(Received for publication, October 30, 1996, and in revised form, January 16, 1997)
From the Cancer Research Program and the
¶ Neurobiology Research Program, Garvan Institute of Medical
Research, St. Vincent's Hospital, Sydney, New South Wales 2010, Australia and the § Growth Regulation Laboratory, Ludwig
Institute for Cancer Research, P. O. Royal Melbourne Hospital,
Victoria 3050, Australia
The Src homology 2 (SH2) domain-containing
protein Grb7 and the erbB2 receptor tyrosine kinase are overexpressed
in a subset of human breast cancers. They also co-immunoprecipitate
from cell lysates and associate directly in vitro. Whereas
the Grb7 SH2 domain binds strongly to erbB2, the SH2 domain of Grb14, a
protein closely related to Grb7, does not. We have investigated the
preferred binding site of Grb7 within the erbB2 intracellular domain
and the SH2 domain residues that determine the high affinity of Grb7 compared with Grb14 for this site. Phosphopeptide competition and
site-directed mutagenesis revealed that Tyr-1139 of erbB2 is the major
binding site for the Grb7 SH2 domain, indicating an overlap in binding
specificity between the Grb7 and Grb2 SH2 domains. Substituting
individual amino acids in the Grb14 SH2 domain with the corresponding
residues from Grb7 demonstrated that a Gln to Leu change at the D6
position imparted high affinity erbB2 interaction, paralleled by a
marked increase in affinity for the Tyr-1139 phosphopeptide. The
reverse switch at the
D6 position abrogated Grb7 binding to erbB2.
This residue therefore represents an important determinant of SH2
domain specificity within the Grb7 family.
Src homology 2 (SH2)1 domains are conserved noncatalytic regions of approximately 100 amino acids found in a variety of cytoplasmic signaling proteins. They bind to specific phosphotyrosine-containing sequences within autophosphorylated RTKs and intracellular phosphoproteins and, along with SH3 and pleckstrin homology domains, mediate inter- and intramolecular interactions involved in signal transduction from activated RTKs (1, 2).
Grb7 is an SH2 domain-containing signaling protein encoded by a gene commonly co-amplified with the ERBB2 gene in human breast cancer cell lines and primary breast cancers (3). Overexpression of the erbB2 receptor tyrosine kinase occurs in approximately 20% of breast cancers, where increased expression correlates with poor patient prognosis (4-6). Grb7 also associates strongly with erbB2 via its SH2 domain in co-immunoprecipitation experiments (3), although the Grb7 binding site on erbB2 has not been determined. The simultaneous overexpression of Grb7 with erbB2 in breast cancer cells therefore suggests greatly amplified signaling through these proteins. Although the precise function of Grb7 is not known, it probably serves as an "adapter" protein, linking tyrosine phosphorylated proteins to downstream effectors, since it lacks a known catalytic region but possesses multiple domains capable of mediating intermolecular interactions.
Grb7 is a member of an emerging family of signaling proteins that also includes Grb10 (7) and Grb14 (8). Another related protein, Grb-IR, is highly conserved with mouse Grb10 and may represent an alternatively spliced version of the human homologue (9). The family members share a common overall structure consisting of an N-terminal region harboring a conserved proline-rich motif (8), a central region exhibiting homology to the Caenorhabditis elegans protein Mig10 that also contains a pleckstrin homology domain, and a C-terminal SH2 domain. The SH2 domains of Grb7 and Grb14 share 67% amino acid identity, yet despite this high conservation, fusion proteins containing these SH2 domains display differences in binding preference for RTKs in whole cell extracts. In particular, the Grb7 SH2 domain interacts strongly with erbB2 in vitro, whereas the affinity of the Grb14 SH2 domain for erbB2 is very weak (8). This therefore represents a novel system in which to study the determinants of SH2 specificity that enable two highly conserved SH2 domains to retain distinct RTK binding preferences.
The specificity of high affinity SH2 binding is conferred both by amino
acids in the SH2 domain and by residues flanking the phosphotyrosine in
the target sequence. In the target sequence, the amino acids
immediately C terminal to the phosphotyrosine are most important in
controlling specificity, especially the first three to six residues, as
shown by structural studies (10, 11) and experiments using
phosphopeptide libraries randomized at +1 to +3 (12, 13). Within the
SH2 domain, selectivity is determined by variation of specific residues
that interact with these positions of the phosphopeptide, whereas the
overall structure is well conserved. All SH2 domain structures solved
to date consist of a large central sheet with an associated smaller
sheet flanked by two
helices (10, 11, 14-16). One residue
important for specificity is
D5, which interacts with both the +1
and +3 positions (10). Group I SH2 domains are distinguished from other types by possessing a residue with a bulky hydrophobic side chain at
the
D5 position (Tyr or Phe) that acts as a divider between two
pockets that bind the phosphotyrosine and +3 positions. In Group III
SH2 domains this is absent, creating a groove between the binding
pockets that can extend out beyond +3 to the +6 position (11, 15).
Introduction of a tyrosine at
D5 into a Group III SH2 domain is
sufficient to shift its phosphopeptide specificity to resemble that of
a Group I SH2 domain (17).
In this study we investigated the preferred binding site of the Grb7
SH2 domain in erbB2 and identified amino acids in the SH2 domain
important for determining the high affinity binding of Grb7 to this
site compared with Grb14. Phosphopeptide competition and site-directed
mutagenesis revealed that Tyr-1139 in erbB2 is specifically bound by
the Grb7 SH2 domain. Interestingly, this also corresponds to the major
Grb2 binding site (18). Six SH2 domain residues implicated in
determination of SH2 domain binding specificity were replaced in Grb14
with the corresponding amino acid from Grb7. Two substitutions
increased binding of the Grb14 SH2 domain to erbB2, and the reciprocal
change at one of these positions (D6) abrogated Grb7 binding to
erbB2. The
D6 amino acid therefore plays an important role in
determining the affinities of the Grb7 and Grb14 SH2 domains for
Tyr-1139 of erbB2, and possible mechanisms for its action are
discussed.
HER14, HER1-2, and HEK 293 cells were maintained and subjected to growth factor stimulation as described previously (8, 19, 20). SK-BR-3 human breast cancer cells were obtained from the American Type Culture Collection (Rockville, MD) and maintained according to Janes et al. (21). Lysates were prepared as described previously (21) and normalized for protein content using a Bradford-based protein assay (Bio-Rad).
Transfection ProceduresHEK 293 cells were plated at a density of 1.6 × 106/dish on 10-cm-diameter tissue-culture dishes. Prior to transfection the medium was changed to Dulbecco's modified Eagle's medium (CSL Biosciences, Parkville, Victoria, Australia) containing 5% fetal calf serum (CSL Biosciences). The cells were then transfected for 6 h using the calcium phosphate co-precipitation method (22), treated with 15% glycerol in 5% fetal calf serum/Dulbecco's modified Eagle's medium for 1 min at 37 °C, and then returned to maintenance medium for 24 h. The precipitates contained 5 µg of expression vectors encoding HER1-2 (23) or HER1-2 Y1139F, made up to 20 µg with vector DNA, whereas control cells received 20 µg of vector DNA alone. Prior to EGF treatment the cells were starved overnight in medium containing 0.5% fetal calf serum. Receptor expression and stimulation was confirmed by Western blotting cell lysates with anti-erbB2 (Novocastra, Newcastle, United Kingdom) and anti-phosphotyrosine (PY20, Transduction Laboratories, Lexington, KY) antibodies. Visualization of bound antibodies was by ECL (Amersham Corp.).
Generation of GST Fusion ProteinsA DNA fragment
corresponding to the SH2 domain of Grb14 (amino acids 426-540) was
amplified from GRB14 cDNA (8) using flanking primers
containing BamHI (forward primer) and EcoRI
(reverse primer) restriction sites to enable subcloning into the pGEX2T
expression vector (Amrad-Pharmacia Biotech Inc., Melbourne, Victoria,
Australia). DNA encoding the human Grb7 SH2 domain (amino acids
415-532) and the mouse Grb7 SH2 domain (amino acids 418-535) was
amplified by reverse transcription-polymerase chain reaction from
SK-BR-3 human breast cancer cell RNA and mouse liver RNA, respectively, using primers containing appropriate restriction sites and cloned into
pGEX2T. The mouse and human Grb7 SH2 domains were used interchangeably with identical results. Recombinant pGEX plasmids were transformed into
Escherichia coli DH5 cells and verified by DNA
sequencing. Fusion proteins were expressed and purified from
isopropyl-
-D-thiogalactopyranoside-induced bacterial
cultures as described previously (24), and purity was confirmed by
SDS-PAGE and silver staining.
Binding experiments were typically performed by mixing 2.5-5 µg of GST-SH2 fusion protein bound to glutathione-Sepharose beads (Sigma) with 150-300 µl of lysate (approximately 5 mg/ml total protein) for 2 h at 4 °C. The beads were then washed three times with cell-lysis buffer (21) and boiled for 3 min in SDS-PAGE sample buffer. Bound proteins were separated by SDS-PAGE, transferred to nitrocellulose, and Western blotted with an anti-erbB2 monoclonal antibody. In experiments using different GST fusion proteins, equal loading was verified by SDS-PAGE followed by Coomassie Blue staining.
Peptide Competition AssaysSynthetic phosphopeptides were
synthesized by Chiron Mimotopes (Clayton, Victoria, Australia). These
were purified by reverse phase high pressure liquid chromatography to
95% and their identity confirmed by ion spray mass spectrometry.
Peptide competition of GST fusion protein binding to the erbB2
intracellular domain was performed as follows. 5 µg of fusion protein
on glutathione-Sepharose beads was preincubated for 30 min in 300 µl
of lysis buffer with each individual peptide or with no peptide
(control), and this was then added to 300 µl of HER1-2 cell lysate
and mixed for 2 h at 4 °C (final peptide concentration of 50 µM). The beads were then washed, and bound proteins were
analyzed as described for the fusion protein binding assays.
Densitometric analysis of autoradiographs was performed using the IP
Lab Gel analysis program (Signal Analytics Corp., Vienna, VA).
Site-directed mutagenesis of recombinant pGEX plasmids encoding the Grb14 and Grb7 SH2 domains and the HER1-2 expression plasmid was performed using a method that incorporates both the mutagenic primer and a selection primer, allowing selection of mutated plasmids due to alteration of a restriction site in the vector sequence (Transformer System, Clontech, Palo Alto, CA). Sequences of mutagenic and selection primers will be made available on request. All mutated SH2 domains were fully sequenced to verify correct incorporation.
BIAcore Affinity MeasurementsThe general operation
principles of the BIAcore biosensor (Pharmacia) have been described
previously (25). Parallel channels of CM-5 sensor chips were
derivatized (45 µl at 2 µl/min) with phosphorylated or
nonphosphorylated peptide solutions (2 mg/ml in 50 mM
HEPES, pH 7.5, 150 mM NaCl) to yield a 250-350 resonance units response increase. The integrity and accessibility of the phosphopeptide was confirmed routinely by monitoring the response to an
anti-phosphotyrosine antibody (Upstate Biotechnology, Lake Placid, NY).
Interaction of soluble GST fusion proteins (0.01-10 µM)
with immobilized peptide was measured at 5 µl/min in 20 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM
dithiothreitol, 0.005% Tween 20 with kinetic constants derived from
raw data of the BIAcore sensograms using kinetic models included in the
BIAevaluation software (version 2.1, 1995, Pharmacia). Equilibrium
constants were derived from steady state binding responses by Scatchard analysis, according to Req/C = KaRmax KaReq (where
Req is the equilibrium response, C is
the analyte concentration, Rmax is the
saturation response, and Ka is the association constant), as outlined in the BIAcore user manual.
In addition, the interactions of the fusion proteins with Tyr-1139 phosphopeptide were studied in solution by reacting constant fusion protein concentrations with increasing concentrations of soluble peptide (0.016-20 µM). Free analyte concentrations, estimated from the BIAcore responses of known samples (BIAevaluation software, version 2.1) were used to calculate the concentration of SH2 domain-bound receptor peptide and yielded the equilibrium constant Kd by Scatchard analysis as described previously (25).
Molecular ModelingStructures of the Grb7 and Grb14 SH2 domains were modeled on the Src SH2 domain structure by substituting sequences of the SH2 domain and the complexed peptide. We assumed that the overall fold was strictly conserved in terms of secondary structure elements and did not attempt to move the backbone in these regions but rather considered compatible rearrangements of side chains (according to Homology Users Guide version 95.0, Molecular Simulations Inc., San Diego, CA). Sequence alignment of Grb7 and Grb14 with Src showed differences in the lengths of the CD and EF loops, which in the case of the EF loop led to clashes with the peptide, so it was assumed that structural homology was conserved in this area and that the insertion in Grb7 and Grb14 is later in the EF loop than is shown in the alignment. Side chains identical between Src and Grb7 or Grb14 were copied from the Src structure, whereas all nonidentical side chains were adjusted using a side chain rotamer library (26), an automated procedure that attempts to minimize the energy of groups of side chains by selecting rotamers in a systematic fashion. Figures were generated using the programs MOLSCRIPT (27) and Insight II (Molecular Simulations Inc.).
The specific site (or sites) within the erbB2 receptor recognized by Grb7 has not been clearly determined. The Grb7 binding site in Shc is likely to be the YVNV motif at Tyr-317, since Grb2, which recognizes this motif, can compete with Grb7 for binding to Shc (3). A Grb2 binding site comprising a similar motif (YVNQ) is also present at Tyr-1139 of erbB2 (18) and is therefore a likely candidate as a Grb7 binding site. However, Stein et al. (3) found that Grb2 was unable to compete for binding of Grb7 to erbB2, suggesting that either Grb7 binds to this site in erbB2 at a much higher affinity than Grb2 or that Grb7 recognizes a different site.
Five in vivo tyrosine autophosphorylation sites have been
identified in the C terminus of erbB2, at positions 1139, 1196, 1221/1222, and 1248 (28, 29), and an additional site of in vitro phosphorylation maps to Tyr-1023 (29) (Fig.
1A). We therefore used synthetic
phosphopeptides spanning these tyrosine residues to compete for binding
of a GST-Grb7 SH2 fusion protein to the autophosphorylated erbB2
intracellular domain in cell lysates. Lysates of HER1-2 cells were used
that express a chimera (HER1-2) of the extracellular domain of the EGF
receptor fused to the intracellular domain of erbB2, enabling study of
erbB2 interactions in a ligand (EGF)-inducible manner (20). Lysates
from EGF-treated cells were incubated with a Grb7 SH2 fusion protein
coupled to Sepharose beads, and associated receptor was detected by
SDS-PAGE and Western blotting. When the Grb7 SH2 fusion protein was
preincubated with each peptide individually, only the Tyr-1139 peptide
inhibited binding (Fig. 1B). This strongly suggests that the
preferred binding site of the Grb7 SH2 domain is the YVNQ motif at
Tyr-1139 of erbB2.
To verify this result, the Tyr-1139 residue of the HER1-2 chimera was
mutated to Phe to create HER1-2 Y1139F, and the receptor was then
tested for its association with the Grb7 SH2 domain. Plasmid constructs
encoding wild type HER1-2 and HER1-2 Y1139F were transiently
transfected into HEK 293 cells and lysates prepared from control and
EGF-stimulated cells. Western blotting of these lysates with either
anti-erbB2 or anti-phosphotyrosine antibodies confirmed equivalent
expression and tyrosine phosphorylation of the wild type and mutated
receptors (Fig. 2A). Incubation of the lysates with GST-Grb7 SH2 fusion protein immobilized on Sepharose beads
followed by detection of bound receptor with anti-erbB2 antibodies
revealed that introduction of the Y1139F mutation reduced binding by
approximately 95% compared with the wild type receptor (Fig.
2B), confirming that Tyr-1139 is the major Grb7 binding site
in erbB2.
Substitution of Specific Grb14 SH2 Domain Residues with the Corresponding Amino Acids from Grb7 Increases erbB2 Association
Having demonstrated binding of Grb7 to Tyr-1139 of
erbB2, we attempted to identify SH2 domain residues determining the
high affinity interaction of Grb7, but not Grb14, with this site. Fig. 3 shows an alignment of the SH2 domains from Grb7,
Grb14, and Src in the context of the secondary structure of the Src SH2
domain as deduced by Waksman et al. (10). Secondary
structure elements are labeled using the notation of Eck et
al. (30). All three SH2 domains fall into Group I since they
contain Phe or Tyr at D5. The six amino acids that were replaced in
the Grb14 SH2 fusion protein by the corresponding amino acid in Grb7
are indicated. It should be noted that the EF loop in Grb7 and Grb14 is
extended compared with that in Src, and the EF1 position refers to that in the latter SH2 domain. The choice of the targeted residues was based
on 1) being nonidentical between Grb14 and Grb7, and 2) their likely
interaction with the +1 to +3 positions of target sequences, as
suggested by crystal structures of other Group I SH2 domains complexed
with a high affinity peptide (10, 30). Thus,
D5 interacts with +1,
D
1 interacts with +2, and
D5,
E4, and EF1 interact with +3.
D6 forms part of the phosphotyrosine binding site and also
contributes to a hydrogen-bonded network involving water molecules that
interact with the +1 and +2 positions.
B9 forms part of the +3
binding pocket.
Amino acid changes were incorporated into the GST-Grb14 SH2 fusion
protein by site-directed mutagenesis, and the resulting fusion proteins
were coupled to Sepharose beads and tested for their ability to bind
the intracellular domain of erbB2 in control or EGF-treated HER1-2 cell
lysates. Association of receptor with the beads was detected by Western
blotting. As described previously (8), binding of the wild type Grb14
SH2 domain was not detected. However, two mutations increased Grb14 SH2
domain affinity for the EGF receptor/erbB2 chimera (Fig.
4A). A Gln to Leu change at D6 caused a
marked increase in binding, whereas a Phe to Tyr change at
D5
resulted in a more modest increase. To evaluate the combined effect of
the
D5 and
D6 substitutions, both mutations were introduced into
the Grb14 SH2 fusion protein. This resulted in a further increase in
affinity above that achieved by the single mutations (Fig.
4B).
Introduction of Specific Grb14 Residues into Grb7 Abrogates Grb7 SH2 Domain Binding to erbB2
To further investigate the
significance of the D5 and
D6 residues in controlling Grb7 SH2
domain specificity toward erbB2, the reverse experiment was performed
by introducing corresponding Grb14 residues, either separately or in
combination, into the Grb7 SH2 fusion protein. The resulting mutant
Grb7 SH2 domains were then analyzed for their ability to bind the erbB2
intracellular domain in HER1-2 cell extracts. The Leu to Gln change at
D6 consistently abrogated binding, whether introduced on its own or
simultaneously with the
D5 change (Fig. 5). However,
binding of the mutant Grb7 SH2 domain incorporating the
D5 change
alone was not reduced compared with the wild type Grb7 SH2. Similarly,
switching the
B9 residues did not affect binding of the Grb7 SH2
domain (Fig. 5). These results highlight the importance of
D6 in
Grb7 binding to erbB2.
Grb14 SH2 Mutants Incorporating
To confirm that the
interaction of the D6 and
D5/
D6 mutant Grb14 SH2 domains with
erbB2 was mediated via the same binding site as Grb7, phosphopeptide
competition analysis was performed as described previously. Only the
phosphopeptide corresponding to Tyr-1139 of erbB2 significantly
inhibited binding to the HER1-2 chimera, reducing the interaction of
the
D6 mutant to 11.5% and that of the
D5/
D6 mutant to 9% of
control values (Fig. 6). These mutants therefore display
Grb7-like SH2 domain specificity in that they bind Tyr-1139 of
erbB2.
Determination of the Relative Affinities of Wild Type Grb7 and Grb14 and Mutant Grb14 SH2 Domains for Tyr-1139 of erbB2
The
interaction of these SH2 domains with the erbB2 Tyr-1139 site was
further investigated using real time biosensor (BIAcore) analysis of
the GST-SH2 fusion proteins binding to sensor chip-immobilized Tyr-1139
phosphopeptide. Since GST dimerization can lead to an overestimation of
binding affinities by this method (31), these experiments enabled
investigation of binding in the absence of interference from other
autophosphorylation sites but only estimated the relative, not
absolute, affinities of the SH2 domains for the phosphopeptide.
Parallel channels of a sensor chip coupled to either the Tyr-1139
phosphopeptide or the nonphosphorylated version were exposed to
increasing concentrations of soluble fusion proteins. Table
I shows apparent affinity constants estimated from
equilibrium responses by Scatchard analysis of BIAcore progress data
(see "Materials and Methods"). No binding of the SH2 domain fusion
proteins to the nonphosphorylated peptide was detected, and the
interaction of the GST-Grb14 SH2 fusion protein with phosphorylated Tyr-1139 was too weak for kinetic analysis. Equilibrium responses of
the D5 mutant binding to the phosphopeptide demonstrated an apparent
Kd of 438 nM, whereas a higher affinity
was derived for the
D6 mutant (Kd = 224 nM). The apparent Kd for interaction of
the
D5/
D6 mutant was slightly lower again (203 nM),
approaching that of wild type Grb7 (126 nM). These results
therefore parallel those obtained upon binding of the fusion proteins
to the intracellular region of erbB2 (Fig. 4) and provide additional
evidence for the interactions being mediated via the Tyr-1139
autophosphorylation site.
|
From all the
available structures of Group I SH2 domains in complex with a high
affinity phosphopeptide, the SH2 domains of Grb7 and Grb14 share the
highest overall homology with that of Src, with 33 and 31% amino acid
identity, respectively. For this reason the structure of the Src SH2
domain, determined by x-ray crystallography (10), was used to model the
structure of the Grb7 and Grb14 SH2 domains. The Src sequence was
substituted by the sequences of Grb7 and Grb14, and the sequence of the
complexed phosphopeptide was replaced by the Tyr-1139 peptide sequence. It was assumed that the backbone conformations of the peptide and of
the SH2 domain secondary structure elements were the same as in the Src
structure, and identical side chains were copied and compatible
rearrangements of nonidentical side chains were considered (see
"Materials and Methods"). In this way the relative influences of
the D6 residues from Grb7 and Grb14 on SH2 interaction with the
Tyr-1139 peptide could be modeled.
The overall structure deduced for the Grb7 SH2 domain is shown in Fig.
7A with the position of the bound Tyr-1139
peptide. Comparison of the Grb7 and Grb14 SH2 domains revealed
differences in the predicted orientation of the D6 side chain with
respect to the peptide, whereas conformations of adjacent side chains were identical. In this model the Leu of Grb7 is directed away from the
Asn at +2 of the peptide, superimposable with the Lys at
D6 of Src,
whereas the Gln in Grb14 is predicted to extend toward the Asn side
chain (Fig. 7B). Without further refinement of the model,
the distance between the side chain amide heavy atoms is 3.95 Å. This
is suggestive of the opportunity for hydrogen bonding between these
side chains, so this distance may in fact be smaller. The conformation
of the Asn BC3 side chain is also predicted to change in Grb14 to a
position sufficiently close (3.24 Å between heavy atoms) for hydrogen
bonding with Gln
D6 (Fig. 7B). Hydrogen bonding of Gln
D6 with Asn at +2 of the peptide may add rigidity to the structure
of the complex that could be detrimental to efficient binding. In
addition, hydrogen bonding and/or the orientation of the
D6 Gln may
disrupt other interactions required for high affinity phosphopeptide
binding.
The receptor tyrosine kinase erbB2 and the signaling protein Grb7 are overexpressed in a subset of human breast cancer cell lines and primary breast cancers due to co-amplification of their respective genes. They are also functionally linked as they co-immunoprecipitate from cell lysates and interact directly in vitro (3). This study investigated the interaction between these two proteins by first determining the binding site of Grb7 and then identifying Grb7 SH2 domain residues conferring specificity toward this site.
Five putative in vivo autophosphorylation sites and an additional site of in vitro autophosphorylation have been identified in the erbB2 C terminus (28, 29). Since point mutation of the five tyrosines phosphorylated in vivo reduced receptor tyrosine phosphorylation by 95% (28), the existence of major autophosphorylation sites other than the six pictured in Fig. 1A is unlikely. Of these six sites, peptide competition revealed Tyr-1139 as the preferred binding site of the Grb7 SH2 domain (Fig. 1B). This was supported by binding studies involving the HER1-2 Y1139F receptor, in which mutation of this site reduced association of the Grb7 SH2 domain by 95% compared with that observed with the wild type receptor (Fig. 2B).
Interestingly, Tyr-1139 also represents the major Grb2 binding site on
erbB2 (18), and there is evidence for interaction of Grb7 with Grb2
binding sites on other proteins. Both bind Tyr-580 of the tyrosine
phosphatase SH-PTP2 (32), and Grb7 probably also recognizes the Grb2
binding site at Tyr-317 of Shc (3). The potential interaction with Grb2
binding sites on other signaling molecules such as FAK (33), IRS-1
(34), and RPTP (35) awaits further investigation. The binding of
different SH2 proteins to a single phosphotyrosine site is not
unprecedented. Nck and the p85 subunit of phosphatidylinositol
3
-kinase both bind and compete for Tyr-751 in the platelet-derived
growth factor
receptor (36), and the Met receptor has a multiple
SH2 protein docking site at Tyr-1349 (37). Since the downstream
effectors of the Grb7 family have yet to be identified, the
consequences of these interactions are not clear. However, since Grb7
exhibits a relatively tissue-specific expression profile (38) and the
GRB7 gene resides on the same breast cancer amplicon as
ERBB2 (3), it is clear that competition between Grb2 and
Grb7 provides a potential mechanism for modulation of the Ras signaling
pathway in specific tissues and/or cancer cells. In the latter context,
amplification and overexpression of the ERBB2 gene in a
series of breast cancer cell lines that exhibit concomitant Grb7
overexpression correlates with increased erbB2-Grb2 interaction and
mitogen-activated protein kinase stimulation, indicating that the Ras
pathway is not markedly down-regulated by Grb7 competition for Grb2
binding sites (3, 21). However, since either proliferation or
differentiation can be specified by the kinetics of mitogen activated
protein kinase activation (39), the consequences of erbB2 signaling via
the Ras pathway may be different in the presence or absence of Grb7.
This is currently under investigation.
The Grb7 and Grb14 SH2 domains are classified into Group I due to the
presence of Phe or Tyr at the D5 position. Most members of this
group exhibit a preference for the target amino acid sequence pTyr-hydrophilic-hydrophilic-Ile/Pro, where the selectivity at the +2
position is lower or similar to that at +1 and +3 (12, 13). However,
Grb2 has an atypical Group I SH2 domain that selects pY-Q/Y/V-N-Y/Q/F,
with the strongest preference at the +2 position (13). As expected,
this consensus exhibits homology to the Grb7 binding sites on Shc
(YVNV) (3), SHPTP2 (YENV) (32), and erbB2 (YVNQ), with the conservation
of Asn at +2 particularly apparent. The poor selectivity of the Grb2
SH2 domain toward the +3 position is thought to be partly due to the
bulky Trp residue at the EF1 position closing up the +3 binding pocket
(12, 40). Two features of the Grb2 and Grb7 SH2 domains may contribute
to their similar phosphopeptide selectivity. First, the insertion in
the Grb7 SH2 EF loop region relative to Src (Fig. 3) may restrict the
+3 binding pocket in a manner similar to Trp EF1 in the Grb2 SH2.
Second, selectivity of the Src SH2 domain at the +2 position is
influenced by the
D
1 residue, which in Src is Arg but in Grb2 and
Grb7 is Leu, which may favor selectivity for Asn at +2. The low
affinity of the Grb14 SH2 domain, which has Ile at
D
1, for such
phosphopeptide sequences may be due to structural changes induced by
the nonconservative change at the
D6 position relative to Grb7, as
discussed below.
Residues within the Grb7 SH2 domain that determine the specificity of
its interaction with erbB2 were identified by substitution analysis
using the closely related Grb14 SH2 domain, which exhibits a low
affinity for erbB2. A Phe to Tyr change at D5 in the Grb14 SH2
domain caused a modest increase in binding to the intracellular domain
of erbB2 (Fig. 4A), whereas the reverse change in the Grb7 SH2 domain had no discernible effect on binding (Fig. 5). At the adjacent
D6 position a Gln to Leu change caused a marked increase in
Grb14 SH2 binding to erbB2, and the reverse change in Grb7 abrogated
its ability to bind the receptor. The erbB2 affinity of the
D6
mutant was further augmented by simultaneously substituting the
D5
residue (Fig. 4B). Substitution at the other four candidate positions, which included
D
1, had minor or undetectable effects on
erbB2 binding. Importantly, phosphopeptide competition analysis demonstrated that the increases in affinity toward erbB2 exhibited by
the Grb14
D6 and
D5/
D6 mutants were directed toward the same
autophosphorylation site as that preferentially bound by Grb7,
Tyr-1139. Moreover, the apparent equilibrium affinity constants of
these mutants for interaction with the Tyr-1139 phosphopeptide were
determined by BIAcore analysis and paralleled their binding to erbB2.
Minimal binding was observed with the Grb14 SH2, whereas the apparent
equilibrium constants for the
D6 and
D5/
D6 mutants approached
that of wild type Grb7. Taken together, our studies of SH2 domain
interactions with both the erbB2 intracellular domain and the Tyr-1139
peptide pinpoint the
D6 Leu as an important determinant of Grb7 SH2
specificity.
We have modeled the interaction of the Grb14 and Grb7 SH2 domains with
the Tyr-1139 phosphopeptide, using the structure determined for the Src
SH2 domain complexed with a high affinity peptide, in an attempt to
understand the impact of the different D6 residues. The conformation
of the peptide was assumed to be the same as in the Src structure,
lying perpendicular to the central
sheet, between the BG and EF
loops (Fig. 7A). Indeed, this peptide position is very
similar in the SH2 domain structures of Lck, PLC-
1, and Syp,
although in Shc the peptide is positioned lower, between the BG and DE
loops (16). Recalculation of side chain configurations using a rotamer
library revealed a difference in the predicted position of the
D6
side chain of Grb14 compared with Grb7, whereas adjacent residues,
including
D5, were superimposable (Fig. 7B). The side
chain of Asn BC3 in the phosphate-binding BC loop also changed
conformation in the Grb14 model. Therefore, since Gln at
D6
abrogates Grb7 binding to erbB2, the orientation of this
D6 residue
might introduce rigidity and/or interference and hence be detrimental
to binding. However, in the recently described crystal structure of the
Grb2 SH2 domain complexed with a high affinity peptide (41), the latter
adopts a novel conformation, incorporating a
turn, and the +2 Asn
hydrogen bonds to the backbone carbonyl and amide groups of the SH2
domain
D6 residue. This is intriguing in light of the overlap in SH2
specificity between Grb2 and Grb7 and our data highlighting the
D6
residue as an important selectivity determinant of the Grb7 SH2 domain,
and will be investigated further once the Grb2 structural coordinates become available.
In summary, the nature of the D6 residue is a critical determinant
of the high affinity interaction of the Grb7 SH2 domain with Tyr-1139
of the erbB2 intracellular domain. Switching this single amino acid
between the Grb14 and Grb7 SH2 domains exchanges their respective
affinities for erbB2, analogous to the effect of swapping the EF1
residue between the Src and Grb2 SH2 domains (40). These observations
therefore provide a clear example of how heterogeneity at a single
residue allows two highly conserved SH2 domains to retain distinct
preferences for RTK binding.
We thank Dr. J. Schlessinger for the HER1-2 cell line and plasmid encoding the Grb2 fusion protein and Dr. Axel Ullrich for the HER1-2 expression vector.