Targeting single oncogenes is not likely to be effective in all cases for the treatment of human cancers 
. Murine models provide a preclinical strategy to identify which combination of oncogenes are most likely to be effective 
. To our knowledge, our study is the first to examine experimentally using conditional transgenic model systems if the combined inactivation of two oncogenes is more likely to be effective in the treatment of cancer in situ
. Using our models we interrogate the role of MYC and K-rasG12D
alone or in combination for the initiation and maintenance of lung and hematopoietic tumorigenesis. The inactivation of K-rasG12D
but not MYC could induce complete tumor regression in lung adenocarcinomas; whereas in marked contrast, single K-rasG12D
- or MYC-inactivation both succeeded in inducing sustained regression in lymphomas. However, the combined inactivation of both K-rasG12D
and MYC was capable of inducing complete regression in both lung tumors and lymphomas.
Our data highlight two important considerations in targeted therapeutics: first, initiation of tumorigenesis by a specific oncogene does not mean inactivation of that specific oncogene will be sufficient to induce tumor regression; and second, that the consequences of the inactivation of a particular oncogene are pointedly dependent on tissue context. Specifically, we have demonstrated that the K-Ras pathway and its down-stream effector Stat3 are correlated with the ability of K-rasG12D or MYC to initiate lung tumorigenesis and that down regulation of the K-Ras/Stat pathway is strongly correlated with lung tumor and lymphoma regression. We conclude that the K-Ras/Stat3 pathways have a dominant role in the initiation and maintenance of lung tumors.
Our experimental model system re-examines the classic experiments first demonstrating the cooperation between c-myc
for malignant transformation in vivo 
. Identical to previous results using conventional transgenic models 
, MYC and K-rasG12D
cooperated to induce tumorigenesis in lymphocytes (compare LM, LR and LMR mice; ). In contrast, MYC failed to cooperate with K-rasG12D
to induce lung adenocarcinomas (compare CM, CR and CMR mice; ). Thus, whether or not MYC and K-rasG12D
functionally cooperate to activate critical tumor promoting pathways appears to depend upon the specific tissue context. In lung adenocarcinomas induced by MYC and/or K-rasG12D
, tumors exhibited activation of the K-Ras/Stat3 signaling pathway. Apparently, MYC activation is not capable of initiating lung tumorigenesis without activation of the mediators of the K-Ras/Stat3 pathway, perhaps accounting for why MYC does not appear to cooperate with K-rasG12D
to induce lung tumorigenesis. For lung tumorigenesis, there must be an essential role for activation of the K-Ras pathway or downstream mediators such as the Stat pathway. Similarly, the combined inactivation of MYC and K-rasG12D
was now capable of reversing lung tumorigenesis in contrast to MYC-induced lung tumors, because under these circumstances the K-Ras pathway and presumably the downstream Stat3 pathway can be conditionally inactivated.
Notably, our results are highly consistent with several elegant studies that illustrated that mutation of the K-Ras pathway in breast tumorigenesis can reduce the dependence of tumors on sustained MYC overexpression 
. We acknowledge that the differences seen between tumor initiation and maintenance could be secondary to differences in the expression of MYC and K-ras
from the two different tissue specific promoters and/or may be confined to the particular genetic background of the mice used in our study. Nevertheless, we speculate that the combined inactivation of both the MYC and K-Ras pathways in these breast tumor models will also result in complete tumor regression.
In our lung tumor model system other genes are likely to be somatically activated in the EGFR/BRAF/KRAS
pathway or parallel pathways that may also contribute to the escape from the requirement of MYC expression. This is evidenced in our study by the inactivated CM () and CMR () lung tumors that did not demonstrate aberrant signaling in any of the pathways we examined. Possible candidates to undergo such mutations include a multitude of gene products described in studies of human lung tumors 
, some of these studies have implicated the EGFR/IL-6/Stat3 pathway in the pathogenesis of lung adenocarcinomas 
. Stat3 and Stat5 transcription factors have been widely implicated in the pathogenesis of tumors 
and are known to be downstream targets of K-Ras 
. Phosphorylation of Stat3/5 promotes homotetramerization, followed by nuclear translocation and increased transcription of target genes critical for cell growth, survival, and angiogenesis 
. Persistent activation of Stat3/5 was found in the majority of our inactivated MYC-induced lung tumors, as evidenced by elevated phosphorylation and nuclear localization by IHC (). Consistent with a role of Stat3/5 in oncogene-addiction, phosphorylation of Stat3/5 has been shown to diminish in tumor cells undergoing apoptosis upon oncogene inactivation in vitro 
. Phosphorylated Stat3/5 appears to be particularly important for survival of human lung adenocarcinoma cells harboring certain EGFR
. Similar to the lung cancer mouse models described above, in these EGFR
mutated lung cancers behave in an oncogene-addicted fashion following treatment with EGFR tyrosine kinase inhibitors 
Our observations illustrate that the combined inactivation of multiple oncogenes is more likely to be effective to treat some cancers 
. The potential of targeting multiple oncogenic pathways in the treatment of human cancer has recently been illustrated in brain tumor cell lines in vitro 
. The identification of the best gene products to therapeutically target in cancers is very likely to be much more complicated than simply identifying the genes mutated in a given tumor, as has recently been illustrated in human lung cancer patients who become resistant to tyrosine kinase inhibitors 
. In this work, we illustrate that even the knowledge of the oncogene that initiated tumorigenesis is not necessarily sufficient to identify a gene product whose inactivation will result in tumor regression. The generation of transgenic mice with multiple conditional oncogenes is a tractable preclinical platform to define the combination of oncogenic targets most likely to be effective in the treatment of cancer.