CD81 associates with Rac1 through its cytoplasmic C-terminal region

Searching for novel interactions of CD81 implicated in cell migration, we performed mass spectrometry using pull downs of T-lymphoblast cell lysates with biotinylated peptides corresponding to the C-terminal cytoplasmic region of CD81 as bait. The analysis identified two peptides belonging to the GTPase Rac1 and five corresponding to Rac2 (Figure 1A) in the bound fraction. The association of CD81 with Rac was also detected by pull-down assays with adherent cells (Figure 1B). Interestingly, the C-terminal cytoplasmic regions of the tetraspanins CD9, CD151, and CD82 did not associate with Rac (Figure 1B; unpublished data). Similarly, biotinylated peptides corresponding to the cytoplasmic sequence of other tetraspanin-associated membrane molecules (EWI-2, ICAM-1, VCAM-1) and other membrane receptors (CD147, CD69, CXCR4, CCR7) were unable to pull down Rac (Figure 1B; unpublished data). Direct interaction with the C-terminal sequence of CD81 was confirmed by direct binding of Rac1–glutathione S-transferase (GST) protein obtained in Escherichia coli cultures (Figure 1C). Interaction between the endogenous molecules was confirmed by coimmunoprecipitation in serum-starved, serum-induced, or epidermal growth factor (EGF)-stimulated primary human umbilical vein endothelial cells (HUVEC) or SUM159 breast carcinoma cells (Figure 1D).

CD81-Rac molecular complexes were detected in situ by total internal reflection microscopy (TIRFM)-based fluorescence image cross-correlation analysis of mCherry-CD81 and green fluorescent protein (GFP)-tagged wild-type Rac (WT-Rac1; Figure 2A). Correlation studies rely on the analysis of fluorescence intensity fluctuations from fluorescently tagged molecules in an image time series. The fluctuations, in this case, likely arise from diffusion and/or membrane binding–unbinding kinetics. The decay of the autocorrelation function for CD81 and wild type Rac (WT-Rac1) indicates that both molecules are producing fluorescence fluctuations over the timescale of the measurement (Figure 2B, top panels). This observation is typical for transmembrane receptors such as CD81, which diffuse in the cell membrane on the seconds timescale (faster than cytosolic proteins), or proteins with slow exchange with the membrane, and suggests that we are measuring a Rac population that is either diffusing in the membrane or exchanging with a membrane-bound complex (^{Moissoglu et al., 2006}).

As we have previously demonstrated, correlation analysis can also be applied to detect molecular complexes (Choi et al., 2011). If two molecules tagged with different fluorophores (e.g., GFP and mCherry) reside within the same molecular complex, they will produce similar temporal fluorescence intensity fluctuation patterns, which will be revealed by calculating the cross-correlation function (see Materials and Methods; ^{Wiseman et al., 2004}; ^{Brown et al., 2006}; ^{Digman et al., 2009}). In the absence of a complex, the intensity fluctuations are independent, and the cross-correlation function will be featureless and indistinguishable from the noise level. The cross-correlation function of CD81 and WT-Rac-1 therefore confirms that the membrane-targeted Rac1 resides with CD81 in the same complex. CD81 also showed high cross-correlation with V12-Rac, a constitutively active mutant of Rac (Figure 2B, middle panels). Coexpression of WT-Rac with a CD81 mutant lacking the cytoplasmic region (CD81-ΔCyt) resulted in reduced Rac autocorrelation, suggesting a faster exchange of the GTPase with the membrane. Moreover, both molecules showed no detectable cross-correlation, further indicating that Rac specifically binds to the CD81 C-terminal region (Figure 2B, bottom panels).

CD81 colocalizes with Rac1 and localizes anterior to nascent adhesions at the leading edge of migrating cells

We next studied the subcellular localization of CD81/Rac complexes in live cells. Stimulation of SUM159 cells or HUVEC with EGF induced partial colocalization of endogenous Rac and CD81 at the cell edge (Figure 3, A and B), and similar results were observed during cell spreading (unpublished data). Colocalization at the leading lamella was not observed for tetraspanin CD151, which concentrated at cell–cell contacts, or for the glycophosphatidylinositol (GPI)-linked receptor CD59 (Figure 3A). Analysis of similar samples using TIRFM, illuminating only ∼100 nm of the ventral surface of the cell, revealed the colocalization of CD81 and Rac1 at the edges of HUVEC cells (Figure 3C).

Loading of Rac with GTP is normally concomitant with its translocation to the plasma membrane. To determine whether association of CD81 with Rac depends on the activation state of the GTPase, we loaded SUM159 cell lysates with an excess of GDP or the nonhydrolyzable GTP analogue GTP-γS. CD81-biotinylated peptides pulled down similar amounts of Rac regardless of the GTP or GDP load, suggesting that CD81 association with Rac is independent of the Rac activation state (Figure 4A).

The dynamic behavior of the CD81/Rac complexes in migrating cells was analyzed with a Förster resonance energy transfer (FRET)-based Raichu-Rac plasmid that allows the dynamic visualization of Rac activity (^{Ouyang et al., 2008}). In human microvascular endothelial cells cotransfected with DsRed-CD81 and Raichu-Rac, EGF stimulation induced a clear colocalization of Rac activity with CD81 at the leading edges of the cells (Figure 4B and Supplemental Video S1). In resting cells, FRET signal was almost undetectable. To define more precisely the location of CD81-Rac complexes at the leading lamella, we cotransfected cells with AqGFP-CD81 and orange-paxillin and monitored them using time-lapse TIRFM. CD81 concentrated at the edges of the cells (Figure 4C, arrowhead, and Video S2), ahead of the nascent adhesions marked by paxillin (Figure 4C, arrow) suggesting that CD81 may prime these foremost areas for Rac activation.

CD81 controls Rac activation dynamics during cell adhesion and migration

The functional significance of the CD81/Rac interaction was explored in cells with reduced endogenous CD81 expression by transfection of two different siRNA sequences: one against the coding sequence and one against the 3′ untranslated region (3′UTR; see Material and Methods). After 48 h of siRNA transfection, down-regulation was assessed by flow cytometry and Western blotting (Figure 5, A and B); experiments were performed only when down-regulation was higher than 50%. We monitored the subcellular location of GFP-coupled Rac1 during migration on fibronectin in control cells and CD81-depleted cells (Figure 5C). Rac concentrated at the leading edge in both conditions, suggesting that CD81 is not required for Rac translocation to the plasma membrane (Figure 5C and Video S3).

Interestingly, active (GTP-bound) Rac levels, as detected by pull down with GST-PAK-CRIB, were significantly higher in unstimulated CD81-silenced cells (p < 0.05 in Student’s t test). Moreover, in CD81-silenced cells, Rac-GTP levels remained largely unaffected by EGF stimulation. Indeed, Rac activity remained high and almost constant, being also significantly higher at 30 min of EGF stimulation compared with control cells (p < 0.05 in Student’s t test). In contrast, no significant differences were observed in RhoA activity (detected with GST-C21), which was only slightly reduced in CD81-silenced cells (Figure 5D).

Cell protrusion during spreading depends mainly on Rac-induced actin poly­merization (^{Choi et al., 2008}). Consistent with the higher levels of activated Rac, CD81-deficient cells displayed a faster rate of protrusion and spreading than control cells (Figure 6, A and B). Importantly, this effect was rescued using full-length CD81 but not with the C-terminus truncated form (CD81-ΔCyt) fused to GFP. Moreover, overexpression of WT-CD81 had the opposite effect, slowing cell protrusion (Figure 6A).

The effects of CD81 depletion in actin polymerization and adhesion formation were visualized by staining F-actin and the adhesion protein paxillin in control and CD81-deficient cells. To normalize cell morphology, we used micropatterned substrates (^{Thery et al., 2006}) that permit averaging the staining of multiple cells with the same morphology, as well as analyzing the response to geometric cues (^{Tan et al., 2004}; Figure 6C). CD81-depleted cells showed increased paxillin staining of focal adhesions (Figure 6C). Interestingly, adhesions in control cells were restricted to the rim of the lamellar edge, but formed in a broader region in CD81-silenced cells. CD81 depletion also induced a defective cell polarization in both F-actin and focal adhesions in polarizing micropatterns. CD81 silencing in primary HUVECs plated onto fibronectin also increased the size of focal adhesions (Figure 6D), an effect reverted by cotransfection with full-length CD81, but not with CD81-ΔCyt-GFP. It has been reported that protrusion rate presents a direct correlation with adhesion assembly rate (^{Choi et al., 2008}). We analyzed adhesion assembly and disassembly rates in cells interfered for CD81, with or without reconstitution of the expression with a complete or truncated CD81 form. For adhesion assembly rate, results were consistent with protrusion rates: silencing of CD81 accelerated adhesion assembly compared with control cells, and this phenotype was rescued by complete CD81 but not by the truncated form (Figure 6E). In contrast, no statistically significant differences were found between any of the groups for the rate of adhesion disassembly, indicating CD81 is able to modify the turnover of focal adhesions, exerting an effect on the assembly of the adhesions, but not on the disassembly.

To directly corroborate that CD81 regulated Rac activation by their interaction through the C-terminal cytosolic domain of the tetraspanin, we incubated cells with a fluorescently labeled, cell-permeable peptide corresponding to the C-terminal sequence of CD81 or with a scrambled combination of the same eight amino acids as a control. These peptides were readily internalized; at 1 h after incubation, 100% of the cells were fluorescently labeled. The mean fluorescence intensity of the cells increased up to 3 h and declined thereafter, although 100% of the cells retained detectable fluorescence after 24 h (Figure 7A; unpublished data). Both peptides faintly labeled the plasma membrane and strongly accumulated at intracellular vesicles (Figure 7B). Incubation with the CD81 C-terminal peptide, but not with the control, increased the basal levels of Rac activity, but prevented its activation in response to EGF (Figure 7C), similar to what was observed when silencing CD81 in these cells. In the same manner, CD81 C-terminal peptides induced a higher protrusion rate (Figure 7D) and increased the cell area after spreading (Figure 7E).

Therefore our data suggest that perturbation of the Rac-CD81 interaction alters the normal Rac activation/inactivation cycle, resulting in increased but delocalized protrusion and increased adhesion formation. Time-lapse video microscopy experiments revealed that these alterations in Rac activity by CD81 depletion decreased the rate of cell migration, an effect avoided by cotransfection with full-length CD81 but not with the CD81ΔCyt mutant (Figure 8, A and B). Again, overexpression of WT-CD81 had the opposite effect, facilitating cell migration (Figure 8C), while treatment of cells with the permeable peptides recapitulated the phenotype of CD81 siRNA depletion, slowing cell motility (Figure 8D).

In sum, the binding of Rac to the C-terminal cytoplasmic domain of CD81 is an important mechanism responsible for tetraspanin regulation of cell migration. Furthermore, insertion into TEM emerges as a novel regulatory mechanism for Rac activity turnover.

Cells

HUVEC were obtained and cultured as described elsewhere (^{Barreiro et al., 2008}). SUM159 breast carcinoma cell line was cultured in DMEM/F-12 (Life Technologies, Invitrogen, Carlsbad, CA), supplemented with 5% fetal bovine serum (FBS), 1% penicillin/streptomycin, nonessential amino acids, 5 μg/ml insulin, and 1 μg/ml hydrocortisone. HEK and U2OS cells were cultured in DMEM (Lonza, Basel, Switzerland) or McCoy’s 5A medium (Life Techno­logies), respectively, both supplemented with 10% FBS and 1% penicillin/streptomycin. HMEC-1 microvascular endothelial cell line was grown in MCDB131 medium (Life Technologies) supplemented with 20% FBS, 1% penicillin/streptomycin, 20 mM HEPES, 2.5 μg/ml fungizone, 10 ng/ml EGF, and 1 μg/ml hydrocortisone.

siRNA

Control siRNA was purchased from Dharmacon (Thermo Scientific, Lafayette, CO). Two different siRNAs for CD81 were purchased from Eurogentec (Seraign, Belgium): CD81b (CACCTTCTATGTAGGCATC) and CD81c (CACGTCGCCTTCAACTGTA). CD81c sequence corresponds to a nontranslated (3′UTR) region of CD81 mRNA, which does not interfere with the expression of exogenous CD81. This sequence was used in rescue experiments.

Antibodies

VJ1/20 (anti-CD9), LIA1/1 (anti-CD151), VJ1/12 (anti-CD59), and HP2/9 (anti-CD44) monoclonal antibodies have been previously described (^{Yáñez-Mó et al., 1998}; ^{Barreiro et al., 2005}). Anti-CD81 5A6 was kindly provided by S. Levy (Stanford University, CA), and I.33.22 by R.Vilella (Hospital Clinic, Barcelona). Monoclonal anti-Rac1 (clone 23A8) was purchased from Upstate (now Millipore, Billerica, MA); rabbit policlonal anti-Rac antibody was from Santa Cruz Biotechonology (Santa Cruz, CA), anti-vimentin clone VIM 3B4 from Millipore (Billerica, MA); monoclonal antibody anti-paxillin, reference 03-6100, was from Zymed Laboratories (now Invitrogen, Carlsbad, CA), and clone 177 anti-paxillin was from BD Biosciences (Franklin Lanes, NJ).

Plasmids and reagents

The truncated CD81-ΔCyt-GFP was obtained using the specific primers: CTCGAGATGGGAGTGGAGGGCTGC (5′) and AAGCTTACAGCACAGCACCATGCT (3′) by PCR onto a CD81 cDNA plasmid. The PCR product was first introduced into the TopoTA vector (Invitrogen) and then subcloned into pAcGFP-N1 or mRFP-N1 (Invitrogen). GFP-wild-type-Rac1, GFP-V12-Rac1 and Orange-paxillin have been previously described (^{del Pozo et al., 1999}).

GST-Rac was provided by X.R. Bustelo (Centro de Investigación del Cancer, Salamanca, Spain); GST-PAK-Crib and C21-GST by J.G. Collard (The Netherlands Cancer Institute, Amsterdam, The Netherlands); and Enhanced Cyan Fluorescent Protein (ECFP)/Ypet-based Raichu-Rac by Y. Wang (University of Illinois, Urbana–Champaign, IL; ^{Ouyang et al., 2008}). pTRIP-CD81 (DsRed-CD81) and AqGFP-CD81 (^{Harris et al., 2008}) were provided by J. McKeating (University of Birmingham, UK) and mEmerald-GFP-CD81 (GFP-CD81) and mCherry-CD81 by M.W. Davidson (Florida State University, Tallahassee, FL). All CD81 plasmids had the fluorescent tag in the N-terminal region of the tetraspanin to avoid interference with C-terminal cytoplasmic region conformation or putative function. pEGFP-N1 was from Clontech (Mountain View, CA). EGF was purchased from R&D (Minneapolis, MN) and used at 100 ng/ml at the indicated times. Fibronectin, GTP-γS and GDP were from Sigma-Aldrich (St. Louis, MO).

Tetramethylrhodamine (TAMRA) N-terminal–labeled peptides with the sequences RRRRRRRCCGIRNSSVY (CD81) or RRRRRRRYSVNICRGCSS (Scrambled) were purchased from LifeTein (South Plainfield, NJ).

Pull-down assays

N-terminally biotinylated peptides containing a SGSG linker sequence connected to the cytoplasmic C-terminal domains of the proteins of interest were purchased from Ray Biotech, (Norcross, GA): CD81, biotin-SGSG-CCGIRNSSVY; CD9, biotin-SGSG-CCAIRRNREMV; CD151, biotin-SGSG-YRSLKLEHY; CD82, biotin-SGSG-CRHVHSEDYSKVRKY; EWI-2, biotin-SGSG-CCFMKRLRKR; ICAM-1, biotin-SGSG-RQRKIKKYRLQQAQKGTPMKPTQATPP. Each peptide (30 nmol) was conjugated to 40 μl streptavidin-Sepharose (GE Healthcare, Uppsala, Sweden). Pull-down assays with T-lymphoblasts or HEK extracts were carried out as previously described (^{Sala-Valdes et al., 2006}) and analyzed by Western blotting or mass spectrometry. Briefly, cells were washed once with ice-cold phosphate-buffered saline (PBS) and then lysed in 1% NP-40 in PBS containing protease and phosphatase inhibitors (Complete, PhosSTOP; Roche, Basel, Switzerland). Lysates were precleared for 2 h at 4ºC with streptavidin-Sepharose (GE Healthcare) and then incubated for 2 h at 4ºC with biotinylated peptides immobilized on streptavidin-Sepharose beads.

For preferential loading of Rac with GDP or GTP, cells were lysed in 5 mM EDTA-containing lysis buffer, and the lysates were incubated with 50 nM of GTP-γS or GDP at 30ºC for 5 min before addition of 6 mM MgCl. Alternatively, cells treated, or not, with EGF were lysed in the presence of 2 mM MgCl before incubation with the Sepharose beads coated with the biotinylated peptides, PAK-GST or C21-GST.

Mass spectrometry

Sepharose beads from the pull-down assays were directly resuspended in Laemmli buffer, applied onto a SDS–PAGE gel, and subjected to the one-step in-gel trypsin digestion method (^{Bonzon-Kulichenko et al., 2011}). The resulting peptides were analyzed by liquid chromatography–tandem mass spectrometry (LC-MS/MS) using a Surveyor LC system coupled to an LTQ linear ion-trap mass spectrometer (Thermo Fisher, San Jose, CA), as previously described (^{Lopez-Ferrer et al., 2004}; ^{Jorge et al., 2009}). The LTQ was operated in a data-dependent MS/MS mode using the 15 most intense precursors detected in a survey scan from 400 to 1600 m/z, as previously described (^{Lopez-Ferrer et al., 2004}). Proteins were identified using the SEQUEST algorithm (Bioworks 3.2 package; Thermo Finnigan, Cambridge, MA); the raw MS/MS files were searched against the Human Swissprot database (Uniprot release 14.0, 19929 sequence entries for human) supplemented with the sequence of porcine trypsin. SEQUEST results were validated using the probability ratio method (^{Martinez-Bartolome et al., 2008}), and false discovery rates (FDR) were calculated by the refined method (^{Navarro and Vázquez, 2009}). Peptide and scan counting was performed, assuming as positive events those with a FDR equal to or lower than 5%.

Immunoprecipitation assays

Cells were plated onto 2 μg/ml of fibronectin and serum-starved for at least 6 h when indicated. EGF (100 ng/ml) was applied at the indicated times before lysis in 0.5% Triton-X-100, 60 mM octylglucoside (Sigma-Aldrich) in Tris 10 mM (pH 8), 150 mM NaCl supplemented with protease and phosphatase inhibitors. Antibodies were precoupled to protein G–Sepharose (GE Healthcare), incubated with cell lysates, and washed several times with lysis buffer before being loaded onto an SDS–PAGE gel.

Fluorescence microscopy

For immunofluorescence, cells were fixed with 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA), permeabilized with 0.5% Triton X-100 (5 min), and stained with corresponding primary antibodies followed by species-matching secondary antibodies (Invitrogen). Samples were analyzed in a Leica TCS-SP5 confocal laser-scanning unit equipped with Ar and He/Ne laser beams and attached to a Leica DMIRBE inverted epifluorescence microscope (Leica Microsystems, Heidelberg, Germany); in a TIRF microscope (Leica Microsystems, on a Leica DMI 6000B) using ∼100-nm depth penetration; or a wide-field fluorescence Leica DMIRE2 microscope coupled to a monochromator (Polychrome IV; Till Photonics, Munich, Germany) and a CCD camera (CoolSNAP HQ; Photometrics, Tucson, AZ). For time-lapse fluorescence microscopy, cells were maintained at 37ºC in 5% CO₂.

For quantification of focal adhesions, subconfluent HUVEC cultures were plated onto 2 μg/ml fibronectin 72 h after transfection, fixed, permeabilized, and stained for paxillin. The area of focal adhesions was selected and measured by Image J using a thresholding method.

Transfection experiments

For DNA or siRNA transfection, cells were electroporated with 2 μM of either negative siRNA or CD81 siRNA duplexes or 20 μg of plasmid DNA, in a Gene Pulser II device (Bio-Rad, Hercules, CA). Conditions used were 200 V, 0.975 × 10⁻⁹ Faradays, in 200 μl of standard medium (5 μl of 1.5 M NaCl had been previously supplemented). For rescue experiments, the siRNA duplexes were electroporated together with 20 μg of GFP, GFP-CD81, or CD81-ΔCyt-GFP.

Flow cytometry

CD81 membrane expression was routinely analyzed 48 or 72 h after transfection with siRNA duplexes by immunostaining with anti-CD81 mAb in a FACScan Canto II cytofluorometer (BD Biosciences, Franklin Lanes, NJ), as previously described (^{Yáñez-Mó et al., 2008}). Silencing of CD81 resulted in a reduction of mean fluorescence intensity consistently greater than 50%. In rescue experiments, membrane expression of CD81, stained with an APC-labeled goat anti-mouse, was analyzed after gating GFP-positive cells.

Incorporation of cell-permeable, fluorescently TAMRA-labeled peptides was analyzed in trypsinized unstained cells by flow cytometry.

Fluorescence image cross-correlation analysis

U2OS cells were cotransfected with mCherry-CD81 (wild-type or cytoplasmic deletion mutant) and GFP-Rac1 (wild-type or V12 active mutant). TIRF images were acquired on an Olympus IX71 microscope using a 60×/1.45 numerical aperture (NA) Plan-Apo oil objective and a Ludl controller (Ludl Electronic Products, Hawthorne, NY), and controlled by MetaMorph software (Molecular Devices, Downingtown, PA). GFP and mCherry were excited using a 488-nm line of an argon ion laser (Melles Griot, Albuquerque, NM) and a 561-nm Cobolt Jive ion laser (Market Tech, Scotts Valley, CA), respectively. For simultaneous dual-color imaging, a polychroic mirror (Z488/568 rpc) and dual-emission filter (Z488/568 nm; Chroma Technology, Bellows Falls, VT) were used. Image time series were acquired with a charge-coupled device camera (Retiga Exi; QImaging, Surrey, Canada). Image processing and correlation analysis was performed in MATLAB 7.7.0 (MathWorks, Natick, MA). Images were corrected for background and photobleaching intensity effects. Cell regions (25 × 25 – 64 × 64 pixels) from both channels were selected for analysis. Single-channel auto- (a = b = 1 or 2) and cross-correlation (a = 1; b = 2) functions were calculated following ^{Wiseman et al. (2000}):

where the fluctuation in the fluorescence intensity is defined as i_a is the fluorescence intensity at a given pixel (x,y) at time t. τ is the temporal lag time. Angular brackets denote spatial averaging over the analyzed cell region. The correlation function is normalized by , the average intensity of the selected region at time t.

Activated Rac measurements

GTP-Rac loading was measured by pulling down GTP-loaded Rac using a GST-coupled PAK-CRIB construct immobilized on Sepharose beads, as described elsewhere (^{Sander et al., 1999}).

Alternatively, endothelial cells were cotransfected with ECFP/Ypet Raichu-Rac and DsRed-CD81 and plated onto 2 μg/ml fibronectin. Cells either untreated or stimulated with 100 ng/ml EGF were imaged using time-lapse confocal microscopy. FRET efficiency is presented as the yellow fluorescent protein (YFP)/cyan fluorescent protein (CFP) ratio after CFP stimulation, as previously described (^{Ouyang et al., 2008}).

Protrusion rate and adhesion turnover measurements

SUM159 cells were either transfected with GFP or CD81-GFP, cotransfected with control or CD81 siRNA and GFP, or transfected with GFP and incubated 2 h with 1 μM permeable CD81 or scrambled peptides. Cells were then trypsinized, washed, and plated onto 35-mm dishes (MatTek) coated with 10 μg/ml of fibronectin. Images were acquired for 10 min with 1 frame every 5 s using a Leica AM TIRF MC mounted on a Leica DMI6000B microscope fitted with a 100×/1.46 NA oil-immersion objective with 1× magnification and ∼90-nm penetration depth. The protrusion rate was measured with Image J’s Multiple Kymograph plug-in, as described elsewhere (^{Choi et al., 2008}).

For the study of focal adhesion turnover, cells cotransfected with Orange-paxillin, together with control or CD81siRNA and GFP, GFP-CD81, or CD81-ΔCyt-GFP, were allowed to spread, and adhesion assembly and disassembly rates were analyzed with Image J, as previously described (^{Webb et al., 2004}; ^{Choi et al., 2008}).

Spreading assays on micropatterned substrates

Micropatterned slides were purchased from CYTOO (Grenoble, France). Cells were trypsinized and replated onto micropatterned slides (CYTOO Starter), according to the manufacturer’s instructions. Cells were allowed to attach and spread for 3 h and were then fixed with paraformaldehyde, permeabilized, and stained for paxillin or F-actin. Images were acquired using wide-field fluorescence microscopy and analyzed with the Image J Reference-cell macro. Briefly, images were cropped following the micropatterned image and micropatterns containing multiple cells were discarded by nuclei counting. Average projections of the label image of several individual cells (paxillin or F-actin) are presented in a pseudocolor intensity-scale image.

Migration experiments

SUM159 cells were plated on fibronectin (2 μg/ml) 2 d after transfection and serum-starved for 12 h. Alternatively, SUM159 were plated on fibronectin (2 μg/ml) and incubated for 1 h with 1 μM permeable CD81 or scrambled peptides. Images were acquired at 10-min intervals for 24 h in a Nikon Eclipse Ti-E video microscope (Tokyo, Japan). Tracking of transfected cells was performed with Image J and MetaMorph software.

Statistical analysis

Statistical significance was calculated using Student’s t test or one-way analysis of variance (ANOVA), and significant differences were labeled as: *, p < 0.05; **, p < 0.01; and ***, p < 0.001.