Caspases Cleave Focal Adhesion Kinase during Apoptosis to Generate a FRNK-like Polypeptide*

Focal adhesion kinase (Fak) is a non-receptor protein-tyrosine kinase that stimulates cell spreading and motility by promoting the formation of contact sites between the cell and the extracellular matrix (focal adhesions). It suppresses apoptosis by transducing sur-vival signals that emanate from focal adhesions via the clustering of transmembrane integrins by components of the extracellular matrix. We demonstrate that Fak is cleaved by caspases at two distinct sites during apoptosis. The sites were mapped to DQTD 772 , which was preferentially cleaved by caspase-3, and VSWD 704 , which was preferentially cleaved by caspase-6 and cytotoxic T lymphocyte-derived granzyme B. The cleavage of Fak during apoptosis separates the tyrosine kinase domain from the focal adhesion targeting (FAT) domain. The carboxyl-terminal fragments that are generated sup-press phosphorylation of endogenous Fak and thus re-semble a natural variant of Fak, FRNK, that inhibits Fak activity by preventing the localization of Fak to focal adhesions. The cleavage of Fak by caspases may thus play an important role in the execution of the suicide program by disabling the anti-apoptotic function of Fak. Interestingly, rodent Fak lacks an optimal caspase-3 consensus cleavage site although it is cleaved in murine cells undergoing apoptosis

Focal adhesion kinase (Fak) is a non-receptor proteintyrosine kinase that stimulates cell spreading and motility by promoting the formation of contact sites between the cell and the extracellular matrix (focal adhesions). It suppresses apoptosis by transducing survival signals that emanate from focal adhesions via the clustering of transmembrane integrins by components of the extracellular matrix. We demonstrate that Fak is cleaved by caspases at two distinct sites during apoptosis. The sites were mapped to DQTD 772 , which was preferentially cleaved by caspase-3, and VSWD 704 , which was preferentially cleaved by caspase-6 and cytotoxic T lymphocyte-derived granzyme B. The cleavage of Fak during apoptosis separates the tyrosine kinase domain from the focal adhesion targeting (FAT) domain. The carboxyl-terminal fragments that are generated suppress phosphorylation of endogenous Fak and thus resemble a natural variant of Fak, FRNK, that inhibits Fak activity by preventing the localization of Fak to focal adhesions. The cleavage of Fak by caspases may thus play an important role in the execution of the suicide program by disabling the anti-apoptotic function of Fak. Interestingly, rodent Fak lacks an optimal caspase-3 consensus cleavage site although it is cleaved in murine cells undergoing apoptosis at an upstream site. This appears to be the first example of a caspase substrate where the cleavage sites are not conserved between species.
Fak is a 125-kDa non-receptor protein-tyrosine kinase that is recruited to and activated by the engagement of transmembrane integrins by components of the extracellular matrix (e.g. fibronectin) during integrin-mediated cell adhesion. The activation of Fak leads to (i) activation of the Ras-mitogen-activated protein kinase pathway and (ii) recruitment and phosphorylation (either directly or indirectly via the recruitment of Src-like tyrosine kinases) of a number of cytoskeletal proteins, resulting in the formation of contact sites between the cell surface and the extracellular matrix called focal adhesions. By activating the Ras-mitogen-activated protein kinase pathway and promoting the assembly of focal adhesions, Fak mediates multiple cellular responses to cell adhesion including cell sur-vival and proliferation as well as cell spreading and motility (for review, see Ref. 1). Fak function is dependent on two distinct domains: the tyrosine kinase domain within the aminoterminal half of the protein and a focal adhesion targeting (FAT 1 ) domain within the carboxyl-terminal half of the protein. Fak is positively regulated by autophosphorylation, which allows the recruitment of signaling molecules like Src (2, 3), phosphatidylinositol 3-kinase (4,5), and subsequently Grb2 (6). Fak is negatively regulated by the expression of FRNK (p41/ p43 FRNK ; Fak-related non-kinase), a truncated isoform of Fak that contains a FAT domain but lacks the kinase domain (7). FRNK inhibits the cellular responses to adhesion by preventing the localization of Fak to sites of integrin clustering (8).
The importance of Fak in transducing an anti-apoptotic signal upon integrin engagement is underscored by a number of studies: (i) constitutively activated forms of Fak prevent epithelial cell death upon cell detachment (anoikis) (9); (ii) inhibition of Fak in cultured fibroblasts results in apoptosis (10); (iii) Fak is overexpressed in some types of cancers (11)(12)(13)(14); and (iv) antisense oligonucleotides to Fak induce apoptosis in tumor cells (15). It is therefore not surprising that Fak has been shown to be the target of proteolysis in chicken embryo fibroblasts undergoing apoptosis (16). More recently, these observations were extended in human cell lines undergoing apoptosis and the caspase family of cysteinyl aspartate-specific proteases were directly implicated in Fak cleavage (17) and disassembly of focal adhesions (18). The caspases are responsible for initiating and executing apoptotic cell death by cleaving critical homeostatic, repair, and structural proteins in the dying cell (for review see Ref. 19). In this report, we identify two sites within chicken Fak that are cleaved by caspases in vitro and in apoptotic cells. Cleavage at either site results in the separation of the kinase domain from the FAT domain. Fragments containing the FAT domain, when expressed in HeLa cells, inhibit phosphorylation of endogenous Fak, suggesting that they act like FRNK and interfere with the function of uncleaved Fak. Although the molecular masses of the fragments released from the cleavage of human Fak in cells undergoing apoptosis are consistent with the sites identified in chicken Fak, cleavage of murine Fak appears to occur at a different site.
GAT CAA ACA GCT TCC TGG AAC 3Ј and 5Ј GTT CCA GGA AGC TGT TTG ATC 3Ј were used in combination with the internal oligonucleotides 5Ј AAGCCCTTCCAGGGAGTG 3Ј and 5Ј ATGCTGATACTTCCT-GAGG 3Ј. From the large PCR fragment, a 417-base pair BstEII fragment containing the mutation was removed and cloned back into the Fak wild-type sequence. The region corresponding to the BstEII fragment was fully sequenced to ensure that no other mutation was introduced inadvertently. A similar strategy was used to generate the D704A mutation using the complementary inverse oligonucleotides 5Ј GTC ACA GTA TCC TGG GCC TCA GGA GGA TCA GAT G 3Ј and 5Ј CAT CTG ATC CTC CTG AGG CCC AGG ATA CTG TGA C 3Ј in combination with the internal oligonucleotides 5Ј AAG CCC TTC CAG GGA GTG 3Ј and 5Ј TCC TCC ATG TTT GGC TGC 3Ј. The mutated fragment was subcloned into the Fak wild-type sequence using the restriction sites PflMI and NdeI, and the clone was sequence-verified.
The introduction of a flag epitope (DYKDDDDK) at the carboxyl terminus of Fak was performed by PCR with the following oligonucleotides: 5Ј CGC TCG AGT TAC TTG TCA TCG TCG TCC TTG TAG TCG TGG GGC CTG GAC TGG CTG ATC ATT TTC AG 3Ј and 5Ј CCA GAT CAT GCC GCT CCA CC 3Ј.
For stable expression of Fak in K562 cells, wild-type and mutant Fak cDNAs were subcloned from the pBluescript II KSϪ digested with the restriction enzymes NotI and EcoRV into the episomal eukaryotic expression vector pCep4␤ (21) digested with BamHI (rendered bluntended with T4 DNA polymerase) and NotI. cDNAs for expression of FRNK and caspase generated FRNK-like fragments in HeLa cells were engineered by PCR using the following oligonucleotides: 5Ј AAT TAA CCC TCA CTA AAG GG 3Ј and either 5Ј CAG GAA TTC TAG CAA AAC CAT GGA ATC CAG GCG ACA AGT CAC AG 3Ј for FRNK, 5Ј CAG GAA TTC TAG CAA AAC CAT GTC AGG AGG ATC AGA TGA AGC TC 3Ј for C 705-1053 , or 5Ј CAG GAA TTC TAG CAA AAC CAT GTC CTG GAA CCA TCG ACC TCA GG 3Ј for C 773-1053 . C 705-1053 and C 773-1053 correspond exactly to the fragments generated by caspase cleavage at VSWD 704 and DQTD 772 , respectively, except that an initiating methionine was introduced immediately upstream of the P 1 Ј amino acid. The fragments generated by PCR were first subcloned into pBluescript II KSϪ digested with the restriction enzyme EcoRI, sequence-verified, and then subcloned into pCep4␤ (21) as described above.
Cell Lines-Cell lines stably expressing the various Fak sequences were created by transfection of the pCep4␤ constructs in the human K562 lymphoid cell line (ATCC GM05372E, NIGMS Human Genetic Mutant Cell Repository, NIH) with LipofectAMINE TM (Life Technologies, Inc.) according to the manufacturer's instructions. Cell lines were propagated and selected in RPMI containing 0.5 mg/ml hygromycin B (Boehringer), 10% (v/v) fetal bovine serum, penicillin (100 units/ml), and streptomycin (100 g/ml).
Induction of Apoptosis-Jurkat cells were serum-starved in 0.5% (v/v) fetal bovine serum for 48 h and then treated for 2 h with anti-CD95 (Fas/Apo-1) antibodies (MBL) at 1 g/ml. HeLa and 3T3 L1 cells were rendered apoptotic by treatment with 1 M staurosporine for 2 h. K562 cells were treated for 18 h with camptothecin at 5 g/ml. Cell lysates were prepared for SDS/PAGE by boiling in Laemmli buffer at a concentration of 5-10 ϫ 10 6 cells/ml.
Immunoblotting-Cell extracts (approximately 2 ϫ 10 5 cells per extract) were analyzed by SDS-PAGE (Novex) and transferred to nitro-cellulose at 40 V for 16 h. The blots were first incubated in blocking buffer (Tris-buffered saline, pH 7.4 (TBS), 5% (w/v) milk (Blotting grade blocker non-fat dry milk; Bio-Rad), 0.05% (v/v) Tween 20) for 1 h at room temperature and then incubated for an additional hour in primary antibody diluted in blocking buffer. After washing three times in 1 ϫ TBS with 0.1% (v/v) Tween 20 for 5 min, blots were incubated for 1 h at room temperature in goat anti-mouse IgG coupled to horseradish peroxidase diluted 1:3000 in blocking buffer. Blots were washed three times in 1 ϫ TBS, 0.3% (v/v) Tween 20 for 5 min, and three times in 1 ϫ TBS, 0.1% (v/v) Tween 20 for 5 min. Detection was performed using enhanced chemiluminescence (ECL TM system from Amersham).
In Vitro Cleavage of Fak-[ 35 S]Methionine-labeled Fak was obtained by coupled in vitro transcription/translation using the Promega TNT reticulocyte lysate system. One g of the cDNA construct obtained by Qiaprep purification (Qiagen) was incubated with T7 polymerase, rabbit reticulocyte lysate, amino acid mixture minus methionine, and [ 35 S]methionine (Ci/mmol) for 1 h at 30 O C. Cleavage of the in vitro transcribed/translated radiolabeled product was performed by incubation at 37 O C in the presence of either apoptotic extract or purified recombinant human (rh)-caspases in cleavage buffer (50 mM Hepes/ KOH (pH 7.0), 2 mM EDTA, 0.1% (w/v) CHAPS, 10% (w/v) sucrose, 5 mM dithiothreitol). The final volume of the reaction was 25 l. The cleavage reaction was terminated by the addition of SDS Laemmli loading buffer and analyzed by SDS-PAGE and fluorography. Seventy-five g of cytosolic extracts prepared from apoptotic Jurkat T-cells lysed in cleavage buffer (see above) were used in the cleavage assays. Purified human caspases-1, -3, -6, and -8 and granzyme B were prepared as described previously (22). The protease inhibitor sensitivity profile was performed by preincubating the protease inhibitors with the apoptotic extracts for 20 min at room temperature prior to the addition of [ 35 S]methioninelabeled Fak generated by in vitro transcription/translation and purified by FPLC on a Superdex 75 column (Amersham Pharmacia Biotech). The proteolytic cleavage was quantitated by laser densitometry of the resulting fluorograms.
Kinetic Evaluation of Fak Cleavage-FPLC-purified radiolabeled Fak was incubated for 60 min at 37 O C with various concentrations of purified caspases in the cleavage buffer as described above. Reaction products were separated by SDS-PAGE, visualized by fluorography, and quantitated by phosphorimaging. All reactions were carried out using levels of substrate well below K m , where the appearance of product is a first-order process. Values for k cat /K m were calculated from the relationship S t /S o ϭ e Ϫkobs*t where S t is the concentration of substrate remaining at time t, S o is the initial substrate concentration, and k obs ϭ k cat *[enzyme]/K m .
Tyrosine Phosphorylation of Endogenous Fak Protein-HeLa cells were transiently transfected with FRNK, C 705-1053 , or C 773-1053 cDNAs in the pCep4␤ expression vector. Cells were collected by scraping 24 h after transfection, washed once in phosphate-buffered saline, and lysed in TNE buffer (50 mM Tris, pH 8.0, 1% Nonidet P-40, 2 mM EDTA) containing 10 g of each of the following protease inhibitors: leupeptin, aprotinin, N-tosyl-L-phenylalanine chloromethyl ketone, N ␣ -p-tosyl-Llysine chloromethyl ketone, and phenylmethylsulfonyl fluoride, as well as the phosphatase inhibitors sodium fluoride (50 mM) and sodium orthovanadate (1 mM). Protein concentrations were determined using the Bio-Rad protein assay. One mg of each lysate was used to dilute The cleavage reactions were carried out for 1 h at 37°C and visualized by 10% SDS-PAGE and fluorography. Arrows indicate full-length Fak (p125 Fak ) and cleaved fragments (p90, p35). 5-fold a 5 ϫ RIPA buffer stock (750 mM NaCl, 5% Nonidet P-40, 2.5% sodium deoxycholate, 0.5% SDS, 500 mM Tris, pH 8.0, 25 mM EDTA). Immunoprecipitation of endogenous Fak protein was carried out with 4 g of an antibody directed against residues 748 -1053 of human p125 Fak (Upstate Biotechnology). Immune complexes were collected on protein A-Sepharose beads, washed three times in 1 ϫ RIPA buffer containing 1 mM sodium orthovanadate, and eluted in Laemmli buffer by boiling. Tyrosine phosphorylation of endogenous Fak was visualized by SDS-4 -20% PAGE, transfer to nitrocellulose, and immunoblotting with the RC20 anti-phosphotyrosine antibody (Transduction Laboratories) using enhanced chemiluminescence (ECL TM system from Amersham Pharmacia Biotech).

RESULTS
Serum deprivation of chicken embryo fibroblasts was shown recently to result in the proteolysis of Fak into fragments of 70 -90 kDa (16). To identify the enzyme responsible for the cleavage of Fak during apoptosis, we set up an in vitro assay whereby [ 35 S]methionine-labeled chicken Fak generated by in vitro transcription/translation was incubated with an apoptotic extract derived from Jurkat T lymphocytic cells treated with an antibody to the CD95 (Fas/Apo-1) death receptor (these cells were chosen because their ability to undergo apoptotic cell death has been extensively characterized). Two fragments of 90 and 35 kDa were observed after incubation with apoptotic but not with non-apoptotic cell extracts (Fig. 1, lanes 1 and 2). The protease inhibitor sensitivity profile of Fak cleavage was characteristic of that of the caspase family of cysteine proteases in that proteolysis was abolished by the cysteine-alkylating reagent iodoacetamide (lane 21), but not by the cysteine protease inhibitor E-64 (lane 14) nor by serine, aspartate, or metalloprotease inhibitors (lanes 3-13 and 15-20).
To confirm that a caspase was responsible for the cleavage of Fak and to define more precisely the group of caspases in-  (p90, p35). B, quantitation of Fak cleavage, assessed by laser densitometric scanning of the 35-kDa proteolytic fragment, is expressed as a percentage of the control to which no drug was added. From this graph, the IC 50 values for the various tetrapeptide aldehydes were determined.

FIG. 3. Cleavage of Fak by rhcaspase-3 in vitro.
[ 35 S]Methionine-labeled Fak was incubated with various amounts of rh-caspase-3 for 1 h at 37°C and visualized by 10% SDS-PAGE and fluorography. Arrows indicate full-length Fak (p125 Fak ) and proteolytic fragments generated (p90, p35). Fak cleavage was measured as described in Fig. 2 to determine the k cat /K m value (see "Materials and Methods").
volved, we performed the cleavage assays in the presence of three different tetrapeptide aldehydes, each of which preferentially inhibits a caspase subgroup (22). The interpretation of the results was based on the known selectivity of these inhibitors for all 10 human recombinant caspases. Up to 10 M Ac-YVAD-CHO did not prevent the generation of the 90-and 35-kDa proteolytic fragments ( Fig. 2A, lanes 15-21), excluding the possibility that one of the Group I caspases (caspase-1, -4, -5; Ref. 22) was responsible for cleavage. In contrast, as little as 100 nM Ac-DEVD-CHO completely abolished Fak cleavage (lanes 1-7; IC 50 10 nM) whereas Ac-IETD-CHO inhibited Fak cleavage only when present at moderately high concentrations (lanes 8-14; IC 50 200 nM). The only caspase that has an inhibitor specificity consistent with these results is caspase-3, strongly suggesting that this enzyme is responsible for cleaving Fak in apoptotic extracts.
Caspase-3 is one of the most abundant Group II caspases in apoptotic Jurkat cell extracts. 2 It has been implicated directly in the cleavage of most of the proteins targeted for proteolysis during apoptosis. To assess whether caspase-3 cleaves Fak and whether this cleavage event is of physiological relevance, [ 35 S]methionine-labeled Fak was incubated with various amounts of purified recombinant human (rh) caspase-3. Fragments with molecular masses identical with those generated by the apoptotic Jurkat cell extract were produced when Fak was incubated with rh caspase-3 (Fig. 3), consistent with the inhibitor studies implicating a Group II caspase such as caspase-3. The k cat /K m for this substrate was 3.4 ϫ 10 5 M Ϫ1 s Ϫ1 , only slightly less than that for poly(ADP)-ribose polymerase (k cat /K m ϭ 15.6 ϫ 10 5 M Ϫ1 s Ϫ1 ; data not shown), the first identified and most extensively characterized substrate of caspase-3 (23,24).
Analysis of the fragments generated by the cleavage of [ 35 S]cysteine-labeled Fak by rh caspase-3 revealed the presence of the 90-kDa fragment but the absence of the 35-kDa fragment (data not shown). Because cysteine residues are absent from the carboxyl-terminal portion of Fak, we predicted that the 35-kDa fragment was derived from the carboxyl terminus. Analysis of the chicken Fak cDNA sequence revealed an excellent consensus cleavage site for caspase-3, DQTD 772 , present 282 amino acids from the carboxyl terminus of the protein.
To confirm that DQTD 772 was the site being recognized by caspase-3, we substituted the P 1 aspartic acid at position 772 for an alanine by site-directed mutagenesis (the presence of an aspartic acid in P 1 is absolutely required for cleavage by all caspases). [ 35 S]Methionine-labeled Fak carrying the D772A mutation was not cleaved by caspase-3 (Fig. 4A, lane 4) under conditions where the conversion of wild-type Fak to 90-and 35-kDa fragments occurred (lane 2).
To confirm that the aspartic acid at position 772 was being cleaved in cells undergoing apoptosis, we subcloned wild-type and mutant Fak (D772A) into the eukaryotic expression vector pCEP4␤ and established polyclonal stable K562 lymphoid cell lines expressing these proteins. The proteins were engineered with a Flag-epitope tag at the carboxyl terminus to allow visualization of the carboxyl-terminal fragments released by proteolysis. Large amounts of the wild-type (wt) and D772A mutant proteins were expressed in these stable cell lines, as assessed by immunoblot analysis using a monoclonal antibody to the Flag epitope (Fig. 4B, lanes 1 and 3). When the cells were rendered apoptotic by treatment with the topoisomerase inhibitor camptothecin, Fak cleavage occurred, as visualized by the appearance of the Flag-tagged 35-kDa carboxyl-terminal fragment (lane 2). As expected, the 35-kDa fragment was not detected in cells expressing the D772A mutant Fak, confirming that cleavage at Asp 772 was responsible for the liberation of the 35-kDa fragment. Instead, low amounts of a fragment of 40 kDa were detected in apoptotic K562 cells expressing the D772A mutant Fak (lane 4). The corresponding 85-kDa fragment was visualized by Western blot analysis of extracts from apoptotic cells expressing a D772A mutant protein with a Flagepitope tag at the amino terminus (data not shown). We predicted that the potential cleavage site VSWD 704 present upstream of the aspartic acid at position 772 was also being cleaved under these conditions. This was confirmed by the absence of Fak cleavage in camptothecin-treated K562 cells expressing a Fak mutant whereby the aspartic acids at both positions 704 and 772 were mutated to alanines (lanes 5 and 6).
Cleavage assays using recombinant human caspase-3 excluded the possibility that the VSWD 704 site was recognized by caspase-3 (Fig. 3). The substrate specificity profile of the caspase family members (22) suggested that VSWD 704 represented a good consensus site for the Group III activator caspases and a poor substrate for Group II enzymes such as caspase-3. Although one of the Group III enzymes, caspase-8, did not cleave at VSWD 704 , caspase-6 cleaved both VSWD 704 and DQTD 772 (Fig. 5) respectively. The serine protease granzyme B, which is present in the granules of cytotoxic T lymphocytes and has been shown to have a very similar substrate specificity to that of Group III activator caspases (22), also cleaved these sites with k cat /K m values similar to those determined for caspase-6 (k cat /K m for VSWD 704 ϭ 3.1 ϫ 10 4 M Ϫ1 s Ϫ1 ; k cat /K m for DQTD 772 ϭ 8.9 ϫ 10 4 M Ϫ1 s Ϫ1 ). Although both caspase-3 and caspase-6 cleave DQTD 772 , it is noteworthy that the catalytic efficiency of caspase-3 for this site is approximately 6 times greater than that of caspase-6, suggesting that caspase-3 (or another Group II effector caspase) is more likely to be the protease responsible for cleavage at this site in vivo.
The physiological significance of the presence of two distinct caspase cleavage sites in Fak was underscored by examining the cleavage of endogenous protein in human cell lines undergoing apoptosis (both DQTD 772 and VSWD 704 are conserved in human Fak). Total cell lysates were prepared from Jurkat cells at various times after stimulation of the CD95 (Fas/Apo-1) death receptor by antibodies to CD95 (Fas/Apo-1) (Fig. 6A). As early as 2 h after the addition of anti-CD95 (Fas/Apo-1), Fak cleavage to a 90-kDa fragment was detected by Western blot analysis using two different antibodies to Fak, one directed against residues 748 to 1053 (␣748 -1053) and one directed against residues 354 -533 (␣354 -533). The antibody raised against residues 354 -533 recognized an additional fragment of 85 kDa. These results are consistent with the cleavage of Fak at both VSWD 704 and DQTD 772 sites in apoptotic Jurkat cells (see Fig. 6C). Curiously, only the 90-kDa fragment was generated in the human cervical carcinoma cell line HeLa in response to treatment for 2 h with staurosporine, a nonspecific kinase inhibitor that induces apoptosis (Fig. 6B). A similar result was obtained when we induced apoptosis by serum starvation or by treatment with camptothecin (data not shown). We confirmed that only DQTD 772 is recognized in apoptotic HeLa cells by demonstrating that cleavage of transiently transfected chicken Fak was abolished when the P 1 aspartic acid at position 772 was substituted for alanine without appearance of fragments corresponding to cleavage at VSWD 704 (data not shown). The absence of cleavage at VSWD 704 may be due to the absence of caspase-6 (or another caspase capable of cleaving VSWD 704 ) or to the inaccessibility of VSWD 704 in HeLa cells.
Cleavage at either DQTD 772 or VSWD 704 results in the separation of the kinase domain from the focal adhesion targeting (FAT) domain. Noteworthy is the fact that the naturally occurring inhibitor of Fak, FRNK, corresponds to the carboxyl-terminal half of Fak, as illustrated in Fig. 6D. When overexpressed, FRNK prevents the localization of Fak to sites of integrin engagement resulting in decreased Fak phosphorylation and inhibition of Fak-mediated cellular responses to cell adhesion (25). In view of the presence of an intact FAT domain within the 35-and 40-kDa fragments generated by caspase cleavage of Fak during apoptosis, we reasoned that these fragments may also act as competitive inhibitors of Fak. To test this hypothesis, Fak deletion mutants identical with the carboxyl-terminal fragments generated during apoptosis were engineered and transiently transfected into HeLa cells. The levels of phosphorylated Fak protein were measured by immunoprecipitation with an anti-Fak antibody and immunoblotting with an anti-phosphotyrosine antibody 24 h after transfection (Fig.  7). Expression of either the 35-(lane 4) or the 40-kDa fragment (lane 3) suppressed the phosphorylation of endogenous Fak, as was observed after transfection of FRNK (lane 2). Like FRNK, these fragments acted as competitive inhibitors because their inhibitory effects were abrogated by overexpressing full-length Fak protein (data not shown).
If the disabling of Fak by caspase cleavage represents a critical step in the execution of the apoptotic program, the two cleavage sites identified should be conserved in other species. Whereas the caspase-6 cleavage site VSWD 704 is present in all species examined (human, rat, mouse, chicken, frog; Fig. 8), the caspase-3 cleavage site DQTD 772 is present in chicken and human but not rat or mouse ( Fig. 8; the DHMD 772 sequence present in frog Fak is predicted to be cleaved efficiently by a Group II effector caspase). Given that one of the caspase cleavage sites (DQTD 772 ) is absent in mouse Fak, we examined whether Fak cleavage occurred in murine cells in response to an apoptotic stimulus. Lysates were prepared from staurosporine-treated 3T3 L1 fibroblasts and analyzed by immunoblot-  4). C 705-1053 and C 773-1053 correspond exactly to the carboxyl-terminal fragments generated by caspase cleavage at VSWD 704 and DQTD 772 , respectively, except that an initiating methionine was introduced immediately upstream of the P 1 ' amino acid. Twenty-four h after transfection, cells were lysed, and tyrosine-phosphorylated Fak was detected by immunoprecipitation using an antibody directed against the carboxyl terminus of Fak followed by immunoblotting using the antiphosphotyrosine antibody RC20 (␣PY; Transduction Laboratories). In parallel, total amounts of Fak were assessed by immunoprecipitation as described above followed by immunoblotting with an anti-Fak antibody (␣Fak; anti-kinase domain from Transduction Laboratories). ting using the ␣ 354 -533 antibody that recognizes both 85-and 90-kDa fragments in human Fak (Fig. 6B). Neither 85-nor 90-kDa fragments of Fak could be detected in apoptotic murine cells. Instead, murine Fak was cleaved at an upstream site, the identity of which remains to be determined, which results in the appearance of a 75-kDa fragment (we have not excluded the possibility that murine Fak is cleaved first at VSWD 704 and then at an upstream site). DISCUSSION The cleavage of Fak observed in both adherent and suspension cell lines in response to various apoptotic inducers suggests that this proteolytic event plays an important role in the execution of the suicide pathway. Preliminary results by Wen et al. (17) and Levkau et al. (18) demonstrated recently the cleavage of Fak by caspases during apoptosis. We have identified two caspase cleavage sites within the carboxyl-terminal half of Fak; DQTD 772 which is preferentially cleaved by the group II effector caspase, caspase-3, and VSWD 704 , which is preferentially cleaved by the group III activator caspase, caspase-6. Although caspase-3 does not recognize VSWD 704 , caspase-6 cleaves both DQTD 772 and VSWD 704 with virtually identical kinetics (k cat /K m ϭ 5.9 ϫ 10 4 M Ϫ1 s Ϫ1 and 6.5 ϫ 10 4 M Ϫ1 s Ϫ1 , respectively). The relative promiscuity of caspase-6 versus caspase-3 is consistent with results obtained from a combinatorial tetrapeptide library used to define the substrate specificity of the caspase family members (22). This combinatorial approach revealed that in the critical specificity-determining P 4 subsite, caspase-6 tolerates a number of amino acids whereas caspase-3 exhibits a very strict requirement for aspartic acid.
Our studies underscore the importance of determining the catalytic efficiency of cleavage for a given substrate with a given caspase. Indeed, although caspase-6 can cleave DQTD 772 (k cat /K m ϭ 5.9 ϫ 10 4 M Ϫ1 s Ϫ1 ), this site is clearly preferred by caspase-3 (k cat /K m ϭ 3.4 ϫ 10 5 M Ϫ1 s Ϫ1 ). Of interest is our observation that whereas DQTD 772 is cleaved in all human cell lines examined, presumably because Group II effector caspases are present in these cell lines, cleavage at VSWD 704 appears to be cell type-dependent, possibly because Group III activator caspases capable of cleaving this site are not ubiquitously expressed. It is of interest that the caspase cleavage sites lie on either side of a proline-rich region shown to interact with two SH3-containing proteins: Graf (a Rho and Cdc42 GTPase-activating protein; Ref. 26) and p130 Cas (an adapter molecule phosphorylated by various tyrosine kinases; Ref. 27). It remains to be determined whether protein-protein interactions modulate the sensitivity of Fak to caspase cleavage.
It is curious that the predominant caspase-3 consensus cleavage site identified in chicken and human Fak is absent in rodent Fak. Species alignments for several identified caspase substrates have revealed that in all cases examined, the P 1 aspartic acid in the caspase cleavage site was conserved. It remains to be determined whether the absence of this predominant caspase cleavage site in rodent Fak modulates the cellular responses to cell adhesion. Of interest is the report that the murine cell line, NIH3T3, is relatively resistant to anoikis (9).
Cleavage at either DQTD 772 or VSWD 704 generates carboxylterminal fragments that inhibit Fak phosphorylation and thus act like FRNK, the naturally occurring variant of Fak (Fig. 6D). Results presented in this report suggest two mechanisms by which Fak-mediated cellular responses to cell adhesion are abrogated during apoptosis: (i) by decreasing the overall amount of Fak in the cell and (ii) by generating fragments that act as competitive inhibitors of the remaining full-length Fak protein. That the cell has devised two mechanisms to inactivate Fak underscores the importance of this cleavage event in the execution of apoptotic cell death. We propose that the disabling of Fak liberates the cell from anti-apoptotic signals generated by the extracellular matrix and allows removal of the apoptotic cell from the tissue.