Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Structure and regulation of Src family kinases

Abstract

Src family kinases are prototypical modular signaling proteins. Their conserved domain organization includes a myristoylated N-terminal segment followed by SH3, SH2, and tyrosine kinase domains, and a short C-terminal tail. Structural dissection of Src kinases has elucidated the canonical mechanisms of phosphotyrosine recognition by the SH2 domain and proline-motif recognition by the SH3 domain. Crystallographic analysis of nearly intact Src kinases in the autoinhibited state has shown that these protein interaction motifs turn inward and lock the kinase in an inactive conformation via intramolecular interactions. The autoinhibited Src kinase structures reveal a mode of domain assembly used by other tyrosine kinases outside the Src family, including Abl and likely Tec family kinases. Furthermore, they illustrate the underlying regulatory principles that have proven to be general among diverse modular signaling proteins. Although there is considerable structural information available for the autoinhibited conformation of Src kinases, how they may assemble into active signaling complexes with substrates and regulators remains largely unexplored.

The domain structure of Src kinases

Src family nonreceptor tyrosine kinases are present in essentially all metazoan cells, where their regulated activation by diverse growth factor, cytokine, adhesion, and antigen receptors is critical for generating an appropriate cellular response to external stimuli (Brown and Cooper, 1996; Thomas and Brugge, 1997). The nine members of the Src family include Src, Lck, Hck, Fyn, Blk, Lyn, Fgr, Yes, and Yrk. Src kinases share a conserved domain structure consisting of consecutive SH3, SH2, and tyrosine kinase (SH1) domains (Figure 1). All family members also contain an SH4 membrane-targeting region at their N-terminus, which is always myristoylated and sometimes palmitoylated (Koegl et al., 1994; Resh, 1999). The SH4 region is followed by a ‘unique’ domain of 50–70 residues, which is divergent among family members.

Figure 1
figure1

The domain structure of Src family kinases. The Src kinase architecture consists of four domains: the unique region, which varies among family members, followed by the SH3, SH2, and tyrosine kinase domains. The approximate extent of each domain is indicated, with the unique, SH3, SH2, and kinase domains colored dark blue, salmon, light green, and light blue, respectively, here and throughout the review. The activation loop of the kinase domain is colored red, and the activating (Tyr 416) and autoinhibitory (Tyr 527) phosphorylation sites are indicated. Conserved residue Arg 175 in the SH2 domain is critical for phosphotyrosine recognition; Trp 260 at the extreme N-terminus of the kinase domain is important for autoinhibition (see text). In the autoinhibited form of Src kinases, the SH2 domain binds the phosphorylated C-terminal tail, and the SH3 domain binds the linker segment between the SH2 and kinase domain, which forms a polyproline type II helix (see Figure 3). By convention, amino-acid residues are numbered as in chicken Src. In the lower panel, Protein Data Bank (PDB) accession codes for selected Src family structures are given, and the approximate region included in each three-dimensional structure is indicated. PDB accession codes correspond to: Lck unique domain in complex with the CD4 (1Q68) and CD8α (1Q69) cytoplasmic tails (Kim et al., 2003); 1SHG, crystal structure of αspectrin SH3 domain (Musacchio et al., 1992); 1SHF, the Fyn SH3 domain (Noble et al., 1993); 1SRM, the NMR structure of Src SH3 domain (Yu et al., 1992); 1SHA, 1SPR, and 1LCJ, crystal structures of SH2 domains of Src and Lck, respectively (1SHA and 1LCJ are in complex with tyrosine-phosphorylated peptides) (Waksman et al., 1992, 1993; Eck et al., 1993); 1LCK and 1G83, crystal structures of SH2–SH3 fragments of Lck (Eck et al., 1994) and Fyn (Arold et al., 2001); 3LCK, Lck kinase domain in its active conformation, phosphorylated on the activation loop tyrosine (Yamaguchi and Hendrickson, 1996); 2SRC (Xu et al., 1999), 1FMK (Xu et al., 1997), 2HCK (Sicheri et al., 1997), Src and Hck crystal structures incorporating SH2-SH3-Kinase domains in the autoinhibited conformation

A hallmark of Src kinases is a short C-terminal tail, which bears an autoinhibitory phosphorylation site (Tyrosine 527 in Src) (Cooper et al., 1986). Like most protein kinases, Src family members require phosphorylation within a segment of the kinase domain termed the activation loop for full catalytic activity. In Src, this autophosphorylation site is Tyrosine 416 (for consistency, we use chicken Src numbering throughout) (Smart et al., 1981). In vivo, Src kinases are phosphorylated on either Tyr 416 (in their active state) or Tyr 527 (in the inactive state). The inactivating phosphorylation on Tyr 527 is carried out by the Src-specific kinase Csk (Nada et al., 1991) or its homolog Chk (Hamaguchi et al., 1996; Davidson et al., 1997). Phosphorylation of the C-terminal tail promotes assembly of the SH2, SH3, and kinase domains into an autoinhibited conformation maintained by intimate interactions among these domains (Sicheri et al., 1997; Williams et al., 1997; Xu et al., 1997).

The discovery of the SH2 domain as a functionally important, but noncatalytic segment of 100 residues conserved in v-Fps, Abl, and Src kinases (Sadowski et al., 1986) gave rise to the concept of ‘modular signaling domain’ (Pawson, 1995). Studies of the function and specificity of the SH2 and SH3 domains of Src and of their effect on the activity of the adjacent kinase domain have provided a powerful paradigm for the functional dissection of a host of modular signaling domains (see Pawson (2004) for an excellent review and historical perspective). Likewise, structural dissection of Src kinases into their component domains and eventual elucidation of the structure of the essentially intact Src kinases have provided a paradigm for structural analysis of many multidomain signaling proteins. In the following paragraphs, we briefly describe the structure of the isolated domains of Src kinases. In later sections, we consider in more detail the structure of autoinhibited Src kinases and the implications of the structure for Src function and regulation.

The SH2 domain

The architecture of the SH2 domain and the structural basis for its recognition of phosphorylated tyrosine was first revealed in near-simultaneous studies of the domains derived from Abl (Overduin et al., 1992), the p85 subunit of PI 3-OH kinase (Booker et al., 1992), and of Src itself (Waksman et al., 1992). The now-familiar fold consists of a central β-sheet, with a single helix packed against each side. These elements of secondary structure and the loops that connect them form two recognition pockets, one that coordinates phosphotyrosine, and the second on the opposite side of the central sheet that typically binds one or more hydrophobic residues just C-terminal to the phosphotyrosine (Eck et al., 1993; Waksman et al., 1993). The phosphotyrosyl recognition pocket is rather highly conserved among SH2 domains, and contains a universally conserved arginine residue (Arg 175 in Src) that forms requisite electrostatic interactions with the phosphorylated tyrosine (Waksman et al., 1992). Not surprisingly, the C-terminal recognition pocket is much more divergent, accounting for the differences among SH2 domains in recognition of specific phosphotyrosine-bearing sequences (Kuriyan and Cowburn, 1997). Historically, phosphopeptide library-binding studies of SH2 specificity established a paradigm for such studies of modular signaling domains, and revealed distinct classes of specificity for the three to five residues following the phosphorylated tyrosine (Songyang et al., 1993). Src-family SH2 domains bind preferentially to the pY-E-E-I motif, coordinating the phosphotyrosine and isoleucine residues in the canonical recognition pockets (Figure 2). Polar and electrostatic interactions favor glutamic acid residues at the pY +1 and +2 positions, but many other residues can be accommodated in these positions in Src SH2 domains.

Figure 2
figure2

The Lck SH2 domain bound to a high-affinity phosphopeptide. The Lck SH2 domain binds the pYEEI -motif phosphopeptide in a two-pronged manner. The phosphorylated tyrosine and the isoleucine at the pY+3 position insert into well-defined recognition pockets on the surface of the domain. The intervening glutamic acid residues extend across the surface of the domain. The Lck SH2 domain is shown as a solvent-accessible surface represented by white dots, the peptide is shown as a CPK model. Figure adapted from Eck et al. (1993)

The common-fold and phosphotyrosyl-binding properties of the SH2 domain belie its versatility as a protein–protein recognition module. SH2 domains have evolved to function in diverse multidomain proteins, including transcription factors, ubiquitin ligases, GTPase-activating proteins in addition to tyrosine kinases and phosphatases (Pawson and Nash, 2003). In each of these contexts, the SH2 domain has evolved critical interdomain interactions that confer particular recognition and/or regulatory properties. Thus, as with most signaling domains, ‘modular’ does not imply ‘interchangeable’. A more comprehensive discussion of SH2 structure and specificity is beyond our scope, but many excellent reviews are available (Sawyer, 1998; Bradshaw and Waksman, 2002; Schlessinger and Lemmon, 2003; Waksman and Kuriyan, 2004).

The SH3 domain

The SH3 domain has a β-barrel architecture consisting of five antiparallel β-strands and two prominent loops, termed the RT and n-Src loops (Musacchio et al., 1992; Yu et al., 1992; Noble et al., 1993). These loops lie at either end of a surface composed of aromatic and hydrophobic residues that comprise the recognition site for proline-rich sequences bearing the ‘PxxP’ motif. These sequences adopt a polyproline type II helical conformation in complex with the SH3 domain. The PPII helix has a triangular cross section, the conserved prolines lie on the base of the triangle and intercalate with aromatic side chains on the SH3 surface. In the case of Src-family SH3 domains, additional specificity and binding affinity are conferred by coordination of a lysine or arginine residue just N- or C-terminal to the PxxP core. The pseudosymmetry of the PPII helix allows two high-affinity-binding modes, which differ in the N- to C-terminal orientation of the bound peptide. Src SH3 domains bind class I ligands with the motif RxxPxxP in an orientation opposite to that of class II ligands with the motif XpxxPxR (Feng et al., 1994, 1995; Lim et al., 1994). As with the SH2 domain, structural analysis of numerous SH3 domains has revealed considerable variation in structure and binding modes (Mayer, 2001). For example, the SH3 domain of the Gads T-cell adapter protein binds an RxxK motif rather than the characteristic PxxP motif (Berry et al., 2002; Liu et al., 2003). Within the Src family, the Fyn SH3 can bind the proline-independent motif RKxxYxxY found in the immune cell adaptor SKAP55 (Kang et al., 2000). Additionally, the Fyn kinase is regulated by a non-proline ‘surface–surface’ interaction of its SH3 domain with the SH2 protein SAP (Chan et al., 2003; Latour et al., 2003).

The tyrosine kinase domain

Src kinases share the bilobal protein kinase fold (Knighton et al., 1991a) characteristic of all tyrosine kinases and ser/thr kinases (Figure 3). The N-terminal (or small) lobe is composed of five β-strands and a single α-helix, termed the C helix, which is an important component of the regulatory mechanism deployed in Src kinases. The C-terminal (or large) lobe is predominantly α-helical, and contains the regulatory activation loop, which is the site of activating tyrosine phosphorylation in Src and other kinases. Nucleotide binding and phosphotransfer occur in the cleft between the two lobes. The adenine moiety of the bound nucleotide is coordinated largely by interactions with the N lobe and a short hinge segment that connects the two lobes. Bound nucleotide phosphates are in part coordinated by the glycine-rich G loop (also termed the P-loop, for phosphate binding) (Taylor et al., 1992; Hubbard and Till, 2000).

Figure 3
figure3

The three-dimensional structure of Src family tyrosine kinases. Ribbon diagrams representing crystal structures of autoinhibited c-Src and the active, Tyr 416-phosphorylated Lck kinase domain are shown on the left and right, respectively. In the assembled, autoinhibited structure, the SH3 domain makes intramolecular interactions with the linker and the N-terminal lobe of the kinase, while the SH2 domain binds the phosphorylated C-terminal tail. These interactions help to stabilize an inactive conformation in which the active site is disrupted by expulsion of the C helix. Also, the activation loop folds into a short helix with unphosphorylated tyrosine 416 oriented towards the active site cleft and buried. This activation loop arrangement further stabilizes the inactive conformation of the kinase domain and may also protect tyrosine 416 from inappropriate phosphorylation. In the active conformation of a Src catalytic domain, phosphorylation of tyrosine 416 stabilizes a very different conformation in the activation loop and also allows the C helix to assume its active position. Ribbon diagrams drawn from the coordinates of PDB depositions 2SRC (autoinhibited Src) and 3LCK (Lck kinase domain). Note that Lck residues are numbered as in Src for simplicity

The first direct structural information for an Src-family tyrosine kinase domain was obtained from the kinase domain of Lck in an active conformation, with Tyr 394 phosphorylated (the equivalent of Tyr 416 in Src) (Yamaguchi and Hendrickson, 1996). This structure showed how phosphorylation of Tyr 394 pinned the activation loop in an active conformation very similar to that observed in cyclic AMP-dependent kinase (Knighton et al., 1991a, 1991b), but quite different from that observed in the autoinhibited insulin receptor kinase (Figure 3, right panel) (Hubbard et al., 1994). Furthermore, it has provided an important reference point for comparisons with the conformation of the inactive kinase and in efforts to understand mechanistically the rearrangements that lead to activation (Schindler et al., 1999; Sicheri et al., 1997; Williams et al., 1997; Xu et al., 1997, 1999). Crystallographic analysis of numerous tyrosine and serine/threonine kinases in both active and inactive conformations has revealed a diverse set of structural deformations that lead to loss of catalytic function in inactive kinases (Huse and Kuriyan, 2002). These deformations involve key elements of the active site, including the C helix (which contains a critical glutamic acid residue, Glu 310 in Src), the relative orientations of the N- and C-lobes, the G-loop, and the activation loop itself. Not surprisingly, the active site conformations in active kinases are strikingly similar, as they must be competent to catalyse the phosphotransfer reaction. Comparison of inactive kinases shows substantial conformational variability, as distinct regulatory mechanisms involving phosphorylation, interdomain, and protein–protein interactions impinge upon these plastic elements of the catalytic core to control kinase activity in particular pathways. In Src, the SH3 and SH2 domains and phosphorylation of the activation loop and C-terminal tail comprise the key regulatory elements (Sicheri et al., 1997; Williams et al., 1997; Xu et al., 1997, 1999; Schindler et al., 1999).

Structure of the unique domain of Lck

The most N-terminal segment of the Src family architecture, the unique domain, has been the last to yield to structural analysis, perhaps in part because this region in isolation may lack a defined structure in many cases. The function of this domain is unclear for most members of the family, but in Lck it has long been known to mediate association with the cytoplasmic tails of T-cell coreceptors CD4 and CD8α (Rudd et al., 1988; Veillette et al., 1988; Shaw et al., 1989). The Lck/coreceptor interactions require conserved cysteine motifs: a CxCP motif in CD4 and CD8α and a CxxC motif in the Lck unique domain (Shaw et al., 1990; Turner et al., 1990). These four conserved cysteine residues (two from Lck and two from either of the coreceptors) coordinate a Zn2+ ion that is critical for complex formation (Huse et al., 1998; Lin et al., 1998). Recent solution NMR analysis of the Lck unique region and the CD4 and CD8α cytoplasmic tails has demonstrated that each of these sequences are unstructured in isolation, but that they fold together to form stable, ternary Lck : Zn : CD4 and Lck : Zn : CD8α complexes in the presence of Zn2+ (Kim et al., 2003). The elucidation of the three-dimensional structures of these complexes (Kim et al., 2003) reveals that they share a similar zinc-binding core, composed of a β-hairpin in Lck and a short extended segment in CD4 or CD8α (Figure 4). In the CD4 complex, this core region is augmented by hydrophobic interactions between two helices, one in Lck and the other in CD4. Interestingly, this interaction buries a dileucine motif that is critical for internalization of CD4 by the clathrin adapter protein AP-2 (Shin et al., 1990). Internalization of CD4 is known to require phosphorylation of Ser 408 in CD4, followed by dissociation of Lck (Shin et al., 1991; Pitcher et al., 1999). The structure of the complex rationalizes this observation, as Ser 408 is exposed and accessible for phosphorylation, but the mechanism by which this phosphorylation promotes dissociation of the complex remains obscure. Although the sequences required for co-receptor interaction are found only in Lck, the unique regions of other Src family members may mediate distinct binding interactions. In Fyn, for example, this region may associate with the cytoplasmic chains of the T-cell receptor via its amino-terminal region (Timson Gauen et al., 1992), but this interaction is weak and not well characterized. Additionally, studies of chimeric Src/Yes proteins show that the Yes unique region cannot substitute for the corresponding region of Src, supporting the notion of distinct interactions and function for this domain (Summy et al., 2003).

Figure 4
figure4

Structures of the Lck unique region in complex with CD4 and CD8α cytoplasmic tails. The three-dimensional NMR structures of the Lck unique domain in complex with the cytoplasmic tails of T-cell coreceptors CD4 (left) and CD8α (right) are shown in a ribbon representation. Lck is shown in dark blue and the cytoplasmic tails of CD4 and CD8α are shown in gray. The bound Zn2+ ion is indicated as a blue sphere. Portions of the Lck unique domain and coreceptor tails that are unstructured and not included in the complex are represented by dotted lines. The Zn2+-binding core of the complexes is formed by two cysteine residues in Lck (Cys 20 and Cys 23) and two in either of the coreceptors (cysteines 420 and 422 in CD4, 194, and 196 in CD8α), small but significant hydrophobic cores also help stabilize each complex. The interaction between Lck and CD4 is regulated; phosphorylation of serine 408 in CD4 promotes coreceptor dissociation and internalization. The dileucine internalization motif in CD4 (leucines 413 and 414) that is required for recognition by the clathrin adaptor complex is buried by the interaction with Lck, explaining the need for dissociation of Lck prior to receptor internalization

The autoinhibited conformation of Src

Crystal structures of both Src (Williams et al.,. 1997; Xu et al., 1997, 1999) and Hck (Sicheri et al., 1997; Schindler et al., 1999) in the autoinhibited, tail-phosphorylated state show that the SH3 and SH2 domains turn inward and make intramolecular interactions that lock the catalytic domain in an inactive conformation. The SH3 and SH2 domains pack against the catalytic domain on the side opposite the catalytic cleft. The SH3 domain packs against the N lobe of the kinase, while the SH2 domain packs against the C lobe. This ‘assembled’ conformation is stabilized by intramolecular interactions of the SH3 and SH2 domains with flexible polypeptide segments; the SH3 domain binds the linker segment that connects the SH2 and kinase domains, and the SH2 domain binds the phosphorylated C-terminal tail (Figure 3, left panel). Importantly, the intramolecular sequences coordinated by the SH3 and SH2 domains are not ‘high-affinity’ binding sites. The linker segment or Src does not contain the PxxP motif, but the linker of both Src and Hck does adopt a PPII helical conformation in association with their respective SH3 domains. Similarly, the C-terminal tail bears a phosphorylated tyrosine, but has a glycine, rather than the preferred isoleucine, at the pY+3 position. As discussed below, the fact that these targeting domains use their ligand-binding surfaces to maintain the autoinhibited conformation has important implications for controlling the activation of Src kinases.

Why is this assembled conformation of Src catalytically compromised? The position of the SH2 and SH3 domains does not sterically occlude the catalytic cleft. Rather, these domains lock the kinase domain in an inactive conformation (Figure 5). The principle distortions of the active site that render it inactive are (i) the displacement of the C helix, which removes the catalytically important Glu 310 from the active site cleft, (ii) the activation loop adopts an α-helical conformation, which precludes binding of peptide substrates and also sequesters Tyr 416, making it inaccessible for phosphorylation, and (iii) the relative orientation of the N and C lobes is constrained in an orientation that may not be optimal for catalysis (Sicheri et al., 1997; Williams et al., 1997; Xu et al., 1997, 1999; Schindler et al., 1999).

Figure 5
figure5

Detailed schematic of the active site in inactive vs active Src kinases. The active site of a Src kinase in an inactive conformation (left panel, Src) and in the active conformation (right panel, Lck). Portions of the N lobe are shown in blue, the activation loop in green and the catalytic loop in red. Changes in the positioning of Glu-310 and the C helix between the active and inactive conformations are evident. In the inactive state, the C helix is rotated outward, and Glu 310 is coordinated by Arg 409 and Tyr 382. Leu 407 and flanking residues in the activation loop also help to fix the C-helix in the inactive position. Phosphorylation of Tyr 416 restructures the activation loop, allowing the C helix to rotate inward and Glu 310 to form the requisite interaction with Lys 295 (which in turn coordinates the phosphates of the bound ATP, not shown). Note that in the active state Arg 409 forms a salt bridge with the phosphorylated Tyr 416 rather than with Glu 310; this electrostatic switch is one component of the activation mechanism. Figure adapted from Xu et al. (1999)

The precise structural mechanism by which the SH3 and SH2 domain assembly promotes displacement of the C helix is not completely clear. However, one can gain insight into this question by considering the C helix to have two preferred conformations, ‘active’ and ‘displaced’, which may differ little in intrinsic stability. Thus, the switch from active to displaced (or vice versa) could be effected by factors that either stabilized the displaced position or destabilized the active conformation. The SH3 and SH2 domains may in part promote the displaced conformation by pinning the linker segment against the N lobe of the kinase domain. Tryptophan 260 lies at the junction between the linker and the N lobe, and packs against the carboxy-terminal end of the C helix in the assembled conformation, where it helps to stabilize the displaced, inactive position of the C helix. In support of this idea, in the structure of the active Lck kinase domain, this residue adopts an alternate conformation and does not contact the C helix (Yamaguchi and Hendrickson, 1996). Also, mutagenesis studies support the notion that Trp 260 helps to communicate the position of the SH2 and SH3 domains to the active site (LaFevre-Bernt et al., 1998). Additional residues in the linker may also participate (Gonfloni et al., 1997). In particular, Leu 255 inserts into a hydrophobic pocket on the back of the N lobe and may stabilize the displaced conformation (Gonfloni et al., 1999).

The SH2 and SH3 domains also communicate to the C helix via the activation loop. In the assembled conformation, the activation loop forms a short helix, which is propped between the N and C lobes (Schindler et al., 1999; Xu et al., 1999). In this conformation, Tyr 416 is buried in the cleft between the lobes of the kinase, where it is not accessible for phosphorylation. Furthermore, the more N-terminal portion of the activation loop, including Leu 407, creates a steric barrier that blocks the active position of the C helix (Figure 5). Release of the SH3/SH2 clamp is likely to induce disordering of the helical activation loop and removal of the blockage created by Leu 407. Subsequent phosphorylation of Tyr 416 creates an electrostatic switch that further promotes the active position of the C helix and induces a new conformation in the activation loop (Figure 5). The activation loop conformation in the active Lck kinase structure closely resembles that in the insulin receptor tyrosine kinase (IRTK) (Yamaguchi and Hendrickson, 1996). In IRTK, the phosphorylated activation loop forms a β-sheet interaction with a bound substrate peptide (Hubbard, 1997). By analogy with IRTK, the phosphorylated and subsequently reorganized activation loop in Src kinases is thought to form a considerable portion of the binding surface for substrate peptides; thus, autoinhibited Src is unable to bind peptide substrate.

Finally, the restriction of interdomain mobility caused by the SH3/SH2 clamp may also directly inhibit the kinase, both because the relative orientation of the N and C lobes induced by the clamp do not appear to be ideal for catalysis and because greater flexibility may be required for efficient catalysis and for turnover of the nucleotide substrate. The effectiveness of the clamp appears to depend upon a rigid coupling between the SH3 and SH2 domains (Young et al., 2001). Although NMR and crystallographic analysis of SH3–SH2 fragments of Src kinases indicated that there is no defined orientation between the two domains when they are ‘excised’ from the intact kinase (Eck et al., 1994; Arold et al., 2001), computational and biochemical analyses show that, in the context of the intact kinase, a relatively rigid coupling between these two domains is required in order to maintain the inactive conformation (Young et al., 2001). Molecular dynamics studies of the assembled Src kinase reveal tight correlation in the movements of the N lobe and SH3 domain, consistent with the intimate association observed in crystal structure (Young et al., 2001). Furthermore, these studies indicate that these correlated motions extend from the SH3 domain into the SH2 domain, suggesting that conformational coupling between these domains is required for effective autoinhibition. This prediction has been experimentally confirmed, glycine mutations in the short turn connecting the SH3 and SH2 domain deregulate Src (Young et al., 2001).

In summary, in an assembled state, Src kinases are catalytically compromised by the displacement of the C helix. The SH3 and SH2 domains are turned inward, their recognition surfaces sequestered by weak intramolecular interactions. Additionally, the helical conformation of the activation loop precludes binding of peptide substrates and also protects Tyr 416 from phosphorylation. The interdependencies in the positions and conformations of the SH3 and SH2 domains, the kinase lobes, the activation loop, and the C helix suggest that maintenance of the autoinhibited conformation relies on this set of cooperative, self-consistent interactions. The autoinhibited Src kinase is a precariously set mousetrap, small perturbations in key interdomain interactions can be sufficient to spring the kinase into its active conformation. Indeed, the oncogenic v-Src protein lacks the autoinhibitory phosphorylation site in the C-terminal tail (Takeya and Hanafusa, 1983; Cooper et al., 1986). Additionally, mutations in the RT loop of the SH3 domain that perturb the SH3-kinase interaction (Kato et al., 1986) or mutations in the SH2 domain that block binding of the phosphorylated tail are sufficient to activate Src.

Insights into Src activation

Why such a complex, multicomponent mechanism of autoinhibition? The architecture of the assembled Src kinase appears to have evolved to create an intrinsic coupling of activation of Src with targeting to its appropriate cellular substrates. Src kinases are known to be activated by binding of cognate ligands to their SH2 and/or SH3 domains (Brown and Cooper, 1996). Src is both recruited to, and activated by, the PDGF receptor via interaction of its SH2 domain with the tyrosine-phosphorylated receptor (Kypta et al., 1990; Alonso et al., 1995). A number of Src substrates contain binding motifs for both the SH3 and SH2 domains of Src; focal adhesion kinase (Fak) is one example (Thomas et al., 1998; Sieg et al., 1999). The closely spaced proline and phosphotyrosine motifs in Fak have been shown to bind simultaneously to an SH3/SH2 fragment of Fyn, although the tandem interaction does not appear to confer an avidity effect (Arold et al., 2001). Additionally, Src kinases can be activated by displacement of their SH3 domain, while the SH2 domain remains engaged with the C-terminal tail. This effect has been demonstrated in vitro by activation of Hck with the HIV protein Nef (Moarefi et al., 1997). Nef contains a PxxP motif within a loop that is preconfigured in a PPII conformation and confers tight binding to the Hck SH3 domain (Lee et al., 1995).

More recently, Fyn has been discovered to be recruited to SLAM-family receptors and activated by interactions of its SH3 domain with the adapter protein SAP (SLAM-associated protein) (Latour et al., 2001) (Figure 6). SLAM receptors are present on T and NK cells, as well as B cells and other antigen-presenting cells, where they make homotypic ‘self-ligand’ interactions that are critical for a proper immune response (Engel et al., 2003). SAP is an unusual adapter protein in that it consists of a single SH2 domain, but it can nevertheless function as an adapter protein because it can bind the SLAM cytoplasmic tail with its phosphopeptide-binding site and simultaneously bind the Fyn SH3 domain via a distinct binding surface (Chan et al., 2003). SAP is also an unusual SH2 domain in that it binds SLAM in a ‘three-pronged’ fashion which tolerates, but does not require, phosphorylation of SLAM (Poy et al., 1999). Binding of SAP to the Fyn SH3 domain is expected to disrupt the autoinhibitory interaction of the SH3 domain, and therefore to unleash the catalytic activity of Fyn (Chan et al., 2003; Latour et al., 2003). The activation of Fyn by SAP and SLAM is critical for proper immune modulation; inherited mutations in SAP which abolish or weaken binding to SLAM are the basis of X-linked lymphoproliferative syndrome (Engel et al., 2003).

Figure 6
figure6

Insights into the activation of Src kinases: recruitment and activation of Fyn by SLAM/CD150 coreceptors in immune cells. In FynT, as in all Src kinases, activation is thought to be accomplished by release of the inhibitory intramolecular interactions of the SH2 and SH3 domains, accompanied by autophosphorylation of the activation loop. FynT is a lymphocyte-specific isoform of Fyn, and is brought to the cytoplasmic tail of SLAM via interaction of its SH3 domain with the adapter protein SAP. SAP is an SH2-only protein that can simultaneously bind SLAM and the FynT SH3 domain. The structure of the entire complex has not been elucidated, but the crystal structure of a SLAM/SAP/Fyn-SH3 complex is known (inset). In the assembled conformation of FynT (left), the SAP-binding site on the SH3 domain is occluded by intramolecular interactions; recruitment by SAP is expected to activate the kinase by displacement of the SH3 domain (right). The active Fyn kinase phosphorylates indicated tyrosines in the SLAM tail. The FynT SH2 domain may also bind one of these sites, further stabilizing the active signaling complex. Also, SAP may interact directly with the catalytic domain of FynT. The ribbon diagram of the SLAM/SAP/FynSH3 complex was drawn from the coordinates of PDB deposition 1M27

The SLAM/SAP/Fyn interaction also highlights significant gaps in our structural understanding of Src kinase activation (Figure 6). Although the active conformation of the Lck kinase domain is known, relatively little is known about how full-length Src kinases may assemble in active signaling complexes. Are there precisely defined interdomain or interprotein interactions in the active state akin to those observed in autoinhibited Src? How do Src kinases recognize their protein substrates? Do the SH2 and SH3 domains ‘direct’ substrate to the kinase domain in a precise manner, or do they simply co-localize the kinase with substrate? To date, no Src kinase has been crystallized in complex with a peptide/protein substrate. In the case of the SAP/Fyn interaction, SAP may bind to the Fyn kinase domain in addition to the SH3 domain (Chan et al., 2003). Additionally, phosphorylation of additional tyrosines in SLAM is required for stable association with Fyn, suggesting that the Fyn SH2 domain may bind SLAM following initial recruitment and phosphorylation of SLAM by Fyn (Latour et al., 2003). These additional interactions hint at the existence of a conformationally defined, active signaling complex (Figure 6), but further structural analysis is needed to address these questions.

Also, it is unclear whether the unique domain of some Src kinases may participate in catalytic regulation. The available structures of Src and Hck lack the unique regions. One clue that the unique region of Src may serve a regulatory function is the presence of serine phosphorylation sites (Gould et al., 1985; Chackalaparampil and Shalloway, 1988; Shalloway et al., 1992). In Lck serines 42 and 59 are phosphorylated by protein kinase C, and these phosphorylations affect the function and catalytic activity of the kinase (Watts et al., 1993; Winkler et al., 1993; Kesavan et al., 2002). Additionally, in the Src-family cousin Abl, the N-terminal ‘cap’ region contributes to the autoinhibited conformation (see below).

Src: turned on by touch

The structural investigations of Src kinases over the past dozen years have provided insights that go well beyond Src itself. Most obviously, they helped to elucidate the general mechanisms of recognition by SH2 and SH3 domains, which are near-ubiquitous in eukaryotic signal transduction. Additionally, the SH3–SH2-kinase assemblage has been reused in evolution with only minor modifications. The Abelson kinase (Abl), and likely Tec family kinases as well, adopt an assembled conformation very similar to that of Src kinases (Nagar et al., 2003). More generally, the underlying principle revealed by the structural organization of Src – the use of protein interaction domains to regulate catalytic activity in a manner that inextricably couples targeting with catalytic activation – applies quite frequently, if not universally, in modular signaling proteins (Pawson and Nash, 2003).

Recent structural analysis of Abl in an autoinhibited conformation reveals that its SH3, SH2, and kinase domains adopt an assembled conformation extremely similar to that of Src kinases (Nagar et al., 2003), despite the fact that Abl lacks the autoinhibitory tail phosphorylation site that is crucial for Src inhibition. This similarity was expected, and reinforced by mutagenesis studies based upon the Src structures and sequence comparisons, but it remained unclear how Abl was maintained in an inactive conformation in the absence of the ‘latch’ provided by the SH2–tail interaction in Src kinases. Structure/function studies of Abl by Superti-Furga and colleagues defined an N-terminal ‘cap’ region in Abl that appeared to stabilize its autoinhibited conformation (Pluk et al., 2002). The crystal structure of Abl confirmed this prediction, and quite unexpectedly revealed that the N-terminal myristoyl group contributes to autoinhibition by inserting into a hydrophobic pocket within the C lobe of the kinase domain. This deep hydrophobic pocket appears to be unique to Abl, and insertion of the acyl chain induces a conformational rearrangement on the back of the C lobe that creates the docking site for the SH2 domain (Nagar et al., 2003). Thus, the myristoylated N-terminus functionally replaces the SH2–tail interaction in Abl, and, as in Src, disruption of this latch leads to oncogenic activation (Azam et al., 2003; Hantschel et al., 2003; Harrison, 2003; Nagar et al., 2003). In the oncogenic BCR-Abl (the chimeric product of the translocated Philadelphia chromosome in chronic myelogenous leukemia), the N-terminal myristoylation site and cap regions are truncated and replaced with BCR-derived sequences. Also, in parallel with Src kinases, appropriate targeting interactions are thought to disrupt the assembled state and induce kinase activation (Nagar et al., 2003).

The ability of Src to be ‘turned on by touch’ is indispensable for regulating and maintaining fidelity in its signaling pathways. Diverse multidomain signaling proteins share this property, in spite of the fact that their domain architectures are unrelated to that of Src. For example, in the protein tyrosine phosphatase SHP-2, which contains two adjacent SH2 domains followed by a the catalytic phosphatase domain, the N-terminal SH2 domain binds back to the PTPase domain and sterically occludes the active site (Hof et al., 1998). The SH2 domain uses a surface that does not overlap with its phosphopeptide-binding groove. The surface–surface interaction with the phosphatase domain is nevertheless disrupted by binding of cognate phosphoproteins to the SH2 domains due to a conformational change within the N-terminal SH2 domain that disrupts its PTPase-binding surface. As with Src and Abl, mutations in SHP-2 that uncouple targeting and catalytic activation (without directly compromising either) result in disease (Tartaglia et al., 2001; Neel et al., 2003).

A feature that is implicit in the design of these proteins, but perhaps not immediately obvious, is an enhancement of specificity that arises from the use of targeting domains for autoinhibition. This property is readily explained in autoinhibited Src – the SH3 and SH2 domains have intramolecular binding partners of modest affinity, thus they only bind and become activated by specific high-affinity binding partners that can outcompete these intramolecular interactions. More generally, in any ‘allosteric’ system, a portion of energy of ligand binding is used to induce conformational change, and is therefore ‘lost’ as intrinsic binding energy. Regulatory intramolecular interactions effectively ‘raise the bar’ for specific binding. Thus, Src is only turned on by the touch of its intended, specific partners.

One might expect release of the SH3/SH2 clamp alone to be sufficient to activate Src, absent Tyr 416 phosphorylation, because Tyr 416 is thought to be autophosphorylated. Strikingly, recent crystallographic analysis of an SH3-SH2-kinase fragment of Src in an unphosphorylated state bears out this prediction. This structure reveals a heretofore unseen conformation in which the SH3, SH2, and linker segments remain assembled with each other, but are greatly displaced from the autoinhibitory position on the back of the kinase domain. Essentially, the clamp is intact, but released from the kinase domain, because the C-terminal tail is unphosphorylated and does not bind the SH2 domain. In this structure, the kinase domain adopts a conformation very similar to that in the active Lck kinase domain, despite the fact that Tyr 416 is not phosphorylated (D Fabbro and S Cowan-Jacob, personal communication). Thus, this structure provides an important snapshot of the Src activation pathway, and it provides direct support for the notion that Src kinases have evolved to intrinsically couple activation with targeting.

References

  1. Alonso G, Koegl M, Mazurenko N and Courtneidge SA . (1995). J. Biol. Chem., 270, 9840–9848.

  2. Arold ST, Ulmer TS, Mulhern TD, Werner JM, Ladbury JE, Campbell ID and Noble ME . (2001). J. Biol. Chem., 276, 17199–17205.

  3. Azam M, Latek RR and Daley GQ . (2003). Cell, 112, 831–843.

  4. Berry DM, Nash P, Liu SK, Pawson T and McGlade CJ . (2002). Curr. Biol., 12, 1336–1341.

  5. Booker GW, Breeze AL, Downing AK, Panayotou G, Gout I, Waterfield MD and Campbell LD . (1992). Nature, 358, 684–687.

  6. Bradshaw JM and Waksman G . (2002). Adv. Protein Chem., 61, 161–210.

  7. Brown MT and Cooper JA . (1996). Biochim. Biophys. Acta, 1287, 121–149.

  8. Chackalaparampil I and Shalloway D . (1988). Cell, 52, 801–810.

  9. Chan B, Lanyi A, Song HK, Griesbach J, Simarro-Grande M, Poy F, Howie D, Sumegi J, Terhorst C and Eck MJ . (2003). Nat. Cell Biol., 5, 155–160.

  10. Cooper JA, Gould KL, Cartwright CA and Hunter T . (1986). Science, 231, 1431–1434.

  11. Davidson D, Chow LM and Veillette A . (1997). J. Biol. Chem., 272, 1355–1362.

  12. Eck MJ, Atwell SK, Shoelson SE and Harrison SC . (1994). Nature, 368, 764–769.

  13. Eck MJ, Shoelson SE and Harrison SC . (1993). Nature, 362, 87–91.

  14. Engel P, Eck MJ and Terhorst C . (2003). Nat. Rev. Immunol., 3, 813–821.

  15. Feng S, Chen JK, Yu H, Simon JA and Schreiber SL . (1994). Science, 266, 1241–1247.

  16. Feng S, Kasahara C, Rickles RJ and Schreiber SL . (1995). Proc. Natl. Acad. Sci. USA, 92, 12408–12415.

  17. Gonfloni S, Frischknecht F, Way M and Superti-Furga G . (1999). Nat. Struct. Biol., 6, 760–764.

  18. Gonfloni S, Williams JC, Hattula K, Weijland A, Wierenga RK and Superti-Furga G . (1997). EMBO J, 16, 7261–7271.

  19. Gould KL, Woodgett JR, Cooper JA, Buss JE, Shalloway D and Hunter T . (1985). Cell, 42, 849–857.

  20. Hamaguchi I, Yamaguchi N, Suda J, Iwama A, Hirao A, Hashiyama M, Aizawa S and Suda T . (1996). Biochem. Biophys. Res. Commun., 224, 172–179.

  21. Hantschel O, Nagar B, Guettler S, Kretzschmar J, Dorey K, Kuriyan J and Superti-Furga G . (2003). Cell, 112, 845–857.

  22. Harrison SC . (2003). Cell, 112, 737–740.

  23. Hof P, Pluskey S, Dhe-Paganon S, Eck MJ and Shoelson SE . (1998). Cell, 92, 441–450.

  24. Hubbard SR . (1997). EMBO J, 16, 5572–5581.

  25. Hubbard SR and Till JH . (2000). Annu. Rev. Biochem., 69, 373–398.

  26. Hubbard SR, Wei L, Ellis L and Hendrickson WA . (1994). Nature, 372, 746–754.

  27. Huse M, Eck MJ and Harrison SC . (1998). J. Biol. Chem., 273, 18729–18733.

  28. Huse M and Kuriyan J . (2002). Cell, 109, 275–282.

  29. Kang H, Freund C, Duke-Cohan JS, Musacchio A, Wagner G and Rudd CE . (2000). EMBO J, 19, 2889–2899.

  30. Kato JY, Takeya T, Grandori C, Iba H, Levy JB and Hanafusa H . (1986). Mol. Cell Biol., 6, 4155–4160.

  31. Kesavan KP, Isaacson CC, Ashendel CL, Geahlen RL and Harrison ML . (2002). J. Biol. Chem., 277, 14666–14673.

  32. Kim PW, Sun ZY, Blacklow SC, Wagner G and Eck MJ . (2003). Science, 301, 1725–1728.

  33. Knighton DR, Zheng JH, Ten EL, Ashford VA, Xuong NH, Taylor SS and Sowadski JM . (1991a). Science, 253, 407–414.

  34. Knighton DR, Zheng JH, Ten EL, Xuong NH, Taylor SS and Sowadski JM . (1991b). Science, 253, 414–420.

  35. Koegl M, Zlatkine P, Ley SC, Courtneidge SA and Magee AI . (1994). Biochem. J., 303, 749–753.

  36. Kuriyan J and Cowburn D . (1997). Annu. Rev. Biophys. Biomol. Struct., 26, 259–288.

  37. Kypta RM, Goldberg Y, Ulug ET and Courtneidge SA . (1990). Cell, 62, 481–492.

  38. LaFevre-Bernt M, Sicheri F, Pico A, Porter M, Kuriyan J and Miller WT . (1998). J. Biol. Chem., 273, 32129–32134.

  39. Latour S, Gish G, Helgason CD, Humphries RK, Pawson T and Veillette A . (2001). Nat. Immunol., 2, 681–690.

  40. Latour S, Roncagalli R, Chen R, Bakinowski M, Shi X, Schwartzberg PL, Davidson D and Veillette A . (2003). Nat. Cell Biol., 5, 149–154.

  41. Lee CH, Leung B, Lemmon MA, Zheng J, Cowburn D, Kuriyan J and Saksela K . (1995). EMBO J, 14, 5006–5015.

  42. Lim WA, Richards FM and Fox RO . (1994). Nature, 372, 375–379.

  43. Lin RS, Rodriguez C, Veillette A and Lodish HF . (1998). J. Biol. Chem., 273, 32878–32882.

  44. Liu Q, Berry D, Nash P, Pawson T, McGlade CJ and Li SS . (2003). Mol. Cell, 11, 471–481.

  45. Mayer BJ . (2001). J. Cell Sci., 114, 1253–1263.

  46. Moarefi I, LaFevre-Bernt M, Sicheri F, Huse M, Lee CH, Kuriyan J and Miller WT . (1997). Nature, 385, 650–653.

  47. Musacchio A, Noble M, Pauptit R, Wierenga R and Saraste M . (1992). Nature, 359, 851–855.

  48. Nada S, Okada M, MacAuley A, Cooper JA and Nakagawa H . (1991). Nature, 351, 69–72.

  49. Nagar B, Hantschel O, Young MA, Scheffzek K, Veach D, Bornmann W, Clarkson B, Superti-Furga G and Kuriyan J . (2003). Cell, 112, 859–871.

  50. Neel BG, Gu H and Pao L . (2003). Trends Biochem. Sci., 28, 284–293.

  51. Noble M, EM, Musacchio A, Saraster M, Courtneidge SA and Wierenga RK . (1993). EMBO J., 12, 2617–2624.

  52. Overduin M, Rios CB, Mayer BJ, Baltimore D and Cowburn D . (1992). Cell, 70, 697–704.

  53. Pawson T . (1995). Nature, 373, 573–580.

  54. Pawson T . (2004). Cell, 116, 191–203.

  55. Pawson T and Nash P . (2003). Science, 300, 445–452.

  56. Pitcher C, Honing S, Fingerhut A, Bowers K and Marsh M . (1999). Mol. Biol. Cell, 10, 677–691.

  57. Pluk H, Dorey K and Superti-Furga G . (2002). Cell, 108, 247–259.

  58. Poy F, Yaffe MB, Sayos J, Saxena K, Morra M, Sumegi J, Cantley LC, Terhorst C and Eck MJ . (1999). Mol. Cell, 4, 555–561.

  59. Resh MD . (1999). Biochim. Biophys. Acta, 1451, 1–16.

  60. Rudd CE, Trevillyan JM, Dasgupta JD, Wong LL and Schlossman SF . (1988). Proc. Natl. Acad. Sci. USA, 85, 5190–5194.

  61. Sadowski I, Stone JC and Pawson T . (1986). Mol. Cell Biol., 6, 4396–4408.

  62. Sawyer TK . (1998). Biopolymers, 47, 243–261.

  63. Schindler T, Sicheri F, Pico A, Gazit A, Levitzki A and Kuriyan J . (1999). Mol. Cell, 3, 639–648.

  64. Schlessinger J and Lemmon MA . (2003). Sci. STKE, 2003, RE12.

  65. Shalloway D, Bagrodia S, Chackalaparampil I, Shenoy S, Lin PH and Taylor SJ . (1992). Ciba Found. Symp., 170, 248–265; discussion 265–275.

  66. Shaw AS, Amrein KE, Hammond C, Stern DF, Sefton BM and Rose JK . (1989). Cell, 59, 627–636.

  67. Shaw AS, Chalupny J, Whitney JA, Hammond C, Amrein KE, Kavathas P, Sefton BM and Rose JK . (1990). Mol. Cell Biol., 10, 1853–1862.

  68. Shin J, Doyle C, Yang Z, Kappes D and Strominger JL . (1990). EMBO J., 9, 425–434.

  69. Shin J, Dunbrack RLJ, Songjae L and Strominger JL . (1991). J. Biol. Chem., 266, 10658–10665.

  70. Sicheri F, Moarefi I and Kuriyan J . (1997). Nature, 385, 602–609.

  71. Sieg DJ, Hauck CR and Schlaepfer DD . (1999). J. Cell Sci., 112, 2677–2691.

  72. Smart JE, Oppermann H, Czernilofsky AP, Purchio AF, Erikson RL and Bishop JM . (1981). Proc. Natl. Acad. Sci. USA, 78, 6013–6017.

  73. Songyang Z, Shoelson SE, Chaudhuri M, Gish G, Pawson T, King F, Roberts T, Ratnofsky S, Lechleider RJ, Neel BG, Birge RB, Fajardo JE, Chou MM, Hanafusa H, Schaffhausen B and Cantley LC . (1993). Cell, 72, 767–778.

  74. Summy JM, Qian Y, Jiang BH, Guappone-Koay A, Gatesman A, Shi X and Flynn DC . (2003). J. Cell Sci., 116, 2585–2598.

  75. Takeya T and Hanafusa H . (1983). Cell, 32, 881–890.

  76. Tartaglia M, Mehler EL, Goldberg R, Zampino G, Brunner HG, Kremer H, van der Burgt I, Crosby AH, Ion A, Jeffery S, Kalidas K, Patton MA, Kucherlapati RS and Gelb BD . (2001). Nat. Genet., 29, 465–468.

  77. Taylor SS, Knighton DR, Zheng J, Ten EL and Sowadski JM . (1992). Annu. Rev. Cell Biol., 8, 429–462.

  78. Thomas JW, Ellis B, Boerner RJ, Knight WB, White 2nd GC and Schaller MD . (1998). J. Biol. Chem., 273, 577–583.

  79. Thomas SM and Brugge JS . (1997). Annu. Rev. Cell Dev. Biol., 13, 513–609.

  80. Timson Gauen LK, Kong AN, Samelson LE and Shaw AS . (1992). Mol. Cell Biol., 12, 5438–5446.

  81. Turner JM, Brodsky MH, Irving BA, Levin SD, Perlmutter RM and Littman DR . (1990). Cell, 60, 755–765.

  82. Veillette A, Bookman A, Horak EM and Bolen JB . (1988). Cell, 55, 301–308.

  83. Waksman G, Kominos D, Robertson SC, Pant N, Baltimore D, Birge RB, Cowburn D, Hanafusa H, Mayer BJ, Overduin M, Resh MD, Rios CB, Silverman L and Kuriyan J . (1992). Nature, 358, 646–653.

  84. Waksman G and Kuriyan J . (2004). Cell, 116, S45–S48, 3 pp following S48.

  85. Waksman G, Shoelson SE, Pant N, Cowburn D and Kuriyan J . (1993). Cell, 72, 779–790.

  86. Watts JD, Welham MJ, Kalt L, Schrader JW and Aebersold R . (1993). J. Immunol., 151, 6862–6871.

  87. Williams JC, Weijland A, Gonfloni S, Thompson A, Courtneidge SA, Superti-Furga G and Wierenga RK . (1997). J. Mol. Biol., 274, 757–775.

  88. Winkler DG, Park I, Kim T, Payne NS, Walsh CT, Strominger JL and Shin J . (1993). Proc. Natl. Acad. Sci. USA, 90, 5176–5180.

  89. Xu W, Doshi A, Lei M, Eck MJ and Harrison SC . (1999). Mol. Cell, 3, 629–638.

  90. Xu W, Harrison SC and Eck MJ . (1997). Nature, 385, 595–602.

  91. Yamaguchi H and Hendrickson WA . (1996). Nature, 384, 484–489.

  92. Young MA, Gonfloni S, Superti-Furga G, Roux B and Kuriyan J . (2001). Cell, 105, 115–126.

  93. Yu H, Rosen MK, Shin TB, Seidel-Dugan C, Brugge JS and Schreiber SL . (1992). Science, 258, 1665–1668.

Download references

Author information

Affiliations

Authors

Corresponding author

Correspondence to Michael J Eck.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Boggon, T., Eck, M. Structure and regulation of Src family kinases. Oncogene 23, 7918–7927 (2004). https://doi.org/10.1038/sj.onc.1208081

Download citation

Keywords

  • Src regulation
  • modular signaling
  • CD4/CD8α

Further reading

Search

Quick links