Muc5b-deficient mice develop early histological lung abnormalities

Mucus gel is the first line of defense in the lung. Gel-forming mucins are high-molecular-weight macromolecules that make up the major organic component of mucus. These molecules are heavily O-glycosylated and dimerize through their carboxy-terminal region and polymerize through their amino-terminal region. Long polymers are secreted at the cell surface and form mucus gel when in contact with water (Thornton et al., 2008). The central region of gel-forming mucins is enriched with proline and hydroxyl amino acids (Ser/Thr/Pro regions) and is extensively O-glycosylated. Five gel-forming mucins, MUC2, -6, -5AC, -5B and -19, have been cloned and characterized and are highly conserved between humans and rodents. They all contain a large exon encoding the Ser/Thr/Pro region.

In lung, the mucus layer covers and protects the cell surface of the airway epithelium and traps exogenous particles and microorganisms for mucociliary clearance. In humans, MUC5AC is secreted by goblet cells, whereas MUC5B is secreted by submucosal glands. MUC5B is also secreted by the salivary glands, nasal mucosa, lacrimal glands, gallbladder, middle ear, submucosal glands of the trachea and esophagus, and the epithelium and glands of the endocervix. A similar pattern of expression was found in mouse tissues (Portal et al., 2017a; Valque et al., 2011) with the early expression of MUC5B during human development (Buisine et al., 1999, 2000) and in mouse lungs at embryonic day (E)12.5 or earlier (Portal et al., 2017a).

The redundancy of the two gel-forming mucins in the lung make it difficult to understand the precise function of each mucin. Dysregulation of MUC5B expression has been reported in airway diseases (Fahy and Dickey, 2010; Rose and Voynow, 2006). Genetic polymorphism of the human MUC5B promoter sequence has been associated with diffuse panbronchiolitis and mucous hypersecretion (Kamio et al., 2005). A single nucleotide polymorphism in the promoter region of the MUC5B gene has been linked to the development of familial interstitial pneumonia and sporadic idiopathic pulmonary fibrosis (Fingerlin et al., 2013; Noth et al., 2013; Seibold et al., 2011; Stock et al., 2013; Zhang et al., 2011) and it has been suggested that this polymorphism might be associated with overexpression of MUC5B in the lung. More recently, a major function of MUC5B has emerged based on the findings of a unique in vivo study showing that MUC5B but not MUC5AC is essential for mucociliary clearance (Roy et al., 2014).

We generated a mouse strain genetically deficient for Muc5b by deleting exons 12 and 13 of the 49 exons of the gene, exon 31 being the large central exon that codes for the Ser/Thr/Pro region (Desseyn, 2009). Here we report that no homozygous mice deficient for Muc5b were obtained, while heterozygous mice were viable and fertile. Mice with Muc5b haplo-insufficiency displayed early lung inflammation that could lead to respiratory distress. In view of the embryolethality of full gene deletion, lung-restricted Muc5b-deficient mice (homozygous and heterozygous) were generated, which showed abnormalities of bronchial structure that could also lead to respiratory distress.

A targeting construct was developed to flank exons 12 and 13 of the Muc5b gene by the loxP sites (Figs S1 and S2) located at the 5′ part of the gene, upstream of the large exon encoding the Ser/Thr/Pro region. Mice with the floxed Muc5b allele were intercrossed with the Cre deleter transgenic line MeuCre40. Mice carrying the Cre transgene and the Muc5b-floxed allele were backcrossed with C57BL/6 wild-type (WT) mice, and their progeny with the Muc5b-floxed allele but without the Cre transgene were retained and studied. Muc5b^ko/+ mice were fertile. Body mass was identical between Muc5b^ko/+ and control WT mice (Muc5b^+/+). Analysis of the progeny from 19 intercrosses of heterozygous Muc5b^ko/+ mice (41 litters, 292 mice) was not consistent with Mendelian ratios as we observed 80 WT (27.4%) mice, 212 Muc5b^ko/+ (72.6%) mice and no Muc5b^ko/ko progeny (P<0.0001). No homozygous embryos or resorption sites from E8 until birth (six pregnant mothers) were found, suggesting that the complete tissue-disruption of both Muc5b alleles was embryolethal at a very early stage of embryogenesis. We then investigated the pulmonary phenotypes of the heterozygous systemic Muc5b^ko/+ mice and lung-specific Muc5b^ko/ko.

Of 63 Muc5b^ko/+ mice analyzed, 12 (19%) displayed severe respiratory distress between 12 and 22 weeks of age including hunched posture, reduced locomotor activity, polypnoea (Movie 1), a thumping respiration, squeaks and a discreet cough (Movie 2), while neither WT mice nor Muc5b-floxed mice displayed this pulmonary phenotype. Muc5b^ko/+ mice developing respiratory distress displayed a dramatically abnormal lung morphology as illustrated in Fig. 1. Lung sections were stained with Mason's Trichrome stain, which readily identifies, for Muc5b^ko/+ mice, deposition of collagen (blue color), a feature of pulmonary fibrosis. We also noted fibrin deposition, epithelial hyperplasia, interstitial inflammation and collapsed alveoli in comparison to WT mice (Fig. 1). A decrease of interstitial vessel density was observed, as visualized by a reduction in CD31 endothelial positive cells (Fig. 1). The inflammatory cell infiltrate was also increased in Muc5b^ko/+ mice around the bronchi and vessels and was composed of both mononuclear and polynuclear cells (Fig. 1). Airways were lined with cuboidal shaped cells in Muc5b^ko/+ mice showing bronchial metaplasia and hyperplasia (Fig. 2A). Increased Muc5b staining was observed in the bronchi and bronchioles of adult mice suffering from respiratory distress in comparison to WT control mice (Fig. 2B). Obstructive Muc5b-positive material was occasionally observed in the lumen of the bronchioles in Muc5b^ko/+ mice. In WT mice, Club cell secretory protein (CCSP) was present throughout the cytoplasm, whereas it was limited to the apical portion of Club cells in Muc5b^ko/+ mice, consistent with a morphological change in the Club cells (Fig. 2C) in agreement with Boucherat et al. (2012). Total cell number of bronchial epithelium was increased (P=0.008; Fig. 2D) in Muc5b^ko/+ mice with respiratory distress. We then assessed the number of proliferative cells by immunofluorescence using anti-PCNA antibodies; this was significantly higher in Muc5b^ko/+ mice than in Muc5b^+/+ mice (P<0.0001; Fig. S3A,B) in agreement with epithelial cell hyperplasia. Bronchi of Muc5b^ko/+ mice exhibited a decrease of CCSP-positive Club cells (P=0.03; Fig. 2D) with an increase of Muc5b-positive Club cells for Muc5b^ko/+ mice with respiratory distress (P=0.03). No modification of the number of acetylated-tubulin (ACT)-positive ciliated cells were observed (Fig. 2C,D).

In WT mice, elastin fibers, which play a mechanical role in supporting and maintaining the lung tissue structure (Rocco et al., 2009), ran longitudinally along the alveolar walls and were present at the tips of the alveolar septa of WT mice. By contrast, elastin fibers in the lungs of Muc5b^ko/+ mice appeared disorganized and fragmented, suggesting that the pulmonary tissue may be less elastic than in WT mice (Fig. S4A). Tight junctions play an important role in maintaining the epithelial barrier integrity in the lung. Because elevation of expression of tight junction proteins may represent a potential biological marker of lung injury severity (Jin et al., 2013), we examined by immunofluorescence the expression of occludin as one major tight junction protein. Occluding expression was increased in the lung of Muc5b^ko/+ mice supporting airway injury.

To quantify histological changes including meta- and hyperplasia, inflammatory cell filtration of the parenchyma and fibrosis, lung sections were stained and coded, and then blindly scored (Madtes et al., 1999). The mean score was significantly higher (P=0.008) in Muc5b^ko/+ mice with respiratory distress than in WT mice (Fig. 3A). In Muc5b^ko/+ mice, α-smooth muscle actin (ASMA) secreting extracellular matrix component was increased around large (data not shown) and small airways (P=0.004; Fig. 3B,C). ASMA was also accumulated in the alveolar septae (Fig. 3D) suggesting the presence of activated fibroblasts. Overall, these data suggest that deletion of one Muc5b allele may lead to severe respiratory distress, lung damage and airway remodeling and affects lung histology with hyperplasia and metaplasia of bronchial cells and with fibrotic signs.

To determine whether mice without signs of respiratory distress had abnormal lungs, we studied 8–16-week-old mice. The lung morphology of 109 Muc5b^ko/+ mice was analyzed macroscopically and 49 mice (45%) displayed an abnormal lung morphology, characterized by the presence of abnormal gray areas and areas of necrosis and hemorrhage (Fig. S5A), which were smaller than those present for mice with respiratory distress (Fig. S5B). Histological quantification showed that the lung score was significantly higher (P=0.018, Fig. 3A) for Muc5b^ko/+ mice without respiratory distress than for WT mice. We then assessed the deposition of ASMA by immunohistochemistry followed by quantification of immunostaining. Increased deposition of ASMA was observed in the small bronchi of Muc5b^ko/+ mice in comparison to WT mice (P=0.018; Fig. 3C) but the difference between mice with lung disease was not significant.

No macroscopic change was observed in other organs examined, with the exception of the salivary glands, since 13 of  51 (25%) Muc5b^ko/+ mice exhibited atrophy of one salivary gland only (Fig. S5C). This organ was not investigated further.

To demonstrate functional abnormalities in the lungs from young Muc5b^ko/+ mice, the lung mechanics of 10 WT/Muc5b^+/+ and 10 Muc5b^ko/+ 6-week-old male mice were analyzed using Flexivent at baseline and after metacholine administration. No significant baseline changes in dynamic resistance, Newtonian resistance, elastance, tissue damping and tissue elastance were observed in Muc5b^ko/+ mice compared to WT mice (Fig. 4). We then administered metacholine, a smooth muscle agonist, to assess the effects of transient bronchoconstriction. An increase in doses of metacholine caused a decrease in maximum dynamic resistance in Muc5b^ko/+ mice compared to WT mice, reflecting a decrease in level of constriction in the lungs (Fig. 4). Moreover, increasing the doses of metacholine caused a significant decrease in Newtonian resistance, which represents the resistance of the central airways in the constant phase model, elastance, tissue damping and tissue elastance in Muc5b^ko/+ mice compared to WT mice.

As an inflammatory cell infiltrate was observed in both the lungs of 8–16-week-old Muc5b^ko/+ mice (data not shown) and in older mice with lung disease, we assessed the levels of the pro-inflammatory chemoattractant chemokine CXCL1/KC in bronchoalveolar lavage (BAL) of adult WT and Muc5b^ko/+ mice, which displayed no signs of respiratory insufficiency. While no CXCL1/KC was found in the 13 WT mice studied, CXCL1/KC was detectable in the BAL of 6/23 Muc5b^ko/+ mice (26%; Fig. 5).

To determine whether the lung defects in Muc5b^ko/+ mice may appear early in life, we measured the CXCL1/KC level in whole lungs of 10 WT and 25 Muc5b^ko/+ 2-day-old pups. A baseline CXCL1/KC level was detected in the lungs of WT mice which was significantly higher in Muc5b^ko/+ mice as 6/25 (24%) Muc5b^ko/+ mice displayed high levels of CXCL1/KC (P=0.03, Fig. 6A). Histology revealed that the lungs of 2-day-old Muc5b^ko/+ pups had pathological changes at different levels (Fig. 6B). On whole-lung sections, Muc5b^ko/+ mice had a denser parenchyma, with condensed alveolae, congested vessels and an inflammatory infiltrate. The bronchial epithelium of Muc5b^ko/+ mice was enlarged with an increased number of mucus-containing epithelial cells and numerous inflammatory cells in the lamina propria. Fibrin deposits, which are signs of tissue damage in lung injury, were found in the subepithelial region with an increased inflammatory cell infiltrate with abundant neutrophils in Muc5b^ko/+ mice around the bronchi and vessels.

Because Muc5b is not just expressed in the lung, the Muc5b mutation may cause an ectopic phenotype leading to a lethal phenotype. We then restricted the deletion of Muc5b in the lung using CCSP transgenic Cre mice crossed with Muc5b-floxed mice (Bertin et al., 2005). CCSP, also referred to as CC10 and SCGB1A, is transcriptionally activated within the bronchi of neonatal mouse lungs starting at E16.5 (Reynolds et al., 2002). Muc5b-floxed mice on one or two alleles were viable and fertile. No respiratory distress was observed in the 30 mice that were inspected by histology and which carried the CCSP-Cre transgene and no Muc5b-floxed allele (sacrificed between 40 and 50 weeks of age). Of the 124 mice studied and carrying the CCSP-Cre transgene and with at least one Muc5b-floxed allele, 27 (22%) were sacrificed as they showed signs of respiratory distress. Mice with conditional lung deletion of one or two Muc5b alleles, termed Muc5b^{lung ko}, developed respiratory failure (Table 1; P<0.0001). The frequency of respiratory failure was increased when CCSP-Cre:Muc5b-floxed mice were between 20 and 30 weeks old (Fig. S6) with a higher probability in mice that carried the two mutated alleles (65% versus 35% for a single mutated allele).

Lung histology revealed that the walls of the bronchi and bronchioles of 3-week-old Muc5b^{lung ko} mice but not adult mice were thinner than those of control floxed mice (CCSP-Cre-negative) with a flattened appearance in the ciliated cells (Fig. 7A). We then determined the total number of epithelial cells/mm² of the cell wall of the bronchi of CCSP-positive cells and the number of ACT-positive cells by immunohistochemistry (Fig. S7). Reduced numbers of epithelial cells (P=0.007), CCSP-positive cells (P=0.02) and ACT-positive cells (P=0.02) were observed in Muc5b^{lung ko} mice (Fig. 7B; Fig. S7). As expected, low levels of Muc5b polypeptide were produced in Muc5b^{lung ko}/CCSP-Cre mice when both alleles were floxed (heterozygous mice not shown) in contrast to homozygous floxed mice that were CCSP-Cre-negative (Fig. S7).

The two polymeric mucins, MUC5AC and MUC5B, represent the main gel-forming mucins in the mucus layer of the lung (Kesimer et al., 2013). The precise function of each mucin is not well understood. Numerous reports link a dysregulation of the gel-forming mucin MUC5B in human lung diseases like asthma, cystic fibrosis, diffuse panbronchiolitis and familial and sporadic and interstitial pneumonia idiopathic pulmonary fibrosis (Fahy and Dickey, 2010; Kamio et al., 2005; Rose and Voynow, 2006; Seibold et al., 2011; Zhang et al., 2011). In the current study, we characterized histological and functional parameters in lungs of mice that are genetically Muc5b-deficient. Alterations to lung, salivary glands and during embryogenesis are consistent with the distribution of Muc5b within WT mice (Portal et al., 2017a; Roy et al., 2014). The phenotype between the Muc5b^ko/+ mice and the CCSP-Cre:Muc5b-floxed mice are not fully similar. Several reasons may explain this difference. First, the CCSP gene is abundantly expressed in the developing airway and trachea at E18.5, but is attenuated after birth and seems not to be expressed before E16.5 (Reynolds et al., 2002). Second, we cannot rule out an ectopic expression of the the CCSP-Cre transgene used that has also been reported to be expressed in the kidney and the testis. Furthermore, all tissues were not examined. Last, it has been reported that the CCSP-Cre transgene exhibited low efficiency for the Igf1r-floxed mouse (López et al., 2016) suggesting that a non-uniform homologous recombination of the Muc5b-floxed sequence may have occurred.

Bronchi are lined by a single columnar epithelium consisting of few goblet cells and ciliated and Club cells. Club cells can differentiate into ciliated and goblet cells. Club cell metaplasia/hyperplasia, to goblet cells, is a hallmark of lung damage in many airway diseases (see Tompkins et al., 2009). Goblet cell metaplasia and hyperplasia in bronchi of Muc5b^ko/+ mice were evident in Alcian Blue-Periodic Acid Schiff (AB-PAS) staining and immunohistochemical studies using anti-CCSP and anti-PCNA antibodies suggesting epithelial remodeling and hypertrophy in agreement with an increase production of Muc5b observed by immunohistochemistry (Fig. 2) and an increase of the inflammatory state. Invalidation of both alleles in lung only, starting at E16.5, led to minimal or no evidence of goblet cell metaplasia/hyperplasia. However, one-fifth to one-quarter of lung-specific Muc5b-deficient mice also exhibited signs of respiratory distress, in agreement with previous reports showing the essential role of Muc5b for mucociliary clearance (Livraghi-Butrico et al., 2017; Roy et al., 2014). This suggests that the respiratory distress observed for conditional and whole-tissue knockout mice comes from two distinct mechanisms and that Muc5b may play an important role in Club cell differentiation.

Both Muc5b^ko/+ with and without respiratory distress showed immune cell infiltrates, tissue injury and remodeling, differentiation of fibroblasts to myofibroblasts (seen by ASMA expression by activated fibroblasts in the lung parenchyma) and increased extracellular matrix deposition such as collagen, which may explain the impaired ventilation and respiratory insufficiency we observed. These modifications are features of pulmonary fibrosis (Camelo et al., 2014; Li et al., 2011; Phan, 2002; Scotton and Chambers, 2007; Seibold et al., 2011; Selman and Pardo, 2002; Wuyts et al., 2013; Wynn, 2011), whereas airflow obstruction with progressive deterioration of lung function, mucus cell hyperplasia and immune cell infiltrates are hallmarks of chronic obstructive pulmonary disease (Jeffery, 1998). We cannot rule out that hypertrophy and hyperplasia of the existing airway smooth muscle cells can also explain the increase in airway smooth muscle mass as observed in airway remodeling following lung injury (Durrani et al., 2011). Furthermore, whether or not Muc5b deficiency may lead to fibrosis was not directly demonstrated in our study and remains to be assessed.

Not all Muc5b-mutated mice developed lung abnormalities, outlining a complex phenotype; 19% of Muc5b^ko/+ adult mice and 22% of Club cell-restricted Muc5b-deficient adult mice that were 40 weeks old or less suffered from severe respiratory distress. In the absence of respiratory distress, 45% of adult Muc5b^ko/+ mice showed abnormal lung morphology and 26% showed abnormal levels of the pro-inflammatory chemokine CXCL1/KC in BAL, consistent with inflammation in the lungs (Fig. 5). We have considered the possibility that abnormal pulmonary phenotype occurs early in life. At 2 days of age, 24% of Muc5b^ko/+ pups already displayed abnormally high levels of CXCL1/KC (Fig. 6A), although it is not possible to determine if the pups that displayed elevated CXCL1/KC levels corresponded to mice that would develop respiratory distress at adulthood.

Agents responsible for the initiation of the lung phenotype are unknown. In lung fibrosis, it is believed that repeated lung injury could initiate inflammation cascades followed by overproduction of pro-fibrotic cytokines. Among the environmental triggers, viruses (Wuyts et al., 2013) in our protected animal facility that do not belong to the Federation of European Laboratory Animal Science Associations (FELASA) list (Mähler et al., 2014) could be suspected to be initiators of mouse fibrosis. Another possibility is mouse exposure to dust from bedding such as birch and hardwood. This has been reported in a case study to increase the risk of idiopathic pulmonary fibrosis in humans (Gustafson et al., 2007). Further studies will be needed to investigate these hypotheses.

The development of a first Muc5b-deficient mouse line, term Muc5b^–/–, has been reported (Roy et al., 2014). The phenotype of Muc5b^–/– mice differs from that described in this current study. Muc5b^–/– mice showed impaired growth, survival (∼40% at 12 months old) and mucociliary clearance accompanied by abnormal breathing and material obstruction impeding the airflow in the upper airways. Inflammatory infiltrates and viable bacteria in the lung, especially streptococci and staphylococci, were also common in these mice. In the current study, we showed abnormal breathing and inflammatory infiltrates in Muc5b^ko/+ mice and lung-deficient Muc5b mice, but we never found any obstructive material during autopsy or culturable bacteria in the lung (data not shown) suggesting that the two different strategies used to mutate Muc5b and/or the animal environment may explain the variable phenotypes observed in the two models. Variable penetrance is common among mice deficient for the same gene and sometimes the phenotype may differ greatly, as reported, for example, for genetically deficient mouse models for Nedd4-2, which has been shown to be essential for fetal and postnatal lung function (Boase et al., 2011) and for CCSP-deficient mouse models (Reynolds et al., 1999; Zhang et al., 1997). The Muc5b gene has been mapped with the three other mucin genes, Muc2, Muc6 and Muc5ac, to mouse chromosome 7 band F5 in a cluster of genes conserved in humans (Desseyn and Laine, 2003). The genomic organization and the deduced polypeptide sequence of the genes (amino- and carboxy-terminal regions), especially for MUC5AC, exhibit remarkable sequence similarities (Buisine et al., 1998; Desseyn et al., 1997b, 1998, 2000). We cannot exclude cis-effects resulting from genetic modification itself of Muc5b and neighborhood effects to other adjacent unrelated genes or to Muc5ac, which is, with Muc5b, a major gel-forming mucin in the lung.

Environmental conditions and variable genetic background in the two approaches may explain in part the different predisposition of the two models to development of disease (DeMayo, 1999; Reynolds et al., 1999). The two different gene-targeting strategies chosen to obtain Muc5b^–/– (team of C. M. Evans, Roy et al., 2014) and Muc5b^ko/+ (and Muc5b^{lung ko}, current investigation) mice may have yielded the different phenotypes. Because we deleted exons 12 and 13 and not the initiating transcription site, we cannot rule out that they are aberrant transcripts generating a hypomorph phenotype as previously reported for other genetically deficient mouse models (DeMayo, 1999) or truncated peptides translated from aberrant transcripts. We cannot rule out that other splicing events may have occurred too. Studies to understand the basis for differential phenotypes observed for the works on Muc5b-deficient mice should help to better understand the physiological functions of mucin. The recent creation in our laboratory of a new conditional Muc5b-deficient mouse by floxing the last two exons of the gene (Portal et al., 2017b) should also confirm in the near future that the lack of Muc5b leads to pulmonary distress without any bacterial infection of the lung.

The mouse Muc5b gene has been mapped on chromosome 7 band F5 in a cluster of genes conserved between humans and mice (Desseyn and Laine, 2003). The two exonic oligonuccleotides 5′-GACGTCTTCCGCTTCCCTGGCCT-3′ and 5′-TCTTCATTCCACAGGAAGGT-3′, respectively, were designed from the Muc5b gene (Chen et al., 2001). These two primers flank intron 3 and were used to amplify genomic DNA extracted from mouse embryonic stem (ES) cells by PCR. A genomic sequence of 437 bp was cloned and sequenced showing that it contained an intron of 144 bp belonging to Muc5b. The 437 bp insert was used as a probe to screen a mouse 129Sv bacterial artificial chromosome (BAC) clone library (Incyte Genomics). Two positive BAC clones were identified. One BAC clone was purchased and was shown to contain the full genomic sequence of Muc5b (data not shown). Both human and mouse Muc5b genes consist of 49 exons, with exon 31 being the largest [10.7 kb in humans (Desseyn et al., 1997a)], and encoding the O-glycosylated regions of mucin.

The general three-loxP strategy used to invalidate Muc5b is summarized in Fig. S1A,B. Deletion of exons 12 and 13 should lead to a frameshift introducing a premature stop codon. The loxP sequence (34 bp) was subcloned into the pKS+ plasmid (Stratagene) between the unique EcoRI and XhoI restriction sites. The Muc5b targeting construct utilized 5.5 kb and 3.9 kb genomic fragments of the 5′-end of Muc5b that flanked the NheI restriction site in intron 13 as the left and right arms, respectively. A unique loxP site was introduced in intron 11 and a blunted- XhoI–XhoI loxP-flanked neomycin expression phosphotransferase (NEO) cassette (Howe et al., 2006) was inserted in the blunted-NheI site of intron 13. A recombinant plasmid carrying three loxP sites with the same orientation was selected. Deletion of the genomic region carrying exons 12 and 13 with Cre recombinase should introduce a frameshift leading to a premature stop codon. The correct orientation of the three loxP sites was verified by transforming bacteria carrying the plasmid construct with a plasmid encoding Cre recombinase (New England Biolabs; Fig. S2). Plasmid DNA was then linearized using the unique EcoRV restriction site located within intron 17 and analyzed on a 0.6% agarose gel.

The targeting vector was linearized with EcoRV and digested using the PvuI restriction enzyme found twice in the pKS plasmid in order to excise the plasmid insert. The 10.6 kb insert was electroporated in CK35 ES cells (SEAT, Villejuif, France). The genomic DNA of ES cells and mouse tissues was extracted, purified and 10 µg of DNA was digested with KpnI and subjected to electrophoresis on a 0.8% agarose gel. DNA was transferred onto a nylon membrane (Roche Applied Science) and probed with the labeled PCR product described above. Probe hybridization was performed at 42°C overnight with shaking. Detection was carried out by chemiluminescence using an anti-digoxigenin Fab antibody and CDP-star according the manufacturer's instructions (Roche Applied Science). The 689 bp Muc5b probe was obtained by PCR amplification using the two oligonucleotides 5′-TGGGCATCCCACTTGCTG-3′ (forward) and 5′-GTAGAGAGGGTCAACTGATGC-3′ (reverse) and labeled with digoxigenin (DIG)-labeled 11-dUTP (Roche Applied Science). One positive ES cell clone was obtained (Fig. S1C) and microinjected. Chimeric male mice were obtained and intercrossed with C57BL/6 WT mice purchased from Charles River, France. The resulting offspring with the mutated Muc5b locus carrying three loxP sites and their progeny were kept in a specific pathogen-free animal facility. DNA extracted from tail biopsies was analyzed by PCR amplification using the two oligonucleotides 5′-GAGAGGCCTCCACTCTTTCTCCAAGC-3′ (P1; forward) and 5′-CCAAATGTGCATGGCGTGTAAATGAC-3′ (P2; reverse) that flanked the loxP site located within intron 11. Heterozygous mice were intercrossed with the MeuCre40 strain (C57BL/6 genetic background). MeuCre40 genotyping was performed as described elsewhere (Leneuve et al., 2003). Pups carrying both the mutated Muc5b locus and the Cre sequence were bred with WT C57BL/6 mice. After two generations, mice without the Cre sequence but carrying the Muc5b allele deleted for exons 12 and 13 (Muc5b^ko/+) without the NEO cassette were kept and intercrossed. The Muc5b genotype of mice was determined by PCR amplification using the oligonucleotides P1 coupled to the oligonucleotide 5′-GAGAAGAAAGTCCCCGCCCAGTGTTT-3′ (P3; reverse) to amplify the knockout allele. Deletion of the selective cassette was performed using the primer P2 coupled to the specific NEO cassette oligonucleotide 5′-TGTTGTGCCCAGTCATAGCCGAATAG-3′ (P4; forward). Genotyping was confirmed by Southern blotting with the external 5′ probe and an internal probe using the two oligonucleotides 5′-GTGGAGAGGCTATTCGGCTATG-3′ and 5′-CTCTTCAGCAATATCACGGGTAG-3′ amplifying a 648 bp NEO nucleotide sequence. Muc5b^ko/+ heterozygous mice were backcrossed for at least five generations into the C57BL/6 genetic background. Muc5b floxed mice that do not carry the MeuCre40 transgene were also bred with CCSP-Cre mice to obtain lung-specific Muc5b-deficient mice (Bertin et al., 2005). The genotype of mice carrying CCSP-Cre was determined by PCR using the two oligonucleotides used to genotype the MeuCre40 transgene.

Mice were maintained by breeding heterozygous mice after at least six backcrosses in C57BL/6 genetic background. In all experiments, mutated mice were compared with their WT littermates. The animal procedure followed in this study was in accordance with French Guidelines for the Care and Use of Laboratory Animals and with the guidelines of the European Union. The creation and use of the Muc5b floxed strain and progeny were approved by the French Biotechnologies Committee and registered under file number 5288.

Mice were anesthetized by injection of 200 µl pentobarbital. The lungs and salivary glands were gently removed, rinsed in PBS and fixed in 4% paraformaldehyde in PBS for 20 h. Formalin-fixed tissues were dehydrated through a series of increasing ethanol washes and embedded in paraffin. Paraffin blocks were brought to room temperature and sectioned on a rotary microtome. 5 µm thick sections were floated onto water at 40°C before being transferred to Superfrost/Plus microscope slides (Thermo Fisher Scientific). Sections of paraffin-embedded lung tissue were dewaxed with xylene, rehydrated through a series of decreasing ethanol washes and stained with Hematoxylin and Eosin (H&E), AB-PAS and Masson's Trichrome for microscopic examination. For the elastin fiber study, sections were stained with orcein (Sigma-Aldrich). Tissue sections were analyzed blindly by two different pathologists unaware of the genotypes on a motorized Z-axis microscope (BX 61 Olympus, Tokyo, Japan), using epi-fluorescent light. Microscope pictures were obtained with a digital camera ColorView III using Olympus-SIS Cell F software (Olympus, Tokyo, Japan). For immunohistochemistry, sections of paraffin-embedded lung tissue were dewaxed with xylene and rehydrated through a series of decreasing ethanol washes and rinsed three times in PBS. To block non-specific binding, slides were incubated with 1% bovine serum albumin (BSA) in PBS for 45 min. The immunolabeled sections were dried and mounted with Mowiol mounting medium and stored at 4°C. Slides were incubated overnight at 4°C with antibodies against Muc5b (Valque et al., 2011) (1:50), CCSP (1:500; R42AP) (Ryerse et al., 2001), occludin (1:100) (Invitrogen, 71-1500), proliferative cell nuclear antigen (PCNA; 1:100) (Abcam, PC10), α-smooth muscle actin (ASMA; 1:300) (Abcam, ab5694) or ACT (1:400) (Sigma-Aldrich, T7451) in PBS/1%BSA. For anti-PCNA and anti-ASMA antibodies, citrate buffer antigen retrieval was performed as described previously (Valque et al., 2012). For anti-occludin antibodies, protease antigen retrieval was performed according to manufacturer's instructions before BSA incubation. After three washes in PBS, slides were incubated with FITC-conjugated secondary antibodies (1:150) diluted in PBS/1%BSA for 2 h in a dark room at room temperature, rinsed three times in PBS and nuclei were counterstained with Hoescht 33258 (1:1000) for 5 min. For CD31 immunohistochemistry, chromogenic staining using horseradish peroxidase-3,3′ diaminobenzidine (HRP-DAB) staining was performed using an anti-CD31 antibody (Abcam, ab8364) diluted 1:50 according to the manufacturer's instructions. For Muc5b, occludin and CD31 immunohistochemistry, lung sections of three control WT mice, seven 14–22-week-old Muc5b^ko/+ mice with respiratory distress and five 10–16-week-old Muc5b^ko/+ mice without respiratory distress were analyzed. Images were acquired and were minimally processed as described previously (Gouyer et al., 2010).

Bronchi with similar diameters from five WT mice (five bronchi/mouse), three Muc5b^ko/+ mice (three to six bronchi/mouse) without respiratory distress and six Muc5b^ko/+ mice with respiratory distress (four or five bronchi/mouse) were analyzed using an anti-ASMA antibody. ASMA immunofluorescence was measured using ImageJ software and expressed relative to the area of bronchial lumen. Five to seven bronchi per adult mouse with similar diameters from four WT, five Muc5b^ko/+ with respiratory distress and five Muc5b^ko/+ without respiratory distress were analyzed using anti-Muc5b, anti-CCSP and anti-ACT antibodies. Concerning lung-specific Muc5b knockout mice, bronchi with similar diameters from seven WT mice (five to seven bronchi/mouse) and from eight 3-week-old Muc5b floxed mice carrying the CCSP-Cre transgene (five to eight bronchi/mouse) were analyzed using anti-ACT and anti-CCSP antibodies. The area of the bronchial wall was determined using ImageJ software by subtracting the area of the bronchial lumen from the total area of the bronchus. Total epithelial cells from bronchi, Muc5b-immunopositive (goblet cells), CCSP-immunopositive cells (Club cells), ACT-immunopositive cells (ciliated cells) and unknown cells (neither ACT nor CCSP-positive) were counted and expressed relative to the area of the epithelial cell wall of the bronchi.

The number of PCNA-immunopositive cells was counted by observing 10 randomly selected bronchi or bronchioles in lung sections from three WT and three heterozygous mice.

Lung sections from three control WT mice, seven 14–22-week-old Muc5b^ko/+ mice with respiratory distress and five 10–16-week-old Muc5b^ko/+ mice without respiratory distress were scored blindly for inflammation and fibrosis from 0 to 11 (adapted from Madtes et al., 1999) and metaplasia and hyperplasia of the bronchi. Lung inflammation, metaplasia and hyperplasia of the bronchi was scored on H&E-stained lung sections on a scale of 0 to 3 for inflammation (0, no inflammatory involvement; 1, inflammatory cell infiltration of 3–29% of the parenchyma; 2, inflammatory cell infiltration of 30–59% of the parenchyma; and 3, inflammatory cell infiltration of 60–100% of the parenchyma), on a scale of 0 to 2 for metaplasia (0, no metaplasia; 1, slight metaplasia of epithelial cells of the bronchus; 2, strong metaplasia) and on a scale of 0 to 2 for hyperplasia (0, no hyperplasia; 1, slight hyperplasia of epithelial cells of the bronchus; 2, strong hyperplasia). Lung fibrosis was scored on Trichrome-stained lung sections on a scale of 0 to 4 (0, no increase in connective tissue; 1, fine connective-tissue fibrils in less than 50% of the area occupied by inflammatory cells, without coarse collagen; 2, fine fibrils in 50–100% of the same area, without coarse collagen; 3, fine fibrils in 100% of the area, with coarse collagen bundles in 10–49% of the area; 4, fine fibrils in 100% of the area, with coarse collagen in 50–100% of the area). Scores were established on three lobes (left lobe and right cranial and middle lobes) of the lung. Three different sections of the lung (upper, middle and lower parts) were analyzed.

Lung mechanics were assessed in Muc5b^ko/+ mice and Muc5b^+/+ sibling mice using Flexivent (Scireq, Montreal, Canada) as follows. Six-week-old mice were anesthetized by intraperitoneal injection of medetomidine (5 ml/kg; Pfizer, Paris, France) and 10% ketamine (Merial, Lyon, France), paralyzed by intraperitoneal injection of 1% pancuronium bromide (5 ml/kg; Organon) and immediately intubated with an 18-gauge catheter, followed by mechanical ventilation. Respiratory frequency was set at 150 breaths/min with a tidal volume of 0.2 ml and a positive-end expiratory pressure of 2 ml H₂O was applied. Mice were exposed to nebulized PBS followed by increasing concentrations of nebulized metacholine (3–50 mg/ml in PBS; Sigma-Aldrich) using an ultrasonic nebulizer (Aeroneb, Aerogen, Galway, Ireland). For each dose, 10 cycles of nebulization and measurements were performed. Nebulization was conducted during the first cycle and consisted of 20 puffs per 10 s, with each puff of aerosol delivery lasting 10 ms. For each cycle, measurements were obtained for 15 s followed by ventilation for 5 s. The maximum dynamic resistance (R_L), Newtonian resistance (Rn), elastance (E), tissue damping (G) and tissue elastance (E) were recorded before and after increasing doses of aerosolized metacholine.

BAL was performed prior to sacrifice by two consecutive injections (500 µl and 1 ml) of PBS through a tracheal cannula. Lavage fluid was centrifuged at 1000× g for 10 min and the supernatant was stored at −80°C until use. Chemokine KC was measured using an ELISA kit (R&D systems, DY453) according to the manufacturer's instructions. Absorbance at 450 nm was determined using a microplate reader. 50 µl of the BAL supernatant was used to evaluate the KC level of adult mice. To measure the KC level in lungs at post-natal day 2, newborn mice were sacrificed, lungs were removed and homogenized in PBS. The homogenate was then centrifuged at 1000× g for 10 min and 50 µl of the supernatant was used to quantify the KC level. Lungs used for BAL were not used for other investigations.

For histological score, ASMA quantification, KC level quantification in PCNA-positive cells, cell number quantification (ACT-, CCSP-positive cells and total epithelial cells of bronchi), the graphs show the median value. The Wilcoxon–Mann–Whitney and Pearson's Chi² tests were performed using StatXact 6.0 (Cytel Studio, Cambridge, USA) to compare unpaired data. For the lung mechanics, all results are expressed as mean±s.e.m. Two-way analysis of variance was performed using GraphPad Prism software (La Jolla, USA) and was used to analyze lung parameters. A P-value of ≤0.05 was considered statistically significant.

We thank M. Holzenberger (Inserm UMRS 938, Paris, France) for the MeuCre40 mouse strain, M. Tauc (CNRS FRE 3093, Nice, France) for the CCSP-Cre mouse strain, C. Goujet-Zalc (CNRS, SEAT UPS44, Villejuif, France) for the generation of Tg mice, J. S. Ryerse (Dept. of Pathology, St Louis University, MO, USA) for the anti-CCSP antibody, M. H. Gevaert and R. M. Siminski (Service Commun-Morphologie Cellulaire, University of Lille, France) for slides, J. Devassine and R. Dehaynin from the EOPS animal facility (University of Lille, France) for mouse colony management and P. Roussel for critical reading of the manuscript and useful discussions.