Liprin-α1 promotes cell spreading on the extracellular matrix by affecting the distribution of activated integrins

During cell motility on the extracellular matrix (ECM), actin-driven protrusion is stabilized by new adhesive sites forming at the extending edge (Geiger et al., 2009; Le Clainche and Carlier, 2008). Lamellipodia are stabilized by clustering and activation of integrins that steady the link between their extracellular ligands and actin (Galbraith et al., 2007; Vicente-Manzanares et al., 2009). This link requires the dynamic regulation of focal adhesions (FAs) (Broussard et al., 2008). Although many of the molecular events involved in the formation of FAs have been identified, understanding of the molecular mechanisms underlying their dynamics during motility is still limited.

The liprin family includes α and β dimeric multidomain adaptor proteins (Serra-Pagès et al., 1998). Liprin-α1 interacts with leukocyte common antigen-related (LAR) tyrosine phosphatase receptors at FAs (Serra-Pagès et al., 1995) and is implicated in the regulation of cell motility (Shen et al., 2007), whereas, in neurons, liprin-α is required for the assembly of pre-synaptic active zones (Stryker and Johnson, 2007). The role of this protein in cell motility remains to be defined. Here, by using COS7 cells plated on fibronectin (FN), we show that liprin-α1 is localized at the edge of spreading cells, where it affects active integrin distribution and is essential for efficient protrusive activity.

In transfected COS7 cells, liprin-α1 localized at the periphery (Fig. 1A). In ventral plasma membranes (VPMs) prepared from ECM-attached transfected cells (Cattelino et al., 1997; Cattelino et al., 1999), liprin-α1 remained associated with the substrate-bound plasma membrane (Fig. 1B). Endogenous liprin was enriched together with talin in VPMs prepared from adherent fibroblasts (Fig. 1C). Staining with purified anti-liprin-α1 antibodies (supplementary material Fig. S1) showed that endogenous liprin partially colocalized with larger, more-central FAs (Fig. 1D). At up to 3 hours of spreading on FN, endogenous liprin was enriched at the cell periphery (supplementary material Fig. S2). Interestingly, co-staining with antibodies for liprin and either paxillin or talin clearly showed little colocalization of endogenous liprin with peripheral FAs. In fact, during spreading, liprin was localized just behind the small nascent FAs present at the cell edge (Fig. 1E; supplementary material Fig. S2). The overlap with older FAs and poor localization with newly formed FAs at the cell edge was observed also for exogenous liprin expressed at low levels (Fig. 1F). Exogenous liprin was excluded also from FAs concentrated at the edge of transfected cells with higher levels of exogenous liprin-α1 (supplementary material Fig. S3).

To study the function of liprin-α1 in the response of cells to the ECM, we altered its expression by either overexpression or downregulation by short interfering RNA (siRNA) and analyzed the effects on spreading. Overexpression of liprin-α1 affected cell morphology and evidently increased spreading (Fig. 2A,B) by strongly enhancing the extent of actin-positive lamellipodia around the cells (Fig. 2C,D), and by inducing the redistribution of newly formed, nascent FAs that became densely packed at the cell edge (Fig. 2E; supplementary material Fig. S4). Reduced spreading of cells that were co-transfected with liprin-α1 and dominant-negative N17-Rac1 indicated the dependence of liprin-enhanced spreading on Rac GTPase (supplementary material Fig. S5A). Moreover, we found that activation of endogenous Rac1 was increased after 30 minutes of spreading on FN upon liprin-α1 overexpression (supplementary material Fig. S5B).

Depletion of liprin-α1 by two specific siRNAs (88±6% s.d., n=4) inhibited spreading (Fig. 2F,G) by affecting lamellipodia formation and the concentration of newly formed FAs at the cell edge (Fig. 2H), as confirmed by time-lapse analysis during spreading (supplementary material Fig. S5C). Conversely, cell adhesion was not affected by alteration of the cellular levels of liprin-α1 (Fig. 2I,J). Time-lapse analysis of cells spreading on FN (supplementary material Movies 1 and 2) showed that GFP-paxillin persisted for shorter times in newly formed FAs at the periphery of cells overexpressing liprin-α1 compared with controls (Fig. 2K). By contrast, liprin-α1 overexpression induced an increase in the number of FAs formed during spreading (Fig. 2L). Overall, these data strongly support a function of liprin-α1 in the turnover of FAs at the periphery of spreading cells.

Integrins can exist in an inactive or activated state, with low or high affinity for their ligands, respectively (Calderwood, 2004). Talin-1 and talin-2 are important regulators of integrin activation (Critchley and Gingras, 2008). The interaction of the talin head with the cytoplasmic region of integrin β-subunits causes conformational changes that increase the affinity of integrins for their ligands (Calderhood et al., 1999; Tadokoro et al., 2003), whereas the rod domain of talin links integrins to the actin cytoskeleton (Jiang et al., 2003; Tanentzapf and Brown, 2006) and is necessary for persistent cell spreading (Zhang et al., 2008).

Here, we analyzed the effects of liprin and/or talin overexpression on cell spreading and on the activation of endogenous β1 integrins. To detect integrin activation, we used surface labeling of the transfected cells with the 9EG7 antibody specific for activated β1 integrins (Lenter et al., 1993). Interestingly, enhanced cell spreading that was induced by liprin overexpression (Fig. 3A,B) was accompanied by the accumulation of activated β1 integrin at the edge (Fig. 3A,D), with a reduction of the fraction of projected cell area occupied by active 9EG7-positive integrins compared with control cells (Fig. 3A,C). We found that transfection with either talin-1 or talin-1H did not affect cell spreading (Fig. 3A,B), but induced an increase in the activation of β1 integrins (Cluzel et al., 2005) (Fig. 3A,C). Moreover, the co-transfection of talin-1 or of its isolated head domain (talin-1H) with liprin prevented both liprin-enhanced spreading (Fig. 3A,B) and talin-induced integrin activation (Fig. 3A,C). These results suggest that the levels of liprin and talin influence the functions of each other during cell spreading. In particular, our findings show that an excess of talin prevents the ability of liprin to enhance cell spreading, possibly by enhancing integrin activation to form more central, mature FAs. By contrast, the coexpression of liprin with talin reduced the hyperactivation of integrins at the cell surface compared with cells transfected with talin-1 or talin-1H alone (Fig. 3A,C), in agreement with the hypothesis that the cellular levels of liprin and talin reciprocally modulate the function of each other on cell spreading and integrin activation.

COS7 cells mainly express talin-1 (Fig. 3E). Three independent siRNAs for talin-1 were used to deplete talin (65±9% depletion ± s.d., n=9) (Fig. 3F; supplementary material Fig. S6). Talin-1 depletion confirmed its requirement for spreading on FN (Zhang et al., 2008) and showed that talin was needed for liprin-enhanced spreading (Fig. 3G,H). Conversely, liprin-α1 depletion inhibited talin-mediated spreading (supplementary material Fig. S7). In talin-depleted cells, there was more activated β1 integrins when liprin was overexpressed (Fig. 3G). One possible explanation is that some residual talin-1 that was left after knockdown of talin by siRNA might be more efficiently used for integrin activation when the cellular levels of liprin are increased by overexpression. This would be consistent with the hypothesis that liprin-α1 is required for efficient talin-mediated integrin activation during cell motility.

Our results point to the fact that, although talin is necessary for integrin activation, it is not sufficient to induce cell spreading in the absence of liprin-α1. By contrast, liprin overexpression cannot rescue cell spreading in the absence of talin. Therefore, liprin and talin are important for the function of each other in spreading, during which they play cooperative and complementary roles in the regulation of cell-edge motility.

We tested whether the interaction between liprin-α1 and LAR (Serra-Pages et al., 1998) was required for the effects of liprin on spreading. LAR depletion (82±18% s.d., n=3) (Fig. 4A) reduced spreading to the same extent of depletion of both liprin and LAR (Fig. 4B,D), suggesting that these proteins participate in the same pathway. LAR depletion in cells overexpressing liprin-α1 prevented also liprin-enhanced spreading (Fig. 4C), but did not prevent the redistribution of FAs at the edge of spreading cells (Fig. 4E), nor did it preclude the localization of endogenous liprin at the edge of spreading cells (data not shown).

LAR interacts with liprin by binding to the sterile alpha motif 2 (SAM2) domain (Olsen at al., 2006). The liprin-ΔSAM2 mutant, which cannot bind LAR (Fig. 4F) (Serra-Pagès at al., 1994), could not induce increased spreading (Fig. 4G) but, similar to liprin, was still able to induce the formation of large lamellipodia (data not shown) and the increase in concentration of FAs at the cell edge (Fig. 4H,I). The cell perimeter positive for activated β1 integrins was increased both in liprin-ΔSAM2- (79.3±3.1%, n=20 perimetral fields) and liprin- (91.5±1.1%, n=20 perimetral fields) expressing cells compared with control cells (67.9±2.5%, n=20 perimetral fields). Moreover, the projected cell area occupied by FAs in liprin-ΔSAM2-transfected cells was similar to control cells but higher than in liprin-transfected cells (Fig. 4J), reflecting the higher density of central FAs in liprin-ΔSAM2-transfected cells (Fig. 4H). These data indicate the uncoupling of the liprin-mediated reorganization of the cell edge from the LAR-mediated effects. Liprin-α1 and LAR cooperate in cell spreading, and two distinct activities can be identified in this process: on one hand the binding of LAR to liprin is needed for cell spreading, but not for the ability of liprin to enhance either the formation of FAs or integrin activation at the periphery of spreading cells. In this way, liprin can enhance the activation of peripheral integrins independently of its binding to LAR. On the other hand, liprin requires the interaction with LAR to induce the actin-mediated protrusive events necessary to move the lamellipodia forwards.

Cell motility requires the dynamic regulation of actin and FA turnover at the protruding edge of the cell (Vicente-Manzanares et al., 2009). We have shown here that changes in the cellular levels of liprin-α1 have profound effects on the morphology and organization of cells spreading on ECM. Our data indicate reciprocal effects of talin and liprin on each other in the control of cell spreading, and suggest that a balance between the cellular levels of these proteins is required for proper dynamics at the cell edge during cell motility. The positive role of liprin-α1 in protrusion was reflected by the increase of lamellipodia and FAs at the periphery (Fig. 2). The concentration of liprin near the cell edge and its poor colocalization with activated peripheral β1 integrins suggest a role of liprin in the dynamic turnover of FAs during motility, which is supported by the dynamic analysis presented in this study of the effects of liprin expression on FAs.

We propose a model in which ubiquitous liprin-α1 acts as a dynamic regulator of protrusion by impinging on the mechanism of talin-mediated integrin activation (Fig. 4K). The effects on protrusion observed by changing the cellular levels of liprin-α would require the specific recruitment of liprin near the cell edge through unknown molecular interactions. This mechanism, by recruiting liprin at the centripetal side of newly formed FAs, would somehow enhance their turnover, and result in a more efficient protrusive activity at the cell edge (supplementary material Movies 1 and 2). The action of liprin-α on cell-edge dynamics cannot be simply explained by an effect of liprin on integrin inactivation and/or FA formation, because this would result in the inhibition of protrusion rather than in its potentiation in cells overexpressing liprin. By contrast, a simple potentiation of FA formation by liprin would result in FAs with a longer life, which is opposite to what we found in cells overexpressing liprin. In fact, liprin overexpression actually enhanced both the formation of newly formed FAs accumulating at the cell periphery and increased their turnover, whereas liprin-α1 depletion inhibited both FA formation and cell spreading. We therefore propose that liprin-α plays a complex role in the dynamic events occurring at the edge of a moving cell. We expect that liprin functions at the cell edge are mediated by a molecular network that is assembled by a multi-domain adaptor protein such as liprin-α. We have shown that LAR is probably part of this network, but it is not sufficient for the full action of liprin in cell motility. Future work will be aimed at identifying the other players linking liprin-α to talin-mediated integrin activation and lamellipodia formation.

The steady association of liprin to the plasma membrane suggests that liprin interacts with as-yet-unidentified components of the plasma membrane. Because liprin localization at the cell edge is independent from interaction with LAR, we propose that liprin, by binding to both the plasma membrane and LAR, affects protrusive activity by recruiting the molecular machinery needed to regulate efficient FA turnover and actin dynamics. Therefore, liprin is an important regulator of cell-edge dynamics, where balanced levels of expression of liprin and talin are important for productive protrusion during motility.

Monoclonal antibodies (mAbs) for FLAG, α-actinin, talin, tubulin (Sigma-Aldrich); vinculin (Upstate); paxillin, phosphotyrosine, LAR 150- and 200-kDa forms (BD Biosciences); TS2/16 (Hemler et al., 1984) (American Type Culture Collection) and 9EG7 (Lenter et al., 1993) (BD Biosciences) recognizing total and activated human β1 integrin, respectively, were used. Polyclonal antibodies against the cytoplasmic domain of β1 integrin (Tomaselli et al., 1988); FLAG and actin (Sigma-Aldrich); FAK and LAR 85-kDa P subunit (Santa Cruz Biotechnology); GFP (Molecular Probes) were used. The anti-liprin-α1 rabbit polyclonal antibody raised to a GST-fusion protein including the C-terminal human liprin-α1 (amino acids 818-1202) was affinity purified on a maltose-liprin(818-1202) fusion protein adsorbed to amylose resin affinity matrix (New England Biolabs).

GFP–liprin-α1 and FLAG–liprin-ΔSAM2 plasmids were obtained from the FLAG–liprin-α1 plasmid. Human GFP–talin-1 and mouse GFP–talin-1H (amino acids 1-433) were from Anna Huttenlocher (University of Wisconsin-Madison, WI). The pSP65-SRα2-HPTPδ and pSP65-SRα2-LAR plasmids (Pulido et al., 1995) were from Robert Kypta (Imperial College, London). The GFP-paxillin plasmid was from Victor Small (Austrian Academy of Sciences, Vienna, Austria).

siRNAs used for silencing (from Sigma or Qiagen) were: siRNA liprin-1a (targeting 5′-TTCCAAGGTACAAACTCTTAA-3′) and liprin-1b (targeting 5′-CACGAGGTTGGTCATGAAAGA-3′); LAR siRNA (HS-PTRF_6 HP validated, from Qiagen); and talin-1a (targeting 5′-AATCGTGAGGGTACTGAAACT-3′) (Manevich et al., 2007), talin-1b (targeting 5′-CAGCTCGAGATGGCAAGCTTA-3′) and talin-1c siRNAs (targeting 5′-CCGCATTGGCATCACCAATCA-3′). Control siRNA targeting the luciferase sequence 5′-CATCACGTACGCGGAATAC-3′ was used. Cells transfected with Lipofectamine 2000 (Invitrogen) and 2-3 μg of plasmids, or siRNAs (50-100 nM) were used after 1-2 days as specified.

Cells lysed with 0.5-1% Triton X-100, 20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM sodium orthovanade, 10 mM sodium fluoride, and protease inhibitors were immunoprecipitated from equal amounts of lysate using antibodies coupled to protein A Sepharose (Amersham). For immunoblotting, primary antibodies were visualized by ECL or ¹²⁵I-anti-mouse Ig or protein A (Amersham). For evaluation of Rac-GTP, transfected COS7 cells were lysed (50 mM Tris-Cl pH 7.5, 100 mM NaCl, 10% glycerol, 2 mM MgCl₂, 1% NP-40) 30 minutes after plating on FN, and cleared lysates were incubated with GST-PAK1-CRIB-adsorbed glutathione Agarose beads for 30 minutes at 4°C. Washed beads were used for immunoblotting with anti-Rac1 mAb (Upstate).

VPMs were prepared from chicken embryo fibroblasts (CEFs) or COS7 cells on 10 μg ml^–1 FN, as described (Cattelino et al., 1999). Briefly, VPMs were prepared with a jet of ice-cold buffer with anti-proteases and anti-phosphatases (20 mM HEPES-NaOH pH 7.6, 0.3 mM PMSF, 10 mM NaF, 1 mM NaV), and used for immunofluorescence. Triton-soluble and -insoluble fractions from VPMs were used for immunoblotting.

COS7 cells were trypsinized 1-2 days after transfection. A total of 25,000-30,000 cells were plated on each 13-mm-diameter coverslip coated with 10 μg ml^–1 FN. Cells were fixed after 1 hour and processed for immunofluorescence. Images were analyzed with ImageJ. Cell-adhesion assays were as described (Cattelino et al., 1995).

Cells and VPMs were incubated with the indicated antibodies after fixation. For 9EG7 and TS2/16 antibodies recognizing extracellular epitopes of the β1 integrin, cells were incubated 15 minutes on ice with 5 μg ml^–1 of purified IgG before fixation. F-actin was revealed by FITC- or TRITC-conjugated phalloidin. Cells were observed with Axiophot or Axiovert microscopes (Zeiss), or confocal microscopes (PerkinElmer and Leica Microsystems SpA). Images were processed using Photoshop (Adobe) and analyzed with ImageJ. The analysis of the fraction of the projected cell area occupied by active β1 integrins was performed on thresholded images, by measuring the total area occupied by FAs and/or integrin clusters larger than 0.5 μm². The values obtained were represented as percentages of the total projected cell area on the substrate. The percentage of cell perimeter positive for FAs was calculated from different perimetral fields by measuring the fluorescence intensity of 9EG7-positive integrin clusters. Data in the bar graphs are expressed as mean ± s.e.m. of one of at least two or three repetitions in which 70-150 cells per experimental condition were analyzed. P-values were calculated by Student's t-test (two-tailed distribution, two-sample unequal variance). Live cells were observed with an UltraVIEWERS microscope (PerkinElmer). Measurements of FA lifespan during cell spreading on FN were made by counting the amount of time lapsed between the first and last frame in which an individual adhesion could be observed. Adhesion formation was evaluated by counting the number of new GFP-paxillin-positive FAs appearing during 30 minutes in cells spreading on FN.