Rap1 regulates mechanical signaling required for alveolar myofibroblast contraction

Postnatal alveolar morphogenesis depends largely on the contractile force produced by alveolar myofibroblasts⁹. Hence, to investigate whether alveolar myofibroblast differentiation is affected in Rap1^iECKO mice, we employed 3D immunostaining with a myofibroblast marker antibody [α-smooth muscle actin (α-SMA)] and found that myofibroblasts were present in terminal sacs at P4 and localized to the alveolar entry ring at P9 in both control and Rap1^iECKO mice (Fig. 2a and Supplementary Fig. ^2a). Consistently, autonomic nerves, which have recently been shown to regulate myofibroblast proliferation and migration⁴, developed normally in Rap1^iECKO mice (Supplementary Fig. ^2b). Flow cytometric analysis also revealed normal numbers of alveolar myofibroblasts [α-SMA(+)/PDGFRα(+) cells], myofibroblast progenitors [PDGFRα(+) cells], and smooth muscle cells [α-SMA(+) cells] in the lungs of Rap1^iECKO mice (Fig. 2b). These findings indicate that myofibroblast differentiation occur normally in Rap1^iECKO mice. We also found that the appearance of AT1 cells labeled with their marker antibody [Advanced Glycosylation End-Product Specific Receptor (RAGE)], did not differ between control and Rap1^iECKO mice at either P4 or P9 (Supplementary Fig. ^2c), revealing that AT1 cells are not involved in the impairment of alveologenesis in Rap1^iECKO mice.

Open in a separate window

Fig. 2

Endothelial Rap1 regulates mechanical signaling required for alveolar myofibroblast contraction.

a Confocal z-projection images for α-SMA in alveoli from P9 Rap1^fl/fl and Rap1^iECKO mice. Arrowheads, myofibroblasts surrounding the alveolar entry ring. b Upper, flow cytometry plots of myofibroblasts [α-SMA(+)/PDGFRα(+) cells], their progenitors [PDGFRα(+) cells] and smooth muscle cells [α-SMA(+) cells] in the lungs of P9 Rap1^fl/fl and Rap1^iECKO mice. Plots show Cy3-anti-α-SMA versus APC-anti-PDGFRα fluorescence. Lower, percentages of cell populations as indicated at the bottom (Rap1^f/f and Rap1^iECKO, n=6 and 5 mice). c Confocal z-projection images for α-SMA in alveoli from P9 Rap1^fl/fl and Rap1^iECKO mice. Arrowheads, α-SMA filaments. d Score of α-SMA bundles in alveoli (arbitrary units, A.U.), as in (c) (Rap1^f/f and Rap1^iECKO, n=8 and 6 mice). e Confocal z-projection images for pMLC(S19) (green), α-SMA (magenta), and Nuclei (DAPI, blue) in alveoli from P9 Rap1^fl/fl and Rap1^iECKO mice. Arrowheads, pMLC signal in myofibroblasts. f pMLC fluorescence intensity in α-SMA(+) areas (arbitrary units, A.U.), as in (e) (n=5 mice/each). g Confocal z-projection images for YAP (green), α-SMA (magenta), and Nuclei (DAPI, blue) in alveoli from P9 Rap1^fl/fl and Rap1^iECKO mice. Arrowheads, the nuclei of α-SMA(+) myofibroblasts. h Percentages of α-SMA(+), nuclear YAP(+), α-SMA(+)/nuclear YAP(+) cell populations, as in (g) (Rap1^f/f and Rap1^iECKO, n=6 and 8 mice). i Western blot analysis of YAP and β-actin in lungs of Rap1^fl/fl and Rap1^iECKO mice at P9 (3 mice/each). j, Expression levels of YAP relative to those of β-actin, as in i (Rap1^f/f and Rap1^iECKO, n=6 and 5 mice). k Representative images of collagen gels containing PDGFRα(+) cells from the lungs of P9 Rap1^fl/fl and Rap1^iECKO mice and cultured without (Minus) or with 10μM blebbistatin (Bleb.). l, Quantification of collagen gel areas (Rap1^f/f and Rap1^iECKO without/with Bleb., n=6/1 and 6/1 mice). Each dot represents an individual mouse, and data are presented as means±s.d. (b, d, f, h, j, l). N.S., no significance; *P<0.05 by two-tailed Student’s t-test (d, f, h, j, l). Boxed areas are enlarged in the insets (c, e, g). Scale bars; 80μm (a, e), 20μm (c, g and enlarged in c and e), 10μm (enlarged in g). Source data are provided as a Source data file.

However, we observed the α-SMA filaments in myofibroblasts to be markedly thinner in Rap1^iECKO mice than in controls (Fig. 2c, d and Supplementary Movie ¹, ²), pointing to a functional impairment of myofibroblasts in Rap1^iECKO mice. Alveolar myofibroblasts generate contractile force through activation of mechanical signaling involving MLCK-dependent phosphorylation of myosin light chain (MLC) and nuclear localization of YAP^10,11. Therefore, to investigate whether mechanical signaling in alveolar myofibroblasts is affected in Rap1^iECKO mice, we employed 3D immunostaining with antibodies against phosphorylated-myosin light chain (pMLC) and YAP. In control mice, myofibroblasts localized to the alveolar entry ring showed a strong pMLC signal (Fig. 2e, f). However, the pMLC signal in alveolar myofibroblasts of Rap1^iECKO siblings was dramatically decreased. Furthermore, the number of alveolar myofibroblasts exhibiting the nuclear YAP signal in Rap1^iECKO mice was also significantly lower than that in control mice (Fig. 2g, h). Cells other than myofibroblasts in the lungs of Rap1^iECKO mice also tended to have a weak nuclear YAP signal. Consistently, YAP protein levels in the lungs of Rap1^iECKO mice were lower than those in control mice, suggesting reduced nuclear YAP localization in the Rap1^iECKO mice, since YAP retained in the cytoplasm undergoes proteasomal degradation³¹ (Fig. 2i, j). These findings indicate that mechanical signaling in alveolar myofibroblasts is impaired in Rap1^iECKO mice.

Next, we investigated whether mechanical signaling in myofibroblasts of Rap1^iECKO mice is compromised in a cell-autonomous or a non-cell-autonomous manner. To this end, PDGFRα(+) cells including myofibroblasts and their progenitors were sorted by FACS from P8-9 lungs of control and Rap1^iECKO mice, and subjected to collagen gel contraction assay. The contractile ability of PDGFRα(+) cells derived from Rap1^iECKO mice was similar to that of control mouse-derived cells (Fig. 2k, l). Furthermore, PDGFRα(+) cells derived from Rap1^iECKO mice and those from control mice exhibited a similar degree of nuclear YAP localization when plated on a glass-base dish (Supplementary Fig. ^{2d, e}), indicating the mechanotransduction machinery to be preserved even in the Rap1^iECKO mouse-derived cells. These results suggest that mechanical signaling in myofibroblasts is non-cell-autonomously impaired in Rap1^iECKO mice.

ECs induce Col-4 accumulation to generate BMs acting as a scaffold for myofibroblasts in a Rap1-dependent manner

The next question to address was how ECs regulate mechanical signaling in alveolar myofibroblasts in a Rap1-dependent manner. To tackle this issue, we first analyzed spatial localizations of ECs and myofibroblasts in alveoli (Supplementary Fig. ^3a). Capillary ECs surrounded and created intimate contacts with AT1 cells to form alveoli in P9 lungs. Furthermore, myofibroblasts covered the outside of the alveolar space by making tight contacts with ECs. It can reasonably be assumed that BMs mediate an indirect contact between ECs and myofibroblasts. In addition, BMs are known to act as scaffolds for cells to regulate development and tissue morphogenesis^32–34. Importantly, the stiffness of the extracellular matrix forming BMs has also been shown to stimulate mechanical signaling involved in various cellular functions^35,36. Therefore, we hypothesized that ECs might regulate alveologenesis by constructing BMs acting as a scaffold for myofibroblasts. To address this hypothesis, we analyzed BMs during alveologenesis by visualizing Collagen Type IV Alpha 1 (Col4a1), a major BM component. Col4a1 was distributed throughout the alveoli in both control and Rap1^iECKO mice (Fig. 3a and Supplementary Fig. ^3b). In controls, Col4a1 was well assembled and formed sheet-like BMs, which closely surrounded the alveolar myofibroblasts (Fig. 3a). In clear contrast, Col4a1 resulted in the construction of disorganized BMs that were loosely attached to the alveolar myofibroblasts in Rap1^iECKO mice (Fig. 3a). Consistently, the single slice images revealed that control mice showed continuous BMs tightly attached to the alveolar myofibroblasts, while Rap1^iECKO mice exhibited a discontinuous appearance of BMs surrounding the myofibroblasts (Fig. 3b–d and Supplementary Fig. ^3c). We further investigated whether alveolar ECs in Rap1^iECKO mice fail to form continuous BMs by conducting mosaic gene deletion analysis. To achieve mosaic deletion of Rap1a/b in ECs, a low dose of tamoxifen was delivered from the lactating mother into Rap1a^fl/fl;Rap1b^fl/fl;Cdh5-CreERT2 postnatal mice with the mTmG reporter background (Rap1^iECKO;mTmG mice) at P1, enabling the identification of ECs lacking Rap1a/b with membrane GFP (mGFP) fluorescence (Supplementary Fig. ^3d). Rap1a^fl/+;Rap1b^fl/+;Cdh5-CreERT2 or Rap1a^fl/fl;Rap1b^fl/fl;Cdh5-CreERT2 mice with the mTmG reporter background (Rap1^iECHet;mTmG mice) served as a control. Col4a1 located around mGFP-labeled Rap1a/b-deficient ECs showed less assembly, forming discontinuous BMs, whereas that around the mGFP-labeled control ECs constructed continuous BMs (Supplementary Fig. ^{3e, f}). These results raise the possibility of defective assembly of Col4a1 into stable BMs in Rap1^iECKO mice. Additionally, we observed lower Col4a1 protein levels in the lungs of Rap1^iECKO mice than in control lungs, despite similar mRNA levels of Col4a1 and Col4a2 in these two groups (Fig. 3e–g). Given that matrix metalloproteinase-2, also known as 72kDa type IV collagenase, is known to be active during alveologenesis³⁷, it is reasonable to assume that Col4a1 in the alveoli of Rap1^iECKO mice undergoes digestion by collagenolytic enzymes, possibly due to impaired Col4a1 assembly. We also analyzed laminin, another BM component, and found that its localization in BMs was relatively well preserved even in Rap1^iECKO mice (Fig. 3b, c and Supplementary Fig. ^{3b, c}), revealing that ECs recruit Col-4, but not laminin, into BMs in a Rap1-dependent manner.

Open in a separate window

Fig. 3

ECs induce Col-4 accumulation to generate BMs acting as a scaffold for myofibroblasts in a Rap1-dependent manner.

a Confocal z-projection images for Col4a1 (green), α-SMA (red), ICAM2 (white), and Nuclei (DAPI, blue) in alveoli from Rap1^fl/fl and Rap1^iECKO mice at P9. Arrowheads, Col4a1-based BMs surrounding myofibroblasts. b Single slice images for CD31 (green), Laminin (red), Col4a1 (white), and Nuclei (DAPI, blue) in alveoli from Rap1^fl/fl and Rap1^iECKO mice at P9. Arrowheads, Col4a1 and Laminin associated with ECs. c Line lengths of Col4a1 (left) and Laminin (right) in CD31(+) areas, as in (b) (n=4 mice/each). d Single slice images for Col4a1 (green), α-SMA (red), ICAM2 (white), and Nuclei (DAPI, blue) in alveoli from Rap1^fl/fl and Rap1^iECKO mice at P9. Arrowheads, BMs localizing between ECs and myofibroblasts. e Western blot analysis of expressions of Col4a1 and β-actin in lungs of Rap1^fl/fl and Rap1^iECKO mice at P9 (3 mice/each). f Expression levels of Col4a1 relative to those of β-actin, as in (e) (Rap1^f/f and Rap1^iECKO, n=6 and 5 mice). g Expression levels of Col4a1, Col4a2, Rap1a, and Rap1b mRNA relative to those of Actb in lungs of Rap1^fl/fl and Rap1^iECKO mice at P9 (n=3 mice/each). h Confocal z-projection images for Elastin in alveoli from Rap1^fl/fl and Rap1^iECKO mice at P9. i Thicknesses of Elastin fibers, as in (h) (n=3 mice/each). j Expression levels of Eln mRNA relative to those of Actb in lungs of Rap1^fl/fl and Rap1^iECKO mice at P9 (n=3 mice/each). k Schematic diagram illustrating how ECs regulate the function of alveolar myofibroblasts. ECs induce BM formation by accumulating Col-4 in a Rap1-dependent manner. BMs constructed by ECs appear to serve as a scaffold for myofibroblasts to induce alveolar formation. Each dot represents an individual mouse, and data are presented as means±s.d. (c, f, g, i, j). N.S. no significance; **P<0.01, *P<0.05, by two-tailed Student’s t-test (c, f, g, i, j). Boxed areas are enlarged on the right (a, d, b) and in the insets (h). Scale bars; 10μm (a, d), 7μm (enlarged in a), 15μm (b), 50μm (h). Source data are provided as a Source data file.

Alveolar myofibroblasts secrete and assemble elastin into elastic fibers which are mechanically required for alveolar formation⁹. Thus, we next analyzed elastic fibers in alveoli and found that they localized around the alveolar entry ring in both control and Rap1^iECKO mice (Fig. 3h). However, quantitative analysis revealed that the elastic fibers surrounding the alveolar entry ring were thinner in Rap1^iECKO mice than in control mice, although expression of Elastin mRNA in the lungs differed minimally between these two groups (Fig. 3i, j and Supplementary Movie ³, ⁴), supporting the hypothesis that functions of alveolar myofibroblasts are impaired in Rap1^iECKO mice. Collectively, these findings suggest that ECs Rap1-dependently generate BMs required for myofibroblasts to induce mechanical signal-mediated alveologenesis (Fig. 3k).

Rap1 stimulates adhesive activity of integrin β1 in alveolar ECs

We next investigated the molecular mechanism by which endothelial Rap1 induces BM formation. Rap1 stimulates the adhesive activity of integrins to promote cell adhesion to the extracellular matrix^38,39. Importantly, a previous report showed that RAP-3, a C. elegans orthologue of mammalian Rap1, activates PAT-2 (α-subunit)/PAT-3 (β-subunit) integrins, which in turn promote Col-4 recruitment into BMs in a laminin-independent manner⁴⁰. Based on these findings, we hypothesized that endothelial Rap1 activates integrins to induce Col-4 recruitment to BMs. To test this hypothesis, we first evaluated the adhesive activity of integrins in ECs of control and Rap1^iECKO mouse lungs. To this end, ECs were isolated from the lungs of control and Rap1^iECKO mice, plated onto dishes coated with Collagen Type I (Col-1), and subjected to analysis of integrin-mediated cell adhesion. Rap1^iECKO mouse-derived ECs exhibited a significantly smaller cell area than control ECs (Fig. 4a, b). In addition, the number and size of focal adhesions (FAs) and focal complexes (FCs) were reduced in the ECs derived from Rap1^iECKO mice as compared to those of control ECs (Fig. 4a, c), suggesting decreased adhesive activity of integrins in Rap1-deficient lung ECs. Consistently, the number of activated integrin β1-positive FAs/FCs was significantly decreased in the lung ECs isolated from Rap1^iECKO mice as compared to those from control mice (Fig. 4d, e). These findings suggest that Rap1 is required for activation of integrins in lung ECs.

Open in a separate window

Fig. 4

Rap1 stimulates adhesive activity of integrin β1 in alveolar ECs.

a–c Formation of FAs and FCs in ECs isolated from the lungs of P9 Rap1^fl/fl and Rap1^iECKO mice and cultured on collagen-coated dishes for 72h. a Confocal fluorescence images for ICAM2 (green), F-actin (red), vinculin (white), and Nuclei (DAPI, blue). b Violin plots of the average cell area (Rap1^f/f and Rap1^iECKO, n=90 and 71 cells from 3 mice). c Number of FAs/FCs per cell, within the indicated size range at the bottom. The dots represent individual cells. Data are means±s.d. (Rap1^f/f and Rap1^iECKO, n=90 and 71 cells from 3 mice). d Confocal fluorescence images for activated integrin β1 (green), CD31 (red), vinculin (white), and Nuclei (DAPI, blue) in ECs isolated from the lungs of P9 Rap1^fl/fl and Rap1^iECKO mice and cultured for 72h. Arrowheads, the activated integrin β1-positive FAs/FCs. e Violin plots indicate the average number of activated integrin β1-positive FAs/FCs per cell, as in d (Rap1^f/f and Rap1^iECKO, n=23 and 24 cells from 3 mice). f Single slice images for mGFP (green), activated integrin β1 (magenta), and integrin β1 (blue) in alveoli from Rap1^iECHet;mTmG and Rap1^iECKO;mTmG mice at P9. Arrowheads, mGFP-labeled plasma membranes of ECs. g Violin plots showing the ratio of the number of activated integrin β1 to the total integrin β1 per cell (arbitrary units, A.U.), as in (f) (Rap1^iECHet;mTmG and Rap1^iECKO;mTmG, n=433 and 217 ROIs from 3 mice). h Single slice images for activated integrin β1 (green), Col4a1 (red), isolectin B4 (blue), and Nuclei (DAPI, white) in alveoli from Rap1^fl/fl and Rap1^iECKO mice at P9. Arrowheads, isolectin B4(+) ECs. In violin plots (b, e, g) bold and thin dashed lines indicate the median and quartiles, respectively. **P<0.01, by two-tailed Student’s t-test (b, c, e, g). Boxed areas are enlarged in the insets (a, d) and on the right (f, h). Scale bars; 20μm (a, d, h), 10μm (f and enlarged images in a, d, h) and 7μm (enlarged images in f). Source data are provided as a Source data file.

To further confirm the Rap1-mediated activation of integrins in alveolar ECs, we performed 3D immunofluorescence analysis of the lungs of Rap1^iECHet;mTmG and Rap1^iECKO;mTmG mice with the antibodies against integrin β1and its activated form. In both groups, integrin β1 was localized at the mGFP-labeled plasma membrane of alveolar ECs (Fig. 4f). However, the plasma membrane localization of the activated form of integrin β1 was significantly diminished in alveolar ECs of Rap1^iECKO;mTmG mice as compared to those of Rap1^iECHet;mTmG mice (Fig. 4f, g), indicating that Rap1 is required for activation of integrins in alveolar ECs. Furthermore, we observed that Col4a1 was assembled near the plasma membrane-localized activated integrin β1 in alveolar ECs of control mice (Fig. 4h). In contrast, alveolar ECs of Rap1^iECKO mice exhibited a reduced plasma membrane localization of activated integrin β1 and Col4a1 assembly was only slightly induced around these ECs (Fig. 4h). Considered collectively, our results suggest that Rap1 activates integrin β1 in alveolar ECs to induce BM formation.

Rap1 generates BMs through integrin-mediated recruitment of Col-4

We next aimed to elucidate whether Rap1 induces Col-4 recruitment by activating integrin in ECs. For this purpose, lung ECs isolated from control and Rap1^iECKO mice were cultured in Col-1-coated dishes for 72h and then subjected to immunofluorescence staining with Col4a1 antibody. Col4a1 was deposited and assembled on the surface of the dish by control lung ECs (Fig. 5a, b and Supplementary Fig. ^4a). However, the amount of Col4a1 assembled on the dish was significantly decreased when lung ECs derived from Rap1^iECKO mice were cultured (Fig. 5a, b and Supplementary Fig. ^4a). We also conducted the same experiments using human umbilical vein endothelial cells (HUVECs). Control HUVECs deposited and assembled COL4A1 on the culture dish surface (Fig. 5c, d). However, depletion of both RAP1A and RAP1B led to a reduction of COL4A1 assembled on the dish without affecting the mRNA levels of COL4A1 and COL4A2 (Fig. 5c, d and Supplementary Fig. ^4b–f). We also tested the possibility that the decreased amount of Col-4 assembled on the dish cultured with Rap1-depleted ECs is due to its degradation by performing Western blot analysis and found that the cellular protein levels of COL4A1 differed minimally between control and RAP1A/RAP1B-depleted HUVECs (Fig. 5e, f). In addition, similar amounts of COL4A1 protein were present in the conditioned media cultured with control and RAP1A/RAP1B-depleted HUVECs (Fig. 5e, f). These findings indicate that the depletion of Rap1 does not affect either the degradation or the secretion of Col-4 and thereby suggest that endothelial Rap1 actively regulates the assembly of Col-4.

Open in a separate window

Fig. 5

Endothelial Rap1 induces integrin-mediated recruitment of Col-4 to generate BMs.

a Confocal fluorescence images for Col4a1 (green), ICAM2 (magenta), and Nuclei (DAPI, blue) in ECs from the lungs of P9 Rap1^fl/fl and Rap1^iECKO mice and cultured on collagen-coated dishes for 72h. b The amounts of Col4a1 assembled on the dish, as in (a), are expressed as percentages of Col4a1-positive area relative to ICAM2-labeled total cellular area (Rap1^f/f and Rap1^iECKO, n=39 and 44 cells from 3 mice). c Confocal fluorescence images for COL4A1 (green), F-Actin (magenta), and Nuclei (DAPI, blue) in HUVECs transfected with control siRNA or two siRNA mixtures targeting both RAP1A and RAP1B (RAP1 KD#1, RAP1 KD#2) and cultured without or with 0.5mM MnCl₂ for 24h. d The amounts of COL4A1 assembled on the dish, as in (c), are expressed as a COL4A1-positive area divided by the number of cells in each image (Numbers of analyzed cells are provided in Supplementary Table ³). e Western blot analysis of COL4A1 and β-actin. Upper, COL4A1 in the conditioned media cultured without (Medium) or with HUVECs transfected with the indicated siRNAs; lower, COL4A1 and β-actin in the corresponding whole cell lysates. f The amounts of COL4A1 in the conditioned media and whole cell lysates normalized by those of β-actin, as in (e), are expressed relative to the levels observed in the control groups (n=3 independent experiments). g Confocal z-projection images for COL4A1 (green), F-Actin (magenta), and Nuclei (DAPI, blue) in vascular cord structures constructed by HUVECs transfected with the indicated siRNAs and cultured on Matrigel without or with 0.5mM MnCl₂ for 24h. Boxed areas are enlarged on the upper right. The cross-sectional images of the areas indicated by dotted lines are shown at the bottom. h Coverage of vascular cord structures with COL4A1, as in (g), are expressed as percentages relative to total vascular cord areas (Numbers of experiments are provided in Supplementary Table ³). Each dot represents an individual cell (b), image (d, h), and experiment (f). Data are presented as means±s.d. (b, d, f, h). N.S. no significance; **P<0.01, *P<0.05, by two-tailed Student’s t-test (b, d, h) and by one-way ANOVA followed by Turkey’s post-hoc test (f). Scale bars; 20μm (a), 30μm (c and upper enlarged images in g), 70μm (g), 15μm (lower enlarged images in g). Source data are provided as a Source data file.

Next, we investigated whether Rap1 induces Col-4 assembly through the activation of integrins. Manganese ions (Mn²⁺) reportedly increase the ligand-binding affinity of integrins⁴¹. Indeed, the treatment of HUVECs with Mn²⁺ increased the active form of integrin β1 localized at FAs (Supplementary Fig. ^4g). Thus, we examined the effect of Mn²⁺ treatment on COL-4 assembly and showed that it partially rescued the assembly of COL4A1 by RAP1A/RAP1B-depleted HUVECs (Fig. 5c, d). These results suggest that ECs deposit and assemble Col-4 through Rap1-mediated activation of integrins.

To further confirm this notion, we performed a Matrigel-based in vitro cord formation assay using HUVECs. HUVECs were plated on Matrigel, allowed to form capillary-like network structures for 36h, and then stained with anti-COL4A1 antibody to analyze BM formation. COL4A1 was accumulated around HUVEC-constructed cord-like structures to form BMs (Fig. 5g, h), a process completely inhibited by knockdown of COL4A1, indicating that COL4A1 assembled on the vessel wall had been deposited by HUVECs (Supplementary Fig. ^4h–j). Depletion of both RAP1A and RAP1B severely inhibited accumulation of COL4A1 around the cord-like structures, while not impacting the expression levels of COL4A1 and COL4A2 mRNAs in HUVECs (Fig. 5g, h, Supplementary Fig. ^4d). Furthermore, the Mn²⁺ treatment partially, yet significantly, rescued the diminished accumulation of COL4A1 around the cord-like structures in HUVECs depleted of both RAP1A and RAP1B (Fig. 5g, h). To further confirm the role of integrin β1 in BM formation, we treated HUVECs with blocking antibody against integrin β1 (mAb13)⁴², and found that COL4A1 accumulation around the cord-like structures was decreased by integrin β1 blocking antibody (Supplementary Fig. ^{4k, l}). Taken together, these findings indicate that Rap1 induces Col-4 accumulation around vessel walls to form BM by stimulating the adhesive activity of integrins.

HE staining

Lungs, liver, stomach, and small intestine were perfused with phosphate-buffered saline (PBS) through the right ventricle of the heart, then inflated by injecting 4% paraformaldehyde (PFA)/PBS containing NaN₃ (PBS/NaN₃) through the trachea, and fixed for 15min. Then, the whole organs were harvested and kept overnight in 4% PFA/PBS/NaN₃ at 4°C. After fixation, they were dehydrated and embedded in paraffin wax. The paraffin-embedded blocks of the organs were cut into 4μm tissue sections and stained with a standard HE staining procedure using 1% Eosin Y Solution (Wako) and Mayer’s Hematoxylin Solution (Wako) according to a previously described method³⁷. Then, the specimens were mounted in Canada Balsam (KANTO CHEMICAL CO., INC.) and imaged employing fluorescence microscopy (KEYENCE, BZ-X710).

Cell isolation from lungs

Cell isolation using flow cytometry was performed in accordance with a method described previously⁷⁰. Briefly, the lungs were perfused with ice-cold PBS via the right ventricle of the heart and inflated by injecting 1ml of 25U/ml Dispase (Thermo Fisher Scientific) dissolved in PBS through the trachea. The lungs were then excised, followed by removal of bronchi and other tissues. All lobes were transferred into a 6-well plate filled with 5ml of digestion buffer containing 100μg/ml Liberase^TM (Roche), 0.1mg/ml DNaseI (Sigma–Aldrich), and 25mM HEPES, and then extensively minced with scissors. Subsequently, the minced lungs were mixed by pipetting several times using a P1000 pipet equipped with a truncated blue tip, transferred into 50ml conical tubes, and incubated in a 37°C water bath for 30min, during which the lungs were extensively pipetted, every 5min, using a P1000 pipet tip. After the incubation, 5ml of FACS buffer (PBS containing 2% fetal calf serum) were added to stop the digestion reaction. Then, the suspended cells were passed through a 100μm mesh cell strainer and collected by centrifugation at 330×g for 5min. The cell pellet was resuspended with 5ml of red blood cell lysing buffer (Funakoshi) and incubated for 4min at room temperature (RT). After the incubation, the lysing reaction was stopped by adding 5ml of FACS buffer. The suspended cells were passed through a 40μm mesh cell strainer, collected by centrifugation at 330×g for 5min, and resuspended in 800μl of FACS buffer.

To achieve sorting of CD31(+) cells and PDGFRα(+) cells, PE-conjugated anti-CD31(BioLegend, 1/200) and APC-conjugated anti-PDGFRα (eBioscience, 1/200) antibodies were used for positive selection, while FITC-conjugated anti-CD45 (BioLegend, 1/200), FITC-conjugated anti-TER119 (BioLegend, 1/200), and FITC-conjugated anti-CD326/EpCAM (BioLegend, 1/200) antibodies were used for negative selection (Supplementary Table ¹). For immunostaining, the cells were incubated in FACS buffer containing the above-mentioned antibodies for 30min on ice with intermittent mixing. After washing with a 10-fold volume of FACS buffer, the cells were resuspended in 1ml of FACS buffer containing SYTOX Blue (Thermo Fisher Scientific, 1/500) to label the dead cells. Then, more than 5×10⁶ cells showing FITC(−), APC (−), SYTOX Blue(−), and PE(+) and those showing FITC(−), PE(−), SYTOX Blue(−), and APC(+) were collected as CD31(+) and PDGFRα(+) cell populations, respectively, employing a cell sorter, BD FACSAria^TM Fusion (BD biosciences), using a 100μm nozzle and FACS Diva software. Then, the sorted cells were pelleted by centrifugation at 330×g for 10min, and used for the experiments or stored at −80°C for RNA purification.

RNA preparation

To extract total RNA from mouse lungs, the lobes excised from P9 mice were homogenized in TRIzol RNA Isolation Reagents (Thermo Fisher Scientific) using a Physcotron homogenizer (MICROTEC CO., LTD.) for 30s and subjected to RNA isolation steps according to the manufacturer’s instructions. The extracted total RNA was further purified using the RNeasy Plus Kit (QIAGEN) and reverse-transcribed using SuperScript^TM III Reverse Transcriptase (Thermo Fisher Scientific) together with 10mM dNTPs, RNase OUT ribonuclease inhibitor (TOYOBO), and random primers (Thermo Fisher Scientific) according to the manufacturer’s instructions.

To isolate total RNA from FACS-sorted cells, 0.5–1×10⁷ cells were lysed in 300μl TRIzol RNA Isolation Reagents by employing extensive pipetting. Then, the cell lysates were mixed with the same volume of 100% EtOH, vortexed vigorously, and subjected to a further purification procedure using a Direct-zolTM RNA MicroPrep kit (Zymo Research) according to the manufacturer’s protocol. Purified RNA was reverse-transcribed using a SuperScript VILO cDNA synthesis kit (Thermo Fisher Scientific) according to the manufacturer’s instructions.

qPCR

Quantitative PCR (qPCR) was performed employing a CFX96 Touch Deep Well real-time PCR detection system (Bio-Rad) using KOD SYBR qPCR Mix (TOYOBO) or Thunderbird Next SYBR qPCR Mix (TOYOBO). The primers used for qPCR are listed in Supplementary Table ^S2.

3D immunofluorescence staining of alveoli

3D immunofluorescence staining of the lungs was conducted according to a previously described method⁷⁰. Briefly, P9 mice were anesthetized with an anesthetic reagent containing 75μg/ml medetomidine, 400μg/ml midazolam, and 500μg/ml butorphanol. After opening the chest cavity, the lungs were refluxed with PBS through the right ventricle to remove blood and subsequently perfused with 1ml of pre-warmed 6% gelatin in PBS. Then, the lungs were inflated by injecting pre-warmed 1% low melting agarose (LMT) (Nacalai tesque, Inc.) in PBS through the trachea, allowing the filling of alveolar tissues with the gel. After this process, the lungs were immediately chilled with crushed ice for 5min to solidify the agarose, excised, and fixed in pre-cooled 2% PFA in PBS at 4°C for 30min. After confirming that the gel-filled lungs had solidified, the lobes were separated from each other and placed directly on a cutting stage with glue. Then, 250–300μm-thick sections were cut from the lobes with a VT1200S vibratome (Leica), fixed in 4% PFA in PBS for 1h at 4°C, and washed with PBS/NaN₃. Subsequently, the tissue sections were warmed to 55°C for 30min in PBS/NaN₃ to melt out the agarose gel. After repeating this step once, the specimens were blocked with blocking solution (5% normal goat serum and 0.25% TritonX-100 in PBS/NaN₃) for 30min at 4°C. Then, individual tissue sections were placed in each well of a 24-well plate and immunostained with primary antibodies in 250μl blocking solution at 4°C overnight. After washing with PBS/NaN₃ at 4°C three times, 30min each time, the tissue sections were further incubated with secondary antibodies in 250μl blocking solution at 4°C overnight and washed with 500μl of PBS/NaN₃ at 4°C three times, 30min each time. The primary and secondary antibodies used for 3D immunofluorescence staining of alveoli are listed in Supplementary Table ¹. To mount the tissue sections on slides, mending tape with a square hollow for placing specimens was attached to the slides (Matsunami Glass). The tissue samples were placed in the hollows on the slides filled with handmade mounting media containing 2.5% DABCO, 1,4-diazabicyclo[2.2.2]octane, 50mM Tris–HCl (pH 8.0), and 20% glycerol or aqueous mounting media (Abcam), then covered with coverslips (Matsunami Glass), and sealed with nail polish. The stained alveoli were imaged employing a FLUOVIEW FV3000 confocal microscope (Olympus) equipped with an ×20 water objective lens (XLUMPLFLN) (Olympus) or a ×60 silicone objective lens (UPLSAPOXS2) (Olympus) operated with FLUOVIEW FV31S-SW software (Olympus). The obtained images were processed and analyzed employing Imaris software (Bitplane).

Flow cytometry-based quantification of myofibroblasts

Alveolar myofibroblasts, their progenitors, and smooth muscle cells in the lungs of postnatal mice were quantified by flow cytometry with modification of a previously described method³⁷. The cells were isolated from the lungs of P9 mice following the method described above in “Cell isolation from lungs” section. Then, the cells were collected by centrifugation at 330×g for 5min, blocked with FACS buffer containing 2% BSA at 4°C for 30min, and stained with APC-conjugated anti-PDGFRα antibody (Thermo Fisher Scientific, 1/100) in FACS buffer for 30min on ice. After washing with a 10-fold volume of FACS buffer, the cells were fixed and permeabilized with Foxp3 Fixation/Permeabilization Staining Buffer Set (eBioscience^TM) according to the manufacturer’s protocol. Then, the permeabilized cells were incubated with Cy3-conjugated anti-α-SMA antibody (Sigma–Aldrich, 1/200) in FACS buffer for 30min at RT. After washing with a 10-fold volume of FACS buffer, the cells were analyzed employing a flow cytometer, BD FACSAria^TM III or FACSAria^TM Fusion (BD Biosciences). The α-SMA(+)/PDGFRα(+) cells, PDGFRα(+) cells, and α-SMA(+) cells were defined as alveolar myofibroblasts, alveolar myofibroblast progenitors, and smooth muscle cells, respectively.

Western blot analysis

Western blotting was carried out basically according to a protocol described previously⁷¹. The lungs of P9 mice were harvested from the chest cavity. Then, the lobes were isolated by removing the trachea and bronchi, homogenized in RIPA buffer containing Tris–HCl (pH7.4), 150mM NaCl, 1% NP-40, 0.1% SDS, and 0.5% DOC for 30s with a Physcotron homogenizer (MICROTEC CO., LTD.) equipped with a microshaft, NS-4A (MICROTEC CO., LTD.), and centrifuged at 15,000rpm at 4°C for 15min. The supernatants were mixed with 5×SDS sample buffer containing 50% glycerol, 0.3M Tris–HCl (pH6.8), 11.5% SDS, 25% β-mercaptoethanol, and 0.5% bromophenol blue (diluted to 1×concentration), and boiled for 3min.

For the detection of RAP1A and RAP1B expression in HUVECs, the cells were trypsinized, collected by centrifugation, and lysed in 1×SDS sample buffer as described above. The samples were supplemented with DNaseI (final concentration, 100µg/ml, Sigma–Aldrich), boiled for 3min, and centrifuged at 15,000rpm for 5min, and the supernatants were collected. To detect COL4A1 secreted from HUVECs, conditioned media were mixed with 5×SDS sample buffer to achieve a 1× concentration. To analyze intracellular COL4A1, the HUVECs were collected and lysed in 1×SDS sample buffer as described above. Then, the samples were mixed with DNaseI, boiled for 3min, and centrifuged at 15,000rpm for 5min, and the supernatants were collected as described above.

Then, the samples were separated by SDS-polyacrylamide gel electrophoresis using TGX Stain-Free Precast Gel (Bio-Rad) or hand casting polyacrylamide gels using 40% Acrylamide/Bis Solution 37.5:1 (Bio-Rad), transferred to Immobilon-P PVDF Transfer Membrane (Merk Millipore) and immunoblotted with anti-YAP (Novus Biologicals, 1/1000), anti-Col4a1 (Merk Millipore, 1/1000), β-actin antibody (BD Bioscience, 1/2000), and Horseradish peroxidase (HRP) -linked 2nd antibodies as listed in Supplementary Table ^S1. Detection was performed using HRP substrate, Immobilon^Ⓡ Forte (Millipore) or ImmunoStar Reagents (Wako) using Amersham^TM ImageQuant^TM 800 (IQ800) or ChemiDoc Tough MP Imaging system (Bio-Rad). All uncropped immunoblot scans can be found in the Source Data file.

Collagen gel contraction assay

PDGFRα(+) cells (1×10⁶ cells) isolated from the lungs of P7-8 mice were resuspended in 60μl of growth medium [DMEM (Wako or GIBCO) containing 10% FBS and antibiotics (Wako)]. Collagen gel solution was prepared by mixing 100μl of 5mg/ml Rat tail Collagen I (R&D SYSTEMS SYSTEMS), 2.3μl of 1N NaOH, and 11.36μl of 10×PBS (pH 7.4) and then kept at 4°C. Then, 68μl of collagen gel solution, 172μl of growth medium, and 60μl of the cell suspension were sequentially added to make 300μl of cell mixture (final concentration of collagen I, 1mg/ml). The cell mixture was gently transferred into a 24-well plate, dried for 1h at 37°C under 5% CO₂, and covered by 500μl of growth medium. To suppress myosin-mediated cell contraction, 10μl of M blebbistatin (Sigma–Aldrich) were included in the growth media when indicated. After the culture for 12h at 37°C under 5% CO₂, the gels containing cells were released from the edge of the wells by using a sterile needle. After further incubation for 72h, collagen gels containing the cells were fixed with 4% PFA in PBS at RT for 15min. Images of the gels were acquired with a SZX16 stereomicroscope (Olympus) equipped with a digital camera (Tough, Olympus). To quantify the contractile ability of PDGFRα(+) cells, the area of collagen gels was measured by using Fiji software (NIH).

Analysis of cellular localization of YAP in PDGFRα(+) cells derived from the lungs

The PDGFRα(+) cells (1×10⁶ cells) isolated from the lungs of P9 mice were resuspended in 100μl of growth medium (DMEM containing 10% FBS and antibiotics), plated onto a 35-mm glass-base dish (Iwaki, ASAHI GLASS Company, Ltd.), and cultured for 48h at 37°C under 5% CO₂. Then, the cells were fixed with 4% PFA in PBS for 15min at RT, washed with PBS/NaN₃, and permeabilized 0.1% TritonX-100/PBS/NaN₃ for 5min at RT. After blocking with 5% normal donkey serum/ PBS/NaN₃ for 30min at RT, the cells were stained with anti-YAP antibody (Novus Biologicals, 1/200) in 5% normal donkey serum/PBS/NaN₃ at 4°C overnight. After washing with PBS/NaN₃ three times, the cells were stained with AlexaFluor ^TM 488-labeled Goat anti-Rabbit IgG (Thermo Fisher Scientific, 1/500), Cy3-conjugated anti-α-SMA antibody (Sigma–Aldrich, 1/400), and DAPI (Thermo Fisher Scientific, 1/1000) at 4°C overnight. Subsequently, the cells were washed with PBS/NaN₃ three times. Fluorescence images were obtained using a FV1200 confocal inverted microscope equipped with a 60× oil objective lens (UPlanSAPO 60x/1.35na oil Objective, Olympus) and GaAsP photomultiplier tubes operated with FLUOVIEW Ver. 4.2c software (Olympus).

The cellular localization of YAP was quantified using the Fiji software as described in Supplementary Fig. ^7b. Total cellular regions and nuclear regions (ROI^nuc) were determined based on the fluorescence signals derived from α-SMA and DAPI, respectively. The cytoplasmic regions (ROI^cell) were defined by subtracting ROI^nuc from the total cellular regions. Each area size and signal intensity were quantified by the measure function (Area, mean gray value) within overlayed ROI^cell and ROI^nuc. Mean gray value within ROI^nuc was used as nuclear intensity. For calculating intensities in cytoplasm, each mean gray value was multiplied by that area size, and then the products were all added as total intensities for ROI^cell and ROI^nuc. Subtracting total intensities and total area sizes of ROI^nuc from ROI^cell, mean gray values for cytoplasm was calculated per fields (Supplementary Fig. ^7b).

Analysis of focal adhesion and focal complex formation in ECs derived from the lungs

The cell suspension from the lungs of P9 mice was prepared as described in “Cell isolation from lungs” section. Subsequently, CD31(+) cells were purified using the MACS system, following the protocol described in the previous report with slight modifications⁷². The lung cells were collected by centrifugation at 330×g for 5min and resuspended with 1ml of MACS buffer [0.1% BSA (Wako)/D-PBS (−) (Nacalai tesque, Inc.)]. Subsequently, the cell suspensions were mixed with 25μl CD31 Microbeads (Miltenyi Biotec) and incubated for 15min at RT with gentle mixing on a bioshaker (TAITEC). After adding 5ml of MACS buffer, the cell suspensions were applied onto MACS^Ⓡ column placed in a magnetic field for positive selection and washed three times with 1ml of MACS buffer. Then, the retained cells were eluted from the column by removing it from the separator, collected by centrifugation at 330×g for 5min, and resuspended with EC media containing 20% FBS, DMEM (Wako), Penicillin/Streptomycin (Wako), 100μg/ml heparin, 1×ECGS (Sigma–Aldrich). Then, the cells were placed on a 24-well plate coated with 0.1% Gelatin (Wako). After culturing for 4 days with a medium change every other day, the cells were replated on a 35-mm glass-base dish coated with Col-1 and cultured for an additional 3 days in EC media.

To analyze formation of FAs and FCs, the cells were fixed with 4% PFA/PBS at RT for 15min, permeabilized with 0.1% TritonX-100/PBS at RT for 5min, and washed with PBS. Then, the cells were blocked with PBS containing 2% normal donkey serum at RT for 1h, and stained with either anti-vinculin (Sigma, 1/400) and anti-ICAM2 (BD biosciences, 1/200) antibodies or anti-activated integrin β1 (BD biosciences, 1/200), anti-vinculin, and anti-PECAM1 antibodies (R&D SYSTEMS SYSTEMS, 1/400) at 4°C overnight. After washing with PBS five times, vinculin, ICAM2, activated integrin β1, and PECAM1 were visualized by incubating with their corresponding fluorescence-labeled secondary antibodies [AlexaFlor^TM647 labeled Donkey anti-mouse IgG (Jackson ImmunoResearch, 1/500), AlexaFlor^TM488 labeled Donkey anti-rat IgG (Jackson ImmunoResearch, 1/500), AlexaFlor^TM488 labeled Donkey anti-goat IgG (Jackson ImmunoResearch, 1/500)] diluted in blocking buffer at 4°C overnight. Rhodamine-phalloidin (Thermo Fisher Scientific, 1/500) and DAPI (1/1000) were also included in the reaction buffer to visualize F-actin and nuclei.

Fluorescence images were obtained using a FV1200 confocal inverted microscope as described in the above section. The sizes and numbers of cells and FAs/FCs were calculated by exploiting the surface function equipped in the Imaris software. Surfaces for cell size and FAs/FCs were created based on the positive staining area for ICAM2 and vinculin, respectively, and calculated automatically.

siRNA-mediated gene knockdown

HUVECs obtained from KURABO were cultured in HuMedia-EG2 (KURABO) as described previously⁷³. For siRNA-mediated gene knockdown, 6×10³ HUVECs were seeded on 60-mm dishes, cultured for 24h, and transfected with control siRNA (Dharmacon), two different sets of siRNA mixtures targeting both RAP1A and RAP1B (Dharmacon) or for COL4A1 using Lipofectamine^TM RNAiMAX Transfection Reagent (Thermo Fischer Scientific) according to the manufacturer’s instructions. The target sequences of siRNA used in this study are listed in Supplementary Table ². siRNA-transfected HUVECs were used for the experiments after 24h culture.

Col-4 deposition assay

ECs isolated from the lungs as described above were seeded on a 35-mm glass-base dish coated with Col-1 at the density of 5×10³ cells/dish and cultured in EC media for 1-2 days. Then, the cells were fixed with 4% PFA/PBS at RT for 15min, blocked with PBS containing 5% normal donkey serum at RT for 1h, and stained with anti-Col4A1 (Millipore, 1/400) and anti-ICAM2 (1/200) antibodies for 24h at 4°C. After washing with PBS three times, the cells were stained with Alexa Fluor^TM488 labeled Donkey antibody against goat IgG (Jackson ImmunoResearch, 1/1000) and Alexa Fluor^TM647 labeled Donkey antibody against rat IgG (Jackson ImmunoResearch, 1/1000), and DAPI (1/1000) diluted in blocking buffer at 4°C overnight. In the experiments shown in Supplementary Fig. ^4a, the cells were permeabilized with 0.1% TritonX-100/PBS at RT for 5min to detect cytoplasmic Col4a1.

Similarly, HUVECs transfected with control siRNA or a mixture of RAP1A and RAP1B siRNAs were plated on a Col-1-coated glass-base dish and cultured in HuMedia-EG2 without or with MnCl₂ for 1–2 days. After the fixation, the cells were blocked and stained with anti-Col4A1 (Millipore, 1/400) antibody for 24h at 4°C. Subsequently, the cells were stained with Alexa Fluor^TM488 labeled Donkey antibody against goat IgG (Jackson ImmunoResearch, 1/1000), rhodamine-phalloidin (1/500), and DAPI (1/1000) diluted in blocking buffer at 4°C overnight. In the experiments shown in Supplementary Fig. ^4e, HUVECs were permeabilized with 0.1% TritonX-100/PBS at RT for 5min to detect cytoplasmic COL4A1.

Fluorescence images were obtained using a FV1200 confocal inverted microscope equipped with 40× objective lens (UPlanSApo 40x/0.95na Objective, Olympus) or 60× oil objective lens (UPlanSAPO 60x/1.35na oil Objective) and GaAsP photomultiplier tubes operated with FLUOVIEW Ver. 4.2c software. Deposited area of Col4a1 was measured by Fiji software. The images for Col4a1 were converted into 8-bit images, binalized by adjusting threshold at 25, changed scale based on the scale bar, and measured the area by executing Analyze particles. In addition, the images for DAPI and ICAM2 were analyzed to quantify the number of cells and total cellular areas, respectively.

Matrigel-based in vitro cord formation assay

For siRNA-mediated gene knockdown, 6×10³ HUVECs were seeded on 60-mm dishes, and transfected with control siRNA, two different sets of siRNA mixtures targeting both RAP1A and RAP1B or for COL4A1. siRNA-transfected HUVECs were used for the experiments after 24h culture.

To induce the formation of cord-like structures, 6×10³ HUVECs were placed on the Matrigel (BD biosciencess) poured onto a 35-mm glass-base dish, incubated for 45min to achieve adherence to the gel, and gently covered with 2ml of growth medium. To inhibit the adhesive activity of integrin β1, 0.5μg/ml blocking antibody against integrin β1 (mAb13) (Millipore) was added to the growth medium 3h after cell plating on the Matrigel. The cells were cultured for 36h to establish cord-like structures, fixed with 4% PFA/PBS at RT for 15min, and permeabilized with 0.1% TritonX-100/PBS at RT for 5min. After washing with PBS, the cells were blocked with PBS containing 2% normal donkey serum at RT for 1h and stained with anti-Col4A1 antibody (Millipore, 1/400) for 3h at RT. After washing with PBS several times, the cells were stained with Alexa Fluor^TM488 Donkey antibody against goat IgG (Thermo Fisher Scientific, 1/1000), 4′,6′-diamidino-2-phenylindole (DAPI) (Dojindo, 1/1000), and Alexa Fluor^TM546 Phalloidin (Thermo Fisher Scientific, 1/500). Images were acquired employing a FV3000 or FV1200 confocal upright microscope equipped with a 20× water objective lens (XLUMPLFLN) and GaAsP photomultiplier tubes and operated with FLUOVIEW FV31S-SW software.

To quantify the coverage ratio of endothelial tubes with COL4A1, the positive area stained with anti-COL4A1 or rhodamine-phalloidin was separately delineated using Fiji software. Subsequently, the area positive for COL4A1 was divided by that of rhodamine-phalloidin.

Quantitative analysis of imaging data

The mean linear intercept (MLI) was measured to quantify alveologenesis in postnatal mice. To calculate MLI, lungs of P14 mice were stained with anti-Col4a1 antibody (Millipore, 1/400), AlexaFluor^TM 647 conjugate of Wheat germ agglutinin (WGA) (Thermo Fisher Scientific, 1/500), and Rhodamine or 488-Phalloidin (Thermo Fisher Scientific) as described in “3D immunofluorescence staining of alveoli” section. Proteins reacting with anti-Col4a1 antibody were visualized with AlexaFluor^TM 488, 546, 633 or 647-conjugated secondary antibody (Thermo Fisher Scientific or Jackson ImmunoResearch). Fluorescence images were obtained employing a FV3000 confocal microscope equipped with a ×20 water objective lens (XLUMPLFLN) as described in “3D immunofluorescence staining of alveoli” section. Six randomly selected areas per mouse were imaged. MLI was quantified based on the imaging data according to a method described previously³⁷.

To quantify the fluorescence intensity of pMLC(Ser19) in myofibroblasts, α-SMA-positive areas excluding arteriole regions were extracted from the fluorescence images of α-SMA using the Fiji image processing software package and defined as myofibroblast areas. Fluorescence intensity of pMLC(Ser19) within myofibroblast areas was measured and expressed as average pixel intensity.

The connectivity of Col4a1 and Laminin was calculated with the Fiji program using the plug-in function of Ridge detection that automatically draws the maximum line based on the fluorescence intensity. The average lengths of continuous signals for Col4a1 or Laminin in alveolar ECs were calculated using the plug-in function designated as Ridge detection function implemented in the Fiji software (NIH). Three slices at the different z positions were selected from 3D images stained with the antibodies against Col4a1, Laminin, and CD31 or Isolectin B4 by use of Imaris software (Bitplane). The signals of Col4a1 and Laminin within the CD31 or Isolectin B4-positive area were extracted using the surface function with which Imaris software is equipped. The images were taken with the snapshot tool, saved in TIFF format, and then quantitatively analyzed with the Fiji software. First, the images were converted in color mode from RGB to 8-bit grayscale. Ridge detection was conducted without binarization. The values of Sigma, Lower Threshold, and Upper Threshold were 1.94, 0.50, and 1.53, respectively. The SLOPE method was not used for this analysis. All lines (270–880 lines) in 3 images per individual mouse were measured and averaged as the Line length. In addition, the connectivity of Col4a1 in the plasma membrane of lung ECs shown in Supplementary Fig. ^3f was similarly quantified. The ROI^mGFP were applied to the channel images of Col4a1. Then, the connectivity of Col4a1 within the ROI^mGFP was calculated using the plug-in function of Ridge detection as described above.

α-SMA-positive cells, nuclear YAP-positive cells, and α-SMA-positive cells exhibiting the nuclear YAP signal were quantified using Imaris software (Bitplane). Numbers of α-SMA-positive cells and nuclear YAP-positive cells were quantified by counting the nuclei within α-SMA-positive areas and within YAP-positive areas, respectively. The nuclei exhibiting both the α-SMA signal and the YAP signal were also counted to quantify the number of α-SMA-positive cells exhibiting the nuclear YAP signal.

The average sizes and numbers of FAs and FCs in CD31(+) cells were quantified as described in “Analysis of focal adhesion and focal complex formation” section.

The quantification of α-SMA bundles was performed with Fiji software as described in Supplementary Fig. ^7a. Using an external plug-in, FeatureJ, compute the smallest eigenvalue of Hessian tensor to extract outlines of alveolar rings, using external plug-in, FeatureJ. Gaussian derivatives (the standard deviation σ=20 pixels) were calculated in the process. The total outlines of alveoli were extracted from the images stained with Cy3-α-SMA employing the constant parameters (Smallest eigenvalue of Hessian tensor, Smoothing scale, 20 pixels). The extracted image further binarized with the adjustment of threshold<−0.16 with a default mode, linearized by using skeletonize and saved as Outlines of α-SMA ring. Aside from that, modifying smoothing scale up to 10 pixels, the bundles of Cy3-α-SMA were extracted using the original image, adjusted threshold<−3.0 in a default mode, linearized by using skeletonize function, and saved the image as Outlines of α-SMA bundles. The length of the both outlines was measured by a manual tracking for each alveolus, i.e., the length of Outlines of α-SMA ring and Outlines of α-SMA bundles. The score was estimated by the division of the length of Outlines of α-SMA bundles by the corresponding Outlines of α-SMA ring in each ring.

The thickness of elastin fibers was quantified as described in Supplementary Fig. ^7c. The volume positively stained with AlexaFluor^TM633-Hydrazide was estimated using Imaris software (Bitplane). Using the projection image from the corresponding 3D images, the area of Hydrazide-positive area was calculated by exploiting Analyze particles function implemented in Fuji software. Then, the volume was divided by the area per field to obtain the total thickness per field. All images used for the quantification were obtained by a FV3000 confocal microscope equipped with a ×20 water objective lens (XLUMPLFLN) with a ×3 magnification.

The ratio of activated to total integrin β1 in the plasma membrane of lung ECs shown in Fig. 4f was quantified as described in the Supplementary Fig. ^7d. Initially, fluorescence images of mGFP were converted into an 8-bit scale. Subsequently, mGFP-positive areas were defined through the binarization, setting the lower limit as 30. After establishing a pixel scale, the Analyzed particles function was applied with a lower limit of 1.5, and the ROIs for mGFP (ROI^mG) were automatically added. The ROI^mGFP were applied to both channels for activated integrin β1 and total integrin β1. Intensities within the ROI^mGFP were calculated without binarization.

Statistics and reproducibility

All experiments in this study were independently repeated more than three times with similar results. All confocal z-projection images and single slice images are representative of more than three independent experiments (Figs. 1f and ​and2a,2a, more than five independent experiments; Figs. 3a, ​a,3d,3d, ​d,4h,4h, and ​and7a,7a, three independent experiments; Fig. 6c, four independent experiments). All data are presented as means±SD unless otherwise noted. Statistical analyses were performed employing the unpaired two-tailed Student’s t-test and one-way ANOVA followed by Turkey’s post-hoc test. Survival outcomes in Rap1^iECKO mice were estimated by the Kaplan–Meier method followed by the log-rank test using GraphPad Prism 10 10.0.2 (GraphPad software, Inc). Data were considered statistically significant if the p-value was less than 0.05. No significant difference, p<0.05 and p<0.01 are shown as N.S., *, and **, respectively. Detailed statistic information is provided in Supplementary Table ³.

We thank K. Matsumoto (Nippon Medical School) for technical support in the preparation of paraffin sections, K. Sawamato (Nagoya City University Graduate School of Medical Sciences) and N. Kaneko (Doshisya University) for their aid in the experiments with floxed Itgb1 mice, Y. Taguchi (BD Biosciences) for supporting the experiments using FACSAria^TM Fusion, and A. Sawaguchi (University of Miyazaki) for his advice regarding the experiments requiring electron microscopy. We also thank H. Ichimiya, Y. Matsushita, K. Kato, M. Fukumoto, and M. Sone for excellent technical assistance. This work was supported by Grants-in-Aid for Scientific Research (B) to S.F. (21H02665), Scientific Research (C) to H.W.-T. (23K06325), for Exploratory Research to S.F. (21K19358, 23K18245) from the Japan Society for the Promotion of Science; by FOREST Program from Japan Science and Technology Agency (JST) under Grant Number JPMJFR220T to H.W.-T.; by research grants from Takeda Science Foundation to H.W.-T. and S.F., from the Mochida Memorial Foundation to H.W.-T., from the Cell Science Research Foundation to H.W.-T., from the Naito Foundation to S.F., from Daiichi Sankyo Foundation of Life Science to S.F., from Astellas Foundation for Research on Metabolic Disorders to S.F., from The Uehara Memorial Foundation to S.F., from the TERUMO LIFE SCIENCE FOUNDATION to S.F., from the NOVARTIS Foundation (Japan) for the Promotion of Science to S.F. and H.W.-T.