The crystallographic structures of functional fragments of ErbBs have provided excellent insights into the geometry of growth factor binding and receptor dimerization. By placing together receptor fragments to build structural models of entire receptors, we expect to understand how these enzymes are allosterically regulated; however, several predictions from these models are inconsistent with experimental evidence from cells. The opening of this gap underlines the need to investigate intact ErbBs by combining cellular and structural studies into a full picture.
The EGFR (epidermal growth factor receptor) family
EGFR (ErbB1) is the founding member of the ErbB family. ErbBs are prototypical examples of the growth factor RTK (receptor tyrosine kinase) superfamily, which also comprises 18 subgroups of cell-surface receptors for many growth factors, cytokines and hormones . Reflecting the complexity of the organisms, the EGFR family has evolved from one receptor/one ligand in Caenorhabditis elegans, through one receptor/multiple ligands in Drosophila, to a family comprising four receptors (EGFR/ErbB1 and ErbB2–ErbB4, known as HER1–HER4 respectively in humans) and 13 extracellular ligands in mammals . ErbBs are key regulators of cell–cell inductive processes and cell fate . Their function is to transmit growth factor signals from the outside to the inside of the cell where changes in gene expression allow the cell to respond to new circumstances. Because of their physiological importance, ErbBs are ubiquitously expressed in many tissues, including skin, lung, heart, liver, breast and prostate .
Since the discovery that an oncogenic erythroblastosis retrovirus (v-erbB) encoded a mutated homologue of EGFR , intense research has shown that, in adulthood, excessive ErbB signalling upsets the balance between cell growth and apoptosis, resulting in the development of a wide variety of solid tumours . In particular, the expression and/or activation of EGFR and ErbB2 are altered in many tumours of epithelial origin, and clinical studies indicate that EGFR and ErbB2 have important roles in tumour aetiology and progression . The ErbB family is therefore a key target of the pharmaceutical industry, and several cancer drugs targeting EGFR and ErbB2 are in different stages of pre-clinical and clinical trials.
Structural insights into the vertebrate EGFR family
The primary structure of EGFR is shared by all members of the family. EGFR is a single 170 kDa polypeptide chain (1186 amino acids) containing an extracellular ligand-binding domain that is connected to the cytoplasmic domain by a single TM (transmembrane) helix (Figure 1). The cytoplasmic domain contains a conserved tyrosine kinase core .
A model of the 2:2 complex of EGF and EGFR
Ligand binding induces the dimerization of ErbB ectodomains (Figure 1). This leads to the formation of an asymmetric dimer by the two associated intracellular kinases  that is stabilized by the inner JM (juxtamembrane) region [8,9]. Active kinases then proceed to phosphorylate tyrosine residues in the C-terminal tail of the receptor.
Over the last decade, crystallographic studies of ErbB ectodomain fragments have shown that receptor dimerization also involves major extracellular structural rearrangements. Unliganded EGFR, ErbB3 and ErbB4 monomers are held in a closed conformation by an intramolecular tether formed by loops in subdomains II and IV [10–12]. In ligand-occupied receptor dimers, the intramolecular tether is broken, and the receptor is opened into an extended conformation which interacts with another monomer to form a back-to-back dimer [13,14]. These structures showed that receptor dimerization was not, as previously thought, mediated directly by the binding of ligand, but was achieved exclusively via receptor–receptor contacts and suggested a mechanism for how unliganded receptor monomers are maintained in an ‘autoinhibited’ configuration.
Interestingly, unliganded ErbB2 has an extended configuration that resembles the structure of ligand-bound ‘activated’ EGFR . This may explain the unique properties of ErbB2, which has no known ligand and can cause cell transformation (and tumorigenesis) by simple overexpression. The latter appears to force the equilibrium towards spontaneous homodimer formation, which leads to receptor activation in the absence of ligands. This is the situation observed in a variety of human cancers .
ErbB dimers and intracellular signalling
A defining characteristic of the EGFR family is that only EGFR and ErbB4 can bind ligands and also signal autonomously via homodimerization and transactivation of their tyrosine kinases . In contrast, ErbB2 and ErbB3 are not autonomous. As outlined above, ErbB2 lacks the intrinsic ability to interact with known ligands, whereas the kinase of ErbB3 is defective . ErbB2 and ErbB3 can therefore only initiate signals through heterodimer formation. Accounting for heterodimers, the number of possible functional inter-receptor complexes involving in normal signalling increases from four (homodimers) to eight .
The pattern of phosphorylated tyrosine motifs is specific to each ErbB . The combinatorial variety of homo/hetero-ErbB1–ErbB4 complexes therefore provides a mechanism by which different cell responses can be induced via recruitment of different combinations of signalling effectors to phosphorylated ErbB complexes, activating intracellular pathways for signal attenuation (receptor desensitization, down-regulation, endocytosis and trafficking), amplification and processing . This may explain how subtle differences in the expression pattern of ErbB1–ErbB4 and their ligands may ultimately regulate development and homoeostasis in many tissue types.
Early in situ studies of the quaternary structure of EGFR complexes
About two decades separated the sequencing of the EGFR from the first high-resolution structures of EGFR fragments. A wealth of data was nevertheless accumulated during this time that yielded the first signalling pathway that could be traced without obvious gaps from the activated EGFR kinase domain at the plasma membrane, through modular domains and multiple isoforms of interconnected effector units, followed by docking proteins and signalling cascades, and ultimately to transcription in the nucleus .
In the absence of crystal structures, it proved more difficult to obtain a clear picture of the initial steps of EGFR activation on the extracellular side of the plasma membrane. However, substantial progress was made by early covalent cross-linking experiments which showed that both ligand binding and receptor dimerization were critical for the activation of EGFR [20–24], results that were backed by rotational diffusion studies of EGF (epidermal growth factor) complexed to cell-surface EGFR [25,26]. Functional analysis, X-ray scattering, CD, fluorescence spectroscopy and electron microscopy in turn showed that domain III is a major ligand-binding domain in EGFR, that EGF binds the EGFR extracellular region in solution with a stoichiometry of 2:2  and that EGF induces a conformational change in the soluble EGFR ectodomain  and stimulates its oligomerization . All of these results were later backed by crystallographic data.
Confirming the existence of EGFR dimers before stimulation by exogenous ligand, predicted two decades ago in the context of the ligand-induced allosteric transactivation mechanism outlined above, was also an achievement that pre-dated the first EGFR structures. Sucrose density gradient centrifugation and chemical cross-linking demonstrated the occurrence of unliganded EGFR dimers in A431 cells, an established human epidermoid cancer cell line overexpressing cell-surface EGFR to a level of up to 2×106 receptors/cell [22,30,31]. EGFR overexpression facilitated EGFR detection in photon-hungry techniques such as FRET (Förster resonance energy transfer) imaging, which is a very sensitive measure of intermolecular distances in the range 2–8 nm . Early FRET experiments were able to report preformed EGFR dimers in these cells, at both the ensemble and the single-molecule level [33–35] from the short FRET-derived distances (5–8 nm) found between receptor-bound EGF ligands labelled with FRET donor and acceptor fluorophores. However, these studies did not exclude the possibility that EGFR dimers were induced by overexpression or autocrine EGFR ligands provided by the A431 cells .
Currently, no crystal structures of unliganded extracellular domain dimers for human EGFR exist, and therefore the nature of preformed dimers is unclear. FRET results also suggested that EGF binding was followed by conformational changes in the pre-formed EGFR dimer [33–35,37,38]. The unliganded tethered conformation of the ectodomain is normally observed as a monomer, but the tethered ErbB3 reveals an antiparallel dimer (PDB code 1M6B) with a strong interface. The dimer is generated by crystal symmetry, with domain I of one molecule in contact with domain III of the other and vice versa [15,39].
Structural studies fail to explain FRET results in cells
The preferred (back-to-back) crystallographic dimer shows N-terminal inter-EGF distances of >11 nm (Figure 1). This distance is not detectable by FRET, even after applying extreme deformations to this crystal structure; for example, by changing the angle between domains I and III and domain II and/or applying extreme perturbations along low-frequency normal modes. However, FRET between fluorophores bound to the N-termini of receptor-bound EGF ligands is found ubiquitously in cells [33–35,40], suggesting that other interfaces between receptor monomers must also be possible.
Pioneering work by Clayton et al.  using fluorescence correlation microscopy and FRET in intact BaF/3 cells (which do not overexpress EGFR), showed tetramers of activated cell-surface EGF–EGFR complexes and very short distances (<4 nm) between EGFR-bound ligands. These distances were later confirmed by single-molecule FRET imaging . These results set the foundation for two new models of EGFR activation that sought to explain the very short distances between receptor-bound ligands. In one model, the tetramers were generated through side-by-side contacts between two adjacent back-to-back dimers; however, there is no crystallographic evidence yet for this arrangement. In the second model (Figure 2), tetramers were made of two back-to-back dimers joined by a weak, asymmetric ‘head-to-head’ interface seen in crystal structures [39,41]. This interface also results in short distances between the N-termini of the two bound EGF molecules of <4 nm. Molecular dynamics simulations showed that the head-to-head interaction stabilizes appreciably when the tetramer is relaxed on the membrane [its interface area increased from 443 Å2 (1 Å=0.1 nm) seen in the crystal structure to 604 Å2] . The two dimers in this tetramer remained stable, but lost the approximate two-fold symmetry of the crystal structure. The average dimerization interface area rose from 1197 Å2 to 1483 Å2, largely as a result of additional interactions between the N-terminus and the dimerization arm, whereas the two ligands buried 1166 Å2 and 1211 Å2 at distinct binding sites.
The back-to-back–head-to-head–back-to-back tetramer on the membrane
FRET-derived in situ structural data explains the ligand-binding properties of EGFR
Possibly the biggest disappointment of the ErbB crystal structures is that they did not explain the characteristic concave-up Scatchard plots first reported for EGF binding to cell-surface receptors three decades ago [42,43]. These plots were initially interpreted as indicating the presence of two receptor populations: a small minority of high-affinity receptors with dissociation constants (Kd) of <1 nM that mediate most signalling events, and a majority of low-affinity receptors with Kd values of >1 nM [44,45]. These populations were believed to be the tethered monomers and extended dimers revealed by the crystallographic structures; however, in this model, stabilization of the extended dimer configuration by one ligand molecule would facilitate the binding of a second ligand to the remaining receptor in the dimer, resulting in positive co-operativity and concave-down Scatchard plots .
Using an inducible expression system to control EGFR expression and global modelling of EGF-binding data as a function of receptor number, Macdonald and Pike  showed that EGF-binding heterogeneity could be explained by negative co-operativity in EGFR dimers. However, this requires that ligand binding to one subunit of the EGFR dimer decreases the ligand affinity of the other subunit. This would, in turn, require the interactions between ligand and the two subunits of the EGFR dimer to be asymmetric, inconsistent with the symmetry observed in the back-to-back structure (Figure 1). A series of crystallographic structures recently revealed that in the dEGFR (Drosophila EGFR), the extracellular domain asymmetry is induced by the binding of the first ligand, which structurally restrains the unoccupied binding site, reducing the affinity for binding of the second ligand . Interestingly, concave-up Scatchard plots were observed in preparations of the isolated extracellular region of the dEGFR, in direct contrast with its human counterpart, in which concave-up plots are observed only when ligands bind to full-length receptors in cells. These differences suggest that other receptor regions and/or unknown cellular components must be involved in the regulation of ligand affinity in EGFR.
Using a combination of FRET microscopy and Monte Carlo and molecular dynamics simulations, Tynan et al.  showed recently a high-affinity ligand-binding human EGFR conformation consistent with the extracellular region aligned flat on the plasma membrane (Figure 3). Interestingly, the asymmetry of this structure shares key features with that of dEGFR , suggesting that the structural basis for negative co-operativity is conserved from invertebrates to humans, but that, in human EGFR, the extracellular region asymmetry requires interactions with the plasma membrane.
EGFR ectodomain dimer with two bound ligands
The discrepancies between crystallography and data from cells highlight the need to study the structural properties of whole receptors if we are to understand the details of their function. Unlike crystallized ErbB fragments, intact ErbBs in cells are influenced by the action potential, hydrophobic environment and geometrical restrictions of the two-dimensional plasma membrane. To understand the input layer in the ErbB network, we need to describe in situ the state of basic ligand–ErbB1–ErbB4 units to reveal their structure and topology and determine how steric mechanisms transduce external cues via JM and TM domains.
Signalling 2011: a Biochemical Society Centenary Celebration: A Biochemical Society Focused Meeting held at the University of Edinburgh, U.K., 8–10 June 2011. Organized and Edited by Nicholas Brindle (Leicester, U.K.), Simon Cook (The Babraham Institute, U.K.), Jeff McIlhinney (Oxford, U.K.), Simon Morley (University of Sussex, U.K.), Sandip Patel (University College London, U.K.), Susan Pyne (University of Strathclyde, U.K.), Colin Taylor (Cambridge, U.K.), Alan Wallace (AstraZeneca, U.K.) and Stephen Yarwood (Glasgow, U.K.).
We gratefully acknowledge funding from the Biotechnology and Biological Sciences Research Council [grant number BB/G006911/1].