PDGF-A signaling is required for secondary alveolar septation and controls epithelial proliferation in the developing lung

The presence of alveoli at the distal ends of the bronchial tree hugely increases the surface for gas exchange in the mammalian lung, and is a prerequisite for efficient breathing. Loss of alveolar structure leads to emphysema and chronic obstructive pulmonary dysplasia (COPD), which are disabling and life-threatening diseases. Alveolar structures comprise alveolar epithelial type 1 (AEC1) and type 2 (AEC2) cells in close contact with the capillary network, lined by mesenchymal cells such as lipofibroblasts and myofibroblasts (Beers and Morrisey, 2011; Chapman, 2011; Morrisey and Hogan, 2010; Warburton et al., 2010).

The process of alveolar maturation is known as alveogenesis. This final stage of lung development is initiated prenatally in humans, but postnatally in mice (McGowan, 2014; Shi et al., 2007; Warburton et al., 2010). The acinar buds at the end of the primitive bronchiolar saccules, which later give rise to the mature alveoli, are formed by thinning of the mesenchyme and continuing differentiation of both mesenchymal and epithelial cells during the sacculation stage of lung development (Treutlein et al., 2014; Warburton et al., 2010). At the end of this stage, lungs can support gas exchange but it is not efficient enough to fulfill the oxygen requirement (Pinkerton et al., 2014). To increase the surface area for gas exchange, the early postnatal alveolar structures are further divided by fishnet-like structures called septal ridges (Branchfield et al., 2016). The septal ridges restrict the ongoing epithelial expansion by creating secondary septae in the forming alveolar structures. This process is dependent on alpha smooth muscle actin (α-SMA) and elastin deposition from myofibroblasts surrounding the airways (Barron et al., 2016; McGowan, 2017; Moiseenko et al., 2017). Anomalous alveolar formation is a distinctive feature of bronchopulmonary dysplasia (BPD), a chronic illness usually observed in premature babies (Popova, 2013).

The platelet-derived growth factor (PDGF) family consists of four ligands (PDGF-A, -B, -C and -D) and two receptor tyrosine kinases (PDGFRα and β). They play important roles in development, homeostasis and disease, and downstream signaling from the PDGFRs is known to regulate cell differentiation, proliferation, migration, actin reorganization and apoptosis (reviewed by Andrae et al., 2008). In mice, PDGFRα and PDGF-A are expressed in the lung from embryonic time points to adulthood (Branchfield et al., 2016; Endale et al., 2017; Gouveia et al., 2017; Ntokou et al., 2015). Pioneering studies of global Pdgfa knockout (Pdgfa^−/−) mice demonstrated a role for PDGF-A in lung development. These Pdgfa-negative mice exhibited abnormally enlarged distal airspaces without signs of secondary septation, but because the mice died soon after birth a detailed analysis of lung morphology was not possible (Boström et al., 1996; Lindahl et al., 1997). The full Pdgfra knockout is early embryonic lethal (Soriano, 1997), but postnatally viable heterozygous mice combined with a hypomorphic Pdgfa allele also demonstrated dilated distal airspaces (Andrae et al., 2013). Other studies have shown that stromal cells obtained from premature infants with BPD have reduced levels of Pdgfra expression (Popova et al., 2014). Together, these observations suggest a role for PDGF-A signaling via PDGFRα in alveolar development.

Previous analyses of the Pdgfa^−/− mice suggested several possible roles for PDGF-A in alveogenesis: (1) PDGFRα⁺ myofibroblast progenitors require PDGF-A to differentiate; (2) PDGF-A induces myofibroblast migration to the correct position in the alveolar niche; (3) myofibroblasts can migrate independently of PDGF-A but require it to execute their functions; (4) myofibroblast growth, proliferation and survival depend on PDGF-A (Boström et al., 1996). However, owing to the embryonic lethality of Pdgfa^−/− mice it has not been possible to prove those hypotheses, even though they are often referred to as dogmas.

Here, we present a lung-specific Pdgfa knockout mouse (Pdgfa^fl/−; Spc-cre), which live into adulthood, thereby surpassing the early lethality and putative systemic confounders of the global Pdgfa knockout mice. We show that deletion of PDGF-A in lung epithelial cells leads to impaired alveolar development resulting in enlarged alveoli lacking secondary septation. We suggest that PDGF-A signaling is required for the correct positioning and proliferation of myofibroblasts rather than their differentiation. In addition, as an indirect consequence of loss of the myofibroblast population, proliferation of AEC2s is increased, resulting in a higher number of AEC2s in mutant lungs.

In order to study the effect of Pdgfa deficiency on postnatal lung development in mice, it was necessary to circumvent the embryonic/perinatal lethality observed in Pdgfa^−/− mice (Boström et al., 1996). We have previously shown that Pdgfa is expressed in airway epithelial cells and alveolar epithelial type-2 cells (AEC2s) in the lung (Gouveia et al., 2017), so we decided to use the Spc-cre mouse (Okubo and Hogan, 2004) to generate a lung-specific knockout mouse of Pdgfa. The breeding strategy (illustrated in Fig. 1A) generated four different genotypes: lung-specific Pdgfa knockout mice (Pdgfa^fl/−; Spc-cre), two variants of Pdgfa heterozygous mice (Pdgfa^fl/+; Spc-cre and Pdgfa^fl/−) and mice equivalent to Pdgfa wild-type mice (Pdgfa^fl/+). Initial analysis revealed a normal Mendelian distribution of the offspring genotypes (Fig. 1A) and no mutation-dependent lethality was observed. Mouse weight was followed until adulthood and no significant differences were observed between the different genotypes (Fig. 1B). Here, and in all further analyses of the Pdgfa^fl/−; Spc-cre mice, we used littermate Pdgfa^fl/+ mice as controls.

These results indicated that Pdgfa^fl/−; Spc-cre mice did not display the perinatal lethality observed in the Pdgfa^−/− mice, and that they were a good model to study the role of PDGF-A during the development of the lung.

We compared the lungs of Pdgfa^fl/−; Spc-cre mice with Pdgfa^fl/+ mice before alveogenesis [postnatal day (P) 3], during alveogenesis (P7 and P10) and at the end of alveogenesis (P21). Quantitative PCR analysis of lung homogenates revealed an approximately 50% reduction in Pdgfa expression in Pdgfa^fl/−; Spc-cre lungs at each of the time points (Fig. 2A). These time points were also used for histological analysis of Hematoxylin and Eosin-stained lung sections, revealing morphological alterations in the lungs of Pdgfa^fl/−; Spc-cre mice. At P3, lungs of Pdgfa^fl/−; Spc-cre mice were histologically indistinguishable from littermate controls (Fig. 2B-D). Even though Pdgfa mRNA expression levels were significantly decreased at P3, neither the distal airspace area nor the surrounding walls were phenotypically different from the control littermates. At P7, a significant increase of the alveolar airspace was detected in mutant lungs (Fig. 2E-G). At P10, mutant lungs failed to assemble alveoli correctly, demonstrated by a significant increase in the alveolar airspace in combination with a reduced number of forming alveolar septa (Fig. 2H-J). By P21, when alveolar formation was nearly completed, the mean linear intercept (MLI) of Pdgfa^fl/−; Spc-cre lungs was considerably increased with significant enlargement of the alveolar airspaces and fewer secondary septa compared with control lungs (Fig. 2K-M).

It was previously shown that Pdgfa signaling is not required for branching morphogenesis and that Pdgfa^−/− lungs do not exhibit histological alterations before the onset of alveogenesis (Boström et al., 2002). However, it was unclear whether the observed phenotype in Pdgfa^−/− lungs was an effect of an embryonic requirement for PDGF-A. To pinpoint the time point at which PDGF-A signaling is required, we used an inducible transgenic cre recombinase mouse (Spc-cre^ERT2) to knockout Pdgfa postnatally. Pups received tamoxifen from P0 for 5 days through the mother's milk and analysis was performed at P10 (Fig. 3A). The ability to induce Pdgfa deletion was confirmed with qPCR. The Pdgfa^fl/−; Spc-cre^ERT2 knockout lungs exhibited a 50% decrease in Pdgfa mRNA expression levels (Fig. 3B), which was similar to that observed for Pdgfa^fl/−; Spc-cre mice (Fig. 2A). Pdgfa^fl/−; Spc-cre^ERT2 lungs also presented a significant increase in the alveolar airspace (Fig. 3C) together with alveolar simplification compared with control lungs (Fig. 3D,E). These results indicate that postnatal deletion of Pdgfa is sufficient to mimic the lung phenotype seen when Pdgfa is deleted during embryonic lung development, and that the effects of Pdgfa deletion are important during secondary septation.

To investigate whether epithelial deletion of Pdgfa affects the number of Pdgfra-positive cells, we combined Pdgfa^fl/−; Spc-cre mice with Pdgfra-GFP reporter mice. Pdgfra^H2B:GFP mice carry an H2B:GFP fusion protein knocked into the Pdgfra gene, thus PDGFRα⁺ cells display nuclear GFP expression. Pdgfra^H2B:GFP mice are also heterozygous null for Pdgfra (Hamilton et al., 2003).

We previously observed that the introduction of Pdgfra^GFP/+ worsened the lung phenotype in Pdgfc^−/− mice (Andrae et al., 2016), and we therefore predicted a similar effect on the Pdgfa^fl/−; Spc-cre mice. However, we could not find any evidence for an aggravated lung phenotype in Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lungs compared with Pdgfa^fl/−; Spc-cre at P10. There was no significant difference in either lung morphology (Fig. 4A-D), nor in the mean linear intercept (Fig. 4E).

The effect on Pdgfra expression was examined by counting Pdgfra^GFP-positive cells in the alveolar region. Already at P3, Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lungs exhibited reduced GFP expression (Fig. 4F-I) and a significant decrease in the number of Pdgfra^GFP-positive cells (5.4±0.6%) compared with control lungs (16.4±1.3%) (Fig. 4J). During alveogenesis, there is normally an increase in Pdgfra⁺ cells (Branchfield et al., 2016; Endale et al., 2017; Gouveia et al., 2017) in the alveolar region, reflecting a necessity for Pdgfra⁺ cells during this stage of development. At P7, Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lungs sustained a significant decrease in the expression of Pdgfra^GFP-positive cells, compared with control lungs (Fig. 4L-O). The number of Pdgfra^GFP-positive cells per total number of cells was 25.9±3% in control lungs, whereas in the mutant lungs there were only 14.9±1.6% (Fig. 4P). The decreased Pdgfra expression at both time points was confirmed with qPCR (Fig. 4K,Q).

We analyzed the expression of alpha smooth muscle actin (α-SMA) at P7. α-SMA expression in the alveolar region was detected in both control and knockout lungs, showing that even in the knockout conditions myofibroblasts were present (Fig. 5A-D). However, we observed striking differences between the patterns of expression. In control lungs, myofibroblasts showed high expression of Pdgfra-driven nuclear GFP, and these cells formed ring structures of α-SMA surrounding the forming alveoli, presumably creating the septal ridges responsible for alveolar septation (Fig. 5E,F). However, in Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lungs, expression of both nuclear GFP and α-SMA were markedly decreased (Fig. 5G,H). Pdgfra has previously been described as a marker also for lipofibroblasts in the mouse lung (Ntokou et al., 2015). To investigate the presence of lipofibroblasts in the mutant lungs, we stained for ADFP (perilipin 2). Expression of ADFP was lower in Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lungs than in the control lungs (Fig. 5I-L). These results indicate that the epithelial deletion of Pdgfa most likely affected more than one population of Pdgfra⁺ cells in the lungs.

To understand better the morphology of the α-SMA rings that surround the developing alveoli, we analyzed thick vibratome lung sections (100 μm) of mutant and control lungs. To our surprise, we found that α-SMA rings were present also in Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lungs (Fig. 6B,D), although this could not be appreciated in the thinner sections (Fig. 5C,D). At P3, both control and mutant lungs exhibited large α-SMA rings (≥50 µm in diameter) that surrounded the alveolar ducts with similar volume of actin per ring (Fig. 6A,B,I). The number and average diameter of the rings were different, however, with a significantly lower diameter in control lungs (65.8±2.7 µm) compared with mutant lungs (73.68±1.9 µm) (Fig. 6J). At P7, control lungs contained α-SMA rings with both low diameter (<50 µm) and large diameter (≥50 µm) surrounding the alveolar ducts (Fig. 6C,E,F,L). In contrast, mutant lungs never exhibited low diameter α-SMA rings (Fig. 6D,G,H,L). Moreover, in knockout lungs, the α-SMA rings were thinner and the volume of α-SMA per ring was significantly lower compared with the rings in control lungs (2403±162.5 µm³ versus 3455±252.1 µm³, respectively) (Fig. 6K).

The thick vibratome sections also enabled us to locate the Pdgfra⁺ cells that formed the α-SMA rings. Fewer GFP⁺ cells formed the α-SMA rings in Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lungs (Fig. 6G,H) than in controls (Fig. 6E,F).

Interestingly, in Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lungs very few GFP⁺ cells were observed outside of the α-SMA rings (Fig. 6D), but such cells were more frequently observed in control lungs (Fig. 6C,F). This observation supports the hypothesis of a decrease in other Pdgfra⁺ cell types, besides myofibroblasts.

Analysis of the cellular content of Pdgfa^fl/−; Spc-cre lungs at P3 revealed no difference in the number of cells per field image (Fig. S1A-C). However, at P7 Pdgfa^fl/−; Spc-cre lungs exhibited a significant reduction in the total number of cells per image field (Fig. S1D-F). The reduced number of cells in Pdgfa^fl/−; Spc-cre lungs was not a consequence of apoptosis, as there was no difference in expression of cleaved caspase 3 between control and Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lungs (data not shown). Proliferation was investigated by examining the expression of Ki67 (Mki67), but we could not detect any significant difference in the percentage of Ki67⁺ cells between control and Pdgfa^fl/−; Spc-cre lungs (12.85±1.3% and 12.72±1.9%, respectively; Fig. 7A,C,E,G,I). However, the number of Pdgfra⁺/Ki67⁺ cells was significantly reduced in the Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lungs (Fig. 7B, arrowheads and 7D). Only 3.5±0.6% of Pdgfra⁺ cells were also Ki67⁺ in the knockout lungs, compared with 7.1±0.9% in Pdgfa^fl/+; Pdgfra^GFP/+ lungs (Fig. 7J). These results indicate that Pdgfra⁺ cells were less proliferative in Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lungs. Interestingly, we observed an increase in the relative number of Spc (Sftpc)⁺ cells in knockout lungs at P7 (Fig. 7H,K). Moreover, proliferation of AEC2s (Spc⁺/Ki67⁺) in Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lungs was higher (13±3.1%) compared with controls (3.3±0.5%) (Fig. 7L) thus explaining why no decrease in overall proliferation was observed. AEC2s function as stem cells (Barkauskas et al., 2013) with the capacity of alveolar regeneration. It has been shown that EGFR/KRAS signaling can induce proliferation of AEC2s and is important for the self-renewal capacity of AEC2s (Desai et al., 2014). Quantitative PCR analysis revealed a significant increase in expression of the genes encoding the EGFR ligand HBEGF and KRAS in Pdgfa^fl/−; Spc-cre; Pdgfra^GFP/+ lung homogenates compared with control lungs (Fig. 7M,N), indicating activation of this pathway in the absence of Pdgfa signaling.

There is an increasing interest in deciphering the mechanisms of alveolar formation, motivated by the need to find therapies for diseases such as BPD and COPD. Recently, there have been several reports published describing PDGFRα expression in the lung (Branchfield et al., 2016; Endale et al., 2017; Gouveia et al., 2017; Green et al., 2016; Ntokou et al., 2015) highlighting the wish to understand better the role of PDGFRα signaling during lung development.

In this study, we created a new mouse model for studying the role of PDGF-A in the developing lung. By specifically deleting Pdgfa in the lung, we have been able to follow up on the seminal studies of Pdgfa^−/− mice (Boström et al., 1996, 2002; Lindahl et al., 1997) by bypassing the perinatal lethality associated with the global knockout of Pdgfa. In addition, we obtained data that provide new information about the role of PDGFRα signaling in assembly of the actin cytoskeleton of myofibroblasts, as well as for mesenchymal-epithelial crosstalk during lung development.

The Pdgfa^fl/−; Spc-cre mice survived until adulthood, but they still exhibited the same lung phenotypes as observed in the Pdgfa^−/− mice, suggesting that the lethality of the latter reflects other organ-specific or systemic effects of Pdgfa. Indeed, PDGF-A signaling has been implicated in developmental processes in many organs in addition to the lung, including the brain, gut, skin and testis (Fruttiger et al., 1999; Gnessi et al., 2000; Karlsson et al., 1999, 2000). Enlargement of the alveolar airspace and defective secondary septa formation were penetrant phenotypes in the Pdgfa^fl/−; Spc-cre mice even though Spc-cre-mediated deletion of Pdgfa only reduced the total expression of Pdgfa to approximately 50%. This reveals the importance of PDGF-A expression by Spc⁺ cells for alveolar formation. The remaining 50% Pdgfa expression could be attributed to expression in the proximal lung airway epithelia and smooth muscle cells, as previously described (Gouveia et al., 2017).

The observed phenotype was not a consequence of embryonic effects of Pdgfa deletion. We observed the same phenotype when inducing the deletion at birth as when deleting Pdgfa in the embryo. Moreover, mutant lungs were histologically indistinguishable from controls at P3, supporting our hypothesis that PDGF-A signaling is important mainly during the alveolar stage. Nevertheless, we observed a lower number of Pdgfra⁺ cells at P3, combined with a larger diameter of the α-SMA rings surrounding the forming alveolar structures. As these observations did not result in evident deficiency in the primary septa or duct formation, we believe that signaling pathways other than PDGF-A/PDGFRα are more important for the developmental processes preceding alveogenesis. For example, Notch inactivation in the distal epithelium causes enlargement of the alveolar structures combined with a decrease in proliferative Spc⁺ cells at P3 (Tsao et al., 2016), which we did not observe in our model.

We show in our model that Pdgfra expression decreased upon lung-specific deletion of Pdgfa, which seemed to affect different populations of Pdgfra⁺ cells such as myofibroblasts and lipofibroblasts. Our observations indicate that myofibroblasts are indeed capable of differentiating in Pdgfa^fl/−; Spc-cre lungs, as α-SMA expression was detected in the alveolar region at P7. The small α-SMA⁺ rings formed by Pdgfra⁺ myofibroblasts in control lungs had a diameter that fits with the forming septal ridges. These α-SMA⁺ rings were formed by multiple myofibroblasts similar to the multicellular contractile structures described in the healing of epithelial wounds. During the wound closure process, the cells at the border of a wound form one thick actin ring that contracts, thereby bringing the edges of the wound together (Tamada et al., 2007; Vedula et al., 2015). In a similar way, the small α-SMA⁺ rings formed by multiple myofibroblasts surrounding the alveoli contract and create a tension that restricts alveolar expansion, thus creating the septal ridges. The lack of formation of small α-SMA⁺ rings in Pdgfa^fl/−; Spc-cre lungs reflects the reduced ability of the myofibroblasts to surround the alveoli, contract and create the septal ridges in the correct manner. In the absence of contraction, the alveoli are unrestrained and free to expand and create the observed enlarged alveolar airspaces.

The mechanisms that trigger Pdgfra⁺ myofibroblasts to express α-SMA and form the contractile rings are still unclear. It is known that inhalation and exhalation cause mechanical stretch in the lung (Roan and Waters, 2011). Recently, it was proposed that fibroblast mechanosensitivity in vitro depends on Pdgfra expression (McGowan and McCoy, 2017). A likely hypothesis is therefore that respiration after birth induces mechanical stretch, which triggers Pdgfra⁺ myofibroblasts to migrate, spread around the developing alveoli and form multicellular rings of α-SMA. In our model (in the absence of PDGF-A signaling from AEC2s), Pdgfra was downregulated in myofibroblasts, which became less responsive to mechanical stretch and failed to occupy their correct position and create the necessary small α-SMA rings.

Pdgfra⁺ fibroblasts produce extracellular matrix proteins and analyses of gene expression revealed changes in, for example, Col1a1 and Eln in the Pdgfa^fl/−; Spc-cre mice (data not shown). This is being followed up in an upcoming project with a focus on adult mice and their response to injury.

We propose that the enlargement of the alveolar airspace in Pdgfa^fl/−; Spc-cre lungs is caused by unresponsive myofibroblasts in combination with deregulated AEC2 proliferation. In a recent study, a population of Pdgfra⁺ cells (named mesenchymal alveolar niche cells or MANCs) identified in close proximity to AEC2s played a role in alveolar epithelial self-renewal (Zepp et al., 2017). The authors identified several signaling pathways (with ligands expressed by MANCs and receptors expressed by AEC2s) supporting the crosstalk between these two cell types. We suggest that the deletion of Pdgfa affects this crosstalk and results in an indirect effect on the proliferation of Spc⁺ cells.

The increased proliferation of AEC2s together with the increased gene expression of EGFR ligand in Pdgfa^fl/−; Spc-cre mutant lungs point to a role for PDGF-A in the stem cell function of AEC2s. It has been shown that Pdgfra⁺ fibroblasts accelerate the conversion of AEC2s to AEC1s in vitro (Barkauskas et al., 2013). Additionally, Desai and colleagues showed that only a small fraction of AEC2s have the capacity to both self-renew and convert to AEC1s. The authors hypothesized that a dominant AEC2 stem cell prevents neighboring AEC2s from being activated and participating in the renewal of AECs (Desai et al., 2014). We further hypothesize that PDGF-A is involved in that neighboring cell suppression, as deletion of Pdgfa expression in Spc⁺ cells resulted in an increased number of AEC2s and altered proliferation. Further studies will be necessary regarding the stem cell capacity of Pdgfa⁺ AEC2s.

All experiments were approved by the Uppsala animal ethics committee (permit number C224/12 and C115/15) and performed according to Swedish legislation.

The following mice were used in this study: conditional Pdgfa knockout mice (Pdgfa^fl/fl) (Pdgfa^tm1Vlcg; Andrae et al., 2014), heterozygous global Pdgfa knockout mice (Pdgfa^+/−) (Pdgfa^tm1Cbet; Boström et al., 1996), Pdgfra^GFP reporter mice (Pdgfra^{tm11(EGFP)Sor}; Hamilton et al., 2003) constitutive Spc-cre-expressing mice (Spc-cre) [Tg(Sftpc-cre)^tm1Blh; Okubo and Hogan, 2004] and inducible Spc-cre-expressing mice (Spc-cre^ERT2) [Sftpctm1^{(cre/ERT2)Blh}; Rock et al., 2011]. Pdgfa^+/− mice were first crossed with either Spc-cre or Spc-cre^ERT2 to generate Pdgfa^+/−; Spc-cre and Pdgfa^+/−; Spc-cre^ERT2 mice, which were further crossed with Pdgfa^fl/fl. Genotypes were confirmed by PCR using the primers described in Table S1.

Pups received five doses of tamoxifen [100 µl of 20 mg/ml solution (Sigma-Aldrich, H6278) administered to the lactating mother on consecutive days] and lungs were harvested at the desired time points after induction.

Mice were anesthetized with a cocktail of ketamine (75 mg/kg; Ketaminol^® vet., Intervet AB, Sweden) and Dexdomitor (0.5 mg/kg; Orion Corporation, Finland) and perfused with PBS through the right ventricle of the heart, followed by perfusion with 4% formaldehyde (9713.5000, VWR). Lungs were inflated through the trachea with 4% formaldehyde and fixed at room temperature for 30 min (if X-gal staining was performed) or 4 h and washed in PBS at 4°C overnight. X-gal staining was performed as previously described (Gouveia et al., 2017). Lungs for cryosections were soaked in 30% sucrose and embedded in Neg-50 OCT (Richard Allan Scientific, 6502) whereas lungs for paraffin sections were dehydrated in a series of ethanol (70%, 95%, 99%) and xylene, followed by paraffin embedding. Lungs for vibratome sections were inflated with 2% low melting agarose, immersion fixed for 4 h in 4% formaldehyde at 4°C and washed in PBS overnight.

Paraffin sections (5 µm) were rehydrated in xylene and a series of ethanol (99%, 95%, 80%) for 4 min each. Sections were stained with Hematoxylin and Eosin and mounted in Neo-mount (109016, Merck). To calculate the MLI, bright-field images were obtained with a Zeiss Axioimager Microscope at 40× magnification. Eleven horizontal lines were superimposed on each image, and the intersections between the lines and the alveolar walls were counted. The number of lines was multiplied by the length of the lines and divided by the number of interception points. Thirty random images were taken from each lung and the MLI was manually calculated.

Rehydrated paraffin sections (5 μm) were subjected to antigen retrieval for 20 min in a water bath at 95°C with Target Retrieval Solution, citrate pH 6.0 (Dako Cytomation, S2369). Sections were washed in dH₂O afterwards and then subjected to the same treatment as cryosections. Cryosections (20 μm) were washed in PBS and blocked overnight in 1% bovine serum albumin (BSA) and 0.5% Triton X-100 at 4°C. Sections were incubated at 4°C overnight with primary antibodies, washed with PBS and incubated for 1 h at room temperature with secondary antibodies. After washes in PBS, sections were mounted in Prolong Gold anti-fade reagent with DAPI (Invitrogen, P36931). Vibratome sections (100 μm) were blocked overnight in 1% BSA and 0.5% Triton X-100 at 4°C, incubated with primary antibodies for 48 h at 4°C, washed in PBS and incubated with secondary antibodies overnight at 4°C. Sections were stained with Hoechst (1:10,000, Invitrogen, 33342) for 10 min and mounted in Prolong Gold anti-fade reagent (Invitrogen, P36930). Primary antibodies used were: mouse anti-human Asma (alpha-smooth muscle actin) conjugated to Cy3 (1:200, Sigma-Aldrich, C6198, Lot# 037M4783V), goat anti-SPC (1:100, Santa Cruz, sc7706, Lot# A2712), rabbit anti-Ki67 (1:100, Abcam, 15580, Lot# GR3184863-1), rabbit anti-cleaved caspase 3 (1:100, Cell Signaling Technologies, 9661, Lot# 45) and rabbit anti-ADFP (1:100, Thermo Fisher Scientific, PA1-16972, Lot# Sh2431603A). Secondary antibodies used were: Alexa Fluor 555-conjugated donkey anti-goat (1:200, Molecular Probes, Lot# 1697092); Alexa Fluor 647-conjugated donkey anti-rabbit (1:200, Molecular Probes, Lot# 1182665) and Alexa Fluor 568-conjugated donkey anti-rabbit (1:200, Molecular Probes, Lot# 1558655).

Confocal images were acquired using a TCS SP8 Laser Scanning Microscope (Leica) and images were processed using ImageJ software (NIH). Quantifications were performed using Imaris image analysis software (Bitplane, Oxford) with spots and surface modules. A total of ten images of 40× fields per lung were analyzed.

For gene expression analysis, whole left lung lobes were harvested and immersed in RNAlater (Qiagen). To extract RNA, the RNeasy Mini Kit (Qiagen) was used and cDNA was synthesized from 1 µg of total RNA using the iScript cDNA synthesis kit (Bio-Rad). Quantitative real-time PCR was performed using the CFX-96 Real Time System (Bio-Rad) and the following Taqman probes: Pdgfa (Mm00435540_m1, Thermo Fisher Scientific), Pdgfra (Mm00440701_m1, Thermo Fisher Scientific), Kras (Mm00517492_m1, Thermo Fisher Scientific) and Hbegf (Mm00439306_m1, Thermo Fisher Scientific). Each qPCR was replicated three times and sample results were normalized to 18 s rRNA endogenous control (X03201.1, Applied Biosystems). Fold change quantifications were performed using the Livak method (2^−ΔΔCt) (Schmittgen and Livak, 2008).

All quantification data assembly and statistical analyses were performed using GraphPad Prism version 6 for Mac (GraphPad Software). Student's t-test (unpaired and two tailed) was used to analyze significant differences between the means of two groups (control versus knockout). For each time point, all the analysis was performed with three different mice per group (three controls and three mutants). All results are presented as mean+s.e.m. P<0.05 was considered statistically significant.

We thank Jana Chmielniakova, Cecilia Olsson, Pia Peterson, Helene Leksell and Veronika Sundell for technical assistance with the mice. We also thank the BioVis core facility at Uppsala University for microscopy technical support.