* This work was supported in part by U. S. Army Breast Cancer Research Program Grant DAMD17-94-J-444 and National Science Foundation Biotechnology Program Grant BES-9421773.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ Recipient of a predoctoral fellowship from the U. S. Army Breast Cancer Research Program.
ErbB-2/HER2 is an important signaling partner for the epidermal growth factor receptor (EGFR). Overexpression of erbB-2 is also associated with poor prognosis in breast cancer. To investigate how erbB-2 amplification affects its interactions with the EGFR, we used a human mammary epithelial cell system in which erbB-2 expression was increased 7–20-fold by gene transfection. We found that amplification of erbB-2 caused constitutive activation of erbB-2 as well as ligand-independent activation of the EGFR. Overexpression of erbB-2 strongly inhibited erbB-2 down-regulation following transactivation by EGFR. Significantly, down-regulation of activated EGFR was also inhibited by erbB-2 amplification, resulting in enhanced ligand-dependent activation of the EGFR. The rate of EGFR endocytosis was not affected by erbB-2 overexpression, but the rate of lysosomal targeting was significantly reduced. In addition, erbB-2 overexpression promoted rapid recycling of activated EGFR back to the cell surface and decreased ligand dissociation from the EGFR. Our data suggest that overexpression of erbB-2 inhibits both its down-regulation and that of the EGFR. The net effect is increased signaling through the EGFR system.
). No ligand has been found that binds erbB-2, although both EGF and heregulin can activate erbB-2 in-trans through ligand-induced heterodimerization and subsequent tyrosine phosphorylation of erbB-2 (
). Once internalized, occupied receptors are sorted in recycling endosomes and subsequently degraded in lysosomes, effectively reducing the number of activated receptors. Overexpression of EGFR has been shown to impair their down-regulation, apparently because of limiting levels of regulatory molecules that mediate rapid endocytosis and lysosomal targeting (
Although activation of erbB-2, erbB-3, and erbB-4 has been well characterized, the negative regulation of these receptors has not been extensively studied. There are conflicting reports regarding the trafficking of erbB-2 following transactivation with EGF. We, as well as other investigators, have described efficient down-regulation and lysosomal targeting of erbB-2 in three different cell types (
). It was reported that although erbB-2 was transactivated following EGF treatment, there was no measurable change in erbB-2 half-life.
One way to reconcile these disparate results is to postulate that although lysosomal targeting is normally involved in erbB-2 down-regulation, it is impaired when erbB-2 is overexpressed. To test this hypothesis, we examined the negative regulation of erbB-2 in mammary epithelial cells that overexpress the protein due to introduction of a transgene. We found that overexpression of erbB-2 severely inhibits its own down-regulation. Intriguingly, we also found that down-regulation of the EGFR was inhibited even though EGFR levels were similar between parental and erbB-2 overexpressing cells. Consistent with the findings of other investigators, we found that overexpression of erbB-2 results in an elevated basal level of activated receptors (
). However, we not only observed constitutive activation of erbB-2, but also of EGFR. These results show that receptor cross-talk is not only important in receptor activation, but also plays a role in negative regulatory processes.
N-13 polyclonal antibody directed against the amino-terminal 13 residues of the EGF receptor was a gift from Dr. Debora Cadena. 1917 polyclonal antibody directed against the carboxyl-terminal 18 residues of erbB-2 was provided by Dr. Gordon Gill. Ab5 mouse monoclonal antibody against the extracellular domain of erbB-2 was from Oncogene Sciences. 225 mouse monoclonal antibody against the EGFR was purified from hybridomas obtained from the American Type Culture Collection. RC20 anti-phosphotyrosine/horseradish peroxidase conjugate was purchased from Transduction Laboratories. Antibodies 13A9 against the EGFR and 4D5 against human erbB-2 were gifts from Genentech. These were directly labeled with Alexa 488 and Alexa 594 dyes, respectively, following the manufacture's protocol (Molecular Probes, Inc.).
B82 mouse L cells transfected with the gene for human EGF receptor were grown in Dulbecco's modified Eagle's medium containing dialyzed 10% calf serum (HyClone) and 5 μm methotrexate. The human mammary epithelial cell lines MTSV1–7, ce1, and ce2 have been described previously (
) and were obtained from Dr. Joyce Taylor-Papadimitriou. They were grown in Dulbecco's modified Eagle's medium containing 10% calf serum (HyClone), 1 μm insulin, and 5 μmdexamethasone. Selection for erbB-2 expression was maintained using 500 μg/ml G418.
Quantification of ErbB-2, EGFR, and Phosphotyrosine Levels
), debris was removed by centrifugation and samples were brought to 2% SDS, 1% β-mercaptoethanol and heated to 100 °C for 10 min. Equal amounts of protein from each sample was separated on a 5–7.5% polyacrylamide gradient gels and transferred to nitrocellulose. EGFR and erbB-2 were detected by N-13 and 1917 polyclonal sera, respectively, using125I-labeled protein A as described (
). The concentrations and incubation times with 125I-labeled protein A were in the linear range of the protein load of the gels. The blots were analyzed by storage phosphor plates using the Bio-Rad G250 Molecular Imager. The Bio-Rad Molecular Analyst software package was used to quantify the amount of radioactivity associated with each band.
To determine the phosphotyrosine content of erbB-2 and the EGFR, cells were lysed for 10 min at 0 °C in RIPA buffer supplemented with 100 μm Na3VO4. The lysate was clarified by centrifugation for 10 min at 10,000 × g, and receptors were immunoprecipitated with 5 μl of either the 1917 or 4D5 antibodies (erbB-2) or 2 μg of 225 (EGFR) and 100 μl of protein A-Sepharose (50% slurry). The beads were washed several times in lysis buffer and boiled in SDS sample buffer prior to gel electrophoresis and transfer to nitrocellulose. Phosphotyrosine levels were measured by detection with RC20 antibody and the Renaissance ECL kit from NEN Life Science Products Inc. Blots were exposed to film, and analyzed using a Bio-Rad Imaging Densitometer and the Molecular Analyst software package.
Distribution of both EGFR and erbB-2 was evaluated using fixed and permeabilized cells as described previously (
). Cells were labeled for 1 h with a mixture of 13A9 and 4D5 antibodies directly labeled with Alexa 488 and Alexa 594 dyes, respectively, at a final concentration of 1 μg/ml. After rinsing, the cells were mounted in Prolong anti-fade medium (Molecular Probes, Inc.) and viewed with a Nikon inverted fluorescence microscope with a ×60, 1.4 N.A. oil immersion objective. Images were acquired as described below. Specificity of the labeling was verified by using control cells (B82) lacking EGFR and human erbB-2.
To follow the transfer of EGFR to lysosomes, the lysosomes were first labeled by incubating cells for 15 min at 37 °C with 5 mg/ml fluorescein-labeled dextran (Mr 10,000, anionic, lysine fixable; Molecular Probes, Inc.), then chased for an additional 2 h. During the chase period, cells were pulsed for 15 min with 1.5 × 10−8m EGF-Texas Red-streptavidin complex (Molecular Probes, Inc.) and chased for up to 105 min. Total chase time following the initial fluorescein-labeled dextran treatment was 2 h for all samples. Cells were fixed with 3.6% paraformaldehyde and mounted in Prolong. Coverslips were viewed with a Nikon inverted fluorescence microscope with a ×60, 1.4 N.A. oil immersion objective. Sets of 3 images at 3 different focal planes spaced 0.5 μm apart centered on the perinuclear endosomes were acquired at 520 and 615 nm (for fluorescein and Texas Red, respectively). The images (12 bit, 656 × 517) were acquired using a Princeton Instruments cooled CCD camera attached to a Macintosh workstation running Openlab software (Improvision, Inc). The image triplets were deconvolved using nearest-neighbor subtraction (
). The deconvolved image of the lysosomes (fluorescein) was then used to generate a binary mask using grayscale values between 700 and 4095. This mask was then applied to the deconvolved image of the EGF (Texas Red) to identify all lysosomal “objects” that contained EGF. The integrated intensity of all of these objects was then taken as the amount of EGF within lysosomal structures. A mask of the EGF image was generated in the same way and applied to the EGF image to determine the total integrated intensity of all EGF-containing objects within the cell. The fraction of all EGF colocalized in lysosomes was then calculated. At each time point, four random fields of cells were analyzed which contained between 100 and 200 vesicles per field.
Number of surface-associated erbB-2 molecules was determined by steady state analysis (
). Scatchard analysis of EGF binding to cells used125I-EGF at a specific activity of 1.6 × 106 cpm/pmol and an incubation period of 4.5 h at 0 °C using ligand concentrations from 1.7 × 10−11to 1.7 × 10−8m as described (
). Measurements were made using a ligand concentration of 10 ng/ml, and each rate constant determination was derived from a 5-min incubation period with ligand. Specific internalization rates were determined by plotting the integral surface-associated ligand against the amount internalized, and the slopes were determined by linear regression (
Cells were grown to confluence in 35-mm dishes and switched to serum-free Dulbecco's modified Eagle's medium containing 20 mm HEPES (pH 7.4) and no bicarbonate (D/H/B) 12 h before experiments. The cells were incubated at 37 °C in 0.1–30 ng/ml 125I-EGF for 3 h to allow the sorting process to reach a steady state (
), rinsed twice with phosphate-buffered saline, and returned to 37 °C in D/H/B containing 1 μg/ml unlabeled ligand to prevent rebinding and reinternalization of recycled ligand. The medium was collected at 10 min and an aliquot counted for total radioactivity. Cells were solubilized with 2% sodium dodecyl sulfate and the amount of radioactivity remaining was determined. The medium was loaded on a 15% native polyacrylamide slab gel and the intact and degraded EGF was separated by isotachophoresis (
). After drying the gel, the relative amount of radioactivity in the bands corresponding to intact and degraded EGF was quantified using a Bio-Rad G250 Molecular Imager. Cell number per plate was determined by counting parallel plates. Fraction of intact ligand was then plotted as a function of ligand in the cells at the start of the chase (lost into the medium + amount remaining) as described previously (
Cells were labeled to steady state (24 h) with cysteine and methionine-free Dulbecco's modified Eagle's medium (ICN) supplemented with 250 μCi/ml EXPRE35S35S from NEN Life Science Products Inc. which contains both radiolabeled methionine and cysteine. Cultures were rinsed 6 times with normal culture medium and chased with medium with or without 100 ng/ml EGF. At 0, 1, 3, 5, and 7 h chase, cells were lysed in RIPA buffer and equal amounts of protein were subjected to immunoprecipitation with 5 μl of 1917 antibody. Samples were then separated by SDS-gel electrophoresis and the gels were then dried on 3MM paper followed by quantitation of erbB-2 bands using a Bio-Rad Molecular Imager as described above.
Characterization of Cells Overexpressing ErbB-2
Ligand binding not only activates the EGFR but also initiates negative regulatory processes. Overexpression of the EGFR, however, has been shown to inhibit this negative regulation. Since erbB-2 acts as a signaling partner of the EGFR, we wanted to determine whether erbB-2 overexpression affected its own negative regulation, or that of the EGFR. We employed a human mammary epithelial cell line (MTSV) and two derivative lines (ce-1 and ce-2) which have been transfected with the gene for erbB-2 (
). The parent cell line, derived from human breast aspirates, was immortalized with SV40 large T antigen, but is not tumorigenic. Transfection with the erbB-2 gene alters the growth characteristics of the ce-1 and ce-2 cells in that the transfectants grow in soft agar can be propagated as tumors in nude mice, and show disorganized growth on collagen gels (
Western blot analysis and binding of radiolabeled anti-erbB-2 (4D5) at 37 °C was used to quantify the expression levels of erbB-2 in these cells. This approach was taken because equilibrium binding could not be achieved at 0 °C due to the very slow binding of 4D5. Steady state binding at 37 °C also allowed us to estimate the relative distribution of erB-2 between cell surface and intracellular pools (
). Preliminary experiments showed that treating the cells overnight with 4D5 did not appear to alter the cellular distribution of erbB-2 (data not shown).
Steady state binding of anti-erbB-2 indicated an accessible pool of 9.8 × 104 and 1.5 × 106 erbB-2 molecules per cell in MTSV and ce2 cells, respectively (Fig.1A), a 15-fold increase in the transfected cell line. Acid stripping of the bound radiolabeled 4D5 showed that surface expression of erbB-2 was 6.3 × 104 in the case of MTSV cells and 6.4 × 105 for ce2 cells. ErbB-2 expression in the ce-1 line varied between 2- and 6-fold higher than the parental cells (data not shown). Western blot analysis indicated a 23-fold increase in erbB-2 mass in the ce2 versus MTSV lines, somewhat higher than the values derived from the steady state analysis, suggesting that not all cellular erbB-2 readily exchanges with the cell surface. The affinity of the 4D5 antibody for erbB-2 was similar for both the MTSV and ce2 cells at 12 and 10 nm, respectively. The percent of 4D5 antibody found internalized at steady state was also similar at 41% (±4%) and 55% (±14%) for MTSV and ce2 cells, respectively.
The number of EGFR in these cells was determined by Scatchard analysis conducted at 0 °C to prevent receptor down-regulation. As shown in Fig. 1B, both MTSV and ce2 cells displayed similar numbers of surface EGFR (5.7 × 105 and 9.2 × 105 per cell, respectively) of predominantly a single affinity class. The affinity of these receptors, 1.4 nm, is similar to what has been described previously for fibroblasts (
). Western blots of detergent extracts of MTSV, ce1, and ce2 cells confirm that they all express a similar number of EGFR (data not shown). Analysis of the Western blots using a molecular imager indicated that relative to cell protein content, ce2 cells express approximately 10% higher EGFR levels whereas the levels of EGFR in ce1 cells was indistinguishable from the parental MTSV cells.
These data indicate that the ratio of EGFR to erbB-2 at the cell surface of MTSV cells is approximately 9:1 whereas the ratio in ce2 cells is approximately 1:1. To characterize erbB-2 and EGFR distribution in these cells, sparse cultures were fixed, permeabilized, and stained using directly labeled erbB-2 and EGFR antibodies. As shown in Fig. 2, erbB-2 was found at both the cell surface and in a collection of intracellular vesicles. The EGFR showed a very similar distribution pattern with EGFR and erbB-2 both colocalized at the cell surface and in intracellular vesicles (arrows in Fig. 2). These data suggest that overexpression of erbB-2 is not accompanied by any striking alteration in either its cellular distribution or affinity for antibodies. In addition, the number, distribution, and affinity of EGFR do not appear to be greatly altered as a result of erbB-2 overexpression.
Amplification of ErbB-2 Inhibits the Down-regulation of ErbB-2
Because the EGFR transactivates erbB-2, changing the ratio of erbB-2 to EGFR may alter EGF-mediated erbB-2 phosphorylation. Both parental MTSV cells and overexpressing ce2 cells were treated with EGF for 5 and 10 min. ErbB-2 and EGFR were then immunoprecipitated followed by Western blot analysis for phosphotyrosine (Tyr(P)), to monitor receptor activation. As shown in Fig. 3, addition of EGF increased Tyr(P) levels of both EGFR and erbB-2. In the case of the parental MTSV cells, very little receptor phosphorylation was observed in the absence of EGF addition. Surprisingly, in ce2 cells, there was a substantial amount of phosphorylation of both erbB-2 and the EGFR in the absence of EGF. Although increased basal activation of erbB-2 as a result of overexpression has been documented by other investigators (
), higher basal activation of the EGFR has not previously been reported. Addition of EGF further increased the level of erbB-2 phosphorylation approximately 3-fold in ce2 cells as compared with 9-fold in the MTSV cells (average of three experiments), indicating that the EGFR was capable of transactivating erbB-2 in both cell types.
It seemed possible that constitutive activation of both the EGFR and erbB-2 could be due to autocrine production of EGF-like ligands. We tested this idea by blocking EGFR activation using antagonistic EGFR antibodies 225 and 13A9 (
). Neither antibody affected the constitutive activation of either the EGFR or erbB-2 (Fig. 3), suggesting that overexpression of erbB-2 alone was responsible.
To determine whether overexpression of erbB-2 affects its down-regulation, we transactivated erbB-2 by treating both MTSV and ce2 cells with 100 ng/ml EGF for 24 h. As a comparison, we also examined a well characterized fibroblast cell line (
). Loss of erbB-2 was assessed by Western blot analysis. As shown in Fig.4, A and B, fibroblasts and MTSV cells showed a 80 and 72% loss of erbB-2, respectively, by 24 h following EGF treatment. However, the ce2 cell line only displayed a 30% loss in erbB-2 mass. Calculating the net amount of erbB-2 loss following 24 h EGF treatment showed that the MTSV cells lost a similar amount of erbB-2 mass relative to ce2 cells (2.0 × 104versus 2.3 × 104 arbitrary molecular imager units). Thus, erbB-2 overexpression affected the relative amount erbB-2 degraded in response to EGF, not the absolute amount.
The kinetics of receptor loss showed that erbB-2 loss in the parental MTSV cells was completed within 6 h (Fig. 4C). The level of erbB-2 in the overexpressing ce2 line was only reduced about 5% at the same time. We confirmed that the accelerated loss of erbB-2 protein was due to enhanced degradation by labeling cells with35S-amino acids and immunoprecipitating erbB-2 following EGF addition (data not shown). EGF decreased the half-life of erbB-2 in MTSV cells from 6 to 1 h, whereas the same treatment decreased erbB-2 half-life from 6 to 5 h in ce2 cells. Reverse transcription polymerase chain reaction analysis of mRNA levels indicated that EGF did not affect the erbB-2 mRNA level (data not shown). These results suggest that overexpression of erbB-2 inhibits its down-regulation by reducing the degradation rate of the transactivated receptor pool.
ErbB-2 Overexpression Increases Signaling through the EGFR
As previously noted, we observed that overexpression of erbB-2 resulted in the constitutive activation of both the EGFR as well as erbB-2 (Fig.3). The 6-fold increase in basal EGFR activation was similar to the 7-fold increased basal activity of erbB-2 (Fig. 3). Because erbB-2 acts as a signaling partner to the EGFR, it seemed possible that erbB-2 overexpression may drive heterodimer formation. Activation of the EGFR, however, is usually followed by several negative regulatory processes, such as desensitization and down-regulation (
). To explore the effect of erbB-2 overexpression on both positive and negative regulation of the EGFR, we examined the time course of EGF-induced EGFR activity.
MTSV and ce2 cells were treated with a high concentration of EGF for up to 2 h. At different time intervals, cells were solubilized and the EGFR were immunoprecipitated. Receptor levels and phosphorylation states were then determined using Western blots. As shown in Fig.5A, there was no phosphorylation of the EGFR in MTSV cells in the absence of EGF. In contrast, ce2 cells displayed a high constitutive level of EGFR phosphorylation which was not affected by the addition of an antagonistic EGFR antibody. In the case of both MTSV cells and ce2 cells, the addition of EGF caused an increase in Tyr(P) content of the EGFR. However, receptor Tyr(P) levels rapidly decreased in the case of MTSV cells, but remained elevated in the ce2 cells. A similar pattern was observed for EGFR mass (Fig. 5A). Receptor levels decreased following EGF addition for MTSV cells, but receptor levels remained relatively constant in ce2 cells.
We analyzed the pattern of EGF-induced EGFR phosphorylation in terms of both Tyr(P) to EGFR ratio and in terms of phosphorylated receptors per cell (Fig. 5, B and C, respectively). This analysis showed that the amount of Tyr(P) as a function of receptor mass was actually depressed following EGF addition in cells overexpressing erbB-2. The kinetics of EGFR phosphorylation in parental MTSV cells showed a rapid rise followed by a subsequent decline, characteristic of receptor desensitization. However, there was no sign of EGFR desensitization in the ce2 cells.
When analyzed in terms of total phosphorylated receptors per cell, the MTSV cells displayed a rapid loss of activated EGFR such that by 2 h following EGF treatment, their levels were similar to the constitutive level of EGFR activation in ce2 cells (Fig.5C). Addition of EGF to ce2 cells caused a persistently high level of activated EGFR, evidently due to the suppression of receptor loss. We repeated these experiments using cells that express only 6-fold higher levels of erbB-2 (ce1 cells) and found an intermediate result in that the constitutive level of EGFR activation and the degree of receptor loss was between that observed for ce2 and MTSV cells (data not shown).
Our results indicate that the overexpression of erbB-2 inhibits down-regulation of the EGFR. To test this idea directly, both MTSV and ce2 cells were treated with 100 ng/ml EGF and at various times the cells were solubilized and the EGFR levels were determined by Western blot analysis. As shown in Fig. 6, the loss of EGFR mass in both MTSV and ce2 cells was biphasic. The parental MTSV cells lost half their EGFR mass in less than 2 h, whereas the ce2 cells lost the same amount in about 10 h. Thus down-regulation of the EGFR is inhibited in cells that overexpress erbB-2.
ErbB-2 Overexpression Inhibits EGFR Down-regulation at Multiple Levels
Ligand-induced down-regulation of the EGFR is regulated at three distinct levels: endocytosis, endosomal sorting, and lysosomal targeting (
). This, in turn, could inhibit EGFR down-regulation. To test for this possibility, we examined the kinetics of both intracellular and cell surface accumulation of EGF. If overexpression of erbB-2 was inhibiting EGFR internalization, then we should observe an increased amount of EGF at the cell surface and a corresponding decrease in intracellular ligand. MTSV and ce2 cells were incubated with 125I-labeled EGF at 37 °C. At various times, the relative amount of ligand either inside the cell or at the cell surface was determined. As shown in Fig.7 (bottom panel), ce2 cells displayed a pronounced increase of 125I-EGF binding to the cell surface relative to MTSV cells, especially at the longer time points (>20 min). The 3–4-fold elevation in binding could not be explained by the relatively small differences in EGFR levels between MTSV and ce2 cells (see Fig. 1B). Paradoxically, we also observed an increased level of intracellular 125I-EGF in ce2 cells, but only after about 20 min incubation with ligand (Fig. 7,top panel). The similar amount of internalized ligand in both MTSV and ce2 cells at early time points is inconsistent with an inhibition of internalization. The accumulation of intracellular125I-EGF at longer incubation times suggests an inhibition of lysosomal degradation.
Since a change at a single point in the EGFR trafficking pathway could not explain the observed accumulation of both surface-associated and intracellular EGF in ce2 cells, it seemed possible that erbB-2 overexpression could cause multiple alterations in EGFR trafficking. We therefore examined the individual steps. The specific internalization rate was determined by incubating cells with radiolabeled EGF for 5 min, during which time surface-associated and internalized ligand was measured. Fig. 8A shows that overexpression of erbB-2 did not significantly alter the internalization rate constant (ke) for the EGFR (0.14 min−1 ± 0.01 versus 0.12 min−1 ± 0.03 for MTSV and ce2 cells, respectively). The kinetics of initial 125I-EGF binding to both MTSV and ce2 cells was also similar (Fig. 8B), which indicates that the forward rate constant (ka) was the same. To determine the dissociation rate constant (kd), the cells were incubated with EGF for 5, 10, and 15 min followed by a chase in a large excess of unlabeled EGF (to prevent rebinding of dissociated125I-EGF). The amount of EGF lost from the surface is a combination of ligand internalization and dissociation from the receptor. Thus, the rate of 125I-EGF loss from the cell surface is equal to kd + ke. Because ke is the same in the two cell types (Fig.7A), differences in loss will reflect kd. As shown in Fig. 8C, loss of EGF was substantially slower from the surface of ce2 cells as compared with MTSV cells (0.15 min−1versus 0.23 min−1, respectively). This did not change appreciably as a function of incubation time. Subtracting the value of kemeasured in parallel experiments (0.135 min−1;dashed line in Fig. 8C) yielded a value ofkd of 0.013 min−1 in ce2 cells and 0.10 min−1 in MTSV cells. Thus overexpression of erbB-2 appears to cause a 7-fold reduction in the EGF dissociation rate constant, which could partially explain the increased levels of EGF at the cell surface.
To assess the effect of erbB-2 overexpression on EGFR endosomal sorting, we used a previously described technique that measures the fraction of internalized receptors that are recycled (
). To measure fractional recycling, cells were brought to steady state with different concentrations of 125I-EGF. Surface-associated EGF was removed with a mild acid strip and the relative amount of intactversus degraded EGF which subsequently appeared in the medium was measured. We have previously shown that the ratio of intactversus degraded EGF indicates the fraction of internalized ligand that is recycled versus targeted to lysosomes (
). When we used this technique on MTSV and ce2 cells, we obtained the results shown in Fig. 8D. The parental MTSV cells showed an increase in fractional ligand recycling from 0.25 to 0.45 as the intracellular ligand increased from 7 × 103 to 4 × 105 molecules per cell. This “saturation” of endosomal sorting is very similar to what has previously been described in fibroblasts (
). The ce2 cells displayed a very similar fractional recycling pattern, but with a greater degree of recycling at all intracellular ligand concentrations. This suggests that overexpression of erbB-2 inhibits sorting of EGFR from endosomes to the lysosomes, and thus promotes recycling.
To directly test if EGFR transfer to the lysosomes was impaired by erbB-2 overexpression, a kinetic analysis of 125I-EGF degradation was done. Cells were incubated for 5 min with125I-EGF followed by a chase in unlabeled medium. The amount of intracellular 125I-EGF remaining at different times was then determined. Under these conditions, almost all of the125I-EGF lost from the cells is degraded ligand (
) (results not shown). As shown in Fig.9A, there was a lag of approximately 15 min in MTSV cells before significant loss of internalized ligand was observed. This lag generally corresponds to the time necessary for internalized EGF to be transferred to lysosomes (
). Thereafter, the 125I-EGF was lost with a of 32 min. In the case of ce2 cells, there was a slightly longer lag before initiation of ligand loss, after which125I-EGF was lost with a of 53 min. The ce1 cells showed an intermediate rate of ligand loss ( of 46 min, data not shown). These data indicate that erbB-2 overexpression interferes with EGFR trafficking to the lysosomes.
To confirm that differences in ligand degradation rates were due to differences in intracellular trafficking, we used immunofluorescence to follow the progression of the EGFR from the cell surface to the lysosome. The lysosomes were labeled with a pulse of fluorescently labeled dextran followed by a 2-h chase. During this chase, the cells were pulsed with Texas Red-labeled EGF to label the EGFR. The cells were fixed at different time periods and colocalization of the EGF with the lysosomes was determined using digital confocal imaging. As shown in Fig. 9B, there was little colocalization of the EGF with lysosomes following the initial 15-min pulse. However, there was progressive colocalization of the two fluorescent labels during the 2-h chase period. Colocalization reached its greatest extent in both MTSV and ce2 cells at 60–80 min (Fig. 9B). The rate at which EGF was transferred to lysosomes was somewhat slower in the ce2 cells as was the extent of transfer (29 versus 35% for ce2 and MTSV cells, respectively). Colocalization of EGF in lysosomes in both cell types declined after 80 min and never involved the majority of the ligand, probably because our protocol only labels a fraction of total lysosomes. The apparent decrease in colocalization is most likely due to EGF being transferred to newly formed lysosomes lacking fluorescein-dextran combined with degradation of previously transferred EGF. Nevertheless, these data confirm that transfer of EGF-containing endosomes to lysosomal structures is slower in ce2 cells relative to MTSV cells. Thus overexpression of erbB-2 results in an inhibition of lysosomal trafficking of EGFR.
We previously reported that erbB-2 is down-regulated following transactivation with EGF (
). However, SKBR-3 cells express very high erbB-2 levels which could interfere with down-regulation. To test this hypothesis, we used a mammary epithelial cell line that expresses similar levels of erbB-2 as SKBR-3 cells due to the introduction of a transgene (2 × 106 molecules/cell) (
). We found that although EGF induces erbB-2 degradation in the parental MTSV cells, it had little effect on erbB-2 levels in the overexpressing ce2 cells. Thus overexpression of erbB-2 inhibits erbB-2 down-regulation in a similar fashion as EGFR overexpression inhibits EGFR down-regulation (
), but it is unclear whether this is the case with erbB-2. Although expression levels of erbB-2 in ce2 cells are high relative to the parental cell line, they are approximately equivalent to the endogenous levels of EGFR (approximately 6 × 105 at the cell surface for both). Because EGF addition can reduce total EGFR levels in the parental cells by >90%, it is unlikely that saturation of some common component of the endocytic pathway (such as coated pits) is responsible for the lack of erbB-2 down-regulation. EGFR-mediated transactivation also does not appear limiting for erbB-2 down-regulation because the total number of transactivated erbB-2 molecules in the overexpressing cells is much greater than the parental lines. Because erbB-2 overexpression inhibits the fraction of receptors undergoing down-regulation, but not the absolute number, it appears likely that a step downstream of transactivation is rate-limiting for erbB-2 degradation.
It is currently unclear whether transactivation causes erbB-2 internalization. Several groups have analyzed erbB-2 endocytosis by using labeled antibodies or EGFR extracellular domain chimeras, and have come to different conclusions regarding whether erbB-2 undergoes activation-induced internalization (
). We have measured the specific internalization rate of erbB-2 using a radiolabeled antibody and have not observed any effect of EGF addition (0.043 min−1 and 0.045 min−1 with or without EGF, respectively).
R. Worthylake, L, K. Opresko, and H. S. Wiley, unpublished observations.
However, antibodies may not accurately reflect any EGF-stimulated change in erbB-2 internalization. Rapid endocytosis is not necessary for down-regulation, however, because constitutive internalization rates are sufficiently quick. For example, if constitutively internalized erbB-2 are directed entirely to the lysosomes for degradation instead of being recycled, their half-life would be approximately 15 min (
). The absence of a major effect of EGF on erbB-2 internalization does suggest that heterodimers between erbB-2 and the EGFR are not sufficiently stable to be internalized as a complex.
We also found that erbB-2 overexpression had a striking effect on regulation of the EGFR. Not only did high levels of erbB-2 induce constitutive activation of the EGFR, but they also inhibited EGF-induced down-regulation of the EGFR. Although constitutive activation of erbB-2 has previously been described as an effect of erbB-2 overexpression, a corresponding effect on EGFR phosphorylation has not been previously reported. The high number of EGFR found in human mammary epithelial cells and the consequent increased sensitivity of our assays could account for our ability to observe this effect. The constitutive phosphorylation of EGFR was quite significant, and on a per cell basis, was higher than that observed following chronic treatment with EGF (Fig. 5C). Monoclonal antibodies that blocked either ligand binding (
) had no effect on this constitutive phosphorylation, indicating that it was probably due to expression-driven heterodimerization with erbB-2. The biological effects of this constitutive EGFR activation are currently unknown.
An unexpected finding was that erbB-2 overexpression also inhibited down-regulation of activated EGFR. This resulted in prolonged signaling through the EGFR. The mechanism by which erbB-2 inhibits EGFR down-regulation is complex. We have found a decrease in EGF dissociation, an increase in recycling fraction, as well as an inhibition of lysosomal targeting. The overall effect of these changes was to greatly enhance ligand-receptor stability, thus increasing the number of activated receptors per cell. These changes are diagrammed in Fig. 10. To determine which specific changes were most important in increasing activated EGFR levels, we calculated their relative contributions using the quantitative model previously described (
). The 7-fold decrease in ligand dissociation rates we observed would increase cell surface levels of activated EGFR only about 1.5-fold. This is because activated receptors are primarily lost by internalization, not dissociation (i.e. ke > kd). The increase in the time necessary to reach the lysosomes (from 15 to 20 min) would increase intracellular pools of receptors 1.33-fold, but in combination with the reduction in degradation rate constant (from 0.021 to 0.13 min−1), intracellular pools would increase 2-fold. This is in good agreement with our observed doubling of intracellular ligand pools (Fig. 7). Fractional recycling from this pool increases from approximately 30 to 45% in the erbB-2 overexpressing cells (Fig.8D). In combination with the expanded intracellular pools, this translates to a 3-fold higher net recycling rate. Because of the much larger intracellular pool of EGFR relative to the cell surface, this would raise surface-associated EGF about 3-fold. The lowered dissociation rate would increase this further, to the 4–5-fold range observed in our experiments. Thus altered postendocytic receptor trafficking is mainly responsible for the enhanced levels of activated EGFR.
) have recently examined the effect of erbB-2 expression on EGFR behavior. They likewise found that erbB-2 overexpression decreased the rate of EGF dissociation from EGFR and delayed their inactivation (
). Although our results are in general agreement with their findings, we did observe some differences. For example, we could find no evidence for an inhibition of EGFR internalization due to erbB-2 overexpression (Fig. 8A). We also found no evidence that erbB-2 overexpression alters EGFR affinity at 0 °C (Fig. 1B) or potentiates signaling through the EGFR (Fig. 5B). The different results are probably due to cell-specific factors and our use of more specific receptor trafficking assays. Nevertheless, we confirm that erbB-2 overexpression increases signaling through the EGFR primarily by delaying receptor inactivation.
The effect of erbB-2 overexpression is to inhibit specific sorting steps in the endocytic pathway. Mechanistically, the simplest explanation for this would be competition between erbB-2 and the EGFR for limiting sorting components in the endocytic pathway. There is strong sequence similarity between erbB-2 and the EGFR in the regions that have been identified as involved in postendocytic trafficking (
). However, the fraction of EGFR involved in heterodimer formation is unknown. It is also uncertain whether the stability of the heterodimers is sufficient to affect receptor trafficking. Experiments are currently in progress to evaluate the role of heterodimer formation in EGFR trafficking.
Although the effects of erbB-2 overexpression on EGFR basal activation and down-regulation were unexpected, they are not inconsistent with the emerging view of the erbB family as a highly interactive group of receptors that coordinately regulate signal transduction (
). Our findings add another layer of complexity to the role of erbB-2 overexpression in breast cancer. Reciprocally, overexpression of the EGFR may alter the normal regulation of erbB-2. It thus appears that regulation of signaling from a single erbB family receptor cannot be completely understood without considering its role as a member a highly interactive signaling group.
We thank Joyce Taylor-Papadimitriou for the MTSV, ce1, and ce2 cells.