Macrophage Mesenchymal Migration Requires Podosome Stabilization by Filamin A

Background: Filamin A is an actin-binding and scaffolding protein. Mutations in the filamin A gene cause developmental anomalies in humans.

Results: Filamin A is required for podosome stabilization, podosome rosette formation, extracellular matrix degradation, and for three-dimensional mesenchymal migration.

Conclusion: New functions are assigned to filamin A.

Significance: Identification of actors involved in cell migration is crucial for understanding human developmental disorders.

Abstract

Filamin A (FLNa) is a cross-linker of actin filaments and serves as a scaffold protein mostly involved in the regulation of actin polymerization. It is distributed ubiquitously, and null mutations have strong consequences on embryonic development in humans, with organ defects which suggest deficiencies in cell migration. We have reported previously that macrophages, the archetypal migratory cells, use the protease- and podosome-dependent mesenchymal migration mode in dense three-dimensional environments, whereas they use the protease- and podosome-independent amoeboid mode in more porous matrices. Because FLNa has been shown to localize to podosomes, we hypothesized that the defects seen in patients carrying FLNa mutations could be related to the capacity of certain cell types to form podosomes. Using strategies based on FLNa knock-out, knockdown, and rescue, we show that FLNa (i) is involved in podosome stability and their organization as rosettes and three-dimensional podosomes, (ii) regulates the proteolysis of the matrix mediated by podosomes in macrophages, (iii) is required for podosome rosette formation triggered by Hck, and (iv) is necessary for mesenchymal migration but dispensable for amoeboid migration. These new functions assigned to FLNa, particularly its role in mesenchymal migration, could be directly related to the defects in cell migration described during the embryonic development in FLNa-defective patients.

Introduction

Filamins are cytoskeletal proteins that organize actin filaments into networks and link these networks to cell membranes. Three isoforms have been identified, FLNa (filamin A), the most abundant and widely expressed isoform localizes to filopodia, lamellipodia, stress fibers, focal contacts, and invadosomes in osteoclasts and tumor cells (1–3). Filamin B is associated to stress fibers but does not normally localize to focal contacts. Filamin C is expressed primarily in muscle cells (4).

FLNa is a cross-linker of actin filaments, which forms orthogonal branches, with the actin networks behaving as weak elastic solids. Such branches cannot be formed by other cross-linkers such as α-actinin or temporary branching proteins such as Arp2/3 (5). FLNa is also a scaffolding protein that binds multiple partners including membrane receptors, enzymes, and signaling intermediates. Many of these partners are involved in the regulation of actin polymerization, and FLNa thus participates in signal transduction related to F-actin polymerization and organization. Integrins interact with FLNa, which mediates a link between the cytoskeleton and the cell membrane to control cell adhesion (3). Over 90 binding partners of FLNa have been identified (4). As a consequence, mutations in the FLNa gene can result in a wide range of anomalies, which include cell adhesion and cell migration defects (4–7). The role of FLNa in cell migration has been emphasized recently in reviews (4, 7), underlining its essential function for embryonic development, organ formation, and homeostasis. For example, it emerges that an appropriate level of FLNa is required for migration of neuron progenitors during embryonic development (4). Indeed, in humans, inactivation of the FLNa gene causes brain malformation and disrupts directed neuronal migration (4, 8). In addition, it has been shown recently that FLNa is required for monocyte two-dimensional migration during in vitro osteoclastogenesis (9). Conversely, cleavage of FLNa by calpain has also been reported to facilitate two-dimensional cell migration, suggesting that the role of FLNa in two-dimensional migration could differ from one cell type to another (1, 7, 10, 11). In vivo, it has been reported that the metastatic capacity of FLNa knockdown tumor cells is modified (11). How FLNa modulates cell migration, especially in three-dimensional environments, has not yet been elucidated.

In three-dimensional environments, macrophages use two distinct migration modes, the amoeboid and mesenchymal modes, depending on the architecture of the matrix (12). The protease-driven mesenchymal migration of macrophages takes place in dense three-dimensional environments and involves podosomes (12, 13). Macrophages and monocyte/macrophage-derived cells (dendritic cells and osteoclasts) constitutively form podosomes when layered on extracellular matrix proteins, a two-dimensional environment while, in other cell types, podosomes can form transiently (14–16). These actin-rich structures are involved in integrin-mediated cell adhesion and in proteolytic degradation of the extracellular matrix (16). Interestingly, in cells cultured in two dimensions, FLNa has been reported to be associated with podosome structures in osteoclasts and with invadopodia, a podosome counterpart, in tumor cells (1, 11, 16). Thus, we decided to investigate whether FLNa regulates the three-dimensional migration process in macrophages, the archetypal migratory cells, and whether it is involved in podosome dynamics and organization.

EXPERIMENTAL PROCEDURES

Antibodies and Reagents

Antibodies against human filamin A (hFLNa)⁶ clone PM6/317 were obtained from Chemicon Intl. (Temecula, CA). FLNa antiserum against mouse filamin A (mFLNa) and ASB2 antiserum were described previously (17, 18). Rabbit polyclonal anti-Hck Abs (sc-72) were from Santa Cruz Biotechnology (TEBU-Bio, France), actin monoclonal Abs, anti-vinculin Abs were from Sigma-Aldrich. Secondary HRP-conjugated Abs were from Bio-Rad (Le Perray en Yvelines, France), secondary anti-mouse and anti-rabbit Abs conjugated to Alexa Fluor 488 or Alexa Fluor 555 and Texas Red/Alexa Fluor 488/Alexa Fluor 633-coupled phalloidins were from Molecular Probes (Invitrogen). The inhibitor of Src kinases SU6656 was purchased from Sigma, recombinant IFN-γ from Immunotools (Friesoythe, Germany). Extracellular matrix proteins: fibrinogen (Sigma-Aldrich), fibronectin (Sigma-Aldrich), vitronectin (Fisher Scientific, Illkirch, France), or FITC-coupled gelatin (Invitrogen).

Cell Culture

Human Monocyte-derived Macrophages (MDMs)

Human monocytes were isolated from the blood of healthy donors as described previously (12). The culture medium RPMI 1640 (Invitrogen) containing 10% heat-inactivated FCS, antibiotics, and 20 ng/ml M-CSF (PeproTech, Rock Hill, NJ) was renewed on the third day of culture. MDMs were used for experiments at day 7 of differentiation. MDMs were distributed on glass coverslips coated with 40 μg/ml fibrinogen as described (13). In some experiments, 1 h after plating, SU6656 was added for 30 min, and cells were fixed in paraformaldehyde-sucrose and stained for immunofluorescence microscopy (see below).

RAW264.7 Macrophages

RAW264.7 macrophages were cultured in RPMI 1640 medium containing 10% heat-inactivated FCS and antibiotics at 37 °C in humidified 5% CO₂ atmosphere. RAW264.7 cells were transfected using the Amaxa electroporation system (19) with the expression vectors encoding for EGFP, EGFP-ASB2a-WT or EGFP-ASB2a-LA protein (20), mouse Hck shRNA (19), mFLNa shRNA (18), or hFLNa (21). Clones stably expressing shRNA against FLNa (OpenBiosystem clone no. V2HS_131780, targeting sequence 5′-ggtgatcactgtggacactaa tagtgaagccacagatgta ttagtgtccacagtgatcacc-3′) or against Hck (5′-ctagttccaaaaa ccgtatgcctcga ccagataat ctcttgaattatctggtcgaggcatacggcggg-3′ designed and cloned as described (19)) were obtained by limiting dilution in parallel to selection with puromycin and characterized by Western blot analysis and podosome content. Among the several clones obtained with similar phenotypes, one, shown under “Results,” was chosen and used for rescue of mFLNa with hFLNa. hFLNa-expressing clones were generated by limiting dilution and selection with geneticin. Two arbitrarily selected clones were characterized by Western blot analysis and podosome content and had similar phenotypes (only one is shown). RAW264.7 cells were seeded either on coverslips coated with 10 μg/ml vitronectin for 24 h or on coverslips coated with FITC-coupled gelatin and subsequently coated with vitronectin for matrix degradation. IFN-γ was added to a final concentration of 100 units/ml for 24 h, cells were fixed using PFA-sucrose and stained for immunofluorescence microscopy (see below).

Fibroblasts

Mouse embryonic fibroblasts (MEF)-3T3 Tet-Off cell clones stably expressing the constitutively active Hck isoforms, p59Hck^ca and p61Hck^ca in fusion or not with EGFP (HckY/F501, Hck^ca) have been described previously (13, 22). These cells optimally expressed Hck after 7 days in doxycycline-free culture medium. Hck-negative MEF-3T3 Tet-Off were used as a negative control. MEF-3T3 fibroblasts stably expressing p59/p61-Hck^ca(-EGFP) were seeded on glass coverslips and fixed 24 h later.

NIH-3T3 stably expressing FLNa shRNA or luciferase shRNA as a control were cultured as described (18) in the presence of 1 mm sodium pyruvate. NIH-3T3 were seeded on fibronectin-coated coverslips, transfected with the p59/61-Hck^ca-EGFP expression vector using calcium phosphate (23, 24), fixed 12 h later, and stained for F-actin.

Bone Marrow-derived Macrophages

Mouse bone marrow-derived macrophages (BMDMs) were differentiated for 7 days as described (13). The culture medium, RPMI 1640 containing 10% heat-inactivated FCS, antibiotics, and 20 ng/ml M-CSF (Immunotools) was renewed on the third day of culture. For immunofluorescence microscopy experiments, cells were seeded on fibronectin-coated coverslips. In some experiments, BMDMs were transduced after 2 days of the differentiation process by adding mCherry-LifeAct lentiviral vector (10⁶ effective viral particles for 10⁶ macrophages) as described (25). At day 7, cells were harvested and embedded in Matrigel, plated in Lab-tek glass base chambers, and kept in a humidified atmosphere at 37 °C and 5% CO₂ for at least 18 h. Cells were then visualized using a Zeiss 710 NLO microscope with a DPPS-laser 561 nm every 10 min for 16 h.

Gelatin FITC Degradation

For the matrix degradation assay, coverslips were coated with 0.2 mg/ml FITC-coupled gelatin (13). Macrophages were seeded and fixed and stained 24 h later. Dark areas in FITC gelatin images were measured using the threshold command of ImageJ software. The degradation index was measured as described previously (13).

Three-dimensional Migration Assay

For migration assays, thick layers of Matrigel or 2.1 mg/ml collagen I were polymerized in Transwell inserts as described previously (12). Macrophages were starved of serum for at least 2 h, harvested, and seeded on top of the matrix at 3·10⁴ cells/Transwell insert. Quantification was performed as described previously (12). The percentage of cell migration was obtained as the ratio of cells within the matrix to the total number of counted cells.

Immunofluorescence Microscopy

Cells were fixed with 3.7% paraformaldehyde, 150 mm sucrose, permeabilized with 0.1% Triton X-100 (Sigma-Aldrich), and stained with primary antibodies: anti-hFLNa (1/1000), anti-mFLNa (1/500), anti-Hck (1/200), and secondary antibody anti-rabbit or anti-mouse conjugated to Alexa Fluor 488 or Alexa Fluor 555 and phalloidin-Texas Red or phalloidin-Alexa Fluor 350 or phalloidin-Alexa Fluor 633 (1 unit/ml). Slides were visualized with a Leica DM-RB fluorescence microscope or using a confocal microscope (Leica SP2). Image stacks were collected using sequential scanning and a standardized 120 nm z-sampling density. Images were processed for brightness and contrast and filtered for noise with Adobe Photoshop, in compliance with the current ethical rules. In some experiments, the fluorescence intensity of FLNa staining was quantified in epifluorescence images acquired with a Leica DM-RB microsocope as a function of the cell size (only spread cells with a surface above 350 μm² were considered). For quantitative analyses of F-actin and FLNa staining shown in Fig. 4, A and B, podosomes were segmented using the auto local threshold function from Fiji software.

FIGURE 4.

The formation of podosome rosettes triggered by expression of Hck^ca in fibroblasts requires FLNa. MEF-3T3 fibroblasts and MEF-3T3 fibroblasts expressing human p59Hck^ca/p61Hck^ca were stained for mFLNa and F-actin and observed by confocal microscopy. A, control cells did not form podosome rosettes, whereas the cells expressing p59 Hck^ca/p61Hck^ca showed FLNa accumulation at podosome rosettes (arrowheads). B, MEF-3T3 fibroblasts expressing p59Hck^ca/p61Hck^ca-GFP and stained for FLNa and F-actin showed accumulation of Hck and FLNa and F-actin at podosome rosettes (arrowheads). Fluorescence intensity profiles along the white dashed line are shown. C, expression of p59Hck^ca/p61Hck^ca-GFP in NIH3T3 stably expressing shRNA against luciferase (control) or against mFLNa were stained for F-actin (scale bar, 10 μm). D, quantification of p59Hck^ca/p61Hck^ca-GFP expressing cells forming podosome rosettes (mean ± S.D. of three independent experiments).

Measurement of Podosome Lifespan

RAW264.7 cells were transfected with the expression vector encoding for mCherry-LifeAct, using the Amaxa® electroporation system. Cells were layered onto vitronectin-coated Lab-Tek chambers and IFN-γ (100 units/ml) was added 4 h later. After 24 h, cells were imaged using an inverted microscope (Leica DMIRB, Leica Microsystems) equipped with a motorized stage and an incubator chamber to maintain the temperature and CO₂ concentration constant. Images were acquired with Metamorph software. In each experiment, time-lapse images were acquired every 15 s in one z-plane over a 15–30-min period for four to five representative fields of view per cell type. Quantification of podosome life-span was measured manually using ImageJ software for podosomes appearing and disappearing during the time course of the experiment, and results were expressed as the mean ± S.D. of >50 podosomes from 10–15 cells from three independent experiments. Cells were screened visually before measurement, and polarized cells were not taken into account.

Western Blot

Proteins were separated with 5–8% SDS-PAGE gels, and proteins were transferred onto nitrocellulose membranes and stained with anti-hFLNa (1/10,000), anti-mFLNa (1/5000), anti-Hck (1/1000: Santa Cruz Biotechnology), anti-actin (1/5000), anti-ASB2 Abs (1/5000), or anti-phosphotyrosine Abs (4G10, 1/2000) revealed by secondary horseradish peroxidase-coupled Abs (1/10,000). Signals were visualized with enhanced chemiluminescence reagents (Amersham Biosciences) and quantified using Adobe Photoshop CS3 software.

Statistical Analysis

Data are reported as means ± S.D. Statistical comparisons between two sets of data were performed with a unilateral Student's unpaired t test. Statistical comparisons between three or more sets of data were performed with analysis of variance, and a Tukey post test. Statistical comparisons of two sets of nominal values were performed with Fisher's exact test. Statistical comparisons of three or more sets of nominal values were performed with a Chi-square test and Bonferonni correction (*, p < 0.05; **, p < 0.01; and ***, p < 0.001).

In Vitro Phosphorylation Assay

hFLNa was immunoprecipated as described in Ref. 20. Recombinant Hck (WT or KD) was produced in Escherichia coli BL21(DE3)pLysS and purifed as described (26). hFLNa was incubated (or not) with Hck-WT or Hck-KD in the presence of 1.5 mm ATP, 1.5 mm MgCl₂, 1.5 mm MnCl₂ in 100 mm Hepes at 30 °C for 15 min, before addition of Laemmli buffer for Western blot analysis.

RESULTS

FLNa Is Involved in Mesenchymal but Not Amoeboid Migration Mode in Macrophages

The migration capacity of BMDMs from conditional knock-out FLNa mice (9) was analyzed using Transwells in which a thick layer of Matrigel matrix was polymerized (12, 13). In dense, poorly porous matrices such as Matrigel, macrophages use the mesenchymal migration mode (12). It is characterized by an elongated and protrusive cell shape and requires proteases, adhesion proteins, the tyrosine kinase Hck, and formation of three-dimensional podosomes, whereas the Rho kinase (ROCK) is dispensable (12, 13, 25). As shown in Fig. 1, FLNa^−/− BMDMs had a reduced mesenchymal migration capacity in Matrigel compared with WT macrophages (Fig. 1A).

FIGURE 1.

FLNa^−/− BMDMs have decreased abilities to perform three-dimensional mesenchymal migration, to form podosome rosettes, and to degrade gelatin FITC. BMDMs from WT or FLNa^−/− mice were seeded on thick layers of matrices of Matrigel (A) or fibrillar collagen I (B), and the percentage of cells migrating into the matrices was quantified (mean ± S.D. of four independent experiments). Pictures of WT (a′ and b′) and FLNa^−/− cells (a″ and b″) migrating, respectively, in Matrigel (a′ and a″) and fibrillar collagen I (b′ and b″) (z is the depth value of cells in the focal plan marked by an arrow). C, FLNa^−/− BMDMs embedded into Matrigel are defective in forming cell protrusions (see supplemental Movies 1 and 2). Scale bar, 10 μm. D, BMDMs from WT or FLNa^−/− mice were seeded on fibronectin-coated coverslips for 16 h and stained for F-actin. The arrow points to a typical podosome rosette. E, quantification of cells making podosome rosettes in conditions of D (arrow) (mean ± S.D. of four independent experiments). F, macrophages seeded on coverslips coated with gelatin FITC for 16 h were then stained for F-actin. The arrow points to a non-degrading cell, and the dashed lines show the degraded areas. G, quantification of cells degrading the matrix, in conditions of F (mean ± S.D. of four independent experiments). H and I, quantification of the surface of degraded gelatin FITC per cell surface, expressed as the mean area degraded per cell (μm²) in H, and the percentage of degraded area per cell surface in I, in condition of F (mean ± S.D. of four independent experiments, at least 25 cells were quantified per condition). ***, p < 0.001.

In porous matrices such as fibrillar collagen I, macrophages migrate using the amoeboid mode characterized by a rounded cell shape, dependent on the Rho/ROCK signaling pathway and independent of proteases and podosomes (12, 25). We observed that in fibrillar collagen I, the migration capacity of FLNa^−/− BMDMs was not affected (Fig. 1B).

During mesenchymal migration and not for the amoeboid mode, we have reported that three-dimensional podosomes are observed at the tip of cell protrusions, where proteolytic degradation of the matrix is undertaken to create paths for cell migration (12, 25). We thus examined the formation of three-dimensional podosomes in live FLNa^−/− macrophages embedded into Matrigel and transduced with mCherry-LifeAct to stain F-actin (25, 27). As shown by video microscopy, the formation of cell protrusions was affected strongly when compared with WT BMDMs (Fig. 1C, supplemental Movies S1 and S2).

To determine whether formation of podosomes was affected in FLNa^−/− BMDMs in two-dimensional environments (layered on coverslips), we examined podosomes by immunofluorescence microscopy, in parallel to their capacity to degrade the extracellular matrix, as assayed by gelatin FITC degradation. Podosomes can either be spread all over the ventral face of adherent macrophages, or limited to specific areas called clusters, or organized as rosettes (22). Although, as expected from our previous report (13), WT BMDMs mostly organized their podosomes as rosettes, FLNa^−/− BMDMs had a defect in podosome rosette formation (Fig. 1, D and E) and in matrix proteolysis (Fig. 1, F–I). Thus, these results show that FLNa^−/− mouse macrophages have a defect in podosomes rosettes formation, matrix degradation, and three-dimensional mesenchymal migration, which is podosome-dependent.

FLNa Is Present at Podosomes and Podosome Rosettes in Human Macrophages

Having observed a defect of podosomes in FLNa^−/− mouse macrophages, we next examined whether, in human macrophages, FLNa is present at podosomes. Human MDM of healthy donors, layered on fibrinogen, formed individual podosomes (Fig. 2A, arrowhead), and ∼25% of the cells spontaneously organized their podosomes as rosettes (Fig. 2A, arrow). FLNa was observed at individual podosomes, forming a ring around the F-actin core (Fig. 2A, arrowhead), similarly to what has been described for vinculin, talin, and paxillin (16). Moreover, FLNa accumulated at podosome rosettes where it co-localized with F-actin (Fig. 2A, arrow, and 2D). The αMβ2 integrin (CD11b-CD18), the major leukocyte fibrinogen receptor (28), and Hck, an Src family tyrosine kinase specifically expressed in phagocytes (29), which regulates the organization of podosomes as rosettes (13), were also present at podosome rosettes (Fig. 2, B–D).

FIGURE 2.

FLNa is localized at podosomes and accumulates with integrin and Hck at podosome rosette of human MDMs. MDMs plated on fibrinogen were stained for microscopy observation for FLNa and F-actin (A), or CD11b and F-actin (B), or Hck and F-actin (C); insets are magnification of areas depicted by the white squares (scale bar, 10 μm). D, normalized fluorescence intensity profiles along the white dotted line (a′, b′, c′) in A, B, and C, respectively. hFLNa was observed at individual podosomes forming a ring around the F-actin core (A, arrowhead) and at podosome rosettes (A, arrow), where CD11b (B, arrow) and Hck (C, arrow) also accumulated.

Thus, in human macrophages FLNa is present at rings of individual podosomes. Furthermore, it accumulates with, β2 integrins and Hck at podosome rosettes, suggesting that FLNa could also play a role in these cell structures in human macrophages.

Filamin A Is Involved in Podosome Stability and Podosome Rosette Formation

As a cross-linker of actin filaments and a scaffold protein involved in the regulation of actin polymerization, FLNa might have a role in the regulation of podosome stability and lifespan, and in organization of podosomes as rosettes.

Thus, different strategies were undertaken to deplete FLNa: transient expression of ASB2α a subunit of an E3 ubiquitin ligase complex, which targets FLNa for proteasomal degradation (20), and stable expression of mouse FLNa shRNA (18). For this, we used the macrophage cell line RAW264.7, which is relatively easy to transfect. When we looked at the localization of endogenous FLNa by immunostaining, we found that, similar to human MDMs (Fig. 2), it was present at the podosome ring and accumulated at podosome rosettes (supplemental Fig. S1A), and we also noticed that the FLNa fluorescence intensity was heterogeneous from one cell to another. Interestingly, we found a positive correlation between the FLNa fluorescence intensity calculated in cells of similar size and the rate of podosome formation and their organization as clusters and rosettes (supplemental Fig. S1B). A similar correlation between the intensity of fluorescence staining with FLNa antibodies and the presence of podosomes and podosome superstructures was obtained in human macrophages (data not shown). In RAW264.7 macrophages expressing GFP-ASB2α, the expression of FLNa was decreased, and the percentage of cells with podosome rosettes was reduced compared with control macrophages expressing GFP or the ASB2α-E3-defective mutant ASB2α-LA (supplemental Fig. S2).

Similarly, in RAW264.7 macrophages stably transfected to express an shRNA specific for mouse FLNa mRNA, we observed that: FLNa was knocked down by ∼60%, the percentage of cells with podosomes was diminished, the density of F-actin in clouds (30) surrounding remnant podosomes was reduced (supplemental Fig. S1C), and podosome rosettes were absent (Fig. 3, A–C). Furthermore, the formation of rosettes was rescued in those cells by expressing human FLNa (Fig. 3, A–C).

FIGURE 3.

Macrophages with inhibited FLNa expression have a defect in podosome and podosome rosettes formation. A, RAW264.7 macrophages (Control) or RAW264.7 cells stably expressing shRNA against mFLNa, and those rescued with stable expression of hFLNa were all seeded on vitronectin-coated coverslips and treated with IFN-γ to enhance the formation of podosomes, fixed and stained for F-actin and FLNa. Cells counted as forming podosomes and podosome rosettes shown by an arrow presented numerous actin dots (≥5) and rosettes (≥1), respectively (scale bar, 10 μm). B, quantification of cells with podosomes or podosome rosettes (mean ± S.D. of three independent experiments). C, Western blot against hFLNa, mFLNa, and actin and quantification of mFLNa in three experiments. AU, arbitrary unit. D, RAW264.7 macrophages control or transfected with mFLNA shRNA alone or together with a vector expressing hFLNA were transiently transfected with mCherry-Lifeact to reveal F-actin. The lifespans of podosomes were then evaluated using time-lapse microscopy and are plotted for each cell type (mean ± S.D. of three independent experiments, five to 10 podosomes analyzed per cell in at least three cells per experiment). ***, p < 0.001.

Next, we used time-lapse videomicroscopy to analyze the role of FLNa on the lifespan of podosomes in RAW264.7 macrophages expressing both the anti-FLNa shRNA and the F-actin binding peptide mCherry-LifeAct (27, 31). As shown in Fig. 3D, the podosome lifespan was decreased in FLNa-depleted cells, and when FLNa-depleted RAW264.7 macrophages were complemented by human FLNa, podosome lifespan was restored. These results indicate that FLNa plays a critical role in podosome formation and/or stability and is required for the organization of podosomes into rosettes.

Hck and FLNa Exhibit Similar Properties on Podosome Stability, Organization, and Functionality

In osteoclasts, the kinase activity of Src has been involved in the control of podosome lifespan (32). In macrophages, the organization of podosomes into rosettes has been shown to be regulated by Hck, and Hck also plays a critical role in the protease-dependent mesenchymal migration (13). In addition, the Src family tyrosine kinase Lck, which is specifically expressed in lymphocytes, has been shown to activate the actin cross-linking property of FLNa (33). So we examined whether the formation of podosome rosettes induced by Hck involves FLNa.

We took advantage of a cellular model that we had established previously to dissect the role of Hck on podosome rosette formation. It involves ectopic expression of constitutively active Hck (Hck^ca) in MEF-3T3 Tet-Off fibroblasts (34). Although MEF-3T3 fibroblasts are unable to form podosomes spontaneously, fibroblasts expressing Hck^ca form podosome rosettes with the classical “donut” shape structure where Hck, F-actin, and FLNa accumulated (Fig. 4, A and B) (34). When NIH-3T3 fibroblasts depleted in FLNa by stable shRNA expression were transfected with the Hck^ca cDNA construct (18), almost no podosome rosettes were formed (Fig. 4, C and D). These results indicate that, in the formation of podosome rosettes, FLNa is either required in the Hck signaling pathway or an essential component of these cell structures.

In the next experiments, we investigated the role of Src kinases in the organization of podosomes and localization of FLNa. The effect of SU6656, a broad inhibitor of Src kinases was examined in human macrophages. We observed that fewer macrophages formed podosomes and podosome rosettes (Fig. 5, A and B, quantified in C) as expected from previous results obtained with PP1, another inhibitor of Src kinases (34). Although FLNa was organized as a duct around the F-actin core of podosomes in control human macrophages (Fig. 5A, see a′ and a″), it was removed partially from the remnant podosomes in the presence of SU6656 (Fig. 5B, see b′ and b″ panels). Rather, FLNa accumulated with F-actin as disorganized patches at the cell periphery (Fig. 5B, b panel). This result suggests that the stability of podosomes could require a proper FLNa localization, which is controlled, at least in part, by Src kinases.

FIGURE 5.

Src kinase activity is required for podosome formation and for filamin A localization to podosomes. MDMs plated on fibrinogen (A) and treated with SU6656 (B) were stained for hFLNa and F-actin, before acquisition of confocal micrograph series (a and b, respectively) (z-step = 0.1 μm) (scale bar, 10 μm). a′ and b′ show the average of the F-actin and FLNa fluorescence staining of at least 100 podosomes from control and SU6656-treated cells (scale bar = 1 μm). a″ and b″ show fluorescence intensity profiles of the averaged podosomes along the white dashed line in (a′) and (b′), respectively. C, quantification of human MDMs with podosomes or podosome rosettes when seeded on coverslips that were either uncoated or coated with fibrinogen (Fg) and treated with the Src inhibitor SU6656 (mean ± S.D. of three independent experiments). D, RAW264.7 macrophages or stably expressing a shRNA against mouse Hck were (or not) transiently transfected with a human Hck-GFP coding vector (to rescue Hck) and were transfected transiently with LifeAct-mCherry coding vector to stain F-actin. Podosome lifespans, measured by time-lapse microscopy, are plotted for each cells type (mean ± S.D. of three independent experiments, five to 10 podosomes analyzed per cell in at least three cells per experiment).

To further examine the role of Hck in the process of podosome formation, RAW264.7 macrophages, in which Hck is specifically down-regulated via stable transfection of shRNA, were used (19). The percentage of cells with podosomes was reduced (21.9 + 6.1% versus 65.3 + 13.5%, mean ± S.D., n = 3), and cells formed fewer podosome rosettes (1.1 + 0.6 versus 10.4 + 3.8%, mean ± S.D., n = 3). Similarly to what had been observed for FLNa (Fig. 3D), podosomes also had a shortened lifetime in shHck-treated cells (Fig. 5D), indicating that, similar to FLNa, Hck is involved in the stability of podosomes. As above in the SU6656 experiments, FLNa was removed partially from the ring of the podosome remnants (data not shown). When Hck depleted mouse macrophages were complemented by human Hck, podosome lifespan was restored (Fig. 5D). In RAW264.7 macrophages expressing shRNAs against either Hck or FLNa, the degradation of the extracellular matrix was found to be reduced (supplemental Fig. S3), most likely as a consequence of the lower number of podosome rosettes formed in those cells. Thus, FLNa and Hck exhibit similar properties on podosome organization and function.

Finally, to determine whether Hck and FLNa also share similar properties in the migration mode of macrophages, the amoeboid migration mode of Hck^−/− BMDMs was examined. Indeed, although we reported previously that Hck^−/− BMDMs are defective in their mesenchymal migration (13), we had not examined their amoeboid migration. Thus, as performed in Fig. 1 for FLNa^−/− BMDMs, we placed Hck^−/− mouse macrophages into fibrillar collagen I, and no difference was observed between Hck^−/− and WT cells (23.0 + 5.7 versus 25.2 + 6.9, percentage of migrating macrophages, n = 5).

Taken together, these results show that FLNa and Hck (i) are two critical components of a signaling pathway leading to podosome rosette formation and extracellular matrix degradation, (ii) are required for podosome stability, and (iii) are essential for the protease-dependent mesenchymal migration but dispensable for the amoeboid mode.

DISCUSSION

Patients carrying defective genes for FLNa have a congenital malformation of the human cerebral cortex called periventricular nodular heterotopia (35, 36). Although the precise molecular mechanisms involved are not yet understood, this pathology has been shown to correlate with a defect in neuronal migration during brain embryogenesis (7, 37–40). The results of the experiments reported here support our working hypothesis that FLNa, which is a well known cross-linker of three-dimensional actin filament networks and a scaffolding protein, is involved in cell migration, and more precisely in the protease-dependent mesenchymal migration of macrophages moving in three-dimensional environments. We also confirm that FLNa is present at podosomes, which are cell structures involved in mesenchymal migration via adhesion and matrix degradation properties, and we have found that FLNa regulates podosome stability, their organization as rosettes and in three dimensions, and the extracellular matrix degradation activity of macrophages. All of these properties are shared with Hck.

Adhesion and migration of cells into tissues are critical processes for organ development, cell-mediated immunity, and wound healing. In vivo, cell migration takes place mostly in three-dimensional environments, which can differ markedly between tissues because of different composition, porosity, stiffness, and viscoelastic properties. In loose and porous environments into which cells can glide and squeeze to find their path in a protease-dispensable manner, macrophages preferentially use the amoeboid migration mode (12). In dense and poorly porous matrices, however, they have to adhere to proteins of the extracellular matrix via their integrins and use the protease-dependent mesenchymal mode to create their own path (12, 25, 41, 42).

The observation that FLNa^−/− macrophages have a reduced capacity for mesenchymal migration as they migrate normally using the amoeboid mode opens the way to the identification of the molecular mechanisms involving FLNa in cell migration. Although FLNa has been described as a regulator of cell migration in two-dimensional environments (9), this is the first report showing that FLNa is involved in three-dimensional cell migration.

In macrophages, proteases involved in matrix degradation are delivered at podosomes (13, 16), and recently, we have shown that the presence of podosome rosettes correlates with the mesenchymal migration of macrophages in three-dimensional environments but not with the amoeboid migration mode (12, 13, 25). Using different approaches to down-regulate FLNa, we show that the lifespan of podosomes, their organization as rosettes, the formation of three-dimensional podosomes when macrophages are embedded into the matrix, and the proteolytic activity on the extracellular matrix are all regulated by FLNa in macrophages. These observations could explain why the mesenchymal migration is altered specifically in FLNa^−/− macrophages. When podosomes are organized as a rosette, they form a structure related to the osteoclast sealing zone that exhibits an efficient degradative activity on the extracellular matrix (13, 22). The sealing zone consists of an array of podosomes communicating through a dense and interconnected network of actin filaments (43, 44). Because FLNa is located at the ring of individual podosomes and not at the core and as it regulates the F-actin density in the cloud (this work), it could be involved in the increased densification of the actin network interconnecting podosome cores, which occurs during rosette formation (43). Paxillin and vinculin also were found at rings and proposed to cooperate in a force-transfer process to the actomyosin complex (45). FLNa could be part of that complex linking integrins to actin to maintain the podosome structure and the degrading and protrusive activities of rosettes when cells penetrate in a three-dimensional environment. In fact, in addition to actin filaments, FLNa binds a large number of other proteins, many of which such as integrins are key players in cell adhesion and migration (4).

Similarly, Hck, a phagocyte-specific Src family tyrosine kinase, has been shown to be involved in the formation of podosome rosettes, in the proteolytic degradation of the matrix, and in mesenchymal migration but has no apparent role in the amoeboid migration mode and two-dimensional cell migration (Ref. 13 and this work). In fibroblasts expressing constitutively active Hck ectopically, the formation of podosome rosettes occurs spontaneously, and this was inhibited by knocking down the expression of FLNa. In macrophages, podosome rosettes also are formed poorly when Hck or FLNa are knocked down, in which case overexpressing Hck or FLNa can rescue podosomes. The organization of podosomes as rosettes triggered by Hck (13, 34) requires Rho, Rac, and Cdc42 (34), three proteins known to interact with FLNa and involved in the process of actin polymerization (9, 46). Interestingly, defective activation of Rac, Cdc42, and Rho in monocytes from the FLNa conditional knock-out mice used here has been reported (9). Taking these findings together, FLNa and Hck could belong to a common signaling pathway involving Rho GTPases. The observation that, in vitro, FLNa is phosphorylated by Hck (supplemental Fig. S4) is consistent with this hypothesis.

The amoeboid movement is a push-and-squeeze type of migration, which helps cells to find their way into porous matrices (47). In such cells, which are poorly adhesive because integrins are not involved, amoeboid migration is driven by RhoA/ROCK-mediated actomyosin contractions (47). Although FLNa has been reported to interact with ROCK (48), we found no evidence that FLNa is involved in the amoeboid movement, suggesting that this property of FLNa is not critical for this migration mode.

The interaction of FLNa with some of its effectors is potentially regulated by mechanical forces (7). In epithelial cells, interaction between FLNa and β1 integrin forms a mechanosensitive complex that can sense the tension of the matrix bidirectionally and, in turn, regulate cellular contractility/morphogenesis and respond to this matrix tension (49). The mechanical forces exerted on macrophages when they infiltrate a dense matrix could be the mechanism directing the cells to use mesenchymal rather than amoeboid migration. In addition, the mechanical forces exerted at the level of podosomes, where integrins and FLNa accumulate, could be contributing to the maintenance of these structures.

In macrophages and monocyte-derived cells such as dendritic cells and osteoclasts, podosomes are constitutive cell structures, but they also can form transiently in other cell types, such as smooth muscle cells stimulated with phorbol esters (14, 50) or endothelial cells stimulated with TGFβ (15). Interestingly, during the process of vascular repair that implicates endothelial cell migration, formation of podosome rosettes also occurs (51), and it recently has been proposed that podosomes may play a role during cell movement in embryogenesis (52). If we consider this hypothesis, it is conceivable that podosome defects could account for the malformations of brain, blood vessels, and several other organs observed in FLNa-null organisms (7, 53). Conversely, as no particular immune disorder has been described in FLNa-null organisms, the partial (∼50%) defects in macrophages mesenchymal migration does not appear sufficient to initiate immune troubles. In support of this, Hck^−/− mice, which also exhibit partial defects in mesenchymal macrophage migration (13), have no apparent immune disorder either. In contrast to the innate immune response, which involves huge numbers of phagocytes moving toward infectious sites, cell migration during embryogenesis generally involves a limited number of cells moving at a precise time of the embryonic development. Under these conditions, migration defects might be more harmful for the organism.

In conclusion, FLNa and Hck are the only two proteins described to date as being involved in mesenchymal migration of macrophages and not in their amoeboid mode. Both proteins regulate the stability of podosomes and their organization as rosettes, which are cell structures involved in mesenchymal movement. In contrast to Hck, which is only expressed in myeloid cells, FLNa is ubiquitously distributed, and null mutations have strong consequences on embryonic development. Our data thus strongly support the hypothesis that FLNa could be involved in mesenchymal migration of embryonic cells, which could in turn explain, at least in part, the organ defects observed in FLNa-null patients.