(Received for publication, February 6, 1997, and in revised form, April 17, 1997)
From the Division of Basic Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington, 98109-1024
Fms is a tyrosine kinase-containing receptor for
macrophage colony-stimulating factor (M-CSF) that regulates survival,
growth, and differentiation of cells along the monocyte/macrophage
lineage. M-CSF stimulation of murine myeloid FDC-P1 cells expressing
Fms resulted in the tyrosine phosphorylation of a number of signal transduction proteins, including an unidentified 100-kDa protein. This
100-kDa protein associated with the tyrosine phosphatase SHP-2 but not
with the related phosphatase SHP-1. The kinetics of tyrosine
phosphorylation of p100 and SHP-2 suggest that p100 may be a direct
substrate of SHP-2. p100 bound directly to the SH2 domains of both
SHP-2 and the p85 subunit of phosphatidylinositol 3-kinase. The
100-kDa protein did not appear to bind directly to Fms, Ship, Cbl, Shc,
or Grb2, although all of these proteins were coimmunoprecipitated with
p85 after M-CSF stimulation. Association of p100 with SHP-2 and p85 did
not require the major autophosphorylation sites on Fms nor binding of
p85 to Fms. A tyrosine phosphorylated protein of 100 kDa also
coprecipitated with SHP-2 from several other myeloid cell lines after
M-CSF stimulation but was not seen in immunoprecipitates from Rat2
fibroblasts expressing Fms. Stimulation of FDC-P1 cells with additional
cytokines also resulted in coprecipitation of a 100-kDa protein with
SHP-2. p100 may therefore be a common component of the signaling
pathways of cytokine receptors in myeloid cells.
Recent studies on the protein tyrosine phosphatases SHP-1 and SHP-2 have established the importance of these phosphatases to intracellular signaling in hematopoietic cells. These two cytoplasmic tyrosine phosphatases are very similar in structure. Each has two tandem phosphotyrosine-binding SH21 domains at the N terminus and a single C-terminal tyrosine phosphatase domain (1, 2). However, their patterns of expression and roles in cellular signaling are distinct.
SHP-1 (previously called SH-PTP1, PTP1C, HCP, and SHP) is expressed at highest levels in hematopoietic cells (3). Mutation of the gene for SHP-1 in mice results in the motheaten phenotype (4). motheaten mice display multiple hematopoietic abnormalities including hyperproliferation of myeloid, erythroid, and lymphoid precursor cells (5). Consistent with this phenotype, SHP-1 negatively regulates signaling from a variety of cytokine receptors, including the receptors for interleukin-3 (6), erythropoietin (EPO) (7), stem cell factor (8, 9), and M-CSF (10). The negative effect on signaling results from dephosphorylation of the cytokine receptor itself (6) or dephosphorylation of one or more tyrosine phosphorylated proteins associated with the receptor (7). For instance, SHP-1 binds to the EPO receptor after EPO stimulation and dephosphorylates the receptor-associated tyrosine kinase JAK2. Cells expressing mutant EPO receptors that cannot bind SHP-1 are hypersensitive to EPO and display prolonged autophosphorylation of JAK2 (7).
In contrast to SHP-1, SHP-2 (previously called SH-PTP2, PTP2C, PTP1D,
and Syp) is widely expressed (11, 12) and is generally a positive
regulator of signals for proliferation. SHP-2 is the mammalian homolog
of the Corkscrew phosphatase in Drosophila (13), which is a
required component of signaling from the Torso and Sevenless receptor
tyrosine kinases (13, 14). In mammalian cells, SHP-2 is involved in the
signaling pathways of many nonhematopoietic receptors, including the
receptors for platelet-derived growth factor (11, 15), epidermal growth
factor (11, 15), insulin (16), and insulin-like growth factor (17).
SHP-2 is also involved in signaling from hematopoietic receptors such
as c-Kit (18), the EPO-R (19), the receptors for IL-3 and granulocyte
macrophage colony stimulating factor (20), thrombopoietin (21), the
gp130 signaling subunit of the receptor for IL-11 (22), and the
interferon /
receptor (23).
The mechanism by which SHP-2 positively regulates signaling is not fully understood. The SH2 domains of SHP-2 may allow it to act in part as an adapter linking two or more signaling proteins. For instance, after binding to the activated platelet-derived growth factor receptor or EPO-R, SHP-2 is itself tyrosine phosphorylated and binds to the SH2 domains of the adapter protein Grb2. Grb2 is constitutively bound to the Ras guanine nucleotide-releasing factor Sos, and by linking Grb2 and the receptor, SHP-2 effectively recruits Sos to the membrane where it can activate Ras (19, 24, 25). However, the phosphatase activity of SHP-2 is also a necessary component of these pathways, because overexpression of a catalytically inactive SHP-2 has been shown to block mitogen-activated protein kinase activation in response to insulin (26, 27), platelet-derived growth factor (28), epidermal growth factor (29), and fibroblast growth factor (30). When the SH2 domains of SHP-2 bind to a tyrosine phosphorylated protein, the phosphatase activity of SHP-2 is greatly increased (31, 32). The catalytic activity of SHP-2 may also be activated by tyrosine phosphorylation of the phosphatase (15, 20) and by association with phospholipids when the phosphatase is recruited to the plasma membrane (33). SHP-2 may dephosphorylate the platelet-derived growth factor receptor to which it is bound (34) and has been shown to dephosphorylate insulin receptor substrate-1 in vitro (35) and the adapter protein Shc in T cells (36). However, most of the in vivo substrates of SHP-2 are still unknown.
The M-CSF receptor (Fms) is a glycoprotein with intrinsic tyrosine
kinase activity (37, 38) and is required for the survival, growth, and
differentiation of cells of the monocyte/macrophage lineage (39).
Murine myeloid FDC-P1 cells expressing an exogenous murine
c-fms gene (FDFms cells) have been used to investigate the
M-CSF-stimulated signals leading to growth and differentiation of these
cells (40-42). M-CSF stimulation of FDC-P1 cells expressing Fms
results in the recruitment of a number of signaling proteins to the
receptor, including Ship (42), Cbl (43, 44), PI 3-kinase (45, 46),
Shc, Grb2, and Sos (40, 46). Only a few proteins have been shown to
bind directly to the receptor. Tyrosine 559 in the membrane proximal
region of the receptor is a potential binding site for Src family
tyrosine kinases (47). The three autophosphorylation sites within the
kinase insert domain of Fms (tyrosines 697, 706, and 721) (45, 48) are
important for protein interactions. The SH2 domains of Grb2 have been
shown to bind to tyrosine 697. Grb2 binding to Fms probably results in
translocation of Sos to the membrane and subsequent Ras activation (40,
46). Tyrosine 706 is required for efficient activation of STAT1 in FDC-P1 cells (49). The p85 regulatory subunit of PI 3
-kinase binds to
tyrosine 721 of Fms (45, 46), thereby activating PI 3
-kinase-mediated
signaling events. Tyrosine 807 within the kinase catalytic domain of
Fms does not bind any known signaling proteins, but is important for
c-myc induction in NIH 3T3 cells (50) and for signals
leading to differentiation of FDC-P1 cells (41).
In this report, we have investigated the roles of SHP-1 and SHP-2 in
Fms signaling in FDC-P1 cells. We find that SHP-2 binds to a novel
100-kDa tyrosine phosphorylated protein after stimulation by M-CSF and
several other cytokines. This 100-kDa protein associates directly with
the SH2 domains of both SHP-2 and the 85-kDa subunit of PI 3-kinase
and is associated with Fms, Ship, Cbl, Shc, and Grb2 in p85 immune
complexes.
Recombinant murine M-CSF and WEHI-3B cell
conditioned medium (a source of IL-3) were described previously (51,
52). Baby hamster kidney cells expressing an exogenous gene for
recombinant murine stem cell factor (BHK/MKL cells) were a gift from
Schickwann Tsai (Fred Hutchinson Cancer Research Center, Seattle, WA).
Rabbit polyclonal antiserum directed against the cytoplasmic domain of murine Fms (53), rabbit antiserum to murine Ship (42), and a rabbit
polyclonal antiserum to the SH2 domain of Shc (42) were prepared in our
laboratory. Monoclonal antibody to phosphotyrosine (clone 4G10) was a
kind gift from Brian Druker (Dana Farber Cancer Institute). Antibody to
the 85-kDa subunit of PI 3-kinase and an antibody specific for SHP-1
were purchased from Upstate Biotechnology Inc. (Lake Placid, NY).
Antibody to Cbl and an antibody specific for SHP-2 were from Santa Cruz
Biotechnology (Santa Cruz, CA). A rabbit polyclonal antibody (206)
recognizing both SHP-1 and SHP-2 was a gift from Ben Neel (Harvard
Medical School). A monoclonal antibody to Grb2 was from Transduction
Laboratories (Lexington, KY). Chicken antiserum recognizing GST was
prepared in our laboratory. Horseradish peroxidase-conjugated secondary
antibodies were purchased from Boehringer Mannheim Corporation or from
Sigma. Reagents for chemiluminescence detection were from Amersham
Corp. or from DuPont NEN. Protein molecular mass markers were obtained
from Life Technologies, Inc., Novagen (Madison, WI), or Novex (San
Diego, CA).
FDC-P1, 32D, EML (54), and Rat2 cells expressing an exogenous murine Fms gene were generated as described previously (52). FDC-P1 cells expressing Fms genes with tyrosine to phenylalanine mutations of the autophosphorylation sites have been described (55). FDC-P1, 32D, and EML cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and 2% WEHI-3B cell conditioned medium as a source of IL-3. BAC1.2F5 cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and 1000 units/ml of murine M-CSF. Rat2 cells were maintained in Dulbecco's modified Eagle's medium with 10% calf serum.
GST Fusion ProteinsDifferent domains of SHP-2 and p85 were
polymerase chain reaction-amplified using oligonucleotides that
included either a BamHI (5 end) or an EcoRI (3
end) restriction site. The templates for polymerase chain reaction
amplification were pGEX-2T-SHP-2 (kindly provided by Zhizhuang Zhao,
Vanderbilt University, Nashville, TN) and a p85 cDNA provided by
Michael Fry (Institute of Cancer Research, Surrey, UK). For SHP-2, the
amplified domains were the N-terminal SH2 domain (amino acids 3-109),
the C-terminal SH2 domain (amino acids 97-223), N and C-terminal SH2
domains (amino acids 3-223), full-length SHP-2 (amino acids 3-593),
and C terminus (amino acids 543-593). For p85, the domains amplified
were the N terminus (amino acids 3-332), N-terminal SH2 domain (amino
acids 300-435), C-terminal SH2 domain (amino acids 589-724), and N
plus C-terminal SH2 domains (amino acids 300-724). The amplified DNA fragments were digested with BamHI and EcoRI, gel
purified, and ligated into the BamHI, EcoRI-cut
plasmid pGEX-3X (Pharmacia Biotech Inc.). Ligated DNAs were transformed
into Escherichia coli strain BL21, and colonies were
screened for the correct insert by restriction digest. GST fusion
proteins were prepared from BL21 lysates by adsorption to
glutathione-agarose as described in the Pharmacia GST Gene Fusion
System manual.
Cells for immunoprecipitation were collected by centrifugation and washed once in phosphate-buffered saline. Cells were stimulated with M-CSF at 10,000 units/ml as described in the figure legends and then immediately lysed in ice-cold Nonidet P-40 lysis buffer (52) containing 10 µM phenylarsine oxide (a phosphatase inhibitor). Antibody and protein A-Sepharose (for rabbit antibodies) or protein G-Sepharose (for mouse or rat antibodies) were added to lysates, and the samples were mixed at 4 °C for 1-4 h. The immune complexes were washed five times in Nonidet P-40 lysis buffer, and then proteins were eluted with either 100 mM phenyl phosphate in 100 mM HEPES, pH 6.8, or SDS sample buffer. Samples were electrophoresed on 7.5% or 10% reducing SDS-PAGE gels and transferred to nitrocellulose membranes on a semi-dry blotting apparatus (Ellard Instrumentation, Ltd., Seattle, WA). The nitrocellulose membranes were incubated in Block buffer (52) for 1-18 h and then incubated 1-4 h with the appropriate primary antibody diluted in Block. The membranes were washed for 1 h with 4-6 changes of Rinse buffer (52) and then incubated 1 h with horseradish peroxidase-conjugated secondary antibody diluted 1:10,000 in Block. The membranes were washed again as described above, and proteins were visualized using enhanced chemiluminescence according to the manufacturer's instructions.
Far Western Blotsp85 was immunoprecipitated from M-CSF-stimulated FDFms cells using a rabbit polyclonal antibody from Upstate Biotechnology Inc. (06-195). Five duplicate immunoprecipitates were run on a 7.5% SDS-PAGE gel and transferred to nitrocellulose. After incubating the membrane with Block overnight at 4 °C, the membrane was cut into five strips. One strip was immunoblotted with antibody to phosphotyrosine as described above. The other strips were incubated separately for 3 h with GST, GST-SHP1-N+C-SH2s, GST-SHP2-N+C-SH2s, or GST-P85-N+C-SH2s (each at 1 µg/ml in Block). The blots were washed 4 × 10 min in Rinse and then incubated for 1 h with a chicken antiserum recognizing GST (1:10,000 in Block). The blots were washed again 4 × 10 min in Rinse and incubated for 1 h with horseradish peroxidase-conjugated rabbit anti-chicken antibody at 1:10,000 in Block. The blots were washed again as above, and signals were detected using enhanced chemiluminescence.
To look for a potential role for tyrosine phosphatases SHP-1 and
SHP-2 in Fms signaling in myeloid progenitor cells such as FDC-P1, we
initially tried to coimmunoprecipitate these phosphatases with antibody
to Fms. Using immunoblots with antibodies specific for SHP-1 or SHP-2,
neither phosphatase was found to be associated with Fms after M-CSF
stimulation (data not shown). Antibodies to SHP-1 or SHP-2 also failed
to coimmunoprecipitate Fms. These results are consistent with those of
Yi and Ihle (56) and Chen et al. (10), who reported that
SHP-1 does not associate directly with Fms. Although Fms was not
present in immunoprecipitates of these phosphatases, a tyrosine
phosphorylated protein of about 100 kDa was coimmunoprecipitated with
SHP-2 from FDFms cells after M-CSF stimulation (Fig. 1).
The 100-kDa protein was by far the most prominent phosphoprotein
associated with SHP-2. In contrast, no tyrosine phosphorylated proteins
were detected in SHP-1 immunoprecipitates either before or after M-CSF
stimulation.
Time course experiments were performed to look at the kinetics of p100
phosphorylation and association with SHP-2. FDFms cells were treated
with M-CSF for 15 s to 4.5 min, and cell lysates were incubated
with an antibody that recognized both SHP-1 and SHP-2 tyrosine
phosphatases. A phosphotyrosine blot of the time course experiment
showed that p100 was phosphorylated within 15 s and was maximally
phosphorylated after 1-1.5 min of M-CSF treatment at 37 °C (Fig.
2). After about 1.5 min of M-CSF stimulation, p100 appeared to be rapidly dephosphorylated or at least dissociated from
SHP-2 (Fig. 2A). A phosphotyrosine blot of whole cell
lysates showed that p100 was among the first proteins tyrosine
phosphorylated after M-CSF stimulation and again decreased in intensity
after about 1.5 min (Fig. 2B). This suggests that p100 was
dephosphorylated at this time and not merely dissociated from SHP-2.
Interestingly, SHP-2 itself was not tyrosine phosphorylated immediately
upon M-CSF stimulation, but was tyrosine phosphorylated after about 1.5 min. Thus, tyrosine phosphorylation of SHP-2 was not required for its
association with p100. Tyrosine phosphorylation of SHP-2 just preceded
dephosphorylation of p100. Because tyrosine phosphorylation of SHP-2
has been shown to activate its phosphatase domain (15, 20), this
suggests that SHP-2 may directly dephosphorylate p100. In contrast to
SHP-2, SHP-1 was constitutively tyrosine phosphorylated in these cells
(Figs. 1 and 2A), and M-CSF treatment did not increase the
level of phosphotyrosine on SHP-1 (Fig. 2).
p100 always appeared as a broad band on SDS-PAGE gels, suggesting that it might be a glycoprotein. However, p100 failed to bind to lectin columns (data not shown), so the broad appearance of the p100 band on SDS-PAGE gels is likely caused by some other modification. The p100 protein also gradually increased in apparent molecular mass with time after M-CSF stimulation. After 15 s of M-CSF stimulation at 37 °C, the protein had an apparent molecular mass of about 90 kDa. After 10 min, the apparent molecular mass had increased to about 100 kDa (Fig. 2A). This increase in molecular mass was not likely due to an increase in tyrosine phosphorylation, because the level of phosphotyrosine seemed to be decreasing at the same time that the molecular mass of p100 was increasing (Fig. 2B). The increase in mass could therefore be due to serine or threonine phosphorylation, ubiquitination, or another modification.
To determine whether p100 was associated with other proteins known to
be involved in Fms signaling, antibodies to Fms, Ship, Cbl, p85, SHP-2,
Shc, or Grb2 were used to immunoprecipitate these proteins from
M-CSF-stimulated or unstimulated FDFms cells, and proteins associated
via phosphotyrosine interactions were eluted with phenyl phosphate. As
shown in Fig. 3A, a 100-kDa tyrosine phosphorylated protein was coprecipitated with both SHP-2 and with p85.
p85 was also associated with tyrosine phosphorylated Fms, Ship, Cbl,
and Shc, as well as unphosphorylated Grb2 (Fig. 3A and data
not shown); however, antibodies directed against these proteins did not
coprecipitate significant amounts of p100. No additional
coprecipitating tyrosine phosphorylated proteins were seen when the
immune complexes were eluted with SDS sample buffer (data not shown).
This suggested that p100 might bind directly to both SHP-2 and p85 but
did not associate directly with the other proteins tested.
Tyrosine phosphorylation of p85 was not detectable on this blot (Fig. 3A). This was expected, because the level of phosphotyrosine on p85 is usually very low after M-CSF stimulation of FDC-P1 cells.2 The level of phosphotyrosine on SHP-2 was highest in the M-CSF-stimulated p85 immunoprecipitates (Fig. 3A) and was also visible in M-CSF-stimulated SHP-2 immunoprecipitates eluted with SDS sample buffer (data not shown). The phosphotyrosine blot in Fig. 3A was then reprobed with antibodies to p85 and to SHP-2. Although p85 had not been detected in the phosphotyrosine blot, a significant amount of p85 was present in the SHP-2 immunoprecipitates after M-CSF stimulation, as well as in the p85 immunoprecipitates (Fig. 3B). (It was evident that a fraction of the proteins bound directly to antibodies were also eluted with phenyl phosphate.) Similarly, a significant amount of SHP-2 was pulled down in the p85 immunoprecipitates after M-CSF stimulation (Fig. 3C). The SHP-2 present in the p85 immunoprecipitates migrated somewhat more slowly than the SHP-2 that was immunoprecipitated directly. Because tyrosine phosphorylation of the SHP-2 protein was detected in the p85 immunoprecipitates but not in the SHP-2 immunoprecipitates, this suggests that the subset of SHP-2 associated with p85 migrated more slowly because it was tyrosine phosphorylated.
To determine which domains of SHP-2 and p85 were required for
association with p100, GST fusion proteins containing different domains
of SHP-2 or p85 were prepared. Fusion proteins bound to glutathione-agarose beads were incubated with lysates of
M-CSF-stimulated FDFms cells, washed extensively with Nonidet P-40
lysis buffer and eluted with phenyl phosphate. p100 coprecipitated with
a GST fusion protein containing both N-terminal and C-terminal SH2
domains of SHP-2 and was efficiently eluted with phenyl phosphate (Fig. 4A). However, GST fusion proteins containing
only the N-terminal or C-terminal SH2 domain of SHP-2 failed to bring
down detectable amounts of p100. This suggests that the association
between p100 and SHP-2 may be stabilized by the binding of two
phosphotyrosines on p100 to the two SH2 domains of SHP-2. p100 was also
coprecipitated with the SH2 domains of p85 and was easily eluted with
phenyl phosphate. Either the N-terminal or the C-terminal SH2 domain of
p85 was sufficient for association with p100 (Fig. 4B).
The results using GST fusion proteins suggested that tyrosine
phosphorylated p100 may bind directly to the SH2 domains of SHP-2 and
p85. However, GST fusion proteins might pull down proteins that bind
indirectly along with proteins that bind directly. We therefore used
far Western blots to demonstrate direct binding of p100 to the SH2
domains of SHP-2 and p85. p85 immunoprecipitates were transferred to
nitrocellulose, and the membranes were incubated with GST or GST-SH2
domain fusion proteins. As shown in Fig. 5, both SHP-2
and p85 SH2 domains bound directly to the p100 band that
coimmunoprecipitated with p85.
Tyrosine phosphorylated p100 could also be coimmunoprecipitated with
SHP-2 from cells expressing Fms mutants lacking four major Fms
autophosphorylation sites (Tyr697, Tyr706,
Tyr721, or Tyr807). Tyrosine phosphorylation of
Fms at these sites was not required for phosphorylation of p100 nor for
association of p100 with SHP-2 (Fig. 6A).
p100 also associated with p85 in cells expressing Fms Y721F (Fig.
6B). This mutant lacks the autophosphorylation site required
for association of p85 with Fms (45, 46). Therefore, in cells
expressing Fms Y721F, p85 did not coprecipitate Fms, and p85 was not
tyrosine phosphorylated. This indicates that tyrosine phosphorylation
of p100 and association of p100 with p85 does not require the tyrosine
phosphorylation of p85 nor binding of p85 to Fms.
In addition to FDC-P1 cells, several other myeloid cells lines were
examined to see whether p100 was a common component of M-CSF signaling.
The myeloid progenitor cell lines FDC-P1, 32D, and EML (54) (each
expressing an exogenous murine c-fms gene) and BAC1.2F5 (a
more differentiated macrophage-like cell line expressing the endogenous
c-fms gene) were stimulated with M-CSF and
immunoprecipitated with antibody to SHP-2. Coprecipitating proteins
were detected by blotting with antibody to phosphotyrosine. In all of
the myeloid cell lines tested, a tyrosine phosphorylated p100 protein
was coimmunoprecipitated with SHP-2 (Fig. 7). However, no p100 was detected after coprecipitation with SHP-2 in Rat2 fibroblasts expressing murine c-fms. Therefore, p100 appears
to be a common component of M-CSF signaling in myeloid cells but not in
all cell types.
Different cytokine and growth factor receptors use many of the same
proteins in their signaling pathways. To determine whether p100 might
be involved in signaling through other receptors, FDFms cells were
stimulated with stem cell factor, IL-3, GM-CSF, or M-CSF and then
immunoprecipitated with antibody to SHP-2. A phosphotyrosine blot of
the immunoprecipitated proteins showed that a tyrosine phosphorylated
protein of 100 kDa was coprecipitated with SHP-2 after stimulation with
each of these cytokines (Fig. 8). (Much more tyrosine
phosphorylated p100 was detected after stimulation with M-CSF, probably
because these cells express much higher levels of Fms than the other
receptors.) These results suggest that p100 is involved in the
signaling pathways of multiple cytokines.
We have characterized a novel 100-kDa tyrosine phosphorylated protein that bound to the SH2 domains of both SHP-2 and p85 in response to M-CSF and several other cytokines. The 100-kDa protein that bound to SHP-2 is very likely the same protein pulled down with p85, because a GST-SHP-2 fusion protein bound in far Western blots to the p100 protein that coimmunoprecipitated with p85. A tyrosine phosphorylated p100 protein was also coimmunoprecipitated with SHP-2 from a variety of Fms-expressing myeloid cells and in response to stem cell factor, IL-3, and GM-CSF as well as M-CSF. However, because we do not yet have an antibody to p100, we cannot be certain that the p100 protein seen in other myeloid cells and in response to other cytokines is identical to the p100 that bound to SHP-2 and p85 in M-CSF-stimulated FDFms cells. Nevertheless, our results suggest that p100 could be a common component of signaling from multiple receptors in hematopoietic cells. Although a tyrosine phosphorylated p100 was seen in all the myeloid cells examined, the p100 protein was not detected in SHP-2 immunoprecipitates from Rat2 fibroblasts expressing Fms. We have shown previously that the 145-kDa Ship protein is tyrosine phosphorylated in response to M-CSF in myeloid cells but not in fibroblasts (42). Our results with p100 further illustrate that Fms signaling in myeloid cells is at least in part different from that in fibroblasts.
p100 bound directly to the SH2 domains of SHP-2 and p85 but did not bind to immunoprecipitated SHP-1 or to GST fusion proteins containing the SH2 domains of SHP-1 (Figs. 1 and 5). Thus, two very similar tyrosine phosphatases, although simultaneously expressed in the same cells, do not interact with the same signaling protein. Binding to different signaling intermediates may be one mechanism by which SHP-1 and SHP-2 exert different effects on cellular signaling.
The consensus sequences for binding to the SH2 domains of SHP-2 and P85 are significantly different. Binding to the SHP-2 N-terminal SH2 domain is favored by Ala, Iso, Leu, or Val at the +1 and +3 positions relative to the phosphotyrosine (57), whereas binding to the SH2 domains of p85 generally requires Met at the +3 position (57, 58). It is likely therefore that SHP-2 and p85 bind to different phosphotyrosines on p100. Whereas either the N-terminal or C-terminal SH2 domain of p85 was sufficient to bind p100, both N- and C-terminal SH2 domains of SHP-2 were required for efficient binding to p100. The reason for this difference is unknown. We cannot exclude the possibility that the GST fusion proteins containing single SH2 domains of SHP-2 were improperly folded. However, these fusion proteins were soluble, undegraded, and very similar in sequence to single SHP-2 SH2 domain constructs used by Tauchi et al. to show binding of individual SH2 domains of SHP-2 to the EPO-R, c-Kit, and Grb2 (18, 19). The two SH2 domains of SHP-2 bind to two different phosphotyrosines on insulin receptor substrate-1 (Tyr1172 and Tyr1222) (31, 32). The two SH2 domains of SHP-2 might therefore bind to two different phosphotyrosines on a single p100 protein.
p100 appears to be a novel signaling protein. The 100-kDa protein did
not cross-react with antibodies to focal adhesion kinase, rasGAP, the
110-kDa catalytic subunit of PI 3-kinase, the adapter protein Eps8,
the insulin receptor
subunit, Vav, Fps/Fes kinase, or STAT1 (data
not shown). Other investigators have seen similar tyrosine
phosphorylated proteins after M-CSF stimulation of myeloid cells, but
these proteins were not further characterized. Sengupta et
al. (59) found that a 99-kDa protein is among the first proteins tyrosine phosphorylated after M-CSF stimulation of BAC1.2F5 cells. Intriguingly, this p99 is more susceptible to dephosphorylation at
37 °C than most other phosphoproteins. This might be particularly true for a phosphoprotein that is bound to a tyrosine phosphatase. Kanagasundaram et al. (44) saw a 95-kDa tyrosine
phosphorylated protein that coprecipitated with p85 after M-CSF
stimulation of BAC1.2F5 cells. As with the p100 protein described here,
this 95-kDa protein exhibits decreasing mobility on SDS-PAGE gels with time after M-CSF treatment. Welham et al. (20) found that
after M-CSF stimulation of FDMAC11/4.5 cells, a tyrosine phosphorylated protein of 100 kDa coimmunoprecipitates with p85. These p99, p95, and
p100 proteins could well be the same as the p100 protein characterized in this report.
A number of SHP-1 or SHP-2-binding proteins have been described that are probably not identical to our p100. Su et al. (60) recently described a 95-kDa protein that is tyrosine phosphorylated after epidermal growth factor receptor stimulation of 293 cells. Unlike our p100 protein, however, this p95 associates with the SHP-1 tyrosine phosphatase. SHPS-1, a recently cloned 115-120 kDa protein, is tyrosine phosphorylated in response to insulin and other mitogens in v-src-transformed rat fibroblasts (61). This is a transmembrane glycoprotein that binds to the SH2 domains of both SHP-1 and SHP-2. SHPS-1 lacks a YXXM consensus sequence, however, and so probably cannot bind to p85. A 115-kDa protein with characteristics similar to SHPS-1 has also been seen in Chinese hamster ovary cells after stimulation with insulin (62). In NIH 3T3 cells expressing the insulin receptor, a tyrosine phosphorylated protein of 120 kDa associates with the SH2 domains of SHP-2 after insulin stimulation. A 115-kDa tyrosine phosphorylated protein also coimmunoprecipitates with SHP-2 from HepG2 and NIH 3T3 cells stimulated with epidermal growth factor (63) and from Chinese hamster ovary or adipocyte cells after stimulation with insulin (64). In the latter case, however, the 115-kDa protein does not coprecipitate with the SH2 domains of SHP-2. It seems likely that many more potential SHP-2 substrates will be found, and these may include several different families of SHP-2-binding proteins.
The Drosophila homolog of SHP-2, Corkscrew, has been shown to bind a 115-kDa cytosolic protein called DOS (65, 66). Upon activation of the Sevenless, Torso, or DER receptor tyrosine kinases, DOS is tyrosine phosphorylated and binds to the SH2 domains of Corkscrew. Binding of DOS to p85 has not been demonstrated, but a YXXM motif is present in DOS (66). The function of DOS is not completely understood, but genetic analysis indicates that DOS acts downstream of growth factor receptors and upstream of or parallel to Ras1 and Raf (66). DOS appears to be a direct substrate of Corkscrew and analysis of DOS mutants suggests that the dephosphorylation of DOS by Corkscrew is a critical component of Sevenless signaling (65).
We do not know whether our p100 protein is similar to DOS, but the data
presented here allow us to make some predictions about its interactions
with other signaling proteins in myeloid cells (Fig. 9).
Tyrosine phosphorylation of p100 is among the earliest events after
M-CSF stimulation of Fms. p100 is probably tyrosine phosphorylated by a
kinase other than Fms, because it is phosphorylated after stimulation
by several different cytokines and in cells that do not express Fms
(Fig. 8 and data not shown). Phosphorylation of p100 allows it to bind
to the SH2 domains of SHP-2 and p85. Tyrosine phosphorylation of p100
and binding of p100 to SHP-2 precedes phosphorylation of SHP-2.
However, the SHP-2 that is coprecipitated with p85 seems to be more
highly tyrosine phosphorylated than the bulk of SHP-2 in the cell. This
suggests that binding of the SHP-2·p100 complex to p85 results in
tyrosine phosphorylation of SHP-2. Because p85 binds directly to Fms,
binding of the SHP-2·p100 complex to p85 could bring SHP-2 into the
vicinity of the receptor and perhaps facilitate phosphorylation of
SHP-2 by Fms. The p85 immunoprecipitates also include a number of other
signaling proteins, including Fms, Ship, Cbl, Shc, Grb2, and Sos (Fig.
3 and data not shown). The rapid dephosphorylation of p100 immediately
following phosphorylation of SHP-2 suggests that SHP-2 may
dephosphorylate p100 directly. The dephosphorylation of p100 occurs
relatively rapidly after M-CSF stimulation at a time when the level of
phosphotyrosine on some cellular proteins is still increasing (Fig.
2B). It is possible that p100, like DOS in
Drosophila, is a required positive effector of
receptor-mediated signaling. If so, then dephosphorylation of p100
by SHP-2 could be a prerequisite for the formation of subsequent
signaling complexes. Additional experiments using antibodies to p100 will be necessary to address these possibilities.
We thank Zhizhuang Zhao for helpful discussions and Ed Giniger, Susan Geier, and David Lucas for critiquing the paper.