Veterans Affairs Medical Center, Tennessee Valley Healthcare System, Nashville, TN 37212, Division of Rheumatology and Immunology, Department of Medicine, Department of Cancer Biology, Department of Cell & Developmental Biology, Vanderbilt-Ingram Cancer Center, Vanderbilt University School of Medicine, Nashville, TN, USA
Eph receptors and their membrane-bound ligands, the ephrins, represent a complex subfamily of receptor tyrosine kinases (RTKs). Eph/ephrin binding can lead to various and opposite cellular behaviors such as adhesion versus repulsion, or cell migration versus cell-adhesion. Recently, Eph endocytosis has been identified as one of the critical steps responsible for such diversity. Eph receptors, as many RTKs, are rapidly endocytosed following ligand-mediated activation and traffic through endocytic compartments prior to degradation. However, it is becoming obvious that endocytosis controls signaling in many different manners. Here we showed that activated EphA2 are degraded in the lysosomes and that about 35% of internalized receptors are recycled back to the plasma membrane. Our study is also the first to demonstrate that EphA2 retains the capacity to signal in endosomes. In particular, activated EphA2 interacted with the Rho family GEF Tiam1 in endosomes. This association led to Tiam1 activation, which in turn increased Rac1 activity and facilitated Eph/ephrin endocytosis. Disrupting Tiam1 function with RNA interference impaired both ephrinA1-dependent Rac1 activation and ephrinA1-induced EphA2 endocytosis. In summary, our findings shed new light on the regulation of EphA2 endocytosis, intracellular trafficking and signal termination and establish Tiam1 as an important modulator of EphA2 signaling.
Eph receptors and their membrane-bound ligands, the ephrins, represent a large and complex subfamily of receptor tyrosine kinases (RTKs). Members of the Eph and ephrin family are integral players of signaling pathways involved during development, axon guidance, cell migration and diseases such as cancer [1, 2]. A unique property of this family is that ligand binding generates bi-directional signaling .
Due to its complexity of signaling, Eph and ephrin family rely on a finely tuned signaling network. Endocytosis has emerged as an essential process in regulating Eph signaling pathways. It plays critical roles in signaling initiation, transduction and termination. Together with the proteolytic cleavage of Eph/ephrinA complexes [4, 5], trans-endocytosis removes Eph/ephrin complexes from the cell surface and allows the switch between cell-adhesion and cell-repulsion [6, 7]. Upon ligand binding, most of RTKs are rapidly internalized and trafficked through endosomes where they can be recycled back to the plasma membrane or degraded. Whether RTKs are degraded or recycled to the cell surface is crucial for the duration and intensity of their signal. With regard to Eph receptors, data about their endocytic sorting and degradation remain very scarce. Concerning Eph/ephrin intracellular trafficking and recycling, it has been reported that several Eph/ephrin complexes traffic through early endosomes [8, 9], but to the best of our knowledge, nothing to date has been investigated about Eph recycling.
A second, central aspect of RTKs endocytosis is the duration of their signals once internalized. Many evidences reveal that RTKs can signal from the endosomes prior to their degradation . As an example, active endosomal EGFR and PDGFR have been shown to recruit signaling molecules and determine physiological responses such as cell proliferation and migration [11, 12]. Do Ephs and ephrins continue to signal into endosomes, and if so, is the signal different from the one emanating from the cell surface? Up to now, only a few studies suggest that EphB receptors may continue to signal into endosomes by demonstrating that they remain phosphorylated in the intracellular vesicles [13, 14].
Some Rho family GTPases are known to regulate the actin cytoskeleton which in turn controls cell morphology, adhesion, motility and also receptor endocytosis and trafficking . Rho-like GTPases function as a molecular switch stimulating the exchange of GDP for GTP via guanine-nucleotide exchange factors (GEFs, ). In addition to activate the Rho GTPases, GEFs also participate to downstream signaling by interacting with scaffold proteins promoting Rho GTPases/specific downstream effectors association. Many reports have shown that Eph/ephrin signaling complexes are related to the Rho family. In particular, Rac1 has been linked to Eph/ephrin internalization and trafficking and its activation is required for Eph/ephrin-induced membrane ruffling and trans-endocytosis [13, 17]. The molecular mechanism by which EphA2/ephrinA1 complexes regulate Rac1 activation remains to be determined. Several Rho GEFs link Ephs to the activation of Rho proteins. Ephexin 4 interacts with EphA2 to activate RhoG and mediate cell migration . The Vav family is involved in EphA2-mediated angiogenesis  and Vav2 has been shown to interact with EphA and B positively regulating Eph/ephrin endocytosis . Recently, Yoo et al. found that Tiam1 (T-Lymphoma invasion and metastasis protein), a Rac1-specific GEF, is required for ephrinA5/EphA8 clathrin-mediated endocytosis . In addition, Tiam1 has been described to mediate EphB-dependent dendritic spine development and EphA-dependent neurite outgrowth [20, 21]. Thus, Tiam1 could play a central role in mediating Eph-dependent cellular behavior.
In this work, we investigated the molecular mechanisms regulating EphA2/ephrinA1 endocytosis, intracellular trafficking and signal termination in MiaPaCa-2 pancreatic carcinoma cells endogenously overexpressing EphA2. We demonstrated that in response to ephrin binding, activated EphA2 receptors internalize, traffic through endosomes and reach the lysosome for subsequent degradation, a process regulated by the ubiquitin ligase c-Cbl. We provide the first insight that a portion of the internalized EphA2 receptors are recycled back to the plasma membrane. In addition, we demonstrated that internalized Eph/ephrin complexes remain active throughout the endocytic pathway and retain the capacity to signal from endosomes by recruiting and activating Tiam1. Silencing Tiam1 impaired both ephrinA1-dependent activation of Rac1 and EphA2 endocytosis, definitively establishing Tiam1 as an important modulator of EphA2 signaling.
Ligand activated EphA2 receptors are degraded through the lysosomal pathway
EphrinA1/Fc is a soluble recombinant ephrinA1 fused to the Fc portion of human IgG. Our laboratory and others have previously shown that ephrin mimetic peptides induce receptor phosphorylation, degradation and downstream signaling as efficiently as ephrinA1 on the surface of ligand-expressing cells through juxtacrine contact [9, 22].
EphrinA1/Fc stimulation of MiaPaCa-2 cells induced degradation of total and cell surface-associated EphA2 receptors in a time-dependent manner (Figure 1A,B). Four hours after stimulation, more than 60% of total EphA2 receptors were degraded whereas only a few receptors were degraded in cells incubated with control IgG. This was not due to a general downregulation of cellular proteins as β-actin and caveolin-1 protein levels remained unchanged (Figure 1A).
Although many RTKs are degraded through the lysosomal pathway [23, 24], EphA2 and EphA3 receptors degradation has been shown to occur through the proteasomal pathway [22, 25, 26]. To clarify the mechanisms of EphA2 degradation we incubated MiaPaCa-2 cells with a lysosomal inhibitor, bafilomycin A1 and proteasomal inhibitors, bortezomib, MG132 and lactacystin. EphA2 degradation was strongly inhibited by bafilomycin, whereas bortezomib, MG132 and lactacystin, simply delayed EphA2 degradation, suggesting that proteasomal activity may influence EphA2 degradation, which mainly occurred through the lysosomal pathway (Figure 2A, data not shown).
Finally, to visualize EphA2 trafficking and intracellular localization after stimulation, we performed immunofluorescence studies. Confocal microscopy images and the co-localization ratio (C.R.) revealed that following stimulation, EphA2 receptors pass through early endosomes within 15 min and reach lysosomes within 30–60 min for subsequent degradation (Figure 2B).
Activated EphA2 receptors are ubiquitinated and interact with Cbl through its TKB domain
To further investigate the molecular mechanisms leading to EphA2 degradation, we assessed receptor activation and ubiquitination. As shown in Figure 3A (upper blot), maximal EphA2 phosphorylation was observed after 20 min of stimulation as well as maximal EphA2 ubiquitination (Figure 3A middle blot).
The c-Cbl adaptor protein has emerged as a negative regulator of RTKs and other tyrosine kinases, a function dependent on its ubiquitin ligase activity following its phosphorylation . Many RTKs, such as EGF, PDGF or EphB1 receptors, activate c-Cbl through phosphorylation by members of the Src family [9, 28, 29]. In MiaPaCa-2 cells, c-Cbl phosphorylation was detected 10 min after ephrinA1/Fc stimulation and increased over the time (Figure 3A lower blot).
Next, we showed by co-immunoprecipitation that c-Cbl was recruited to the EphA2 signaling complex within 10 min following stimulation (Figure 3B). This interaction was dependent on the EphA2 phosphorylation status, as no EphA2/Cbl complex was observed when EphA2 phosphorylation was impaired (data not shown). EphA2/Cbl interaction was confirmed by confocal microscopy and glutathione S-transferase (GST) pull-down assays. Before stimulation, EphA2 localized at the plasma membrane, whereas Cbl was mainly cytoplasmic. However, after 30 min of ligand stimulation, EphA2 and Cbl co-localized into the cytoplasm (Figure 3C).
Cbl has been shown to bind to activated RTKs through its tyrosine kinase binding domain (TKB, ). A GST pull-down assay demonstrated that EphA2 receptors in lysates of stimulated cells strongly bound to GST-Cbl-N whereas no binding was observed with GST control or with GST-Cbl-N-G306E (Cbl-N with a point mutation in the TKB domain). No binding was observed with unstimulated lysates (Figure 3D).
Collectively, these results indicate that activated EphA2 recruits Cbl and induces its phosphorylation, which probably leads to EphA2 ubiquitination and downregulation through the lysosomal pathway.
Part of internalized EphA2 receptors is recycled back to the plasma membrane via both the early and the late recycling pathways
Once internalized, RTKs may become uncoupled from their ligand due to acidic intracellular compartments and recycle back to the cell surface, or continue through the endocytic pathway to be degraded . To investigate the ability of EphA2 receptors to recycle back to the plasma membrane after endocytosis, we adapted the protocol used by Turvy and Blum . This method allowed us to directly quantify cell-surface expression, endocytosis and recycling of EphA2 (protocol in Figure 4A).
To block EphA2 degradation and observe its accumulation in the endocytic pathway, cells were incubated with bafilomycin, an inhibitor of endosomal acidification and thus of lysosomal degradation. Using capture-enzyme-linked immunosorbent assay (ELISA) to quantify internalized biotinylated EphA2, we found that 30 min after ligand stimulation, about 40% of the initial amount of EphA2 was internalized, with the caveats that at this time point, some parts are already recycled back to the plasma membrane (Figure 4B). EphA2 receptors accumulation in endosomes increased when cells were incubated with bafilomycin. The fact that bafilomycin prevents receptor degradation could explain the accumulation during stimulation times of 15–60 min. Before receptors reach lysosomes at the 5 and 10 min time points, bafilomycin also affects EphA2 endocytosis. On the basis of the literature , one possibility is that bafilomycin affects the recycling rate of EphA2, however, more evidence should be provided to confirm this hypothesis. To demonstrate EphA2 recycling, cells were incubated one more time at 37°C for different periods of time (Figure 4A). Capture-ELISA showed that a recycling time of 15–30 min allow 28–35% of the internalized EphA2 to recycle, meaning that 10–12% of total surface EphA2 receptors are recycled after stimulation. The longer the cells are incubated at 37°C, the more EphA2 receptors are recycled back to the plasma membrane. Primaquine, an inhibitor of endosomal recycling, blocked transferrin recycling in our cells (data not shown) and indeed blocked EphA2 recycling (Figure 4C). The same experiments were performed with PC-3 cells, a human prostate cancer cell line overexpressing EphA2, and confirmed that the percentages of internalized and recycled EphA2 were comparable (data not shown).
To validate our experiments, biotinylated proteins were isolated with streptavidin agarose beads. EphA2 immunoblotting confirmed the following findings: cell-surface EphA2 receptors were correctly biotinylated (‘total surface’), the stripping step efficiently removed biotin from cell-surface EphA2 (‘strip’), the longer the cells were incubated at 37°C the more EphA2 were endocytosed (‘internalized’) and finally, less biotinylated EphA2 remained after a second incubation at 37°C and a reducing wash, suggesting that a part of EphA2 receptors are recycled back to the plasma membrane (‘post-recycling’, Figure 4D).
Rab4 and Rab11 belong to the Rab family, a part of the Ras superfamily of small GTPases. They both function as regulators of the recycling pathway. Rab4 is localized to early and recycling endosomes, and has been shown to sort recycling receptors directly from early endosomes to the plasma membrane in a ‘fast recycling’ route [34, 35]. Rab11 can regulate the receptor traffic from the Golgi apparatus to the plasma membrane and control the ‘slow recycling’ route through the perinuclear recycling endosomes . To further characterize the EphA2 recycling pathways, we performed co-immunoprecipitation and immunofluorescence experiments with Rab4 and Rab11. As shown in Figure 5A, a swift EphA2/Rab4 interaction was observed after 5 min of ligand stimulation, whereas EphA2 and Rab11 immunoprecipitated at a later time point, 15–30 min following stimulation. Identical results were obtained with co-immunoprecipitation using PC-3 cells, a prostate cancer cell line endogenously expressing EphA2 (Figure 5B). Confocal microscopy images confirmed that Rab4 and EphA2 co-localized 5 and 15 min after ligand stimulation, whereas Rab11 and EphA2 co-localized 15, 30 and 60 min after ligand stimulation (Figure 5C). Studies measuring the recycling rates of the canonical recycling receptor, the transferrin receptor, confirmed the kinetic of the ‘fast’ recycling pathway (t1/2 ˜5 min) and the ‘slow’ recycling pathway (t1/2 ˜15–30 min, ). In addition, the pathway followed by a part of EphA2 is similar to the one taken by transferrin, as demonstrated by labeled transferrin/EphA2 co-localization experiments (data not shown).
Overall, these results suggest that following stimulation, a part of EphA2 receptors are recycled back to the cell surface via both the early and late recycling routes.
EphA2 receptors remain ligand-associated and phosphorylated in early endosomes
Since the emergence of the signaling endosomes concept coming from neuronal studies [38, 39], numerous examples show the signaling capacity of receptors localized to endosomes [11, 12]. Once internalized, a receptor can remain active if it stays ligand-coupled, phosphorylated and transduces downstream signaling. Although it has been shown that internalized Eph receptors can be tyrosine phosphorylated [13, 14], it is not known how long they stay active and whether they remain associated with their ligands.
Internalized receptor/ligand complexes progressively encounter more acidic environments once they penetrate and advance into the endocytic pathway. In addition, receptor-ligand associations are pH-sensitive. To test at which pH ephrinA1 dissociates from EphA2 receptors, two different methods were applied: one using biotinylated ephrinA1/Fc and capture-ELISA (Figure 6A) and one using immunofluorescence (Figure 6B). As shown in Figure 6A,C, 50% of receptor/ligand complexes were dissociated at a pH slightly lower than 5.5. The internal pH of endosomes decreases the closer they get to lysosomes: early sorting endosomes have a pH of 5.8–6.3, late endosomes a pH of 5–6 and lysosomes a pH of 5–5.5 [40, 41]. In aggregate, our findings suggest that EphA2 and ephrinA1 dissociate when they reach late endosomes and lysosomes, and thus that the bulk of the internalized EphA2 receptors remain ligand-associated within the early endosomes. Immunofluorescence experiments confirmed these results, showing a strong C.R. of EphA2 and ephrinA1 within the early endosomes, 15 min after stimulation (Figure 6D).
To confirm the presence of active EphA2 in early endosomes, we assessed its phosphorylation status throughout the whole internalization process (Figure 6E). Using an antibody specifically recognizing phospho-EphA2 (Y594), we demonstrated that 5 min after stimulation, EphA2 was phosphorylated and starts to be internalized. Fifteen minutes after stimulation, most of phospho-EphA2 has reached the early endosomes and 60 min after stimulation most of phospho-EphA2 receptors were degraded.
Taken together, our results from Figure 6 indicate that most of the internalized EphA2 receptors remain ligand-associated and phosphorylated in the early endosomes, suggesting that internalized EphA2 receptors could retain the capacity to transduce downstream signaling.
EphA2 receptors associate with Tiam1 in the early endosomes
At this point, the most interesting question relates to the potential effects of endosomes-based EphA2 signaling. Signaling from endosomes could be functionally distinct from those emanating from the cell surface, or they could just be an extension of the signal initiated at the plasma membrane .
To examine the ability of EphA2 receptors to associate with specific molecules after endocytosis, we adapted the technique developed by Burke et al. to separate internal from cell-surface proteins . Cells were incubated with ephrinA1/Fc prior to cell-surface biotinylation. As shown in Figure 7B, biotinylated proteins were precipitated using streptavidin agarose, allowing the separation of non-biotinylated, internalized proteins (supernatant) from biotinylated cell-surface proteins (beads). As we have previously shown, internalized EphA2 remained phosphorylated for a considerable length of time. Total EphA2 and tyrosine phosphorylated EphA2 levels decreased at the same time, meaning that receptor dephosphorylation did not precede receptor degradation (Figure 7C, upper blots). Moreover, the ratio between phosphorylated EphA2 and total EphA2 was higher in the internalized fraction than in the cell-surface fraction, showing that phosphorylated receptors were internalized within 30 min (Figure 7C–E). Tiam1, a GEF specific for Rac1, has previously been reported to be involved in EphA8 endocytosis . Although Tiam1 is mainly cytoplasmic, it has been shown to be recruited to early endosomes in Rab5 overexpressing cells . Therefore we hypothesized that following stimulation; EphA2 and Tiam1 could associate into endocytic vesicles and early endosomes. Immunoprecipitation experiments first revealed that Tiam1 is recruited to the EphA2 complex 2–5 min after stimulation in MiaPaCa-2 and PC-3 cells (Figure 7A). Then, co-precipitation experiments demonstrated that internalized EphA2 associated with Tiam1 5–15 min after stimulation, but no Tiam1 associates with cell-surface EphA2 (Figure 7C,D). The Tiam1 blot of the entire biotinylated fraction (prior to the step of precipitation) revealed that no Tiam1 was present in the surface protein fraction in contrast to EphA2 (Figure 7D lower part).
To visualize the EphA2-Tiam1 interaction we performed immunofluorescence studies. Confocal microscopy images confirmed that EphA2 and Tiam1 associate following ephrinA1/Fc stimulation. In addition, more Tiam1 were found in early endosomes after ephrinA1/Fc treatment (Figure 7F).
Taken together, these results demonstrate that EphA2 remains phosphorylated during the endocytic pathway. For technical reasons (use of EEA1 for co-localization) our experiments indeed prove the interactions in early endosomes; however, they cannot rule out that it could also take place at an earlier time point, i.e. in (coated or uncoated) endocytic vesicles already.
Therefore, endosomes could be important sites for EphA2-initiated signal transduction via Tiam1.
Characterization of Tiam1 activation
To further characterize EphA2-Tiam1 interaction, cells were transiently transfected with EphA2 c-myc tagged mutant plasmids and their ability to interact with Tiam1 under ephrinA1/Fc stimulation was tested. We confirmed that the kinase dead (KD) EphA2 mutant and the non-phosphorylatable double-mutant retaining a kinase activity (EE) were not phosphorylated by ephrinA1/Fc stimulation. We found that Tiam1 precipitates with wild type (WT) EphA2 but co-immunoprecipitation with EphA2 mutants lacking EE and/or phosphorylated kinase was significantly decreased (Figure 8A). Since the mutated tyrosines are localized in the juxtamembrane of EphA2 these results suggest that Tiam1 may directly bind the phosphorylated tyrosines in the juxtamembrane region of EphA2. The lack of Tiam1 binding to EphA2 mutants cannot be due to their inability to internalize as their internalization process is as good as for the WT (data not shown). Knowing that Tiam1 and EphA2 interact in a phosphorylation-dependent manner we investigated whether this interaction triggers Tiam1 activation. Using an active Rac-GEF assay kit we succeeded at pulling down active Tiam1 in 2, 5 and 10 min stimulated cell lysates, but not in unstimulated lysates. Cells stimulated with hydrogen peroxide and sodium orthovanadate were used as a positive control of Tiam1 activity (, Figure 8B). To further confirm that ephrinA1/Fc induces Tiam1 activation, we performed co-immunoprecipitation assays between Tiam1-HA and endogenous Rac1. We found that the Rac1/Tiam1 interaction was maximum 2 min after stimulation and diminished in later time points, suggesting that Tiam1 is activated at these time points and facilitates Rac1 activation (Figure 8C).
Having established that Tiam1 is activated following stimulation, we next sought to determine the mechanisms leading to this activation. Since several studies have previously demonstrated that Tiam1 activity can be modulated by tyrosine phosphorylation [20, 21], we examined the Tiam1 tyrosine phosphorylation status in cells treated with ephrinA1/Fc. Using endogenous Tiam1 immunoprecipitation and anti-phosphotyrosine immunoblots we were not able to demonstrate Tiam1 phosphorylation following ephrinA1/Fc stimulation. However, when we used the specific phospho Tiam1 (Y384) antibody in Tiam-HA transfected cells we observed a weak Tiam1 phosphorylation 2, 5 and 10 min after ephrinA1/Fc stimulation (Figure 8D).
Given the weakness of the phosphorylation, we sought to determine whether Tiam1 is activated by another mechanism. An interesting pathway leading to Tiam1 activation is the one mediated by Ras. It has been reported that Ras can directly bind to the Ras-binding domain (RBD) of Tiam1 and facilitate its activation [46, 47]. Therefore, in order to evaluate the role of Ras in Tiam1 activation mediated by ephrinA1/Fc, we performed co-immunoprecipitation assays between Ras and Tiam1. Using this method we did not detect any Ras and Tiam1 association in response to ephrinA1/Fc (Figure 8E). These results suggest that Tiam1 is activated by a mechanism independent of the Ras pathway.
Taken together, our data demonstrate that only a small part of Tiam1 is activated by phosphorylation – how Tiam1 is fully activated following ephrinA1/Fc stimulation is not completely determined. According to the literature , the activation of Tiam1 is complex and involves a myriad of potential mechanisms such as phosphorylation, phosphoinositide binding, modulation of cellular levels and/or interaction with other proteins. However, in the majority of the cases, the precise stimuli leading to Tiam1 activation and determining Tiam1-mediated cellular functions remain to be defined.
Tiam1 is required for Rac1 activation induced by ephrinA1/Fc
Previous studies have reported that ephrinA1/Fc stimulation increases or decreases Rac1 GTPase activation and that Rac1 activity is required for endocytosis of Eph proteins [13, 49]. We thus investigated whether Eph/ephrin signaling-mediated Rac1 activation is affected by Tiam1 downregulation.
First, we performed RNA interference experiments. As shown in Figure 8F, Tiam1 protein expression was reduced by 62% by Tiam1 siRNA compared with scrambled siRNA. As shown in Figure 8G, upon ephrinA1/Fc stimulation, we detected a transient activation of Rac1 GTPase, with a peak at 2–5 min. In Tiam1 knockdown cells, Rac1 activation was significantly decreased 2 and 5 min after stimulation. In contrast, ephrinA1/Fc stimulation decreased RhoA activity. Tiam1 knock down seems to attenuate ephrinA1-induced RhoA-GTP decrease; however, the results are not significant, suggesting that another GEF could be involved in this process. EphrinA1/Fc stimulation increased the Cdc42 GTPases level in WT cells, whereas the GTP-bound-Cdc42 level seems reduced in Tiam1knock down cells. However, this result is not statistically significant. Taken together, these results strongly suggest that Tiam1 specifically mediates ephrinA1-dependent Rac1 activation and probably ephrinA1-dependent Cdc42 activation (Figure 8G,H, *p < 0.05).
Tiam1 is required for ephrinA1/Fc-induced endocytosis of EphA2 receptors
Next, we sought to determine the biological effects of Tiam1/EphA2 association. As it is known that Tiam1 is involved in Rac1 activity and EphA8/ephrinA5 endocytosis , we investigated whether Tiam1 would play a role in ephrinA1-induced EphA2 endocytosis.
Cells were transfected with Tiam1 siRNA or scrambled siRNA and subjected to a biotinylation endocytosis assay, as described in Figure 3A. Using capture-ELISA to measure the amount of internalized EphA2, we found that EphA2 endocytosis was significantly reduced in Tiam1 siRNA transfected cells (Figure 8I, ˜40%, *p < 0.05; **p < 0.01). These data suggest that Tiam1 is required for ephrinA1-induced EphA2 endocytosis.
In this study, we confirmed in MiaPaCa-2 and PC-3 cells endogenously overexpressing EphA2 that upon ephrinA1/Fc stimulation, activated EphA2 recruits Cbl and induces its phosphorylation, which in turn leads to EphA2 ubiquitination and lysosomal degradation. EphA2 internalization is the first step in receptor degradation; however, internalization is not necessarily an attenuation process. Our findings suggest that a part of internalized EphA2 is recycled back to the plasma membrane, increasing the receptor half-life. In addition, as demonstrated for other RTKs, but not to date for Eph receptors, we present evidence that internalized ephrinA1/EphA2 complexes are still engaged in active signaling. We demonstrated that active EphA2 reaches endosomes as early as 5 min following stimulation. In those endocytic compartments, EphA2 induces the activation and recruitment of a Rho GEF, Tiam1, leading to increased Rac1 activity. We also observed that disrupting Tiam1 function with RNA interference impaired ephrinA1-dependent Rac1 activation, suggesting that Tiam1 is an important GEF downstream of Eph receptors. Our data also indicate that Tiam1 is required for ephrinA1-induced EphA2 endocytosis. Taken together, our observations clarify the molecular mechanisms regulating Eph endocytosis, intracellular trafficking and signal termination (Model Figure 9).
Degradation and recycling pathways of Eph receptors
The lysosomal pathway is the preferred route of degradation for most of the RTKs [23, 24]. This is also true for EphA receptors as demonstrated by Kajiho et al. and previous work from our lab [9, 50].
EphA2 and EphA3 are degraded through the proteasomal pathway [22, 25, 26]. We and Kajiho have also found a participation, albeit weak, of the proteasome in the process of EphA degradation, thus the proportion of the proteosomal involvement may depend on receptor or/and on cell type .
It is thought that, upon ligand binding, RTKs are rapidly internalized and trafficked to endosomes. From endosomes, receptors are either transported to lysosomes for degradation, to the trans-Golgi network or to recycling endosomal compartments that return to the plasma membrane. Receptor fate is regulated by ubiquitination, which in turn is dependent on ligand binding. Ligand dissociation in the mildly low pH of early endosomes results in loss of ubiquitin signal and promotes EGFR recycling to the cell surface . In contrast, PDGF receptors have been shown to be efficiently targeted to lysosomes independent of ligand binding . What about Eph receptors? There has been to date, only one study suggesting that Eph receptors can be recycled. VAB-1, the Eph receptor in Caenorhabditis elegans, was shown to transit to the endocytic-recycling compartment independently of ligand binding . Our results demonstrate that Eph receptors can be recycled back to the plasma membrane. However, compared to other receptors such as EGFR (TGF-α or Epi stimulated), met receptors or transferrin receptors which are almost 100% recycled, only 35% of internalized EphA2 receptors take this pathway [54, 55]. Our data show that ephrinA1/Fc binding to EphA2 is relatively stable at the mildly acidic pH found in early endosomes, thus EphA2 is continuously ubiquitinated by c-Cbl and transported to lysosomes. This result could explain why only a small part of Eph receptors can be recycled back to the plasma membrane.
Many Rab GTPases including Rab4, Rab5, Rab7, Rab11 and Rab21, have been identified as critical proteins that regulate RTK endocytosis and recycling and thereafter RTK signaling [54, 56, 57]. To further characterize EphA2 recycling pathways we demonstrated that active EphA2 co-localized and interacted with Rab4 and Rab11 (Figure 5). Although several RTKs such as EGFR [54, 58], PDGFR [57, 59, 60], VEGFR  and MET RTKs , have been shown to co-localized with Rab4 and Rab11 in different cell lines, studies demonstrating that cargos interacted with Rab proteins are scarce. To our knowledge, the only study showing such an interaction demonstrated that Rab21 interacted with EGFR and enhanced its degradation . What could be the role of EphA2/ Rab4 and EphA2/Rab11 interactions? It is well known that Rab4 and Rab11 coordinate vesicles trafficking during recycling . We envision that Rab4 and Rab11 interacted with activated EphA2 to regulate its trafficking and signaling, as it was demonstrated for Rab21 and EGFR . However, more evidence should be provided to define the specific role of EphA2/Rab interactions, which is beyond the scope of this study.
Sustained Eph receptor signaling within the endosomal compartment
The ability of a receptor to signal after endocytosis is important to ensure the sufficient duration and amplitude of signaling. However, this capacity requires receptors to remain active in endosomes. It has been reported that phosphorylated EphBs are internalized [13, 14]. In this work, we have now clearly demonstrated that EphA2 remains ligand bound, phosphorylated and active until later stages of endosomal trafficking. A crucial question is whether these internal active Ephs still transduce signals from endosomes. Our demonstration by fluorescence microscopy and protein isolation that EphA2 and Tiam1 associate in endosomes but not at the plasma membrane, leading to Tiam1 and Rac1 activation, strongly suggest that EphA2 indeed is able to transduce specific signals from the endosomal membrane. It has previously been proposed that following RTK activation, Rac1 could be activated in early endosomes through Tiam1 . While Tiam1 is mainly cytoplasmic and has been shown to translocate to the plasma membrane upon cellular activation, we proposed that active EphA2 recruits soluble Tiam1 to the endosomes (early endosomes and even before to endocytic vesicles). Whether soluble Rac1 is also recruited to the endosomes or whether Rac1 is activated at the plasma membrane remains to be elucidated.
Regulation of Tiam1 activation
Only a few studies have shown an interaction between Eph receptors (including EphA2, EphB2, EphA4 and EphA8) and Tiam1 [8, 20, 21]. In all the reports, the cytoplasmic region of Eph receptors is required for Tiam1 association. Although the precise region is not clearly established, one group found that the juxtamembrane region of EphA8 may be involved in Tiam1 association . In accordance with this observation, we showed that EphA2 tyrosine phosphorylation in the juxtamembrane region is necessary for its association with Tiam1.
How is Tiam1 activated by Eph receptors? Previous studies suggest that phosphorylation of Tiam1 (probably by Src) is an efficient mechanism to trigger its activation [20, 21, 45, 63]. However, the robustness of Tiam1 phosphorylation was very much cell- and context dependent. For example, although downregulation of Tiam1 significantly inhibited nerve growth factor (NGF)-dependent Rac1 activation and neurite outgrowth, Tiam1 phosphorylation was not observed in response to NGF whereas Tiam1 is phosphorylated by TrkB following BDNF stimulation . In our system, we were able to detect phosphorylation of Tiam1 only when it was overexpressed in MiaPaca2 cells. This observation could be explained by the low amount of phosphorylated Tiam1 compared to total Tiam1, or by the fact that Tiam1 may not be significantly phosphorylated.
Several groups have demonstrated that active Ras can cooperate with Tiam1 to activate Rac1 in a mechanism independent of PI3K [46, 64]. Our findings suggest that Ras is not involved in Tiam1 activation. These results open the possibility that ephrinA1/Fc activates Tiam1 through different intracellular mechanisms.
Binding of lipids to the PH domains of Tiam1 is also a mechanism regulating its activity. Interestingly, it has been reported that PI3K products enhanced Tiam1 activity, while constitutively active PI3K increased the amount of active Rac1 . Given that active EphA2 recruits the p85 subunit of PI3K, increases PIP3 levels and induces Rac1 activation [19, 66], ephrinA1/Fc stimulation could activate Tiam1 through an indirect PI3K-dependent mechanism, as it has been described for Vav activation . However, the exact mechanism dependent on Tiam-EphA2 association and leading to direct Tiam1 activation remain to be elucidated. The current literature suggests that the modes of regulation of Tiam1 could involve both intra- and inter-molecular interactions. In general, however, the precise biochemical mechanisms of Tiam1 activation remain poorly understood.
Role of Tiam1 in ephrinA1-dependent Rac1 activation and endocytosis
Although it has been shown that ephrin-dependent endocytosis requires Rac1 activation , the precise mechanisms remain to be defined. Using RNA interference we demonstrated that Tiam1 is required for ephrinA1-induced Rac1 and probably also Cdc42 activation. In the case of RhoA, ephrinA1 stimulation decreases its activation which is not significantly affected by silencing Tiam1. Those data suggest that ligand stimulated EphA receptor activates Tiam1, which in turn leads to Rac1 activation. Although Tiam1 can act directly as an exchange factor, it can also interact with different membrane and intracellular proteins, which control its GEF activity and determine the outcome signaling of Tiam1-mediated Rac1 activation. This may reconcile the different findings demonstrating that ephrins can increase or decrease GTPases activity depending on the recruited GEFs [19, 67-69].
Other GEFs (including Vav2, ephexin, Kalirin) have been reported to mediate ephrinA1-dependent Rac1 activation  and cellular functions. Why are several Rac1-specific GEFs required to regulate Eph-dependent biological responses? One explanation could be that different GEFs may regulate specific biological functions [7, 19]. In our model, Tiam1 mediated ephrin-induced endocytosis in cancer cells. Given the large numbers of Rho GEFs and Rho GTPAses, the ability of many GEFs to activate several Rho GTPases and the potential of RTKs to interact with multiple GEFs, we cannot conclude that Tiam1 is the lonely scaffold protein responsible for ephrin-induced Eph endocytosis. However, we can affirm that Tiam1 is associated with the Eph endocytosis process.
Mechanisms leading to Tiam1-mediated EphA2 endocytosis
It is well established that actin polymerization and rearrangement depend on different Rho GTPase including Cdc42, Rho and Rac [71, 72], that actin plays a crucial role in endocytosis  and Rac1 is involved in endocytosis mediated by ephrinA2 during growth cone collapse . Thus, it is clear that Rac1 regulates endocytosis through actin modulation. Here, we demonstrated that Tiam1 is activated in endosomes promoting Rac1 activation and endocytosis. However, an important question arises as to the localization of such activation. On the basis of the previously published studies, two possible mechanisms could be envisaged. Palamidessi et al. proposed that Rac1 and Tiam1 are simultaneously recruited to early endosomes by an unknown mechanism. From the endosomal membrane, Tiam1 activates Rac1 and thereafter Rac1 is targeted to the plasma membrane in an Arf6-dependent manner to ensure actin dynamics and thus promote cell migration . Similarly, following activation by EphA2/Tiam1 complex Rac1 could be recycled to specialized areas of the plasma membrane where it could promote actin reorganization and thus endocytosis. Another more probable mechanism is that Tiam1 activates Rac1 directly at the plasma membrane. This hypothesis is supported by a study showing that Tiam1 interacts with the Arp2/3 complex at the site of actin polymerization such as membrane ruffles and epithelial cell–cell contacts . In addition, it has been reported that Tiam1 is mainly cytoplasmic and can translocate to the plasma membrane upon cellular activation . This suggest that once activated in endosomes, Tiam1 could be recycled or translocated to the plasma membrane, where it can activate Rac1, thus leading to subsequent actin polymerization and Eph endocytosis.
In conclusion, our results shed new light on two important aspects of EphA2 receptors' fate following ligand stimulation. First, they definitively establish that EphA2 receptors can actively signal while positioned in the endosomal compartment. Second, they highlight the importance of EphA2-Tiam1 interactions in this process. Tiam1 may convert a repulsive response to an adhesive interaction and vice versa, modulating cellular responses. Being involved in Eph-dependent neurite outgrowth, spine development and endocytosis, Tiam1 is an important regulator of Eph-mediated cellular functions, suggesting that these two proteins could interplay in cellular migration, invasion and adhesion, all processes where they have been shown to play a role by their own. In addition, because endocytosis, trafficking, recycling and degradation of Eph receptors are key regulators of the duration, intensity and specificity of the signaling response, Tiam1 could be one of the main actors responsible for the oncogenic properties of EphA2 receptors.
Materials and Methods
Antibodies and reagents
Rabbit polyclonal antibodies against EphA2, ephrinA1, c-Cbl, ubiquitin, caveolin-1, c-myc (A14), Tiam-1 (C-16), monoclonal β-Actin, c-myc (9E10) and anti-phosphotyrosine antibody PY20 were from Santa Cruz Biotechnology. Monoclonal anti-Ras (K-, H-, N-), clone 9A11.2, polyclonal anti-Tiam1 and anti-phosphotyrosine HRP (clone 4G10) were from Millipore. Monoclonal anti-EphA2 (clone D7) was from Sigma-Aldrich and anti-EEA1 (Early Endosomal Antigen 1) from BD Biosciences. Rabbit polyclonal antibodies against phospho-EphA2 (Y594), Rab4 and Rab11 were from Cell Signaling Technology. Polyclonal anti-Tiam1 phospho Y384 was from Abcam and monoclonal anti-HA from Roche Diagnostics. HRP-goat anti-mouse or anti-rabbit secondary antibodies, control rabbit or mouse IgG were from Santa Cruz. For immunofluorescence, Alexa Fluor-conjugated antibodies and Lysotracker® were from Invitrogen AG.
The soluble recombinant mouse ephrinA1/Fc and the biotinylated ephrinA1/Fc were from R&D systems.
The lysosomal inhibitor, bafilomycin A1 and the proteasome inhibitor MG132 were from Calbiochem. The proteasome inhibitor bortezomib (Velcade®) was from Janssen-Cilag. The inhibitor of vesicular transport, brefeldin A (BFA) was from Sigma-Aldrich.
ABTS (2,2′-Azinobis [3-ethylbenzothiazoline-6-sulfonic acid]-diammonium salt), EZ-Link Sulfo-NHS-SS-Biotin, EZ-Link Sulfo-NHS-LC-Biotin and the active Rac1 Pull-down kit were purchased from Thermo Scientific. The active Rac-GEF assay kit was from Cell Biolabs.
MiaPaCa-2 cells, a human pancreatic tumor cell line, and PC-3 cells, a human prostate cancer cell line were grown in Dulbecco's Modified Eagle's Medium (DMEM) and Roswell Park Memorial Institute medium (RPMI-1640), respectively, supplemented with 10% fetal calf serum (FCS), 100 units/mL penicillin and 100 µg/mL streptomycin. Cells were maintained at 37°C in humidified atmosphere containing 5% C02.
RNA interference and transient transfections
The Ambion® Silencer® Tiam siRNA and control siRNA (Life Technologies) were used at a final concentration of 15 nm and reverse transfected using Lullaby® siRNA transfection reagent (Oz Biosciences) for 48 h.
Tiam-HA plasmid was a kind gift of A. Malliri (Paterson Institute for Cancer Research, Manchester, UK). Transient transfection of EphA2-c-myc mutants  and Tiam-HA was performed using 2–4 µg of plasmids and Lipofectamine 2000 (Invitrogen). Cells were harvested 24 h after transfection.
Immunoprecipitation and immunoblotting
Cells were serum-starved for 24 h, then stimulated with 1 µg/mL of ephrinA1/Fc or control IgG protein and incubated at 37°C for the indicated times. In some experiments, prior to stimulation, cells were preincubated 1 h with inhibitors. Cells were lysed in RIPA buffer [50 mm Tris–HCl pH 7.2, 120 mm NaCl, 1 mm EDTA pH 8, 6 mm EGTA pH 8.5, 1% Nonidet P-40, 20 mm NaF, 10 mm sodium pyrophosphate, 0,1% SDS, 2 mm sodium orthovanadate and protease inhibitor cocktail (Roche Diagnostics)]. For co-immunoprecipitations, lysates were incubated overnight with the appropriate antibody and 1 h with protein A/G Plus agarose (Santa Cruz). The resulting proteins were resolved on SDS-PAGE and subjected to immunoblot. To confirm equal loading, membranes were stripped and reprobed with the appropriate antibody.
Immunofluorescence and confocal microscopy
Cells were grown on glass coverslips overnight. When indicated, cells were preincubated 30 min with 20 µm of BFA. After stimulation, cells were fixed for 10 min at room temperature in 4% sucrose, 2% paraformaldehyde (PFA) or 5 min at −20°C in acetone; background staining was blocked by 1 h incubation in PBS, 1% BSA, 5% normal serum, Triton X100 0.2%. Primary antibodies were diluted in PBS, 1% BSA and applied overnight at 4°C. Alexa Fluor antibodies were used to detect primary antibodies. Cells were mounted in Vectashield with DAPI (Vector Laboratories). Confocal images were recorded using a Zeiss LSM510 confocal microscope (Carl Zeiss Microscopy) and analyzed with the Imaris software.
Capture – ELISA analysis
Maxisorb 96-wells plates were coated overnight at 4°C with 1 µg/mL ephrinA1 or EphA2 antibodies and blocked in PBS containing 0,05% Tween-20 (PBS-T) with 5% FCS for 1 h. Wells were incubated with 0.05 mg/mL of biotin-labeled cell lysates for 2 h at 4°C. Unbound material was washed with PBS-T and Avidin-HRP-conjugated was added for 30 min. After extensive washing biotinylated proteins were detected by addition of ABTS substrate. The optical density was read at 405 nm, 20 min after substrate addition.
GST fusion protein pull-down
The vectors encoding GST, GST-cbl-N (encoding amino acids 1–357 of human cbl) and GST-cbl-N/G306E (G306E, amino acids 1–357 of human cbl with a G306E point mutation) were a gift from Hamid Band (ENHRI, Evanston, IL, USA).
GST fusion proteins were expressed and purified as described . Cell lysates were prepared as described above. Binding reactions were carried out with 20 µg of purified GST fusion protein and 500 µg of cells lysates for 2 h at 4°C. Beads were washed, bound proteins were eluted in SDS sample buffer and analyzed by SDS-PAGE. Membranes were blotted for anti-EphA2, equal loading was ensured by Ponceau staining.
Cell-surface biotinylation, EphA2 internalization and recycling assays
Cell-surface biotinylation and recycling assays were adapted from (, Figure 4A). Cells were preincubated at 37°C with 0.5 µm of bafilomycin or DMSO for 1 h. Cell membrane proteins were biotinylated with EZ-Link Sulfo-NHS-SS-Biotin (Pierce) according to manufacturer's instructions; the percentage of surface receptor biotinylation is close to 100%. Control samples (total biotinylation and stripping control) were incubated at 4°C while other samples were incubated at 37°C with ephrinA1/Fc for the indicated times. Cell-surface biotin was stripped at 4°C using a reducing solution [50 mm glutathione (GSH), 100 mm NaCl, 1 mm EDTA, 75 mm NaOH, 1% (w/v) BSA, pH 8.7] and the reaction was quenched in 120 mm iodoacetamide. To demonstrate receptor recycling, cells were subjected to an additional incubation at 37°C for different times prior to reducing washes. Remaining biotinylated endocytosed receptors were quantified by ELISA. The percentages of internalized or recycled receptors were calculated as followed: internalized EphA2 = (EphA2 level after first incubation at 37°C) − (EphA2 level after stripping step without incubation at 37°C)/(total surface EphA2); recycled EphA2 = 100 − [(EphA2 level after re-incubation at 37°C) − (EphA2 levels after stripping step without incubation at 37°C)/(EphA2 level after first incubation at 37°C) − (EphA2 levels after stripping step without incubation at 37°C) × 100]. For western blot, biotinylated proteins were isolated with streptavidin agarose beads (Pierce), washed and released by heating 3 min at 95°C in Laemmli buffer.
Separation of internal from surface EphA2
Stimulated cells were biotinylated 1 h on ice with 0.35 mg/mL of EZ-Link Sulfo-NHS-LC-Biotin, washed and lysed in RIPA buffer. Biotinylated receptors were pulled-down with streptavidin agarose; proteins from the supernatant (internalized proteins) and from the agarose beads (biotinylated cell-surface proteins) were precipitated with anti-EphA2 or anti-Tiam1, isolated using magnetic beads (Millipore) or protein A/G beads and subjected to immunoblot (Figure 7A).
pH sensitivity of ephrinA1 binding
Cells were incubated with 0.5 µg/mL of ephrinA1/Fc biotinylated for 30 min on ice. Ligand was removed by washes with ice-cold buffers of varying pH (50 mm acetic acid, 50 mm Nacl, pH 2.5–6). After acid wash, the percentage of ligands dissociated from the receptors was determined by ELISA.
Active Rac-GEF and Rac1 pull-down assay
Pull-down assay using nucleotide-free Rac1 G15A agarose beads which selectively isolate active form of Tiam1, pull-down assay using Pak1-PBD agarose beads which binds to active state of Rac1 or Cdc42 and pull-down assay using Rhotekin-RBD were performed following the manufacturer's instructions (Cell Biolabs).
The levels of Rac-GTP, RhoA-GTP and Cdc42-GTP were measured by the band intensity quantification using ImageJ software and normalized to the amount of total Rac1, total RhoA and total Cdc42, respectively, present in cell lysates.
All values are presented as mean values ± standard deviation. Student's t-test using GraphPad software was performed for statistical analyses. p-Values are indicated in the figure legends and differences were considered statistically significant when p < 0.05. Results are representative of at least three independent experiments.
We would like to thank H. Band (ENHRI, Evanston, IL, USA) and A. Malliri (Paterson Institute for Cancer Research, Manchester, UK) for providing us the mentioned plasmids. We also thank B. M. Frey and F. J. Frey (Department of Nephrology and Hypertension, Inselspital, University of Bern, Switzerland) for their support. This work was supported by grant 3100A0-118369 of the Swiss National Science Foundation to U. H. D. This work was also supported by a VA Merit Award and NIH RO1 grant CA95004 to J. C.