Molecules called growth factors can stimulate cells to grow, divide, or differentiate into more specialised cell types. Cells detect these molecules via proteins called receptor tyrosine kinases that span their surface membrane. The growth factor binds to the portion of the receptor outside the cell, which makes the receptor send signals to the cell's nucleus that change how the cell grows, divides, or specialises.

Different growth factors and receptor tyrosine kinases affect cell development in different ways. However, it was unclear how this occurred, as the receptors all send signals via the same signalling pathways. Some researchers proposed that specific responses could be triggered if some receptor tyrosine kinases activated these pathways more strongly than other receptors, or if they activated the pathways for different lengths of time.

Now, Villaseñor et al. have looked at a receptor tyrosine kinase for a growth factor called EGF. Activated EGF receptors are marked with a phosphate group (or ‘phosphorylated’) and are then removed from the surface membrane and packaged into structures within the cell called endosomes. Villaseñor et al. found that different endosomes contain the same mean amount of phosphorylated EGF receptor. When exposed to higher EGF concentrations, the cells respond by increasing the number of endosomes, and so the average number of phosphorylated EGF receptors in each endosome remains almost constant. Villaseñor et al. used mathematical modelling to show that this mechanism, which they refer to as an ‘analogue-to-digital conversion’, ensures a robust signal, and can regulate the signalling of the activated receptors in both space and time.

Different growth factors either increase or decrease the number and size of endosomes in various cell types. Moreover, Villaseñor et al. found that changing how the phosphorylated EGF receptor is distributed between endosomes alters how the cells interpret the signal and differentiate when they are exposed to EGF. These findings mean that the signalling activity of a cell could be predicted from the number and size of its endosomes. Moreover, the findings also suggest that interfering with this mechanism could change cell behaviour, for example, it could stop cancer cells proliferating and force them to differentiate instead.

Cells respond to various signals by activating different types of RTKs and committing to specific cell-fate decisions (Katz et al., 2007). A remarkable property of this system is that different RTKs can elicit distinct cellular responses through the same signal transduction machinery (Marshall, 1995; Kholodenko et al., 2006). In several cases, signalling specificity results from differences in amplitude and duration of the intracellular signalling cascades (Marshall, 1995; Maroun et al., 2000; Nagashima et al., 2007). For example, in PC12 cells, EGF stimulation of EGFR leads to transient Erk phosphorylation and cell proliferation, whereas NGF binding to TrkA leads to sustained Erk phosphorylation and cell differentiation (Marshall, 1995). Differences in signalling amplitude and duration can arise from positive or negative feedback loops within the same signalling pathway (Santos et al., 2007) or activation of additional signalling components (York et al., 1998). To explain such differences, it has been proposed that both EGF and NGF stimulation induce a specific ‘molecular context’ that determines the topology of the signal transduction network (Santos et al., 2007). How such a topology is determined for different RTKs and whether it is the sole determinant of signal specificity is unclear (Kholodenko, 2007).

Insights into this problem may be provided by the spatio-temporal distribution of RTKs along the endosomal system. The detection of phosphorylated receptors and signalling adaptors in endosomes (Di Guglielmo et al., 1994; Vieira et al., 1996; Sorkin, 2001; Teis et al., 2002; Lampugnani et al., 2006; Galperin and Sorkin, 2008; Schenck et al., 2008; Coumailleau et al., 2009) led to the concept that signalling is initiated at the plasma membrane but continues in endosomes (Di Guglielmo et al., 1994). Indeed, inhibition of endocytosis by blocking Dynamin function causes significant alterations in signalling specificity (Vieira et al., 1996). However, recent studies challenged this concept arguing that EGFR signalling occurs primarily at the plasma membrane (Damke et al., 1994; Brankatschk et al., 2012; Sousa et al., 2012). Interestingly, a recent systems survey of endocytosis (Collinet et al., 2010) revealed an unexpected tight control in the number, size, and cargo content for EGF-positive endosomes, raising the question of why is EGF packaging in endosomes so accurately controlled? Here, we hypothesized that the tight control of the endosomal distribution of EGF could serve to regulate signal transmission. We tested this hypothesis by quantitatively analysing the endosomal distribution of EGFR as an RTK model system in endosomes and evaluating its impact on cell-fate decisions.

To measure the content of p-EGFR in individual endosomes, we used two independent assays (for a detailed description see ‘Materials and methods’ and Figure 1—figure supplement 1). First, we modified a FRET-FLIM microscopy assay previously used to measure the spatial distribution of p-EGFR at the plasma membrane (Wouters and Bastiaens, 1999; Verveer et al., 2000). The assay measured the FRET signal between EGFR-GFP and an anti-phospho-tyrosine antibody (p-Tyr-ab) labelled with AlexaFluor 555. Since FLIM microscopy lacks the spatial resolution to analyse the receptor activation at a sub-cellular level, we modified the assay into a high-resolution FRET microscopy assay. However, instead of the total cell signal, we measured the distribution of EGFR and p-EGFR at the level of individual endosomes resolved by high-resolution confocal microscopy and quantitative automated image analysis (Rink et al., 2005; Collinet et al., 2010) (Figure 1—figure supplement 2). To avoid artefacts of overexpression, we used HeLa cells transfected with a bacterial artificial chromosome (BAC) transgene stably expressing EGFR-GFP under its endogenous promoter (Poser et al., 2008). In these cells (Figure 1—figure supplement 3A), the uptake of EGF was only ∼twofold higher compared to endogenous (Figure 1—figure supplement 3B). However, the transport kinetics were similar (Figure 1—figure supplement 3C). Second, we measured p-EGFR with an antibody against a specific phospho-tyrosine residue (Tyr1068). Both assays gave very similar results (Figure 1—figure supplement 4). As the FRET assay is not restricted to a single phosphorylation site that can change over time (Morandell et al., 2008), we used it as a primary assay in further experiments. Under the fixation conditions used, we observed no significant difference in the morphology (Figure 1—figure supplement 5A) or area (Figure 1—figure supplement 5B) of EGFR-positive endosomes (Video 1). For every time point, ∼15,000 endosomes from over 200 cells were analysed.

Video 1

Download asset

Continuous stimulation with EGF triggered the internalization of EGFR into endosomes (Figure 1 and Figure 1—figure supplement 2). The total amount of endosomal EGFR peaked after 15 min and decreased, reflecting (1) down-regulation of surface receptors (Wiley et al., 1991) and (2) their degradation (Dunn and Hubbard, 1984) over time (Figure 1A, green curve). On the other hand, the total p-EGFR levels reached a maximum already at 10 min, followed by a phase of decay (Figure 1A, red curve). Comparison of decay kinetics for both curves after 15 min showed that de-phosphorylation of p-EGFR occurred faster than degradation (τ_{decay EGFR} = 88.13 ± 14.49, τ_{decay p-EGFR} = 30.97 ± 1.69, for details see ‘Materials and methods’). Our FRET measurements are thus consistent with previously reported EGFR transport and phosphorylation kinetics determined by biochemical and microscopic methods (Di Guglielmo et al., 1994; Burke et al., 2001).

Figure 1 with 9 supplements see all

Download asset Open asset

We next determined the distribution of EGFR and p-EGFR in individual endosomes. The number of endosomes with p-EGFR decayed with similar kinetics as the total p-EGFR signal (τ_{decay N-p-EGFR} = 45.24 ± 11.39 vs τ_{decay p-EGFR} = 30.97 ± 1.69; compare red with black curve in Figure 1—figure supplement 6). The mean content of total EGFR per endosome increased over time and then rapidly decayed reaching steady state, due to the balance of continuous EGF uptake and degradation (Figure 1B, green curve). After a rapid increase, the mean content of p-EGFR in each endosome stabilized to a fairly constant level after ∼20 min (Figure 1B, red curve). Similar results were obtained when EGF was pulsed for 1 min and chased for different periods of time (Figure 1B, blue and black points).

To determine how the endosomal content of p-EGFR originates over time, we compared the distributions of EGFR and p-EGFR content per endosome. The width of distribution of total EGFR increased with time (Figure 1C), due to the fact that, as EGF continues to flow in, it first enters small early endosomes and progressively accumulates in larger ones (Rink et al., 2005). In contrast, the p-EGFR distribution first widened like that of total EGFR but then became almost twofold narrower than that of EGFR (Figure 1—figure supplement 7A–E compare the red and green curves) and stabilized after 30 min (Figure 1D). These results suggest an unexpected behaviour of p-EGFR, which over time stabilizes at a constant mean level per endosome.

Surprisingly, the mean amount of p-EGFR in endosomes was not ligand dependent. We stimulated cells with different concentrations of EGF for 30 min, when the amount of p-EGFR per endosome reached its steady state (Figure 1B,D). Once again, we found that EGFR and p-EGFR behaved very differently. The number of endosomes containing EGFR saturated already at low concentrations of EGF (0.5–1.0 ng EGF; Figure 1E, green curve) whereas the amount of total EGFR per endosome increased almost linearly (Figure 1F, green curve). This is expected because the higher the concentration of EGF, the higher the internalization of EGFR, whereas the number of receiving endosomes does not change significantly. In contrast, the number of endosomes with p-EGFR augmented with increasing EGF concentrations (Figure 1E, see red curve in semi-logarithmic scale, inset in linear scale). Strikingly, the mean amount of p-EGFR per endosome remained fairly constant, despite the EGF concentration varying almost over three orders of magnitude (Figure 1F, red curve). Therefore, increasing concentrations of EGF resulted in an increase in the number of endosomes with the same mean package of p-EGFR. Importantly, such a packaging is saturable because at high EGF concentrations the mean p-EGFR content per endosome was no longer constant with time (Figure 1—figure supplement 8).

The finding that endosomes contain a constant mean level of p-EGFR is striking. We performed several control experiments to verify that this is not an artefact caused by the FRET method or the assay. First, the mean amount of p-EGFR per endosome did increase at EGF concentrations higher than 10 ng/ml (Figure 1—figure supplement 8), indicating that the value measured is not artificially fixed, for example, by limited antigen accessibility. Second, a similar constant mean value of p-EGFR per endosomes was estimated with an independent method using the Tyr1068 antibody (Figure 1—figure supplement 4). Third, the narrow distribution of p-EGFR per endosome may simply reflect the sorting into endosomes of regular size. Whereas at 10 min the p-EGFR amount per endosome increased with the endosome area (Figure 1—figure supplement 9A, black curve), at 30 min (steady state, Figure 1B), the same mean amount was present in small and large endosomes alike (Figure 1—figure supplement 9A, red curve). Therefore, the amount of activated receptors per endosome is independent of endosome area. Finally, we verified that it is not a phenomenon peculiar to HeLa cells but also occurring in non-immortalized, non-cancer cell lines. Using the anti-phosphoTyr1068 antibody, we found that in primary mouse hepatocytes upon EGF stimulation the mean amount of p-EGFR per endosome saturated at ∼20 min whereas the mean amount of EGF continued to grow (data not shown), indicating that the packaging of p-EGFR in endosomes is not peculiar to a signalling-aberrant cancerous cell line.

In which endocytic compartment was p-EGFR packaged so uniformly? Nearly 80% of p-EGFR colocalized with the early endosomal marker EEA1 throughout the time course (Figure 2—figure supplement 1A). Less than 10% of p-EGFR colocalized with APPL1 after 15 min showing that it passed this endosomal compartment (Miaczynska et al., 2004) (Miaczynska et al., submitted). Very little p-EGFR colocalized with LAMP-1, a marker of late endosomes and lysosomes (Figure 2—figure supplement 1C). One possibility is that the packages of p-EGFR may reflect incorporation into intra-luminal vesicles (ILV) of multi-vesicular bodies (MVB). This possibility was ruled out using a previously described differential detergent solubilisation method (Malerød et al., 2007). We could determine that a large fraction of EGFR was not accessible to antibodies upon digitonin permeabilization, reflecting sequestration into ILVs (Malerød et al., 2007; Piper and Katzmann, 2007) (see Suppl. information and Figure 2A,B). In contrast, p-EGFR was always detectable suggesting that it was not within ILVs.

Figure 2 with 8 supplements see all

Download asset Open asset

How do the kinetics of p-EGFR endosomal packaging compare with the kinetics of receptor dephosphorylation and ubiquitylation? After 10 min of EGF stimulation, the pool of p-EGFR in EEA1-positive endosomes (Figure 2—figure supplement 1A, red curve) decayed faster than total EGFR (Figure 2—figure supplement 1A, green curve; τ_{decay EGFR} = 56.03 ± 5.72, τ_{decay p-EGFR} = 37.08 ± 4.19), possibly due to de-phosphorylation or preferential removal of p-EGFR from early endosomes. The latter can be excluded since the fraction of p-EGFR in EEA1-positive endosomes remained almost constant throughout the time course (Figure 2—figure supplement 1B). EGFR ubiquitylation is required for its internalization into endosomes, and this is dependent on EGFR phosphorylation and the recruitment of the c-Cbl E3 ligase (Sigismund et al., 2013). To compare the levels of ubiquitylated EGFR (ub-EGFR) with those of p-EGFR within the endosomal system, we modified the FRET assay using an anti-ubiquitin antibody (Figure 2—figure supplement 2A). The kinetics of ub-EGFR were significantly different from those of p-EGFR. Whereas the levels of p-EGFR peaked at 15 min, ub-EGFR reached its maximum at 30 min after stimulation (Figure 2C, compare red with blue curves) and decreased more slowly than p-EGFR at later times, probably reflecting deubiquitylation prior to receptor sequestration into ILVs (Piper and Katzmann, 2007). These results are consistent with the fact that the appearance of p-EGFR precedes that of ub-EGFR (Umebayashi et al., 2008). Moreover, ub-EGFR had a similar distribution to that of EGFR (Figure 2—figure supplement 2B) but significantly wider than p-EGFR (Figure 2—figure supplement 2C), suggesting that the mechanisms responsible for stabilizing the mean levels of p-EGFR per endosome are not correlated with receptor ubiquitylation.

Our data suggest the existence of a saturable mechanism adjusting the amount of p-EGFR in each individual endosome. Such a constant mean amount may be due to the formation of small clusters within early endosomes. To test this possibility, we imaged the spatial distribution of p-EGFR in endosomes using the anti-EGFR phosphoTyr1068 antibody by super-resolution microscopy. Using direct Stochastic Optical Reconstruction Microscopy (dSTORM) (Lampe et al., 2012), we could indeed visualize clusters of p-EGFR (Figure 2D, left panel) that decreased in size between 10 and 30 min of EGF internalization, in agreement with the narrowing of p-EGFR distribution over time (Figure 1D). To determine the number of molecules in the clusters, we used two methods. First, we developed a new method to estimate the number of fluorescent molecules in light microscopy images by measuring the intensity fluctuations during photo-bleaching over time (for details see ‘Materials and methods’ and Figure 2—figure supplement 3). Based on the fluorescence signal from the anti-phosphoTyr1068 antibody and EGFR-GFP, we estimated an average of 102 ± 38 and 76 ± 29 (Mean ± SEM) molecules of EGFR and p-EGFR per endosome 30 min after EGF (10 ng/ml) internalization (Figure 2—figure supplement 3), corresponding to 707 ± 265 and 527 ± 202 molecules per μm³ of endosomal volume (apparent, assessed by light microscopy), respectively. A hundred EGFR molecules would require ∼12 clathrin-coated vesicles for delivery to endosomes (see ‘Materials and methods’). We also estimated the total number of GFP-EGFR per cell and found values (29,000) well in agreement with previous estimates for HeLa cells (see ‘Materials and methods’ and Figure 2—figure supplement 1A,B). Second, based on the size of receptor from the PDB database (structure ID: 3NJP), we calculated that 83 ± 25 (Mean ± SEM, N = 1456) receptors could fit in the apparent area of p-EGFR visualized by dSTORM, a value which is remarkably in agreement with the fluorescence intensities estimates.

To further validate that the constant mean amount of p-EGFR per endosome corresponds to receptor clusters, we performed a focused RNAi screen on established components of the endosomal receptor sorting machinery (CHLCb, CHLCa, Hip1, Hip1R, Htt, Tom1, Tollip, Tom1L1, Tom1L2, Hrs, Snf8, Vps24). We found that only Hrs depletion resulted in a continuous accumulation of p-EGFR in endosomes with time (Figure 2E). At the same time, the p-EGFR intensity distribution widened similar to that of total EGFR (Figure 2—figure supplement 4A,B). The silencing of Hrs also caused an increase in the size of p-EGFR clusters within each endosome as revealed by dSTORM (Figure 2D, right panel). Interestingly, this is not due to the inhibition of ILV formation, as down-regulation of Snf8 and Vps24, members of the ESCRT-II and ESCRT-III complexes, respectively (Piper and Katzmann, 2007), reduced the sequestration of EGFR into the endosomal lumen (Figure 2—figure supplement 4D) but did not have significant effects on the amount of p-EGFR per endosome (Figure 2E, blue and green curves). Thus, the constant mean amount of p-EGFR in endosomes likely corresponds to the receptor clusters observed by super-resolution microscopy. This raises the question of whether p-EGFR can be accessible to downstream signalling components. Therefore, we measured the recruitment of a direct downstream effector of p-EGFR, Shc1. The kinetics of Shc1 recruitment to endosomes precisely mimic the kinetics of p-EGFR (Figure 2—figure supplement 5) arguing that the p-EGFR clusters are signalling competent.

Upon internalization, EGF enters the early endosomal network and, similar to LDL (Rink et al., 2005), following endosome homotypic fusion and fission reactions, accumulates in few large endosomes prior to transfer to late endosomes. A mechanism must exist that prevents the continuous accretion of p-EGFR upon endosome fusion. A simple mechanism could be that the de-phosphorylation rate increases with the increase in p-EGFR per endosome. When two endosomes fuse, the resulting endosome should contain the sum of EGFR and p-EGFR of the original endosomes. However, given such de-phosphorylation rate dependency, the amount of p-EGFR would return to the level prior to fusion, thus stabilizing the mean amount of p-EGFR per endosome. A prediction of this hypothesis is that the kinase activity of EGFR in endosomes controls its own dephosphorylation. To test this, we inhibited the EGFR kinase activity pharmacologically with AG1478, lapatinib or gefitinib 10 min after EGF stimulation (to prevent alterations on receptor internalization) and determined the effects on the receptors already internalized and phosphorylated. We compared low with high concentrations of EGF, that is, under conditions of saturation of p-EGFR packaging in endosomes (Figure 1—figure supplement 8). At low EGF concentrations, when the packaging mechanism is not saturated, the total amount of p-EGFR was not significantly reduced by the inhibitors (Figure 2—figure supplement 6 compare black and green curves). This behaviour argues that the packages of p-EGFR in endosomes are protected from the phosphatases. In addition, the inhibitors caused a continuous accumulation of p-EGFR in fewer and larger endosomes over time (Figure 2—figure supplement 7). In contrast, adding the inhibitor after stimulation with high concentrations of EGF caused a sharp reduction in the total amount of p-EGFR (Figure 2—figure supplement 6 compare red and blue curves), as observed previously (Kleiman et al., 2011). This means that the kinase activity of EGFR is necessary to maintain the levels of p-EGFR in individual endosomes. These results support the idea that the dephosphorylation of p-EGFR in endosomes indeed depends on the EGFR activity within the endosomal packages.

Which phosphatases are responsible for controlling p-EGFR packaging in endosomes? To identify them, we performed a focused RNAi screen against 21 protein tyrosine phosphatases (PTP) expressed in HeLa cells (Tarcic et al., 2009). Hits were defined if silencing satisfied three conditions: (1) it increased the total amount of p-EGFR in endosomes and (2) increased the mean amount of p-EGFR per endosome, and (3) the phenotype was observed with at least two siRNAs per gene. Five phosphatases, PTP4A1, PTPN11, PTPN9, PTPN18, and PTPRK, increased the amount of p-EGFR in individual endosomes (Figure 2F and Figure 2—figure supplement 8). Interestingly, PTPN11 is an EGFR interactor (Deribe et al., 2009) whose activity is enhanced upon tyrosine phosphorylation (Agazie and Hayman, 2003), suggesting a molecular mechanism whereby p-EGFR could regulate its own de-phosphorylation in endosomes.

What are the consequences of such mechanism for signal transduction? To address these questions and generate testable predictions, we developed a mathematical model that describes the amount of total intracellular p-EGFR over time. Previously, excellent models have been developed that quantitatively describe EGFR endocytosis and signalling (Felder et al., 1992; French et al., 1994; Kholodenko et al., 1999; Kholodenko, 2002; Resat et al., 2003). However, although all these models described in detail the dynamics of ligand binding, dimer formation and endocytosis, recycling and degradation of the receptor, they did not consider the trafficking dynamics of the phosphorylated receptors with respect to the dynamics of the endosomal network because these data were not available. Our new experimental data brought two new concepts. First, dephosphorylation and degradation of p-EGFR occur sequentially but are uncoupled. Second, the amount of p-EGFR is controlled at the level of individual endosomes. These new concepts require further development of the existing EGFR mathematical models. Our model was formulated as a set of ordinary differential equations (ODE, see ‘Materials and methods’ and Figure 3) describing (1) the total amount of EGFR and p-EGFR at the plasma membrane as a function of ligand binding, (2) endocytosis of p-EGFR and its indirect effects on EGFR endocytosis, and (3) distribution of cargo between early endosomes at different stages of maturation (e.g., formation of MVB). For this, we considered the processes of receptor internalization, dephosphorylation, degradation, recycling, endosome fusion and fission. As in previous models (French et al., 1994; Resat et al., 2003), we described time course kinetics of total cellular p-EGFR, surface and endosomal EGFR and p-EGFR. Importantly, our model also describes the total number of p-EGFR-positive endosomes and mean amount of p-EGFR per endosome (see ‘Materials and methods’ for details). To account for the observed stabilization of the mean amount of p-EGFR per endosome over time (Figure 1), the dependency of p-EGFR dephosphorylation on EGFR kinase activity (Figure 2—figure supplement 6,7) and the fact that the mechanism is saturable (Figure 1—figure supplement 8), we included a sigmoidal dependency of the p-EGFR dephosphorylation rate on the amount of p-EGFR per endosome. The model was then fitted to the experimental data from the p-EGFR time course (Figure 3A,B). Figure 3C shows that this simple theoretical model can reproduce our observations of a constant mean amount of p-EGFR per endosome in a wide range of EGF concentrations when fitted to the experimental data. Importantly, a model without this non-linear dephosphorylation dependency could correctly describe the total amount of EGFR and p-EGFR in endosomes (Figure 3—figure supplement 1A,B) but did not agree with the measurements for the mean amount of p-EGFR per endosome (Figure 3—figure supplement 1C), thus supporting the sigmoidal dependency of the p-EGFR de-phosphorylation rate on the amount of p-EGFR per endosome (Figure 3). Previous models did not include this non-linear term because data on the distribution of p-EGFR in individual endosomes was not available.

Figure 3 with 1 supplement see all

Download asset Open asset

An unexpected prediction of our model is that the total de-phosphorylation rate, and thus the total amount of p-EGFR, is dependent on the fusion/fission rate of the endosomes (Figure 3D). If so, could this have an effect on signal transduction? To test these hypotheses, we reduced early endosome homotypic fusion by lowering the intracellular concentration of established components of the endosome tethering and fusion machinery, EEA1, Rabenosyn5, Vps45 (Christoforidis et al., 1999; Ohya et al., 2010), Syntaxin-6 and Syntaxin-13 (Brandhorst et al., 2006) that play no direct role in signalling. These genes were down-regulated by RNAi in combinations and only partially (∼50–70% depletion for each protein, Figure 4A) to achieve a significant inhibition of endosome fusion and yet prevent or reduce cell toxicity. This procedure caused a mild redistribution of EGFR to endosomes of smaller size (<0.5 μm² cross-section area, for details see ‘Materials and methods’ and Figure 6—figure supplement 2) (Figure 4C,D). Similar results were obtained upon depletion of a second combination of genes (EEA1, Stx13, Stx6, not shown, see below). Note that these treatments generated a pattern of endosomes similar to that observed in different cell types and under different culture conditions (see below, Figure 6) and neither altered the surface levels of EGFR (Figure 4—figure supplement 1) nor its kinetics of uptake (Figure 4B) and exit from endosomes, that is, recycling and degradation (Figure 4—figure supplement 2). We also excluded potential effects on endosome acidification, because blocking it with bafilomycin did increase both p-EGFR and total EGFR (Figure 4—figure supplement 3). Remarkably, under our experimental conditions of mild down-regulation of the early endosomal fusion machinery the packaging of active receptors was unaffected as shown by both the time course and the steady-state mean (constant) amount of p-EGFR per endosome (Figure 4E; see above, Figure 1B). In contrast, the total number of endosomes with p-EGFR and their life-time augmented (Figure 4F), resulting in a net increase in the total amount and life-time of p-EGFR (Figure 4G). Notably, reduction of the endosome fusion rate in the mathematical model (∼40%, in line with the depletion of tethering proteins, Figure 4A) is sufficient to reproduce fairly well the experimental increase in p-EGFR endosomes observed (Figure 4F). These results support the hypothesis that EGFR activation can be modulated by the endosomal system. Since p-EGFR de-phosphorylation precedes EGFR degradation (see above, Figure 1A, Figure 2—figure supplement 2) and EGFR degradation is unaffected (Figure 4B, Figure 4—figure supplement 2), we deduce that the effect on the life-time of p-EGFR caused by reduced endosomal fusion is primarily due to reduced de-phosphorylation.

Figure 4 with 3 supplements see all

Download asset Open asset

Increased EGFR phosphorylation results in sustained Erk signalling (Sasagawa et al., 2005; Nakakuki et al., 2010) and this leads to the phosphorylation and stabilization of the immediate early gene product c-Fos (Nakakuki et al., 2010). We asked whether the redistribution of endosomal EGFR could be sufficient to induce sustained Erk activation and c-Fos phosphorylation. Indeed, upon EGF stimulation, both the amplitude and duration of Erk1/2 phosphorylation were increased in the depleted cells compared to control (Figure 5A,B). Consistently, c-Fos phosphorylation was also higher after 30 min of EGF stimulation (Figure 5C,D). The fact that the amount and life-time of total EGFR in endosomes remained unvaried in these experiments (Figure 4B) eliminates the trivial possibility that the observed changes are due to modulation of receptor degradation.

Figure 5

Download asset Open asset

The experiments on HeLa cells and the theoretical analysis raise the question of whether modulation of early endosome homotypic fusion is a general mechanism to regulate signal amplitude and duration. If this were the case, we would predict that growth factors with different signalling outputs (amplitude and duration) differentially modulate the endosomal distribution (i.e., endosome number, size, and cargo content). To test this prediction, we examined different growth factors and cellular systems. First, we used primary mouse hepatoblasts where HGF promotes their proliferation (Tanimizu et al., 2003). In these cells, HGF but not EGF elicits a sustained Erk response (Figure 6—figure supplement 1). Indeed as predicted, stimulation of hepatoblasts with HGF caused a strong shift in the distribution of early endosomes toward smaller sizes (Figure 6A, red curve Figure 6B), whereas EGF had the opposite effect (Figure 6A, green curve, Figure 6B). Second, we turned to an in vitro model of reference for cell-fate decisions, PC12 cells. In PC12 cells, EGF stimulation leads to transient Erk phosphorylation and cell proliferation, whereas NGF leads to sustained Erk phosphorylation and cell differentiation (Marshall, 1995). Consistent with our results in primary mouse hepatoblasts, NGF stimulation in PC12 cells caused a significant shift in the distribution of early endosomes toward smaller sizes compared with EGF (Figure 6C,D). Moreover, NGF itself was distributed to a larger number of smaller endosomes in comparison with EGF (Figure 6E,F). Altogether, these data argue that the modulation of endosome fusion, reflected by the changes in endosome number and size, is a general property of growth factors. These data further suggest that signalling amplitude and duration can be regulated by changes in the fusion rate of endosomes (see Table 1).

Figure 6 with 2 supplements see all

Download asset Open asset

Finally, we tested whether the differences in endosomal distribution can be, at least in part, causative of the different cell-fates triggered by EGF and NGF in PC12 cells. If so, we would predict that redistributing EGF to a larger number of small endosomes as seen in PC12 stimulated with NGF would be sufficient to switch signalling specificity and induce differentiation of PC12 cells. Therefore, we applied the same protocol of partial protein depletion previously used for HeLa cells (Figure 4) in PC12 cells and consistently observed a mild redistribution of EGF into smaller endosomes (Figure 7—figure supplement 1A,C,D). Also in this case, the partial depletion did not result in major changes in EGF transport kinetics in PC12 cells (Figure 7—figure supplement 1B), but increased the phosphorylation of both Erk (Figure 7—figure supplement 2A,B) and c-Fos (Figure 7—figure supplement 2C,D) upon stimulation with EGF. Next, we stimulated PC12 cells with EGF or NGF for 24 hr and analysed for neurite formation and β-III tubulin expression as markers of differentiation (Ohuchi et al., 2002) and for EdU incorporation as a measure of proliferation (Figure 7A). Stimulation with NGF increased the number of cells with neurites (Figure 7B, quantification in Figure 7C) and positive for β-III tubulin (Figure 7B, quantification in Figure 7D), and reduced cell proliferation (Figure 7A, quantification in Figure 7E), the opposite of the stimulation with EGF. Remarkably, upon redistribution of endosomes, EGF increased process formation (Figure 7B, quantification in Figure 7C), β-III tubulin expression (Figure 7B, quantification in Figure 7D), and reduced cell proliferation (Figure 7A, quantification in Figure 7E). The type of response was therefore similar to that of NGF, although the efficacy was lower. Nevertheless, these results show that a mild reduction of homotypic early endosome fusion was sufficient to modify cell fate and induce neuronal differentiation of PC12 cells.

Figure 7 with 2 supplements see all

Download asset Open asset

Genomic studies have revealed that signalling pathways exert a profound effect on the endosomal system (Pelkmans et al., 2005; Stasyk et al., 2007; Collinet et al., 2010). Parameters such as number of endosomes and size are tightly controlled in the case of EGF endocytosis (Collinet et al., 2010). Our results provide a rationale for such modulation and a novel framework for interpreting and predicting the signalling response of phosphorylated RTKs. In homogeneous assays (e.g., by Western blot), the total levels of active RTKs can be observed to rapidly decay with time in most signalling systems (Dunn and Hubbard, 1984; Burke et al., 2001; Sousa et al., 2012). These methods, however, measure the average steady state of an entire cell population and lack the spatial information. Here, we employed quantitative high- and super-resolution microscopy to resolve details of this process with sub-cellular resolution and high sensitivity. We discovered that the mean amount of p-EGFR per endosome was fairly constant over time and p-EGFR was found in small clusters in early endosomes.

The endosomal network is shaped by the balance of endosome fusion and fission (Foret et al., 2012) and this balance is also necessary for the formation of the endosomal clusters of p-EGFR. Modulation of the endosomal fusion/fission machinery manifests itself as a change in the size of endosomes (Sigismund et al., 2008) (Figure 6). Shifting the balance toward smaller endosomes through inhibition of fusion increased the number and reduced the size of endosomes, consequently expanding the number and life-time of p-EGFR clusters. Although the inhibition of endosome fusion was very mild, we cannot exclude the possibility that it may alter the recruitment and/or activity of signalling components by yet unknown mechanisms. On the other hand, for the interpretation of phenotypes upon perturbations on signalling, it is also important to consider the impact they have on the endosomal network (Collinet et al., 2010).

By analogy with synaptic transmission (Edwards, 2007), the packages of p-EGFR in early endosomes could be considered as quanta of signalling molecules. The concept of phosphorylated RTK quanta is reminiscent of analogue-to-digital communication systems, where a continuous variable (e.g., extracellular growth factor concentration) is transformed into a sequence of binary levels (e.g., phosphorylated RTK quanta in endosomes). An analogue-to-digital switch was described for Ras nanoclusters at the plasma membrane (Tian et al., 2007). In the case of endosomal digital signalling, our mathematical model predicts that it could serve two functions. First, it provides a mechanism to regulate signal amplitude and duration following RTK internalization. As a consequence, the total de-phosphorylation rate becomes dependent on the fusion/fission rate of the endosomes. This is interesting in view of the specific modulation of the endosome fusion/fission rates by growth factors (Figure 6, see below). Second, it acts as a noise dampening system (Ladbury and Arold, 2012), suppressing the noise due to, for example, fluctuations of EGF in the extracellular medium, expression levels of EGFR on the cell surface, etc. An increase in the amount of p-EGFR would result in faster de-phosphorylation rates. In contrast, low concentrations of EGF or EGFR would result in low de-phosphorylation rates. The middle point between the two extremes is the hallmark of signalling resilience. In addition, such a digital system may facilitate the integration of signalling information from different RTKs into a single, correct cell-fate decision. Our results highlight the importance of measuring the spatio-temporal distribution of signalling molecules using quantitative image analysis approaches to gain a deeper understanding of signal transduction regulation.

What is the molecular machinery responsible for the formation of the clusters and how is the number of p-EGFR molecules regulated? Clearly, the clustering mechanism is saturable (Figure 2A,B), as very high concentrations of EGF above some threshold suppress the correct endosomal packaging in addition to changes in the entry routes and signal output (Sigismund et al., 2008). We found that both Hrs and a few phosphatases, notably PTPN11 (SHP2), specifically regulate the amount of receptors within the p-EGFR clusters and their size. Hrs is known to interact with EGFR and regulate its degradation together with other components of the ESCRT machinery (Umebayashi et al., 2008). However, the effect of Hrs on the size of the p-EGFR clusters appears to be independent of the formation of ILVs, as suggested by the fact that Snf8 and Vps24 down-regulation does not produce the same effect.

Our mathematical model revealed that a correlation between p-EGFR dephosphorylation rate and p-EGFR amount per endosome can explain the mean constant size of p-EGFR quanta. We can envisage various non-exclusive mechanisms that can account for this correlation. One possible mechanism is a scaffold with a characteristic size that binds to p-EGFR and protects it from phosphatases. This hypothesis correlates higher total EGFR kinase activity to higher p-EGFR dephosphorylation, but only indirectly. Increasing the concentration of EGF in the medium would lead to a higher rate of delivery of p-EGFR to endosomes through vesicles which have no scaffold. If scaffold formation were rate limiting, the increased flux of p-EGFR into endosomes would reduce the fraction of protected p-EGFR thus exposing it to dephosphorylation. A caveat of this model is that, as the fusion of endosomes proceeds over time, multiple quanta would be expected to be brought together, increasing the mean amount of p-EGFR per endosome. This expectation is in contradiction with our experimental data (Figure 1B,D). With this model, additional factors must thus be taken into account to explain why multiple quanta cannot co-exist on the same endosomes.

The finding that Hrs knock-down increases the levels of p-EGFR suggests a different scaffold-based model. Instead of acting as a p-EGFR protective scaffold (or part of a scaffold), Hrs could exert the opposite function and stabilize the unphosphorylated EGFR, preventing its re-phosphorylation (Kleiman et al., 2011). Since the activity of Hrs is negatively regulated by p-EGFR (Row et al., 2005; Bache et al., 2002), this model is compatible with the data showing loss of quanta and increase in endosomal p-EGFR levels upon Hrs knock-down (Figure 2D,E). However, this hypothesis alone can neither explain the formation of quanta nor the finding that blocking p-EGFR kinase activity does not change the total levels of p-EGFR over time (Figure 2—Figure supplement 6).

Another mechanism is based on Turing Instability (Turing, 1952) (a reaction-diffusion mechanism). This mechanism is perhaps less intuitive but widely spread in biological processes, such as symmetry breaking and pattern formation in morphogenesis (Kondo and Miura, 2010). It is based on the observation that p-EGFR recruits and phosphorylates PTPN11 (SHP2) in a phosphor-tyrosine dependent manner (Deribe et al., 2009), thus enhancing its phosphatase activity (Agazie and Hayman, 2003). Briefly, p-EGFR would recruit and activate the phosphatase SHP2, forming a negative feedback loop. The phosphatase would diffuse on the surface of endosomes, dephosphorylating p-EGFR molecules before being itself inactivated in the absence of further interactions with p-EGFR. Such reaction-diffusion mechanism within a specific parameter range is known to form spatially restricted clusters of active molecular species (Turing Instability) (Turing, 1952), in this particular case quanta of p-EGFR on endosomes. A transient increase in p-EGFR after an endosome fusion event would increase the recruitment and/or activity of SHP2, re-establishing the p-EGFR quanta through dephosphorylation. If the characteristic length of Turing Instability is larger than endosome size, then multiple quanta cannot co-exist within a single endosome. The Turing Instability hypothesis explains the observed increase in p-EGFR quanta size after EGFR kinase inhibition, keeping the total p-EGFR levels unchanged (Figure 2—figure supplement 6,7), as well as the increase in total endosomal p-EGFR upon inhibition of endosome fusion (Figure 4G). However, it does not explain the effect of Hrs knock-down. A combination of the Turing instability and Hrs-mediated (negative) scaffold mechanisms is more consistent with our observations.

The regulation of endosomal packing reported in our study is likely not restricted to EGFR alone but is a general property, as different growth factors affect the endosomal network according to their specific signal output and cellular context (Figure 6). Hrs and SHP2 are also recruited by other RTKs (Agazie and Hayman, 2003; Row et al., 2005). The relative affinity of SHP2 to different receptors could lead to larger or smaller quanta, thus tuning the specificity of the signalling response. RTK quanta with different sizes could also result from differential phosphorylation of Hrs by RTKs (Row et al., 2005), given that the relative amount of Hrs on endosomes depends on its phosphorylation state (Urbé et al., 2000).

By which mechanisms can RTKs regulate the endosomal network? It has been shown that RTKs can modulate the activity of the transport machinery. For example, activation of p38 MAP kinase causes phosphorylation of the Rab5 effectors EEA1 and Rabenosyn-5, enhancing their recruitment to endosomes and consequently stimulating early endosome fusion (Macé et al., 2005; Cavalli et al., 2001). RTK stimulation also modulates the nucleotide cycle of Rab5 via activation of the Rab5 GEF RIN1 (Tall et al., 2001) or inactivation of the Rab5 GAP RN-tre (Lanzetti et al., 2000). Therefore, we predict that in general RTK ligands that stimulate the endosomal fusion machinery (such as EGF) will have a short phosphorylation half-life, whereas ligands that change the fusion/fission balance in favour of smaller endosomes (such as NGF) will have a long phosphorylation half-life. The combined effects of quanta size regulation through Hrs and SHP2 and modulation of fusion/fission will give a specific signalling amplitude and duration in different cell types stimulated with different ligands. We propose that the shape of distribution of the endosomal network can serve as a predictive parameter of the signalling status of the cell.

Our results support the concept of endosomes as signalling platforms (Di Guglielmo et al., 1994), a view recently shared for the β2-adrenoceptor (Irannejad et al., 2013) but opposed by other studies (Brankatschk et al., 2012; Sousa et al., 2012). This apparent contradiction can be explained by the fact that, under normal conditions of endocytosis, only the small fraction of p-RTK in endosomes are protected from inactivation and degradation, and can thus contribute to signal propagation. A feature of the p-EGFR clusters is that, with the increase in the local concentration, the stability of the active EGFR dimer (Chung et al., 2010) and signalling properties (Verveer et al., 2000) would also be increased. By blocking endocytosis, the levels of active receptors are artificially increased at the cell surface (Sousa et al., 2012), bypassing the normal requirement for endosomal regulation.

Our observations raise many more questions concerning the molecular mechanisms of quanta formation and their impact on cell fate decision. Clearly, the variety of models on quanta formation requires future experimental tests to determine the correct mechanism and reveal its molecular details. In addition, it will be important to validate our observations in an in vivo animal model to demonstrate that the dynamics of the endosomal network reflect the signalling activity by RTK under physiological conditions.

1. 1. Lu C
   2. Mi LZ
   3. Grey MJ
   4. Zhu J
   5. Graef E
   6. Yokoyama S
   7. Springer TA

   (2010) The Extracellular and Transmembrane Domain Interfaces in Epidermal Growth Factor Receptor Signaling

   ID 3NJP. Publicly available at RCSB Protein Data Bank (http://www.rcsb.org).

   http://www.rcsb.org

1. 1. Agazie YM
   2. Hayman MJ

   (2003) Molecular mechanism for a role of SHP2 in epidermal growth factor receptor signaling

   Molecular and Cellular Biology 23:7875–7886.

   https://doi.org/10.1128/MCB.23.21.7875-7886.2003

   • Google Scholar
2. 1. Ariotti N
   2. Liang H
   3. Xu Y
   4. Zhang Y
   5. Yonekubo Y
   6. Inder K
   7. Du G
   8. Parton RG
   9. Hancock JF
   10. Plowman SJ

   (2010) Epidermal growth factor receptor activation remodels the plasma membrane lipid environment to induce nanocluster formation

   Molecular and Cellular Biology 30:3795–3804.

   https://doi.org/10.1128/MCB.01615-09

   • Google Scholar
3. 1. Bache KG
   2. Raiborg C
   3. Mehlum A
   4. Madshus IH
   5. Stenmark H

   (2002) Phosphorylation of Hrs downstream of the epidermal growth factor receptor

   European Journal of Biochemistry 269:3881–3887.

   https://doi.org/10.1046/j.1432-1033.2002.03046.x

   • Google Scholar
4. 1. Brandhorst D
   2. Zwilling D
   3. Rizzoli SO
   4. Lippert U
   5. Lang T
   6. Jahn R

   (2006) Homotypic fusion of early endosomes: SNAREs do not determine fusion specificity

   Proceedings of the National Academy of Sciences of USA 103:2701–2706.

   https://doi.org/10.1073/pnas.0511138103

   • Google Scholar
5. 1. Brankatschk B
   2. Wichert SP
   3. Johnson SD
   4. Schaad O
   5. Rossner MJ
   6. Gruenberg J

   (2012) Regulation of the EGF transcriptional response by endocytic sorting

   Science Signaling 5:ra21.

   https://doi.org/10.1126/scisignal.2002351

   • Google Scholar
6. 1. Burke P
   2. Schooler K
   3. Wiley HS

   (2001) Regulation of epidermal growth factor receptor signaling by endocytosis and intracellular trafficking

   Molecular Biology of the Cell 12:1897–1910.

   https://doi.org/10.1091/mbc.12.6.1897

   • Google Scholar
7. 1. Cavalli V
   2. Vilbois F
   3. Corti M
   4. Marcote MJ
   5. Tamura K
   6. Karin M
   7. Arkinstall S
   8. Gruenberg J

   (2001) The stress-induced MAP kinase p38 regulates endocytic trafficking via the GDI: Rab5 complex

   Molecular Cell 7:421–432.

   https://doi.org/10.1016/S1097-2765(01)00189-7

   • Google Scholar
8. 1. Christoforidis S
   2. McBride HM
   3. Buroyne RD
   4. Zerial M

   (1999) The Rab5 effector EEA1 is a core component of endosome docking

   Nature 397:621–625.

   https://doi.org/10.1038/17618

   • Google Scholar
9. 1. Chung I
   2. Akita R
   3. Vandlen R
   4. Toomre D
   5. Schlessinger J
   6. Mellman I

   (2010) Spatial control of EGF receptor activation by reversible dimerization on living cells

   Nature 464:783–787.

   https://doi.org/10.1038/nature08827

   • Google Scholar
10. 1. Collinet C
    2. Stöter M
    3. Bradshaw CR
    4. Samusik N
    5. Rink JC
    6. Kenski D
    7. Habermann B
    8. Buchholz F
    9. Henschel R
    10. Mueller MS
    11. Nagel WE
    12. Fava E
    13. Kalaidzidis Y
    14. Zerial M

    (2010) Systems survey of endocytosis by multiparametric image analysis

    Nature 464:243–249.

    https://doi.org/10.1038/nature08779

    • Google Scholar
11. 1. Coumailleau F
    2. Fürthauer M
    3. Knoblich JA
    4. González-Gaitán M

    (2009) Directional Delta and Notch trafficking in Sara endosomes during asymmetric cell division

    Nature 458:1051–1055.

    https://doi.org/10.1038/nature07854

    • Google Scholar
12. 1. Damke H
    2. Baba T
    3. Warnock DE
    4. Schmid SL

    (1994) Induction of mutant dynamin specifically blocks endocytic coated vesicle formation

    The Journal of Cell Biology 127:915–934.

    https://doi.org/10.1083/jcb.127.4.915

    • Google Scholar
13. 1. Deribe YL
    2. Wild P
    3. Chandrashaker A
    4. Curak J
    5. Schmidt MH
    6. Kalaidzidis Y
    7. Milutinovic N
    8. Kratchmarova I
    9. Buerkle L
    10. Fetchko MJ
    11. Schmidt P
    12. Kittanakom S
    13. Brown KR
    14. Jurisica I
    15. Blagoev B
    16. Zerial M
    17. Stagljar I
    18. Dikic I

    (2009) Regulation of epidermal growth factor receptor trafficking by lysine deacetylase HDAC6

    Science Signaling 2:ra84.

    https://doi.org/10.1126/scisignal.2000576

    • Google Scholar
14. 1. Di Guglielmo GM
    2. Baass PC
    3. Ou W
    4. Posner BI
    5. Bergeron JJ

    (1994)

    Compartmentalization of SHC, GRB2 and mSOS and hyperphosphorylation of Raf-1 by EGF but not insulin in liver parenchyma

    The EMBO Journal 13:4269–4277.

    • Google Scholar
15. 1. Dunn WA
    2. Hubbard AL

    (1984) Receptor-mediated endocytosis of epidermal growth factor by hepatocytes in the Perfused Rat liver: ligand and receptor dynamics

    The Journal of Cell Biology 98:2148–2159.

    https://doi.org/10.1083/jcb.98.6.2148

    • Google Scholar
16. 1. Edwards RH

    (2007) The neurotransmitter cycle and quantal size

    Neuron 55:835–858.

    https://doi.org/10.1016/j.neuron.2007.09.001

    • Google Scholar
17. 1. Faust D
    2. Schmitt C
    3. Oesch F
    4. Oesch-Bartlomowicz B
    5. Schreck I
    6. Weiss C
    7. Dietrich C

    (2012) Differential p38-dependent signalling in response to cellular stress and mitogenic stimulation in fibroblasts

    Cell Communication and Signaling 10:6.

    https://doi.org/10.1186/1478-811X-10-6

    • Google Scholar
18. 1. Felder S
    2. LaVin J
    3. Ullrich A
    4. Schlessinger J

    (1992) Kinetics of binding, endocytosis, and recycling of EGF receptor mutants

    The Journal of Cell Biology 117:203–212.

    https://doi.org/10.1083/jcb.117.1.203

    • Google Scholar
19. 1. Foret L
    2. Dawson JE
    3. Villaseñor R
    4. Collinet C
    5. Deutsch A
    6. Brusch L
    7. Zerial M
    8. Kalaidzidis Y
    9. Jülicher F

    (2012) A general theoretical framework to infer endosomal network dynamics from quantitative image analysis

    Current Biology 22:1381–1390.

    https://doi.org/10.1016/j.cub.2012.06.021

    • Google Scholar
20. 1. French AR
    2. Sudlow GP
    3. Wiley HS
    4. Lauffenburger DA

    (1994)

    Postendocytic trafficking of epidermal growth factor-receptor complexes is mediated through saturable and specific endosomal interactions

    The Journal of Biological Chemistry 269:15749–15755.

    • Google Scholar
21. 1. Galperin E
    2. Sorkin A

    (2008) Endosomal targeting of MEK2 requires RAF, MEK kinase activity and clathrin-mediated endocytosis

    Traffic 9:1776–1790.

    https://doi.org/10.1111/j.1600-0854.2008.00788.x

    • Google Scholar
22. 1. Gordon GW
    2. Berry G
    3. Liang XH
    4. Levine B
    5. Herman B

    (1998) Quantitative fluorescence Resonance Energy transfer measurements using fluorescence microscopy

    Biophysical Journal 74:2702–2713.

    https://doi.org/10.1016/S0006-3495(98)77976-7

    • Google Scholar
23. 1. Hahn AT
    2. Jones JT
    3. Meyer T

    (2009) Quantitative analysis of cell cycle phase durations and PC12 differentiation using fluorescent biosensors

    Cell Cycle 8:1044–1052.

    https://doi.org/10.4161/cc.8.7.8042

    • Google Scholar
24. 1. Hofman EG
    2. Bader AN
    3. Voortman J
    4. van den Heuvel DJ
    5. Sigismund S
    6. Verkleij AJ
    7. Gerritsen HC
    8. van Bergen en Henegouwen PM

    (2010) Ligand-induced EGF receptor oligomerization is kinase-dependent and enhances internalization

    The Journal of Biological Chemistry 285:39481–39489.

    https://doi.org/10.1074/jbc.M110.164731

    • Google Scholar
25. 1. Irannejad R
    2. Tomshine JC
    3. Tomshine JR
    4. Chevalier M
    5. Mahoney JP
    6. Steyaert J
    7. Rasmussen SG
    8. Sunahara RK
    9. El-Samad H
    10. Huang B
    11. von Zastrow M

    (2013) Conformational biosensors reveal GPCR signalling from endosomes

    Nature 495:534–538.

    https://doi.org/10.1038/nature12000

    • Google Scholar
26. 1. Kalaidzidis Y
    2. Kalaidzidis I
    3. Zerial M

    (2015) A probabilistic method to quantify the colocalization of markers on intracellular vesicular structures visualized by light microscopy

    AIP Conference Proceedings 1641:580.

    https://doi.org/10.1063/1.4906025

    • Google Scholar
27. 1. Kamiya A
    2. Kinoshita T
    3. Ito Y
    4. Matsui T
    5. Morikawa Y
    6. Senba E
    7. Nakashima K
    8. Taga T
    9. Yoshida K
    10. Kishimoto T
    11. Miyajima A

    (1999) Fetal liver development requires a paracrine action of oncostatin M through the gp130 signal transducer

    The EMBO Journal 18:2127–2136.

    https://doi.org/10.1093/emboj/18.8.2127

    • Google Scholar
28. 1. Katz M
    2. Amit I
    3. Yarden Y

    (2007) Regulation of MAPKs by growth factors and receptor tyrosine kinases

    Biochimica Et Biophysica Acta 1773:1161–1176.

    https://doi.org/10.1016/j.bbamcr.2007.01.002

    • Google Scholar
29. 1. Kholodenko BN

    (2002) MAP kinase cascade signaling and endocytic trafficking: a marriage of convenience?

    Trends in Cell Biology 12:173–177.

    https://doi.org/10.1016/S0962-8924(02)02251-1

    • Google Scholar
30. 1. Kholodenko BN

    (2007) Untangling the signalling wires

    Nature Cell Biology 9:247–249.

    https://doi.org/10.1038/ncb0307-247

    • Google Scholar
31. 1. Kholodenko BN
    2. Demin OV
    3. Moehren G
    4. Hoek JB

    (1999) Quantification of short term signaling by the epidermal growth factor receptor

    The Journal of Biological Chemistry 274:30169–30181.

    https://doi.org/10.1074/jbc.274.42.30169

    • Google Scholar
32. 1. Kholodenko BN
    2. Hancock JF
    3. Kolch W

    (2006) Signalling ballet in space and time

    Nature Reviews Molecular Cell Biology 11:414–426.

    https://doi.org/10.1038/nrm2901

    • Google Scholar
33. 1. Kleiman LB
    2. Maiwald T
    3. Conzelmann H
    4. Lauffenburger DA
    5. Sorger PK

    (2011) Rapid phospho-turnover by receptor tyrosine kinase impacts downstream signaling and drug binding

    Molecular Cell 43:723–737.

    https://doi.org/10.1016/j.molcel.2011.07.014

    • Google Scholar
34. 1. Kondo S
    2. Miura T

    (2010) Reaction-diffusion model as a framework for understanding biological pattern formation

    Science 329:1616–1620.

    https://doi.org/10.1126/science.1179047

    • Google Scholar
35. 1. Ladbury JE
    2. Arold ST

    (2012) Noise in cellular signaling pathways: causes and effects

    Trends Biochem. Sci 37:173–178.

    https://doi.org/10.1016/j.tibs.2012.01.001

    • Google Scholar
36. 1. Lampe A
    2. Haucke V
    3. Sigrist SJ
    4. Heilemann M
    5. Schmoranzer J

    (2012) Multi-colour direct STORM with red emitting carbocyanianes

    Biology of the Cell 104:229–237.

    https://doi.org/10.1111/boc.201100011

    • Google Scholar
37. 1. Lampugnani MG
    2. Orsenigo F
    3. Gagliani MC
    4. Tacchetti C
    5. Dejana E

    (2006) Vascular endothelial cadherin conrols VEGFR-2 internalization and signaling from intracellular compartments

    The Journal of Cell Biology 174:593–604.

    https://doi.org/10.1083/jcb.200602080

    • Google Scholar
38. 1. Lanzetti L
    2. Rybin V
    3. Malabarba MG
    4. Christoforidis S
    5. Scita G
    6. Zerial M
    7. Di Fiore PP

    (2000) The Eps8 protein coordinates EGF receptor signalling through Rac and trafficking through Rab5

    Nature 408:374–377.

    https://doi.org/10.1038/35042605

    • Google Scholar
39. 1. Macé G
    2. Miaczynska M
    3. Zerial M
    4. Nebreda AR

    (2005) Phosphorylation of EEA1 by p38 MAP kinase regulates mu opioid receptor endocytosis

    The EMBO Journal 24:3235–3246.

    https://doi.org/10.1038/sj.emboj.7600799

    • Google Scholar
40. 1. Malerød L
    2. Stuffers JS
    3. Brech A
    4. Stenmark H

    (2007) Vps22/EAP30 in ESCRT-II mediates endosomal sorting of growth factor and chemokine receptors destined for lysosomal degradation

    Traffic 8:1617–1629.

    https://doi.org/10.1111/j.1600-0854.2007.00630.x

    • Google Scholar
41. 1. Maroun CR
    2. Jaujokas MA
    3. Holgado-Madrugada M
    4. Wong AJ
    5. Park M

    (2000) The tyrosine phosphatase SHP-2 is required for sustained activation of extracelular signal-regulated kinase and epithelial morphogenesis downstream from the met receptor tyrosine kinase

    Molecular and Cellular Biology 20:8513–8525.

    https://doi.org/10.1128/MCB.20.22.8513-8525.2000

    • Google Scholar
42. 1. Marshall CJ

    (1995) Specificity of receptor tyrosine kinase signaling: transient versus sustained extracellular signal-regulated kinase activation

    Cell 80:179–185.

    https://doi.org/10.1016/0092-8674(95)90401-8

    • Google Scholar
43. 1. Miaczynska M
    2. Christoforidis S
    3. Giner A
    4. Shevchenko A
    5. Uttenweiler-Joseph S
    6. Habermann B
    7. Wilm M
    8. Parton RG
    9. Zerial M

    (2004) APPL proteins link Rab5 to nuclear signal transduction via an endosomal compartment

    Cell 116:445–456.

    https://doi.org/10.1016/S0092-8674(04)00117-5

    • Google Scholar
44. 1. Morandell S
    2. Stasyk T
    3. Skvortsov S
    4. Ascher S
    5. Huber LA

    (2008) Quantitative proteomics and phosphoproteomics reveal novel insights into complexity and dynamics of the EGFR signalling network

    Proteomics 8:4383–4401.

    https://doi.org/10.1002/pmic.200800204

    • Google Scholar
45. 1. Nagashima T
    2. Shimodaira H
    3. Ide K
    4. Nakakuki T
    5. Tani Y
    6. Takahashi K
    7. Yumoto N
    8. Hatakeyama M

    (2007) Quantitative transcriptional control of ErbB receptor signaling undergoes graded to biphasic response for cell differentiation

    The Journal of Biological Chemistry 282:4045–4056.

    https://doi.org/10.1074/jbc.M608653200

    • Google Scholar
46. 1. Nakakuki T
    2. Birtwistle MR
    3. Saeki Y
    4. Yumoto N
    5. Ide K
    6. Nagashima T
    7. Brusch L
    8. Ogunnaike BA
    9. Okada-Hatakeyama M
    10. Kholodenko BN

    (2010) Ligand-specific c-Fos expression emerges from the spatiotemporal control of ErbB network dynamics

    Cell 141:884–896.

    https://doi.org/10.1016/j.cell.2010.03.054

    • Google Scholar
47. 1. Ohuchi T
    2. Maruoka S
    3. Sakudo A
    4. Arai T

    (2002) Assay-based quantitative analysis of PC12 cell differentiation

    Journal of Neuroscience Methods 118:1–8.

    https://doi.org/10.1016/S0165-0270(02)00116-4

    • Google Scholar
48. 1. Ohya T
    2. Miaczynska M
    3. Coskun U
    4. Lommer B
    5. Runge A
    6. Drechsel D
    7. Kalaidzidis Y
    8. Zerial M

    (2010) Reconstitution of Rab- and SNARE-dependent membrane fusion by synthetic endosomes

    Nature 459:1091–1097.

    https://doi.org/10.1038/nature08107

    • Google Scholar
49. 1. Pelkmans L
    2. Fava E
    3. Grabner H
    4. Hannus M
    5. Habermann B
    6. Krausz E
    7. Zerial M

    (2005) Genome-wide analysis of human kinases in clathrin- and caveola/raft-mediated endocytosis

    Nature 436:78–86.

    https://doi.org/10.1038/nature03571

    • Google Scholar
50. 1. Piper RC
    2. Katzmann DJ

    (2007) Biogenesis and function of multivesicular bodies

    Annual Review of Cell and Developmental Biology 23:519–547.

    https://doi.org/10.1146/annurev.cellbio.23.090506.123319

    • Google Scholar
51. 1. Poser I
    2. Sarov M
    3. Hutchins JR
    4. Hériché JK
    5. Toyoda Y
    6. Pozniakovsky A
    7. Weigl D
    8. Nitzsche A
    9. Hegemann B
    10. Bird AW
    11. Pelletier L
    12. Kittler R
    13. Hua S
    14. Naumann R
    15. Augsburg M
    16. Sykora MM
    17. Hofemeister H
    18. Zhang Y
    19. Nasmyth K
    20. White KP
    21. Dietzel S
    22. Mechtler K
    23. Durbin R
    24. Stewart AF
    25. Peters JM
    26. Buchholz F
    27. Hyman AA

    (2008) BAC TransgeneOmics: a high-throughput method for exploration of protein function in mammals

    Nature Methods 5:409–415.

    https://doi.org/10.1038/nmeth.1199

    • Google Scholar
52. Book

    1. Press WH
    2. Teukolsky SA
    3. Vetterling WT
    4. Flannery BP

    (1992)

    Numerical recipes in c. the art of scientific computing

    New York: Cambridge University Press.

    • Google Scholar
53. 1. Resat H
    2. Ewald JA
    3. Dixon DA
    4. Wiley HS

    (2003) An integrated model of epidermal growth factor receptor trafficking and signal transduction

    Biophysical Journal 85:730–743.

    https://doi.org/10.1016/S0006-3495(03)74516-0

    • Google Scholar
54. 1. Rink J
    2. Ghigo E
    3. Kalaidzidis Y
    4. Zerial M

    (2005) Rab conversion as a mechanism of progression from early to late endosomes

    Cell 122:735–749.

    https://doi.org/10.1016/j.cell.2005.06.043

    • Google Scholar
55. 1. Row PE
    2. Clague MJ
    3. Urbé S

    (2005) Growth factors induce differential phosphorylation profiles of the Hrs-STAM complex: a common node in signalling networks with signal-specific properties

    The Biochemical Journal 389:629–636.

    https://doi.org/10.1042/BJ20050067

    • Google Scholar
56. 1. Santos SD
    2. Verveer PJ
    3. Bastiaens PI

    (2007) Growth factor-induced MAPK network topology shapes Erk responses determining PC-12 cell fate

    Nature Cell Biology 9:324–330.

    https://doi.org/10.1038/ncb1543

    • Google Scholar
57. 1. Sasagawa S
    2. Ozaki Y
    3. Fujita K
    4. Kuroda S

    (2005) Prediction and validation of the distinct dynamics of transient and sustained ERK activation

    Nature Cell Biology 7:365–373.

    https://doi.org/10.1038/ncb1233

    • Google Scholar
58. 1. Schenck A
    2. Goto-Silva L
    3. Collinet C
    4. Rhinn M
    5. Giner A
    6. Habermann B
    7. Brand M
    8. Zerial M

    (2008) The endosomal protein Appl1 mediates Akt substrate specificity and cell survival in vertebrate development

    Cell 133:486–497.

    https://doi.org/10.1016/j.cell.2008.02.044

    • Google Scholar
59. 1. Sigismund S
    2. Argenzio E
    3. Tosoni D
    4. Cavallaro E
    5. Polo S
    6. Di Fiore PP

    (2008) Clathrin-mediated internalization is essential for sustained EGFR signaling but dispensable for degradation

    Developmental Cell 15:209–219.

    https://doi.org/10.1016/j.devcel.2008.06.012

    • Google Scholar
60. 1. Sigismund S
    2. Algisi V
    3. Nappo G
    4. Conte A
    5. Pascolutti R
    6. Cuomo A
    7. Bonaldi T
    8. Argenzio E
    9. Verhoef LG
    10. Maspero E
    11. Bianchi F
    12. Capuani F
    13. Ciliberto A
    14. Polo S
    15. Di Fiore PP

    (2013) Threshold-controled ubiquitination of the EGFR directs receptor fate

    The EMBO Journal 32:2140–2157.

    https://doi.org/10.1038/emboj.2013.149

    • Google Scholar
61. 1. Sorkin A

    (2001) Internalization of the epidermal growth factor receptor: role in signaling

    Biochemical Society Transactions 29:480–484.

    https://doi.org/10.1042/BST0290480

    • Google Scholar
62. 1. Sorkin A
    2. Goh LK

    (2009) Endocytosis and intracellular trafficking of ErbBs

    Experimental Cell Research 315:683–696.

    https://doi.org/10.1016/j.yexcr.2008.07.029

    • Google Scholar
63. 1. Sousa LP
    2. Lax I
    3. Shen H
    4. Ferguson SM
    5. De Camilli P
    6. Schlessinger J

    (2012) Suppression of EGFR endocytosis by dynamin depletion reveals that EGFR signalling occurs primarily at the plasma membrane

    Proceedings of the National Academy of Sciences of USA 109:4419–4424.

    https://doi.org/10.1073/pnas.1200164109

    • Google Scholar
64. 1. Stasyk T
    2. Schiefermeier N
    3. Skvortsov S
    4. Zwierzina H
    5. Peränen J
    6. Bonn GK
    7. Huber LA

    (2007) Identification of endosomal epidermal growth factor receptor signaling targets by functional organelle proteomics

    Molecular & Cellular Proteomics 6:908–922.

    https://doi.org/10.1074/mcp.M600463-MCP200

    • Google Scholar
65. 1. Tall GG
    2. Barbieri MA
    3. Stahl PD
    4. Horazdovsky BF

    (2001) Ras-activated endocytosis is mediated by the Rab5 guanine nucleotide exchange activity of RIN1

    Developmental Cell 1:73–82.

    https://doi.org/10.1016/S1534-5807(01)00008-9

    • Google Scholar
66. 1. Tanimizu N
    2. Nishikawa M
    3. Saito H
    4. Tsujimura T
    5. Miyajima A

    (2003) Isolation of hepatoblasts based on the expression of Dlk/Pref-1

    Journal of Cell Science 116:1775–17786.

    https://doi.org/10.1242/jcs.00388

    • Google Scholar
67. 1. Tarcic G
    2. Boguslavsky SK
    3. Wakim J
    4. Kiuchi T
    5. Liu A
    6. Reinitz F
    7. Nathanson D
    8. Takahashi T
    9. Mischel PS
    10. Ng T
    11. Yarden Y

    (2009) An unbiased screen identifies DEP-1 tumor suppressor as a phosphatase controlling EGFR endocytosis

    Current Biology 19:1788–1798.

    https://doi.org/10.1016/j.cub.2009.09.048

    • Google Scholar
68. 1. Teis D
    2. Wunderlich W
    3. Huber LA

    (2002) Localization of the MP1-MAPK scaffold complex to endosomes is mediated by p14 and required for signal transduction

    Developmental Cell 3:803–814.

    https://doi.org/10.1016/S1534-5807(02)00364-7

    • Google Scholar
69. 1. Tian T
    2. Harding A
    3. Inder K
    4. Plowman S
    5. Parton RG
    6. Hancock JF

    (2007) Plasma membrane nanoswitches generate high-fidelity Ras signal transduction

    Nature Cell Biology 9:905–914.

    https://doi.org/10.1038/ncb1615

    • Google Scholar
70. 1. Turing AM

    (1952) The chemical basis of morphogenesis

    Philosophical Transactions of the Royal Society of London Series B, Biological Sciences 237:37–72.

    https://doi.org/10.1098/rstb.1952.0012

    • Google Scholar
71. 1. Umebayashi K
    2. Stenmark H
    3. Yoshimori T

    (2008) Ubc4/5 and c-Cbl continue to ubiquitinate EGF receptor after internalization to facilitate polyubiquitination and degradation

    Molecular Biology of the Cell 19:3454–3462.

    https://doi.org/10.1091/mbc.E07-10-0988

    • Google Scholar
72. 1. Urbé S
    2. Mills IG
    3. Stenmark H
    4. Kitamura N
    5. Clague MJ

    (2000) Endosomal localization and receptor dynamics determine tyrosine phosphorylation of hepatocyte growth factor-regulated tyrosine kinase substrate

    Molecular and Cellular Biology 20:7685–7692.

    https://doi.org/10.1128/MCB.20.20.7685-7692.2000

    • Google Scholar
73. 1. van de Linde S
    2. Löschberger A
    3. Klein T
    4. Heidbreder M
    5. Wolter S
    6. Heilemann M
    7. Sauer M

    (2011) Direct stochastic optical reconstruction microscopy with standard fluorescent probes

    Natur Protocols 6:991–1009.

    https://doi.org/10.1038/nprot.2011.336

    • Google Scholar
74. 1. Verveer PJ
    2. Wouters FS
    3. Reynolds AR
    4. Bastiaens PIH

    (2000) Quantitative imaging of Lateral ErbB1 receptor signal propagation in the plasma membrane

    Science 290:1567–1570.

    https://doi.org/10.1126/science.290.5496.1567

    • Google Scholar
75. 1. Vieira AV
    2. Lamaze C
    3. Schmid SL

    (1996) Control of EGF receptor signaling by clathrin-mediated endocytosis

    Science 274:2086–2089.

    https://doi.org/10.1126/science.274.5295.2086

    • Google Scholar
76. 1. Wiley HS

    (2003) Trafficking of the ErbB receptors and its influence on signaling

    Experimental Cell Research 284:78–88.

    https://doi.org/10.1016/S0014-4827(03)00002-8

    • Google Scholar
77. 1. Wiley HS
    2. Herbst JJ
    3. Walsh BJ
    4. Lauffenburger DA
    5. Rosenfeld MG
    6. Gill GN

    (1991)

    The role of tyrosine kinase activity in endocytosis, compartmentation, and down-regulation of the epidermal growth factor receptor

    The Journal of Biological Chemistry 266:11083–11094.

    • Google Scholar
78. 1. Wouters FS
    2. Bastiaens PI

    (1999) Fluorescence lifetime imaging of receptor tyrosine kinase activity in cells

    Current Biology 9:1127–1130.

    https://doi.org/10.1016/S0960-9822(99)80484-9

    • Google Scholar
79. 1. York RD
    2. Yao H
    3. Dillon T
    4. Ellig CL
    5. Eckert SP
    6. McCleskey EW
    7. Stork PJ

    (1998) Rap1 mediates sustained MAP kinase activation induced by nerve growth factor

    Nature 392:622–626.

    https://doi.org/10.1038/33451

    • Google Scholar
80. 1. Zeigerer A
    2. Gilleron J
    3. Bogorad RL
    4. Marsico G
    5. Nonaka H
    6. Seifert S
    7. Epstein-Barash H
    8. Kuchimanchi S
    9. Peng CG
    10. Ruda VM
    11. Del Conte-Zerial P
    12. Hengstler JG
    13. Kalaidzidis Y
    14. Koteliansky V
    15. Zerial M

    (2012) Rab5 is necessary for the biogenesis of the endolysosomal system in vivo

    Nature 485:465–470.

    https://doi.org/10.1038/nature11133

    • Google Scholar
81. 1. Zwang Y
    2. Yarden Y

    (2006) p38 MAP kinase mediates stress-induced internalization of EGFR: implications for cancer chemotherapy

    The EMBO Journal 25:4195–4206.

    https://doi.org/10.1038/sj.emboj.7601297

    • Google Scholar

Author details

1. Roberto Villaseñor

   Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany

   Contribution

   RV, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

2. Hidenori Nonaka

   Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany

   Contribution

   HN, Established procedures for the isolation, culture and staining conditions of the embryonic hepatoblasts, Contributed unpublished essential data or reagents

   Competing interests

   The authors declare that no competing interests exist.

3. Perla Del Conte-Zerial

   Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany

   Contribution

   PDC-Z, Developed the mathematical model and simulations, Analysis and interpretation of data

   Competing interests

   The authors declare that no competing interests exist.

4. Yannis Kalaidzidis

   1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
   2. Faculty of Bioengineering and Bioinformatics, Moscow State University, Moscow, Russia

   Contribution

   YK, Developed the FRET correction algorithm, Developed method to estimate the number of fluorescent molecules in light microscopy images, Supervised the image analysis, Developed the mathematical model and simulations, Conception and design, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

5. Marino Zerial

   Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany

   Contribution

   MZ, Directed Study, Conception and design, Drafting or revising the article

   For correspondence

   zerial@mpi-cbg.de

   Competing interests

   The authors declare that no competing interests exist.

• 5,771

  views

• 1,228

  downloads

• 78

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.