Contact inhibitory Eph signaling suppresses EGF-promoted cell migration by decoupling EGFR activity from vesicular recycling

Ephrin signaling in densely crowded cells alters EGFR recycling to inhibit migration induced by EGF. Limiting movement in a crowded situation The epidermal growth factor receptor (EGFR) mediates the distinct cellular processes of proliferation and migration, which do not always occur concomitantly upon EGFR stimulation. Eph receptors are activated by increasing cell density, and they suppress cell migration, in contrast to EGFR. Stallaert et al. (see also the Focus by Shi and Wang) found that Eph receptors selectively inhibited migration but not proliferation mediated by EGFR. Eph receptor activation prevented the recycling of EGFR to the cell surface (the subcellular compartment from where it mediates migratory signaling) by trapping EGFR in endosomes (the subcellular compartment from where it can continue to promote proliferative signaling). In addition, EGFR-mediated migration was also inhibited by the receptor Kiss1, which not only is structurally unrelated to Eph receptors but also inhibits cell migration by suppressing EGFR recycling. The authors note that this system enables different receptors to regulate a signaling pathway without needing to directly interact with components in that pathway. The ability of cells to adapt their response to growth factors in relation to their environment is an essential aspect of tissue development and homeostasis. We found that signaling mediated by the Eph family of receptor tyrosine kinases from cell-cell contacts changed the cellular response to the growth factor EGF by modulating the vesicular trafficking of its receptor, EGFR. Eph receptor activation trapped EGFR in Rab5-positive early endosomes by inhibiting Akt-dependent vesicular recycling. By altering the spatial distribution of EGFR activity, EGF-promoted Akt signaling from the plasma membrane was suppressed, thereby inhibiting cell migration. In contrast, ERK signaling from endosomal EGFR was preserved to maintain a proliferative response to EGF stimulation. We also found that soluble extracellular signals engaging the G protein–coupled receptor Kiss1 (Kiss1R) similarly suppressed EGFR vesicular recycling to inhibit EGF-promoted migration. Eph or Kiss1R activation also suppressed EGF-promoted migration in Pten−/− mouse embryonic fibroblasts, which exhibit increased constitutive Akt activity, and in MDA-MB-231 triple-negative breast cancer cells, which overexpress EGFR. The cellular environment can thus generate context-dependent responses to EGF stimulation by modulating EGFR vesicular trafficking dynamics.


INTRODUCTION
Activation of epidermal growth factor receptor (EGFR) promotes various cellular responses, including cell growth, proliferation, survival, apoptosis, differentiation, and migration (1), some of which are functionally opposed. To select among these diverse outcomes, the cell requires additional contextual information. This context can be intrinsic (dependent on the cell type or cell cycle stage, for example), or extrinsic, in the form of extracellular signals that provide information about the current (or past) environmental context. Adaptability to a changing environment requires that extrinsic information be integrated through mechanisms that can transform the response to subsequent growth factor stimulation.
Local cell density is one such example of extrinsic context that can influence cellular activity to generate distinct functional states (2)(3)(4). The Eph family of receptor tyrosine kinases act as sensors of cell density, becoming activated at points of cell-cell contact through interactions with membrane-bound ephrin ligands presented on the surfaces of adjacent cells (5). In many ways, Eph receptors operate in functional opposition to EGFR, acting as tumor suppressors (5)(6)(7)(8)(9)(10)(11) and mediating contact inhibition of locomotion to suppress cellular migration and metastasis (12)(13)(14)(15). Moreover, a functional coupling of EGFR and Eph receptor activity controls cell migration (15). Although the precise mechanism through which Eph receptors regulate EGFpromoted migration remains elusive, a convergence of receptor activity on phosphoinositide 3-kinase (PI3K)/Akt signaling has been implicated.
Akt regulates EGFR vesicular trafficking through the endosomal system (16). By stimulating the activity of the early endosomal ef-fector PIKfyve (FYVE-containing phosphatidylinositol 3-phosphate 5-kinase), Akt activity controls the transition of EGFR through early endosomes, regulating both its recycling back to the plasma membrane (PM) and its degradation in the lysosome. Thus, while endocytosis of cell surface receptors has traditionally been viewed as a mechanism to attenuate downstream signaling after ligand stimulation, the notion that signaling molecules downstream of cell surface receptors can, in turn, influence vesicular trafficking (16)(17)(18)(19)(20)(21) generates a reciprocal relationship between receptor activation and vesicular dynamics whose role in shaping the cellular response to stimuli has begun to garner attention (22). Furthermore, this bidirectional relationship could also allow the signaling activity of one receptor to influence the response properties of another through changes in its vesicular trafficking dynamics, generating context-dependent receptor activity.
Here, we demonstrated that Eph receptor activation at cell-cell contacts regulated the vesicular dynamics of EGFR by inhibiting Akt-dependent trafficking. By modulating the spatial distribution of EGFR activity, Eph receptor activation altered the cellular response to EGF stimulation, selectively suppressing EGF-promoted migratory signaling while preserving its effect on proliferation.
A1 stimulation (min) 5 2 0 4 0 Pre   regulating the abundance of EGFR at the PM. Therefore, we hypothesized that Eph receptor activation might reduce PM EGFR abundance by trapping constitutively recycling receptors in endosomes. Activation of Eph receptors decreases the activity of Akt (12,14,26,27), a signaling effector that regulates EGFR vesicular trafficking (16). A1 stimulation of Cos-7 cells, which endogenously express multiple EphA isoforms including EphA2, EphA3, EphA4, EphA7, and EphA8 (fig. S1C and table S1), decreased Akt activity ( Fig. 1C and  fig. S1D). A reduction in Akt activation and a concomitant loss of PM EGFR were also observed after Eph receptor activation in human embryonic kidney (HEK) 293 cells, mouse embryonic fibroblasts (MEFs), MCF10A cells, and MDA-MB-231 cells (fig. S1, E to H), which exhibit a wide range of EGFR expression ( fig. S1I). The abundance of PM EGFR over time followed the changes in Akt activity across the different cell types. Notably, in MEFs, changes in PM EGFR abundance followed a biphasic decrease in Akt activation in response to A1 stimulation (fig. S1F). Pharmacological inhibition of Akt with AktVIII ( Fig. 1D and fig. S1D) or of its downstream, early endosomal effector PIKfyve with YM201636 ( Fig. 1E) (16) reduced PM EGFR abundance in Cos-7 cells, as did knockdown of PIKfyve by small interfering RNA (siRNA) ( fig. S2A). Eph receptor activation and PIKfyve inhibition promoted a similar decrease in PM EGFR abundance (Fig. 1E). In addition, the combination of Eph receptor activation with PIKfyve inhibition did not further reduce PM EGFR abundance (Fig. 1E), suggesting a shared molecular mechanism. Consistent with a suppression of constitutive EGFR recycling, we observed an endosomal accumulation of ectopically expressed EGFR-mCherry in live cells after Akt or PIKfyve inhibition (Fig. 1F, movie S1, and fig. S2B). Time-lapse confocal imaging of Cos-7 cells expressing EGFR-mCherry and EphA2-mCitrine also revealed endosomal accumulation of EGFR with time after soluble A1 stimulation (Fig. 1, G and H, and movie S2) or upon presentation of ephrinA1 ligand on the membrane of adjacent cells at sites of cell-cell contact (movie S3), leading to a decrease in PM EGFR abundance (Fig. 1I). This shift in the spatial distribution of EGFR occurred primarily through the trapping of receptors in Rab5-positive early endosomes, as observed for both ectopically expressed (Fig. 1J, left) and endogenous EGFR (Fig. 1K), consistent with an inhibition of Akt-dependent trafficking (16). Thus, Eph receptor activation alters the subcellular distribution of EGFR before growth factor stimulation by inhibiting Akt/ PIKfyve-dependent vesicular recycling and trapping constitutively recycling receptors in Rab5-positive early endosomes.
We next investigated how Eph receptor activation influences EGFR trafficking during EGF stimulation. The trafficking fate of EGFR through the endosomal system is determined by posttranslational modifications, with the ubiquitination of ligand-bound receptors acting as a molecular signal that diverts EGFR through Rab7-positive late endosomes to lysosomes for degradation (25). Saturating EGF concentrations (>50 ng/ml) (28), therefore, generate a finite temporal signaling response by progressively depleting liganded EGFR through ubiquitin-dependent lysosomal degradation. Stimulation of endogenous receptors in Cos-7 cells with a saturating concentration of EGF (100 ng/ml) induced a ~40% reduction in total EGFR expression after 60 min of stimulation ( fig. S2, C and D), and residual EGFR resided primarily in Rab7-positive late endosomes (Fig. 1J, right). In contrast, A1 pretreatment or direct Akt inhibition impaired Rab5-to-Rab7 endosomal maturation (Fig. 1J, right) (29), leading to a reduction in receptor degradation at saturating EGF concentrations (≥50 ng/ml; fig. S2, C and D).
At subsaturating EGF concentrations typically found in human tissue secretions (0.4 to 20 ng/ml) (30), only a fraction of receptors are ligand-bound, receptor ubiquitination is reduced (31), and unliganded, nonubiquitinated receptors are recycled back to the PM (32). Therefore, during subsaturating EGF stimulation, the recycling of unliganded receptors is necessary to counter the EGF-induced depletion of receptors by endocytosis and maintain sensitivity to persistent growth factor stimulation (25). To assess whether Eph receptors inhibit EGFR recycling during subsaturating EGF stimulation, we exposed endogenous receptors in Cos-7 cells to a pulse of EGF (10 ng/ml) to induce EGFR endocytosis and measured its subsequent return to the PM after EGF washout (Fig. 1L). Whereas we observed a complete recovery of PM EGFR abundance in control cells after EGF washout, A1 pretreatment completely suppressed EGFR recycling. Thus, Eph receptor activation suppresses EGFR trafficking from the early endosome during EGF stimulation, impairing the recycling of nonubiquitinated receptors back to the PM and inhibiting the degradation of ubiquitinated receptors in the lysosome.

Eph receptor activation changes the spatial distribution of EGFR activity
Many functional outcomes to EGFR activation, such as cellular migration, require that cells remain responsive to persistent growth factor stimulation. To ensure sensitivity to stimuli during long periods of exposure, cells must maintain sufficient receptor abundance at the PM despite continuous internalization of activated receptors. We therefore posed the following questions: (i) Does Akt-dependent recycling help maintain cellular responsiveness to EGF during persistent, subsaturating stimulation, and (ii) can Eph receptor activation at cell-cell contacts change the response properties of EGFR by modulating its vesicular trafficking?
To address the impact of Akt-dependent recycling on EGFR activation, measurements of endogenous EGFR phosphorylation and trafficking in Cos-7 cells were obtained by immunofluorescence after subsaturating EGF stimulation in control cells and after Eph receptor activation or inhibition of Akt or PIKfyve. Individual cells were radially segmented to quantify changes in the average spatial distribution of EGFR abundance and activity with time and visualized using three-dimensional (3D) spatiotemporal maps ( Fig. 2A). Through an accumulation of EGFR in endosomal compartments during sustained EGF stimulation, cells pretreated with either A1 or an Akt inhibitor generated less EGFR phosphorylation after 60 min of EGF stimulation relative to control cells (Fig. 2, A and B). Decoupling Akt activation from its effect on trafficking by inhibiting PIKfyve had similar effects as direct Akt inhibition or A1 pretreatment on EGFR phosphorylation and trafficking (Fig. 2, A and B), indicating that Akt activity maintains EGFR activation at the PM during sustained, subsaturating EGF stimulation by promoting vesicular recycling.
To quantify EGFR phosphorylation at the PM and on endosomes during sustained, subsaturating EGF stimulation, we used fluorescence lifetime imaging microscopy (FLIM) to detect Förster resonance energy transfer (FRET) between EGFR-mCitrine and a phosphotyrosinebinding domain fused to mCherry (PTB-mCherry) (33) in Cos-7 cells (Fig. 2, C to F). In control cells, EGFR-mCitrine remained highly phosphorylated at both the PM and in endosomes after 60 min of sustained EGF-Alexa Fluor 647 stimulation (Fig. 2, C and D). In cells pretreated with A1 or after Akt or PIKfyve inhibition, we observed reduced PM EGFR-mCitrine density (Fig. 2E) and EGF-Alexa Fluor 647 binding (Fig. 2F), resulting in diminished EGFR-mCitrine phosphorylation specifically at the PM (Fig. 2D). In conditions in which Akt-dependent recycling was suppressed, ligand-bound and active EGFR-mCitrine accumulated in endosomes ( Fig. 2, C, E, and F), maintaining its phosphorylation in this compartment to the same extent as control cells (Fig. 2D). By inhibiting Akt-dependent recycling, Eph receptor activation thus changes the spatial distribution of EGFR activity during sustained, subsaturating EGF stimulation, selectively reducing EGFR activation at the PM while preserving receptor activity in endosomes.

Eph receptor activation at cell-cell contact alters the EGFR signaling response
Although EGFR continues to activate signaling effectors from endosomal membranes (34)(35)(36)(37)(38)(39)(40)(41)(42), Akt is preferentially activated at the PM ( fig. S3, A to D) (43,44). We therefore investigated how Eph receptor activation, by changing the spatial distribution of EGFR activity, regulates its signaling response during EGF stimulation. By suppressing vesicular recycling and reducing EGFR activity at the PM, A1 pretreatment selectively inhibited Akt activation during sustained, subsaturating EGF stimulation of endogenous ( To confirm that EphA2 inhibits EGF-promoted Akt activation by suppressing EGFR recycling and does not simply reflect the opposed regulation of Akt by EGFR and EphA2 (activation and inhibition, respectively), we assessed whether EGFR trafficking was dispensable for the A1-induced suppression of EGF-promoted Akt activation. Cells were prestimulated with A1, treated with the dynamin inhibitor dynole 34-2 to block subsequent endocytosis, and then stimulated with EGF. When EGFR endocytosis was blocked ( fig. S3C), A1 pretreatment did not reduce EGF-promoted Akt activation ( Fig. 3B, top). Pretreatment with the negative control analog dynole 31-2, to control for off-target effects, did not inhibit EGFR endocytosis ( fig. S3C) and did not affect the A1-induced suppression of EGF-promoted Akt activation (Fig. 3B, bottom), corroborating that intact EGFR vesicular trafficking is required for the inhibitory effect of Eph receptors on EGFR signaling. Increasing concentrations of A1 progressively inhibited EGFmediated Akt activation (Fig. 3C), suggesting that the degree of cell-cell contact might determine the magnitude of Akt activation in response to a given concentration of EGF. Use of a conformational FRETbased sensor of EphA2 activity (24) demonstrated that homotypic cell-cell contact promoted Eph receptor activation through interactions with ephrins presented on neighboring cells (Fig. 3D). Furthermore, activation of Eph receptors at cell-cell contacts promoted the recruitment of the effector SH2 domain tagged with mCherry (46) to phosphorylated receptors (Fig. 3D), indicating their signaling competency. To directly investigate the influence of cell-cell contact on EGFR signaling, we obtained single-cell measurements of Akt and ERK activation in Cos-7 with varying degrees of cell-cell contact.
Akt activation decreased with cell-cell contact both before and after EGF stimulation (Fig. 3E), demonstrating that increasing cell-cell contact reduces the magnitude of EGF-promoted Akt activation. In contrast, ERK activation was unaffected by cell-cell contact, with cells generating similar EGF-promoted increases in ERK activation irrespective of their degree of cell-cell contact (Fig. 3F).

Coupling EGFR activity to vesicular recycling generates positive feedback
Although the inhibition of Akt-dependent recycling resulted in reduced PM EGFR abundance in Cos-7 cells (Fig. 1, A, B, D, and E, and figs. S1, E to H, and S2A), we also observed that increasing cellular Akt activity through the inhibition of its negative regulator PP2A by okadaic acid (Fig. 4A) or ectopic expression of the constitutively active Akt D323A/D325A mutant ( Fig. 4B) (47) resulted in a concomitant increase in PM EGFR. Because EGFR activation itself increased Akt activity in cells (Fig. 3, A to C and E, and fig. S3, C and D), we next asked whether PM EGFR abundance is actively maintained during growth factor stimulation through an EGF-induced increase in Akt-dependent vesicular recycling. Using a fluorescence localization after photoactivation (FLAP) approach to quantify the vesicular recycling of EGFR to the PM after photoactivation of EGFR-paGFP in endosomes, we observed an increase in EGFR-paGFP recycling during EGF stimulation (Fig. 4C). Akt inhibition completely suppressed this EGF-promoted increase in vesicular recycling (Fig. 4C), further demonstrating the contribution of Akt-dependent recycling in sustaining PM EGFR activity. Thus, by stimulating Akt-dependent recycling, EGFR activation generates a positive feedback that actively maintains its PM abundance during EGF stimulation. Positive feedback in combination with inhibitory network motifs can convert graded inputs into switch-like, ultrasensitive signaling responses (48). Because Akt is preferentially activated at the PM (fig. S3, A to D), the EGF-induced increase in EGFR vesicular recycling ( Fig. 4C) might generate a positive feedback for Akt activation (Fig. 4D). To investigate whether this positive feedback can generate a switch-like activation of Akt, we measured Akt phosphorylation in thousands of individual Cos-7 cells by flow cytometry after sustained stimulation with a range of EGF concentrations (Fig. 4, E and F). Cells were stimulated in suspension to negate in situ cell-cell contact as an extrinsic source of variability in Akt activation (Fig. 3E). At concentrations of ≥1 ng/ml, EGF stimulation produced a switchlike activation to a high Akt phosphorylation state in a subpopulation of cells, whose proportion increased with EGF concentration (Fig. 4, E and F, top). Decoupling EGFR activation from its effect on vesicular recycling by PIKfyve inhibition (Fig. 4, E and F, middle) or A1 pretreatment (Fig. 4, E and F, bottom) did not result in a global decrease in cellular Akt activation but rather reduced the proportion of cells generating a high Akt phosphorylation state (Fig. 4F), consistent with the inhibition of a positive feedback that produces this switch-like response. Intrinsic cell-to-cell variability in the EGF threshold required to stimulate Akt-dependent vesicular recycling therefore determines the proportion of cells that transition to a high Akt activity state at a given EGF concentration. Eph receptor activation, by decoupling EGFR activation from its effect on vesicular trafficking, reduces Akt activation within the population by decreasing the proportion of cells transitioning to a high Akt activity state during EGF stimulation.
Eph activation at cell-cell contact suppresses the EGF-promoted transition to a migratory state EGFR signaling to effectors at the PM generates exploratory cellular behaviors (49)(50)(51)(52)(53)(54)(55)(56) that must be maintained to induce a persistent migratory response. Given that contact inhibitory Eph receptor activation selectively suppresses PM signaling during sustained, subsaturating EGF stimulation (Figs. 3, A and E, and 4, E and F), we investigated whether cell-cell contact regulates EGF-promoted migration by inhibiting Akt-dependent recycling. Because Cos-7 cells exhibit limited migratory behavior, we examined MEFs, which generate a haptotactic migratory response to fibronectin that is enhanced by EGF through an increase in exploratory behavior (57). Similar to Cos-7 cells, these cells express several EphA isoforms, including EphA2, EphA3, and EphA5 (58,59), and also exhibited an Eph activity-dependent depletion of PM EGFR abundance (Fig. 5, A and  B, and fig. S1F). After stimulation with a subsaturating EGF concentration (20 ng/ml), we observed a significant increase in the proportion of migratory cells (Fig. 5C, top; movie S4; and fig. S5, A and B) but no change in the average distance traveled per cell (Fig. 5C,  bottom). These findings indicate that EGF promotes the transition of individual cells to a migratory state rather than increasing overall cellular motility. Because EGF binding promotes receptor ubiquitination and degradation, leading to a loss in EGF sensitivity with time, sustained stimulation with supraphysiological, saturating EGF concentrations (100 ng/ml) did not significantly increase the proportion of migratory cells (Fig. 5C, top). Decoupling EGFR activation from its effect on Akt-dependent recycling through the inhibition of PIKfyve or after Eph receptor activation decreased the proportion of migratory cells (Fig. 5C, top). We observed further that increasing concentrations of A1 progressively decreased EGF-induced migration (Fig. 5C, top), consistent with its concentration-dependent effect on EGF-promoted Akt activation (Fig. 3C) and suggesting that the amount of ephrinA1-Eph receptor interactions at points of cellcell contact may determine whether a cell initiates a migratory response to EGF. We found that the number of migratory cells after EGF stimulation was inversely proportional to cell density (Fig. 5D) and that the increase in migration observed at low densities could be countered by treatment with soluble A1 to mimic Eph receptor contact inhibitory signaling (Fig. 5D). Thus, physiological Eph receptor activation at points of homotypic cell-cell contact suppresses EGF-promoted migration by inhibiting Akt-dependent vesicular recycling. The influence of Eph receptor activation on the migration of cells was also assessed using a transwell migration assay. EGF promoted a significant increase in the migration of MEFs, which was completely suppressed by A1 prestimulation or PIKfyve inhibition (Fig. 5E).
We next investigated the influence of Akt-dependent recycling on EGF-promoted migration in pathological contexts in which either Akt activity is dysregulated or EGFR is overexpressed. Knockout of phosphatase and tensin homolog (PTEN) in MEFs (Pten −/− MEFs), which increases cellular Akt activity (60), enhanced both autonomous and EGF-promoted directional migration (Fig. 5F). Similarly, the triple-negative breast cancer cell model MDA-MB-231, in which EGFR is overexpressed ( fig. S1I), exhibited an increase in motility relative to wild-type MEFs (Fig. 5G). Inhibition of Akt-dependent recycling by either A1 prestimulation or PIKfyve inhibition also blocked EGF-induced chemotaxis in both cell lines (Fig. 5, F and G), indicating that decoupling Akt activity from its effect on vesicular trafficking can inhibit cell migration even when Akt activity or EGFR expression is pathologically increased.
Suppression of Akt-dependent recycling selectively inhibited PM signaling while leaving endosomal ERK activation intact (Fig. 3A). Consistent with this result, we found that neither PIKfyve inhibition nor A1 pretreatment reduced EGF-promoted cell proliferation (Fig. 5H) in wild-type MEFs. Thus, by altering the spatiotemporal distribution of EGFR activity, contact inhibitory signaling by Eph receptors influences the cellular outcome to EGF stimulation, preserving a proliferative response while suppressing cell migration.

Modulation of vesicular dynamics may represent a general mechanism to produce context-dependent EGFR signaling
To determine whether environmental signals other than cell-cell contact can influence the cellular response to EGF stimulation through changes in EGFR trafficking, we investigated the effect of activation of the G protein-coupled receptor Kiss1 (Kiss1R), which, similar to Eph receptors, inhibits Akt (61) and suppresses cell migration and metastatic invasion (62). Stimulation with the soluble Kiss1R ligand kisspeptin-10 (Kp-10) reduced Akt activity in HEK293 cells and decreased PM EGFR abundance (Fig. 6A). Similar to the effect of cellcell contact, pretreatment with Kp-10 selectively inhibited EGF-promoted Akt activation (Fig. 6B) while preserving ERK activation (Fig. 6C). Furthermore, activation of Kiss1R completely suppressed EGF-promoted migration of both wild-type and Pten −/− MEFs, as well as MDA-MB-231 breast cancer cells (Fig. 6, D to F). The modulation of EGFR vesicular trafficking dynamics could therefore provide a general mechanism to generate plasticity in the signaling response to EGFR activation, through which diverse environmental signals such as cell-cell contact or soluble stimuli like Kp-10 can influence the cellular response to EGF.

DISCUSSION
Here, we demonstrated that Eph receptor activation at cell-cell contacts generated context-dependent cellular responses to EGF stimulation by modulating EGFR vesicular trafficking dynamics. Chemotaxis requires that cells remain responsive to stimuli for prolonged periods of time as they migrate toward the chemotactic source. EGF stimulation promoted an increase in Akt-dependent recycling (Fig. 4C), which maintains sensitivity to EGF by sustaining unliganded EGFR abundance at the PM to counter the depletion of liganded receptors by endocytosis. A1-promoted inhibition of both the constitutive and EGF-induced recycling of unliganded EGFR thereby changes the signaling output of the receptor and alters the cellular response to EGF stimulation. Because Akt itself is preferentially activated at the PM ( fig.  S3, A to D), the EGF-promoted increase in vesicular recycling generates a positive feedback that switches cells to a high Akt activation state (Fig. 4, E and F). Although Akt has previously been observed on endosomal membranes through interactions with the early endocytic adaptor protein APPL1 (47,63), de novo activation of Akt by EGFR requires the production of phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P 3 ], which is impeded by the low abundance of phosphatidylinositol 4,5-bisphosphate [PI(4,5)P 2 ] in endosomal membranes (43,64). Akt activation may occur, to some extent, on endosomal membranes (65); however, because the coupling of active EGFR to Akt activation will be more efficient at the PM, any perturbations that influence the spatial distribution of EGFR, such as Eph or Kiss1R activation, should influence the capacity of EGFR to activate Akt (Figs. 3, A and E, and 6, A and B). We observed that the switch to a high Akt activity state only occurred in a proportion of cells, even in the absence of in situ cell-cell contacts, and increased with EGF concentration (Fig. 4, E and F  been previously attributed to cell-to-cell variation in PI3K expression (66). Our data suggest that intrinsic variability in the expression of signaling and/or trafficking effectors in individual cells may determine the EGF concentration required to stimulate Akt-dependent trafficking and engage the positive feedback that produces a high Akt activity state. Small differences in EGF concentration substantially influenced the proportion of cells generating a high Akt res ponse (for example, a shift from 5 to 10 ng/ml increased the proportion of cells from 43 to 85%, respectively; Fig. 4, E and F). It may not be coincidental that the concentration range over which this switch occurs corresponds to the physiological range of EGF concentrations (30). By generating a sharp boundary for Akt activation within the physiological EGF concentration regime, even slight changes in the threshold of this switch could have profound implications for tissue dynamics (for example, the initiation of migration). Eph receptor activation decreased the proportion of cells generating a high Akt response from 85 to 41% in response to a given concentration of EGF (10 ng/ml) (Fig. 4, E and F). The dependence of Akt activation on EGFR recycling thus allows the degree of cell-cell contact to regulate the proportion of cells generating a migratory response to EGF stimulation. PI3K/Akt signaling has previously been suggested as the point of convergence for EGFR/Eph control of cell migration (15); however, the molecular mechanism underlying this oppositional relationship remained unclear. Our results indicate that Eph receptor activation inhibits EGF-promoted cell migration by suppressing Akt-dependent recycling. By inhibiting EGFR recycling, Eph activation impedes the spatially maintained positive feedback that generates a high Akt response and decreases the sensitivity of cells to persistent EGF stimulation, which is necessary to maintain an exploratory behavior. However, by changing the spatial distribution of EGFR activity (Fig. 2, C and D), Eph receptor activation selectively suppressed migratory signaling from the PM while leaving proliferative ERK signaling intact (Fig. 3, A, E, and F). This contextual plasticity generates two distinct cellular outcomes to EGF stimulation that may be important in physiological settings such as wound healing. At the tissue boundary, cells with reduced cell-cell contact would increase their exploratory behavior in response to EGF released at the site of the wound. Cells located deeper in the tissue, despite extensive cell-cell contacts, would retain their proliferative response to extracellular EGF and undergo mitosis to fill the vacant space created as exploratory cells migrate to occupy the wound area.
Our observations demonstrate that communication between receptors with opposed functionality can emerge through changes in vesicular trafficking dynamics rather than relying on direct interactions between the receptors or their respective effectors. Such a mechanism also allows different receptors with similar functional roles (for example, EphA2 and Kiss1R) to alter the cellular response to stimuli without having to evolve distinct protein interaction domains to do so. The dependency of EGFR signaling on its vesicular dynamics could confer a general mechanism through which the cell can generate functional plasticity to growth factor stimulation while preserving specificity in cell-cell communication.

Antibodies
The primary antibodies used were as follows: mouse anti-Akt  . Cells were pretreated for 1 hour with serum-free medium or Kp-10 (purple, 100 nM) before seeding (means ± SEM from three independent experiments). Control and EGF data were previously presented in Fig. 5 (E to G).

In-Cell Western and On-Cell Western
Cells were seeded on black, transparent-bottomed 96-well plates (3340, Corning) coated with poly-l-lysine (P6282, Sigma-Aldrich). Cells were fixed with Roti-Histofix 4% (Carl Roth) for 5 min at 37°C. For ICW, cells were permeabilized with 0.1% (v/v) Triton X-100 for 5 min at room temperature. For OCW, cells were not permeabilized. Samples were incubated in Odyssey tris-buffered saline (TBS) blocking buffer (LI-COR Biosciences) for 30 min at room temperature. Primary antibodies were incubated overnight at 4°C, and secondary antibodies (IRDyes, LI-COR Biosciences) were incubated in the dark for 1 hour at room temperature. All wash steps were performed with TBS (pH 7.4). Intensity measurements were made using the Odyssey Infrared Imaging System (LI-COR Biosciences). ICWs were calibrated by Western blots to ensure accurate quantification ( fig. S1D). Quantification of the integrated intensity in each well was performed using the Micro Array Profile plugin (OptiNav Inc.) for ImageJ v1.47 (http://rsbweb.nih. gov/ij/). In each ICW or OCW, two to four replicates per condition were obtained per experiment, and all data presented represent means ± SEM from at least three independent experiments.

Immunofluorescence
Cells were cultured on four-or eight-well chambered glass slides (Lab-Tek) and fixed with 4% (w/v) paraformaldehyde/phosphatebuffered saline (PBS) for 10 min at 4°C. To measure PM EGFR, fixed, nonpermeabilized samples were first incubated with primary antibody directed at an extracellular epitope of EGFR (AF231, R&D Systems) overnight at 4°C followed by secondary antibody for 1 hour at room temperature. For all other immunofluorescence experiments, samples were permeabilized with 0.1% (v/v) Triton X-100 for 5 min at room temperature before incubation with primary antibodies. All wash steps were performed with TBS (pH 7.4). Fixed samples were imaged in PBS at 37°C. For all analyses, an initial background subtraction was performed on immunofluorescence images. To quantify the proportion of EGFR in Rab5 or Rab7 compartments, binary masks were generated from intensity thresholded images of Rab5 and Rab7 staining. To generate a mask of Rab5/Rab7 doublepositive endosomes, the product of their individual masks was used.
The integrated fluorescence intensity of EGFR-mCherry was determined in each of the endosomal masks and divided by the total integrated EGFR fluorescence intensity of the cell. Image analysis was performed using ImageJ. A cell segmentor tool was developed inhouse in Anaconda Python (Python Software Foundation version 2.7; www.python.org/) to quantify the spatial distribution of EGFR and EGFR(pTyr 845 ) in fixed cells. Cells were divided into six equally spaced radial bins emanating from the plasma membrane.

Confocal imaging
Cells were cultured for live-cell confocal imaging on four-or eightwell chambered glass slides (Lab-Tek) and transiently transfected as described above. Confocal images were recorded using an Olympus FluoView FV1000 confocal microscope (Olympus Life Science Europa) or a Leica SP8 confocal microscope (Leica Microsystems).

Leica SP8
The and 12-bit images of 512 × 512 pixels were acquired in a frame-byframe sequential mode.

Analysis of time-lapse confocal imaging
All analysis of live-cell imaging data required an initial background subtraction for images obtained. To quantify the proportion of endosomal EGFR-mCherry or EphA2-mCitrine, binary masks of endosomes were generated from intensity thresholded images. The integrated fluorescence intensity of EGFR-mCherry and EphA2-mCitrine was determined in their corresponding endosomal masks and divided by the total integrated fluorescence intensity of the cell. FLAP experiments were carried out at 37°C on a Leica SP8. EGFR-mCherry was coexpressed to identify and select regions of endosomal EGFR for photoactivation. Background intensity of EGFR-paGFP before photoactivation was measured and subtracted from postactivation images. Photoactivation of EGFR-paGFP was performed with the 405-nm laser at 90% power. After photoactivation, fluorescence images of EGFR-paGFP were acquired every minute for a total of 15 min. PM EGFR-paGFP fluorescence was quantified as the integrated intensity in a five-pixel ring of the cell periphery and, after subtracting preactivation background intensity, was calculated as the proportion of total EGFR-paGFP intensity.
Fluorescence lifetime imaging microscopy EGFR-mCitrine, PTB-mCherry, and HA-c-Cbl-BFP were ectopically expressed in Cos-7 cells. Fluorescence lifetime measurements of EGFR-mCitrine were performed at 37°C on a Leica SP8 equipped with a time-correlated single-photon counting module (LSM Upgrade Kit, PicoQuant) using a 63×/1.4 NA oil objective. EGFR-mCitrine was excited using a pulsed white light laser at a frequency of 20 MHz and wavelength of 514 nm, and fluorescence emission was restricted to 525 to 570 nm with an AOBS. Photons were integrated for a total of approximately 2 min per image using the SymPhoTime software V5.13 (PicoQuant). Data analysis was performed using custom software in Anaconda Python based on global analysis, as described in (67). Fluorescence lifetime measurements of LIFEA2 were performed and analyzed as previously described (24).

Flow cytometry
Cells were detached using Accutase, centrifuged at 200g for 5 min, and resuspended in serum-free DMEM before EGF stimulation. Cells were fixed with 5% (w/v) sucrose/Roti-Histofix for 15 min at 37°C. Ice-cold methanol was added to 90% (v/v) for 30 min on ice. Cells were rinsed once with 0.5% (w/v) bovine serum albumin/TBS and incubated with Odyssey TBS blocking buffer (LI-COR Biosciences) for 30 min at room temperature. Anti-phospho-Akt(Ser 473 )-Alexa Fluor 647 (4075, CST) was added directly to blocking buffer and incubated overnight at 4°C. Anti-Akt-Alexa Fluor 488 (2917, CST) was added for 2 hours before measurement. Samples were analyzed using an LSR II flow cytometer (BD Biosciences). Alexa Fluor 488 was excited with a 488-nm laser, and fluorescence emission was collected using a 505-nm long-pass dichroic and a 530/30-nm filter. Alexa Fluor 647 was excited with 633-nm lasers, and fluorescence emission was collected using a 670/40-nm filter. Samples were analyzed using FlowJo v10 (FlowJo LLC) to obtain single-cell intensity measurements of phosphorylated and total Akt. Population distributions of log(phosphorylated/total Akt) were fitted with a single Gaussian or a sum of two Gaussian distributions using GraphPad Prism (GraphPad Software).

Cell migration
MEFs were seeded onto fibronectin-coated (1.25 g/cm 2 ; F0895, Sigma) 12-well culture dishes (83.3921, Sarstedt) containing twowell Culture-Inserts (80209, Sarstedt) to create a cell-free area. Immediately before stimulation, inserts were removed and cells were incubated with Hoechst to label nuclei. Wide-field images were acquired using an Olympus IX81 inverted microscope equipped with an MT20 illumination system, a 4×/0.16 NA air objective, and an Orca charge-coupled device camera (Hamamatsu Photonics). Transmission and fluorescence images were acquired every 10 min for 16 hours. The cell-free area created by the Culture-Insert was cropped using ImageJ and defined as the migration region. Individual cells were detected and tracked by their nuclear Hoechst staining as they traveled within the migration region using TrackMate ImageJ plugin (68), and the total distance of each track was quantified. To distinguish between migratory cells and cells that moved into the migration region due to population expansion over the course of the experiment, a minimum migration distance threshold that separated the population of nonmigrating cells from migratory cells was determined ( fig. S5, A and B).

Transwell migration assay
Cells were serum-starved overnight, detached with Accutase, and pretreated with compounds in suspension for 1 hour at 37°C, when required. Cells were subsequently seeded in serum-free medium at a density of 20,000 per well into the upper chamber of CIM-Plates (ACEA Biosciences Inc.). EGF (20 ng/ml) or serum-free medium was added to the lower chamber of the CIM-Plates as the chemoattractant and negative control, respectively. Transwell migration was quantified as cells migrated through a microporous membrane toward the chemoattractant in the lower chamber to microelectrode sensors, generating a real-time increase in impedance measured by the xCELLigence RTCA DP Instrument (ACEA Biosciences Inc.).

Reverse transcription quantitative polymerase chain reaction
Cos-7 cells were trypsinized and pelleted, and mRNA extraction was carried out with TRIzol Reagent (Thermo Fisher) according to the manufacturer's instructions. Reverse transcription quantitative polymerase chain reaction (RT-qPCR) was performed with the Luna Universal One-Step RT-qPCR Kit (New England Biolabs) with validated primers (table S1). If the differences between C t values for a given transcript and a matched negative control (NRT, same primer pair without reverse transcription) were not statistically significant using a two-tailed Student's t test, then the transcript was classified as not detected. Transcript abundance was normalized to the housekeeping gene TATA box-binding protein (TBP) and calculated as 2 −(sample Ct − TBP Ct) .

Immunoprecipitation and Western blotting
Cells were lysed in TGH [150 mM NaCl, 2 mM EGTA/EDTA, 50 mM Hepes (pH 7.4), 1% Triton X-100, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, and 10 mM N-ethylmaleimide (NEM)] or RIPA [for immunoprecipitation; 50 mM tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EGTA, 1 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, 0.2% SDS, 2.5 mM sodium pyrophosphate, and 10 mM NEM], supplemented with Complete Mini EDTA-free protease inhibitor (Roche Applied Science) and 100 l of phosphatase inhibitor cocktail 2 and 3 (P5726 and P0044, Sigma-Aldrich). Lysates were sonicated before centrifugation at 14,000 rpm for 10 min at 4°C to pellet nonsoluble material. For immunoprecipitation, cell lysates were incubated with 50 l of washed protein G magnetic beads (10003D, Life Technologies) for 1 hour at 4°C to preclear nonspecific binding proteins from samples. Supernatants were incubated with primary antibody alone for 2 hours followed by the addition and overnight incubation with protein G magnetic beads at 4°C with agitation. SDS-polyacrylamide gel electrophoresis was performed using an XCell II mini electrophoresis apparatus (Life Technologies) according to the manufacturer's instructions. Samples were transferred to preactivated polyvinylidene difluoride membranes (Merck Millipore) and incubated with the respective primary antibodies at 4°C overnight. Detection was performed using species-specific IRDye secondary antibodies (LI-COR Biosciences) and the Odyssey Infrared Imaging System (LI-COR Biosciences). The integrated intensity of protein bands of interest was measured using the ImageJ software, and signals were normalized by dividing the intensities of phosphorylated protein by total protein intensities or by dividing intensities of coimmunoprecipitated proteins by the corresponding immunoprecipitated protein.