Although the EGFR (epidermal growth factor receptor) was discovered over 30 years ago, its mechanism of activation is still the subject of intense study. There are many published studies on the mechanism of EGFR activation and regulation, including biochemical and biophysical analyses and crystallographic structures of EGFR in different activation states and conformations, mutated at various amino acids or bound to different pharmacological inhibitors. The cumulative biochemical, biophysical and structural data have led to a nearly complete account of the mechanism of activation of EGFR. The role of the JXM (juxtamembrane) domain in EGFR structure and activity has only recently begun to be elucidated through biochemical, biophysical and structural studies. In the present article, I review the studies that have highlighted the role of the JXM domain in EGFR activation.
EGF (epidermal growth factor) and its corresponding receptor, EGFR (EGF receptor) (or ErbB1), were initially discovered by Stanley Cohen's group over 30 years ago (reviewed in ). The ErbB denomination was derived from the similarity of the EGFR gene sequence to the ErbB transforming gene of the avian erythroblastosis virus . The role of EGFR in cancer is now well established and several cancer therapies, both approved and in the pipeline, target EGFR [2–7]. EGFR is amplified or overactive in many types of epithelial cancers, including pancreatic, breast, brain, non-small-cell lung and colorectal cancers, and head and neck squamous cell carcinoma [4,5,7]. Aberrant EGFR signalling in cancer has been shown to lead to increased tumour proliferation and tumour growth rates, anchorage-independent growth and enhanced metastasis formation .
EGFR is a member of the ErbB family of RTKs (receptor tyrosine kinases), of which there are four members, ErbB1/HER-1/EGFR, ErbB2/HER-2, ErbB3 and ErbB4. ErbB signalling activates cellular pathways, leading to a plethora of cellular processes, including cell growth, proliferation and survival . In general, the ErbB receptors are activated by ligand binding and dimerization which induces or increases the activity of the intracellular kinase domain (Figure 1A). One exception to this activation mechanism is ErbB3, which exhibits extremely low kinase activity and is therefore regarded as a decoy receptor, acting only as a dimerization partner without ever becoming activated itself . ErbB receptors have the ability to signal from various homo- and hetero-dimers, with ErbB2 being the preferential dimer partner, and downstream signalling is altered depending on the nature of the homo/hetero-dimer . ErbB2, which has no known ligands, only requires dimerization for activity since its extracellular domains exist in the open active state .
Schematic representation of the mechanism of EGFR activation, organization of EGFR domains and homology of the JXM domains across the ErbB family
Each ErbB family member contains extracellular domains (I–IV) with a ligand-binding domain, a transmembrane domain, JXM (juxtamembrane) domain, kinase domain (with a C- and N-lobe) and a C-terminal tail (Figure 1B). The ICD (intracellular domain) consists of the JXM domain, kinase domain and the C-terminal tail. The ErbB JXM domain is proximal to the inner leaflet of the plasma membrane, adjacent to the transmembrane domain, and has two major segments: JMA (EGFR residues 645–663) and JMB (EGFR residues 664–682) (Figure 1). There are amino acids in the JXM region that are conserved across the ErbB family (Figure 1C), which indicate possible functional roles. Many studies have explained the various mechanisms by which the ErbB family of receptors become activated and are reviewed elsewhere [2,3,7,8]. The focus of the present review is to highlight the role of the JXM domain in the activation and regulation of EGFR activity.
The structure of JXM domain in the active EGFR conformation
A study by Zhang et al.  showed that, in the active conformation, EGFR exists as an asymmetrical dimer, where one monomer (the donor) has an activated kinase domain and the other (the receiver) is being phosphorylated (Figure 1A). The receiver is orientated with its N-lobe towards the C-lobe of the donor and allosterically activates its kinase domain. To access this fully active conformation, the restructuring caused by ligand binding and dimerization of the extracellular domain must be translated through the transmembrane and JXM domains to the kinase domain.
Until recently, the published structures of the ICD of EGFR did not include the JXM region, hence the interactions made by the JXM domain were structurally uncharacterized. However, the structure of the ICD of ErbB4 in an active asymmetrical conformation included the JXM region and showed that the JMB region of the receiver forms a latch over the C-lobe of the donor . It was therefore posited that the EGFR JXM region would make similar structural contacts in the active conformation. In a study by Red-Brewer et al. , the crystal structure of the ICD of EGFR, including the JXM region, revealed that the JXM latch formed between the JMB region and the opposing C-lobe was present in the EGFR structure, as observed previously with ErbB4. This was the first report of the structural mechanism which partially accounts for the activity brought about by the presence of the JXM domain of EGFR (discussed in the next section).
In a study by Jura et al. , it was explicitly shown that the JMB latch is formed in an acceptor to donor fashion, through mutational experiments. In the previous structural study by Zhang et al. , EGFR variants were developed by introducing point mutations, such that one construct could behave exclusively as a donor and the other exclusively as the receiver. Each donor or receiver construct is only able to induce kinase activity when mixed together with each other or the wild-type EGFR construct. In the context of this donor–receiver system of constructs, a point mutation (R953A) was introduced in the C-lobe of the kinase, which disrupts latch formation, by disruption of two potential ion pairs . The R953A mutation was introduced into either the receiver or donor impaired construct and expressed with the corresponding wild-type donor or wild-type receiver construct . As a result, the receiver-impaired EGFR construct lost activity when the latch-inhibiting mutation was present, whereas the donor-impaired EGFR construct retained activity, indicating that the JMB latch direction was from acceptor to donor.
The study by Jura et al.  included a crystal structure of the core kinase domain (residues 672–998) which revealed an autoinhibited EGFR conformation as a novel symmetrical dimer. The structure further supported the role of the JXM domain. First, the site where the JMB latch binds on the donor C-lobe is occluded in an inhibitory manner by the AP-2 (activating protein 2) helix of the C-terminal region from the opposite monomer, which binds in a strikingly similar mode as the ErbB4 JXM latch. In a docking experiment, this ‘latch’-binding interaction was shown to likely be present in the inhibitory interaction between EGFR and an adapter protein, Mig6 [12,13]. Also, a six-residue sequence within EGFR-binding motif of Mig6 (residues 323–328: EPLSPS) displays strong sequence homology with a portion of the JMB latch region (residues 666–671: EPLTPS). These results elucidate the context of the JMB latch interaction with the C-lobe in the activation mechanism of EGFR. It also refutes any potential autoinhibitory role, which is a common function of the JXM region in the activation of RTKs belonging to families other than ErbB .
Zhang et al.  resolved the structure of the asymmetrical EGFR ICD dimer in the active conformation using an EGFR construct with an activating mutation. In contrast, the construct used to crystallize the intracellular and JXM domains of EGFR contained a kinase-inactivating K721M mutation  and yet EGFR adopts the active asymmetrical dimer conformation. The authors emphasize that this is likely to be due to the extra interactions contributed by the JXM region which were not present in the previous structures, where EGFR adopted a symmetrical dimeric inactive conformation .
The crystal structure of the complete EGFR ICD revealed that a portion of the JMA domain (residues 653–663) contained a helical segment, whereas the remaining portion (residues 645–653) was relatively disordered. The study by Jura et al.  included an NMR structure of the JMA region, which confirmed that the region contained a helix. Furthermore, by connecting two JMA segments with a long flexible linker, they showed that the helical regions interact with each other in an antiparallel fashion. Jura et al.  reconciled this JMA region structure with the asymmetrical dimer conformation using structural modelling and concluded that one face of the JMA dimer probably interacts with C-lobe of the donor, whereas the opposite face probably interacts with the plasma membrane, as suggested by other studies [16–19].
EGFR kinase domain activity is dependent on the presence of the JXM domain
Recent studies have shown that the presence of the JXM domain is necessary for the full catalytic activity of EGFR. In a study by Thiel and Carpenter , it was shown that an intracellular EGFR construct which lacked the JXM domain was approximately 95% less active compared with the full-length ICD which contained the JXM domain. In a later study from the Carpenter laboratory, scanning alanine mutagenesis experiments showed that mutations at 36 different residues within the JXM of ICD constructs of EGFR reduced activity by more than 50% . Two of these point mutations (V665A and L680A) were shown to significantly reduce activity of full-length EGFR. Many of the amino acids that were shown to be critical for activity are conserved across the ErbB family (Figure 1C), except for ErbB3, which has minute intrinsic kinase activity. These results show that the majority of conserved amino acids within the JXM domain have an important role in the supporting the activation of the kinase domain. That single point mutations were highly disruptive to activity  suggests secondary structures and/or functional interactions comprise the activity contributed by the JXM region.
The importance of the JXM domain for EGFR kinase activity was also demonstrated in the study by Jura et al. . The activity of the ICD lacking the JXM domain (residues 672–998) was compared with that of the ICD containing the JXM domain (residues 645–998). It was observed that the presence of the JXM domain increased activity ~70-fold. Deletion of JMA region reduced catalytic efficiency 10-fold. The activity of the kinase domain (without the JXM domain, residues 672–998) increases ~20-fold when concentrated on lipid vesicles, compared with the activity of the solubilized kinase domain [9,12]. This ~20-fold increase is much less significant when compared with the increase in activity (~70-fold) when the JXM domain is added to the solubilized kinase domain, even without concentration on lipids. An isolated JMA peptide has been shown to interact with phospholipids (discussed in the next section). However, the activity of the full EGFR ICD (residues 645–698) on lipid vesicles was not measured. Comparing the activity of EGFR with and without the JXM region, while concentrated on lipid vesicles, may be valuable to understand the contribution of the JMA–lipid interaction to catalytic activity.
The presence of the JXM domain also contributes to EGFR dimerization. The ICD construct without the JXM domain (residues 672–998) is primarily monomeric in solution with a Kd value of approximately 8 μM, whereas inclusion of the JXM region (residues 645–698) caused the EGFR ICD construct to adopt a fully dimeric state with no detectable monomer. Deletion of the JMA region caused a shift to a monomeric state and decreased the apparent Kd to approximately 200 nM, indicating an increase in affinity .
The interaction of the EGFR JXM domain with phospholipids and Ca2+/CaM (calmodulin) signalling
Several studies have presented evidence that the JMA region of EGFR interacts with the inner leaflet of the plasma membrane [16,17,19] and promotes binding to CaM [16,18,19,21]. Much of the evidence for the interaction with phospholipids is from experiments that analysed the interaction of an EGFR JXM peptide (residues 645–660) with PtdInsP2-containing phospholipid vesicles [16,19]. It has also been shown, using Biacore SPR (surface plasmon resonance), that an immobilized EGFR JXM peptide (residues 645–662) binds to PtdInsP2 in solution and that the JMA region is important for autophosphorylation activity .
Several studies have suggested that the interaction of the EGFR JMA domain with the plasma membrane is autoinhibitory [16,18]. Upon activation, the rapid EGF-induced Ca2+ release and subsequent activation of CaM allows CaM binding to the JMA domain and subsequent release ICD from the inner leaflet of the plasma membrane, allowing active asymmetrical dimer formation. FRET (Förster resonance energy transfer) experiments were used to show that the interaction between the transmembrane JMA peptide (EGFR residues 622–660) and PtdInsP2 on lipid vesicles was reduced in the presence of Ca2+ and CaM . A later study by McLaughlin and colleagues was designed to explain the kinetics of CaM binding to the JMA region . The proposed mechanism suggests that, under physiological conditions, the interaction between the plasma membrane and the JMA region has such high affinity that the proximity of the two JMA domains provided by a pre-formed dimer is necessary to allow for repulsion from the plasma membrane, CaM binding and subsequent EGFR activation .
The structural model by Jura et al.  supports a slightly different view of the inactive EGFR dimer. The structure revealed a conserved cluster of lysine residues on the face of the kinase domain which is proposed to interact with the membrane in a manner that would locate the JXM segments away from the membrane. Also, in the FRET-based model of PtdInsP2 binding to the transmembrane JMA peptide (residues 622–660), the addition of sphingosine, which reverses the charge at the plasma membrane and should therefore inhibit the charge–charge-mediated interaction with the JXM region, reduced the interaction between the peptide and PtdInsP2. However, in the study by Thiel and Carpenter , the addition of sphingosine had no effect on kinase activity of the ICD of EGFR (residues 645–1186).
A role for the JXM domain in EGF binding
The ligand (EGF) binding to EGFR exhibits negative co-operativity, which explains the concave-up Scatchard plot of EGF binding (see  and references therein). Recently, the structural basis of the negative co-operativity was elucidated . It was shown that the first molecule of EGF that binds the receptor causes a conformational shift in the partnering receptor molecule, which renders binding of the second EGF molecule much less energetically favourable . There is a positive linkage between EGF binding and receptor dimerization, but not with kinase activity . In experiments performed by Macdonald-Obermann and Pike , it was shown, using ligand-binding assays of EGFR mutants, that the linkage to dimerization and the negative co-operativity of EGF binding are dependent on the presence of the JXM domain. These results imply that the JXM domain may contribute to the conformational change that occurs upon the first EGF-binding event.
Regulatory feedback signals mediated by the JXM domain of EGFR
There are two threonine residues (Thr654 and Thr669) within the JXM domain of EGFR which have feedback roles when phosphorylated (Figure 2). PKC (protein kinase C) can phosphorylate EGFR at Thr654 which dampens EGF-induced EGFR phosphorylation [25,26] and causes a delay in the down-regulation of the receptor, probably due to the effect of phosphorylation status . Thiel and Carpenter  showed that phosphomimetic mutation (T654D) at the PKC phosphorylation site caused a reduction in the activity of the ICD of EGFR, whereas the mutation that prevents phosphorylation (T654A) increased the phosphorylation activity. In contrast, phosphorylation by ERK (extracellular-signal-regulated kinase) at Thr669  and at a more minor site, Ser671  (Figure 2), causes accelerated down-regulation of the receptor . This is likely to be a positive-feedback mechanism since mutation at the ERK phosphorylation site (T669A) did not reduce kinase activity  and caused a heightened and shortened kinase activity profile .
Schematic diagram of the various roles of the EGFR JXM domain in EGFR activation
A study by Li et al.  supported the result that the T669A EGFR mutant displays increased kinase activity, but specifically showed that it is more acute compared with the wild-type EGFR, displaying a half-life of 1.6 h compared with the 3.2 h half-life of the wild-type. Furthermore, Li et al.  showed that EGF-induced EGFR ubiquitination was significantly increased in the T669A mutant or by treatment with a MEK1 (mitogen-activated protein kinase/ERK kinase 1) inhibitor. Notably, a cell line with an activating Ras mutation and constitutive ERK activity had a slightly more blunted EGFR degradation profile compared with cells expressing wild-type EGFR and wild-type Ras . This means that cancers with mutations that affect the Raf/Ras/ERK pathway may exhibit a different EGFR signalling profile compared with cancers that are wild-type in the Raf/Ras/ERK pathway.
The JXM domain of EGFR and receptor sorting and recycling
Following synthesis and release from the trans-Golgi network, EGFR is sorted to the basolateral membrane, where it is localized in vivo and in most cell models. The JXM domain (residues 652–674) was shown to be involved in the basolateral sorting of EGFR , as truncated EGFR mutants lacking this sequence are sequestered to the apical membrane region . A JXM dileucine motif (Leu679Leu680, Figure 2) has been shown to cause expression of EGFR that is truncated at the C-terminal region . The same dileucine motif was shown to be important for the rate of EGF-occupied receptor recycling . Mutation of the leucine residues to alanine did not have an effect on the internalization rate, but altered the compartmentalization of internalized receptors, directing them towards recycling compartments and away from lysosomes. Hence the JXM region of EGFR contains structural motifs which contribute to the sorting of the newly synthesized receptor and which effect the trafficking of internalized receptors.
There is accumulating evidence of the importance of the JXM region in the regulation of EGFR signalling, which is illustrated in Figure 2. The activating role of the EGFR JXM domain is unique, in that the JXM regions of other RTKs generally allosterically inhibit the kinase domain. The structural and functional characterization of the JXM region combined with its unique role in receptor activity validate this domain as a target for cancer therapies. Targeting the JXM region would offer a unique mode of modulating EGFR activity which is currently the target of several marketed anti-cancer therapies. Furthermore, modulating EGFR activity allosterically at the JXM region may offer an advantage to the more absolute inhibition by tyrosine kinase inhibitors at the kinase domain or by monoclonal antibodies at the ligand-binding inhibition.
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.).
This work was supported by the National Institutes of Health [grant number GM-54508].