The epidermal growth factor receptor family consists of four receptor genes and at least 11 ligands, several of which are produced in different protein forms. They create an interacting system that has the ability to receive and process information that results in multiple outputs. The family has an important role in directing and coordinating many normal processes, including growth and development, normal tissue turnover and wound healing. Its members are also aberrantly activated by overexpression or mutation in many common human tumour types and as such have been the target for anticancer drug development.
The epidermal growth factor receptor (EGFR) family
The epidermal growth factor (EGF) receptor (R) belongs to the ErbB family of receptor tyrosine kinases (RTKs). These receptors possess protein tyrosine kinase (PTK) activity and are found only in metazoans, in contrast to many of the serine/threonine kinase families, which are conserved throughout eukaryotes and are found in both unicellular and multicellular organisms (Stein & Staros 2000). The four receptor genes, encoding the EGF receptor (EGFR/erbB-1), c-erbB-2/HER2, c-erbB3-/HER3 and c-erbB4/HER4, can be alternatively spliced to give rise to various partial protein products; in the case of c-erbB-4, these include four variations of the full-length receptor. Examination of the completed human genome sequence has determined that 58 of the 90 tyrosine kinase sequences identified are RTKs (Robinson et al. 2000). Furthermore, there are no additional sequences that are potentially new members of the growth factor receptor family (Robinson et al. 2000). The ErbB nexus, comprising the four interacting receptor types found in mammals, appears to be an evolutionary elaboration of the fundamental module found in the worm Caenorhabditis elegans (named LET-23 or lethal complementation group 23) and the equivalent single-gene homologue (expressed as three splice variants) found in the fruit fly Drosophila melanogaster, namely, DER (see Moghal & Sternberg (1999) for a comparative review). The roles played by the type 1 growth factors and receptors in normal development, differentiation, migration, wound healing and apoptosis are essential for the viability of multicellular organisms; the effect of aberrant function is demonstrated in disease states such as cancer.
The 11 ligands currently identified for these receptors in mammals are EGF, transforming growth factor-α (TGF-α), HB-EGF (heparin binding), beta-cellulin, amphiregulin, epiregulin, epigen and the neuregulins (NRGs) 1–4 (Olayioye et al. 2000). A significant proportion of these are initially expressed as membrane-anchored proteins that require proteolytic cleavage either to achieve activity in solution or bind to cell surface proteoglycans from where they can act as a reservoir to be made available for receptor binding (Falls 2003). Two other genes, tomoregulin and epiglycan C, have recently been suggested to represent NRG 5 and 6 respectively, but further work is needed to confirm this assertion (Kinugasa et al. 2004). Determination of the ways in which this plethora of proteins interact and the resultant biological effects will be necessary to unravel the temporal and spatial nature of the mechanisms involved in this system (Jones et al. 1999).
A simple model by which one can begin to understand the complex system of growth factor signalling is based upon perceiving the network as three individual, sequential layers (Gullick 2001). The initial, extracellular layer is composed of the ligands, and its nature is therefore determined by their concentration and bioavailability. These two parameters dictate whether the receptors that reside within the cell membrane (and comprise the second layer of the system) will dimerise to become active. If the information in the first layer is sufficient to induce receptor dimerisation and consequently increase catalytic activity, the third, intracellular layer of second messenger proteins can bind to specific sites on the receptors and initiate the signals required to induce the appropriate response. As discussed below, it is now evident that most or all of the ErbB family of receptors further aggregate into oligomers of several hundreds or a few thousand receptors that recruit one or more second messenger proteins. Clearly, there are opportunities for this to allow interactions in the ‘third layer’, creating a system capable of integrating many inputs and producing multiple outputs, in many respects resembling a computational device (Johnson et al. 2004).
Modular structure of the EGFR
The first RTK to be discovered was the EGFR (Carpenter et al. 1978), the study of which has subsequently yielded insight into the underlying mechanisms by which RTKs function (Schlessinger 2000), including the recruitment and activation of second messengers (reviewed by Jorissen et al. 2003). Furthermore, it was this growth factor receptor that first identified a relationship between an activated oncogene and cancer (Downward et al. 1984).
Each member of the ErbB family comprises a conserved protein tyrosine kinase domain that resides within the cytoplasm, a transmembrane domain that makes a single pass through the plasma membrane, and a glycosylated, extracellular ligand-binding domain (Fig. 1). In the EGF receptor family, this last domain exhibits four subdomains denominated L1, S1 (CR1), L2 and S2 (CR2) (or, more simply, I, II, III and IV respectively) (Lax et al. 1989). Of these domains, S1 and S2 are homologous, cysteine-rich regions (CR1 and CR2), while L1 and L2 form the ligand-binding site (Garrett et al. 2002, Ogiso et al. 2002). It is likely that these are derived from an ancient gene duplication event (Stein & Staros 2000), and it is notable that the cysteine residues do not form disulphide bonds between the two S1/S2 domains.
In the absence of ligand, monomeric receptors reside within the cell membrane in an inactive state distributed fairly evenly over the cell membrane (although there are reports of concentrations of receptors in structures called calveolae and at the base of microspikes and in membrane ruffles). However, when ligand is present in the surrounding milieu, the receptor sites become occupied by monomeric EGF. Provided two stable 1:1 EGF:EGFR intermediate complexes persist, sequential receptor dimerisation and oligomerisation ensue (Ferguson et al. 2003). Receptor dimers, which have greater stability and ligand-binding affinity than their corresponding monomers (Ben-Levy et al. 1992, Zhou et al. 1993), are increasingly considered to be the minimal number of clustered receptors necessary for activation (Alroy & Yarden 1997). Such dimerisation is accomplished through the interaction of one EGFR S1 domain with that of a second ErbB receptor in a 2:2 receptor:ligand complex. Each S1 domain projects an extracellular ‘dimerisation loop’, and it is thought that the homophilic interactions between the two loops drive dimerisation. This scenario was elucidated by the crystal structure of ligand-bound EGFR that has recently been solved (Fig. 2) (Garrett et al. 2002, Ogiso et al. 2002).
Dimerisation may also involve direct contact between the helical transmembrane structures of the receptors, the attractive forces involved being either van der Waals interactions (Lemmon et al. 1997, White & Wimley 1999) or hydrogen bonding (Choma et al. 2000, Senes et al. 2001). Such close association between these domains is thought to be permitted by the amino-acid ‘dimerisation motifs’ contained within their 22–26 hydrophobic residues (Sternberg & Gullick 1989, 1990, Lemmon et al. 1994). These motifs were originally described as sequences of five residues, the first of which is a small side chain, the fourth an aliphatic chain, and the fifth either Gly or Ala (Sternberg & Gullick 1990). Particular point mutations within this sequence, such as the substitution of valine by glutamate, lead to constitutive dimerisation that, in the case of the c-erbB-2 receptor, is capable of inducing cell transformation (Brandt-Rauf et al. 1990, 1995, Sternberg & Gullick 1990, Smith et al. 1996, Gullick & Srinivasan 1998, Sharpe et al. 2000). The transmembrane domain is hypothesised to exert a high degree of regulation over receptor association, spatial arrangement and consequently signalling dynamics (Alroy & Yarden 1997, Gullick & Srinivasan 1998).
It is now evident that further oligomerisation of these receptors occurs in response to ligand binding and second messenger recruitment. Even in the seminal experiments of Yarden and Schlessinger (1987), who first reported the phenomenon of dimerisation, it was evident that higher-order aggregates were seen, and the authors ‘concluded that receptor oligomerisation is an intrinsic property of the occupied EGF receptor’. Now, by green fluorescent protein tagging (Hayes et al. 2004), the process can be followed visually (for films and still images, see Gillham et al. (1999) and www.kent.ac.uk/bio/gullick/). We and other groups are investigating the significance of these ‘signalling platforms’, which, by recruiting different receptor types and multiple second messenger proteins, may add further levels of complexity and capability to the process.
Homodimers and heterodimers
Encoded by the ERBB2 proto-oncogene located on chromosome 17q21, c-erbB-2 is a 1255-amino-acid transmembrane glycoprotein of 185 kDa (Coussens et al. 1985), which is thought to function as a ligand-less coreceptor for the other three members of the ErbB family and therefore for the majority (if not all) of the ErbB ligands (Klapper et al. 1999, Rubin & Yarden 2001). In response to transactivation by EGF-like ligands, such as TGF-α or epiregulin, c-erbB-2 receptors can form heterodimers with EGF receptors (Pinkas-Kramarski et al. 1996, 1997), or with c-erbB-3 and c-erbB-4 when activated by the NRGs (Tzahar et al. 1996, Burden & Yarden 1997, Pinkas-Kramarski et al. 1998). Indeed, c-erbB-2 is the preferred dimerisation partner in a distinct hierarchy that determines the interreceptor interactions (Tzahar et al. 1996) that are essential for normal Schwann cell (Morrissey et al. 1995) and certain epithelial cell behaviour (Klapper et al. 1997), and for determining specific cell lineages.
The published crystal structure of the extracellular domain of c-erbB-2 (Cho et al. 2003) reveals that, in the absence of direct ligand binding, c-erbB-2 adopts a fixed conformation resembling a ligand-activated state and, as such, is permanently capable of interacting with other members of the ErbB family. In contrast to c-erbB-2, this conformation is adopted only by the EGFR (Ferguson et al. 2003) and c-erbB-3 (Cho et al. 2002) upon ligand binding, and is characterised by a long, extracellular, finger-like projection from domain II (the ‘dimerisation loop’) that mediates interaction with the corresponding structure on its dimerisation partner (Garrett et al. 2002, Ogiso et al. 2002). In the absence of ligand, the dimerisation loop makes intramolecular contact with a pocket on domain IV, thereby restraining the receptor in an autoinhibited conformation. The fixed, open structure of c-erbB-2 confers upon it an intrinsic capability to interact with other ligand-bound receptors by circumventing the requirement for ligand binding to release the dimerisation loop from an autoinhibited conformation (Cho et al. 2003). This absence of a barrier against auto-activation contributes to the transforming potential of c-erbB-2 when overexpressed in cell culture (Di Fiore et al. 1987, Hudziak et al. 1987, Cho et al. 2003) and may enable c-erbB-2 homodimers to function as a constitutively active kinase (Lonardo et al. 1990). Overexpression of c-erbB-2 and the spontaneous (hetero)- dimerisation that ensues has been shown not only to cause an increase in basal receptor phosphorylation and activation (D’Souza et al. 1993, Samanta et al. 1994, Ram & Ethier 1996, Worthylake et al. 1999), but also to inhibit downregulation by reducing the degradation rate of the transphosphorylated pool of receptors (Worthylake et al. 1999).
It is interesting to note that the NRGs (ligands for the c-erbB-3 and c-erbB-4 receptors) are subject to alternative splicing that results in ligands with differing affinities for their receptors. Consequently, different receptors are recruited, so that this alternative splicing serves as a further mechanism by which the required heterodimers are achieved (Gullick & Srinivasan 1998). Such diversity of interaction between ligands and their receptors dictates the strength of the signalling kinetics and consequently the magnitude of the output and its specificity (as in growth or differentiation) (Marshall 1995).
Activation of the EGFR cytoplasmic PTK domain
The cytoplasmic region of the EGFR comprises three distinct domains: 1. the juxtamembrane domain, required for feedback by protein kinase C (PKC); 2. the noncatalytic carboxy-terminal tail, possessing the six tyrosine transphosphorylation sites mandatory for recruitment of adaptor/effector proteins (e.g. Grb2 and phospholipase Cγ (PLCγ) respectively) containing SH2 domains (src homology domain 2) or PTB (phosphotyrosine binding) domains, plus the motifs necessary for internalisation and degradation of the receptor; 3. the central tyrosine kinase domain (src homology domain 1 (SH1)) that is responsible for mediating transphosphorylation of the six carboxyterminal tyrosine residues.
It is thought that prior to EGF binding, the activation loop that resides within the cytoplasmic PTK domain of the EGFR may adopt an inactive conformation that is inaccessible to both substrate and ATP. Consequent to ligand binding and receptor dimerisation, intracellular tyrosine kinase catalytic activity is increased. The majority of RTKs absolutely require tyrosine phosphorylation within their activation loop for catalytic activity and biological function (reviewed by Hubbard & Till 2000, Dibb et al. 2004). However, the major exception to this is the EGFR, which instead is thought to require dimerisation of the cytoplasmic domain for catalytic enhancement (Mohammadi et al. 1993, Sherrill 1997), and this appears to be sufficient to stimulate cross-phosphorylation between the two receptors at 3–6 specific tyrosine residues. Recently, it has been suggested that in c-erbB-4 the tyrosine residues that become phosphorylated may be determined by the type of growth factor that has bound the receptor (Sweeney & Carraway 2000); this, in turn, determines the identity of the second messenger recruited (Margolis 1992, van der Geer & Pawson 1995). The tyrosine residues and surrounding amino acids of each member of the ErbB family are specifically tailored to interact with a unique collection of second messengers such that the specific biological and biochemical response may be precisely induced (Carraway & Cantley 1994, Alroy & Yarden 1997, Olayioye et al. 2000). Further observations have indicated that these properties are not unique to c-erbB-4 but are characteristic of all the ErbB family members (Crovello et al. 1998, Sweeney & Carraway 2000) and may be required for tissue-specific biological activity.
Subsequent to transphosphorylation of the receptor dimer, second messenger proteins possessing one of the two main classes of domains that recognise site-specific phosphorylation (‘docking sites’) can interact with the receptors. The larger of the two groups of second messenger proteins interact with the receptor via an SH2 (src homology-2) structure, whereas the others exhibit what is known as a PTB domain (Pawson 1995). Both are modular proteins (Kuriyan & Cowburn 1997, Pawson & Scott 1997, Margolis 1999) and serve to transmit the signal received at the receptor to either the cytoplasm or the nucleus. The specificity of PTB domain binding is dictated by the particular sequence of 3-5 amino-acid residues in the N-terminal stretch of activated EGFR that precedes the phosphotyrosine (pTyr) moiety (Prigent & Gullick 1994, Margolis 1999). Conversely, SH2 domains bind to specific amino-acid sequences 1–6 residues long that occur in the C-terminal sequence directly after the pTyr (Songyang et al. 1993). Furthermore, many SH2- and PTB-containing proteins can interact with phospholipids or nucleic acids and/or possess intrinsic enzymatic activity.
An example of a well-characterised second messenger/receptor interaction is the recruitment of the enzyme phospholipase C gamma (PLCγ) (Wang et al. 2001). In its inactive state, PLCγ is normally found in the cytosol. However, upon phosphorylation of the EGFR, the SH2 domain(s) of PLCγ are able to interact with the phosphorylated receptor. This causes not only its tyrosine phosphorylation by the activated receptor, but also relocation to the membrane, where it makes contact with the substrate PtdIns(3,4,5)P3 and ultimately generates the second messengers Ins(1,4,5)P3 and diacylglycerol (Falasca et al. 1998). A very similar redistribution occurs for several other second messengers (Gillham et al. 1999, Hayes et al. 2004). In addition to this effector protein activation by tyrosine phosphorylation, the two alternatives ‘activation by membrane translocation’ and ‘activation by a conformational change’ have been theorised, and the three are reviewed by Schlessinger (2000).
Proteins that contain SH2 or SH3 domains but lack enzymatic activity mediate signalling by acting as a platform upon which a specific complement of signalling proteins can be recruited and assembled, thereby linking the receptor to a specific signalling cascade (Pawson & Schlessinger 1993). This is exemplified by the adaptor protein Grb2 interacting with activated EGFR and recruiting the guanine nucleotide release factor Son of sevenless (Sos). In so doing, Sos is brought into close proximity with the plasma membrane and consequently its target protein Ras (Schlessinger 1994, Pawson 1995). Alternatively, EGFR can phosphorylate docking proteins, such as Gab1 (Mattoon et al. 2004), which then interact with multiple effector proteins.
In spite of the magnitude of information obtained, it has yet to be determined whether receptor dimers can recruit second messengers or whether receptor clustering is a prerequisite. Furthermore, it remains unknown whether a second messenger can bind more than one receptor at once, or indeed whether a receptor can bind multiple second messengers simultaneously (Gullick 2001).
The strength of ligand binding and signalling by heterodimers containing c-erbB-2 is significantly greater than that of either homodimers or heterodimers that do not recruit this receptor (Sliwkowski et al. 1994, Karunagaran et al. 1996). Dissociation of ligand:receptor complexes is decelerated by c-erbB-2 such that perpetuated ligand binding to c-erbB-2, c-erbB-3 or EGFR causes enhanced and prolonged stimulation of the MAP kinase (ERK) and c-Jun kinase (SAPK) pathways by both the NRGs and EGF (Karunagaran et al. 1996). Signal duration and potency is further enhanced by a slow rate of internalisation compared with that of the EGFR (Sorkin et al. 1993, Baulida et al. 1996). Internalisation is a process that serves to reduce the number of functional receptors at the plasma membrane and attenuate the strength of the signal generated. However, the sequences contained within the carboxyl terminus of c-erbB-2 (Sorkin et al. 1993) confer only a weak and ineffective coupling between this receptor and c-Cbl (Levkowitz et al. 1996, Graus-Porta et al. 1997, Muthuswamy et al. 1999) and an inability to associate with AP-2 (a plasma membrane-coated pit adaptor complex) (Gilboa et al. 1995) such that internalisation is impaired.
In terms of cell growth and transformation, the most potent heterodimer of the ErbB receptor family is composed of c-erbB-2 and c-erbB-3 (Wallasch et al. 1995, Pinkas-Kramarski et al. 1996), of which c-erbB-3 is a defective kinase (Guy et al. 1994) capable of binding some of the isoforms of the NRGs (Tzahar et al. 1994). Not only does the dimerisation of c-erbB-2 with c-erbB-3 increase the binding affinity of these sites, but the diversity of potential ligands to which they may bind is also extended to include EGF and betacellulin (Alimandi et al. 1997, Pinkas-Kramarski et al. 1998). This is particularly relevant to c-erbB-2-overexpressing human adenocarcinomas, in which c-erbB-3 is ubiquitously expressed at moderate or high levels (Lemoine et al. 1992) such that the combination of the two promotes ligand-binding promiscuity (Pinkas-Kramarski et al. 1998). An abundant supply of ErbB ligands, including EGF and betacellulin, originates from the mesenchyme-derived stroma underlying the layers of epithelial cells that are the precursors of adenocarcinomas. Consequently, this paracrine mesenchyme–epithelial interaction is thought to play a crucial role in tumour development (Salomon et al. 1995).
The collaboration of c-erbB-2 and c-erbB-3 leads to the activation of several of the major signal transduction pathways. Subsequent to ligand binding and dimerisation, c-erbB-3 becomes transphosphorylated by c-erbB-2, causing several carboxyl-terminus phosphotyrosine residues in each receptor to undergo the phosphorylation (Wallasch et al. 1995) required to permit interaction with intracellular signalling proteins (second messengers). Consequently, many of the c-erbB-3 phosphotyrosine moieties are able to recruit PI-3 kinase and thereby activate the Akt pathway (Prigent & Gullick 1994), whereas a single phosphotyrosine residue of c-erbB-2 is required for the recruitment of Shc and stimulation of the mitogen-activated protein kinase (MAPK) pathway, an interaction sufficient for cell transformation (Ben-Levy et al. 1994).
Downstream signalling pathways
The MAPK and PI3K/Akt pathways promote cell proliferative and survival/antiapoptotic signals via the activation of transcription factors and upregulation of cyclin D1 (Diehl et al. 1998, Lee et al. 2000, Lenferink et al. 2001, Wulf et al. 2001, Citri et al. 2003). Enhanced levels of functional c-erbB-2 cause an increase in cyclin D1 that functions to sequester the cyclin kinase inhibitor p27 and release Cdk2 (Neve et al. 2000). Subsequently, Cdk2 becomes positively regulated via its association with cyclin E and causes deregulation of the G1/S checkpoint such that the cell-cycle progression is promoted and leads to malignant transformation (reviewed in Harari & Yarden 2000). Overexpression of cyclin D1 at a 40% incidence has led to its identification as a significant factor in the development of breast cancer (Bartkova et al. 1994).
The downstream effectors of Akt also serve to sequester p27 such that the constitutive activation of Akt that arises from c-erbB-2-overexpression is thought to confer resistance to tumour necrosis factor-induced apoptosis (Zhou et al. 2000). Antiapoptotic signalling is further mediated by significant upregulation of the CDK inhibitor p21Waf1 (Waf1, Cip1, Sdil) (Fan et al. 1995, 1997, Bacus et al. 1996, Fiddes et al. 1998, Yu et al. 1998). It is this upregulation and consequent decrease in cyclinB-Cdc2 activation that is thought to protect against the paclitaxel-induced cell death exhibited by c-erbB-2-overexpressing cells (Yu et al. 1998).
Additional targets of c-erbB-2/c-erbB-3 signalling are the PLCγ pathway and the JAK-STAT pathway. PLCγ is activated by docking at specific sites solely on c-erbB-2 (Fedi et al. 1994) and hydrolyses PIP2 into the signalling molecules IP3 and DAG required for calcium/calmodulin-dependent kinases and stimulation of protein kinase C.
The full-length c-erbB-2 protein (p185) overexpressed by cultured tumour cells is subject to slow proteolytic cleavage that releases the Mr 110 000 (p110) extracellular domain (ECDc-erbB-2) into the conditioned medium (Lin & Clinton 1991, Zabrecky et al. 1991, Pupa et al. 1993). Soluble ECDc-erbB-2 was subsequently found in the serum of patients with advanced breast cancer in a manner that correlated well with c-erbB-2 overexpression in the primary tumour; in contrast, the sera of those patients in the early stages of the disease appear to be negative for this marker (Pupa et al. 1993). The shedding of ECDc-erbB-2 is thought to be a process regulated by tyrosine phosphorylation (Codony-Servat et al. 1999) that results in the constitutive activation of the remaining membrane-associated c-erbB-2 domains (Segatto et al. 1988, Huang et al. 1997). Such potentially increased signalling may explain the association of high ECDc-erbB-2 serum levels with a poor prognosis and limited response to endocrine therapy and chemotherapy in patients with advanced breast cancer (Di Fiore et al. 1987, Segatto et al. 1988).
A second marker of Mr 95 000 was later discovered in cancer tissues that correlated with the levels of p110 shed from the surface of c-erbB-2-overexpressing tumour cells. This was determined to be an NH2-terminally truncated c-erbB-2 product possessing in vitro kinase activity (Christianson et al. 1998). Subsequently, a relationship was revealed between the expression of this truncated protein and metastasis to the lymph nodes in p95-positive breast cancer patients; p95-negative tumours were associated with node-negative patients (Christianson et al. 1998). However, this study found no correlation between p185 or p95 enrichment in patients and tumour size or hormone receptor status.
The complexity of EGFR signalling is great. However, new technologies are evolving, such as fluorescently labelled receptors, ligands and second messengers expressed by live cells that can be detected by digital microscopy, which offer the opportunity of visualising not only the interplay between these signalling components but also their trafficking and positioning within the cell (Hayes et al. 2004). Moreover, prototype computational models of growth factor receptor clustering have been developed (Johnson et al. 2004) that will in time also contribute to obtaining further information about signalling interactions. Genetic manipulation (knockins and -outs) and the ability to suppress protein levels (knockdowns) will be critical tools in cell biology experiments. Finally, the growing use of antibodies and small molecule receptor tyrosine kinase inhibitors in large numbers of cancer patients will reinforce and inform laboratory experiments. Ultimately, however, the problem is soluble and not only will lead to a comprehensive description of the system and better cancer treatments but also represents an example where the new methods can be applied to other complex systems.
Laura Bazley is the recipient of a postgraduate studentship from the Breast Cancer Campaign.
Laura Bazley declares no conflict of interest. Bill Gullick has received payments from AstraZeneca for work associated with several advisory boards.
BacusSS Yarden Y Oren M Chin DM Lyass L Zelnick CR Kazarov A Toyofuku W Gray-Bablin J Beerli RR Hynes NE Nikiforov M Haffner R Gudkov A & Keyomarsi K 1996 Neu differentiation factor (heregulin) activates a p53-dependent pathway in cancer cells. Oncogene122535–2547.
FediP Pierce JH di Fiore PP & Kraus MH 1994 Efficient coupling with phosphatidylinositol 3-kinase but not phospholipase C gamma or GTPase-activating protein distinguishes ErbB-3 signaling from that of other ErbB/EGFR family members. Molecular Cell Biology14492–500.
GarrettTP McKern NM Lou M Elleman TC Adams TE Lovrecz GO Zhu H-J Walker F Frenkel MJ Hoyne PA Jorissen RN Nice EC Burgess AW & Ward CW 2002 Crystal structure of a truncated epidermal growth factor receptor extracellular domain bound to transforming growth factor alpha. Cell110763–773.
HuangHS Nagane M Klingbeil CK Lin H Nishikawa R Ji XD Huang CM Gill GN Wiley HS & Cavenee WK 1997 The enhanced tumorigenic activity of a mutant epidermal growth factor receptor common in human cancers is mediated by threshold levels of constitutive tyrosine phosphorylation and unattenuated signaling. Journal of Biological Chemistry2722927–2935.
LaneHA Beuvink I Motoyama AB Daly JM Neve RM & Hynes NE 2000 ErbB2 potentiates breast tumour proliferation through modulation of p27Kip1-Cdk2 complex formation: receptor overexpression does not determine growth factor dependency. Molecular and Cellular Biology203210–3223.
LeeRJ Albanese C Fu M D’Amico M Lin B Watanabe G Haines GK 3rd Siegel PM Hung MC Yarden Y Horowitz JM Muller WJ & Pestell RG 2000 Cyclin D1 is required for transformation by activated Neu and is induced through an E2F-dependent signaling pathway. Molecular and Cellular Biology20672–683.
RamTG & Ethier SP 1996 Phosphatidylinositol 3-kinase recruitment by p185erbB-2 and erbB-3 is potently induced by neu differentiation factor/heregulin during mitogenesis and is constitutively elevated in growth factor-independent breast carcinoma cells with c-erbB-2 gene amplification. Cell Growth and Differentiation7551–561.
SliwkowskiMX Schaefer G Akita RW Lofgren JA Fitzpatrick VD Nuijens A Fendly BM Cerione RA Vandlen RL & Carraway KL 3rd 1994 Coexpression of erbB2 and erbB3 proteins reconstitutes a high affinity receptor for heregulin. Journal of Biological Chemistry26914661–14665.
TzaharE Levkowitz G Karunagaran D Yi L Peles E Lavi S Chang D Liu N Yayon A Wen D & Yarden Y 1994 ErbB-3 and ErbB-4 function as the respective low and high affinity receptors of all Neu differentiation factor/heregulin isoforms. Journal of Biological Chemistry26925226–25233.
TzaharE Waterman H Chen X Levkowitz G Karunagaran D Lavi S Ratzkin BJ & Yarden Y 1996 A hierarchical network of inter-receptor interactions determines signal transduction by Neu differentiation factor/neuregulin and epidermal growth factor. Molecular and Cellular Biology165276–5287.