Src Substrate Surprise

Abstract

In this issue of Cancer Cell, Lowy and colleagues show that the SH2 domain of tensin-3 is regulated by phosphorylation by Src, and that this phosphorylation promotes the oncogenic function of tensin-3. Phosphorylation of the SH2 domain represents a novel mechanism for the regulation of SH2 ligand binding.

Src is a non-receptor tyrosine kinase that is activated by a variety of mechanisms in human cancer. Its biological effects are mediated by the phosphorylation of a plethora of protein substrates. A primary role of tyrosine phosphorylation is to generate docking sites for proteins containing SH2 or PTB domains, thereby promoting protein-protein interactions and the formation of macromolecular complexes that function in intracellular signal transduction (Pawson, 2004). The binding activity of SH2 domains is primarily regulated by the phosphorylation of the ligand, although some SH2 domain - target interactions are phosphorylation-independent.

Many prominent Src substrates are found at focal adhesions, including the focal adhesion kinase FAK and the Crk-associated substrate Cas. Focal adhesions are sites of integrin-dependent substrate adhesion generated by Rho-ROCK-activated actomyosin-dependent contractility. They serve as sites for the exertion of force on the substratum. They also serve as sites of intracellular signal transduction, mediating adhesion-dependent signals that result in anchorage-dependent cell proliferation.

Tensins are another family of focal adhesion proteins that can serve as Src substrates. There are four members of the family in mammals (Lo, 2004). Tensins 1-3 contain three distinct regions (Fig 1A): the N-terminal domains bind to the side of F-actin and target the molecule to focal adhesions; the central region of the tensins is non-conserved; while the C-terminus contains an SH2 domain and a PTB domain. The PTB domain interacts, not with tyrosine phosphorylated proteins, but with NPXY motifs in the integrin tails of β1, 3, 5 and 7. Tensin-1-null fibroblasts show migration defects, and the focal adhesion localization and SH2 domains of tensin-1 are required for this pro-migratory function (Chen and Lo, 2003). The tensin SH2 domain binds to tyrosine-phosphorylated Cas and FAK, which are pro-oncogenic. However it also binds to non-phosphorylated RhoGAP7, aka DLC1 (deleted in liver cancer 1), a focal adhesion-localized protein that is a known tumor suppressor, and to the related tumor suppressor DLC3 (Liao et al., 2007, Qian et al., 2007).

Figure 1. Tyrosine phosphorylation of the SH2 domain of tensin-3.

(A) Domain structure of tensin-3. In focal adhesions tensins link integrin heterodimers to the actin cytoskeleton. The N-terminus of tensin-3 contains domains that bind to F-actin and a focal adhesion targeting activity. The C-terminus contains an SH2 domain and a PTB domain. The PTB domain interacts with NPXY motifs in the integrin tails of certain β integrins.

(B) The model of Qian et al. The authors propose that binding of ligands to the tensin-3 SH2 domain is promoted not only by Src-dependent tyrosine phosphorylation of the ligand but also by tyrosine phosphorylation of the SH2 domain itself.

Perhaps consistent with the fact that tensins interact with both pro- and anti-oncogenic proteins, expression profiling and overexpression studies have supported both pro- and anti-oncogenic functions for tensins. In addition, the oncogenic functions of tensins may be isoform specific (Yam et al., 2006, Katz et al., 2007). Intrigued by this complexity, Qian et al. (Cancer Cell, this issue) decided to examine the functions of different tensins in both Src-transformed fibroblasts and a panel of human cancer cell lines. Knockdown of tensin-3 inhibited transformation by Src and cell migration and growth of the human cancer lines. These effects were not observed upon knockdown of the other tensin family members. In addition, knockdown of tensin-3 reduced the tumorigenicity of mammary cells derived from transgenic mice expressing polyoma middle-T (MMTV-PyMT).

Tensin-1 is known to be tyrosine-phosphorylated in Src-transformed chicken embryo fibroblasts (Davis et al., 1991). Qian et al. observed that in their panel of human cancer cell lines the level of phosphotyrosyl-tensin-3 correlated roughly both with malignancy and with the level of Src kinase activity. Furthermore the level of phosphotyrosyl-tensin-3 was strongly reduced by pharmacological inhibition of Src or by Src siRNA knockdown. Tensin-3 was also phosphorylated at tyrosine in the MMTV-PyMT mouse model, in which endogenous Src is activated, and this phosphorylation was reversed by the Src inhibitor PP2. In addition, recombinant Src could phosphorylate tensin-3 in vitro. All these and other observations presented in the paper argue that tensin-3 is an authentic Src substrate in mammalian tumor cells.

If tyrosine phosphorylation of tensin-3 is biologically significant, it might be expected to influence its association with functionally important partners. Indeed, inhibition of Src not only decreased the phosphorylation of two other known Src substrates, Cas and the RNA-binding protein Sam68, but also decreased the level of these two proteins that associated with tensin-3 in co-immunoprecipitation assays. This decrease in association must be due at least in part to decreased tyrosine phosphorylation of the tensin-3 ligands. In addition, mutational analysis indicated that three tyrosine residues in the SH2 domain (Y1173, Y1206 and Y1256) were, unexpectedly, Src phosphorylation sites. Mutation of two of these residues (2F) or all three (3F) in resulted in reduced co-precipitation of a GST-SH2 fusion protein with Src, Cas, FAK and Sam68. In contrast, however, interaction of the tensin-3 SH2 domain with two other ligands, DLC-1 and the integrin-linked kinase ILK, was unaffected by these mutations, indicating that the role of these three tyrosine residues is ligand-specific.

Thus three tyrosine residues in the tensin-3 SH2 domain are phosphorylated by Src, and the same residues appear to promote binding to ligands. Putting two and two together, the most obvious interpretation of these findings is that phosphorylation of these residues promotes binding. And indeed, in both co-precipitation and Far Western assays, tyrosine phosphorylation of the SH2 domain was found to promote interaction with Src, Cas and FAK, as well as with unidentified phosphotyrosyl-proteins in lysates of Src-transformed cells. The authors therefore proposed that phosphorylation of these residues directly affects ligand binding, as shown in Figure 1B. This interpretation is based on homology modeling of the tensin-3 SH2 domain, which suggests that the two tyrosine residues most significantly involved in this regulation are located close to the ligand-binding site of the SH2 domain. The binding sites within the ligands were not defined in this study, so that is formally possible that phosphorylated and non-phosphorylated SH2 domains might bind different phosphotyrosyl-sites in their ligands. In addition the interaction of tensin-3 with Src may be mediated in part by the Src SH2 domain, with the phosphotyrosyl residues in tensin-3 acting as docking sites. More detailed biophysical and structural studies using defined peptide ligands will be needed to sort this out. Nevertheless it is clear that the tyrosine phosphorylation of the tensin-3 SH2 domain is somehow affecting its function, an observation for which there is no precedent.

Given these findings, what is the biological significance of this novel mode of regulation? Does the tyrosine phosphorylation of the tensin-3 SH2 domain affect its oncogenic function? Consistent with this idea, GST fusions containing the tensin-3 SH2 domain functioned as dominant-negative mutants in cell migration and colony growth assays, while the 2F and 3F substitution mutants did not. Furthermore overexpression of full-length tensin-3 promoted cell migration and colony growth, and this effect was again abolished by the 2F mutation. As noted earlier, knockdown of tensin-3 reduced tumorigenicity in the MMTV-PyMT mouse model. It would be very interesting to determine if the effects of tensin-3 knockdown in this system could be reversed by wild-type tensin-3 but not by the 2F or 3F mutant, as would be predicted by the authors’ model, but this rescue experiment was not performed.

In summary, although the precise molecular details remain to be clarified, Lowy and colleagues have uncovered a novel mode of regulation of an SH2 domain that affects both the biochemical properties of the SH2 domain and the biological effects of the molecule in which it resides. It remains to be determined whether other SH2 domains are regulated in a similar manner. The ligand-specificity of this mode of regulation is intriguing, since Src, Cas and FAK, whose association with tensin-3 is tyrosine phosphorylation-dependent, are all pro-oncogenic, whereas the association of tensin-3 with DLC1, a tumor suppressor, is not affected by tyrosine phosphorylation. Thus the biological complexity of tensin-3 function – oncogenic in some contexts, anti-oncogenic in others – may reflect not only the complexity of the targets to which it binds but also the complexity of its regulation.