|Home | About | Journals | Submit | Contact Us | Français|
MYC exerts both positive and negative functions in cancer cells, such that its pro-cancerous effects are unmasked only after its anticancer effects are blocked. Here we used multiple mouse models of lung adenocarcinoma to identify genetic events that can cooperate with MYC activation to promote the genesis of non-small cell lung cancer (NSCLC), the most common form of lung cancer in humans. MYC over-expression targeted to pulmonary alveolar cells was sufficient to induce lung adenomas and carcinomas. Tumorigenesis was assisted by either spontaneous mutations in Kras or experimental introduction of activated RAS, but investigations revealed that additional events were required to circumvent apoptosis, one of the most significant negative functions exerted by MYC. We determined that over-expression of the anti-apoptotic protein MCL1 was sufficient to circumvent apoptosis in this setting. Previous clinical studies have indicated that prognosis of human NSCLC is not associated with MCL1, despite its over-expression in many NSCLCs. In re-examining the prognostic value in this setting, we found that MCL1 over-expression does correlate with poor patient survival, but only when accompanied by MYC over-expression. Our findings therefore produce a convergence of mouse and human results that explain how MCL1 can block an important negative consequence of MYC over-expression in both experimental models and clinical cases of NSCLC.
Over-expression of the proto-oncogene MYC has been implicated in the genesis of multiple human malignancies (1). MYC encodes a transcription factor that regulates gene networks controlling proliferation, metabolism and ribosome biogenesis (2-4). However, MYC over-expression also activates the pro-apoptotic BCL2 family protein BAX (5, 6) leading to the release of cytochrome C and apoptosis (7). Compensatory genetic and/or epigenetic alterations are required to diminish the anti-tumorigenic effects of MYC and tip the balance in favor of malignant growth (8, 9).
Here we have used mouse models to uncover events that enable the genesis of lung tumors initiated by MYC. Targeting expression of MYC to pulmonary alveolar cells gave rise to adenomas and carcinomas. Tumorigenesis was assisted by spontaneous mutations in Kras, but also required an additional event to circumvent apoptosis. One such event proved to be over-expression of the anti-apoptotic protein MCL1.
MCL1 is an anti-apoptotic member of the BCL2 family of proteins that localizes to the outer mitochondrial membrane, as well as other intracellular membranes (10). It inhibits apoptosis by interfering with mitochondrial events that lead to cytochrome C release (11, 12). MCL1 is amplified in human NSCLC (13) and cooperates with MYC in experimental leukemogenesis (14, 15).
Previous work has suggested that MCL1 over-expression does not correlate with prognosis in NSCLC (16, 17). We reexamined the prognostic value of MCL1 expression in NSCLC and found that MCL1 over-expression does correlate with poor patient survival, but only when accompanied by over-expression of the MYC protein. We suggest that the joint over-expression of MCL1 and MYC may be a useful biomarker for both prognosis and treatment in human NSCLC. In particular, inhibition of MCL1 might have a synthetic lethal effect by unleashing the pro-apoptotic effect of MYC.
Protocols were carried out with approval of the Institutional Animal Care and Use Committee (IACUC) of the University of California, San Francisco. Transgenic mice expressing the reverse tetracycline transactivator (rtTA) (18) and mice carrying a human MYC transgene under the transcriptional control of a tetracycline-response element (TRE) promoter (19) have been previously described. Lung phenotypes were evaluated according to criteria recommended by the Mouse Models of Human Cancer Consortium (20).
RNA was extracted using the Absolutely RNA® Miniprep kit (Stratagene). Total RNA was reverse-transcribed using Stratascript reverse transcriptase (Stratagene) and cDNA was analyzed by realtime-PCR (Taqman®, Applied Biosystems). Relative gene expression was normalized to a mouse β-Actin Taqman® probe.
Immunohistochemical staining was carried out using the Vector Elite ABC Kit (Vector Laboratories) and TUNEL staining was done using the ApopTag® Peroxidase In Situ Apoptosis Detection Kit (Chemicon). For BrdU labeling, mice were injected with 100 mg/kg BrdU (BD Pharmingen) 2 hours prior to sacrifice. BrdU+ cells were detected using anti-BrdU mouse monoclonal B44 (BD Biosciences).
Western analysis was carried out using standard protocols. Information regarding the antibodies utilized is available upon request. RAS activity was measured in 250 μg of whole lung or tumor protein lysate using a RAS Activation Assay Kit (Millipore).
The entire coding sequences for Kras, Nras, and Hras were cloned into the vector pCR2.1 using a TA Cloning kit (Invitrogen). A mutation was deemed present when 2 or more clones harbored the same nucleic acid change in comparison to the wild-type allele. We sequenced 10 clones per sample.
Mice were injected once intraperitoneally at 4 weeks of age with 50 mg/kg N-methyl-N-nitrosourea (MNU) and monitored for up to 6 months post injection. Tumors present on the pleural surface were enumerated using a dissecting scope.
Transgenic mice were treated with DOX 24 hours prior to tracheal injection of retrovirus. Anesthetized mice were intubated and concentrated viral particles were directly introduced to the airways via the intubation tube using a syringe and needle. Afterwards, cohorts of transduced mice were either kept free of doxycycline (DOX) or fed a DOX diet to continuously induce MYC. Ecotropic retrovirus was produced using standard techniques. The pMig retroviral plasmid (21) was utilized as it encoded the enhanced green fluorescent protein (EGFP) downstream of an internal ribosomal entry site (IRES). A final volume of 100 μl of viral suspension (>1×106 retroviral particles) was used per injection.
Immunohistochemistry was carried out following procedures previously described (22). The primary antibodies were rabbit polyclonals raised against either the human MYC (N-262) or MCL1 protein (S-19), both from Santa Cruz Biotechnology. Details regarding TMA construction, scoring and statistical analysis are found in Supplementary Methods.
We created mouse models in which MYC expression could be induced with DOX with the aim of examining the earliest stage of MYC-induced lung tumorigenesis. Transgenic mice expressing rtTA under the control of either the human surfactant protein C (SPC) promoter (S transgene) or the rat Clara cell secretory protein (CCSP) promoter (C transgene) were used to achieve lung specific expression (Figure 1A) (18). Endogenous SPC is expressed in alveolar type II pneumocytes and CCSP in Clara cells of the upper airway. However, the fidelity of the CCSP expression pattern is not maintained by the C transgene (18). Both the S and C transgenes are transcriptionally active mainly in type II cells and the main distinction between the two transgenes appears to be the level of DOX-inducible expression that can be achieved. We bred S and C mice to mice carrying a human MYC transgene under the transcriptional control of a TRE promoter (TM transgene) (19). A diet supplemented with DOX was used to stimulate the transactivating function of the rtTA protein and induce MYC in the resulting STM and CTM mice.
Transgenic lungs were examined for MYC mRNA expression after 2, 4, and 7 days of DOX. As expected, expression of MYC was not detected in S mice (data not shown). TM and CTM mice had detectable levels of transcription from the MYC transgene (Figure 1B and data not shown) but MYC protein remained below the level of detection by Western analysis (Figure 1C). In contrast, we observed a substantial but transient rise in MYC transgene mRNA in STM mice, with a concomitant increase in MYC protein. This coincided with the appearance of alveolar hyperplasia, which was abundant in STM mice (Figure 1E) but absent from CTM mice (data not shown) where MYC expression was lower. Alveolar hyperplasia formed as clusters of cells that expanded the alveolar septa, reached a maximum after 4 days of DOX treatment and then resolved (Figure 1F). We pulse-labeled proliferating cells using BrdU. Clusters of BrdU+ cells were common in the alveolar space after 4 days of DOX treatment (Figure 1H), but not after 7 days (Figure 1I). We also detected TUNEL+ cells (Figure 1K). As alveolar hyperplasia resolved, TUNEL staining declined to near background levels (Figure 1L). We concluded that in the alveolar epithelium MYC-induced proliferation was restrained by MYC-induced apoptosis. Eradication of cells by this apoptosis could explain the transience of detectable MYC expression in response to DOX.
Reduced lifespan was observed for STM and CTM mice, both treated with DOX after weaning and in the absence of DOX, but not in a cohort of DOX-treated TM mice (Figure 2A). STM and CTM mice treated with DOX did fare worse than untreated mice, but decreases in survival did not reach statistical significance (Logrank; p=0.108 for STM and p=0.052 for CTM mice).
After 18 months only solid adenomas were found in 73.3% (n=15) of the DOX treated TM mice (Figure 2B, C, D). In contrast, STM and CTM mice developed both papillary adenomas and adenocarcinomas (Figure 2B, E-G) with DOX increasing the penetrance of adenocarcinoma from 31.6% (n=19) to 57.9% (n=19) and 23.8% (n=21) to 56.7% (n=30) in STM and CTM mice respectively (Figure 2B). Tumorigenesis was rarely multi-focal and respiratory distress was observed only when an adenocarcinoma expanded to occlude an entire lobe of the lung (Figure 2H) or the upper respiratory tract (Figure 2I). Metastases (Figure 2J) were observed only in DOX treated STM (1 of 17, 5.9%) and CTM (5 of 30, 16.7%) mice.
Our findings suggest that even MYC expression below the limit of detection by Western analysis can predispose to lung tumorigenesis. Despite the initial low levels of MYC, the resulting adenomas and adenocarcinomas both had higher levels of MYC protein than normal lung samples (100% of tumors; Figure 3A), and tumors from DOX-treated mice always had very high levels of MYC. Given the ability of MYC to elicit apoptosis in pulmonary epithelium, we reasoned that an anti-apoptotic function must emerge during the course of tumorigenesis elicited by activation of the MYC transgene. We sought to define events that occurred in lung tumors to allow bypass of MYC-induced apoptosis.
We immunostained lung tumors from STM and CTM mice for proteins expressed in lung epithelial lineages. Tumors from both genotypes resembled tumors elicited by mutant RAS (23-25) in being TTF-1+, SPC+ and CCSP− (Supplementary Figure 1). This suggested that the pathogenesis of tumors in STM and CTM mice might also involve activated RAS. We assayed RAS activity and found elevated activity in all the MYC transgenic tumors (Supplementary Figure 2). Sequencing of Ras genes revealed activating Kras mutations in 100% of the tumors (Figure 3B). Since adenomas harbored Kras mutation (4/4 adenomas; 7/7 adenocarcinomas; 0/4 control lungs), we presume that the mutations occurred before the transition from adenoma to adenocarcinoma. We concluded that mutant RAS and MYC cooperated in lung tumorigenesis in our MYC transgenic mice.
It is widely held that RAS blocks MYC-induced apoptosis through the PI3K/AKT pathway, while RAS-induced senescence is overcome by MYC-induced proliferation (26). Therefore, we reasoned that signaling downstream of activated KRAS might be responsible for abrogation of MYC-induced apoptosis in lung tumorigenesis.
In an initial effort to explore this possibility, we treated mice with the DNA-alkylating agent N-methyl-N-nitrosourea (MNU) to jump-start tumorigenesis with a synchronous wave of mutagenesis in Kras (27). As expected, all the lung tumors from MNU-treated mice harbored activating Kras mutation (data not shown). The survival of TM mice was unaffected by MNU (Figure 4A and data not shown for DOX-free mice), as these mice only developed a few adenomas (Figure 4B and data not shown). In contrast, treatment of STM mice with MNU substantially accelerated their usual rate of demise (Figure 4A; compare to Figure 2A) as adenocarcinomas developed in both the DOX treated and untreated groups. STM mice fed a DOX diet fared better than untreated mice (Figure 4A; Logrank, p<0.02) and had a significantly lower number of tumors compared to mutagenized STM mice kept on a DOX-free diet (Figure 4B; T-test, p<0.002). Therefore, MYC induction hindered, rather than aided MNU-induced tumorigenesis. These results present an apparent paradox when compared to those in Figure 2, where activation of the MYC transgene exacerbated tumorigenesis. We attribute the paradox to the different time frames of tumorigenesis in the two sets of experiments. The relatively rapid course of tumorigenesis in response to MNU might reduce the likelihood that an anti-apoptotic event could occur in time to protect developing tumor cells from the pro-apoptotic effect of MYC.
We further explored the interaction between MYC and activated RAS through retroviral transduction of alveolar hyperplasia (Figure 4C). In brief, we gave STM mice DOX for one day prior to tracheal instillation of concentrated retrovirus encoding mutant RAS. DOX induced the MYC transgene and the proliferation of alveolar cells, a prerequisite for retroviral infection of the otherwise quiescent alveolar epithelium. Transduction was not successful in DOX-treated CTM mice (data not shown), in which the level of MYC induction was below the threshold required to induce hyperplasia (see above).
Our initial proof-of-concept experiments utilized retrovirus expressing activated human HRAS and EGFP downstream of an internal ribosomal entry site so that cells expressing activated RAS would be marked. We found that activated forms of both Kras and HRAS induced identical tumor phenotypes in this assay. As with NMU-treated animals, transduction with activated RAS accelerated the demise of STM mice (Figure 4D). Continuous DOX administration increased survival (Logrank, p<0.002) and 2/5 mice remained tumor-free (Supplementary Figure 3A). In contrast, mice treated with DOX only for the 24 hours prior to transduction developed multiple adenomas and adenocarcinomas (Figure 4E, F) and adenomas could be detected as early as one month following transduction (Figure 4G). Only solitary transduced cells were found in mice treated with DOX for one month (n=4 mice) (Figure 4H). At least some of these cells maintained their tumorigenic potential, however, because removal of DOX after one month resulted in tumors (Supplementary Figure 3B) and intermediate survival (Figure 4D). The data present the same paradox in comparison with Figure 2 as observed with MNU. We presume that the same explanation applies.
Our results suggest that the anti-tumor effects of MYC are not mitigated by activated RAS. This proved to be the case both in the MNU mutagenesis experiment, where activated Kras was expressed from its endogenous promoter, and when activated RAS was ectopically expressed using retrovirus.
We reasoned that MYC transgenic tumors must harbor additional tumorigenic changes and took a candidate gene approach to search for anti-apoptotic events that could account for the ability of transgenic tumors to over-express MYC. Caspases, members of the inhibitor of apoptosis protein (IAP) family and members of the BCL2 protein family were measured by Western blotting of lysate from tumors and adjacent normal tissue (data not shown and see legend to Figure 5). This approach yielded one candidate, MCL1, an anti-apoptotic member of the BCL2 family of proteins (28). MCL1 was over-expressed in one STM and one CTM adenocarcinoma (2/13 in total) (Figure 5A and data not shown). This suggested that MCL1 over-expression could be an adaptive trait acquired by some tumors to offset MYC-induced apoptosis. Other, unknown anti-apoptotic events presumably fulfilled a similar role in transgenic tumors not over-expressing MCL1.
In order to validate that MCL1 cooperated with MYC, we utilized retroviral transduction to introduce Mcl1 or KrasG12V alone, or both KrasG12V and Mcl1 together into STM mice (Figure 5B). Retrovirus expressing only Mcl1 had no tumorigenic effect in STM mice (n=8, data not shown). Like mice transduced with HRASG12V retrovirus (Figure 4E, F), both KrasG12V and KrasG12V/Mcl1 transduced mice had a high tumor burden when placed on a DOX-free diet (Supplementary Figure 4A, B). There was no significant difference in their survival (Supplementary Figure 4C). In contrast, continuous induction of the MYC transgene (Supplementary Figure 4D) reduced the tumor burden in KrasG12V transduced mice (Figure 5D) but not in KrasG12V/Mcl1 transduced mice (Figure 5E), and the former had significantly better survival relative to the latter (Figure 5C; Logrank, p<0.05).
KrasG12V/Mcl1 tumors from DOX-fed STM mice had a mixed morphology of papillary adenocarcinoma interspersed with areas of large cell undifferentiated carcinoma (Figure 5F) that lacked the expression of differentiation markers (Supplementary Figure 5). This contrasts with the strictly papillary growth in the tumors of the original CTM and STM mice (see Figure 2G). In human NSCLC, lack of differentiation is correlated with poor prognosis (29). It appears that the KrasG12V/Mcl1 tumors from mice given DOX to induce MYC may mimic aggressive human NSCLCs.
Presumably, the effects of MCL1 stemmed from its ability to protect against MYC-induced apoptosis, as MCL1 over-expression did not alter either expression of MYC (Supplementary Figure 4D) or RAS activity (Supplementary Figure 4E). We concluded that MCL1 acted as an oncogene, but its tumorigenic potential was manifest only when MYC was over-expressed.
To explore whether the interaction between MCL1 and MYC has biological significance in human NSCLC, we evaluated their expression by immunohistochemistry. Compared to normal lung tissue, MYC and MCL1 were each over-expressed in a subset of tumors (Supplementary Table 1, Supplementary Figure 6). High expression of MYC and MCL1 was detected in 30.5% and 49.1% of the tumors, respectively. Of the MYC over-expressing tumors, 72.2% also over-expressed MCL1 (Table 1, Figure 6A). Therefore, MYC and MCL1 over-expression overlapped significantly (22% of all the NSCLCs examined).
A trend was observed for both MYC (Table 1) and MCL1 (Supplementary Table 2) to be highly expressed in NSCLCs that were less differentiated, a finding that prompted us to examine the correlation of combined expression with prognosis. Among the 59 NSCLCs with full follow-up (demographics in Table 1), neither MYC nor MCL1 expression alone was prognostic (data not shown). However, two completely different outcomes were observed when tumors expressing MCL1 were stratified by MYC expression. Whereas high expression of MCL1 was marginally associated with better survival when MYC was not expressed (p=0.0738, Figure 6B), it was significantly associated with poorer overall survival when MYC was expressed (p=0.0142, Figure 6C). The data suggest that high MCL1 and MYC expression may cooperate during the genesis of a subset of human NSCLCs and combined expression has implications for patient survival. A likely explanation for our findings is that MCL1 facilitates NSCLC progression by impeding MYC’s pro-apoptotic function.
MYC induction in STM mice led to the formation of alveolar hyperplasia. However, apoptosis soon cleared hyerplastic cells. Pleomorphic, proliferative cells have previously been noted in mice constitutively expressing MYC under control of the SPC promoter (30). These cells also succumbed to apoptosis. Therefore, both the previous study and ours suggest that MYC over-expression in the alveolar epithelium induces hyperplasia that is eliminated by apoptosis.
In CTM mice and following resolution of hyperplasia in STM mice, MYC protein levels remained below the level of detection until the emergence of tumors (data not shown). Presumably stochastic events occurred that enabled tumorigenesis. Our data point to mutation of Kras as one of these stochastic events. We presume that Kras mutation is acquired spontaneously as the animal ages. It has long been known that MYC and activated RAS have the potential to cooperate in transformation (31). However, others have found Kras mutation in only a subset of MYC-induced murine lung tumors (30) or no mutations at all (32). This is in contrast to our finding that 100% of MYC transgenic tumors had an activating mutation in Kras. We cannot account for this discrepancy, other than to note that the composition of transgenes and the genetic backgrounds of the mice differed.
We sought to examine the tumor-initiating potential of MYC-induced hyperplastic cells. Our experiments using retrovirus that encodes mutant RAS to induce lung tumors in STM mice suggest that hyperplasias are, in fact, a cell population from which tumors can emerge. Since pretreatment with DOX was required for the establishment of tumors, retroviral transduction presumably occurred in alveolar cells that responded to MYC-induction by proliferating. Thus, these experiments targeted mutant RAS expression directly to alveolar hyperplasia, supporting the view that cells with tumor-initiating potential reside within the hyperproliferative population.
Stem cells that are SPC+, CCSP+ and localized to the bronchioalveolar duct junction (BADJ) have been suggested to be the cell of origin for mouse adenomas and adenocarcinomas (33). However, in our mice hyperplasia arose at the junctions of the alveolar septa, the anatomical location of type II cells. In addition, tumors from both CTM and STM mice expressed TTF-1 and SPC, proteins expressed in type II cells, but never Clara cell marker CCSP. If type II cells did give rise to alveolar hyperplasia and eventually to tumors, our experiments suggest that the right combination of oncogenic stimuli is able to initiate tumors from cells other than BADJ stem cells.
Our experiments suggest that a small enhancement of MYC activity can augment lung tumorigenesis. This phenomenon was easily discernable in STM mice given MNU or transduced with retrovirus expressing mutant RAS. In both cases, tumorigenesis was enhanced in mice kept free of DOX. This suggests that activated RAS promoted tumorigenesis in uninduced STM mice, where MYC was expressed above endogenous levels, but not above the threshold that induces apoptosis. When STM mice were treated with DOX, however, expression of MYC was induced to high levels and tumorigenesis diminished.
Previous work has defined distinctive thresholds for the various biological outputs of MYC. In particular, it has been shown that sustained expression of the gene at little more than physiological levels can elicit potentially tumorigenic proliferation of cells in diverse tissues (9). In contrast, higher levels of expression elicit tumor suppressor activities, including senescence and apoptosis. In the present instance, however, the MYC transgenic tumors and a subset of human NSCLCs both over-expressed MYC. This raises the question as to what additional genetic/epigenetic events have occurred over time that allow bypass of MYC-induced apoptosis.
We utilized the tracheal delivery of retrovirus to STM mice to demonstrate that the combination of KrasG12V and Mcl1 was more potent than KrasG12V alone in promoting tumor growth. However, the difference was only evident when the MYC transgene was activated by DOX treatment. Our data suggest that MCL1 over-expression is an acquired trait that can promote tumor progression by impeding MYC-induced apoptosis.
Consistent with this supposition, there was a positive correlation between high MCL1 and MYC expression in human NSCLC and the co-expression of these two factors correlated with poor patient survival. No prognostic significance could be attributed to MCL1 expression in the absence of MYC, which was consistent with previous observations (16, 17). We suggest that MCL1 over-expression mediates a shift in the threshold of MYC activity required to initiate apoptosis. This enables the selection of tumor cells expressing MYC and consequent tumor progression.
Our findings have several potential implications in regard to human malignancies. First, co-expression of MCL1 and MYC may serve as a useful biomarker for identifying aggressive forms of NSCLC and for predicting patient outcomes. Second, therapeutics directed at MCL1 may display a synthetic lethal interaction with MYC by exposing cells to MYC-induced apoptosis. Inhibitors of anti-apoptotic BCL2-family proteins are now in clinical trials (34) and some are active against MCL1. Tumors that co-express MYC and MCL1 might respond preferentially to inhibition of MCL1. Conversely, inhibiting anti-apoptotic BCL2 family proteins other than MCL1 might fail in MYC-positive tumors when MCL1 levels are sufficient to protect against MYC-induced apoptosis. Third, to the best of our knowledge, this study is the first to demonstrate that MCL1 can partner with MYC in an epithelial malignancy and this provides incentive to examine interactions between MCL1 and MYC in other such tumors.
We thank Luda Urisman for her help with the mouse colony and Dr. Dean Sheppard and Dr. Jeffrey Whitsett for supplying mice.
T.D.A., J.M.B. - G.W. Hooper Research Foundation
J.M.B. – N.I.H. Grant (CA009043),
M.S.T. - Canadian Cancer Society
N.Y. – C.I.H.R.’s Terry Fox Foundation Training Program in Molecular Pathology of Cancer (STP-53912).
Conflicts of interest: None