Coregulation of Epidermal Growth Factor Receptor/Human Epidermal Growth Factor Receptor 2 (HER2) Levels and Locations

INTRODUCTION

EGFR 3 and HER2 expression levels have distinguished themselves as important factors in contributing to various types of cancers including breast and ovarian cancers (1) , but quantitative linkages between receptor expression levels and aberrant cell behaviors are not well understood. HER2 is a member of the EGFR family of receptor tyrosine kinases consisting of four highly related members (EGFR, HER2, HER3, and HER4) that form a wide array of homo- and heterodimers after the binding of an EGF family ligand, such as EGF, or transforming growth factor-α, to the EGFR, or neuregulin to either HER3 or HER4 (2, 3, 4) .

HER2 is overexpressed in approximately one-third of breast and ovarian cancers and has been negatively correlated with patient prognoses (1 , 5 , 6) . In addition to its clinical presence as a prognostic factor, HER2 overexpression has been shown to contribute to cell transformation, anchorage-independent cell growth, and increased proliferation and mitogenic sensitivity, as well as to increase tumor cell migration and invasiveness (7, 8, 9, 10, 11, 12, 13) .

HER2 is almost ubiquitously expressed, yet it does not bind any of the EGF family ligands. Consequently, the activation of HER2 tyrosine kinase activity and subsequent recruitment of signaling intermediates relies on heterodimerization with other EGFR family members after ligand stimulus or via homodimerization, both of which are facilitated by elevated HER2 expression (14) . The downstream effect of HER2 overexpression on signaling appears to be 2-fold: (a) enhanced cellular sensitivity to growth factor stimulation caused by signal amplification via the recruitment of additional signaling molecules (4 , 15, 16, 17, 18) ; and (b) diminished negative regulatory mechanisms involved in signal attenuation, resulting in prolonged signal duration (19 , 20) .

The trafficking behavior of the EGFR, in the absence of other family members, has been well characterized (21) . Activated receptors are rapidly internalized by receptor-mediated endocytosis, removing them from the cell surface. Once internalized, receptors and ligands are sorted in endosomes and either are targeted to lysosomal degradation or follow the default recycling pathway back to the surface (21) . Both endocytic and endosomal sorting machinery can be impaired via saturation at high receptor levels, apparently because of limiting levels of the regulatory molecules involved in these processes (22, 23, 24) .

The intimate interaction between the EGFR and HER2 should result in reciprocal effects on the trafficking of both receptors when either is overexpressed. Overexpression of HER2 has been demonstrated to inhibit down-regulation of the EGFR and of itself, as well as to increase the fraction of EGF recycled (20) . HER2 expression has been shown to shunt ligand-activated receptors to recycling fates, suggesting that receptor heterodimer species may have superior signaling potency as a consequence of their altered intracellular routing (25 , 26) .

Mounting evidence suggests that receptor compartmental location plays a crucial role in determining cellular responses to external stimuli (21 , 27, 28, 29, 30) . Clearly, changes in receptor trafficking can affect the duration and quality of signals generated. This, in turn, could ultimately be responsible for the pathological behaviors associated with receptor overexpression, such as heightened mitogenic sensitivity and increased cell motility.

To better understand the effects of HER2 overexpression on cell responses in terms of the specific underlying mechanisms, we have developed a computational model geared at capturing EGFR and HER2 trafficking dynamics with a minimal level of complexity. The model presented here is able to generate a priori predictions of EGFR and HER2 levels and compartmental locations based on the empirical measurement of their relative trafficking behaviors. Most importantly, the model yields insight into which rate processes are the most influential in dictating the overall system behavior.

MATERIALS AND METHODS

Experimental

Reagents.

7C2 Fab fragments, 2C4 mAb, 13A9 mAb, and 4D5 mAb were generous gifts from Genentech, Inc. mAb 225 was purified from hybridomas obtained from American Type Culture Collection (31) . 7C2 Fab, 13A9, 225, and EGF were labeled with ¹²⁵I (NEN) using iodobeads (Pierce) as described elsewhere (32) . Human EGF was obtained from Peprotech.

Cell Culture.

184A1 HMECs were a kind gift from Martha Stampfer (Lawrence Berkeley Laboratory, Berkeley, CA) and were maintained in DFCI-1 medium supplemented with 12.5 ng/ml EGF (33) .

Expression of HER2 by Retroviral Transduction.

Different levels of HER2 expression were achieved in HMEC 184A1 cells by retroviral transduction. The retroviral vector was constructed by excising HER2 from the LTR-2 vector (9) with Xho1, adding Not1 linkers and inserting this into the MFG vector (34) that was modified to contain neomycin resistance as well as a Not1 site. The resulting HER2 vector was transfected into the Ψ̃CRIP packaging cell line as described previously (35) . Clones of transfectants were screened for HER2 expression using mAb 4D5, and virus-containing supernatants were collected and screened for high titer. Cells were transfected with retrovirus stock using 4 μg/ml Polybrene for 2 h and were grown for 2 days before plating at clonal density. Individual colonies were isolated using cloning rings and then were screened by immunofluorescence and by flow cytometry. Surface HER2 levels were characterized by flow cytometry and equilibrium-binding studies using ¹²⁵I-labeled 7C2 Fab.

Binding Studies.

Before experiments, cells were grown to confluency and brought to quiescence in binding medium (DFCI-1, without sodium bicarbonate, EGF, bovine pituitary extract, or fetal bovine serum) overnight and placed in an air incubator. Cell number per plate was determined with parallel plates using a Coulter Counter.

Strip Protocol.

Surface-bound ligand was determined by removal with an acid strip solution (36 , 37) . Cells were washed five times with ice-cold wash buffer [1 mg/ml polyvinylpyrrolidone, 130 mm NaCl, 5 mm KCl, 0.5 mm MgCl₂, 1 mm CaCl₂, and 20 mm HEPES (pH 7.4)] and incubated on ice with acid strip solution [50 mm glycine-HCl, 100 mm NaCl, 1 mg/ml polyvinylpyrrolidone, 2 m urea (pH 3.0)]. Surface-bound ligand was quantified by counting acid strip solutions in a gamma counter. Internalized ligand was determined by solubilizing cells after the acid strip with 1 n NaOH and counting on a gamma counter.

Internalization Rate Constant Measurement.

Internalization rate constants were determined as described previously (37) . Unoccupied EGFR, bound EGFR, and HER2 internalization measurements were done with ¹²⁵I-labeled 225, -EGF and -7C2 Fab, respectively (36) .

Endosomal Sorting Assay.

Endosomal sorting trafficking parameters were determined as described previously (24 , 38) . Cells were incubated at 37°C with binding medium containing various levels of ¹²⁵I-labeled EGF or 200 ng/ml ¹²⁵I-labeled 7C2-Fab for 2.5 h to allow the sorting process to reach steady state. Cells were washed with PBS, and surface bound ligand was removed with acid strip solution without urea for 2 min at 4°C. Cells were washed twice with PBS and returned to 37°C with binding medium containing excess ligand (160 nm unlabeled EGF or 1 μg/ml unlabeled 7C2-Fab). After 10 min, degraded and intact (recycled) ligands in the medium were separated via centrifugal ultrafiltration using 5000 MWCO filter units (Millipore).

Receptor Distribution (Inside/Surface) Measurement.

Cells were incubated at 37°C with ¹²⁵I-labeled EGF, ¹²⁵I-labeled 7C2 Fab, or ¹²⁵I-labeled 13A9 in binding medium and allowed to reach steady state (2.5–5 h). Surface bound and internal ligand was quantified as detailed above.

Surface Receptor Down-Regulation.

Cells were treated with or without 100 ng/ml EGF for 5 h. Surface receptor number was then quantified by the addition of saturating quantities (600 ng/ml) of ¹²⁵I-labeled 13A9, for EGFR, or by equilibrium binding studies with ¹²⁵I-labeled 7C2 Fab, for HER2. At equilibrium, surface bound ligand was determined as described above in “Strip Protocol.” Fractional down-regulation is defined as surface receptors (EGFR or HER2) after EGF treatment divided by surface receptors before treatment.

Computational

Model Development.

A mathematical model for HER2/EGFR trafficking dynamics can include different levels of detail. At a maximal extreme, it could explicitly include thermodynamic/kinetic interactions regulating each of the individual trafficking processes and combine them into a very large system with dozens of equations and parameters. At a minimal opposite, as we have chosen, it could account for these interactions implicitly through lumped rate constants describing the various processes, with quantitative experimental measurements providing empirical characterization of these rate constants. This results in a reduced number of equations with a minimized set of system parameters. Thus, we are following a “top-down” approach in which each lumped rate constant for a given process could be expanded into a subordinate model for greater molecular-level detail.

EGFR Model.

At the most macroscopic level, the trafficking of a given receptor can be divided into four steps: synthesis, internalization, recycling, and degradation. Collectively, the behavior of these processes governs both the number of receptors present on a cell and their distribution between surface and internal compartments.

EGFR trafficking can be modeled with a simple set of differential equations describing the motion of empty and occupied receptors from one compartment to another (illustrated in Fig. 1 ⇓ ; Ref. 39 ): Free EGFR (Rs) are synthesized at rate S_R, internalized at the constitutive rate (k_er) and exit endosomes at rate k_xr with fraction f_xr recycling to the surface and fraction (1 − f_xr) being degraded. EGF (L) reversibly binds free EGFR with on and off rate k_f and k_r, respectively. EGF-EGFR complexes (in all forms, including homodimers and heterodimers with HER2) internalize at rate k_ec and are sorted in endosomes with exit rate k_xc, recycling fraction f_xc, and degradation fraction (1 − f_xc). EGFR-HER2 interactions after ligand addition are implicitly incorporated by allowing k_ec and f_xc to depend on HER2 expression level. These dependencies have been empirically determined for this study (see Figs. 3 ⇓ ,4 ⇓ and Table 2 ⇓ ). We assume that all of the EGF remains bound once internalized.

HER2 Model.

For HER2, the model is completely analogous to that of EGFR in the absence of ligand (illustrated in Fig. 1 ⇓ ): HER2 synthesis, internalization, endosomal exit, and sorting fraction are given by S_H, k_eh, k_xh, and f_xh, respectively. HER2-EGFR interaction and the effect of ligand addition are implicitly contained in the k_eh term, where k_eh is a complex function of both HER2 level and EGF stimulation (see Fig. 3b ⇓ and Table 2 ⇓ ). Hs and Hi include both free HER2 and HER2 that is homo or heterodimerized.

RESULTS

The primary focus of our effort here is to understand how expression of HER2 affects the distribution and levels of activated EGFR using a quantitative, systems modeling approach. In general, the distribution of receptors depends on their rate of entry into cells (internalization) and rate of exit (recycling and degradation). To describe internalization, we use the concept of the “endocytotic rate constant” which has been well characterized in a variety of cell types (20 , 32 , 36 , 40 , 41) . This lumped constant encapsulated information of a variety of endocytic parameters, such as number of coated pits, receptor activation state, and binding to proteins within coated pits (36 , 40) . To describe the loss of receptors from cells, we use two parameters: an endosomal exit constant (k_x) and a recycling fraction (f_x). Mechanistically, k_x describes the first order rate of receptor transit through the endosomal compartment per unit time, whereas f_x is simply the fraction of receptors that recycle intact back to the cell surface. These lumped parameters allow us to consolidate several molecular-level events into experimentally accessible parameters and, thus, permit identification of the steps at which the system is altered.

For our approach to be useful, expression of HER2 must cause predictable and reproducible alterations in our model parameters. In addition, these alterations must be reasonable from a mechanistic viewpoint. For this, we used a series of cloned 184A1 HMEC lines expressing various levels of HER2, constructed by retroviral-mediated gene transfer. The uniformity of HER2 expression in each clonal population is shown in Fig. 2 ⇓ . Approximate surface EGFR and HER2 expression levels in these cells are listed in Table 1 ⇓ .

Effect of HER2 Expression on EGFR and HER2 Trafficking

Internalization.

To examine the relationship between HER2 levels and EGFR internalization, we measured EGF internalization in the different clonal cell lines using the method of Wiley [Wiley and Cunningham (36) and Lund et al. (37)] . As shown in Fig. 3a ⇓ , increasing HER2 expression elicited up to a 60% decrease in the EGF endocytotic rate constant from 0.25 min⁻¹, for the parental line, to 0.10 min⁻¹, for HER2 clones 24H and 1. To probe the role of heterodimerization in this effect, we repeated these measurements after pretreatment with 2C4, an anti-HER2 mAb found to block both heterodimerization and transactivation of the EGFR (42, 43, 44) . Pretreatment with 2C4 mAb abrogated the HER2-dependent effect on EGF internalization (Fig. 3a) ⇓ , suggesting that heterodimerization between HER2 and EGFR leads to a reduction in EGFR internalization rates.

To determine whether activation of the EGFR had a reciprocal effect on HER2 internalization, we used 7C2 Fab as an artificial HER2 ligand. 7C2 mAb binds to an extracellular epitope on HER2, but does not interfere with heterodimerization (42) . HER2 internalization was measured in the presence and absence of 10 ng/ml EGF in our set of HER2 clones (Fig. 3b) ⇓ . We found that HER2 is internalized very slowly in the absence of EGF, at a rate consistent with constitutive membrane turnover (∼0.01 min⁻¹; Ref. 37 and 45 ). Activation of the cellular complement of EGFR elicits an increase in HER2 internalization that decreases from 3- to 2-fold with increasing HER2 expression.

HER2 expression had no observable effect on the constitutive internalization rate constant of unoccupied EGFR (data not shown), as measured with ¹²⁵I-labeled 225, an antagonistic mAb that binds to the EGFR (46) . The EGF association (k_f) and dissociation (k_r) rate constants, were also measured and found to be 9.7 × 10⁷ m⁻¹min⁻¹ and 0.24 min⁻¹, respectively, roughly constant across all HER2 levels.

Receptor Recycling.

The values of the endosomal exit constant (k_xc) and a recycling fraction (f_xc) were determined using a steady-state sorting assay that follows the fate of ¹²⁵I-labeled EGF. At steady state, the rate at which EGF is lost from the cells (both degraded and intact) is equal to the rate at which it exits from the endosomes. Because the fraction of EGF recycled is somewhat dependent on the size of the internal pool of ligand, we incubated cells with different concentrations of ¹²⁵I-labeled EGF. By 2.5 h, the intracellular sorting process reached quasi-steady state as evidenced by an approximately constant amount of internal EGF. Surface-bound EGF was removed with a mild acid strip, and the cells were returned to the medium with excess unlabeled EGF to prevent rebinding. After 10 min, intact and degraded ¹²⁵I-labeled EGF in the medium were separated by centrifugal ultrafiltration. The fraction of EGF recycled (f_xc) was calculated from the ratio of intact ¹²⁵I-labeled EGF to the total ¹²⁵I-labeled EGF (intact and degraded) in the medium (24) . The value of k_xc was determined from the rate at which total ¹²⁵I-labeled EGF appeared in the medium.

Consistent with our previously published studies (20) , increasing levels of HER2 expression resulted in an increased fraction of EGF recycling back to the cell surface (Fig. 4) ⇓ , from 0.5 up to 0.7. The extent to which HER2 expression affected recycling was dependent on the size of the intracellular pool of ligand, with the effect being more pronounced at higher levels of internalized EGFR.

Blocking heterodimerization with 2C4 pretreatment was sufficient to reverse the effect of elevated HER2 expression on the EGF sorting fraction (Fig. 4) ⇓ and had no effect on sorting in the parental cell line (data not shown). The observation that HER2 expression affected only the fraction of recycled ligand, but not the transit of ligand through the endosomes suggests that HER2 interferes with lysosomal targeting of the EGFR. The enhanced effect observed at high levels of internalized EGF is consistent with this idea. The endosomal exit constant of EGF (k_xc) was essentially constant across all HER2 clones at 0.036 min⁻¹.

The sorting of HER2 was measured by following the fate of attached 7C2 Fab, in a manner analogous to that of EGF. We found that for all of the HER2-expressing lines, the fraction of recycled 7C2 Fab was very high (0.94). The endosomal exit constant of 7C2 Fab was also very similar in all of the cell lines at 0.07 min⁻¹. This value is about twice that observed for EGF, indicating that the occupied EGFR transits the endosomal apparatus slower than HER2. The addition of EGF had no measurable effect on either the fraction of recycled 7C2 or its endosomal exit constant. Recycling was the predominant fate of 7C2 Fab and was always much faster than internalization. We conclude that HER2 is primarily localized to the cell surface because of a slow internalization rate and relatively fast recycling rate.

The complete set of trafficking parameters is shown in Table 2 ⇓ .

Model Predictions

Effect of HER2 Expression on Receptor Location.

We next determined whether our model could accurately describe the distribution of EGFR and HER2. A priori predictions of EGFR distribution can be made based on experimental data for the individual trafficking steps and tested against an independent set of experimental results. We used the inside:surface receptor ratio as an end point, rather than absolute amounts, to remove the dependence on receptor synthesis (S_R). Thus, we have a model that is completely defined in an experimental sense with relative EGFR and HER2 expression as the relevant parameters. The inside:surface ratio of EGF at steady state is: For the case of HER2, the same relationship would hold in that the inside:surface ratio would be equal to k_eh/k_xh. We determined whether cells displaying a given value of k_ec or k_eh would display an inside:surface ratio predicted from the previously determined values of k_xc and k_xh, which were roughly constant across all HER2 expression levels.

After EGF treatment, the inside/surface distribution of EGF reached a steady state within 2 h (data not shown). The inside:surface EGF ratio was measured and plotted as a function of k_ec for the parental and HER2-expressing lines, shown in Fig. 5a ⇓ . We also determined the inside/surface distribution of HER2 before and after EGF treatment using steady-state ¹²⁵I-labeled 7C2 Fab binding. This was plotted as a function of the measured value of k_eh, also shown in Fig. 5a ⇓ . A diagonal line corresponding to a theoretical endosomal exit constant of 0.1 min⁻¹ has been provided for comparison.

The inside/surface data for both HER2 and EGF from all of the cell lines shows good accordance with a priori model predictions and falls parallel to the line describing a constant endosomal exit constant (Fig. 5a) ⇓ . This suggests that HER2 expression does not change the rate at which either the EGFR or the HER2 transits the endosomal apparatus, but simply alters the sorting pattern within endosomes. This explicit assumption of our model appears to be entirely consistent with the data. The addition of EGF appears to cause a shift of HER2 toward a faster endosomal transit. The approximately 50% faster endosomal transit appears independent of HER2 expression level and may reflect the enhanced membrane turnover stimulated by EGF (47) . The EGF data are shifted considerably from the diagonal line describing the HER2 data, indicating that the EGFR transits the endosomal apparatus much slower than HER2. This is consistent with previous work suggesting that endosomal retention is a major mechanism by which intracellular EGFR is regulated (48) .

Effect of HER2 Expression on Surface Receptor Loss.

A second set of predictions that can be made relates to the loss of receptor at the cell surface as a result of both internalization and accelerated degradation (down-regulation). Fractional surface receptor loss is determined by solving the model at steady state in the presence of 100 ng/ml EGF. The number of receptors remaining on the surface after EGF stimulus is divided by the original number present on the surface to yield the fractional surface loss (D_EFGR):

The effect of EGFR activation on the fractional surface loss of HER2 (D_HER2) is much simpler and of the form: where superscript + or − indicates the parameter values associated with or without EGF treatment, respectively.

Levels of EGFR and HER2 at the cell surface were measured by comparing steady-state binding of ¹²⁵I-labeled 13A9 and ¹²⁵I-labeled 7C2 Fab before and after 5-h incubation with 100 ng/ml EGF in the different cell lines. The data, shown in Fig. 6a ⇓ , are plotted as a function of k_ec and k_eh. We found that increasing HER2 expression inhibited the fractional loss of both itself and EGFR. Because EGFR loss is also a function of the recycling fraction (f_xc), and because HER2 expression causes an increase in this parameter, we have plotted model predictions using values from both high-HER2-expressing cells and the parental cells (Fig. 6a) ⇓ .

Our model is very sensitive to small changes in the fraction of EGF or HER2 recycled. In our experimental measurements, EGF had no discernible effect on HER2 recycling, but changes within experimental error may have large effects on the model output. As such, we consider two cases in Fig. 6a ⇓ : (a) EGF elicits no change in HER2 sorting fraction (f_xh+ = 0.94); and (b) a scenario where EGF treatment elicits a small degree of sorting saturation resulting in a 4% increase in HER2 sorting fraction (f_xh+ = 0.98). An increased recycling fraction would also be compatible with the likely increase in endosomal transit rates of HER2 suggested by the results presented in Fig. 5a ⇓ .

Effects of 2C4 mAb.

Finally, a priori predictions from the model can be tested for the effects of the 2C4 mAb, which interferes with HER2/EGFR heterodimerization. We evaluated predictions of EGFR loss from the cell surface and steady-state inside:surface ratios. The EGFR model was evaluated using parameter values measured after preincubation with 2C4 (Table 2) ⇓ and compared with experimental measurements of inside:surface ratios and surface receptor levels, also after 2C4 treatment (Figs. 5b ⇓ and6b ⇓ ). For both the model and the experiments, 2C4 treatment returned the EGF inside:surface ratio and fractional surface EGFR down-regulation for clone 24H to levels similar to the parental cell line. 2C4 had no effect on the parental cell line (data not shown). The model successfully predicts the direction of change resulting from the addition of 2C4 mAb; however, there are some minor discrepancies in the prediction of magnitudes. It is unclear whether these are attributable to cell-cell variations or to errors in parameter estimations.

DISCUSSION

Changes in HER2 expression levels have significant effects on EGF family ligand-induced signaling that contribute to alterations in cell behavior, such as those found in breast cancer (10 , 12 , 19 , 49) . Because signaling can be simultaneously regulated by receptor type (e.g., EGFR versus HER2), receptor location (e.g., cell surface versus intracellular compartment), and receptor levels, a reliable integrative model for EGFR/HER2 coregulation should be very helpful toward deconvoluting the effects of HER2 overexpression on cell responses.

We have used a quantitative “top-down” approach in the development, validation, and application of a dynamic systems model capable of a priori prediction of HER2 and EGFR levels and compartmental locations. We have predicted the overall effects of HER2 expression on the quantity and distribution of EGFR and HER2 with reasonable accuracy and without fitting any parameters.

The empirical measurement of individual trafficking parameters and their dependence on HER2 expression adds to a growing body of EGFR and HER2 trafficking data. With regard to internalization: (a) we have confirmed that HER2 is indeed internalized, albeit slowly; (b) we have demonstrated that EGF addition accelerates HER2 internalization, although not to the same degree as reported in some other cell types or chimeric constructs (50 , 51) ; and (c) we have found that heterodimerization, as a consequence of increased HER2 expression, reduces but does not block EGF internalization. Our experimental results provide some interesting new insights into HER2 dynamics in mammary epithelial cells. We report here the first measurements of the recycling rate constant and sorting fraction of HER2, and demonstrate that these are essentially independent of HER2 expression and minimally dependent on EGF stimulation. Second, our EGF sorting results demonstrate an increase in fraction recycled with increased HER2 expression, consistent with that observed in HB2 cells with various levels of HER2 expression (20) . The EGFR trafficking model and experiments after 2C4 intervention provide quantitative support for the notion that blocking heterodimerization can significantly affect the EGFR signaling attributable to effects on receptor trafficking. Some of the therapeutic effect of Herceptin 4 could be attributable to such a mechanism.

Our modeling allows us to easily scan a multiparameter space and critically discern which regulatory processes are the most sensitive or insensitive to disruption. From these calculations, it is evident that EGFR compartmental location is primarily controlled at the level of internalization. The down-regulation of EGFR, however, is controlled by a balance of both internalization and endosomal sorting (see Fig. 6a ⇓ ).

HER2 overexpression appears to disrupt EGFR trafficking at both internalization and endosomal sorting. The mechanisms by which HER2 overexpression is able to alter each step is unclear, but heterodimerization appears to be required. This may imply that homodimerization of the EGFR is necessary for efficient trafficking, or that the conformation of the heterodimerized EGFR is not optimally recognized by the endocytic machinery.

The majority of the model error lies in predictions of down-regulation. This suggests that most of the error arises from the sorting fraction parameter, in part from experimental uncertainty but mainly from the tremendous simplification of sorting behavior. The current model does not propose the existence of more than a single recycling compartment and does not account for differential sorting of ligands and receptors within endosomes. Clearly, more detail at the level of endosomal sorting is needed to precisely capture the behavior, but this will come at the expense of having an experimentally measurable parameter set (52) .

This model is specific for HER2 expression effects in the face of EGF stimulus. Because parameter dependencies were empirically determined, they implicitly include interaction with all other EGFR family receptors and not just EGFR. In principle, we could write such a model for HER3 and HER4 and use this approach for the effects of any receptor overexpression in the presence of any combination of stimuli; it simply requires the empirical determination of parameters spanning the desired space.

From both the model and the data, it is clear that the net trafficking effect of elevated HER2 expression, after EGF stimulus, is to increase receptor lifetime and to shift the receptor distribution toward the surface. This should increase the half-life of a given ligand stimulus, resulting in prolonged signaling. The decrease in inside:surface ratio may also impact the quality of the signals generated, preferentially activating signaling pathways that are restricted to the cell surface. Increased surface signaling relative to total signaling may alter cell motility via increased PLC-γ, calpain, and gelsolin activation (53, 54, 55, 56) .