(Received for publication, May 5, 1997)
From the Eppley Institute for Research in Cancer and the Department of Pathology and Microbiology, University of Nebraska Medical Center, Omaha, Nebraska 68198-6805
The c-fes proto-oncogene encodes a non-receptor tyrosine kinase (Fes) that has been implicated in cytokine receptor signal transduction and myeloid differentiation. Previous work from our laboratory has shown that Fes autophosphorylates via an intermolecular mechanism more commonly associated with growth factor receptor tyrosine kinases. Analysis of the Fes amino acid sequence with the COILS algorithm indicates that the N-terminal region of the protein has a very high probability of forming coiled-coil structures often associated with oligomeric proteins. These findings suggest that oligomerization may be a prerequisite for trans-autophosphorylation and activation of Fes. To establish whether the active form of Fes is oligomeric, we performed gel-filtration experiments with recombinant Fes and found that it eluted as a single symmetrical peak of approximately 500 kDa. No evidence of the monomeric, 93-kDa form of the protein was observed. Deletion of the unique N-terminal domain (amino acids 1-450, including the coiled-coil homology region) completely abolished the formation of oligomers. Furthermore, co-precipitation assays demonstrated that an immobilized glutathione S-transferase fusion protein containing the Fes N-terminal region bound to full-length Fes but not to a mutant lacking the N-terminal region. Similarly, a recombinant Fes N-terminal domain protein was readily cross-linked in vitro, whereas the SH2 and kinase domains were refractory to cross-linking. Incubation of wild-type Fes with a kinase-inactive Fes mutant or with the isolated N-terminal region suppressed Fes autophosphorylation in vitro, suggesting that oligomerization may be essential for autophosphorylation of full-length Fes. The presence of an oligomerization function in the Fes family of tyrosine kinases suggests a novel mechanism for non-receptor protein-tyrosine kinase regulation.
The c-fes proto-oncogene encodes a non-receptor protein-tyrosine kinase (Fes) that is expressed predominantly in hematopoietic cells (1-3). Several lines of evidence suggest that Fes may influence myeloid differentiation commitment. Transfection of the myeloid leukemia cell line K-562 with c-Fes resulted in growth suppression and terminal differentiation (4). K-562 cells are derived from chronic myelogenous leukemia and express the p210 form of Bcr/Abl but not Fes. In this context, restoration of Fes is sufficient to suppress transformation by Bcr/Abl in addition to generating a more differentiated phenotype. Recent studies from our laboratory have shown that normal Bcr is a substrate for Fes and suggest that Fes may interact directly with Bcr/Abl as well (5, 6). In related studies, an activated viral homolog of Fes (v-Fps) was shown to induce differentiation of chicken bone marrow cells to mature macrophages in the absence of CSF-1 (7). Thus, activation of Fes is sufficient to induce terminal differentiation of normal as well as leukemic progenitor cells.
Although Fes was originally thought to be restricted to hematopoietic cells, more recent work has uncovered a broader pattern of expression. First, studies of c-fes expression in developing mouse embryos showed that it was not only present in hematopoietic cells but also in non-hematopoietic tissues where expression is not observed in the adult (8). In particular, c-fes transcripts were detected in embryonic lung, liver, spinal cord, skin, gut, heart, and kidney. These results suggest that Fes may play a critical role in the early development of many tissues in addition to its role in hematopoiesis. A second study reports the surprising finding that Fes may be involved in the regulation of the vascular endothelium. Expression of an activated form of Fes in transgenic mice induced hypervascularity which progressed to multifocal hemangiomas (9). Expression of endogenous Fes was also observed in primary human vascular endothelial cells at levels comparable to those detected in myeloid cells. These results show that in addition to hematopoietic cells, Fes expression is also intrinsic to cells of the vascular endothelial lineage and imply that Fes may play a role in angiogenesis. A more recent study demonstrated Fes expression in other adult tissues, including neuronal and epithelial cells (10). The function of Fes in these tissues is currently unknown.
A common feature of receptor tyrosine kinases is that they are activated by oligomerization in response to ligand binding (11, 12). The oligomerized receptor is then capable of trans-autophosphorylation, which is essential for downstream signaling through the recruitment of effector proteins with SH2 domains (13, 14). The mechanism regulating the kinase activity of Fes and other non-receptor tyrosine kinases is less clear. Fes tyrosine kinase activity is tightly regulated in cells, and even under conditions of overexpression, very little tyrosine autophosphorylation is observed (3, 15, 16). Tyrosine autophosphorylation of Fes appears to be essential for biochemical activation and biological function, as c-Fes and v-Fps mutants with Phe substitutions of the major autophosphorylation site show reduced enzymatic and biological activity (16-19). Recently, we demonstrated that Fes autophosphorylation occurs by an intermolecular mechanism, suggesting that Fes activation occurs as a result of transphosphorylation in a manner analogous to receptor tyrosine kinases (16). Consistent with this idea is the observation that Fes associates with multiple hematopoietic cytokine receptors and is activated in response to cytokine binding (20-24), which stimulates the oligomerization of cytokine receptors (25-27). Alternatively, Fes may have an intrinsic oligomerization capacity that may drive autophosphorylation and activation. In this study, we demonstrate that the Fes tyrosine kinase is an oligomeric protein. Amino acid sequences found in the unique N-terminal domain mediate oligomerization and exhibit striking homology to coiled-coil domains associated with other classes of oligomeric proteins (28, 29). Furthermore, we observed that a kinase-inactive mutant of Fes as well as the isolated N-terminal region were able to associate with wild-type Fes and suppress autophosphorylation, consistent with a model for Fes activation that requires oligomerization and trans-autophosphorylation. The presence of an oligomerization domain in Fes suggests a novel mechanism for regulation of its protein-tyrosine kinase activity.
Expression of Fes wild-type, K590E, Y713F,
N,
SH2,
KIN, and isolated N-terminal proteins with C-terminal
FLAG epitope tags using the baculovirus/Sf-9 cell system is described
in detail elsewhere (5, 16). The expression of the Fes N-terminal
domain as well as full-length Fes as GST1
fusion proteins (GST-N-Fes and GST-Fes, respectively) in Sf-9 cells has
also been described (16). To create the
N-KE Fes mutant, a unique
KpnI/EcoRI restriction fragment was isolated from
the Fes K590E mutant cDNA (17) and swapped with the corresponding fragment in the wild-type
N mutant. The resulting
N-KE Fes
cDNA was subcloned into the baculovirus transfer vector pVL1392 and used to generate a recombinant baculovirus as described elsewhere (16,
30).
A Sephacryl S-300 column (1.5 × 75 cm)
was equilibrated with S-300 buffer (50 mM Tris-HCl, pH 7.4, 50 mM NaCl, 1 mM EDTA, 10 mM
MgCl2) and calibrated with thyroglobulin (669 kDa),
ferritin (440 kDa), catalase (232 kDa), and bovine serum albumin (67 kDa). Sf-9 cells expressing full-length Fes-FLAG (16) were sonicated in
S-300 buffer, and the lysate was clarified by centrifugation at
100,000 × g for 1 h at 4 °C. The clarified
supernatant (0.5 ml) was applied to the S-300 column and eluted with
S-300 buffer at a flow rate of 0.25 ml/min. Fractions (1.0 ml) were
collected and the position of Fes in the elution profile was determined by immunoblotting with the anti-FLAG M2 monoclonal antibody (IBI/Kodak) as described elsewhere (16). Similar gel-filtration procedures were
used for the Fes N and N-terminal proteins, except a Sephacryl S-200
column was employed (1.0 × 100 cm); 0.5-ml fractions were collected and standard proteins included catalase, bovine serum albumin, and carbonic anhydrase (29 kDa).
The Fes amino acid sequence was analyzed using the COILS algorithm (28, 29).2 A gliding window of 28 residues was used for the analysis presented. A detailed description of the program is available.2
Co-precipitation Assay for Fes OligomerizationSf-9 cells were co-infected with recombinant baculoviruses encoding the Fes N-terminal region fused to GST (GST-N-Fes) and wild-type or mutant Fes proteins with the FLAG epitope fused to their C termini. Forty-eight hours postinfection, cells were lysed by sonication as described (16), and the GST-N-Fes protein was precipitated with glutathione-agarose. The glutathione-agarose beads were washed with RIPA buffer (16), and the presence of associated Fes was detected by immunoblotting with antibodies to the FLAG epitope (M2 antibody). As a negative control, parallel experiments were conducted in which Fes proteins were co-expressed with GST. Expression of GST and GST-N-Fes was verified by SDS-PAGE and Coomassie staining of an aliquot of each glutathione-agarose precipitate.
Chemical Cross-linkingThe Fes N-terminal domain (amino acids 1-450) and the SH2/kinase region (amino acids 451-822) were expressed as FLAG fusion proteins in Sf-9 cells. Cells were lysed by sonication (16) and diluted 1:10 with lysis buffer, and the bifunctional cross-linking reagent disuccinimidyl suberate (DSS) was added from a concentrated stock solution in Me2SO to final concentrations ranging from 0.1 to 2.0 mM. The reactions were incubated for 5 min at room temperature and were stopped by heating in SDS-PAGE sample buffer. The effect of the cross-linker on the Fes proteins was analyzed by immunoblotting with antibodies to the FLAG epitope. Parallel incubations with Me2SO alone were run as a negative control.
Kinase AssaysLysates from Sf-9 cells expressing a GST-Fes
fusion protein, a kinase-inactive mutant of Fes (Fes-KE) (17), the
isolated Fes N-terminal domain, or a kinase-inactive N-terminal
deletion mutant (N-KE) were mixed in various ratios and incubated
for 1 h at 4 °C. Fes proteins were immunoprecipitated with the
anti-FLAG monoclonal antibody and incubated with 10 µCi of
[
-32P]ATP (3,000 Ci/mmol) in 40 µl of kinase buffer
(50 mM Hepes, pH 7.4, 10 mM MgCl2)
for 10 min at 30 °C. Reactions were stopped by heating in SDS-PAGE
sample buffer, and phosphorylated proteins were resolved by SDS-PAGE.
Relative protein levels were quantitated by two-dimensional laser
densitometry of the Coomassie-stained gel. The relative extent of
autophosphorylation of GST-Fes was analyzed by storage phosphor
technology (Molecular Dynamics PhosphorImager). GST-Fes
autophosphorylation was corrected for protein levels and is plotted as
percent of control activity observed in the absence of the added mutant
Fes proteins. To investigate the interaction of GST-Fes with Fes-KE,
N-terminal, and
N-KE, Sf-9 cells were co-infected with GST-Fes (or
GST as a negative control) and each of these mutants. GST-Fes and GST
were precipitated from infected cell lysates, and associated Fes
proteins were detected by immunoblotting as described above.
Recent work from our
laboratory demonstrated that Fes autophosphorylation occurs by an
intermolecular mechanism (16). This finding suggests that Fes
activation may require oligomerization in a manner analogous to growth
factor receptor tyrosine kinases (11, 12). Gel-filtration experiments
were conducted to determine whether the active form of Fes is
oligomeric. Recombinant Fes was expressed in Sf-9 insect cells, and
clarified lysates from the infected cells were applied to a Sephacryl
S-300 column calibrated with thyroglobulin (669 kDa), ferritin (440 kDa), catalase (232 kDa), and bovine serum albumin (67 kDa). The
position of Fes in the elution profile was determined by immunoblot
analysis. As shown in Fig. 1A, Fes eluted
from the column as a single oligomeric peak of approximately 500 kDa.
No peak was observed in 93-kDa range, which corresponds to the unit
molecular mass of Fes. This result clearly shows that the active form
of Fes is oligomeric.
The Fes kinase has a modular structure consisting of unique N-terminal,
SH2, and kinase domains. To determine which regions of the protein
contribute to oligomerization, Sephacryl S-200 gel-filtration
experiments were performed on recombinant Fes proteins either lacking
the unique N-terminal domain (N; Fig. 1B) or on the
isolated N-terminal region itself without the SH2 and kinase domains
(N-term; Fig. 1C). The Fes N-terminal
deletion mutant eluted from the column as a monomer of approximately 40 kDa, indicating that the SH2 and kinase domains are not involved in
oligomerization. By contrast, the N-terminal region was exclusively
oligomeric, with no evidence for the 50-kDa monomeric form of this
protein. These data suggest that structural features found within the
unique N-terminal domain regulate oligomerization of the Fes
molecule.
Coiled-coil domains
often mediate protein oligomerization (31). This structural element is
comprised of an amphipathic -helix with a characteristic heptad
repeat in which the first and fourth positions are occupied by
hydrophobic amino acids. As a result, these residues are positioned on
the same face of the helix and pack together to form a hydrophobic
core. The remaining amino acids are often hydrophilic and allow for
solvation of the structure. To determine whether a coiled-coil domain
exists within the Fes protein, we analyzed the Fes amino acid sequence
using the COILS algorithm (28, 29). This program measures the
probability that a given amino acid and its surrounding sequence exists
as an amphipathic
-helix capable of forming a coiled-coil and plots the result against the amino acid position in the sequence. As shown in
Fig. 2, a region spanning Fes N-terminal amino acids 128-169 is predicted by the program to have a nearly absolute probability of forming a coiled-coil structure, consistent with the
gel-filtration results (Fig. 1). Other regions within the N-terminal
domain were also predicted to possess coiled-coil forming motifs,
although with lower probability relative to the 128-169 region. By
contrast, the SH2 and kinase domains are essentially devoid of any
predicted coiled-coil domains. As shown in more detail below, the
prediction that the unique N-terminal region of Fes confers its
oligomeric structure is confirmed by several additional lines of
experimental evidence.
Co-precipitation Assays and Chemical Cross-linking Identify an Oligomerization Domain in the Fes N-terminal Region
As described
above, gel-filtration experiments and computer analysis suggest that
the Fes N-terminal region is responsible for oligomerization of the
protein. To test this hypothesis further, a GST fusion protein
containing the N-terminal region of Fes (or GST alone as a negative
control) was co-expressed with Fes in a baculovirus/Sf-9 cell system.
The GST-N-terminal fusion protein was precipitated from the co-infected
cell lysates with glutathione-agarose and the presence of associated
Fes was determined by immunoblotting. As shown in Fig.
3, the GST-N-Fes fusion protein readily associated with
full-length Fes in the co-precipitation assay, whereas no association
was observed with GST itself. This result provides direct evidence that
the Fes N-terminal region is capable of protein-protein interactions
and is likely to mediate the oligomerization of Fes.
To identify the region of the Fes protein that associates with the
GST-N-terminal fusion protein, co-precipitation experiments were
conducted using deletion mutants of Fes lacking the N-terminal (N),
SH2 (
SH2), or kinase (
KIN) domains. As shown in Fig. 3, the
SH2 and
KIN mutants readily associated with the Fes N-terminal region, indicating that these domains are dispensable for
oligomerization. By contrast, the
N mutant was unable to associate
with the GST-N-terminal domain fusion protein, indicating that the
N-terminal region contains the structural elements essential for
oligomerization to occur.
We also investigated whether the phosphorylation state of Fes influenced oligomerization using the co-precipitation assay. For these experiments, we used a kinase-inactive mutant of Fes with a glutamic acid residue in place of the critical lysine residue in the ATP binding site (K590E mutant) (17) as well as a mutant with a phenylalanine in place of tyrosine at the major autophosphorylation site (Y713F mutant). As shown in Fig. 3, both of these mutants readily associated with the Fes N-terminal fusion protein, indicating that kinase activity is not required for oligomerization to occur.
Chemical cross-linking experiments were also performed as an
alternative method to demonstrate the Fes domain responsible for
oligomerization. The Fes N-terminal domain and the SH2/kinase region
were expressed independently as FLAG fusion proteins in Sf-9 cells.
Lysates from the cells were diluted 1:10 and incubated in the presence
and absence of the bifunctional cross-linking reagent DSS. The
reactions were stopped with SDS-PAGE sample buffer and analyzed for the
presence of cross-linked products by immunoblotting. As shown in Fig.
4, cross-linking of the N-terminal region of the Fes
protein resulted in the formation of a new band of approximately 150 kDa, approximately three times the unit molecular mass of this domain
and somewhat lower than anticipated based on the gel-filtration results
that are consistent with at least a pentameric molecule. This
difference may be due to artifactually high electrophoretic mobility of
the cross-linked complex. Identical results were obtained with other
cross-linking reagents with longer or shorter spacer arms and with DSS
concentrations as low as 0.1 mM (data not shown). By
contrast, no cross-linking was observed with the SH2-kinase portion of
Fes under any of these conditions. These results are in good agreement
with the gel-filtration and GST-N-Fes co-precipitation experiments as
well as the N-terminal localization of coiled-coil motifs by the COILS
program (Fig. 2). We also performed cross-linking experiments on the
full-length Fes molecule. A new band of much higher molecular mass was
observed following incubation of Fes with the cross-linking reagent.
Because of the very low electrophoretic mobility of the cross-linked
product, estimation of its molecular mass was not possible (data not
shown).
Suppression of Wild-type Fes Tyrosine Kinase Activity by a Kinase-defective Fes Mutant and the Isolated N-terminal Region
Gel-filtration, co-precipitation, and chemical
cross-linking data presented above demonstrate that Fes is capable of
oligomerization, which may be mediated via coiled-coil oligomerization
motifs localized to the unique N-terminal domain. Previous work from
our laboratory shows that autophosphorylation of Fes is intermolecular
and is required for full activation of the kinase (16, 17). If
oligomerization is required for Fes kinase activation and the mechanism
is intermolecular, then co-expression of Fes with a kinase-inactive
mutant would be predicted to suppress autophosphorylation by forming
oligomers incapable of maximal transphosphorylation. To test this
hypothesis, we expressed full-length Fes as a GST fusion protein as
well as the kinase-inactive form of Fes (Fes-KE) independently in Sf-9 cells. Clarified lysates from the infected cells were mixed together in
various ratios, and the resulting protein complexes were
immunoprecipitated and the extent of GST-Fes autophosphorylation was
measured by addition of [-32P]ATP. As shown in Fig.
5A, the autophosphorylation capacity of Fes
decreased in proportion to the amount of the kinase-inactive mutant
added to the reaction. This result supports the hypothesis that
oligomerization and transphosphorylation are necessary for Fes
autophosphorylation to occur.
To demonstrate that the kinase inhibition shown in Fig. 5A
is dependent upon the Fes N-terminal domain, similar in
vitro kinase assays were performed with GST-Fes following
incubation with either the isolated N-terminal domain or a
kinase-inactive Fes mutant lacking the N-terminal region (N-KE). As
shown in Fig. 5B, the N-terminal region alone was sufficient
to suppress GST-Fes autophosphorylation, and the
N-KE mutant was
without effect.
Data shown in Fig. 5 suggest that GST-Fes and the kinase-inactive
mutant form mixed oligomers. To test this possibility, lysates from
Sf-9 cells co-expressing GST-Fes and Fes-KE, the Fes N-terminal region
or N-KE were incubated with glutathione-agarose. Following incubation and washing, GST-Fes-associated proteins were visualized by
immunoblotting. As shown in Fig. 6, Fes-KE as well as
the isolated N-terminal region readily co-precipitated with GST-Fes,
whereas the
N-KE mutant did not. None of these proteins bound to GST alone. These results are consistent with the conclusion that mixed oligomer formation is required for the inhibition of kinase activity by
the kinase-inactive Fes mutants.
Data presented in this paper demonstrate for the first time that the c-Fes tyrosine kinase is an oligomeric protein and are consistent with a model for Fes activation that requires oligomerization and transphosphorylation. Gel-filtration, chemical cross-linking, and co-precipitation experiments all clearly demonstrate the intrinsic capacity of Fes to form oligomers. Gel-filtration experiments suggest that full-length Fes is capable of forming large oligomers containing as many as five Fes molecules (Fig. 1A). Similarly, gel-filtration of the isolated N-terminal domain of Fes suggests the capacity for formation of at least pentameric structures (Fig. 1C). By contrast, cross-linking studies (Fig. 4) suggest that the isolated Fes N-terminal domain forms homotrimers and are consistent with a previous report for the Fes-related tyrosine kinase, Fer (32). This difference may be due to the artifactually high migration rate of the cross-linked molecule on SDS-PAGE gels. Alternatively, the oligomeric state of Fes may be variable and may be influenced by the presence of other molecules that bind to Fes such as Bcr (5).
Our data clearly indicate that the unique N-terminal region is responsible for the oligomerization of Fes. Computer analysis of this region shows the presence of a heptad repeat pattern of hydrophobic and hydrophilic amino acids typical of coiled-coil oligomerization domains (31). One of these regions (amino acids 128-169) scored with a near absolute probability of being a coiled-coil former. A small FLAG fusion protein containing this putative coiled-coil domain was readily cross-linked by DSS, producing dimers, trimers, and tetramers (data not shown). Other potential coiled-coil-forming sequences were also detected by the COILS program, although with lower probability. The possibility that Fes may contain more than one coiled-coil domain suggests a mode of kinase regulation involving interconversion of a monomeric, inactive form of the protein with the oligomeric, active form of the protein observed here. A precedent for this type of regulation is provided by the heat shock transcription factor, HSF-1 (33). In this case, the latent form of the protein has been proposed to exist as a monomer, in which its multiple coiled-coil domains are occupied via intrastrand interactions that may be stabilized via interactions with HSP70. Activation of the protein by heat stress results in formation of a homotrimer via a triple-stranded coiled-coil structure; this trimeric form of the protein shows greatly enhanced affinity for DNA. In the case of Fes, interaction with a host factor may suppress kinase activity in intact cells by preventing formation of the active oligomer. As with HSF-1, Fes has recently been shown to interact with HSP70 and other molecular chaperones (34).
Oligomerization domains have also been implicated in the biological activation of the c-Abl tyrosine kinase in various human leukemias. The best characterized example is Bcr/Abl, the chimeric oncoprotein that results from the Philadelphia chromosome translocation first associated with chronic myelogenous leukemia (35). The N-terminal, Bcr-derived portion of Bcr/Abl mediates the formation of large oligomeric Bcr/Abl complexes (36, 37). A specific Bcr domain with homology to coiled-coil oligomerization sequences (amino acids 1-63) has been shown to mediate homotetramer formation in vitro (36). Tetramerization of Bcr/Abl via this domain correlates with F-actin binding, leading to association of Bcr/Abl with the actin cytoskeleton. Oligomerization has been proposed to enhance the affinity of Bcr/Abl for F-actin via the Abl C-terminal F-actin binding domain (38). Mutagenesis of the Bcr-derived oligomerization domain impairs both F-actin binding and transforming activity, suggesting that formation of Bcr/Abl oligomers and F-actin localization are required for transformation (38, 39). Another chromosomal translocation associated with myeloid leukemia has been reported recently that activates Abl by a similar mechanism (40). This translocation event fuses c-Abl to the N-terminal region of the TEL transcription factor, resulting in oligomerization, kinase activation, and cytoskeletal localization. Thus, oligomerization may represent an important activating mechanism for both normal and transforming tyrosine kinases of the non-receptor class.