Lina Badimon, Cardiovascular Research Center, C/Sant Antoni Mª Claret 167, 08025 Barcelona, Spain. Tel.: +34 93 556 58 80; fax: +34 93 556 55 59. E-mail:firstname.lastname@example.org
Summary. Background: Tissue factor (TF) is the most relevant physiological trigger of thrombosis. Additionally TF is a transmembrane receptor with cell signaling functions. Objectives: The aim of this study was to investigate TF subcellular localization, function and signaling in human coronary artery smooth muscle cell migration. Methods: Coronary arteries and primary cultures of vascular smooth muscle cells (HVSMC) were obtained from human explanted hearts. Wound repair and Boyden chamber assays were used to measure migration in vitro. TF-pro-coagulant activity (TF-PCA) was measured in extracted cellular membranes. Analysis of TF distribution was performed by confocal microscopy. A nucleofector device was used for TF and protease activated receptor 2 (PAR2) silencing. mRNA levels were analyzed by RT-PCR. Results: In migrating HVSMC TF translocates to the leading edge of the cells showing an intense patch-like staining in the lamellipodia. In the migrating front TF colocalizes with filamin (FLN) in the polarized lipid rafts. TF-PCA was increased in migrating cells. Silencing of the TF gene inhibits RSK-induced FLN-Ser-2152 phosphorylation, down-regulates CDC42, RhoA, and Rac1 protein expression and significantly inhibits cell migration. Silencing PAR2 also inhibits cell migration; however, silencing both TF and PAR2 induces a significantly higher effect on migration. Smooth muscle cells expressing TF have been identified in non-lipid-rich human coronary artery atherosclerotic plaques. Conclusions: TF translocates to the cell front in association with cytoskeleton proteins and regulates HVSMC migration by mechanisms dependent and independent of factor (F)VIIa/PAR2. These results extend the functional role of TF to smooth muscle cell trafficking in vessel wall remodeling.
Tissue factor (TF) is currently considered the single most relevant physiological trigger of blood coagulation in response to vascular injury [1–3]; however, TF has also shown to be involved in various cellular processes associated with vascular smooth muscle cell function [4,5]. Under physiological conditions TF antigen is not detected in the intima or media of human vessels  with its expression restricted to the adventitia [7,8]. However, TF is widely found in atherosclerotic lesions associated with the lipid core and infiltrated inflammatory cells . Vascular smooth muscle cells have been shown to express TF upon induction by growth factors, vasoactive agonists and clotting factors [4,10]. The plasticity of VSMCs for undergoing phenotypic changes is well known. In vascular remodeling processes, from atherogenesis to restenosis, VSMCs migrate from the media to the intima in a process that involves gene reprogramming and protein synthesis. Filamin A (FLN) is essential for mammalian cell locomotion. FLN is important in the stabilization of the cortical skeleton of cells and might influence the processing and expression of many membrane proteins. Previous studies reported that cells lacking FLN have impaired locomotion [11,12].
TF, a member of the cytokine receptor family, is a 47-kDa glycosylated transmembrane protein [13,14] whose activity is modulated by a variety of factors, including its membrane domain association, the phosphatidylserine transmigration to the outer leaflet of the plasma membrane [15,16] and the sphingomyelin levels . TF activity is also modulated by the redox state of its membrane proximal cysteine pair (Cys186/Cys209) , as oxidation of this cysteine pair increases the procoagulant activity of TF. The extracellular domain of TF binds the multi domain coagulation serine protease factor (F)VII with sub-nanomolar affinity leading to its allosteric activation (FVIIa)  and to the resulting TF–FVIIa complex that activates FIX  and FX . Ligand binding not only initiates the coagulation cascade but also induces cell signaling through the cytoplasmic domain of TF by activating G-protein coupled protease-activated receptors (PAR) [21,22]. PAR activation causes the mobilization of cytosolic Ca2+ and induction of the MAPK signal transduction cascade. Moreover, phosphorylation of the cytoplasmic tail of TF by protein kinase C  allows interaction with FLN . In a mice model with a targeted deletion in the cytoplasmic tail, a role for TF in vascular remodeling was reported . However, the cytoplasmic domain is not required for TF-activated protein synthesis  or for FVIIa-dependent activation of the p44/42 MAP-kinase pathway . Cellular TF is a highly regulated molecule owing to its potent effects on hemostasis, and is present in three cellular pools: as surface, encrypted and intracellular protein [28,29].
The mechanisms that control TF subcellular localization during cell migration and TF signaling functions in adult human coronary VSMCs are not yet fully understood. We have hypothesized that TF is involved in vessel wall remodeling by modulating cell migration in the early process of VSMCs activation.
Materials and methods
Primary cultures of human vascular smooth muscle cells (HVSMC) were obtained from coronary arteries of explanted hearts. The role and subcellular localization of TF in HVSMC was analyzed by confocal microscopy. Furthermore, TF-pro-coagulant activity was measured in extracted cellular membranes. A nucleofector device was used for TF and/or PAR2 silencing (siRNA). mRNA levels of TF and PAR2 were analyzed by RT-PCR, and chemotactic migration of cells was measured in a modified Boyden chamber. Additionally, in TF siRNA HVSMC filamin, phospho-filamin (ser 2152), RSK, phospho-p90ser380, CDC42, Rac1 and RhoA were evaluated by western blot. Coronary arteries were obtained from hearts explanted and analyzed by conventional and confocal microscopy. See Data S1 for expanded methods.
TF distribution and mobilization in static HVSMC
TF mRNA was barely detectable in quiescent cells (Fig. S1A). Indeed, in isolated permeabilized quiescent coronary artery derived HVSMC, intracellular TF levels were very low and localized around the cell nucleus showing a vesicular pattern (Fig. 1A). Higher TF levels were observed at cell edges, frequently in cell–cell contact sites often concentrated in lamellipodia (see dotted arrows), with a low colocalization with F-actin. Colocalization analysis revealed a weak TF/F-actin colocalization rate (CR) of 10% (overlap coefficient 0.4, and Pearson’s correlation 0.4). When quiescent HVSMCs were stimulated with PDGF-BB (5 μg mL−1) TF expression increased at border edges of cells and in lamellipodia (Fig. 1B, see dotted arrows). Colocalization studies showed that the TF/F-actin signal at these localizations had a CR of 31% with an overlap coefficient 0.7, and Pearson’s correlation of 0.6 (10% control vs. 31% PDGF-BB treated HVSMCs, P < 0.05).
TF subcellular localization in migrating HVSMCs
After 12-h incubation with 10% FCS confluent HVSMC showed increased TF mRNA levels (time of wounding of the HVSMC monolayer, defined as migrating time 0 h in Fig. S1). At this time point TF mRNA expression levels were about 64% higher than those observed in quiescent HVSMC. TF mRNA expression levels did not show any significant further change in migrating cells (at 4, 8 and 16 h) (Fig. S1A).
Four hours after wounding, cells at the migratory front were spread and extended lamellipodia into the wounded area. Cells showed polarized morphology with lamellipodia localized only at the front and not at the side or rear edge (Fig. 1C, see dotted arrows). An examination of TF staining in permeabilized cells showed that at the migrating front TF expression was almost entirely located in the leading edge of the cells with intense patch-like staining in the lamellopodias and within ring-like structures that colocalized with F-actin (Fig. 1D, see dotted line). These ring-like structures differed from those observed in static cells because of their high colocalization with F-actin ends (CR 77%, overlap coefficient: 0.80, Pearson’s correlation: 0.7), as shown in xyz cytofluorogram (Fig. 1E) of the xzy sections obtained at the doted line in Fig. 1D (Fig. 1F).
To confirm the structural relationship between TF and the F-actin cytoskeleton we used cytochalasin E (cytE) to inhibit actin polymerization and abrogate microtubule formation. Treatment of HVSMCs with 10 μm cytE for 1 h abolished migration directionality and TF/F-actin colocalization, whereas TF antigen was unaffected. A representative image of the effect of cytE on HVSMCs is showed in Fig. 1G.
In order to verify whether TF protein levels were modified in migrating cells, western blot analysis was carried out. Blots confirmed the increase (85%) in TF protein expression in HVSMC (with 10% FCS) compared with quiescent cells (P < 0.05). Therefore in the repair model, migrating cells showed a TF-subcellular translocation to the migratory front, without significant changes in the amount of total protein (Fig. 1H). To confirm an interaction between TF and actin, a TF immunoprecipitation assay was performed. Results show that the β-actin–TF interaction significantly increased (over 60%) in migrating cells (Fig. 1I). To complete the TF system assessment procoagulant activity analysis (TF-PCA) of quiescent, quiescent with PDGF-BB and migrating HVSMC (4 h) was performed. TF-PCA was significantly increased in migrating vs. quiescent HVSMC (P < 0.05); PDGF-BB only induced a slight increase (not significant) over control cells (Fig. 1J).
In quiescent cells, TF was consistently colocalized within the Golgi system (colocalization with Golgin-97) with low expression in other localizations (Fig. 2A). Brefeldin A treatment, which has been reported to inhibit protein transport from the endoplasmic reticulum and to cause disassembly of the Golgi complex , abrogates colocalization confirming the TF localization in the Golgi complex (Fig. 2B). The migrating cells exhibit a polarized morphology that also affects the size and realignment of the Golgi towards the direction of the cell movement. TF colocalizes with golgin-97 in the perinuclear compartment of these cells (Fig. 2C).
TF is in lipid rafts in the leading edge of migrating cells
In order to investigate how migration modulates the surface localization of TF, we studied cells using markers of both caveolar and non-caveolar lipid rafts. Glycosylphosphatidylinositol-anchored proteins (GM1), a ganglioside that can be cross-linked and detected by labeled pentameric cholera toxin B, was used to analyze whether TF was localized in cholesterol-rich membrane microdomains.
In static non-permeabilized (to evidence membrane-associated proteins) cells, lipid raft staining of the cell surface showed a dotted pattern all over the cell (red). TF showed a punctuate pattern with a random distribution (green). Co-localization analysis revealed that TF levels on the cell surface were low but localized within lipid rafts appearing with a random punctuated pattern at the cell edges (Fig. S2A, see dotted arrows). CR was 52% (overlap coefficient: 0.85, and Pearson’s correlation of 0.6) (Fig. S2A).
In migrating (non-permeabilized) cells, lipid raft staining and TF expression showed a high colocalization at the cell boundaries and specifically at the leading edges (Fig. S2B, see dotted arrows). Computer-assisted colocalization analysis showed a high CR of 93% (overlap coefficient of 0.9 and Pearson’s correlation of 0.89) (Fig. S2B). This represents an almost two-fold increase in co-localization compared with static HVSMCs (52% vs. 93%; P < 0.05). The technique for observing lipid rafts in non-permeabilized cells requires cross-linking of membrane components by cholera toxin B. It was described that this chemical process changes the properties of the lipid bilayer and can alter to some extent the exact localization of proteins and lipids . For this reason TF distribution over migrating cells shows a slightly different distribution from that observed in permeabilized cells (Fig. S2B).
In static HVSMCs caveolin-1 was localized in different regions of the cell with a punctuated pattern (non-permeabilized cells). Overlapping images were analyzed to determine the relative colocalization with TF. TF/caveolin-1 colocalization was mainly found in the perinuclear area of the cell. The quantitative colocalization analysis in a scatter plot show negligible results (overlap coefficient: 0.2, Pearson’s correlation: 0.1) (Fig. S3A).
In migrating HVSMCs the punctuated pattern of caveolin-1 appeared in the cell surface and it was especially visible in the peripheral perinuclear regions of the cell. However, it did not appear in the leading edge of the cell where TF staining was concentrated. TF/caveolin-1 colocalization was found all in the rear edge of migrating cells. The quantitative colocalization analysis showed that only about 22% of the two fluorescent signals colocalized (overlap coefficient at this points: 0.5, Pearson’s correlation: 0.5) (Fig. S3B, see dotted arrows).
In additional studies, we evaluated the effect of lipid raft/caveolae disruption on TF localization in quiescent non-permeabilized cells. Cholesterol depletion (mβCD) from the plasma membrane reduced both TF/caveolin expression and abrogated colocalization, without affecting survival. mβCD treatment almost abolished caveolin staining and the TF staining pattern appeared over the entire cell surface. Overlapping images were analyzed to determine the relative colocalization of the two signals as represented in a scatter plot. The quantitative co-localization analysis showed that < 3% of the two fluorescent signals were detected with spatial coincidence (overlap coefficient: 0.80, Pearson’s correlation: 0.3) (Fig. S3C). Filipin altered the physical distribution of cholesterol in the membrane by forming filipin–cholesterol (membrane) complexes. TF/caveolin-1 co-localized in some patches on the HVSMC membranes. Filipin-treated cells showed minimal colocalization (overlap coefficient: 0. 5, Pearson’s correlation: 0.6) of TF/caveolin-1 (Fig. S3D, see dotted arrows). Therefore TF in migrating cells is associated with lipid rafts and only minimally to caveolin-1.
TF co-localizes with Filamin at the leading edge of migrating cells
We next investigated whether the high-density TF expression observed in the llamellipodia of migrating cells was associated with FLN. Indeed, in the migrating front there was a high colocalization of FLN with TF just in the plasma membrane of the elongated lamellipodia, and at the adhesion contacts sites, such as typical stress fibers (Fig. 3A, see arrows). TF/FLN colocalization had a CR of 47% (overlap coefficient 0.80 and Pearson’s correlation 0.6) (Fig. 3B). A TF immunoprecipitation assay was performed to confirm the FLN–TF interaction in migrating cells. Results showed that migration (4 h after wounding) did not modify the FLN–TF interaction (Fig. 3C). A TF immunoprecipitation assay was performed in migrating vs. confluent HVSMC to confirm the p-FLN–TF interaction. Results showed that migration (4 h after wounding) increased the p-FLN–TF interaction (Fig. 3D).
TF co-localizes with PAR-2 at the leading edge of a wound
To evaluate if PAR2 was associated with TF in the cell surface of migrating cells, PAR2 distribution and localization was also analyzed. Interestingly migrating cells showed an elevated signal for PAR-2 that was polarized on the lamellopodias of the migrating cells (Fig. 3E), with high co-localization with TF (CR 65% and overlap coefficient: 0.90 and Pearson’s correlation of 0.7) (Fig. 3F). A TF immunoprecipitation assay was done to confirm the PAR2–TF interaction in migrating cells. Results showed that migration (4 h after wounding) increased the PAR2–TF interaction over 350% (Fig. 3G).
TF and/or PAR2 silencing inhibits HVSMC migration
After finding the remarkable translocation of TF and PAR2 to the HVSMC migrating front we performed silencing experiments in order to evidence whether TF and PAR2 had functional effects driving cell migration. Specific silencing of TF (full length and alternatively spliced TF form ) and PAR2 gene expression with specific TF and PAR2 siRNA significantly reduced the expression of their respective endogenous transcripts (TF siRNA cells: TF 12% ± 1.0; PAR2 siRNA cells: PAR 9% ± 4; and in both TF and PAR2 siRNA cells: TF 8% ± 3 and PAR2 9% ± 4) (Fig. 4 A–C). Scrambled siRNA did not produce any effect on TF or PAR2 expression (95.8% ± 2.7) vs. non- treated HVSMC. TF-silenced HVSMC showed a 55% inhibition in migration (P < 0.05) (Boyden chamber assay), PAR2-silenced HVSMC showed a 50% inhibition in migration (P < 0.05) (Boyden chamber assay), whereas both TF and PAR2 silencing induced a significantly lower inhibitory effect reaching 80% inhibition in HVSMC migration (P < 0.05) (Boyden chamber assay), as shown in Fig. 4D.
The effect of silencing TF on the subcellular localization of FLN was analyzed by immunostaining. FLN was predominantly associated with filaments in the center of the cell with a lower intensity in the cell contact sites. F-actin localization was not altered by TF silencing (Fig. 4E–H). While levels of FLN protein expression were not affected, FLN phosphorylation (p-FLN) was significantly decreased in the TF-silenced cells (Fig. 5A–B). Indeed, FLN phosphorylation on Ser-2152 was reduced to < 10% in TF siRNA cells. These results indicate that TF-dependent signaling is involved in FLN phosphorylation (Fig. 5A–B). Analyzes of PAR2-silenced cells and TF/PAR2-double-silenced cells by western blot showed that FLN protein expression was not affected, and FLN phosphorylation (p-FLN) was significantly decreased in PAR2 and TF/PAR2-depleted cells (Fig. S4A).
FLN is a membrane-associated ribosomal S6 Kinase (RSK) substrate and RSK phosphorylates FLN on Ser2152 in vivo. Western blot of RSK and phospho(P)-p90rsk (ser380) showed that P-p90 RSK/RSK levels were reduced to < 20% in TF siRNA cells (Fig. 5C). Analyzes of PAR2 silenced cells and TF/PAR2-double-silenced cells by western blot showed that RSK protein expression was not affected, and phospho(P)-p90rsk (ser380) was significantly decreased only in TF/PAR2-depleted cells (Fig. S4B).
TF silencing down-regulated Rho proteins family
The Rho GTPase family member CDC42 regulates filopodia formation, the polarization of migrating cells and the stabilization of microtubules, at the leading edge of migrating cells. TF siRNA significantly reduced CDC42 protein expression (by 80%) (Fig. 5D). Rac-1 a downstream effector of CDC42, involved in the regulation of lamellipodia and membrane ruffling formation, was significantly reduced (by 50%) in TF siRNA cells (Fig. 5E). A RhoA protein level, which regulates formation of stress fibers and focal cell adhesion, was also significantly reduced (over 50%) in TF-silenced cells (Fig. 5F). mRNA levels of CDC42, Rac-1 and RhoA were not modified.
TF expression in atherosclerotic lesion in human coronary arteries
Coronary arteries were obtained from hearts explanted because of non-ischemic dilated cardiomyopathy. These arteries had early atherosclerotic lesions (intimal thickenings) that did not show detectable levels of TF (N = 5) (Fig. 6 A,B). However, TF appears in coronary arteries obtained from hearts explanted because of ischemic heart disease. These arteries have atherosclerotic plaques rich in VSMC in their hyperplasic intimas, plaques classified as intermediate (AHA type II/III) (N = 5). Double immunolocalization studies showed that TF expression was found in atherosclerotic lesions co-localizing with VSMC (alpha-smooth muscle [α-SM] actin marker) (Fig. 6C–E). This colocalization is shown in plaques without inflammatory infiltration to exclude the confounding effect of macrophage-derived TF present in lipid-rich plaques.
TF is the most important trigger of arterial thrombosis in atherosclerotic plaque rupture . In arterial injury models, TF is markedly induced in medial VSMCs and accumulates in the VSMCs of the developing intima .
Here we report that the TF subcellular distribution is significantly modified in migrating HVSMC. Migration induces a high interaction between TF and actin, PAR2, and p-FLN. TF-silencing induces a highly significant inhibition of cell migration mediated by Ser2152-FLN phosphorylation through RSK kinase.
Previous results in our group have shown that abrogation of TF expression in endothelial and/or VSMC cells was associated with reduced ‘angiogenesis’ in matrigel plugs implanted into mice. Although silencing TF in HVSMC reduced the formation of the microvessel network, the TF-dependent effects in HVSMC migration were not elucidated .
Confocal microscopy has enabled us to study the cellular migration dynamics and TF polarization patterns on the motile cells. TF is expressed in low levels in static cells appearing in small contact sites predominantly placed within lipid rafts with a small fraction allocated in caveolin-rich areas. Intracellular staining is low and localized around the nucleus in Golgi structures awaiting a signal for exocytosis to the plasma membrane. PDG-BB, which activates the small G protein Rac, and triggers actin polymerization at the cell periphery to produce lamellipodia , induced an increased staining of TF at border edges of cells and in lamellipodia linked to F-actin. Shortly after stimulation of quiescent cells, TF is mobilized to the cell surface. This may be a mechanism of inducing a prothrombotic phenotype in stimulated HVSMC.
Smooth muscle cells from mTF−/−/hTF+ mice exhibited a marked defect in cell migration in a modified Boyden chamber assay, suggesting that the defect in TF was involved in regulating mice smooth muscle cell migration. In migrating human coronary smooth muscle cells TF cellular distribution rapidly changed. In the migrating cells TF is relocated and transported to the front leading edge of the cell, and also in the directionally reoriented Golgi structures. Indeed, we detected up-regulated TF expression in the non-lipid-rich intima of human atherosclerotic coronary arteries with intermediate lesions co-localizing with migrating smooth muscle cells.
Cytoskeletal reorganization is a requisite for leading edge cellular protrusion during locomotion, and VSMC migration requires such rearrangement. Rho GTPase family member CDC42 is a master regulator of cell polarization that influences directional migration and acts to stabilize microtubules at the leading edge of cells moving towards a cell denuded region. In migrating cells the microtubule-organizing center (MTOC) reorients towards the leading edge. It was reported that MTOC reorientation repositions the Golgi towards the front of the cell and contributes to directional migration . This reorientation induces the formation of stabilized microtubules at the leading edge that in our studies with Golgi reorientation seems to help mobilize intracellular TF from the Golgi, to increase TF expression at the cell surface and to drive TF to the leading edge of the migrating cells. Indeed, CDC42 regulates MTOC reorientation  and membrane traffic , and we have seen a significant down-regulation of protein levels of CDC42, Rac1 and RhoA in the TF-silenced cells that show a significantly inhibited motility.
Cholesterol removal impairs TF localization in lipid rafts/caveolin-1; however, caveolin was not associated with TF translocation to the front edge of the migrating cells. An association of TF to lipid rafts allowed its polarization to the leading migrating front. Interestingly a cholesterol-TF positive high-affinity state was previously reported . Sequestration of TF and PAR2 in lipid rafts may facilitate their interaction and could activate signaling pathways. TF caveolae localization (peripheral perinuclear regions in migrating cells) might function as a latent pool, which can rapidly be activated at sites of injury. Our data suggest that localization of TF in membrane lipid rafts is essential to promote the interaction of TF directly or indirectly with microfilament proteins and mediate cell migratory functions.
Because many of the cellular effects of TF involve cytoskeleton reorganization, we analyzed whether FLN polarized to the migrating cell leading edge with TF. FLN is essential in CDC42-induced llamellipodia formation and is a key protein in integrin-mediated cell signaling [40,41]. Our results suggest that TF and FLN are closely connected directly or through integrins, and that this relationship is enhanced by migratory stimuli. The attachment of integral membrane proteins to the cytoskeleton serves as a retention signal and in siRNATF-HVSMCs FLN phosphorylation (ser2152) was drastically diminished. Consistent with our results, different reports show that phosphorylated forms of FLN play an important role in cell migration . Many kinases including PAK, CaM kinase II and PKC can phosphorylate FLN in vitro , but RSK phosphorylates FLN on Ser 2152 in vivo . Our results suggest that TF is linked to this kinase. Many proteins have been reported to interact with FLN: two of these binding proteins are Rho family GTPases and β integrins. We have found diminished CDC42, Rac-1 and RhoA levels in TF siRNA HVSMC showing a close relationship between p-FLN/CDC42/Rac-1/RhoA/TF in the leading edge of migrating cells. An increase in migratory cell response dependent on the cytoplasmic TF-domain through MAPK p38 and Rac1 has also been previously described . However, further studies are needed to understand the mechanisms behind the increased phosphorylation of FLN during migration and the possible role of the TF-cytoplasmic domain in the process. Our preliminary unpublished data show that partial deletion of the TF cytoplasmic domain  does not affect migration and does not induce FLN recruitment; however, additional studies are needed to fully underline the TF–FLN interaction in the migration process.
Consequently, filamin seems to be involved in the integration of external stimuli and functions as a platform that can link to downstream effectors such as small GTPases. Interestingly, silencing TF and PAR2 significantly inhibited HVSMCs migration in the Boyden chamber assays. However, the simultaneous silencing of TF and PAR2 in HVSMCs produced a further reduction in migration that was statistically significant. These results suggest that both proteins contribute to the regulation of cell migration.
Cell polarization results in the specific localization of proteins and lipids to different membrane domains . Silencing TF in HVSMC leads to a decrease in FLN phosphorylation (mediated through RSK), which negatively regulates cell polarization and downstream signaling including CDC42, Rac and RhoA, leading to decreased cell motility.
In summary, our findings show that TF is a central molecule not only in triggering thrombotic responses to repair luminal vascular damage but also in the remodeling of the vascular wall by regulating HVSMC migration.
The authors are grateful to Oriol Juan, Monica Pescador and Olaya Garcia for expert technical assistance.
This work was supported in part by grants from Ministry of Science and Education of Spain (SAF 2010/16549) and Instituto de Salud Carlos III (CIBEROBN-CB06/03) (to L.B.) and (CP07/00224) (to G.A.). We would also like to thank Fundación de Investigación Cardiovascular and Fundación Jesus Serra for their continuous support.
Disclosure of Conflict of Interest
The authors state that they have no conflict of interest.