Regulation of EGFR trafficking and cell signaling by Sprouty2 and MIG6 in lung cancer cells

Epidermal growth factor receptor (EGFR)-mediated signaling must be tightly controlled in order to ensure appropriate cellular outcomes. Such control is achieved through a variety of mechanisms, including feedback. Studies of biological networks have shown that feedback regulation is necessary to generate biologically observed signaling patterns such as adaptation, oscillations and switch-like responses (Ma et al., 2009; Kholodenko et al., 2010). In addition, multiple positive and negative feedback loops can create systems that are tunable and insensitive to noise (Brandman et al., 2005). These and other observations have made it increasingly clear that a complete understanding of signaling initiated by EGFR will require a deeper understanding of the feedback regulators that control EGFR-mediated signaling (Avraham and Yarden, 2011). Two such feedback regulators of EGFR-mediated signaling are Sprouty2 (SPRY2) and mitogen-inducible gene 6 (MIG6; also known as RALT).

SPRY2 belongs to a family of four mammalian Sprouty proteins and regulates signaling downstream of multiple growth factor receptors. SPRY2 expression is induced by extracellular signal-regulated kinase (ERK) activity (Ozaki et al., 2001). When phosphorylated at tyrosine 55 in response to binding of EGF to EGFR, SPRY2 binds and sequesters the E3 ubiquitin ligase CBL, impeding EGFR ubiquitylation and degradation (Egan et al., 2002; Wong et al., 2002; Rubin et al., 2003; Haglund et al., 2005). SPRY2 also inhibits EGFR passage from early to late endosomes in a proposed mechanism involving SPRY2 binding to hepatocyte growth factor-regulated tyrosine kinase substrate (Kim et al., 2007). Downstream of receptors, SPRY2 may antagonize ERK activity by inhibiting RAS through binding to GRB2 or by inhibiting RAF, depending on the cellular context (Yusoff et al., 2002; Lao et al., 2006).

MIG6 is transcriptionally regulated by ERK downstream of EGFR (Hackel et al., 2001; Fiorini et al., 2002) and inhibits EGFR activation by binding at the asymmetric interface between dimerized EGFR kinases (Zhang et al., 2006; Zhang, X. et al., 2007). MIG6 also promotes EGFR endocytosis by coupling the receptor to AP-2 and intersectins (Frosi et al., 2010; Ying et al., 2010).

Although elevated expression and activating mutations of EGFR occur frequently in multiple cancers (Itakura et al., 1994; Hirsch et al., 2003; Mellinghoff et al., 2005; Sheng and Liu, 2011), the functional role of EGFR feedback regulation by proteins such as SPRY2 and MIG6 has not been thoroughly studied. In cases where there is an important role for EGFR feedback regulation in oncogenesis or tumor progression, it could potentially be leveraged to overcome de novo or acquired resistance to EGFR inhibitors (Kobayashi et al., 2005; Mellinghoff et al., 2005; Kosaka et al., 2006; Sheng and Liu, 2011).

We studied the roles of SPRY2 and MIG6 in non-small cell lung cancer (NSCLC) cells, where EGFR is frequently expressed at elevated levels. In a small fraction of NSCLCs, the expression of kinase-activated EGFR mutants confers unusual cellular sensitivity to EGFR inhibitors (Lynch et al., 2004; Paez et al., 2004; Mitsudomi and Yatabe, 2007) and leads to increased expression and phosphorylation of SPRY2 and MIG6 (Rubin et al., 2003; Guo et al., 2008; Nagashima et al., 2009). These EGFR mutations also lead to dramatic impairment of EGFR endocytosis, which has been linked to differential cellular sensitivity to EGFR inhibitors (Hendriks et al., 2006; Lazzara et al., 2010). We hypothesized that SPRY2 and MIG6 could participate in this perturbation to EGFR mutant endocytosis and in turn serve as important determinants of cellular response to EGFR inhibitors.

In two different NSCLC cell lines, one with an EGFR-activating mutation and demonstrating the previously documented impairment in receptor internalization, EGFR endocytosis was augmented by SPRY2 knockdown and reduced by MIG6 knockdown. EGFR recycling, which we quantitatively determined to be roughly twofold more efficient in the cell line with the EGFR mutation, was also reduced by SPRY2 knockdown in both cell lines. Thus, SPRY2 may play two roles that promote EGFR expression in NSCLC cells with or without an EGFR mutation. Interestingly, the effects of SPRY2 knockdown on receptor endocytosis and recycling were explained by a concomitant decrease in EGFR expression, as revealed by EGFR reconstitution experiments. Downstream of the receptor, SPRY2 knockdown significantly reduced ERK phosphorylation. However, MIG6 knockdown had a relatively modest effect on ERK phosphorylation. Moreover, as a result of reduced ERK phosphorylation, SPRY2 knockdown promoted an apoptotic response to the EGFR inhibitor gefitinib. This increase in apoptosis was especially pronounced in PC9 cells, which express a deletion mutant of EGFR. Despite the rescue effects of EGFR reconstitution on top of SPRY2 knockdown on EGFR trafficking, EGFR reconstitution did not rescue the effects of SPRY2 knockdown on ERK phosphorylation or cellular response to gefitinib. Thus, our study identifies SPRY2 and MIG6 as important regulators of EGFR endocytosis and recycling in EGFR-mutant-expressing cells, as well as cells expressing wild-type EGFR. Our results also point to an EGFR-expression-independent function of SPRY2 in the regulation of ERK that impacts cellular sensitivity to EGFR inhibition. These findings provide new insights into the coupling between EGFR trafficking, signaling and feedback regulation and suggest that interference with SPRY2 expression or function could be a useful therapeutic approach in lung cancer cells with acquired resistance to EGFR-targeted therapy.

We reduced SPRY2 and/or MIG6 expression in PC9 (delE746-A750 EGFR) and H1666 (wild-type EGFR) NSCLC cells through stable shRNA expression (supplementary material Fig. S1). As expected based on previous studies of NSCLC cells (Hendriks et al., 2006; Lazzara et al., 2010), the rate constant for EGF-mediated EGFR endocytosis (k_e) was larger in H1666 cells transduced with either of the control shRNA vectors than in corresponding PC9 cells (Fig. 1). SPRY2 knockdown significantly increased k_e in both cell lines (from 0.05 to 0.08/minute in PC9 cells, and from 0.16 to 0.21/minute in H1666 cells). MIG6 knockdown reduced k_e in PC9 and H1666 cells (from 0.06 to 0.03/minute in PC9 cells, and from 0.18 to 0.12/minute in H1666 cells). When SPRY2 and MIG6 were simultaneously depleted, the net result was a decrease in k_e in both cell lines (from 0.07 to 0.045/minute in PC9 cells, and from 0.17 to 0.15/minute in H1666 cells). The effect of combined knockdown suggests a hierarchy between SPRY2 and MIG6 in controlling k_e for both wild-type and mutant EGFR. Data used to generate the k_e values are shown in supplementary material Fig. S2.

At EGF concentrations above or near 10 ng/ml (the condition used in the experiments in Fig. 1), EGFR endocytosis may occur through clathrin-dependent and -independent processes (Sigismund et al., 2005; Sigismund et al., 2008). To determine whether the differences measured in Fig. 1 might reflect effects other than those attributable to clathrin-mediated endocytosis, we repeated k_e measurements in H1666 cells at 1.5 ng/mL EGF. The differences observed at 10 ng/ml were preserved at 1.5 ng/ml (supplementary material Fig. S3A). We also found that inhibition of dynamin, a GTPase required for clathrin-mediated endocytosis of many receptor tyrosine kinases (Kirchhausen et al., 2008), reduced k_e in PC9 and H1666 cells, indicating that clathrin-mediated endocytosis was relevant in both cell lines (supplementary material Fig. S3B). Both CBL- and MIG6-mediated endocytosis pathways are clathrin mediated (Swaminathan and Tsygankov, 2006; Frosi et al., 2010).

Because normal EGFR trafficking is important for complete ERK activation in at least some cellular contexts (Vieira et al., 1996), and because ERK activity is a key determinant of NSCLC cell response to EGFR inhibition (Lazzara et al., 2010; Furcht et al., 2013), we tested whether the changes in endocytosis we measured that were due to SPRY2 and MIG6 knockdown correlated with any changes in ERK phosphorylation. In PC9 and H1666 cells, SPRY2 knockdown resulted in an approximately twofold reduction in ERK phosphorylation with or without gefitinib present (Fig. 2A; supplementary material Fig. S4A,B). ERK phosphorylation was increased by MIG6 knockdown, but the magnitude of this effect was more modest. With combined SPRY2 and MIG6 knockdown, there was a small (but statistically insignificant; P = 0.072 or 0.058 for 0 or 0.001 µM gefitinib conditions, respectively) reduction in ERK phosphorylation in PC9 cells and no change in ERK phosphorylation in H1666 cells. Qualitatively similar, but generally smaller, changes in AKT phosphorylation were observed with SPRY2 and MIG6 knockdown (supplementary material Fig. S4C).

In validating the stable knockdown of SPRY2 and MIG6, we also noticed a reduction in EGFR expression concomitant with SPRY2 knockdown. Western blot analysis revealed that, in PC9 and H1666 cells, SPRY2 knockdown resulted in decreases in EGFR protein levels of 54% and 40%, respectively (Fig. 2B). Differences in absolute counts of ¹²⁵I-EGF binding in PC9 and H1666 cell lines from k_e measurements also confirmed these decreases in EGFR expression. This effect was repeated with an independent SPRY2 shRNA (supplementary material Fig. S5A) and generally held when we probed the effects of SPRY2 knockdown in a larger panel of NSCLC cells (supplementary material Fig. S5B). As will be shown later, we also verified these changes by flow cytometry. Because SPRY2 knockdown led to the most dramatic changes in the activity of the ERK pathway and also appeared to alter EGFR expression, effects that may alter cellular response to EGFR inhibition, we focused most of the remainder of our studies on the effects of SPRY2 knockdown.

To further probe the mechanism of SPRY2-mediated regulation of EGFR internalization and expression, we measured EGFR ubiquitylation and association with CBL. A previous report found that EGFR mutants are poorly ubiquitylated and do not associate with CBL in response to EGF (Padrón et al., 2007), but the effects of SPRY2 on EGFR ubiquitylation and CBL association has not previously been explored in NSCLC cells expressing mutant EGFR. We did not detect significant EGFR ubiquitylation or CBL association in EGFR immunoprecipitates of PC9 cells treated with EGF (Fig. 3A). In contrast, and consistent with previously hypothesized mechanisms of wild-type EGFR regulation by SPRY2 (Egan et al., 2002; Wong et al., 2002; Rubin et al., 2003; Haglund et al., 2005), EGFR was ubiquitylated and CBL-associated in response to 10 ng/ml EGF in H1666 cells, and these effects were augmented by SPRY2 knockdown (Fig. 3B). Even with EGF treatments for longer times and at higher concentrations than used in the experiments in Fig. 3, these differences between PC9 and H1666 cells persisted (supplementary material Fig. S6A–D). The absence of non-specific immunoprecipitation of EGFR, ubiquitin and CBL was confirmed in a separate experiment (supplementary material Fig. S6E).

Because no differences in EGFR ubiquitylation or CBL association were detected in PC9 cells with SPRY2 knockdown (where EGFR expression was substantially reduced), we examined the effect of SPRY2 knockdown on EGFR mRNA levels. In PC9 cells, SPRY2 knockdown reduced EGFR mRNA levels by ∼75% (Fig. 4A). A reduction in EGFR mRNA was also measured for a second non-overlapping SPRY2 shRNA (supplementary material Fig. S7). In contrast, no change in EGFR mRNA level was detected in H1666 cells with SPRY2 knockdown (Fig. 4A; supplementary material Fig. S7).

Because EGFR expression can be regulated by ERK activity (Grassian et al., 2011), and since a large effect on ERK phosphorylation was found in cells with SPRY2 knockdown, we tested the effect of MEK inhibition on EGFR expression. As determined by western blotting, MEK inhibition decreased EGFR expression in PC9 cells, but had no effect in H1666 cells (Fig. 4B). Thus, changes in EGFR expression with SPRY2 knockdown are likely to occur through transcriptional effects in PC9 cells (due to decreased ERK phosphorylation) and through changes in trafficking (increased endocytosis and degradation) in H1666 cells.

Because we found that SPRY2 knockdown increased k_e in PC9 cells without increasing EGFR ubiquitylation or CBL association, we examined whether SPRY2 knockdown could affect k_e through changes in EGFR expression. This hypothesis was motivated by knowledge that clathrin-mediated endocytosis is a saturable process wherein relatively large numbers of receptors may decrease k_e as clathrin-mediated machinery becomes limiting and receptors are forced to internalize through slower, non-clathrin-mediated pathways (Lund et al., 1990; Sigismund et al., 2005). In PC9 and H1666 cells, reconstitution of mutant and wild-type EGFR, respectively, rescued the effect on k_e observed with SPRY2 knockdown (Fig. 5A,B). In the H1666 EGFR-reconstituted cells, k_e was slightly lower than in the appropriate control cells (transduced with SPRY2 shRNA and empty expression vector). This difference may have occurred because EGFR reconstitution increased EGFR levels beyond those seen in the control cells. Data used to calculate these k_e values are shown in supplementary material Fig. S2. Despite the ability of EGFR reconstitution to rescue the effects of SPRY2 knockdown on EGFR k_e, EGFR reconstitution did not augment ERK phosphorylation basally, in the presence of gefitinib, or with EGF stimulation compared with SPRY2 knockdown without EGFR reconstitution (Fig. 5C; supplementary material Fig. S8). We verified that effects due to EGFR reconstitution were not occurring in a small sub-population of cells or as a result of mis-localized EGFR expression by measuring EGFR surface expression with flow cytometry (supplementary material Fig. S9).

We further hypothesized that the reduction in ERK phosphorylation observed with SPRY2 depletion could be the result of changes in recycling of endocytosed EGFR that, unlike the effects on k_e, might not be rescued by EGFR reconstitution. The general notion of a connection between receptor endocytic recycling and ERK activation has been previously discussed (Robertson et al., 2006; Parachoniak et al., 2011). In support of this hypothesis, treatment of PC9 or H1666 control cells with the trafficking inhibitor monensin reduced ERK phosphorylation to a similar degree as SPRY2 knockdown in cells with or without EGFR reconstitution (Fig. 6A). We verified an effect of monensin on trafficking by showing that PC9 and H1666 cells pre-treated with 10 µM monensin released ∼60% and 80% less internalized ¹²⁵I-EGF compared with untreated controls (data not shown). We also directly measured the recycling fraction of internalized EGF (f_r) in PC9 and H1666 cells with SPRY2 knockdown and EGFR reconstitution (Fig. 6B). It has been suggested previously, based on receptor localization, that mutant EGFR is preferentially recycled (Chung et al., 2009). However, enhanced recycling of mutant EGFR has never been quantified, as far as we are aware. The f_r for mutant EGFR in PC9 cells was significantly higher than for wild-type EGFR in H1666 cells (over 0.9 for control PC9 cells and 0.5 for H1666 cells). Interestingly, f_r was reduced by SPRY2 knockdown in both cell lines, but this change was at least partially rescued by EGFR reconstitution. Thus, it seems unlikely that the change in EGFR recycling is responsible for the observed reductions in ERK phosphorylation with SPRY2 knockdown.

To test whether the changes in EGFR trafficking, EGFR expression and ERK phosphorylation due to SPRY2 and MIG6 knockdown altered cellular response to EGFR inhibitors, we first treated PC9 and H1666 cells with gefitinib at appropriate doses to induce cell death. For PC9 cells treated with 0.1 µM gefitinib for 24 hours, Annexin V staining (as a measure of apoptosis) increased from low basal levels for shRNA controls to >25% with SPRY2 knockdown (Fig. 7A). In contrast, Annexin V staining did not change significantly in response to gefitinib in PC9 cells as a result of MIG6 knockdown. There were significant increases in Annexin V staining in PC9 cells, however, when SPRY2 and MIG6 were simultaneously depleted. Qualitatively similar changes were observed in H1666 cells, but they were small in comparison to those observed in PC9 cells. This is probably a result of wild-type EGFR expression in H1666 cells, which generally confers a much greater degree of cellular resistance to interference with survival signaling. Increased cell death in response to gefitinib was also observed in H1975 and H358 cell lines with SPRY2 knockdown compared with controls (supplementary material Fig. S10).

To determine whether the reduction in ERK phosphorylation observed with SPRY2 knockdown could cause increased cell death response to gefitinib, we used U0126 to reduce ERK phosphorylation to a similar degree in control PC9 and H1666 cells as observed with SPRY2 knockdown. Appropriate U0126 concentrations were found by U0126 titration in the presence of gefitinib to recapitulate the conditions in Fig. 3 (supplementary material Fig. S11). PC9 and H1666 cells expressing control shRNA, treated with the concentrations of U0126 we identified, showed the anticipated augmentation in cell death response to gefitinib (Fig. 7B). Consistent with the fact that EGFR reconstitution did not rescue the decrease in ERK phosphorylation due to SPRY2 knockdown (Fig. 5C), there was no effect of EGFR reconstitution on cellular response to gefitinib in H1666 or PC9 cells (Fig. 7C).

To our knowledge, the data presented here constitute the first study of the functional roles of SPRY2 and MIG6 in EGFR endocytosis and recycling, EGFR-mediated signaling, and cellular response to EGFR kinase inhibitors in cells expressing the constitutively active EGFR mutants that arise in NSCLC. Our data demonstrate that feedback regulation can indeed play important roles in determining receptor trafficking and signaling in cells characterized by EGFR overexpression and activating mutations. Since the effects we measured were generally largest in a cell line with an EGFR mutation, our results suggest that the role of feedback regulation through SPRY2 and MIG6 could be especially important in the context of receptor mutation. The main trends we found are illustrated schematically in Fig. 8 and discussed in further detail below.

We first hypothesized that SPRY2 and MIG6 might regulate cell signaling and cell fate determination through their effects on EGFR trafficking. Several studies have reported impaired ligand-mediated EGFR endocytosis in NSCLC cells expressing EGFR mutants versus those expressing wild-type EGFR (Hendriks et al., 2006; Lazzara et al., 2010). These studies point to EGFR mutation itself as playing a direct role in defective endocytosis in a receptor expression-dependent manner, but the molecular basis for this defect is not fully understood. Consistent with the reported roles of MIG6 and SPRY2 in the endocytosis of wild-type EGFR in other cell types, we found that the endocytosis of mutant EGFR is promoted by MIG6 and antagonized by SPRY2. Since MIG6 expression tends to be elevated in NSCLC cells with EGFR mutations (Guo et al., 2008; Nagashima et al., 2009), the finding that EGFR k_e is greatly reduced in mutant-expressing cells relative to wild-type cells could indicate that MIG6-mediated endocytosis does not occur as efficiently for mutant EGFR as for wild-type EGFR. We do not have direct evidence for this, but it should be noted that even with SPRY2 knockdown the measured k_e value in PC9 cells was still lower than typical values in cell lines expressing wild-type EGFR. It is also worth noting that MIG6 is functionally impaired in other cancers through mechanisms including downregulation, deletion or loss-of-function mutation (Anastasi et al., 2005; Zhang, Y. W. et al., 2007; Ying et al., 2010; Li et al., 2012). Thus, disruption of MIG6 function may be a general feedback perturbation across several cancer types.

Focusing on SPRY2, we also found that SPRY2 exerts control over EGFR endocytosis rates in PC9 and H1666 cells by influencing EGFR expression. Specifically, the increased k_e values we measured in PC9 and H1666 cells with SPRY2 knockdown were accompanied by reduced EGFR expression, and rescuing EGFR expression on top of SPRY2 knockdown returned k_e values to their levels prior to SPRY2 knockdown. These data indicate that some component of the EGFR endocytosis rate process is at or near saturation in the context of the elevated EGFR expression characteristic of PC9 and H1666 cells. These results suggest that future studies of SPRY2-mediated EGFR regulation should control for EGFR levels to determine whether the effects of SPRY2 are an indirect consequence of changes in EGFR expression.

We also explored the basis of reduced EGFR expression with SPRY2 knockdown. In H1666 cells, SPRY2 knockdown promoted EGFR-CBL association and EGFR ubiquitylation, effects that are consistent with reduced EGFR expression. In PC9 cells, where SPRY2 knockdown also reduced EGFR expression, we did not detect any EGFR-CBL association or EGFR ubiquitylation with or without SPRY2 knockdown. We did, however, find that impaired ERK activity reduced EGFR expression in PC9 cells, but not in H1666 cells. Consistent with this, PC9 cells with SPRY2 knockdown also displayed significant reductions in EGFR mRNA levels. Thus, reduced EGFR expression with SPRY2 knockdown resulted from the effect on ERK phosphorylation in PC9 cells uniquely. The fact that EGFR reconstitution did not restore ERK phosphorylation levels in PC9 cells identifies the ERK/EGFR connection as a one-way coupling and suggests that SPRY2 controls ERK phosphorylation in an EGFR-expression-independent manner.

In addition to the finding of receptor expression-level-dependent endocytosis, we also found that the recycling of endocytosed EGFR was SPRY2 and EGFR expression-level-dependent. In both cell lines, SPRY2 knockdown reduced f_r, and this change was at least partially reversed by EGFR reconstitution in cells with SPRY2 knockdown. This is the first examination of the relationship between SPRY2 and EGFR expression on EGFR recycling in NSCLC cells. Although it was previously reported that mutant EGFR colocalized with transferrin (Chung et al., 2009), suggesting that mutant EGFR is preferentially recycled, our measurements are the first quantitative comparison of recycling between wild-type and mutant EGFR, as far as we are aware. We found that mutant EGFR was almost entirely sorted for recycling (f_r>0.9), whereas wild-type EGFR was split more evenly between degradation and recycling (f_r∼0.5). Overall, these recycling results demonstrate an additional mode of regulation utilized by SPRY2 in setting EGFR expression levels.

Part of our initial hypothesis was that SPRY2 and MIG6 could play a role in the previously documented impairment of ERK activation in the context of mutant EGFR expression (Lazzara et al., 2010). This hypothesis was based in part upon the previously reported requirement of normal EGFR endocytosis for complete activation of ERK downstream of EGFR (Vieira et al., 1996; Lazzara et al., 2010). The trends we identified in this study did not align in a straightforward way with this previously established relationship between k_e and ERK activity, owing to the complex, coupled, and multi-faceted processes governed by SPRY2 and MIG6. For example, MIG6 knockdown decreased k_e but promoted ERK activity. This net effect of MIG6 knockdown may have been observed because, in addition to participating in EGFR endocytosis, MIG6 plays an important role as an inhibitor of the EGFR kinase. Although SPRY2 knockdown increased k_e in PC9 and H1666 cells, ERK phosphorylation was reduced rather than enhanced. As already mentioned, this effect on ERK was not the result of reduced EGFR expression, but instead could have resulted from perturbations to signaling downstream of other RTKs such as c-MET, FGFR and PDGFR, that are also regulated by CBL (Petrelli et al., 2002; Swaminathan and Tsygankov, 2006). Our EGFR recycling results could also partially explain the reduction in ERK phosphorylation observed in PC9 cells since f_r did not fully return to its control value with EGFR reconstitution.

We briefly investigated the effects of SPRY2 and MIG6 knockdown on AKT phosphorylation as well, and found qualitatively similar but smaller effects to those on ERK. Interestingly, previous studies in HeLa cells showed that SPRY2 expression promoted PTEN expression and reduced AKT phosphorylation (Edwin et al., 2006). However, our data indicate that SPRY2 promotes AKT phosphorylation in NSCLC cells, which would be inconsistent with decreased PTEN expression. Indeed, we found no change in PTEN expression with SPRY2 knockdown in any of the cell lines we studied (supplementary material Fig. S12).

Our findings regarding SPRY2 expression and NSCLC cellular sensitivity to gefitinib are distinct from the findings of Feng et al. in colon cancer cells where SPRY2 expression correlated with cellular sensitivity to gefitinib (Feng et al., 2010). Although the authors observed a relationship between SPRY2 and EGFR expression similar to what we observed, they did not investigate any possible perturbations to downstream signaling. Thus, the net effects of SPRY2 expression may have been qualitatively different from those we observed in NSCLC cells.

Our study utilized NSCLC cells as a model cell background, however, feedback regulation has recently emerged as an important determinant of response to clinically relevant inhibitors in many different cancer cell types. ERK signaling drives oncogenic processes (e.g. proliferation and migration) and is often dysregulated in cancer, but MEK inhibitors have been largely unsuccessful in clinical trials (Rinehart et al., 2004; Adjei et al., 2008; Haura et al., 2010). It has been proposed that inhibition of the ERK pathway relieves negative feedback loops generated by ERK activity and that inhibition therefore has a net neutral effect or may even promote other signaling pathways such as AKT (Mirzoeva et al., 2009; Pratilas et al., 2009). Similarly to MEK inhibition, AKT inhibitors can induce the expression and phosphorylation of several receptor tyrosine kinases due to relief of feedback inhibition (Chandarlapaty et al., 2011). Along with the data presented in our study, these findings suggest that much more must be known about the mechanisms of feedback regulation of cell signaling in order to design successful therapies for disease.

H1666 cells (EGFR wild type) were obtained from the American Type Tissue Collection (Manassas, VA, USA) and maintained in ACL4 (Lazzara et al., 2010). PC9 cells (EGFR delE746-A750) were a generous gift from Dr Douglas Lauffenburger (MIT, Cambridge, MA, USA) and were maintained in RPMI supplemented with 10% fetal bovine serum (FBS), 1 mM L-glutamine, 100 units/ml penicillin, and 100 µg/ml streptomycin (Life Technologies, Grand Island, NY, USA). For EGFR or MEK inhibition experiments, gefitinib or U0126 (both from LC Laboratories, Woburn, MA, USA) were added to cells in complete medium. PC9 cells are extremely sensitive to gefitinib, with an IC₅₀ for cellular proliferation <<1 µM (Tracy et al., 2004; Noro et al., 2006; Guo et al., 2008). H1666 cells are moderately sensitive to gefitinib with an IC₅₀ of ∼2 µM (Tracy et al., 2004). Consistent with previously reported trends (Guo et al., 2008; Nagashima et al., 2009), PC9 cells expressed more MIG6 than H1666 cells (supplementary material Fig. S1A). SPRY2 levels were similar for both cell lines. These general trends held in a broader panel of NSCLC cell lines (supplementary material Fig. S1A).

The pSuper retroviral shRNA vector with neomycin resistance was purchased from Oligoengine (Seattle, WA, USA), and the pSicoR lentiviral shRNA vector with puromycin resistance was a generous gift from Dr Tyler Jacks [MIT Koch Institute for Integrative Cancer Research, Cambridge, MA, USA; (Ventura et al., 2004)]. An oligonucleotide encoding a hairpin targeting nucleotides 1649–1667 of human MIG6 was cloned into the pSuper plasmid. Oligonucleotides encoding hairpins targeting nucleotides 2061–2079 (main sequence used) or 1195–1213 of human SPRY2 were cloned into pSicoR. Controls were created for each vector using hairpins that do not target a known human mRNA, and control cells for simultaneous knockdown of SPRY2 and MIG6 expressed both control shRNAs. All oligonucleotides were purchased from IDT (Coralville, IA, USA). Retrovirus was produced by calcium-phosphate-mediated transfection of amphotropic Phoenix cells (Dr Gary Nolan, Stanford University, Stanford, CA, USA) with pSuper plasmids. Lentivirus was produced by transfection of 293FT cells (Life Technologies) with pSicoR, pCMV-VSVg, pMDL-gp-RRE, and pRSV-Rev plasmids (Dr Marilyn Farquhar, UCSD, La Jolla, CA, USA) using calcium phosphate. Virus-containing supernatant was passed through 0.45 µm syringe filters prior to addition to target cells, which were selected in 1–2 µg/ml puromycin (Sigma, St. Louis, MO, USA) and/or 100–500 µg/ml geneticin (Life Technologies). Efficient stable knockdown of SPRY2 and MIG6 in PC9 and H1666 cells was confirmed by western blotting (supplementary material Fig. S1B).

The pBabe.hygro wild-type human EGFR retroviral expression plasmid was constructed by sub-cloning from a pCDNA4/TO/Myc-HisB vector with a wild-type human EGFR insert (a gift from Dr Yi-Rong Chen, National Health Research Institutes, Taiwan). Retrovirus was prepared and cells were infected as described above. Cells were selected in 100 µg/ml hygromycin B (Sigma). To express the human EGFR delE746-A750 mutant, a lentiviral expression plasmid (pLenti6/V5-DEST) with the appropriate insert was used (a gift from Dr Daniel Haber, Harvard Medical School, Boston, MA, USA). Lentivirus was prepared and cells were infected as described above. Target cells were selected in 2 µg/ml blasticidin (Life Technologies).

Whole-cell lysates were prepared in a standard cell extraction buffer (Life Technologies) supplemented with protease and phosphatase inhibitors (Sigma). Lysates were cleared by centrifugation at 13,200 r.p.m. for 10 minutes, and total protein concentrations were determined by micro-bicinchoninic assay (Thermo Scientific, Rockford, IL, USA). Approximately 20 µg of total protein was loaded per lane on 4–12% gradient polyacrylamide gels (Life Technologies) under denaturing and reducing conditions and transferred to 0.2 µm nitrocellulose membranes (Life Technologies). After probing with antibodies, membranes were imaged on a LI-COR Odyssey scanner (LI-COR, Lincoln, NE, USA). Membranes were stripped with 0.2 M NaOH as needed.

Whole cell lysates were prepared using a lysis buffer optimized for immunoprecipitation (Cell Signaling Technology, Danvers, MA, USA) supplemented with protease and phosphatase inhibitors. 600 µg of total protein was incubated overnight with protein-G–agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA, USA) that were pre-conjugated to 400 ng of EGFR antibody. Immunoprecipitates were analyzed by western blotting, as described above.

Antibodies against EGFR (no. 2232), AKT (no. 9272), pAKT S473 (no. 9271), ERK (no. 4695), ubiquitin (no. 3933), and pERK T202/Y204 (no. 4377) were purchased from Cell Signaling Technology. The EGFR immunoprecipitation antibody was from Thermo Scientific (Ab-12). The CBL antibody was purchased from Epitomics (no. 1486; Burlingame, CA, USA). The SPRY2 antibody was purchased from Sigma (no. S1444). The MIG6 antibody was purchased from Santa Cruz Biotechnology (no. sc-137155), and the actin antibody was purchased from Millipore (no. MAB1501; Billerica, MA, USA). Infrared dye-conjugated secondary antibodies were purchased from Rockland Immunochemicals (Gilbertsville, PA, USA). All antibodies were used according to manufacturers' recommendations.

Floating and adherent cells were pooled and stained with FITC-conjugated Annexin V (Southern Biotech, Birmingham, AL, USA). Cells were analyzed within 1 hour of staining using a Becton Dickinson FACS-Calibur cytometer, and data were analyzed using FlowJo.

Rate constants of EGF-mediated EGFR endocytosis (k_e) were measured using ¹²⁵I-EGF and corrected for the effects of non-specific binding and surface spillover, as described previously (Wiley and Cunningham, 1982; Lund et al., 1990). Steady-state EGFR recycling fractions (f_r), defined as the fraction of intact internalized ligand that is returned to and released from the plasma membrane, were measured as described previously (French et al., 1995), with intact and degraded ¹²⁵I-EGF separated with 5 kDa molecular mass cutoff centrifugal filters (Millipore).

Relative amounts of EGFR mRNA were determined using the comparative C_T method. RNA samples were prepared using the RNeasy Mini Kit (Qiagen, Valencia, CA, USA) with on-column DNase I digestion. Equal amounts of RNA from each sample were reverse transcribed using the High Capacity cDNA Reverse Transcription Kit (Life Technologies). qPCR was performed using SYBR Green PCR Master Mix (Life Technologies) on an Applied Biosystems 7300 Real-Time PCR System.

The authors are grateful to Dr Douglas Lauffenburger (MIT, Cambridge, MA, USA), Dr Eric Haura (Moffitt Cancer Center, Tampa, FL, USA), Dr Pasi Jänne (Dana Farber Cancer Institute, Boston, MA, USA), Dr Daniel Haber (Harvard Medical School, Boston, MA, USA), Dr Yi-Rong Chen (National Health Research Institutes, Taiwan), Dr Tyler Jacks (MIT, Cambridge, MA, USA), and Dr Marilyn Farquhar (UCSD, San Diego, CA, USA) for generously providing reagents. The authors are also grateful to Dr Ben Neel for helpful technical discussions. The authors are also grateful to Ms Janine Buonato for a careful review of the manuscript.