Through the development of new genetic tools in mice, we provide strong evidence to support the notion that human NSCLC can originate from alveolar type II cells. Our results confirm and extend the findings in a recent report 
, which primarily analyzed mice carrying Kras
mutations. No evidence was presented in the published article 
to demonstrate tumor invasion or metastasis. It is worth noting that earlier reports using adenoviral-Cre to infect lung epithelium globally showed that mice carrying the Kras
mutation and one loss-of-function allele of p53
in the lung exhibited phenotypes similar to mice carrying the Kras
mutation alone 
. This suggests that loss of both copies of p53
is critical for invasive tumors to develop. Our experimental setting enabled us to achieve simultaneous Kras
activation and p53
ablation. As a result, tumor invasiveness was observed in Kras/p53
mutants but not in mice carrying activated Kras
alone under our experimental conditions. Importantly, taking advantage of the leaky expression of our SPCCreER
line only in alveolar type II cells but not in BASCs, we were able to attribute lung adenocarcinoma developed in these mice to type II cells and not BASCs. Taken together, these new findings strongly support the idea that development of lung adenocarcinoma (beyond adenoma or carcinoma in situ
) originates from SPC+
alveolar type II cells.
A previous report 
relied on a CCSP-rtTA
transgene to investigate how EGFR mutations in different lung cell types could influence tumor development. The CCSP-rtTA
transgene utilizes a short fragment of the rat CCSP
promoter to drive rtTA expression. The short rat CCSP
promoter does not recapitulate mouse CCSP
) expression and displays significant leaky expression in SPC-expressing cells, raising the possibility that it is expressed in BASCs. How EGFR mutations affect BASC proliferation/transformation was not addressed in previous work 
. Our SPC-rtTA
mouse line in which rtTA is under the control of the endogenous SPC
promoter provides the first mouse model to assess the effects of EGFR mutations on transforming SPC+
type II cells.
While our results suggest that NSCLC could initiate in type II cells, the exact nature of tumor-initiating cells requires further investigation. Without tracing the early events at the molecular level, it is not possible to decipher how differentiated type II cells that carry oncogenic mutations undergo progressive transformation. It is possible that all SPC-expressing type II cells share a similar potential of being transformed. Alternatively, tumors may only initiate in a small population of uncharacterized type II cells, for instance, those expressing low levels of SPC and presumably possessing proliferative potential in various conditions 
or those residing at specific anatomical locations. Without markers currently available to define subpopulations of type II cells, it is not possible to assess the transformation potential of distinct subpopulations of type II cells, if they do exist. It is also uncertain how the transformation potential of type II cells is compared to other lung epithelial cell types such as Clara cells, BASCs, ciliated cells, goblet cells and type I cells under various genetic perturbations. Perhaps, distinct cell types are susceptible to specific insults and a systematic investigation is required to delineate the molecular basis of proliferative potential among distinct lung cell types or populations. These studies would shed light on the source of heterogeneity or subtypes of NSCLC, the very properties of which could originate from distinct cell types carrying different mutation profiles. Finally, we cannot rule out the possibility that cells with altered p53 and Kras activities may exert non-cell autonomous effects on neighboring cells and induce tumor development in these cells containing wild-type p53
We found some degree of leakiness present in both SPCCreER and SPCrtTA mouse strains that we used to manipulate gene activity in type II cells. In particular, more sensitive reporter lines (e.g., ROSA26mTmG) appear to detect even very low levels of leaky expression. In some cases, variations in background expression among individual mice were also noted. Consistent with these findings, when leaky CreER expression reached a critical threshold in SPCCreER/+; KrasLSL-G12D/+ or SPCCreER/+; p53f/f; KrasLSL-G12D/+ mice, cell hyperplasia and even tumor development was observed without TM administration albeit at a lower frequency than in mice injected with TM. Interestingly, even in mice in which leaky CreER expression led to tumor development, tumors were never detected at the BADJ. By contrast, SPCrtTA/+; tetO-EGFRL858R mice without dox do not develop hyperplasia or tumors. This probably reflects differential sensitivities of type II cells to diverse perturbations.
Given that SPCCreER confers the ability to manipulate gene activity in BASCs, it is somewhat surprising that hyperplasia at the BADJ was not observed in SPCCreER/+; KrasLSL-G12D/+ or SPCCreER/+; p53f/f; KrasLSL-G12D/+ mice when CreER is activated by TM. This raises the possibility that BASCs are not as sensitive to Kras/p53 perturbation as other SPC-expressing cells. Similarly, active EGF signaling in SPC-expressing cells only led to alveolar tumor development and not BASC hyperplasia or neoplasia. Since BASCs express both SPC and CC10, it would be interesting to compare the transformation potential between BASCs and Clara cells. Whereas our results do not provide evidence to support a role for BASCs in tumor development, it is conceivable that additional genetic perturbation not utilized in this study may be required to transform BASCs. It is also possible that a longer time period is needed for tumors to develop in BASCs. This is consistent with the observation that human lung tumors usually develop after many years or even decades of repeated epithelial injury such as from cigarette smoking. Though BASC proliferation is rare in our experimental setting, we cannot formally rule out the possibility that BASCs can be transformed, expanded and differentiated into type II-like cells. Lineage-tracing studies in which Kras/p53 mutant BASCs and type II cells are labeled and followed during early tumor development will provide new insight into this important issue.
Human NSCLC is a heterogeneous group of diseases that share common histology but carry different mutations and perhaps even originate from different cell types 
. For instance, some NSCLC cell lines express markers of alveolar type II cells (). By contrast, the mouse models we developed likely mimic only a subset of human lung cancer. A key feature of the system is its ability to mark alveolar type II cells and trace the early events of tumor development. Equally important is the opportunity that this system offers to isolate transformed type II cells induced by different genetic perturbations at various stages of progression for whole-genome analysis (including RNA-Seq and Exome sequencing). These studies will yield key insight into the molecular mechanisms of transformation. Such analysis will also provide candidates for early diagnosis and targeted therapies of NSCLC 
. We speculate that our mouse model offers a defined genetic system with a shorter time frame of tumor development in which fewer mutations are accumulated in the process of transformation compared to human lung cancer. Employing mouse models would facilitate the identification of relevant candidates for diagnosis and treatment through comparative genomic analysis between human and mouse tumors.