Epidermal Growth Factor Receptor Internalization through Clathrin-Coated Pits Requires Cbl RING Finger and Proline-Rich Domains But Not Receptor Polyubiquitylation

Authors

  • Xuejun Jiang,

    1. Department of Pharmacology, University of Colorado Health Sciences Center, Denver, Colorado 80111, USA
    Search for more papers by this author
    • Present address: Center for Molecular Immunology, Institute of Microbiology, Chinese Academy of Science, Beijing 10080, PR china.

  • Alexander Sorkin

    Corresponding author
    1. Department of Pharmacology, University of Colorado Health Sciences Center, Denver, Colorado 80111, USA
    Search for more papers by this author

Corresponding author: Alexander Sorkinalexander.sorkin@uchsc.edu

Abstract

Cbl proteins have been implicated in the regulation of endocytic trafficking of epidermal growth factor receptor. However, the precise role of Cbl in epidermal growth factor receptor endocytosis is not defined. To directly visualize Cbl in cells and perform structure-function analysis of Cbl's role in epidermal growth factor receptor internalization, a yellow fluorescent protein-fusion of c-Cbl was constructed. Upon epidermal growth factor receptor activation, Cbl-yellow fluorescent protein moved with epidermal growth factor receptor to clathrin-coated pits and endosomes. Localization of Cbl-yellow fluorescent protein to these endocytic organelles was dependent on a proline-rich domain of c-Cbl that interacts with Grb2 as shown by fluorescence resonance energy transfer microscopy. In contrast, direct binding of Cbl to phosphotyrosine 1045 of the epidermal growth factor receptor was required for epidermal growth factor receptor polyubiquitination, but was not essential for Cbl-yellow fluorescent protein localization in epidermal growth factor receptor-containing compartments. These data suggest that the binding of Cbl to epidermal growth factor receptor through Grb2 is necessary and sufficient for Cbl function during clathrin-mediated endocytosis. Overexpression of c-Cbl mutants that are capable of Grb2 binding but defective in linker/RING finger domain function severely inhibited epidermal growth factor receptor internalization. The same dominant-negative mutants of Cbl did not block epidermal growth factor receptor recruitment into coated pits but retained receptors in coated pits, thus preventing receptor endocytosis and transport to endosomes. These data suggest that the linker and RING finger domain of Cbl may function during late steps of coated vesicle formation. We propose that the RING domain of Cbl facilitates endocytosis either by epidermal growth factor receptor monoubiquitylation or by ubiquitylation of proteins associated with the receptor.

Binding of epidermal growth factor (EGF) to the receptors (EGFR) at the cell surface dramatically enhances endocytosis of ligand–receptor complexes via clathrin-coated pits (1). The elevated internalization in combination with the effective sorting of internalized EGFRs to the lysosomal degradation pathway results in down-regulation of these receptors. By regulating receptor levels in the cell, endocytic trafficking controls the intensity and duration of EGF signaling (2). However, the molecular mechanisms of clathrin-dependent endocytosis of EGFR are not well understood. A number of models of ligand-induced internalization through coated pits and lysosomal targeting of EGFR have been proposed, and several proteins have been implicated in this process. The molecular mechanisms in the proposed models are often difficult to reconcile with each other.

Cbl proteins are the phosphorylation substrates of tyrosine kinases and represent an example of extensively studied molecules regulating endocytosis of growth factor receptors (3). Cbl proteins are thought to control the lysosomal targeting and degradation of the EGFR (4,5) and, more recently, Cbl was also implicated in the regulation of EGFR internalization (6–8). Mammalian cells contain three Cbl proteins: c-Cbl, Cbl-b and Cbl-3 (3). All isoforms bear the amino-terminal tyrosine kinase binding (TKB) domain that contains a unique Src homology (SH) 2 domain (9). The TKB domain is connected by a linker to the RING finger and extensive proline-rich domains. The RING finger domain confers the E3 ligase activity of Cbl as it is capable of recruiting E2 ubiquitylation enzymes, thus allowing ubiquitylation of the associated receptor tyrosine kinase (10). The proline-rich region is involved in interactions with proteins containing SH3 domains, such as Grb2 and Nck adaptors (11–14). The carboxyl-terminal part is present in c-Cbl and Cbl-b, but absent in Cbl-3 as well as in the worm Cbl ortholog (3). The carboxyl-terminus of c-Cbl and Cbl-b becomes tyrosine phosphorylated after stimulation of various receptor tyrosine kinases, cytokine receptors, and integrins on at least three tyrosine residues.

The role of Cbl in clathrin-mediated endocytosis of EGFR is supported by several lines of evidence. First, c-Cbl can be detected in coated pits after EGFR activation (15). Second, in some cells, overexpression of c-Cbl or Cbl-b can increase the uptake rates of EGF, and several c-Cbl mutants can slow down the endocytosis when overexpressed (6,16). Such effects, however, were not observed in other studies (4,5). A possible explanation for this discrepancy is that the high concentrations of labeled EGF used in the above-cited studies to measure internalization rates favor measurements of the clathrin-independent component of EGF uptake. In addition, many studies utilized cells with transiently expressed EGFR, which also results in saturation of the clathrin pathway in cells overexpressing receptors. Furthermore, mutation of a docking site for c-Cbl in EGFR, tyrosine 1045, did not result in significant inhibition of EGFR internalization (7). In this case, it was proposed that in the absence of direct interaction, Cbl binds to EGFR through Grb2, and thus may function in the endocytic process.

If Cbl is involved in coated-pit internalization, what might be its role? At least two models have been proposed. According to one model, Cbl promotes ubiquitylation of the EGFR and this allows for receptor internalization (7). How exactly ubiquitylation targets receptors to coated pits is unknown. Another model suggests that the interaction of Cbl carboxyl-terminus with the SH3-containing protein CIN85 links the EGFR/Cbl complex to coated pits (8).

In this study we utilized live-cell microscopy, Cbl fusion proteins and various Cbl mutants to examine which Cbl domains are required for its role in endocytosis. The experiments showed that the interaction of Cbl with Grb2 together with the RING finger domain of Cbl but not the direct interaction of Cbl with EGFR or receptor polyubiquitylation are essential for Cbl function during EGFR internalization through clathrin-coated pits.

Results

Cbl-YFP recruitment into clathrin-coated pits requires proline-rich but not TKB domain

To directly visualize c-Cbl during EGFR activation and endocytosis, fusion proteins of yellow fluorescent protein (YFP) with wild-type c-Cbl and various truncated and point mutants of c-Cbl were generated (Figure 1). Two cell lines, PAE and HeLa, were employed in the structure-function analysis of Cbl in EGFR endocytosis. PAE cells have been our preferred system for stable EGFR expression and fluorescence imaging analysis. These cells do not have endogenous EGFRs. The transfected EGFRs are internalized, down-regulated and degraded in PAE cells with the same efficiency as in many other cell types (17). Unlike HeLa cells, the PAE cells are flattened and have very low background autofluorescence. However, transient transfection of PAE cells does not allow high levels of protein expression in a large fraction of cell population. Therefore, HeLa cells, that express endogenous EGFR and allow high efficiency of the transient transfections, were primarily used in biochemical experiments with radiolabeled ligands.

Figure 1.

Figure 1.

Schematic representation and immunodetection of wild-type and truncated c-Cbl fusion proteins.A, Depicted are wild-type and mutant c-Cbl proteins fused to the amino-terminus of YFP. Tyrosine kinase binding (TKB) domain (including SH2-like domain) is connected to the RING finger domain by a linker (L). Major tyrosine phosphorylation sites in the carboxyl-terminus of c-Cbl are indicated (Y). B, To confirm the correct size of Cbl-YFP truncation mutants, HeLa cells expressing c-Cbl, 70Z-Cbl, C′480 and C′655 fusions were lysed, and Cbl fusion proteins were detected in lysates by Western blotting with anti-Cbl antibody.

To investigate whether Cbl-YFP is recruited into clathrin-coated pits in EGF-activated cells, Cbl-YFP was coexpressed in PAE/EGFR cells with β2-adaptin fused to CFP (β2-CFP), a marker of coated pits described in our previous studies (18). The cells were stimulated with low, physiological concentrations of EGF-rhodamine conjugate (EGF-Rh) at 4 °C for 1 h, conditions allowing coated pit recruitment but restricting endocytosis, and the localization of EGF-Rh, Cbl-YFP and β2-CFP was compared. Figure 2(A) shows significant colocalization of fluorescent dots containing all three proteins, indicating that Cbl-YFP is effectively targeted together with EGFR to coated pits.

Figure 2.

Figure 2.

Localization of Cbl-YFP in clathrin-coated pits. A and B, PAE cells expressing wild-type (wt) or Y1045F mutant EGFR were transiently transfected with β2-CFP and one of either wild-type Cbl-YFP (A), Cbl-G306E-YFP (A), Cbl-C′655-YFP (B) or Cbl-C′480-YFP (B). Two days after transfection, the cells were incubated with 2 ng/ml EGF-Rh at 4 °C for 1 h, fixed and the images were acquired through Cy3 (red), YFP (green) and CFP (blue) filter channels. Insets, higher magnification images of the cell regions shown by white rectangles. The ‘white’ indicates the overlap of rhodamine, YFP and CFP fluorescence. Bar, 10 μm. C, PAE/EGFR cells were transiently transfected with wild-type Cbl-YFP, Cbl-G306E-YFP, Cbl-C′655-YFP or Cbl-C′480-YFP and treated with EGF as in A. The cells were then lysed and subjected to immunoprecipitation (IP) with polyclonal anti-GFP. Immunoprecipitates were then probed by Western blotting (WB) with monoclonal anti-GFP and polyclonal anti-Grb2. The data are representative of three independent experiments.

To test whether the direct interaction of the tyrosine-kinase binding (TKB) domain of c-Cbl with the docking site on EGFR, Tyr1045, is necessary for c-Cbl recruitment into coated pits, two approaches were employed. Firstly, the recruitment into coated pits of a Cbl mutant, in which the SH2-like domain was inactivated by G306E mutation (Cbl-G306E-YFP), was examined. This mutant was effectively targeted to coated pits with EGF-Rh (Figure 2A). Secondly, β2-CFP and Cbl-YFP were coexpressed in PAE cells stably expressing Y1045F EGFR mutant. As demonstrated in Figure 2(A), Cbl-YFP was well colocalized with β2-CFP in these cells. Therefore, TKB/Tyr1045 interaction is not necessary for c-Cbl recruitment into coated pits.

To determine which region of c-Cbl molecule is critical for coated pit recruitment, Cbl carboxyl-terminal truncation mutants were prepared (Figure 1) and transiently coexpressed with β2–CFP. As shown in Figure 2(B), Cbl-C′655-YFP mutant could be readily detected in coated pits containing EGF-Rh, although a considerable pool of this mutant was located in the cytosol (Figure 2B). In contrast, a mutant lacking the proline-rich domain, Cbl-C′480-YFP, was restricted to the cytosol and not detected in coated pits containing EGF-Rh (Figure 2B). Similarly to Cbl-C′480-YFP, shorter c-Cbl mutants, such as Cbl-C′420 and v-Cbl, as well as the carboxyl-terminal fragment corresponding to amino acid residues 655–906 of c-Cbl, were all incapable of efficient translocation into coated pits but did not prevent EGFR recruitment into coated pits (data not shown). These data showed that the proline-rich region between 480 and 655 residues is critical for Cbl localization in clathrin-coated pits.

Cbl–Grb2 interactions in cell lysates and in living cells

Previous studies used in vitro binding and co-immunoprecipitation assays with ectopically expressed proteins to demonstrate the importance of the proline-rich region of Cbl (residues 480–655) for binding to the SH3 domains of several proteins (11–14). Hence, we showed in Figure 2 that this region of the Cbl molecule is important for Cbl localization in coated pits. One of the major Cbl interactors, Grb2 is located in coated pits together with EGFR and important for EGFR internalization (19). Therefore, we tested whether the ability of Cbl mutants to localize in coated pits correlates with their ability to bind Grb2. Co-immunoprecipitation experiments revealed that wild-type Cbl-YFP, Cbl-G306E-YFP and Cbl-C′655-YFP mutants are associated with endogenous Grb2 in cell lysates (Figure 2C). In contrast, no co-immunoprecipitation of Grb2 with Cbl-C′480-YFP mutant was detected (Figure 2C).

The role of the Cbl sequences encompassed by residues 480–655 in the interaction with Grb2 in endocytic compartments was further confirmed using fluorescent resonance energy transfer (FRET) microscopy in living cells. To this end, Grb2-CFP was coexpressed with Cbl-YFP in PAE/EGFR cells. Before EGF treatment Grb2-CFP was seen in the cytosol and nucleus, whereas Cbl-YFP was restricted to cytosol (Figure 3A). After EGF-Rh treatment at 37 °C, both proteins were seen clustered at the cell surface and then internalized into endosomes containing EGF-Rh (Figure 3B). Because of the transiency of receptor localization in coated pits and a small number of receptor complexes per coated pit, FRET signals were measured at the postendocytic stage of trafficking when significant amounts of Grb2-CFP and Cbl-YFP were accumulated in endosomes (Figure 3B). FRETC values measured on a pixel-by-pixel basis were highest in large perinuclear endosomes, where the largest concentration of both proteins was observed. Regardless of the expression levels, about 14% of total cellular Cbl-YFP was associated with endosomes (Figure 3C). FRETC signals were also detected in unstimulated cells coexpressing CFP-Grb2 and Cbl-YFP (Figure 3A), indicative of constitutive interaction of two proteins.

Figure 3.

Figure 3.

Interaction of Grb2-CFP and Cbl-YFP in endosomes. Grb2-CFP was coexpressed with wt Cbl-YFP, Cbl-G306E-YFP, Cbl-C′655-YFP or Cbl-C′480-YFP) in PAE/EGFR cells. YFP, CFP, and FRET images were acquired before (A) and after (B) incubation of cells with 100 ng/ml EGF-Rh for 15 min at 37 °C from living cells at room temperature. EGF-Rh was detected using Cy3 filter channel. FRETC is presented as pseudocolor YFP-intensity-modulated images (FRETC/YFP). A. l.u.f.i, arbitrary linear units of fluorescence intensity. Bars, 10 μm. Total amount of wild-type Cbl-YFP, Cbl-G306E-YFP, Cbl-C′655-YFP and Cbl-C′480-YFP per cell was 15.3 × 107, 16.1 × 107, 8.6 × 106 and 8.6 × 106 a.l.u.f.i., respectively. C, The amount of wild-type and mutant Cbl-YFP (open bars) or Grb2-CFP (filled bars) in endosomes containing EGF-Rh was determined in 4–5 images similar to those presented in B as described in Materials and Methods and expressed as per cent of the total cellular YFP or CFP. The error bars represent standard deviations.

Figure 3.

Figure 3.

Interaction of Grb2-CFP and Cbl-YFP in endosomes. Grb2-CFP was coexpressed with wt Cbl-YFP, Cbl-G306E-YFP, Cbl-C′655-YFP or Cbl-C′480-YFP) in PAE/EGFR cells. YFP, CFP, and FRET images were acquired before (A) and after (B) incubation of cells with 100 ng/ml EGF-Rh for 15 min at 37 °C from living cells at room temperature. EGF-Rh was detected using Cy3 filter channel. FRETC is presented as pseudocolor YFP-intensity-modulated images (FRETC/YFP). A. l.u.f.i, arbitrary linear units of fluorescence intensity. Bars, 10 μm. Total amount of wild-type Cbl-YFP, Cbl-G306E-YFP, Cbl-C′655-YFP and Cbl-C′480-YFP per cell was 15.3 × 107, 16.1 × 107, 8.6 × 106 and 8.6 × 106 a.l.u.f.i., respectively. C, The amount of wild-type and mutant Cbl-YFP (open bars) or Grb2-CFP (filled bars) in endosomes containing EGF-Rh was determined in 4–5 images similar to those presented in B as described in Materials and Methods and expressed as per cent of the total cellular YFP or CFP. The error bars represent standard deviations.

FRET analysis of Grb2–CFP interaction with Cbl-G306E-YFP mutant revealed that the Cbl SH2-like domain does not play a significant role in Cbl–Grb2 interaction and Cbl recruitment to endosomes (Figure 3A,B). Furthermore, the bulk of Cbl-C′480-YFP mutant remained in cytosol in the presence of activated EGFR and Grb2 in endosomes, with only a small amount of Cbl-C′480-YFP recruited to endosomes (less than 2%), presumably through the direct interaction with pTyr1045 (Figure 3B). In contrast, Cbl-C′655-YFP mutant was efficiently bound to Grb2-CFP and targeted to endosomes in EGF-stimulated cells. Thus, the region of Cbl responsible for coated-pit localization [480–655] is also important for the interaction with Grb2 and for efficient docking of Cbl-YFP to endosomes. In contrast, TKB–Tyr1045 interaction is dispensable for Cbl association with EGFR and Cbl localization in endosomes containing EGFR. The recruitment of Grb2-CFP to endosomes was independent of the presence of Cbl mutants (Figure 3C) and was not affected by the Y1045F mutation in the EGFR (data not shown). This is consistent with the equal efficiency of Grb2 coimmunoprecipitation with the wild-type EGFR and Y1045F mutant (7).

TKB domain-mediated EGFR polyubiquitylation is not essential for EGFR internalization

The experiments presented in Figures 2 and 3 demonstrated that the direct interaction of Cbl with EGFR is not necessary for targeting of EGFR and c-Cbl to coated pits and endosomes in PAE cells. However, it is well established that this interaction is critical for Cbl-mediated poly/multiubiquitylation of EGFR (5). Because ubiquitylation has been implicated in internalization of several cargoes (20), EGFR ubiquitylation was examined under conditions of coated-pit recruitment assays. As shown in Figure 4(A), EGFR was ubiquitylated at 4 °C in an EGF-dependent manner, although the extent of ubiquitylation was higher at 37 °C. We have previously described several PAE cell clones of Y1045F mutant of EGFR that differ in receptor expression levels but all exhibit negligible EGFR ubiquitylation (19). Hence, the clone was chosen that has an essentially similar amount of EGFRs as PAE/EGFR cells to properly compare the kinetics of EGF endocytosis. In these cells Y1045F mutant was poorly ubiquitylated under any conditions (Figure 4A), suggesting that poly/multiubiquitylation is not necessary for receptor sorting to coated pits.

Figure 4.

Figure 4.

Ubiquitylation and internalization of Y1045F EGFR mutant.A, PAE cells stably expressing wild-type EGFR (wt) or Y1045F mutant were incubated with or without EGF (20 ng/ml) for 1 h at 4 °C or for 2 min at 37 °C (conditions of maximal ubiquitylation of EGFR), lysed, and EGFRs were precipitated using antibody 528. The immunoprecipitates were resolved by electrophoresis and then probed by Western blotting with ubiquitin antibody and anti-EGFR (Ab 2319). The very faint signals of ubiquitin immunoreactivity associated with EGF-activated Y1045F receptor could be detected on overexposed blots (data not shown). B, The internalization of 125I-EGF (1 ng/ml) by wtEGFR and Y1045F mutant in cells used in A. The ratio of internalized/surface 125I-EGF was determined as described in Materials and Methods. The amount of surface-occupied receptors during the time-course experiment was essentially similar in both cell lines. C, Degradation and recycling of 125I-EGF (5 ng/ml) by wtEGFR and Y1045F mutant in cells used in A. The per cent of degraded or recycled radioactivity relative to the total amount of radioactivity associated with cells and cell medium was determined as described in Materials and Methods.

Furthermore, the 125I-EGF internalization rate of Y1045F mutant was only slightly reduced (Figure 4B), thus confirming that Y1045F is not essential for the entire process of EGFR internalization. Comparison of 125I-EGF internalization rates in several clones of Y1045F and wild-type EGFR-expressing cells under conditions of similar receptor occupancies by 125I-EGF revealed essentially equivalent endocytic capacities of wild-type and mutant receptors. Collectively, these data suggest that EGFR Tyr1045 and poly/multiubiquitylation are not essential for EGFR internalization through clathrin-coated pits in PAE cells. In contrast, Tyr1045 was essential for efficient degradation of 125I-EGF (Figure 4C) in agreement with the data previously obtained in other cells (5).

Linker and RING finger mutants of c-Cbl inhibit EGFR internalization

Despite extensive studying, the mechanisms by which Cbl proteins regulate clathrin-mediated endocytosis of EGFR are not understood and the data are often contradictory (4,6,8,16). One of the reasons for variability in the experimental results is that the role of Cbl in endocytosis was analyzed mainly through overexpression of Cbl mutants. The internalization rates were often analyzed using high concentrations of labeled EGF over long time-courses in cells transiently overexpressing EGFRs (4,6,8,16). Based on mathematical modeling, it was earlier proposed that such conditions do not favor measurements of the rates of clathrin-mediated pathway due to saturation of this pathway by large amounts of activated EGFRs (21). To test this hypothesis experimentally, the internalization rates were measured in HeLa cells expressing dominant-negative K44A dynamin-1 mutant, a well-established inhibitor of clathrin-mediated endocytosis, under control of an inducible promoter (22). When a low concentration of 125I-EGF(1 ng/ml) was used, the internalization was significantly inhibited in the presence of K44A dynamin (Figure 5A). However, the apparent internalization rates were low and not significantly affected by K44A-dynamin when a saturating concentration of 125I-EGF(20 ng/ml) was used. These data show that large concentrations of EGF-receptor complexes are mainly endocytosed through the slow clathrin-independent pathways. The clathrin-independent endocytosis diminishes the contribution of the rapid internalization pathway through coated pits in the overall uptake of these complexes into the cell. Therefore, in experiments with dominant-interfering Cbl mutants, all measurements of EGF endocytosis were performed using low concentrations of 125I-EGF to avoid saturation of the clathrin-mediated rapid pathway of EGFR endocytosis, as described in our previous studies (18).

Figure 5.

Figure 5.

Linker region and RING finger domain are important for Cbl function in EGFR internalization.A, HeLa cells that express K44A-dynamin under control of the inducible promoter (tet-off) were grown for 3 days with (+) or without (–) tetracycline (Tet), thus restricting or allowing, respectively, the expression of mutant dynamin. The internalization rate constants ke were measured by incubating cells with 1 or 20 ng/ml 125I-EGF for 0.5–4 min at 37 °C as described in Materials and Methods. B–D, HeLa cells transiently transfected with wild-type or various mutant Cbl-YFP were incubated with 1 ng/ml 125I-EGF (B and C) or 2 μg/ml 125I-transferrin at 37 °C (D). ke values were measured and expressed as per cent of this value obtained for the vector-transfected cells (control). The representative example of internalization experiments in HeLa cells expressing control vector (mock, closed circles), wild-type c-Cbl-YFP (open circles), Cbl-70Z-YFP (open squares), ΔH398-YFP (deletion of H398, closed squares) and ΔC381 (deletion of C381, closed triangles) is presented in B. The level of Cbl expression and transfection efficiency was monitored by imaging the YFP fluorescence of confluent cells prior to internalization assays (data not shown). Mutations are made in TKB, linker, RING finger domain and carboxyl-terminal tail (CT). pTyr, mutations of tyrosine phosphorylation sites. The data represent mean values from 3–4 experiments and the error bars represent standard deviations. The differences between EGF internalization rates in mock-transfected cells and cells expressing Linker and RING mutants were statistically significant (p < 0.01). E, The rates of 125I-EGF internalization were measured in PAE/EGFR cells that were mock-transfected (control) or transiently expressed Cbl-70Z-YFP, ΔC381 and ΔH398-YFP as described in B, and expressed as per cent of the mean values obtained for mock-transfected cells.

Several truncated, deletion or site-point mutants of c-Cbl were generated as YFP-fusion proteins and expressed in HeLa cells (Figure 1A). Overexpression of wild-type c-Cbl did not alter the rate of 125I-EGF internalization, whereas 70Z-Cbl had a strong inhibitory effect (Figure 5B,C). Single-amino acid deletions within the linker, ΔY368 and ΔY371, had similar strong effects on internalization, whereas conserved substitutions Y368F and Y371F produced only moderate inhibitory effects. Deletions and point mutations within the RING finger, ΔC381, ΔH398 or C381A, all affected EGF internalization to a similar extent, comparable to the maximum effect of 70Z-Cbl. Because only a fraction of cells in the population expressed levels of Cbl mutants that are significantly high to uncouple the interactions of endogenous Cbl proteins, 40–50% reduction of the internalization rates measured for the whole cell population is indicative of a substantial inhibition. These data are consistent with the model whereby a functional RING domain is critical for EGFR internalization. Proper conformation of the linker between RING and TKB domains is also critical for Cbl function in EGFR internalization.

Inactivation of the TKB domain of c-Cbl did not yield a dominant-negative mutant (Figure 5B,C). This result is in agreement with the data demonstrating that the TKB binding site in EGFR is not necessary for EGFR internalization (Figures 2–4). Furthermore, the carboxyl-terminus truncation mutants had no dominant-negative effect (Cbl-C′655-YFP), suggesting that the interactions mediated by the carboxyl terminus play a redundant or no role in Cbl function in EGFR internalization (Figure 5C). C′480-YFP also did not affect EGF endocytosis, presumably, because this mutant does not bind to Grb2 and therefore cannot displace endogenous Cbl from Grb2. Finally, a c-Cbl mutant in which three tyrosine phosphorylation sites in the carboxyl-terminus were eliminated, had no effect on EGFR internalization (Figure 5C). As in HeLa cells, Cbl mutants inhibited EGFR endocytosis when transiently expressed in PAE cells (Figure 5E). The inhibitory effects were less pronounced in PAE cells, because the transfection efficiency and expression levels of Cbl proteins were significantly lower in PAE than in HeLa cells. Importantly, none of the dominant-negative Cbl fusion proteins inhibited endocytosis of 125I-transferrin, a cargo that is constitutively internalized through coated pits (Figure 5D).

Effects of CIN85 mutants on endocytosis

Recent studies proposed that the carboxyl-terminal tail of Cbl links receptor tyrosine kinases, such as EGFR and c-MET, to CIN85 protein, which in turn binds to endophilin, a protein found in coated pits (8,23). However, our experiments presented in Figure 5 suggested that the carboxyl-terminal domain is dispensable for Cbl function in clathrin-mediated internalization of EGFR. To directly test the role of CIN85 in clathrin-mediated endocytosis, YFP-fused CIN85 mutants were expressed in HeLa cells, and the rates of 125I-EGF (1 ng/ml) internalization were measured. Figure 6 shows that the expression of the carboxyl-terminal fragment of CIN85 (C-CIN85) did not have a significant dominant-negative effect on 125I-EGF or 125I-transferrin endocytosis, although the level of C-CIN85 expression was higher than that of 70Z-Cbl or other dominant-interfering mutants of c-Cbl in experiments presented in Figure 5. An overexpressed fragment of CIN85 containing SH3 domains (SH3-CIN85) had a moderate effect on 125I-EGF endocytosis (Figure 5). However, this effect was not specific to EGFR, because transferrin internalization was also inhibited to a similar extent by this CIN85 mutant. These experiments suggest that CIN85 does not seem to play an essential and specific role in EGFR internalization through clathrin-coated pits.

Figure 6.

Figure 6.

Effects of CIN85 on EGFR and transferrin receptor internalization. A, Schematic representation of the domain structure and mutated forms of CIN85 used in internalization experiments presented in B. Wild-type and fragments of CIN85 are fused to the carboxyl-terminus of YFP. SH3 and coiled-coil (CC) domains of CIN85 are shown. B, HeLa cells transiently transfected with wild-type or mutant YFP-CIN85 were incubated with 1 ng/ml 125I-EGF or 2 μg/ml 125I-transferrin at 37 °C. ke values were measured and expressed as per cent of the mean values obtained for vector-transfected cells. The data represent mean values from 3 experiments and the error bars represent standard deviations.

RING domain of Cbl is necessary for late stages of coated vesicle budding

The experiments described above established an important role of the Cbl RING finger domain and Grb2–Cbl interaction in clathrin-dependent endocytosis of EGFR. To investigate which stage of internalization is inhibited by 70Z-Cbl and therefore regulated by Cbl, the transition of EGF-Rh through clathrin-coated pits was analyzed using a coat-recruitment assay described in Figure 2. PAE/EGFR cells were transfected with β2-CFP and wild-type Cbl-YFP or 70Z-Cbl-YFP. When cells were incubated with EGF-Rh (2 ng/ml) for 1 h at 4 °C, both wild-type c-Cbl (Figure 2A) and 70Z-Cbl (Figure 7A) were well colocalized with EGF-Rh in coated pits. Subsequent incubation of cells expressing wild-type Cbl-YFP for 5 min at 37 °C resulted in re-distribution of EGF-Rh to large vesicular compartments with the characteristic endosomal morphology. In contrast, the localization of EGF-Rh in large endosomes was minimal in 70Z-Cbl-YFP expressing cells incubated at 37 °C (Figure 7B). Instead, the bulk of labeled ligand remained in coated pits together with 70Z-Cbl-YFP. The large amounts of EGF-Rh and 70Z-Cbl remained in β2-CFP-positive structures even after 10–20-min incubations of cells at 37 °C (data not shown). These experiments suggest that Cbl proteins may function at the stage of endocytosis between coated pits and early endosomes, and therefore, late stages of vesicle formation following EGFR recruitment into coated pits.

Figure 7.

Figure 7.

Expression of 70Z-Cbl-YFP inhibits post-recruitment stages of EGFR internalization through coated pits. PAE/EGFR cells transfected with β2-CFP and Cbl-YFP (low panel) or 70Z-Cbl-YFP (upper and middle panel). 36–48 h after transfection, the cells were incubated with EGF-Rh (2 ng/ml) at 4 °C for 1 h and fixed, or further incubated at 37 °C for 5 min before fixation. Rhodamine, YFP and CFP images were acquired through Cy3, YFP and CFP filter channels. Insets, high magnification images of the small regions of the cell shown by white rectangles. The ‘white’ indicates the overlap of rhodamine, YFP and CFP fluorescence, and the ‘yellow’ means the overlap of rhodamine and YFP. Bars, 10 μm.

Discussion

In this study we performed structure-function analysis on the role of Cbl in clathrin-dependent endocytosis of EGFR. The negative regulatory role of Cbl in growth factor signaling in several types of mammalian cells and genetic-model organisms prompted extensive investigation of the function of Cbl proteins in endocytosis and down-regulation of growth factor receptors, particularly, EGFR. However, the literature describing molecular mechanisms of Cbl regulation of EGFR endocytosis is contradictory, and the physical and functional links of Cbl proteins to the endocytic machinery are not fully elucidated. That Cbl directly binds to EGFR through the TKB domain and causes poly/multiubiquitylation of the receptor is well established (5,9,10,24). This ubiquitylation is proposed to link EGFR to sorting machineries in endosomes (5). Whether EGFR ubiquitylation plays a role in the internalization step of EGFR trafficking has not been directly demonstrated. On the contrary, no correlation between receptor poly/multiubiquitylation and endocytosis in cells with inducible expression of dynamin mutant, that blocks clathrin-dependent endocytosis, was found (25). Moreover, it has been reported that receptor ubiquitylation is not sufficient for EGFR endocytosis and that Cbl proteins link EGFR to coated pits through the interaction of Cbl with CIN85/endophilin complex (8).

In order to analyze Cbl function in clathrin-mediated endocytosis, we used YFP fusion protein of c-Cbl, and experimental conditions designed to specifically follow the process of EGFR internalization through coated pits (Figures 2 and 5). As in the case of endogenous c-Cbl (15), Cbl-YFP was colocalized with activated EGFR in coated pits (Figure 2). The colocalization of Cbl with EGFR in coated pits and endosomal compartments was dependent on the proline-rich domain of Cbl responsible for Grb2 binding rather than on the interaction of Cbl TKB domain with Tyr1045 of the EGFR (Figures 2 and 3). These observations differentiate the function of Cbl in EGFR internalization from its function in endosomal sorting; the latter step does require direct pTyr1045–TKB interaction (Figure 4C) (5).

In contrast to the dispensable role of TKB domain in internalization, the RING finger domain of Cbl and a linker between TKB and RING domain were found essential for both the internalization and lysosomal targeting of EGFR in our experiments (Figures 4 and 5). In general, our data obtained by overexpression of linker/RING mutants in HeLa cells (Figure 5) were consistent with the results reported by Thien and coworkers (6). The stronger inhibitory effects of several point mutations in our experiments could be due to the use of low 125I-EGF concentrations and short-time internalization assays, conditions that are optimized for the measurement of the rates of clathrin-dependent endocytosis.

The experiments with Cbl and CIN85 mutants did not demonstrate the specific role of Cbl–CIN85 interactions in clathrin-dependent EGFR endocytosis in PAE and HeLa cells. Moreover, we did not observe localization of YFP-CIN85 in coated pits in PAE cells, although a small pool of CIN85-YFP could be seen in endosomes [data not shown; also see (26)]. A moderate effect of the CIN85 SH3-containing fragment on EGFR and transferrin internalization could be due to the disruption of interactions of SH3-binding proteins involved in the regulation of general clathrin endocytic machinery. One explanation for this disagreement with the results reported by Dikic and coworkers (8) is that these authors measured endocytic rates in cells transiently overexpressing EGFRs, used high concentrations of EGF and long time-courses of the uptake. It is likely that under these conditions, clathrin-independent endocytosis and recycling were the main determinants of the rates of 125I-EGF uptake. In fact, in the recent publication, CIN85 was implicated in the postendocytic step of EGFR trafficking (27).

Combining the results of our work with published findings, we propose the following working model of Cbl function in clathrin-dependent endocytosis of EGFR (Figure 8). Binding of Grb2 to the EGFR is necessary for receptor recruitment into clathrin-coated pits (19). Cbl proteins bind to EGFR and are recruited to coated pits mainly through the interaction of the proline-rich region with the SH3 domains of Grb2 (Figure 2). The direct binding of the Cbl TKB domain to Tyr1045 of the EGFR plays a minor role in Cbl localization and its function in the internalization step. This notion is supported by the observations of high endocytic rates of Y1045F receptor mutant by others (7) and ourselves (Figure 4), and by the absence of the inhibitory effect of c-Cbl-G306E mutant on endocytosis (Figure 5). However, for proper EGFR trafficking (maximum high rates of internalization and degradation), Cbl must interact with both Grb2 and pTyr1045, suggesting that the mechanisms controlled by these interactions might be synergistic. It is also possible that one Cbl molecule can be positioned to interact with both the SH3 domain of receptor-bound Grb2 and pTyr1045 on the same EGFR molecule. Importantly, the data obtained in experiments with EGFR and Cbl mutants revealed no correlation of EGFR internalization rates with the extent of receptor polyubiquitylation (Figure 4).

Figure 8.

Figure 8.

Hypothetical model of Grb2- and Cbl-dependent endocytosis of EGFR through clathrin-coated pits. UIM, ubiquitin-interaction motifs; R, RING finger domain of Cbl. See Discussion for detailed explanations.

If Cbl-induced EGFR polyubiquitylation is not essential for the internalization step, what is the role of RING finger domain in EGFR internalization? Three possibilities can be considered. Firstly, clathrin-coated pit endocytosis may be mediated through the monoubiquitylation of the EGFR. Although Y1045F mutant was very weakly ubiquitylated in our experiments, immunoblot detection using a ubiquitin antibody is not sufficiently sensitive to rule out receptor monoubiquitylation or minimal multiubiquitylation. It is well established that monoubiquitinin serves as an internalization signal in yeast cells (28). The second possibility is that the RING finger domain mediates mono- or polyubiquitylation of proteins associated with the EGFR or the components of coated pits containing EGFR. For instance, c-Cbl and Cbl-b themselves have been shown to be ubiquitylated in cells stimulated with EGF (29). Cbl-dependent monoubiquitylation of CIN85 has been recently demonstrated and implicated in the degradation of EGFR (27). Other ubiquitylated proteins in coated pits are Eps15, Eps15R and epsin; all are also tyrosine phosphorylated in response to EGFR activation (30,31). Finally, the RING finger domain may function in endocytosis through the interaction with proteins other than E2 ligases, independently of its function in ubiquitylation.

The coated pit localization experiments (Figure 7) suggested that overexpression of 70Z-Cbl substantially reduces the endocytosis of EGF-Rh that had been preclustered into coated pits. We therefore speculate that monoubiquitin moiety on the EGFR or poly/monoubiquitylated associated proteins may interact with ubiquitin-binding motifs of other proteins, such as Eps15/15R or epsin, located in coated pits, thus modifying the interactions and the activities of proteins, which regulate coated vesicle budding. For example, the formation of complexes that include Eps15/Eps15R may affect the function of epsin and dynamin, both capable of interaction with Eps15, which may subsequently promote coat constriction and the fission of the coated vesicle. We cannot rule out, however, that Cbl is also involved in the recruitment of receptors into coated pits because it is possible that very high levels of overexpression would be required to demonstrate the inhibition of this step by Cbl mutants. Because transferrin internalization was not affected by Cbl mutants, the Cbl-dependent pathway may be specific for EGFR and similar receptors. Our data recall the observation reported by Di Fiore and coworkers (32) that the dominant-negative Eps15 mutant inhibits endocytosis of EGFR by retaining receptors in coated pits but does not affect the endocytosis of transferrin receptor. Therefore, the results of experiments with Cbl and Eps15 mutants are consistent with the view that there is a population of ‘specialized’ coated pits that are responsible for EGFR endocytosis.

In conclusion, the targets of Cbl RING domain-dependent ubiquitylation and the molecular details of protein–protein interactions during Grb2-Cbl-mediated endocytosis remain to be elucidated. Nevertheless, these studies showed that the Grb2–Cbl pathway represents the major internalization route of the EGFR in PAE cells. Additional mechanisms that are independent of EGFR tyrosine phosphorylation and Grb2/Cbl association appear to mediate EGFR internalization in other cells.

Materials and Methods

Reagents

Pfu polymerase and QuickChange site-directed mutagenesis kit were from Strategene Cloning Systems (La Jolla, CA, USA). EGF-Rh was from Molecular Probes (Eugene, OR, USA). Mouse monoclonal anti-GFP antibody was obtained from Zymed Laboratories (San Francisco, California, USA); monoclonal antibodies 528 to EGFR from ATCC (Manassas, VA, USA); monoclonal anti-ubiquitin antibodies from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Monoclonal antibody to c-Cbl was from Transduction Laboratories (San Diego, CA, USA); rabbit polyclonal antibodies to GFP from Abcam Limited (Cambridge, UK); rabbit polyclonal anti-Grb2 antibodies from Santa Cruz Biotechnology. Polyclonal rabbit antibody to EGFR, Ab2913, was described previously (33). Iron-saturated human transferrin was purchased from Sigma and iodinated using Iodo-Beads (Pierce, Inc, Rockford, IL, USA). Mouse receptor-grade EGF was obtained from Collaborative Research Inc. (Bedford, MA, USA) and iodinated using a modified chloramine T method as described previously (34). The specific activity of 125I-transferrin and 125I-EGF was 3.0 × 105 cpm/μg and 1.5–2.0 × 105 cpm/ng, respectively.

Plasmids

Human c-Cbl cDNA was provided by Dr Y. Yarden (Weizmann Institute of Science, Israel). To generate YFP fusion proteins, full-length c-Cbl or 70Z-Cbl was amplified by PCR and ligated into pEYFP-N1 (Clontech, Palo Alto, CA, USA) using Kpn I and Bam HI restriction sites. A stop codon was created at the position corresponding to the amino acid residue 481 and 656 in the full-length c-Cbl to generate Cbl-C′480-YFP, and Cbl-C′655-YFP. Human CIN85 cDNA was kindly provided by Dr Kajigaya (National Institutes of Health). Full-length YFP-CIN85 and fragments of CIN85 fused to YFP (CIN85-SH3, amino acid residues 1–334 aa; C-CIN85, residues 334–666) were generated by cloning corresponding PCR fragments into pEYFP-C1 using EcoRI and BamHI restriction sites.

Site-directed mutagenesis of EGFR and c-Cbl was performed using a QuickChange kit according to the manufacturer's protocol. β2-CFP was prepared by subcloning from β2-YFP cDNA (18). Grb2-CFP was described previously (35). All constructs and point mutations were verified by dideoxynucleotide sequencing.

Cell cultures and transfections

Cell lines of PAE cells stably expressing wild-type (PAE/EGFR) or Y1045F mutated receptor were established using G418 standard selection procedures. Single-cell clones expressing 1.5–2 × 105 receptors/cell were used in the present study. PAE cell lines were grown in F12 medium containing 10% serum, antibiotics and glutamine, and supplemented with G418. HeLa cells were grown in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum, antibiotics and glutamine. HeLa or PAE cells were transiently transfected using Effectine (Qiagen, Valencia, CA, USA) with various fusion proteins and used for experiments 24–48 h later. HeLa cells expressing K44A mutant of dynamin 1 under control of tetracycline-dependent promoter (tet-off) (22) were kindly provided by Dr S. Schmid (Scripps Institute, La Jolla, CA). The cells were maintained with G418, puromycin and tetracycline as described (22). To allow overexpression of K44A-dynamin, the cells were thoroughly washed with DMEM and grown without tetracycline for 3 days before the internalization experiments.

Internalization of 125I-EGF and 125I-transferrin

To monitor 125I-EGF internalization at 37 °C, cells were grown in 12-well dishes and incubated with 125I-EGF (0.5–1 ng/ml) or 125I-transferrin (2 μg/ml) in binding medium (DMEM, 0.1% bovine serum albumin) for 1–6 min. Low concentrations of 125I-EGF were used to avoid saturation of the rapid internalization pathway (21). After indicated times, the medium was aspirated, the monolayers were rapidly washed 3 times with cold DMEM to remove unbound ligand and then incubated with 0.2 m acetic acid (pH 2.8) containing 0.5 m NaCl at 4 °C for 5 min. The acid wash was combined with another short rinse with the same solution to determine the amount of surface-bound 125I-EGF. Finally, the cells were lysed in 1 N NaOH to quantitate internalized radioactivity. The ratio of internalized and surface radioactivity was plotted against time, and the specific rate constant for internalization ke was calculated as linear regression coefficient. Non-specific binding was measured for each time-point in the presence of 200-fold molar excess of unlabeled EGF, and was not more than 3–10% of the total counts.

125I-EGF recycling and degradation was measured as described (36). Briefly, cells in 35-mm culture dishes were incubated with 5 ng/ml of 125I-EGF for 8 min at 37 °C and washed with cold DMEM. 125I-EGF that had not been internalized during 37 °C incubation was removed from the cell surface by a 2.5-min mild acid wash (0.2 m sodium acetate, 0.5 m NaCl, pH 4.5). At this point, cells are referred to as ‘125I-EGF-loaded cells’. Trafficking of 125I-EGF-receptor complexes in these loaded cells was then initiated by incubating the cells in fresh binding medium containing 200 ng/ml of unlabeled EGF at 37 °C for 0–120 min. The excess of unlabeled EGF in the medium and at the surface prevented re-binding and re-internalization of recycled 125I-EGF. At the end of the chase incubation, the medium was collected to measure the amount of intact and degraded 125I-EGF by precipitation with trichloracetic (TCA) and phosphotungstic acids. The cells were then subjected to the acid wash (pH 2.8) as described in internalization experiments to determine the amount of surface-bound 125I-EGF. This amount was negligible at time point ‘0’ (1–2% of total cell associated radioactivity, suggesting that the first acidic treatment (pH 4.5) efficiently removed surface 125I-EGF. Finally, cells were solubilized in 1 N NaOH to measure the amount of intracellular 125I-EGF. The amount of recycled 125I-EGF was estimated by summing the radioactivity counted on the cell surface and the TCA-precipitated radioactivity in the medium during chase incubation. The recycling and degradation rates were expressed as the per cent of, respectively, recycled and degraded 125I-EGF of the total ligand associated with cells and medium at each time point.

Immunoprecipitation and Western blotting

The cells were lysed in Triton X-100/glycerol solubilization buffer as described previously (35). In experiments where ubiquitylation of the EGFR was analyzed, 1% Na deoxycholate and 10 mm N-ethyl-maleimide were included in the lysis buffer to minimize coprecipitation of other proteins and inhibit deubiquitination enzymes, respectively. Wild-type or mutant Cbl-YFP proteins were immunoprecipitated using polyclonal anti-GFP for 5–12 h at 4 °C. EGFR were immunoprecipitated using Ab528 for 3 h at 4 °C following by a 1- h incubation with Protein A-Sepharose (Sigma). Immunoprecipitates and cell lysates were electrophoresed on SDS-PAGE, transferred to nitrocellulose membranes, and Western blotting was performed with several antibodies indicated in each experiment, as described previously (35).

Coated pit recruitment and internalization of EGFR by fluorescence microscopy

Cells transfected with β2-CFP and YFP-tagged Cbl constructs were grown on coverslips. The coverslips were incubated with 1–2 ng/ml EGF-Rh for 1 h at 4 °C. The cells were then washed and either fixed immediately with freshly prepared 4% paraformaldehyde (Electron Microscopy) for 30 min at 4 °C, or further incubated at 37 °C and then fixed. The coverslips were mounted into a microscopy chamber with phosphate buffer, and the rhodamine, YFP and CFP images were acquired through Cy3, YFP and CFP filter channels using epifluorescence imaging workstation as described (35).

FRET microscopy

For FRET experiments, PAE/EGFR cells transfected with Grb2-CFP and YFP-fused Cbl mutants were incubated with 100 ng/ml EGF-Rh for 30 min at 37 °C. The methodology of FRET measurements has been described previously (35). Briefly, images were acquired through YFP, CFP and FRET filter channels, and corrected FRET (FRETC) images were calculated using the FRET module of SlideBook 3 software (Intelligent Imaging Innovation, Denver, Colorado). Binning 2 × 2 mode was used.

To estimate the efficiency of Cbl and Grb2 recruitment to endosomes, the amount of YFP and CFP colocalized with endosomal EGF-Rh was calculated using the SlideBook ‘AND data w/mask’ algorithm as described (18). All images were background-subtracted. The images were then segmented using a minimal intensity of EGF-Rh vesicles as a low threshold, and new images were generated from the segmented images, so that all pixels that did not overlap with EGF-Rh vesicles were assigned ‘0’ fluorescence intensity. The integrated intensity of YFP and CFP fluorescence that overlapped with EGF-Rh in the segmented image was considered as YFP and CFP localized in endosomes. Finally, the extent of YFP or CFP localization in endosomes was calculated as the ratio of the integrated YFP fluorescence of the segmented image to that of the original image. The basal background of overlap was determined in cells coexpressing YFP or CFP (diffusely distributed in the cell), and treated with EGF-Rh. These measurements yielded a mean value of 4.7% and 3.3% for YFP and CFP overlap, respectively. These values were subtracted from colocalization values obtained in cells expressing Cbl-YFP and Grb2-CFP fusion proteins.

Acknowledgments

The authors thank Drs Y. Yarden for c-Cbl cDNA, S. Kajigaya for CIN85 cDNA, S. Schmid for HeLa-K44A-dynamin cells and F. Huang for preparation of β2-YFP/CFP. This work was supported by grants CA089151 from NCI and RPG-00-247-01-CSM from ACS.

Ancillary