Proteomic Identification of the Cerebral Cavernous Malformation Signaling Complex Thomas L. Hilder,† Michael H. Malone,† Sompop Bencharit,†,‡ John Colicelli,§ Timothy A. Haystead,| Gary L. Johnson,*,† and Christine C. Wu⊥ Department of Pharmacology and the Lineberger Comprehensive Cancer Center, and Department of Prosthodontics, School of Dentistry, University of North Carolina, Chapel Hill, CB #7365, Chapel Hill, North Carolina 27599-7365, Department of Biological Chemistry, Molecular Biology Institute, and the Jonsson Comprehensive Cancer Center, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, California 90095, Department of Pharmacology and Cancer Biology, Duke University Medical Center, C119 LSRC Research Drive, Durham, North Carolina 27710, and University of Colorado School of Medicine, Department of Pharmacology/Mail Stop 8303, Fitzsimons RC1 South, 12801 East 17th Avenue, L18-6117, P.O. Box 6511, Aurora, Colorado 80045 Received July 11, 2007
Cerebral cavernous malformations (CCM) are sporadic or inherited vascular lesions of the central nervous system characterized by dilated, thin-walled, leaky vessels. Linkage studies have mapped autosomal dominant mutations to three loci: ccm1 (KRIT1), ccm2 (OSM), and ccm3 (PDCD10). All three proteins appear to be scaffolds or adaptor proteins, as no enzymatic function can be attributed to them. Our previous results demonstrated that OSM is a scaffold for the assembly of the GTPase Rac and the MAPK kinase kinase MEKK3, for the hyperosmotic stress-dependent activation of p38 MAPK. Herein, we show that the three CCM proteins are members of a larger signaling complex. To define this complex, epitope-tagged wild type OSM or OSM harboring the mutation of F217 f A, which renders the OSM phosphotyrosine binding (PTB) domain unable to bind KRIT1, were stably introduced into RAW264.7 mouse macrophages. FLAG-OSM or FLAG-OSMF217A and the associated complex members were purified by immunoprecipitation using anti-FLAG antibody. OSM binding partners were identified by gel-based methods combined with electrospray ionization-MS or by multidimensional protein identification technology (MudPIT). Previously identified proteins that associate with OSM including KRIT1, MEKK3, Rac, and the KRIT1-binding protein ICAP-1 were found in the immunoprecipitates. In addition, we show for the first time that PDCD10 binds to OSM and is found in cellular CCM complexes. Other prominent proteins that bound the CCM complex include EF1A1, RIN2, and tubulin, with each interaction disrupted with the OSMF217A mutant protein. We further show that PDCD10 binds phosphatidylinositol di- and triphosphates and OSM binds phosphatidylinositol monophosphates. The findings define the targeting of the CCM complex to membranes and to proteins regulating trafficking and the cytoskeleton. Keywords: cerebral cavernous malformation • KRIT1 • OSM • CCM3 • PDCD10 • MudPIT
Introduction Cerebral cavernous malformations (CCM) are sporadic or inherited vascular lesions in the brain characterized by endothelial-lined sinusoids and a sub-endothelial layer of connective tissue distinct from the neural parenchyma (reviewed in refs 1 and 2). The lesions are devoid of mature vessel wall elements like smooth muscle and elastic tissue. The blood-brain barrier is severely compromised, as gaps at endothelial cell tight * To whom correspondence should be addressed. Gary L. Johnson: Phone, (919) 843-3107; Fax, (919) 966-5640; E-mail,
[email protected]. † Department of Pharmacology and the Lineberger Comprehensive Cancer Center. ‡ Department of Prosthodontics. § University of California Los Angeles. | Duke University Medical Center. ⊥ University of Colorado School of Medicine. 10.1021/pr0704276 CCC: $37.00
2007 American Chemical Society
junctions are observed and no astrocytic foot processes border the vessels.3 Hemorrhaging of the vessels in affected individuals can occur, leading to chronic headaches, seizures, focal neurological deficits, and/or stroke.4,5 The familial form of CCM is autosomal dominant and has been mapped to three loci (ccm1, ccm2, and ccm3),6-9 encoding the proteins KRIT1, OSM/malcavernin, and PDCD10/CCM3, respectively.10-14 KRIT1 contains ankyrin repeat domains and a FERM domain, both of which coordinate protein-protein interactions;15,16 however, proteins that interact with KRIT1 remain to be defined. KRIT1 also contains multiple NPxY motifs that are bound by the phosphotyrosine binding (PTB) domain of OSM and of the integrin binding protein ICAP-1.17,18 Additionally, KRIT1 contains a functional nuclear localization signal that enables it to undergo nuclear-cytoplasmic shutJournal of Proteome Research 2007, 6, 4343-4355
4343
Published on Web 09/27/2007
research articles tling.17 OSM, in addition to binding KRIT1, is a scaffold for Rac and MEKK3 at membrane ruffles for the activation of p38 MAP kinase in response to sorbitol.19 A fraction of OSM colocalizes with actin and OSM binds F-actin in in vitro binding assays,19 suggesting that OSM organizes a complex capable of coupling Rac-dependent actin reorganization to p38 activity. Finally, PDCD10 is a relatively small protein with no discernible protein-protein interaction domains or enzymatic activity. PDCD10 was initially characterized as a gene whose expression was upregulated upon the induction of apoptosis in human myeloid cell lines.14 Mutations in ccm1, ccm2, or ccm3 genes are nonsense, frameshift, or splice site mutations, resulting in loss-of-function alleles.12-14,20 One missense mutation (L198 f R) has been described in the ccm2 gene,13 which lies within the PTB domain of OSM and significantly disrupts the binding of OSM to KRIT1.17 We hypothesized that OSM, KRIT1, and PDCD10 are scaffold- or adaptor-like proteins to organize and localize a macromolecular complex, and mutations in any of the three CCM proteins could disrupt the function of the CCM complex. To test this hypothesis, we stably expressed a FLAG-tagged version of OSM in RAW264.7 macrophages and identified the proteins that specifically co-immunoprecipitated with OSM using nanoelectrospray mass spectrometry and multidimensional protein identification technology (MudPIT). Our results indicate that OSM, KRIT1, and PDCD10 form a CCM protein complex. Previously identified binding partners for OSM and KRIT1 (Rac, MEKK3, and ICAP-1) were identified in our proteomic analyses. An engineered point mutation within the PTB domain of OSM (F217 f A), which like the L198R patient mutation disrupts OSM binding to KRIT1,17 resulted in the loss of numerous proteins associated with the CCM complex. The results provide novel insight into the proteins organized by the OSM-KRIT1 complex and demonstrate that functional disruption of the OSM PTB domain has profound effects on the protein network assembled by the CCM complex.
Experimental Procedures Plasmids and Production of Recombinant Proteins. FLAGtagged KRIT1, OSM, OSMF217A in pRK5, FLAG-LAD in pcDNA3.1, and pEYFPC1-Krit and pECFPN1-OSM were previously described.17-19,21 For the generation of retroviral constructs, FLAG-OSM and FLAG-OSMF217A were subcloned into the pMSCVpuro vector (Clontech) using standard PCR cloning methods. Full length cDNA for EF1A1 (I.M.A.G.E. Consortium CloneID 3948601)22 was purchased from ATCC and subcloned in-frame into the pcDNA4/myc-His B vector (Invitrogen). Recombinant 6×His-tagged OSM was described previously.19 Full-length murine PDCD10 was PCR amplified from a mouse fibroblast cDNA library and cloned into pMCSG7-His. This fulllength 6×His-PDCD10 was expressed in BL21 cells and purified by nickel affinity chromatography. The purified recombinant 6×His-PDCD10 protein was then coupled to CNBr-Sepharose beads (GE Biosciences) according to the manufacturer’s protocol. Cell Culture and Generation of Stable Cell Lines. Phoenix cells and COS7 cells were maintained in Dulbecco’s Modified Eagle Medium (Invitrogen) supplemented with 10% fetal bovine serum, 100 U/mL penicillin, and 100 µg/mL streptomycin. RAW264.7 mouse macrophages and 1321N1 human astrocytoma cells were maintained in the same medium, but heat inactivated serum was used and all components were filtered through a sterile 0.22 µm vacuum flask (Denville Scientific). 4344
Journal of Proteome Research • Vol. 6, No. 11, 2007
Hilder et al.
Amphotropic Phoenix cells were used as packaging cells for the retroviral gene transfer system.23,24 FLAG-OSM or FLAGOSMF217A in pMSCVpuro were transfected into Phoenix cells using Lipofectamine and PLUS reagents (Invitrogen). Two days after transfection, the supernatants were collected, diluted with an equal volume of fresh medium, and Polybrene was added to a final concentration of 8 µg/mL. The infection mixtures were added to RAW264.7 or 1321N1 cells. The next day, the cells were reinfected using the same method. Twenty-four hours after the second infection, selection was started with 4 µg/mL puromycin in complete DMEM. Stable cells were fully selected following 1-2 weeks exposure to puromycin. Cell Lysis and Immunoprecipitation. FLAG pulldowns were performed essentially as described.25 Briefly, RAW264.7 or 1321N1 cells stably expressing pMSCVpuro empty vector, FLAG-OSM, or FLAG-OSMF217A were lysed in pulldown buffer (50 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 0.5% NP40, 1.5 mM MgCl2, 1 mM EGTA, 10 mM sodium pyrophosphate, 100 mM sodium fluoride, and 20 mM β-glycerophosphate) and supplemented with a complete protease inhibitor cocktail tablet (Roche) and 1 mM sodium vanadate. Lysates were clarified by centrifugation, and 30 mg of protein, diluted to 2 mg/mL in pulldown buffer, was used per pulldown. Protein G-sepharose resin (Invitrogen) was used to preclear the lysates, and FLAGtagged proteins were immunoprecipitated with anti-FLAG M2agarose (Sigma). Following extensive washing in pulldown buffer, FLAG-OSM was competitively eluted with FLAG peptide (Sigma). For protein identification by nanoelectrospray mass spectrometry, the eluates were resolved by SDS-PAGE and visualized by Sypro Ruby protein staining (Invitrogen) according to manufacturer’s directions or by silver staining. For the MudPIT analysis, proteins in the eluates were precipitated using MeOH/CHCl3 as previously described.26 Immunoprecipitation of EF1A1-myc was performed in essentially the same manner. COS7 cells transfected with FLAGtagged constructs and EF1A1-myc/His were lysed in pulldown buffer, and EF1A1 was immunoprecipitated with the monoclonal anti-c-myc 9E10 antibody (Santa Cruz Biotechnology). After washing, EF1A1 was eluted from the beads in 2× SDSloading buffer and immunoblots were performed as described below. To determine the binding of PDCD10 to OSM and KRIT1, PDCD10-Sepharose or uncoupled Sepharose was tumbled overnight with lysates from RAW264.7 cells stably expressing empty vector or FLAG-OSM, or with lysates from COS7 cells transiently expressing FLAG-OSM, FLAG-OSMF217A, or FLAGKRIT1. The beads were then washed four times in pulldown buffer, proteins were eluted into 2× SDS-loading buffer, and immunoblots were performed as described below. The amount of PDCD10 present in the pulldowns was measured by staining the nitrocellulose membrane with Memcode protein stain (Pierce). Nanoelectrospray Ionization Mass Spectrometry. Bands excised from SDS-PAGE gels were diced into 1 mm2 pieces. Silver-stained bands first destained for 10 min in a 1:1 mixture of 100 mM sodium thiosulfate:30 mM potassium ferricyanide, then were washed three times in water. The silver-stained or Sypro Ruby-stained gel slices were then incubated in a 1:1 solution of acetonitrile:100 mM ammonium bicarbonate, and dried in a speed-vac. The gel pieces were re-swelled on ice in 30 µL 25 mM ammonium bicarbonate containing 0.5 µg sequencing grade modified trypsin (Promega) for 45 min, then 30 µL 25 mM ammonium bicarbonate was added incubated
Proteomic Identification of the CCM Complex
overnight at 37 °C. Peptides were recovered from the gel slices by two successive extractions in 60% acetonitrile with 5% formic acid. The acetonitrile was removed, and the peptides were concentrated to ∼5 µL by speed-vac. In preparation for nanoelectrospray analysis, the peptides were bound to Poros R2 resin (Applied Biosystems), washed twice in 5% methanol with 5% formic acid, and eluted into an electrospray needle (Proxeon) with 60% methanol with 5% formic acid. Identification of the bands was performed by nanospray electrospray ionization mass spectrometry (nESI-MS) on an Applied Biosystems QSTAR Pulsar mass spectrometer with an ion source voltage of 8 kV. Peaklists were generated (m/z 4001600 for precursor ions, m/z 130 to twice the mass of the doubly charged precursor ion for fragment ions) and database searches were performed with Analyst QS software, version 1 (build 7051; Applied Biosystems), with oMALDI source support, service pack 8, and Bioanalyst extensions. For database searches, the following parameters were used: (i) enzyme specificity of trypsin; (ii) no missed cleavages were allowed; (iii) no fixed or variable modifications were allowed; and (iv) mass tolerances of precursor and fragment ions were set at 1.1 and 0.2 Da, respectively. For de novo peptide sequencing, the y-ion with the highest detectable m/z was selected, and Analyst QS was used to select the next 3-5 y-ions in the series to generate a sequence tag. This sequence tag, along with the masses of the y- and b-ions, with or without NH3, were searched against the National Center for Biotechnology Information nonredundant (NCBInr) database (Apr. 27, 2006 download; 1 596 370 entries). No species restriction was applied, but for all identifications a mouse or human homologue was identified (from RAW264.7 or 1321N1 cells, respectively). Based on these parameters, greater than 50% of these y- and b-ions fell within the mass tolerance specified above. The thresholds applied to these sequence tags resulted in a single identification with an Analyst QS score between 585 and 696 (below this cutoff, peptides either had a residue that did not match the tandem MS spectrum or were not tryptic). At least two peptides meeting these parameters were required for a positive identification. Immunoblotting. Following anti-FLAG M2-agarose pulldowns or immunoprecipitations, immunoblots were performed for members of the complex following transfer of the proteins to nitrocellulose. Primary antibodies were used at manufacturer’s recommended dilutions and included the following: c-myc 9E10, 6×His, CCT3, CCT6, and KRIT1 (Santa Cruz Biotechnology), R- and β-tubulin, actin, and rabbit anti-FLAG (Sigma Aldrich), HSP70 and HSP90 (Cell Signaling Technology), and EF1A1 (Upstate). Rabbit anti-RIN2 was raised against a GST fusion protein comprising residues 425-895 of human RIN2. Donkey anti-rabbit, donkey anti-goat, and sheep anti-mouse secondary antibodies coupled to HRP were purchased from Jackson ImmunoResearch, Santa Cruz Biotechnology, and Amersham Biosciences, respectively. Detection of HRP was performed using SuperSignal West Pico Chemiluminescent Substrate from Pierce. Lipid Arrays. PIP arrays (Echelon Biosciences, Inc.) were blocked in 0.1% ovalbumin in Tris buffered saline with 0.05% Tween-20 (TBS-T) for 1 h and then were incubated with 0.5-1 µg/mL of recombinant 6×His-PDCD10 or OSM for 2 h. After washing unbound CCM proteins using TBS-T, the bound proteins were detected by immunoblotting with an anti-6×His antibody. Live Cell Imaging. COS7 cells were plated on 25 mm glass coverslips, and then the cells were transfected with the
research articles indicated constructs fused to mCherry, EYFP, or ECFP. The coverslips were placed in an imaging chamber (Molecular Probes) with medium and were imaged using a Zeiss Axiovert 200M inverted microscope with a 125-W xenon arc lamp (Sutter Instrument Company, Novato, CA), digital charge-coupled device camera (CoolSNAP HQ; Roper Scientific, Tucson, AZ), and Slidebook 4.0.10 software (Intelligent Imaging Innovations, Denver, CO). An objective lens (63× oil 1.25-numerical aperture, Plan-Neofluar [Zeiss]) was coupled with immersion oil to the bottom face of glass coverslips. Images from three planes were taken for each of the three channels (CFP [a band-pass excitation filter of 436/20 nm, a 455DCLP band beamsplitter, and a band-pass emission filter of 480/40 nm], YFP [a bandpass excitation filter of 500/20 nm, a 515DCLP band beamsplitter, and a band-pass emission filter of 535/30 nm], and Cy5 [a band-pass excitation filter of 620/60 nm, a 660DCLP band beamsplitter, and a band-pass emission filter of 700/75 nm]; Chroma). The three planes were deconvolved using the nearest neighbors algorithm. MudPIT Analysis. The precipitated FLAG peptide eluates described above were sonicated and resuspended in 0.1% Rapigest (Waters Corp, Milford, MA), reduced with 5 mM dithiothreitol at 60 °C for 15 min and alkylated with 15 mM iodoacetamide at room temperature for 30 min in the dark. The samples were then digested with sequence grade modified trypsin (Promega, Madison, WI) at a 1:50 enzyme/protein concentration at 37 °C overnight with gentle shaking in a thermomixer (Eppendorf, Westbury, NY). The Rapigest was hydrolyzed by the addition of concentrated HCl to a final concentration of 200 mM and incubation at 37 °C for 45 min. Insoluble particulates were removed by centrifugation at 20 000 g for 10 min. The peptide supernatant was loaded onto a desalting column (1 cm C18-reverse phase) and then connected to a two-phase, 100 µm inner diameter microcapillary column (10 cm C18-reverse phase, 3 cm strong cation exchange) with a 5 µm tip using an in-line filter assembly (Upchurch). A sixstep MudPIT was run using a linear MeCN gradient (5-60%) containing 0.1% formic acid over 120 min with salt pulses (0, 50 µL 300 mM, 50 µL 500 mM, 50 µL 700 mM, 50 µL 1 M, 100 µL 5M NH4OAc) injected at the beginning of each step. This multidimensional separation was performed in a manner similar to those described previously.27 However, instead of using a quaternary pump to deliver the salt pulses, an autosampler was used to deliver a plug of ammonium acetate directly in the path of the HPLC flow. Briefly, an Agilent 1100 binary pump was run at 200 µL/min and the flow was split immediately distal to the multiphasic capillary column using a microtee as previously described.28 A 50 µm ID capillary was added to the waste of the microtee and the length of the split capillary was adjusted to produce a flow rate through the capillary column of ∼250 nL/min. The autosampler was placed in-line between the HPLC and column and a flow splitter is used to add 50 µL ammonium acetate pulses (of variable concentrations) into the HPLC flow. By running the pump at 200 µL/min, the salt pulse reaches the column almost instantly and only a fraction of the total salt injected (∼0.1%) makes it onto the column. Final salt concentrations required per injection for optimal distribution of peptides between the individual steps were derived empirically from prior experiments on similar mixtures. Fractions were eluted directly onto a ThermoElectron LTQ mass spectrometer. A spray voltage of 2.4 kV was applied and mass spectra were acquired in a datadependent mode, whereby a single full mass scan (m/z 400Journal of Proteome Research • Vol. 6, No. 11, 2007 4345
research articles
Hilder et al.
1400) was followed by five tandem MS scans of the most intense peaks. The instrument raw files were converted to text files in ms2 format29 using libraries supplied by the manufacturer (ThermoElectron). Dynamic exclusion was enabled. The acquired tandem MS were searched against a mouse RefSeq database (downloaded from NCBI on July 11, 2005 and containing 26 996 entries and modified to include the programmed cell death 10 (CCM3) and recombinant FLAG-OSM protein sequences) concatenated with a shuffled decoy database as described30 using a normalized version of Sequest 27.31,32 Database selection was determined by cell type (mouse RAW264.7 macrophages). Database search parameters were as follows: no proteolytic enzyme specificity, no variable modifications were considered, and a static modification of +57 was assumed for all cysteines. Mass tolerance was (1.5 m/z for precursor ions (average masses), and fragment ions (monoisotropic masses) were binned to the nearest integer. The data was reassembled using DTASelect 1.9,33 and thresholds were set up as follows: Xcorr cutoff for singly, doubly, and triply charged peptides was set at 0.3,31 the ∆Cn was set to 0.1, only loci with unique peptides were allowed, proteins that are subsets of others were removed, peptides must have six or more residues, and three tryptic peptides were required for a positive identification. False discovery rates for protein identifications were maintained at
