INTRODUCTION

Cortactin, an F-actin¹ binding and cross-linking protein (1, 12), is a major target for tyrosine phosphorylation in response to signaling mediated by fibroblast growth factor (2), epidermal growth factor (3), integrin activation (4), bacteria-mediated phagocytosis (5), and v-src oncogene (6). Overexpression or amplification of the human cortactin gene (also called EMS1) is often associated with human malignancies (7, 8). In v-Src-transformed cells, cortactin has been found to co-localize with Src oncoproteins within podosomes, membrane-substratum contact structures (6). Analysis of cortactin phosphorylation in cells lacking the c-src gene (9) or following overexpression of c-Csk (10), a negative regulator for pp60^c-src, has provided further compelling evidence that cortactin is an intrinsic substrate for pp60^c-src.

The protein sequence of cortactin is unique because it contains six and one-half 37-amino acid tandem repeats near the NH₂ terminus, and a Src homology 3 (SH3) domain at the carboxyl-terminal end. Between the repeat and the SH3 domain is an -helix, a proline-rich region, and multiple tyrosine residues. The amino acid sequence of human cortactin within the repeat domain shares nearly 100% identity with the chicken and murine homologues and 70% with HS1, a cortactin-related gene product (11), indicating that the repeat domain plays a fundamental role for cortactin (12). Indeed, the repeat domain has been demonstrated as the binding site for F-actin (12). In contrast, the sequence between the -helix and the SH3 domain exhibits less than 33% identity to HS1, but the function of this region has not yet been identified.

We recently reported that there is abundant expression of cortactin in megakaryocytes and platelets (13). While tyrosine phosphorylation of cortactin has been described as a major phenomenon in thrombin-stimulated platelets (14, 15), the significance of the tyrosine phosphorylation is unknown. In the present study, we examined the fate of cortactin in platelets stimulated by collagen. We found that cortactin is degraded following tyrosine phosphorylation and that the protease responsible for the cortactin degradation is a calpain-related enzyme, which requires integrin _IIb3. Furthermore, we provide in vitro evidence that the sequence containing the proline-rich region and multiple tyrosine residues targeted by pp60^c-src is required for the calpain-mediated cleavage. Finally, we demonstrated that tyrosine phosphorylation of cortactin by pp60^c-src dramatically alters its susceptibility to calpain. These data suggest that tyrosine phosphorylation may play a role in the calpain-mediated proteolysis of cortactin.

EXPERIMENTAL PROCEDURES

Monoclonal antibody (mAb) 4F11 was purchased from Upstate Biotechnology, Inc. (Lake Placid, NY). Monoclonal antibody against phosphotyrosine (RC20) was from Transduction Laboratories (Lexington, KY). The polyclonal antibody against the C-terminal part of murine cortactin was derived from mice immunized with a recombinant protein corresponding to amino acids 323-546. Prostaglandin E₁, phenylmethylsulfonyl fluoride, Triton X-100, EGTA, benzamidine, leupeptin, and aprotinin were from Sigma. SDS-PAGE markers were from Bio-Rad. Protein A-Sepharose was from Pharmacia Biotech Inc. Purified µ-calpain derived from pig erythrocytes was obtained from ICN (Costa Mesa, CA). Calpeptin was from Biomol (Plymouth Meeting, PA). Sodium orthovanadate was from Fisher. Type I tendon collagen was from Chrono-Log Co. (Havertown, PA).

Human blood (500 ml) from healthy volunteers was collected into 70 ml of CPD solution, containing 1.84 mg of sodium citrate, 1.78 mg of dextrose, 209 mg of citric acid, and 155 mg of monobasic sodium phosphate. Platelet-rich plasma was obtained by centrifugation at 200 × g for 16 min at ambient temperature. Blood from a Glanzmann's patient (female) was kindly provided by Robert Abel (Christina Hospital, Wilmington, DE). Citric acid and prostaglandin E₁ were added to platelet-rich plasma to final concentrations of 4 mM and 1 µg/ml, respectively. The platelet-rich plasma was then centrifuged at 700 × g for 10 min. The platelet pellet was resuspended in washing buffer (4.26 mM NaH₂PO₄, 7.46 mM Na₂HPO₄, pH 6.5, containing 5.5 mM dextrose, 128 mM NaCl, 4.77 mM sodium citrate, 2.35 mM citric acid, and 3.5 mg/ml of bovine serum albumin) and centrifuged at 700 × g for 10 min. The pellet was then resuspended in a modified Tyrode-Hepes buffer (10 mM Hepes, pH 7.35, containing 136.7 mM NaCl, 5 mM glucose, 2.6 mM KCl, 13.8 mM NaHCO₃, 1.0 mM MgCl₂, 0.36 mM NaH₂PO₄, and 3.5 mg/ml bovine serum albumin) at 1 × 10⁹ platelets/ml.

Collagen at a final concentration of 2.5 µg/ml was added to the washed platelets (0.6 ml of 1 × 10⁹ cells/ml) in the presence of 1 mM CaCl₂ in an aggregometer cuvette at 37 °C for the times indicated. Activated platelets were immediately lysed by adding 200 µl of 4 × Triton lysis buffer (200 mM Tris-HCl, pH 7.2, containing 4% Triton X-100, 20 mM EGTA, 40 µg/ml leupeptin, 40 µg/ml aprotinin, 4 mM phenylmethylsulfonyl fluoride, 4 mM benzamidine, and 4 mM Na₃VO₄). The lysates were centrifuged at 15,000 × g for 10 min. The pellet (insoluble fraction) was solubilized by adding an equal volume of 2 × SDS sample buffer (16). The soluble fractions were subjected to immunoprecipitation by mAb 4F11 (2.5 µg/ml) as described previously (2). The immunoprecipitates were washed once with 1 × Triton-lysis buffer, resuspended in 2 × SDS sample buffer, and analyzed by immunoblotting analysis with either mAb 4F11 or RC20 as described previously (2).

Stirred platelets (0.4 ml of 1 × 10⁹ cells/ml) were stimulated with collagen at 37 °C and lysed by adding an equal volume of 2 × SDS sample buffer including 10 mM EGTA and 20 µM calpeptin. The platelet proteins were separated by SDS-PAGE, and cortactin was detected by immunoblot analysis.

To prepare GST-cortactin, a DNA fragment of 182 bp was generated by polymerase chain reaction. The oligonucleotide ACTCGTGGATCCTGGAAAGCCTCTGCA was used as the 5 primer and contained a BamHI site; the oligonucleotide CTTGAGCGTCTGGTGTT was used as the 3 primer and contained an Xmn1 site. The amplified fragment was ligated with a DNA fragment derived from the digestion of a cDNA clone encoding the murine cortactin (2) with Xmn1 and EcoRI, and the ligated product was then cloned into BamHI and EcoRI sites of PGEX-2T (Pharmacia). The cortactin variant Cort_496-546 was prepared as follows. A DNA fragment of 370 base pairs was prepared by polymerase chain reaction. The oligonucleotide CGAGAGAGCTCAGCGGATGGCC was used as the 5 primer and contained a SacI site; the oligonucleotide ACTGCAGAATTCTAGATGGCTGTGATGCC was used as the 3 primer and contained an EcoRI site and a stop codon. The amplified fragment was cloned into a DNA fragment derived from the digestion of GST-cortactin with SacI and EcoRI. Cort_375-546 was prepared in the same way as Cort_496-546 except that the oligonucleotide CGAGAGAGCTCAGTAACGGATCGCCAAAGAA was used as the 5 primer and contained a stop codon and a SacI site. All polymerase chain reaction-generated fragments were confirmed by DNA sequencing. Cortactin and its variants were expressed in Escherichia coli as glutathione S-transferase fusion proteins and purified by affinity chromatography using glutathione-Sepharose as described previously (17). The purified glutathione S-transferase fusion proteins were further digested with thrombin, and the glutathione S-transferase-free proteins were purified using glutathione-Sepharose and DEAE-Sepharose (1).

Purified recombinant cortactin (3.6 µg) was incubated with µ-calpain for 90 min at different concentrations in 40 µl of reaction buffer (50 mM Tris-HCl, pH 7.36, containing 134 mM KCl, 1 mM MgCl₂, 75 µM EGTA, and 75 µM CaCl₂). The reaction was terminated by adding an equal volume of 2 × SDS sample buffer, and the proteins were separated by a gradient SDS-PAGE gel (4-20%, w/v). The digested proteins were visualized by either Coomassie Blue staining or immunoblotting with mAb 4F11 or a polyclonal antibody directed against a peptide encoding the amino acid sequence from the -helix to the SH3 domain.

RESULTS

To evaluate the role of cortactin in platelet aggregation, we examined tyrosine phosphorylation of cortactin in collagen-stimulated platelets. Activated platelets were lysed using a Triton X-100-containing buffer. The soluble fractions were subjected to immunoprecipitation with 4F11, a mAb recognizing the repeat domain of cortactin (6). The pellets were solubilized in SDS sample buffer. Proteins in both fractions were immunoblotted using either a polyclonal antibody against cortactin or a mAb against phosphotyrosine. As shown in Fig. 1A, stimulation of stirred platelets with collagen caused a dramatic increase in the level of tyrosine phosphorylation of cortactin after 15 s and a maximum phosphorylation at 45 s, which was concomitant with platelet aggregation (data not shown). However, the level of phosphorylated cortactin declined slightly after 1 min of stimulation, and this coincided with a decrease in the level of cortactin in the soluble fraction (Fig. 1B).

We examined the possibility that a reduced amount of cortactin in the soluble fraction in response to collagen could be a consequence of the cytoskeletal translocation that has been described previously in thrombin-stimulated platelets (15, 18). Immunoblot analysis of cortactin in the insoluble fraction demonstrated that the stimulation of platelets with collagen enhanced tyrosine phosphorylation of multiple proteins including those that migrated at the positions for cortactin (Fig. 1C). However, the amount of cortactin associated with the insoluble fraction was only transiently increased during the period from 30 to 45 s and diminished afterward (Fig. 1D), suggesting that platelet cortactin, in either the soluble or insoluble fractions, was degraded after collagen stimulation. The degradation of cortactin appears not caused by a nonspecific proteolysis because pp60^c-src associated with the pellets was not degraded under the same conditions even after a prolonged stimulation (Fig. 1D).

To confirm that the apparent degradation of cortactin was not the result of a protease released during the Triton-mediated lysis, we analyzed cortactin in platelet lysates that were prepared by direct lysis in SDS-sample buffer. The results from these experiments were compared with the pattern of cortactin degradation prepared in Triton X-100 buffer. As shown in Fig. 2A, significant amount of degraded cortactin was detected in the SDS-lysed whole platelets after collagen stimulation, although the extent of the degradation, especially at early phases of stimulation (30 and 45 s), appeared to be less than that of Triton-lysed platelets. However, the degradation patterns in both lysates are similar (Fig. 2B).

Calpain is a family of calcium-dependent cysteine proteases that are abundantly present in platelets and are activated during platelet aggregation (19, 20). As shown in Fig. 3A, EGTA treatment of platelets significantly inhibited the degradation of cortactin as compared with untreated platelets. Furthermore, treatment with calpeptin, a specific membrane-permeable peptide-derivative inhibitor for calpain (21), resulted in the same reduction of cortactin degradation (Fig. 3B, part a). However, when a lysis buffer containing either EGTA or calpeptin was used to lyse activated platelets, no significant inhibition of cortactin degradation was observed (Fig. 3B, parts b and c). This result further confirms that the degradation of cortactin primarily occurs prior to platelet lysis.

Degradation of Cortactin in Platelets Is Dependent on _IIb3

In platelets, the influx of calcium can be regulated by the activation of _IIb3 (22, 23), a major integrin on the surface of platelets. To evaluate the role of _IIb3 in the proteolysis of cortactin, we examined tyrosine phosphorylation of cortactin in platelets from a Glanzmann's patient. As shown in Fig. 4, normal platelets exhibited a 60% reduction in the amount of intact cortactin after 3 min of collagen stimulation. In contrast, no significant reduction was found with the Glanzmann's platelets, suggesting that the degradation of cortactin requires _IIb3 under identical conditions. Interestingly, the induction of tyrosine phosphorylation of cortactin in response to collagen in the Glanzmann's platelets was not impaired (Fig. 4), implying that tyrosine phosphorylation of cortactin is a process independent of _IIb3.

Proteolysis of platelet cortactin requires _IIb3.

Purified µ-calpain (calpain-I) digests recombinant murine cortactin in vitro in a dose-dependent manner (Fig. 5A). At a concentration of 6.2 µg/ml of calpain, approximately 90% of the cortactin proteins were digested to multiple fragments. Interestingly, many of the digested fragments were reactive to mAb 4F11, which specifically recognizes the repeat domain of cortactin (12), but not to an antibody directed against the region between the repeat and the carboxyl terminus (Fig. 6A). This implies that the sequence in this region may be more susceptible to calpain. To verify this, we analyzed two cortactin variants, Cort_496-546, which lacks the SH3 domain, and Cort_375-546, which lacks the sequence from the proline-rich region to the carboxyl terminus (Fig. 6B, upper part). As with the wild-type cortactin, the mutant Cort_496-546 was efficiently digested by calpain (Fig. 6B, lower part). In contrast, little digestion of the mutant Cort_375-546 was detected under the same conditions, indicating that the sequence of amino acids 375-496, which contains the proline-rich region and multiple tyrosine residues, may be involved in the calpain-mediated proteolysis.

Amino acids 375-496 contain multiple tyrosine residues that can be targeted for phosphorylation by pp60^c-src.² Thus, we performed a calpain digestion of cortactin phosphorylated by pp60^c-src. Fig. 7 shows that most phosphorylated cortactin proteins were digested nearly completely within 2 min. In contrast, significant amounts of nonphosphorylated cortactin remained even after 20 min of digestion under the same conditions. However, when partially digested phosphorylated cortactin was analyzed by SDS-PAGE and compared with nonphosphorylated cortactin, we did not observe any significant difference in the two patterns (Fig. 7). Therefore, it is likely that tyrosine phosphorylation enhances the efficiency of calpain-mediated digestion without altering its cleavage sites.

DISCUSSION

It is unclear whether platelet calpain-mediated proteolysis occurs during lysis of cells or within aggregated platelets (24, 25). Our data indicate that the proteolysis of cortactin occurs within activated platelets. We detected the degradation of cortactin in whole platelets prepared by direct lysis in a SDS-sample buffer (Figs. 2 and 4). In addition, the calpain inhibitors EGTA and calpeptin block the degradation when they are applied before but not after platelet activation. Finally, it appears that the calpain-mediated proteolysis of cortactin is not the result of a nonspecific proteolysis, because pp60^c-src, another substrate for calpain (26), was not degraded under the same conditions that allow cortactin proteolysis (Fig. 1D). However, it should be pointed out that the degree of the proteolysis of cortactin in activated platelets appears to vary depending on the method of lysing platelets. There is more extensive degradation found in Triton-lysed platelets than in SDS sample buffer (Fig. 2). This may be due to the fact that the soluble cortactin becomes more vulnerable to calpain released after lysis.

In agreement with other reports (15, 18), we observed that cortactin undergoes a transient translocation into the Triton-insoluble fraction between 30 and 45 s after collagen stimulation (Fig. 1D); however, the role of the translocation in this proteolysis of cortactin is not clear. While cortactin is a potent F-actin binding and cross-linking protein, the presence of F-actin does not apparently change the efficiency of the proteolysis of cortactin in vitro (data not shown). Furthermore, a cortactin mutant able to bind to F-actin but lacking the sequence from the proline-rich region to the carboxyl terminus is not efficiently digested by calpain (Fig. 6B). Hence, it is unlikely that the F-actin binding is a rate-limiting step for the calpain digestion.

Many cytoskeleton-associated proteins have been reported to be substrates for calpain. These include actin-binding proteins (27), vitronectin (28), protein-phosphotyrosine phosphatase 1B (29), integrin ₃ subunit (30), talin (27), spectrin (31), and protein kinase C (32). As with many of those substrates, proteolysis of cortactin appears to be dependent on _IIb3 because it does not occur in Glanzmann's platelets lacking functional _IIb3 (Fig. 4). We have found, however, that the absence of _IIb3 does not affect collagen-induced tyrosine phosphorylation of cortactin. Our finding is in agreement with a previous report, which also showed increased tyrosine phosphorylation in thrombin-treated platelets derived from Glanzmann's patients (15). Furthermore, tyrosine phosphorylation of cortactin can be detected after 15 s of stimulation (Fig. 1A). This is prior to platelet aggregation, which occurs 30-45 s after stimulation. Thus, tyrosine phosphorylation of cortactin is kinetically correlated with the activation of pp60^c-src, which occurs in the early phase prior to the activation of _IIb3 during platelet stimulation (33). These data indicate that the Src-mediated tyrosine phosphorylation of cortactin could be involved in the calpain-mediated digestion. The importance of Src in the digestion of cortactin is further highlighted by our findings that the digestion of recombinant cortactin by µ-calpain is dependent on the presence of a sequence containing multiple tyrosine residues targeted by pp60^c-src (Fig. 6), and the efficiency of the digestion of cortactin in vitro is dramatically increased by pp60^c-src (Fig. 7).

Calpain-digested cortactin in vitro has significantly less F-actin cross-linking activity (data not shown). Interestingly, the F-actin cross-linking activity can be also down-regulated by tyrosine phosphorylation without degradation (1). These dual mechanisms regulating cortactin may be required to ensure the irreversible shape change associated with activated platelets. It is also noteworthy that both mechanisms involve the same structural region between the proline-rich motif and the SH3 domain. This may suggest the importance of this region in the regulation of cortactin function. Since calpain and cortactin are widely expressed in many mammalian cells, future studies using a structure-function approach should reveal the significance of calpain-mediated cleavage of cortactin in cellular cytoskeletal reorganization.