p38gamma regulates the localisation of SAP97 in the cytoskeleton by modulating its interaction with GKAP

Abstract

Activation of the p38 MAP kinase pathways is crucial for the adaptation of mammalian cells to changes in the osmolarity of the environment. Here we identify SAP97/hDlg, the mammalian homologue of the Drosophila tumour suppressor Dlg, as a physiological substrate for the p38gamma MAP kinase (SAPK3/p38gamma) isoform. SAP97/hDlg is a scaffold protein that forms multiprotein complexes with a variety of proteins and is targeted to the cytoskeleton by its association with the protein guanylate kinase-associated protein (GKAP). The SAPK3/p38gamma-catalysed phosphorylation of SAP97/hDlg triggers its dissociation from GKAP and therefore releases it from the cytoskeleton. This is likely to regulate the integrity of intercellular-junctional complexes, and cell shape and volume in response to osmotic stress.

Figure 1

Phosphorylation of SAP97/hDlg by SAPK3/p38γ depends on the interaction of these two proteins. (A) HeLa cells (I, II, III), differentiated PC12 (IV–IX) and differentiated SH5-SY5Y neuroblastoma (X, XI, XII) were stained with anti-SAPK3 or anti-SAP97 antibody, and subjected to fluorescence microscopy. SAPK3/p38γ and SAP97/hDlg are shown in green and red, respectively. In merged images, the co-localised signal is shown in yellow. Scale bar, 10 μM. (B) Co-immunoprecipitation of SAPK3/p38γ with SAP97. HEK293 cells were transfected with GFP-SAPK3(FL) or GFP-SAPK3 (lacking the last four amino acids (ΔC)). Endogenous SAP97 in the lysates was immunoprecipitated, and immunoblotted with anti-SAPK3 or anti-SAP97 antibodies. GFP-SAPK3 was immunoprecipitated using an anti-GFP antibody, and immunoprecipitates immunoblotted with anti-SAP97 or anti-SAPK3 antibodies. (C) Interaction of SAPK3/p38γ with SAP97PDZ domains. HEK293 cells were transfected with GFP-SAPK3(FL) and either GST, GST-SAP97(PDZ1), GST-SAP97(PDZ2) or GST-SAP97(PDZ3). After transfection the cells were lysed and GST-fusion proteins purified by affinity chromatography on GSH-Sepharose beads. The GFP-SAPK3 was immunoprecipitated using an anti-GFP antibody. The proteins were immunoblotted using an anti-SAPK3/p38γ antibody to detect GFP-SAPK3/p38γ, or anti-GST antibody to detect expression of GST-SAP97 fusion proteins. (D) Phosphorylation of SAP97 by SAPK3/p38γ is dependent on the carboxy-terminal four amino acids of the kinase. GST-SAP97 or MBP, both at 1 μM, were phosphorylated for the times indicated with 2.0 U/ml of either GST-SAPK3(FL) or GST-SAPK3(ΔC). The results are shown as the mean±s.e.m. of four experiments. (E) GST-SAP97 (filled bars) or MBP (open bars), each at 1 μM, were incubated for 30 min at room temperature with synthetic peptides (300 μM) corresponding to the C-terminal six (PKETAL) or eight (RVPKETAL) amino acids of rat SAPK3. GST-SAPK3/p38γ (black bars) or GST-SAPK4/p38δ (grey bars) were added to 0.2 U/ml and the reactions initiated with Mg[γ-³²P]ATP. Substrate phosphorylation is plotted as a percentage of that measured in the absence of each peptide. Results in (E) are shown as the mean±s.e.m. for triplicate determinations from a single experiment.

Figure 2

Hyperosmotic stress induces phosphorylation of endogenous SAP97/hDlg. (A) Identification of the sites on SAP97/hDlg phosphorylated in vitro by SAPK3/p38γ or SAPK4/p38δ. Rat GST-SAP97 was incubated for 1 h at 30°C with Mg[γ-³²P]ATP in the presence of 0.5 U/ml of SAPK3/p38γ or SAPK4/p38δ, and subjected to SDS–PAGE. The phosphorylated SAP97 was excised from the gel, digested with trypsin and the peptides separated by chromatography. The column was developed with an acetonitrile gradient (broken line) and ³²P-radioactivity is shown in full line. The phosphopeptides P1–P4 are indicated. To identify the residue phosphorylated in P2, it was subdigested with the protease Asp-N to give a smaller phospho-peptide (residues 427–445). P3 and P4 are a mixture of two peptides, each phosphorylated at a single residue. All residues were identified by a combination of techniques MALDI-TOF, Q-TOF, MS/MS, solid phase sequencing and phospho-amino-acid analysis. (B) Phosphorylation of SAP97/hDlg in HEK293 cells after cellular stress. Cells were incubated for 1 h with or without 10 μM SB203580 and/or 5 μM PD184352, then exposed for 15 min to 0.5 M sorbitol or to UV-C radiation (200 J/m²), followed by a 30 min incubation. Endogenous SAP97/hDlg was immunoprecipitated from 1–5 mg of cell lysate, the pellets immunoblotted using an antibody that recognises SAP97 phosphorylated at S¹⁵⁸ (Phos-Ser158), T²⁰⁹ (Phos-Thr209), S⁴³¹ (Phos-Ser431), S⁴⁴² (Phos-Ser442) and an antibody that recognises unphosphorylated and phosphorylated SAP97 equally well. The lanes in this panel are duplicates. (C) HEK293 were transfected with SAP97 WT or GST-SAP97 mutant in which S¹²² has been mutated to Ala. GST-SAP97 was immunoprecipitated from 50 μg of lysate, and the pellets were immunoblotted using Phos-Ser122 antibody. (D) HEK293 cells were incubated for 1 h with or without 400 μM TatSAPK3C(WT) or TatSAPK3C(AA) peptide, and then exposed for 15 min to 0.5 M sorbitol. The immunoprecipitated SAP97/hDlg was immunoblotted as above.

Figure 3

Generation of SAPK3/p38γ and SAPK4/p38δ knockouts. Diagram illustrating the targeting vector for the knockout of SAPK3/p38γ (A) and SAPK4/p38δ (B). The black boxes represent exons, the positions of the probes and the PCR primers used for genotyping are indicated by white boxes and black arrows, respectively. Genomic DNA purified from indicated ES cell lines were digested with BamHI, electrophoresed on a 1% agarose gel, transferred to nitrocellulose for Southern blotting. Genomic DNA purified from tail biopsy sample was used as a template for PCR, electrophoresed on a 1% agarose gel and examined by ethidium bromide staining. (C) Lysates from MEF WT, SAPK3/p38γ, SAPK4/p38δ and SAPK3/4 double knockout (30–50 μg of protein) were immunoblotted with antibodies that recognise specifically each p38. The lanes in this panel are duplicates. (D) MEF SAPK3/p38γ, SAPK4/p38δ and SAPK3/4 double knockout were exposed for 15 min to 0.5 M sorbitol. To examine the activation of p38s, SAPK3/p38γ or SAPK4/p38δ were immunoprecipitated from 2 mg of cell lysates, and immunoblotted with the p38α phospho-specific antibody that also recognises phosphorylated SAPK3/p38γ and SAPK4/p38δ. Alternatively, 50 μg of cell lysates was immunoblotted with the same phospho-specific antibody to detect active p38α.

Figure 4

Phosphorylation of endogenous SAP97/hDlg in mouse embryonic fibroblasts. MEF from WT, SAPK3/p38γ(−/−) (A), SAPK4/p38δ(−/−) (B) or SAPK3/4 double knockout mice (C) were incubated for 1 h with or without 10 μM SB203580 or 5 μM PD184352, then exposed for 15 min to 0.5 M sorbitol. Endogenous SAP97 was immunoprecipitated from 1–5 mg of cell lysate, the pellets immunoblotted using an antibody that recognises SAP97 phosphorylated at S¹⁵⁸ (Phos-Ser158), T²⁰⁹ (Phos-Thr209) or S⁴⁴² (Phos-Ser442), or with an antibody that recognises both unphosphorylated and phosphorylated SAP97. (D) Mouse embryonic fibroblasts from WT, SAPK3/p38γ(−/−) or SAPK4/p38δ(−/−) were incubated for 1 h with or without 400 μM TatSAPK3C(WT) or TatSAPK3C(AA), and then exposed for 15 min to 0.5 M sorbitol. The immunoprecipitated SAP97 was immunoblotted as above.

Figure 5

Association of SAP97/hDlg with the protein GKAP is regulated by phosphorylation. (A) Structural organisation of SAP97/hDlg, indicating that the protein CASK binds to the L27 domain, SAPK3/p38γ to the PDZ1 and PDZ3 domains, PBK to PDZ2 and GKAP to the GK domain. HEK293 cells were transfected with PBK or SAPK3/p38γ. Cells were left unstimulated or exposed for 15 min to 0.5 M sorbitol or to UV-C radiation (200 J/m²), followed by a 30 min incubation, and endogenous SAP97/hDlg was immunoprecipitated from 0.2–15 mg of cell lysate. The pellets were immunoblotted using antibodies that recognise CASK, PBK, SAPK3/p38γ, GKAP or SAP97/hDlg. (B) Endogenous SAP97/hDlg was immunoprecipitated from 0.2 mg of undifferentiated PC12 or 15 mg of HEK293 cell lysate. The pellet and 100 μg of protein from both, total lysates (T) (as loading control) or the supernatants (Sup.) were immunoblotted, using antibodies that recognise GKAP or SAP97/hDlg (upper panel). Quantification of the amount of protein detected is shown in the lower panel. (C) Recombinant SAP97 (500 ng) unphosphorylated or phosphorylated with SAPK3/p38γ was incubated with recombinant GST-GKAP (500 ng). After 10 min at 4°C, GKAP or SAP97 were immunoprecipitated, and the washed pellets and supernatants immunoblotted with an anti-GKAP antibody or anti-SAP97 antibody. (D) Endogenous SAP97 was immunoprecipitated from 0.5 mg of SH-SY5Y or undifferentiated PC12 lysate treated as in (A). The pellets were immunoblotted with the antibodies indicated in the figure. (E) HEK293 cells were transfected with WT GKAP and either WT GST-SAP97 or different GST-SAP97 mutants, and then exposed for 15 min osmotic shock (0.5 M sorbitol). Expressed SAP97 was immunoprecipitated from 0.5 mg of protein lysates and the pellets were immunoblotted with the antibodies indicated.

Figure 6

SAP97/hDlg is recruited to the cytoskeleton through GKAP. (A) HEK293 cells transfected with SAP97/hDlg either FL (GST-SAP97) or lacking the GK domain (GST-SAP97(ΔGK)) with or without GKAP were subjected to cellular fractionation, and 40 μg of protein from each cell fraction was immunoblotted using the antibodies indicated in the figure. (B) HEK293 cells co-transfected with SAP97/hDlg and GKAP or undifferentiated PC12 cells were left unstimulated or exposed for 15 min to 0.5 M sorbitol or to UV-C radiation (200 J/m²), followed by a 30 min incubation, and subjected to cellular fractionation. Then, 40 μg of protein from the cytoskeleton was immunoblotted using the antibodies indicated in the figure. (C) Untransfected undifferentiated PC12 cells (upper panel) or HEK293 cells (lower panel) transfected with SAP97/hDlg either FL (GST-SAP97) or lacking the GK domain (GST-SAP97(ΔGK)) or GST with GKAP (as indicated in the figure) and then exposed for 15 min osmotic shock (0.5 M sorbitol). Samples were then immunoprecipitated using anti-SAP97 or anti-GKAP antibodies or IgG as control, and the pellets were blotted with anti-cytokeratin, anti-GKAP or anti-GST antibody. (D) HEK293 cells were transfected with either WT GST-SAP97 or different GST-SAP97 mutants, and then exposed for 15 min osmotic shock (0.5 M sorbitol). Cytoskeletal fraction was extracted and 40 μg of protein blotted using anti-SAP97 antibody or anti-cytokeratin antibody as loading control. Results from two different experiments (Exp. 1 and Exp. 2) are shown. Quantification of the protein detected in panels B, C and D are shown in Supplementary Figure 4. (E) GKAP-SAP97/hDlg complex is associated with the cytokeletal fraction under normal physiological conditions. Changes in the osmolarity of the environment activate SAPK3/p38γ, which phosphorylates SAP97/hDlg, causing a conformational change and its dissociation from GKAP and therefore from the cytoskeleton.