Contact inhibitory Eph signalling decouples EGFR activity from vesicular recycling to generate contextual plasticity

The ability of cells to adapt their behaviour to growth factors in relation to their environment is an essential aspect of tissue development and homeostasis. Here we show that cell-cell contact can change the outcome of EGFR activation by altering its vesicular trafficking. EGFR promotes Akt-mediated vesicular recycling to maintain its plasma membrane (PM) abundance during EGF stimulation. This self-sustained vesicular recycling of EGFR generates a positive feedback that maintains PM Akt signalling and promotes migration. By decoupling EGFR activation from its vesicular recycling, Eph activity at cell-cell contacts impedes the positive feedback that maintains PM signalling and traps EGFR in endosomes. Through this change in the spatial distribution of EGFR activity, cell-cell contact selectively suppresses migratory PI3K/Akt signalling from the PM, while preserving proliferative ERK signalling from endosomes. Thus, by altering the vesicular trafficking of EGFR, the cellular environment can modulate its signalling to generate diverse outcomes to EGF stimulation.

cell contacts impedes the positive feedback that maintains PM signalling and traps EGFR in 23 endosomes. Through this change in the spatial distribution of EGFR activity, cell-cell 24 contact selectively suppresses migratory PI3K/Akt signalling from the PM, while 25 preserving proliferative ERK signalling from endosomes. Thus, by altering the vesicular 26 trafficking of EGFR, the cellular environment can modulate its signalling to generate 27 diverse outcomes to EGF stimulation. sustained stimuli by maintaining PM receptor abundance (Baumdick et al. 2015). EGF 50 binding promotes receptor ubiquitination (Sigismund et al. 2013), which re-routes 51 internalized EGFR to lysosomal degradation (Baumdick et al. 2015). Therefore, at 52 subsaturating EGF concentrations typically found in human tissue secretions (0.4-20 53 ng/ml) (Konturek et al. 1989), the cell maintains its sensitivity to persistent stimulation by 54 continually recycling unliganded, non-ubiquitinated receptors to the PM. Saturating EGF 55 concentrations (> 50 ng/ml) (Macdonald and Pike 2008), on the other hand, generate a 56 finite temporal response by progressively depleting PM and total EGFR abundance through 57 ubiquitin-dependent lysosomal degradation. 58 59 (Fig. 1f). Therefore, by impeding trafficking from the early endosomal compartment, Akt 122 regulates both the recycling of unliganded, non-ubquitinated receptors to the PM and the 123 degradation of liganded, ubiquitinated receptors in the lysosome. 124 125 We next investigated how these effects of Akt on EGFR vesicular trafficking influence 126 receptor response properties during EGF stimulation. By decreasing PM receptor 127 abundance prior to and during EGF stimulation (Fig. 1b,e), pretreatment with an Akt 128 inhibitor significantly decreased the proportion of phosphorylated receptors at the 129 autocatalytic Y845, the effector docking site Y1068 and the Cbl-mediated ubiquitination 130 regulatory site Y1045 (Baumdick et al. 2015) following 30 min of sustained EGF stimulation 131 ( Fig. 1g). At higher EGF concentrations (≥ 50 ng/ml), Akt inhibition also suppresses 132 receptor degradation (Fig. 1f), trapping internalized EGFR in endosomal compartments 133 accessible to ER-bound PTPs with high activity (Yudushkin et al. 2007; Haj et al. 2002), 134 further decreasing the proportion of phosphorylated receptors in the entire cell (Fig. 1g). Y845, which stabilizes their active conformation (Baumdick et. al, 2017), autocatalytic 141 activation can further propagate to additional unliganded EGFR. This autocatalytic 142 component in EGFR activation increases non-linearly with PM receptor density. We 143 therefore hypothesized that Akt activity might regulate EGFR autocatalysis by controlling 144 receptor abundance at the PM. As receptor occupancy increases at higher EGF 145 concentrations, EGFR phosphorylation becomes less dependent on autocatalysis. 146 Therefore, to specifically measure the effect of Akt on EGFR autocatalytic activation, we 147 stimulated cells with increasing, subsaturating concentrations of EGF-Alexa647 (1, 3, 10 148 and 20 ng/ml) and simultaneously measured EGF binding and EGFR activation (by 149 translocation of the EGFR phospho-tyrosine binding (PTB) domain) at the PM in individual 150 cells by time lapse imaging ( Fig. 1h-j, Supplementary Fig. 2). Compared to control cells 151 (Fig. 1h), pretreatment with an Akt inhibitor decreased both EGF binding and EGFR 152 phosphorylation (Fig. 1i), consistent with a reduction in PM receptor abundance prior to 153 EGF stimulation (Fig. 1b,e). Plotting the relationship between the number of liganded 154 receptors (i.e. EGF-Alexa647 binding) and the extent of EGFR phosphorylation revealed 155 that phosphorylation increases non-linearly with EGF binding above a certain threshold in 156 control cells, consistent with an autocatalytic activation (Fig. 1j). Pretreatment with an Akt 157 inhibitor, on the other hand, generated less PM EGFR phosphorylation for equivalent 158 numbers of liganded receptors by impeding its autocatalytic activation (Fig. 1j). Therefore, 159 by promoting the constitutive recycling of EGFR to maintain its PM density prior to EGF 160 stimulation ( Fig. 1a-e), Akt activity fosters the autocatalytic activation of EGFR at 161 subsaturating EGF concentrations. 162

163
Since EGF stimulation itself increases Akt-dependent vesicular recycling (Fig. 1e), we 164 further reasoned that EGFR might self-sustain its activation during persistent, 165 subsaturating EGF stimulation. To further investigate how Akt-dependent recycling 166 influences PM EGFR autocatalytic activation during EGF stimulation, measurements of 167 EGFR phosphorylation and trafficking were obtained by immunofluorescence prior to and 168 following subsaturating EGF stimulation (20 ng/ml) for 5, 30 and 60 min. Cells were 169 radially segmented to quantify changes in the spatial distribution of EGFR with time and 170 visualized using 3D spatial-temporal maps (Fig. 1k, Stanoev et al. 2017). By reducing PM 171 receptor density prior to EGF stimulation (Fig. 1k, top), pretreatment with an Akt inhibitor 172 decreased the proportion of phosphorylated EGFR at the PM after 5 min of EGF stimulation 173 (Fig. 1k, bottom), consistent with reduced EGFR autocatalysis (Fig. 1j). By continually 174 delivering internalized, non-ubiquitinated receptors to the PM to both bind additional EGF 175 and act as a substrate for autocatalysis, the EGF-induced increase in vesicular recycling 176 ( Fig. 1e), maintains a high proportion of phosphorylated receptors at the PM during 177 sustained stimulation (Fig. 1k, bottom). Akt inhibition, on the other hand, by suppressing 178 the EGF-induced increase in receptor recycling (Fig. 1e) also reduced EGFR autocatalytic 179 activation at the PM at later time points (Fig. 1k, bottom) by trapping internalized 180 receptors in endosomes (Fig. 1k, top). Akt promotes vesicular trafficking through the 181 activation of the Rab5 effector PIKfyve (Er et al. 2013). PIKfyve inhibition led to an 182 endosomal accumulation of EGFR (Supplementary Fig. 3a,b) and decreased PM EGFR 183 abundance (Fig. 1k, top), consistent with a suppression of Akt-dependent recycling. 184 Decoupling Akt activation from its effect on trafficking by PIKfyve inhibition had 185 indistinguishable effects from direct Akt inhibition on EGFR phosphorylation and 186 trafficking (Fig. 1l), further validating that Akt activity sustains autocatalytic EGFR 187 activation at the PM through its effects on vesicular recycling. 188 189 Therefore, by stimulating an increase in Akt-dependent recycling, EGFR maintains its 190 sensitivity to EGF, effectively generating a positive feedback that sustains autocatalytic 191 EGFR activation at the PM during persistent, subsaturating stimulation (Fig. 1l). led to Rab5-positive early endosomal accumulation of EGFR ( Fig. 2a-c, Supplementary  211 Movie 2) and a reduction in PM EGFR abundance in Cos-7 (Fig. 2d, Supplementary Fig.  212 4c-d), HEK293 (Supplementary Fig. 4e), MCF10A (Supplementary Fig. 4f) and MDA-MB-213 231 cells (Supplementary Fig. 4g). EGFR also accumulated in endosomes upon 214 physiological presentation of ephrinA1 ligand at sites of cell-cell contact (Supplementary 215 Movie 3). Consistent with the delayed endosomal accumulation of EGFR relative to EphA2 216 ( Fig. 2b,d), we confirmed that the loss of PM EGFR does not result from the increased 217 formation of a heterodimer (Supplementary Fig. 5) or through the transactivation of 218 EGFR by EphA2 (Fig 2e). To directly assess whether Eph receptors decrease PM EGFR 219 through an inhibition of Akt-mediated EGFR recycling, we stimulated cells for 15 min with 220 a subsaturating concentration of EGF to induce EGFR endocytosis and measured the 221 subsequent recycling of internalized receptors back to the PM after EGF washout (Fig. 2f). 222 In control cells, the PM abundance of EGFR recovered to pre-EGF levels following washout, 223 while in A1-pretreated cells, EGFR recycling was significantly inhibited (Fig. 2f), consistent 224 with an EphA2-mediated inhibition of Akt-dependent recycling. and PIKfyve inhibition (Fig. 1k), A1 pretreatment reduced autocatalytic activation of EGFR 232 at the PM during persistent, subsaturating EGF stimulation, while preserving EGFR 233 phosphorylation in endosomes (Fig. 3a). Although EGFR can continue to activate some  (Fig 3b bottom, Supplementary Fig. 6). To confirm that 242 EphA2 inhibits EGF-promoted Akt activation by suppressing EGFR recycling and does not 243 simply reflect the opposed regulation of Akt by EGFR and EphA2 (activation vs inhibition, 244 respectively), we assessed whether EGFR trafficking was dispensable for the A1-induced 245 suppression of EGF-promoted Akt activation. Cells were prestimulated with A1 for 30 246 minutes, followed by treatment with the dynamin inhibitor dynole 34-2 to block 247 subsequent endocytosis, and then stimulated with EGF. When EGFR endocytosis was 248 blocked (Supplementary Fig. 1c), A1 pretreatment had no effect on EGF-promoted Akt 249 activation ( Fig. 3c top), while the negative control analogue, dynole 31-2, which does not 250 inhibit EGFR endocytosis (Supplementary Fig. 1c), had no effect on A1-induced 251 suppression of EGF-promoted Akt activation (Fig. 3c bottom), corroborating that intact 252 decreased PM EGFR abundance (Fig. 3g). Similar to the effect of cell-cell contact, 274 pretreatment with Kp-10 selectively inhibited EGF-promoted Akt activation (Fig. 3h), 275 while preserving ERK activation (Fig. 3i). Changes in EGFR vesicular trafficking dynamics, 276 therefore, provides a general mechanism to generate plasticity in the signalling response to 277 EGFR activation, through which diverse environmental signals such as cell-cell contact or 278 soluble stimuli like Kp-10 can influence the cellular response to EGF. high Akt response, consistent with an inhibition of the positive feedback that produces this 300 switch-like response. Intrinsic cell-to-cell variability in the EGF threshold required to 301 stimulate Akt-dependent vesicular recycling, therefore, determines the proportion of cells 302 that transition to a high Akt activity state at a given EGF concentration. 303

304
If the transition to a high Akt activity state is dependent on the positive feedback produced 305 by coupling EGFR activation to vesicular recycling, we should also observe similar cell-to-306 cell variability in PM EGF binding during sustained stimulation. We therefore performed 307 single cell time lapse imaging and quantified PM EGF binding during persistent stimulation 308 with a subsaturating concentration of EGF-Alexa647 (20 ng/ml) ( Fig. 3l-n). Cells that 309 successfully engage Akt-dependent recycling should exhibit a more sustained EGF binding 310 at the PM during stimulation. Cells were grouped by their rate of loss in PM EGF binding 311 using k-means clustering into two groups exhibiting a slow (Group 1) or fast (Group 2) loss 312 of PM EGF during stimulation ( Fig. 3l- to induce a persistent migratory response. Given that contact inhibitory Eph receptor 333 activation selectively suppresses PM signalling during persistent, subsaturating EGF 334 presentation (Fig. 3b,e,f), we investigated if cell-cell contact regulates EGF-promoted 335 migration by inhibiting Akt-dependent recycling. NIH 3T3 mouse embryonic fibroblast 336 (MEF) cells were grown on fibronectin in the presence of a silicone barrier that, once 337 removed, creates a cell-free area into which individual cells can be tracked following 338 uniform stimulation with EGF. When grown on fibronectin in 2-D cultures, fibroblasts 339 generate an autonomous, haptotactic migratory response which is enhanced by an EGF-340 induced increase in their exporatory behaviour (Li et al. 1999). Following stimulation with 341 subsaturating EGF concentrations (20 ng/ml), we observed a signficant increase in the 342 proportion of migratory cells (Fig. 4a, top, Supplementary Movie 4, Supplementary Fig.  343   7), but no change in the average distance travelled by migrating cells (Fig. 4a, bottom), 344 indicating that EGF promotes the transition of individual cells to a migratory state rather 345 than increasing overall cellular motility. Since ligand binding promotes receptor 346 ubiquitination and degradation, persistent stimulation with saturating, supraphysiological 347 EGF concentrations (100 ng/ml) induces a loss in EGF sensitivity with time and thus does 348 not significantly increase the proportion of migratory cells (Fig. 4a, top). Decoupling EGFR 349 activation from its effect on trafficking through the PIKfyve inhibition or A1 pretreatment 350 decreased the proportion of migratory cells (Fig. 4a), demonstrating that the positive 351 feedback generated by the EGF-induced increase in Akt-dependent recycling promotes the 352 transition to a migratory state. We observed further that increasing concentrations of A1 353 progressively decreased EGF-induced migration (Fig. 4a), consistent with its 354 concentration-dependent effect on EGF-promoted Akt activation (Fig. 3d) and suggesting 355 that the amount of ephrinA1-Eph receptor interactions at points of cell-cell contact may 356 determine whether a cell initiates a migratory response to EGF. Consistent with this 357 observation, we found that the number of migratory cells following EGF stimulation was 358 inversely proportional to cell density (Fig. 4b) and that the increase in migration observed 359 at low densities could be countered by treatment with soluble A1 to mimic Eph receptor 360 contact inhibitory signalling (Fig. 4b). Thus, physiological Eph receptor activation at points 361 of cell-cell contact suppresses EGF-promoted migration by inhibiting Akt-dependent 362 vesicular recycling. However, by preserving endosomal ERK activation following EGF 363 stimulation (Fig. 3b, bottom), we found that neither PIKfyve inhibition nor A1 364 pretreatment led to a reduction in EGF-promoted cell proliferation (Fig. 4c) (Fig. 1e). Saturating, supraphysiological EGF concentrations, on the other hand, 381 increase the proportion of liganded, ubiquitinated receptors that unidirectionally traffic, in 382 an Akt-dependent manner, from early to late endosomes to be degraded in lysosomes (Fig.  383  Since Akt itself is preferentially activated at the PM (Supplementary Fig. 1), the EGF-393 promoted increase in vesicular recycling generates a positive feedback that switches cells 394 to a high Akt activation state (Fig. 3j-k). Although Akt has been previously observed on the coupling of active EGFR to Akt activation will be more efficient at the PM, any 401 perturbations that influence the spatial distribution of EGFR will influence the capacity of 402 EGFR to activate Akt (Fig. 3b, Supplementary Fig. 1c). 403

404
We observed that the switch to a high Akt activity state only occurs in a proportion of cells, 405 which increases with EGF concentration (Fig. 3j-k). Population heterogeneity in Akt 406 activation has been previously attributed to cell-to-cell variation in PI3K expression (Yuan 407 et al. 2011). Our data suggest that intrinsic variability in the expression of signalling and/or 408 trafficking effectors may determine, for a given cell, the EGF concentration required to 409 stimulate Akt-dependent trafficking and engage the positive feedback that produces a high 410 Akt activity state. Small differences in EGF concentration substantially influence the 411 proportion of cells generating a high Akt response (e.g. a shift from 5 to 10 ng/ml increases 412 the proportion of cells from 43 to 85%, respectively, Fig. 3j-k) Eph receptor activation, for example, by suppressing EGFR recycling, decreased the 420 proportion of cells generating a high Akt response from 85 to 41% in response to 10 ng/ml 421 EGF (Fig. 3j-k). The dependence of Akt activation on EGFR recycling allows the degree of proliferative ERK signalling intact (Fig. 3b,e-f, Fig. 4)

Competing financial interests 472
The authors declare that no competing interests exist. was restricted with an AOBS as follows: Cerulean @ 468-505 nm, Alexa488 @ 498-551 nm, 593 mCitrine @ 525-570 nm, mCherry @ 570-650 nm and Alexa647 @ 654-754 nm. The 594 pinhole was set to 250 μm and 12-bit images of 512x512 pixels were acquired in a frame-595 by-frame sequential mode. Photoactivation experiments were carried out at 37°C on a Leica SP8. EGFR-mCherry was 615 co-expressed to identify and select regions of endosomal EGFR for photoactivation. 616 Background intensity of EGFR-paGFP prior to photoactivation was measured and 617 subtracted from post-activation images. Photoactivation of EGFR-paGFP was performed 618 with the 405nm laser at 90% power. Following photoactivation, fluorescence images of 619 EGFR-paGFP were acquired every minute for a total of 15 minutes. PM EGFR-paGFP 620 fluorescence was quantified as the integrated intensity in a 5-pixel ring of the cell 621 periphery and, after subtracting pre-activation background intensity, was calculated as a 622 proportion of total EGFR-paGFP intensity. condition). Control data previously presented in Figure 1k. (b