Epithelial β1 integrin is required for lung branching morphogenesis and alveolarization

Embryonic lung development occurs as a result of epithelial-mesenchymal interactions that are initiated when the bronchial buds grow into the surrounding mesoderm. The buds undergo iterative branching morphogenesis to give rise to the respiratory tree, terminal saccules and alveoli, whereas the mesoderm forms the lung fibroblasts and the pulmonary vasculature (Cardoso and Lu, 2006; Hogan, 1999). This complex pattern of development depends on a variety of factors, including expression of mesenchymal-derived signaling molecules and their receptors on epithelium, as well as cell-extracellular matrix (ECM) interactions (McGowan, 1992; Morrisey and Hogan, 2010).

The lung basement membrane (BM) is composed primarily of collagen IV, laminins, nidogen and proteoglycans (Thibeault et al., 2003; Wasowicz et al., 1998), and several previous studies have shown that distinct laminin chains are required for normal lung branching morphogenesis, lobar septation and alveolarization (Nguyen et al., 2005,, 2002; Nguyen and Senior, 2006; Willem et al., 2002). The major integrins that affect cell-ECM interactions during lung development are the laminin receptors α3β1, α6β1 and α6β4 (De Arcangelis et al., 1999; Georges-Labouesse et al., 1996; Has et al., 2012; Kim et al., 2009; Kreidberg et al., 1996; Nicolaou et al., 2012). Of the laminin receptors, integrin α3β1 has been most extensively investigated. Lungs from constitutive integrin α3-null mice fail to branch into bronchioles (Kreidberg et al., 1996), and mice null for both integrin α3 and α6 have severe lung hypoplasia (De Arcangelis et al., 1999). Surprisingly, mice with a lung epithelial-specific deletion of α3 integrin have a normal lifespan with normal airway branching but subtle defects in epithelial differentiation (Kim et al., 2009). Recently, lung hypoplasia due to integrin α3 mutations has been identified as a cause of early childhood mortality (Has et al., 2012; Nicolaou et al., 2012). The RGD-binding integrin α8β1 has also been shown to regulate lung development as constitutive integrin α8-null mice have abnormally fused medial and caudal lobes of the lung, as well as subtle abnormalities in airway division (Benjamin et al., 2009).

β1 integrin is the major isoform of the eight β-integrin subunits, and is the β-integrin subunit present in 12 of the 24 known α-β integrin heterodimers (Pozzi and Zent, 2011). When β1 integrin is selectively deleted in lung epithelium at the time of endodermal tracheal outgrowth (E9.5) using a Shh promoter-driven Cre, mice develop a severe branching morphogenesis defect that blocks development at the level of primary bronchi, making it impossible to determine the role of β1 integrin at later stages of development (Chen and Krasnow, 2012). To elucidate possible crucial functions of β1 integrin beyond the initial stages of lung development, we generated mice with an epithelial-targeted deletion of β1 integrin using a surfactant protein C (SP-C; also known as SFTPC) promoter-driven Cre, which is expressed in the lung epithelium at E10.5 (Okubo et al., 2005). Analysis of these mice showed that β1 integrin regulates a variety of crucial epithelial cell processes required for normal lung development and homeostasis.

We deleted β1 integrin in the lung epithelium from E10.5 onwards by crossing SP-C-Cre mice with integrin β1^f/f (β1^f/f) mice (hereafter called β1^SP-C.Cre mice). By E13, β1 integrin was selectively deleted in the proximal and distal lung epithelium as shown by immunostaining of embryonic airways (Fig. 1A,B). β1^SP-C.Cre mice were born in the normal Mendelian ratio; however, few survived longer than 4 months (Fig. 1C). β1^SP-C.Cre mice grew poorly and were significantly smaller than age-matched littermate control β1^f/f (Fig. 1D) and heterozygote SP-C-Cre; β1^f/WT mice (data not shown). Interestingly, β1^SP-C.Cre mice had normal oxygen saturation at birth, but by 8 weeks they were severely hypoxemic with a mean oxygen saturation of 79% (Fig. 1E). Thus, deleting β1 integrin in the lung epithelium results in a lethal phenotype with severe hypoxia.

Despite having lower body weight, adult β1^SP-C.Cre mouse lungs were larger and weighed more (35.8±3 mg versus 24.2±2 mg) than littermate β1^f/f mice (Fig. 2A,B). Histological examination of lung parenchyma from P28 and older β1^SP-C.Cre mice revealed dilated airspaces surrounded by thickened, hypercellular alveolar septa with type II epithelial cell hyperplasia and increased numbers of macrophages (Fig. 2C-K). β1^SP-C.Cre lungs showed sustained airspace enlargement at 3 months of age (Fig. 2H-K). Alveolar size, as measured by calculating the mean linear intercept, in P28 β1^SP-C.Cre mice was approximately double the value in controls (Fig. 2G). Heterozygotes were indistinguishable from littermate control β1^SP-C.Cre mice in histological examinations at both time points (data not shown). Thus deleting β1 integrin from the lung epithelium results in large and dilated airspaces, thickened inter-alveolar septa, abnormal epithelial cell differentiation and influx of alveolar macrophages, probably explaining the hypoxemia and premature mortality observed in these mice.

Analysis of the airways in adult β1^SP-C.Cre mice revealed a paucity of terminal bronchioles, indicative of a branching morphogenesis defect. We therefore examined lung histology near the initiation, mid-point and completion of branching morphogenesis. Although no branching defect was identified at E13 during early branching morphogenesis (Fig. 3A,B), abnormal branching was seen at E15. At E15, β1^SP-C.Cre airways were similar in size to controls but fewer in number and separated by a thickened interstitium (Fig. 3C,D). This branching defect was even more evident by E18, where there was a marked reduction of airways in β1^SP-C.Cre mice (Fig. 3E,F) as verified by quantification of histological sections (147±14 airways per mm² in β1^SP-C.Cre lungs compared with 259±13 airways per mm² in β1^f/f mice; mean±s.e.m.) (Fig. 3I). In contrast to the proximal to distal airway narrowing seen in normal mice, there were a number of large airways found in the peripheral lung of the β1^SP-C.Cre mice (Fig. 3F, asterisks). Consistent with observations indicating fewer airways in β1^SP-C.Cre mice, the airspace volume density of E18 β1^SP-C.Cre lungs was significantly less (29±1%) than β1^f/f lungs (46±1%) (Fig. 3J). Airspaces in lungs of β1^SP-C.Cre mice at P0 were fewer in number and larger in histological examinations (Fig. 3G,H).

In addition to abnormal airway branching at E18 in β1^SP-C.Cre mice, hypercellularity of the interstitium was evident. By Ki67 (also known as MKI67) staining, we identified increased cellular proliferation in the interstitium of E18 β1^SP-C.Cre lungs compared with β1^f/f lungs (29±2% versus 7±1% Ki67-positive cells) (Fig. 3K-M). The proliferating cell population was primarily non-epithelial in β1^SP-C.Cre lungs as verified by co-immunostaining for the mitotic marker phosphohistone H3 (PHH3) and the epithelial marker E-cadherin (Fig. 3N-P). A total of 58.7±0.5% and 39.2±0.5% of proliferating (PHH3+) cells were non-epithelial in β1^SP-C.Cre and β1^f/f lungs respectively (P<0.05). By contrast, there was no significant difference in apoptosis between β1^SP-C.Cre and β1^f/f lungs at this time point (supplementary material Fig. S1E). Thus, deleting β1 integrin in the lung epithelium results in a branching morphogenesis defect and increased interstitial cell proliferation.

To better define the mechanism of branching defects in β1^SP-C.Cre mice, we examined E15 fetal lung explants. Compared with controls, significantly less new branching occurred in β1^SP-C.Cre lung explants cultured for 12 and 48 h (Fig. 4A-E). When the velocity of the leading epithelial edge of peripheral airways was tracked in these cultures using time-lapse microscopy, we measured a decreased outward velocity of the epithelial edge in β1^SP-C.Cre explants (3.2±0.05 μm/h versus 2.9±0.06 μm/h in β1^f/f explants) (Fig. 4F). Although no difference in explant size was identified between β1^SP-C.Cre and β1^f/f lung explants at 48 h (data not shown), the distance from the epithelial basement membrane to the explant edge was greater in β1^SP-C.Cre lung explants compared with controls (2.08±0.5 μm versus 0.13±0.6 μm), indicative of decreased outward epithelial growth and/or increased mesenchymal proliferation (Fig. 4G-I). These data suggest that loss of epithelial β1 integrin disrupts epithelial migration, a key cellular process during branching morphogenesis. To further investigate the migratory phenotype of β1 integrin-null lung epithelial cells, we isolated type II epithelial cells from P28 β1^SP-C.Cre and β1^f/f mice and confirmed β1 integrin deletion by immunoblotting (Fig. 4J). These cells had a marked adhesion defect (74±11% β1^f/f versus 18±18% β1^SP-C.Cre cells adhered) on purified laminin 511, a major laminin isoform found in the fetal lung basement membrane that is required for normal lung development (Nguyen et al., 2005) (Fig. 4K). In addition, these cells had a major haptotactic migration defect on laminin 511 (Fig. 4L). Taken together, these data indicate a role for β1 integrin in alveolar epithelial adhesion and migration, and suggest a potential mechanism for the branching defect seen in β1^SP-C.Cre mice.

In addition to an airway branching defect, adult β1^SP-C.Cre mice had dilated alveoli with thickened alveolar septa, suggestive of an additional alveolarization defect. We therefore examined β1^SP-C.Cre lungs at P7 and P14 to identify defects in secondary septation. Airspace enlargement was seen by P7 (Fig. 5A,B) in β1^SP-C.Cre lungs and this was even more obvious at P14 (Fig. 5C,D). In addition, P14 β1^SP-C.Cre alveolar septa were hypercellular and thickened suggesting increased ECM deposition. We quantified the differences in alveolar size by measuring the average alveolar diameter in P14 β1^SP-C.Cre and β1^f/f mice (50±2 μm in β1^SP-C.Cre lungs versus 30±3.5 μm in β1^f/f lungs; mean±s.e.m.) (Fig. 5G). To determine whether loss of β1 integrin disrupted secondary septation, we quantified the number of secondary crests and found that β1^SP-C.Cre lungs had significantly fewer secondary crests per high-power field than β1^f/f control littermates (26±2 crests in β1^SP-C.Cre lungs versus 92±5 crests in β1^f/f lungs) (Fig. 5H). Similar to the above studies in embryonic lungs, we found that the hypercellularity noted in the interalveolar spaces during alveolarization was due to increased cell proliferation and not decreased apoptosis, as Ki67 staining was increased in β1^SP-C.Cre mice compared with β1^f/f controls (25±2% versus 10±1%), whereas immunostaining for cleaved caspase 3 was not different between the two genotypes (Fig. 5I,J). When we defined the composition of interstitial ECM at P14, we noted a marked increase in elastin and collagen I, but not collagen IV, in β1^SP-C.Cre alveolar septa compared with that amount in alveolar septa from β1^f/f lungs (Fig. 5K-P). Despite disruptions to the epithelium and ECM, the vasculature developed in close proximity to the alveolar epithelium in β1^SP-C.Cre mice because staining for the endothelial marker CD31 demonstrated that there were intact capillary networks adjacent to the alveolar surface in both β1^f/f and β1^SP-C.Cre mice (supplementary material Fig. S1C,D). Taken together, these data suggest that deleting β1 integrin disrupts alveolar septal structure by reducing secondary septation and increasing mesenchymal cell proliferation and ECM deposition.

The histological features of β1^SP-C.Cre lungs suggested that there were β1 integrin-dependent defects in epithelial cell differentiation. Abundant cuboidal epithelial cells lined the peripheral airways of E18 β1^SP-C.Cre lungs (Fig. 6B, arrows), compared with a more differentiated flattened epithelium in control mice (Fig. 6A, arrowheads). After birth, periodic acid schiff staining revealed that there were epithelial cells with a marked increase in glycoprotein deposition in large airways of P14 β1^SP-C.Cre mice, which was not present in the β1^f/f mice (Fig. 6C,D). In addition to airways, we evaluated epithelial structure in lung parenchyma. Alveoli of P14 β1^SP-C.Cre mice showed increased Nkx2.1 and pro-SP-C immunostaining consistent with type II cell hyperplasia (Fig. 6E-J). Transmission electron microscopic images of representative type II cells from adult β1^SP-C.Cre and β1^f/f mice demonstrated that type II alveolar epithelial cells lining the thickened septa (Fig. 6L, arrows) in β1^SP-C.Cre mice were not only hyperplastic, but also had abnormal cellular morphology (Fig. 6K-M) characterized by fewer lamellar bodies (demarcated by arrowheads), a larger nucleus and diminished cytoplasm. As these cells appeared to have a hybrid phenotype (i.e. between that of a type I and type II cell), we immunostained P14 lungs for the type I marker T1α (also known as PDPN) and the type II cell marker pro-SP-C and found numerous dual-positive β1^SP-C.Cre epithelial cells lining the dilated airspaces (Fig. 6N,O), suggesting that β1 integrin regulates alveolar epithelial cell differentiation.

To address whether the β1 integrin-dependent airway branching and alveolarization defects were separable, we generated triple transgenic mice (SP-C rtTA; tetO-Cre; β1^f/f) with a doxycycline-inducible lung epithelial-specific β1 integrin deletion at P0, which is after the completion of branching but prior to the start of alveologenesis. These mice retained the alveolarization defect seen in β1^SP-C.Cre mice at P28, characterized by dilated airspaces, thickened alveolar septa, type II cell hyperplasia and increased alveolar macrophages (Fig. 7A-D). As seen in Fig. 7B, the subpleural airspaces in the peripheral lung were most severely affected, with more modest dilation of the proximal airspaces, consistent with proximal-distal maturation of the lung. These findings indicate that regulation of alveolarization by epithelial β1 integrin is independent of its role in branching morphogenesis.

An increase in alveolar macrophage numbers was confirmed in the bronchoalveolar lavage fluid (BALF) of 8-week-old β1^SP-C.Cre mice (Fig. 8A-C). Increased macrophage accumulation, as measured by CD68 immunostaining, was seen as early as E18 in β1^SP-C.Cre lungs (Fig. 8D-G); however, there were no difference in macrophage numbers in E15 β1^f/f and β1^SP-C.Cre lungs (supplementary material Fig. S1A,B). No differences in lung neutrophils or lymphocytes were identified between genotypes on immunostained lung sections (Fig. 8H-K).

The massive recruitment of macrophages, but not other types of immune or inflammatory cells, seen in adult β1^SP-C.Cre lungs prompted us to investigate the contribution of increased alveolar macrophages to the alveolarization defect seen in β1^SP-C.Cre mice. We therefore depleted macrophages during alveolarization by treating β1^f/f and β1^SP-C.Cre mice with intranasal clodronate from P5 to harvest at P14 or P28. Although clodronate-treated P28 β1^SP-C.Cre mice still had airway branching defects (asterisks in Fig. 9B), this treatment rescued the alveolarization defect seen in β1^SP-C.Cre mice. In contrast, P28 β1^SP-C.Cre mice treated with vehicle (liposomes containing PBS) retained the phenotype of dilated airspaces, thickened alveolar septa, type II cell hyperplasia and increased alveolar macrophages (Fig. 9A-F). Normalization of the β1^SP-C.Cre phenotype was quantified by measuring the average alveolar diameter (Fig. 9G) and quantification of secondary crests at P14 (Fig. 9H). Taken together, these data indicate that deficiency of β1 integrin disrupts alveolarization through recruitment and/or activation of macrophages.

Given that inflammation that is predominantly mediated by macrophages impairs alveolarization in β1^SP-C.Cre mice, we next sought to identify the mechanism of alveolar macrophage recruitment to the lungs of mice with epithelial β1 integrin deficiency. We first measured the BALF levels of the chemoattractant monocyte chemoattractant protein 1 (MCP-1; also known as CCL2), the major chemokine responsible for monocyte and macrophage recruitment. We found that MCP-1 levels were markedly increased in BALF from β1^SP-C.Cre mice compared with control mice (438±52 pg/ml versus 142±5 pg/ml) (Fig. 10A). In contrast, there was no difference in the BALF levels of KC (also known as CXCL1), a neutrophil-specific chemoattractant (Fig. 10B). To determine whether the MCP-1 was derived from the epithelium, we measured MCP-1 levels in primary type II alveolar epithelial cells isolated from β1^f/f and β1^SP-C.Cre mice. As for BALF, levels of MCP-1, but not KC, were significantly increased in medium from β1^SP-C.Cre epithelial cells compared with that from control β1^f/f cells (Fig. 10C,D). These data suggest that intact β1 integrin suppresses the expression of chemokines in the lung epithelium.

MCP-1 expression has been previously shown to be induced by reactive oxygen species (ROS) in different cell types and disease states (Chen et al., 2004b; Lakshminarayanan et al., 2001; Vlahos et al., 2011; Xing and Remick, 2007a,,b). To determine whether epithelial β1^SP-C.Cre MCP-1 secretion was associated with an increase in ROS production, we performed both in vivo and in vitro ROS assays. In vivo, we assayed in lungs of β1^SP-C.Cre and control mice for ROS production by bioluminescence imaging following injection of a modified luminol derivative (Han et al., 2013). We found increased ROS production over the chest of β1^SP-C.Cre mice compared with β1^f/f control mice (Fig. 10E-G). To define whether the β1 integrin-deficient epithelium directly contributed to ROS production, we examined epithelial lipid peroxidation, as an indicator of oxidative stress. Primary type II alveolar epithelial cells isolated from β1^SP-C.Cre mice had a fivefold increase in lipid peroxidation over β1^f/f epithelial cells (Fig. 10H). These findings suggest that deleting β1 integrin in epithelial cells alters ROS production, which in turn regulates macrophage influx.

We found that deleting β1 integrin in lung epithelium beginning at E10.5 results in abnormal lung development with mortality by 4 months of age. A variety of specific defects were identified, including altered branching morphogenesis, impaired alveolarization with epithelial cell differentiation defects and persistent macrophage-mediated inflammation, indicating that β1 integrin expression in the lung epithelium plays multiple roles in distinct phases of lung development.

Our findings are complementary to a prior study that found that epithelial depletion of β1 integrin beginning at E9.5 causes a serious branching defect and neonatal death (Chen and Krasnow, 2012). Although the disparity in these phenotypes could be due to differences in potency between the Cre-expressing mice (Shh and the SP-C), it is more likely that β1 integrin has a more profound role in early rather than late lung development. This is consistent with the crucial role for β1 integrin in the early development of other branched organs. In the kidney, β1 integrin deletion at initiation (E10.5) of ureteric bud development results in a lethal phenotype but has no effect when it is removed at E18.5 (Zhang et al., 2009). In addition, fetal β1 deletion in the basal epithelial cells in the mammary gland causes a branching defect, whereas postpubertal deletions do not (Faraldo et al., 1997; Naylor et al., 2005; Taddei et al., 2008). Thus, epithelial cell β1 integrin expression appears to be a crucial requirement for early embryonic development of branched organs.

In contrast to our findings, mice with epithelial-targeted deficiency of the integrin α3 subunit have a normal lifespan, preserved alveolar architecture and no differences in total lung capacity and airway resistance (Kim et al., 2009). These mice have only mild type II alveolar epithelial cell hyperplasia, mild fibrosis and increased numbers of macrophages in BALF (Kim et al., 2009). In contrast to the epithelial-specific α3-null mice, global integrin α3 deletion results in an airway branching defect (Kreidberg et al., 1996), and mice mutant for both integrin α3 and α6 die at E16 with a single-lobed right lung and an almost absent left lung (De Arcangelis et al., 1999). Consistent with findings in constitutive integrin α3-null mice, functional mutations of integrin α3 in humans cause pulmonary hypoplasia, severe fibrosis with thickened alveolar septa lined with reactive type II pneumocytes, and increased numbers of alveolar macrophages (Has et al., 2012; Nicolaou et al., 2012). Although integrins expressed by the epithelial and mesenchymal compartments of the lung probably play distinct roles in different aspects of lung development, available information suggests that several of the α-β1 integrin heterodimers expressed in lung epithelium regulate lung branching morphogenesis and alveolarization.

We show that integrin β1-null type II alveolar epithelial cells are increased in number and morphologically abnormal with enlarged nuclei and fewer mitochondria and lamellar bodies, suggesting abnormal type II alveolar epithelial cell differentiation. Consistent with this abnormal morphology, the epithelial cells lining dilated airspaces in β1^SP-C.Cre lungs are positive for both type I and type II cell markers at P14. These cells appear similar to the bipotent epithelial progenitors cells recently identified during late embryonic lung development (Yamashita et al., 2004), suggesting that β1 integrin is required for normal alveolar epithelial cell differentiation. Abnormal epithelial morphology is also seen in constitutive integrin α3-null mice (Kreidberg et al., 1996) and lung epithelial-specific laminin α5-null mice (Nguyen et al., 2005). Different in vitro ECM culture conditions have been shown to alter lung epithelial cell differentiation with laminins promoting a type II cell phenotype, and fibronectin and collagen I inducing type I cell characteristics (Isakson et al., 2001; Lwebuga-Mukasa, 1991; Olsen et al., 2005; Rannels and Rannels, 1989). In addition to the ECM type, physical force mediated through integrin-ECM interactions regulate lung epithelial cell differentiation in vitro through unknown mechanisms (Huang et al., 2012; Sanchez-Esteban et al., 2004; Wang et al., 2013,, 2006,, 2009). Thus, it is likely that abnormally differentiated type II cells in β1^SP-C.Cre mice result from impaired integrin-dependent attachment and altered interactions with the ECM. This explanation is consistent with abnormalities in epithelial cell differentiation induced by deletion of integrin β1 in other organs, such as the kidney proximal tubule (Elias et al., 2014), enterocytes (Jones et al., 2006), keratinocytes (Brakebusch et al., 2000), mammary epithelium (Naylor et al., 2005) and the submandibular gland (Menko et al., 2001).

Many studies have focused on the role of mesenchymal growth factors and epithelial receptor signaling in lung development, and there are many similarities between mice with growth factor and/or receptor deletions and mice lacking β1 integrin in the lung epithelium. Lungs from FGF10- and FGFR2-null mice fail to branch beyond the trachea (Min et al., 1998; Sekine et al., 1999), FGFR3/4-null mice have an alveolarization defect (Weinstein et al., 1998), and mice null for both FGFR3 and FGFR4 have a mild branching defect and thickened alveolar septa, similar to the β1^SP-C.Cre mice (Miettinen et al., 1995). We have previously shown in the kidney that β1 integrin is required for FGF2 and FGF10 signaling in ureteric bud development (Zhang et al., 2009), and that β1 integrin regulates FGF- and EGF-dependent signaling in renal collecting duct cells (Mathew et al., 2012). Thus many of the phenotypical characteristics observed in the β1^SP-C.Cre mice might be caused by both alterations in integrin-dependent growth factor signaling as well as adhesion and migration defects.

Our studies point to an important role for β1 integrin in maintaining alveolar homeostasis, which is required for normal alveolarization during the early post-natal period. In the mammary gland, epithelial β1 integrin deletion results in epithelial detachment from the basement membrane without inflammation (Li et al., 2005; Naylor et al., 2005). In contrast, increased numbers of macrophages were observed in the lungs of mice with laminin α3 chain mutations (Urich et al., 2011) and in the lungs of humans with integrin α3 mutations (Nicolaou et al., 2012), suggesting that β1 integrin-mediated regulation of inflammation is specific to the lung epithelium. Whereas β1 integrin deficiency results in increased ROS production and MCP-1 secretion from alveolar epithelial cells, the molecular mechanisms accounting for these findings will require further study. Increased ROS production has been described in integrin α1-null glomerular mesangial cells (Chen et al., 2007); however, this has not previously been shown to occur in epithelial cells, and β1 integrins have not been linked to MCP-1 expression or ROS production in other systems. Although the role of macrophages during alveolarization is not well understood, we and others have shown that macrophages and macrophage-derived products disrupt branching morphogenesis (Blackwell et al., 2011; Nold et al., 2013). Specifically, we have found that products of activated macrophages can impair expression of molecules by epithelial and mesenchymal cells that are important for control of airway branching, including BMP4, Wnt7b and FGF10 (Benjamin et al., 2010; Blackwell et al., 2011; Carver et al., 2013). We speculate that mediators secreted by activated macrophages might also disrupt crucial epithelial-mesenchymal interactions required for normal septation and remodeling of the interstitium.

In conclusion, this study shows that β1 integrin expression in lung epithelium is required during different stages of lung development for airway branching morphogenesis, alveolarization and maintenance of homeostasis. In addition to its well-known functional role in epithelial cell adhesion and migration, β1 integrin modulates alveolar cell differentiation and alveolar septal ECM deposition. Finally, we have also shown that β1 integrin regulates epithelial chemokine and ROS production, which appears to be a new mechanism whereby cell-ECM interactions regulate lung inflammation and alveolarization.

All experiments were approved by the Vanderbilt University Institutional Animal Use and Care Committee. Transgenic mice expressing Cre recombinase under control of the surfactant protein C promoter (SP-C-Cre) were generated by Brigid Hogan (Duke University, Durham, NC, USA) (Okubo et al., 2005) and integrin β1^flox/flox mice were a generous gift from Elaine Fuchs (Howard Hughes Medical Institute, The Rockefeller University, New York, NY) (Raghavan et al., 2000). β1^SP-C.Cre mice (C57BL/6J background) were generated by crossing a male heterozygote (SP-C-Cre; β1^f/WT) with a female control (β1^f/f). This mating strategy deletes β1 integrin specifically from the lung epithelium from E10.5 in SP-C-Cre; β1^f/f mice (called β1^SP-C.Cre mice). Littermate β1^f/f mice were used as controls.

Triple transgenic mice (SP-C rtTA; tetO-Cre; β1^f/f) were obtained by breeding (Perl et al., 2002; Tanjore et al., 2013) and lung epithelial-specific deletion of β1 integrin was induced by the ingestion of doxycycline in the drinking water (2 g/dl). β1^f/f littermates given identical dosage of doxycycline were used as controls.

Lungs were dried overnight at 75°C. Non-anesthetized oxygen saturations were obtained using MouseOx pulse oxymetry (Starr Life Sciences, Holliston, MA). Lungs were embedded, sectioned and stained as indicated. Ki67 staining was used for analysis of proliferation (Abcam ab16667, Cambridge, MA). Lung morphometry was performed on images viewed using a 40× objective for six sections per mouse, with a minimum of four mice in each group (Kauffman, 1977) using Image Pro Plus software. The airspace volume density was measured by dividing the sum of the airspace area by the total area. P14 alveolar secondary crests were quantified as previously described (Nicola et al., 2009). For electron microscopy, adult lungs were processed, post-fixed with potassium ferrous cyanide, dehydrated with graded acetone, sectioned at 1 μm and imaged using a Philips FEI T-12 transmission electron microscope.

Lung explants from E15 β1^SP-C.Cre and β1^f/f mice were cultured on permeable supports and on an air-liquid interface at 37°C in 95% air with 5% CO₂ overnight to allow adherence to the filter, as previously described (Blackwell et al., 2011; Prince et al., 2005,; Prince et al., 2004). Peripheral branch counts (all branches touching the edge of the explant) were obtained from images taken at 12 and 48 h in culture using a Leica DFC450 microscope. Time-lapse images were obtained every 10 min for 8 h using a Nikon Ti Eclipse microscope. The epithelial edge was tracked on the first ten consecutive peripheral airways from the 12:00 position using ImageJ analysis software.

Clodronate (dichloromethylene diphosphonic acid; Sigma-Aldrich) and sterile PBS-containing liposomes (vehicle) were prepared as previously described (Everhart et al., 2005; Zaynagetdinov et al., 2011). β1^f/f and β1^SP-C.Cre mice were treated with intranasal clodronate or PBS-containing liposome vehicle every 5 days from P5 until harvest at P14 or P28. Intranasal clodronate was given at doses of 12 μl, 15 μl, 18 μl, 21 μl and 24 μl at P5, P10, P15, P20 and P25, respectively. Duplicate experiments were performed using PBS-containing liposome vehicle control.

For immunofluorescence, 8 μm frozen sections were fixed with 10% formalin for 20 min at room temperature, permeabilized with 0.1% Triton X-100, blocked in 5% donkey serum in PBS, and stained with the following primary antibodies: anti-β1 integrin (1:250, Millipore MAB1997), anti-E-cadherin (1:500, Invitrogen 131900), anti-cleaved caspase 3 (1:500, Cell Signaling 9661), anti-Nkx2.1 (1:250, Santa Cruz Biotechnology sc-13040), anti-pro-SP-C (1:500, Abcam ab90716), anti-T1α (1:250, Developmental Studies Hybridoma Bank 8.1.1), anti-CD68 (1:500, Abcam ab53444), anti-CD31 (1:500, BD Pharmingen 550274), anti-CD3 (1:500, Abcam ab5690) and FITC-conjugated anti-Gr1 (also known as Gsr; 1:500, eBioscience 11-5931) antibodies. Detection of bound antibodies was accomplished with Alexa Fluor 488-conjugated donkey anti-rat IgG or Alexa Fluor 555-conjugated donkey anti-rabbit IgG (both 1:500, Life Technologies A31572). Slides were analyzed using a Leica SPE confocal microscope.

Type II epithelial cells were isolated from P28 β1^SP-C.Cre and β1^f/f mice as previously described, yielding >90% type II cells (Rice et al., 2002; Young et al., 2012). Tissue was digested using dispase and the epithelial cell population was separated from leukocyte and monocyte populations using antibody-coated plates (CD45 and CD32, BD Pharmingen) for negative selection. After incubation for 2 h at 37°C in 5% CO₂, type II cells were harvested, centrifuged and were resuspended in lysis buffer or cultured. Conditioned medium was collected at 24 h from 10⁶ isolated  epithelial cells per well cultured in serum-free medium on laminin-511.

Cell adhesion and migration assays were performed with isolated epithelial cells from β1^SP-C.Cre and β1^f/f mice in serum-free bronchial epithelial growth medium (BEGM) (Chen et al., 2004a). Plates and filters were coated with purified laminin 511 (1 μg/ml, made by Eugenia Yazlovitskaya), a prominent lung basement membrane component, which we purified from HEK cells stably transfected with the α5β1γ1 using methods previously described (McKee et al., 2009).

For the adhesion assay, 10⁵ cells were added per well, and cells that were adherent at 6 h were fixed, stained with Crystal Violet and solubilized. The optical density was read at 540 nm. Each mouse was analyzed in triplicate and five independent experiments were averaged. For the migration assay, 2×10⁵ cells were added to the upper chamber of a filter. After 6 h, cells that migrated through the filter were stained and counted. Five random fields were analyzed and three independent experiments were performed.

Sterile saline lavages were performed with 35 μl/g of PBS after killing, using a 20 g blunt-tipped needle inserted into the trachea. Samples were centrifuged at 400 g for 10 min, and the supernatant was collected. Cell counts were performed using a Bio-Rad TC10 Automated Cell Counter.

Total protein (40 μg) collected from isolated type II cell lysate was electrophoresed in a 10% gel and transferred onto nitrocellulose membranes. Membranes were blocked and incubated with different primary antibodies [anti-SP-C (1:5000, Abcam 90716), anti-β1 integrin from Millipore (1:10,000, 1952) and anti-β-actin (1:5000, Sigma A5316) antibodies], followed by the appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies. Immunoreactive bands were identified using enhanced chemiluminescence according to the manufacturer's instructions.

Enzyme-linked immunosorbent assay (ELISA) for MCP-1 and KC on BALF and medium was performed according to the manufacturer's instructions (R&D Systems, MJE00 and MKC00B, respectively). Each sample was run in triplicate.

Lipid peroxidation in freshly isolated epithelial cells from β1^SP-C.Cre and β1^f/f mice was detected by measuring the reaction of thiobarbituric acid with malondialdehyde colorimetrically at 540 nm (Cayman Chemical Company).

ROS production was detected in β1^SP-C.Cre and β1^f/f mice in vivo as previously described (Han et al., 2013). L-012 (Wako Chemicals), a luminol analog, was injected retro-orbitally. Upon reaction with superoxide, luminescence was produced and detected by an IVIS system device (Xenogen). Luminescence data was collected using Living Image software v.4.1 (Xenogen).

The Student's t-test was used for comparisons between two groups with results representing mean±s.e.m. P<0.05 was considered statistically significant.

We thank Brigid Hogan, Elaine Fuchs and Eugenia Yazlovitskaya for generously supplying reagents.