Wnt/β-catenin signaling accelerates mouse lung tumorigenesis by imposing an embryonic distal progenitor phenotype on lung epithelium (original) (raw)
Wnt/β-catenin activation in mouse lung Clara cells is not sufficient to induce tumors but promotes carcinogenesis induced by the KrasG12D oncogene. Previous studies have shown that activation of Wnt/β-catenin signaling in bronchiolar epithelium of the adult mouse lung does not lead to increased tumor formation or increased cell proliferation (12). However, additional studies have demonstrated a link between increased Wnt/β-catenin signaling and lung tumor metastasis and cell proliferation in human lung tumor cell lines (15–17). To address this apparent discrepancy, we induced expression of an activated form of β-catenin by itself or in combination with the well-characterized KrasG12D oncogenic form of Kras, which has been demonstrated to promote lung tumorigenesis (18, 19), specifically in postnatal bronchiolar epithelium of the lung. The Ctnnb1ex3flox allele of β-catenin has loxP sites surrounding exon 3, which contains the regulatory phosphorylation sites that control β-catenin protein stability. Cre-mediated loss of exon 3 leads to increased β-catenin accumulation and increased Wnt signaling (20). In these experiments, we used a previously described CC10-Cre line that expresses Cre recombinase exclusively in postnatal Clara cells in the lung (21). This Cre line exhibits an intermediate rate of recombination efficiency, with 30%–50% of Clara cells exhibiting recombination using the R26R-lacZ reporter (data not shown and ref. 21).
In agreement with previous studies (12), expression of the Ctnnb1ex3flox mutant in Clara cells did not result in an increase in lung tumor development or proliferation up to 18 months of age (Figure 1, A–C, and data not shown). After 3 months, expression of the KrasG12D mutant in Clara cells resulted in extensive formation of hyperplastic regions in the adult lung, many of which were located in the bronchioalveolar duct junction (BADJ) region (Figure 1, A, E, I, and M). Some of these had developed into adenomas by 3 months (Figure 1, I and M). In contrast, expression of both Ctnnb1ex3flox and KrasG12D for 3–4 months resulted in a significant increase in both the number of lung tumors and lung tumor size (Figure 1, A and B). This resulted in increased net weight of the lungs (Figure 1C). To independently verify the changes in tumor number, size, and stage, we performed a blinded analysis with a pathologist who graded histological samples without knowing the genotype using previously published criteria (22). These results show that CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutant tumors represented a more aggressive and advanced stage of development including the presence of adenocarcinomas, whereas expression of Ctnnb1ex3flox did not produce any tumors, and expression of KrasG12D generated primarily hyperplastic regions and a small number of focal adenomas after 3–4 months (Table 1). Of note, most of the CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutants died before 6 months of age due to respiratory distress (data not shown). Tumors from both the CC10-Cre:KrasG12D and CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutants were lung epithelial in origin, as indicated by expression of the transcription factor Nkx2.1 (Figure 1, D–F). Tumors in the CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutants also expressed higher mucus levels, as indicated by increased PAS staining, which correlates with progression to the adenocarcinoma stage, in comparison to the CC10-Cre:KrasG12D single mutants (Figure 1, O–R). Expression of the mesenchymal marker vimentin was not changed between controls and the CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutants, suggesting no increase in epithelial-mesenchymal transition caused by coexpression of oncogenic Kras and activated β-catenin (Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI44871DS1). Together these data indicate that coexpression of the Ctnnb1ex3flox and KrasG12D oncogenes leads to an acceleration of lung tumorigenesis, which differs significantly from expression of Ctnnb1ex3flox or KrasG12D alone.
Coactivation of KrasG12D and Ctnnb1ex3flox leads to a synergistic increase in tumor development in the mouse lung. (A) Whole mount picture of CC10-Cre:KrasG12D:Ctnnb1ex3flox, CC10-Cre: KrasG12D, and wild-type lungs at 3 months of age. (B) Tumor numbers in lungs of mice of the indicated genotypes at 3 months of age. (C) Left lung weight from animals of the indicated genotypes at 3 months of age. Number of animals assessed in B and C is indicated in the graphs. Data in B and C are presented as mean ± SEM. (D–F) Nkx2.1 immunostaining of lung and tumor samples of animals of the indicated genotypes at 3 months of age. (G–N) H&E staining of histology sections from animals of the indicated genotypes at 3 months of age. Note that K–N show higher-magnification pictures of the same samples shown in G–J. (O–R) PAS staining of histological sections from animals of the indicated genotypes at 3 months of age. Scale bars: D–F, 100 μm; G–J, 400 μm; K–R, 200 μm.
Summary of developmental genes altered in CC10-Cre:KrasG12D:Ctnnb1ex3flox tumor epithelium
Coexpression of Ctnnb1ex3flox and KrasG12D alters the phenotype of bronchiolar epithelial cells of the lung. To further assess the phenotype of the lung tumors resulting from coexpression of Ctnnb1ex3flox and KrasG12D oncogenes, we performed immunostaining for the lung epithelial markers SP-C, SP-B, CC10, and Hopx, which represent different lung epithelial lineages and stages of development. SP-C and SP-B are surfactant proteins expressed primarily in alveolar epithelial cells in the adult lung. In contrast, CC10 expression is confined to Clara cells found in the airways of the adult mouse lung, and recombination mediated by the CC10-Cre line is restricted to these cells in the lung (21). Hopx is a transcriptional regulator that is expressed in lung epithelium after E13.5 and has recently been shown to act as a tumor suppressor in the lung (23, 24). Since we expressed both the KrasG12D and the Ctnnb1ex3flox oncogenes in Clara cells of the postnatal lung, we predicted that the resulting tumors would represent a bronchiolar airway phenotype. Surprisingly, tumors in CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutant animals expressed high levels of SP-C and did not express CC10 or Hopx (Figure 2, C, F, and I). This is in contrast to CC10-Cre:LSL-KrasG12D single mutant tumors, which expressed a mix of SP-C– and CC10-positive cells and were Hopx positive (Figure 2, B, E, and H). Quantitative PCR (Q-PCR) confirmed the changes in SP-C and CC10 gene expression (Figure 2M). These data suggested that tumors in CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutants represented a more immature epithelial phenotype similar to early distal or alveolar epithelium. Loss of Hopx may also contribute to the increased tumor development in Ctnnb1ex3flox:LSL-KrasG12D double mutants given its role as a tumor suppressor. Sca1 is also expressed in lung tumors from LSL-KrasG12D mutants (18). Sca1/CC10 double immunostaining showed that although the tumors in CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutants did not express CC10, they did express Sca1 (Figure 2, J–L). Recent data have demonstrated that the bronchiolar epithelium represents a separate and distinct lineage that does not generate alveolar epithelium during homeostasis or after lung injury (25). Thus, these data suggest that coexpression of Ctnnb1ex3flox and KrasG12D in Clara cells of the postnatal lung results in a cell lineage switch from a bronchiolar lineage to an immature alveolar or distal epithelial lineage that resembles developing lung epithelium prior to E13.5.
Bronchiolar and alveolar marker gene expression in lungs expressing KrasG12D and Ctnnb1ex3flox oncogenes. SP-C (A–C), CC10 (D–F), and Sca1/CC10 (J–L) immunostaining of lungs from animals of the indicated genotypes at 3 months of age. (G–I) In situ hybridization for expression of Hopx in lungs from animals of the indicated genotypes at 3 months of age. (M) Q-PCR for expression of SP-C and Hopx in parallel lung cDNA samples from animals of the indicated genotypes at 3 months of age. Three individual animals of the indicated genotypes were used for the Q-PCR analysis shown in M, and the data are presented as mean ± SEM. Scale bar: 100 μm.
Tumors observed in CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D mutants originate from Clara cells. Given the difference in tumor phenotype in CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutants compared with their presumed cell of origin, we wanted to determine whether these tumors did indeed derive from Clara cells of the postnatal lung. Using the Rosa26R lacZ indicator line, we performed lacZ histochemical staining as well as β-galactosidase immunostaining of control, CC10-Cre:LSL-KrasG12D:R26R, and CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D:R26R lungs. The data showed that the tumors in both CC10-Cre:LSL-KrasG12D:R26R and CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D:R26R mutants were derived from CC10-expressing Clara cells (Figure 3, A–F). Expression of SP-C on adjacent slides confirmed that the resulting tumors had altered their phenotype to a distal epithelial or immature alveolar lineage (Figure 3, G–I). Since CC10-expressing cells do not generate alveolar or distal epithelial cell types in the lung in lineage tracing experiments (25), these data are consistent with a cell lineage switch from CC10-positive bronchiolar epithelium to immature alveolar epithelium in CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D mutants. Alternatively, the previously identified bronchioalveolar stem cells (BASCs), which have been suggested to act as cancer stem cells, could also be the cancer-initiating cell of origin in these studies since they express CC10 (18).
Coactivation of KrasG12D and Ctnnb1ex3flox leads to an imposition of an embryonic distal progenitor phenotype on tumor epithelium derived from Clara cells of the adult lung. (A–C) β-Galactosidase staining of CC10-Cre:R26RlacZ fated cells showing that the tumors in CC10-Cre:KrasG12D and CC10-Cre:KrasG12D:Ctnnb1ex3flox animals at 3 months of age are derived from Clara cells. (D–F) Immunostaining of epithelium of CC10-Cre, CC10-Cre:KrasG12D, and CC10-Cre:KrasG12D:Ctnnb1ex3flox mice for β-galactosidase expression at 3 months of age. (G–I) Immunostaining of epithelium of CC10-Cre, CC10-Cre:KrasG12D, and CC10-Cre:KrasG12D:Ctnnb1ex3flox lungs for SP-C expression. Assessment of Sox9 (J–L), Sox2 (M–O), Gata6 (P–R), Wnt7b (S–U), and Id2 (V–X) expression using immunostaining (J–O and V–X) or in situ hybridization (P–U) in animals of the indicated genotypes at 3 months of age. (Y) Q-PCR of these same genes. Three individual animals of the indicated genotype were used for Q-PCR analysis shown in Y, and data are presented as mean ± SEM. Changes in gene expression for Sox9 (P < 0.001), Sox2 (P < 0.05), Gata6 (P < 0.07), Wnt7b (P < 0.005), and Id2 (P < 0.004) were considered significant as indicated. Scale bars: A–C, 400 μm; D–X, 100 μm.
Coexpression of Ctnnb1ex3flox and LSL-KrasG12D alleles leads to increased expression of markers of distal embryonic lung epithelium. To further determine the developmental state of the tumor epithelium generated by coactivation of KrasG12D and Ctnnb1ex3flox, we performed immunostaining and in situ hybridization of multiple markers of both distal and proximal progenitors found in the embryonic lung. Tumors from CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutants expressed high levels of Sox9, Wnt7b, and Id2, which are expressed exclusively in embryonic distal endoderm progenitors within the developing lung (Figure 3, J–L and S–X, and refs. 26–31). CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutant tumors also expressed high levels of Gata6, which is expressed at higher levels in distal progenitors than in more proximal progenitors during lung development (Figure 3, P–R, and refs. 32–35). However, Ctnnb1ex3flox:LSL-KrasG12D double mutant tumors did not express appreciable levels of Sox2, which is a marker of proximal epithelial progenitors in both the developing and adult lung (Figure 3, M–O, and refs. 28, 36, 37). These changes in expression were confirmed by Q-PCR (Figure 3Y). In a direct comparison, CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D tumors were more similar to E12.5 lungs than adult lungs in their expression of Sox9, Wnt7b, Gata6, Hopx, CC10, and Sox2 (Figure 4, A–F). The increased levels of Sox9 and decreased levels of Sox2 in CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D mutants in comparison to E12.5 lung buds further supports a phenotypic change to a distal epithelial progenitor phenotype in these tumors.
Comparison of CC10-Cre:KrasG12D and CC10-Cre:KrasG12D:Ctnnb1ex3flox tumor epithelium with embryonic lung epithelium and identification of molecular pathways altered in CC10-Cre:KrasG12D:Ctnnb1ex3flox lung epithelium. A comparison of CC10 (A), Sox2 (B), Hopx (C), Sox9 (D), Wnt7b (E), and Gata6 (F) expression in WT (CC10-Cre), CC10-Cre:KrasG12D, and CC10-Cre:KrasG12D:Ctnnb1ex3flox tumor epithelium. Three individual animals of the indicated genotypes were used for Q-PCR analysis shown in A–F, and data are presented as mean ± SEM. Changes in gene expression for CC10 (P < 0.001), Sox2 (P < 0.01), Hopx (P < 0.003), Sox9 (P < 0.005), Wnt7b (P < 0.001), and Gata6 (P < 0.04) were considered significant as indicated. (G) Microarray analysis showing that 673 genes were uniquely altered in CC10-Cre:KrasG12D:Ctnnb1ex3flox tumor epithelium. (H) Biological process enrichment graph showing molecular pathways altered in CC10-Cre:KrasG12D:Ctnnb1ex3flox tumor epithelium.
Coactivation of Wnt/β-catenin and oncogenic Kras signaling results in a distinct transcriptional profile representing an embryonic distal epithelial progenitor. The above data suggested that coactivation of oncogenic Kras and Wnt/β-catenin signaling leads to a transdifferentiation of adult Clara cells to an embryonic distal progenitor phenotype. To further address this hypothesis, we performed microarray analysis of control, CC10-Cre:KrasG12D, and CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutant microdissected tumors to determine their phenotype. From these studies we identified more than 600 genes whose deregulation was unique to the coexpression of Ctnnb1ex3flox and LSL-KrasG12D alleles (Figure 4G and Supplemental Table 1). Of these genes, several important molecular pathways were specifically affected in CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutant tumors. As expected, GTP-mediated signal transduction and cell proliferation were significantly affected (Figure 4H). This analysis also showed increased enrichment of factors expressed during lung respiratory system development, during divalent cation transport, and in cell-cell adhesion (Figure 4H). As predicted from the data above, important developmental regulators such as Wnt7b, Id2, Sox9, and Mycn were all increased, whereas Hopx was decreased, in CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D tumors (Table 1).
To determine whether such embryonic regulators were also upregulated in human lung tumor samples, we performed Sox9 immunostaining on a human lung tumor array. This array consisted of histological sections from various phenotypically distinct tumor types including squamous cell carcinoma, small cell undifferentiated carcinoma, bronchioalveolar carcinoma, and adenocarcinoma. The majority (62%) of adenocarcinomas were positive for Sox9 immunostaining, while normal lung tissue contained only rare Sox9-positive cells (Figure 5, A and B). Of note, Sox9 expression has been associated with prostate tumor development and basal cell carcinoma and is positively regulated by the Wnt/β-catenin pathway (38–40). To determine whether these same tumor samples also expressed elevated levels of Wnt/β-catenin signaling, we immunostained the same human lung tumor array to detect expression of axin2, a known target of Wnt signaling that can be used to measure signaling activity (41). Overall, 42% of all adenocarcinomas were positive for axin2 expression (Figure 5E). Moreover, 52% of the Sox9-positive samples were axin2 positive, indicating a substantial correlation between Sox9 and increased Wnt/β-catenin signaling activity (Figure 5E).
Sox9 expression in human lung adenocarcinomas. The human lung tumor array was immunostained for Sox9 and axin2 expression as described in Methods and scored by 3 independent researchers. Normal human lung contains only rare Sox9-positive cells (A) and almost no axin2-immunostaining cells (C). (B, D, and E) 62% of adenocarcinoma samples had high levels of Sox9, while 42% of adenocarcinoma samples were positive for axin2 immunoreactivity. The percentage of Sox9-positive samples that were axin2 positive was 52% (E). The samples used for Sox9 and Axin2 immunostaining were from the same normal lung and tumor samples. (F) LiCl and Wnt3a induce SOX9 and ID2 expression in A549 and H441 cells but not H552 and H1299 cells. Untreated cells served as controls in these assays. Original magnification, ×100.
To assess whether increased Wnt/β-catenin signaling along with Kras signaling would increase expression of the distal progenitor marker genes SOX9 and ID2 in human lung cancer cell lines, we activated Wnt/β-catenin signaling in two human lung tumor cell lines that harbor mutant forms of Kras, A549 and H441, and two cell lines that harbor wild-type Kras alleles, H552 and H1299 (42–44). Untreated cells served as negative controls in these assays. Activation of Wnt/β-catenin signaling by Wnt3a and LiCl, a known agonist of glycogen synthase kinase 3β (GSK-3β) and activator of Wnt signaling (45), resulted in increased expression of both Sox9 and Id2 in A549 and H441 cells but not H552 or H1299 cells (Figure 5D). Together, these data support the imposition of a distal embryonic lung progenitor phenotype by coactivation of Wnt/β-catenin and Kras signaling in both mouse and human lung epithelium.
Ctnnb1ex3flox:LSL-KrasG12D double mutant tumors exhibit decreased E-cadherin expression. A recent study showed that activation of Wnt/β-catenin signaling in human lung tumor lines leads to increased metastatic potential (15). Disruption in cell-cell adhesion can lead to increased metastasis in cancer, and E-cadherin plays an important role in regulating this process in lung cancer (46–50). Moreover, alterations in β-catenin levels that occur upon Wnt activation can lead to changes in cadherin-catenin interactions and a loss of cadherin stability (46, 47, 51, 52). Therefore, we wanted to determine whether the integrity of cell-cell interactions, in particular E-cadherin–related complexes, was disrupted in CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutant tumors. We used confocal microscopy to assess the expression of both E-cadherin and β-catenin in wild-type, CC10-Cre:Ctnnb1ex3flox, CC10-Cre:LSL-KrasG12D, and CC10-Cre:Ctnnb1ex3flox:LSL-KrasG12D double mutant tumors. Although β-catenin was still present in the membrane of the compound mutant cells, expression of E-cadherin was markedly reduced at the cell surface as well as throughout the rest of the mutant epithelium (Figure 6 and Supplemental Figure 2). E-cadherin and β-catenin colocalization was also reduced in Ctnnb1ex3flox:LSL-KrasG12D double mutant tumors (Figure 6 and Supplemental Figure 2). These data suggest that loss of E-cadherin may underlie part of the increased metastatic potential of lung tumor cell lines with increased Wnt/β-catenin signaling activation as previously reported, which is consistent with the function of Wnt/β-catenin in other cancers such as pancreatic cancer (46–50). Despite the decreased E-cadherin expression in the Ctnnb1ex3flox:LSL-KrasG12D double mutant tumors, we did not observe metastasis in these animals, which is likely explained by the premature demise of the majority of the compound mutants by 4 months of age (data not shown).
Loss of E-cadherin expression in CC10-Cre:KrasG12D:Ctnnb1ex3flox tumor epithelium. (A–D) Confocal microscopy showing loss of E-cadherin expression by immunostaining and _z_-axis scanning of confocal data. (E) E-cadherin expression quantified by relative intensity. Relative intensity changes were only significant for CC10-Cre:KrasG12D:Ctnnb1ex3flox tumor epithelium (P < 0.01). Scale bar: 50 μm.