Mig6 is a sensor of EGF receptor inactivation that directly activates c-Abl to induce apoptosis during epithelial homeostasis

Abstract

A fundamental aspect of epithelial homeostasis is the dependence on specific growth factors for cell survival, yet the underlying mechanisms remain obscure. We found an "inverse" mode of receptor tyrosine kinase signaling that directly links ErbB receptor inactivation to the induction of apoptosis. Upon ligand deprivation Mig6 dissociates from the ErbB receptor and binds to and activates the tyrosine kinase c-Abl to trigger p73-dependent apoptosis in mammary epithelial cells. Deletion of Errfi1 (encoding Mig6) and inhibition or RNAi silencing of c-Abl causes impaired apoptosis and luminal filling of mammary ducts. Mig6 activates c-Abl by binding to the kinase domain, which is prevented in the presence of epidermal growth factor (EGF) by Src family kinase-mediated phosphorylation on c-Abl-Tyr488. These results reveal a receptor-proximal switch mechanism by which Mig6 actively senses EGF deprivation to directly activate proapoptotic c-Abl. Our findings challenge the common belief that deprivation of growth factors induces apoptosis passively by lack of mitogenic signaling.

Graphical abstract

Figure 1

Deletion of Errfi1 Leads to Impaired Cell Death and Propagation of Luminal Epithelial Cells during Mammary Morphogenesis (A) Mammary glands from BrdU-injected 6-week-old Errfi1^−/− or wild-type littermate control mice were either subjected to whole-mount carmine staining (upper panels) or terminal end buds and mammary ducts were subjected to immunostaining for cleaved caspase 3 (Ccasp-3), or BrdU as indicated. Mammary ducts from a 10-week-old Errfi1^−/− or wild-type littermate control mice were stained for keratin-14 (K14) or keratin-18 (K18) as indicated. (B) pMECs were isolated from 14- to 15-week-old Errfi1^−/− (KO) or wild-type (WT) littermate mice and subjected to flow cytometry with antibodies against indicated proteins. The number of CD24^high mature luminal cells (circled red) and CD24^low basal/myoepithelial cells (circled gray) are represented in the inlayed graphs. (C) Quantification of cleaved caspase 3⁺ cells in terminal end buds of KO or WT littermate mammary glands, and % BrdU⁺ cells in TEBs and ducts analyzing longitudinal sections from mice aged 6–8 weeks. Data were analyzed by two-tailed unpaired Student's t test (p = 0.0023, denoted three asterisks), while error bars represent SEM. n = 14 (Ccasp-3) and n = 12 (BrdU) independent TEBs from five mice (WT), and n = 9 (Ccasp-3) and n = 9 (BrdU) TEBs from five mice (KO); and for ductal BrdU incorporation: n = 15 (WT) and n = 12 (KO) ducts from six and five mice, respectively. (D) Epithelial cell autonomous function of Mig6. Cleared mammary fat pad transplantation of pMECs isolated from WT or KO littermates into WT hosts and analysis of mammary outgrowths by carmine staining or mammary ducts by H&E or immunostaining for K14 and K18 as indicated. (E) The degree of filling of primary ducts was quantified as the ratio of higher intensity areas (orange) of carmine staining over the total ductal area (middle panels show representative sections). Dots in the left hand graph represent the mean high intensity area/total ductal area per mouse. The number of branchpoints per gland was determined comparing glands from Errfi1^−/− (KO) mice or littermate wild-type (WT) controls plotted individually (n = 9 for WT and KO mice; n = 6 for WT and KO transplants. Failed transplant outgrowths were excluded (in parenthesis). Data were analyzed by two-tailed unpaired Student's t test. The black bar indicates the mean while red bars represent SEM. Scale bars: 1 mm (whole mounts) and 25 μm (H&E and immunofluorescence images). See also Figure S1.

Figure 2

c-Abl Is Activated upon EGF Deprivation in a Mig6-Dependent Manner and Required for Apoptosis (A) Errfi1^−/− or wild-type pMECs were cultured in the presence (+) or absence (−) of 10 ng/ml EGF for indicated times and cell lysates subjected to immunoblotting as indicated. (B) Trypan blue cell viability assay on Errfi1^−/− or wild-type pMECs cultured in the presence (+) or absence (−) of 10 ng/ml EGF for 8 hr. The % of cells taking up the dye is indicated (n = 3 independent experiments; ^∗∗∗p = 0.00015). (C) Longitudinal sections of a TEB or duct from wild-type (WT) or Errfi1^−/− (KO) mice (5 and 6 weeks of age, respectively) subjected to immunostaining for active phospho-Y877-ErbB2 and counterstained with DAPI. Note the absence of active ErbB2 in the interior luminal cell layers in the wild-type TEB or Errfi1^−/− duct. Representative images of eight analyzed TEBs and ducts from a 6-week-old mouse. (D) Errfi1^−/− (KO) or littermate wild-type pMECs (WT) were deprived of EGF for 6 hr in the absence (−) or presence (+) of 1 μM lapatinib, 1 μM nilotinib, prior to western analysis for indicated proteins. The right panels constitute a separate independent experiment. Note that inhibition of c-Abl disrupts EGF-deprivation induced cell death, while inhibition of EGFR and ErbB2 fails to rescue the cell death defect of the KO pMECs. (E) pMECs subjected to shRNAi-mediated knock-down of c-Abl using two independent constructs were deprived of EGF for 6 hr and cell lysates immunoblotted for indicated proteins. (F and G) pMECs from Errfi1^−/− or littermate wild-type mice were cultured in the presence or absence of EGF for 2 hr prior to harvest, followed by immunoprecipitation of c-Abl and immunoblotting for total phospho-tyrosine (F) or tyrosine 245 (G, independent experiment), indicating that c-Abl is rapidly activated upon EGF withdrawal in a Mig6-dependent manner. The numbers indicate fold increase in intensity as determined by densitometric analysis. (H and H′) pMECs from KO or littermate WT mice were cultured in the presence or absence of EGF for 2 hr prior to harvest, followed by immunoprecipitation of c-Abl and an in vitro kinase assay using a specific Abl substrate peptide (Abltide) as a substrate. (H′) Quantification of kinase activity as fold change over the EGF stimulated WT samples (n = 4, ^∗∗∗p = 0.00017). See also Figures S2 and S3). Indicated p values were determined by two-tailed unpaired Student's t test, while the error bars represent SEM.

Figure 3

Mig6 Interacts with and Activates c-Abl (A) c-Abl or Mig6 were immunopurified from HEK293T cells ectopically coexpressing Mig6 and wild-type (WT), catalytically inactive (KD), or constitutively active (PP) forms of c-Abl and examined by immunoblotting as indicated. (B and C) Immunopurification of endogenous Mig6 from pMECs cultured in the presence or absence of EGF for 2 hr and examined for interaction with c-Abl (B) or EGFR (C). (D) Immunoprecipitated EGFR from HEK293T cells ectopically coexpressing EGFR and Mig6 or mutant forms of Mig6 lacking the AH1 domain (deletion aa 328–361) or the Grb2 binding site (PP319/20AA) and examined by immunoblotting as indicated. (E) Immunoprecipitation of c-Abl coexpressed with Mig6 and EGFR in HEK293T cells followed by immunoblotting as indicated.

Figure 4

Mig6 Binds the Abl Kinase Domain via the Conserved EGFR Binding Region (A) Coimmunopurification of ectopically expressed c-Abl and full length (WT) Mig6 or fragments of HA-tagged Mig6 constituting either the minimal ErbB interaction aa 335–365 (Ack homology region 1, AH1) or the complete Ack homology region aa 314–412 (AH1+2). (B) Outline of the minimal region of Mig6 required for Abl activation with point mutations generated. (C) Molecular modeling of the predicted c-Abl/Mig6 interface based on the structural similarity with EGFR bound to Mig6 (EGFR/Mig6: PDB 2RF9 and c-Abl: PDB 1OPK). View of the c-Abl kinase domain (light blue) and Mig6 (red) with residues F352 and Y358 (highlighted red) known to contribute to the Mig6/EGFR interface. The regulatory helix I′ of c-Abl is shown in its inactive closed conformation (blue, PDB 1OPK) or active open conformation (yellow, PDB 1OPJ). Note the apparent clash between Mig6 and the inactive but not active I′ helix conformation. c-Abl residues predicted to participate in the interaction with Mig6 that were mutated in d are highlighted yellow. (D) Coimmunopurification of ectopically expressed Mig6 with mutant forms of c-Abl with indicated point mutations, implicating the predicted C-lobe interface is in the activation of Abl by Mig6. (E) Identification of the Abl phosphorylation sites on Mig6. Immunopurification of ectopically expressed wild-type or Y394/395F mutants of Mig6 and constitutively active c-Abl-PP. (F) Coimmunopurification of ectopically expressed HA-tagged inactive c-Abl kinase domains with indicated C-lobe mutation with WT or mutated forms of Mig6 in the presence or absence of nilotinib (2 μM). See also Figure S3.

Figure 5

Phosphorylation of Abl on Y488 by Src-Kinases Regulates Mig6-Induced Activation of Abl (A) Coimmunopurification of ectopically expressed Mig6 with wild-type (WT), Y488E, or Y488F forms of c-Abl in HEK239T cells followed by western analysis as indicated. Note that the phosphomimic Y488E and partially the Y488F mutation disrupts the ability of Mig6 to activate c-Abl. (B) In vitro kinase assay with recombinant active c-Src or EGFR with wild-type (WT), Y488E or Y488F mutant forms of the inactive catalytic domain of c-Abl immunopurified from overexpressing HEK293T cells. (C) Immunoprecipitation of WT or Y488F mutant c-Abl coexpressed with Mig6 and active c-Src, followed by western analysis as indicated. (D) Western analysis of endogenous phosphorylation of Abl on Y488 in pMECs cultured in the presence or absence of EGF for 2 hr prior to cell harvest. Values denote relative intensities of p488 bands, normalized against total c-Abl levels. (E) Western analysis of endogenous phosphorylation of Abl on Y488 in control wild-type MEFs or MEFs lacking Src-family members. Values denote relative intensities of p488 bands, normalized against total c-Abl levels. (F) Immunopurification of c-Abl coexpressed with Mig6 and EGFR with or without 0.8 μM of the Src inhibitor PP2 and western analysis as indicated. Note the Src-dependent inhibition of Mig6-induced Abl activation by EGFR overexpression. (G) Western analysis of immunoprecipitated endogenous c-Abl for Y phosphorylation in pMEFs cultured in the presence of EGF, treated with 0.4 or 0.8 μM PP2 for 2 hr. Values denote relative intensities of pY bands, normalized against total immunoprecipitated c-Abl levels, as determined by densitometry. (H) Induction of cell death by Mig6/Abl but not Src/Abl. Ectopic expression of Abl with or without Mig6 or Src in pMEFs for 16 hr, followed by western analysis as indicated. Graph shows cell death determined as % of cells taking up the viability dye trypan blue (n = 3). (I) siRNA-mediated silencing of p73 followed by ectopic coexpression of Abl and Mig6 in pMEFs for 16 hr, and western analysis as indicated. Graph shows cell death determined as % of cells taking up the viability dye trypan blue (n = 3). (J) C-Abl immunofluorescence in WT or Errfi1 KO pMECs cultured in absence of EGF for 2 hr. Note the reduced nuclear staining of c-Abl in the absence of Mig6. Images are representative of three independent experiments. Indicated p values were determined by two-tailed unpaired Student's t test, while the error bars represent SEM.

Figure 6

Abl Inhibition Impairs Luminal Cell Death and Cause TEB Filling (A) Immunostaining of TEBs for proliferating (BrdU-positive), cleaved caspase 3 (Ccasp-3) positive cells, or epithelial markers keratin 14 (K14) or keratin 18 (K18) of mice treated with nilotinib or vehicle (representative of three independent experiments). Scale bars = 20 μm. (B) Quantification of cleaved caspase 3 positive cells in TEBs of mock or nilotinib-injected mammary glands, n = 5 TEBs each sample group from three independent experiments. (C–D′) Whole-mount carmine staining of mammary glands from mock- or nilotinib-injected littermate mice. (C) The dotted line indicates the estimated position of the TEBs when treatment was initiated. Note filling of TEBs and reduced outgrowth of the secondary side branches upon nilotinib treatment. Scale bars = 500 μm. (D) The degree of filling of TEBs was quantified as the ratio of higher intensity areas of carmine staining over the total TEB area (as outlined, D′). (E) Quantification of degree of TEB filling in nilotinib-treated mice, n = 16 (nilotinib) and 12 (mock). In (B) and (E), data were analyzed by two-tailed unpaired Student's t test while error bars represent SEM.

Figure 7

Silencing of c-Abl Leads to Impaired Mammary Lumen Formation and Branching Morphogenesis (A–C) pMECs were subjected to lentivirus-mediated shRNA silencing of c-Abl and cultured inside (A and A′) or on top of (B and B′) Matrigel. (A) The degree of luminal filling of individual acini was determined in a blind fashion from stacks of phase contrast and GFP fluorescence images (see A′ for representative acini). The proportion of spheres within the indicated subcategories are represented (n = 3 independent experiments with a total of 120 control and 118 Abl knockdown (k.d.) acini analyzed). (B) The number of branches of individual spheres cultured on top of Matrigel are plotted, n = 3 independent experiments with a total of 83 (ctrl) and 72 (Abl k.d.) acini analyzed. p = 0.006 for difference in mean number of branches per ctrl versus acini. (B′) Phase contrast/GFP images of representative acini. (C) Western analysis for indicated proteins of pMECs infected with the nontargeting or c-Abl-targeting shRNAs, confirming efficient knockdown. p values were determined by two-tailed unpaired Student's t test (A) or Mann-Whitney test (B), while error bars indicate SEM. (D) Model illustrating the molecular switch mechanism by which Mig6 senses ErbB receptor inactivation. In the presence of EGF the ErbB-binding region of Mig6 (red) binds to the active ErbB receptors interfering with kinase domain dimer formation (Zhang et al., 2007a). C-Src is recruited to and activated by active ErbB receptors and phosphorylates c-Abl on Y488, thereby preventing Mig6 from activating c-Abl to induce p73-dependent cell death. Upon ErbB receptor inactivation c-Abl and Mig6 dissociate from the receptor, leading to de-phosphorylation of c-Abl on Y488, which enable Mig6 to activate c-Abl to trigger p73-dependent cell death.