EGFR juxtamembrane domain, membranes, and calmodulin: kinetics of their interaction

Abstract

Calcium/calmodulin (Ca/CaM) binds to the intracellular juxtamembrane domain (JMD) of the epidermal growth factor receptor (EGFR). The basic JMD also binds to acidic lipids in the inner leaflet of the plasma membrane, and this interaction may contribute an extra level of autoinhibition to the receptor. Binding of a ligand to the EGFR produces a rapid increase in intracellular calcium, [Ca2+]i, and thus Ca/CaM. How does Ca/CaM compete with the plasma membrane for the JMD? Does Ca/CaM directly pull the JMD off the membrane or does Ca/CaM only bind to the JMD after it has dissociated spontaneously from the bilayer? To answer this question, we studied the effect of Ca/CaM on the rate of dissociation of fluorescent JMD peptides from phospholipid vesicles by making kinetic stop-flow measurements. Ca/CaM increases the rate of dissociation: an analysis of the differential equations that describe the dissociation shows that Ca/CaM must directly pull the basic JMD peptide off the membrane surface. These measurements lead to a detailed atomic-level mechanism for EGFR activation that reconciles the existence of preformed EGFR dimers/oligomers with the Kuriyan allosteric model for activation of the EGFR kinase domains.

Figure 1

Cartoon of the dominant model for EGFR activation based on similar figures in previous studies (6,12,13,15). Binding of a ligand to the EGFR exposes a dimerization arm; lateral diffusion and binding of dimerization arms result in formation of a dimer. This allows the C-terminal or large lobe of one kinase domain to bind to the N-terminal lobe of the adjacent kinase domain and activate it by an allosteric mechanism (12). The activated kinase then phosphorylates tyrosine residues in the C-terminal tail region of the adjacent EGFR. The JMD is assumed to be free, flexible, and extended into the cytoplasm. An alternative “twist” model for activation (not shown) assumes that binding of EGF to each EGFR in a preformed dimer causes a rearrangement of the complex, perhaps a relative rotation of subunits (see text for discussion). The extracellular region of EGFR (orange), TM helix (purple), natively unfolded JMD (black line), kinase domain (green), and C-terminal tail (black) including tyrosine (dark blue circles) or phosphotyrosine (orange) residues are shown.

Figure 2

Ca/CaM enhances the rate of desorption of the EGFR JMD peptide from a PC/PS vesicle. (A) Kinetics of the transfer of acrylodan-EGFR(645–660) from PC/PS vesicles (15% PS) to Ca/CaM. Vesicles with membrane-bound peptide were mixed rapidly with a solution containing CaM to obtain a final Ca/CaM concentration in the mixing chamber of 0.5 μM (triangles) or 4 μM (squares). The experimental data points are the average of 10 shots. The curves are the single exponential fits to the data. Note that increasing [Ca/CaM] from 0.5 to 4 μM decreases the relaxation time constant ∼4-fold. Additional experimental details are in the Methods section of the Supporting Material. (B) The reciprocal of the time constant (1/τ) obtained from experiments shown in A and similar experiments is plotted versus [Ca/CaM] in the final mixing chamber. Note that both the intercept and slope increase as the mole fraction of PS in the vesicles decreases. The intercept is one measure of the spontaneous dissociation rate of the peptide from the vesicle; the slope is a measure of the transfer rate constant between the vesicle and Ca/CaM. The average values of these parameters are reported in Table 1.

Figure 3

Transfer rate constant, k_trans (slope of the curves in Fig. 2B), increases as the mole % PS in the membrane decreases. Data for EGFR(645–660) from Table 1 are shown as open triangles; data for MARCKS(151–175) peptide are shown as open circles. The rate constants for the interaction between Ca/CaM and the membrane-bound peptide, k₃ (see Section 7 of the Supporting Material), as calculated for EGFR(645–660) are shown as solid triangles; data for MARCKS(151–175) peptide are shown as solid circles.

Figure 4

Binding of a myristoylated basic (net charge +5) peptide corresponding to the N-terminal 18-residue region of Src (myr-Src peptide) to a PC/PS phospholipid vesicle increases the rate at which Ca/CaM can dissociate the EGFR(645–660) peptide from the surface by 10-fold. Kinetics for the transfer of acrylodan-EGFR(645–660) from vesicles with 17% PS to 3 μM Ca/CaM, with 0 (A), 0.5 μM (B), and 1 μM (C) myr-Src peptide also present in the mixing chamber. The solid curves designated b show single exponential fits with the following relaxation times: (A) τ = 100 ms, (B) τ = 23 ms, and (C) τ = 13 ms. Additional experimental details are provided in the Methods section of the Supporting Material.

Figure 5

Proposed mechanism of activation of EGFR. (A) Proposed structure of an EGFR monomer in a quiescent cell. The extracellular domain has the structure determined from x-ray analysis, as reviewed by Ferguson (6). The N-terminal portion of the intracellular JMD (residues 645–660 of the JMD), with the eight basic residues shown in blue, is postulated to bind electrostatically to the inner leaflet of the plasma membrane. The basic face of the kinase domain (see Fig. 2 of McLaughlin et al. (33)) for the electrostatic potential profile) is also postulated to bind to the negatively charged inner leaflet of the plasma membrane, but less strongly than the JMD. For simplicity, the C-terminal tail of the EGFR is not shown. (B) View from the cytoplasm. Two monomers form a dimer (of known extracellular structure) upon binding of EGF. Alternatively, binding of EGF mediates rotation of the subunits of a preformed dimer to the same extracellular structure. We hypothesize that in either case the intracellular basic JM regions come into proximity. Calmodulin is shown in the cytoplasm. (C) The electrostatic repulsion between the two highly charged JMDs weakens their attraction to the membrane. This causes one of the JMDs and its associated kinase domain to dissociate. (D) Alternatively, and in addition to the spontaneous dissociation shown in panel C, Ca/CaM binds to one of the JMD and rips it from the membrane, resulting in the conformation shown in panel E. The proximity of the second JMD greatly facilitates this process. Our measurements are consistent with the hypothesis that Ca/CaM can only compete with the membrane for binding of the JM region when the two JMDs are close together as in panel B. (E) Ca/CaM binding to the JMD of one of the EGFR molecules in a dimer causes it and the associated kinase region to dissociate from the membrane. Activation of the kinase domain by an allosteric mechanism (12) can proceed from the conformations shown in either panel E or C. In panel F, we assume activation proceeds from the conformation shown in panel C. (F) The N-terminal lobe (small lobe, colored green) binds to the C-terminal lobe of the membrane-bound kinase domain and is activated by an allosteric mechanism (12).