Regulation of intercellular adhesion strength in fibroblasts.

The regulation of adherens junction formation in cells of mesenchymal lineage is of critical importance in tumorigenesis but is poorly characterized. As actin filaments are crucial components of adherens junction assembly, we studied the role of gelsolin, a calcium-dependent, actin severing protein, in the formation of N-cadherin-mediated intercellular adhesions. With a homotypic, donor-acceptor cell model and plates or beads coated with recombinant N-cadherin-Fc chimeric protein, we found that gelsolin spatially co-localizes to, and is transiently associated with, cadherin adhesion complexes. Fibroblasts from gelsolin-null mice exhibited marked reductions in kinetics and strengthening of N-cadherin-dependent junctions when compared with wild-type cells. Experiments with lanthanum chloride (250 microm) showed that adhesion strength was dependent on entry of calcium ions subsequent to N-cadherin ligation. Cadherin-associated gelsolin severing activity was required for localized actin assembly as determined by rhodamine actin monomer incorporation onto actin barbed ends at intercellular adhesion sites. Scanning electron microscopy showed that gelsolin was an important determinant of actin filament architecture of adherens junctions at nascent N-cadherin-mediated contacts. These data indicate that increased actin barbed end generation by the severing activity of gelsolin associated with N-cadherin regulates intercellular adhesion strength.

Cadherin-mediated adherens junctions are critically involved in tumorigenesis and metastasis as well as in the maintenance of mature tissue architecture, formation of distinct tissue boundaries, tissue differentiation, and cell sorting events such as epithelial-mesenchymal transitions during embryogenesis (1)(2)(3)(4)(5). The classical cadherins are a group of singlepass, transmembrane, calcium-dependent glycoproteins that mediate homotypic intercellular adhesions termed adherens junctions (5). Adherens junctions are linked to the actin cytoskeleton, and their formation or dissolution is tightly regulated. N-cadherin (adherens junction-specific cell adhesion molecule) is a member of the classical cadherin family, and its expression and function are necessary for maintenance of embryonic vitality (6) and for establishment of embryonic asymmetry (7). Further, N-cadherin is the predominantly expressed cadherin of mesenchymal tissues and is known to play an important role in mediating tissue organization and cell differentiation in muscle (8 -10), cartilage (11), bone (12,13), and neural tissues (14,15). However, the regulation of N-cadherinmediated intercellular adhesions, particularly in connective tissue fibroblasts, is poorly characterized.
N-cadherins are tethered to the cortical actin cytoskeleton by ␣-catenin, a member of the armadillo family of proteins, which binds indirectly to the cytoplasmic tail of cadherins via ␤-catenin (16,17). Tethering of cadherins to cortical actin filaments is required for cadherin-mediated adhesion and adhesion strengthening (18 -21). Recently, cadherins have been shown to function as adhesion-activated cell surface receptors (reviewed in Ref. 22). Ligation of cadherins on opposing cell surfaces generates signals that induce recruitment to the adherens junctions of several actin-binding proteins that mediate localized remodeling of the actin cytoskeleton (reviewed in Ref. 23). Notably, vasodilator-stimulated phosphoprotein, zyxin, mena, Arp2/3, and cortactin 1 are localized to nascent cadherin-mediated intercellular adhesions, further underlining the importance of actin filament networks in cadherin function (24). After initial intercellular contact and cadherin ligation, the dramatic reorganization of the actin cytoskeleton at adherens junctions is likely mediated by actin-severing proteins, proteins that can generate new barbed ends and promote rapid assembly of new actin filaments (25,26). However, this has not been shown. Notably, actin-severing proteins such as gelsolin are activated by localized increases of calcium (27,28), which suggests a possible regulatory mechanism for intercellular adhesion.
The precise mechanism and regulatory molecules involved in cadherin-mediated adhesions are currently unclear, but gelsolin appears to be an important candidate. Notably, gelsolin is expressed at high levels in connective tissue fibroblasts (29) and is the only known calcium-dependent severing protein with a Ca 2ϩ K d in the micromolar range (30). Intracellular calcium transients localize to areas of nascent cadherin-mediated contacts and can regulate remodeling of cortical actin filaments (24,31). We used the donor-acceptor intercellular adhesion model (32), Ncad-Fc 2 recombinant chimeric protein, and gelsolin-null fibroblasts to study the role of gelsolin in mediating the formation of intercellular contacts. Our results indicate that gelsolin transiently associates with nascent N-cadherin-mediated adherens junctions, and there, in a calcium-dependent manner, remodels cortical actin filaments. This novel association may serve to explain the observations related to the importance of calcium in cadherin-mediated intercellular adhesions noted previously (31).
Donor-Acceptor Model and Wash-off Assay-The donor-acceptor model was used to analyze nascent cadherin-mediated intercellular adhesions in fibroblasts as described previously (31,32). Briefly, for quantification, donor and acceptor cells were incubated overnight in growth medium containing spectrally discrete 1 mg/ml of dextranconjugated fluorochromes (Sigma). Following designated treatments, donor cell suspensions were prepared with 0.01% trypsin supplemented with CaCl 2 (2 mM) and seeded onto acceptor monolayers at ratios of 1:1 unless otherwise indicated. Cells were incubated for the indicated times before jet-washing in a logarithmic series to estimate the strength of intercellular adhesions (34). Attached donor cells were fixed with paraformaldehyde and quantified in three randomly chosen ϫ40 fields with an inverted fluorescent microscope.
Immunofluorescence, Confocal, and Video Microscopy-Acceptor cells as non-confluent monolayers were overlaid with donor cells to establish nascent adherens junctions and incubated for discrete time periods. Non-confluent acceptor monolayers were used in these analyses to permit contrasts between nascent donor-acceptor intercellular adhesions and donor-substratum adhesions. Cells were fixed for 10 min with 2% paraformaldehyde, 5% sucrose solution in phosphate-buffered saline, permeabilized for 5 min in a 0.02% Triton-X solution in phosphate-buffered saline, and stained with monoclonal N-cadherin (GC-4, Sigma), TRITC-conjugated ␤-catenin (14: BD Transduction Laboratories), fluorescein isothiocyanate-conjugated pan-cadherin (CH-19, Sigma), or rabbit polyclonal gelsolin antibodies (kind gift of D. J. Kwiatkowski) followed by Cy3 or fluorescein isothiocyanate-tagged secondary antibodies (Sigma). Immunofluorescence was visualized by confocal microscopy (Leica, Heidelberg, Germany; ϫ40 oil immersion lens) using 1-m transverse optical sections. For fluorescein isothiocyanate labeling, excitation was set at 488 nm, and emission was collected with a 530-/20-nm barrier filter. For TRITC or Cy3, excitation was set at 530 nm, and emission was collected at 620/40 nm.
Transfection-Transient transfections were performed with Fu-GENE 6 transfection reagent (Roche Applied Science). Cells at 50% confluence were incubated with 3 l of FuGENE6 reagent and 1 g of pEGFP-gelsolin in 100 l of serum-free medium at 24°C for 30 min and assayed 48 h after transfection.
Flow Cytometry-EGFP-gelsolin-transfected GsnϪ/Ϫ cells were harvested with 0.01% trypsin supplemented with 2 mM CaCl 2 . Transfected cells were sorted from untransfected cells (FACSTAR Plus; BD Biosciences) with excitation at 488 nm. Sorted cells were washed three times with phosphate-buffered saline and electronically counted prior to kinetic and strengthening experiments.
For characterization of cadherin expression, whole cell lysates were prepared with 2% SDS Laemmli sample buffer. Protein concentrations of samples were standardized using the RC DC Bio-Rad protein assay (Bio-Rad Laboratories), and equivalent amounts of proteins were analyzed by Western blotting. Membranes were probed with N-cadherin antibody (GC-4; Sigma) or with anti-P-cadherin antibody (BD Transduction Laboratories) or with anti-E-cadherin antibody (G-10: Santa Cruz Biotechnology, Santa Cruz, CA). ␤-Actin (AC-15: Sigma) antibody was used to co-blot.
Preparation of N-cadherin-Fc Dishes-The Ncad-Fc (chicken N-cadherin ectodomain fused to the Fc fragment of mouse IgG2b) protein was expressed in HEK-293 cells and collected as described (35). Ncad-Fccoated microbiological plastic dishes were coated with protein G-purified Ncad-Fc protein reconstituted in sodium bicarbonate buffer in yields of at least 100 g/ml was adsorbed onto plates following overnight incubation at 4°C and used immediately. Protein adsorption was quantified by dot blot and was estimated at 1.25 g/cm 2 based on densitometric comparison with purified mouse IgG Fc fragment controls (Jackson Laboratories, West Grove, PA).
Severing Assay-The ability of gelsolin to sever actin filaments was measured as described (36). Briefly, rhodamine phalloidin (1 M; Molecular Probes, Eugene, OR) was added to actin filaments (0.4 M), and the rate of fluorescence loss at 570 nm was measured by fluorimetry. Reduction of fluorescence is caused by gelsolin severing of actin and displacement of phalloidin after adding CaCl 2 (1 mM). Affinity-purified rabbit muscle actin (1.0 mg/ml; Cytoskeleton, Denver, CO) was resuspended in polymerization buffer (50 mM KCl, 2 mM MgCl 2 , 0.5 mM ATP, 2 mM Tris, pH 8.0) and sedimented with an Airfuge (Beckman; 30,000 rpm for 20 min) to remove unpolymerized actin. Cell lysates from wild-type and GsnϪ/Ϫ cells attached onto non-tissue culture plates coated with Ncad-Fc protein were prepared with detergent plus protease inhibitors in buffer containing 50 mM KCl, 2 mM MgCl 2 , 0.5 mM ATP, 2 mM Tris, pH 8.0, 1 mM EGTA, and 1% Triton X-100. The lysates were dialyzed with several changes of buffer containing 2 mM MgCl 2 , 50 mM KCl, 2 mM Tris-HCl, and 1 mM EGTA, 0.5 mM ␤-mercaptoethanol. The volume of the dialyzed cell lysate was adjusted to 400 l in dialysis buffer. Labeled F-actin in polymerizing buffer (200 l; 50 mM KCl, 2 mM MgCl 2 ) was added to a final concentration of 400 nM. Severing assays were performed in calcium (2 mM CaCl 2 , 1 mM EGTA).
Actin Assembly-Gelsolin wild-type or gelsolin-null cells were allowed to attach to Ncad-Fc-coated non-tissue culture plates for specific incubation times. In permeabilized cells incubated with rhodamine actin monomers, increases of rhodamine fluorescence due to incorporation into nascent actin filaments were measured (37)(38)(39). Cells were permeabilized for 20 s using 0.1 volume of octyl glucoside buffer (PHEM buffer (60 mM PIPES, 24 mM HEPES, 5 mM EGTA, 1 mM MgSO 4 , pH 6.9) containing 2% octyl glucoside and 2 M phalloidin). Permeabilization was stopped by diluting the detergent with buffer. Immediately thereafter, freshly sedimented rhodamine actin monomer (0.23 M) in buffer containing 120 mM KCl, 2 mM MgCl 2 , 3 mM EGTA, 10 mM PIPES, and 0.1 mM ATP was added to the samples for 10 s followed by fixation with 3.7% formaldehyde. The samples were observed with a Nikon TE 300 microscope, and rhodamine fluorescence in single cells was quantified using the PCI imaging program. For background correction, detergent treatments were omitted, fluorescence was quantified, and background signal were subtracted from experimental samples.
Ca 2ϩ Fluxes-Donor cells were loaded with fluo-4/acetoxymethyl ester (3 M) according to the manufacturer's instructions (Molecular Probes) and plated on acceptor cells or onto Ncad-Fc-coated microbiological plastic dishes. Peripheral membrane Ca 2ϩ influx was measured in donor cells immediately after attachment and visualized by z axis optical sectioning by confocal microscopy. For wash-off assays, embryonic fibroblasts were preincubated with lanthanum chloride (250 M), and donor-acceptor cultures were established for 15 min in growth medium containing the same concentration of the calcium channel blocker. For experiments to detect near plasma membrane calcium transients, a lipophilic calcium ion indicator fura-C18 was loaded into substratum-bound cells according to the manufacturer's instructions, and Ncad-Fc-coated or bare beads were incubated onto cells. Plasma membrane calcium ion measurements were conducted as described previously (27).
Magnetic Bead Pull-off Assays-Proteins enriched at sites of Ncadherin ligation through recombinant protein-coated bead-associated adhesion complexes were prepared. Briefly, after designated incubation times, cells and attached N-cadherin-coated magnetic beads (Sphero-tech, Libertyville, Il) were collected by scraping into ice-cold extraction buffer (cytoskeleton extraction buffer). Beads were pelleted using a side-pull magnetic isolation apparatus (Dynal, Lake Placid, NY), and supernatants were collected. Isolated beads were resuspended, sonicated, homogenized, and washed three times in CSKB prior to gel fractionation and Western blot analysis.
Electron Microscopy of Cytoskeletons and Quantification of Images-Gelsolin-null or gelsolin wild-type cells were allowed to attach onto Ncad-Fc-coated glass coverslips for 3 min prior to detergent extraction, fixation, and processing as described elsewhere (40). Briefly, cells were extracted for 5 min with 1% Triton X-100, 4% polyethylene glycol in PEM buffer (100 mM PIPES, pH 6.9; 1 mM MgCl 2 ), 1 mM EGTA supplemented 10 M phalloidin, washed three times in PEM buffer, and fixed in 2% glutaraldehyde (electron microscopy grade) in 0.1 M sodium cacodylate, pH 7.3, for 20 min at room temperature and overnight at 4°C. Samples were subsequently fixed in 0.1% aqueous tannic acid and uranyl acid solutions, respectively, 20 min prior to dehydration and critical point drying. Samples were gold-coated using a Polaron sputter coater with a rotary planetary stage. Samples were visualized and digital images were acquired with a Hitachi S-570 scanning electron microscope. Filament lengths and branching frequency were quantified using inclusion criteria as described elsewhere (41). Briefly, filament lengths were quantified in a 1.76 by 1.76 m image using the Simple PCI software (Compix Inc., Imaging Systems, Cranberry Township, PA). Filaments longer than the field of view were excluded. Included filaments were traced from the cell edge until they emerged from a mother filament or were lost in the network. Branching frequency of stereotypical 70°Y-branches was measured from an 880 by 880 nm image at the cell edge. Normalized measurements of number of branches/m of actin filament length were obtained to correct for differences in filament density. The number of branches per image was divided by the total length of all visible filaments.
Statistical Analyses-For continuous variables, means and S.E. were computed. Comparisons between two groups were evaluated by Student's t test, and for multiple samples, analysis of variance was used with statistical significance set at p Ͻ 0.05.

Cadherin Expression and Importance of Polymerized Actin in
Intercellular Adhesions-We assessed the expression of classic cadherin family members in embryonic mouse fibroblasts and Rat-2 fibroblasts by immunoblotting whole cell lysates with antibodies specific for extracellular epitopes of P-, E-, and Ncadherins. Only N-cadherin was detected in these cells (Fig.  1A, wild-type embryonic fibroblasts shown). Control cell lysates were used to verify specificity of antibodies used for cadherin expression profiling (results not shown). Immunoprecipitations for N-cadherin in wild-type and null fibroblasts showed that FIG. 2. Gelsolin co-localizes with Ncadherin adhesion complex proteins and actin filaments at nascent contacts. Optical sections at the donor-acceptor interface of stained donor Rat-2 cells were incubated onto non-confluent acceptor monolayer for 15 min; images were acquired by confocal immunofluorescence. A, gelsolin is enriched at nascent donor-acceptor interfaces and co-localizes with ␤-catenin (yellow in merged overlay). Donor-substratum interfaces show weaker gelsolin staining and lack of ␤-catenin staining (green in merged overlay). The outline of the underlying acceptor cell is indicated by a gray outline. A differential interference contrast (DIC) image is provided to reveal the location of the acceptor cell. B, gelsolin co-localizes with distinct F-actin staining at sites of nascent donor-acceptor interfaces (open arrow; yellow in merged overlay). The lower optical section of F-actin staining is provided to reveal the location of the acceptor cell and spatial interaction between donor and acceptor cell. The bars indicate 20 m. ␣-catenin and ␤-catenin association with the cytoplasmic domain of N-cadherin was unaffected by the presence or absence of gelsolin (Fig. 1B).
An intact cortical actin cytoskeleton is known to be important for the formation and maintenance of cadherin-mediated intercellular adhesions (18 -21). We verified the functional importance of actin filament assembly and organization in cadherin-mediated adhesion strengthening in mouse embryonic fibroblasts using the donor acceptor model shear wash-off assays using latrunculin B (1 M). As expected, we found that in a logarithmic series of jet washes, latrunculin B-treated samples exhibited significantly reduced levels of donor cell adhesive strength when compared with vehicle control samples throughout the wash-off series (p Ͻ 0.01; results not shown).
Gelsolin Co-localizes with N-cadherin Adhesion Complexes-As remodeling of cortical actin filaments by actin-binding proteins may be important for the formation of adherens junctions, we determined whether gelsolin was localized to intercellular adhesions. Immunofluorescence analysis of nascent, N-cadherin-dependent intercellular adhesions of Rat-2 fibroblasts was conducted using the donor-acceptor model, a systemthatgenerateslargenumbersofsynchronizedN-cadherindependent intercellular adhesions (31,32). There was significant enrichment of gelsolin at donor-acceptor interfaces that co-localized with ␤-catenin ( Fig. 2A, merged images). Incompletely established acceptor monolayers were used to contrast donor-acceptor and donor-substratum interfaces as demonstrated in the differential interference contrast mage. Although ␤-catenin staining was limited to donor-acceptor interfaces, gelsolin was also found peripherally at donor-substratum interfaces, albeit with greatly reduced staining intensity ( Fig. 2A,  merge image). Indeed, when donor cells were incubated on completely confluent monolayers of acceptor cells, there was a peripheral ring of gelsolin staining corresponding to sites of donor-acceptor cell interfaces (Fig. 3B). Gelsolin also co-localized with cortical actin filaments at donor-acceptor interfaces and at nascent intercellular contacts (Fig. 2B).
Gelsolin Transiently Associates with Nascent Intercellular Contacts-Immunoprecipitations were conducted using antibodies directed against a cytoplasmic cadherin epitope to evaluate whether gelsolin is recruited and physically associates with N-cadherin-mediated intercellular adhesions. Donor-acceptor samples containing equivalent amounts of protein showed that gelsolin was recruited to, and transiently associated with, the N-cadherin adhesion complex of proteins (Fig.  3A). The cadherin-gelsolin physical interaction was detected most prominently during the first 30 min of intercellular adhesion and subsequently decreased to baseline levels (Fig. 3A, time 0 is acceptor monolayer without donor cells).
To evaluate the gelsolin-cadherin interaction spatially over time, donor cells were incubated over completely confluent acceptor monolayers and subsequently fixed and stained at distinct time points. Merged images demonstrate that the greatest degree of co-localization between the cadherin and gelsolin staining occurred during 15 and 30 min with a subsequent decrease as the timeline progressed (Fig. 3B). Collectively, these data suggest that gelsolin is actively recruited to sites of early intercellular contact and progressively disassociates as junctions mature.
Gelsolin Regulates N-cadherin Adhesion Kinetics and Strengthening-Donor-acceptor cultures were established using wild-type fibroblasts, gelsolin-null fibroblasts, or gelsolinnull fibroblasts reconstituted with EGFP-gelsolin by transient transfection (gelsolin rescue) to study the effects of gelsolin on the kinetics of N-cadherin-mediated intercellular adhesion. Gelsolin-null cells showed markedly reduced N-cadherin-de-pendent intercellular adhesion kinetics when compared with gelsolin wild-type cells (Fig. 4A). The greatest kinetic differences between the wild-type and null cell types were detected between 5 and 30 min with very minor increases in total numbers of intercellular adhesions subsequent to that, suggesting an important role of gelsolin in the initial formation of intercellular adhesions. After transient transfection of null donor cells with a full-length EGFP-gelsolin construct, the adhesion kinetics were restored to that of wild-type cells (Fig. 4A,  rescued).
We next investigated the strength of intercellular binding using a logarithmic wash-off assay performed on donor-acceptor cultures incubated for 15 min to evaluate the possible role

FIG. 4. Gelsolin severing activity is required for N-cadherin adhesion and strengthening.
A, temporal increase of intercellular adhesion between murine wild-type (ϩ/ϩ), gelsolinnull (Ϫ/Ϫ), and gelsolin-null reconstituted with gelsolin (gelsolin rescue) fibroblasts in the donor-acceptor model. By 30 min, gelsolin-null cells exhibit 2-fold less intercellular attachment (p Ͻ 0.01). This effect was reversed after gelsolin reconstitution. B, wash-off assay was conducted to evaluate the strength of nascent intercellular adhesions in the donor-acceptor model. Cells were incubated for 15 min and were jet-washed in a logarithmic series. There is ϳ50% reduced binding in gelsolin-null versus wild-type fibroblasts at all time points (p Ͻ 0.01). After gelsolin reconstitution in gelsolinnull fibroblasts, the level of intercellular adhesion was equivalent to that of wildtype cells. The means Ϯ S.E. are shown for three separate samples. C, gelsolin has no effect on N-cadherin dependent strengthening by the wash-off assay in cells incubated for 180 min time period. of gelsolin in nascent intercellular adhesion strengthening. The estimated strength of attachment of gelsolin-null fibroblasts was considerably lower than that of wild-type cells (Fig. 4B). Similar to the kinetics assay, gelsolin-null fibroblasts that were transiently transfected with gelsolin exhibited restoration of intercellular binding strength to that of wild-type cells.
To evaluate the potential role of gelsolin in the maintenance of established intercellular adhesions, the strength of intercellular binding of donor-acceptor cultures incubated for 180 min was investigated using the shear wash-off assay. No differences of intercellular adhesion strength were detected between wildtype and null cell types throughout the wash-off series (Fig.  3C). These data suggest that the role of gelsolin is limited to the strengthening of nascent intercellular contacts with no effect on the maintenance of mature contacts.
External Calcium Influx and Gelsolin Activation-Gelsolin severing and capping activity are critically dependent on calcium ion concentration (42)(43)(44). We examined calcium transients subsequent to N-cadherin ligation using the donor-acceptor model (31). Donor gelsolin wild-type cells were loaded with fluo-4/AM (3 M) harvested and imaged in real time by confocal microscopy as the donor cells attached to underlying acceptor cells. There was a localized increase of calcium at sites of donor-acceptor adhesion directly opposed to the donor cell plasma membrane (Fig. 5A, i). We verified that this calcium response was due specifically to N-cadherin ligation as opposed to other, intercellular adhesion-associated, surface-expressed proteins. Donor cells were loaded with fluo-4/AM and were allowed to attach onto non-tissue culture plates coated with recombinant Ncad-Fc chimeric protein. Similar to the results found with the donor-acceptor model, distinct submembranous calcium signals were found after fibroblast attachment (Fig.  5A, iii and v) when compared with unattached cells (Fig. 5A, ii and iv). Cells did not attach to control plates coated with mouse IgG-Fc fragments.
To further validate the submembranous or near plasma membrane localization of calcium transients at sites of Ncadherin ligation, Fura-C18 dye-loaded substratum-bound cells were observed following N-cadherin-Fc and control bare bead binding. N-cadherin-Fc-coated beads elicited a sharp rise in near plasma membrane [Ca 2ϩ ] at sites of bead to cell binding when compared with bare controls (Fig. 5B).
As the apparent submembranous localization of the calcium signal suggested an influx of calcium through plasma membrane channels as described previously (45), we evaluated the effect of lanthanum chloride (a blocker of external calcium channels) on gelsolin wild-type and gelsolin-null fibroblast nascent intercellular adhesion strengthening using the donor-acceptor model (15-min incubation). Lanthanum chloride-treated wild-type donor-acceptor samples demonstrated significantly reduced adhesion strength when compared with vehicle controls and with lanthanum-treated and control gelsolin-null cells. Notably, there was no significant difference of adhesion between lanthanum-treated and untreated gelsolin-null cells (Fig. 5C).
Gelsolin Regulates Cadherin-localized Actin Filament Assembly-To determine whether cadherin-mediated adhesion kinetics and strengthening were related to the severing activity of gelsolin, we conducted actin severing assays (36). Gelsolin wild-type and null fibroblasts were plated on Ncad-Fc-coated, non-tissue culture dishes for 30 min as our immunoprecipitation data showed that this time period was coincident with the largest amount of cadherin associated with gelsolin. Cell lysates were prepared, and the severing activity from the wildtype fibroblasts (slope ϭ Ϫ35.4; R 2 ϭ 0.817; n ϭ 3; p Ͻ 0.01) was significantly higher than that of the gelsolin-null cells (slope ϭ Ϫ11.5; R 2 ϭ 0.734; n ϭ 3; Fig. 6A). As anticipated, this finding indicated that more barbed ends would be generated during intercellular adhesion in the wild-type cells than the gelsolin-null cells.
We determined whether greater numbers of barbed ends would be generated at sites of N-cadherin ligation as a result of cadherin-associated gelsolin activity in the wild-type cells. Accordingly, actin monomer addition assays were conducted (37)(38)(39) in wild-type and gelsolin-null fibroblasts plated on Ncad-Fc-coated non-tissue culture plates. The rate of monomer incorporation was quantified by fluorescence image analysis (Fig. 6B). We found significantly faster incorporation of actin monomers in wild-type cells than gelsolin-null cells samples during the first 60 min of adhesion with a subsequent lack of activity after 60 min (p Ͻ 0.05). By 180 min, gelsolin-null samples incorporated ϳ2.5-fold less actin monomer at sites of N-cadherin ligation ( Fig. 6B; p Ͻ 0.01).
Effect of Gelsolin on N-cadherin-associated Cytoskeletal Architecture-We determined whether gelsolin expression affects the relative proportion of N-cadherin adhesions formed between cells. Ncad-Fc-conjugated protein A-coated magnetic beads were allowed to bind to monolayers of wild-type or gelsolin-null fibroblasts for 5 min. Saturation of available bead binding sites was verified by separate immunoblotting experiments that demonstrated minimal amounts of chimeric protein remaining in the supernatant (data not shown). Proteins eluted from the beads were co-immunoblotted for ␤-catenin and ␤-actin. The immunoblots showed that approximately equivalent amounts of ␤-catenin were recruited to N-cadherin-mediated adhesions (Fig. 7A). As there is equimolar stoichiometry of ␤-catenin and ␣-catenin in the cadherin complex (17,46) we considered that there were equivalent numbers of actin binding sites and N-cadherin adhesions formed between the two cell types. As actin filaments bind to N-cadherin through associated ␣-catenin, we co-immunoblotted for ␤-actin. There were dramatically reduced amounts of ␤-actin in the bead preparations from the gelsolin-null cells despite equivalent amounts of ␤-actin in the whole cell lysates (Fig. 7A, lysates not shown). Densitometry revealed that there was 45% less ␤-actin associated with the cadherin adhesions of null cells when compared with wild-type cells (Fig. 7A, right panel), as determined when the densities of ␤-actin bands were standardized with those of ␤-catenin bands. Bare beads were included to show the absence of associated adherens junctions proteins (Fig. 7A).
Scanning electron microscopy was used to visualize the cytoskeletal architecture of wild-type and gelsolin-null fibroblasts that adhered to N-cadherin-Fc-coated plates for 5 min. The gelsolin-null fibroblasts (Fig. 7, B-D) showed longer actin filaments, reduced cross-linking, reduced branching, and a generally lower level of actin filaments than wild-type cells (Fig. 7,  E-G). Quantification of the structural features seen in these images verified that indeed the filament length and 70°b ranching frequency in null cells was significantly less at nascent N-cadherin adhesions when compared with wild-type cells (Fig. 7, H and I). These data suggest that despite an equivalent amount of ␤-actin expression in both cell types and a tendency to form approximately the same number of junctions based on a bead rip-off assay, the underlying cytoskeletal architecture of N-cadherin adhesions is critically dependent on gelsolin expression.

DISCUSSION
Down-regulation of gelsolin expression is an important prognostic marker for progression of malignancy and tumor invasion (47), indicating that gelsolin may play a crucial role in regulating physiological intercellular adhesion. Our findings provide a mechanistic basis by which gelsolin contributes to the maturation of cadherin-mediated adhesions, and consequently, is an important regulator of tissue integrity. We have used the donor-acceptor model and a recombinant Ncad-Fc chimeric pro-tein to study early events involved in the organization of the actin filament network to which N-cadherin molecules are tethered. We demonstrated that in vitro, gelsolin spatially co-local- izes to, and transiently associates with, N-cadherin adhesions. Gelsolin regulates the kinetics and strength of early N-cadherin adhesions through a mechanism that involves actin severing, barbed end generation, and subsequent filament remodeling that is important for the maturation of intercellular adhesions (Fig. 8).
To our knowledge, this is the first report that demonstrates a local role for gelsolin in the regulation of nascent, N-cadherinmediated adhesions. These findings support the developing notion that cadherin ligation marks intercellular junctions as sites of dynamic actin reorganization through the recruitment of various components of the actin assembly machinery (22,24,48). Our data also underline the importance of localized intracellular calcium transients in regulating the formation of intercellular junctions. Thus, although extracellular calcium is crucial in the structural requirements for maintenance of the extracellular domains of cadherins (49,50), intracellular calcium concentration is also likely important for gelsolin severing (28) and the subsequent adhesion strengthening that is effected through remodeling of the actin cytoskeleton.
Calcium and N-cadherin Adhesion Strengthening-Millimolar levels of extracellular calcium are required for maintenance of the structural integrity and the association of lateral dimers of cadherin extracellular domains (49,51). In contrast, micromolar intracellular calcium transients are generated after Ncadherin ligation in fibroblasts (31). These transient increases of intracellular calcium are required for actin polymerization and regulate cadherin-mediated intercellular adhesion through an unidentified mechanism (24). Our results demonstrated an influx of calcium ions through the plasma mem- brane that was localized to sites of N-cadherin ligation and that appears to be important for activating cadherin-associated gelsolin severing activity. Previous reports have shown that in fibroblasts undergoing cadherin ligation, calcium influx occurs through plasma membrane channels located at nascent but not mature N-cadherin-mediated intercellular junctions (45). We found that blockade of this calcium entry resulted in marked reductions of nascent adhesion strength in gelsolin wild-type cells but had no effect on gelsolin-null cells. The absence of any effect in gelsolin-null cells is likely due to the adaptation of these cells to gelsolin deficiency, possibly through compensatory increases of other actin-severing proteins such as actindepolymerizing factor/cofilin that are not calcium-dependent. Indeed, the gelsolin-null cells exhibited measurable severing activity and actin assembly when plated on N-cadherin substrates, albeit at greatly reduced rates when compared with wild-type cells. We suggest that the intracellular calcium transients following N-cadherin ligation (45) are important in intercellular adhesion because of their role in locally activating gelsolin severing activity.
N-cadherin Adhesion and Gelsolin Activation-Cadherin association with the actin cytoskeleton is thought to be the ratelimiting step in epithelial intercellular adhesion (23,24). Inhibition of this association significantly reduces cadherin adhesive function (Refs. 52 and 53 and reviewed in Ref. 54). An expanding list of regulators of actin assembly has been localized to developing cadherin-mediated junctions including the Arp 2/3 complex, cortactin, and the vasodilator-stimulated phosphoprotein/Ena and vinculin/zyxin family members. These findings further support the importance of local actin polymerization in the regulation of cadherin adhesion. Indeed, our data show that the severing activity of gelsolin is an important component of actin filament dynamics at the earliest stages of N-cadherin-dependent adhesions. We found that gelsolin-null mouse fibroblasts exhibited significantly reduced adhesion kinetics and strengthening when compared with wild-type controls. Gelsolin transiently associated with nascent N-cadherin adhesions, suggesting that gelsolin plays an important local role in breaking down existing actin filament networks at nascent contacts. This localized remodeling facilitates locally required configurations of adherens junctions to promote intercellular adhesion.
In addition to increasing the pool of actin monomers, the severing activity of gelsolin followed by uncapping generates a large number of free barbed ends that are necessary for actin assembly (25). Thus, the association of gelsolin with cadherins and its local activation may be required for efficient actin assembly by actin nucleators such as the Arp 2/3 complex, as has been shown in platelets and fibroblasts (26). Indeed, we found that gelsolin wild-type cells incorporated significantly larger amounts of actin monomer and subsequently exhibited considerably more polymerized actin at sites of early N-cadherin ligation than gelsolin-null cells. The greatest amount of actin monomer addition occurred during the first 60 min of N-cadherin ligation with very little addition occurring subsequent to that. This profile corresponds quite closely with the transient association of gelsolin with cadherin adhesion complex noted in the donor-acceptor model. Further, our electron microscopy showed that the actin network of gelsolin wild-type cells was significantly more cross-linked with a shorter average filament length than gelsolin-null cells. The contrast in the microfilament architecture between gelsolin wild-type and null FIG. 8. Proposed mechanism of the role of gelsolin in N-cadherin-mediated intercellular fibroblast adhesion. Intercellular contact via surface-expressed N-cadherin molecules induces adhesion complex formation and tethering to actin cytoskeleton. Gelsolin is recruited to nascent contacts and is activated by locally induced influx of calcium. Gelsolin severs existing actin filaments and generates barbed ends. Actin nucleation sites for polymerization and actin reorganization are necessary for adhesive strength. cells underlines the important functional differences of cadherin adhesion noted above and highlights the importance of gelsolin as a regulator of intercellular adhesions.
Overexpression of gelsolin in Madin-Darby canine kidney cells disrupts intercellular contacts by an unknown mechanism that maintains the composition of the E-cadherin-catenin complex (55). We found that the level of gelsolin did not influence ␤-catenin association with N-cadherin; however, our functional assessments are inconsistent with these findings. This discrepancy may be attributable to differences in cellular background and the type of cadherin that was expressed (56), in addition to variations in gelsolin levels between studies. Although structurally similar (5), different classical cadherin family members mediate functionally distinct adhesions in different tissue types (57). Further, overexpression of gelsolin may produce different effects when compared with cells lacking gelsolin altogether. Gelsolin null cells reconstituted with gelsolin showed the importance of actin severing in the formation of nascent intercellular contacts and intercellular adhesion strengthening.
Loss of cadherin-mediated intercellular adhesion has been implicated in malignant transformation (58). The reduction in cadherin-mediated adhesion strength, which compromises the integrity of intercellular contacts prior to metastasis, may result from reduced gelsolin expression or function. Indeed, down-regulation of gelsolin expression coincides with tumor invasiveness and has been implicated as a prognostic indicator for therapeutic interventions in cancer (59). We found that gelsolin-null cells demonstrated a defect in the maturation of cadherin-mediated adhesions, which resulted in reductions in the strength of intercellular contacts. The rescue of this deficiency by reconstituting gelsolin in these cells underscores the importance of gelsolin as a critical regulator of adhesion strength for nascent contacts. This is particularly relevant for N-cadherin-expressing mesenchymal cells as they exhibit rapid rates of lamellipodial extension and high turnover of cadherinmediated intercellular adhesions (60,61).
Cadherin ligation rapidly increases GTP-bound rac without affecting other Rho family GTPases such as Rho or cdc42 (62,63). As active rac is required for actin filament assembly and insertion into N-cadherin adhesions (16,64), and as gelsolin is an important rac-dependent effector of actin assembly (39), we suggest that N-cadherin ligation may mediate rac-dependent activation of gelsolin that is required for adherens junction formation. Collectively, our data demonstrate the importance of gelsolin and actin remodeling in mediating intercellular adhesion following cadherin ligation.