Identification of cancer initiating cells in K-Ras driven lung adenocarcinoma

Significance

K-RAS oncogene-driven lung adenocarcinomas is one of the most malignant human tumors for which there are no efficacious therapeutic strategies. Here, we have used a mouse tumor model that closely recapitulates this human disease to illustrate that adult lung cells are uniquely sensitive to transformation by this oncogene. Monitoring lung cells at the single-cell level revealed that they respond differently to K-Ras oncogenic signals. Whereas K-Ras–expressing Clara cells required an inflammatory response to yield hyperplasias and adenomas, alveolar type II cells or their committed precursors led to the generation of malignant adenocarcinoma regardless of their surrounding microenvironment.

Abstract

Ubiquitous expression of a resident K-Ras^G12V oncogene in adult mice revealed that most tissues are resistant to K-Ras oncogenic signals. Indeed, K-Ras^G12V expression only induced overt tumors in lungs. To identify these transformation-permissive cells, we induced K-Ras^G12V expression in a very limited number of adult lung cells (0.2%) and monitored their fate by X-Gal staining, a surrogate marker coexpressed with the K-Ras^G12V oncoprotein. Four weeks later, 30% of these cells had proliferated to form small clusters. However, only SPC⁺ alveolar type II (ATII) cells were able to form hyperplastic lesions, some of which progressed to adenomas and adenocarcinomas. In contrast, induction of K-Ras^G12V expression in lung cells by intratracheal infection with adenoviral-Cre particles generated hyperplasias in all regions except the proximal airways. Bronchiolar and bronchioalveolar duct junction hyperplasias were primarily made of CC10⁺ Clara cells. Some of them progressed to form benign adenomas. However, only alveolar hyperplasias, exclusively made up of SPC⁺ ATII cells, progressed to yield malignant adenocarcinomas. Adenoviral infection induced inflammatory infiltrates primarily made of T and B cells. This inflammatory response was essential for the development of K-Ras^G12V–driven bronchiolar hyperplasias and adenomas, but not for the generation of SPC⁺ ATII lesions. Finally, activation of K-Ras^G12V during embryonic development under the control of a Sca1 promoter yielded CC10⁺, but not SPC⁺, hyperplasias, and adenomas. These results, taken together, illustrate that different types of lung cells can generate benign lesions in response to K-Ras oncogenic signals. However, in adult mice, only SPC⁺ ATII cells were able to yield malignant adenocarcinomas.

The K-RAS oncogene is frequently activated in some of the most aggressive human tumor types including lung carcinomas (25% incidence), colorectal carcinomas (40% incidence), pancreatic ductal adenocarcinomas (90% incidence), and endometrial carcinomas (15% incidence) (1). Other tumor types also contain K-RAS oncogenes, albeit with lower incidence (1). Accumulating evidence suggests that K-RAS activation might be one of the key initiating events in this tumor type, hence the recent interest in identifying the cell type(s) susceptible to K-RAS–driven transformation. Most studies described thus far have used a genetically engineered mouse (GEM) model that carried a knocked-in K-Ras^G12D allele whose expression can be activated by various Cre-dependent strategies (2). Using this model, Kim et al. first identified stem cells, designated as BASCs and located at the bronchioalveolar duct junctions (BADJ), as the cancer initiating cells (3). However, subsequent studies using tool strains that expressed the Cre recombinase under the control of cell type-specific promoters have identified the cancer-initiating cells as alveolar type II cells (ATII), a main component of the alveoli responsible for the production and secretion of surfactant molecules (4, 5). Other investigators using a similar experimental approach have concluded that the cancer initiating cells are not ATII but Clara cells, the main cell type that lines the bronchiolar epithelium (6).

In this study, we have used a K-Ras–driven GEM tumor model that upon Cre-mediated recombination coexpresses a resident K-Ras^G12V oncoprotein along with a bacterial β-Geo protein that serves as a surrogate marker (7). This strategy permits the identification of cells expressing the K-Ras^G12V oncogene at the single-cell level, thus allowing us to monitor the earliest stages of unscheduled proliferation in the lung without biasing expression of the resident K-Ras oncogene in any particular cell type. Our results indicate that although different lung cell types can become transformed and yield hyperplasias and benign adenomas, only SPC⁺ ATII cells are able to yield malignant adenocarcinomas.

Results

Most Adult Mouse Cells Are Resistant to Transformation by Endogenous K-Ras Oncogenes.

To determine the consequences of unbiased expression of a resident K-Ras oncogene in adult mouse tissues, we exposed young K-Ras^+/LSLG12Vgeo;RERT^ert/ert mice to 4-hydroxy-tamoxifen (4OHT) for 24 wk to activate the CreERT2 recombinase knocked-in within the ert alleles. As illustrated in Fig. 1A, this treatment led to widespread expression of the resident K-Ras^G12V oncogene along with its surrogate β-Geo marker (7). Rosa26^+/LSLLacZ;RERT^ert/ert mice were used as controls. Whereas in some tissues, such as colon and testis, most of their cells expressed K-Ras^G12V, other tissues, including kidney, liver, and lung, displayed a more limited pattern of expression (Fig. 1).

Fig. 1.

Widespread expression of an endogenous K-Ras^G12V oncogene in adult mice only leads to tumor development in lung tissue. (A) K-Ras^+/LSLG12Vgeo;RERT^ert/ert mice and control Rosa26^+/LacZ;RERT^ert/ert animals were treated at weaning with 4OHT (0.5 mg, three injections per week) for 24 wk. Mice were killed at the end of the treatment (open bars) or 8 wk later (filled bars), and their tissues were submitted to FACS analysis after incubation with FDG. Analyzed tissues included bone marrow (BM), colon (CO), kidney (KI), liver (LI), lung (LU), pancreas (PA), skin (SK), spleen (SP), testis (TE), and thymus (TH). Results are represented as percentage of FDG⁺ cells in tissues of K-Ras^+/LSLG12Vgeo;RERT^ert/ert mice respect to those of Rosa26^+/LacZ;RERT^ert/ert control animals. (B) Representative images of X-Gal–stained sections from the indicated tissues obtained from K-Ras^+/LSLG12Vgeo;RERT^ert/ert mice exposed to 4OHT. Sections were counterstained with Nuclear Fast Red. (Scale bars: 50 μm.)

When these mice were analyzed at 1 yr of age, we only found bona fide tumors in lung tissue, despite the limited frequency of K-Ras^G12V–expressing cells (Fig. 1). These tumors were characterized as adenocarcinomas. Other tissues, mainly the gastric epithelium, displayed premalignant lesions including papillomas, but none of them developed into overt tumors, at least by 1 yr of age. In agreement with previous studies, cells of the intestinal track and the pancreas were completely resistant to transformation by the resident K-Ras^G12V oncogene (8, 9). These observations indicate that lung cells possess unique properties that make them particularly permissive to transformation by this oncogene.

Expression of a Resident K-Ras Oncogene Induces Unscheduled Proliferation of Lung Cells.

To identify those lung cells permissive for K-Ras transformation and to shed light on the earliest steps of lung tumor development, we treated K-Ras^{lox/LSLG12Vgeo};RERT^ert/ert animals with a single, limiting dose of 4OHT (1 mg per mouse) to activate expression of the K-Ras^G12V oncoprotein in a limited number of cells so we could monitor their individual fate before they yielded anatomically identifiable lesions. We introduced the conditional K-Ras^lox allele to accelerate tumor development. No other differences were observed between K-Ras^+/LSLG12Vgeo;RERT^ert/ert and K-Ras^{lox/LSLG12Vgeo};RERT^ert/ert mice (10). Under these experimental conditions, approximately 0.2% of all lung cells became positive for X-Gal staining just 1 wk after 4OHT treatment. At this time, most of these X-Gal–positive cells (>98%) appeared as individual cells, indicating that they had not undergone cell division (Fig. 2). However, these cells were endowed with proliferative properties because their numbers increased linearly during the weeks ensuing the initial 4OHT treatment (Fig. S1A).

Fig. 2.

Selective expansion of K-Ras^G12V-expressing cells in the alveolar region. K-Ras^{lox/LSLG12Vgeo};RERT^ert/ert mice were treated with a single 4OHT injection (1 mg) at P21 and killed at the indicated time points after 4OHT exposure. Lungs were submitted to whole-mount X-Gal staining. Sections were counterstained with Nuclear Fast Red. Areas marked by a rectangle are amplified in the upper right corner for better vision. (Scale bar: 200 μm.)

When mice were killed 2 wk after 4OHT injection, approximately 30% of the X-Gal positive cells appeared in small clusters of 4–8 cells (Fig. 2). Groups of two cells were not counted because of their high variability. All cell counts were scored in 2D sections and, hence, represent an underestimate of the actual number of cells in each cluster. These small clusters were uniformly distributed throughout the four main structures of the lung including bronchi, bronchioles, BADJs, and alveoli, indicating that Cre-mediated recombination occurred in an unbiased spatial fashion (Fig. 2). As a control, X-Gal–positive cells from Rosa26^{LSLLacZ/LSLLacZ};RERT^ert/ert mice treated with the same dose of 4OHT exhibited a similar distribution but failed to expand (Fig. S2). These observations illustrate that a significant fraction of adult lung cells, regardless of their location, have the capacity to initiate a proliferative response upon expression of an endogenous K-Ras^G12V oncoprotein.

Expansion of K-Ras^G12V-Expressing Cells Only Occurs in the Alveolar Region.

When mice were killed 4 wk after 4OHT exposure, most X-Gal–positive cells appeared in clusters. Approximately half of them contained 4–20 cells, whereas the rest had >20 cells, including few clusters with as many as 100 cells per section (Fig. 2 and Fig. S1B). Small clusters (4–20 cells) were uniformly distributed among the main lung structures. However, those containing >20 cells were almost exclusively located in the alveolar region (Fig. 2 and Fig. S1C). When mice were examined 12 wk after 4OHT exposure, none of the clusters located in the large bronchi, the distal bronchiolar area, or the BADJ region had expanded beyond 20 cells per section. In contrast, many of the X-Gal–positive clusters located in the alveoli had grown into hyperplastic areas characterized by cells with uniform nuclei that resemble the atypical adenomatous hyperplasias (AAH) observed in human patients (11, 12). These lesions covered as much as 8% of the total lung area (Fig. 2 and Fig. S1A). We also observed some well-defined adenomas characterized by enlarged nuclei along with a compact and round shape (11) (Fig. S1D). At 20 wk after 4OHT treatment, some of these adenomas had progressed into more malignant lesions characterized by pleomorphic nuclei with nuclear atypia, abnormal mitotic figures, and multinucleated giant cells (11). Finally, at 24 wk after 4OHT treatment, the last time point used in this study, we observed a significant increase in the number of high-grade adenocarcinomas reaching an average of four tumors per mouse (Fig. S1D). At this time, the small clusters of X-Gal–positive cells present in other areas of the lung remained unaltered. These observations indicate that whereas a large fraction of lung cells can be enticed to divide a few times upon expression of the K-Ras^G12V oncoprotein, only those cells located in the alveolar region maintained a sustained proliferative response.

Proliferating Cells Display Markers of Alveolar Type II Cells.

To identify those cells that responded to K-Ras^G12V, we screened lung sections for the expression of specific markers. At 4 wk after 4OHT treatment, small clusters in the epithelium lining the bronchioles and in the BADJs stained for CC10, a protein also known as Scgb1a/uteroglobin that serves as a marker for Clara cells (Fig. S3 A–D) (13). All X-Gal–positive clusters containing >20 cells stained for the surfactant protein C (SPC/Sftpc), a marker for ATII cells, and were located in the alveolar region (Fig. 3A and Fig. S3A) (13). None of these larger clusters were positive for CC10 or for aquaporin 5 (AQP5), a marker of alveolar type I cells (ATI) (Fig. 3B) (13). Some bronchiolar and BADJ clusters contained cells positive for CC10 and SPC antibodies, suggesting that they might correspond to BASC progenitor cells (Fig. S3 A and D) (3). Finally, none of the X-Gal–positive cells stained with CGRP antibodies, a marker for neuroendocrine (NE) cells (Fig. S3E). At 8 wk after 4OHT treatment, all AAH lesions as well as all adenomas were uniformly positive for SPC staining (Fig. S4). Likewise, all high-grade adenocarcinomas observed at later time points were also exclusively SPC⁺. These observations suggest that advanced tumor lesions originate from those proliferating clusters of SPC⁺ ATII cells located in the alveolar region.

Fig. 3.

Clusters of proliferating K-Ras^G12V-expressing cells are SPC⁺. K-Ras^{lox/LSLG12Vgeo};RERT^ert/ert mice were treated with a single injection of 4OHT (1 mg) at P21 and killed 4 wk later. Representative sections of X-Gal–stained lungs were incubated with antibodies against SPC and CC10 (A) and SPC and AQP5 (B). Merged images represent the overlay of X-Gal staining and IF images. Nuclei were counterstained with DAPI. (Scale bars: 20 μm.)

Activation of K-Ras^G12V Expression with Adenovial Cre Vectors Induces Bronchiolar Hyperplasias.

We next activated expression of the resident K-Ras^G12V oncogene by intratracheal infection of K-Ras^{lox/LSLG12Vgeo} mice with adenoviral vectors expressing the Cre recombinase (Ad-Cre). Four weeks after infection, we observed small clusters of cells expressing the K-Ras^G12V oncoprotein, as determined by X-Gal staining primarily located in the bronchial and bronchiolar epithelium and in the BADJs. However, we also observed some small X-Gal–positive clusters in the alveolar region, indicating that some Ad-Cre particles also reached this region (Fig. S5A). Most of these alveolar clusters (97%) only stained with SPC antibodies. Likewise, most clusters present in the bronchioles were exclusively made up of CC10⁺ cells. However, a percentage of these clusters (6–8%) contained some SPC⁺ and double CC10⁺/SPC⁺ cells (Fig. 4A). Similar results were observed in the BADJs (Fig. 4).

Fig. 4.

Clusters of K-Ras^G12V-expressing CC10⁺ cells in bronchiolar and BADJ areas contain SPC⁺ and CC10⁺/SPC⁺ cells. K-Ras^{lox/LSLG12Vgeo};RERT^ert/ert mice were infected with Ad-Cre particles at 8–10 wk of age and killed 4 wk later. (A) Representative X-Gal–stained section depicting a bronchiolar area was incubated with SPC and CC10 antibodies. The enlarged merged image depicts the overlay of SPC and CC10 staining to better illustrate the presence of (arrows) SPC⁺ and (arrowheads) CC10⁺/SPC⁺ cells. (B) Representative X-Gal–stained section depicting a BADJ was incubated with SPC and CC10 antibodies. The merged image depicts the overlay of SPC and CC10 staining to better illustrate the presence of SPC⁺ (arrows) and CC10⁺/SPC⁺ cells (arrowheads). (Scale bar: 20 μm.)

Unlike in mice exposed to 4OHT, some of the clusters located in the bronchiolar and BADJ areas continued proliferating beyond the first rounds of unscheduled cell division. Indeed, 24 wk after infection, Ad-Cre–infected mice displayed papillary hyperplasias in these regions (Fig. 5A and Fig. S5B). No hyperplasias were observed in the proximal bronchi. All lesions were positive for X-Gal staining, indicating that they expressed the K-Ras^G12V oncoprotein (Fig. 5A and Fig. S5B). Although these bronchiolar papillary hyperplasias were mainly made up of Clara cells as determined by CC10 staining, some of them contained SPC⁺ cells (Fig. 5A). Small clusters of double CC10⁺/SPC⁺ were also observed (Fig. 5A). Similar results were observed in the BADJs. Finally, we also observed hyperplasias in the alveolar region made up of SPC⁺ cells and a characteristic AAH morphology (Fig. S5B).

Fig. 5.

Induction of bronchiolar hyperplasias required an adenovirus-induced inflammatory response. (A) Activation of K-Ras^G12V via intratracheal infection with Ad-Cre particles induces bronchiolar hyperplasias. (Left) Representative sections of X-Gal whole mount-stained lungs of K-Ras^{lox/LSLG12Vgeo} mice killed 4 wk (Upper) and 24 wk (Lower) after Ad-Cre infection depicting bronchiolar hyperplasias. (Center) IHC analysis of SPC (Upper) and CC10 (Lower) expression. Images show a bronchus with bronchiolar branches depicting Clara cell hyperplasia. Images also show hyperplastic areas of SPC⁺ cells in the terminal bronchiolar region. (Right) IF analysis using SPC⁺ and CC10⁺ antibodies. The merged image depicting the overlay between SPC and CC10 signals revealed the presence of double CC10⁺/SPC⁺ cells (Inset). Nuclei were counterstained with DAPI. (B) Infection of 4OHT-treated K-Ras^{lox/LSLG12Vgeo} mice with control Ad-GFP particles also developed bronchiolar hyperplasias. (Left) Representative sections of X-Gal whole mount-stained lungs of 4OHT-treated K-Ras^{lox/LSLG12Vgeo} mice killed 4 wk (Upper) and 24 wk (Lower) after Ad-GFP infection. (Center) IHC analysis of SPC (Upper) and CC10 (Lower) expression. Images show bronchiolar branches depicting Clara cell hyperplasia as well as hyperplastic areas of SPC⁺ cells in the terminal bronchiolar region. (Right) IF analysis using SPC⁺ and CC10⁺ antibodies. The merged image depicting the overlay between SPC and CC10 signals revealed the presence of a small cluster of double CC10⁺/SPC⁺ cells (Inset). Nuclei were counterstained with DAPI. (Scale bars: A Center, 200 μm; A Left, A Right, and B, 100 μm.)

Ad-Cre–infected mice also developed bronchiolar adenomas. A percentage of these low-grade tumors contained a mix population of SPC⁺ and CC10⁺ cells. Small clusters of double CC10⁺/SPC⁺ cells were occasionally observed (Fig. S6 A and B). Similar results were observed in the BADJs. In contrast, adenomas located in the alveoli were almost exclusively made up of SPC⁺ cells. These alveolar adenomas were the only lesions that progressed to yield malignant adenocarcinomas. We never observed high-grade tumors in the bronchiolar or BADJ areas. Interestingly, the levels of p-ERK staining appeared to be more intense in the alveolar lesions (Fig. S5C). Whether the oncogenic signaling mediated by the resident K-Ras^G12V oncoprotein is mitigated in adult CC10⁺ Clara cells is an interesting possibility that needs to be further investigated. In any case, our findings indicate that the SPC⁺ ATII cells are the only cell type capable of acquiring malignant properties to yield high-grade tumors.

Development of Bronchiolar Hyperplasias Is Associated with an Inflammatory Response.

Mice infected with Ad-Cre, regardless of whether they expressed the K-Ras^G12V oncogene, displayed inflammatory infiltrates not observed in 4OHT-treated mice (Fig. S7A). These infiltrates were primarily made of T and B lymphocytes (Fig. S7B). No significant numbers of macrophages or megakaryocytes were observed. To determine whether this inflammatory response, caused by the adenoviral infection, might be responsible for the generation of bronchiolar and BADJ lesions, we infected K-Ras^{lox/LSLG12Vgeo} mice previously exposed to 4OHT with adenoviral particles expressing a control green fluorescent protein (Ad-GFP). As illustrated in Fig. 5B, these mice also developed bronchiolar hyperplasias, albeit with lower frequency because of the significantly lower numbers of K-Ras^G12V–expressing cells in the bronchiolar and BADJ epithelia in 4OHT-treated mice. As expected, these hyperplasias expressed the K-Ras^G12V oncoprotein as determined by X-Gal staining (Fig. 5B). These hyperplasias were indistinguishable from other observed in Ad-Cre–infected mice. That is, they had papillary morphology and were primarily made up of CC10⁺ cells (Fig. 5B). Some of them also contained SPC⁺ and double CC10⁺/SPC⁺ cells (Fig. 5B). Similar results were obtained in the BADJ region. A limited number of 4OHT-treated K-Ras^{lox/LSLG12Vgeo} mice infected with Ad-GFP also developed adenomas similar to those observed in Ad-Cre–infected mice (Fig. S6 C and D). As expected, these Ad-GFP–infected mice displayed inflammatory infiltrates primarily made of T and B cells similar to those observed in mice infected with Ad-Cre particles (Fig. S7).

Clara Cells Are the Main Target for K-Ras–Induced Transformation During Embryonic Development.

Sca1 appears to be a marker for lung epithelial precursor cells (3, 14). Thus, we interrogated whether expression of K-Ras^G12V in Sca1⁺ cells may activate different types of cancer-initiating cells either during embryonic development or in adult mice. To this end, we crossed K-Ras^+/LSLG12Vgeo mice to a Sca1-Cre^+/T transgenic strain that expresses the Cre recombinase under the control of the Sca1 promoter (SI Materials and Methods). These promoter sequences begun driving Cre expression in the developing lung at embryonic day (E) 13.5 as determined by analysis of LacZ expression in Rosa26^+/LSLLacZ;Sca1-Cre^+/T embryos (Fig. S8A). At this time, the developing lung contains SPC⁺ cells thought to be progenitors of the mature ATI, ATII, and Clara cells (13). Expression of the CC10 marker appears during late embryonic development (E18.5). At E17.5, the K-Ras^G12V oncogene is expressed in SPC⁺ cells (Fig. S8B). However, just a day later, we observed the appearance of small clusters of X-Gal–positive cells in the bronchiolar region. These cells now stained with CC10 and not with SPC antibodies, suggesting that they had differentiated into the Clara cell lineage (Fig. S8C). In addition, we observed some small clusters in the alveoli that remained positive for SPC staining. At postnatal day (P) 1, we started to observe small hyperplastic areas that could be easily recognized by histological analysis (Fig. S9A). Most of these small lesions (64%) were located in the bronchiolar region, although we also observed some at the BADJs (20%) and the alveoli (16%) (Fig. S9B).

When we analyzed these mice at 8–10 wk of age, their lungs displayed large bronchiolar hyperplasias as well as multiple adenomas, all of which were made up of CC10⁺ Clara cells (Fig. 6 A and B). Some of these mice displayed some breathing difficulties possibly due to obstruction of the airways caused by the CC10⁺ bronchiolar hyperplasias (Fig. 6A). Occasionally, we also observed areas containing an increased density of SPC⁺ cells within the alveolar regions (Fig. 6C). However, these areas did not progress to form either hyperplasias or adenomas, at least during the limited lifespan of these animals.

Fig. 6.

Expression of K-Ras^G12V under the control of a Sca1 promoter during embryonic development induces Clara cell tumors. (A–C) Representative lung sections obtained from a 10-wk-old K-Ras^+/LSLG12Vgeo;Sca1-Cre^+/T mouse were incubated with SPC and CC10 antibodies. (A) Bronchiolar hyperplastic area. The normal bronchiolar epithelium is indicated by dotted lines. Arrowheads indicate hyperplastic areas inside the lumen. (B) Bronchiolar adenomas. (C) Representative area depicting increased density of SPC⁺ cells (arrows). (D) Representative lung section obtained from a K-Ras^+/LSLG12Vgeo;Sca1-CreERT2^+/T mouse killed 8 wk after 4OHT treatment and incubated with SPC and CC10 antibodies. Tumor area is defined by a dotted line. Nuclei were counterstained with DAPI. The merged figures depict the overlay between the SPC and CC10 images. (Scale bars: 50 μm.)

Finally, we crossed K-Ras^+/LSLG12Vgeo mice to Sca1-CreERT2^+/T, another transgenic strain that expresses the inducible Cre recombinase CreERT2 under the control of the same Sca1 promoter sequences (SI Materials and Methods). X-Gal staining of lung sections obtained from K-Ras^+/LSLG12Vgeo;Sca1-CreERT2^+/T mice exposed to 4OHT at weaning revealed K-Ras^G12V-expressing cells at the alveoli, bronchioles, and BADJs (Fig. S9C) that expanded to yield clusters preferentially located in the alveolar region (Fig. S9C). Indeed, as previously observed in adult K-Ras^{lox/LSLG12Vgeo};RERT^ert/ert animals, only those clusters containing SPC⁺ ATII cells continued to proliferate to yield tumors (Fig. 6D). These results illustrate the differential nature of cancer-initiating cells in embryonic and adult lungs.

Discussion

In this study, we provide experimental evidence illustrating that adult lung cells are uniquely sensitive to transformation by K-Ras oncogenes. However, within the lung, not all cells are equally permissive to transformation. Cre-mediated activation of a resident K-Ras^G12V oncogene by passive diffusion of 4OHT led to expression of the K-Ras^G12V oncoprotein in cells distributed throughout the major regions of the lung. A significant percentage of these cells underwent few rounds of unscheduled cell division, except in the large bronchi. However, only SPC⁺ ATII cells were able to sustain a proliferative response leading to the formation of AAHs, some of which progressed to adenomas and malignant adenocarcinomas. Whether these cells are mature ATII or committed precursors will require the development of specific markers to allow a more precise characterization of the SPC⁺, K-Ras^G12V-expressing cells identified in this study.

Activation of the resident K-Ras^G12V oncogene with Ad-Cre vectors through the lung airways yielded somewhat different results. In this case, K-Ras^G12V expression was preferentially observed in bronchiolar and BADJ cells. However, unlike in mice exposed to 4OHT, a percentage of these K-Ras^G12V-expressing cells formed large clusters that progressed to papillary hyperplasias and adenomas. These differential results are likely to be a direct consequence of the inflammatory response induced by the adenoviral particles. Indeed, infection of mice treated with 4OHT and infected with control adenoviral particles expressing GFP also induced bronchiolar and BADJ hyperplasias that, in some cases, progressed to benign adenomas. No hyperplasias were ever observed in control mice not exposed to adenoviral particles. These observations are reminiscent of the effect of pancreatic injury on adult acinar cells expressing the same K-Ras oncogene (9). This inflammatory response may alter the differentiation program of the CC10⁺ cells, leading to a less mature, progenitor-like state that becomes responsive to K-Ras oncogene-induced transformation. Alternatively, the inflammatory response may inhibit oncogene-induced senescence as observed in pancreatic acinar cells (15).

Bronchiolar lesions primarily contained CC10⁺ cells; however, some of them had limited numbers of SPC⁺ and double CC10⁺/SPC⁺ cells. It is possible that these cells migrated from the alveoli and the BADJs, respectively. However, it is more likely that they originated by differentiation of K-Ras^G12V-expressing CC10⁺ cells. Clusters located in the BADJs were more heterogeneous, containing a significant number of CC10⁺/SPC⁺ cells. Whether these cells also arose by differentiation of K-Ras^G12V expressing CC10⁺ cells or represented clonal expansion of BASC precursors remains to be determined. Despite the proliferative properties displayed by bronchiolar and BADJ cells in Ad-Cre–infected lungs, only alveolar clusters made of SPC⁺ ATII cells progressed to yield malignant adenocarcinomas. These observations were independent of the method used to activate the resident K-Ras^G12V oncogene. Hence, the ability of SPC⁺ ATII cells to generate malignant tumors appears to be an intrinsic property of these cells. Finally, it is important to notice that none of the cells clusters or hyperplastic lesions observed in this study contained significant numbers of neuroendocrine or ATI cells.

A third scenario emerged when we expressed the resident K-Ras^G12V oncogene during embryonic development under the control of a Sca1 promoter. Although K-Ras^G12V was expressed in both SPC⁺ ATII cells and CC10⁺ Clara cells, only the latter led to the formation of bronchiolar hyperplasias and adenomas, even in the absence of tissue injury. No SPC⁺ hyperplastic regions could be observed in these mice. Unfortunately, the limited lifespan of these animals prevented us from determining whether these transformed CC10⁺ Clara cells could progress to yield malignant adenocarcinomas. However, when we activated a Sca1-driven inducible Cre recombinase in adult mice, they only developed lesions containing SPC⁺ ATII cells. These observations suggest that the intrinsic resistance of Clara cells to become transformed by K-Ras^G12V oncogenes is acquired during postnatal development.

In summary, our results illustrate that different types of lung cells can initiate tumor development depending on the timing of K-Ras^G12V activation (embryonic vs. postnatal) and the microenvironment (inflammatory response). However, only the SPC⁺ ATII cells or their committed precursors were able to yield malignant adenocarcinomas. Similar conclusions were obtained in a recent study in which expression of the resident K-Ras oncogene was mediated by expression of the Cre recombinase under the control of cell type-specific promoters (4). It is also possible that the CC10⁺ or the mixed lesions observed in mice infected with Ad-Cre vectors might have progressed into malignant tumors if we had introduced additional mutations such as loss/inactivation of P53 or LKB1, two tumor suppressors frequently mutated in human cancer. Indeed, a recent study has suggested that the cancer-initiating cells in lung tumors differed based on the nature of the initiating mutations (16).

The ultimate goal of GEM tumor models is to provide an experimental scenario that closely recapitulates the natural history and physiopathological properties of human tumors. Early studies have indicated that human and mouse lung adenocarcinomas containing K-RAS oncogenes share common signatures (17). Detailed comparative analysis of hyperplastic and adenomatous lesions obtained from cancer patients with those generated in the experimental paradigms described here, as well as in related mouse models (4, 5), should provide key information regarding the early events that lead to the development of K-RAS–driven lung adenocarcinomas.

Materials and Methods

Mouse Strains.

The K-Ras^LSLG12Vgeo, RERT^ert, and K-Ras^lox alleles were described (7, 10). The Sca1-Cre and the Sca1-CreERT2 transgenic lines were generated in I.S.-G.’s laboratory (SI Materials and Methods). All animal experiments were approved by the Centro Nacional de Investigaciones Oncológicas Ethical Committee and performed in accordance with the guidelines for ethical conduct in the care and use of animals as stated in The International Guiding Principles for Biomedical Research involving Animals, developed by the Council for International Organizations of Medical Sciences. All alleles were genotyped by Transnetyx.

Cre-Mediated Recombination.

CreERT2-mediated recombination was induced at weaning by i.p. administration of 4-OHT (1 mg per dose). Intratracheal Ad-Cre instillation (2.5 × 10⁸ pfu per mouse) was carried out under anesthesia (i.p. injection of 75 mg/kg ketamine, 12 mg/kg xylazine) (18) in 8- to 10-wk-old mice.

FACS Analysis.

Organs were mechanically disaggregated to obtain single-cell suspensions, stained with the fluorescent substrate of β-galactosidase fluorescein di-β-d-galactopiranosyde (FDG) (Molecular Probes) and detected with a FACScan cytometer (BD Biosciences).

Histopathology, Immunohistochemistry, and Immunofluorescence.

X-Gal and hematoxilin/eosin (H&E) stainings were carried out as described (7). For immunohistochemical (IHC) and immunofluorescence (IF) analyses, 3-μm tissue sections were incubated with antibodies directed against SPC/Sftpc (Merck Millipore), CC10 (Santa Cruz Biotechnology), AQP5 (Abcam), and CGRP (Sigma). Alexa Fluor 555 or Alexa Fluor 647-conjugated secondary antibodies (Molecular Probes) were used for IF. Sections were mounted with Vectashield medium containing 4',6-diamidino-2-phenylindole (DAPI; Vector Laboratories). Quantification of areas from histological sections was performed by using the Pannoramic Viewer software (3DHISTECH).