Talin phosphorylation by Cdk5 regulates Smurf1-mediated talin head ubiquitylation and cell migration

Abstract

Cell migration is a dynamic process that requires temporal and spatial regulation of integrin activation and focal adhesion assembly/disassembly. Talin, an actin and beta-integrin tail-binding protein, is essential for integrin activation and focal adhesion formation. Calpain-mediated cleavage of talin has a key role in focal adhesion turnover; however, the talin head domain, one of the two cleavage products, stimulates integrin activation, localizes to focal adhesions and maintains cell edge protrusions, suggesting that other steps, downstream of talin proteolysis, are required for focal adhesion disassembly. Here we show that talin head binds Smurf1, an E3 ubiquitin ligase involved in cell polarity and migration, more tightly than full-length talin does and that this interaction leads to talin head ubiquitylation and degradation. We found that talin head is a substrate for Cdk5, a cyclin-dependent protein kinase that is essential for cell migration, synaptic transmission and cancer metastasis. Cdk5 phosphorylated talin head at Ser 425, inhibiting its binding to Smurf1, thus preventing talin head ubiquitylation and degradation. Expression of the mutant tal(S425A), which resists Cdk5 phosphorylation thereby increasing its susceptibility to Smurf1-mediated ubiqitylation, resulted in extensive focal adhesion turnover and inhibited cell migration. Thus, talin head produced by calpain-induced cleavage of talin is degraded through Smurf1-mediated ubiquitylation; moreover, phosphorylation by Cdk5 regulates the binding of Smurf1 to talin head, controlling talin head turnover, adhesion stability and ultimately, cell migration.

Figure 1. Cdk5 phosphorylates talin at Ser₄₂₅

a, Direct phosphorylation of TH by Cdk5 in vitro. Recombinant His-tagged TH (2.8 µg) was incubated with active Cdk5/p25 in 40 µl of kinase buffer containing 40 µM [γ-³²P]ATP (10 µCi). A 1:8 stoichiometry of phosphorylation of the TH was achieved in the experiments. b, Phosphorylation of Ser₄₂₅ on talin by Cdk5 in vitro. HA-talin, -tal_S425A or -tal_S425D were immunoprecipitated from transfected CHO-K1 cells, and incubated with active Cdk5/p35 in 50 µl of kinase buffer containing 20µM [γ-³²P]ATP (10 µCi). The numbers on the top indicates phosphorylation levels (% of wt talin) quantified as described in “Methods”. c, 2-D phospho-peptide mapping analysis of wt talin, tal_S425A, and tal_S425D. d, Talin is phosphorylated at Ser₄₂₅ by Cdk5 in vivo. Cells were transfected with HA-talin or –talS425A in the presence or absence of Cdk5/p35. After 18 hr of transfection, cells were incubated with [³²P]orthophosphoric acid (0.8 mCi/ml) for 12 hr. Cells were then harvested and analyzed as described in “Methods”. Over-expression of Cdk5 promotes phosphorylation of talin but not of tal_S425A. e, 2-D phospho-peptide mapping analysis of talin and tal_S425Ain the absence and presence of recombinant Cdk5.

Figure 2. Expression of phosphorylation-deficient talin mutants inhibits cell migration

a, Inhibition of PDGF-stimulated migration of SH-SY5Y neuroblastoma cells by expressing EGFP-tal_S425A or –tal_S425D. SH-SY5Y cells were transfected with EGFP-talin, -talS425A or –talS425D, using Nucleofactor Kit V (Amaxa). At 24 hr post-transfection, cells were trypsinized for transwell migration assays employing collagen I (10µg/ml)-coated filters and 20 nM PDGF in lower charmber. c and d, Inhibition of wound closure in CHO cells expressing EGFP-tal_S425A or –tal_S425D. Confluent monolayers of CHO cells stably expressing EGFP-talin, tal_S245A or –tal_S425D were wounded and images were taken when the wounds were made (left panels) and after incubating for 18 hr (right panels). Data are expressed as mean ± SEM of three independent experiments.

Figure 3. Talin phosphorylation by Cdk5 is important for the disassembly of focal adhesions

a, CHO cells stably expressing EGFP–talin, EGFP–tal_S425A, or EGFP-tal_S425D were plated on fibronectin (5µg/ml)-coated glass-bottom dishes (MatTek) and analyzed by time-lapse TIRF microscopy. Arrowheads point to disassembling focal adhesions. b, Quantification of the rate constants of EGFP–talin, EGFP–tal_S425A, or EGFP-tal_S425D assembly into and dissociation from focal adhesions. (c) Effects of roscovitine on the rate constants of EGFP–talin, assembly into and dissociation from focal adhesions. Quantifications are expressed as means ± SEM of 30 focal adhesions from 7 cells.

Figure 4. Talin phosphorylation by Cdk5 stabilizes lamellipodia

a, CHO cells expressing EGFP–talin, or EGFP–tal_S425A were plated on 5 µg/ml fibronectin and analysed by time-lapse TIRF microscopy. Arrow heads points to lamellipodium formation and retraction. b, Kymograph of lamellipodium dynamics in CHO cells expressing EGFP–talin or EGFP–tal_S425A. c, The lamellipodial persistence in cells expressing wild-type Talin is significantly higher than that of mutant S425A. d, There is no significant difference in protrusion velocities between cells expressing the wild-type and mutant S425A. Quantifications are expressed as means ± SEM.

Figure 5. Cdk5 phosphorylation inhibits Smurf1-mediated TH ubiquitination

a, Talin-Smurf1 interaction. Binding of EGFP-Smurf1 to immobilized GST-profilin, -tal1-433, -tal393-605, or -tal393-605_S425D. Smurf1 was detected by immunoblotting with anti-GFP. b, TH but not full-length talin binds Smurf1 tightly,. Binding of purified His-tagged TH or FL talin to immobilized GST-Smurf1 was detected by immunoblotting with anti-His tag mAb. c, TH and catalytically-inactive Smurf1(Sm_C699A) interact in cells. EGFP-TH and VSVG-Sm_C699A were co-transfected in combination with H-Ras_G12V into CHO-K1 cells. Anti-GFP immunoprecipitated EGFP-TH and bound Smurf1_C699A was detected with anti-VSVG. d, Inhibition of TH-Smurf1 interaction by Cdk5-mediated phosphorylation. HA-TH was purified from CHO-K1 cell lysates and phosphorylated in vitro with Cdk5/p35 in vitro, and binding to GST-Smurf1 was analyzed as in panel b. e, Inhibition of Smurf1-mediated ubiquitination Cdk5-mediated phosphorylation. Purified His-tagged TH was incubated 120 min with the indicated quantity of Cdk5/p25 and then with Myc-ubiquitin, E1, and UbcH7 in the presence or absence of GST-Smurf1. f, Effect of TH phosphorylation on ubiquitination in vivo. CHO-K1 cells were co-transfected with EGFP-TH, -TH_S452A, or -TH_S425D and HA-ubiquitini. Twenty-four hr post-transfection, cells were re-plated on fibronection for 5 hr and then were treated with MG132 (40µM) for an additional 4 hr before ati-HA immunoblotting. g, The poly-ubiquitination of EGFP-TH, -TH_S452A and -TH_S425D was quantified by densitometry of the rectangular area (see lane 3). Depicted are the mean and range of two independent experiments. h, CHO cells expressing EGFP–tal_S425A, were transfected with vector or DsRed-Smurf1_C699A and plated on fibronectin for 4 hr. The rate constants of EGFP–tal_S425A dissociation from focal adhesions were quantified by time-lapse microscopy (mean ± SEM, n=30). i, Proposed mechanism whereby talin phosphorylation by Cdk5 regulates TH ubiquitination, focal adhesion disassembly, lamellipodial dynamics, and cell migration. Full scans of western blots shown in b, c, e, and f are available in Supplementary Information, Fig. S7.