The GIT-Associated Kinase PAK Targets to the Centrosome and Regulates Aurora-A  Zhuo-shen Zhao, Jet Phey Lim, Yuen-Wai Ng, Louis Lim, Ed Manser  Molecular Cell  Volume 20, Issue 2, Pages 237-249 (October 2005) DOI: 10.1016/j.molcel.2005.08.035 Copyright © 2005 Elsevier Inc. Terms and Conditions

Figure 1 The Focal Adhesion Protein GIT1 Localizes to Centrosomes (A) Time-lapse confocal images of a live NIH3T3 cell stably expressing GFP-GIT1 show that GFP-GIT1 protein is concentrated at the centrosome (arrowhead). Images from two time points are shown. (B) Endogenous GIT1 in NIH3T3 cells was detected with affinity-purified anti-GIT1 antibody (in green). Centrosomes in these cells were visualized with γ-tubulin mAb (in red and highlighted by an arrowhead). (C) Endogenous PIX in NIH3T3 cells was stained with affinity-purified anti-PIX antibody (in green, gift from Dr. Eung-Gook Kim) as in (A). The cell nucleus was stained with DAPI (in blue). This PIX antibody stains both centrosomes (arrowhead) and focal adhesions in interphase cells. Scale bars, 5 μm. Molecular Cell 2005 20, 237-249DOI: (10.1016/j.molcel.2005.08.035) Copyright © 2005 Elsevier Inc. Terms and Conditions

Figure 2 GIT Targets PIX and PAK to the Centrosome (A) Centrosome localization of PIX and PAK requires GIT binding. GFP-tagged GIT2, PAK1, PAK1-KD (kinase-dead mutant), GIT1ΔC (C-terminal paxillin binding domain deletion), βPIX, and βPIXΔC (GIT binding domain deletion) were transiently expressed in an RFP-centrin expressing NIH3T3 cell line. Scale bar, 5 μm. (B) PAK phosphorylation of GIT1 takes place at the centrosome. PAK1 phosphorylation of GIT1 Ser517 was assessed in HeLa cells by a phospho-specific antibody costained for γ-tubulin. Phospho-GIT1 staining (green) is shown in the left panels, and the merged images are shown in the right panels; γ-tubulin in red and DNA in blue. Alignment of this PAK phosphorylation site in GIT1 with the corresponding GIT2 region shown. The position of phospho-serine is marked in red. Note adjacent sequences N-terminal to S517 are not conserved in GIT2. The centrosome in the interphase cell is marked by arrowheads. Molecular Cell 2005 20, 237-249DOI: (10.1016/j.molcel.2005.08.035) Copyright © 2005 Elsevier Inc. Terms and Conditions

Figure 3 Targeting PAK to the Centrosome Activates the Kinase (A) Constructs encoding PAK1, centrosome-targeted AKAP-PAK1 (A), GIT1(1–119)-PAK1 (G), AKAP-PAK1(S76P), GIT1(1–119)-PAK1(S76P), or nontargeted controls were transiently expressed in Cos7 cells. The PAK S76P mutant lacks Cdc42 or Rac1 binding. The status of the kinase was assessed by Western analysis using phospho-PAK (pT422) antibody: protein loading is indicated by Coomassie brilliant blue staining (CBB). (B) Removal of centrosomes prevents activation of PAK fused with a centrosome targeting motif. A schematic diagram of experimental procedures is shown. The centrosome in each NIH3T3 cell expressing RFP-centrin was identified by fluorescent microscopy, and the portion of cytoplasm containing centrosome (or equivalent region without centrosome) was excised with a glass microfilament (Murata-Hori et al., 2004). After 60 min recovery, the nucleus of the centrosome-free cell or control cell was microinjected with plasmid DNA encoding GFP-CenPAK1. Two hours later, cells were fixed and stained for anti-pPAK1(S198). The top panels show typical results (total n = 12) obtained with control cells (a–c) versus centrosome-excised cells (d–f). In controls, the expression of GFP coincided with pPAK detected but centrosome-excised cells did not generate pPAK staining cells. Lines in (a) and (d) indicate the position of the glass cutter. Centrosomes in cut cells are indicated by solid arrowheads and in surrounding uncut cells by open arrowheads. Note, centrosomes are present in the “cytoplast” cut off from the experimental cell in (d). Enlarged images of centrosomes in the control cell and in excised cytoplast are shown in the insets. The dotted lines in (b) and (e) indicate outline of the cytoplast from either cell (more details are in Figure S3). Molecular Cell 2005 20, 237-249DOI: (10.1016/j.molcel.2005.08.035) Copyright © 2005 Elsevier Inc. Terms and Conditions

Figure 4 Active PAK Interacts with Aurora-A (A) Constructs encoding FLAG-tagged wild-type (wt) full-length PAK1 (lane 1) PAK1(1–250) (lane 2), PAK1K298A(206–544) (lane 3), and centrosome-targeted CenPAK1 (lane 4) were coexpressed with GST-Aurora-A in Cos7 cells. GST-Aurora-A was recovered on glutathione Sepharose, and associated PAK proteins were detected with anti-FLAG antibody. Recovered GST-Aurora-A was visualized with Coomassie brilliant blue stain. FLAG-tagged p53 (lane 5) and GFP-FLAG-NM23-H1 (lane 6) were used as positive controls for Aurora-A binding proteins. (B) GST-tagged full-length Aurora-A and deletions as indicated were coexpressed with a kinase-dead but “open” form of PAK1 with L107F/K298A substitutions; GST fusions were analyzed for coprecipitated PAK by using anti-FLAG antibody. Molecular Cell 2005 20, 237-249DOI: (10.1016/j.molcel.2005.08.035) Copyright © 2005 Elsevier Inc. Terms and Conditions

Figure 5 Active PAK Selectively Interacts with Aurora-A, but Not PIX (A) Endogenous PAK coimmunoprecipitates with Aurora-A. HeLa cells were synchronized by double thymidine block and released for 20 hr in the presence of 40 ng/ml nocodazole. Endogenous PAK was immunoprecipitated with anti-PAK2 polyclonal antibody, and associated Aurora-A was detected by the anti-Aurora-A mAb 35C1. (B) Aurora-A selectively interacts with active PAK. GST-Aurora-A was coexpressed in Cos7 cells with FLAG-tagged wt PAK, centrosome-targeted PAK (Cen-PAK), β1PIX, or GIT1 as indicated. GST-Aurora-A was recovered on glutathione Sepharose, and associated protein was detected by Western blotting. Aurora-A loading was assessed by Coomassie staining, and expression levels of FLAG-tagged constructs were determined by Western analysis. (C) Active and inactive PAK1 interact with different partners. Aurora-A interacts with active forms of PAK1, whereas GIT-PIX prefers inactive PAK1. GST-Aurora-A (lanes 1 and 2) or GST-GIT1 (lanes 3, 4, and 5) was coexpressed with FLAG-tagged PAK, PIX, and centrosome-targeted PAK1 (CenPAK) in Cos7 cells as indicated in the table in the upper left. GST-Aurora-A and GST-GIT1 were recovered on glutathione Sepharose beads, and associated proteins were analyzed by Western blotting. Activated PAK was detected by antibody to phospho-S198/203 PAK1. The Coomassie staining shows recovered GST-fusion protein (black arrowhead pointing right). Coprecipitating βPIX (black arrowhead head pointing left) and PAK1 (asterisk; lane 2, activated CenPAK; lane 4, PAK1; and lane 5, CenPAK). Open arrowhead points to endogenous GIT1, recovered with GST-GIT1. Activated CenPAK is recovered in sufficient quantity with GST-Aurora to be seen by Coomassie staining (lane 2). Molecular Cell 2005 20, 237-249DOI: (10.1016/j.molcel.2005.08.035) Copyright © 2005 Elsevier Inc. Terms and Conditions

Figure 6 PAK Phosphorylates and Activates Aurora-A (A) PAK1 phosphorylates Aurora-A in vitro. E. coli-expressed 6 His-tagged kinase-inactive Aurora-AD274A was incubated with E. coli-expressed GST-PAK1(107F) and 0.1 mM ATP (left) or with 10 μM γ32P-ATP (center and right). The left panel shows GST-PAK and Aurora-A levels, detected on a Coomassie-stained PVDF blot (CBB), which was probed for phospho-T288 Aurora A (bottom box). Similarly, the right panel shows a Coomassie-stained gel (CBB) with corresponding autoradiograph in the middle panel to detect 32P-labeled PAK and Aurora-A. (B) Active PAKs promote activation of Aurora-A in Cos7 cells. GST-Aurora-A was cotransfected with various FLAG-tagged PAK constructs encoding PAK1(L107F), GIT1(1-119)-PAK, GIT1(1–119)PAK1(S76P), AKAP-PAK1(S76P), and GIT1(1–119) PAK1(K298A). GST-Aurora-A activation status was assessed by Western blotting with the phospho-T288 antibody. GST-Aurora-A on the Western blot filter was stained with Coomassie (middle). PAK1 expression levels were determined by Western blotting with anti-FLAG antibody (bottom). (C) Identification of PAK1 phosphorylation sites on Aurora-A. A peptide array displaying human Aurora-A peptides (13-mers, nine amino acids overlap, A1–I3) was incubated with 10 μg/200 μl of recombinant PAKl(L107F) and γ32P-ATP at 30°C for 1 hr. After washing, the peptide array was exposed to film for 10 min. Histones 1, 2A, 3, and 4, MBP, α/β-casein, and Tau proteins were included as control substrates. The sequences of phosphorylated peptides (asterisks) are indicated below, with likely phosphorylated residues marked in gray. (D) Thr288 and Ser342 phosphorylation contribute to Aurora-A kinase activity in mitotsis. GST-Aurora-A was transiently transfected into Cos7 cells subsequently blocked at the mitosis phase with 40 ng/ml nocodazole for 20 hr. Aurora-A activity was determined with 10 μM γ32P-ATP and GST-histone H3(1–21) as substrate (30°C for 30 min). The amount of Aurora-A kinase was normalized by Coomassie staining (as shown); T288 phosphorylation appears responsible for the size shift. The bottom panel shows activity toward histone H3(1–21) quantified after autoradiography (as average with SEM) relative to wt Aurora-A from three independent experiments. Molecular Cell 2005 20, 237-249DOI: (10.1016/j.molcel.2005.08.035) Copyright © 2005 Elsevier Inc. Terms and Conditions

Figure 7 PIX Depletion Affects Aurora-A at the Centrosome (A) PIX RNAi-mediated knock down of endogenous βPIX in nonsynchronized and synchronized HeLa cells. In these experiments, control siRNA (lanes 1 and 2) or PIX siRNA (lanes 3 and 4) were added at 20 nM (two identical plates shown) and βPIX expression in nonsynchronized cells was analyzed by Western blotting after 48 hr by using an anti-PIX mAb. In lanes 5 and 6, βPIX expression was monitored at 49 hr after transfection and 6 hr after release from a double thymidine block. Rho-GDI levels provide a loading control. (B) Synchronized HeLa cells as for lanes 5 and 6 of (A) were stained for Aurora-A (green) and DNA (DAPI in blue). A typical microscopy field from each sample is presented. (C) Quantification of Aurora-A-positive centrosomes in βPIX siRNA-transfected cells. Control versus PIX siRNA-transfected cells (n = 50) were examined by fluorescent microscopy and scored for Aurora-A-positive centrosomes. Black bars represent a similar experiment in which glutathione S-transferase (GST), the centrosome-localized domain GIT1(1–119) (GST-GIT1-N), kinase inhibitory domain PAK83–149 (GST-PAK-KID), or full-length kinase-inactive PAK1(K298A) designated FL-PAK-KD. After 6 hr release from a thymadine block, HeLa cells were stained for GST expression and scored for Aurora-A-positive centrosome staining (>50 transfected cells). All transfected cells contained centrosomes that were γ-tubulin positive. The data represent average numbers and SEM in three independent experiments. Molecular Cell 2005 20, 237-249DOI: (10.1016/j.molcel.2005.08.035) Copyright © 2005 Elsevier Inc. Terms and Conditions