Homozygous deletion, but not heterozygous deletion, of Apc resulted in ectopic activation of Wnt/Ctnnb1 in embryonic lung mesenchyme
Using a Tbx4 lung enhancer-driven Tet-On transgenic system generated in our lab [11], we were able to induce Cre expression specifically in mouse embryonic lung mesenchymal cells (Fig. 1a). The Apc CKO mice were induced during lung branching morphogenesis by administering doxycycline (Dox) from E10.5. Deletion of Apc exon 14 in lung tissue was verified at both genomic DNA and mRNA levels (Fig. 1b,1c). Since Apc is a negative regulator for the Wnt/Ctnnb1 canonical pathway [4], loss of Apc function is expected to result in abnormal activation of Ctnnb1. In our homozygous Apc CKO embryos, hyperactivation of Ctnnb1 was detected in embryonic lung mesenchyme, reflected by accumulation of non-phospho (Ser37/Thr41, also called active) Ctnnb1 at as early as E11.5 and significantly increased expression of Axin2 (a Ctnnb1 downstream target gene) from E12.5 (Fig. 1d–1f and Additional file 1). In contrast, staining of Ctnnb1 in airway epithelial cells, mainly localized on apical cell membranes, was comparable between Apc CKO and WT lungs (Fig. 1d and Additional file 1), confirming mesenchymal specificity of altered Wnt signaling activity due to loss of Apc function. Interestingly, like the wild-type (WT) controls, heterozygous Apc CKO (HT) mice did not display detectable Wnt/Ctnnb1 activation in embryonic lung mesenchyme (Fig. 1d), suggesting that a single allele of WT Apc gene is sufficient to suppress abnormal activation of Wnt/Ctnnb1 signaling in these cells.
Lung mesenchyme-specific Apc conditional knockout resulted in abnormal lung morphogenesis and fetal lethality at mid-gestation
Abrogation of Apc in lung mesenchyme starting from E10.5 did not affect the overall growth of the embryos/fetuses by comparison of their body sizes among different genotypes (Fig. 2a). However, the Apc CKO fetuses had severe chest hemorrhages with dark blue coloration by gross view at E14.5 (Fig. 2a), and died soon after E15.5. In order to determine the dynamic changes of the phenotypes, lungs of Apc CKO embryos were isolated from E11.5 to E14.5. As shown in Fig. 2b, most of the E11.5 Apc CKO lungs had normal epithelial domain branches surrounded with appropriate mesenchyme comparable to WT littermate controls. However, one day later (E12.5), airway branches of the Apc CKO lungs became difficult to see under a dissecting microscope due to a thickened and condensed mesenchymal compartment. The Apc CKO lung growth seemed fully arrested by E13.5, with a size and appearance similar to Apc CKO lungs at E12.5. In contrast, WT lungs grew rapidly with many airway branches. At E14.5, lung tissue destruction with massive hemorrhage was detected in Apc CKO mice, explaining the lung contusion observed within the whole body view (Fig. 2a). Interestingly, heterozygous Apc CKO did not result in any of the defects as described above, suggesting that one allele of Apc in lung mesenchymal cells is sufficient for proper early lung development, which is consistent with the unaltered Ctnnb1 activation shown above.
The abnormal histology of the Apc CKO lung was further analyzed in H&E-stained tissue sections (Fig. 3). Although no significant difference was observed between E11.5 WT and Apc CKO lungs, alteration of the mesenchymal structure in Apc CKO lungs was obvious starting from E12.5, with circumferentially orientated and condensed mesenchymal cells around epithelial tubes (see Additional file 2 for high magnification). There were fewer airway epithelial tubes in Apc CKO lungs starting from E12.5 (Fig. 3). At E14.5, a massive hemorrhage with peripheral lung tissue destruction was observed in Apc CKO fetuses (Fig. 3 and Additional file 3).
As mentioned above, Apc is a major negative regulator for Wnt/Ctnnb1 signaling, and the abnormal lung development in Apc CKO fetuses could be mediated by increased Ctnnb1 activation, which was indeed detected in E12.5 Apc CKO lung mesenchyme (Fig. 1). To further determine the potential molecular mechanism, we performed a rescue experiment by generating Apc CKO in combination with heterozygous or homozygous Ctnnb1 knockout in developing lung mesenchyme. Surprisingly, reduction of Ctnnb1 gene dosage by single allele deletion could not rescue lung phenotypes in Apc CKO fetuses (Apc
fx/fx/Ctnnb1
fx/wt/Tbx4-rtTA/TetO-Cre in Fig. 4), while null deletion of Ctnnb1 alone (Apc
wt/wt/Ctnnb1
fx/fx/Tbx4-rtTA/TetO-Cre) resulted in reduced branching morphogenesis without change in mesenchymal cell density. Furthermore, null mutation of Ctnnb1 in combination with Apc CKO (Apc
fx/fx/Ctnnb1
fx/fx/Tbx4-rtTA/TetO-Cre) resulted in complicated lung growth arrest, but no mesenchymal condensation (Fig. 4), suggesting that abnormal mesenchymal condensation of early Apc CKO fetal lung may be mediated by hyperactivation of Ctnnb1 following Apc deletion.
Loss of Apc function in lung mesenchymal cells results in dynamic changes in cell proliferation that eventually lead to arrest of lung growth
Given that Apc is a negative regulator for Wnt/Ctnnb1 signaling, loss of Apc function could have an important impact on mesenchymal cell proliferation. Short-term 5-ethynyl-2-deoxyuridine (EdU) incorporation was used to identify cells with active DNA synthesis. Consistently with activation of Wnt/Ctnnb1 signaling (Fig. 1), increased 2-hour EdU labeling was detected in lung mesenchymal cells rather than epithelial cells at E11.5, one day after Apc knockout induction (Fig. 5a,5b). However, while both epithelial and mesenchymal cells in WT embryonic lungs retained a steady rate of cell proliferation shown by 2-hour EdU labeling at E12.5 and E13.5, cell proliferation in both epithelial and mesenchymal cells of Apc CKO lungs was then significantly reduced (Fig. 5a,5b). Interestingly, mouse embryos with a lung mesenchyme-specific activation of Ctnnb1 from E10.5 (Tbx4-Cre
ERT2/Cnntb1
floxed-ex3 or Ctnnb1 Δex3/+) had a sustained increase of lung mesenchymal cell proliferation, as detected by 2-hour EdU incorporation during early lung development (Fig. 5c,5d and Additional file 4), suggesting that subsequent reduction of cell proliferation in Apc CKO lungs after E12.5 is likely independent of aberrant Ctnnb1 activation.
To further understand the related mechanisms, we performed a pulse-chase experiment with EdU labeling at E11.5 and detection of the labeled DNA at E13.5. In WT lung, only a few cells were still positive for early EdU labeling due to continuous cell division and rounds of DNA synthesis (Fig. 6a). In contrast, the majority of lung mesenchymal cells in Apc CKO lungs were still EdU-positive. Interestingly, the sizes of mesenchymal cell nuclei in E13.5 Apc CKO lungs (30.0 ± 3.7 μm2) were significantly larger than those of WT littermate controls (23.8 ± 2.7 μm2, P < 0.05) and those of E11.5 Apc CKO lungs (23.0 ± 3.2 μm2, P < 0.05). These data suggested that the cells with Apc deletion were subsequently arrested in the cell cycle. Since interaction between Apc and the plus-end of microtubules is reported to be essential for spindle formation and chromatin segregation at metaphase in cultured cells [4], cell cycle analysis of Apc knockout lung mesenchyme was then performed in order to provide a potential mechanism for altered cell proliferation. Surprisingly, most Apc CKO lung mesenchymal cells were negative for phosphorylated Histone 3 (PH3), a marker of metaphase. Instead, from E12.5, they had more primary cilia as detected by acetylated α-tubulin staining (Fig. 6b and Additional file 5A), which is a post-mitotic cellular structure existing in G0/G1 phase [12, 13]. These findings indicate that lung mesenchymal cells with Apc deletion are arrested at G0/G1 phase rather than metaphase. To further verify this, single cell suspensions prepared from E13.5 lung were assessed for their DNA contents using propidium iodide staining and fluorescence-activated cell sorting (FACS) analysis (Fig. 6c). 83 % of Apc CKO lung mesenchymal cells were in G0/G1 phase, a significantly higher percentage than that in the WT controls (63 %), while the cells at S and G2/M phases were relatively lower in Apc CKO lung (13.9 % and 2.8 %) than those (26.7 % and 10.2 %) in the WT controls. Therefore, the significant reduction of cell proliferation seen in the Apc CKO lung after E12.5 was not due to metaphase arrest, but rather due to more cells exiting the cell cycle.
Additional studies were performed to determine the cellular mechanisms underlying lung growth arrest, including cell senescence and apoptosis. The dynamic expression of key genes involved in the cell cycle was measured using real-time PCR (Fig. 6d). Interestingly, c-Myc expression was increased at E11.5, not changed at E12.5, and decreased at E13.5. The increased c-Myc protein expression in E11.5 Apc CKO fetal lung mesenchymal cells was also verified by immunostaining (Additional file 5B). Moreover, the mRNA level of Ccnd1 (encoding cyclin D1) was significantly reduced from E12.5 to E13.5 in Apc CKO lungs. The changes of gene expression at E13.5 were further verified at the protein level by western blot (Fig. 6e). However, expression or activation of genes associated with cell senescence, Glb1 (encoding SA-β-galactosidase) and phospho-p53 [14, 15], were not altered (Fig. 6d,e). In addition, activation of the apoptotic pathway and number of apoptotic cells, evaluated by caspase 3 activation and nuclear DNA fragmentation (Fig. 6e and Additional file 6), were not changed between Apc CKO lung and WT controls. Therefore, with apoptosis and senescence ruled out, the impaired growth of embryonic lung in Apc CKO fetuses is likely due to G0/G1 cell cycle arrest.
Apc is essential for appropriate mesenchymal cell differentiation and some extracellular matrix production
During branching morphogenesis, mesenchymal cells surrounding proximal airways or vasculature, marked by epithelial Sox2 or endothelial PECAM1, differentiate into airway or vascular smooth muscle cells. However, abrogation of Apc significantly inhibited both proximal airway and vasculature smooth muscle cell differentiation, as detected by α-smooth muscle actin (SMA) staining in Fig. 7a. Interestingly, some mesenchymal cells in E13.5 peripheral lung of Apc CKO fetuses expressed Sox9, while only peripheral airway epithelial cells in the WT lung were positive for Sox9 staining at this stage (Fig. 7b). Although the exact identity of these Sox9-positive mesenchymal cells is unknown, commitment and differentiation of lung mesenchymal cell lineages in Apc CKO lung appear to be severely disrupted.
In addition to cellular changes, some extracellular matrix protein production and deposition were also altered in Apc CKO embryos after E12.5. One of these was versican (Vcan), a large chondroitin sulfate proteoglycan. Expression of Vcan in normal embryonic lung is restricted to a thin mesenchymal cell layer that surrounds large airways after E12.5 (Fig. 8a). However, abrogation of Apc in embryonic lung mesenchyme induced ubiquitous and high expression of Vcan, which overlapped with the pattern of excessive Ctnnb1 activation. Moreover, simultaneous abrogation of Ctnnb1 in Apc CKO lung mesenchyme blocked the abnormal expression of Vcan (Fig. 8b), suggesting that hyperactivation of the Wnt/Ctnnb1 pathway in Apc CKO lung mesenchyme is responsible for this phenotypic change. In order to determine the related molecular mechanism, the Vcan promoter DNA sequence was analyzed. At least two Ctnnb1/TCF consensus binding sites (5′-CTTTGAT-3′ or 5′-ATCAAAG) were identified in the Vcan promoter (−3197 to −3191 and −911 to −905, Fig. 8c). To further confirm the direct binding of Ctnnb1/TCF to the Vcan promoter, chromatin immunoprecipitation (ChIP) using anti-Ctnnb1 antibody was performed for E18.5 normal lung tissue. As shown in Fig. 8d, both Ctnnb1/TCF consensus binding sites mentioned above have been shown to interact with Ctnnb1/TCF specifically, indicating that excessive expression of Vcan is mediated by hyperactivation of the Wnt/Ctnnb1 pathway due to loss of Apc inhibitory function.
Abrogation of mesenchymal Apc function inhibited epithelial branching morphogenesis by a paracrine mechanism
Lung epithelial branching morphogenesis requires coordinated signaling between epithelial and mesenchymal cells. Mesenchymal Apc CKO resulted in smaller lung size at E13.5 (Fig. 2). The change in airway epithelial structure was visualized by whole mount E-cadherin immunofluorescence staining (Fig. 9a). Compared to WT lungs, elongation of the primary epithelial tubes and terminal bud sprouting were drastically inhibited in the Apc CKO lungs. In order to understand the molecular mechanisms by which altered mesenchymal Apc/Ctnnb1 activity disrupts epithelial branching, expression of genes encoding key epithelial growth factors was examined. A substantial decrease of Fgf10 expression (>5 fold) and a striking increase of Bmp4 expression (>20 fold) at the mRNA level were detected in Apc CKO lungs as early as E11.5, before morphological changes had occurred (Fig. 9b). Furthermore, whole-mount in situ hybridization showed that Fgf10 expression domains at the distal tips of the lung bud mesenchyme were reduced or even absent (Fig. 9c), compared to those in the WT controls. In contrast, pronounced Bmp4 expression was detected throughout the entire lung mesenchyme at E11.5, compared to the relatively restricted pattern of Bmp4 expression to the tips of epithelial cells in WT lung (Fig. 9c). Changes of Bmp4 and Fgf10 protein expression in the entire lung tissue at E13.5 were also confirmed by western blot (Fig. 9d). Since Bmp4 expression can be directly upregulated by Wnt/Ctnnb1 signaling in embryonic lung mesenchyme [16], we then wondered whether defective Fgf10 expression in the Apc CKO lung could be mediated by increased Bmp4. Therefore, the regulatory effect of Bmp4 on Fgf10 gene expression in cultured human fetal lung fibroblast line HLF1 was examined. High concentrations of BMP4 (50–100 ng/ml) were able to inhibit Fgf10 expression at both mRNA and protein levels (Fig. 9e,f).
Loss of mesenchymal Apc function also disrupts lung vasculogenesis and formation of the pulmonary circulation by a paracrine mechanism
Massive lung hemorrhage in Apc CKO fetuses occurred around E14.5 (Figs. 2 and 3), when the pulmonary circulation should have started. This suggests that mesenchymal Apc function is essential for regulating lung vasculogenesis and/or angiogenesis as well as for building intact pulmonary circulation networks. In order to determine the vascular continuity in Apc CKO lungs, fluorescein isothiocyanate (FITC)-labeled lectin, an endothelial tracer, was injected into the right cardiac ventricle of E13.5 live fetuses 5 min before lung harvest, and pulmonary vascular perfusion was visualized by FITC-lectin binding. In addition, all the vasculature within the same lung tissue section was labeled by PECAM1 staining (Fig. 10a). We found that the proximal large vessels in Apc CKO lungs were readily perfused (Lectin+/PECAM1+), while the distal small vessels were not (Lectin−/PECAM1+). In contrast, a plexus-like pattern of lectin-labeled vasculature perfectly matched the PECAM1-stained endothelial cells in the WT lungs, indicating thorough perfusion and a well-established pulmonary circulation at this stage. Therefore, lung mesenchymal Apc CKO fetuses had disrupted pulmonary circulation due to disconnection between proximal vessels and the distal vasculature, which could result in the massive pulmonary hemorrhage and loss of blood, and eventually fetal lethality.
To further understand these related mechanisms, we then investigated the dynamic changes of lung vascular formation from E12.5 to E14.5 in Apc CKO mice by detecting two different endothelial cells markers: Flk1 and PECAM1. Flk1 is present in both pre-mature and mature endothelial cells, while PECAM1 is only expressed in mature endothelial cells [17]. Thus, Flk1+/PECAM1− cells represent endothelial progenitors (angioblasts) and Flk1+/PECAM1+ cells are mature endothelial cells. At E12.5, although both WT and Apc CKO lungs had comparable patterns of Flk1 and PECAM1 protein immunostaining, with the majority of cells being Flk1+/PECAM1+, Flk1 expression at the mRNA level in Apc CKO lung was already reduced significantly (Fig. 10b,c). One day later (E13.5), Flk1+/PECAM1− cells in Apc CKO lung were decreased, accompanied by an overall reduction and simplification of the peripheral vascular network (Fig. 10c). Furthermore, by E14.5, a marked reduction of the entire vascular network was seen in Apc CKO lung, including reductions in both Flk1+/PECAM1+ and Flk1+/PECAM1− cells.
Studies have shown that the pulmonary circulation network may be formed by two coordinated mechanisms, proximal angiogenesis and distal vasculogenesis [2]. Our recent study using the Tbx4-rtTA/TetO-Cre/mT-mG reporter mice with Dox induction from E6.5 suggests that early lung mesenchymal progenitor cells give rise to the endothelial progenitor cells for vascular formation [11]. However, using the same driver/reporter line, we found that Dox induction from E10.5 did not mark both premature and mature endothelial cells with green fluorescent protein (GFP) (Additional file 7), and therefore, deletion of the Apc gene in our Apc CKO mice should not occur in these endothelial cells. This was further verified by no change of Ctnnb1 activation in Flk1+ cells of E12.5 Apc CKO lungs induced from E10.5 (Fig. 10d). Thus, the effect of mesenchymal Apc conditional knockout on vascular development was indirect, possibly through a paracrine mechanism. We then screened for changes of key growth factors that are important in vasculogenesis. Surprisingly, there was no difference in Vegf-a expression between WT and Apc CKO lungs (Fig. 10e). However, expression of Igf1 and Angpt1 at the mRNA level was markedly reduced in E12.5 Apc CKO lungs, suggesting that deregulation of multiple growth factors may disrupt pulmonary vascular formation in this model system.