Tyrosine 335 and proline-rich C-terminal sequence are required for Src-induced tyrosine phosphorylation of Dgk-α, and for interaction respectively with Src-SH2 and -SH3 domain
We and others have previously shown that Dgk-α is activated by growth factors in a Src-dependent manner, and that it is tyrosine phosphorylated and activated upon coexpression with either Src or Lck (Cutrupi et al., 2000; Cipres et al., 2003). To verify that Dgk-α could be directly phosphorylated by Src tyrosine kinase activity, we incubate partially purified glutathione-S-transferase (GST)-Dgk-α with recombinant Src in presence of Mg and ATP. In these conditions, Src promotes a strong tyrosine phosphorylation of GST-Dgk-α as verified by western blot with anti-phosphotyrosine antibodies (Figure 1a). Observing Dgk-α sequence, we noted two tyrosine residues featuring isoleucine in the -1 position, FLKIYLEVDN and PPSSIYPSVLA (Figure 2), suggesting strong substrate selection by Src (Songyang and Cantley, 1995; Schmitz et al., 1996). To verify whether these two tyrosine residues are substrates of Src tyrosine kinase activity, we coexpressed Src in COS cells with myc-tagged Dgk-α, either wt, Y60F, or Y335F. Tyrosine phosphorylation of Dgk-α was evaluated by anti-phosphotyrosine western blot of anti-myc immunoprecipitates (Figure 1b). Upon coexpression with Src, Myc-Dgk-α-Y335F does not feature any detectable tyrosine phosphorylation, while both Myc-Dgk-α wt and Myc-Dgk-α-Y60F mutant are tyrosine phosphorylated. Anti-myc and anti-Src western blots confirmed uniform expression of transfected Src and Dgk-α proteins. Thus, this experiment indicates that Y335 is the major site of phosphorylation of Dgk-α upon coexpression with Src, suggesting that contribution of Y60 is negligible.
As optimal protein-substrate sequences for Src tyrosine kinase activity provides optimal consensus sequences for binding of SH2 domain of Src itself (Songyang et al., 1993; Songyang and Cantley, 1995), we decided to investigate the ability of Y335 of Dgk-α to mediate interaction with Src-SH2 domain in an in vitro pull-down assay. Immobilized GST-Src-SH2 fusion protein was incubated with cell lysates obtained from serum cultured COS cells transfected with either empty vector or Myc-Dgk-α wt or mutants. Myc-Dgk-α wt was pulled down by GST-SrcSH2, but not by GST alone, indicating that Dgk-α interacts with Src-SH2 domain (Figure 3a). The interaction between Dgk-α and the SH2 domain is specific, as the GST-Src-SH2 R175L mutant, unable to recognize the phosphorylated tyrosine (Yeo et al., 2006), does not interact with Myc-Dgk-α (Table 1). Furthermore, Myc-Dgk-α-Y335F, which shows a dramatically reduced phosphorylation upon coexpression with Src, fails to associate with GST-Src-SH2 in the pull-down assay, while Myc-Dgk-α-Y60F interacts with GST-Src-SH2 as well as Myc-Dgk-α wt. In summary, these experiments demonstrate that Src-SH2 domain interacts selectively with the phosphorylated Y335 of Dgk-α. The interaction of Dgk-α is not limited to Src-SH2 domain, as, at least in vitro, Dgk-α interacts also at similar or lower efficiency, with SH2 domains of Bruton's tyrosine kinase (Btk), c-phospholipase C (PLC)γ, Grb2 and Lck, but not with SH2 domains of Abl, n-PLCγ and p85n (Table 1).
As several Src substrates, such as p130Cas, become tyrosine phosphorylated upon interaction of their proline-rich motif with Src-SH3 domain (Pellicena and Miller, 2001), we verified whether Dgk-α interacts with Src-SH3 domain in a pull-down assay. Immobilized GST-Src-SH3 was incubated with cell lysates obtained from serum cultured COS cells, either control or expressing Myc-Dgk-α-wt or mutants. Myc-Dgk-α-wt and Myc-Dgk-α-Y335F were specifically pulled down by immobilized GST-Src-SH3, but not by GST alone (Figure 3b), indicating that indeed Dgk-α interacts with Src-SH3 domain. The interaction between Dgk-α and the SH3 domain is specific, as the GST-Src-SH3-D99N a SH3 mutant, which is impaired in poly-proline binding (Weng et al., 1995), does not interact with Dgk-α. Although Dgk-α does not contain a consensus sequence for SH3 interaction (PxxP), it features a highly conserved C-terminal proline-rich sequence (PMLMGPPPR, Figure 2). Thus, we generated two deletion mutants lacking respectively the entire C-terminal half of Dgk-α (Myc-Dgk-α-STOP) or the last 13 amino acids PPPRSTNFFGFLS (Myc-Dgk-α-ΔP). Both mutants were assayed in the GST-Src-SH3 pull-down assay. Figure 3b shows that both Myc-Dgk-α-ΔP and Myc-Dgk-α-STOP mutants, different from Myc-Dgk-α-wt and Myc-Dgk-α-Y335F, are not pulled down by immobilized GST-Src-SH3 fusion protein. These data indicate that the proline-rich region is required for Dgk-α interaction with Src-SH3 (Figure 3a). The interaction of Dgk-α is not limited to the SH3 domain of Src, but SH3 domains of both Lck and Abl interact as well with Dgk-α (Table 1).
Based on the model proposed for tyrosine phosphorylation of p130Cas (Pellicena and Miller, 2001), we verified whether proline-rich tail of Dgk-α is required for Src-mediated tyrosine phosphorylation. We coexpressed in 293 T cells Myc-Dgk-α either wt, ΔP or Y335F with Src-Y527F, an activated form of Src. Tyrosine phosphorylation of Myc-Dgk-α in anti-myc immunoprecipitates was assayed by anti-phosphotyrosine western blot. Figure 4 shows that Myc-Dgk-α-ΔP and Myc-Dgk-α-Y335F mutants are not tyrosine phosphorylated upon coexpression with Src-Y527F, while Myc-Dgk-α-wt is tyrosine phosphorylated. Anti-myc and anti-Src western blots confirm uniform expression of transfected proteins, either wt or mutant.
Finally, these data demonstrate, both in intact cells and in vitro, that the proline-rich tail of Dgk-α is required for interaction with Src-SH3 domain as well as for its tyrosine phosphorylation, suggesting that interaction of Dgk-α with Src SH3 domain may precede its tyrosine phosphorylation.
The data presented so far clearly indicate that Y335 and the pro-rich C-terminal sequence of Dgk-α are the major determinants for its Src-mediated tyrosine phosphorylation, and provide the reagents to investigate whether phosphorylation of Y335 is required for Src- and HGF-induced enzymatic activation of Dgk-α. Indeed, while several evidence have firmly showed that activation of Dgk-α by growth factors depends on Src family tyrosine kinases, the putative role of its tyrosine phosphorylation in growth factor-induced enzymatic activation has been elusive (Cutrupi et al., 2000; Cipres et al., 2003; Baldanzi et al., 2004; Bacchiocchi et al., 2005).
The enzymatic activity of Myc-Dgk-α either wt, Y335F or ΔP, were assayed upon co-incubation with Src, in an in vitro activation assay performed with crude lysates obtained from either Src- or Dgk-α-transfected cells. Through this assay, we had previously shown that enzymatic activity of Myc-Dgk-α wt is significantly increased upon co-incubation with Src cell lysates (dark column), as compared with control lysates (white columns) (Cutrupi et al., 2000; Figure 5). Conversely, the enzymatic activities of either Myc-Dgk-α-Y335F or Myc-Dgk-α-ΔP mutant are not significantly stimulated upon co-incubation with Src in vitro (Figure 5). This finding provides the first direct demonstration that both Y335 and proline-rich sequence are required for activation of Dgk-α in vitro.
Next, we investigated whether both Y335 and proline-rich sequence are also required for HGF-induced activation of Dgk-α in intact cells. We assayed the enzymatic activity of Myc-Dgk-α-wt, Y335F or ΔP mutant (Figure 6a), transiently transfected in COS cells, either control or HGF-stimulated. The enzymatic activity was measured in whole-cell lysates, as described previously; under these conditions, the contribution of endogenous Dgk to the total Dgk activity is negligible (Cutrupi et al., 2000; and data not shown). Figure 6a indicates that, while enzymatic activity of Myc-Dgk-α-wt is stimulated by HGF, the enzymatic activities of either Myc-Dgk-α-Y335F or Myc-Dgk-α-ΔP mutants are not stimulated on HGF cell stimulation. Expression of Myc-Dgk-α-wt and mutants was verified by anti-myc western blot (Figure 6a, lower panel).
Consistently, the enzymatic activity of the double mutant Myc-Dgk-α-Y335F-ΔP, featuring a lower basal activity, is not further activated upon HGF stimulation, as assayed in anti-myc immunoprecipitates (Figure 6b). The expression of Myc-Dgk-α-wt and Myc-Dgk-α-Y335F-ΔP was verified by anti-myc western blot (Figure 6b, lower panel).
To provide further evidence for the role of Y335 and proline-rich sequence as major determinants of Src-mediated activation of Dgk-α in intact cells, we investigated tyrosine phosphorylation and activation of Myc-Dgk-α either wt, Y335F or ΔP in transiently transfected Madin-Darby canine kidney (MDCK)-ts-v-Src epithelial cells (Figure 7). In these cells, ts-v-Src tyrosine kinase activity is impaired at 40°C, and is activated upon shifting the cell culture to 35°C (Behrens et al., 1993). Under these conditions, differently from COS and 293T cells, Myc-Dgk-α is expressed at low level, and it does not significantly affect total Dgk activity assayed in whole-cell lysates (Figure 7b).
Shifting MDCK-ts-v-Src cells to the permissive temperature results in both tyrosine phosphorylation (Figure 7a) and enzymatic activation (Figure 7b) of Myc-Dgk-α wt, as evaluated respectively by anti-phosphotyrosine western blot of anti-myc immunoprecipitates and in vitro Dgk-α assay. Next, we verified whether v-Src induces tyrosine phosphorylation and stimulates enzymatic activity of both Myc-Dgk-α-Y335F and Myc-Dgk-α-ΔP. Activation of ts-v-Src fails to induce tyrosine phosphorylation of both Myc-Dgk-α-Y335F and Myc-Dgk-α-ΔP (Figure 7a), and fails to stimulate their enzymatic activity (Figure 7b). Expression of both mutants is comparable to the wild type (Figure 7).
In summary, these results, providing the first evidence in vivo that Dgk-α is a target of oncogenic Src, demonstrate that Src regulates Dgk-α in vivo through phosphorylation of Y335. In addition, as both enzymatic activation and tyrosine phosphorylation of Dgk-α depend on its proline-rich sequence, these data suggest that interaction of Dgk-α proline-rich sequence with Src-SH3 domain is a prerequisite for its phosphorylation and enzymatic activation.
As Dgk-α is a cytosolic enzyme which associates to the plasma membrane upon growth factor stimulation (Flores et al., 1996; Sanjuán et al., 2003), we investigated whether phosphorylation of Dgk-α on Y335 regulates its recruitment to the membrane upon HGF stimulation. To address this question, we investigated the subcellular localization of GFP tagged Dgk-α wt, Y335F and ΔP mutants, transiently transfected in MDCK cells. We observed that in most of control transfected cells, GFP-Dgk-α wt is localized exclusively in the cytosol, and that upon HGF stimulation it translocates at the plasma membrane in the majority of transfected cells (70%) (Figure 8a, Figure 9c). In addition, the kinase dead mutant (GFP-Dgk-α-k-) behaves as the wild type, being diffuse in the cytoplasm in control cells and associates to the plasma membrane in HGF-stimulated cells (Figure 8b). HGF-induced membrane recruitment was dependent on Src activity, as it was reduced of 50% by pharmacological inhibition of Src with 10 μ M PP2 (Figure 8).
To verify whether tyrosine phosphorylation of Dgk-α mediates HGF-induced membrane recruitment of Dgk-α, we investigated the subcellular localization of both Y335F and ΔP mutants. Surprisingly, in most of control-transfected cells, GFP-Dgk-α-Y335F is associated to intracellular vesicles. Similarly, GFP-Dgk-α-ΔP is also associated to intracellular vesicles, albeit of different shape and size, in all transfected cells. Upon HGF stimulation, neither mutant translocates at the plasma membrane, while their vesicular localization is not affected (Figure 9).
These observations demonstrate that Y335 and proline-rich sequence are required for proper localization of Dgk-α, and suggest that phosphorylation of Y335 is a key event for HGF-induced recruitment to the plasma membrane. In addition, the vesicular localization of both GFP-Dgk-α-Y335F and GFP-Dgk-α-ΔP suggest that the recruitment of Dgk-α to the plasma membrane may occur through vesicular traffic. If this holds true, we should expect that specific inhibition of vesicular traffic between the inner cytosol and the plasma membrane by Brefeldin A (BFA) treatment, would result in accumulation of GFP-Dgk-α-wt in intracellular vesicles (Lippincott-Schwartz et al., 1989). Indeed, upon 15 min of treatment with 10 μ M BFA, even GFP-Dgk-α-wt associates to intracellular vesicles in unstimulated cells and fails to translocate to the membrane following HGF stimulation (Figure 10).
These observations strongly suggest that HGF-induced recruitment of Dgk-α to the plasma membrane depends on the integrity of the vesicular transport network, requires phosphorylation of Y335 by Src, but does not require its enzymatic activity.
As we previously showed that activation of Dgk-α is required for HGF- and VEGF-induced cell migration (Cutrupi et al., 2000; Baldanzi et al., 2004), we investigated whether Y335 contributes to the transduction of HGF pro-migratory signaling. Although HGF does not stimulate chemotaxis of COS-7 cells, transient overexpression of Myc-Dgk-α-wt makes COS-7 cells able to migrate in response to HGF in a transwell chemotaxis quantitative assay (Figure 11a). This observation provides a functional assay to verify the requirement for phosphorylation of Y335 to transduce HGF-induced migratory signaling. Figure 11 indicates that the expression of Myc-Dgk-α-Y335F mutant impairs HGF-induced motility of COS cells, as compared with wild type. These data lend further support to the hypothesis that activation and membrane recruitment of Dgk-α, occurring through its phosphorylation on Y335, are required for HGF-induced migratory signaling.
Next, we asked whether Dgk-α constitutive recruitment to the plasma membrane provides sufficient signaling to stimulate cell motility. Sanjuan et al. (2001) had previously shown that myristylated Dgk-α is constitutively active and associated to the plasma membrane. Transient expression of myr-Dgk-α in COS cells, enhances threefold spontaneous migration of serum-starved COS cells in absence of HGF in transwell chemotaxis assay and enhanced spontaneous cell migration in a wound healing assay (Figures 11b and c). These observations carried out in two different migration assays indicate for the first time that constitutive activation of Dgk-α at the cell membrane provides rate limiting intracellular signals, both necessary and sufficient to stimulate cell migration.