p18^Ink4c Collaborates with Men1 to Constrain Lung Stem Cell Expansion and Suppress Non–Small-Cell Lung Cancers

Mutant mice lacking both cyclin-dependent kinase (CDK) inhibitors p18^Ink4c and p27^Kip1 develop a tumor spectrum reminiscent of human multiple endocrine neoplasia (MEN) syndromes. To determine how p18 and p27 genetically interact with Men1, the tumor suppressor gene mutated in familial MEN1, we characterized p18-Men1 and p27-Men1 double mutant mice and showed that p18, but not p27, functionally collaborates with Men1 in suppressing lung tumorigenesis. Lung tumors developed in both Men1^+/− and p18^−/−;Men1^+/− mice at a high penetrance and contain both neuroendocrine and nonneuroendocrine cells. The remaining wild-type Men1 allele was lost in most lung tumors from Men1^+/− mice but was retained in most tumors from p18^−/−;Men1^+/− mice, showing a functional collaboration between p18 and Men1 in lung tumor suppression. Phosphorylation of Rb protein at both CDK2 and CDK4/CDK6 sites were significantly increased in normal bronchial epithelia and tumor cells derived from p18^−/−;Men1^+/− mice compared to those from single p18^−/− or Men1^+/− mice. Lung tumors developed in p18^−/−;Men1^+/− mice were multifocal, more heterogeneous, and highly invasive compared to those developed in either p18^−/− or Men1^+/− mice. Bronchioalveolar stem cells are expanded in normal and tumorigenic lungs of p18^−/− mice and are further expanded in p18^−/−;Men1^+/− lung tumors. These results reveal a previously unrecognized function of p18 in lung tumor suppression through collaboration with Men1 to control lung stem cell proliferation. [Cancer Res 2007;67(7):3162–70]

Lung cancer is the leading cause of cancer death worldwide and accounts for more than a quarter of all cancer deaths in the United States (1). Clinically, lung cancer can be divided into two major histopathologic groups: non–small-cell lung cancer (NSCLC), which accounts for 80% of lung cancer cases, and small-cell lung cancer (SCLC), which, although accounting for <20% of the cases, are highly proliferative and malignant with a close to complete mortality. Compared with the investigation of other major tumor types such as cancers of the breast, prostate, colon, and blood, much less is known about the molecular and cellular pathways involved in lung cancer development including, in particular, the lack of appropriate animal models. Whereas enormous efforts have been made to generate mouse models for human lung cancer over the past decade, nearly all presented thus far have been gain-of-function mutant mouse models in which mice were transgenically overexpressing an activated oncogene (2).

All SCLC and ∼8% of NSCLC exhibit neuroendocrine features. Endocrine glands include seven so-called classic tissues or organs (pituitary, thyroid, parathyroid, pancreatic islets, adrenals, testes, and ovaries), as well as various other tissue systems such as lung and gastrointestinal carcinoid that are not usually considered endocrine glands but contain cells that produce and secrete hormones, cytokines, and secondary messengers (3). The recognition of Mendelian inheritance of hereditary multiple endocrine neoplasia (MEN) syndromes has led efforts, over the past decade, to identify causal genes by linkage analysis and positional cloning, including the type 1 multiple endocrine neoplasia (MEN1) gene MEN1, which encodes a nuclear transcriptional regulator protein, known as menin (4). Despite its broad tissue expression in nearly all tissues during embryonal development and in adults (5), loss of function of the MEN1 gene perplexingly causes a unique and restricted pattern of endocrine tumors in both human and mice. Germ line mutations in MEN1 predispose human individuals to endocrine tumor development, predominantly in the anterior pituitary, parathyroid, and pancreatic islet cells (6). Likewise, mice heterozygous for Men1 developed multiple endocrine tumors with a spectrum very similar to human MEN1 kindreds (7, 8). Both the biochemical function of menin and the cellular pathway Men1 are currently under investigation. In particular, how Men1 may genetically interact with other genes involved in tumor suppression has not been examined.

p18^Ink4c is a member of the mammalian INK4 family of cyclin-dependent kinase (CDK) inhibitors, encoding a haploinsufficient tumor suppressor in mice (9–11). Loss or reduced expression of the p18 gene has also been observed in several types of human cancers, including Hodgkin lymphomas (12), parathyroid tumors (13), hepatocellular carcinomas (14), and medulloblastomas (15). Mice deficient for p18 (10, 11, 16) or another CDK inhibitor, p27^Kip1, develop characteristic intermediate lobe pituitary tumors (17–19). More strikingly, mice simultaneously deleted for both CDK inhibitors p18 and p27 develop tumors in multiple endocrine organs, including the pituitary, thyroid, parathyroid, adrenal glands, pancreas, and testis (11, 20), a tumor spectrum that is congruous with the human multiple endocrine neoplasia (MEN) syndrome. These observations led us to determine how Men1 genetically interacts with p18 and p27 and uncovered a previously unrecognized function of p18, but not p27, in suppression of NSCLC tumors and in regulation of lung stem cells through collaboration with Men1.

Mouse strains, histopathology, and immunohistochemistry. The generation of p18 and Men1 mutant mice and conditions for genotyping were previously described (7, 10). p18 mutant mice in the C57BL/6 background were backcrossed for nine generations with C57BL/6 mice, and Men1 mutant mice in the mixed background C57BL/6 × 129/SvEv were backcrossed for two generations with C57BL/6 mice. Mice heterozygous for both p18 and Men1 were then intercrossed to generate all of the genotypes analyzed in this study. Cohorts were housed and analyzed in a common setting. The Institutional Animal Care and Use Committee of the University of North Carolina approved all procedures.

Tissues of most organs were removed, fixed in 10% neutral buffered formalin, and examined histologically by two pathologists after H&E staining. Lesions were photographed and additional sections were taken for immunohistochemical analyses. Immunohistochemistry was done as previously described (9). To detect lung stem cells, sections were immunostained with a goat polyclonal antibody against CC10 (Santa Cruz Biotechnology, Santa Cruz, CA) and a rabbit polyclonal antibody against pro-surfactant apoprotein-C (SP-C; Chemicon, Temecula, CA). Other primary antibodies used were as follows: synaptophysin (Lab Vision, Fremont, CA), neurospecific enolase (Lab Vision), Chromogranin A (Lab Vision), phosphorylated histone H3 (Upstate Biotechnology, Charlottesville, VA), phospho-Rb(Ser⁶⁰⁸) (Cell Signaling, Danvers, MA), phospho-Rb (Thr⁸²¹) (Biosource, Camarillo, MA), and menin (Bethyl Laboratories, Montgomery, TX). Immunocomplexes were detected using the Vectastain ABC alkaline phosphatase kit according to the manufacturer's instructions (Vector Laboratories, Burlingame, CA) or using rhodamine-conjugated secondary antibody. For apoptosis assays, terminal deoxynucleotidyl transferase–mediated dUTP-biotin nick-end labeling (TUNEL) assays were carried out with an in situ ApopTag kit (Chemicon) according to the manufacturer's protocol.

Laser capture microdissection and loss of heterozygosity analysis. Laser capture microdissection was used to obtain pure cell populations of selected areas from formalin-fixed, paraffin-embedded tissue sections. Ten-micrometer sections were deparaffinized and lightly stained with hematoxylin. Using a PixCell lle Laser Capture Microdissection system (Arcturus, Mountain View, CA), the lesions that were clearly separated from normal tissues were isolated from the slides. Particular care was taken to avoid contamination by surrounding tissue. DNA isolation from the microdissected tissue samples was done as described before (9).

Development of lung tumors in p18-Men1 mutant mice. Men1^−/− mice die in utero (between E11.5 and 13.5) and Men1^+/− mice develop endocrine tumors later in their life with a spectrum similar to those in MEN1 patients (7, 8). Whereas conditional alleles have been generated to allow tissue-specific homozygous deletion of Men1, we carried out our studies in Men1^+/− mice to gain a broad view on functional interaction between Men1 with p18 and p27 because of the broad expressions of all three genes in many tissues. Genotype analysis of >250 offspring did not identify any p18^−/− and p27^−/−;Men1^−/− double null mice, nor were there any viable p18^−/−;Men1^−/− and p27^−/−;Men1^−/− embryos beyond E15.5 (data not shown). These results indicate that homozygous deletion of p18 or p27 did not rescue the embryonic lethality caused by Men1 loss. p18^−/−;Men1^+/−, p27^−/−;Men1^+/−, and Men1^+/− mice developed normally and did not show significant phenotypic abnormalities at a young age (<3 months) but subsequently developed various hyperplastic and tumor phenotypes in the pituitary, thyroid, parathyroid, pancreas, adrenal, and testis. Compared with these single mutant mice, tumor phenotypes in nearly all these endocrine organs were accelerated in double mutant p18^−/−;Men1^+/− mice but not in p27^−/−;Men1^+/− mice (data not shown). Unexpectedly, lung tumors developed in both Men1^+/− and p18^−/−;Men1^+/− mice at a high frequency and exhibited different pathologic and morphologic changes and are characterized in detail in this study.

Before 1 year of age, no lung tumors were found in any mutant mice lacking p18, p27 or Men1. When the mice aged beyond 1 year, 1 of 12 (8%) p27^−/− mice developed adenocarcinoma in the lung and 2 of 10 p18^−/− (20%) mice developed lung tumors (1 adenoma and 1 adenocarcinoma). Five of 12 (42%) Men1^+/− mice developed lung tumors somewhat higher than previously reported (7). All five lung tumors developed in Men1^+/− mice were adenocarcinomas including two neuroendocrine carcinomas (see below). Only 1 of 16 (6%) p27^−/−;Men1^+/− mice developed lung adenomas. Notably, 4 of 12 (33%) p18^−/−;Men1^+/− mice developed lung adenomas before 1 year of age and 5 of eight (63%) p18^−/−;Men1^+/− mice developed adenocarcinomas after 1 year of age, including three carcinomas with neuroendocrine characteristics (Table 1; Fig. 1). Together with pathologic and cellular changes when p18 and Men1 mutations were combined (see below), these results provide the first evidence revealing a function of p18 in suppressing lung tumor development and a functional collaboration between p18 and Men1 in this process. No evidence was seen that lung tumor incidence was increased in p27^−/−;Men1^+/− mice than in Men1^+/− mice.

The p18^−/−;Men1^+/− mice also developed tumors in multiple endocrine organs (21). To rule out the possibility that the lung tumors were metastasized from other endocrinal organ tumors, we did immunohistochemical staining with the antibodies against adrenocorticotropic hormone, prolactin, 3β-hydroxysteroid dehydrogenase, insulin, and glucagon. Only two lung tumors developed from p18^−/−;Men1^+/− mice with Leydig cell adenocarcinoma in testis were partly positive for 3β-hydroxysteroid dehydrogenase staining, suggesting the metastasis from the testis tumor (data not shown). The remaining lung tumors developed in these two mice, and in other mice, were stained negative for these hormones. Together with the typical pathologic morphology, CC10 and SP-C staining, and detection of bronchioalveolar stem cells (BASC) in the tumors, we conclude that most of lung tumors were primary tumors.

p18^−/−;Men1^+/− lung tumors are more aggressive and contain both neuroendocrine and nonneuroendocrine cells. Detailed pathologic analysis revealed that lung tumors arising from p18^−/− and p27^−/− mice usually display a solitary, uniform nodule with a clear border and homogeneous tumor cell type. Lung tumors from Men1^+/− mice, however, frequently contained several nodules, and individual nodules often consisted of more than one tumor cell type that displayed blurring or disrupted borders, indicative of invasion (Fig. 1A and B). Importantly, when combined with loss of p18, but not p27, Men1^+/− mice developed multiple lung carcinomas. Even in a single tumor lesion, a high degree of cell type heterogeneity was frequently observed. These tumors consisted of densely packed and highly pleomorphic tumor cells with scant cytoplasm and finely granular chromatin that form nests with fine vascular stroma, resembling neuroendocrine carcinomas according to recent recommended classification for diagnosis of mouse proliferative pulmonary lesions (ref. 22; Fig. 1A and B). Morphologically, the p18^−/−;Men1^+/− lung tumor cells contain a higher nuclear grade and were more aggressive and invasive than those from single mutant mice. Most of these tumors were moderately or poorly differentiated and possessed strong potential to invade into blood vessels and surrounding tissues, such as bronchus/bronchiole (Figs. 1 and 2). These results suggest that Men1 suppresses neuroendocrine cell proliferation and tumor formation in bronchi and lungs and that p18 cooperates with Men1 to suppress pulmonary neuroendocrine carcinoma development.

Various neuroendocrine markers, including neurospecific enolase, synaptophysin, and Chromogranin A, were detected in carcinomas, confirming the neuroendocrine features of the Men1^+/− and p18^−/−;Men1^+/− lung tumors. Notably, the immunoreactivity to different neuroendocrine markers varied in lung tumors derived from different mutant mice: p18^−/− tumors stained negatively with all three markers; Men1^+/− tumors stained strongly with synaptophysin but weakly with neurospecific enolase and Chromogranin A; and p18^−/−;Men1^+/− tumors stained with similar intensity with all three markers (Fig. 2). Individual p18^−/−;Men1^+/− lung tumors often contained both neuroendocrine and nonneuroendocrine cells (Fig. 2, bottom row), further indicating the heterogeneity of the tumor cell types.

Increased proliferation and apoptosis in lung tumors of p18^−/−;Men1^+/− mice. The neuroendocrine carcinomas from p18^−/−;Men1^+/− lungs seemed to be more proliferative. Whereas lung tumors developed, Men1^+/− mice contained less than two mitotic cells per 10 high-power fields (or 2 mm²); all p18^−/−;Men1^+/− lung tumors had more than two mitotic cells per 10 high-power fields; and two of three tumors displayed more than 10 mitotic cells per 10 high-power fields (data not shown). To further determine the increased proliferative activity in the lung tumors with combined mutations of p18 and Men1, we examined mitotic index by immunostaining of anti–phospho-histone H3. Phospho-histone H3–positive cells were low in lung tumors developed in Men1^+/− or p18^−/− mice and were substantially increased in lung tumors developed in p18^−/−;Men1^+/− mice (Fig. 3A). The increase of cell proliferation following loss of p18 and Men1 correlated with an increased level of apoptosis in the lung tumors. A higher number of TUNEL-positive cells were detected in p18^−/−;Men1^+/− lung tumors than in either p18^−/− or Men1^+/− mice (Fig. 3A). These results provide further evidence supporting a functional collaboration between p18 with Men1.

p18 deficiency protects the loss of wild-type Men1 allele. MEN1 patients typically inherit loss-of-function mutations in the Men1 gene and tumors arise after loss of the remaining wild-type allele, fitting the classic “two-hit” model for inactivating a tumor suppressor. Development of lung tumors from Men1 heterozygotes offered an opportunity to address the issue of Men1 haploinsufficiency in lung tumor suppression. We took two approaches to this end: immunohistochemistry to examine menin expression and PCR analysis for the remaining Men1 wild-type allele. As determined by immunohistochemistry, menin expression was lost in three of three lung tumors analyzed (Fig. 3B and D). In contrast, menin protein expression was retained in four of five lung tumors developed in p18^−/−;Men1^+/− mice. Similarly, PCR analysis of genomic DNA extracted from laser-capture microdissected tumors revealed that two of two lung tumors developed in Men1^+/− mice exhibited loss of the wild-type Men1 allele, whereas two of three lung tumors derived from p18^−/−;Men1^+/− mice retained the wild-type Men1 allele (Fig. 3C and D). Together, these results show that complete loss of Men1 function renders strong growth advantage for lung tumor development to the Men1^+/− cells, but not the p18^−/−;Men1^+/− cells, and provide additional evidence supporting the notion that p18 and Men1 regulate two separate pathways that collaboratively suppress lung tumorigenesis.

Phosphorylation of Rb protein at CDK2 and CDK4/CDK6 sites was increased in p18^−/−;Men1^+/− bronchiolar epithelium and lung tumor cells. To search for the underlying biochemical basis for the different genetic interaction between p18 and Men1, we examined in both normal and tumor cells of different genotypes the kinase activity of CDK4 and CDK6, two exclusive targets of p18, as well as CDK2 whose activity is indirectly affected by the INK4 proteins. Taking advantage of the availability of several well-characterized antibodies specifically recognizing Rb proteins phosphorylated by CDK at different sites and their suitability for immunostaining of embedded tissue samples, we directly examined normal lung tissues and tumors. Two well-characterized phosphorylation sites in the Rb protein, Ser⁶⁰⁸ phosphorylated by CDK4 and CDK6 and Thr⁸²¹ phosphorylated by CDK2 (23, 24), were examined. A visible increase of pRb phosphorylation at CDK4/CDK6 site Ser⁶⁰⁸ was detected in normal bronchial epithelia of p18^−/− and Men1^+/− mice (Fig. 4, top). Combined mutation of p18^−/− and Men1^+/− substantially increased phosphorylation of Rb at Ser⁶⁰⁸ sites in both normal bronchial epithelia and adenocarcinoma tumor cells (Fig. 4). CDK2 site Thr⁸²¹ was highly phosphorylated in the wild-type normal bronchial epithelia, producing almost saturated staining signals and preventing us from determining how different mutations in p18 and Men1 genes affect Thr⁸²¹ phosphorylation in normal bronchial epithelia proceeding to neoplastic transformation. However, a clear increase of Thr⁸²¹ phosphorylation of Rb was seen in adenocarcinoma tumor cells developed in p18^−/−;Men1^+/− mice than in those developed in either p18^−/− or Men1^+/− mice (Fig. 4, bottom). Together, these results are consistent with the pathologic analyses showing that cell proliferation and tumor growth were synergistically stimulated in p18^−/−;Men1^+/− lungs.

BASCs are expanded in p18^−/− mice. p18 plays an important role in constraining the self-renewal of hematopoietic stem cells (25) and neural stem cells.¹

Injury studies have shown the existence of functional epithelial stem cell population in the lung (26). The mouse lung contains anatomically and functionally distinct epithelial stem cell populations. A putative regional pulmonary stem cell population of distal lung, referred to as BASCs, was recently identified in mice at the bronchioalveolar duct junction as double positive for both SP-C [an alveolar type 2 (AT2) cell-specific marker] and CC10 (a marker for Clara cell; ref. 27). These SP-C/CC10 double-positive cells share many characteristics with adult stem cells, including their localization in a putative stem cell niche, resistance to and increased proliferation after both bronchiolar and alveolar damages, ability to self-renew in clonal assays and multipotent differentiation in culture, and expansion at a very early stage of K-ras–induced lung tumorigenesis. Following the terminology used in literature, we refer to these SP-C/CC10 double-positive cells as BASCs but note that the abilities of these cells to generate differentiated lung epithelia and to self-renew in vivo have not been formally shown.

We first examined BASCs in normal lung tissues from 3-month-old mice when there were no microscopically detectable lesions in the lungs of any of the genotypes we analyzed. Of 20 terminal bronchioles we have examined, BASCs were detected in seven wild-type and eight Men1^+/− terminal bronchioles, respectively, and all BASCs detected were located at bronchioalveolar duct junction of terminal bronchioles where lung stem cell niches were previously suggested by cell injury experimentation (28) and later identified by immunostaining (ref. 27; Fig. 5A). These numbers are in close agreement with a previously reported distribution of BASC-positive terminal bronchioles for the wild-type lung at similar age (35%; ref. 27). This result also suggests that haploid loss of Men1 gene had no significant effect on BASC proliferation or expansion. In contrast, p18 loss resulted in an expansion of both the number of BASC-containing terminal bronchioles and the number of BASCs in individual terminal bronchioles. Of 20 p18^−/− terminal bronchioles examined, 15 (75%) contained BASCs, and a total of 25 BASCs were detected in the 15 terminal bronchioles, including 8 terminal bronchioles with one BASC, 4 terminal bronchioles with two BASCs, and 3 terminal bronchioles with three BASCs. As comparison, of 7 BASC-containing wild-type terminal bronchioles, 5 had one BASC and 2 had two BASCs (Fig. 5C). This result revealed a role of p18 in constraining BASC self-renewal. Of 20 p18^−/−;Men1^+/− terminal bronchioles examined, 21 BASCs were detected in 14 terminal bronchioles, including 9 terminal bronchioles containing one BASC, 3 terminal bronchioles containing two BASCs, and 2 terminal bronchioles containing three BASCs, suggesting that combined Men1 heterozygosity did not further expand BASCs in p18^−/− background at an early age.

BASCs are expanded in p18^−/− and p18^−/−;Men1^+/− lung tumors. We next examined BASCs in lung tumors developed in different mice at ∼1 year of age (Fig. 5B and C). Lung tumors from Men1^+/− mice contain predominantly SP-C–positive cells and sparsely located CC10-positive cells (1.6%; 36 positive cells of total 2,190 cells counted), suggesting that Men1^+/− lung tumors contain mixed cell origins and include a small population of cells of Clara origin. The majority of CC10-positive cells in these tumors were negative for SP-C. Only 0.23% (5 of total 2,190 cells counted) SP-C/CC10 double-positive cells were detected in Men1^+/− lung tumors. Because there were no lung lesions developed in the wild-type mice in the study, we are not able to conclude whether Men1 heterozygosity affects BASCs at this age. Given the old age of the tumors (∼1 year), the distribution of SP-C/CC10 double-positive BASCs in these Men1^+/− lung tumors is probably not significantly different from that reported in total lung cell preparation from younger wild-type mice (0.34%; ref. 27).

Lung tumors derived from p18 null mice, on the other hand, contained almost exclusive SP-C–positive cells and no CC10-positive cells were identified, indicating that cells in the p18^−/− lung tumors originated from AT2 cells and providing further evidence for dissimilar cell origins of p18^−/− and Men1^+/− lung tumors. Ten of 1,854 (0.54%) cells were doubly stained for both SP-C and CC10, indicating a greater expansion of putative BASCs in p18^−/− lung tumors than in Men1^+/− lung tumors. Significantly, the CC10/SP-C double-positive BASCs were substantially increased in p18^−/−;Men1^+/− lung tumors to 2.2% (42 of total 1,895 cells counted), whereas the number of CC10 single-positive cells (1.4%, 27 of 1,895) was similar to Men1^+/− lung tumors (Fig. 5), indicating that combined Men1 heterozygosity further expands BASCs in p18^−/− background during lung tumorigenesis at a late stage. The mechanism underlying the contribution by Men1 heterozygosity to the BASC expansion in p18^−/− mice in an age- and/or tumor stage–dependent manner is not clear. We suspect that either other genetic mutation(s) or epigenetic change(s) accumulate during lung tumorigenesis or an increase of Clara cells is initiated by Men1 heterozygosity and facilitates the expansion of BASCs in p18^−/− background.

p18-Men1 mouse as a system for studying NSCLC. Much effort has been made to generate mouse models for human lung cancer and the only loss-of-function mouse model for lung cancer is the conditional deletion of both Rb1 and p53 that predominantly develops SCLC (29). To the best of our knowledge, the p18-Men1 mouse represents the first loss-of-function mouse model that develops NSCLC. The identity of the cell(s) of origin is largely unknown for both human and mouse NSCLC, which, unlike SCLC, contain multiple cell types. Murine pulmonary tumors are predominantly adenomas and adenocarcinomas and are derived from SP-C–positive AT2 cells (30, 31). CC10-positive nonciliated bronchiolar epithelial Clara cells have also been suggested to be the cell origin of murine lung tumors (32, 33) and are infrequently found in human NSCLC (34, 35). Our studies show that the cells of origin in p18^−/− lung tumors are AT2 cells. Men1^+/− lung tumors, on the other hand, are morphologically more heterogeneous (Fig. 1 and 2) and contain both SP-C–positive AT2 and CC10-positive Clara cells (Fig. 5), suggesting that either Men1^+/− lung tumors were developed from different cellular origins (polyclonal origin) or, more likely, from BASCs that later differentiated into different cell lineages (AT2, Clara, and neuroendocrine cells), implicating a role for Men1 in controlling lung progenitor cell proliferation and differentiation.

Lung tumors developed in p18-Men1 mice are distinct from those developed in K-ras and p53-Rb mice. It is interesting to compare p18-Men1 mice with two other recently developed conditional mouse lung tumor models, one expressing from its native promoter a oncogenic K-ras allele and the other deleting both Rb1 and p53 tumor suppress genes (29, 36). Lung tumors developed in p18-Men1 and K-ras mice are either adenomas or adenocarcinomas and contain mostly AT2 cells. There are two major differences between the lung tumors developed in these mice: the cell type heterogeneity and tumor onset. p18^−/−;Men1^+/− lung tumors are more heterogeneous, containing both neuroendocrine and nonneuroendocrine cells and a low percentage of Clara cells, whereas neither neuroendocrine nor Clara cells were identified in K-ras–induced lung tumors. Somatic activation of oncogenic K-ras resulted in lung tumor development at a near full penetrance after 16 weeks, whereas about two thirds of p18^−/−;Men1^+/− mice did not develop lung tumors before 1 year of age and about one third remained free of lung tumors after 1 year of age. This late onset and incomplete penetrance may result from the function of the remaining Men1 allele.

Lung tumors developed in p18^−/−;Men1^+/− mice are substantially different than those developed in Rb-p53 mice in three major features: tumor type (NSCLC versus SCLC), cell type heterogeneity, and tumor malignancy. Histopathologically, SCLC tumors developed in p53^−/−;Rb^−/− conditional mice contain one homogenous cell type with a neuroendocrine feature, whereas p18^−/−;Men1^+/− mice often develop multiple and different NSCLC tumors per lung, and individual tumors are often highly pleomorphic and contain multiple cell types. SCLC tumors developed in p53^−/−;Rb^−/− conditional mice are more metastatic than the NSCLC tumors developed in p18^−/−;Men1^+/− mice, resulting in decreased survival rate and early death. This is consistent with the worse prognosis associated with SCLC in humans. Considering the genetic difference between two mice, several factors could have contributed to the lower malignancy of NSCLC tumors that developed in p18-Men1 mice. Deletion of p18 only partially reduced, but did not completely abolish, the function of the Rb pathway. Retention of a partial Rb pathway and/or an intact p53 pathway, or possibly even the remaining wild-type Men1 allele in p18^−/−;Men1^+/− cells, although insufficient to suppress tumor initiation and progression, prevents the transition into the metastatic state.

p18 and Men1 perform different and collaborative functions in regulating lung stem/progenitor cells. The role of stem cells in lung tumorigenesis and the gene(s) involved in regulating the stem cells during this process remain largely unknown. In this study, we discovered that p18 loss results in BASC expansion and that this expansion was detected in p18^−/− mice at an early stage, preceding the development of any microscopically visible lesion or histologic abnormality, continuing late in lung tumors. These results are consistent with the idea that the stem/progenitor cells are the target cell population in lung tumor initiation and are involved in maintenance of advanced lung tumors. Equally significant is the finding that Men1 heterozygosity, although not directly affecting BASC proliferation, can functionally collaborate with p18 loss and cause further BASC expansion during lung tumor progression. In this model, Men1 seems to play a rate-limiting role in the initiation of lung tumorigenesis. An expansion of stem/progenitor cell populations resulting from p18 loss contributes to the increased aggressiveness and heterogeneity of lung tumors developed in p18^−/−;Men1^+/− mice. Characterization of these mouse strains provides not only an opportunity for determining the role of stem/progenitor cells in lung tumorigenesis and elucidating the origin of lung cancer cells but also a system to determine the role of stem/progenitor cells in lung tumor maintenance and their therapeutic value.

Grant support: U.S. Department of Defense Career Postdoctoral fellowship (F. Bai), Diane Emdin Sachs Lung Research Award (Y. Xiong), and NIH grant CA68377 (Y. Xiong).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Drs. Francis Collins and Judy Crabtree for providing Men1 mutant mouse, Drs. Jerrold M. Ward and Virginia Godfrey for histologic examination, and Drs. Seung Kim, Matthew Meyerson, and Ned Sharpless for reading the manuscript.