|Home | About | Journals | Submit | Contact Us | Français|
Alveolar capillary dysplasia (ACD) is a congenital, lethal disorder of the pulmonary vasculature. Phosphatase and tensin homologue deleted from chromosome 10 (Pten) encodes a lipid phosphatase controlling key cellular functions, including stem/progenitor cell proliferation and differentiation; however, the role of PTEN in mesodermal lung cell lineage formation remains unexamined. To determine the role of mesodermal PTEN in the ontogeny of various mesenchymal cell lineages during lung development, we specifically deleted Pten in early embryonic lung mesenchyme in mice. Pups lacking Pten died at birth, with evidence of failure in blood oxygenation. Analysis at the cellular level showed defects in angioblast differentiation to endothelial cells and an accompanying accumulation of the angioblast cell population that was associated with disorganized capillary beds. We also found decreased expression of Forkhead box protein F1 (Foxf1), a gene associated with the ACD human phenotype. Analysis of human samples for ACD revealed a significant decrease in PTEN and increased activated protein kinase B (AKT). These studies demonstrate that mesodermal PTEN has a key role in controlling the amplification of angioblasts as well as their differentiation into endothelial cells, thereby directing the establishment of a functional gas exchange interface. Additionally, these mice could serve as a murine model of ACD.
Understanding of the molecular mechanisms regulating formation of the pulmonary vascular system has advanced in recent years (1–4). However, the complex regulatory network that controls lung vasculogenesis and involvement of mesenchymal progenitors in pulmonary vascular diseases remains unclear. Developmental abnormalities of the pulmonary circulation contribute to significant neonatal morbidity due to disorders such as alveolar capillary dysplasia (ACD) and bronchopulmonary dysplasia (BPD) (5). Congenital ACD is a disorder of pulmonary vascular development seen in neonates, with irreversible persistent pulmonary hypertension and 100% mortality (6). The initial presentation is identical to that of severe idiopathic pulmonary hypertension of the newborn. However, infants with ACD do not respond or respond only transiently to therapies that are usually effective in reversing this condition. Infants with ACD do not improve despite maximal support in the intensive care nursery, including mechanical ventilation, nitric oxide, and extracorporeal membrane oxygenation (ECMO). ACD is also associated with BPD, a chronic lung disease that affects almost 10,000 babies every year in the United States. To date, the underlying etiological factors of these diseases are unknown, and therefore there are no available diagnostic tests and therapeutic options. For these reasons, a better understanding of this process is necessary and mandatory in order to establish urgently needed clinical treatment remedies.
Phosphatase and tensin homologue deleted from chromosome 10 (PTEN) is an important modulator of cell-fate determination and governs normal vascular development. PTEN is a lipid phosphatase involved in the PI3K pathway and is therefore connected with growth factor signaling pathways (7). By controlling the amount of phosphatidylinositol-3,4,5-triphosphate (PIP3), PTEN is involved in several cellular functions, such as cell migration, organ size control, proliferation, and apoptosis (8).
In previous studies, we have shown that absence of Pten in the lung epithelium leads to an increase of epithelial progenitor cells and possibly increased susceptibility to lung cancer, but also to protection against lung injury (9). Hamada et al. (10) reported that mice with an endothelial cell–specific mutation of Pten displayed embryonic lethality due to bleeding and cardiac failure caused by impaired recruitment of pericytes and VSMCs to blood vessels. In addition, Deleris et al. (11) showed that PTEN is also expressed and active in VSMCs controlling the level of PIP3 and therefore potentially controlling VSMC proliferation. To date, the role of Pten in the lung mesenchyme has remained elusive.
In this study, we examined the consequences of mesenchymal-specific deletion of Pten in the embryonic lung using a Dermo1-Cre mouse line that drives Cre expression in the lung mesenchyme as early as E11.5 (2, 12). We observed that the Ptenfl/fl;Dermo-Cre animals died at birth for lack of blood oxygenation. We show that mesodermal PTEN plays a key role in controlling the amplification of angioblasts as well as their differentiation into endothelial cells, thereby allowing the establishment of a functional gas exchange interface. The Ptenfl/fl;Dermo-Cre mice generated in the current study represent a potential murine model of ACD.
We used Dermo1-Cre to inactivate Pten via deletion of exon 5. The expression pattern and efficiency of Dermo1-Cre in the lung were examined by crossing Dermo1-Cre mice with ROSA26R-LacZ reporter mice (Supplemental Figure 1G; supplemental material available online with this article; doi: 10.1172/JCI61334DS1). As reported (1), LacZ activity was widespread throughout the pulmonary mesenchyme, but did not include the endothelial cells in the lung vasculature (Supplemental Figure 1G).
To accomplish mesodermal inactivation of Pten in the lung, heterozygous Ptenfll/+;Dermo-Cre males were crossed with Ptenfl/fl females. The offspring (n = 121) were genotyped using PCR analysis of tail DNA at 3 weeks of age. No Ptenfl/fl;Dermo-Cre (i.e., homozygous deletion) offspring were detected. We therefore determined the percentile of homozygous mutants and WT embryos at different gestational ages (Table (Table1).1). At E12.5 and E15.5, the mutants accounted for 29% (19 out of 65) and 25% (16 out of 63), respectively, of the total number of embryos, while at E18.5, their number was reduced to 21% (67 out of 311), indicating embryonic lethality between E15.5 and E18.5. During embryonic stages, the mutants showed a wide range of phenotype, from a lack of vascularization in entire embryos at E15.5 (Supplemental Figure 2, A and B; 7 out of 16 mutant embryos: 44%) to a hemorrhagic phenotype at E18.5 (Supplemental Figure 2, C and D; 10 out of 67 mutant embryos: 15%). All other mutant embryos with less severe phenotype died within 2 to 3 hours postnatally, displaying cyanosis, chest retractions, and dyspnea (Supplemental Figure 2E). Measurements of blood oxygenation (Figure (Figure1I)1I) showed statistically significant differences between Ptenfl/fl (controls) and Ptenfl/fl;Dermo-Cre newborns (97% ± 3.7% vs. 73% ± 8.2%, P < 0.01). A careful study of the embryos at E15.5 showed lack of vascularization in other organs, such as limbs and liver (Supplemental Figure 3, A–J).
To validate Pten inactivation in the pulmonary mesoderm, we compared expression patterns and levels of PTEN and phosphorylated protein kinase B (p-AKT) in mutant versus WT lungs by immunohistochemistry (IHC) in E18.5 embryos (Supplemental Figure 1, A–D). When compared with controls, mutant lungs showed an overall decreased PTEN (Supplemental Figure 1, compare B and A). However, recombination was not complete, as some mesodermal-derived cells remained PTEN positive, displaying nuclear immunoreactivity (Supplemental Figure 1B). Consistent with the specificity previously documented for Dermo1-Cre activity, epithelial PTEN immunoreactivity was unperturbed (Supplemental Figure 1B). Using E18.5 lung RNA, quantitative RT-PCR (qRT-PCR) was performed to confirm interruption of Pten gene expression (Supplemental Figure 1E). In Ptenfl/fl;Dermo-Cre lungs, Pten mRNA was decreased (0.58% ± 0.07%, P < 0.01) compared with Ptenfl/fl controls (n = 4 per group). Finally, Western blot analysis (Supplemental Figure 1F) showed that PTEN was reduced in the mutant lungs compared with controls (0.14 ± 0.01 vs. 0.51 ± 0.09, n = 4 per group, P < 0.05).
PTEN is a lipid phosphatase that regulates PIP3 levels, thus negatively modulating the PI3K/AKT pathway. IHC revealed an increase of p-AKT in mutant lungs compared with controls (Supplemental Figure 1, D vs. C), further confirming deregulation of the PI3K pathway. Western blot analysis of E18.5 whole-lung extracts (n = 4 for each group) confirmed increased p-AKT in mutant versus control lungs (0.03 ± 0.004 vs. 0.012 ± 0.00053 P < 0.05; Supplemental Figure 1F).
To determine the potential structural causes underlying reduced blood oxygenation in the mutant newborns, histology of E18.5 lungs using H&E staining was examined (Figure (Figure1,1, A–D). While no branching or other significant gross structural abnormalities were observed (Figure (Figure1,1, B and D vs. A and C), closer examination of mutant lungs (Figure (Figure1D)1D) revealed a hypercellular mesenchymal compartment compared with the controls (Figure (Figure1C).1C). This observation was confirmed by detection of mesenchymal-specific increased cell proliferation, as documented by E-cadherin (E-CAD)/PH3 immunofluorescence (Figure (Figure1,1, E and F). Due to increased mesodermal cells, the total number of E-CAD–negative cells was higher in mutant lungs compared with controls (Figure (Figure1G,1G, 363.5 ± 20.7 vs. 223.1 ± 10.2, P < 0.01). The number of E-CAD–negative/PH3-positive cells was nearly 6-fold higher in mutant versus control lungs (17.8% ± 1.1% versus 3.15% ± 0.8% respectively n = 3, P < 0.01) (Figure (Figure11H).
Differentiated endothelial cells, identified by PECAM, are derived from mesodermally derived progenitor cells, which express Flk1 (Vegfr2). These progenitors can be identified in embryonic lungs starting as early as E10.5 (13). Their formation is tightly associated with mesenchymal and epithelial development during branching morphogenesis (1). As defective pulmonary vascular formation can cause hypoxemia, we examined PECAM in E15.5 lungs by IHC (Figure (Figure2,2, A and B). In Ptenfl/fl;Dermo-Cre lungs, there was significant reduction in distal capillary network density compared with Ptenfl/fl controls, the latter exhibiting normal dense capillary plexus adjacent to the developing airway epithelium. In E18.5 control lungs (Figure (Figure2C),2C), red blood cell–containing capillaries were found in close spatial proximity to the alveolar-air interface. This coupling is critical for normal gas exchange. In contrast, the capillary network in Ptenfl/fl;Dermo-Cre lungs was distinctly abnormal, with clear uncoupled capillary/airway network, rendering pulmonary units incapable of efficient gas exchange (Figure (Figure2D).2D). Statistical analysis of the distance between the capillaries and the lumen of airways showed significant increase in the mutant versus control lungs (Figure (Figure2E,2E, 3.3 μm ± 0.3 μm vs. 0.9 μm ± 0.1 μm, P < 0.01). More detailed study of the lung ultrastructure showed mutant alveolar spaces were frequently lined by cuboidal cells with “immature” lamellar bodies, while the type 2 cells in control lungs contained surfactant. To better understand the mechanism underlying this phenotype, we assessed the levels of Vegfa, Flt1 (Vegfr1), and Flk1 (Vegfr2) at E18.5 by qRT-PCR. Vegfa was reduced in mutant lungs compared with controls (Figure (Figure3A,3A, 0.39 ± 0.03 vs. 1, P < 0.01) as was Flt1 (Figure (Figure3A,3A, 0.73 ± 0.08 vs. 1, P < 0.05). There was also a quantifiable increase in Flk1 mRNA in Ptenfl/fl;Dermo-Cre lungs (Figure (Figure3A,3A, 11.61 ± 2.6 vs. 1, P < 0.05). Furthermore, Pecam mRNA was decreased in the mutant lungs versus controls (Figure (Figure3A,3A, 0.47 ± 0.1 vs. 1, P < 0.01), supporting the IHC data shown in Figure Figure2,2, A and B.
Finally, qRT-PCR for Vegfa, using mRNA from E15.5 and E18.5 isolated fibroblasts, revealed no statistically significant difference between control and mutant cells (Figure (Figure3F,3F, E15.5: 1.067 ± 0.5 vs. 1, P > 0.05; E18.5: 1.16 ± 0.6 vs. 1, P > 0.05). Thus, as the mesenchyme does not appear to be involved, the data suggest the epithelium as the compartment in which the decrease in Vegfa occurs. This conclusion was validated via immunofluorescent (IF) for VEGFA plus E-CAD, which showed markedly decreased VEGFA in E-CAD–positive cells in the mutant lungs (compare Figure Figure3,3, H and G).
One possibility for the changes in Flk1 and Flt1 levels is that mesodermal-specific Pten inactivation affects endothelial cell differentiation even though Dermo1-Cre as shown in this and other reports is not active in this cell lineage (1, 2). Therefore, we examined the differentiation of endothelial cells by generating triple-transgenic mice consisting of the Ptenfl/fl;Dermo-Cre embryos in a Flk1LacZ reporter background (14). In these mice, LacZ is under the control of the endogenous Flk1 promoter. Flk1 is an early marker of angioblasts, and its expression is maintained in mature endothelial cells together with Flt1 (14). Significant increase in Flk1-driven LacZ activity was found throughout the entire mutant lungs compared with controls (Figure (Figure3,3, C vs. B). Ablation of mesodermal Pten therefore does not interfere with specification and amplification of angioblasts. However, analysis of PECAM, a marker for mature endothelial cells in Ptenfl/fl;Dermo-Cre lungs, revealed impaired differentiation of angioblasts into mature endothelial cells and blood vessels when compared with controls. We therefore analyzed sections from E18.5 Ptenfl/fl;Dermo-Cre;Flk1LacZ and Ptenfl/fl;Flk1LacZ lungs and examined them for PECAM. The mutant lungs showed increased FLK1-positive/PECAM-negative cells (Figure (Figure3,3, E vs. D), confirming the conclusion that angioblasts are indeed increased in the mutant lungs. Finally, we examined the tridimensional structure of lung vasculature by injecting FITC leptin in the left ventricle of mutant and control hearts, observing again a significant defect in vasculogenesis in mutant versus controls (Supplemental Video 1). In sum, the results strongly support the conclusion that lack of PTEN in the mesodermal lineages inhibits differentiation of angioblasts into mature endothelial cells.
To determine the impact of mesodermal Pten inactivation on the emergence and differentiation of lung mesenchymal cell lineages, expression of a number of cell markers was examined by IHC and qRT-PCR. Immunofluorescence showed decreased α-SMA, a smooth muscle differentiation marker in E18.5 mutant lungs (Figure (Figure4,4, B versus A). Adipose differentiation-related protein (ADRP), a lipid droplet–associated protein expressed early during adipose differentiation and a marker of lipofibroblasts in the lung, was decreased in Ptenfl/fl;Dermo-Cre lungs compared with Ptenfl/fl (Figure (Figure4,4, D vs. C). Finally, in E18.5 lungs, the capillary network was misaligned with corresponding respiratory airways (i.e., airway/capillary uncoupling or dysplasia) in the mutants, as shown by IHC for PECAM (Figure (Figure4,4, F vs. E). Interestingly, double IHC for PDGFrα (marker for fibroblasts) and E-CAD (marker for epithelial cells) and Sirius red staining for collagen, a product of fibroblasts, showed increased PDGFrα-positive/E-CAD–negative cells (Figure (Figure4,4, H vs. G, statistical analysis: Figure Figure4L,4L, 2% ± 0.2% vs. 11% ± 0.02%, P < 0.01). Collagen deposition was also increased in the Ptenfl/fl;Dermo-Cre lungs compared with controls (Figure (Figure4,4, J vs. I). These IHC results were validated by qRT-PCR for the different cell lineage markers, which showed a trend toward reduced expression of α-Sma (Figure (Figure4K,4K, 0.52 ± 0.3 vs. 1, P > 0.05) and Adrp (Figure (Figure4K,4K, 0.03 ± 0.00001 vs. 1, P < 0.01) in primary cultures of E18.5 mutant fibroblasts compared with controls. Pecam was also reduced in the mutant versus control lungs (Figure (Figure4K,4K, 0.47 ± 0.1 vs. 1, P < 0.01). Collagen 1 by Sircol assay was increased in the mutant lungs versus controls (Figure (Figure4M,4M, 83.13 μg/ml ± 7.2 vs. 39.23 μg/ml ± 6.3, P < 0.01).
Using flow cytometry to confirm these observations, we gated the mesenchymal progenitor cells by Hoechst staining in E17.5 mutant and control lungs (n = 14 for each). In the embryonic lung, 2 subtypes of side populations are thought to exist (15). One is identified as CD45–CD31+E-SP cells, believed to differentiate into endothelial cells. The other is identified by markers CD45–CD31–E-SP, with a gene profile consistent with a smooth muscle precursor (15). At E17.5, we observed more than a 5-fold increase in the number of CD45–CD31+E-SP and CD45–CD31–E-SP cell populations in Ptenfl/fl;Dermo-Cre lungs versus controls (respectively 1.1% versus 0.2% and 2.9% versus 0.5%) (Figure (Figure4N).4N). Therefore, mesenchymal PTEN controls the size of the 2 putative lung mesenchymal progenitor cell populations.
AKT-mediated phosphorylation of β-catenin (β-CAT) on Ser552 drives nuclear localization of β-CAT and activation of WNT target genes. To determine whether mesodermal Pten inactivation increases phosphorylation of β-CAT on Ser552 in the mesenchyme, we performed IHC for E-CAD and β-CATSer552 (Figure (Figure5,5, G and H), the form of β-CAT specifically phosphorylated by the PTEN/AKT pathway (16). We found significantly increased E-CAD–negative/ β-CATSer552–positive cells in mutant lungs compared with controls (Figure (Figure5J,5J, 1.3% ± 0.1 vs. 0.3% ± 0.1, P < 0.01). Co-IF for β-CAT and mesenchymal markers including PECAM or PDGFRα showed that these cells are negative for these markers (data not shown). We did not observe β-CATSer552–positive cells in the epithelium. In our model, absence of mesenchymal Pten affected lipofibroblast, VSMC, and parabronchial SMC (PSMC) differentiation likely due to increased levels of Ser552 β-CAT, a recognized positive regulator of stem cell homeostasis downstream of the AKT pathway.
The pulmonary mesenchyme is the source of many signaling factors, the most important of which is FGF10, which interacts with receptor tyrosine kinases (RTKs). Fgf10 expression by the mesenchyme is positively regulated by WNT and FGF9 signaling and is negatively regulated by sonic hedgehog (SHH) (2, 4). Deletion of Fgf10 leads to lung agenesis, suggesting a key role in mediating mesenchymal-epithelial interactions that are necessary for lung morphogenesis (17, 18). FGF10 is the major FGFR2b ligand during the pseudoglandular stage. FGF10 is secreted exclusively by the mesenchyme and acts on the epithelium through FGFR2b.
Because of the increase in β-CATSer552, Fgf10 expression was analyzed by in situ hybridization (ISH). Fgf10 mRNA was increased in mutant lungs compared with controls (Figure (Figure5,5, B vs. A). qRT-PCR analysis for Fgf10 confirmed the ISH results (Figure (Figure5I,5I, 2.09 ± 0.48 vs. 1, P < 0.05). A downstream pathway activated by FGF10 in the epithelium is β-CAT signaling (3). To examine whether the observed increased FGF10 and increased mesenchymal β-CATSer552 stimulates β-CAT signaling, we generated triple transgenic mice composed of Ptenfl/fl;Dermo-Cre mice and Batgal, the latter a WNT signaling reporter (19), and analyzed the embryos at E15.5 (Figure (Figure5,5, C–F). β-CAT signaling was increased in lung epithelium and mesenchyme of the Ptenfl/fl;Dermo-Cre;Batgal lungs compared with controls (Figure (Figure5,5, D and F vs. C and E).
Since activated β-CAT is recognized as controlling stem cell homeostasis in lung epithelial cells, we performed IF for epithelial markers in E18.5 control and mutant lungs (Figure (Figure6,6, C–H). Double IF with antibodies to E-CAD and FGFR2 showed increased FGFR2 in E18.5 Ptenfl/fl;Dermo-Cre lungs compared with controls (Figure (Figure6,6, B vs. A). qRT-PCR confirmed this observation by showing a 4-fold increase of Fgfr2b mRNA in mutant lungs (Figure (Figure6I,6I, 4.1 ± 0.3 vs. 1, P < 0.01). qRT-PCR showed increased expression for Fgf9 and Fgf7, 2 additional growth factors made by the lung mesothelium/epithelium (Fgf9) and by the mesenchyme (Fgf7), (Figure (Figure6I,6I, Fgf9: 2.6 ± 0.5 vs. 1, P < 0.05; Fgf7: 5.7 vs. 1 ± 1.6, P < 0.05). To validate these results, we also examined the expression of FGF10/FGFR2b downstream and positive regulator targets by qRT-PCR. As expected, compared with the controls, Wnt2b, Spry2, Etv4, Etv5, and Bmp4 were increased, while Shh, Ptch1, and Gli1 decreased in the mutant lungs (Figure (Figure6I,6I, Wnt2b: 3.03 ± 1.2 vs. 1, NS; Spry2: 2.46 ± 0.58 vs. 1, P < 0.05; Etv4: 1.26 ± 0.15 vs. 1, NS; Etv5: 1.56 ± 0.24 vs. 1, P < 0.05; Bmp4: 1.26 ± 0.06 vs. 1, P < 0.01; Figure Figure6I,6I, Shh: 0.6 ± 0.3 vs. 1, P < 0.05; Ptch1: 0.47 ± 0.1, P < 0.05; Gli1: 0.5% ± 0.06 vs. 1, P < 0.01).
As we observed increased FGF10 activity in the epithelium of the mutant lungs, we investigated whether this change affected epithelial differentiation along the pulmonary proximal and distal domains (20). IHC for SPC, T1α, and CC10 markers for alveolar type 2 (distal domain), alveolar type 1 (distal domain) and Clara cells (proximal domain), respectively, showed an increase of alveolar type 2 (SPC-positive) cells, while the Clara cells (CC10-positive) and type 1 cells (T1α-positive) were unchanged in the Ptenfl/fl;Dermo-Cre lungs compared with Ptenfl/fl controls (Figure (Figure6,6, C–H). Western blot analysis demonstrated a statistically significant increase in SPC expression (marker for alveolar type 2 cells) in the mutant compared with controls (Figure (Figure6J,6J, SPC: 0.2 ± 0.003 vs. 0.09 ± 0.003, in arbitrary units, P < 0.05). In addition, comparison of CC10 and T1α expression showed no statistical difference between the 2 groups (data not shown). Finally, we performed IHC for TTF1 (NKX2.1), the first transcriptional factor present in the lung during embryogenesis (21), and for ID2, a marker of distal lung progenitor cells (22). The increase in TTF1-positive cells (Figure (Figure6,6, L vs. K) and ID2/E-CAD–double-positive cells (Figure (Figure6,6, N vs. M) in the mutant lungs compared with the control confirmed the expansion of the distal domain. Western blot results further confirmed the latter results, showing an increase of TTF1 (0.13 ± 0.006 vs. 0.23 ± 0.006, P < 0.01) in the mutants compared with controls (Figure (Figure6J).6J). Overall, these data suggest that the increase in FGF10 signaling in the epithelium, due to Pten mesenchymal inactivation, is the likely cause of distal epithelial domain expansion.
As the phenotype of the lungs with mesenchymal Pten inactivation resembled what has been described in newborn infants as ACD, we examined the hypothesis that the PTEN/PI3K/AKT pathway is activated in the lungs of ACD patients. For this purpose, we performed PTEN and p-AKT IHC on 5 different and independent lung samples from newborn human infants who died with ACD diagnosis and compared these to lungs of newborns who died of causes unrelated to ACD. ACD patients showed a decrease of PTEN staining and a corresponding increase in p-AKT–positive cells compared with lungs from control patients. While PTEN staining was readily detectable (Figure (Figure7,7, B vs. A), the number of p-AKT–positive cells was limited in the control lungs (Figure (Figure7,7, D vs. C). Statistical analysis showed an increase in the percentage of p-AKT–positive cells in the ACD lungs compared with controls (Figure (Figure7F,7F, 1.4 ± 0.4 vs. 0.07 ± 0.05, P < 0.05).
ACD has been associated with mutations in Forkhead box protein F1 (FOXF1) in humans and heterozygous Foxf1 and Foxc2 loss of function in mice (23, 24). Accordingly, Foxf1 and Foxc2 mRNA were measured by qRT-PCR. We found decreased expression of both genes in mutant compared with control lungs (Figure (Figure7E,7E, Foxf1: 0.59 ± 0.015 vs. 1, P < 0.01; Foxc2: 0.57 ± 0.13 vs. 1, P < 0.05). These data support the involvement of the PTEN/PI3K/AKT pathway in the pathogenesis of ACD and justify further analysis of this pathway in our newly described mouse model of a lethal human congenital lung disease.
A well-organized and functional capillary network is essential for air-blood gas exchange and therefore for oxygenation of tissues and organs. Impaired development of the vascular capillary network in the lung can cause life-threatening deficiencies in pulmonary function present in neonatal diseases such as BPD and ACD (25, 26).
The results of the present work show that inactivation of Pten in the mesodermal lineages that contribute to lung morphogenesis disrupts normal capillary plexus formation as well as mesenchymal and epithelial differentiation.
The role of PTEN in mesodermal lineage formation during lung development had not been hitherto adequately examined. Pten was previously shown to be important in regulating proliferation of endothelial cells (10). Endothelial-specific inactivation of Pten using the Tie2-Cre driver mice led to early embryonic lethality due to fatal vascular remodeling defects secondary to an impaired recruitment of VSMCs. Hamada and colleagues (10) showed also that the absence of PTEN in endothelial cells causes hypersensitivity to VEGFs, resulting in enhanced angiogenesis and bleeding due to a failure in vascular remodeling. Deleris et al. (11) found that PTEN is also expressed and active in VSMCs controlling the level of PIP3 and therefore potentially controlling VSMC proliferation. Another report (27) showed that absence of PTEN in VSMCs leads to vascular remodeling and increased progenitor cell recruitment. However, to date, the specific deletion of Pten in lung mesenchyme during development has not been reported.
In the present work, we used the Dermo1-Cre driver line to inactivate Pten from the onset of lung morphogenesis (i.e., E9.5) in all mesodermally derived cells (refs. 12, 28, and Supplemental Figure 1G) except for endothelial progenitors. The resulting mutant mice exhibited perinatal lethality that was largely due to vascular defects and mesenchymal hypercellularity. The newborn mice were clearly hypoxemic and could not oxygenate their blood. Detailed examination of their lungs revealed decreased capillary formation, but more importantly, misalignment of the existing capillary network with the airways, due also to the increased number of mesenchymal cells in the parenchyma. Interestingly, our genetic manipulation, which enabled the embryos to survive until birth, is significantly different from the one resulting from endothelial-specific deletion of Pten (10). Thus, the use of the Dermo1-Cre driver line to delete Pten in all mesodermally derived cells except for endothelial cells provides new insight into the key role played by PTEN in mesodermal lineage formation.
To investigate the underlying mechanisms for the observed vascular defects in Ptenfl/fl;Dermo-Cre mutants, we measured a number of relevant parameters. Vegfa was markedly reduced in the mutant lungs at E18.5. Vegfa is first expressed in the distal subepithelial mesenchyme, but begins to be expressed also in the epithelium at E14.5 (1). Thereafter, Vegfa becomes progressively restricted to the epithelium (1). It is therefore likely that the main source of Vegfa, whose secretion is decreased in our mutants, is the epithelium. Interestingly, the decrease in VEGFA occurs in spite of increased FGF10 signaling to the epithelium, which we had previously shown to be a positive regulator of Vegfa, using Fgf10 hypomorphic mice (3). However, in these mutants, this decrease is likely the consequence of reduced epithelial cell numbers rather than being caused by a direct transcriptional regulation of Vegfa by FGF10. In support of this observation, the decrease in TTF1 expression (marker of lung epithelium) in the lungs of Fgf10 hypomorphic embryos was also apparent within the same range as was observed for Vegfa (3). The change in Vegfa therefore is indirect and may reflect changes in the epithelial/mesenchymal ratio, which was drastically changed in favor of the mesenchyme in Ptenfl/fl;Dermo-Cre lungs (higher FGF10 signal leads to only modest epithelial progenitor cell increase, whereas mesodermal Pten inactivation causes a marked increase in mesenchymal proliferation). This, in turn, will automatically lead to a global decrease in Vegfa expression in the lung. Reduced VEGFA will undoubtedly cause corresponding defects in vascular formation. In addition, SHH and FGF9 are reported to be critical for vascular development (1). Therefore, as Shh is decreased, vascular (endothelial) development is also expected to be impaired in Ptenfl/fl;Dermo-Cre lungs.
Interestingly, the lack of differentiation in the epithelial compartment was associated with increased angioblasts in the mesenchyme. This was supported by the increase in Flk1 (marker of vascular progenitors), but also by concomitant decrease in Flt1 and Pecam (both markers of differentiated endothelial cells) mRNAs. As previously demonstrated, absence of Flt1 leads to disorganized angiogenesis due to the role of this receptor in vascular commitment (29). Moreover, PTEN may not be necessary for differentiation of angioblasts from the ventral mesoderm and for their migration within the lung, as has been shown previously (10). Accordingly, we observed that the absence of PTEN did not affect the formation of the major vessels or the angioblast population. Nevertheless, our studies identified an important role for mesenchymal Pten in the differentiation of angioblasts to endothelial cells. As Pten is not deleted in endothelial cells, this role is likely indirect, being an outcome of alterations in mesenchymal-endothelial or mesenchymal-epithelial-endothelial crosstalk mediated by as yet unknown signaling molecules.
In the course of this study, we also noticed that PTEN in the lung mesenchymal cells was present both in the cytoplasm and in the nuclei, while in the epithelium, PTEN was confined only to the cytoplasm. A recent report investigating the role of PTEN in cancer cells showed that nuclear-localized PTEN interacts with APC/C (anaphase promoting complex) and promotes its association with Fizzy-related protein 1 (FZR1), thereby enhancing tumor-suppressive activity of the APC/C-FZR1 complex in a phosphatase-independent manner (30). In this way, both proliferation and senescence are affected by reduction of APC/C activity upon Pten loss. Whether the lack of PTEN in the nuclei or in the cytoplasm of mesenchymal cells is responsible for some aspect of the phenotype observed in Ptenfl/fl;Dermo-Cre lungs remains to be clarified by future studies.
The mesenchyme is essential during lung development, as it contains the progenitors for several mesenchymal cell types that include the PSMCs, the VSMCs, the lipofibroblast, the alveolar myofibroblasts, and the stromal fibroblasts (31, 32). The proper generation of these cells relies on the controlled amplification of progenitors followed by the exit of these cells from the cell cycle and their timely differentiation into appropriate progeny (2). Several studies (9, 33) have shown that epithelial inactivation of Pten in the lung leads to an increase in epithelial progenitor cells and impaired epithelial cell differentiation. In our model, absence of mesenchymal Pten affected lipofibroblast, VSMC, and PSMC differentiation, likely via an increase in β-CAT Ser552–mediated WNT signaling, a recognized regulator of epithelial stem cell homeostasis. Indeed, we have previously shown that mesenchymal-specific deletion of β-CAT results in lack of amplification and premature differentiation of PSMC progenitor cells (2). Moreover, we have recently shown that reactivation of Wnt signaling in mature PSMC after airway epithelial injury results in their dedifferentiation and adoption of a PSMC progenitor-like state and reexpression of Fgf10 (34). Interestingly, in contrast to our findings, others have reported increased SMC differentiation associated with increased WNT signaling (35). One difference is that in the latter study, SMC-specific targeting was employed, while in our study, Dermo1-Cre driver line targeted the mesenchymal progenitor cells. In addition, treatment of lung explants in the pseudoglandular stage with LiCl to increase Wnt signaling in both epithelium and the mesenchyme was associated with increased α-SMA around bronchi and blood vessels (35). Preliminary data from our group show that after Dermo1-Cre deletion of the negative regulator of WNT signaling, adenomatous polyposis coli (leading to increased β-CAT signaling) differentiation of α-SMA cells was blocked (P. Minoo et al., unpublished data). These data are consistent with our findings in the present manuscript. Thus, lack of β-CAT in the mesenchyme led to premature differentiation of PSMC and lack of amplification of the progenitors, while increased WNT signaling blocked differentiation and increased mesenchymal progenitors. Finally, Pten deletion during embryonic stages led to increased collagen production. This could occur either by induction of collagen expression in mesenchymal progenitors or by promotion of fibroblast differentiation. Interestingly, several studies have reported decrease or absence of PTEN in fibrotic foci in idiopathic pulmonary fibrosis (IPF) (36).
Epithelial-mesenchymal interactions mediated via key signaling factors during lung development are critical for proper patterning of the epithelium and mesenchyme along the proximal-distal axis. The mesenchyme is the source of important signaling factors, such as FGF10, required for airway epithelial development and branching. Also, the epithelium produces signaling molecules essential for mesenchymal differentiation and proliferation, such as bone morphogenetic protein 4 (BMP4), SHH (18, 37, 38), and FGF9 (1). The observations in this study showed that mesodermal Pten inactivation caused increased Fgf9, Fgf10, and Fgf7 expression and decreased expression of Shh, Ptch1, and Gli1. We further found that increase in FGF10 caused increased FGFR2b signaling, supported by upregulation of FGF10 downstream targets Fgfr2b, Bmp4, and Spry2. Such changes in Ptenfl/fl;Dermo-Cre lungs would undoubtedly have an impact on lung development and cell differentiation. Thus, mesenchymal-specific inactivation of Pten led to arrested differentiation of mesoderm-derived cells. Concomitantly, we observed an increase in the number of SMC progenitors, which include the Fgf10-positive cells. The latter have been described as giving rise to PSMCs. The increased number of Fgf10-expressing cells is likely responsible for increased expression of FGF10 downstream targets Fgfr2b, Bmp4, and Spry2 in the mutant lungs. These cells proliferate and give rise to an increased number of distal SPC-positive cells. IHC analysis for ID2, a distal cell marker, suggested alterations in proximal-distal differentiation of epithelial cells; increased numbers of ID2-positive cells were found to be localized proximally in the mutant lung. Interestingly, we did not detect any defect in epithelial differentiation in the proximal lung; CC10 and T1α staining indicated a normal pool of Clara and type 1 cells in both mutants and control lungs. Therefore, absence of PTEN in the mesenchyme leads to the expansion of the pool of distal epithelial progenitor cells likely secondary to increased FGF10/FGFR2b signaling in the epithelium.
Our observations suggested vascular similarities in phenotype between Ptenfl/fl;Dermo-Cre lungs and the lungs of human neonates with ACD. FGF9 and SHH pathways were previously shown to be important for vasculogenesis, and disruption of their function leads to ACD-like phenotypes (1). Consistent with the latter findings,we found alterations in both FGF9 and SHH pathways in Ptenfl/fl;Dermo-Cre lungs. qRT-PCR revealed increased Fgf9 mRNA, while Shh and its downstream targets, Ptch1 and Gli1, were decreased in the mutant lungs (Figure (Figure6I).6I). Thus, both the vascular phenotype and the molecular alterations in Ptenfl/fl;Dermo-Cre lungs are consistent with the features associated with human ACD. Furthermore, a significant decrease in Foxf1 and Foxc2 mRNAs (Figure (Figure7E)7E) was also documented in the Ptenfl/fl;Dermo-Cre lungs, also consistent with previous findings in human and murine ACD phenotype (23, 24). Finally, to directly ascertain the relevance of our findings in human tissues, we conducted IHC for PTEN and p-AKT in 5 independent human lung samples from ACD patients and newborns who died of causes other than ACD. The absence of PTEN staining and the increase in p-AKT–positive cells in ACD lungs demonstrated that the PI3K/AKT pathway is activated and that loss of PTEN may be relevant to this activation. It has been reported that Foxf1 expression is directly inhibited by AKT in liver myofibroblasts (39). Therefore, it is possible that a similar situation occurs also in the lung where Foxf1 could be a direct transcriptional target for PTEN signaling in the lung mesenchyme through AKT activation. Whether loss of PTEN can be recognized as primum movens in the pathogenesis of ACD and whether activation of PTEN can be used as a therapeutic target remains to be determined.
In conclusion, mesodermal-specific inactivation of Pten by Dermo1-Cre leads to a phenotype resembling ACD, a dramatic neonatal disease with 100% mortality at birth. The cellular mechanisms underlying the phenotype are connected with an increase in mesenchymal progenitor cells and a reduced differentiation of these cells into PSMCs and lipofibroblasts. This also leads indirectly to an increase in the angioblast pool as well as defective differentiation of these cells into endothelial cells. We propose a model whereby increased FGF10-expressing PSMC progenitor cells activate the FGF10/FGFR2b pathway in the epithelium, leading to an expansion of the distal progenitor pools containing ID2-positive cells. Lack of epithelial differentiation also leads to decreased Vegfa during early embryonic stages that in turn causes defects in capillary development around the primitive alveoli and, therefore, a lack of oxygenation at birth, with subsequent perinatal lethality. Finally, blocked differentiation of mesenchymal progenitors to SMC and lipofibroblasts drives progenitor cells toward an alternative fibroblast phenotype, leading to increased collagen production (Figure (Figure77G).
Ptenfl/fl (C57BL/6 background; stock no. 006440), Flk1LacZ/+ (stock no. 002938), Dermo1-Cre (stock no. 008712), and Batgal (stock no. 005317) mice were from Jackson Laboratory. Pten knockouts were obtained by mating Ptenfl/fl females with Dermo1-Cre male mice (C57BL/6 background) for 2 generations to obtain Dermo1-Cre; Ptenfl/fl (heretofore Ptenfl/fl;Dermo-Cre) and Dermo1-cre;Ptenfl/+ (heretofore Ptenfl/+;Dermo-Cre) on a C57BL/6 background. Ptenfl/fl mice were used as control. Genotyping of the Dermo1-Cre mice containing Ptenfl, PtenΔ, and PtenWT alleles was as previously described (12, 40).
Heart rate and transcutaneous oxygen saturation were monitored every 10 seconds for 1 minute in newborn mice via a pulse oximeter (MouseOx; STARR Life Sciences Corporation) by placing noninvasive sensors on the back of the newborn mice.
Lungs from embryos were collected at E15.5 and E18.5 and fixed overnight, dehydrated through increasing ethanol concentrations, embedded in paraffin, and prepared as 5-μm sections.
E18.5 lungs (6 controls and 6 mutants) were mechanically disrupted and incubated at 37°C with 1:1 Trypsin 0.025%/EDTA: DMEM/F12. Dissociated cells were passed through a 19.5- to 27-gauge needle, incubated in DMEM/F12 with 10% FBS (2 hours), and washed 10 times with DMEM/F12. Adherent fibroblasts were grown for 24 hours, detached by scraping, and centrifuged (at 1,700 g for 5 minutes). Cell pellets were stored at –80°C.
Data were presented as means ± SEM unless stated otherwise. The data were statistically analyzed using 2-tailed Student’s t tests to compare the 2 groups. P < 0.05 was considered significant.
All animal experiments were approved by the University of Southern California Animal Use and Care Committee. Post mortem unidentified human ACD tissue was contributed by P. Sen and C. Langston (Baylor College of Medicine, Houston, Texas, USA) (IRB number, H-8712).
This work was supported by NIH HL56594 (to P. Minoo), HL086322 (to S. Bellusci), HL074832 (to S. Bellusci), the Excellence Cluster Cardio-Pulmonary System, LOEWE (to S. Bellusci), HL092967 (to S. De Langhe), the Hastings Foundation (to P. Minoo), the CIRM Clinical Fellowship (to C. Tiozzo and G. Carraro), NIH K12-HD-052954 (to B. Chan), and the American Lung Association as well as the American Heart Association (to D.A. Alam). P. Minoo is Hastings Professor of Pediatrics. S. Bellusci holds the chair of Lung Matrix Remodelling at Justus Liebig University. Z. Borok is Edgington Chair in Medicine. The authors also thank the ACD Association and P. Sen and C. Langston (Baylor College of Medicine) for providing human ACD lung samples. We thank Hiroyuki Shimada (Children’s Hospital, Los Angeles) for discussion and technical support, Esteban Fernandez for help with tridimensional reconstruction of lung vasculature, and Clarence Wigfall for help with the processing of the videos. C. Tiozzo acknowledges C. Fagan and L. Barrow.
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J Clin Invest. 2012;122(11):3862–3872. doi:10.1172/JCI61334.
Caterina Tiozzo’s present address is: Nassau University Medical Center, Pediatrics Department, East Meadow, New York, USA.