Role of Focal Adhesion Kinase Ser-732 Phosphorylation in Centrosome Function during Mitosis*

Focal adhesion kinase (FAK) is the major cytoplasmic tyrosine kinase in focal adhesions and a critical mediator of integrin signaling in a variety of cells, including endothelial cells (ECs). Here we describe a new function for FAK in the regulation of centrosome functions in a Ser-732 phosphorylation-dependent manner during mitosis. Deletion of FAK in primary ECs causes increases in centrosome numbers, multipolar and disorganized spindles, and unaligned chromosomes during mitosis. Re-expression of wild-type FAK, but not S732A mutant, rescued these mitotic defects, suggesting a role for Ser-732 phosphorylation in the regulation of centrosomal functions. Consistent with this possibility, Ser-732-phosphorylated FAK was found to co-localize in centrosomes in mitotic cells. FAK also associated with cytoplasmic dynein in a Ser-732 phosphorylation-dependent manner. Further analysis in FAK-null primary ECs showed that S732A mutant could rescue EC migration but not proliferation or tubulogenesis in vitro. Last, we showed that deletion of FAK in ECs reduced tumor angiogenesis in vivo, which could be restored by re-expression of wild-type FAK but not S732A mutant. Together, these studies demonstrated a novel role for Ser-732 phosphorylation of FAK in the regulation of centrosome function during mitosis, which may contribute to EC proliferation and angiogenesis.

Focal adhesion kinase (FAK) 2 is a cytoplasmic tyrosine kinase that is a major mediator of signal transduction by integrins and also participates in signaling by other cell surface receptors in a variety of cells, including endothelial cells (ECs) (1)(2)(3)(4). In most adherent cells FAK is activated upon integrin-mediated cell adhesion to extracellular matrix proteins through disruption of an intramolecular inhibitory interaction between its amino-terminal FERM domain and the kinase domain (5,6). Once it is activated, FAK undergoes autophosphorylation at Tyr-397, which creates a binding site for several SH2 (Src homology 2) domain-containing proteins including Src family kinases (1,2). The associated Src could then phosphorylate additional sites on FAK, including residues Tyr-576 and Tyr-577 in its activation loop, to further activate FAK and Tyr-925 to create binding sites for the adaptor molecule Grb2 (7)(8)(9). FAK also functions as a scaffold to mediate Src family kinase phosphorylation of several proteins, including paxillin (10,11), p130cas (12,13), and endophilin A2 (14), which bound to the carboxyl-terminal region of FAK. This cascade of phosphorylation events and protein-protein interactions has been shown to trigger several signaling pathways in the regulation of a variety of cellular functions in different cells.
Besides these well characterized tyrosine phosphorylations, recent studies have identified FAK phosphorylation on several serine residues (15,16). In the post-mitotic neurons, Ser-732 has been shown to be phosphorylated by Cdk5, which plays an important role in microtubule organization and proper nuclear movement during neuronal migration (17,18). Indeed, Ser-732-phosphorylated FAK is enriched in centrosome-associated microtubule fork that abuts the nucleus and a perinuclear region around the centrosome, consistent with its regulation of these functions in neurons. Ser-732 of FAK has also been shown to be phosphorylated by Rho-dependent kinase (ROCK) in ECs, which has been suggested to play a role in vascular endothelial growth factor-stimulated EC migration (19). In addition to Ser-732, Ser-722, Ser-843, and Ser-910 in the carboxyl-terminal domain of FAK have also been found to be phosphorylated and regulate cell spreading and migration in recent studies (16, 20 -24). Despite these findings, our understanding of serine phosphorylation of FAK is very limited in contrast to the wealth of information on the regulation and function of tyrosine phosphorylation of FAK. In particular, it is not clear whether and how serine phosphorylation is involved in the regulation of cell cycle progression and proliferation by FAK.
Focal adhesion localization of FAK in adherent cells is essential for its functions in the regulation of cell migration as well as proliferation (1,2). During mitosis, however, focal adhesion complexes dissociate as cells round up and detach from extracellular matrix. Interestingly, serine phosphorylation of FAK is increased during mitosis, and this has been suggested to cause FAK dissociation from p130Cas and Src to inactivate signaling at focal adhesions (25), although the relevant sites of phosphorylation were not mapped in this study. It is not known whether FAK is localized to any specific subcellular structures and/or plays a role in mitosis and whether these are regulated by serine phosphorylation of FAK in mitotic cells.
Consistent with its critical importance in the regulation of various cellular functions, deletion of FAK gene leads to early embryonic lethality at embryonic day 8.5 (E8.5) due to defects in the axial mesodermal tissues including the cardiovascular system with incomplete development of both the blood vessels and the heart (26). Using a conditional mouse KO approach, we and others have recently shown a role of FAK in vascular angiogenesis through its regulation of multiple functions of ECs including their survival, proliferation, migration, and tubulogenesis (27)(28)(29). The availability of the floxed FAK mice and ECs isolated from these mice also allowed us to investigate FAK signaling pathways involved in the regulation of EC functions and angiogenesis in vivo by a reconstitution strategy, where endogenous FAK is deleted via recombinant adenoviruses encoding Cre followed by re-expression of FAK or its various mutants in ECs both in vitro and in vivo. In this study we present data showing a novel function of FAK in the regulation of centrosomal functions in a Ser-732 phosphorylation-dependent manner in ECs during mitosis, which plays a role in the regulation of EC proliferation and tubulogenesis in vitro and tumor angiogenesis in vivo.
Isolation and Infection of ECs-ECs were isolated from 4 -6week-old homozygous FAK floxed mice using the magnetic bead (Dyanbead M-450; Dynal Corp.) purification protocol with rat anti-mouse PECAM-1 (BD Biosciences), as described previously (27,30,31). EC population was ϳ90% pure as determined by anti-PECAM-1 staining. Isolated ECs were infected at a multiplicity of infection of 100 with Ad-lacZ or Ad-Cre. To increase efficiency, a second infection was performed after 9 -12 h and incubated for 48 h. For the rescue experiments, cells infected with Ad-Cre were re-infected with recombinant adenoviruses encoding FAK, its mutants, or Ad-GFP 2 days after infection of Ad-Cre at a multiplicity of infection of 100. No detectable cell toxicity was observed.
Flow Cytometry Analysis-ECs were fixed with 70% ice-cold ethanol at 4°C for more than 2 h. After fixation cells were stained with 50 g/ml propidium iodide (Sigma-Aldrich) with 100 g/ml RNase A in phosphate-buffered saline containing 0.1% Triton X-100. Flow cytometry analysis was performed by a BD Biosciences BD-LSR II flow cytometer.
Immunoprecipitation and Western Blotting-Immunoprecipitation and Western blotting analysis was performed as described previously (32). BrdUrd Incorporation Assay-ECs were serum-starved for 18 h to arrest the cells in G 0 . A BrdUrd incorporation assay was performed as described previously (27). Three independent experiments were performed, and the percentage of cells positive for BrdUrd was quantified using a microscope (model BX41; Olympus) with UplanF1 ϫ 10/0.3 objective lens at room temperature. Approximately 100 cells were examined for each condition in each independent experiment.
Wound Closure Cell Migration Assay-ECs were plated on gelatin-coated dishes (60 mm) and stimulated with 50 ng/ml vascular endothelial growth factor and then subjected to assays. Wound closure assays were performed as described previously (27,33).
Tube Formation Assay-ECs were plated on a layer of Matrigel (growth factor-reduced, BD Biosciences) and allowed to form a tubular structure as described previously (27). Tubulogenesis in each condition was examined on a microscope (model IX70; Olympus) with a UplanF1 ϫ 10/0.3 objective lens and photographed with a progressive 3CCD camera (Sony) at room temperature. The length and branch points were determined using Image-Pro Plus, Version 3.0.00.00 as described previously (34).
Matrigel Plug Assays-Matrigel (growth factor-reduced, BD Biosciences) was supplemented with 5 ϫ 10 8 plaque-forming units/ml recombinant adenoviruses and 5 ϫ 10 5 B16F10 melanoma cells in a final volume of 0.5 ml. Matrigel mixture was then injected subcutaneously into the flank region of 8-weekold floxed FAK mice. Mice were sacrificed 10 days after injection, and Matrigel weights were determined. Vascularization in Matrigel plugs was visualized by immunohistological examination using anti-PECAM-1 antibody (M-20, 1:200 dilution; Santa Cruz Biotechnology, Inc.) as described previously (27). They were then examined under a microscope (model BX41; Olympus) with an UplanF1 ϫ 10/0.3 objective lens at room temperature, and the images were captured using a camera (model DP70; Olympus) with DP Controller, Version 1.2.1.108. Five representative images were obtained from each Matrigel plug, and vessel density was quantified using Image-Pro Plus, Version 3.0.00.00.

Deletion of FAK Causes Spindle and Centrosomal Abnormalities in Mitotic
ECs-Previous studies of EC-specific FAK knock-out mice and isolated ECs showed a role of FAK in angiogenesis and vascular development due to its essential role in the regulation of EC functions, including proliferation, migration, cell survival, and tubulogenesis (27). To further investigate mechanisms of FAK in the regulation of EC proliferation, we performed flow cytometric analyses to get more detailed information on the cell cycle profile in FAK Ϫ/Ϫ ECs. Primary ECs were isolated from floxed FAK mice and were infected by a recombinant adenovirus encoding Cre recombinase (Ad-Cre) to delete endogenous FAK or by a control recombinant adenovirus encoding lacZ to produce FAK Ϫ/Ϫ ECs or control FAK ϩ/ϩ ECs, respectively, as described previously (27). Analysis of these cells by flow cytometry showed an altered cell cycle profile for FAK Ϫ/Ϫ ECs compared with the control FAK ϩ/ϩ ECs (Fig. 1A). In consistent with our previous results showing an increased apoptosis in FAK Ϫ/Ϫ ECs, a significant increase in SubG 1 population was found for these cells compared with FAK ϩ/ϩ ECs (from about 10 to 22%). We also found a significantly increased G 2 /M population in FAK Ϫ/Ϫ ECs compared with FAK ϩ/ϩ ECs (from about 20 to 32%), suggesting a possibly increased mitotic arrest upon FAK deletion in ECs.
To investigate the mechanisms of mitotic abnormalities in FAK Ϫ/Ϫ ECs, we examined the effect of FAK deletion on mitotic spindle organization and chromosome alignment and segregation in these cells. As shown in Fig. 1B, normal mitotic spindles and chromosome alignment were detected in FAK ϩ/ϩ ECs during mitosis by staining for ␣-tubulin and DNA, respectively (a, f, and k). In contrast, the analysis of FAK Ϫ/Ϫ ECs revealed various defects including multiple and randomly positioned spindles (bd), loosely congregated chromosomes (g-i), and unaligned chromosomes (i and n, arrowheads). In some cells anaphase proceeded with an unattached chromosome (j and o, arrows). Because centrosomes are the primary microtubule organization center and play an essential role in mitotic spindle organization and chromosome segregation, we next evaluated possible defects in centrosome organizations in FAK Ϫ/Ϫ ECs by immunostaining for ␥-tubulin, a centrosome marker. Fig. 1C shows that whereas FAK ϩ/ϩ ECs have a typical staining of two centrosomes on the opposite sides of the condensed chromosomes during mitosis (a, f, and k), deletion of FAK in ECs (FAK Ϫ/Ϫ ECs) caused various centrosomal defects in the number, size, and position of centrosomes (be), which are associated with abnormal chromosome condensation in metaphase (g-i) and segregation in anaphase (j). Quantitation of ϳ200 mitotic cells for each group showed centrosomal defects in about 60% of FAK Ϫ/Ϫ ECs compared with less than 10% of the control FAK ϩ/ϩ ECs (Fig. 2B). Together, these results suggest that FAK plays a role in the regulation of mitotic spindle assembly, chromosome alignment and centrosome integrity during mitosis and that the deregulation of these functions caused by the deletion of FAK may be responsible for mitotic arrest in FAK Ϫ/Ϫ ECs.
Ser-732 Phosphorylation of FAK Is Required for Its Regulation of Centrosome Function during Mitosis in Primary ECs-To investigate the mechanisms of FAK regulation of centrosome function during mitosis, we generated recombinant adenoviruses encoding several FAK mutants and analyzed their ability to rescue the mitotic defects in ECs upon deletion of endogenous FAK. Primary ECs isolated from floxed FAK mice were infected by Ad-Cre to delete endogenous FAK followed by infection with recombinant adenoviruses encoding FAK (Ad-FAK), kinase-defective (Ad-KD), Tyr-397 to Phe (Ad-Y397F), Pro-712 and Pro-715 to Ala (Ad-P712A/P715A), or Ser-732 to Ala (Ad-S732A) mutant. As expected, infection of Ad-Cre, but not Ad-LacZ, resulted in the deletion of FAK (see Fig. 1A, right panel). Re-infection of FAK Ϫ/Ϫ ECs with recombinant adenoviruses encoding FAK or its mutants led to expression of exogenous FAK and mutants to comparable levels in these cells ( Fig.  2A). As expected, restoration of FAK expression in FAK Ϫ/Ϫ ECs significantly rescued the centrosomal abnormalities caused by deletion of endogenous FAK (Fig. 2B).
Consistent with previous studies on the critical roles of Tyr-397 in FAK downstream signaling pathways initiated by autophosphorylation of this site, re-expression of either FAK Y397F or KD mutant did not rescue the centrosomal defects (Fig. 2B). In contrast to well characterized tyrosine phosphorylation of FAK, the role of serine phosphorylation of FAK is relatively less investigated, although several serine residues of FAK have also been shown to be phosphorylated (15,16). In particular, phosphorylation of Ser-732 in FAK by Cdk5 has been shown to play a role in nuclear translocation during neuronal migration through regulation of microtubule networks (17). As the centrosome-associated microtubule structure, the mitotic spindle, plays a crucial role during mitosis in non-neuronal cells, we examined the potential role of Ser-732 phosphorylation in FAK regulation of centrosome function by analysis of FAK S732A mutant in the rescue experiments. We found that re-expression of S732A mutant did not rescue centrosomal abnormalities in FAK Ϫ/Ϫ ECs (Fig. 2B). Analysis of another FAK mutant, P712A/ P715A, which is deficient in binding to p130Cas, rescued centrosomal defects in FAK Ϫ/Ϫ ECs to a comparable level as the wild-type FAK (Fig. 2B), suggesting that FAK signaling through p130Cas is not involved in the regulation of centrosome function. Together, these mutational analyses suggest a novel role of Ser-732 phosphorylation in mediating FAK regulation of centrosome functions during mitosis.
Localization of Ser-732-phosphorylated FAK in Centrosomes during Mitosis-FAK is localized in focal adhesions in adherent cells, which are disassembled during mitosis. Previous studies have shown an increase of serine phosphorylation of FAK concomitant with cell rounding-up and disassembly of focal adhe-  ECs, panel A) or Ad-Cre (FAK Ϫ/Ϫ ECs, panel B) and then co-stained with Ser-732 phosphospecific FAK antibody (green) and ␥-tubulin antibody (red), as indicated. Representative images show the co-localization of Ser-732phosphorylated FAK with ␥-tubulin at centrosome in the different phases during mitosis in FAK ϩ/ϩ ECs (A) but not in FAK Ϫ/Ϫ ECs (B and data not shown). Chromosomes were revealed by Hoechst staining (blue). C, murine embryonic fibroblasts (MEF; top panels) and Cos-7 (bottom panels) cells were processed for immunofluorescence staining with Ser-732-phosphospecific FAK antibody (green) and ␥-tubulin antibody (red) as described in A. D, HeLa cells were co-stained with Ser-732-phosphospecific FAK antibody (green) and ␣-tubulin antibody (red), as indicated. Representative metaphase cells were shown.
sions in mitotic cells (25). It is not clear, however, whether serine-phosphorylated FAK are evenly distributed in the cytoplasm or are localized to particular subcellular structures in mitotic cells. In light of the above observation suggesting a potential role of Ser-732 phosphorylation of FAK in the integrity of centrosomes, we examined the possibility of a centrosomal localization of Ser-732-phosphorylated FAK in mitotic ECs. Primary FAK ϩ/ϩ ECs at various phases of mitosis were subjected to double-label immunofluorescent staining with antibodies against phospho-Ser-732 of FAK (Ser(P)-732) and the centrosomal marker ␥-tubulin. As shown in Fig. 3A, Ser-732-phosphorylated FAK was detected in the centrosomes throughout mitosis. The lack of staining in FAK Ϫ/Ϫ ECs by anti-Ser(P)-732 confirmed the specificity of the antibody against FAK Ser(P)-732 (Fig. 3B). Furthermore, localization of Ser-732-phosphorylated FAK in the centrosomes was also detected in several other cell types, including murine embryonic fibroblasts, COS7 cells (Fig. 3C), and HeLa cells (Fig. 3D). These results suggest that FAK may regulate centrosome functions through acting on some components of centrosomes directly in a Ser-732 phosphorylation-dependent manner during mitosis.
Ser-732 Phosphorylation-dependent Association of FAK with Cytoplasmic Dynein-To explore potential FAK targets, we examined various proteins localized in centrosomes for their potential association with FAK in a Ser-732 phosphorylation-dependent manner. FAK Ϫ/Ϫ ECs were infected with Ad-FAK, Ad-S732A, or Ad-GFP as a control, and lysates from these cells were immunoprecipitated by anti-Myc for the Myc-tagged FAK and S732A mutant and their associated proteins. Analysis of the immunoprecipitates by anti-dynein showed that it was associated with wild-type FAK but not S732A mutant (Fig.  4A). Previous studies suggested that Ser-732 of FAK can be phosphorylated by Cdk5 and ROCK1 in different cells (17,19). We, therefore, examined the effect of down-regulation of Cdk5 and ROCK1 on the association of FAK with cytoplasmic dynein. Lysates were prepared from 293T cells that had been transfected with vectors encoding Cdk5 shRNA, ROCK1 shRNA, or the vector alone control. Fig. 4B shows that interaction of FAK with cytoplasmic dynein was reduced by knockdown of expression of Cdk5, but not ROCK1, when compared with cells treated with control shRNA. Last, colocalization of Ser-732-phosphorylated FAK with dynein at centrosomes was also confirmed by double-label immunofluorescent staining of mitotic cells (Fig.  4C). Together, these results suggest that Cdk5-dependent Ser-732 phosphorylation of FAK and its binding to cytoplasmic dynein may play a role in the regulation of centrosome function during mitosis.

Ser-732 Phosphorylation of FAK Is Required for Cell Proliferation and Tubulogenesis in Primary ECs-Our previous studies
showed that inactivation of FAK in primary ECs caused increased apoptosis, reduced proliferation and migration, and reduced capillary formation on Matrigel, suggesting essential function of FAK in the regulation of multiple EC activities (27). The inability of S732A mutant to rescue the centrosomal defects in FAK Ϫ/Ϫ ECs raised the possibility that this mutant will not be able to rescue the deficiency in proliferation of these cells, consistent with a role for Ser-732 phosphorylation of FAK in the regulation of EC proliferation. To test such a possibility, FAK Ϫ/Ϫ ECs were infected with Ad-FAK or Ad-S732A, and the re-expressions of FAK and its mutant to comparable levels were verified in these cells (Fig. 5A). The cells were then subjected to analysis for proliferation by BrdUrd incorporation assays. As expected, re-expression of FAK in FAK Ϫ/Ϫ ECs rescued their deficiency in proliferation compared with those cells infected by Ad-GFP control virus. In contrast, however, re-expression of S732A mutant did not restore the reduced proliferation of FAK Ϫ/Ϫ ECs (Fig. 5B), suggesting a role of Ser-732 phosphorylation for FAK regulation of cell cycle progression.
We next examined whether Ser-732 phosphorylation is required for FAK stimulation of EC migration and/or tubulogenesis by analysis of S732A mutant in FAK Ϫ/Ϫ ECs. As shown in Fig. 5C, re-expression of S732A mutant in FAK Ϫ/Ϫ ECs restored migration of these cells to a comparable level as wildtype FAK, suggesting that Ser-732 phosphorylation of FAK is not necessary for regulation of EC migration by FAK. In contrast, re-expression of S732A in FAK Ϫ/Ϫ ECs was not able to rescue tubulogenesis deficiency of FAK Ϫ/Ϫ ECs (Fig. 5D). Together, these results demonstrate that Ser-732 phosphorylation-dependent functions of centrosomal-localized FAK in mitotic cells is important for FAK regulation of proliferation and tubulogenesis of primary ECs.
Role of FAK Ser-732 Phosphorylation in Tumor Angiogenesis-Previous studies using EC-specific FAK knock-out mice indicated a role for FAK in embryonic angiogenesis in vivo (27,28). However, the embryonic lethality of these mice prevented their usage for analysis of a role for FAK in angiogenesis in adult mice. A mouse model with inducible EC-specific deletion of FAK in adult mice was generated very recently and used to demonstrate a role for FAK and its related kinase Pyk2 in angiogenesis using inhibitors for FAK and Pyk2 (29). Because of the compensatory up-regulation of Pyk2 in these mice, this induc-ible EC-specific FAK knock-out mouse model could not be used to assess the specific role of FAK and its mutants. Therefore, to examine a potential role of Ser-732 phosphorylation of FAK in angiogenesis in vivo, we developed a tumor angiogenesis assay using floxed FAK mice in which Ad-Cre is included in Matrigel to induce Cre-mediated deletion of endogenous FAK in ECs migrating into the Matrigel plugs in response to angiogenic stimulation of Matrigel-containing tumor cells. Floxed FAK mice were subcutaneously injected with Matrigels containing Ad-Cre or Ad-lacZ as a control as well as B16F10 melanoma cells to induce angiogenesis, as described under "Experimental Procedures." Ten days later the Matrigel plugs containing tumors were dissected and analyzed. As shown in Fig. 6A, Matrigel plugs from mice with Ad-Cre infection were smaller and appeared less red when compared with those from mice infected by the Ad-LacZ control. Quantification of multiple samples showed a significant reduction in weight (Fig.  6B) as well as size (data not shown) for the plugs containing Ad-Cre than those with Ad-LacZ control. The plugs were also sectioned and subjected to immunohistochemical analysis using anti-PECAM-1 antibody to detect blood vessels. Consistent with the reduced tumor growth, we found a significantly reduced density of blood vessels in the Matrigel plugs containing Ad-Cre than those with Ad-LacZ control (Fig. 6C). Together, these results suggest that Ad-Cre-mediated deletion of the floxed FAK in ECs significantly reduced tumor angiogenesis and growth in vivo.
We then assessed the role of Ser-732 phosphorylation of FAK in tumor angiogenesis by re-expression of FAK or S732A mutant in FAK Ϫ/Ϫ ECs using this floxed FAK mouse model. Ad-FAK or Ad-S732A was included in Matrigel containing Ad-Cre as well as B16F10 melanoma cells injected into floxed FAK mice. As shown in Fig. 6, re-expression of wild-type FAK restored tumor growth as well as angiogenesis in Matrigel, as expected. In contrast, re-expression of S732A mutant did not rescue the decreased tumor growth or angiogenesis caused by deletion of endogenous FAK in ECs. Therefore, consistent with results from in vitro analysis, Ser-732 phosphorylation of FAK is required for angiogenesis in vivo due to its role in the regulation of centrosome functions and proliferation in ECs.

DISCUSSION
As the principal cytoplasmic tyrosine kinase located in focal adhesions, FAK is well established as a major mediator of signaling cascades triggered by clustering of integrins in these sites in the regulation of various cellular functions, including G 1 -S transition in cell cycle (1,2). In this report, we present data suggesting a novel function for FAK in the regulation of centrosome integrity, spindle pole formation, and chromosome segregation during mitosis in primary ECs. Besides being required for G 1 -S transition, cell adhesion to extracellular matrix was known to control other phases of cell cycle such as cytokinesis (35)(36)(37). Indeed, a recent study showed that inhibition of integrin function disrupted centrosome functions, spindle assem-bly, and cytokinesis in mitotic cells (38). Thus, FAK may play a role in both focal adhesions and centrosomes during different phases of cell cycle progression. In this regard it is interesting to note that several other focal adhesion proteins, including HEF1 (39), paxillin (40), zyxin (41), and ILK (42) have been shown to localize and function in centrosomes. Given the known connections with FAK for at least some of these molecules (1)(2)(3)(4), FAK may work together with these other focal adhesion proteins to provide a mechanistic link for the control of mitotic events in the nucleus by integrins localized on the plasma membrane.
Centrosomes are composed of two paired centrioles surrounded by pericentriolar material, which comprised hundreds of structural and signaling proteins. They undergo structural modifications during cell cycle, including duplication, maturation, and separation, which are tightly coordinated with the chromosome duplication and segregation (43). The abnormal centrosomal phenotype in FAK Ϫ/Ϫ ECs could result from deregulation of centrosomal duplication, incomplete centrosome separation, or loss of cohesion in mitotic centrosomes, resulting in premature splitting of mother and daughter centrioles. Centrosome number and splitting are regulated by protein kinases, including Cdk2/cyclinE, the Polo-like kinases, Aurora-A, and Nek2 (44 -47). Two focal adhesion-associated proteins, HEF1 and ILK, may regulate centrosomal functions through Aurora-A, as previous studies showed that HEF1 associates with and activates Aurora-A (39,42) and that ILK regulates spindle organization by modulating Aurora A/TACC3/ ch-TOG interaction (39,42). However, we found that deletion of FAK in primary ECs did not affect Aurora-A activation (data not shown), suggesting that FAK may use a different mechanism from HEF1 and/or ILK in its regulation of centrosome functions.
Our results showed that centrosomal-localized FAK is phosphorylated on Ser-732 and that Ser-732 phosphorylation is required for FAK to rescue the centrosomal defects in FAK Ϫ/Ϫ ECs, suggesting that FAK phosphorylation at Ser-732 is crucial for its distinct functions in G 2 /M phase of mitosis. Interestingly, Ser-732 of FAK was identified as a physiological substrate for Cdk5, and its phosphorylation was shown to promote organization of microtubule structures in the post-mitotic neurons (17,18). Centrosomes are the major microtubule-organizing center in mammalian cells, which regulate spindle bipolarity, spindle positioning, and cytokinesis (43). Therefore, Ser-732phosphorylated FAK may regulate these mitotic events through its functions in the microtubule organization in the proliferating primary ECs in a similar manner as its regulation of nuclear translocation in the post-mitotic neurons. Indeed, our preliminary studies showed that treatment of ECs with roscovitine, a specific Cdk5 inhibitor, abolished Ser-732 phosphorylation of FAK and induced similar centrosomal abnormality as that in FAK Ϫ/Ϫ ECs (data not shown). Although the potential target proteins localized in centrosomes that mediate FAK regulation of centrosomal functions are not clarified, we have found an interaction of FAK with cytoplasmic dynein in a Ser-732 phosphorylation-dependent manner. Interestingly, Nudel, a substrate for Cdk5, and its binding partner Lis1 form a complex with cytoplasmic dynein localized around the centrosomes and in the growth cones of neuronal cells (48). Perturbation of cyto- plasmic dynein or Lis1 has been shown to cause various defects in chromosome alignment, spindle organization, and centrosome separation (49 -51). In addition, Nudel participates with Lis1 in the regulation of cytoplasmic dynein function via Cdk5 phosphorylation (48). Together with our findings of disruption of FAK interaction with cytoplasmic dynein upon down-regulation of Cdk5, these results raised the possibility that Cdk5induced FAK phosphorylation at Ser-732 could stimulate activation of cytoplasmic dynein and participate in cytoplasmic dynein function. Future studies will be necessary to determine how interaction of FAK with cytoplasmic dynein regulates cytoplasmic dynein function in mitosis and also to identify potentially other target proteins of FAK in centrosomes that mediate centrosome functions of FAK during mitosis.
Consistent with a role of Ser-732-phosphorylated FAK in centrosome functions, Ser-732 phosphorylation of FAK was found to be required for its promotion of cell cycle progression and tubulogenesis of primary EC in vitro as well as tumor angiogenesis in vivo. However, we found that Ser-732 phosphorylation is dispensable for FAK stimulation of migration in the primary ECs, although Le Boeuf et al. (19) reported recently that Ser-732 can be phosphorylated by ROCK and that S732A mutant was unable to stimulate migration of HUVEC and murine embryonic fibroblasts as wild-type FAK. It is not clear whether the use of different cells is responsible for the discrepancies. Nevertheless, our data suggest that a defective cell cycle progression, caused by deregulation of centrosome functions rather than migration, contributes to the inability of S732A mutant to promote EC tubulogenesis in vitro and tumor angiogenesis in vivo.