Talin phosphorylation sites mapped by mass spectrometry

Talin, a 270 kDa protein containing 2541 amino acids, forms a direct link between integrins and the cytoskeleton (Critchley, 2004; Critchley, 2000). Protein domains of talin include an N-terminal globular head and a C-terminal rod. The globular head contains sequences that resemble the FERM domain found in the ERM (exrin, radixin and moesin) family of proteins (Rees et al., 1990). FERM domains contain three subdomains (F1, F2 and F3). The major integrin-binding site of talin lies within the F3 subdomain and involves a PTB (phosphotyrosine-binding) domain that interacts with the NPxY motif present in the cytoplasmic tails of the integrin β subunit (Calderwood et al., 2002; Garcia-Alvarez et al., 2003). The F3 subdomain of talin also binds type I PIP kinase γ (phosphatidylinositol-4P 5-kinase) at a site whose sequence overlaps with that found in the cytoplasmic tails of the integrin β subunit (Barsukov et al., 2003; Di Paolo et al., 2002; Ling et al., 2002; de Pereda et al., 2004). Binding of the integrin cytoplasmic tails to the FERM domain of talin induces conformational changes in the integrin extracellular domains, thus increasing their affinity for ligands (Calderwood et al., 2004; Tadokoro et al., 2003; Calderwood et al., 2002; Kim et al., 2003). Conversely, PIP kinase γ inhibits integrin binding to talin F3 (Barsukov et al., 2003; Di Paolo et al., 2002; Ling et al., 2002). The talin FERM domain also contains a binding site for focal adhesion kinase (FAK), a non-receptor tyrosine kinase involved in integrin outside-in signaling, and layilin, a hyaluronon-binding membrane protein (Bono et al., 2001; Borowsky and Hynes, 1998). Both the FERM and rod domains of talin possess binding sites for actin (Hemmings et al., 1996; Senetar et al., 2004; McCann and Craig, 1997; Lee et al., 2004). Binding sites on the rod domain of talin include one thought to be involved in homodimerization (Molony et al., 1987; Muguruma et al., 1995), three for interaction with the cytoskeletal protein, vinculin (Bass et al., 2002; Bass et al., 1999; Izard et al., 2004; Papagrigoriou et al., 2004), and one for a second interaction with integrins (Xing et al., 2001; Tremuth et al., 2004).

Phosphorylation of the FERM family of proteins regulates their interactions with transmembrane proteins (Tsukita and Yonemura, 1999). Talin is also phosphorylated (Qwarnstrom et al., 1991; Tidball and Spencer, 1993; Turner et al., 1989; Watters et al., 1996; Beckerle et al., 1989) and the integrin-binding site (Yan et al., 2001; Martel et al., 2001) and at least one vinculin-binding site (Papagrigoriou et al., 2004) in talin are masked. This suggests that conformational rearrangements in talin could be required for integrin activation and interactions with vinculin in a manner analogous to other ERM proteins. Talin is an in vitro substrate of protein kinase C (PKC) and is phosphorylated on both serine and threonine residues in blood platelets (Litchfield and Ball, 1990; Litchfield and Ball, 1986; Murata et al., 1995). These observations point to the possibility that talin phosphorylation could be of importance for regulating its biological functions such as integrin activation. Talin is also cleaved between the head and the rod domains by the enzyme calpain. This enhances the binding of talin to the β3 integrin subunit cytoplasmic tail (Yan et al., 2001). Phosphorylation of talin could be a mechanism for altering the availability of the cleavage site to calpain. In addition, given the number of protein partners that talin interacts with, talin phosphorylation could control when and where these interactions occur. Here we use a combination of immobilized metal affinity chromatography and tandem mass spectrometry to map the phosphorylation sites of talin isolated from human blood platelets.

We chose to isolate talin from platelets because it is highly abundant in this source and easily separated from other platelet proteins by SDS-PAGE. Talin gives rise to a complex mixture of peptides upon digestion with proteolytic enzymes. To cover 95.1% of the talin sequence and 96.4% of the serine, threonine and tyrosine residues, we performed three separate digests with three different combinations of enzymes (Lys C/trypsin, Lys C/Asp-N and Lys C/GluC [for data, see the Cell Migration Gateway (http://www.cellmigration.org/resource/discovery/proteomics/talin_total.shtml)].

We used cell-permeable phosphatase inhibitors, calyculin A and sodium peroxivanadate to suppress dephosphorylation of the serine/threonine and tyrosine residues, respectively, and, thus, to increase the abundance of the corresponding phosphopeptides. Platelets were pretreated with the phosphatase inhibitors and stimulated with agonist (TRAP) for 30 seconds. Only three sites of phosphorylation, T144, T150 and S446, were detected when the mixture of peptides from each of the three enzymatic digests of talin was analyzed directly by tandem mass spectrometry. Stoichiometries for each these three sites were in the range of 40-60%. The location of these high stoichiometry sites in relation to the domain structure of talin is shown in Fig. 1. TRAP is the N-terminal peptide sequence liberated as a consequence of cleavage of PAR-1, the major human platelet thrombin receptor. It is a potent agonist of platelets, mimicking the action of thrombin. PAR-1 is a G-protein-coupled receptor that activates G_q, G₁₂ and perhaps G_i family members, which leads to the activation of PLCβ, PI 3-kinase and the monomeric G proteins Rho, Rac and Rap1, causes an increase in the cytosolic Ca²⁺ concentration and inhibits cAMP formation (Brass, 2003). The phosphorylations reported were an aggregate of those observed in talin from platelets stimulated for 5, 10, 30 and 60 seconds with TRAP.

Subsequently, we employed immobilized metal affinity chromatography (IMAC) to enrich the above mixtures for phosphopeptides and then repeated the analysis by tandem mass spectrometry. The 30 phosphorylation sites detected by this approach are shown in Table 1. Note that the additional 27 sites have stoichiometries that are 10-1000-fold lower than the sites detected without enrichment by IMAC. We used NetPhos 2.0 (http://www.cbs.dtu.dk/services/NetPhos/) and Scansite (http://scansite.mit.edu/) to predict the kinases that might be responsible for phosphorylation of particular sites (Table 1). The identified sites are potential targets for kinases involved in a variety of physiological processes. For example, T144, T150 and S425 are potentially phosphorylated by protein kinase C (PKC), whereas S446 is predicted to be phosphorylated by PKA, although with a low score by NetPhos (Table 1). S425 can also be potentially phosphorylated by CDK5.