(Received for publication, November 13, 1996, and in revised form, January 29, 1997)
From the Department of Physiology and
§ Department of Microbiology and Immunology, University of
Maryland School of Medicine, Baltimore, Maryland 21201
The 14-3-3 proteins have been implicated as
potential regulators of diverse signaling pathways. Here, using
two-hybrid assays and in vitro assays of protein
interaction, we show that the isoform of 14-3-3 interacts with the
insulin-like growth factor I receptor (IGFIR) and with insulin receptor
substrate I (IRS-1), but not with the insulin receptor (IR).
Coprecipitation studies demonstrated an IGFI-dependent
in vitro interaction between 14-3-3-glutathione S-transferase proteins and the IGFIR. In similar studies no
interaction of 14-3-3 with the IR was observed. We present evidence to
suggest that 14-3-3 interacts with phosphoserine residues within the
COOH terminus of the IGFIR. Specifically, peptide competition studies combined with mutational analysis suggested that the 14-3-3 interaction was dependent upon phosphorylation of IGFIR serine residues 1272 and/or
1283, a region which has been implicated in IGFIR-dependent transformation. Phosphorylation of these serines appears to be dependent upon prior IGFIR activation since no interaction of 14-3-3 was observed with a kinase-inactive IGFIR in the two-hybrid assay nor
was any in vitro interaction with unstimulated IGFIR derived from mammalian cells. We show that the interaction of 14-3-3 with IRS-1 also appears to be phosphoserine-dependent. Interestingly, 14-3-3 appears to interact with IRS-1 before and after
hormonal stimulation. In summary, our data suggest that 14-3-3 interacts with phosphoserine residues within the COOH terminus of the
IGFIR and within the central domain of IRS-1. The potential functional
roles which 14-3-3 may play in IGFIR and IRS-1-mediated signaling
remain to be elucidated.
The insulin and IGFI1 receptors share
a high degree of homology (1), and both are believed to signal in part
via the IRS and SHC proteins (2-6). Despite the apparently identical
signaling events which are initiated by the IR and IGFIR via these
signaling proteins, the primary physiological role of insulin is to
regulate metabolic events, while IGFI primarily regulates cellular
growth, differentiation, and transformation (7). It is therefore likely that divergent signaling pathways exist that are responsible for mediating these different cellular effects. Several studies have suggested the existence of such divergent pathways. First, receptor chimeras in which the cytoplasmic domain of the IGFIR was fused to the
IR, were reported to have significantly increased mitogenic potential
compared with the IR (8). Another study showed that the IR was
incapable of mediating cellular transformation when expressed in cells
derived from IGFIR-deficient mouse embryos, whereas the IGFIR was
capable of mediating this response (9). Studies on signal divergence
have largely focused on the COOH-terminal region of the receptors,
since this region is the most disparate (10-15). These studies
concluded that the COOH terminus of the IR was important for regulation
of a variety of cellular effects but few conclusions regarding the role
of the COOH terminus of the IGFIR were proposed. However, several
studies have suggested a role for the COOH terminus of the IGFIR. For
example, the region of the IGFIR between residues 1245 and 1310 has
been implicated in transforming activity of the IGFIR (16), and some
studies have suggested that residues 1293-1294 and the tyrosine at
position 1251 are involved in the transforming response (16). A recent study showed that four serines within the COOH-terminal region of the
IGFIR (1280-1283) were necessary for the transforming activity of the
IGFIR, since a mutant IGFIR in which these serines were mutated to
alanine was unable to transform cells (termed R cells) which
do not contain any endogenous IGFIR due to gene knockout via homologous
recombination (17). Since the IRS-1 and SHC signaling pathways were
unaffected by these serine mutations it was proposed that this region
of the IGFIR COOH terminus may activate an uncharacterized signaling
pathway required for cellular transformation.
The 14-3-3 proteins have been identified in numerous organisms
including yeast, plants, and humans (18). Seven different isoforms have
been identified, and these proteins have been associated with a diverse
number of biological activities including activation of tyrosine and
tryptophan hydroxylases, inhibition of protein kinase C, and an
uncharacterized role in cell cycle control (18, 19). More recently, the
14-3-3 proteins have been found to associate with a number of signaling
proteins including Raf (20-23), bcr-abl (24), polyoma virus middle T
antigen (25), PI 3-kinase (26), the protooncogene product Cbl (27), and
the cdc25 phosphatase (28). Together, these findings suggest a
potential for 14-3-3 proteins in the regulation of mitogenesis or
cellular transformation. The structure of the and
isoforms of
14-3-3 have recently been reported (29, 30) revealing a number of
interesting findings. First, as expected from previous biochemical
studies, the 14-3-3 proteins form a dimer via interactions at the
extreme NH2 terminus. Second, each monomer of 14-3-3 was
found to have a large negatively charged channel which theoretically
could serve to bind phosphorylated peptides. In support of this idea,
it has been shown that 14-3-3 binds specifically to serine 621 within
the Raf molecule in a phosphorylation-dependent manner
(31).
In this report, we examined the interaction between 14-3-3 and the insulin and IGFI receptors and IRS-1 using the yeast two-hybrid system and assays of in vitro protein interaction. We show that 14-3-3 interacts with the IGFIR, but not the insulin receptor, and that 14-3-3 also interacts with IRS-1. We demonstrate that serine-phosphorylated peptides corresponding either to the COOH-terminal region of the IGFIR or to the 14-3-3-binding site within Raf were able to compete the interaction of 14-3-3 with the IGFIR and IRS-1, whereas phosphotyrosine-containing peptides corresponding to known sites of tyrosine phosphorylation within the IGFIR had no effect upon either interaction. We show that an IGFIR mutant in which serines 1272 and 1280-1283 within the COOH terminus were mutated to alanine showed no interaction with 14-3-3 in the two-hybrid assay, but had no effect upon IGFIR interaction with IRS1, SHC, or p85. Taken together, these data suggest that 14-3-3 interacts with phosphoserine residues within the IGFIR COOH terminus. Since this region of the IGFIR has been shown to be necessary for IGFIR-dependent cellular transformation (17, 32), it is possible that 14-3-3 is involved in mediating the transforming activity of the IGFIR.
The yeast strain EGY40 and all
yeast expression plasmids were provided by the laboratory of Roger
Brent and have been previously described (33-35). All routine growth
and maintenance of yeast strains was as described (36). Plasmid
transformation of yeast was by the lithium acetate method (37). The IR
and IGFIR two-hybrid fusions and IGFIR K1003A mutant have been
previously reported (38-40). The IGFIR receptor mutants (4S, S1272A,
and 4S/S1272A) were made by site-directed mutagenesis according to the
method of Kunkel (41). The chimeric receptors joined within the kinase domain were constructed using a conserved XmnI site within
the amino terminus of the kinase domains. The COOH-terminal chimera was
generated from the previously reported construct (10) which was a gift
from Derek LeRoith. To obtain the final 14-3-3 clone, we utilized PCR
and amplified cDNA derived from RNA obtained from a cell line
derived from a polyoma virus-induced mammary tumor. We used a 5 PCR
primer containing an EcoRI site and a 3
primer with a
BamHI site. The PCR primers were as follows:
5
-CCGGAATTCATGGATGATCGGGAGGATCTG-3
and
5
-CGGGATCCCGTGATTCTCATCTTCCAC-3
. Note that the final amino acid
(Gln-256) was inadvertently left off of this PCR-derived cDNA and
thus our 14-3-3 protein contains only amino acids 1-255. The PCR
product was first cloned into the pGBT9 vector
(CLONTECH) as an EcoRI/BamHI
fragment and subsequently transferred to the pJG4-5 interaction trap
plasmid as an EcoRI/SalI fragment by cloning into
the EcoRI/XhoI sites within the pJG4-5 plasmid.
The 14-3-3 deletion constructs were generated by insertion of linkers
into the following restriction sites: 1-203 (MunI), 1-184
(ApoI), 1-118 (StyI), and 118-255
(StyI). All two-hybrid and GST constructs were produced by
standard methods and detailed cloning strategies are available upon
request.
The solution -galactosidase
assays were performed as described (42) and the units of
-galactosidase activity were calculated by the method of Miller
(43). The values shown for the positive colonies represent the average
of 3 assays (each assay representing an independent colony).
GST fusion proteins were
generated with 14-3-3 and an IRS-2 fragment which we have previously
demonstrated to interact with the IR (44) using pGEX-5X-1 (Pharmacia
Biotech, Inc.). All GST fusions were expressed in DH5 bacteria and
purified onto glutathione-agarose beads using standard techniques (45).
The beads that contained immobilized fusion protein were then incubated
with cell lysates derived from CHO.T cells (which overexpress the IR)
(46) or rat-1 cells (which overexpress the IGFIR) (47) prior to or
after insulin or IGFI stimulation (100 nm, 10 min, 37 °C). Lysates
were prepared as described previously (44) in lysis buffer (50 mM HEPES (pH 7.6), 1% Triton X-100, 1 mM EGTA,
10 mM sodium fluoroide, 20 mM sodium
pyrophosphate, 1 mM phenylmethylsulfonyl fluoride, 1 mM sodium orthovanadate, and 10 µg/ml aprotinin and
leupeptin). The lysates were then incubated with the immobilized GST
proteins at 4 °C overnight. After washing four times with 50 mM HEPES (pH 7.6), 150 mM NaCl, 0.1% Triton
X-100, the proteins that coprecipitated with 14-3-3 were analyzed by
SDS-PAGE followed by immunoblotting with anti-phosphotyrosine
antibodies (PY20, Transduction Labs), anti-IGFIR (IGFIR 1-2, a gift
from Ken Siddle), or anti-IRS1 (UBI).
Peptides were made which correspond either to the region of Raf which interacts with 14-3-3 or to the various regions of the IGFIR, both with or without serine or tyrosine phosphorylation. The peptides were added to both the immobilized 14-3-3 GST protein and the cellular lysates at concentrations of 0, 1, 5, 10, 50, and 100 µM for the Raf peptide competition, and at 100 µM for the IGFIR peptide competitions. Incubation of the GST with the cellular lysates and analysis of precipitated proteins are as outlined above. In addition to the serine-phosphorylated peptides shown in the figures herein, the following peptides which correspond to regions of the IGFIR were also tested (data not shown) either with or without phosphorylated tyrosine residues: juxtamembrane, GVLpYASVNPEpYFSAA; catalytic loop, NTRDIpYETDpYpYRKGG; and COOH terminus, RASFDERQPpYAHMNG.
We have previously demonstrated the utility of the yeast
two-hybrid system in the characterization of the interactions between the insulin and IGFI receptors and their known intracellular substrates including IRS-1, IRS-2, SHC, and GRB10 (38-40, 44, 48, 49). Here we
show that the 14-3-3 proteins also interact with the IGFIR and IRS-I.
The 14-3-3 proteins have been implicated in a number of signaling
pathways through their association with signaling proteins including
Raf, cdc25, polyoma middle T antigen, cbl, and Bcr-Abl (20-28). We
initially set out to examine the interaction of 14-3-3 with the polyoma
middle T antigen (25) using the two-hybrid system. Although we were
unable to demonstrate an interaction between these
proteins,2 we did note a strong interaction
of 14-3-3 with the IGFIR in the yeast two-hybrid system. The results of
our initial experiments are shown in Fig. 1A.
In an attempt to identify the minimal domain of 14-3-3 which was
sufficient to interact with the receptor, we made five constructs of
the isoform of 14-3-3 and tested them for interaction with the
insulin receptor or IGFIR in the two-hybrid system. We saw high levels
of activity in both the colony color and solution assays with both
full-length 14-3-3 and the 1-203 activation domain hybrids when tested
against the IGFIR-LexA bait. Conversely, 14-3-3 showed no interaction
with the insulin receptor in the two-hybrid system. A much lower amount of activity was seen with the 14-3-3 (1-184) activation domain hybrid
and the IGFIR compared with full-length 14-3-3, and no interaction was
detected with either the 1-118 or 118-255 constructs. We conclude
that an important element for 14-3-3 binding to the IGFIR is located
between amino acids 184 and 203 of 14-3-3, although this region by
itself is insufficient for interaction. The inability of the
COOH-terminally truncated 14-3-3 to interact with the receptor is
consistent with the structure of 14-3-3, in which the COOH terminus of
each of the protein monomers forms a binding groove which has been
proposed to function as a protein-binding domain (29, 30). We examined
whether the interaction between 14-3-3 and IGFIR was dependent upon the
kinase activity of the IGFIR by testing interaction with a
kinase-inactive IGFIR in which the critical lysine (Lys-1003) has been
mutated to alanine (denoted K1003A). We detected no interaction between
any of the 14-3-3 activation domain hybrids and the kinase-dead mutant
in the two-hybrid system.
Since 14-3-3 interacts with the IGFIR, but not the insulin receptor, we constructed three chimeric receptors with regions of the insulin and IGFI receptor cytoplasmic domains and tested them for interaction with 14-3-3 in the two-hybrid system. We were able to detect interaction of 14-3-3 only with the chimeras which contained the COOH terminus of the IGFIR (Fig. 1B). We saw no interaction between 14-3-3 and either of two chimeras which contained the COOH-terminal region of the insulin receptor. Since the COOH-terminal chimera replaces the 107 carboxyl residues of the IGFIR with 99 amino acids derived from the IR (10), we conclude that 14-3-3 requires the COOH-terminal 107 amino acids of the IGFIR for interaction.
The COOH-terminal region of the IGFIR contains two tyrosines at
position 1250 and 1251 which are not present on the insulin receptor
and have been implicated in IGFIR signaling. Specifically, mutation of
tyrosine 1251 to phenylalanine results in an IGFIR which is mitogenic,
but unable to transform R cells (16). Although it is unclear
whether or not these tyrosines are in fact phosphorylated in
vivo, we hypothesized that 14-3-3 could be interacting with these
tyrosines in a phosphorylation-dependent manner, since
14-3-3 did not interact with the kinase-inactive IGFIR. To test this,
we made site-directed mutants in which either of these tyrosines was
mutated to phenylalanine and tested them for interaction with 14-3-3 in
the two-hybrid system. We saw undiminished levels of activity for
interaction of both mutants with 14-3-3 (Fig. 1B),
indicating that these tyrosines are not required for interaction with
14-3-3.
To further examine the interaction between the IGFIR and 14-3-3, we
generated a 14-3-3-GST fusion protein and used it to precipitate proteins from lysates derived from IGFI-stimulated or unstimulated IGFIR-overexpressing cells (47). The 14-3-3-GST fusion protein was able
to precipitate phosphorylated IGFIR from rat1-IGFIR cells, as well as a
second phosphorylated band of approximately 180 kDa (Fig.
2A, top panel, lane 4). As expected, the
IRS-2 GST fusion protein was also able to precipitate the
phosphorylated IGFIR from the rat1-IGFIR cell lysates (Fig. 2A,
top panel, lane 2). In contrast, the 14-3-3-GST fusion protein was
unable to precipitate any detectable IR from insulin-stimulated CHO-IR
cell lysates, although it did interact with the higher molecular weight
band seen in the rat1-IGFIR precipitates (Fig. 2A, bottom panel,
lane 4).
To determine whether the 95-kDa phosphoprotein precipitated from the rat1-IGFIR cellular lysates was the IGFIR, we probed an identical blot with antibodies to the IGFIR. The phosphorylated protein precipitated from rat1-IGFIR cells by both the IRS-2 and 14-3-3-GST proteins was indeed the IGFIR (Fig. 2B, lanes 2 and 4). The identity of the higher molecular weight band in the IGFIR immunoblots is unclear. We also probed an identical blot with antibodies to IRS-1, and determined that the larger phosphoprotein was at least in part IRS-1 (Fig. 2C). Interestingly, a larger amount of IRS-1 coprecipitated with the 14-3-3-GST fusion prior to stimulation of the cells with IGFI (Fig. 2C, compare lanes 5 and 6). No IRS-1 was detected in precipitates using GST alone (Fig. 2C, lanes 8 and 9). We also demonstrated a direct interaction between 14-3-3 and IRS-1 in the two-hybrid assay (Table I). Using human IRS-1-LexA fusion proteins as bait, we were able to detect significant interaction with 14-3-3. We were unable to utilize a full-length IRS-1 bait hybrid because of intrinsic activation of the reporter genes which caused expression of lacZ in the absence of an activation domain hybrid (not shown). However, the two IRS-1 bait fusions shown (45-516 and 516-865) did not autoactivate and were therefore useful for the two-hybrid assay. Both of these constructs showed some interaction with 14-3-3 in the colony color assay although the strongest interaction occurred with the hybrid protein containing amino acids 516-865.
|
It has been shown that 14-3-3 associates with the Raf protein (22),
although the physiological role of this interaction remains unclear
(50). Recently, one domain of Raf which interacts with 14-3-3 has been
identified, and has been shown to be dependent upon serine
phosphorylation of Raf (31). The recently reported crystallographic
structure of 14-3-3 suggests the presence of a groove within 14-3-3 which forms a possible binding pocket for phosphorylated amino acids
(29, 30). We tested whether this same region of 14-3-3 might be
responsible for the interaction of 14-3-3 with the IGFIR and/or IRS-1
by asking whether the interaction between 14-3-3 and either the IGFIR
or IRS-1 could be blocked using phosphopeptides corresponding to the
Raf-binding site (Fig. 3A). As shown in Fig.
3B, 14-3-3 interaction with both the IGFIR and IRS-1 was
reduced by addition of the phosphorylated peptides in a
concentration-dependent manner (Fig. 3B, left
panel). No competition was seen when increasing amounts of
unphosphorylated peptide were utilized (Fig. 3B, right
panel). These data indicate that the IGFIR and IRS-1 interact with
14-3-3 at the same site as does Raf and also suggest that phosphoserine
residues within the IGFIR and IRS-1 may also be necessary for
interaction of 14-3-3.
We next examined the possibility that 14-3-3 may interact with phosphoserine residues within the IGFIR COOH terminus. Using the Raf sequence as a guide, we noted two serines within the COOH-terminal region of the IGFIR that were potential sites for interaction with 14-3-3. Specifically, the serines at positions 1272 and 1283 contain a motif similar to the Raf region which binds 14-3-3, since both have the COOH-terminal Ser-X-Pro. In addition, these serines are good candidates for interaction with 14-3-3 because both are unique to the IGFIR and do not have analogous residues within the IR. We also noted that serine 1283 was located within a serine quartet which had been shown to be essential for IGFIR-dependent transformation of cells (17). To determine whether serines 1272 and 1280-1283 were responsible for 14-3-3 interaction with the IGFIR, we mutated these serines to alanine, in combinations of S1272A alone, S1280A/S1281A/S1282A/S1283A (termed 4S), or both S1272A and S1280A/S1281A/S1282A/S1283A. These mutant IGFI receptors were expressed as LexA fusion proteins and tested for interaction with 14-3-3 in the two-hybrid system. The results are shown in Table I. While both the S1272A and the 4S mutants interacted efficiently with 14-3-3, the S1272A/4S combination mutant showed no interaction with 14-3-3. The S1272A/4S IGFIR mutant maintained high levels of interaction with IRS-1, SHC, and the p85 subunit of PI 3-kinase in the two-hybrid assay, which indicated that the mutant receptor was functioning properly and that the decreased interaction with 14-3-3 was not due to a kinase-dead or otherwise structurally-impaired receptor.
We next examined whether serines 1272 and 1283 were responsible for
14-3-3 interaction with the IGFIR, and whether the phosphorylation state of these serines was critical for the interaction. To do this, we
designed peptides to the COOH-terminal region of the IGFIR spanning
both serine 1272 and serine 1283, with or without phosphorylation of
these serine residues (Fig. 4A).
IGFI-stimulated rat 1-IGFIR cellular lysates were incubated with these
peptides prior to incubation with the 14-3-3 GST protein, and the
results of the 14-3-3 precipitation are shown in Fig. 4B.
When the non-phosphorylated peptides were added to the lysates
(lane 2), 14-3-3 precipitated both the phosphorylated IGFIR
and IRS-1 in a manner identical to the control precipitation in
lane 1. However, when phosphorylated peptides were added to
the lysates, the interaction between 14-3-3 and both the IGFIR and
IRS-1 was almost completely eliminated (Fig. 4B, lane 3).
This result suggests that the interaction between 14-3-3 and the IGFIR
is mediated by the phosphorylation of serine 1272 and/or 1283 on the
IGFIR COOH terminus. These results also suggest that both IRS-1 and the
IGFIR interact with 14-3-3 at the same site within 14-3-3, since the
IGFIR peptides blocked both interactions. The interaction of 14-3-3 with IRS-1 derived from unstimulated cells is likely to be due to basal
serine phosphorylation of IRS-1 (51-53). The subsequent tyrosine
phosphorylation of IRS-1 may reduce 14-3-3 interaction due to
structural alterations or steric mechanisms. Alternatively, tyrosine
phosphorylation of the IRS-1 may result in interaction of IRS-1 with
other cellular proteins which have greater affinity for IRS-1 and
therefore block interaction with 14-3-3 in vitro.
Interestingly the fragment of IRS-1 which appears to interact best with
14-3-3 (516-865) contains five Ser-X-Pro motifs, one of
which (amino acid 641) has the sequence KSVSAP which is very close to
the Raf consensus 14-3-3-binding site RSXpSXP.
Further experiments will be needed to determine the exact region or
regions within the IRS-1 molecule which interact with 14-3-3.
Although the IGFIR serine phosphopeptides were able to compete for the interaction of 14-3-3 with the IGFIR and IRS-1, it remained possible that 14-3-3 may interact with phosphotyrosine residues. This possibility seemed plausible since we observed interaction (a) only with the kinase-active IGFIR in the two-hybrid assay and (b) only with IGFI-stimulated receptors in rat1-IGFIR lysates. To address the possibility that 14-3-3 could be interacting with a tyrosine-phosphorylated site on the IGFIR, we investigated whether tyrosine-phosphorylated peptides corresponding to known tyrosine phosphorylation sites within the IGFIR could block interaction of the 14-3-3-GST fusion protein with the IGFIR or IRS-1 from IGF-stimulated cell lysates. Tyrosine-phosphorylated peptides corresponding to the juxtamembrane, catalytic loop, or the COOH-terminal tyrosine were unable to block the interaction between 14-3-3 and the IGFIR or IRS-1 (data not shown). These results suggest that 14-3-3 binding to the IGFIR is not directly dependent upon tyrosine phosphorylation, but rather that activation or autophosphorylation of the IGFIR is necessary to allow an unknown serine kinase to phosphorylate the IGFIR, thereby allowing 14-3-3 to interact with the receptor.
We have shown that the isoform of 14-3-3 interacts with the
IGFIR in the two-hybrid system and in vitro, and that this
interaction is dependent upon the kinase activity of the receptor. We
have localized this interaction to the COOH-terminal region of the IGFIR, specifically to serines 1272 and 1283 in a region which differs
from the insulin receptor and has been implicated as being essential
for the transforming activity of the IGFIR (17). Our phosphopeptide
competition experiments suggest that this interaction is dependent upon
serine phosphorylation. IGFI-dependent serine phosphorylation of the IGFIR has been previously demonstrated (17, 54, 55), although it is unclear whether serines 1272 and 1283 are
phosphorylated. The IGFIR serine phosphorylation is most likely due to
a serine kinase which has yet to be identified. One candidate serine
kinase is protein kinase C, since phorbol esters can stimulate serine
phosphorylation of the IGFIR. However, the serines within this region
of the IGFIR are not contained within recognized protein kinase C
consensus sequences. In addition, the pattern and extent of IGFIR
serine phosphorylation induced by protein kinase C activation within
cells differs from that induced by IGFI and it is unlikely that
IGFI-induced serine phosphorylation is entirely due to protein kinase C
activity (54, 55). If an unidentified serine kinase is responsible for
phosphorylation of the IGFIR, our data suggest that this kinase must be
present in a variety of cells, including the rat1 fibroblasts used in our GST experiments and even, perhaps fortuitously, in the yeast used
in the two-hybrid system. A second less likely possibility is that the
IGFIR may autophosphorylate on serine residues as has been proposed for
the insulin receptor (56, 57).
Recent reports suggest the importance of serine phosphorylation of
cellular proteins to 14-3-3 binding. The recently reported crystal
structure of the 14-3-3 dimer suggests the presence of binding grooves
which contain a cluster of arginine and lysine residues which may be
involved in binding to phosphorylated peptides (29, 30). In fact, one
study has proposed a phosphoserine-dependent 14-3-3 binding
motif based upon the site within Raf which interacts with 14-3-3 (31).
The consensus motif which was proposed was RSXpSXP. Although our findings are consistent
with some parts of this consensus, we have found that this exact
sequence is not absolutely necessary for 14-3-3 interaction with the
IGFIR in vitro. We have shown that 14-3-3 interacts with
serine residues 1272 and 1283 on the IGFIR, both of which lack an
NH2-terminal arginine residue but do contain
S-X-P motifs. Additionally, although serine 1272 on the
IGFIR lacks a serine residue at the 2 position, serine 1283 does have
a serine in the
2 position. The differences between our findings and
the proposed consensus may be due to many reasons. It may be that
14-3-3 interaction with the IGFIR is more complex than its interaction
with Raf. For example, we have demonstrated that serines 1272 and 1283 are each sufficient for 14-3-3 interaction with the IGFIR, whereas
interaction with Raf only requires a single serine residue. It may also
be true that different isoforms of 14-3-3 may bind to distinct
phosphoserine motifs, since we examined the
isoform and others have
examined the binding of the
,
,
, and
isoforms to the Raf
phosphopeptides (31). Clarification of the exact amino acids necessary
for 14-3-3
interaction with the IGFIR will require further
study.
Although the insulin and IGFI receptors share a high degree of
homology, insulin is primarily a metabolic hormone while IGFI is
implicated in differentiation, mitogenesis, and cellular
transformation. The fact that 14-3-3 interaction is specific to the
IGFIR compared with the IR is at least suggestive that this interaction
may play a role in mitogenesis or cellular transformation. Since the
receptors differ most highly in the COOH-terminal region, it is likely
that this region is involved in the signal divergence between the two receptors. We have localized the site of 14-3-3 interaction with the
IGFIR to serines 1272 and 1283 on the IGFIR, residues which are not
present within the insulin receptor. Intriguingly, serine 1283 is
located within a serine quartet which has recently been shown to be
critical for IGFIR-dependent cellular transformation (17).
In these experiments IGFIR () (R
) cells that were
transfected with IGFI receptors in which serines 1280-1283 were
mutated to alanine were unable to form colonies in soft agar, although
they were fully responsive to IGFI-stimulated mitogenesis under normal
culture conditions. Phosphorylation of IRS-1 and SHC was not impaired
in these cells, and PI 3-kinase activity was unaffected. Our data are
in agreement with this observation, since IRS1, SHC, and the p85
subunit of PI 3-kinase all interact normally with the S1272A/4S mutant
IGFIR in the two-hybrid system. The implication of these findings is
that the serine quartet on the IGFIR is important for cellular
transformation via a transforming pathway which is separate from the
mitogenic pathways (17). Our data suggest that this pathway may involve
the 14-3-3 proteins, and that the interaction involves serine
phosphorylation of the IGFIR.
A role for 14-3-3 in IGFI-mediated transforming activity is an
attractive possibility, but 14-3-3 may play other roles in cellular
function. It has been shown that 14-3-3 binds to and activates Raf (20,
22, 23, 58), although whether 14-3-3 is essential for Raf activity has
since come into question (59). One role which has been suggested for
14-3-3 is the formation of structural "bridges" connecting
signaling proteins in the cell, which is possible since the 14-3-3 proteins form both hetero- and homodimers (60). One study suggests that
14-3-3 may allow Raf to stimulate the phosphatase activity of cdc25
phosphatase by facilitating interaction between these two proteins
(28). In the case of IGFIR signaling, 14-3-3 may function to bring Raf into close proximity with the membrane by virtue of its interaction with the IGFIR and/or IRS-1. This may allow Raf to interact more efficiently with upstream effector proteins such as Ras. Another recent
report implicates 14-3-3 proteins as modulators of PI 3-kinase activity
by facilitating association between the protein tyrosine kinase
substrate Cbl and PI 3-Kinase in Jurkat T-cells (27). 14-3-3 has been
shown to associate with and inhibit both PI 3-kinase and protein kinase
C in these cells (26, 61). It is possible that 14-3-3 is involved in
regulating PI 3-K activity resulting from IGFI stimulation by linking
PI 3-kinase with the IGFIR and/or IRS1. Last, it is interesting to note
that expression of the epsilon isoform of 14-3-3 has been shown to be
regulated during embryonic development (62). In these studies it was
noted that expression of the isoform of 14-3-3, which was high
during early mesenchyme development, dropped dramatically during late
development in mesenchyme condensations which were destined to become
cartilage, bone, and muscle. This is intriguing since these tissues are
known to be very responsive to the IGFs, and therefore 14-3-3 may play
a role in the regulation of IGFI-dependent
differentiation.
The interaction we observed between 14-3-3 and IRS-1 is also interesting. We noted that this interaction is competed efficiently with phosphoserine-containing peptides, and in addition appears to be stronger prior to tyrosine phosphorylation of IRS-1 by the IGFIR, which may suggest a phosphoserine-dependent interaction since IRS-1 is known to be basally phosphorylated upon serines (52). In addition, our finding that IRS-1 can interact with 14-3-3 in the two-hybrid assay supports the role of serine phosphorylation but not tyrosine phosphorylation since it is unlikely that IRS-1 proteins are tyrosine phosphorylated in yeast whereas it is possible for IRS-1 to be serine phosphorylated, perhaps by the same kinases which phosphorylate the IGFIR. The function of the 14-3-3/IRS-1 interaction is unclear, but again may be related to the structural "bridging" role suggested for 14-3-3. It is possible that the interaction between 14-3-3 and IRS-1 is in part responsible for localizing 14-3-3 to the membrane and/or the IGFIR, thus allowing 14-3-3 to interact with the IGFIR or other signaling components. We have not yet identified the site or sites on IRS-1 which bind to 14-3-3.
Last, much recent work has suggested that serine phosphorylation of
either the insulin receptor and/or IRS-1 may play an important role in
mediating insulin resistance (63-65). In these papers tumor necrosis
factor has been suggested to lead to serine phosphorylation of
IRS-1 resulting in a poorly understood decrease in the ability of the
insulin receptor to phosphorylate IRS-1 upon tyrosines, thus leading to
impaired insulin signaling. Our data, showing that proteins such as
14-3-3 can interact with IRS-1 in an apparently phosphoserine-dependent manner provides a potential
molecular mechanism by which serine phosphorylation might down-regulate IRS-1-mediated signaling. In this model, serine phosphorylation of
IRS-1 may lead to interaction with phosphoserine-binding proteins such
as 14-3-3 which may hinder signaling, perhaps by sterically blocking
access of IRS-1 to the insulin receptor or by blocking access to
downstream signaling proteins. Identification of the serine residues
within IRS-1 which mediate this effect will be essential to
understanding the molecular mechanisms of this response.
We thank Derek LeRoith for the IGFIR/IR COOH-terminal chimeric cDNA, Ken Siddle for the anti-IGFIR antibody, Michael Weber for the rat1-IGFIR cell line, and Thomas J. O'Neill for consultation and technical assistance on this project.