Stretch regulates alveologenesis and homeostasis via mesenchymal G_αq/11-mediated TGFβ2 activation

Normal alveologenesis requires tight spatiotemporal control of numerous molecular signalling pathways, and coordinated crosstalk between multiple cell types. Any perturbation to these complex processes can disrupt alveolar formation, resulting in structural and functional abnormalities in the gas exchange regions of the lungs. Such abnormalities contribute to perinatal death and lifelong lung function disturbances in survivors (Lovering et al., 2014). The alveolar stage is the final phase of lung development, during which primitive pulmonary sacculi are divided by newly formed secondary septa to form mature alveoli. Alveolarisation occurs between 36 weeks gestation and around 6 years of age in humans (Donahoe et al., 2016), and from postnatal day (P) 3 to P30 in mice (Beauchemin et al., 2016; Li et al., 2015; Pozarska et al., 2017); therefore, postnatal exposures and stimuli are key influences in alveolar development. Many pathways that drive normal lung development are also instrumental in adult lung repair (Chanda et al., 2019); therefore, understanding normal lung development could have implications for numerous pulmonary diseases.

Mesenchymal cells include various cell types that are integral to normal developmental processes, and pericytes are perivascular cells widely considered to be mesenchymal precursors in the lung (Barron et al., 2016; Kato et al., 2018; Ricard et al., 2014). Pericytes express platelet-derived growth factor receptor-β (PDGFRβ), PDGFRα, and neural/glial antigen 2 (NG2; also known as CSPG4), among other markers. However, the most specific marker for pericytes is PDGFRβ (Riccetti et al., 2020), and PDGFRβ co-expression with other pericyte markers correlates with the expected location of pericytes in the lung (Henderson et al., 2013). Pericytes migrate and differentiate into parenchymal myofibroblasts in the lung, as well as other mesenchymal cell types. Myofibroblast-driven deposition of extracellular matrix (ECM) proteins, such as collagen and elastin, provide the scaffolds for secondary septation during lung development and lung repair (Mecham, 2018; Mizikova and Morty, 2015). Therefore pericytes, and the mesenchymal cells that are derived from them, are instrumental in alveologenesis and lung homeostasis.

The pleiotropic growth factor transforming growth factor-β (TGFβ) regulates numerous developmental and repair processes, including the proliferation, migration and differentiation of pericytes and other mesenchymal cells (Bartram and Speer, 2004), as well as the generation of ECM. TGFβ signalling is tightly regulated in vivo by the production of TGFβ in latent form, and the three mammalian TGFβ isoforms, TGFβ1, TGFβ2 and TGFβ3, must be activated to exert their biological effects. Although it is known that TGFβ signalling is essential for multiple processes in alveolar development and repair (Bartram and Speer, 2004), the mechanisms that control TGFβ activation in alveologenesis are unclear.

Latent TGFβ is activated when a conformational change to the large latent complex alters the relationship between TGFβ and the latency-associated peptide, allowing TGFβ to interact with its receptor. The G protein α subunits G_αq and G_α11 (G_αq/11) mediate TGFβ activation in response to G protein-coupled receptor (GPCR)–ligand binding as well as mechanical stretch in epithelial cells (John et al., 2016; Xu et al., 2009). GPCR signalling has also been implicated in normal alveologenesis (Funke et al., 2016). Cyclical mechanical stretch (CMS) has been shown to induce TGFβ activation in lung slices via a Rho-associated kinase (ROCK)- and αv integrin-dependent process (Froese et al., 2016), although the contribution to this by individual cell types is unknown. Although stretch secondary to foetal breathing movements in utero has been shown to be essential for early lung development (Donahoe et al., 2016), the role of breathing-related CMS specifically in mesenchymal cells in alveolar development and the maintenance of adult alveoli has not been investigated.

We hypothesised that G_αq/11 would mediate CMS-induced TGFβ activation via ROCK and integrin signalling in mesenchymal cells, and that this would be important in alveologenesis and lung homeostasis. Here, using mesenchymal G_αq/11 knockout mouse models and an in vitro CMS system, we show that mesenchymal G_αq/11 is essential for normal alveologenesis and maintenance of adult alveoli via CMS-induced TGFβ signalling, but that this occurs in a ROCK- and integrin-independent manner via a pathway likely to involve the TGFβ2 isoform.

To understand whether mesenchymal G_αq/11 deletion resulted in detrimental effects in vivo, gross phenotypes and genotype frequencies of offspring from the Pdgfrb-Cre^+/−×Gnaq^fl/fl;Gna11^−/− crosses were analysed. Fewer mesenchymal G_αq/11 knockout (Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/−) pups reached genotyping age (P14) than was expected (6.6% observed compared with 12.5% expected, Chi squared value=22.03, P<0.005; Fig. 1A). Conversely, mice with at least one functional mesenchymal Gnaq or Gna11 allele reached genotyping age at rates closer to the expected Mendelian frequencies (Fig. 1A). Furthermore, Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− pups were notably smaller than littermates with at least one intact mesenchymal Gnaq or Gna11 allele. Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− animals had a mean weight 1.9-3.2 g lower than all other genotypes (5.4 g versus 7.3-8.4 g, P<0.03; Fig. 1B). Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− pups were also smaller in physical size compared with control animals (Fig. 1C). There was no sex-related difference in weight across genotypes (Fig. 1D). These findings indicate that mesenchymal G_αq/11 deletion causes a detrimental developmental phenotype, leading to death in utero or in early life.

The first two Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice from this breeding programme were humanely killed because of poor physical condition at P21 and P24. Therefore, all further analyses were performed in P14 mice, before evidence of ill health was observed. Gnaq^fl/fl;Gna11^−/− mice develop normally and do not express a phenotype (John et al., 2016); therefore, Pdgfrb-Cre^−/−;Gnaq^fl/fl;Gna11^−/− littermates were used as controls for all analyses to ensure that control mice had an identical genotype to the mesenchymal G_αq/11 knockout mice other than Cre expression, and to facilitate the use of age-matched littermate controls. From here, mice with the Pdgfrb-Cre^−/−;Gnaq^fl/fl;Gna11^−/− genotype will be referred to as Gna11^−/− controls.

To understand the role of mesenchymal G_αq/11 signalling in lung development, the lungs of Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice and Gna11^−/− controls were examined histologically. Pdgfrb-Cre^+/−; Gnaq^fl/fl;Gna11^−/− mouse lungs exhibited clear abnormalities consistent with impaired alveolar development at P14 (Fig. 2A). Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs contained enlarged airspaces with a mean linear intercept distance of 63.47 µm compared with 36.43 µm in Gna11^−/− mice (P=0.03; Fig. 2B), thickened alveolar walls of 12.2 µm compared with 7.0 µm in Gna11^−/− controls (P=0.03; Fig. 2C), and fewer secondary crests (53.7 versus 107.2 per field, P=0.03; Fig. 2D) relative to Gna11^−/− littermate controls. Mice with at least one expressed Gnaq or Gna11 allele (i.e. any genotype other than Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/−) had normal lung histological appearances and similar morphometric measurements that differed from those of Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice (Fig. S1), indicating that a complete absence of Gnaq and Gna11 in mesenchymal cells is required for the abnormal lung phenotype to be observed. These data also demonstrate that Cre expression or presence of the floxed Gnaq alleles alone do not influence the lung phenotype observed in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice.

In addition to these structural abnormalities, Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs expressed lower levels of the proliferative marker Ki67 (Mki67) than Gna11^−/− controls, with 16% of cell nuclei staining positively for Ki67 in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs compared with 26% in Gna11^−/− controls (P=0.03; Fig. 2A,E). Furthermore, Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs contained a lower proportion of cells staining positively for the type II epithelial cell marker pro-surfactant protein C (pro-SPC) than Gna11^−/− control lungs, at 8.9% and 12.8% of all cells, respectively (P=0.03; Fig. 2A,F).

Finally, Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs were heavier relative to total body weight compared with lungs from Gna11^−/− mice (16.5 versus 14.3 mg/g total body weight, P<0.01; Fig. 2G), which could be because of elevated lung density or interstitial oedema in these animals. This trend of increased lung weight in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice was also observed compared with all other genotypes possible from this breeding strategy (Fig. S1). Overall, these structural, proliferative and cellular differentiation abnormalities indicate a disturbance to alveologenesis in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice.

Myofibroblasts are essential for normal alveolar development; therefore, studies were undertaken to assess myofibroblast differentiation and function in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs.

Immunohistochemical staining for the myofibroblast marker α-smooth muscle actin (αSMA; Acta2) demonstrated fewer myofibroblasts in the lungs of P14 Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice compared with Gna11^−/− littermate controls (Fig. 3A). Although overall αSMA staining was decreased in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs, there was no significant reduction in the proportion of αSMA-positive secondary crests compared with Gna11^−/− lungs (0.69 versus 0.84 in controls, P=0.2; Fig. 3B).

To investigate whether G_αq/11 knockout influences myofibroblast differentiation, murine embryonic fibroblasts (MEFs) that were wild-type (WT), G_αq/11 deficient (Gnaq^−/−;Gna11^−/−) or G_α12/13 deficient (Gna12^−/−;Gna13^−/−) were assessed for αSMA protein and Acta2 mRNA expression. MEFs with a stable G_αq/11 knockout had lower Acta2 mRNA (Fig. 3C) and αSMA protein expression than WT MEFs, whereas MEFs lacking G_α12/13, another G_α subunit family, did not have significantly different αSMA expression compared with WT cells (Fig. 3D,E). This implies a key role for G_αq/11 signalling in the differentiation of myofibroblasts from mesenchymal precursor cells.

Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs also showed evidence of defective myofibroblast synthetic function. Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs contained fewer elastin foci (7.4 versus 24.9 foci per field, P=0.03; Fig. 3A,F) and fewer elastin-positive secondary crests (57.5% versus 84.8%, P=0.03; Fig. 3G) than Gna11^−/− mouse lungs. Furthermore, Picrosirius Red staining revealed that P14 Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mouse lungs contained less collagen than the lungs of Gna11^−/− controls (Fig. 3A). These data were supported by lower Eln, Col1a1 and Col3a1 mRNA expression in Gnaq^−/−;Gna11^−/− MEFs than WT MEFs (Fig. 3H-J). These findings imply a failure of myofibroblast differentiation in the lungs of mice lacking mesenchymal G_αq/11 associated with a reduction in myofibroblast function, leading to a reduction in subepithelial matrix deposition.

Pericytes are Pdgfrb-expressing cells that originate in the perivascular region (Henderson et al., 2013), and vasculogenesis is an important driver of normal lung development. Therefore, we examined the pulmonary vasculature histologically to assess for abnormalities caused by mesenchymal G_αq/11 deletion. P14 Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs contained markedly abnormal peripheral pulmonary vessels (Fig. 4A-G), with significantly thicker walls (mean ratio of vessel wall thickness:vessel diameter 0.45 versus 0.33 µm, P=0.03; Fig. 4H) and reduced vessel lumen diameter (mean ratio of internal vessel diameter:whole vessel diameter of 0.54 compared with 0.67, P=0.03; Fig. 4I) than the peripheral pulmonary vessels of Gna11^−/− controls. These vessels consisted of a thin CD31 (Pecam1)-positive endothelial layer (Fig. 4B) surrounded by a thickened αSMA-positive vascular smooth muscle layer (Fig. 4C) without increased proliferation (marked by Ki67; Fig. 4D), indicating that the smooth muscle layer was hypertrophic rather than hyperplastic. These abnormal vessels did not contain significant collagen or elastin layers (Fig. 4E-G). CD31 staining of the alveoli of Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs had a similar appearance to those seen in Gna11^−/− lungs at high magnification (Fig. 4J). This argues against there being gross abnormality of the small alveolar vessels; however, this cannot be completely ruled out.

Given that there were some similarities in appearance of the abnormal peripheral pulmonary vasculature in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs to those seen in pulmonary arterial hypertension, we assessed the hearts from these animals for evidence of right ventricular hypertrophy. We found no difference in right:left ventricular wall ratio in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice relative to controls (Fig. 4K,L). These data suggest a primary Pdgfrb⁺ cell-driven defect, rather than secondary pulmonary hypertension due to impaired alveologenesis.

As Pdgfrb expression is not exclusive to lung mesenchymal cells, the kidneys, hearts, livers and bowel of Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice were assessed for extrapulmonary abnormalities.

We observed an expansion and prominence of medullary mesenchymal cells in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− kidneys demonstrated by αSMA and PDGFRβ staining (Fig. 5A), with associated thinning of the cortex (median cortex:medulla ratio 0.31 in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− kidneys and 0.43 in Gna11^−/− controls, P<0.03; Fig. 5B,C). The relative kidney to total body weight values of Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mouse kidneys were not different to those of Gna11^−/− controls (median kidney:total body weight ratio 7.3 in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice and 6.5 in Gna11^−/− controls, P=0.55; Fig. 5D). These data suggest that mesenchymal G_αq/11 is important in normal kidney development.

Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice had normal heart, liver and bowel histology (Fig. S2), suggesting that mesenchymal G_αq/11 signalling is not required for normal heart, liver or bowel development or homeostasis from conception to P14 in mice.

To assess whether the abnormalities seen in Pdgfrb-Cre^+/−;Gnaq^fl/fl; Gna11^−/− mice were related solely to disturbed organ developmental processes or could also affect mature lungs, a tamoxifen-inducible mesenchymal G_αq/11 knockout model (Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/−) was established in adult mice (Fig. 6A).

Tamoxifen-naïve Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice were born at the expected frequency. According to the supplier, it is expected that 20% of offspring from breeding of the Cre-expressing hemizygous mice with WT mice will express the Pdgfrb-Cre/ERT2 transgene rather than the 50% Cre-expression rate observed in the germline Pdgfrb-Cre^+/− mouse colony. The frequency of Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice reaching genotyping age was 6.4%, compared with the expected 5% (total number of mice born 109; Fig. 6B). This indicates that having the Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− genotype, without administration of tamoxifen, does not cause any gross developmental defects.

When a 3-week course of tamoxifen was administered to P49 Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice [n=4 (1 female, 3 male)], no detrimental effect to health status was observed compared with littermate controls. Furthermore, Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice gained weight at the same rate as littermate controls with the other genotypes during the tamoxifen protocol (median weight on day 21 of tamoxifen 104.3% of baseline in Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice compared with 106.2% of baseline in other genotypes, P=0.71; Fig. 6C). A small reduction in weight was observed early in the tamoxifen protocol that was independent of genotype and was in keeping with a change in diet (Kiermayer et al., 2007). These data suggest that short-term mesenchymal G_αq/11 knockout does not cause gross physiological disturbances in vivo.

Histological analysis revealed that the lungs of Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice treated with tamoxifen had increased airspace size compared with Gna11^−/− controls (mean linear intercept distance 52.5 µm in Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice compared with 39.3 µm in Gna11^−/− controls, P=0.03; Fig. 6D,E), suggestive of emphysema. Pdgfrb-Cre/ERT2^+/−; Gnaq^fl/fl;Gna11^−/− lungs contained fewer elastin foci than Gna11^−/− controls after 3 weeks of tamoxifen (median number of elastin foci per high-powered field 13.0 in Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice compared with 26.9 in Gna11^−/− controls, P=0.03; Fig. 6D,F), similar to the constitutive knockout. In contrast, Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− lungs did not exhibit altered collagen deposition or evidence of fewer myofibroblasts (αSMA) compared with Gna11^−/− controls (Fig. 6D). Three of the four Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice also exhibited abnormal pulmonary mononuclear cellular aggregates, which were mainly localised at the pleural surfaces (Fig. 6G) and were not observed in littermate control mice. Despite these abnormalities, Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice did not exhibit signs of respiratory distress.

In contrast with Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice, Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice administered tamoxifen did not exhibit any renal abnormalities on histology (Fig. S3). This implies that mesenchymal G_αq/11 is needed for normal kidney development, but not maintenance of the normal kidney.

Given the crucial roles of TGFβ in alveolar development, lung repair, and mesenchymal cell migration and differentiation, we investigated the role of mesenchymal G_αq/11 in a cyclical stretch model of TGFβ activation. Mesenchymal cells with and without intact G_αq/11­ signalling were subjected to breathing-related CMS and TGFβ signalling was assessed. CMS-induced TGFβ signalling, as assessed by Smad2 phosphorylation, was significantly reduced in Gnaq^−/−;Gna11^−/− MEFs compared with WT MEFs (Fig. 7A,B). This finding was specific to the G_αq/11 family of G proteins, as there was no effect of G_α12/13 knockdown on stretch-induced TGFβ signalling in MEFs (Fig. 7A).

To validate the role of G_αq/11 in stretch-induced TGFβ signalling in mesenchymal cells across species, human lung fibroblasts (HLFs) with and without siRNA-induced GNAQ and GNA11 knockdown were subjected to breathing-related CMS. GNAQ and GNA11 siRNA led to substantial reductions in both G_αq and G_α11 protein expression in HLFs, and significantly reduced CMS-induced TGFβ signalling compared with scrambled control (Scr) siRNA as measured by phosphorylation of Smad2 (Fig. 7C,D). These data indicate that G_αq/11 is a key component of CMS-induced TGFβ signalling in both murine and human fibroblasts.

Previous studies have reported that G_αq/11 induces TGFβ activation via the Rho-ROCK cascade and αv integrins in epithelial cells (Froese et al., 2016; Xu et al., 2009). As αvβ1, αvβ3 and αvβ5 integrins are expressed by myofibroblasts and are involved in TGFβ activation (Pakshir et al., 2020), we utilised chemical inhibition of these integrins and ROCK in our CMS model. When human fibroblasts were subjected to breathing-related CMS in the presence of a ROCK1/2 inhibitor (Y27632), a pan αv integrin inhibitor (CWHM-12) or a β1 integrin-specific inhibitor (NOTT199SS), CMS-induced TGFβ signalling was not reduced (Fig. S4). These data imply the existence of a pathway for CMS-induced TGFβ signalling in mesenchymal cells that requires G_αq/11, but is independent of ROCK and integrin signalling.

Proteases can activate latent TGFβ independently of integrins; therefore, we assessed the effect of protease inhibitors in our CMS- induced TGFβ signalling system. A pan serine protease inhibitor 4-(2-aminoethyl) benzenesulfonyl fluoride (AEBSF), decreased CMS-induced Smad2 phosphorylation in HLFs (Fig. 8A,B), whereas the matrix metalloproteinase (MMP) inhibitor GM-6001 had no effect on CMS-induced TGFβ signalling even at high concentrations (Fig. 8C,D). These findings indicate that serine proteases mediate CMS-induced TGFβ signalling in mesenchymal cells.

As TGFβ2 is the only TGFβ isoform that is not activated by integrins (Jenkins, 2008), we hypothesised that breathing-related CMS would predominantly activate the TGFβ2 isoform in mesenchymal cells. Although CMS did not influence TGFβ2 protein expression in HLFs, HLFs with siRNA-induced GNAQ and GNA11 knockdown expressed less TGFβ2 than HLFs with intact G_αq/11 signalling (Fig. 8E,F), suggesting that G_αq/11 plays a role in TGFβ2 production. Conversely, TGFβ1 protein expression was not affected by GNAQ and GNA11 knockdown in HLFs (Fig. 8G), suggesting an isoform-specific effect.

To evaluate the role of this CMS-induced TGFβ2 signalling pathway in alveologenesis, we assessed TGFβ2 expression in the lungs of mice from our mouse models. Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs had a significantly lower TGFβ2 content than Gna11^−/− control lungs (median immunostaining score 0.8 in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs, compared with 2.7 in Gna11^−/− controls, P<0.03; Fig. 8H,I). Similarly, Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mouse lungs also had reduced TGFβ2 deposition compared with Gna11^−/− controls after 3 weeks of tamoxifen (median immunostaining score 0.8 in Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− lungs compared with 2.2 in Gna11^−/− controls, P<0.03; Fig. 8H,J). These data demonstrate that lungs lacking mesenchymal G_αq/11 have less TGFβ2 available for breathing-related CMS-induced activation, and this may be important in alveologenesis and the maintenance of normal lung structure in vivo.

Platelet-derived growth factor (PDGF) signalling is known to be important in alveolar development, and this pathway interacts with TGFβ signalling in normal development and disease (Gouveia et al., 2017, 2018). We therefore investigated how G_αq/11 signalling influences the expression of PDGF signalling components in fibroblasts.

Gnaq^−/−;Gna11^−/− MEFs expressed significantly lower levels of Pdgfb and Pdgfd mRNA compared with WT cells (P=0.03; Fig. 9B,D). There was not a statistically significant difference in the expression of Pdgfa, Pdgfc, Pdgfra or Pdgfrb mRNA expression between Gnaq^−/−;Gna11^−/− and WT MEFs (Fig. 9A,C,E,F), although there was a trend to reduced Pdgfa expression in Gnaq^−/−;Gna11^−/− MEFs (P=0.06; Fig. 9A). These data imply that mesenchymal G_αq/11 deletion influences the expression of PDGF signalling components, and thus may regulate PDGF signalling.

In this study, we used mice with a targeted deletion of G_αq/11 in Pdgfrb-expressing cells to demonstrate that mesenchymal G_αq/11 is essential for the development and maintenance of normal alveoli. Loss of G_αq/11-mediated signalling in mesenchymal cells caused failure of the myofibroblast differentiation and ECM synthetic function required for alveolar development and the maintenance of the adult lung, and reduced mesenchymal cell TGFβ2 production is a key factor in these processes. In the absence of mesenchymal G_αq/11, TGFβ2 is unavailable for activation by CMS-induced serine proteases, thereby diminishing downstream TGFβ signalling in both developing and adult lungs. These findings establish a previously undescribed role for breathing-related CMS in TGFβ2 generation and suggest a role for TGFβ2 in alveolar development and lung homeostasis.

The role of G_αq/11 in alveolar development has not previously been investigated, primarily because germline G_αq/11 deletion is embryonically lethal (Offermanns et al., 1998) and murine alveolarisation occurs entirely postnatally (Beauchemin et al., 2016). Cell type-specific Gnaq and Gna11 deletion in neural, cardiovascular and haematological tissues have various manifestations ranging from no phenotype to profound cardiac abnormalities associated with perinatal death (Hoyer et al., 2010; Sassmann et al., 2010; Wettschureck et al., 2007, 2004, 2005, 2001, 2006). However, alveolar abnormalities have not been described in germline or conditional G_αq/11 knockout mice, suggesting a unique role for mesenchymal G_αq/11 in alveolar development and maintenance.

We propose that the key mechanisms underlying the abnormal alveologenesis and emphysema in mice with mesenchymal G_αq/11 deletion present from conception or induced in adulthood, respectively, are failure of myofibroblast differentiation and synthetic function. Both Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− and Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice had lower lung elastin deposition than controls. Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− lungs also contained fewer myofibroblasts and less collagen compared with controls, and mesenchymal cells lacking G_αq/11­ express less Col1a1, Col3a1 and Eln mRNA than cells with intact G_αq/11. As myofibroblasts induce secondary septation by depositing ECM proteins, particularly elastin, at the tips of developing secondary septa (Dabovic et al., 2015), and loss of elastin is a key feature of emphysema (Ito et al., 2019; Suki et al., 2012), these data suggest that mesenchymal G_αq/11-induced myofibroblast differentiation and function are required for alveolar development and homeostasis. In addition, as elastin is a key factor governing lung compliance (Dolhnikoff et al., 1999; Hilgendorff et al., 2012), and mechanical forces themselves may alter availability of elastin-binding sites (Jesudason et al., 2010; Suki et al., 2012), elastin may influence the response to stretch-related forces. Therefore, the impact of mesenchymal G_αq/11 deletion on elastin deposition and distribution may influence the response to and generation of mechanical forces within the lungs.

Secondary crest myofibroblasts (SCMFs) are known to derive from PDGFRα-expressing precursors (Boström et al., 1996; Li et al., 2018; Lindahl et al., 1997; McGowan et al., 2008); however, the role of PDGRFβ⁺ precursors in the development of SCMFs has not been described. Although this study cannot definitively conclude that PDGFRβ⁺ precursors differentiate into SCMFs, it does show a role for PDGFRβ⁺ cells in alveolarisation. Whether this occurs via direct differentiation of SCMFs from PDGFRβ⁺ precursors, or via paracrine signalling from PDGFRβ⁺ cells, is currently unknown.

Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mouse lungs contained abnormal peripheral pulmonary vessels, with thickened vessel walls and reduced lumen diameter associated with muscularisation of the media (indicated by αSMA staining). These findings could be explained by pulmonary arterial hypertension (PAH), which could relate to hypoxaemia secondary to the profound pulmonary defects, in combination with disturbed GPCR signalling, resulting in vascular remodelling (Cheng et al., 2012; Patel et al., 2018). However, cardiac histology did not show thickening of the right ventricular wall in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice, nor was any intimal or adventitial fibrosis observed, findings inconsistent with substantial PAH. A limitation of this study is that assessments such as the Fulton index to assess for right ventricular hypertrophy were not possible and so it is not possible to determine conclusively whether there was any PAH. An alternative explanation for the abnormal vasculature in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice could be that altered activity of Pdgfrb-expressing cells influences vascular development or the growth, differentiation and activity of constituent cells, such as vascular smooth muscle cells. Furthermore, the observed increased lung weights in Pdgfrb-Cre^+/−; Gnaq^fl/fl;Gna11^−/− mice could be linked to interstitial oedema and cardiac malfunction. However, it is not currently possible to define the precise cause of the vascular abnormalities observed.

Altered CMS-induced TGFβ activation is likely to be a key driver of the lung phenotypes observed in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− and Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice. TGFβ drives myofibroblast differentiation, cellular migration and ECM protein production (Harrell et al., 2018), and deficiencies and genetic polymorphisms in TGFβ signalling pathway components have been associated with emphysema (Bonniaud et al., 2004; Celedón et al., 2004; Hersh et al., 2009; Li et al., 2011). Both lung stretch and tightly controlled TGFβ signalling are important for normal lung development and regeneration (Alejandre-Alcázar et al., 2008; Belcastro et al., 2015; Bonniaud et al., 2004; Chen et al., 2005, 2008; Deng et al., 2019; Donahoe et al., 2016; Gauldie et al., 2003; Nakanishi et al., 2007; Pieretti et al., 2014; Sterner-Kock et al., 2002; Vicencio et al., 2004), and CMS has been demonstrated to induce TGFβ signalling in a number of models and organ systems (Froese et al., 2016; Fujita et al., 2010; Furumatsu et al., 2013; John et al., 2016; Maeda et al., 2011; Russo et al., 2018; Wang et al., 2013). Using the same Gnaq^fl/fl;Gna11^−/− mice used in our study, John et al. described age-related emphysema related to reduced stretch-induced TGFβ signalling in mice lacking G_αq/11 in type II alveolar epithelial cells (John et al., 2016). Open access RNA-sequencing data on the LungMAP and IPF Cell Atlas databases show that in human and mouse lung, PDGFRβ-positive cells include pericytes, fibroblasts and myofibroblasts (www.ipfcellatlas.com: Kaminski/Rosas dataset; www.lungmap.net: mouse: LungMAP ID LMEX0000001602; human: LungMAP ID LMEX0000004388). We therefore used human lung fibroblasts and murine embryonic fibroblasts to assess the role of mesenchymal G_αq/11 in CMS-induced TGFβ signalling.

CMS-induced TGFβ signalling in HLFs and MEFs was dependent on serine proteases and independent of αv integrins, contrary to previous work in lung slices and epithelial cells (Froese et al., 2016; Xu et al., 2009). This indicated that TGFβ2, an isoform that is activated by proteases but not integrins (Jenkins, 2008), may be the primary TGFβ isoform activated by mesenchymal cell stretch. G_αq/11-deficient human lung fibroblasts expressed less TGFβ2, but had unchanged levels of TGFβ1, compared with cells that express G_αq/11, suggesting a TGFβ isoform-specific effect of G_αq/11 deletion. These data suggest a pathway in which mesenchymal G_αq/11 drives TGFβ2 production, which is then available for protease-mediated activation. Although the use of MEFs and HLFs does not precisely recapitulate the fibroblast cell populations present during lung development, these data demonstrate conservation of stretch-induced TGFβ signalling in fibroblasts across species.

This is the first study to propose an isoform-specific role for TGFβ2 in mammalian alveolar development and lung homeostasis. The three TGFβ isoforms are highly expressed during lung development with distinct spatial and temporal expression patterns (Schmid et al., 1991); however, little is known about the specific regulation of TGFβ2 signalling. Tgfb2^−/− mice die shortly after birth from developmental abnormalities distinct from those seen in Tgfb1^−/− or Tgfb3^−/− mice (Kaartinen et al., 1995; Sanford et al., 1997; Shull et al., 1992). Tgfb2^−/− mice have no gross lung morphological abnormalities in late intrauterine gestation; however, collapsed conducting airways are found postnatally (Sanford et al., 1997). Although the Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice generated in the present study did not share phenotypic features withTgfb2^−/− mice, it is possible that TGFβ2 production by non-mesenchymal cells is sufficient for normal prenatal development. Additionally, as alveolarisation occurs entirely postnatally in mice, the role of TGFβ2 in alveolar development that we describe could not be observed in Tgfb2^−/− mice owing to perinatal death. Our data demonstrate that loss of mesenchymal G_αq/11 causes a loss of the precise control of TGFβ signalling in the lungs, resulting in abnormal alveologenesis and loss of lung homeostasis in developed lungs. Further work is required to understand the precise roles of individual TGFβ isoforms in these processes.

The PDGF family is known be important in lung development and regeneration, with PDGF-A being particularly important in alveolar development (Gokey et al., 2021; Gouveia et al., 2018, 2020). We found a trend towards reduced Pdgfa expression in MEFs with G_αq/11 deletion, as well as Pdgfb and Pdgfc, suggesting that G_αq/11 signalling may interact with PDGF-related pathways. Postnatal deletion of Pdgfra, which encodes the major receptor for PDGF-A, reduces lung Tgfb2, but not Tgfb1, transcripts (Li et al., 2019), further supporting a role for PDGF signalling in G_αq/11- and TGFβ2-driven alveolar development and regeneration. However, elastin deposition during alveologenesis may not be dependent on PDGF-A (Gouveia et al., 2020); therefore, PDGF-independent pathways are also likely to be involved in driving the abnormalities in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− and Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mouse lungs. As pulmonary mesenchymal cells are predominantly PDGF receptor expressing, rather than PDGF ligand producing (Gouveia et al., 2017), and G_αq/11 deletion did not alter Pdgfra or Pdgfrb expression, we hypothesise that mesenchymal G_αq/11 deletion reduces lung TGFβ2 signalling, which subsequently alters PDGF ligand expression by other cell types. However, it was beyond the scope of this work to dissect the interactions between G_αq/11, TGFβ2, and PDGF signalling.

As PDGFRβ is a mesenchymal cell marker found outside of the lung, the other organs of Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− and Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice were examined histologically. Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− kidneys demonstrated expansion and prominence of medullary mesenchymal cells. However, the kidneys of Pdgfrb-Cre/ERT2^+/−; Gnaq^fl/fl;Gna11^−/− mice were normal, supporting the hypothesis that abnormalities observed in Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− kidneys were developmental in nature.

There are a number of limitations of this study. The poor condition of Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice limited the analyses to a single time point and precluded the study of CMS in vivo. The absence of detailed lineage tracing or single-cell RNA sequencing of the Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mouse lungs means it cannot be confirmed that the abnormalities were driven by any particular mesenchymal cell subtype. Pdgfrb-Cre^+/− mice have been used to investigate the role of pericytes in various organ development and disease models (Diéguez-Hurtado et al., 2019; Eilken et al., 2017; Foo et al., 2006; Gong et al., 2018; He et al., 2020; Henderson et al., 2013; Ivanova et al., 2021; Wang et al., 2017; Zaitoun et al., 2019; Zhang et al., 2021), including a study that used a GFP reporter mouse to demonstrate Pdgfrb-Cre-induced gene recombination in lung pericytes during time points relevant to alveolarisation (Kato et al., 2018). However, previous studies have shown that Pdgfrb-Cre-induced gene recombination can occur in other cell types, including myofibroblasts induced by injury, fibroblasts, smooth muscle cells and renal interstitial cells (Cuttler et al., 2011; Gong et al., 2018; Henderson et al., 2013; Schiessl et al., 2018; Ulvmar et al., 2016; Wang et al., 2019; Zou et al., 2018). Furthermore, Pdgfrb expression may vary at different stages of organ development and cellular differentiation (Salter et al., 2019), and it is possible that altered gene expression in a Pdgfrb-expressing precursor cell could influence the characteristics of non-Pdgfrb-expressing cells that derive from them. Therefore, although we hypothesise that abnormalities in pericyte activity, differentiation and migration underlie the defective alveologenesis and emphysema in mesenchymal G_αq/11 knockout mice, the role of other mesenchymal cells in this process cannot be ruled out. However, we can conclude that G_αq/11 signalling in Pdgfrb-expressing cells is important in lung development and homeostasis.

The growth restriction of Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice may indicate a nutritional deficiency that could have contributed to delayed alveolar development. Although these animals did have renal abnormalities which may have contributed to the poor condition and failure to thrive of Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice, the bowel appeared normal and mice with mesenchymal G_αq/11 deletion induced in adulthood had normal kidneys. This suggests a true pulmonary phenotype in mesenchymal G_αq/11 knockout mice. Additionally, our in vitro data provide compelling evidence for a role for mesenchymal G_αq/11 in a key lung developmental signalling pathway, suggesting that mesenchymal G_αq/11 deletion generates a true lung developmental phenotype.

Finally, this study has not investigated the role of lung inflammation in mesenchymal G_αq/11 knockout mice. TGFβ regulates inflammation, and John et al. showed that emphysema in mice with a type II epithelial G_αq/11 deletion was associated with lung inflammation and M2 macrophage polarisation (John et al., 2016). The mononuclear cellular aggregates in the lungs of mice with mesenchymal G_αq/11 deletion induced in adulthood could indicate abnormal inflammation in these mice. However, these cellular aggregates were not observed in mice with a germline mesenchymal G_αq/11 knockout, and it was not possible to fully define the role of inflammation and the immune response in the emphysema observed in Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− mice in our study.

In conclusion, this is the first study to generate mesenchymal G_αq/11 deleted mice, and has demonstrated a previously unappreciated signalling pathway for CMS-induced TGFβ2 signalling in murine embryonic and mature human mesenchymal cells that is important for alveologenesis and maintenance of the normal lung. These findings could have implications for the treatment of several conditions associated with dysregulated developmental and repair pathways, including fibrosis and emphysema.

All reagents and resources are listed in Table S1.

Mice were housed under specific pathogen-free conditions, with standard food and water available ad libitum. All animal experiments were performed in accordance with the Animals (Scientific Procedures) Act 1986, and approved by the Animal Welfare and Ethical Review Board at the University of Nottingham.

For the germline mouse studies, mice with floxed alleles for Gnaq and germline deficiency in Gna11 (Gnaq^fl/fl;Gna11^−/−) were crossed with mice that express Cre recombinase under the control of the Pdgfrb gene (Pdgfrb-Cre^+/−). Pdgfrb-Cre^+/−;Gnaq^+/fl;Gna11^+/− offspring from this F1 generation were then bred with Gnaq^fl/fl;Gna11^−/− founders to produce an F2 generation, including Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− mice. The genetic background for all mice was predominantly C57BL6, with a minimum of a six backcross generations. The generation of Gnaq^fl/fl;Gna11^−/− and Pdgfrb-Cre^+/− mice has been described previously (Foo et al., 2006; Offermanns et al., 1998; Wettschureck et al., 2001).

For the tamoxifen-inducible mouse gene knockout studies, the same breeding strategy was used as for the germline studies but substituting Pdgfrb-Cre/ERT2^+/− mice [029684, B6.Cg-Tg(Pdgfrb-cre/ERT2)6096Rha/J, The Jackson Laboratory] for Pdgfrb-Cre^+/− animals. The generation of these mice has been described elsewhere (Gerl et al., 2015).

Mice were genotyped using DNA isolated from ear-notch biopsies by PCR analysis with allele-specific primers (Eurofins Scientific). Primer sequences were: Cre transgene 5′-GCGGTCTGGCAGTAAAAACTATC-3′, 5′-GTGAAACAGCATTGCTGTCACTT-3′ (product 100 bp); internal positive control 5′-CTAGGCCACAGAATTGAAAGATCT-3′, 5′-GTAGGTGGAAATTCTAGCATCATCC-3′ (product 324 bp); Gna11 WT 5′-AGCATGCTGTAAGACCGTAG-3′, 5′-GCCCCTTGTACAGATGGCAG-3′ (product 820 bp); Gna11 knockout 5′-CAGGGGTAGGTGATGATTGTG-3′, 5′-GACTAGTGAGACGTGCTACTTCC-3′ (product 450 bp); Gnaq WT and floxed alleles 5′-GCATGCGTGTCCTTTATGTGAG 3′, 5′-AGCTTAGTCTGGTGACAGAAG-3′ [products: 600 bp (WT), 700 bp (floxed)]. For Cre-ERT2, the following primers were used: 5′-GAACTGTCACCGGGAGGA-3′, 5′-AGGCAAATTTTGGTGTACGG-3′ (400 bp product).

PCR products were analysed by electrophoresis on ethidium bromide-stained agarose gels.

Mice were genotyped at 2 weeks old (P14). Genotype ratios of F2 mice from the Gnaq^fl/fl;Gna11^−/− and Pdgfrb-Cre^+/− crosses were compared with the expected Mendelian frequency (12.5% per genotype). Similarly, genotype ratios of F2 mice from the Gnaq^fl/fl;Gna11^−/− and Pdgfrb-Cre/ERT2^+/− crosses were assessed, with an expected frequency of 5% for each Cre-expressing genotype.

For in vitro experiments using human lung fibroblasts, cells from four to six donors were used per group. Informed consent was obtained from all donors and work was approved by the East Midlands Nottingham 1 Research Ethics Committee (reference 08/H0407/1). Cells were used at passage 5-6 for all in vitro experiments.

HLFs were isolated from donated post-mortem or surgical lung biopsy samples, from male and female donors with and without pulmonary fibrosis. For non-fibrotic fibroblasts, cells were isolated from regions of lung distant from the area of primary diagnosis. Tissue was cut into 1 mm×1 mm pieces and placed 10 mm apart in a 10 cm cell culture dish. Tissue was cultured in DMEM supplemented with 10% foetal calf serum (FCS, Fisher Scientific, 11573397), L-glutamine (4 mM, G7513, Sigma-Aldrich), penicillin (200 units/ml) and streptomycin (0.2 mg/ml) (P4458, Sigma-Aldrich) and amphotericin B (2.5 µg/ml, A2942, Sigma-Aldrich). Fibroblast outgrowth could be seen after 6-8 days. Tissue was removed from the cell culture dish if it became detached, or when cells had reached 80% confluency and were ready for passage. Cells were maintained in a humidified incubator at 37°C, 5% CO₂/95% air, in Dulbecco's Modified Eagle's Medium (DMEM, Sigma-Aldrich), supplemented with 10% FCS, L-glutamine (4 mM), penicillin (100 units/ml) and streptomycin (0.1 mg/ml).

WT, Gna12^−/−;Gna13^−/− and Gnaq^−/−;Gna11^−/− MEFs were a gift from Professor Stefan Offermanns (Max Planck Institute for Heart and Lung Research, Bad Nauheim, Germany), and their generation has been described elsewhere (Gu et al., 2002; Zywietz et al., 2001). Gnaq, Gna11, Gna12 and Gna13 gene expression was also confirmed in-house prior to these studies (Fig. S5). Cells were maintained in a humidified incubator at 37°C, 5% CO₂/95% air, in DMEM supplemented with 10% FCS, L-glutamine (4 mM), penicillin (100 units/ml) and streptomycin (0.1 mg/ml).

Litters were observed for signs of ill health daily from birth. Mice were weighed at P14. Male and female mice were included in all analyses. Mice had not undergone any previous procedures. All mice that survived to P14 were phenotyped and had organs collected. Mouse phenotyping analyses were performed by an observer unaware of the genotype. Genotype information was not available to the phenotyping observer until all phenotyping and health status data had been recorded.

For comparisons, Pdgfrb-Cre^+/−;Gnaq^fl/fl;Gna11^−/− (mesenchymal G_αq/11 knockout mice) were compared with littermate Pdgfrb-Cre^−/−;Gnaq^fl/fl;Gna11^−/− control mice (referred to as Gna11^−/− controls).

Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− offspring and their littermates were kept under standard conditions until 7 weeks of age (P49), when tamoxifen-containing chow (400 mg/kg tamoxifen citrate, Envigo) was introduced ad libitum. Health scoring and weights were measured daily for 3 weeks as tamoxifen was administered. Animals were humanely killed after 3 weeks of tamoxifen administration (at 10 weeks old, P70).

For comparisons, Pdgfrb-Cre/ERT2^+/−;Gnaq^fl/fl;Gna11^−/− (tamoxifen-inducible mesenchymal G_αq/11 knockout mice) were compared with littermate Pdgfrb-Cre/ERT2^−/−;Gnaq^fl/fl;Gna11^−/−control mice (referred to as Gna11^−/− controls).

Mice were humanely killed by intraperitoneal injection of pentobarbital, and organs collected for histological analyses. The lungs were perfused by injecting 40 units/ml heparin sodium in PBS (Wockhardt) into the right ventricle, and inflated by cannulating the trachea and filling the lungs with 10% formalin (VWR) under gravity. The trachea was ligated, and the heart and lungs removed en bloc. Livers and kidneys were also collected. Organs were kept in 10% formalin for 24 h before paraffin embedding and sectioning.

Formalin-fixed, paraffin-embedded tissue sections of lung (3 µm), kidney (3 µm), heart (5 µm) and liver (5 µm) were deparaffinised in xylene and rehydrated in graded alcohols. Haematoxylin and Eosin (H&E), Verhoeff–Van Gieson (elastin) and Picrosirius Red staining were performed as per standard protocols using buffers and stains prepared in-house and mounted in DPX.

The following histology solutions were generated in house: Weigert's iodine [2 g potassium iodide (03124, Sigma-Aldrich), 1 g iodine (326143, Sigma-Aldrich), 100 ml distilled water]; Verheoff's solution [20 ml 5% alcoholic Haematoxylin (H3136, Sigma-Aldrich), 8 ml 10% ferric chloride (157740, Sigma-Aldrich), 8 ml Weigert's iodine]; Van Gieson's solution [5 ml aqueous acid fuschin (F8129, Sigma-Aldrich), 100 ml saturated aqueous picric acid (84512.260, VWR)]; Picrosirius Red solution [0.5 g Direct Red 80 (365548, Sigma-Aldrich), 500 ml saturated aqueous picric acid (84512.260, VWR)]; Weigert's Haematoxylin (1:1 ratio of Weigert's solution A and Weigert's solution B); Weigert's solution A [1% Haematoxylin (H3136, Sigma-Aldrich) in 100% ethanol]; Weigert's solution B [4 ml 30% ferric chloride (157740, Sigma-Aldrich), 1 ml 12 N hydrochloric acid, 95 ml water]; acidified water (5 ml glacial acetic acid, 1 l distilled water); acid/alcohol solution (70% ethanol, 0.1% hydrochloric acid).

After being deparaffinised and rehydrated, tissue sections were submerged in Mayer's Haematoxylin (51275, Sigma-Aldrich) for 2 min, acid/alcohol solution for 1 min, then 1% Eosin solution (101411-524, VWR) for 3 min. Sections were rinsed with tap water between each step, then dehydrated and mounted.

Lung sections were deparaffinised and hydrated to distilled water, then stained in Verhoeff's solution for 1 h until the tissue was completely black. Sections were differentiated in 2% ferric chloride (157740, Sigma-Aldrich) until elastin foci were seen on a grey background, incubated in 5% sodium thiosulphate (72049, Scientific Laboratory Supplies) for 1 min, and then washed in running tap water for 5 min. Sections were then counterstained in Van Gieson's solution for 5 min, dehydrated and mounted as above.

Lung, kidney and heart sections were deparaffinised and hydrated. Nuclei were stained with Weigert's Haematoxylin for 8 min, and then washed in running tap water for 5 min. Sections were incubated in Picrosirius Red for 1 h, washed in two changed of acidified water, then dehydrated and mounted.

Tissue sections were deparaffinised in xylene and rehydrated in graded alcohols. Heat-mediated antigen retrieval was performed by boiling sections in a microwave for 20 min in 10 mM citric acid buffer (pH 6.0). Endogenous peroxidase activity was blocked by incubating sections in 3% hydrogen peroxide in methanol for 30 min. Non-specific binding was blocked with 5% goat serum (G9023, Sigma-Aldrich) in 0.1% bovine serum albumin in PBS. Sections were incubated with primary antibody in 5% goat serum overnight at 4°C in a humidified chamber, followed by incubations for 60 min with secondary antibody and 30 min with avidin-biotin complex (SP2001, Vector Laboratories). Sections were then stained with diaminobenzidine (SIGMAFAST, D4418, Sigma-Aldrich), counterstained with Mayer's Haematoxylin (51275, Sigma-Aldrich), and mounted in DPX (06522, Sigma-Aldrich). Slides were washed in PBS between incubation steps.

The following antibodies were used for immunohistochemistry: rabbit anti-αSMA (Abcam, ab5694; 1:500), rabbit anti-CD31 (Abcam, ab182981; 1:2000), rabbit anti-Ki67 (Abcam, ab15580; 1 µg/ml), rabbit anti-pro-surfactant protein C (Sigma-Aldrich, Ab3786; 1:2000), rabbit anti-TGFβ2 (Proteintech, 19999-1-AP; 1:3000), rabbit anti-elastin (Atlas Antibodies, HPA056941; 1:100) and biotinylated goat anti-rabbit IgG (Vector Laboratories, BA1000; 1:200).

Images of H&E, elastin and immunohistochemical staining were taken using a Nikon 90i microscope and NIS-Elements software v3.2 (Nikon). Polarised light imaging of Picrosirius Red-stained samples was performed using a Zeiss Axioplan microscope (Zeiss) and MicroManager 1.4 software (Vale Lab, UCSF).

For all analyses of histology images, Pdgfrb-Cre^+/− ;Gnaq^fl/fl;Gna11^−/− or Pdgfrb-Cre/ERT2^+/− ;Gnaq^fl/fl;Gna11^−/− mice were compared with Pdgfrb-Cre^−/−;Gnaq^fl/fl;Gna11^−/− or Pdgfrb-Cre/ERT2^−/− ;Gnaq^fl/fl;Gna11^−/− littermate controls (referred to as Gna11^−/− controls), respectively. For histological analyses, four animals per genotype were assessed to allow differences in histological appearances to be detected. All image quantification was performed by an observer unaware of the genotype. This observer was not informed of the genotypes until all image quantification data had been recorded.

For quantitative analyses of the lungs, five to ten images were assessed per set of lungs, covering all lobes and avoiding major airways and central blood vessels. All morphometric analyses were performed using NIS Elements software v3.2 (Nikon), with the exception of peripheral pulmonary vessel thickness measurements and kidney measurements, which were performed using CaseViewer 2.3 software (3D Histech).

For quantification of immunohistochemistry and elastin staining, the ‘count’ feature of ImageJ (National Institutes of Health) was used. Elastin foci were identified as black fibres on Verhoeff–Van Gieson staining, and secondary crests were considered elastin positive if they had black staining that was not clearly a cell nucleus on Verhoeff–Van Gieson staining. For immunohistochemistry staining, a cell was counted if it stained brown. Only nuclear DAB staining was counted for Ki67 quantification. For αSMA quantification, the number of αSMA-positive secondary crests per 40× field was counted. For Ki67 and pro-SPC staining, the total number of cells per 40× field was quantified by counting nuclei, and the proportion of Ki67- or pro-SPC-positive cells calculated by dividing the number of stained cells per image by the total number of cells per image.

For quantification of TGFβ2 staining, the following scoring system was used and seven fields (20× magnification) per mouse were analysed: Score 0, no cells stained; Score 0.5, 1-25 cells stained at low intensity; Score 1.0, 1-25 cells stained at high intensity; Score 1.5, 26-50 cells stained at low intensity; Score 2.0, 26-50 cells stained at high intensity; Score 2.5, >50 cells stained at low intensity; Score 3.0: >50 cells stained at high intensity.

Mean linear intercept analysis of airspace size was performed as previously described (John et al., 2016). Briefly, 10× magnification images were overlaid with a grid comprising 100 µm squares, and ‘intercepts’ between gridlines and airspace walls counted. The mean linear intercept was calculated by dividing the length of each gridline by the intercept count. For alveolar wall thickness measurements, 40× magnification images were overlaid with five equally spaced horizontal lines and the alveolar wall thickness measured at points where lung tissue crossed each line using the ‘measure’ function of NIS Elements. Mean linear intercept and alveolar wall thickness values were calculated for each mouse from all measurements across all images and data presented as median±interquartile range. For secondary crest counts, 10× magnification images were used and secondary crests counted for each image.

Cells were seeded at 2×10⁵ cells per well on collagen I-coated Bioflex^® 6-well culture plates (3001-C, Dunn Labortechnik) in DMEM supplemented with 10% FCS, L-glutamine (4 mM), penicillin (100 units/ml) and streptomycin (0.1 mg/ml) and allowed to adhere for 24 h. The culture medium was changed to 1% FCS in DMEM with 4 mM L-glutamine for 24 h before stretching commenced. The Flexcell^® FX-5000T system (Flexcell International Corporation) was used to apply cyclical stretch to cells in vitro, according to the manufacturer's instructions. MEFs were stretched at a frequency of 1 Hz, and HLFs at 0.3 Hz to mimic breathing in the relevant organism. 15% elongation and a sine waveform were used for all cyclical stretch experiments. Cyclical stretch was applied for 48 h, except for experiments using siRNA-induced GNAQ and GNA11 knockdown, for which 24 h of cyclical stretch was used. Unstretched control cells were cultured in identical conditions alongside the Flexcell^® apparatus. Cells were lysed in protein lysis buffer (9803, Cell Signaling Technology) supplemented with phosphatase (Phos-Stop, 04906837001, Sigma-Aldrich) and protease (cOmplete Mini, 04693124001, Sigma-Aldrich) inhibitors, and 20 µM PMSF in isopropanol (P7626, Sigma-Aldrich). All experimental replicates were performed independently.

When used, inhibitor compounds were applied in DMEM supplemented with 1% FCS and 4 mM L-glutamine 30 min before stretching commenced. The activin receptor-like kinase (ALK5)/type I TGFβ-receptor kinase inhibitor SB-525334 (S8822, Sigma-Aldrich) was used at a concentration of 50 µM. The ROCK inhibitor Y27632 (Y0503, Sigma-Aldrich), the pan-αv integrin inhibitor CWHM-12 [a gift from Dr David Griggs (St Louis University, MI, USA)], the β1 integrin inhibitor NOTT199SS [a gift from Dr Thomas McInally (University of Nottingham, UK)], the MMP inhibitor GM6001 (CC1010, Sigma) and the serine protease inhibitor AEBSF (SBR00015, Sigma-Aldrich) were used at varying concentrations. Where inhibitors were dissolved in DMSO, the negative-control cells were treated with a DMSO concentration equivalent to that used in the highest inhibitor concentration.

siRNAs for human GNAQ (ON-TARGET-plus SMARTpool GNAQ, L-008562-00-0005, Dharmacon) and GNA11 (ON-TARGET-plus SMARTpool GNA11, L-010860-00-0005, Dharmacon) were used to induce GNAQ and GNA11 knockdown. A non-targeting siRNA pool was used as a control (ON-TARGET-plus non-targeting pool, D-001810-10-05, Dharmacon).

Cells were seeded at 1.5×10⁵ cells per well of collagen I-coated Bioflex^® 6-well culture plates (3001-C, Dunn Labortechnik) in antibiotic-free DMEM supplemented with 10% FCS and 4 mM L-glutamine. The following day, GNAQ and GNA11 siRNA was applied at a concentration of 15 nM each with 4 µl/ml DharmaFECT 1 transfection reagent (T-2001-01, Dharmacon) as per the manufacturer's protocol. At 48 h after transfection, the media was changed to DMEM supplemented with 1% FCS and 4 mM L-glutamine. Cyclical stretch was applied for 24 h from 72 h post-transfection. G_αq/11 knockdown was confirmed by western blotting and qPCR.

Protein concentrations were determined by BCA assay using a commercially available kit (PN23227, Thermo Fisher Scientific), according to the manufacturer's instructions. Equal amounts of protein (15-25 µg) were loaded per lane of a 10% SDS-polyacrylamide gel and subjected to electrophoresis, then transferred onto a polyvinylidene fluoride membrane (1620177, Bio-Rad). Membranes were blocked for 1 h in either 5% non-fat milk (pSmad2, Smad2/3, αSMA, G_αq, G_α11, GAPDH) or 3% bovine serum albumin (TGFβ1, TGFβ2) in Tris-buffered saline containing 0.1% Tween 20, pH 7.4 (TBST). Membranes were incubated overnight at 4°C in blocking buffer with the appropriate primary antibody. Membranes were washed in TBST, then incubated for 1-2 h in the appropriate horseradish peroxidase (HRP)-conjugated secondary antibody in blocking buffer. Western blots were analysed using chemilluminescence and exposure to film (28-9068-35, GE Healthcare). Where membranes were probed for two different proteins of the same molecular weight, i.e. pSmad2 and Smad2, the membrane was stripped after analysis of pSmad2 using Western Restore Stripping Buffer (21059, Thermo Fisher Scientific) for 5 min and re-blocked with 5% non-fat milk before application of the second primary antibody.

The following antibodies were used for western blots: rabbit anti-phospho-Smad2 (pSmad2) (Cell Signaling Technology, 3108; 1:1000), rabbit anti-Smad2/3 (Cell Signaling Technology, 3102; 1:1000), rabbit anti-αSMA (Abcam, ab5694; 0.5 µg/ml), rabbit anti-GAPDH (Abcam, ab181603; 1:10,000), rabbit anti-TGFβ1 (ab92486; 4 µg/ml), mouse anti-TGFβ2 (Abcam, ab36495; 1:1000), rabbit anti G_α11 (Abcam, ab153951; 1:1000), goat anti-G_αq (Abcam, ab128060; 0.1 µg/ml), HRP-conjugated goat anti-rabbit (Agilent, P044801-2; 1:3000), HRP-conjugated rabbit anti-goat (Agilent, P016002-2; 1:3000), HRP-conjugated rabbit anti-mouse (Agilent, P0260022-2, 1:3000).

Densitometry was performed using ImageJ on scanned western blot images. JPEG images were converted into greyscale images, and the software used to calculate densitometry values for each band relative to the other bands. These relative densitometry values were used to calculate the expression of protein relative to loading control using the equation: protein relative to loading control=protein densitometry value/loading control protein densitometry value.

RNA was isolated from in vitro experiments using the Machery-Nagel Nucleospin RNA isolation kit (740955.250) according to the manufacturer's instructions. Complementary DNA (cDNA) was reverse transcribed from 200 µg RNA using Superscript IV Reverse Transcriptase (18090050, Thermo Fisher Scientific) according to the manufacturer's protocol. Quantitative PCR was performed on cDNA using gene-specific primers (see below), and an MXPro3000 qPCR machine (Stratagene) at an annealing temperature of 60°C for 40 cycles. KAPA SYBR FastTaq (KK4618, Sigma-Aldrich) was used for qPCR of all genes other than Pdgfa, Pdgfb, Pdgfc, Pdgfd, Pdgfra and Pdgfrb, for which PerfeCTa SYBR Green Fastmix (733-1382, VWR) was used. Amplification of a single PCR product was confirmed by melting curve analysis. The ΔΔCt method was used to quantify gene expression relative to the housekeeping genes Hprt (mouse samples) or B2M (human samples).

Primers were obtained from Eurofins. Sequences for mouse genes were: Hprt forward 5′-TGAAAGACTTGCTCGAGATGTCA-3′, Hprt reverse 5′-CCAGCAGGTCAGCAAAGAACT-3′; Acta2 forward 5′-GGGATCCTGACGCTGAAGTA-3′, Acta2 reverse 5′-GACAGCACAGCCTGAATAGC-3′; Eln forward 5′ GATGGTGCACACCTTTGTTG-3′, Eln reverse 5′-CAGTGTGAGCCATCTCA-3′; Col1a1 forward 5′-AGCTTTGTGCACCTCCGGCT-3′, Col1a1 reverse 5′-ACACAGCCGTGCCATTGTGG-3′; Col3a1 forward 5′-TTTGCAGCCTGGGCTCATTT-3′, Col3a1 reverse 5′-AGGTACCGATTTGAACAGACT-3′; Pdgfa forward 5′-GAGATACCCCGGGAGTTGA-3′, Pdgfa reverse 5′-TCTTGCAAACTGCAGGAATG-3′; Pdgfb forward 5′-TGAAATGCTGAGCGACCAC-3′, Pdgfb reverse 5′-AGCTTTCCAACTCGACTCC-3′; Pdgfc forward 5′-AGGTTGTCTCCTGGTCAAGC-3′, Pdgfc reverse 5′-CCTGCGTTTCCTCTACACAC-3′; Pdgfd forward 5′-CCAAGGAACCTGCTTCTGAC-3′, Pdgfd reverse 5′-CTTGGAGGGATCTCCTTGTG-3′; Pdgfra forward 5′-CAAACCCTGAGACCACAATG-3′, Pdgfra reverse 5′-TCCCCCAACAGTAACCCAAG-3′; Pdgfrb forward 5′-TGCCTCAGCCAAATGTCACC-3′, Pdgfrb reverse 5′-TGCTCACCACCTCGTATTCC-3′.

Primer sequences for human genes were: GNAQ forward 5′-GGACAGGAGAGGGTGGCAAG-3′, GNAQ reverse 5′-TGGGATCTTGAGTGTGTCCA-3′; GNA11 forward 5′-CCACTGCTTTGAGAACGTGA-3′, GNA11 reverse 5′-GCAGGTCCTTCTTGTTGAGG-3′; B2M forward 5′-AATCCAAATGCGGCATCT-3′, B2M reverse 5′-GAGTATGCCTGCCGTGTG-3′.

Statistical analyses were performed using GraphPad Prism 8.2 software. For experiments with group sizes of ≤5, a non-parametric test was used. For experiments with group sizes of >5, data were assessed for normality and a parametric test used if data followed a normal distribution.

We thank Dr Thomas McInally (University of Nottingham, UK) and Dr David Griggs (Saint Louis University, MI, USA) for supplying the compounds NOTT199SS and CWHM-12, respectively. NOTT199SS was identified as part of an MSci Chemistry undergraduate integrin drug discovery collaboration between the School of Chemistry at the University of Nottingham and GSK, supervised by Dr Simon Macdonald (GSK) and Dr Thomas McInally (University of Nottingham). We also thank Dr Tim Kendall (University of Edinburgh), for his opinion on the liver histology. We would also like to acknowledge the University of Nottingham Bio-Support and School of Life Sciences Imaging (SLIM) units for their support with the animal work and polarised light microscopy, respectively. In vitro experiments were performed using facilities at the Clinical Sciences Building (University of Nottingham), Nottingham City Hospital.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201046.reviewer-comments.pdf