Systems Biological Analysis of Epidermal Growth Factor Receptor Internalization Dynamics for Altered Receptor Levels

Epidermal growth factor (EGF) receptor (EGFR) overexpression is a hallmark of many cancers. EGFR endocytosis is a critical step in signal attenuation, raising the question of how receptor expression levels affect the internalization process. Here we combined quantitative experimental and mathematical modeling approaches to investigate the role of the EGFR expression level on the rate of receptor internalization. Using tetramethylrhodamine-labeled EGF, we established assays for quantifying EGF-triggered EGFR internalization by both high resolution confocal microscopy and flow cytometry. We determined that the flow cytometry approach was more sensitive for examining large populations of cells. Mathematical modeling was used to investigate the relationship between EGF internalization kinetics, EGFR expression, and internalization machinery. We predicted that the standard parameter used to assess internalization kinetics, the temporal evolution r(t) of the ratio of internalized versus surface-located ligand·receptor complexes, does not describe a straight line, as proposed previously. Instead, a convex or concave curve occurs depending on whether initial receptor numbers or internalization adaptors are limiting the uptake reaction, respectively. To test model predictions, we measured EGF-EGFR binding and internalization in cells expressing different levels of green fluorescent protein-EGFR. As expected, surface binding of rhodamine-labeled EGF increased with green fluorescent protein-EGFR expression level. Unexpectedly, internalization of ligand· receptor complexes increased linearly with increasing receptor expression level, suggesting that receptors and not internalization adaptors were limiting the uptake in our experimental model. Finally, determining the ratio of internalized versus surface-located ligand·receptor complexes for this cell line confirmed that it follows a convex curve, supporting our model predictions.The epidermal growth factor receptor (EGFR)³ belongs to the family of transmembrane receptor tyrosine kinases and mediates diverse actions, including proliferation, differentiation, and apoptosis (1, 2). Overexpression and/or mutations of the EGFR occur in ∼40% of neoblastomas (3) and correlate with poor prognosis (4–6). Unstimulated EGFR is located at the plasma membrane as a monomer and pre-formed dimer (7). Upon ligand binding, EGFR forms a dimer, and trans-phosphorylation occurs at specific residues of the cytoplasmic domain (8). Phosphorylated EGFR recruits adaptor proteins from which different conserved signaling pathways are activated, namely the MAPK (9), phosphatidylinositol 3-kinase, and protein kinase C pathways (10).Furthermore, activated EGFR recruits various adaptor proteins that mediate receptor internalization by endocytosis (2). Endocytosis occurs via the recruitment of adaptor proteins to phosphorylated tyrosine residues of the receptor and formation of membrane invaginations, which eventually pinch off to form internalized early endosomes (2, 11) (see Fig. 1). Both constitutive endocytosis and ligand-induced EGFR endocytosis are critical events in EGF signal regulation (2, 12). Endosomal EGFR can be transited back to the plasma membrane or to the late endosome/lysosome for degradation (2). As the majority of internalized receptors are targeted for lysosomal degradation upon EGF stimulation (13), endocytic entry of active EGFR is a crucial step for signal attenuation, which is also highlighted by the findings that impaired or delayed internalization is highly oncogenic (14, 15).Open in a separate windowFIGURE 1.Scheme of ligand-induced internalization. EGF binds membrane-located EGFR to give rise to surface-bound EGF·EGFR complex RE_s. Via diffusion events, the activated receptor binds internalization adaptors IC, which leads to internalized receptors R_i.In light of the role of endocytosis in EGFR signal attenuation and the oncogenicity of EGFR overexpression, it is important to elucidate the relationship between high receptor expression levels relative to internalization pathway capacity and their effect on internalization dynamics.Mathematical modeling is an important tool in elucidating EGFR signaling, at the level of EGFR internalization (16–19) and, more recently, at the level of the integration of input signals into signaling events downstream of the EGFR, such as the MAPK cascade (20, 21). In earlier models, pioneering concepts such as the nonlinearity of the uptake reaction, because of the existence of alternative pathways that are entered with different affinities, were developed (16, 19). Also, the notion of saturability of the EGFR endocytosis system, in contrast to internalization of the transferrin receptor, for example, was introduced (18).Importantly, in mathematical formulations of EGFR endocytosis, the standard parameter used to estimate the rate of the internalization step (16) and to assess the effect of certain perturbations on internalization (22–24) is the temporal evolution of the ratio of internalized versus surface-located ligand·receptor complexes r(t). In Refs. 16, 17, it was mathematically determined that, under certain assumptions, this ratio describes a straight line with the slope corresponding to the rate of the internalization step. These assumptions were as follows: (i) that the number of surface-bound ligand·receptor complexes (RE_s) remains approximately constant during the measurements, and (ii) that the internalization step is a first-order process, i.e. it is directly proportional to RE_s and independent of a potentially limiting availability of internalization adaptors.The presence of multiple endocytotic routes (23, 25) and different EGFR affinities for EGF (26) argue against first-order kinetics. Moreover, the possible limited capacity of internalization adaptors may restrict EGFR internalization in cells expressing abnormally high numbers of EGFR (18). In this work we investigated the potential of EGFR internalization to occur as a nonlinear process by combining mathematical modeling with novel quantitative, live cell measurements of EGF internalization.We extended the previous derivation of the ratio of internalized versus surface-located ligand·receptor complexes r(t) (16, 17, 19) by eliminating above assumptions i and ii, which allowed us to investigate in silico different scenarios for the shape of r(t) as a function of the relative concentrations of EGFR and internalization adaptors. We predicted that r(t) is not a straight line as derived previously but is a convex or concave curve depending on whether receptors or internalization components are limiting the reaction, respectively.In earlier studies, quantitative measurements of parameters of EGFR endocytosis have been performed using classical biochemical techniques to detect cellular ligand uptake using radioactively labeled EGF (16, 24, 27) or biotin-labeled EGF (28). Importantly, both methods do not reach single cell precision and instead yield an integrated signal over a population of cells. To test our mathematical predictions we combined the following: (i) quantitative laser scanning confocal microscopy, and (ii) multiple parametric flow cytometry, using a custom Beckman Coulter FC500 equipped with a 488 and 561 nm laser excitation, to quantitatively measure the temporal and spatial dynamics of EGFR endocytosis using tetramethylrhodamine-tagged EGF (Rh-EGF) and GFP-EGFR. We show that both quantitative imaging and flow cytometry measurements were highly sensitive, allowing for live cell investigations and confirmation of the mathematical predictions.