|Home | About | Journals | Submit | Contact Us | Français|
Exposure of cells to certain cytokines can alter how these same cells respond to later cues from other agents, such as extracellular matrix or growth factors. Interferon (IFN)-γ pre-exposure inhibits the spreading of fibroblasts on fibronectin. Expression of the IFN-γ–induced GTPase murine guanylate-binding protein-2 (mGBP-2) can phenocopy this inhibition and small interfering RNA knockdown of mGBP-2 prevents IFN-γ–mediated inhibition of cell spreading. Either IFN-γ treatment or mGBP-2 expression inhibits Rac activation during cell spreading. Rac is required for cell spreading. mGBP-2 also inhibits the activation of Akt during cell spreading on fibronectin. mGBP-2 is incorporated into a protein complex containing the catalytic subunit of phosphatidylinositol 3-kinase (PI3-K), p110. The association of mGBP-2 with p110 seems important for the inhibition of cell spreading because S52N mGBP-2, which does not incorporate into the protein complex with p110, is unable to inhibit cell spreading. PI3-K activation during cell spreading on fibronectin was inhibited in the presence of mGBP-2. Both IFN-γ and mGBP-2 also inhibit cell spreading initiated by platelet-derived growth factor treatment, which is also accompanied by inhibition of Rac activation by mGBP-2. This is the first report of a novel mechanism by which IFN-γ can alter how cells respond to subsequent extracellular signals, by the induction of mGBP-2.
The ability of a cell to identify and respond to cues from the extracellular environment provides important points of regulation for many cellular processes, including proliferation, differentiation, adhesion, and migration (Gonzalez-Amaro and Sanchez-Madrid, 1999 ; Coppolino and Dedhar, 2000 ). Although much is known about how cells respond to individual cytokines, growth factors, and extracellular matrix (ECM) components, cells within whole organisms are frequently exposed to multiple extracellular signals either sequentially or simultaneously. This exposure to more than one signal is what is believed to occur in environments containing multiple proinflammatory cytokines. Although such an environment is believed to occur frequently, experimentally little is known about how exposure to one signal modulates the subsequent exposure to another.
Interferon (IFN)-γ is a proinflammatory cytokine frequently found in environments containing other cytokines and growth factors. IFNs have profound and pleiotropic effects on cells. Originally studied for their antiviral and antimicrobial activities, they also have antitumor, antiangiogenic, and proapoptotic effects on some cells (Boehm et al., 1997 ; Der et al., 1998 ; Stark et al., 1998 ). Other consequences of IFN exposure include changes in adhesion (Shaw and Mercurio, 1989 ; Shaw et al., 1990 ), migration (Adelmann-Grill et al., 1987 ), cytoskeletal arrangement (Pfeffer et al., 1980 ), and cellular morphology (Schuger et al., 1990 ). Although it is apparent that IFNs elicit these changes primarily through the transcriptional regulation of hundreds of genes (Boehm et al., 1997 ; Der et al., 1998 ), it remains unclear which induced or repressed proteins are responsible for which phenotypic changes.
The guanylate-binding proteins (GBPs) are a family of 67- to 69-kDa GTPases that are induced by IFN-α/β and IFN-γ, as well as tumor necrosis factor (TNF)-α and interleukin (IL)-1 (Vestal, 2005 ; Olszewski et al., 2006 ; Degrandi et al., 2007 ). GBPs are unusual because their third region of nucleotide contact is different from that of other GTPases (Cheng et al., 1991 ; Praefcke et al., 1999 ). Consequently, GBPs hydrolyze guanosine triphosphate (GTP) to both guanosine diphosphate (GDP) and guanosine monophosphate (GMP) (Schwemmle and Staeheli, 1994 ; Neun et al., 1996 ). Considerable structural and biochemical information has been obtained for one of the human (h) family members, hGBP-1. In addition to solving the crystal structures of full-length hGBP-1 without nucleotide (Prakash et al., 2000a ) and with nonhydrolyzable GTP analogue (Prakash et al., 2000b ), and the amino-terminal globular GTP binding domain alone (Ghosh et al., 2006 ), much is known about the biochemical properties of hGBP-1 in solution and the contribution of individual amino acids to those properties (Praefcke et al., 1999 , 2004 ; Kunzelman et al., 2005 ; Abdullah et al., 2009 ). Although some GBPs are robustly induced by IFN-γ treatment (Boehm et al., 1998 ), information on the role of GBPs in IFN responses is only recently emerging and is still incomplete.
hGBP-1 inhibits growth factor-induced proliferation of endothelial cells (Guenzi et al., 2001 ) and the expression of matrix metalloproteinase (MMP)-1 (Guenzi et al., 2003 ). This inhibition of MMP-1 results in decreased endothelial cell invasion and capillary tube formation (Guenzi et al., 2003 ). In addition, hGBP-1 protects tumor cells from killing by the chemotherapeutic drug paclitaxel (Duan et al., 2005 ). It is also suggested to inhibit apoptosis of endothelial cells upon growth factor withdrawal (Guenzi et al., 2001 ) and intestinal epithelial cells in a model of inflammatory bowel disease (Schnoor et al., 2009 ). hGBP-1 has modest antiviral activity against encephelomyocarditis virus (EMCV) and vesicular stomatitis virus (VSV) (Anderson et al., 1999 ) and inhibits replication of the hepatitis C virus replicon (Itsui et al., 2006 ). In addition, antimicrobial activity is suggested by the discovery of secreted hGBP-1 in cerebral spinal fluid of patients with bacterial meningitis (Naschberger et al., 2006 ). Recently, forced expression of hGBP-1 in human umbilical vein endothelial cells (HUVECs) was shown to inhibit their spreading on fibronectin (FN) (Weinlander et al., 2008 ).
New information is also available on the properties of some murine GBPs. mGBP-2, the putative murine orthologue of hGBP-1, also alters cell proliferation (Gorbacheva et al., 2002 ), modestly inhibits VSV and EMCV (Carter et al., 2005 ), and protects cells from killing by paclitaxel (Balasubramanian et al., 2006 ). Murine GBPs are also suggested to have antimicrobial properties. Murine GBP-5 promotes pyroptosis in human macrophage-like RAW 264.7 cells upon Salmonella infection, which might also be antimicrobial (Rupper and Cardelli, 2008 ). Mice infected with Listeria monocytogenes up-regulate expression of several of the murine GBPs, particularly mGBP-1, mGBP-2, and mGBP-7 (Degrandi et al., 2007 ). When mGBP expression was measured in murine lung and spleen after Toxoplasma gondii infection, the levels of all GBPs were increased by day 5 of infection (Degrandi et al., 2007 ). When RAW 264.7 macrophages were transfected with DsRed fusion constructs for some mGBPs (mGBP-6, mGBP-7, and mGBP-9) and treated with IFN-γ before infection with T. gondii type II strain ME49, mGBP-1, -2, -3, -6, -7, and -9 all relocalized to the parasitophorous vacuole within 30 min (Degrandi et al., 2007 ). IFN-γ treatment was required for this localization, suggesting that one or more other IFN-γ–induced proteins are required for this localization. These data suggest that at least some of the mGBPs may be involved in modulating T. gondii infection (Degrandi et al., 2007 ).
When mGBP-2 was cloned, NIH 3T3 fibroblasts were used to analyze its functional properties because they respond well to both IFN-α/β and IFN-γ. Early in the characterization of mGBP-2–expressing NIH 3T3 cells, it was noticed that the cells took longer to adhere and spread when transferred. The data from this study provide the first direct evidence that IFN-γ alters both fibroblast cell spreading on FN and after platelet-derived growth factor (PDGF) treatment. The study also identifies mGBP-2 as the IFN-regulated protein responsible for this inhibition of cell spreading and shows that the inhibition of cell spreading is not limited to fibroblasts but also occurs in B16 melanoma cells. mGBP-2 inhibits cell spreading by inhibiting signals from integrins or PDGF receptor that should activate Rac to promote cytoskeletal rearrangement. The inhibition of cell spreading by mGBP-2 is accompanied by the inhibition of phosphatidylinositol 3-kinase (PI3-K) activation.
Rabbit polyclonal antiserum against mGBP-2 was described previously (Vestal et al., 2000 ). Anti-FLAG monoclonal antibody M2 and rabbit anti-actin (A2066) were purchased from Sigma-Aldrich (St. Louis, MO). Anti-myc (9E10) antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Purified human fibronectin and mouse laminin were purchased from Roche Molecular Biochemicals (Indianapolis, IN). Mouse type IV collagen was purchased from Invitrogen (Carlsbad, CA). Mouse anti-Rac was purchased from BD Biosciences Transduction Laboratories (Lexington, KY). Murine recombinant (r)IFN-γ was purchased from Calbiochem (San Diego, CA) or PBL InterferonSource (Piscataway, NJ), and PDGF was purchased from Calbiochem.
The plasmids pRK5, pRK5 myc-Rac1(G12V), pRK5 myc-Cdc42(G12V), and pRK5 myc-RhoA(G14V) were a gift from Dr. Amy Wilson-Delfosse (Case Western Reserve University, Cleveland, OH). The plasmid for the p21-activated kinase (PAK) binding domain (PBD) of PAK1 as a glutathione transferase (GST)-fusion protein was obtained from Richard Cerione via Lawrence Quilliam (Indiana University-Purdue University Indianapolis, Indianapolis, IN). The generation of the plasmid c-myc-pcDNA3.1hygro(−) was described previously (Gorbacheva et al., 2002 ). The plasmid c-myc-mGBP-2-pcDNA3.1 was generated by excising mGBP-2 from FLAG-mGBP-2-pCMV2(NH) (Vestal et al., 1998 ) by using Not1/Kpn1 and ligation into Not1/Kpn1 cut c-myc-pcDNA3.1hygro(−). mGBP-1 was cloned using reverse transcription-polymerase chain reaction (RT-PCR) from RAW 264.7 cells treated with 500 U/ml IFN-γ for 18 h using 5′-ATATGCGGCCGCGCCTCAGAGATCCACATGTCG-3′ and 5′-GGTACCTTAAAGTATGGTGCATGATCGAGGTGGAGGCATGTTC-3′ as forward and reverse primers, respectively.
NIH 3T3, RAW 264.7, WEHI-3, and B16 cells were obtained from American Type Culture Collection and are maintained in DMEM (Mediatech, Herndon, VA) containing 10% fetal bovine serum (BioSource International, Camarillo, CA), 2 mM glutamine (Mediatech), and 50 U/ml penicillin/streptomycin (Mediatech). The FLAG epitope-tagged eukaryotic expression vectors containing either wild type or S52N mGBP-2 were generated and used to make stable cell lines in NIH 3T3 cells as described previously (Vestal et al., 2000 ; Gorbacheva et al., 2002 ). Characterization of the expression levels and growth characteristics of these clones were described previously (Gorbacheva et al., 2002 ). To generate B16 cells that stably express mGBP-2, cells were transfected with either pcDNA 3.1 myc or pcDNA 3.1 myc mGBP-2 using ExpressFect (Denville Scientific, Denville, NJ). Cells were selected in 300 μg/ml hygromycin, and pools of expressing and nonexpressing cells were analyzed for mGBP-2 expression and cell spreading as described below. Adhesion assays were performed using two different methods. First, cells were gently suspended in 0.05% trypsin/0.53 mM EDTA (Invitrogen) followed by DMEM containing 200 μg/ml soybean trypsin inhibitor (Sigma-Aldrich). After pelleting, cells were resuspended in DMEM containing 1% bovine serum albumin (BSA) and plated at 6 × 105 cells/well in 24-well dishes. The assays were also performed in 96-well format by plating 1 × 105 cells/well. After 20 min, the cells were washed and fixed with 2.5% glutaraldehyde. The attached cells were stained with 0.5% crystal violet, washed, and extracted with 1% SDS. The results were quantified by absorbance at 538 nm. The second method involved labeling cells with 4 μM calcein calcein AM (AM) (Invitrogen) per manufacturer's instructions before plating cells. After removal of nonadherent cells, adherent cells were measured by absorbance at 530 nm. Dishes were also coated with purified human fibronectin per manufacturer's instructions before adhesion assays. After coating, wells were blocked with 1% BSA in phosphate-buffered saline (PBS) overnight at 4°C.
Coverslips (12-mm circles) in 24-well dishes were coated with FN (10 μg/ml). Cells were gently suspended by treatment with 0.05% trypsin/0.53 mM EDTA (Invitrogen); trypsin inhibitor (Sigma-Aldrich) was added to 0.2 mg/ml; and then cells were washed twice in DMEM, diluted in DMEM, and allowed to recover in suspension for 30 min at 37°C. Cells (2.5–5 × 104) were plated per coverslip and allowed to adhere for 20, 40, or 60 min. Unattached cells were removed by gentle washing, and attached cells were fixed for 45 min with 4% paraformaldehyde. Coverslips were mounted in 50% VECTASHIELD (Vector Laboratories, Burlingame, CA) in PBS. Images were collected using either a MicroMax-1300Y camera (Roper Scientific, San Diego, CA) on a DMR upright microscope or a DM IRB HC inverted microscope (both Leica Microsyste, Deerfield, IL) with a cooled Retiga EX monochrome digital camera and analyzed with Image-Pro Plus software. For surface area determination, calibration was performed with a stage micrometer to provide the information of pixels per micrometer for the appropriate objective. To measure the surface areas, the perimeter of each cell was carefully outlined and its surface area was calculated by Image-Pro Plus, using the pixels per micrometer provided. The control transfectants (empty vector) varied in surface area from 200 μm2 to >3000 μm2. Surface areas of 100 cells per type were measured, grouped into 14 increments of 200 μm2, and expressed as the percentage of the adherent population of cells. The smallest increment (1) contained cells that had surface areas between 200 and 400 μm2 and the largest increment (14) contained cells with areas between 2800 and 3000 μm2. Cells were classified as “spread” or “unspread” based on measured size cutoffs of either 600 or 800 μm2. The association between cell type or treatment and proportion of unspread cells was assessed using Cochran–Mantel–Haenszel (CMH) tests of association, stratified by experiment repetition. The significance level of each test was 0.05; a Bonferroni correction was applied to the significance for a particular comparison when three tests were done (i.e., 0.05/3 = 0.017). Statistical analyses were performed using the SAS system (SAS Institute, Cary, NC). Once it had been decided that an unspread cell would be defined as one with a surface area of ≤600 μm2, then the means, SDs, and p values were calculated using Excel (Microsoft, Redmond, WA). Although results from a single transfection and isolation are shown, they were confirmed using transfectants from a second transfection process and isolation.
Control transfectants and mGBP-2–expressing cells (5 × 105/10-cm dish) were transfected with 6 μg of myc-tagged Rac1(G12V), RhoA(G14V), Cdc42(G12V), or pRK5 using ExpressFect (Denville Scientific), according to manufacturer's instructions. Twenty-four hours later, cells were processed for spreading as described above. After spreading on FN-coated coverslips for 25 min, cells were fixed and processed for two-color indirect immunofluorescence as described below using anti-Myc and anti-mGBP-2 as primary antibodies. Surface areas of stained cells were quantified by Image-Probsoftware, as described above.
NIH 3T3 cells (3.25 × 104) were plated into individual wells of a 24-well dish and cotransfected with 50 nM RISC-FREE transfection control and either 50 nM mGBP-2 SMART Pool siRNA (a pool of 5 siRNAs: 5′-GCUGUGUGGUGAAUUUGUAUU-3′, 5′-GUUGAAACACUUCUACAGAUU-3′, 5′-AGACGAUUCCGCUAACUUUUU-3′, 5′-CAAAACAAAUCGUCUGGGAUU-3′), 50 nM cyclophilin B siRNA (5′-GGAAAGACUGUUCCAAAA-3′), or 50 nM nontarget siRNA (5′-UAGCGACUAAACACAUCAA-3′) using Dharmafect #1. All siRNAs were from Dharmacon RNA Technologies (Lafayette, CO). Thirty hours after transfection, the cells were treated with IFN-γ (500 U/ml) for 18 h. For Western blot analysis, cells from six wells of the 24-well dish were lysed, combined, and 20 μg of cell lysates were analyzed as described by Western blots. For spreading assays, multiple wells were transfected and treated. The cells were suspended, and 1.5 × 105 cells of each type were plated onto FN-coated coverslips for 40 min. The surface areas of cells containing RISC-FREE reagent (red perinuclear staining) were determined.
Cells were plated onto coverslips in complete media overnight and then incubated in serum-free media for 2 h. After treatment with PDGF (10 ng/ml) for 0, 5, or 10 min, cells were washed, and actin filaments were labeled by fixing with 4% paraformaldehyde in PHEM buffer [60 mM piperazine-N,N′-bis(2-ethanesulfonic acid), 25 mM HEPES, 5 mM EGTA, 1 mM MgCl2, 3% sucrose, and 0.1% Triton-X 100] containing 10 μl of Alexa 594-conjugated phalloidin (Invitrogen) per 200 μl for 20 min at 37°C. Cells were washed three times with PBS and mounted in VECTASHIELD with 4,6-diamidino-2-phenylindole. Surface areas were determined as described above.
Cells were processed for cell spreading as described above. Cells were fixed and stained as described previously (Vestal et al., 2000 ). The antibodies used were rabbit polyclonal anti-mGBP-2 (1851; 1:200), mouse anti-vinculin (1:100; Clone VIN-11-5; Sigma-Aldrich), Alexa 488-conjugated goat anti-mouse and anti-rabbit, and Alexa 594-conjugated goat anti-rabbit and anti-mouse (Invitrogen). Images were captured with a DM IRB HC inverted microscope (Leica Microsystems) with a cooled Retiga EX monochrome digital camera. Confocal images were collected using aTCS-SP spectophotometric laser scanning confocal microscope (Leica Microsystems). For staining with antiphosphotyrosine antibody (1:800; 4G10; Millipore, Billerica, MA), cells were processed for indirect immunofluorescence as described previously (Vestal et al., 2000 ), except for the inclusion of phosphatase inhibitors in the wash buffer. Images were all collected keeping the gain, offset, and zoom constant. No adjustments were made in image intensity when the figures were prepared.
To examine cells for the expression of integrin α4, RNA was isolated from cells using RNeasy mini kit (QIAGEN, Valencia, CA) per the manufacturer's instructions. cDNA was generated using 1.5 μg of total RNA using the SuperScript First Strand Synthesis kit (Invitrogen) and the random hexamers provided. PCR amplification was performed using 1.5 μg of cDNA and two different PCR primer pairs for α4. The two different primer pairs were designed using the Pick Primers function at www.ncbi.nlm.nih.gov and the cDNA sequence at NM_010576. Each of these primer pairs was chosen to amplify a region of α4 that spanned at least one intron, to control for possible DNA contamination of the samples and the primer pairs did not recognize any other sequences in the mouse database. The first primer pair has 5′-AGCTCCAGCTGGGTAGCCCC-3′ as forward and 5′-GGCTCCTCCCACGACTTCGGT-3′ as reverse primer. The second primer pair has 5′-TGGAAAGGGTCCTTGTCGGAAGTGA-3′ as forward and 5′-AGGCCTTGTCCTTAGCAACACTGC-3′ as reverse primer. The mouse glyceraldehyde-3-phosphate dehydrogenase primers used were 5′-CCAGGTTGTCTCCTGCGACT-3′ as forward and 5′-ATACCAGGAAATGAGCTTGACAAAGT-3′ as reverse.
To analyze Rac activation during cell spreading, control tranfectants and mGBP-2–expressing cells (5 × 106) were plated onto two 150-mm dishes and allowed to adhere overnight. After serum starving for 2 h, the cells were allowed to spread for 30 min in serum-free media onto two 100-mm fibronectin-coated (10-μg/ml) plates. Attached cells were lysed in 250 μl of lysis buffer per dish (50 mM Tris, pH 7.5, 10 mM MgCl2, 1% IGEPAL CA-630 [Sigma-Aldrich], 150 mM NaCl, 1 mM sodium vanadate, 1 μl/ml protease inhibitor cocktail [Sigma-Aldrich], and 1 mM phenylmethylsulfonyl fluoride [PMSF]; Mira et al., 2000 ). Cell lysates were clarified by centrifugation at 13,000 × g for 3 min at 4°C. Cell lysates (500 μg) were added to 30 μg of PAK1 PBD/GST on beads in a final volume of 500 μl, and the samples were rotated at 4°C for 45 min. The resulting complexes were washed twice with 500 μl of lysis buffer and then twice with 500 μl of lysis buffer without IGEPAL. Complexes were dissociated and size fractionated on 10% PAGE gels. Total cell lysates (TCLs; 20 μg) were also included. Rac levels were determined by Western blot analysis and were quantified using 1D Scientific Imaging software (Eastman Kodak, Rochester, NY). The ratios of Rac from pull-downs and total cell lysates were determined and normalized to control cell samples for each blot.
To analyze Rac activation by PDGF, control transfectants and mGBP-2–expressing cells (5 × 106) were plated, allowed to adhere overnight, serum starved for 2 h, and treated with 10 ng/ml PDGF for 5 min. To analyze Rac activation by PDGF subsequent to IFN-γ treatment, cells were plated as described above, allowed to adhere for 4 h, and then serum starved in the presence or absence of IFN-γ (500 U/ml) for 18 h. Cells were lysed and processed for analysis of Rac activity as described above.
Control transfectants and mGBP-2–expressing NIH 3T3 cells were serum starved for 1 h and processed for cell spreading on FN. Adherent cells were lysed in 200 μl of cold radioimmunoprecipitation assay lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 1 μl/ml Protease inhibitor cocktail [Sigma-Aldrich], 1 mM PMSF, 25 mM sodium fluoride, 10 mM sodium vanadate, 100 μM sodium molybdate, and 50 mM sodium pyrophosphate) per 100-mm FN-coated dish on ice. Samples of the cells were also taken 1) after the 1-h serum starvation and 2) before plating but after the recovery in suspension. Lysates were clarified by centrifugation at 10,000 × g for 5 min at 4°C and analyzed by Western blot with anti-phospho-Akt antibodies as described below. The membranes were then stripped and immunodetected with antiserum against total Akt. Films were quantified using 1D Scientific Imaging software (Eastman Kodak) and the ratios of phospho-Akt and total Akt were determined.
Control, mGBP-2-, and/or S52N mGBP-2–expressing NIH 3T3 cells were plated at 5 × 106 cells per 150-mm dish and allowed to adhere overnight. Four 150-mm dishes were plated for each cell line. Cells were suspended and allowed to recover as described previously. The cells from all four 150-mm dishes were divided and plated onto two 150-mm FN-coated dishes for 35 min. Unattached cells were removed by gentle washing with Hanks' balanced salt solution containing 1 mM sodium vanadate. Adherent cells were lysed in 250 μl of Tris-buffered saline/Tween 20 lysis buffer (10 mM Tris, pH 7.4, 1 mM EDTA, 150 mM NaCl, 1.0% Triton X-100, 1 μl/ml Protease Inhibitor Cocktail [Sigma-Aldrich], 0.1 mM PMSF, 25 mM sodium fluoride, 10 mM sodium vanadate, 100 μM sodium molybdate, and 50 mM sodium pyrophosphate) per dish. Lysates were clarified by centrifugation at 10,000 × g for 5 min at 4°C. Cell lysates (500 μg) were added to 100 μl of FLAG M2 agarose (Sigma-Aldrich) or 5 μl of anti-PI3-K p110 antisera or FLAG M2 antisera in a final volume of 500 μl of lysis buffer and rotated overnight at 4°C, followed by protein G-Sepharose for 1 h. The immune complexes were washed three times with 500 μl of lysis buffer and the immune complexes were dissociated in 5× Laemmli sample buffer by heating at 95°C for 4 min. Samples were size fractionated on SDS-PAGE gels and analyzed by Western blotting as described below. TCLs (20 μg) also were included.
PAGE and Western analysis of cell lysates were performed as described previously (Gorbacheva et al., 2002 ). Antibodies were used at 1:500 mouse anti-FLAG (M2), 1:1000 mouse anti-Rac (BD Biosciences Transduction Laboratories), 1:800 mouse anti-β1 (BD Biosciences Transduction Laboratories), 1:5000 rabbit anti-α5 (Millipore Bioscience Research Reagents, Temecula, CA), 1:800 rabbit anti-mGBP-2 (1851), 1:1500 rabbit anti-cyclophilin B (Ab16045; Santa Cruz Biotechnology), 1:500 mouse anti-IFN-γ–induced GTPase (IGTP; BD Biosciences Transduction Laboratories), 1:200 mouse anti-myc (Santa Cruz Biotechnology), 1:500 phospho-Akt (Cell Signaling Technology, Danvers, MA), 1:1000 anti-Akt (Cell Signaling Technology), 1:500 rabbit monoclonal anti-PI3-K p85α (04-403; Millipore; recognizes the C terminus), 1:200 rabbit polyclonal anti-PI3-K p110 (Sc-7189; Santa Cruz Biotechnology; reacts with p110α, -β, -γ, and -δ), and 1:300 for rabbit anti-actin (Sigma-Aldrich). Secondary antibodies used were horseradish peroxidase-conjugated goat anti-rabbit (1:10,000; Rockland Immunochemicals, Gilbertsville, PA), and anti-mouse (1:800; Invitrogen) and detected using enhanced chemiluminescence (GE Healthcare, Little Chalfont, Buckinghamshire, United Kingdom). Antibodies were removed using Restore Western blot stripping buffer (Pierce Chemical, Rockford, IL).
To analyze PK3-K activity during cell spreading on FN, control transfectants and mGBP-2–expressing cells were resuspended and processed for cell spreading on FN as described above. Twenty minutes after plating on FN, the nonadherent cells were removed by gentle washing and the adherent cells were lysed in 20 mM Tris-HCl, pH 7.4, 137 mM NaCl, 1 mM CaCl2, 1 mM MgCl2, 0.1 mM sodium vanadate, 1% NP-40, and 1 mM PMSF. Cell lysates (500 μg) were incubated with anti-p85 PI3-K antibody (06-195; Millipore) overnight at 4°C. Protein A-Sepharose was added for 1 h, and the immune complexes were washed and processed using a PI3-kinase activity enzyme-linked immunosorbent assay (ELISA) kit (Pico; Eschelon Biosciences, Logan, UT) and following the manufacturer's instructions. The level of phosphatidylinositol (3,4,5)-trisphosphate (PIP3) generated by serum-starved control transfectants in each assay was set to 100%, and the values from all other samples were normalized to this.
The GBPs are an incompletely characterized family of large GTPases (Vestal, 2005 ; Olszewski et al., 2006 ; Degrandi et al., 2007 ). The murine family member mGBP-2 is not detected in NIH 3T3 cells before treatment with IFN-γ and only becomes detectable after 4 h of treatment (Vestal et al., 2000 ). Stable NIH 3T3 fibroblast cell lines expressing mGBP-2 have been established, and the expression levels of mGBP-2 in several of these cell lines are comparable with the level of mGBP-2 observed in untransfected cells after IFN-γ treatment for 24 h. This information has been published previously (Gorbacheva et al., 2002 ). During the characterization of these cell lines, we observed that cells expressing mGBP-2 did not attach as quickly to tissue culture dishes as the control transfectants (Figure 1). To explore the possibility that there was a defect in the ability of cells expressing mGBP-2 to adhere to FN, calcein AM was used to confirm that cells expressing mGBP-2 do not adhere as rapidly as control transfectants (empty vector) (Figure 1A). This observation was followed by experiments to determine whether there was also a defect in the ability of mGBP-2–expressing cells to spread on FN.
Cell spreading is a complex process that follows cell attachment and involves the generation of new substrata adhesion sites, cytoskeletal reorganization, and lamellipodia formation (Small et al., 1999 ). mGBP-2–expressing NIH 3T3 cells were plated on FN-coated coverslips and allowed to attach/spread for 20 min (Supplemental Figure 1). Control transfectants (empty vector) exhibited surface areas ranging from 200 to 400 μm2 (increment 1) to 2800–3000 μm2 (increment 14). Comparison of the distribution of surface areas of the cells indicates that there are more mGBP-2–expressing cells in the smaller three size increments than control transfectants (Figure 1B). Only 16 ± 5% of the control cells had surface areas ≤600 μm2 (increments 1 + 2), whereas 42 ± 10% of the mGBP-2–expressing cells failed to spread to surface areas >600 μm2 (Figure 1C). Using a less stringent measure of an unspread cell (the first 3 increments, or ≤800 μm2), there were still approximately twice as many mGBP-2–expressing cells that failed to spread compared with control cells (Figure 1C). Because the relative proportion of cells expressing mGBP-2 that remained poorly spread was the same with either the ≤600 or ≤800 μm2 as a measurement of small, poorly spread cells, the more stringent cut-off of ≤600 μm2 was chosen to represent unspread cells in subsequent experiments. Comparable results were obtained for other clones expressing mGBP-2 at similar levels (data not shown). mGBP-2 does not abolish the ability of NIH 3T3 cells to spread on FN, it just slows the process considerably. When the cells were plated on FN-coated coverslips and allowed to adhere for 24 or 48 h, both control and mGBP-2–expressing cells were able to spread to sizes >2400 μm2 (Supplemental Figure 2, A and B).
NIH 3T3 cell lines that expressed S52N mGBP-2 at comparable levels to the mGBP-2–expressing cells evaluated in the previous experiments were generated and characterized previously (Gorbacheva et al., 2002 ). S52N mGBP-2 has a single amino acid substitution in the GTP binding domain of mGBP-2 that is analogous to the S17N substitution of Ras. Because this amino acid residue is involved in coordination of the magnesium necessary for GTP binding, when this mutation is generated it results in proteins with reduced affinity for GTP (Feig, 1999 ). The crystal structure of the highly homologous, putative human orthologue of mGBP-2, hGBP-1, showed that S52 of hGBP-1 is involved in coordination of the magnesium ion as observed for other GTPases (Prakash et al., 2000 ). In vitro studies of hGBP-1 suggest that it will be nucleotide free at cellular concentrations of guanine nucleotides (Praefcke et al., 2004 ). Although it has not been confirmed for mGBP-2, we expect that because of the high homology between hGBP-1 and mGBP-2 that S52N mGBP-2 will behave the same. NIH 3T3 cells expressing S52N mGBP-2 were examined for their ability to spread on FN (Figure 1, B and C, and Supplemental Figure 1). No significant difference was observed between the spreading of S52N mGBP-2–expressing cells and the control transfectants.
We examined the effect of IFN-γ treatment on NIH 3T3 cell spreading on FN. NIH 3T3 cells treated with IFN-γ for 8 and 18 h were allowed to spread on FN (Figure 1D and Supplemental Figure 1). By 8 h of IFN treatment, NIH 3T3 cells were retarded in their ability to spread (70 ± 10% had surface areas ≤600 μm2 compared with 31 ± 12% for untreated cells) (Figure 1D). No additional reduction in spreading was observed after 18 h of IFN-γ treatment (70 ± 15% had areas ≤600 μm2). IFN-γ treatment results in maximal mGBP-2 expression by 8 h, and expression remains high through 24 h (Gorbacheva et al., 2002 ).
To ask whether mGBP-2 can inhibit the spreading of other cells, murine B16 melanoma cells were generated that contained empty vector (control) or vector with myc epitope-tagged mGBP-2 (mGBP-2) (Figure 1E). The surface areas of the B16 cells after 20 min of spreading on FN is shown (Figure 1F). The B16 cells had spread to surface areas ranging from <100 μm2 to just >600 μm2 in 20 min. However, just over twice as many of the B16 cells expressing mGBP-2 still had surface areas in the smallest two increments (<100–200 μm2) after 20 min on FN compared with the control transfects. When B16 control and mGBP-2–expressing cells are allowed to spread on FN for 24 h (Supplemental Figure 2C), the number of mGBP-2–expressing B16 cells in the smaller increments is not more than observed for the control transfectants. From this, we conclude that mGBP-2 inhibits cell spreading in the B16 melanoma cells.
To further characterize the differences in control transfectant and mGBP-2–expressing NIH 3T3 cells during cell spreading, cells were allowed to spread for 20 min on FN and were examined by staining for phosphotyrosine (Figure 2A). Examination of phosphotyrosine staining at the basal surface of spreading cells allows visualization of cell–substrata contact sites where active signaling events are occurring. The images presented are from the basal surface of the cells and have been captured under the same conditions. Although there was significant heterogeneity in the phosphotyrosine staining even within the control or mGBP-2–expressing cells, when cells of comparable sizes were compared the staining at the basal surface tended to be less bright and the foci of staining somewhat smaller in the mGBP-2–expressing NIH 3T3 cells. Whether mGBP-2 was concentrated at sites of focal adhesions was examined by double-label indirect immunofluorescence for mGBP-2 and vinculin in cells spreading on FN (Figure 2B). mGBP-2 has a punctate distribution throughout the cytoplasm and is not appreciably colocalized with vinculin, suggesting that little, if any, mGBP-2 is found in focal adhesions.
To determine whether mGBP-2 is required for IFN-γ-mediated inhibition of cell spreading, mGBP-2–specific siRNAs were used to knockdown the expression of mGBP-2 in IFN-γ–treated NIH 3T3 cells (Figure 3). NIH 3T3 cells were cotransfected with RISC-FREE transfection control and mGBP-2 siRNAs, cyclophilin B siRNA, or nontarget siRNA. After 30 h, the cells were treated with IFN-γ, and the specificity of the knockdowns were evaluated (Figure 3A). As expected, in the absence of IFN-γ no mGBP-2 is detected. In the presence of the mGBP-2 siRNAs, the level of mGBP-2 after IFN-γ treatment is reduced by >80% (Figure 3A). Nontarget and cyclophilin B siRNAs have no effect on mGBP-2 expression. The mGBP-2 siRNAs have no effect on the expression of another IFN-γ–induced GTPase, Irgm3 (formerly IGTP) (Bekpen et al., 2005 ), or on the level of cyclophilin B. To determine whether IFN-γ treatment could still inhibit cell spreading in the absence of mGBP-2, NIH 3T3 cells were transfected with RISC-FREE transfection control and either nontarget siRNA or mGBP-2–specific siRNAs; treated with IFN-γ; and analyzed for cell spreading (Figure 3B). The knockdown of mGBP-2 completely rescued IFN-γ–induced inhibition of cell spreading (Figure 3B). Although IFN-γ induces the expression of all of the murine GBPs (Degrandi et al., 2007 ), mGBP-1 is the most closely related murine GBP to mGBP-2, with ~80% amino acid identity (Olszewski et al., 2006 ). All other murine GBPs are significantly different from mGBP-2 (identities in the 50% range). No good commercial antibodies against mGBP-1 are available to determine whether the siRNAs against mGBP-2 have any effect on the expression of mGBP-1. To rule out a contribution of mGBP-1 to IFN-γ–initiated inhibition of cell spreading, cells were transiently transfected with mGBP-1 or mGBP-2 and their relative abilities to inhibit spreading on FN were evaluated. mGBP-1 was unable to inhibit cell spreading (Figure 3C). Therefore, we conclude that mGBP-2 is the major contributor to IFN-γ–mediated inhibition of cell spreading.
IFN-γ treatment of cells can result in the transcriptional regulation of hundreds of genes, including a number of integrins (Boehm et al., 1997 ; Der et al., 1998 ). Adhesion of NIH 3T3 cells to FN is mediated primarily by the integrin heterodimer α5β1. IFN-γ treatment for as long as 48 h did not alter the amount of α5 or β1 expressed by NIH 3T3 cells (Supplemental Figure 3, A and B). In addition, IFN-γ treatment for up to 48 h did not alter the steady-state expression of the linker proteins talin, vinculin, paxillin, or α-actinin (data not shown). Similar to treatment with IFN-γ, mGBP-2 expression did not reduce the steady-state levels of α5 or β1 (Supplemental Figure 3, B and C) or the linker proteins talin, vinculin, paxillin, or α-actinin (data not shown). In fact, they have slightly increased levels of surface α5 or β1.
To determine whether increasing the FN concentration could restore cell spreading, control and mGBP-2–expressing cells were allowed to spread on concentrations of FN ranging from 2 to 20 μg/ml (Supplemental Figure 2D). Increasing the concentration of FN beyond 10 μg/ml did not increase cell spreading, suggesting that mGBP-2 does not inhibit cell spreading by reducing integrin affinity.
While preparing this manuscript, it was reported that the putative human orthologue of mGBP-2, hGBP-1, inhibits HUVEC cell spreading and that this phenotypic change is correlated with the up-regulation of integrin α4 (Weinlander et al., 2008 ). Examination of NIH 3T3 cells showed mGBP-2 expression did not induce the expression of integrin α4 (Supplemental Figure 3D).
The initial observation that mGBP-2–expressing cells took longer to stably adhere to uncoated culture dishes suggested that the inhibition of spreading might not be FN dependent. The spreading of mGBP-2–expressing cells was examined on both collagen IV and laminin. Cell spreading on collagen IV and laminin took more time than on FN (40 vs. 20 min), but the same trend toward slower spreading by mGBP-2–expressing cells was observed. On collagen IV, fewer control transfectants remained unspread (≤600 μm2; 50.5 ± 10%) compared with mGBP-2–expressing cells (68 ± 2%; n = 2) (data not shown). On laminin, again fewer control cells (26.6 ± 4%) remained unspread compared with mGBP-2–expressing cells (59.3 ± 9%; n = 2) (data not shown). mGBP-2 inhibition of cell spreading is not dependent on a particular (Zhang et al., 2006 ) integrin; therefore, it might be through the inhibition of intracellular signals regulating the actin cytoskeleton and be common to many integrins.
Members of the Rho family of small GTPases are master regulators of the actin cytoskeleton, mediating cell spreading (Hall, 1998 ; Price et al., 1998 ; Ridley, 2000 ). To determine whether Rac1, Cdc42, or RhoA are downstream of mGBP-2–mediated inhibition, their ability to rescue cell spreading was examined. Control transfectants (pCMV) or mGBP-2–expressing NIH 3T3 cells were transfected with either Rac1(G12V), RhoA(G14V), Cdc42(G12V), or empty pRK5 and spreading on FN was analyzed (Figure 4A). After transfection with empty pRK5, control mGBP-2–expressing cells were still significantly inhibited in their ability to spread on FN compared with control transfectants (75.8 ± 4.2 vs. 23.97 ± 4.4%). As shown in other cells, expression of Rac1(G12V) in control transfectants promotes cell spreading (13.1 ± 1.5 vs. 23.97 ± 4.4% unspread cells) (Zhang et al., 22006 ). As expected, both RhoA(G14V) (88.9% unspread cells) and Cdc42(G12V) (42.0 ± 5.4%) inhibited cell spreading. However, only Rac1(G12V) was capable of restoring cell spreading in the presence of mGBP-2 (13.4 ± 0.9% unspread cells) (Figure 4A). These data suggest that mGBP-2 is upstream of Rac1. If this is correct, cell spreading in the presence of mGBP-2 should result in lower levels of activated Rac. Consistent with this prediction, less activated Rac is detected in mGBP-2–expressing cells as they spread on FN (Figure 4, B and C). Because the cells were serum-starved to quiet signaling before adhesion/spreading, the levels of active Rac was below the level of detection in cells before suspension and restimulation by plating onto FN. Although there is sometimes some variability in the level of Rac observed in the TCLs from the spreading cells, these differences are compensated for when the quantifications for active Rac are normalized for differences in Rac in the TCLs. mGBP-2 does not itself alter the level of Rac in NIH 3T3 cells (Figure 4D).
Consistent with Rac inhibition being critical to the inhibition of spreading by IFN, IFN-γ treatment also inhibits Rac activation during cell spreading (Figure 5, A and B).
Integrin engagement is not the only physiological stimulus for cell spreading. PDGF can also induce flattening and spreading in several cell types, a process that is Rac dependent (Symes and Mercola, 1996 ; James et al., 2004 ; Petroll et al., 2008 ; Abramovici et al., 2009 ). To ask whether mGBP-2 could inhibit the spreading of cells induced by PDGF treatment, control transfectants and mGBP-2–expressing cells on uncoated coverslips were serum starved for 2 h and treated with PDGF for 5 or 10 min and stained with phalloidin (Figure 6D). The surface areas were measured and the size distributions of untreated control transfectants and mGBP-2–expressing cells are shown in Figure 6A. Before treatment with PDGF, the cells seem to shrink in surface area compared with cells in complete media (Figure 6A vs. Supplemental Figure 2), but there are no striking differences in the surface areas of control transfectants versus mGBP-2–expressing cells. After 5 min of PDGF treatment the control cells have increased in size, whereas the mGBP-2–expressing cells remained comparable in size to before treatment (Figure 6, B and D). After 10 min of PDGF treatment, the differences between the surface areas of the PDGF-treated control and mGBP-2–expressing cells persists (Figure 6, C and D). Together, these data demonstrate that PDGF treatment of NIH 3T3 cells does increase their surface areas and that the expression of mGBP-2 inhibits this spreading.
Rac can be activated by a variety of stimuli, including integrin engagement, growth factors, cadherins, immunoglobulin superfamily receptors, and G protein-coupled receptors (Schwartz, 2004 ). Rac activation by PDGF was examined in the presence and absence of mGBP-2 (Figure 7A). As expected, PDGF treatment of control transfectants resulted in Rac activation (Figure 7A). In the presence of mGBP-2, this activation was attenuated by ~50% (Figure 7B). Consistent with this, IFN-γ treatment of untransfected cells also resulted in inhibition of Rac activation upon PDGF treatment (Figure 7, C and D). Together, these data show that sustained IFN-γ treatment inhibits Rac activation by both integrin engagement and PDGF treatment, as a consequence of mGBP-2 induction. PDGF treatment itself does not induce the expression of mGBP-2 and does not alter the level of mGBP-2 induced by IFN-γ (Figure 7E).
Having demonstrated that Rac is inhibited by mGBP-2 subsequent to plating on FN or PDGF treatment, we wanted to know where in the pathway(s) from these two stimuli mGBP-2 acts. Operating on the hypothesis that the protein(s) directly inhibited by mGBP-2 would need to come into contact with mGBP-2, immunoprecipitations were used to determine whether mGBP-2 interacts directly with Rac during cell spreading (data not shown). No direct interaction between the two proteins was detected, suggesting to us that mGBP-2 probably does not act directly on Rac. Forced overexpression of the Rac guanine exchange factor Tiam-1 was able to rescue cell spreading in mGBP-2–expressing cells, but no direct interaction between mGBP-2 and Tiam-1 was detected (data not shown). This suggested that the point of inhibition by mGBP-2 was more proximal to integrin or PDGF receptor.
PI3-K is a major contributor to the activation of Rac upon either integrin engagement or PDGF treatment (Bokoch et al., 1996 ; Symes and Mercola, 1996 ; Welch et al., 2003 ). For this reason, PI3-K is a critical component of the signaling cascade(s) involved in lamellipodia formation and cell spreading (Wennstrom et al., 1994 ; Berrier et al., 2000 ; Hermanto et al., 2002 ; Welch et al., 2003 ; Qian et al., 2005 ; Lasunskaia et al., 2006 ). PI3-K activation was analyzed indirectly by monitoring changes in the levels of active Akt (phospho-Akt) during cell spreading in the presence or absence of mGBP-2 (Figure 8, A and B). pAkt levels were reduced significantly in the presence of mGBP-2, suggesting that mGBP-2 may inhibit PI3-K activity.
To determine whether PI3-K is the point of interaction with mGBP-2, we asked whether mGBP-2 forms a protein complex with one or more of the subunits of PI3-K. To answer this, cell lysates from control or mGBP-2–expressing cells spreading on FN were immunoprecipitated with anti-FLAG antibody to isolate mGBP-2. Western blot analyses showed that the p110 subunit of PI3-K is part of a protein complex containing mGBP-2 (Figure 8C). We confirmed this interaction by immunoprecipitation with anti-p110 antisera (Figure 8D). In cell lysates immunoprecipitated with anti-FLAG to isolated mGBP-2 and subsequently immunodetected with antisera for p85, the level of p85 in the immune complexes generated in the presence of mGBP-2 was not above background (Figure 8C). We conclude from this that little, if any, p85 is in the same protein complex as mGBP-2 and p110.
To confirm the importance of the interaction between the p110 subunit of PI3-K and mGBP-2 to the inhibition of cell spreading, the ability of S52N mGBP-2 to interact with p110 was examined (Figure 8E). Immunoprecipitation of p110 from spreading cells demonstrated the expected interaction with mGBP-2 but S52N mGBP-2 did not interact with p110. Even at longer exposure times, no interaction was detected between S52N mGBP-2 and p110.
Control transfectants and mGBP-2–expressing cells were analyzed for PI3-K activity during cell spreading. After 2 h of serum starvation, the basal PI3-K activity in mGBP-2–expressing cells was modestly inhibited compared with that of control transfectants (72.4 ± 3.3% of control transfectants) (Figure 9). That this could be observed when no Akt phosphorylation was observed in serum starved cells is believed to reflect the higher sensitivity of the PI3-kinase activity ELISA. Spreading of control transfectants on FN activated PI3-K significantly (157.8 ± 16.7% of control). In the presence of mGBP-2, the activation of PI3-K by spreading on FN was significantly inhibited (81.6 ± 18.6% of control). In fact, the PI3-K activity in mGBP-2 cells spreading on FN was not significantly different from that of serum-starved mGBP-2–expressing cells (p = 0.48).
Together, these data suggest that IFN-γ inhibits cell spreading by inhibiting Rac activation through induction of mGBP-2. This inhibition of Rac activation would be expected to slow the cytoskeletal rearrangements needed for cell spreading, consistent with our data. mGBP-2 inhibition of Rac activation is accompanied by interaction of mGBP-2 with the p110 subunit of PI3-K, a well characterized activator of Rac. Although the cellular fraction of p110 that interacts with mGBP-2 seems to be relatively small, the observation that S52N mGBP-2 neither interacts with p110 nor is able to inhibit cell spreading suggests that this interaction is physiologically significant. In fact, the presence of mGBP-2 inhibits PI3-K activation during NIH 3T3 cell spreading on FN.
Cell adhesion plays a role in normal tissue development and maintenance, as well as such processes as hematopoiesis, lymphocyte homing, and immune recruitment. It also plays a central role in disorders such as cancer, inflammation, and vascular disease. To facilitate these many processes, binding of adhesion molecules to ECM must result in the productive transduction of intracellular signals from the plasma membrane to sites as divergent as the actin cytoskeleton and the nucleus. The IFNs are involved in a variety of these processes and disorders, but their roles in influencing cell adhesion are incompletely understood.
IFNs are a family of cytokines to which many, if not most, cells are responsive (Rubinstein et al., 1987 ). An area of IFN-mediated cellular changes that has been poorly explored is the role of IFNs in the induction of cytoskeletal changes. Although IFN treatment of some cancer cells can result in increased actin stress fiber formation and adhesion (Brouty-Boye, 1982 ), the opposite is observed for other cells (Hovi et al., 1985 ). It seems that all IFN-treated cells undergo some cytoskeletal changes, but studies suggest that the cell background significantly influences these changes.
Both IFN-α/β and IFN-γ can alter cell adhesion. They do this in a variety of ways, all of which seem to show some cell type variations. In some cells, IFNs alter the expression levels of ECM proteins, thereby altering the adhesive properties of the cell's microenvironment. Whether the expression of individual components of the ECM are up- or down-regulated is very dependent on cell type (Maheshwari et al., 1990 ; Schuger et al., 1990 ; Diaz and Jimenez, 1997 ). IFNs can also influence adhesion by altering the expression levels of the receptors that mediate both ECM and cell–cell adhesion, particularly integrins. Again, these responses vary between cell types (Maruguchi et al., 1991 ; Tsao et al., 1994 ; Bauvois et al., 1996 ). In addition, IFN-γ modulation of integrin-mediated adhesion in some cells does not seem to be the result of changes in levels of integrin receptors but the consequence of altering the ability of existing integrins to support adhesion. This modulation is the result of altered cytoskeletal association, phosphorylation, and reduced recruitment to focal adhesion sites (Shaw and Mercurio, 1989 ; Shaw et al., 1990 ; Wills et al., 1999 ).
One group of IFN-induced proteins for which functional information has only recently been acquired is the GBP family (reviewed in Vestal, 2005 ). They were identified almost 20 yr ago as some of the most abundantly IFN-induced proteins (Cheng et al., 1983 , 1985 ). GBPs are a family of 67- to 69-kDa GTPases that are induced by both type I (α/β) and type II (γ) IFNs (Cheng et al., 1986 ; Decker et al., 1991 ; Vestal et al., 2000 ), as well as TNF-α and IL-1 (Cardozo et al., 2001 ; Guenzi et al., 2001 ). This profile of cytokines that induce GBPs suggests a role for GBPs in more generalized responses to proinflammatory cytokines. GBPs are unusual because their primary amino acid sequences contain only two of the three highly conserved consensus nucleotide binding motifs found in other GTPases (Cheng et al., 1991 ). The third region of nucleotide contact identified for GBPs is different from that observed for other GTPases (Praefcke et al., 1999 ; Prakash et al., 2000a ). Consequently, GBPs have both GDP and GMP as end products of GTP hydrolysis (Schwemmle and Staeheli, 1994 ; Neun et al., 1996 ; Schwemmle et al., 1996 ). Several members of this protein family also have a CaaX sequence at their extreme carboxy termini. CaaX sites are potentially involved in the recruitment of isoprenyl transferases that are responsible for the attachment of either C-15 farnesyl or C-20 geranylgeranyl isoprenoids to the C (cysteine) of the CaaX site (Casey and Seabra, 1996 ). Several of the GBPs have been shown to be isoprenylated in vivo (Nantais et al., 1996 ; Vestal et al., 1996 , 1998 ; Stickney and Buss, 2000 ).
Despite the observation that one of the murine GBPs, mGBP-2, is robustly induced after IFN-γ treatment (Boehm et al., 1998 ), almost nothing was known about the role of these proteins in mediating IFN responses until recently. Our initial discovery that the IFN-induced GTPase mGBP-2 could inhibit cell adhesion and spreading (Figure 1) suggests that we have identified a new mechanism by which IFN-γ could modulate interaction of cells with their extracellular environment. Because Rac controls lamellipodia formation in cells as they spread, we asked whether mGBP-2 inhibited Rac activity during cell spreading on FN (Figure 4). The reduction in active Rac facilitated by mGBP-2 would be expected to slow rearrangement of the actin cytoskeleton and consequently slow shape change. Depending on where in the pathways leading to Rac activation this inhibition occurs, the slowing of shape changes could be common to more than one external signal. The observation that cell spreading is inhibited by mGBP-2 on more than one matrix molecule and that adhesion is reduced on uncoated tissue culture dishes (data not shown) suggested that mGBP-2 inhibits a pathway to Rac activation that is common to multiple stimuli and is not the result of the inhibition of signals from a single integrin. This was confirmed when mGBP-2 was demonstrated to inhibit Rac activation by PDGF (Figure 7).
PI3-K is activated by almost all integrins, growth factor and cytokine receptors, and some G protein-coupled receptors (Vanhaesebroeck et al., 1997 ; Ballou et al., 2003 ; Welch et al., 2003 ). The best studied of the class I PI3-K enzymes is a heterodimer containing the regulatory subunit p85 and the catalytic subunit p110 (Vanhaesebroeck et al., 1997 ). Activation can be facilitated by the binding of the SH2 domain of the p85 regulatory subunit to tyrosine-phosphorylated residues on receptors or associated molecules and recruitment of the p85/p110 dimer to the membrane (Vanhaesebroeck et al., 1997 ; Cantrell, 2001 ). Members of the Rho family (Rac and Cdc42) also can bind to p85 and contribute to PI3-K activation (Chan et al., 2002 ; Wu et al., 2007 ). In addition, GTP-bound Ras can bind to the p110 catalytic subunit (Rodriquez-Viciana et al., 1996 ; Vanhaesebroeck et al., 1997 ). PI3-K can then activate Rac by several mechanisms (Fleming et al., 2000 ; Innocenti et al., 2003 ; Welch et al., 2003 ). The finding that mGBP-2 is found within a protein complex containing the p110 subunit of PI3-K suggested that mGBP-2 might inhibit PI3-K activity as a consequence of this interaction (Figure 8). Consistent with this conclusion, S52N mGBP-2, which is not found within a complex containing p110, is unable to inhibit cell spreading (Figure 8). In fact, mGBP-2 inhibits PI3-K activation during cell spreading on FN (Figure 9). Inhibiting PI3-K by binding to the p110 subunit also has been observed for activated Gαq, which displaces Ras when it binds (Ballou et al., 2003 , 2006 ). Future studies will determine whether mGBP-2 binds directly to p110 and the significance of mGBP-2 binding to p110 to PI3-K activity.
Rac and PI3-K are both activated rapidly after binding of either type I or type II IFNs to their respective receptors (reviewed in Platanias, 2005 ). They are also important in signaling initiated by a wide variety of growth factors, cytokines, G protein-coupled receptors, and integrins. This study provides evidence of a novel mechanism for IFN-γ–mediated modulation of intracellular signals that impact how cells interact with their extracellular environment. IFN-γ exposure for times sufficient to induce the expression of mGBP-2 can inhibit subsequent Rac activation by cell spreading on FN or PDGF treatment. Further experiments will determine whether this reflects a more global method to inhibit signals from integrins and growth factors.
We thank Drs. Judith Drazba, Amit Vasanji, and Andrea Kalinoski for assistance with image analysis. We are grateful to Drs. Ulla Knaus and Maria Diakanova for helpful discussions about the PAK pull-downs and to Dr. Fan Dong for helpful comments on the manuscript. This work was supported by American Cancer Society grant RPG-CIM-88031 (to D.J.V.).
This article was published online ahead of print in MBoC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E09-04-0344) on May 26, 2010.