Loss of MIG6 Accelerates Initiation and Progression of Mutant Epidermal Growth Factor Receptor–Driven Lung Adenocarcinoma

Lung cancer is the leading cause of cancer mortality in the United States, accounting for about 27% of cancer-related deaths. EGFR and KRAS are among the most commonly mutated genes associated with the initiation and maintenance of lung adenocarcinomas. The most prevalent EGFR mutations associated with lung cancer are two hotspot mutations, a leucine to arginine substitution at position 858 (L858R; 40%–45%) and an in-frame deletion mutation eliminating the conserved sequence LREA in exon 19 (e.g., Del E746–A750; 45%; refs. 1–4). These mutations render the EGFR protein–tyrosine kinase constitutively active. Lung adenocarcinomas harboring these mutations are sensitive to EGFR-directed tyrosine kinase inhibitors (TKI), such as erlotinib and gefitinib. Unfortunately, patients undergoing TKI treatment eventually develop acquired resistance. A mutation in the gatekeeper residue, T790M, accounts for 50% to 60% of acquired drug resistance (5, 6). Other mechanisms of resistance to TKIs include MET amplification, with or without concomitant T790M mutation (7, 8); HER2 amplification (9); CRKL amplification (10); NF1 loss (11); small cell lung cancer transformation (12, 13); epithelial–mesenchymal transformation (EMT; refs. 14–16); and low-frequency mutations in BRAF (17) and HER2 (18). It is therefore important to understand the signaling pathways activated downstream of mutant EGFRs in TKI-sensitive and TKI-resistant lung adenocarcinoma cells.

Aberrant EGFR signaling that leads to activation of downstream signaling components, such as AKT and ERK, is associated with increased cellular proliferation and development of cancer (19–21). Recently, several groups, including ours, have performed global phosphoproteomic profiling of lung adenocarcinoma tumor tissue from patients and in cell lines, particularly TKI-sensitive lung adenocarcinoma cell lines, and have identified a large number of sites that are tyrosine phosphorylated (22, 23). We previously employed stable isotope labeling with amino acids in cell culture (SILAC) and quantitative phosphoproteomics to elucidate the differences in use of phosphorylation targets of wild-type and mutant EGFRs in isogenic human bronchial epithelial cells (24). One of the candidates that was hyper-phosphorylated on tyrosines in cells expressing mutant EGFRs was MIG6 (gene symbol ERRFI1, also known as RALT; Gene 33), an immediate early response gene that is induced by growth factors, including EGF and stress stimuli (25, 26). MIG6 functions as a negative feedback regulator of ERBB family members, including EGFR and ERBB2 (27). Ablation of Mig6 in mice leads to tumors of various tissues, including lung, implicating Mig6 as a potential tumor suppressor gene (28–30). Several studies have reported that MIG6 inhibits EGFR by blocking its kinase activity, as well as by promoting its degradation (29, 31, 32). It has also been demonstrated that MIG6 RNA is increased in EGFR-mutant lung adenocarcinoma cell lines (33). These observations raise questions about whether MIG6 is a tumor suppressor for mutant EGFR–driven lung adenocarcinoma and, if so, how mutant EGFR induces lung adenocarcinomas in the presence of MIG6.

In this study, we sought to establish whether MIG6 deficiency would accelerate tumorigenesis induced by the common mutant alleles of EGFR, thus demonstrating its tumor-suppressive role. We generated doxycycline-inducible mutant EGFR transgenic mice on different Mig6 genetic backgrounds and demonstrate that MIG6 deficiency accelerates the initiation and progression of mutant EGFR–driven tumorigenesis in vivo. MIG6 also functions as a haploinsufficient tumor suppressor in this model. To further examine the mechanisms of tumor suppression by MIG6 and to elucidate how mutant EGFR can circumvent MIG6 function in human lung tumors, we studied the consequences of tyrosine phosphorylation of MIG6 in human cancer cell lines. Using global quantitative mass spectrometry–based phosphoproteomics, we identified Y394/Y395 as constitutively phosphorylated sites on MIG6 in lung adenocarcinoma cells expressing mutant EGFRs; these sites are inhibited by erlotinib in TKI-sensitive lung adenocarcinoma cells but not in drug-resistant cells. Increased phosphorylation of MIG6 increases the interaction of mutant EGFRs and MIG6. However, contrary to its effects on wild-type EGFR, MIG6 does not promote degradation of mutant EGFR. We propose a model in which mutant EGFR may circumvent the tumor suppressor function of MIG6 by constitutively phosphorylating Y394/Y395. However, the attenuated inhibitory function of MIG6 in the context of mutant EGFRs is still sufficient to delay tumorigenesis in a mouse model of mutant EGFR–driven lung adenocarcinoma.

Tissue-specific knockout of Mig6 increases EGFR signaling and the proliferation of epithelial cells in mouse lungs, suggesting that MIG6 is essential for lung homeostasis (34). Deletion of Mig6 in mice also promotes adenomas and adenocarcinomas in the lung, gallbladder, and bile duct, albeit at low penetrance (30). However, the role of MIG6 in mutant EGFR–driven lung tumorigenesis has not been studied. To test this, we crossed Mig6 heterozygous mice (Mig6^+/−; ref. 30) with doxycycline-inducible mutant EGFR transgenic mice (tetO-EGFR^mut; ref. 35) and CCSP^rtTA mice (36). The resulting tetO-EGFR^mut/Mig6^+/− and CCSP^rtTA/Mig6^+/− mice were further bred to generate transgenic mice with conditional, doxycycline-inducible expression of EGFR^L858R or EGFR^Del in type II lung epithelial cells in Mig6^+/+, Mig6^+/−, and Mig6^−/− backgrounds. After induction of transgenic mutant EGFRs, we monitored mice for the appearance of lung tumors by serial MRI. CC10^rtTA/EGFR^L858R/Mig6^−/− mice developed tumors earlier than CC10^rtTA/EGFR^L858R/Mig6^+/+ mice (Fig. 1A; Supplementary Fig. S1A). The same was true for CC10^rtTA/EGFR^Del/Mig6^−/− mice (Supplementary Fig. S1B). The Mig6^−/− mice carrying mutant EGFR transgenes were euthanized earlier than mice without EGFR transgenes because of progressive disease. The Mig6^−/− mice without the EGFR transgene had to be euthanized between 3 and 6 months of age, not due to lung tumor formation, but because of osteoarthritis affecting food intake (data not shown). Although there were transgenic line–specific differences, histopathology of the tumors at the survival endpoint indicated a higher incidence of adenocarcinoma in Mig6^−/− mice compared with Mig6^+/+ mice (Fig. 1B and Supplementary Fig. S1C, Table). Lungs of CC10^rtTA/EGFR^L858R/Mig6^+/+ mice showed only pulmonary adenomas or adenomas with infrequent adenocarcinomas. There were no signs of invasion. The surrounding alveolar compartment showed type II cellular hyperplasia and variable amounts of macrophages (Supplementary Fig. S1D; A–C). The neoplastic lesions induced by both EGFR mutants in Mig6^+/− or Mig6^−/− mice were more advanced with features of adenocarcinoma (Supplementary Fig. S1E; A–C). Lungs were often completely effaced with hyperplastic and dysplastic alveolar type II epithelial cells and had intense infiltration of macrophages and other inflammatory cells. These mice also demonstrated marked abnormalities of the airway lining epithelium with Clara cell hyperplasia or dysplasias and proliferation at bronchioalveolar duct junctions of the terminal bronchioles (Supplementary Fig. S1E; D and E).

To examine the effect of Mig6 deletion on the survival of mice harboring mutant EGFRs (EGFR^L858R and EGFR^Del), we generated tumors by doxycycline induction of mutant EGFR transgenes and euthanized the mice when they displayed specific criteria related to lung tumor burden, such as labored breathing, weight loss, and failure to thrive. We performed Kaplan–Meier survival analyses in two separate transgenic lines of each of the mutant EGFRs in Mig6^+/+, Mig6^+/−, and Mig6^−/− backgrounds (Fig. 1C–F). In all the tested lines of mutant EGFR mice, the survival time of Mig6^−/− mice was significantly shorter than that of Mig6^+/+ mice. The median survival of CCSP-rtTA/TetO-EGFR^L858R/Mig6^−/− mice after doxycycline induction was 13 days (both lines) compared with 60 (line 57) to 100 days (line 56) for CCSP-rtTA/TetO-EGFR^L858R/Mig6^+/+ mice (Fig. 1C and D). The median survival of CCSP-rtTA/TetO-EGFR^Del/Mig6^−/− mice after doxycycline induction was 16 (line 9) to 45 days (line 11) compared with 143 (line 9) to 337 days (line 11) for CCSP-rtTA/TetO-EGFR^Del/Mig6^+/+ mice (Fig. 1E and F). Interestingly, the median survival of Mig6^+/− mice expressing EGFR^Del in EGFR^DelL9 (line 9) was only 34.5 days after doxycycline induction and was significantly shorter than that of the Mig6^+/+ mice. A similar trend was observed in EGFR^DelL11 (line 11) mice, although the EGFR^Del-driven tumors appeared later than EGFR^L858R-driven tumors. Thus, the survival of mice with EGFR^Del-induced tumors in a Mig6^+/− background appears more curtailed than in a Mig6^+/+ background due to the longer latency of tumor induction by transgenic EGFR^Del. We euthanized littermates with various genotypes 9 days after doxycycline induction to demonstrate the possible early appearance of tumors in Mig6-deficient mice. Although there was mild hyperplasia of type II cells in Mig6^+/+ mice (Fig. 2A), Mig6^+/− mice showed intermediate histopathology with increased type II hyperplasia, adenomas, and adenocarcinomas (Fig. 2B; Supplementary Fig. S2A; A–D), and Mig6^−/− littermates exhibited dramatic effacement of lung alveoli with diffuse adenocarcinoma (Fig. 2C). We performed immunohistochemistry on lung tissue sections from Mig6^+/+, Mig6^+/−, and Mig6^−/− mice 9 days following doxycycline induction, using antibodies against TTF1 (a type II epithelial cell marker and hence also a marker for lung cancer cells), EGFR^L858R, the proliferation marker Ki67, and phosphorylated ERK (pERK). There was an increase in TTF1, EGFR^L858R, Ki67, and pERK immunoreactive cells in both Mig6^+/− and Mig6^−/− mice, compared with the lungs of Mig6^+/+ mice (Fig. 2D–O), confirming significantly increased lung tumor burden 9 days following doxycycline induction of mutant EGFRs in the Mig6-deficient background.

Loss of Mig6 cooperates with loss of Pten for endometrial cancer initiation and progression in a mouse model (37). This acceleration of tumorigenesis was shown to be due to the prevention of apoptosis. Normal mammary gland development in Mig6-null mice demonstrated that MIG6 promotes apoptosis in terminal end buds (38). MIG6 has also been shown to be an inducer of replicative or oncogene-induced senescence in fibroblasts (39–41). To further investigate whether mutant EGFR-driven tumor cells in Mig6^−/− mice escape apoptosis and/or senescence, we performed terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) staining and senescence-specific p19^ARF staining (42, 43) of lung tissue sections from Mig6^+/+, Mig6^+/−, and Mig6^−/− mice 9 days after doxycycline induction. However, we did not observe any significant apoptosis or senescence in mutant EGFR–induced mouse lung tumorigenesis (Supplementary Fig. S2B; A–H), further confirming that the rapid progression of mutant EGFR-driven tumorigenesis is likely a result of increased proliferation as a result of increased ERK/MAPK pathway activation.

Because we observed that median survival of EGFR^L858R/Mig6^−/− mice is about 13 days, we performed hematoxylin and eosin (H&E) staining and immunohistochemistry on lung tissue sections from Mig6^+/+, Mig6^+/−, and Mig6^−/− mice 7 days following doxycycline induction, using antibodies against TTF1 and EGFR^L858R. We also observed an increase in TTF1 and EGFR^L858R immunoreactive cells in Mig6^−/− mice, relative to lungs of Mig6^+/− and Mig6^+/+ mice both in EGFR^L858R Line 56 (Supplementary Fig. S3A; A–I) and EGFR^L858RLine 57 (Supplementary Fig. S3B; A–I).

We then investigated whether there was an LOH of the Mig6 gene in mutant EGFR–driven lung tumors developing in Mig6^+/− mice. We performed quantitative genomic PCR for Mig6 copy number in lung tumor DNA and compared it with that of germline copy number in tail DNA. Our results showed that the Mig6 copy number did not change in lung tumors from a group of Mig6^+/− mice that had to be euthanized early, arguing against LOH (Supplementary Fig. S4). Hence, MIG6 is a haploinsufficient tumor suppressor for initiation and progression of tumorigenesis, at least in the context of expression of transgenic EGFR^Del in mice.

To examine whether accelerated tumor growth in Mig6^−/− mice is correlated with increased phosphorylation and activity of transgenic EGFR mutants, we measured the levels and phosphorylation status of mutant EGFR proteins in mouse lung tissue lysates and lung tissue sections by immunoblotting and immunohistochemistry. Early after doxycycline induction, cells reactive with TTF1 and EGFR^L858R antisera were significantly more abundant in lungs from Mig6^−/− and Mig6^+/− as opposed to Mig6^+/+ mice (Fig. 2D–I), indicating early tumor initiation due to MIG6 deficiency. This was also associated with increased immunoreactivity to Ki67 and pERK, suggesting increased proliferation due to activation of the MAPK pathway (Fig. 2J–O). However, we found both by immunohistochemistry and Western blotting that EGFR^L858R levels were paradoxically reduced in tumors collected at survival endpoint from Mig6-deficient animals (Fig. 3A and B). When adjacent tumor sections were stained with anti-TTF1 and anti-EGFR antibodies, levels of EGFR^L858R were significantly lower in TTF1⁺ cells from EGFR^L858R/Mig6^−/− mice than in those from EGFR^L858R/Mig6^+/+ mice (Fig. 3B).

To confirm this unexpected result, we immunoprecipitated EGFR from mouse tumor lysates and examined the immunoprecipitated EGFR with anti-pEGFR (Y1068), anti-EGFR^L858R, and anti-EGFR antibodies (Fig. 3C and D). Although the levels of EGFR^L858R were reduced in established tumors, the levels of pEGFR remained high; the ratio of pEGFR to total EGFR was higher in tumors from Mig6^−/− mice than in tumors from Mig6^+/+ or Mig6^+/− mice (Fig. 3D). We also observed decreased EGFR^Del levels in tumors from EGFR^Del/Mig6^−/− mice compared with protein levels in tumors from EGFR^Del/Mig6^+/+ or EGFR^Del/Mig6^+/− mice at the survival endpoint (Fig. 3E). Likewise, as with Mig6^−/− mice expressing the EGFR^L858R transgene, the pEGFR:EGFR ratio was higher in the EGFR^Del/Mig6^−/− mice (Fig. 3F). Thus, the higher proportion of pEGFR appears to be sufficient for tumor maintenance in Mig6^−/− mice.

We further measured components of the MAPK and PI3K signaling pathways in established tumors. Immunoblots of whole lung extracts from Mig6^+/+, Mig6^+/−, and Mig6^−/− mice expressing either EGFR^L858R or EGFR^Del showed highly variable levels of pERK and pMEK. The level of pERK was higher in whole lung lysates from Mig6^−/− mice than from Mig6^+/+ or Mig6^+/− mice, particularly those with the transgene encoding EGFR^Del (Fig. 3G and H). Furthermore, immunohistochemistry demonstrated increased pERK reactivity in distinct, comparable areas of lung tumors in Mig6^−/− mice compared with those from Mig6^+/+ mice (Fig. 3I–L). Interestingly, we observed increased levels of both p4EBP and 4EBP protein synthesis factor in Mig6^−/− mice (Fig. 3G and H).

The reduced levels of mutant EGFR in established lung tumors at survival endpoint was surprising. We next determined whether this reduction in protein levels could be due to reduced levels of EGFR RNA. We examined whole-lung lysates for levels of Mig6, transgenic human EGFR, mouse Egfr, rtTA, and mouse Sftpc RNAs in tumors obtained at the survival endpoint (Supplementary Fig. S5). We confirmed complete loss of Mig6 transcripts in Mig6^−/− mice (Supplementary Fig. S5A). Although the transcript level of endogenous mouse Egfr was the same regardless of MIG6 status, at this late time-point, the levels of the human EGFR transgene (EGFR^L858R or EGFR^Del) RNAs were slightly lower in tumors from Mig6^−/− mice (Supplementary Fig. S5B and S5C). This was accompanied by a slight decrease in transgenic rtTA RNA levels (Supplementary Fig. S5D) but no significant change in levels of Sftpc RNA (Supplementary Fig. S5E). We also examined transcript levels in lung lysates obtained early (9 or 14 days) after doxycycline induction of mutant EGFRs and found no significant differences in EGFR^L858R or rtTA expression at these early time points among mice with different Mig6 genotypes (Supplementary Fig. S6A–S6E).

As part of a larger screen to identify proteins that may be regulated by tyrosine phosphorylation downstream of mutant EGFRs in lung adenocarcinoma cells, we employed a global phosphoproteomic approach in PC9 and H1975 cells that express TKI-sensitive EGFR^{Del(E746–A750)} and TKI-resistant EGFR^L858R+T790M, respectively. Although EGFR mutants are constitutively active in these cells, they can be further stimulated by EGF treatment. The cells were metabolically labeled with “light, medium, or heavy” isotope forms of arginine and lysine to measure the relative abundance of tyrosine phosphorylation in cells expressing the EGFR mutants in three states: serum-starved (“light”), EGF-stimulated (“medium”), or EGF-stimulated with prior erlotinib treatment (“heavy”; Fig. 4A). We performed this triple-SILAC experiment to identify the changes in tyrosine phosphorylation by maximal activation of mutant EGFR signaling and inhibition by TKIs. We expected increased tyrosine phosphorylation upon EGF stimulation in both PC9 and H1975 cells. However, although the identified phospho-sites were expected to show decreased phosphorylation after erlotinib pretreatment in the sensitive PC9 cells, they were predicted to remain unchanged in the erlotinib-resistant H1975 cells. Phospho-sites showing such changes in phosphorylation are the likely targets for phosphorylation by mutant EGFR. We identified a total of 309 and 488 unique tyrosine phospho-sites with quantitative phosphorylation data in PC9 and H1975 cells, respectively (data not shown). Based on this screen, a large number of phosphopeptides were identified as potential targets of mutant EGFRs.

MIG6 was identified as a target of mutant EGFR kinase in both lung adenocarcinoma cell lines examined. Y394 or both Y394/Y395 sites were found to be constitutively phosphorylated in these cells. A representative mass spectrometry (MS) spectrum demonstrates that the relative abundance of the phosphorylated Y394-containing MIG6 peptide from PC9 cells decreased significantly upon erlotinib treatment (Fig. 4B). There was no significant change in phosphorylation at Y394/Y395 sites in H1975 cells upon erlotinib treatment (Fig. 4C). MS-MS analysis of this peptide revealed that the Y394 residue was phosphorylated in PC9 cells, whereas both Y394 and Y395 were phosphorylated in H1975 cells (Supplementary Fig. S7). To validate whether Y394/Y395 is the major site of MIG6 phosphorylation in vivo, we carried out phosphoproteomic analysis of pTyr peptides from tumor lysates of EGFR^L858R transgenic mice and identified the MIG6 phosphopeptide with phosphorylation at the Y394 residue (Fig. 4D). Using purified proteins for in vitro kinase assays, it has been shown that EGFR can phosphorylate MIG6 directly (44). Our results in this study and in human bronchial epithelial cells (HBEC) expressing mutant EGFRs (24) provide in vivo evidence consistent with three conclusions: that MIG6 is a direct target of mutant EGFR; that Y394 and Y395 are sites constitutively phosphorylated by mutant EGFRs; and that erlotinib inhibits such phosphorylation in TKI-sensitive cells, but not TKI-resistant cells.

We have demonstrated that MIG6 residues Y394 and Y395 are constitutively phosphorylated in mutant EGFR–expressing lung adenocarcinoma cells. Using in vitro kinase assays with purified proteins, others have shown that EGFR can directly phosphorylate MIG6 on tyrosines (44). A 77–amino acid region of MIG6 segment 1 (aa336–412) has been shown to be necessary for EGFR inhibition (31). The structural determinants of MIG6 required for binding to EGFR have been previously mapped to an EGFR-binding region (EBR), spanning residues 323 to 411 at the C-terminus of MIG6 protein (31, 45). The binding domain contains six tyrosine residues at positions 341, 358, 394, 395, 403, and 407, all of which have been shown to be phosphorylated in mass spectrometry–based experiments (data from Phosphosite Plus; refs. 24, 46). The crystal structure of the EGFR kinase domain bound to part of segment 1 of MIG6 indicates that Y358 resides in the binding interface, and mutation of this residue to alanine (Y358A) disrupts binding (31). Frosi and colleagues (32) also showed that, unlike wild-type (WT) MIG6, the Y358A mutant failed to promote endocytosis of EGFR, indicating that the Y358 residue is important for MIG6 function. However, Y358 is not a major site of phosphorylation in vivo. Moreover, the effect of phosphorylation of the major sites, Y394/Y395, on MIG6 function has not been studied in detail. Recently, using purified proteins and in vitro kinase assays, it has been shown that phosphorylation of MIG6^Y394 reduces the inhibitory function of MIG6 on the EGFR kinase (47). We postulate that phosphorylation of Y394/Y395 and other tyrosine sites within the EBR domain of MIG6 affects the binding of EGFR and MIG6 and regulation of EGFR kinase activity by MIG6.

To determine whether phosphorylation of MIG6 at Y394 and Y395 is important for the interaction of MIG6 with EGFR, we replaced these tyrosines with phenylalanine to mimic unphosphorylated tyrosine. Expression vectors containing WT or mutant MIG6 cDNAs were cotransfected into HEK293 cells with vectors containing WT EGFR, EGFR^L858R, or an empty vector as a control. After serum starvation for 18 hours, some cultures were stimulated with EGF for 10 minutes. Cell extracts were examined for interacting proteins by immunoprecipitation with EGFR-specific antibodies, followed by Western blotting with second antibodies against MIG6 and other proteins (Fig. 5A). These studies indicated that MIG6 interacts with both WT and mutant EGFR. EGF stimulation increased the interaction of MIG6 with WT EGFR. Furthermore, MIG6 interacted more efficiently with EGFR^L858R than with WT EGFR in unstimulated cells, and the increase was associated with increased Tyr phosphorylation of MIG6. More importantly, mutation of residues Y394/Y395 to phenylalanine, abolishing phosphorylation of these sites, impaired the ability of MIG6 protein to bind both WT and mutant EGFR proteins. However, it is possible that Y-F mutants may affect hydrogen-bonding interactions due to the loss of a hydroxyl in the phenylalanine, and hence influence interactions beyond just the loss of phosphorylation.

Because HEK293 cells express endogenous MIG6, we also performed coimmunoprecipitation studies in H322M cells, a lung adenocarcinoma cell line that does not express detectable levels of endogenous MIG6 because of a homozygous MIG6 nonsense mutation, E83Stop (30). H322M cells stably producing WT or Y-F MIG6 mutants were transiently transfected with either WT EGFR or EGFR^L858R expression vectors and analyzed by EGFR coimmunoprecipitation assays (Fig. 5B). These experiments demonstrated a reduced interaction of MIG6 Y-F mutants with both WT EGFR and EGFR^L858R, except for the MIG6^Y358F mutant. To examine the overall tyrosine phosphorylation of MIG6, we immunoprecipitated MIG6 from lysates of H322M cells transfected with WT EGFR or EGFR^L858R and probed the blots with pTyr and FLAG (MIG6) antibodies. There was increased tyrosine phosphorylation of MIG6 in EGFR^L858R-expressing cells. Y394/Y395 were again demonstrated to be the predominant sites of tyrosine phosphorylation in MIG6 (Fig. 5C).

Previous studies have shown that EGF promotes activation-dependent endocytosis and degradation of EGFR, potentially regulating the duration of downstream signaling (48, 49). However, mutant EGFRs are ineffectively internalized (50, 51) and exhibit diminished downregulation following ligand activation (52, 53). To assess differences in the degradation of WT EGFR and mutant EGFR, we first used isogenic HBECs. Lysates from HBECs treated with EGF and cycloheximide for 0 minutes, 10 minutes, 30 minutes, 1 hour, 2 hours, and 3 hours were immunoblotted with antibodies that specifically recognize pan-EGFR (both WT and L858R EGFR), EGFR^L858R, and EGFR^L858 (epitope surrounding the L858 residue in WT EGFR; Supplementary Fig. S8). Substantial degradation of WT EGFR occurred within 2 hours of EGF-induced activation of EGFR. In contrast, degradation of EGFR^L858R was significantly reduced in HBECs expressing the mutant EGFR (Fig. 6A), indicating that EGFR^L858R is more resistant to degradation than WT EGFR upon EGF stimulation.

MIG6 inhibits EGFR kinase activity and promotes WT EGFR trafficking to the degradation pathway (29, 32). Kinetic modeling based on EGFR endocytosis experiments performed on lung adenocarcinoma cells expressing WT and mutant (Del 746–750) EGFRs has suggested that MIG6 promotes WT EGFR, but not mutant EGFR, internalization (54, 55). To investigate the effect of MIG6 on EGF-induced degradation of EGFR, we performed an EGFR degradation assay with cycloheximide on HEK293 cells stably transfected with WT or Y394/Y395F MIG6, together with WT EGFR or EGFR^L858R (Fig. 6B) and quantified retained EGFRs (Fig. 6C). As predicted, WT EGFR was degraded upon EGF stimulation. However, there was little effect of EGF on mutant EGFR degradation. Furthermore, expression of MIG6 had no effect on the degradation of mutant EGFR, and it slowed the degradation of activated WT EGFR. To ascertain whether WT EGFR degradation utilizes the lysosomes, we performed these degradation experiments in the presence of chloroquine, a lysosomotropic inhibitor. We observed that chloroquine significantly delayed degradation of WT EGFR (Fig. 6D). However, there was no additional effect of chloroquine on the retention of EGFR^L858R, suggesting that the lysosomal pathway is utilized for WT EGFR, but probably not for mutant EGFR degradation (Fig. 6E). We further performed these degradation assays in WT or Y394/Y395F MIG6-transduced H322M lung adenocarcinoma cells that contain WT EGFR and no endogenous MIG6. WT EGFR was efficiently degraded upon EGF stimulation in the absence or the presence of Y394/Y395F MIG6; however, WT EGFR appeared more stable in the presence of MIG6, findings similar to those observed in HEK293 cells (Fig. 6F).

MIG6 is known to inhibit EGFR kinase activity and promote the degradation of WT EGFR, so it is considered a suppressor of tumors with active WT EGFR signaling. However, the role of MIG6 in regulating lung cancer–specific mutant EGFRs has not been studied in detail. Here, we show for the first time that Mig6 deficiency, even haploinsufficiency, accelerates the initiation and progression of tumorigenesis and lethality driven by mutant EGFR in mouse models. We have previously shown in isogenic HBECs that mutant EGFR enhances tyrosine phosphorylation of MIG6 more efficiently than WT EGFR (24). Here, we demonstrate constitutive phosphorylation of MIG6 at Y394/Y395 in lung adenocarcinoma cells harboring EGFR mutations. Phosphorylation at these sites is inhibited by erlotinib in TKI-sensitive, but not TKI-resistant, cells, suggesting that mutant EGFRs directly phosphorylate and possibly regulate MIG6. This conclusion is consistent with published evidence that purified EGFR protein can directly phosphorylate MIG6 in vitro (44). We have further examined the functional consequences of tyrosine phosphorylation of Y394/Y395 on MIG6 and how this affects its tumor suppressor function. We show that phosphorylation increases binding of MIG6 to mutant EGFRs; but, in contrast to WT EGFR, the increased interaction does not direct mutant EGFR to the degradation pathway. Our observation is consistent with another study in which MIG6 was shown to be a poor inhibitor of the kinase activity of nearly full-length mutant EGFR in vitro (44). A phosphorylated Y394-containing fragment of MIG6 was also shown to be a poor kinase inhibitor of WT EGFR compared with its unphosphorylated counterpart using kinase assays with purified proteins in vitro (47). Taken together, our findings of constitutive phosphorylation of MIG6 Y394/Y395 in lung adenocarcinoma cell lines and the increased binding of MIG6 with mutant EGFRs (possibly leading to stability of mutant EGFR), along with the published in vitro studies of inadequate EGFR kinase inhibition by phosphorylated MIG6, provide strong evidence that mutant EGFRs can partially circumvent inhibition by MIG6 in lung adenocarcinoma cells through tyrosine phosphorylation of MIG6 on key residues. However, most importantly, we show that the residual inhibitory activity of MIG6 is still tumor suppressive in mutant EGFR-driven lung tumor models, because MIG6 deficiency reduces survival of mice due to accelerated tumorigenesis. This was not expected based on the results of enzymologic studies published to date (44, 47). It is also possible that MIG6 is capable of inhibiting the formation of heterodimers of mouse WT EGFR and human mutant EGFR, a likely scenario in the early stages of doxycycline induction in our mouse model, especially when levels of transgenic mutant EGFR are low. A schematic of our model of inhibitory activity of MIG6 against WT and mutant EGFRs is depicted in Fig. 7A and B.

The accelerated initiation and progression of mutant EGFR–driven tumorigenesis in Mig6-deficient background was quite striking. The manifestation of early tumorigenesis was dramatic in the early time periods of doxycycline induction of mutant EGFRs. There was almost complete effacement of normal alveoli by type II cells 7 to 9 days after doxycycline induction in Mig6^−/− mice, the earliest times at which tissue was analyzed in this study, at a time when only focal type II cellular hyperplasia was observed in Mig6^+/+ littermates. Increased proliferation was associated with increased pERK immunoreactivity in the lungs of Mig6^−/− mice. Interestingly, our experiments also demonstrate a statistically significant difference in the survival of Mig6^+/− compared with Mig6^+/+ mice in the presence of an EGFR^Del transgene, at least in line 9, with a strong trend toward decreased survival in EGFR^Del line 11 mice. In two EGFR^L858R mouse lines, we saw a statistically insignificant trend toward decreased survival for Mig6^+/− mice. The EGFR^Del mice develop tumors much later than EGFR^L858R mice, which may allow for the difference in survival between the Mig6^+/+ and Mig6^+/− mice to manifest. We did not observe Mig6 LOH in a select group of Mig6^+/− mice we studied. However, it is still possible that there could be LOH in individual mice that we did not analyze. The fact that there is strong evidence of MIG6 being a haploinsufficient tumor suppressor in our studies has implications in mutant EGFR-driven human lung adenocarcinoma biology. A recent study evaluated the relative expression level of MIG6 and EGFR in a small cohort of patients with lung cancer treated prospectively with gefitinib. This study concluded that a lower MIG6:EGFR ratio is associated with sensitivity to TKIs, whereas a higher MIG6:EGFR ratio is a predictor of TKI resistance (56). In another study, the ratio of MIG6 and miR200c RNA levels correlated with EMT and resistance to erlotinib (57). However, these studies were performed primarily with patients harboring WT EGFR. Further prospective clinical studies are warranted to ascertain whether absolute MIG6 levels can influence the initiation, progression, and EGFR-TKI response in mutant EGFR-driven human lung adenocarcinoma.

We saw comparable induction of steady-state levels of transgenic mutant EGFR transcripts in the early time periods of doxycycline induction, regardless of MIG6 status. However, there were more TTF1 and EGFR^L858R immunostained cells at these time periods, consistent with enhanced stimulation of growth by mutant EGFR and early tumorigenesis in Mig6-null transgenic mice. Surprisingly, mutant EGFR protein levels in tumor lysates collected at the survival endpoint were significantly lower in Mig6^−/− mice than in Mig6^+/+ mice. We demonstrated that the low level of transgenic EGFR protein was still hyperphosphorylated in these late-stage tumors, and the pEGFR:EGFR ratio was higher in tumors from Mig6^−/− mice than in tumors from Mig6^+/+ mice, suggesting that the residual mutant EGFR activity was sufficient to maintain these aggressive tumors. We did not find any difference in the mutant EGFR and rtTA transcript levels at earlier time periods. However, there was a slight decrease of rtTA and mutant EGFR transcript expression in the late-stage tumors. This could be a result of decreased CCSP promoter activity in the lungs of older Mig6^−/− mice, reducing mutant EGFR transcripts. This is corroborated by the fact that MIG6 is essential for normal lung development (34). However, the modest decrease in rtTA or mutant EGFR mRNA does not explain the reduced levels of mutant EGFR protein. We speculate that during the progression of tumorigenesis, there is selection for lower transgenic EGFR–expressing cells. However, the residual mutant EGFR signal strength is still enough to maintain these aggressive tumors. We also speculate that MIG6 inhibits heterodimers of mutant and WT EGFR and promotes their degradation in the early stages of doxycycline induction of mutant EGFRs in these models, thus explaining the dramatic tumor-suppressive role. At later stages, because of increased transgenic mutant EGFR levels, the mutant EGFRs exist predominantly as homodimers. MIG6 is unable to traffic these homodimers to degradation pathways because of increased feedback tyrosine phosphorylation; instead, MIG6 binds more strongly and stabilizes mutant EGFR homodimers at this stage. Hence, MIG6 deficiency results in lower levels of transgenic mutant EGFRs in end-stage tumors.

We postulate a two-pronged mechanism by which mutant EGFRs dampen inhibition by MIG6; one acts to regulate the levels and the other modulates the function of MIG6. The first is by a downregulation of MIG6 protein levels similar to the regulation of a classic tumor suppressor. In a recent study, lung cancer–specific EGFR mutations correlated with loss of MIG6 protein; 12 of 16 EGFR-mutant tumors lacked MIG6 protein (58). At least one lung adenocarcinoma cell line, H322M, harbors a homozygous nonsense mutation in MIG6, with undetectable MIG6 protein in the context of WT EGFR expression (30). Around 50% of primary glioblastoma multiforme (GBM) tumor samples and cell lines have reduced MIG6 RNA and protein expression (29). MIG6 levels may also be regulated by epigenetic mechanisms. MIG6 promoter methylation was observed in 79% of papillary thyroid carcinomas (59), and histone deacetylation (HDAC) inhibition upregulates MIG6 in lung cancer cell lines (60). We confirmed that MIG6 functions as a potent tumor suppressor in the initiation and progression of tumorigenesis induced by mutant EGFRs in mouse models. Furthermore, our studies indicate that MIG6 can act as a haploinsufficient tumor suppressor in the context of mutant EGFRs.

The second mechanism of reduced MIG6 inhibitory function is the increased tyrosine phosphorylation of Y394/Y395 by mutant EGFRs, leading to decreased kinase inhibition (44, 47) and increased constitutive binding of MIG6 to mutant EGFRs, possibly stabilizing mutant EGFRs presented in this study. Recently, Ying and colleagues (29) also observed that the EGFRvIII mutant does not undergo MIG6-mediated endocytosis and degradation in lysosomes of GBM cell lines, unlike WT EGFR. A recent study showed a modest decrease in EGFR internalization upon MIG6 knockdown in PC9 cells, a lung adenocarcinoma cell line harboring the EGFR^Del mutant (54). The study by Walsh and colleagues does not distinguish between mutant EGFR and WT EGFR in PC9 cells, which are heterozygous for the EGFR^Del mutant. Furthermore, we also noticed that mutant EGFRs could be effectively internalized into early endosomes in HBECs (Supplementary Fig. S9A–B). Interestingly, we found strong colocalization of mutant or WT EGFR and MIG6 in discrete vesicles upon EGF stimulation of HBECs. However, there was less colocalization of mutant EGFR compared with WT EGFR and LAMP1, a lysosomal marker even after 2 hours of EGF stimulation, suggesting that mutant EGFR may not traffic through the lysosomal degradation pathway (Supplementary Fig. S9C–D). Our data suggest that mutant EGFR degradation is inhibited in spite of the increased interaction of MIG6 and mutant EGFRs (Fig. 6 and Supplementary Fig. S9).

Prospective biomarker-validation studies are warranted to establish the role of MIG6 expression or phosphorylation in the overall prognosis of patients harboring WT EGFR or mutant EGFRs. Such clinical studies are needed to ascertain whether absolute MIG6 levels can influence the initiation, progression, and EGFR-TKI response in mutant EGFR-driven lung adenocarcinoma.

Additional methods are described in the Supplementary Materials and Methods section.

RPMI and DMEM tissue culture media and FBS were obtained from Invitrogen. Defined FBS for H3255 adenocarcinoma cell culture was obtained from Hyclone. All chemicals were obtained from Sigma-Aldrich, unless stated otherwise. Fugene X-tremeGENE 9 DNA transfection reagent, complete minitab protease inhibitor, and PhosStop phosphatase inhibitor were obtained from Roche Applied Science. Nitrocellulose Western transfer sandwich was obtained from Invitrogen, and nitrocellulose membrane was obtained from GE Healthcare Life Sciences. EGF was obtained from Millipore and Peprotech. The tyrosine kinase inhibitor erlotinib was obtained from Beta Pharma, Inc. Mouse anti-MIG6 mAb was a kind gift from Dr. Oreste Segatto (Regina Elena Cancer Institute, Italy) and was also obtained from Abnova. Mouse anti-EGFR mAb was obtained from BD Biosciences. Rabbit polyclonal antibodies to EGFR^{Del(E746–A750)}, pEGFR (Y1068), EGFR, AKT, ERK, RSK, 4EBP, pAKT, pERK, pRSK, and p4EBP, as indicated in the figures, were obtained from Cell Signaling Technology. Rabbit polyclonal antibodies to MIG6 (H125) were obtained from Santa Cruz Biotechnology. Mouse mAbs to EGFR^L858R (18D1), EGFR^L858 (9D3), and EGFR^E746 (13D6) were made in collaboration with nanoTools. Anti-EEA1 and anti-LAMP1 antibodies were obtained from Abcam and Cell Signaling Technology, respectively. Anti-TTF1 antibody was obtained from Dako, Inc., and rabbit anti–Rho-GDI polyclonal antibodies, protein A and G sepharose, were obtained from Sigma. Ki67 and p19^ARF-specific antibodies were obtained from Abcam. TUNEL staining was performed using the ApopTag Peroxidase In Situ Apoptosis Detection Kit (Millipore).

H1975 and HEK293 cell lines were purchased from the ATCC, the PC9 cell line was obtained from the Varmus Laboratory, and the H322M cells were obtained from the Division of Cancer Treatment and Diagnosis (DCTD) Tumor Cell Line Repository (NCI, Frederick, MD). All human lung adenocarcinoma cells were maintained in RPMI supplemented with 10% FBS, 100 units/mL penicillin, and 100 μg/mL streptomycin. The human embryonic kidney cell line HEK293 was cultured in DMEM supplemented with 10% FBS, 100 units/mL penicillin, and 100 μg/mL streptomycin. The HBECs were a kind gift from Dr. John D. Minna (University of Texas Southwestern, Dallas, TX), and were maintained in keratinocyte serum-free medium (Invitrogen) supplemented with bovine pituitary extract (BPE) and EGF. Cells were authenticated by short tandem repeat (STR) profiling using the AmpFℓSTR Identifiler kit at the Protein Expression Laboratory (NCI, Frederick, MD) in February 2015.

Site-directed mutagenesis of human EGFR and MIG6 and subcloning of wild-type and mutant constructs for lentivirus production were performed at the Protein Expression Laboratory, a Frederick National Laboratory for Cancer Research (FNLCR) core facility.

Tissue culture or mouse tissue lysates used for immunoblot were prepared in RIPA lysis buffer (150 mmol/L NaCl, 1.0% IGEPAL CA-630, 0.5% sodium deoxycholate, 0.1% SDS, and 50 mmol/L Tris, pH 8.0). For immunoprecipitation, cell extracts were prepared in NP-40 lysis buffer. Mouse tissue extracts were prepared in RIPA buffer using a tissue lyser (Qiagen) following the manufacturer's protocol. For phosphoproteomic analysis, mouse tissue extracts were prepared in urea lysis buffer (20 mmol/L Hepes pH 8.0, 9 mol/L urea, 1 mmol/L sodium orthovanadate, 2.5 mmol/L sodium pyrophosphate, and 1 mmol/L β-glycerophosphate). All lysis buffers contained protease and phosphatase inhibitor cocktails from Roche, and 1 mmol/L sodium orthovanadate to inhibit protease and phosphatase activities. Protein concentrations were quantified using a modified Lowry method (BioRad). For immunoprecipitation, 800 to 1,000 μg of lysate was incubated overnight at 4°C with 2 to 5 μg of mouse anti-EGFR (MAB108) or mouse anti-Flag (MIG6) monoclonal antibody. The antigen–antibody complex was then captured by incubating the mixture with protein G beads for an additional 1 hour. The immunocomplexes were washed with NP-40 lysis buffer twice and once with PBS buffer containing 1 mmol/L sodium orthovanadate. The bound proteins were then extracted with 2× SDS loading buffer [65.8 mmol/L Tris–HCl, pH 6.8, 2.1% SDS, 26.3% (w/v) glycerol, 0.01% bromophenol blue, 5% 2-mercaptoethanol] by incubating at 95ºC for 5 minutes and fractionated by SDS-PAGE (4%–15%). The proteins were transferred to nitrocellulose membrane using either the semidry or wet transfer method, and probed with the specified antibody.

Cells were serum starved for 18 hours and treated with 100 μmol/L cycloheximide (Sigma-Aldrich) for 1 hour to inhibit new protein synthesis. Cells were either mock stimulated or stimulated with EGF (100 ng/mL) for different time points at 37ºC. Following EGF treatment, the cells were quickly chilled and washed with cold PBS, and lysed in RIPA buffer supplemented with protease and phosphatase inhibitors, as previously described. For the degradation assay in the presence of chloroquine, cells were serum starved and treated with cycloheximide as described above. Fifteen minutes before EGF addition, cells were treated with or without 100 μmol/L of chloroquine. Extracts for Western blot analysis were prepared at various time points after EGF stimulation.

The doxycycline-inducible EGFR^L858R and EGFR^Del transgenic models have been described previously (35). All mice were maintained in a pathogen-free facility approved by the NCI and Memorial Sloan Kettering Cancer Center (MSKCC) Animal Care and Use Committees (ACUC). Animal studies were carried out with the approval of research protocols by the ACUC. We bred Mig6^+/− mice with CCSP-rtTA mice, and Mig6^+/− mice with TetO-EGFR^mut mice. Resulting Mig6^+/−/CCSP-rtTA and Mig6^+/−/TetO-EGFR^mut offspring were then crossed to generate mutant EGFR–expressing mice in a Mig6 wild-type (Mig6^+/+), Mig6 heterozygous (Mig6^+/−), or Mig6-null (Mig6^−/−) background.

Transgenic mice were fed with doxycycline-impregnated food pellets (625 ppm; Harlan-Teklad) to induce mutant forms of human EGFR from a doxycycline-regulated promoter. Mice were monitored for EGFR-driven tumor development by MRI of the lungs. MRI was carried out with respiratory gating at the MRI Core Facility of NCI or MSKCC. Serial MRI analyses were performed, and the tumor burden was quantified by ImageJ software. Regions of interest tool (ROI) was used to outline the lung and tumor within the lung. ROI measurements provided the area and mean intensity of lung and tumor. For each MRI time point, the percentage of tumor burden was calculated from the total lung and tumor measurements obtained. Mice were selected to be euthanized for survival analysis primarily using clinical criteria such as hunched posture, trachypnea, weight loss, and decreased movement. Lungs were perfused with PBS, and a representative portion of the tumor tissue was frozen in liquid nitrogen for further analysis. Another representative lung tumor tissue was processed by perfusion with phosphate-buffered 4% paraformaldehyde for histopathology evaluation.

For replicate experiments, SD or SE was calculated to indicate the variation between experiments, and values given represent the mean ± SD. Statistical analyses of the results to assess the significance of differences were performed using an unpaired Student t test. A threshold of P ≤ 0.05 was used for significance. Kaplan–Meier survival analyses were performed on tumor-bearing mice using GraphPad Prism.

R.M. Simpson is president-elect at the American College of Veterinary Pathologists. K. Politi reports receiving a commercial research grant from AstraZeneca, has ownership interest (including patents) in Molecular MD/MSKCC, and is a consultant/advisory board member for Takeda. No potential conflicts of interest were disclosed by the other authors.

Conception and design: T.K. Maity, A. Venugopalan, H.E. Varmus, U. Guha

Development of methodology: T.K. Maity, A. Venugopalan, C.M. Cultraro, R.M. Simpson, R. Biswas, U. Guha

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T.K. Maity, A. Venugopalan, I. Linnoila, C.M. Cultraro, A. Giannakou, R. Nemati, X. Zhang, J.D. Webster, D. Ritt, H. Hoschuetzky, R.M. Simpson, K. Politi, U. Guha

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T.K. Maity, A. Venugopalan, I. Linnoila, X. Zhang, R.M. Simpson, D.K. Morrison, H.E. Varmus, U. Guha

Writing, review, and/or revision of the manuscript: T.K. Maity, A. Venugopalan, I. Linnoila, J.D. Webster, R.M. Simpson, R. Biswas, K. Politi, D.K. Morrison, H.E. Varmus, U. Guha

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T.K. Maity, A. Venugopalan, C.M. Cultraro, A. Giannakou, R. Nemati, D. Ritt, S. Ghosal, H.E. Varmus, U. Guha

Study supervision: H.E. Varmus, U. Guha

The authors thank George Vande Woude for Mig6 knockout mice, Oreste Segatto for MIG6 antibody and helpful discussions, and Philip Cole for helpful discussions.

This work was supported by the National Cancer Institute Center for Cancer Research (NCI-CCR) Intramural Research Program ZIA BC 011259 (to U. Guha), a K99 Career Transition Award (K99CA140792; to U. Guha), and PO1 CA129243 (to H.E. Varmus).

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.

Received July 16, 2014; revised February 18, 2015; accepted February 20, 2015; published OnlineFirst March 3, 2015.