Role for Fes/Fps tyrosine kinase in microtubule nucleation through is Fes/CIP4 homology domain.

We have previously demonstrated that Fes/Fps (Fes) tyrosine kinase is involved in Semaphorin3A-mediated signaling. Here we report a role for Fes tyrosine kinase in microtubule dynamics. A fibrous formation of Fes was observed in a kinase-dependent manner, which associated with microtubules and functionally correlated with microtubule bundling. Microtubule regeneration assays revealed that Fes aggregates colocalized with gamma-tubulin at microtubule nucleation sites in a Fes/CIP4 homology (FCH) domain-dependent manner and that expression of FCH domain-deleted Fes mutants blocked normal centrosome formation. In support of these observations, mouse embryonic fibroblasts derived from Fes-deficient mice displayed an aberrant structure of nucleation and centrosome with unbundling and disoriented filaments of microtubules. Our findings suggest that Fes plays a critical role in microtubule dynamics including microtubule nucleation and bundling through its FCH domain.

The Fes/Fps (Fes) proto-oncogene encodes a structurally unique member of the non-receptor protein-tyrosine kinase (PTK) 1 family (1). The Fes expression was first detected in cells of the myeloid lineage, but recent work (2) has shown that Fes exhibits a more wide spread expression pattern including developing neurons and vascular endothelial cells. Transgenic mice overexpressing v-Fps showed a neurological disorder that included a marked trembling, correlated with the expression of v-Fps in the brain, and a striking bilateral enlargement of the trigeminal nerves (3). In our previous works, Fes was found to be involved in semaphoring-mediated signaling during neural development (4,5), but the Fes function remains largely unknown.
Distinct from Src and other non-receptor PTK, Fes has a long N-terminal unique region containing Fes/CIP4 homology (FCH) domain followed by three coiled-coil domains, a central SH2 domain and a C-terminal kinase domain (6). The FCH domain was first described as a region of homology between Fps/Fes/Fer PTKs and a Cdc42-interacting protein, CIP4 (7). Amino acid sequence homology searches have detected this FCH domain in numerous proteins, many of which are implicated in the regulation of cytoskeletal rearrangements (8). Although the FCH domain of CIP4 was reported to bind to microtubules, the exact role of FCH domain in Fes remains obscure.
In this study we report that Fes plays an important role in microtubule dynamics including microtubule nucleation and bundling through its FCH domain. Implications of Fes in the regulation of microtubules and neural development are briefly discussed.
Immunoprecipitation and Immunoblotting Procedures-Mouse brains or cultured cells were lysed in the lysis buffer (20 mM Tris/HCl, pH 8.0, 1% Nonidet P-40, 150 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin, and 10 M Na 3 VO 4 ) at 4°C. The lysates (1 mg of protein) were clarified by centrifugation at 100,000 ϫ g for 10 min and immunoprecipitated with the appropriate Ab. Immunoprecipitates were washed three times with lysis buffer, once with 10 mM HEPES/NaOH, pH 8.0, containing 0.5 M NaCl and finally with 10 mM HEPES/NaOH, pH 8.0. Immunoprecipitates were boiled with SDS-PAGE sample buffer for 3 min, separated by SDS-PAGE, and transferred to polyvinylidene difluoride membranes (Millipore), followed by detection with the appropriate Ab as described previously (9).
Expression Constructs-Cloning of mouse Fes cDNA was performed as described previously (5). Fes cDNA tagged with FLAG epitope at the N terminus were subcloned into pCMV5 expression vectors. Sequencing of all strands of the cloned reverse transcriptase-PCR fragments were performed using an ABI-310 automated DNA sequencer. Fes kinasenegative mutant was generated by missense mutation K562R as described previously (10). Point mutation of Arg 482 of Fes cDNA in SH2 domain to Lys (R482K) was performed by the site-directed mutagenesis kit from Strategene. A sense primer (5Ј-GGACTTCCTGGTTAAGGAG-AGCCAGGGC-3Ј) was used. For the FCH-domain deletion mutant Fes (amino acid residues 1-95), a sense primer (5Ј-GAAGATCTAACTCGG-GGCCCTTGAGC-3Ј) was used. For FCH and coiled-coil-domain deletion mutant Fes (amino acid residues 452-820), a sense primer (5Ј-G-AAGATCTAAGCCTCTCTATGAGCAGCTG-3Ј) was used. Both Fes and Fes kinase-negative cDNAs were subcloned into pEGFP-C1 (Clontech) for green fluorescence. FCH-domain deletion mutant, FCH, coiled-coildomain deletion mutant, and SH2 mutant were also subcloned into pEGFP-C1.
Cell Culture and Stimulation-COS-7 cells were cultured in Iscove's modified Dulbecco's medium (Sigma) containing 10% fetal bovine serum. Hippocampal cultures were prepared as described previously (11). For mouse embryonic fibroblasts (MEFs), fetuses from embryonic day 12 were cut into small size and homogenized in 0.05% trypsin-PBS solution with a Dounce homogenizer. After trypsin digestion for 1 h at 37°C, the supernatants were collected and centrifuged at 1000 ϫ g for 5 min. The pellet was resuspended in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and plated onto 6-well plates.
Transfection of Plasmid into Cells and Treatment with Chemicals-COS-7 cells and MEFs were cultured in dishes at about 50% confluence. Expression vectors (1 g) were mixed with serum-free medium and transfected using a LipofectAMINE Plus kit (Invitrogen). After 48 h, the medium was changed to serum-free medium for about 4 h, and COS-7 cells were then treated with colcemid (0.8 g/ml) for 1 h at 37°C. After washing twice with PBS, COS-7 cells were incubated with Iscove's modified Dulbecco's medium for 45 min at 37°C to polymerize microtubules.
Immunofluorescence Microscopy-COS-7 cells transiently coexpressing cDNAs or hippocampal cells were fixed with 4% paraformaldehyde in PBS for 10 min, washed twice with PBS, permeabilized with 0.2% Triton X-100 in PBS for 10 min, washed three times with PBS, and blocked with 3% bovine serum albumin in PBS, all at room temperature. For double staining, the cells were incubated with appropriate Abs for 2 h at room temperature and washed three times with 0.2% Triton X-100 in PBS and then with appropriate secondary Ab (Cy3-labeled goat anti-mouse IgG, Alexa Fluor 594 goat anti-rabbit IgG, Alexa Fluor 647 goat anti-mouse IgG) for 30 min. The samples were washed as before, mounted using SlowFade-Light (Molecular Probes), and analyzed using LSM510 META (Carl Zeiss).

Association of Fes with Microtubules and Correlation with
Microtubule Modifications-To investigate the subcellular distribution of Fes, Fes was visualized as green fluorescent protein (GFP)-tagged fusion protein (GFP-Fes) in COS-7 cells. As shown in Fig. 1A, GFP-Fes showed three characteristic expression patterns in COS-7 cells: aggregate, fibrous, and diffuse. On the other hand, GFP-Fes(kϪ) revealed either a diffuse or an aggregate distribution without the fibrous structures. The ratio of each expression pattern was measured in 100 Fes-transfected cells, and each percentage is shown in the panels of Fig.  1A. GFP-Fes revealed a fibrous formation at 30ϳ40% in transfected cells but in a kinase-dependent manner. Since these fibrous Fes structures appeared very similar to microtubules, we examined whether these Fes structures colocalized with microtubules. Expectedly, immunofluorescence analysis using anti-␣-tubulin Ab showed a complete colocalization of Fes with microtubules ( Fig. 1B, top panels). In addition, Fes structures were also immunostained with anti-phoshotyrosine (PY) Ab (Fig. 1B, bottom panels). To confirm the association of Fes with microtubules, immunoprecipitation assay was performed. As shown in Fig. 1C, tyrosine-phosphorylated ␣-tubulin was detected in GFP-Fes immunoprecipitates but not in GFP-Fes(kϪ) immunoprecipitates. To examine whether Fes associates with microtubules in developing brain, whole lysate of neonatal rat brain was immunoprecipitated with anti-Fes Ab or normal rabbit serum as control. As shown in Fig. 1C, lower panels, tubulin was detected in Fes immunoprecipitates but not in control immunoprecipitates. In addition, colocalization of endogenous Fes with microtubules was detected in neuronal cells (see Supplemental Fig. S1). These results demonstrated that Fes colocalized and associated with microtubules in a kinase activity-dependent manner and that tubulin might be phosphorylated by Fes.
The immunofluorescence signals from microtubules seemed to be enhanced by colocalization with Fes structures. To confirm this observation, the effect of Fes expression on microtubule bundling activity was compared by densitometric analysis of immunofluorescence signals from normal-and Fes-colocalized microtubules immunostained with anti-␣-tubulin Ab. As shown in Fig. 1D, immunofluorescence signals from microtubules colocalized with Fes were significantly enhanced, and the space between microtubules became wider than that of normal microtubules. All Fes fibrous structures observed in other cells exhibited similar results. In control cells, these strong micro- Immunofluorescence analysis of GFP-Fes (wild-type (wt)) or GFP-Fes kinase-negative mutant (kϪ) showed three distinct subcellular distributions including an aggregate and a fibrous and a diffuse pattern. The proportion of each pattern was calculated from 100 Fes-expressing cells. B, colocalization of Fes with microtubules. COS-7 cells expressing GFP-Fes (green) were labeled with anti-␣-tubulin (red) or anti-phosphotyrosine (PY) Abs (red). C, association of Fes with microtubules. COS-7 cells expressing GFP-Fes(wt) or GFP-Fes(kϪ) were immunoprecipitated with anti-GFP Ab, and the immunoprecipitates were immunoblotted with anti-GFP, anti-phosphotyrosine, or anti-␣-tubulin Ab (upper panels). Whole brain lysates were immunoprecipitated with anti-Fes Ab or normal rabbit serum as control (NRS), and immunoprecipitates were immunoblotted with anti-␣-tubulin or anti-Fes Abs (lower panels). D, COS-7 cells expressing GFP-Fes (green) and control cells were labeled with anti-␣-tubulin (red) Ab (upper panels). Microtubule bundling activity by Fes was determined by densitometric analysis between Fes-colocalized (panel 1) and not colocalized (panel 2) microtubules immunostained with anti-␣-tubulin Ab (middle and bottom panels). The fluorescence intensities of GFP-Fes (green) and ␣-tubulin (red) immunostaining including control cells were analyzed by using NIH Image Analysis software. E, partial colocalization of Fes with acetylated or detyrosinated microtubules. COS-7 cells expressing GFP-Fes (green) were labeled with anti-acetylated or anti-detyrosinated tubulin Abs (red). Each bar is 20 m.
tubule bundling was not observed (Fig. 1D). Therefore, we concluded that Fes was actively involved in the processes of microtubule bundling.
We next examined whether Fes was involved in the posttranslational modifications of tubulins such as acetylation and detyrosination. Acetylated tubulins are generally found in stable microtubules. As shown in Fig. 1E, fibrous Fes structures were partially colocalized with acetylated and detyrosinated tubulins. Similar results were obtained in almost all other Fes-expressing cells. To examine whether Fes associates with acetylated tubulin, immunoprecipitation assay was performed. GFP-Fes, but not control GFP, was found to be associated with acetylated tubulin (Fig. 2B). However, this result showed a weak coimmunoprecipitation of Fes with acetylated tubulin, as with ␣-tubulin in Fig. 1C, suggesting that this interaction may be partial and limited. At present, it is uncertain whether Fes was actively involved in these posttranslational modifications.
Role for Fes in Microtubule Nucleation and Effects of Various Fes Mutants on Microtubule Dynamics-To understand the role for Fes in microtubule dynamics, COS-7 cells expressing GFP-Fes were treated with colcemid, and the relationship between Fes and microtubule dynamics was examined during the time-dependent process of microtubule regeneration. When microtubules were depolymerized by colcemid treatment Fes existed in an aggregate form without any microtubule formation (not shown). Upon removal of colcemid to start microtubule nucleation and polymerization, microtubule nucleation of centrosomes was observed at 45 min after removal of colcemid and Fes aggregations were found to be colocalized with microtubule nucleation sites ( Fig. 2A, upper panels). Thereafter, fibrous Fes structures which colocalized with newly polymerized microtubules were detected after 60 min ( Fig. 2A, lower panels). To examine whether Fes associates with ␥-tubulin, immunoprecipitation assay was performed. GFP-Fes, but not control GFP, was found to be associated with ␥-tubulin, although at a very weak level of association (Fig. 2B). Equal amount of immunoprecipitated ␥-tubulin or acetylated tubulin was confirmed by immunoblot analysis using anti-␥-tubulin Ab (not shown). To confirm the localization of Fes to microtubule nucleation sites, COS-7 cells expressing GFP-Fes were double labeled with anti-␣-tubulin and anti-␥-tubulin Abs. As shown in Fig. 2C, Fes aggregates were detected at microtubule nucleation sites with ␥-tubulin aggregations.
To investigate which domain of Fes was required for its localization to microtubule nucleation, effects of several deletion mutants on microtubule nucleation were examined. As shown in Fig. 2D, GFP-FCH domain fusion protein was able to localize to microtubule nucleation sites (Fig. 2D, upper panels). Interestingly, in some cells, aggregations of GFP-FCH domain were colocalized with multiple ␥-tubulin aggregations (Fig. 2D, bottom panels), suggesting the close relationship between Fes FCH domain and ␥-tubulin. On the other hand, FCH domaindeleted mutants never colocalized with microtubule nucleation sites. Rather, this mutant blocked normal formation of microtubule nucleation and centrosome. Surprisingly, a fibrous formation of FCH domain deleted Fes mutant was detected, but this mutant Fes showed little colocalization with microtubules (Fig. 2E). Furthermore, expression of FCH domain-deleted mutants blocked normal ␥-tubulin localization at microtubule nucleation sites (Fig. 2F). How could FCH domain-deleted mutant lead to the collapse of microtubule structure? One explanation is that in addition to interfering with Fes-mediated signaling, the FCH domain-deleted mutant is also disrupting Fer signaling because Fes and Fer may perform redundant signaling functions in cells. Another explanation is that the FCH domain-deleted mutant could still bind to putative Fes substrates that were involved in microtubule formation, but without the FCH domain these substrates could not distribute to their normal subcellular location such as colocalization to microtu- bules or nucleation. Therefore, it is important to determine how this FCH domain might mediate functional modification of microtubules and which molecules interact with this domain.
Overexpression of Fes FCH protein also blocked normal formation of microtubule nucleation and centrosome without fibrous structures. In this regard, we should consider that the involvement of FCH domains of CIP4 and other FCH domain containing proteins with microtubules is of relevance to the Fes FCH overexpression results. Indeed, a previous work (12) has shown that truncation of Fes which retains the N-terminal domain, but lacks both the SH2 and the tyrosine kinase domain via gene targeting results in embryonic lethality. Thus, mice expressing Fes N-terminal domain showed far much severe phenotype when compared with Fes-deficient mice. This result is highly suggestive that FCH domain-containing proteins also may be involved in microtubule dynamics such as nucleation.
Coiled-coil domains-deleted Fes mutant and Fes SH2 mutant revealed either a diffuse or an aggregate distribution but no fibrous formation, same as Fes(kϪ) mutant (see Supplemental  Fig. S2). A striking feature of Fes/Fer PTKs is their ability to form oligomers which is believed to be mediated by the coiled coil domains (13). In addition, a previous work (14) has indicated that the SH2 domain of Fes can bind to autophosphorylation sites. Therefore, intramolecular or intermolecular interactions may be required for fibrous formation of Fes. These results are briefly summarized in Table I.
Aberrant Structure of Centrosome in Fes-deficient Fibroblasts-To determine the physiological relevance of Fes in microtubule dynamics, we compared MEFs derived from wildtype and Fes-deficient mice. Light microscopic analysis indicated that Fes-deficient cells were smaller than wild-type fibroblasts (not shown). These FesϪ/Ϫ and Fesϩ/ϩ MEFs were labeled with anti-acetylated tubulin Ab, and the structures of centrosome were compared. In contrast to wild-type, Fes-deficient cells displayed an aberrant centrosome structure with unbundling and disoriented filaments of acetylated microtubules. Furthermore, in Fes-deficient cells, the immunofluorescence signals for acetylated microtubules were significantly decreased and microtubule bundling was also severely inhibited. We examined microtubule nucleation in these cells further by ␥-tubulin immunostaining. Normal microtubule nucleation colocalized with ␥-tubulin was observed in control cells, but microtubule nucleation and centrosome were not normally developed in Fes-deficient cells (Fig. 3A, right panels). These phenotypes were commonly observed in almost all MEFs derived from three independent Fes-deficient mice. Furthermore, the transfection of FLAG-tagged Fes into Fes-deficient cell restored the normal structure of centrosome with oriented and bundled acetylated microtubules (Fig. 3B). This experiment was performed independently three times, and similar results were obtained. Thus, these results suggested that Fes was required for normal microtubule formation.
We examined whether these abnormal microtubule structures observed in Fes-deficient cells were due to an indirect effect such as cell adhesion. As shown in Fig. 3C, analysis using focal adhesion and F-actin staining showed no significant difference between wild-type and Fes-deficient cells. In addition, confocal laser microscopy scanning along the z axis revealed similar morphology between wild-type and Fes-deficient cells. These results suggested that abnormal microtubule formation of Fes-deficient cells was not due to cell adhesion. However, these findings cannot completely deny the indirect effect.
Although an abnormal microtubule structure was detected in cultured Fes-deficient cells, Fes-deficient mice have been reported to be born and developed normally without any obvious defects (15). Biochemical microtubule assembly analysis using brain lysates of wild-type and Fes-deficient mice did not show  3. Comparative analysis of mouse embryonic fibroblasts derived from wild-type and Fes-deficient mice. A, Fes-deficient cells display an aberrant structure of centrosome with disoriented fragments of acetylated microtubules. MEFs were labeled with anti-acetylated tubulin Ab (white and green) or anti-␥-tubulin Ab (red). These three MEFs were obtained from three independent Fes-deficient mice. Arrowheads indicate microtubule nucleation sites colocalized with ␥-tubulin. B, expression of FLAG-tagged Fes into Fes-deficient cells restored normal structure of centrosome with oriented and bundled acetylated microtubules. MEFs were labeled with anti-FLAG Ab (white and green) and anti-acetylated tubulin Ab (red). This experiment was performed independently at three times, and similar results were obtained. Each bar is 20 m. C, comparison of cell adhesion between wild type and Fes-deficient cells. MEFs were labeled with anti-focal adhesion kinase (FAK) Ab (green) and phalloidin staining (red). Cell morphology analysis along the z axis is shown in the bottom panels. Each horizontal or vertical bar is 20 or 5 m, respectively. a significant change in microtubule polymerization activity. Since Fer is known as another member of Fes family PTKs, one possibility is that Fer may play a redundant role in microtubule dynamics during development of Fes-deficient mice. Generation of Fes and Fer double deficient mice will answer that question and reveal the physiological importance of this PTK family.
Role for Fes in Neural Development-Transgenic mice overexpressing v-Fps showed a neurological disorder that included a striking bilateral enlargement of the trigeminal nerves (3). This result suggested an important role for Fes in neural differentiation and growth. In our previous works, Fes was found to be associated with CRMP/CRAM and involved in semaphoring-mediated signaling during neural development (4,5). It has been recently shown that CRMP-2 binds to tubulin heterodimers and promotes microtubule assembly (16). Thus, both Fes and CRMPs are closely involved in microtubule dynamics. A previous report (17) suggested that conditions that stabilize microtubules could lead to bundle formation and allow microtubule assembly by a mechanism different from that employed by microtubule-associated proteins. This means that additional mechanisms besides the action of tau on tubulin exist to organize microtubules in the axon. One attractive hypothesis is that Fes and CRMP/CRAM may play a role in microtubule nucleation, serving as microtubule organizing center during neuronal development.
In summary, our findings demonstrated that Fes is involved in microtubule nucleation through FCH domain and that FCH domain is required for normal formation of microtubule nucleation and centrosome. Further studies will be needed to characterize the mechanism of microtubule dynamics mediated by Fes.