Author response: A molecular mechanism for the generation of ligand-dependent differential outputs by the epidermal growth factor receptor
Year: 2021
Type: peer-review
Abstract: Article Figures and data Abstract Editor's evaluation Introduction Results and discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract The epidermal growth factor receptor (EGFR) is a receptor tyrosine kinase that couples the binding of extracellular ligands, such as EGF and transforming growth factor-α (TGF-α), to the initiation of intracellular signaling pathways. EGFR binds to EGF and TGF-α with similar affinity, but generates different signals from these ligands. To address the mechanistic basis of this phenomenon, we have carried out cryo-EM analyses of human EGFR bound to EGF and TGF-α. We show that the extracellular module adopts an ensemble of dimeric conformations when bound to either EGF or TGF-α. The two extreme states of this ensemble represent distinct ligand-bound quaternary structures in which the membrane-proximal tips of the extracellular module are either juxtaposed or separated. EGF and TGF-α differ in their ability to maintain the conformation with the membrane-proximal tips of the extracellular module separated, and this conformation is stabilized preferentially by an oncogenic EGFR mutation. Close proximity of the transmembrane helices at the junction with the extracellular module has been associated previously with increased EGFR activity. Our results show how EGFR can couple the binding of different ligands to differential modulation of this proximity, thereby suggesting a molecular mechanism for the generation of ligand-sensitive differential outputs in this receptor family. Editor's evaluation This is an impressive study providing solid evidence for a molecular mechanism by which two related, high-affinity growth factors, binding in exactly the same site, can achieve differential signaling outputs through a dimerized receptor tyrosine kinase, and represents an important advance in the field. https://doi.org/10.7554/eLife.73218.sa0 Decision letter Reviews on Sciety eLife's review process Introduction Human EGFR can be activated by seven related ligands that generate different signaling outputs from the receptor. While the classical mechanism for receptor tyrosine kinase activation, ligand-induced dimerization, is essential for all these ligands to activate the receptor (Kovacs et al., 2015; Leahy, 2004; Lemmon et al., 2014), it cannot explain how the extracellular module of EGFR is able to respond differently to different ligands once the receptor is dimerized (Ronan et al., 2016; Singh et al., 2016; Wilson et al., 2009). The extracellular module of EGFR consists of four domains, denoted Domains I, II, III, and IV (Figure 1). Domains I and III sandwich the ligands between them (Garrett et al., 2002; Ogiso et al., 2002). Domain II, which bridges the ligand-binding domains, contains the ‘dimerization arm’, a loop that mediates the principal interaction with the other subunit in an activated dimer (Dawson et al., 2005). Domains I, II, and III form a compact unit that we refer to as the ligand-binding ‘head’, and Domain IV forms a rigid and elongated ‘leg’ that connects the ligand-binding head to the single transmembrane helix (Figure 1). Figure 1 Download asset Open asset A structural collage of a ligand-bound EGFR dimer. Left, a schematic diagram of ligand-bound dimeric EGFR. Right, structures of EGFR fragments, determined previously: a dimeric extracellular module bound to EGF (PDB code: 3NJP Lu et al., 2010), a transmembrane helix dimer (PDB code: 2M20 Endres et al., 2013), and an asymmetric dimer of kinase domains (PDB code: 2GS6 Zhang et al., 2006). Previous crystallographic analyses of fragments of the EGFR extracellular module have revealed different conformations of the ligand-binding head when bound to EGF, TGF-α, epiregulin, and epigen (Freed et al., 2017; Garrett et al., 2002; Lu et al., 2010; Ogiso et al., 2002). It appeared that these conformational differences might explain how different EGFR ligands generate different signals (Wilson et al., 2009; Scheck et al., 2012), but the significance of these structural differences are unclear because only fragments of the receptor were analyzed. For weakly binding ligands, such as epiregulin and epigen, the lifetime of the dimeric receptor is shortened relative to that for the tighter-binding ligands, such as EGF, and the reduction in dimer lifetime affects signaling output (Freed et al., 2017). However, EGF and TGF-α are both high-affinity ligands (Jones et al., 1999), and they also generate different outputs from EGFR. How this happens is still unclear. Cell-based studies using fluorescent reporters of EGFR conformation have shown that the binding of EGF and TGF-α results in different conformations of the intracellular juxtamembrane segments of the receptor (Scheck et al., 2012; Doerner et al., 2015; Lowder et al., 2015). The balance between alternative conformations of the juxtamembrane segments can be modulated by point mutations in the transmembrane helices (Sinclair et al., 2018) and oncogenic mutations in the kinase domains (Lowder et al., 2015). A more recent study reveals that simply switching the conformation of the juxtamembrane segments by mutation suffices to switch EGFR signaling outputs (Mozumdar et al., 2021). These studies demonstrate that the extracellular module, the transmembrane helix, the intracellular juxtamembrane segments, and the intracellular kinase domains are coupled allosterically, and that they respond differently to EGF and TGF-α. How the two ligands generate these differences remains unknown. To gain structural insight into this mechanism, we analyzed full-length human EGFR in complex with EGF and TGF-α by cryo-EM. Results and discussion Linkages between the extracellular modules and the transmembrane helices are not rigidly defined We purified full-length human EGFR expressed in a human HEK293S GnTI- cell line (Reeves et al., 2002; Qiu et al., 2009), and reconstituted the complex of EGFR with EGF into four membrane-mimetic environments: detergent micelles, lipid nanodiscs (Bayburt and Sligar, 2010), amphipols (Gohon et al., 2006), and peptidiscs (Angiulli et al., 2020; Carlson et al., 2018; Figure 2—figure supplement 1). Cryo-EM reconstructions of the full-length EGFR:EGF complex in all four reconstitutions look very similar, and only the extracellular modules are resolved in the cryo-EM densities (Figure 2). The lack of resolvable density for the transmembrane helices and the intracellular kinase domains suggests that the linkage between the extracellular modules and the transmembrane helices is not rigidly defined, consistent with 2D class averages of negative-stain EM images (Mi et al., 2011; Figure 2—figure supplement 1). Thus, mechanisms for ligand-dependent differential responses in EGFR that require strict stereochemical coupling between the extracellular ligand-binding modules and the transmembrane helices appear to be ruled out. Recently published cryo-EM analyses of intact insulin receptor and intact type 1 insulin-like growth factor receptor also did not resolve the transmembrane and intracellular modules (Li et al., 2019; Uchikawa et al., 2019; Zhang et al., 2020), suggesting that a lack of rigidity in the connection between the extracellular module and the transmembrane helices is a feature common to receptor tyrosine kinases. Figure 2 with 1 supplement see all Download asset Open asset Linkage between the extracellular module and transmembrane helices in full-length human EGFR is not rigid. (A, C–E) In each panel, the cryo-EM density map for a reconstituted full-length EGFR:EGF complex is shown at two contour levels. Note the lack of EM density for the transmembrane and kinase domains of EGFR at higher contour level (left) in all these complexes. (A) Full-length EGFR:EGF complex reconstituted in nanodiscs. Map shown at two contour levels, high (σ: ~ 5.8) and low (σ:~0.6). (B) Superposition of the cryo-EM map shown in A with a model for full-length EGFR embedded in the lipid bilayer, described previously (Arkhipov et al., 2013). (C) Full-length EGFR:EGF complex reconstituted in peptidiscs. Map shown at two contour levels, high (σ: ~ 5.9) and low (σ:~0.8). (D) Full-length EGFR:EGF complex reconstituted in amphipols. Map shown at two contour levels, high (σ: ~ 12.0) and low (σ:~1.2). (E) Full-length EGFR:EGF complex reconstituted in detergents. Map shown at two contour levels, high (σ: ~ 12.8) and low (σ:~1.0). Cryo-EM single-particle analysis reveals a range of conformations for the EGF-bound extracellular module After focusing the cryo-EM reconstructions on the extracellular module alone, the dataset for the full-length EGFR:EGF complex reconstituted in peptidiscs yielded the highest resolution reconstruction of the dimeric extracellular module, with an overall resolution of 2.9 Å (Figure 3A, Figure 3—source data 1; Figure 3—video 1). In this reconstruction, cryo-EM density for the Domain IV legs is less well resolved than for the heads. To better understand the origin of this apparent structural heterogeneity, we used 3D classification to further partition the entire dataset of 765,883 particles into 10 subclasses (Punjani et al., 2017). Cryo-EM reconstructions corresponding to individual subclasses resulted in overall resolutions ranging between 3.1 Å and 3.7 Å (Figure 3A, Figure 3—figure supplement 1). Figure 3 with 6 supplements see all Download asset Open asset Cryo-EM analysis reveals a range of conformations for the EGF-bound extracellular module. (A) Summary of the cryo-EM data processing workflow for the full-length EGFR:EGF complex reconstituted in peptidiscs. The two extreme conformations of the ensemble, the juxtaposed and the separated, are highlighted by red dotted circle. The intersubunit rotational angles and the separations between the tips of the extracellular module are summarized below each subclass. See Materials and methods for more details. (B–D) Two orthogonal views of the cryo-EM maps for three representative states of the conformational ensemble of the EGF-bound extracellular module: the tips-juxtaposed conformation (B), an intermediate conformation (C), and the tips-separated conformation (D). These conformations differ from one another in the rotations between the ligand-binding heads, as well as in the distances between the two Thr 614 residues located at the membrane-proximal ends of the Domain IV legs. Figure 3—source data 1 Statistics of cryo-EM data collection and structure determination. https://cdn.elifesciences.org/articles/73218/elife-73218-fig3-data1-v2.doc Download elife-73218-fig3-data1-v2.doc The resulting structures for different subclasses differ from one another in the extent of a ‘scissor-like’ rotation between the two ligand-binding heads of the dimeric extracellular module (Figure 3A, Figure 3—videos 1–3). The rotation between the two ligand-binding heads is due to a rigid-body rotation of each head about an axis parallel to the dimerization arms, and is correlated with a slight twist in the dimerization arm of each EGFR subunit (Figure 3B–D, Figure 3—videos 1–3). The extent of this intersubunit rotation can be estimated by defining a dihedral angle corresponding to rotation about a virtual bond connecting the Cα atoms of Thr 249 in each EGFR subunit (we number amino acid residues in EGFR without including the 24-residue signal sequence). The positions of the Cα atoms of Ile 190 in each subunit are used to further specify the dihedral angle, and the value of this dihedral angle ranges from ~10o to ~25o in the 10 representative structures (Figure 3A). The scissor-like rotation between ligand-binding heads in the cryo-EM structures is reminiscent of differences noted previously between crystal structures of fragments of EGFR extracellular modules bound to different ligands (Liu et al., 2012). At one extreme, the EM-derived structure with the largest intersubunit rotation (~25o) resembles closely the crystal structure of the ligand-binding head of EGFR bound to EGF (PDB code: 3NJP (Lu et al., 2010; Ogiso et al., 2002), root-mean-square deviations (RMSDs) in Cα positions calculated over EGFR residues 1–501: 0.8 Å, Figure 4A). The EM-derived structure with the smallest intersubunit rotational angle (~10o) resembles the crystal structure of the EGFR ligand-binding head bound to TGF-α (PDB code: 1MOX (Garrett et al., 2002), RMSDs in Cα positions calculated over EGFR residues 1–501: 1.5 Å, Figure 4B). Thus, the partitioning of the cryo-EM data into subclasses reflects an intrinsic conformational variability of the ligand-bound EGFR extracellular module. Importantly, the cryoEM results demonstrate that the conformations ascribed previously to specific binding of EGF or TGF-α to the receptor are actually part of an ensemble of conformations that are readily accessible to EGFR when bound to either one of these ligands. Figure 4 Download asset Open asset Comparison of Cryo-EM structures of EGFR:EGF complex in two extreme conformations with previous crystal structures of the ligand-binding head of EGFR bound to EGF and TGF-α. In each panel, a cryo-EM structure of the extracellular module of the full-length EGFR:EGF complex (juxtaposed or separated) is compared to a crystal structure of the ligand-binding head of EGFR bound to EGF or TGF-α. In each comparison, structures are superimposed on the right-hand EGFR subunits (gray). (A) Top, the cryo-EM structure of juxtaposed conformation of EGFR:EGF (cyan) is compared to a crystal structure of the ligand-binding head of EGFR bound to EGF (pink, PDB code: 3NJP). Bottom, the cryo-EM structure of juxtaposed conformation of EGFR:EGF (cyan) is compared to a crystal structure of the ligand-binding head of EGFR bound to TGF-α (light purple, PDB code: 1MOX). Note that the cryo-EM structure of the juxtaposed conformation of the EGFR:EGF complex resembles closely the crystal structure of the ligand-binding head of EGFR bound to EGF (PDB code: 3NJP), but deviates from the crystal structure of the ligand-binding head of EGFR bound to TGF-α (PDB code: 1MOX). (B) Top, the cryo-EM structure of separated conformation of EGFR:EGF (yellow) is compared to a crystal structure of the ligand-binding head of EGFR bound to EGF (pink, PDB code: 3NJP). Bottom, the cryo-EM structure of separated conformation of EGFR:EGF (yellow) is compared to a crystal structure of the ligand-binding head of EGFR bound to TGF-α (light purple, PDB code: 1MOX). Note that the cryo-EM structure of the separated conformation of the EGFR:EGF complex resembles closely the crystal structure of the ligand-binding head of EGFR bound to TGF-α (PDB code: 1MOX), but deviates from the crystal structure of the ligand-binding head of EGFR bound to EGF (PDB code: 3NJP). Different dimeric conformations of the ligand-bound extracellular module have different separations between the tips of the two legs The two Domain IV legs in the dimer behave differently when the scissor-like rotational angle between the ligand-binding heads changes. The structure of the Domain IV leg from a previously determined crystal structure of the EGFR extracellular module (PDB code: 3NJP Lu et al., 2010) can be docked into our cryo-EM densities (Figure 3—figure supplement 2A, B). This allowed us to determine the position and orientation of the Domain IV legs associated with each representative conformation from the ensemble revealed by cryo-EM. When the rotational angle between the two heads is at its largest value (~25o), the two legs adopt similar conformations, and the membrane-proximal tips of the two legs are juxtaposed, with a separation of ~5 Å between the Cαatoms of the last residue in the legs (Thr 614) (the ‘juxtaposed’ conformation, Figures 3B and 5). When the rotation between the two subunits is small, maintenance of the same conformation in both legs would lead to steric clashes between the two legs (Figure 3—figure supplement 2C). The clashes are relieved by one of the two Domain IV legs undergoing a hinge rotation around the connection between the head and the leg, and we refer to this leg as ‘flexible’. This hinge rotation causes the flexible leg to swing upwards, closer to the ligand-binding head, increasing the separation between the membrane-proximal tips of the two Domain IV legs in the dimer. When the angle between the heads is at its minimal value (~10o), the flexible leg swings maximally toward the ligand-binding head, increasing the distance between the membrane-proximal tips of the extracellular module by nearly 15 Å (the ‘separated’ conformation, Figures 3D and 5). Figure 5 Download asset Open asset Comparison between the cryo-EM structures of the juxtaposed and the separated conformations of the extracellular module of the EGFR:EGF complex. (A) The cryo-EM structures representing the juxtaposed conformation (cyan) and the separated conformation (yellow) are superimposed on the right-hand EGFR subunits which have similar Domain IV leg conformations (light gray). A vector (dark gray) is used to indicate the displacement of Cα atoms between the two structures, pointing from the juxtaposed structure to the separated structure. (B–E) Side-by-side comparison between the juxtaposed conformation and the separated conformation of the extracellular module of the EGFR:EGF complex. Molecular surface of the cryo-EM models of the juxtaposed and separated conformations are shown from the side (B), the front (C), the top (D), and the bottom (E). EGF molecules are shown in slate blue. The same EGFR subunit used for superposition in (A) is colored gray and the other EGFR subunit is light pink. The tips-separated conformation of the dimeric extracellular module can couple more readily to the N-terminal dimer of the transmembrane helices The EGFR transmembrane helix contains two dimerization motifs, one located toward the N-terminal end of the transmembrane helix, near the extracellular face of membrane, and the other located toward the C-terminal end (Mendrola et al., 2002). These dimerization motifs are involved in close association between the transmembrane helices in alternative structures defined by NMR, in which the helices are splayed apart at either the N-terminal or the C-terminal ends (Bocharov et al., 2016; Bocharov et al., 2017; Endres et al., 2013). To understand how these alternative dimeric configurations of the transmembrane helices connect to the different conformations of the extracellular module, with different separation between the tips of the Domain IV legs, we analyzed two structures of EGFR transmembrane helix dimers determined previously by NMR. One structure shows N-terminal association of the transmembrane helices (PDB code: 5LV6 Bocharov et al., 2017) and the other shows C-terminal association (PDB code: 2M0B Bocharov et al., 2016; Figure 6A). Figure 6 Download asset Open asset The tips-separated conformation of the extracellular module can couple more readily to the N-terminal dimer of the transmembrane helices. (A) Models of the transmembrane helices in two configurations based on previous NMR structures. One of the structures in the NMR-based ensemble for the N-terminal dimer configuration (PDB code: 5LV6) is shown in yellow (chain A) and gray (chain B), and the structure for the C-terminal dimer configuration (PDB code: 2M0B) is shown in cyan (chain A) and gray (chain B). Shown in spheres are superimposed Cα atoms of the residue Thr 614, located at the end of the extracellular module, from the 20 structures in the NMR-based ensemble for the N-terminal dimer configuration (left) and for the C-terminal dimer configuration (right), respectively. The black dotted square indicates the linker region that is imposing a constraint on the orientation of the linkers. In the N-terminal dimer configuration of the transmembrane helices (left), the linkers from each subunit are oriented away from one another. In the C-terminal dimer configuration of the transmembrane helices (right), the linkers are oriented toward each other. (B) Enlarged view of the linker located at the junction with the transmembrane helices, zoomed in from the areas highlighted by dotted black square in A, showing well-defined conformations for this region in both NMR structures. (C), Probability distribution of the distances between the Cα atoms of the two Thr 614 residues, sampled every 250 picoseconds over 2.5 μs of MD simulations of the N-terminal dimer configuration (purple) and the C-terminal dimer configuration (cyan) of the transmembrane helices. Initial structures of the transmembrane helices dimers with the linker are shown as cartoons. Residues Thr 614 located at the end of the extracellular module are shown as spheres. Also shown as spheres are residues Ile 619 and Ala 623, located at the junction between the linker and the transmembrane helix, and the interaction between these two residues is maintained over 2.5 μs MD simulations (see E and F). (D) Free-energy change of the system when the distance between the Thr 614 residues from each subunit is driven from 27 Å to 8 Å, when the transmembrane helices are in the N-terminal dimer configuration. Results from three independent umbrella-sampling calculations are shown in green, orange, and purple. The dotted gray line indicates the value of kBT at room temperature, 2.479 kJ/mol. (E and F) Probability distribution of the distances between the Cα atoms of Ile 619 and Ala 623, sampled every 250 picoseconds over 2.5 μs of MD simulations of the N-terminal dimer configuration (E) and the C-terminal dimer configuration of the transmembrane helices (F). The seven-residue linker connecting the extracellular module and the transmembrane helices (Asn 615-Ser 621) is modeled in both NMR structures. The three C-terminal residues of the linker (Ile 619-Ser 621), located at the junction with the transmembrane helix, adopt well-defined conformations in both NMR structures. The hydrophobic sidechain of Ile 619 packs against the sidechain of Ala 623 in the first turn of the transmembrane helix. Ile 619 is followed by a 310-helical turn spanning Pro 620 to Ala 623 (Figure 6B). These structural constraints cause the linker-transmembrane helix junction to orient parallel to the membrane surface in both structures (Figure 6A). As a consequence, in the N-terminal dimer of the transmembrane helices, the two linkers are oriented away from each other, with the two Ile 619 residues separated by ~20 Å. In contrast, the C-terminal dimer has the two linkers oriented toward each other, with the distance between the two Ile 619 residues at ~10 Å. Thr 614 is at the junction between Domain IV leg of the extracellular module and the linker. In the ensemble of NMR structures for the N-terminal dimer of the transmembrane helices (Bocharov et al., 2017), the two threonine residues (Thr 614) are separated by an average distance of ~35 Å. In comparison, the separation between the two threonine residues is ~20 Å in the ensemble of structures for the C-terminal dimer of the transmembrane helices (Bocharov et al., 2016; Figure 6A). These observations suggest that the junctions between the extracellular module and the transmembrane helices are more separated in the N-terminal dimer of the transmembrane helices than the connection in the C-terminal dimer. We used molecular dynamics simulations to explore the constraints on the separation between the two threonine residues at position 614 in the linker, starting either the N-terminal dimer configuration (PDB code: 5LV6 Bocharov et al., 2017) or the C-terminal dimer configuration (PDB code: 2M0B Bocharov et al., 2016). Following equilibration (see Materials and methods for full details), we initiated 25 independent molecular dynamics trajectories for both configurations, each 100 ns long, for a total simulation time of 2.5 µs for each configuration. The configuration of the transmembrane helices remains close to the starting configuration over this timescale. The conformation of the junction between the linker and the transmembrane helices is also very stable in both sets of simulations, with the sidechain of Ile 619 remaining associated with Ala 623 over the course of the simulations (Figure 6E and F). In the simulations of the N-terminal dimer the linkers explore a range of configurations but are oriented such that they point away from one and the distance between Thr 614 from each subunit ranges between 20 Å and Å (Figure In the simulations starting from the C-terminal dimer the Thr 614 residues from each subunit are closer to each other, distances between Å and Å (Figure These results further the that the junction between the linker and the transmembrane helices a constraint on the orientation of the linkers. When the transmembrane helices are in the N-terminal dimer this linker a connection to the tips-separated conformation of the extracellular module, and when the transmembrane helices are in the C-terminal dimer a connection to the tips-juxtaposed conformation of the extracellular module is We the change of the system when the distance between the Thr 614 residues from each subunit is driven from 27 Å (the in the distribution of distances between these residues in simulations of the N-terminal see Figure to 8 Å (the corresponding in simulations of the C-terminal see Figure when the transmembrane helices are in the N-terminal dimer configuration. We used umbrella-sampling simulations and with 1 Å apart a total of 20 and starting structures for each from a molecular dynamics independent umbrella-sampling calculations were with the starting structures for each from different molecular dynamics These calculations the of not when each sampled ns of three umbrella-sampling simulations similar results (Figure with the when the distance between the Cα atoms of Thr 614 residues from each subunit is Å. this distance to 8 Å to the of the system increasing by kBT at room temperature, Figure This indicates that in the N-terminal dimer it is for the linkers between the extracellular modules and the transmembrane helices to conformations the distance between the Thr 614 residues is Thus, when the transmembrane helices are in the N-terminal dimer it is that the tips-separated conformation of the extracellular module with the tips of Domain IV at a is over the tips-juxtaposed oncogenic mutation in the kinase the tips-separated conformation of the extracellular module of the transmembrane helices through the N-terminal dimerization is associated with increased kinase et al., Lu et al., 2010; et al., 2018; et al., our structural analysis of the connection between the extracellular module and the transmembrane helices suggests that the tips-separated conformation of the extracellular module is associated with a more conformation of the kinase domains, through its coupling to the N-terminal dimer of the transmembrane helices. To this we an oncogenic of EGFR in in which a residue in the loop of the intracellular kinase is by this is residue when the signal is The loop is a at the of the kinase and the mutation the of the asymmetric dimer of the kinase domains, to of EGFR et al., et al., 2012). We that this mutation the of EGFR (Figure supplement consistent with previous results et al., 2013). We carried out cryo-EM analysis of full-length bound to EGF, using the same as for EGFR (Figure 3—figure supplement 3A). The two structures representing the tips-juxtaposed and the tips-separated conformations of the complex were determined at resolution of Å and Å, (Figure 3—source data 1). The overall structures of the tips-juxtaposed and tips-separated conformations of the complex are very similar to for the corresponding complex of the receptor. The of the apparent when the of the cryo-EM densities in the Domain IV legs for the and EGFR. For the tips-juxtaposed conformations, the cryo-EM densities are of (Figure important when the tips-separated conformations for EGFR and the oncogenic the oncogenic mutation a residue that is located in the intracellular kinase this mutation the flexible leg in the tips-separated conformation of the extracellular module on the other side of the membrane, compared to EGFR. The cryo-EM density for the flexible leg is better resolved in the complex than for with
Source: eLife
Language: English
Topic: HER2/EGFR in Cancer Research
Subfield: Oncology
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $2,500
OpenAlex ID: W4200453481