 129S1 mouse with an agouti coat color. Chimeric males were mated with C57BL/6J females, and agouti progeny were screened for targeting allele transmission. Positive mice were subsequently bred with Rosa26-Flpe recombinase knock-in animals to remove the neomycin selection cassette. Progeny were crossed with CMV-Cre transgenic mice to generate full knockout animals, resulting in animals that were heterozygous mosaics for Muc5b (Muc5b^Δ/−) in all tissues, including the germline. Germline transmission was confirmed by breeding Cre positive Muc5b^Δ/− and wild type C57BL/6J mice. Muc5b recombinant progeny were thus identified as Muc5b^Δ/+. F1 Muc5b^+/− mice were intercrossed, and resulting litters were viable with typical Mendelian inheritance distributions. Mice were continuously backcrossed by breeding heterozygotes with wild type C56BL/6J mice for >5 generations using strain-specific microsatellite marker facilitated analysis (speed congenics) to confirm congenic status (>99%). For the studies reported here, Muc5b and Muc5ac targeted mice were maintained by heterozygous crossing using littermates for comparisons.

Muc5b transgenic mice were generated by insertion of the full length 34 kb genomic coding region of the mouse Muc5b gene into a rat Scgb1a1 promoter-IRES-AcGFP transgenic targeting cassette using recombinegenic cloning³¹. Founders were generated by injecting the targeting vector into C57BL/6N pronuclei at the National Jewish Health transgenic mouse core. Due to the size of the DNA, only 2 lines of fertile mice were made. One line demonstrated significant overproduction of Muc5b protein by Western blot, and all signals were >1 MDa on agarose gels, indicating that both native and transgenic Muc5b were extensively glycosylated. Expression was also examined immunohistochemically, and extensive expression throughout the lungs was found (Extended Data Fig. 1). These mice were maintained by continuous hemizygous outcrosses with C57BL/6J mice.

Immunohistochemistry

Loss of expression of Muc5b was confirmed immunohistochemically using rabbit-anti-mouse Muc5b antisera as described previously^32,33. The peptide used to generate this antisera is repeated 5 times in the full length protein, and the DNA regions encoding it are >20-25 kb 3’ to the excised region in Muc5b^−/− mice. Tissues from wild type mice demonstrated expression in the nasal turbinates, tracheal submucosal glands, and throughout the bronchial tree (see Fig. 1). No signal was detected in tissues from knockout mice, thus indicating that disruption of exon 1 was sufficient to abolish Muc5b protein production.

To detect cytospun neutrophils, macrophages, and apoptotic cells, NIMP-R14 (1:500, Abcam, Cambridge, MA) followed by Alexa-594 conjugated anti-rat IgG (Life Technologies, Carlsbad, CA), FITC conjugated F4/80 (1:100, BM8, eBioscience, San Diego, CA), and polyclonal anti-activated caspase 3 (1:500, ab13847, Abcam) followed by Alexa-647 conjugated anti-rabbit IgG (Life Technologies) were used, respectively. NIMP-R14 was also used to individually assess neutrophil accumulation in tissue sections using HRP-conjugated secondary antisera followed by DAB for detection. Antibodies were diluted in 0.01% Tween 20 diluted in PBS. Hoechst 33342 or hematoxylin was used to counterstain nuclei. Cells and tissues were imaged on an upright Olympus BX60 or BX63 microscope and imaged using Olympus DP70 and DP72 cameras (Olympus America, Center Valley, PA). Subsequent analysis of colocalization in fluorescent stained cells was made using cellSens digital imaging software (Olympus). Isotype control antibodies were used to confirm specificity of immunolabeling experiments.

Mucociliary clearance in vivo

Tracheal ciliary beat frequency was measured as described previously³⁴. Mucociliary clearance was measured by assessing the elimination of fluorescent microspheres from the nose as described previously³⁵ and from the lungs using a method adapted from Coote et al³⁶. For pulmonary clearance 5 × 10⁵ yellow-green fluorescent carboxylate-modified microspheres (Fluospheres, Life Technologies, Carlsbad, CA) were instilled intratracheally using a Penn-Century microsprayer via a transoral route in urethane anesthetized mice (2 g/kg, ip). Mice were euthanized by exsanguination immediately or 15 min after Fluosphere administration. Lungs and trachea (up to but not including the larynx) were excised, placed in vials containing PBS/0.1% Tween 20 and 1 mm diameter glass beads, and homogenized by agitating at 4,800 oscillations/min for 10 s using a Mini-Beadbeater-1 (Bio-Spec Products, Bartlesville, OK). Homogenized material was placed on a hemacytometer, and beads remaining in the lungs were quantitated under epifluorescence on an Olympus BX60 microscope (Olympus America, Inc., Center Valley, PA). Baseline deposition was assessed by averaging the numbers of Fluospheres detected in the lungs in mice euthanized immediately after installation. Out of 5 × 10⁵ beads administered, means ± sem’s of 3.98±0.50 × 10⁵ beads were observed in Muc5b^+/+ animals, and 3.66±0.4.3 × 10⁵ beads were observed in Muc5b^−/− animals at time zero (79.6% and 73.2% deposition fractions, respectively). These baseline values were not significantly different (t-test p-value = 0.64). Numbers of beads remaining in mice euthanized 15 min after installation were then measured. Fractional clearance was calculated by subtracting the ratio of the numbers of beads remaining in the lungs at 15 min over the average at time zero for each genotype (n = 8-11 for time zero data). This ratio was subtracted from 1 and the difference was multiplied by 100 to calculate percent clearance.

Mucociliary transport in vitro

Mucociliary transport was measured in vitro by culturing mouse tracheal epithelial cells (mTEC’s) as described previously^37,38. For these studies cultures were grown on collagen coated 4.67 cm² transwell inserts in 6 well plates. Five tracheas were used to derive epithelial cells to seed each well at a density of 5 × 10⁵ cells/well. Confluent cultures were transitioned to air-liquid interface and maintained for an additional 21-28 days. Mucociliary transport was assessed by adapting methods previously used in human airway epithelial cell cultures³⁹. Briefly, red carboxylate-modified Fluospheres (4 μm diameter) were diluted to a concentration of 2 × 10³/μl in 300 mOsm mannitol⁴⁰. These were added to the centers of transwells in a volume of 5 μl and placed in a 37°C incubator for 30 min to allow bead dispersion and airway surface liquid normalization. Wells were observed on an Olympus IX-81 inverted microscope under red fluorescence. Time lapse images were acquired at 30 s intervals under 10 × magnification for 15 min and were analyzed using Slidebook 4.0 software (Olympus). Transport distance and displacement measured for all particles in two separate mTEC cultures from 10 Muc5b^+/+ and 10 Muc5b^−/− mice. Values were calculated for 124 Fluospheres in Muc5b^+/+ and 95 Fluospheres Muc5b^−/− cultures, respectively. Data were analyzed for particles that traveled a distance greater of than 4 μm (the diameter of a single bead) to exclude effects of vibration and Brownian motion.

Mouse physiology

Ventilation was measured in conscious mice using a head out plethysmograph (Scireq, Montreal, Canada). Tidal volume, respiratory rate, and minute ventilation were measured for each animal. Mice were restrained in chambers separated with a custom fit neck seal made of 0.02” thick silicone rubber (35A durometer rating; McMaster-Carr, Robbinsville, NJ). Plethysmography was assessed in real time and recordings were made after a 2-5 min acclimation period. Data were averaged over 5-10 min periods in which resting breath patterns were observed.

Arterial blood oxygen saturation was assessed in mice using a MouseOx mouse pulse oximeter (STARR Life Sciences Corp, Oakmont, PA). For studies in conscious mice, fur around the neck was removed with clippers in order to use a collar clip LED sensor. Percent oxygen saturation (SaO₂) of functional arterial hemoglobin was measured in room air and in the presence of 100% O₂ (fraction of inspired oxygen, FiO₂ = 1.0). Data were analyzed from n = 4 Muc5b^+/+ and n = 5 Muc5b^−/− mice. For arterial oxygen saturation measurements pre and post tracheostomy, mice were anesthetized by inhalation of ~2% isoflurane in room air and SaO₂ was measured with the MouseOX using a thigh clip. When a stable reading was obtained, the anesthetized mouse was tracheostomized by exposing the trachea and cutting a ~1 mm opening with micro scissors just below the larynx, through which the mouse breathed. The SaO₂ was again measured following the tracheostomy. Despite their chronic hypoxemic states, which would be expected to increase hematocrit, Muc5b^−/− mice had slightly reduced circulating red blood cell numbers compared to Muc5b^+/+ littermates (~7% decrease), consistent with the development of anemia of chronic disease^41-43.

Pulmonary function was analyzed using a flexiVent (Scireq, Montreal, Canada). Mice were anesthetized with urethane (2.0 g/kg, i.p.). This dose is sufficient to provide 6-8 h of sedation, though experiments lasted only 20-30 min. Mice were tracheostomized with a blunt beveled 18 G Luer stub adapter, and once ventilated (150 breaths/min, 10 ml/kg, against 3 cmH2O positive end expiratory pressure - PEEP), they were paralyzed by continuous infusion of succinylcholine chloride (10 mg/kg/min, i.p.). Total lung resistance (R_L) and airway resistance (R_aw) were assessed at baseline and in response to successive increasing methacholine (MCh) doses (0.1, 0.3, 1, 3, and 10 mg/ml) administered by an in-line ultrasonic nebulizer. At the highest dose of MCh (10 mg/ml) heart rates assessed by electrocardiogram often fell below 100 bpm and did not recover. At this point, no further doses responses were performed.

Micro-CT imaging

Mice were anesthetized with isoflurane, positioned on a warming bed and the head imaged with a GE eXplore CT120 scanner, 70 kV, 50 mA, 100 μm isotropic scans. All mice recovered uneventfully from this procedure.

Examination of middle ear inflammation and infection

To visualize the ear drum in vivo, the mouse was anesthetized with isoflurane and positioned under a dissecting microscope. An “otoscope” was fashioned using a pipette tip, the large end of which was interfaced with a fiberoptic light. The small end of the pipette tip, cut to a diameter of ~1 mm, was inserted into the external auditory ear canal. Using this procedure the ear drum was visualized and photographed.

For middle ear lavage (MEL), mice were euthanized by exsanguination under anesthesia. The mandible was removed. Tympanic bullae were exposed by blunt dissection and thoroughly wiped with sterile cotton swabs and betadine to sterilize the external surface. A small incision was made on the ventral side of the tympanic bulla using a 23G needle and a blunt 30G needle was inserted. Middle ear lavage was performed by instilling and retrieving 2 × 10 μl of sterile PBS. MEL total cells were counted with a hemocytometer, cytospin slides were obtained (StatSpin Cytofuge 2, Norwood, MA), air dried, and stained with modified Giemsa (NewComer Middleton, WI) for differential cell counts of at least 200 cells per slide.

Bacterial, isolation, growth and sequencing

Bacteria were cultured from lungs of mice homogenized as described above for Fluosphere isolation. Aliquots of homogenates of MEL lavage cells were serially diluted and grown on tryptic soy agar plates as described previously⁴⁴ in humidified 37°C incubators under atmospheric oxygen, under nitrogen charged hypoxic (<1% O₂) conditions and in the presence of 5% CO₂, or in a candle jar on Columbia anaerobe sheep blood agar (Becton Dickinson, NJ). Plates were incubated at 37°C for 24 h, and colony forming units (CFU) were enumerated. Colonies were counted 18-24 h after incubation. Ten colonies were picked at random from each sample and grown in 2 ml tryptic soy broth under the aerobic/anaerobic conditions above, except when <10 colonies were present, in which case all colonies were picked and grown. Expanded cultures were pelleted and DNA was isolated using Fast DNA Spin kits (MP Bio, Solon, OH). Ribosomal 16S sequencing was conducted in a 25 μl volume with 2 μl bacterial DNA (Bioline MyTaq^™ HS Red Mix), using forward primer 27F 5’-AGAGTTTGATCMTGGCTCAG and reverse primer 1492R 5’-TACGGYTACCTTGTTACGACTT (at 0.4 pM; where M = A or C; Y = C or T). Cycling conditions were 95°C for 5 min followed by 35 cycles of 95°C for 30 sec, 58°C for 30 sec, and 72°C for 30 sec, followed by one 7 min extension at 72°C. PCR products were purified using ExoSAP-IT (Affymetrix, Santa Clara, CA). Samples were prepared for sequencing (GENEWIZ, Inc., South Plainfield, NJ) by mixing 10μL of each PCR product with 5 μL of the sequencing primer 5’-CCGTCAATTCCTTTRAGTTT. Sequences were aligned using nucleotide BLAST (http://blast.ncbi.nlm.nih.gov/), microbial genomic BLAST (http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi?organism=microb), Ribosomal Database Project (http://rdp.cme.msu.edu/), and Greengenes http://greengenes.lbl.gov/cgi-bin/nph-index.cgi⁴⁵ search tools. Microbial genera were identified in cases in which results demonstrated >97% sequence similarities, and species were identified in cases where sequence similarities were >99%⁴⁶. To confirm findings in some cases, sequencing of the RNase P rnpB gene was amplified using the strF 5’-YGTGCAATTTTTGGATAAT and strR 5’-TTCTATAAGCCATGTTTTGT 3’ primers, and products were sequenced and analyzed as described above. For Staphylococcus aureus pneumonia experiments, bacteria were grown as described previously and administered by 10 μl intranasal inoculations or 50 μl intratracheal bolus challenges¹⁸.

Leukocyte identification and functional analysis

Blood was obtained by cardiac puncture in order to measure hematocrit and circulating leukocytes. No differences were seen in hematocrit in Muc5b^−/− mice compared to Muc5b^+/+ littermates at 12 weeks of age. There was a small increase in circulating leukocyte numbers in Muc5b^−/− mice compared to Muc5b^+/+ mice (7.53 ± 1.62 vs. 5.12 ± 0.12 × 10³μl), but these differences were not statistically significant. To assess lung leukocytes, the lungs were lavaged using 2 × 1 ml PBS introduced via a 20 G tracheal cannula. Cells were counted on a hemacytometer. Samples were centrifuged (300 × g, 5 min), supernatants were stored in liquid N₂, and pellets were resuspended in 1 ml PBS. To identify leukocytes in lavage fluid, cells were cytospun and stained with Wright-Giemsa. In a small number of wild type C57BL/6J mice, an aerosolized lysate of a clinical isolate of Non-typeable Haemophilus influenzae (NTHi) was administered to induce inflammation and cytokine production for use as positive controls in the current studies⁴⁷. In some cases, cells stained with anti-macrophage (F4/80 clone BM8, eBioscience, San Diego, CA) and anti-neutrophil (NIMP-R14, Abcam, Cambridge, MA) markers were used to confirm differential identification. NIMP-R14 recognizes Ly6G and Ly6C antigens (like GR1, see below). NIMP-R14 specificity was tested in lavage from NTHi exposed mice 24 h post exposure. Following NTHi, neutrophils had high fluorescence intensity and typical polymorphonuclear appearances, while macrophages were negative. Staining in Muc5b^−/− lavage samples was performed simultaneously, and image acquisition settings were identical to those used in wild type NTHi exposed samples.

For flow cytometry, mice were lavaged with 5 × 1 ml PBS with 1mM EDTA. Cell suspensions were incubated with Fc blocking antibodies for 20 minutes and then incubated with fluorochrome-conjugated monoclonal antibodies for 30 minutes. Cell were washed twice, fixed with 1% paraformaldehyde and assayed using a BD LSR-II flow cytometer. Data were analyzed with FloJo Software (Tree Star, Ashland, OR). Cells were identified with forward and side scatter and doublets were excluded. Alveolar macrophages were identified with F4/80 (6F12) and CD11c (HL3). Neutrophils were defined as GR1^high CD11b^high F4/80^neg CD11c^neg events. Macrophage activation was assessed with monoclonal antibodies directed at MHCII (In-1), CD86 (GL1) and CD40 (3/23). All antibodies were obtained from BD Biosciences (Franklin Lakes, NJ). Lavage supernatants were analyzed for cytokine production by a commercial service (Aushon Biosystems, Billerica, MA; Extended Data Table 1), and IL-23 was confirmed by ELISA (eBioscience, cat no. BMS6017).

Leukocyte phagocytosis was assessed in separate animals using a method adapted from Steinkamp et al.⁴⁸. Lavage fluid was collected from 12 week old animals as described above, with the exception that RPMI 1640 medium (Life Technologies, Carlsbad, CA) was used for lavage instead of PBS. Lavage volume was normalized to 1.75 ml with RPMI, and 250 μl of PBS containing 10⁷ 2 μm diameter yellow-green Fluospheres was added to each lavage sample and incubated for 30 min in a 37°C shaking incubator under gentle agitation (50 rpm). Samples were then placed on ice for 1-2 min, centrifuged for 10 min at 4° C at 300 × g, washed in 1 ml PBS, and recentrifuged for 10 min at 4°C at 300 × g. The pellet was then resuspended in a 200 μl solution of 10 μg/ml Hoechst 33342 in PBS. Cells were imaged and counted on a hemacytometer under epifluorescence as described above. Due to the presence of an efflux pump in macrophages, 88% of Fluosphere positive cells were Hoechst dye low or negative.

To detect total bacteria and methicillin resistant S. aureus 7 d post-infection, lung lavage cell pellets were analyzed by quantitative PCR as described previously⁴⁹. Total DNA was isolated from pellets using a MO-BIO Soil DNA purification kit, per the manufacturer’s directions. 16S rRNA gene DNA was amplified as described²¹. To identify the presence of the methicillin resistant S. aureus strain used here, a separate SYBR green qPCR detection strategy was employed to quantify mecA gene copy number using the following primers: sense 5’-GATGGTATGTGGAAGTTAGATTGG-3’; antisense 5’-CGGTTTTAAGTGGAACGAAGG-3’). Amplified mecA copy number was normalized to lavage volume and 16S rRNA gene content. Standard curves were generated using purified PCR products, and water was used as a negative control. 16S PCR was sensitive to a lower detection limit of 100 copies, and mecA was sensitive to a single copy.

Statistical analyses

All analyses were carried out using GraphPad Prism v5.04 (GraphPad Software, Inc., La Jolla, CA). Numbers of mice used to detect changes in survival were based upon prior studies from our lab using models of infectious pneumonia and cancer^44,50. All other studies were based upon prior experience using inflammation, infectious pneumonia, mucous clearance, and lung function from our labs^36,51. Means and variances from prior findings in wild type mice served as the basis to determine sample sizes required to detect significant differences at 0.80 power.