Tyr-863 phosphorylation enhances focal adhesion kinase autophosphorylation at Tyr-397

Transcription

1 (2002) 21, ª 2002 Nature Publishing Group All rights reserved /02 $ Tyr-863 phosphorylation enhances focal adhesion kinase autophosphorylation at Tyr-397 Tzeng-Horng Leu*,1 and Ming-Chei Maa 2 1 Department of Pharmacology, College of Medicine, National Cheng Kung University, Tainan, Taiwan 70101, Republic of China; 2 Institute of Biochemistry, Chung Shan Medical University, Taichung, Taiwan, Republic of China Tyr-397 phosphorylation is important for focal adhesion kinase (FAK)-mediated signalling. In vitro FAK immunocomplex kinase experiments demonstrated that both FAK Tyr-576/577 and Tyr-863 phosphorylation regulated FAK Tyr-397 phosphorylation. While the former increased the intermolecular transphosphorylation activity of FAK, the latter was crucial for its cisphosphorylation. This observation was further supported by the reduced complex formation between Src and 3F- FAK (576F/577F/863F-FAK) as compared to that between Src and 576F/577F-FAK or Src and 863F- FAK. Regulation of cis- and transphosphorylation activities of FAK by such a differential tyrosyl phosphorylation mechanism is unprecedented. Furthermore, in fibronectin-stimulated cells, both Tyr-576/577 and Tyr-863 phosphorylation could enhance FAK Tyr- 397 phosphorylation. This observation implies that integrin-mediated FAK Tyr-397 phosphorylation was also regulated through both FAK cis- and transphosphorylation mechanisms. (2002) 21, doi: /sj.onc Keywords: Src; FAK; tyrosine kinase; tyrosol phosphorylation Introduction Integrin is a family of widely expressed transmembrane receptors that are composed of a and b subunits. Integrins mediate cellular adhesion to extra-cellular matrices (ECM) that are important for many cellular processes including growth, transformation, migration, survival, and differentiation (Kumar, 1998). In cultured cells, upon ligand binding, the integrins are clustered and their cytoplasmic domains will associate with several cytoskeleton-associated proteins including a-actinin and talin as well as tyrosine kinases (such as FAK and Src) that may promote the actin filament assembly and the formation of focal adhesion (or focal contact). It is well established that FAK becomes tyrosyl phosphorylated and activated in response to the engagement of integrins (Guan et al., 1991; Hanks et *Correspondence: T-H Leu; tzengleu@mail.ncku.edu.tw Received 6 May 2002; revised 11 July 2002; accepted 18 July 2002 al., 1992; Guan and Shalloway, 1992) and activation of Src (Kanner et al., 1990; Schaller et al., 1992; Guan and Shalloway, 1992; Calalb et al., 1995). Upon integrin activation, FAK Tyr-397 becomes phosphorylated and recruits SH2-containing proteins such as PI-3 kinase (Chen and Guan, 1994) and Src family kinases (Schaller et al., 1994; Eide et al., 1995). Src family kinases then might further mediate phosphorylation of Tyr-407, -576/577, -863 and -925 on chicken FAK (Schlaepfer et al., 1994; Calalb et al., 1995; Maa and Leu, 1998). Previous studies have indicated that Tyr- 576/577 phosphorylation can enhance the kinase activity of FAK on exogenous substrate poly (Glu- Tyr) (Calalb et al., 1995; Maa and Leu, 1998), while Tyr-925 phosphorylation provides the binding site for Grb2 SH2 domain and activates the Ras signalling pathway (Schlaepfer et al., 1994; Schlaepfer and Hunter, 1996). Mutation of Tyr-397 to phenylalanine not only abrogated the association between Src and FAK, but also ablated the increased tyrosol phosphorylation and kinase activity of FAK (Calalb et al., 1995; Schaller et al., 1994; Maa and Leu, 1998). To date, which tyrosine kinases, FAK and/or its associated Src family kinases, are responsible for integrin-mediated phosphorylation of cytoskeletal proteins remains obscure. Receptor tyrosine kinases (i.e. EGFR) exist as monomers in unstimulated cells. However, upon ligand binding, they dimerize, and become activated and autophosphorylated. This autophosphorylation is attributable to intermolecular transphosphorylation. In contrast, activation of the nonreceptor tyrosine kinases (i.e. Src) is mediated by intramolecular autophosphorylation (i.e. cis-phosphorylation). Thus, autophosphorylation could occur in an inter- or intramolecular manner. Accumulated evidence indicates that through multiple phosphorylation, the enzymatic activity of many kinases can be delicately regulated (Anderson et al., 1990; Keranen et al., 1995; Konishi et al., 1997). Since the consequences of FAK Tyr-863 phosphorylation are still unclear, we were interested in defining the role of Tyr-863 phosphorylation in FAK activity. To address this issue, a GST fusion protein containing FAK Tyr-397 and its neighboring sequence was utilized to mimic the physiological substrate of FAK and, in addition to 397F-, 576F/577F- and 863F- FAK, a 3F-FAK (i.e. 576F/577F/863F-FAK) mutant was also generated for use in this study. Interestingly,

2 in our in vitro immunocomplex kinase experiments, we observed that while Tyr-576/577 phosphorylation increased the intermolecular transphosphorylation activity of FAK, Tyr-863 phosphorylation enhanced its cis-phosphorylation activity. Both events led to the enhancement of FAK Tyr-397 phosphorylation. Similar results were also demonstrated in fibronectinstimulated cells and cells expressing active Src. Additionally, this is the first reported demonstration of regulation of FAK cis- and transphosphorylation activity by differential tyrosol phosphorylation. Results Phosphorylated Tyr-863 is crucial for FAK autophosphorylated activity FAK Tyr-863 has been demonstrated as an Srcmediated phosphorylation site both in vitro and in vivo. However, the biological role of this process is still unclear. To study whether FAK Tyr-863 phosphorylation (Pi-Tyr-863) could affect its kinase activity, CE cells expressing wt-fak and various FAK mutants were obtained. The in vitro kinase reaction of FAK immunocomplexes prepared from the lysates of these cells was performed, and the 32 P-incorporation of FAK (i.e. autophosphorylation) in the absence of any exogenous substrate was measured. Though the amount of FAK in each immunoprecipitate tested was similar (Figure 1a), 32 P-incorporated 397F- or KR-FAK was not detected in the gel as expected (Figure 1b, lanes 5, 6, Figure 1 Significant reduction of FAK autophosphorylation activity by Tyr-863 mutation. Equal amounts (500 mg) of lysates prepared from various FAK overexpressing cells, i.e. wt-fak (wt, lane 2), 397F-FAK (397F, lane 3), 576F/577F-FAK (576F/ 577F, lane 4), 863-FAK (863F, lane 5) and KR-FAK (KR, lane 6) as well as half amount of the wt-fak lysate (1/2wt; lane 1) were immunoprecipitated with FAK antibody. Half amounts of all the immunocomplexes were resolved in an SDS PAGE followed by Western immunoblotting with FAK specific antibody (FAK blot) (a). The rest of the FAK immunocomplexes were further divided into two equal parts and subjected to in vitro kinase reaction in the presence of [g- 32 P]ATP as described in the Materials and methods (b). The 32 P-incorporated amount of 1/ 2wt- (lanes 1 and 2), wt- (lanes 3 and 4), 397F- (lanes 5 and 6), 576F/577F- (lanes 7 and 8), 863F- (lanes 9 and 10) and KR FAK (lanes 11 and 12) reflected their relative autophosphorylation activity in vitro 11 and 12). Surprisingly, while *50% reduction of 32 P- incorporation was detected in 863F-FAK relative to wt- FAK (Figure 1b, lanes 9, 10 vs 3, 4), no significant reduction of 32 P-incorporation was observed in 576F/ 577F-FAK (Figure 1b, lanes 7, 8 vs 3, 4). Nevertheless, when poly (Glu-Tyr) is utilized as an exogenous substrate, phosphorylation of Tyr-576/577 (Pi-Tyr- 576/577) is required for FAK kinase activity (Calalb et al., 1995; Maa and Leu, 1998). This finding suggested that either Tyr-863 played a significant role in the regulation of FAK autophosphorylation at Tyr-397, or that Tyr-863 was another FAK autophosphorylation site. The latter seemed unlikely since no 32 P-incorporation of 397F-FAK was observed (Figure 1b, lanes 5 and 6). To further confirm the involvement of Pi-Tyr-863 in FAK Tyr-397 phosphorylation, the in vitro kinase reaction of various FAK immunocomplexes was carried out. Unlike the previous in vitro kinase assay, we substituted cold ATP for [g- 32 P]ATP in this specific experiment, and the level of FAK Tyr-397 phosphorylation (Pi-Tyr-397) was determined by direct Western immunoblotting with its specific antibody. To prove this method was feasible, the in vitro kinase reaction of immunoprecipitated Src was performed in the presence of cold ATP and various GST fusion proteins (i.e. GST- 397Y, -CTF and -CTF[863F]; Figure 2a). Then, the phosphorylated GST fusion proteins were analysed by Western immunoblotting with either Pi-Tyr397 Ab (Figure 2b, left) or anti-phosphotyrosine (ptyr) Ab (Figure 2b, right). To reveal the specificity of Pi-Tyr397 Ab, only 1/10 of the original and the kinase reaction products of GST-397Y (Figure 2b, lanes 1 and 2) were purposely loaded in the gel. As shown in Figure 2b, only phosphorylated GST-397Y (left, lane 2) was specifically detected by Pi-Tyr397 Ab, though much less tyrosyl phosphorylation of GST-397Y (right, lane 2) was detected by the ptyr Ab as compared to that of GST- CTF (right, lane 3) and -CTF (863F) (right, lane 4). Thus, the sensitivity and specificity of the Pi-Tyr397 Ab was adequate for detection of the Pi-Tyr-397 level of FAK, and this modified method was used to study the influence of Pi-Tyr-863 on FAK Tyr-397 phosphorylation. As demonstrated in Figure 3a, while a similar amount of FAK was present in each FAK immunocomplex tested, increased Pi-Tyr-397 was detected with time in both wt- and 576F/577F-FAK (Figure 3b). However, much less Pi-Tyr-397 was observed in 863F- FAK, and no Pi-Tyr-397 was detected in 397F-FAK (Figure 3b). There was about 40 60% reduction of the normalized level of Pi-Tyr-397 in 863F-FAK relative to wt- or 576/577F-FAK, indicating that Pi-Tyr-863 was indeed crucial for FAK autophosphorylation at Tyr-397 (Figure 3c). However, owing to the similarity between the sequence bearing FAK Tyr-863 and that bearing Src Tyr-527 (whose phosphorylation confers its ability to bind to Src SH2), one might argue that the reduced Pi- Tyr-397 observed in 863F-FAK was attributable to the absence of FAK-associated Src. However, three lines of evidence indicated that this was unlikely: (1) No detectable c-src was present in these FAK immunocomplexes, which were thoroughly washed (data not 6993

3 6994 Figure 2 Pi-Tyr397 Ab could specifically recognize the Tyr-397 phosphorylated FAK. In the presence of cold ATP in the kinase buffer, equal amount (2 mg) or purified bacterially expressed GST-397Y, GST-CTF, and GST-CTF(863F) fusion proteins were subjected to in vitro kinase reaction with or without active Src immunocomplexes. Then the supernatant containing the GST fusion proteins were collected and resolved in an SDS PAGE followed by Coomassi-blue staining (a) or Western immunoblotting with Pi-Tyr397 antibody or ptyr antibody (b). It should be noted that only 1/10 of the original (lane 1) and the kinase reaction products (lane 2) of GST-397Y were intentionally loaded in the gel for Western blot. The position of GST-CTF or GST-CTF (863F) fusion protein is marked with an arrow shown); (2) These FAK immunocomplexes failed to phosphorylate GST-CTF (an exogenous Src, but not FAK, substrate) (data not shown); (3) No significant decrease of 32 P-incorporation of FAK was observed when Src-specific inhibitor PP2 was present in the in vitro kinase cocktail (data not shown and Figure 6b). Thus, Pi-Tyr-397 of FAK is regulated by Pi-Tyr-863- mediated autophosphorylation. Tyr-576/577 and Tyr-863 phosphorylation can increase the trans- and cis-phosphorylation of FAK, respectively To further confirm the importance of Tyr-863 phosphorylation on FAK autophosphorylation, Tyr- 863 was replace by phenylalanine in the 576F/577F- FAK to obtain 3F-FAK. Then the relative activities of FAK in cells expressing both active Src and various FAKs including wt, 576F/577F-, 3F- and KR-FAK were determined by in vitro kinase reaction. In an attempt to measure their relative transphosphorylation activity at the same time, bacterially expressed GST- Figure 3 Mutation of Tyr-863 to phenylalanine led to the reduction of FAK autophosphorylation at Tyr-397. (a) The relative amounts of FAK in various FAK immunocomplexes. Same amounts of lysates (1.2 mg) prepared from each FAK overexpressing cells were immunoprecipitated with FAK antibody and one quarter of the immunoprecipitates was analysed by SDS PAGE followed by Western immunoblotting with FAK antibody. The rest of the immunocomplexes were further divided for in vitro kinase reactions (b). Instead of [g- 32 P]ATP, cold ATP (1 mm) was present in the kinase buffer and the reaction was carried out in triplicate for 0, 10 and 20 min. Then, the tyrosyl phosphorylated FAK was resolved in an SDS PAGE followed by Western immunoblotting with Pi-Tyr397 Ab. The level of Pi-Tyr-397 was quantified by densitometer and normalized by the FAK amount present in the immunocomplexes (c). For comparison, the value of wt-fak in the 0 mine was arbitrarily assigned as 1. Values were means+s.d. from a representative experiment as shown in (b). Similar results were obtained from at least three experiments 397Y fusion protein, which contains a FAK Tyr-397 autophosphorylation site and its neighboring sequence and is designed to mimic the physiological substrate of FAK, was included as an exogenous substrate. As previously described, in the presence of active Src, wt- FAK exhibited higher tyrosyl phosphorylation and autophosphorylation activity as well as ability to transphosphorylate GST-397Y (Figure 4a, lanes 1 and 2). Since no 32 P was incorporated into GST (data not shown and Figure 5b), we envisioned FAKmediated GST-397Y phosphorylation to be quite specific. Remarkably, as demonstrated in Figure 4a, 32 P-incorporated GST-397Y, but not FAK, was reduced in the presence of 576F/577F-FAK (lane 3). To our surprise, when Tyr-863 in 576F/577F-FAK was mutated to phenylalanine (i.e. 3F-FAK), the amount of 32 P-incorporation in both GST-397Y and FAK was significantly reduced (Figure 4a, lane 4), while negligible signals were detected in KR-FAK (Figure 4, lane 5). Further quantitative analysis revealed that under the influence of Src, though 576F/577F-FAK lost most of its transphosphorylation activity (Figure 4b, right), it still retained autophosphorylation activity comparable to that of wt-fak (Figure 4B, left). This finding suggested that mutation of Tyr-576/577 was not sufficient to abolish the enhancement of FAK autophosphorylation and that phosphorylation of other

4 6995 Figure 4 Mutation of Tyr-863 to phenylalanine dramatically reduces the Src-mediated enhancement of 576F/577F-FAK autophosphorylation in vitro. (a) The FAK immunocomplexes prepared from cells expressing wt-fak alone (lane 1) or cells expressing both active Src (Src 518amb) and wt-fak (lane 2), 576F/ 577F-FAK (lane 3), 3F-FAK (lane 4), or KR-FAK (lane 5) were subjected to in vitro kinase reactions in the presence of [g- 32 P]ATP for 15 min. The 32 P-labeled FAK and GST-397Y were resolved in an SDS PAGE and detected by autoradiography as described in Materials and methods. The relative phosphotyrosyl levels and amounts of FAK in each cell line were further analysed by immunoblotting of the FAK immunocomplexes with either ptyr antibody or FAK antibody. The amount of active Src (Src 518amb) and endogenous c-src (indicated by an arrow) in each cellular lysate was detected by Western blotting with Src specific antibody (b) The 32 P-incorporated FAK and GST-397Y derived from above were excised and counted. Then the cpm values were normalized based on the FAK amounts present in the immunocomplex and used to represent the auto- (left) and transphosphorylation (right) activity of FAK respectively. The results were shown in means+s.d. for one representative experiment performed in triplicate Figure 5 Src-mediated phosphorylation of Tyr-576/577 and Tyr- 863 increases FAK Tyr-397 phosphorylation. (a) PP2 inhibits the kinase activity of Src. Active Src was immunoprecipitated from the lysates (500 mg) of Src overexpressors and divided into five portions for in vitro kinase reaction. The kinase reaction was carried out under the influence of various concentrations of PP2 (0, 10, 50, 100 nm) utilizing 0.1 mg of GTS-CTF or GST-CTF (863F) as the exogenous substrates. The 32 P-labeled GTS fusion proteins and Src were resolved in an SDS PAGE and detected by autoradiography. (b) The FAK immunocomplexes prepared from cells expressing both active Src and wt-fak (wt, lanes 1 to 4), 576F/ 577F-FAK (576F/577F, lanes 5 to 7), 863F-FAK (863F, lanes 8 to 10) or 3F-FAK (3F, lanes 11 to 13) were subjected to a similar in vitro kinase reaction as described in Figure 4 except that various concentrations of PP2 (0, 50, 100 mm) were included in the kinase buffer. Also, GST (0.1 mg, lane 1) was used as the negative control for GST-397Y (0.1 mg, lanes 2 to 13) in this experiment. The 32 P-labeled GST-397Y and FAK were resolved in an SDS PAGE and detected by autoradiography. (c) Equal portions of FAK immunocomplexes carried out in (b) were Western immunoblotted with FAK Ab tyrosine residues, i.e. Tyr-863, in 576F/577F-FAK could compensate for its reduced Tyr-397 phosphorylation by engaging in cis-phosphorylation. Indeed, the level of Pi-Tyr-863 in 576F/577F-FAK was similar to that in wt-fak (Figure 6c, middle and lower) and the FAK cis-phosphorylation was abolished when Tyr-863 was mutated to phenylalanine (Figure 4b, left). Similar results were obtained when poly (Glu-Tyr) was utilized as an exogenous substrate (data not shown). However, we could not exclude the possibility that the abrogation of autophosphorylation of 3F-FAK might be attributable to the mutation-induced conformational change that severely impaired its catalytic activity. Both phosphorylated Tyr-863 and Tyr-576/577 are crucial for FAK Tyr-397 phosphorylation The FAK immunoprecipitates prepared in the experiment described in Figure 4 were from cells

5 6996 Figure 6 3F-FAK exhibited reduced level of Pi-Tyr-397 and failed to complex with Src. The FAK immunoprecipitates prepared from cells expressing active Src alone (lane 1), or cells expressing both active Src and various FAK mutants (lanes 2 to 7 as indicated) were resolved in an SDS PAGE and Western immunoblotted with either Pi-Tyr397 Ab (a, upper), FAK Ab (a, lower), or anti-src antibody (b). (c) The relative amounts of Src (upper), FAK Pi-Tyr-863 (middle) and FAK (lower) present in these cell lines were determined by immunoblotting of the whole lysate with anti-src, anti-pi-tyr863, and anti-fak antibodies, respectively co-expressing active Src and various FAKs. To exclude the possibility that Src might be present in the FAK immunoprecipitates and phosphorylate FAK at Tyr-397 and other residues, its specific inhibitor PP2 was included in the cocktail of the in vitro kinase reactions (Figure 5). Since PP2 treatment (100 nm) wiped out almost all the Src-mediated 32 P- incorporation into GST-CTF (Figure 5a), the 32 P- labeled FAK and GST-397Y in PP2-treated FAK immunocomplex kinase reaction should unambiguously reflect the FAK kinase activity. As shown in Figure 5b, a similar amount of FAK present in each FAK precipitate tested (Figure 5c), was able to phosphorylate GST-397Y and to autophosphorylate, similar to the results shown in Figure 4. Since inclusion of PP2 in the reaction mixture did not alter the overall experimental outcomes, we concluded that the influence of Src on phosphorylation of FAK Tyr-397 could be negligible. With this finding, we confirmed that Pi-Tyr-863 indeed regulated FAK cisphosphorylation activity. Interestingly, 863F-FAK prepared from cells expressing active Src retained no less kinase activity on both poly (Glu-Tyr) and FAK itself than wt-fak (Figure 5b and data not shown). Thus, Pi-Tyr-576/577 could also regulate the intermolecular transphosphorylation of FAK Tyr-397 in vitro. The diminished Pi-Tyr-397 of 3F-FAK abrogated its association with Src To confirm the importance of both Pi-Tyr-863 and Pi- Tyr-576/577 on FAK Tyr-397 phosphorylation in vivo, the Pi-Tyr-397 level of FAK in cells coexpressing various FAKs and active Src was further examined by Western immunoblotting with Pi-Tyr397 Ab. Consistent with our previous studies, slightly reduced Pi-Tyr- 397 was detected in 567F/577F- and 863F-FAK in comparison with that in wt-fak, and almost no Pi- Tyr-397 was observed in 3F-FAK, though similar amounts of FAK were detected in these FAK immunoprecipitates (Figure 6a). Since Pi-Tyr-397 of FAK is a strong Src SH2-binding site, the amounts of Src present in the FAK immunocomplexes can be used to measure the relative abundance of Pi-Tyr-397 in FAK. As shown in Figure 6b, a significant amount of Src was detected in the immunocomplexes of wt-, 575F/577F-, and 863F-FAK, but not in that of 397F- FAK, though similar amounts of Src were present in all Src expressing cells (Figure 6c, upper). Consistent with previous results (Figure 5b), 863F-FAK retained Pi-Tyr and was able to associate with Src, indicating that Pi-Tyr-397 might occur through FAK-FAK intermolecular transphosphorylation in vivo. In support of this hypothesis, low levels of Pi-Tyr-397 detected in KR-FAK might be attributable to endogenous FAKmediated transphosphorylation (Figure 6a). Furthermore, mutation of Tyr-576/577 in 863F-FAK (i.e. 3F- FAK) led to the significant reduction of FAK Pi-Tyr- 397 (Figure 6a) and Src present in 3F-FAK immunocomplexes (Figure 6b). On the basis of all these findings, we concluded that both Pi-Tyr-863 and Pi- Tyr-576/577 regulated the Pi-Tyr-397 level of FAK in vivo. The fibronectin-induced FAK Tyr-397 phosphorylation is regulated by both Pi-Tyr-576/577 and Pi-Tyr-863 It has been well documented that FAK, an important player in integrin-mediated signaling, becomes autophosphorylated when cells adhere to fibronectin-coated plates. However, the mechanisms responsible for this integrin-induced FAK autophosphorylation remain unclear. Since we have demonstrated that both Pi- Tyr-576/577 and Pi-Tyr-863 could affect FAK Tyr-397 phosphorylation both in vitro and in cells coexpressing active Src and FAK, we wondered whether similar mechanisms could also be applied to regulate the fibronectin-induced FAK autophosphorylation. To address this point, cells overexpressing wt, 576F/ 577F-, 863F-, and 3F-FAK were trypsinized and plated on fibronectin (FN)-coated dishes for 10 or 30 min. In addition, cells plated on poly-lysine-coated dishes were used as the control in this study. The amount of Tyr-397 phosphorylated FAK in each cell lysate was determined by Western immunoblotting utilizing its specific antibody. Though similar levels of FAK were observed in all these FAK-overexpressing cells (Figure 7a), the level of Pi-Tyr-397 was

6 significantly reduced in poly-lysine (PL)-stimulated cells than in cells stimulated by fibronectin. This integrin-mediated phosphorylation reached plateau shortly after 10-min activation in cells expressing wt- FAK but not in cells expressing 576F/577F- or 863F- FAK (Figure 7b). To accurately compare the level of FN-induced phosphorylation of FAK Tyr-397 in these cells, results derived from Figure 7a,b were quantified by densitometry. The normalized Pi-Tyr-397 of FAK was plotted and is shown in Figure 7c. The substantial reduction of Pi-Tyr-397 cells expressing 3F-FAK indicated that Pi-Tyr-576/577 and Pi-Tyr-863 not only were responsible for integrin-mediated FAK autophosphorylation, but also worked in a synergistic manner Discussion To date, Src family members are the only known tyrosine kinases binding to and phosphorylating FAK at Tyr-407, -576/577, -863 and -925 both in vitro and in vivo. To analyse the influence of each tyrosol phosphorylation on FAK enzymatic activity, poly (Glu-Tyr) has been used as an exogenous substrate to conduct the in vitro immunocomplex kinase assay, and the correlation between increased FAK tyrosyl phosphorylation and its elevated enzymatic activity has been observed. This phosphorylation-mediated FAK activation was abrogated when Tyr-576 and -577 were mutated to phenylalanines (Calalb et al., 1995; Maa and Leu, 1998). However, no effect of FAK Tyr-863 phosphorylation on FAK transphosphorylation activity was observed in this specific experiment (Maa and Leu, 1998). Interestingly, when phosphorylation of FAK itself was used to assess FAK kinase activity, we observed that FAK autophosphorylation at Tyr-397 was not significantly influenced by 576F/577F mutation, but was reduced to about one half by 863F mutation (Figure 1). This finding seemed to conflict with what we observed previously, and could be explained by combined cis- and transphosphorylation of Tyr-397. In an attempt to distinguish the trans- from the cisphosphorylation activity of FAK without ambiguity, GST fusion protein containing Tyr-397 and its neighbouring sequence (GST-397Y) was utilized as an exogenous substrate, whose incorporation of 32 P was a measure of the transphosphorylation activity of FAK. In the meantime, the extent of 32 P-incorporated FAK was determined to reflect both the cis- and transphosphorylation activity of FAK. When the 32 P-incorporation of FAK and GST-397Y mediated by wt-, 576F/ 577F and 3F-FAK was compared, we observed that mutation at Tyr-576/577 inhibited the transphosphorylation, but not the cis-phosphorylation, activity of FAK. However, when Tyr-863 was further mutated, the ability of FAK to phosphorylate itself was almost completely abolished. Thus, Pi-Tyr-863 might regulate the level of Pi-Tyr-397 probably by cis-phosphorylation, and the cis- and transphosphorylation activity of FAK could be differentially regulated by phosphorylation of its different tyrosine residues. This observation was Figure 7 Both Tyr-576/577 and Tyr-863 phosphorylation modulate the level of Pi-Tyr-397 of FAK in response to fibronectin. (a) The relative amounts of FAK in various FAK expressing cells were determined by Western blot analysis. (b) Various FAK expressing cells were detached and held in suspension for 30 min, then plated on fibronectin-coated dishes for 10 and 30 min. Cells plated on poly-lysine-coated dishes for 30 min were used as a control. Cell lysates were analysed by Western immunoblotting with Pi-Tyr397 Ab to assess the level of FAK-Tyr-397 phosphorylated FAK. This experiment was repeated at least three times with similar results. (c) Densitometric quantification of Tyr-397 phosphorylated FAK normalized for FAK protein levels; values were means+s.d. from three measurements of b without precedent, and might be explained by the position of Tyr-397 (the autophosphorylation site) outside the FAK catalytic domain (see below). Biochemical and structural analysis of several protein kinases have revealed that autophosphorylation within the catalytic domain can stabilize the kinases in their active conformation (Hubbard et al., 1998). For example, Tyr-416, the Src autophosphorylation site, is located in the activation loop, which when phosphorylated maintains Src in its enzymatically active site. A comparison of the amino acid sequences of the catalytic domains of FAK and Src revealed that FAK Tyr-576 or -577 is analogous to Src Tyr-416 (Schaller et al., 1992), and indeed the phosphorylation of FAK Tyr-576 or 577 can up-regulate the transphosphorylation activity of FAK (Calalb et al., 1995; Maa and Leu, 1998). However, Tyr-576 and -577 are not the major FAK autophosphorylation sites. Instead, FAK autophosphorylates itself at Tyr-397, which resides at the FAK N-terminal region, close to the catalytic domain. Thus, the cis- and transphosphorylation activity of FAK might be regulated by separate and different mechanism. Indeed, this hypothesis was

7 6998 supported by our findings that Pi-Tyr-576/577 and Pi- Tyr-863 up-regulated FAK trans- and cis-phosphorylation activity respectively. On the basis of our in vitro results conducted in cells overexpressing FAK alone (Figures 1 3), we concluded that Pi-Tyr-863, but not Pi-Tyr-576/577, regulated FAK cis-phosphorylation at Tyr-397. Additionally, we speculated that this might be attributable to the blockade of intermolecular interaction among the individual protein A Sepharose bead-immobilized FAK molecules and/or low level of Src-mediated Pi- Tyr-576/577. Interestingly, when the FAK immunocomplex was prepared from cells co-expressing FAK and Src, its transphosphorylation activity was enhanced owing to the presence of Src-mediated Pi- Tyr-576/577. In this context, Pi-Tyr-863 and Pi-Tyr- 576/577 regulated FAK cis- and trans-phosphorylation, respectively (Figures 4 and 5). Then, we asked how FAK Tyr-397 phosphorylation is regulated in vivo. Within cells, FAK is located in the integrin-containing multiple protein complex where FAK molecules can mutually interact. Under this circumstance, the Pi-Tyr- 397 level of FAK was regulated not only by cisphosphorylation but also by transphosphorylation. This hypothesis could be supported by the following findings: (1) Relatively high abundance of Src was present in either 576F/577F- or 863F-FAK immunocomplex, and only when all these three tyrosine residues were mutated to phenylalanines was complex formation between Src and FAK abrogated (Figure 6b); (2) In response to Src activation or fibronectin stimulation, the enhancement of Pi-Tyr-397 levels in FAK was not completely inhibited by either Tyr-576/ 577 or Tyr-863 mutation alone but by mutation of both sites to phenylalanines (Figures 6a, 7b,c); (3) Pi- Tyr-576/577 was required for the maximal phosphorylation of Tyr-397 that follows fibronectin stimulation (Owen et al., 1999; Ruest et al., 2000; Figure 7) or Src activation (Figure 6a). Figure 8 presents a simple model proposed to illustrate the mechanisms modulating FAK Tyr-397 phosphorylation. As demonstrated in Figure 8a, when the intermolecular interaction among FAK is obstructed, such as the in vitro FAK immunocomplex prepared from the lysates of FAK overexpressing cells, phosphorylation of Tyr-397 could only be mediated by cis-phosphorylation and regulated by Pi-Tyr-863. The enhancement of FAK cis-phosphorylation activity by Pi-Tyr-863 might be attributable to the release of the constraint on the Tyr-397-bearing kinase domain imposed by its N- terminal portion as suggested by Giancotti (1997), and/ or to a phosphorylation-induced conformational change such that Tyr-397 becomes accessible to the kinase domain. However, when the intermolecular interaction of FAK was promoted (i.e. as by FAK with Pi-Tyr-576/577 or FAK located in focal contacts), phosphorylation of Tyr-397 might occur through both intra- and intermolecular interactions. Since Pi-Tyr-863 and its neighboring sequence is a potential Src-binding site, the transient association between Src and FAK might facilitate Src-mediated phosphorylation of FAK Tyr-576/577 upon integrin activation. Under these circumstances, both Pi-Tyr-576/577 and Pi-Tyr-863 could cooperatively regulate the phosphorylation of FAK Tyr-397 at the same time (Figure 8b). Src-mediated sites on FAK have been demonstrated to play crucial roles in integrin signalling and in the process of Src-induced transformation. Sequence comparison between FAK and its family member PYK2 revealed that except for Tyr-863, PYK2 retained all the Src-mediated sites (i.e. Tyr-407, -576/577 and -925) on FAK. Since we have demonstrated that Tyr- 863 is important in the regulation of FAK autophosphorylation, the absence of an analogous Tyr-863- bearing sequence in PYK2 implies that the mechanisms modulating PYK2 activity should be different from the one described for FAK. Indeed, PYK2 was originally identified as a kinase whose tyrosyl phosphorylation and catalytic activity were regulated by the intracellular concentration of Ca 2+ (Lev et al., 1995). How Ca 2+ and tyrosyl phosphorylation affects PYK2 enzymatic activity thus becomes an intriguing issue and needs further investigation. Figure 8 Model of regulation of FAK Tyr-397 phosphorylation by differential tyrosyl phosphorylation. For the aspect of autophosphorylation, the FAK molecule may exist as an inactive or active conformation. When FAK was immobilized in protein A Sepharose bead in the in vitro kinase reaction, Pi-Tyr-863 alone could promote the intramolecular autophosphorylation at Tyr- 397 (a). In contrast, in fibronectin-stimulated cells or cells expressing active Src, the mutual interaction among FAK molecules could occur. Therefore, the level of Tyr-397 phosphorylated FAK not only could be regulated by Pi-Tyr-863-mediated cisphosphorylation but also by Pi-Tyr-576/577-mediated transphosphorylation (b)

8 Materials and methods Cell, viruses and preparation of cell lysates Primary chicken embryo (CE) cells were prepared as previously described (Maa and Leu, 1998). cdnas encoding FAK and its mutants were cloned into retroviral expression vector RCAS A (BH), while cdna encoding active c-src mutant (518amb) was cloned into RCAS B (BH) vector as described previously (Hughes et al., 1987; Maa and Leu, 1998). These constructs were individually transfected into CE cells by calcium phosphate precipitation method for generating FAK or Src overexpressing cells. Cells expressing both Src and FAK and the preparation of their lysates were carried out as previously described (Maa and Leu, 1998). Protein amount in each lysate was determined by Bio-Rad protein assay kit (Bio-Rad). Plasmids and bacterial fusion protein The construction of plasmid DNA expressing 397F-, 576F/ 577F-, 863F- and KR-FAK mutants was described previously (Maa and Leu, 1998). To generate 3F (i.e. 576F/577F/863F)- FAK expression construct, the fak cdna encoding 576F/ 577F-FAK was first subcloned into the plasmid palter 71 mutagenesis vector (Promega). Then, Tyr-863 was changed to phenylalanine by oligonucleotide-directed mutagenesis according to the provider s manual. After the sequence of fak cdna encoding 3F-FAK was confirmed by Sangerdideoxy DNA-sequencing method, it was further subcloned into RCAS A (BH) vector. To generate the bacterial GST fusion protein containing FAK Tyr-397 and its neighboring sequence (GST-397Y), DNA fragment encoding FAK amino acid 385 to amino acid 406 was generated by PCR (Polymerase Chain Reaction) amplification of chicken fak cdna by utilizing an optimal pair of DNA primers. At their 5 ends, the sense and antisense primers were linked to BamHI and EcoRI restriction enzyme-cutting sites respectively. The resulting fragment was digested with BamHI/ EcoRI and cloned into the BamHI/EcoRI site of pgex-2t to generate pgex-2t-397y. GST-CTF (amino acids ) and GST-CTF (863F) were encoded by pgex-cterm (Hildebrand et al., 1995) and pgex-cterm (863F) respectively. The pgex-cterm (863F) was derived from pgex- Cterm whose NcoI fragment was replace by its counterpart obtained from 863F-fak cdna. The induction and purification of GST fusion protein was done as previously described (Maa et al., 1999). Antibodies, immunoprecipitation and immunoblotting Rabbit polyvalent antibody against FAK (CTF3) was generated by immunization of the male New Zealand rabbits with the C-terminal FAK-containing fusion protein. Immunoprecipitation and immunoblotting of FAK was performed as previously described (Maa and Leu, 1998). For detection of phosphotyrosine content, Western immunoblotting was performed with horseradish peroxidase-conjugated mouse monoclonal antibody PY20 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and detected by Enhanced Chemiluminescence method (Amersham). For detection of Src, Western immunoblotting was performed with anti-src monoclonal antibody (2 17) followed by 125 I-labeled goatanti-mouse IgG (Amersham) blotting and X-ray autoradiography. For detection of FAK and the level of its Tyr-397 or Tyr-863 phosphorylation, CTF3 and purified rabbit serum specifically against phosphorylated FAK Tyr-397 (UBI, Lake Placid, NY, USA) or phosphorylated FAK Tyr-863 (BioSource International, Camarillo, CA, USA) were used respectively in the primary blotting. Then 125 I-labeled protein A was used in the secondary immunoblotting and detected by X-ray autoradiography. In vitro kinase reaction FAK immunocomplexes were prepared from the cells tested. After washing at least three times with RIPA buffer and twice with Tris-saline solution (50 mm Tris-Cl, 150 mm NaCl, ph 7.2), the FAK immunocomplexes were ready for in vitro kinase reaction. The kinase reaction was performed in a 20 ml kinase mixture, containing 20 mm piperazine-n,n -bis(2- ethanesulfonic acid) (PIPES) (ph 7.2), 10 mm MnCl 2,5mM ATP, and 10 mci [g- 32 P]ATP (6000 Ci/mmol, NEN), for various times as indicated in each experiment at 258C. Occasionally, the bacterial GST fusion protein GST-397Y (0.2 mg) was added as an exogenous substrate for the measurement of FAK transphosphorylation activity. Then, by adding equal volume of 26-concentrated Laemmli sample buffer (Laemmli, 1970) and boiling for 5 min, the kinase reactions were terminated. The 32 P-labeled products were resolved by sodium dodecyl sulphate (SDS)-polyacrylamide gel electrophoresis (PAGE), dried down, and visualized by autoradiography. For further quantitative analysis, the labelled FAK and GST-397Y were individually excised from the gel and counted. Fibronectin stimulation To study the effect of integrin engagement on FAK Tyr-397 phosphorylation, 16-h serum-starved cells expressing various FAK were trypsinized and held in suspension for *30 min. Then, cells were plated on poly-lysine-coated dishes (10 mg/ cm 2 ) for 30 min as a control or fibronectin-coated culture dishes (2.5 mg/cm 2 ; Guan et al., 1991) for 10 or 30 min. Their lysates were harvested and analysed by Western immunoblotting Acknowledgements We would like to thank Drs Sarah J Parsons and J Thomas Parsons for providing us with the Src mabs and cdna constructs of fak and Src. We appreciate the technical help provided by Ruey-Wen Lin and Ya-Chun Chuang. This work was supported by National Science Council Grant (NSC B to MC Maa), NHRI (NHRI- EX SL to TH Leu) and by MOE Program for Promoting Academic Excellence of Universities (91-B- FAO9-1-4 to TH Leu) in Taiwan, Republic of China. References Anderson NG, Maller JL, Tonks NK and Sturgill TW. (1990). Nature, 343, Calalb MB, Polte TR and Hanks SK. (1995). Mol. Cell. Biol., 15,

9 7000 Chen HC and Guan JL. (1994). Proc. Natl. Acad. Sci. USA, 91, Eide BL, Turck CW and Escobedo JA. (1995). Mol. Cell. Biol., 15, Giancotti FG. (1997). Curr. Opin. Cell Biol., 9, Guan JL, Trevithick JE and Hynes RO. (1991). Cell Regul., 2, Guan JL and Shalloway D. (1992). Nature, 358, Hanks SK, Calalb MB, Harper MC and Patel MK. (1992). Proc. Natl. Acad. Sci. USA, 89, Hildebrand JD, Schaller MD and Parsons JT. (1995). Mol. Biol. Cell, 6, Hubbard SR, Mohammadi M and Schessing J. (1998). J. Biol. Chem., 273, Hughes SH, Greenhouse JJ, Petropoulos CJ and Sutrave P. (1987). J. Virol., 61, Kanner SB, Reynolds AB, Vines RR and Parsons JT. (1990). Proc. Natl. Acad. Sci. USA, 87, Keranen LM, Dutil EM and Newton AC. (1995). Curr. Biol., 5, Konishi H, Tanaka M, Takemura Y, Matsuzaki H, Ono Y, Kikkawa U and Nishizuka Y. (1997). Proc. Natl. Acad. Sci. USA, 94, Kumar CC. (1998)., 17, Laemmli UK. (1970). Nature, 227, Lev S, Moreno H, Marinez R, Canoll P, Peles E, Musacchio JM, Plowman GD, Rudy B and Schlessinger J. (1995). Nature, 376, Maa MC and Leu TH. (1998). Biochem. Biophys. Res. Commun., 251, Maa MC, Lai JR, Lin RW and Leu TH. (1999). Biochim. Biophys. Acta, 1450, Owen JD, Ruest PJ, Fry DW and Hanks SK. (1999). Mol. Cell. Biol., 19, Ruest PJ, Roy S, Shi E, Mernaugh RL and Hanks SK. (2000). Cell Growth Differ., 11, Schaller MD, Borgman CA, Cobb BS, Vines RR, Reynold AB and Parsons JT. (1992). Proc. Natl. Acad. Sci. USA, 89, Schaller MD, Hildebrand JD, Shannon JD, Fox JW, Vines RR and Parsons JT. (1994). Mol. Cell. Biol., 14, Schlaepfer DD, Hanks SK, Hunter T and van der Geer P. (1994). Nature, 372, Schlaepfer DD and Hunter T. (1996). Mol. Cell. Biol., 16,