Identification of Protein Networks Associated with the PAK1-βPIX-GIT1-Paxillin Signaling Complex by Mass Spectrometry Mark W. Mayhew,*,† Donna J. Webb,‡ Mykola Kovalenko,† Leanna Whitmore,† Jay W. Fox,§ and Alan F. Horwitz† Department of Cell Biology, University of Virginia, Charlottesville, Virginia 22908, Department of Biological Sciences and Vanderbilt Kennedy Center for Research on Human Development, Vanderbilt University, Nashville, Tennessee 37235, and W.M. Keck Biomedical Mass Spectrometry Laboratory and Biomedical Research Facility, University of Virginia, Charlottesville, Virginia 22908 Received April 3, 2006
The process of cell motility involves coordinate signaling events among proteins associated in interactive integrin-linked networks. Mass spectrometric analysis of immunoprecipitation-derived protein mixtures have provided efficient means of identifying proteomes. In this study, we investigate strategies to enhance the detection of interactome proteins for the known signaling module: PAK1, βPIX, GIT1, and paxillin. Our results indicate that near-endogenous expression levels of bait protein enhances the identification of associated proteins, and that phosphatase inhibition augments the detection of specific protein interactions. Following the analysis of a large pool of spectral data, we have identified and mapped clusters of proteins that either share common interactions among the four bait proteins of interest or are exclusive to single bait proteins. Taken together, these data indicate that biochemical manipulations can enhance the ability for LC-MS/MS to identify interactome proteins, and that qualitative screening of multiple samples leads to the compilation of proteins associated with a known plexus. Keywords: PAK • PIX • GIT • paxillin • proteomics • mass spectrometry • MS/MS • cell migration • binding partners
Introduction Cell migration utilizes a complex signaling network that regulates and integrates a multitude of individual processes and signals that comprise the migration signaling plexus.1,2 A key aspect of this plexus is the multitude of protein interactions that spacially and temporally organize numerous kinases, lipases, adaptors, phosphatases, and G-proteins with their respective activators and effectors and, thus, serve to locally regulate migratory phenomena like polarization, adhesion, protrusion, and contraction.2,3 Many of these proteins operate in complexes that are linked to other complexes. One goal of proteomics is to provide protein-binding data that aid in piecing together the interactions that comprise the proteome responsible for various cellular phenomena. One approach to identifying the proteome for a process of interest is to immunoprecipitate tagged bait proteins and analyze the associated proteins via LC-MS/MS.4 This approach has been * To whom correspondence should be addressed. Mark W. Mayhew, University of Virginia, Department of Cell Biology, P.O. Box 800732, Charlottesville, VA 22908-0732. Phone, 434-243-6812; fax, 434-982-3912; e-mail,
[email protected]. † Department of Cell Biology, University of Virginia. ‡ Vanderbilt University. § W.M. Keck Biomedical Mass Spectrometry Laboratory and Biomedical Research Facility, University of Virginia. 10.1021/pr060140t CCC: $33.50
2006 American Chemical Society
highly successful in yeast and used to construct its interactome.5 However, analyses of vertebrate cell proteomes have not yet been as successful due in part to the complexity of the vertebrate proteome and lack of easy genetic manipulations. Rather than attempting a global analysis, we have approached the migration proteome by developing mass spectrometry strategies for identifying binding partners for defined complexes that mediate migration. Our initial focus has been on the identification of binding partners for the PAK1-βPIX-GIT1-paxillin signaling module. This is a known complex, and therefore, it enabled us to develop and validate our procedures by first confirming interactions among proteins in the complex, and then identifying novel interactions that have not been reported. The PAK1βPIX-GIT1-paxillin module plays a critical role in adhesion and protrusion formation.6-9 Paxillin is an important focal adhesion adaptor protein that bridges FAK and GIT1 and is critical in focal adhesion assembly and cell protrusion.10,11 PAK1 (p21 activated kinase) is activated via the binding of active cdc42 and Rac and is involved in the regulation of myosin contraction and MAP kinase cascades.12 βPIX (PAK Interacting Exchange Factor) is a RhoGEF implicated in Rac and cdc-42 activation, which also binds and regulates PAK activity.13,14 GIT1 (G-protein-receptor kinase-interactor 1) is an ArfGAP that integrates various signaling pathways to regulate cell migration Journal of Proteome Research 2006, 5, 2417-2423
2417
Published on Web 08/17/2006
research articles and protrusion. Additionally, the binding of the GIT1 to βPIX provides a focal adhesion anchor for this complex through the paxillin-FAK interface.8 In this study, we optimized aspects of the protocol, including issues of protein overexpression and mislocalization, nonspecific interactions, phosphorylation levels, and sample purity to accommodate the sensitivity and limited dynamic range of the ion trap mass spectrometer. Our results demonstrate the importance of lowering bait protein expression to near endogenous levels, the use of phosphatase inhibitors for the enhancement of some binding interactions, and purity of the eluted complex. From these data, we were able to construct an interactome based on commonalities between shared and exclusive immunoprecipitated protein associations.
Materials and Methods Materials. All reagents were of analytical purity grade. Penicillin/streptomycin, lipofectamine, FBS, MEM/Opti-MEM, and the pCDNA3 plasmid were from Invitrogen (Carlsbad, CA). pBluescript was from Stratagene (La Jolla, CA). Protease, phosphatase inhibitor cocktails, and FLAG (DYKDDDDK) M2 agarose were purchased from Sigma (St. Louis, MO). Sodium vanadate was a product from Fisher Scientific (Fairlawn, NJ), and calyculin A was from Calbiochem (San Diego, CA). Plasmids. Human GIT1 cDNA was cloned as previously described.8 Chicken pCDNA3-FLAG-paxillin was provided by Dr. Tom Parsons (University of Virginia, Charlottesville, VA), and mouse βPIXa cDNA was provided by Dr. Chris Turner (SUNY Upstate Medical University, Syracuse, NY). Human PAK1 cDNA was a gift from Dr. Jonathan Chernoff (Fox Chase Cancer Center, Philadelphia, PA). PAK1 and βPIX were PCRamplified and subcloned into pCDNA3-FLAG and pCDNA3FLAG-GFP using EcoRI and XhoI restriction site oligonucleotide cassettes. Paxillin was subcloned into pCDNA3-FLAG-GFP via BamHI and EcoRI, and GIT1 was subcloned using NotI and SalI. All constructs were sequenced to confirm point of in-frame fusion and correct cDNA sequence. Cell Culture and Transfection. HEK293 cells were obtained from American Type Culture Collection and grown in DMEM F12 + HEPES supplemented with 10% FBS in the presence of penicillin/streptomocyin at 37 °C with 5% CO2. Prior to transfection, plasmids were serially diluted to 10 ng/µL and premixed with pBluescript plasmid (carrier DNA) providing a final DNA concentration of 10 ng FLAG plasmid/3 µg pBluescript per 100 mm plate. Lipofectamine/DNA complexes at a ratio of 1:5 were incubated in Opti-MEM for 45 min prior to addition to cells. Cells were seeded at ∼80% confluency on (4) 100 mm dishes and transfected in the absence of serum for 4 h in Opti-MEM media. Cells were then washed with PBS and reconditioned with MEM + 10% FBS overnight. Cell Lysis and Immunoprecipiation (IP). All steps of the IP were performed at 4 °C. If phosphatase inhibitors were used, the cells were treated for 30 min with 1 mM peroxovanadate (VO4) and 5 nM calyculin A with FBS supplement. Cells were harvested on ice 24 h post-transfection in the presence of 1 mL of lysis buffer (25 mM Tris, pH 7.4, 100 mM NaCl, 0.5% NP-40, and protease inhibitor cocktail 100×). If cells were treated with phosphatase inhibitors, phosphatase inhibitor cocktails 1 and 2 were added to lysis buffer. Cells were scraped and lysates passed through a 26-gauge needle, followed by a 30 min incubation on ice. Lysates were then centrifuged at 11 000 rpm and supernatants subjected to 50 µL of anti-mouse IgG agarose for 1 h. Lysates were again centrifuged, and the 2418
Journal of Proteome Research • Vol. 5, No. 9, 2006
Mayhew et al.
supernatant was incubated with 100 µL of anti-FLAG M2 agarose for 1 h. Immunocomplexes were collected via brief centrifugation and washed three times in 400 µL of wash buffer (25 mM Tris, pH 7.4, 100 mM NaCl). Bait protein interactomes were eluted via FLAG (DYKDDDDK) peptide challenge in 100 µL of elution buffer (25 mM Tris, pH 7.4, 200 µg/mL FLAG peptide) for 20 min. Western Blot Analysis and Antibodies. Standard procedures after immunoprecipitation included SDS-PAGE silver stain of washes and eluted protein fractions to confirm purity of elution fraction. Second, Western blot analysis was used to confirm the presence of FLAG-tagged protein using the monoclonal M2 FLAG anti-mouse antibody (Stratagene, La Jolla, CA). In instances where phosphatase inhibitors were used, an additional blot for phosphotyrosine using monoclonal anti-mouse 4G10 (Upstate Biotech, Charlottesville, VA) was immunoconfirmed for an enhancement of bait phosphorylation (data not shown). The GIT1 antibody was developed in-house8 and raised by Biosource International (Camarillo, CA). The antibodies against paxillin and CSK were from BD Biosciences (San Diego, CA), and the antibody for βPIX was from Chemicon (Temecula, CA). The N-20 PAK1 antibody was from Santa Cruz (Santa Cruz, CA). For Western blot analysis, in brief, 10% of elution was separated using SDS-PAGE and blotted to (polyvinylidene dichloride) (PVDC) membranes. Blots were blocked for 1 h in 5% nonfat milk (3% BSA for 4G10 antibody), and incubated overnight in primary antibody at 4 °C. Blots were then washed in TBS-Tween for 10 min, and secondary anti-mouse conjugated horseradish peroxidase was applied for 1 h prior to a TBSTween wash and chemiluminescence using Western Lightening Reagent (Perkin-Elmer). Blots were developed on Kodak XB-1 film. Mass Spectrometry. Samples were reduced with 10 mL of 10 mM dithiothreitol in 0.1 M ammonium bicarbonate for 30 min at room temperature (RT), and alkylated with 30 µL of 50 mM iodoacetamide in 0.1 M ammonium bicarbonate for 30 min at RT before digestion with 1 µg of Promega modified trypsin for 24 h at RT. A second 1 µg of trypsin was added, and digestion was allowed to proceed for an additional 24 h. The sample was then desalted on a self-packed 6 cm × 150 mm i.d. C18 column. Approximately 25% of the digest was introduced into the mass spectrometer for analysis. The mass spectrometry system consisted of a Thermo Electron LCQ DecaXP mass spectrometer with a Protana nanospray ion source interfaced to a self-packed 8 cm × 75 µm i.d. Phenomenex Jupiter 10 µm C18 reversed-phase capillary column. The peptides were eluted from the column by an acetonitrile/0.1 M acetic acid gradient over 2 h at a flow rate of 0.3 µL/min. The nanospray ion source was operated at 2.8 kV. Digests were then analyzed using the double play capability of the instrument by acquiring a full scan mass spectrum to determine peptide molecular weights followed by four product ion spectra to determine amino acid sequence in sequential scans. MS/ MS spectra were searched using the Sequest algorithm against the NCBI nonredundant (NR) database and validated using inhouse software (Proteofarm).
Results Low Expression Levels Increase Detection of Binding Partners. Our initial goal was to optimize protocols for the detection of bait-associated proteins in mammalian cells by mass spectrometry. Since the PAK1-βPIX-GIT1-paxillin sig-
PAK1-βPIX-GIT1-Paxillin Network Associations by MS/MS
research articles
Figure 1. Near endogenous levels of bait protein augments MS/MS detection of associated proteins and known binding partners. (A) The PAK1-βPIX-GIT1-paxillin known binding complex is demonstrated in order of known direct interactions. (B) HEK 293 cells were transfected with GFP-βPIX or GFP-paxillin at 100, 10, and 1 ng/100 mm plate. Lysates were subjected to Western blot and probed with either βPIX or paxillin antibodies. Comparisons between ectopic and endogenous expression levels are demonstrated. (C) Lowering bait expression levels below 100 ng resulted in detection of associated proteins, demonstrated here with FLAG pull-downs of paxillin and βPIX. (D) Confirmation that the known PAK1-βPIX-GIT1-paxillin complex is detected in our MS/MS analyses. Note: The (+) indicates that statistical criteria were met to signify the presence of associated protein peptides.
Figure 2. Phosphatase inhibitors augment the detection of known binding partners via MS/MS and Western blot analysis. HEK 293 cells were transfected with 10 ng of plasmid DNA and 3 µg of pBluescript carrier plasmid overnight and treated with (peroxovanadate VO4 (1 mM) and calyculin A (5 nM) in the presence of serum for 30 min prior to lysis. Bait proteins were immunoprecipitated and elutants blotted for their prospective binding partners. (A) For FLAG-paxillin pull-downs (n ) 8), we observed increased MS/MS total peptides for CSK and GIT1 in the presence of phosphatase inhibitors. (B) For FLAG-PAK1 pull-downs (n ) 6), we observed more MS/MS total peptides for βPIX in the presence of inhibitors. Biochemical analysis was confirmed via Western blot with FLAG immunoreactivity of bait proteins compared to binding partners as shown.
naling module (Figure 1A) plays a central role in regulating migration, we began by examining binding partner interactions with a key component of this signaling complex. When HEK cells were transfected with FLAG-tagged paxillin using a standard transfection protocol (1 µg of plasmid DNA per 100 mm dish), GIT1 peptides were not detected in the paxillin immunoprecipitates as analyzed by MS/MS. Western blot analysis showed that this transfection protocol resulted in very high expression levels of FLAG-paxillin (greater than 20-fold over endogenous). These data suggested that overexpression can lead to mislocalization of bait proteins and/or alter the
stoichiometry of protein-protein interactions, which in turn can mask binding partner detection. Clearly, paxillin bait expression that was closer to endogenous levels was needed, but the challenge was to maintain the high transfection efficiencies under these conditions. To lower expression levels and maintain transfection efficiencies, a range of paxillin cDNA concentrations (100, 10, and 1 ng) were used in the presence of carrier plasmid DNA such that the total DNA concentration remained at 3 µg. One and 10 ng transfection levels resulted in FLAG-paxillin and FLAG-βPIX expressions that were near endogenous (Figure 1B). GIT1 peptides were detected with Journal of Proteome Research • Vol. 5, No. 9, 2006 2419
research articles
Mayhew et al.
Table 1. FLAG-Bait Proteins Share MS/MS-Detected Associated Complex Proteinsa FLAG-BAIT
endogenous complex proteins
paxillin
>3 Shared Bait Interaction 14-3-3 Multiple Isoforms * R/β-Tubulin (Q71U36, Q9H4B7) * Actin (P60709) * β2 Centaurin (Q15057) * Calmodulin (P62158) Casein Kinase II R (P68400) * Cofilin (P23538) * DDX-3 Dead Box Protein (O00571) * Dynactin 2 p50 Subunit (Q13561) * Importin R2 (Q64292) Myosin LC/HC Multiple Isoforms * PP1G Phosphatase (P36873) Profilin (P07737) * Ras GAP (P20936) * RIO Kinase (Q9BRS2) Transgelin (Q01995) * Tropomyosin R3 (P06753) * 2 Shared Bait Interactions A32A Lucine-rich SET protein (P39687) A32B Lucine0rich SET protein (Q92688) A32C Lucine-rich SET protein (O43423) R-Centractin (Arp1) (P61163) Coronin 3 (Q3WUM4) Dynactin 1 p150 Subunit (Q14203) Dyncin HC (Q92815) Gelsolin (P06396) Grb-2 (P62993) * Importin R6 (O15131) Nck (P16333) PP2A Phosphatase (P67775) Ran GAP (P46060) Spectrin R2 (Q13813) *
GIT-1
βPIX
PAK-1
* * * * * * *
* * * * * * * * * * * * * * * * *
* * * * *
* * * * * * *
* * * * * * * * * * * * * *
* * * * * * * * * * * * * * * * * * * * *
a Shared endogenous complex proteins are listed in two groups: 2 shared bait interactions, and >3 shared bait interactions. These data represent the presence of the associated peptides/proteins within the bait proteome via MS/MS analysis, but do not indicate direct interactions. These data offer confirmation of the reliability of the protocol, and the presence of these proteins in the PAK1-βPIX-GIT1-paxillin signaling complex. Primary accession numbers are listed with each protein.
paxillin immunoprecipitates at these lower expression levels (Figure 1C). Interestingly, lower expression levels of FLAGpaxillin also resulted in the detection of peptides for talin, profilin, and β2 centaurin (Figure 1C). Similar results were also obtained with another molecule in this signaling module, βPIX (Figure 1B,C). At lower expression levels, peptides for scribble, dynactin, profilin, and β2 centaurin were detected in FLAGβPIX immunoprecipitates. Two of these binding partners, profilin and β2 centaurin, were detected in both paxillin and βPIX immunoprecipitates providing corroborating evidence that they interact with a part of this migration complex. Taken together, these results show that expression levels close to endogenous enhance binding partner detection by mass spectrometry. We extended this analysis to include each of the four components (PAK1, βPIX, GIT1, and paxillin) individually. Peptides from the other components of the signaling module were detected in the immunoprecipitates using each of these four molecules, individually, as bait. For example, peptides for GIT1, βPIX, and PAK1 were detected in paxillin immunoprecipitates (Figure 1D). These results suggest that this approach detects not only direct binding partners, but also components that interact indirectly with proteins of the signaling complex. 2420
Journal of Proteome Research • Vol. 5, No. 9, 2006
Figure 3. Mapping of associated proteins shared among bait protein complex. Shared interacting proteins of the PAK1-βPIXGIT1-paxillin complex are depicted here with asterisks indicating previously reported direct interactions. This map provides a means to examine shared associated proteins. Connection lines represent the presence of these associated proteins within the proteome, but not direct binding interactions.
Phosphorylation Affects the Interaction of Bait Proteins with Binding Partners. Post-translational modifications, such as phosphorylation, can regulate protein-protein interactions. To determine whether phosphorylation affected binding partner interactions with the paxillin-GIT1-βPIX-PAK1 signaling module, cells were pretreated with tyrosine and serine/threonine phosphatase inhibitors, peroxovanadate and calyculin A, respectively. These inhibitors should increase basal phosphorylation levels and thus enhance detection of binding partner interactions that are phosphorylation-dependent. As seen in Figure 2A, MS analyses of FLAG-paxillin immunoprecipitation showed an increased number of CSK and GIT1 peptides from cells treated with the phosphatase inhibitors, suggesting an increased concentration of the protein. Figure 2B demonstrates a similar increase in βPIX peptides seen in immuprecipitations with FLAG-PAK1 in the presence of phosphatase inhibitors. We verified the mass spectrometry observations by immunoblot analyses. FLAG-paxillin immunoprecipitations were immunoblotted with antibodies against CSK and GIT1. Both showed increased immuno-reactivity in immunoprecipitates. In addition, βPIX immuno-reactivity was similarly increased in FLAG-PAK1 immunoprecipitation in the presence of phosphatase inhibitors (Figure 2, panels A and B, respectively). These results suggest that the phosphorylation states of these molecules in the signaling module can affect binding partner interactions. Network Maps Generated from Analysis of Bait Complex Protein Associations. We used our dataset of 120 individual mass spectrometric analyses utilizing the four bait proteins to assemble a list of proteins that were either shared associations among multiple bait proteins or exclusive to a single bait protein. Individual bait pull-down n-values are as follows: PAK1 ) 16, βPIX ) 18, GIT1 ) 26, paxillin ) 60. Our strategy was to list the associated proteins that were observed by identification criteria of more than one peptide/sample, detected in more than one sample, and had a SEQUEST-Xcorr
research articles
PAK1-βPIX-GIT1-Paxillin Network Associations by MS/MS
Figure 4. MS/MS detection of proteins exclusively associated with bait complex. Depicted are lists of proteins that were observed exclusively with individual proteins of the bait complex. Asterisks represent previously reported interactions. Accession numbers are included.
value >5 (cross-correlation scoring routine). In addition, we removed those proteins that were present in the FLAG mock controls. This better ensures that the assignment is correct and that the protein is indeed present in the immunoprecipitation. The putative binding proteins were organized into three classes: (1) those that shared an interaction with three or more bait proteins, (2) those that shared an interaction with two bait proteins, and (3) those that were exclusive for a single bait protein. The list of shared interacting proteins is shown in Table 1. Additionally, to better visualize the shared interactions, a map was created showing all of the proteins that associated with two or more shared interactions among the four bait proteins (Figure 3). Asterisks have been placed on previously reported binding interactions. The lines connecting the proteins do not necessarily reflect direct binding interactions, but instead reveal protein network associations. Several interesting proteins emerged in these analyses and will be highlighted in Discussion. In Figure 4, proteins that associated exclusively with a single bait protein are shown. Previously reported protein-protein interactions or kinase/substrate interactions are labeled with an asterisk. For PAK1, exclusive binding partners were C-Myc BP, Diaphanous, G-protein 1, GTP-BP SAR1A, KIF11 Kinesinlike, KIP-2, Integrin BP, C-Jun Kinase 1, NDR (S/T Kinase), Rab1, Rab-GDI, and Talin-2. For βPIX, proteins detected were Arp A (Arp10), BAL-B, G3BP RasGAP BP, GBAF GTPase, *Scribble, Vimentin, and VRK1 Kinase. Exclusive proteins for GIT1 were
*Aurora A, CDK5 RAP2, DOCK-4, Fascin, Flightless-1, Moesin, and Rab-3 BP. Proteins found exclusively in paxillin precipitates include *Abl-1, Arf-1, *Crk, *Csk, *Erk 1, *FAK, FAS-R, MAPKAPK 5, *Nek-9 Kinase, *Rho G, *Src, *Talin-1, and VASP.
Discussion The goal of this investigation was to use mass spectrometry to characterize interactions mediated by the PAK1-βPIXGIT1-paxillin signaling module and to identify novel proteins associated with it. The study greatly benefited from several biochemical manipulations. Immunoprecipations were adjusted to increase sample purity, as assessed by the number of major bands observed on silver-stained gels. The adjustments included faster elution times and appropriate wash procedures. Another major factor was low expression levels of the bait protein. This decreases mislocalization and other overexpression artifacts, and increased the detection of binding partners. While we used transfection procedures with carrier DNA to achieve low expression at high efficiency, other approaches including viral vectors and knockdown or knockout cell lines would also be useful. Finally, the use of cell phosphatase inhibitors, like peroxovanadate and calyculin A, preserved protein phosphorylation levels and enhanced coprecipitation of some binding partners, most notably, CSK and GIT1 in paxillin immunoprecipitations, and βPIX in PAK1 immunoprecipitations. This result is not surprising since tyrosine phosphorylation can activate SH2 binding domains, and other Journal of Proteome Research • Vol. 5, No. 9, 2006 2421
research articles phosphorylations, in general, can also regulate interactions through a variety of mechanisms, including conformational alterations. The proteins that co-purified with more than one member of the complex likely interact with the complex either by binding directly with one of the core complex members, for example, GIT1, paxillin, and so forth, or indirectly through interactions with another component that does bind directly to a core member of the complex. While many proteins are represented, it is likely that there are others that were not detected due to either low affinity or stoichiometry. Many proteins were observed but excluded from the list of interacting proteins. These include all of the abundant proteins commonly detected in controls that lacked the bait protein, for example, heat shock proteins, ribosomal proteins, RNA/DNA binding proteins, and metabolic proteins. Thus, this analysis represents a first step in defining a functional signaling “network” for the PAK1-βPIX-GIT1paxillin complex. Proteins that co-purified with all four components of the module include profilin, cofilin, importin, and 14-3-3. Another scaffolding adaptor protein, Nck, co-purified with PAK1 and GIT1, and the adaptor Grb-2 co-purified with both paxillin and GIT1. It has been reported that paxillin is associated with the microtubule organizing center (MTOC) in lymphocytes.15 Interestingly, we identified the centrosome-associated protein Aurora A in GIT1 immunoprecipitates, and the spindle/ centrosome-associated protein, KIF11, in PAK1 immunoprecipitates. Furthermore, we identified dynactin, dynein, centractin, and spectrin as major constituents interacting with paxillin, βPIX, and PAK1, agreeing with previously reported biochemical data demonstrating a GIT1-dependent localization of βPIX and PAK1 to centrosomal compartments, and PAK1 phosphorylation of the centrosomal kinase, Aurora A.16 Additionally, we have also observed GFP-βPIX and GFP-paxillin in centrosome-like structures in MEF cells by video imaging (data not shown). Phosphatase activity is known to be an intrinsic factor of protein signaling, so it was no surprise that phosphatases would be tightly coupled to such a dynamic protein complex involved in focal adhesion turnover. PHAPI and PHAPII, which form the A32 SET protein complex, are potent PP2A phosphatase inhibitors and were detected in βPIX and PAK1 immunoprecipitates.17-19 Moreover, PP1G and PP2A coprecipitated with βPIX, GIT1, and PAK1. On the basis of these data, it appears that the PAK1βPIX-GIT1-paxillin complex is under stringent phosphoregulation, not only by phosphatases, but also by phosphatase protein inhibitors. As previously reported, GIT1 contains a known ArfGAP domain that is thought to be linked to Rac1-dependent reorganization of actin,20 and neurtite outgrowth.21 Activated Arf6 has been shown to activate Rac1-dependent cell motility,22,23 and GIT1 has been shown to play a role in the regulation of Arf6-dependent Rac1 cell polarity.24 We detected an Arf6specific β2 centaurin ArfGAP (ACAP2) as a prominent associated protein among βPIX, PAK1, GIT1, and paxillin.25 Because the β isoforms of the centaurin ArfGAP family are the only members that comprise coiled-coil domains, we speculate that coiled-coil binding domains specific to proteins such as βPIX, GIT1, and β2 centaurin might provide anchors for their interactions.26 Spectral data of peptide sequences confirmed specificity to the β2 centaurin isoform in 13 different tryptic peptides, and none of these sequences contained homology 2422
Journal of Proteome Research • Vol. 5, No. 9, 2006
Mayhew et al.
to the GIT1/PKL ArfGAP domain (data not shown). Therefore, it appears that another Arf6-GAP protein is present in the complex. It is understood that a majority of the associated proteins are not directly linked to the bait proteins of the core plexus and, thus, cannot be classified as direct binding partners, but members of the proteome, nonetheless. Calmodulin (CaM) was detected in βPIX, PAK1, and GIT1 precipitates. However, we linked the PAK1-CaM interaction through the previously reported spectrin-CaM interaction, via our detected PAK1spectrin association.27 Other interesting observations were RasGAP and transgelin peptides co-purified with βPIX and paxillin. It has been reported that RasGAP is coimmunoprecipitated with endogenous paxillin in human mesangial cells.28 A possible scaffold linker for transgelin could be its reported interaction with actin.29 As a major adaptor protein in focal adhesions, we found it interesting that paxillin, among other bait proteins, coprecipitated with such an array of proteins with various function. The paxillin linkage between the mitotic progression regulator Nek kinase could form a complex with the previously reported binding of Nek and RAN, thus, coupling to GIT1, PAK1, and R-importin.30 Interestingly, of the many unique proteins screened during our analysis, the cell cycle regulator Rio kinase was identified in precipitates of βPIX, GIT1, and PAK1.31 The mapping of exclusively associated proteins (Figure 4) provides possible bridging points within the interactome and, in some cases, might lead to interactions with other proteome clusters. Specifically, binding partners labeled with an asterisk indicate a previously reported association based on information from the Human Protein Reference Database (http:// www.hprd.org/). New proteins reported here shed light on possible functionality within the PAK1-βPIX-GIT1-paxillin module. PAK1-associated proteins include c-myc binding protein, diaphanous, G-protein 1, GTP binding protein SAR1A, the spindle/centrosomal protein KIF11, integrin-binding protein KIP-2, c-Jun kinase 1, NDR kinase, Rab-1, Rab-GDI, and talin-2. For βPIX, the binding partner, scribble, has previously been identified by mass spectrometry and is reported to play a role in neurotransmission;32 while other proteins such as Arp10, BAL-B, GBAF GTPase, vimentin, and VRK1 serine/ threonine kinase are also interesting, associated proteins. The GIT1 exclusively associated proteins include Aurora A, DOCK4, fascin, flightless-1, moesin, Rab-3 interacting protein, and the Cdk5 activator C48, which might regulate PAK1 T212 phosphorylation by Cdk5.33 Paxillin-associated proteins include Abl-1, Arf-1, Crk, Csk, Erk1, FAK, FAS1 receptor, MAPKAPK 5, Nek-9 kinase, RhoG, Src, talin-1, and VASP. Taken together, our investigation demonstrates the use of mass spectrometry to assemble a working interactome for a migration-related signaling module. While the details of the association of each protein with the core signaling complex remain to be studied, these proteins represent starting points for their detailed characterization in the context of the PAK1βPIX-GIT1-paxillin signaling module.
Conclusion The ectopic expression of FLAG-tagged bait proteins together with immunoprecipitation, FLAG peptide elution, and tandem mass spectrometry has led to discoveries of mammalian cell proteomes. In this study, we used LC-MS/MS to identify key proteins associated with the cell migration, signaling core complex PAK1, βPIX, GIT1, and paxillin. Our data suggest that
research articles
PAK1-βPIX-GIT1-Paxillin Network Associations by MS/MS
optimization of controlled bait expression levels, elution purity, and the use of phosphatase inhibitors results in enhanced detection of known and novel proteins of this cell migration protein nexus. We have determined numerous core complexassociated proteins with various functions, and further biochemical studies of these network proteins will achieve advancements in the field of proteomics discovery.
Acknowledgment. We thank Dr. Tom Parsons for helpful discussions of this work, and for providing the FLAGpaxillin construct. We also thank Dr. William Pearson for the mass spectrometry database management of this work. We graciously thank Dr. Jonathan Chernoff for providing the PAK1 cDNA, and Dr. Chris Turner for providing βPIX cDNA. We also thank Dr. Nicholas Sherman for his help with methodological aspects of the work and Dr. Li Ma and Dr. Kristina Nelson for their invaluable LC-MS/MS technical assistance. This work was supported by NIH-The Cell Migration Consortium (U54 GM064346), and grant GM37537 (A.F.H.). References (1) Ridley, A. J. Pulling back to move forward. Cell 2004, 116 (3), 357358. (2) Zamir, E.; Geiger, B. Molecular complexity and dynamics of cellmatrix adhesions. J. Cell Sci. 2001, 114 (Pt 20), 3583-3590. (3) Lauffenburger, D. A.; Horwitz, A. F. Cell migration: a physically integrated molecular process. Cell 1996, 84 (3), 359-369. (4) Romijn, E. P.; Krijgsveld, J.; Heck, A. J. Recent liquid chromatographic-(tandem) mass spectrometric applications in proteomics. J. Chromatogr., A 2003, 1000 (1-2), 589-608. (5) Kolkman, A.; Slijper, M.; Heck, A. J. Development and application of proteomics technologies in Saccharomyces cerevisiae. Trends Biotechnol. 2005, 23 (12), 598-604. (6) Manser, E.; Huang, H. Y.; Loo, T. H.; Chen, X. Q.; Dong, J. M.; Leung, T.; Lim, L. Expression of constitutively active alpha-PAK reveals effects of the kinase on actin and focal complexes. Mol. Cell. Biol. 1997, 17 (3), 1129-1143. (7) Zhao, Z. S.; Manser, E.; Loo, T. H.; Lim, L. Coupling of PAKinteracting exchange factor PIX to GIT1 promotes focal complex disassembly. Mol. Cell. Biol. 2000, 20 (17), 6354-6363. (8) Manabe, R.; Kovalenko, M.; Webb, D. J.; Horwitz, A. R. GIT1 functions in a motile, multi-molecular signaling complex that regulates protrusive activity and cell migration. J. Cell Sci. 2002, 115 (Pt 7), 1497-1510. (9) Brown, M. C.; West, K. A.; Turner, C. E. Paxillin-dependent paxillin kinase linker and p21-activated kinase localization to focal adhesions involves a multistep activation pathway. Mol. Biol. Cell 2002, 13 (5), 1550-1565. (10) Turner, C. E. Paxillin and focal adhesion signalling. Nat. Cell Biol. 2000, 2 (12), E231-E236. (11) Brown, M. C.; Turner, C. E. Paxillin: adapting to change. Physiol. Rev. 2004, 84 (4), 1315-1339. (12) Bokoch, G. M. Biology of the p21-activated kinases. Annu. Rev. Biochem. 2003, 72, 743-781. (13) Feng, Q.; Albeck, J. G.; Cerione, R. A.; Yang, W. Regulation of the Cool/Pix proteins: key binding partners of the Cdc42/Rac targets, the p21-activated kinases. J. Biol. Chem. 2002, 277 (7), 56445650. (14) Ten Klooster, J. P.; Jaffer, Z. M.; Chernoff, J.; Hordijk, P. L. Targeting and activation of Rac1 are mediated by the exchange factor {beta}-Pix. J. Cell Biol. 2006, 172 (5), 759-769. (15) Herreros, L.; Rodriguez-Fernandez, J. L.; Brown, M. C.; AlonsoLebrero, J. L.; Cabanas, C.; Sanchez-Madrid, F.; Longo, N.; Turner, C. E.; Sanchez-Mateos, P. Paxillin localizes to the lymphocyte microtubule organizing center and associates with the microtubule cytoskeleton. J. Biol. Chem. 2000, 275 (34), 26436-26440.
(16) Zhao, Z. S.; Lim, J. P.; Ng, Y. W.; Lim, L.; Manser, E. The GITassociated kinase PAK targets to the centrosome and regulates Aurora-A. Mol. Cell 2005, 20 (2), 237-249. (17) Vaesen, M.; Barnikol-Watanabe, S.; Gotz, H.; Awni, L. A.; Cole, T.; Zimmermann, B.; Kratzin, H. D.; Hilschmann, N. Purification and characterization of two putative HLA class II associated proteins: PHAPI and PHAPII. Biol. Chem. Hoppe-Seyler 1994, 375 (2), 113-126. (18) Li, M.; Makkinje, A.; Damuni, Z. Molecular identification of I1PP2A, a novel potent heat-stable inhibitor protein of protein phosphatase 2A. Biochemistry 1996, 35 (22), 6998-7002. (19) Li, M.; Makkinje, A.; Damuni, Z. The myeloid leukemia-associated protein SET is a potent inhibitor of protein phosphatase 2A. J. Biol. Chem. 1996, 271 (19), 11059-11062. (20) Di Cesare, A.; Paris, S.; Albertinazzi, C.; Dariozzi, S.; Andersen, J.; Mann, M.; Longhi, R.; de Curtis, I. p95-APP1 links membrane transport to Rac-mediated reorganization of actin. Nat. Cell Biol. 2000, 2 (8), 521-530. (21) Albertinazzi, C.; Za, L.; Paris, S.; de Curtis, I. ADP-ribosylation factor 6 and a functional PIX/p95-APP1 complex are required for Rac1B-mediated neurite outgrowth. Mol. Biol. Cell 2003, 14 (4), 1295-1307. (22) Santy, L. C.; Casanova, J. E. Activation of ARF6 by ARNO stimulates epithelial cell migration through downstream activation of both Rac1 and phospholipase D. J. Cell Biol. 2001, 154 (3), 599-610. (23) Santy, L. C.; Ravichandran, K. S.; Casanova, J. E. The DOCK180/ Elmo complex couples ARNO-mediated Arf6 activation to the downstream activation of Rac1. Curr. Biol. 2005, 15 (19), 17491754. (24) Nishiya, N.; Kiosses, W. B.; Han, J.; Ginsberg, M. H. An alpha4 integrin-paxillin-Arf-GAP complex restricts Rac activation to the leading edge of migrating cells. Nat. Cell Biol. 2005, 7 (4), 343352. (25) Jackson, T. R.; Kearns, B. G.; Theibert, A. B. Cytohesins and centaurins: mediators of PI 3-kinase-regulated Arf signaling. Trends Biochem. Sci. 2000, 25 (10), 489-495. (26) Jackson, T. R.; Brown, F. D.; Nie, Z.; Miura, K.; Foroni, L.; Sun, J.; Hsu, V. W.; Donaldson, J. G.; Randazzo, P. A. ACAPs are arf6 GTPase-activating proteins that function in the cell periphery. J. Cell Biol. 2000, 151 (3), 627-638. (27) Dubreuil, R. R.; Brandin, E.; Reisberg, J. H.; Goldstein, L. S.; Branton, D. Structure, calmodulin-binding, and calcium-binding properties of recombinant alpha spectrin polypeptides. J. Biol. Chem. 1991, 266 (11), 7189-7193. (28) Shikata, Y.; Shikata, K.; Matsuda, M.; Sugimoto, H.; Wada, J.; Makino, H. Signaling transduction pathway of angiotensin II in human mesangial cells: mediation of focal adhesion and GTPase activating proteins. Biochem. Biophys. Res. Commun. 1999, 257 (1), 234-238. (29) Fu, Y.; Liu, H. W.; Forsythe, S. M.; Kogut, P.; McConville, J. F.; Halayko, A. J.; Camoretti-Mercado, B.; Solway, J. Mutagenesis analysis of human SM22: characterization of actin binding. J. Appl. Physiol. 2000, 89 (5), 1985-1990. (30) Roig, J.; Mikhailov, A.; Belham, C.; Avruch, J. Nercc1, a mammalian NIMA-family kinase, binds the Ran GTPase and regulates mitotic progression. Genes Dev. 2002, 16 (13), 1640-1658. (31) Laronde-Leblanc, N.; Guszczynski, T.; Copeland, T.; Wlodawer, A. Structure and activity of the atypical serine kinase Rio1. FEBS J. 2005, 272 (14), 3698-3713. (32) Audebert, S.; Navarro, C.; Nourry, C.; Chasserot-Golaz, S.; Lecine, P.; Bellaiche, Y.; Dupont, J. L.; Premont, R. T.; Sempere, C.; Strub, J. M.; Van Dorsselaer, A.; Vitale, N.; Borg, J. P. Mammalian Scribble forms a tight complex with the betaPIX exchange factor. Curr. Biol. 2004, 14 (11), 987-995. (33) Rashid, T.; Banerjee, M.; Nikolic, M. Phosphorylation of Pak1 by the p35/Cdk5 kinase affects neuronal morphology. J. Biol. Chem. 2001, 276 (52), 49043-49052.
PR060140T
Journal of Proteome Research • Vol. 5, No. 9, 2006 2423
