Oncogene (2000) 19, 6093 ± 6101
ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00
www.nature.com/onc
HER2/Neu: mechanisms of dimerization/oligomerization
Patrick J Brennan1, Toru Kumogai1, Alan Berezov1, Ramachandran Murali1 and
Mark I Greene*,1
1
Department of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine, Philadelphia, PA 19104,
USA
Oncogene (2000) 19, 6093 ± 6101.
Keywords: erbB; kinase; growth factor; signaling;
tumor
Introduction
Within a multi-celled organism, regulation and organization require that biological signals be transmitted
from one cell to another, across cell membranes. Such
is the case with growth factors, which originate from
one site in an organism, yet need to be distributed
throughout the organism to many cell types in order to
exert their pleiotropic e€ects. Systems have evolved to
allow a soluble signal, a growth factor for example, to
be conveyed from the extracellular space to the
cytoplasm and nucleus of a cell, thus directing protein
synthesis, cellular growth, and cellular proliferation. To
solve this problem of transmitting signals across
membranes, growth factors appear to induce aggregation of their receptors, leading to activation of these
receptors and signal propagation. Unregulated, however, these growth signals have the potential to
promote inappropriate proliferation; receptor aggregation in the absence of signal represents a means of
unregulated growth factor receptor activity which is
postulated to occur in human neoplasia.
The neu gene encodes a 185-kDa transmembrane
glycoprotein, referred to as p185neu, HER2, or erbB-2,
possessing intrinsic protein tyrosine kinase activity. The
receptor consists of an extracellular domain, with four
subdomains including two cysteine rich domains, a
transmembrane domain, and an intracellular domain,
consisting of a juxtamembrane region, a tyrosine kinase
domain, and a carboxyl tail harboring autophosphorylation sites (Figure 1). HER2 is homologous to, but
distinct from, other members of the erbB family, which
includes the epidermal growth factor receptor (EGFR
or erbB-1), erbB-3, and erbB-4. The binding of cognate
growth factors to these receptors regulates cell growth,
proliferation, and di€erentiation through the activation
of receptor tyrosine kinases, triggering an incompletely
de®ned signal transduction cascade. Signal transduction by these receptors is believed to involve dimerization and oligomerization, both in the form of homooligomers and hetero-oligomers in various erbB
receptor combinations. Though this oligomerization
has been recognized for more than 20 years, and
though much progress has been made in de®ning the
critical aspects of this process, many of the mechanistic
*Correspondence: MI Greene
principles involved remain to be de®ned. Key studies
which have led to our current understanding of the
mechanism and function of the oligomerization process
in erbB receptors will be outlined below, as will the
outstanding questions raised by these studies.
Mechanisms of dimerization/oligomerization
Though dimerization/oligomerization has been recognized as an integral component of signaling by erbB
family receptors since shortly after the discovery of
neu, the mechanism by which these receptors aggregate
is still not fully understood. Extracellular, transmembrane, and intracellular domains of the protein have all
been implicated in dimerization/oligomerization; the
potential contribution of each of these domains to
receptor aggregation will be discussed below.
Insights from the transmembrane domain
A role for neu in transformation and tumorigenesis has
been suspected since its discovery as a di€erentially
expressed antigen in a chemically induced rat neuroglioblastoma (Padhy et al., 1982; Schechter et al., 1984; Shih
et al., 1979, 1981), hence the name `neu'. Upon
comparison of the normal, protooncogenic form of the
gene with oncogenic neu in the rat model, a single
causative point mutation in the putative transmembrane
region of the protein revealed itself as the culprit lesion
(Bargmann et al., 1986a,b). Mutation occurred at
position 664 of the protein sequence, encoding a change
from the hydrophobic amino acid valine to the
negatively charged glutamate. Further studies suggested
that this mutation garnered its transforming e€ect by
increasing the propensity of the receptor to form
aggregates (Weiner et al., 1989b). Although the mutation
is not observed in human tumors, the information gained
through the study of this mutation has provided
invaluable clues to the mechanism of activation and
transformation utilized by HER2/neu in human disease.
Early studies identi®ed that the e€ects of the point
mutation seen in rat neu were mediated by dimerization. Using nondenaturing gel electrophoresis with cell
lysates prepared from cells overexpressing normal or
oncogenic (containing the transmembrane point mutation) forms of p185, Weiner et al. (1989a) found that
the majority of oncogenic p185 existed in a dimeric
form, while normal p185 was predominantly in a
monomeric state. Co-transfection of short neu transmembrane sequences lacking both intracellular and
extracellular domains was shown to diminish the
transforming potential of full-length p185neu protein,
presumably by preventing the formation of produc-
HER2/Neu: mechanisms of dimerization/oligomerization
PJ Brennan et al
6094
Figure 1 Schematic diagram of structural organization of p185
receptors. Epidermal growth factor receptor (EGFR), Her2/neu/
erbB2, erbB3 and erbB4 are members of the super family. The
numbers indicate the sequence homology among members
compared to EGF receptor. They share a high degree of primary
and tertiary structural homology in the tyrosine kinase domain
and a moderate degree of homology in ectodomain. C-terminus is
the most variable region among the receptors
tively arranged oligomers (Lofts et al., 1993). Further
supporting a role for oligomerization within the erbB
family, EGFR dimerization had been demonstrated
using chemical crosslinking (Fanger et al., 1989),
sucrose gradients (Boni-Schnetzler and Pilch, 1987),
and ¯orescent energy transfer (Carraway et al., 1989).
Though much e€ort has been expended toward
de®ning the mechanism by which the observed
transmembrane mutation leads to transformation, a
clear answer has remained elusive. That this point
mutation leads to increased dimerization, increased
receptor tyrosine kinase activity, and increased transformation potential is well accepted; how this mutation
leads to these e€ects remains a matter for debate.
Three models have been proposed to explain the
mechanism by which the transmembrane point mutation leads to oligomerization. The ®rst and second
model, which di€er only slighlty, favor stabilization of
interreceptor transmembrane association via additional
hydrogen bonds provided by the mutation; the ®rst
model proposes that adjacent 664E residues interact,
while the second proposes that the 664E residue
Oncogene
interacts with the backbone of the adjacent receptor
(Smith et al., 1996; Sternberg and Gullick, 1989).
Recent e€orts in molecular modeling support these ®rst
two models, but have not provided a means to
distinguish between the two (Sajot and Genest, 2000).
Structural studies using NMR and IR have suggested
that the 664E mutation is involved in a hydrogen
bonding type of interaction with its partner transmembrane domain, supporting the ®rst model (Smith et al.,
1996). Mutational studies show that only Glu and Gln
residues at position 664 lead to transformation, Asp
and Tyr having weak e€ects, results consistent with a
primary role for hydrogen bonding (Bargmann and
Weinberg, 1988). The third model proposes that the
V664E mutation leads to a global conformational
change within the transmembrane domain allowing for
a more stable dimeric structure. Initially, molecular
modeling and other structural analyses of the protooncogenic and oncogenic forms of the neu transmembrane domain suggested that the protooncogenic form
of the receptor contained a kinked transmembrane
domain, while the oncogenic form of the receptor was
primarily helical (Brandt-Rauf et al., 1989, 1990, 1995).
This model, however, has not been supported by later
moleuclar dynamics simulations or other structural
studies (Sajot and Genest, 2000; Smith et al., 1996).
Because the transmembrane mutation seen in neu
seems sucient to cause constitutive dimerization, a
similar mutation in EGFR might be expected to lead
to dimerization, unregulated signaling, and transformation. There exist con¯icting data on this point,
however, bringing the autonomy of the transmembrane
mutation into question. Simply mutating the equivalent
Val in EGFR to Glu led to a receptor complex with
increased, yet still ligand-dependent, transforming
potential (Kashles et al., 1988). A more recent study,
however, found con¯icting results, suggesting that a
point mutation in the EGFR transmembrane domain
could lead to a ligand-independent transforming
receptor complex (Beguinot et al., 1995)
Production of a functional neu complex requires
more than stable dimer formation, as a steric
constraint on this dimerization also appears to be
crucial. Experimentally relocating the position of the
point mutation within the transmembrane domain only
leads to a transforming receptor complex if the
mutation is placed in certain positions. Bargmann
and Weinberg (1988) noted that mutation of either
residues 663 or 665, the residues immediately adjacent
to the mutation observer in rat neu, to Glu did not
lead to transformation. Other groups have pursued this
observation, investigating the importance of residues
surrounding the point mutation and more closely
examining the location constraints for a transforming
mutation (Burke et al., 1997; Cao et al., 1992). By
generating a battery of both random and rationally
designed potential transmembrane sequences, Chen et
al. (1997) found a critical limitation for spacing
between residues with the potential for hydrogen
bonding, suggesting a possible rotational constraint.
A transgenic mouse overexpressing protooncogenic
erbB-2 develops tumors predominantly after spontaneous mutation of juxtamembrane residues, leading to
disul®de-mediated constitutive dimerization (Siegel and
Muller, 1996). In addition to providing proof that
dimerization leads to transformation, this observation
HER2/Neu: mechanisms of dimerization/oligomerization
PJ Brennan et al
provided for the use of juxtamembrane mutations as a
readily controlled arti®cial experimental system for the
induction of dimerization (Sorokin et al., 1994).
Surprisingly, it was found that, depending on the
location of the cysteine mutations introduced, stable
dimers could be induced without concomitant transforation (Sorokin et al., 1994), a result supported by a
later study utilizing a transplanted dimerization-inducing transmembrane domain (Burke et al., 1997). Upon
further investigation, receptors with mutations that led
to transforming receptor complexes all formed dimers
involving the same predicted helical interface, strongly
suggesting a steric constraint for p185neu activation
(Burke and Stern, 1998).
Extracellular domain
ErbB receptor extracellular domains are responsible
ligand binding and seem to mediate dimerization/
oligomerization in the majority of biological settings.
Natural variants and experimental mutations of these
extracellular domains have provided clues to the nature
of ligand binding and interaction of these domains.
While other members in the receptor family have
natural ligand, erbB2 has no known natural ligand to
date. Despite an exhaustive search e€ort to ®nd a
natural ligand, erbB-2 remains an orphan receptor, and
it may very well be that this receptor functions mainly
in complex with other family members. HER2/neu
shows both ligand-independent and ligand-dependent
activity, and investigation of each of these activities has
provided valuable information regarding the mechanisms of transformation utilized by this receptor.
Ligand-dependent dimerization
Though erbB-2 is now thought to be an orphan
receptor, it plays an active role in ligand-mediated
signaling through heterodimerization with other erbB
family members. Dimer formation between multiple
erbB family members, in most combinations, in
response to a growing list of ligands, constitutes a
complex signaling network (Klapper et al., 2000;
Tzahar et al., 1996). The EGFR was the ®rst receptor
tyrosine kinase proposed to dimerize in response to
ligand binding (Yarden and Schlessinger, 1987a,b), and
this system has been well studied; an understanding of
the ligand binding mechanism in EGFR should serve
as a template for receptor ± ligand and receptor ±
receptor interactions within the erbB family, and
perhaps beyond. Ligand binding mechanisms utilized
by other receptor families may also provide valuable
clues to the mechanisms used within the erbB family. A
general scheme based on current understanding of
EGF induced oligomerization is shown in Figure 2.
Multiple studies have concluded that EGF binds
EGFR with a 1 : 1 stoichiometry (Brown et al., 1994;
Domagala et al., 2000; Lemmon et al., 1997; Odaka et
al., 1997; Weber et al., 1984). Subdomains I and
especially III of the extracellular portion of EGFR
have been identi®ed as important in ligand binding
(Lax et al., 1998, 1989, 1991; Lee et al., 1995;
Summer®eld et al., 1996), as has subdomain IV, one
of two cysteine rich domains (Saxon and Lee, 1999).
Recent structural modeling proposes that EGF binds
6095
Figure 2 Schematic diagram of putative dimerization of EGF
receptors. Dimerization of EGFR viewed down from the top. The
scheme was conceived based on our molecular modeling of the
EGF receptor. The proposed dimerization model assumes a 2 : 2
association of EGFR : EGF
in a cleft formed by subdomains I, II and III (Jorissen
et al., 2000). Studies investigating the binding of EGF
to EGFR show a high and a low anity binding site,
and kinetic studies show that receptor activity is second
order with respect to ligand (Canals, 1992). In a recent
study of EGF-sEGFR interactions, a complex binding
mechanism has been observed, which could not be
adequately described by a simple Langmuirian interaction (Domagala et al., 2000). Two sEGFR populations have been detected in solution: one with a high
anity (2 ± 20 nM) and the other with a low anity
(400 ± 550 nM) for EGF (Domagala et al., 2000). The
results of these studies are complicated by the fact that
some groups have used soluble receptor fragments,
while others have used full-length receptor present on
the surface of intact cells; the anity of EGF for a
soluble extracellular EGFR fragment has been shown
to be markedly lower than for the full-length receptor
(Brown et al., 1994). Furthermore, interpretation of
binding studies with erbB receptors must take into
account that the receptors are in a monomer/oligomer
equilibrium which is not governed solely by ligand
concentration. Using a soluble EGFR extracellular
domain, positive cooperativity with increasing EGF
concentration has been demonstrated (Hurwitz et al.,
1991; Lemmon et al., 1997), while studies of intact cells
have demonstrated negative cooperativity (Chamberlin
and Davies, 1998).
Lemmon et al. (1997) investigating EGF-induced
dimerization using soluble EGFR extracellular domain,
proposed a model in which one EGF binds an EGFR
monomer, followed by dimerization with a similar
EGF/EGFR complex. One possibility assumes that the
bound EGF molecule mediates subsequent dimerization; alternatively, EGF may bind to EGFR causing a
conformational change favoring dimerization, without
a requirement for EGF at the dimer interface (Green®eld et al., 1989; Lemmon et al., 1997). While this
model may accurately represent the events involved in
ligand-induced oligomerization in solution, it may not
re¯ect what is happening at the surface of intact cells.
A second model which could explain the observed
high and low binding activities of EGF for EGFR
involves the induction of dimerization by the binding
of one EGF molecule to two EGFR, a low anity
interaction, followed by the binding of a second EGF
Oncogene
HER2/Neu: mechanisms of dimerization/oligomerization
PJ Brennan et al
6096
to the previously formed EGF : 2EGFR complex in
which the EGFR now bind EGF with high anity. In
such a model, high and low anity binding sites may
be explained by the asymmetrical binding of a bivalent
EGF in which each EGF binds with high anity to
one member of the EGFR dimer and with low anity,
via a distinct epitope, to a second EGFR (Summer®eld
et al., 1996; Tzahar et al., 1997); a two binding site
model is consistent with a recent structural model
developed for the EGFR extracellular domain (Jorissen
et al., 2000).
In support of this model, recent electron microscopic
analysis suggests that EGFR may exist in a predimerized state, and that this state seems to correspond
with observed high anity EGF binding sites (Gadella
and Jovin, 1995). A theoretical kinetic analysis
proposed by Chamberlin and Davies (1998) which
accounts for possible clustering that may occur on the
surface of intact cells also supports a two site model in
which EGF molecules bind sequentially. Recent singlemolecule imaging of EGFR on the cell surface (Sako et
al., 2000) suggests that the predominant mechanism of
dimerization involves the formation of a cell-surface
complex of one EGF molecule and an EGFR dimer,
followed by the direct arrest of a second EGF
molecule. The EGFR dimers are thus preformed before
the binding of the second EGF molecule.
Ligand-mediated binding involving erbB-2-containing heterodimers has not been as extensively studied as
EGF/EGFR binding. Given the high degree of
homology between erbB family members, models
developed through the study of EGF/EGFR binding
should prove a useful frame for understanding
interactions within the erbB family. With respect to
erbB-2 ligand-dependent dimerization, the data that
exist suggest a model in which heterodimerization is
mediated by ligand bivalence. ErbB-2 has been shown
to be the preferred dimerization partner of the other
erbB family members (Tzahar et al., 1996), and
demonstrates synergistic transforming e€ect when
expressed with EGFR (Kokai et al., 1989). Using
soluble chimeric receptors, Jones et al. (1999) recently
showed that heterodimerization with erbB2 enhances
ligand binding for all other erbB family members to
their respective ligands. Tzahar and colleagues (Tzaher
et al., 1997), in studies of erbB-2/erbB-3 heterodimerization driven by NDF/neuregulin, propose that NDF
has a high anity interaction with erbB-3 and a low
anity interaction with erbB-2, and that NDF
simultaneously binds both receptors. Furthermore, this
low anity binding site, while possessing relatively
broad speci®city, may favor erbB-2 binding, providing
a model which could explain the observed preference
for heterodimerization and possibly even erbB-2
oncogenicity.
Ligand-independent dimerization
As discussed previously, the transmembrane mutation
seen in rat neu confers the ability for ligandindependent signaling. Transmembrane mutation observed in rat is not seen in human tumors, probably
because such a mutation would require a two
nucleotide mutation in the human gene (Dougall et
al., 1994). Nevertheless if arti®cially introduced into
human HER2/neu, an equivalent mutation leads to
Oncogene
increased dimerization and transforming ability (Segatto et al., 1988). Overexpression of the protooncogenic form of HER2/neu, however, either with or
without ampli®cation, is observed in human tumors
(Dougall et al., 1994; Klapper et al., 2000). It might be
expected, from data obtained with oncogenic neu, that
increased dimerization/oligomerization, thus leading to
increased signaling, would be seen in human tumors
associated with HER2/neu overexpression; indeed, this
turns out to be the case. In cell culture models,
moderate overexpression (receptor number *105) of
the protooncogenic form of p185, using an SV40
promoter-driven expression vector, failed to cause
transformation in NIH3T3 cells. However, overexpression (receptor number 4106) of this receptor at higher
levels, using a retroviral promoter, was transforming
(Di Fiore et al., 1987b; Di Marco et al., 1990). Hence,
it seems that once a critical threshold level of HER2/
neu expression is reached, transformation is achieved, a
phenomenon thought to be due to spontaneous
dimerization/oligomerization. To directly show that
overexpression itself can lead to spontaneous oligomerization, we demonstrated that increasing the expression
of protooncogenic p185neu shifted the receptor equilibrium from a monomeric to an aggregated state
(Samanta et al., 1994). It appears that overexpression,
much like the rat neu point mutation discussed
previously, leads to unregulated oligomerization and
unregulated signaling.
Extracellular domain of p185 is sucient to induce
dimers. We have shown that mutant forms of p185neu
lacking the intracellular portion of the receptor, both
with and without the oncogenic transmembrane point
mutation, are sucient to oligomerize with full-length
EGFR and p185neu receptors (Qian et al., 1994a,b,
1996). This binding is ligand-independent, and that
truncated forms both with and without the oncogenic
transmembrane point mutation eciently oligomerize
suggests that this binding is mediated by the extracellular domain. In the reverse experiment, an
intracellular truncation mutant of EGFR also heterodimerizes with p185neu (Kashles et al., 1991), suggesting again that the extracellular domains of erbB
receptors are sucient for dimerization.
Analysis of extracellular domain mutants derived
from erbB-1 and erbB-2 have provided clues as to
which subdomains are important for ligand binding
and oligomerization. A truncated EGFR seen in
human glioblastoma, for example, can dimerize and
transform mammalian cells in a ligand-independent
fashion (Moscatello et al., 1996) Homologous to
mammalian erbB proteins, avian viral v-erbB truncation mutants display ligand-independent e€ects, though
the signi®cance of the dimerization status of these
proteins is unclear (Adelsman et al., 1996). Herstatin, a
recently discovered erbB-2 alternate transcript, consisting of 340 amino acids of the erbB-2 extracellular
domain and 79 amino acids of novel sequence, binds
full-length erbB-2, inhibiting its function (Doherty et
al., 1999). Studies with the deletion mutants suggest
that the extracellular domains of erbB receptor that
small structural components in a subdomain may
function not only to mediate dimerization in response
to ligand, but also to prevent surreptitious dimerization. Recently we have modeled the three dimensional
structure of the erbB-2 extracellular domain using the
HER2/Neu: mechanisms of dimerization/oligomerization
PJ Brennan et al
insulin-like growth factor-1 receptor (Garrett et al.,
1998) as a template and found that the fourth domain
may be critical for dimer formation. Based on this
model, we have developed a small peptide with cystineknot topology which shows varying degree of anity
to di€erent members of erbB receptors. This small
peptide is also able to block the dimer formation
(Berezov, Murali and Greene, unpublished data). These
results suggests that a relatively small fragment of the
p185 extracellular domain has the ability to homodimerize.
Intracellular domain
Although the extracellular domain appears to be
responsible for the majority of the interactions
governing erbB receptor dimerization/oligomerization,
evidence suggests that the intracellular domain plays
more than a passive role in dimerization. For example,
if both full-length EGFR and an intracellular domain
truncated EGFR are coexpressed, full-length receptor
favors binding available full-length receptor; this
suggests that the intracellular domain may stabilize
dimer formation (Kashles et al., 1991).
To directly address the role of the intracellular
domain in oligomerization, Chantry coexpressed truncated EGFR forms lacking the extracellular domain
along with full-length receptor containing the intracellular domains of EGFR or HER2/neu, measuring the
resultant dimerization, kinase activity, and autophosphorylation (Chantry, 1995); it was found that the
intracellular domain, if membrane associated, was
sucient for dimerization and kinase activation.
Analysis of a panel of deletion mutants, created to
ascertain which portions of the intracellular domain
were involved in the observed dimerization, suggests
that the kinase domain is critical. Interestingly,
intracellular domain-mediated interactions were also
observed between EGFR and PDGF receptor as well
as HER2/neu and PDGF receptor, suggesting that
dimerization may not be speci®c, but may represent
interactions between `dimerization motifs' present
within the highly conserved kinase domains of these
receptors (Chantry, 1995). The reverse experiment,
however, in which the intracellular domain of p185
was expressed along with full-length EGFR, yielded
con¯icting results, as an interaction was not seen (Qian
et al., 1999). Modeling from our laboratory nevertheless supports kinase domain interaction. Dimerization and tetramerization models of EGFR and p185
kinase domains suggest that these domains can
associate as energetically stable structures (Murali et
al., 1996). In addition, the models suggest that the
heterodimer interaction is more favorable than that of
either homodimer, an observation which could contribute to observed synergistic signaling (Kokai et al.,
1989), and even to the favored binding partner status
enjoyed by p185 (Graus-Porta et al., 1997; Tzahar et
al., 1996). The model also proposed dimerization of
erbB receptors may be in¯uenced by electrostatic
charge distribution near membrane proximal regions
(Murali et al., 1996). This hypothesis suggests that
proper orientation of receptors is critical for dimerization. This feature has been observed in the EPO
receptor complex (Livnah et al., 1998).
Aggregation of erbB kinase domains seem to be
intrinsically linked to their activity. Activation of
kinase activity was observed in response to forced
aggregation of a soluble EGFR intracellular domain by
various methods, including cross-linking antibodies,
polycation, or chemical reduction of protein solubility
(Mohammadi et al., 1993). Using a soluble version of
the p185neu intracellular domain, we have shown that
aggregation leads to an increase in the Vmax of the
receptor (LeVea et al., 1993). It is possible, then, that
the binding activity inherent in the kinase domains of
erbB and other growth factor receptors may serve to
mediate interactions of these domains after extracellular domain-mediated juxtaposition, the end result
being the same as with forced aggregation.
6097
Functional consequences of oligomerization
It is clear that both homodimerization and heterodimerization occur within the erbB receptor tyrosine
kinase family. To de®ne the function and the essential
features of this phenomenon, a variety of receptor
mutants have been generated, and their function
subsequently evaluated both alone and in combination
with other erbB receptors. The picture that has
emerged suggests that oligomerization may allow
kinase domains to be positioned next to one another,
thereby allowing cross-phosphorylation, activation,
and substrate phosphorylation. Several of the key
experiment which have led to these conclusions are
outlined below.
Homodimerization
Though the transforming activity of p185neu was
identi®ed along with the receptor itself, and though
receptor oligomerization was also quickly identi®ed as
a feature of this receptor, the link between dimerization
and transformation was not immediately accepted. The
functional consequences of homodimerization of
EGFR have also been studied, and this dimerization,
which can be regulated by the presence or absence of
ligand, has provided evidence in support of a role for
homodimerization of erbB-2. Two possible mechanisms
for the activation of EGFR signaling by EGF have
been proposed; the ®rst suggests that EGF signal
transmission occurs through a single EGFR monomer,
possibly through an induced conformational change
(Koland and Cerione, 1988). The second, initially
alternative hypothesis, supports a functional role for
dimerization in ligand-responsive signal transduction
(Yarden and Schlessinger, 1987b). The kinetics of both
dimerization and activation of full-length, solubilized
EGFR were found to be second-order, with an
identical rate constant, a correlation which strongly
suggests that the two processes are linked (Canals,
1992). EGFR has been shown to transform NIH3T3
cells when overexpressed, though this transformation is
ligand-dependent (Di Fiore et al., 1987a). An engineered EGFR mutant which no longer binds EGF
with high anity, yet is constitutively dimerized, is
transforming when expressed in NIH3T3 cells, suggesting that dimerization may be the critical function of
ligand binding (Sorokin, 1995).
Further evidence that the function of EGFR is
dependent on dimerization/oligomerization has come
Oncogene
HER2/Neu: mechanisms of dimerization/oligomerization
PJ Brennan et al
6098
Oncogene
from the use of antibodies. Cross-linking antibodies are
capable of activating EGFR, an observation which has
been documented by a number of groups using
antibodies which recognize distinct epitopes of the
EGFR extracellular domain (Schreiber et al., 1981;
Spaargaren et al., 1991; Yarden and Schlessinger,
1987b); Non-crosslinking Fab fragments of these
antibodies, however, do not activate the receptor
(Spaargaren et al., 1991). In contradiction with the
above results, another antibody, capable of reducing
EGFR dimerization, had little e€ect on EGFR
autophosphorylation (Carraway and Cerione, 1993).
As with the EGFR receptor, overexpression of erbB2 at sucient levels transforms NIH3T3 cells (Di Fiore
et al., 1987b); unlike the EGFR, however, this
transformation is ligand-independent. As discussed
previously, overexpression leads to spontaneous receptor oligomerization, and as with the EGFR receptor,
oligomerization is thought to be necessary for signaling
and subsequent transformation. In order to establish a
role for oligomerization, and to de®ne the mechanism
of activation and signaling utilized by erbB receptors,
we and others have generated a variety of mutant
forms of these proteins. A full-length kinase-inactive
mutant of oncogenic neu, T757, when expressed in a
background de®cient of both EGFR and erbB-2
proteins, retains the ability to oligomerize, yet is not
transforming, indicating the critical importance of
kinase activity to transformation (Qian et al., 1995).
Such a kinase-inactive mutant has been shown to have
the ability to competitively inhibit the function of wildtype p185neu when co-expressed (Akiyama et al., 1991;
Messerle et al., 1994), likely by preventing the
formation of productive dimers. TAPstop, a mutant
in which the carboxy terminus has been truncated,
including potential autophosphorylation sites, retains
transforming ability, although at a reduced level
compared to full-length p185neu. TD5, an ectodomain-deleted form of neu, also retains transforming
ability, though like TAPstop, a lower level than
p185neu.. Interestingly if either TAPstop or TD5 is
coexpressed with the kinase-inactive mutant, T757, an
increase in transforming potency is observed. While in
cells expressing only T757, autophosphorylation of this
mutant is not observed, double mutant cells contain
T757 which has been phosphorylated at the carboxyl
tail. Furthermore, the autophosphorylated T757 is
capable of binding SH2-containing signal transduction
proteins. These experiments suggest a model in which
oligomerization allows trans-phosphorylation of binding partners which is dependent on the intrinsic kinase
activity of the erbB receptor. That the introduction of
a kinase-de®cient neu leads to an increase in
transforming potency in the double mutant cell lines
suggests that transformation may not be directly
dependent on the intrinsic kinase activity of the erbB
receptor, but rather on the docking of other signaling
molecules at erbB autophosphorylation sites.
The critical aspect of erbB oligomerization may be
juxtaposition of the kinase domains of binding
partners. Studies of mutant neu forms in which the
intracellular domain has been deleted, such as
T691stop from our laboratory, support the notion that
kinase domain interaction is critical. Coexpression of
T691stop along with oncogenic neu results in a loss of
transforming activity, the inhibition observed being
dependent on the ratio of truncated and full-length
receptor (Qian et al., 1996). Stable dimers consisting of
T691stop and p185neu are observed, and likely
preclude the interaction of p185neu kinase domains,
preventing transphosphorylation. The end result of this
situation is the formation of dimers which do not
contain phosphorylation-dependent docking sites,
much the same as in cells expressing only kinaseinactive neu forms.
Heterodimerization of HER2/neu and EGFR
Early observations suggested that protooncogenic p185
activation is EGF dependent, despite that EGF does
not bind p185, implying that EGFR may be involved
in p185 signal transduction (Dougall et al., 1994).
Transformation by either protooncogenic p185 or
EGFR alone requires high receptor expression; however, expression of both receptors at moderate levels is
sucient to cause transformation (Kokai et al., 1989),
which is dependent on continuous cell surface expression of both receptors (Wada et al., 1990). The
apparent synergy in transformation observed with
erbB-2 and EGFR may be particularly signi®cant, as
these two receptors have been shown to be upregulated
concomitantly in breast as well as other tumors
(Dougall et al., 1994; Klapper et al., 2000).
Investigating the link between p185 and EGFR
signaling, our laboratory identi®ed a protein ± protein
interaction between the two receptors using crosslinking reagents (Wada et al., 1990). The observed
interaction was found to be stimulated by EGF (Qian
et al., 1992; Spivak-Kroizman et al., 1992; Wada et al.,
1990), though heterodimers were also detected in the
absence of EGF (Qian et al., 1994a). These results were
con®rmed by non-denaturing gel electrophoresis, and
the interactions found to be non-covalent in nature
(Qian et al., 1992). Heterodimerization of soluble
extracellular domains of HER2/neu and EGFR has
also been demonstrated (Spivak-Kroizman et al., 1992).
In addition to the two EGF anities observed for cells
expressing only the EGFR, three anity states were
identi®ed in cells expressing heterodimers, the additional anity state being higher than those observed
with EGFR only (Wada et al., 1990); this suggests that
oligomers which contain erbB-2 have a higher anity
for ligand, an observation which has been con®rmed
and extended to other erbB-2-containing erbB family
heterodimers (Jones et al., 1999). Heterodimers of
EGFR and p185 are also formed preferentially over
either homodimeric form (Qian et al., 1994b), which
may further contribute to their transforming potential.
Since this initial observation, a series of mutants
have been constructted in order to investigate the
functional interaction between p185neu and EGFR.
When expressed with EGFR, a kinase-inactive HER2/
neu mutant, N757, was found to be trans-phosphorylated in and EGF-dependent fashion, suggesting that
EGFR may be responsible for HER2/neu phosphorylation in the wild type oligomer (Qian et al., 1994b).
An intracellular truncation of p185, N691stop, also
underwent EGF-dependent dimerization with EGFR,
though the resulting complex was not active (Qian et
al., 1994b); this suggests that activation is a transprocess, and that the absence of a binding partner
precludes activation, auto or otherwise. These results
HER2/Neu: mechanisms of dimerization/oligomerization
PJ Brennan et al
provided evidence that erbB-2 is involved in the
reciprocal activation of EGFR, showing that it is
more than a mere substrate of the EGFR. Further
studies with the N757 and N691stop mutants provide
evidence for the functional importance of heterodimer
formation; each of these mutations abolished the
transformation and tumorigenicity seen with formation of the wild type heterodimer (Qian et al., 1994a).
The dominant-negative interaction of the truncated
N691stop, e€ective in both cis and trans, could
potentially be used therapeutically. T691stop, a
version of this mutant with the transmembrane
mutation observed in rat neu, binds in a ligandindependent fashion, and should therefore be an even
more e€ective inhibitor of erbB receptor function. To
test the therapeutic potential of this construct in a
pathologic setting, O'Rourke et al. (1997) expressed
T691stop in a cell line established from human
glioblastoma; inhibition of transformation and tumorigenicity was observed.
The interactions observed between p185/EGFR
heterodimers appear to be conceptually parallel to
those observed with homodimers of either receptor.
Similar to the studies outlined above, experiments have
been done utilizing EGFR mutants, and the results are
consistent with those obtained using mutant erbB-2
receptors. Coexpression of HER2/neu with an EGFR
lacking most of the intracellular domain demonstrated
that the kinase activity of HER2/neu could be inhibited
in a trans-negative fashion, similar to the e€ects of the
N691stop p185 mutant. Coexpression of a kinasede®cient EGFR with HER2/neu led to trans-phosphorylation of the mutant EGFR (Spivak-Kroizman et al.,
1992). Interestingly, tyrosine-phosphorylated dimers of
kinase-de®cient EGFR were observed in these cells,
which is surprising, given the inability of these
receptors to trans-phosphorylate. It is possible that
the receptors were phosphorylated as part of a larger
HER2/neu-containing oligomeric complex. A second
possibility which would explain this observation
involves secondary dimerization, which involves an
activated receptor dissociating from an oligomeric
complex and subsequently phosphorylating a previously inactive receptor; this idea has been supported
for erbB family proteins (Gamett et al., 1997).
Oligomerization
As discussed above, activation of erbB-2 receptors
clearly involves the formation of dimeric receptor
complexes, the dimer serving as the minimal functioning unit for erbB receptor signaling. Several lines of
evidence, however, suggest that higher order aggregated structures, in addition to dimers, may function
in erbB signaling. The idea that aggregation state may
function in signal regulation is not novel; for example,
it has been suggested that E. coli Tar receptor, which
exists as a dimer, forms tetramers in its highest
activity state (Cochran and Kim, 1996). Though no
de®nitive evidence has been produced demonstrating a
functional role for higher-order aggregation states in
the erbB family, the evidence which exists is extremely
provocative. LeVea et al. (1993) working with the fulllength neu receptor generated in baculovirus, found
that the transmembrane mutation led to an increase in
aggregated receptor forms. In further studies using
these puri®ed receptors, higher order oligomerization
states were again observed (Samanta et al., 1994).
Similar results have been seen using soluble EGFR
fragments, in which ligand-induced dimers and trimers
were demonstrated after chemical crosslinking (Hurwitz et al., 1991). Using a structure-based analysis,
Lax et al. (1991) have also demonstrated the
formation of dimers, trimers, and higher order
oligomers using a soluble EGFR extracellular domain
fragment. Though these results using puri®ed receptors are suggestive, there are obvious limitations to
such in vitro studies.
Additional evidence for oligomerization comes from
an early study into the kinetics of EGF binding to
p185/EGFR heterodimers. As mentioned previously,
Wada et al. (1990) described three binding anities for
EGF in cells expressing both EGFR and p185.
Monoclonal antibody treatment, led to the removal
of p185 from the surface of the cells without a€ecting
EGFR levels, as this antibody is thought to bind only
to p185 homodimers. After treatment with this antibody, only two anity states remained. A possible
model which could explain this binding behavior would
predict that the high anity binding site was provided
by a heterooligomeric complex consisting of two
homodimers. The situation, however, is complex and
would require further, more directed studies in order to
con®dently draw such a conclusion.
More physiologically relevant data supporting receptor oligomerization has come from imaging studies
of EGFR receptors, using both electron microscopy
(van Belzen et al., 1988) and time-resolved ¯uorescence
imaging microscopy (Gadella and Jovin, 1995). Fluorescence imaging microscopy suggests that EGF treatment quickly causes EGFR aggregation/clustering, and
that a subset of EGFR receptors are pre-dimerized
(Gadella and Jovin, 1995). These studies raise the
possibility that erbB signaling in intact cells, while
requiring dimer formation, may actually be functionally activated by aggregation of receptor complexes,
complexes which need not contain only members of the
erbB family. Although no microscopic studies have
directly addressed the aggregation of erbB-2 receptors,
oligomerization would likely represent a general
mechanism of activation within and likely beyond the
erbB family, and thus its demonstration with any
member of the erbB family would greatly support the
importance of this phenomenon.
To further test the possibility that erbB protein can
undergo higher order aggregation, we used molecular
modeling to assess the energetic and structural
feasibility of such a complex (Murali et al., 1996).
Modeling using the intracelluar domains of EGFR and
p185neu predicts that a tetrameric complex with a 2 : 2
EGFR : p185neu stoichiometry would be energetically
favorable. An interesting aspect of this model is the
prediction that the interactions between subunits di€ers
between the dimeric and the tetrameric forms; in the
tetrameric form, the interactions are predicted to
involve the autophosphorylation tyrosines. The possibility exists, then that tetramerization occurs only after
dimerization and cross-phosphorylation, and that this
transition is involved in signal regulation. Tetramerization/oligomerization could prove to be a central
principle in receptor signaling, and its further investigation should prove fascinating.
6099
Oncogene
HER2/Neu: mechanisms of dimerization/oligomerization
PJ Brennan et al
6100
Conclusions and prospectus
Though much has been discovered regarding the
mechanisms employed by the erbB family of receptors,
many basic questions remain. An erbB crystal structure
has been elusive, and though modeling insights have
been helpful, crystallographic data would likely greatly
advance current notions. Resolving the normal and
tumorigenic receptor and ligand combinations involved
in the erbB signaling network should also be of value.
Additonally, although many candidate downstream
signal transduction pathways have been identi®ed, the
relative importance of each of these pathways to
normal function and malignancy remains largely
unknown.
Our current understanding of HER2/neu signaling
has already translated into clinical success. That a
threshold of HER2/neu overexpression is necessary for
transformation suggested that any reduction in HER2/
neu signaling may lead to phenotypic reversion.
Therapy directed toward this end, Herceptin, an antiHER2/neu monoclonal antibody, has proven clinical
e€ectiveness in the treatment of a subset of breast
cancers (Shak, 1999). Current and future research into
the mechanisms by which HER2/neu promotes tumorigenesis will most likely lead to additional therapeutics.
For example, since it appears that oligomerization is
necessary for erbB signal transduction, agents capable
of preventing oligomerization should also prevent
signaling. Tumors which overexpress both HER2/neu
and EGFR represent another clinical target, possibly
involving bispeci®c antibodies directed at heterooligomers of these receptors. If future trends follow those
that have already passed, further elucidation of erbB
signaling should prove both scienti®cally and medically
rewarding.
References
Adelsman MA, Huntley BK and Maihle NJ. (1996). J. Virol.,
70, 2533 ± 2544.
Akiyama T, Matsuda S, Namba Y, Saito T, Toyoshima K
and Yamamoto T. (1991). Mol. Cell. Biol., 11, 833 ± 842.
Bargmann CI, Hung MC and Weinberg RA. (1986a). Cell,
45, 649 ± 657.
Bargmann CI, Hung MC and Weinberg RA. (1986b).
Nature, 319, 226 ± 230.
Bargmann CI and Weinberg RA. (1988). EMBO J., 7, 2043 ±
2052.
Beguinot L, Vass WC and Miloso M. (1995). J. Cell.
Biochem., 19A (Suppl), 11.
Boni-Schnetzler, M and Pilch PF. (1987). Proc. Natl. Acad.
Sci. USA, 84, 7832 ± 7836.
Brandt-Rauf PW, Pincus MR, Chen JM and Lee G. (1989).
J. Protein Chem., 8, 679 ± 688.
Brandt-Rauf PW, Pincus MR and Monaco R. (1995). J.
Protein Chem., 14, 33 ± 40.
Brandt-Rauf PW, Rackovsky S and Pincus MR. (1990).
Proc. Natl. Acad. Sci. USA, 87, 8660 ± 8664.
Brown PM, Debanne MT, Grothe S, Bergsma D, Caron M,
Kay C and O'Connor-McCourt MD. (1994). Eur. J.
Biochem., 225, 223 ± 233.
Burke CL, Lemmon MA, Coren BA, Engelman DM and
Stern DF. (1997). Oncogene, 14, 687 ± 696.
Burke CL and Stern DF. (1998). Mol. Cell. Biol., 18, 5371 ±
5379.
Canals F. (1992). Biochemistry, 31, 4493 ± 4501.
Cao H, Bangalore L, Bormann BJ and Stern DF. (1992).
EMBO J., 11, 923 ± 932.
Carraway KLD and Cerione RA. (1993). J. Biol. Chem., 268,
23860 ± 23867.
Carraway KLD, Koland JG and Cerione RA. (1989). J. Biol.
Chem., 264, 8699 ± 8707.
Chamberlin SG and Davies DE. (1998). Biochim. Biophys.
Acta, 1384, 223 ± 232.
Chantry A. (1995). J. Biol. Chem., 270, 3068 ± 3073.
Chen LI, Webster MK, Meyer AN and Donoghue DJ.
(1997). J. Cell. Biol., 137, 619 ± 631.
Cochran AG and Kim PS. (1996). Science, 271, 1113 ± 1116.
Di Fiore PP, Pierce JH, Fleming TP, Hazan R, Ullrich A,
King CR, Schlessinger J and Aaronson SA. (1987a). Cell,
51, 1063 ± 1070.
Di Fiore PP, Pierce JH, Kraus MH, Segatto O, King CR and
Aaronson SA. (1987b). Science, 237, 178 ± 182.
Di Marco E, Pierce JH, Knicley CL and Di Fiore PP. (1990).
Mol. Cell. Biol., 10, 3247 ± 3252.
Oncogene
Doherty JK, Bond C, Jardim A, Adelman JP and Clinton
GM. (1999). Proc. Natl. Acad. Sci. USA, 96, 10869 ±
10874.
Domagala T, Konstantopoulos N, Smyth F, Jorissen RN,
Fabri L, Geleick D, Lax I, Schlessinger J, Sawyer W,
Howlett GJ, Burgess AW and Nice EC. (2000), Growth
Factors, 18, 11 ± 29.
Dougall WC, Qian X, Petereson NC, Miller MJ, Samantha A
and Greene MI. (1994). Oncogene, 9, 2109 ± 2123.
Fanger BO, Stephens JE and Staros JV. (1989). FASEB J., 3,
71 ± 75.
Gadella Jr TW and Jovin TM. (1995). J. Cell. Biol., 129,
1543 ± 1558.
Camett DC, Pearson G, Cerione RA and Friedberg I. (1997).
J. Biol. Chem., 272, 12052 ± 12056.
Garrett TP, McKern NM, Lou M, Frenkel MJ, Bentley JD,
Lovrecz GO, Elleman TC, Cosgrove LJ and Ward CW.
(1998). Nature, 394, 395 ± 399.
Graus-Porta D, Beerli RR, Daly JM and Hynes NE. (1997).
EMBO J., 16, 1647 ± 1655.
Green®eld C, Hiles I, Water®eld MD, Federwisch M,
Wollmer A, Blundell TL and McDonald N. (1989). EMBO
J., 8, 4115 ± 4123.
Hurwitz DR, Emanuel SL, Nathan MH, Sarver N, Ullrich A,
Felder S, Lax I and Schlessinger J. (1991). J. Biol. Chem.,
266, 22035 ± 22043.
Jones JT, Akita RW and Sliwkowski MX. (1999). FEBS
Lett., 447, 227 ± 231.
Jorissen RN, Epa VC, Treutlein HR, Garrett TP, Ward CW
and Burgess AW. (2000). Protein Sci., 9, 310 ± 324.
Kashles O, Szapary D, Bellot F, Ullrich A, Schlessinger J and
Schmidt A. (1988). Proc.Natl. Acad. Sci. USA, 85, 9567 ±
9571.
Kashles O, Yarden Y, Fischer R, Ullrich A and Schlessinger
J. (1991). Mol. Cell. Biol., 11, 1454 ± 1463.
Klapper LN, Kirschbaum MH, Sela M and Yarden Y.
(2000). Adv. Cancer Res., 77, 25 ± 79.
Kokai Y, Myers JN, Wada T, Brown VI, LeVea CM, Davis
JG, Dobashi K and Greene MI. (1989). Cell, 58, 287 ± 292.
Koland JG and Cerione RA. (1988). J. Biol. Chem., 263,
2230 ± 2237.
Lax I, Bellot F, Howk R, Ullrich A, Givol D and
Schlessinger J. (1989). EMBO J., 8, 421 ± 427.
Lax I, Burgess WH, Bellot F, Ullrich A, Schlessinger J and
Givol D. (1988). Mol. Cell. Biol., 8, 1831 ± 1834.
HER2/Neu: mechanisms of dimerization/oligomerization
PJ Brennan et al
Lax I, Mitra AK, Ravera C, Hurwtiz DR, Rubinstein M,
Ullrich A, Stroud RM and Schlessinger J. (1991). J. Biol.
Chem., 266, 13828 ± 13833.
Lee DC, Fenton SE, Berkowitz EA and Hissong MA. (1995).
Pharmacol Rev., 47, 51 ± 85.
Lemmon MA, Bu Z, Ladbury JE, Zhou M, Pinchasi D, Lax
I, Engelman DM and Schlessinger J. (1997). EMBO J., 16,
281 ± 294.
LeVea CM, Myers JN, Dougall WC, Qian X and Greene MI.
(1993). Receptor, 3, 293 ± 309.
Livnah O, Johnson DL, Stura EA, Farrell FX, Barbone FP,
You Y, Liu KD, Goldsmith MA, He W, Krause CD,
Pestka S, Jolli€e LK and Wilson IA. (1998). Nat. Struct.
Biol., 5, 993 ± 1004.
Lofts FJ, Hurst HC, Stermberg MJ and Gullick WJ. (1993).
Oncogene, 8, 2813 ± 2820.
Messerle K, Schlegel J, Hynes NE and Groner B. (1994).
Mol. Cell. Endocrinol, 105, 1 ± 10.
Mohammadi M, Honegger A, Sorokin A, Ullrich A,
Schlessinger J and Hurwitz DR. (1993). Biochemistry,
32, 8742 ± 8748.
Moscatello DK, Montgomery RB, Sundareshan P, McDanel
H, Wong MY and Wong AJ. (1996). Oncogene 13, 85 ± 96.
Murali R, Brennan PJ, Kieber-Emmons T and Greene MI.
(1996). Proc. Natl. Acad. Sci. USA, 93, 6252 ± 6257.
O'Rourke DM, Qian X, Zhang HT, Davis JG, Nute E,
Meinkoth J and Greene MI. (1997). Proc. Natl. Acad. Sci.
USA, 94, 3250 ± 3255.
Odaka M, Kohda D, Lax I, Schlessinger J and Inagaki F.
(1997). J. Biochem. (Tokyo), 122, 116 ± 121.
Padhy LC, Shih C, Cowing D, Findelstein R and Weinberg
RA. (1982). Cell, 28, 865 ± 871.
Qian X, Dougall WC, Fei Z and Greene MI. (1995).
Oncogene, 10, 211 ± 219.
Qian X, Dougall WC, Hellman ME and Greene MI. (1994a).
Oncogene, 9, 1507 ± 1514.
Qian X, LeVea CM, Freeman JK, Dougall WC and Greeene
MI. (1994b). Proc. Natl. Acad. Sci. USA, 91, 1500 ± 1504.
Qian X, O'Rourke DM, Fei Z, Zhang HT, Kao CC and
Greene MI. (1999). J. Biol. Chem., 274, 574 ± 583.
Qian X, O'Rourke DM, Zhao H and Greene MI. (1996).
Oncogene, 13, 2149 ± 2157.
Qian XL, Decker SJ and Greene MI. (1992). Proc. Natl.
Acad. Sci. USA, 89, 1330 ± 1334.
Sajot N and Genest M. (2000). Eur. Biophys, J., 28, 648 ± 662.
Sako Y, Minoghchi S and Yanagida T. (2000). Nat. Cell.
Biol., 2, 168 ± 172.
Samanta A, LeVea CM, Dougall WC, Qian X and Greene
MI. (1994). Proc. Natl. Acad. Sci. USA, 91, 1711 ± 1715.
Saxon ML and Lee DC. (1999). J. Biol. Chem., 274, 28356 ±
28362.
Schechter AL, Stern DF, Vaidyanathan L, Decker SJ,
Drebin JA, Greene MI and Weinberg RA. (1984). Nature,
312, 513 ± 516.
Schreiber AB, Lax I, Yarden Y, Eshhar Z and Schlessinger J.
(1981). Proc. Natl. Acad. Sci. USA, 78, 7535 ± 7539.
Segatto O, King CR, Pierce JH, Di Fiore PP and Aaronson
SA. (1988). Mol. Cell. Biol., 8, 5570 ± 5574.
Shak S. (1999). Semin. Oncol., 26, 71 ± 77.
Shih C, Padhy LC, Murray M and Weinberg RA. (1981).
Nature, 290, 261 ± 264.
Shih C, Shilo BZ, Goldfarb MP, Dannenberg A and
Weinberg RA. (1979). Proc. Natl. Acad. Sci, USA, 76,
5714 ± 5718.
Siegel PM and Muller WJ. (1996). Proc. Natl. Acad. Sci.
USA, 93, 8878 ± 8883.
Smith SO, Smith CS and Bormann BJ. (1996). Nat. Struct.
Biol., 3, 252 ± 258.
Sorokin A. (1995). Oncogene, 11, 1531 ± 1540.
Sorokin A, Lemmon MA, Ullrich A and Schlessinger J.
(1994). J. Biol. Chem., 269, 9752 ± 9759.
Spaargaren M, De®ze LH, Boonstra J and de Laat SW.
(1991). J. Biol. Chem., 266, 1733 ± 1739.
Spivak-Kroizman T, Rotin D, Pinchasi D, Ullrich A,
Schlessinger J and Lax I. (1992). J. Biol. Chem., 267,
8056 ± 8063.
Sternberg MJ and Gullick WJ. (1989). Nature, 339, 587.
Summer®eld AE, Hudnall AK, Lukas TJ, Guyer CA and
Staros JV. (1996). J. Biol. Chem., 271, 19656 ± 19659.
Tzahar E, Pinkas-Kramarski R, Moyer JD, Klapper LN,
Alroy I, Levkowitz G, Shelly M, Henis S, Eisenstein M,
Ratzkin BJ, Sela M, Andrews GC and Yarden Y. (1997).
EMBO J., 16, 4938 ± 4950.
Tzahar E, Waterman H, Chen X, Levkowitz G, Karunagaran D, Lavi S, Ratzkin BJ and Yarden Y. (1996). Mol. Cell.
Biol., 16, 5276 ± 5287.
van Belzen N, Rijken PJ, Hage WJ, de Laat SW, Verkleij AJ
and Boonstra J. (1988). J. Cell. Physiol., 134, 413 ± 420.
Wada T, Qian XL and Greene MI. (1990). Cell. 61, 1339 ±
1347.
Weber W, Bertics PJ and Gill GN. (1984). J. Biol. Chem.,
259, 14631 ± 14636.
Weiner DB, Kokai Y, Wada T, Cohen JA, Williams WV and
Greene MI. (1989a). Oncogene, 4, 1175 ± 1183.
Weiner DB, Liu J, Cohen JA, Williams WV and Greene MI.
(1989b). Nature, 339, 230 ± 231.
Yarden Y and Schlessinger J. (1987a). Biochemistry, 26,
1443 ± 1451.
Yarden Y and Schlessinger J. (1987b). Biochemistry, 26,
1434 ± 1442.
6101
Oncogene