Podosomes are specialized plasma-membrane actin-based microdomains that combine adhesive and proteolytic activities to spatially restrict sites of matrix degradation in in vitro assays, but the physiological relevance of these observations remain unknown. Inducible rings of podosomes (podosome rosettes) form in cultured aortic cells exposed to the inflammatory cytokine TGFβ. In an attempt to prove the existence of podosomes in living tissues, we developed an ex vivo endothelium observation model. This system enabled us to visualize podosome rosettes in the endothelium of native arterial vessel exposed to biologically active TGFβ. Podosomes induced in the vessel appear similar to those formed in cultured cells in terms of molecular composition, but in contrast to the latter, arrange in a protruding structure that is similar to invadopodia. Local degradation of the basement membrane scaffold protein collagen-IV, is observed underneath the structures. Our results reveal for the first time the presence of podosome rosettes in the native endothelium and provide evidence for their capacity to degrade the basement membrane, opening up new avenues to study their role in vascular pathophysiology. We propose that podosome rosettes are involved in arterial vessel remodeling.
Podosomes are specialized plasma-membrane actin-based microdomains consisting of a core of actin filaments associated with the Arp-2/3-based actin polymerization machinery and surrounded by a ring of vinculin, talin, paxillin and integrins. They can be distinguished from other focal adhesions complexes by the presence of `podosomal markers', such as gelsolin, cortactin, dynamin 2 and WASP/NWASP proteins (Gimona et al., 2008; Linder and Aepfelbacher, 2003). Therefore, double staining for F-actin and cortactin is routinely used to identify podosomal structures. In addition, podosomes are enriched with metalloproteases, which endow them with matrix-degradation activities. In physiological settings, podosomes form spontaneously in certain cells such as macrophages and immature dendritic cells, which share the common feature of travelling across tissue boundaries. Podosomes can also be seen in non-myeloid cells, where they are induced upon stimulation with specific receptor ligands and/or second messenger stimulators (Hai et al., 2002; Osiak et al., 2005). Podosome-related structures named invadopodia assemble in cultured cells transformed by the viral src oncogene and in melanoma or carcinoma in response to oncogenic signals (Buccione et al., 2004; Stylli et al., 2008).
We have established that TGFβ stimulates podosome formation in cultured bovine arterial endothelial (BAE) cells (Varon et al., 2006). In this model, podosomes always arise assembled together into large ring-shaped structures. TGFβ is a multifunctional cytokine that is secreted from cells in a latent form which is unable to bind the TGFβ receptor and is inactive (Lawrence et al., 1985). Primary regulation of TGFβ activity occurs through factors that control the processing of the latent form into the biologically active molecule. TGFβ is active on most cells and the specificity of the response is determined by expression levels of the various TGFβ receptors and co-receptors, as well as the TGFβ concentration. TGFβ first binds to type II receptors (TβRII), which then recruit a type I receptor (TβRI). Formation of the complex leads to TβRI kinase activation, which subsequently propagates the signal into the cell by phosphorylating members of the Smad family of transcription factors (Derynck and Zhang, 2003; ten Dijke and Hill, 2004).
Because TGFβ is deeply involved in the maintenance of vascular homeostasis during adult life (Goumans and Mummery, 2000), in particular in the arterial tree, we wondered whether podosomal structures assemble in vivo. Indeed, podosomes have only been described in cultured cells in vitro and convincing evidence for the existence of such structures in intact tissues is still lacking. Here, we show the presence of rosettes of podosomes in the native endothelium of arterial vessel segments exposed to the inflammatory cytokine TGFβ, but not in the normal resting endothelium. The structures formed in arterial vessel segments exposed to TGFβ strongly resemble those formed in TGFβ-treated cultured cells and the only difference found was in the topology and architectural organization of proteins in the rosettes, which might be accounted for by differences in terms of rigidity, roughness or molecular composition of the substratum. Importantly, in the context of the vessel, TGFβ-induced podosome rosettes exhibit collagen-directed matrix-degrading activities. Our results reveal for the first time the presence of podosome rosettes in the native endothelium, and open new avenues to study their role in vascular pathophysiology.
Detection of actin rings in the living endothelium
Endothelial cells in culture do not normally form podosomes. However, BAE cells assemble podosomes rosettes upon TGFβ exposure (Varon et al., 2006). In vitro studies indicate that TGFβ is activated upon contact between endothelial and smooth muscle cells in vitro (Flaumenhaft et al., 1993), suggesting that local activation of TGFβ regulates vessel stability in vivo. Since the endothelium is constantly exposed to TGFβ present in circulating blood or trapped in the underlying matrix, we wondered whether endothelial cells lining the interior surface of the aorta display podosomal structures. We developed a protocol to perform such screening in the aortic endothelium in mice. Tissues were fixed in situ through intracardiac injection of paraformaldehyde (ivc) in anesthetized animals. Next, vessel segments were harvested, cut along their long axis and stained for F-actin, cortactin, VE-cadherin or CD31 whilst nuclei were highlighted with Hoechst 33342 stain. The specimens were then subjected to confocal microscopy, with endothelial cells facing upward, for `en face' viewing (Fig. 1). F-actin and Hoechst staining distinguished between endothelial cells with large round nuclei (Fig. 1A) and vascular smooth muscle cells (vSMCs) with thin and elongated nuclei (Fig. 1B). Along the z-axis, the two cell types were seen separated by the first elastic lamina (Fig. 1C), as illustrated in Fig. 1D. VE-cadherin and CD31 staining was detected in the endothelial layer (Fig. 1E,F), but absent in the deeper layer of smooth muscle cells (Fig. 1B), clearly delimiting the cellular boundaries and revealing the integrity of the endothelium in the specimen. F-actin and cortactin were consistently concentrated at cell margins within the endothelium and never showed the characteristic arrangement of podosomal structures (Fig. 1G-H).
We therefore reasoned that, under normal physiological conditions, the bioavailability of TGFβ in the endothelium was too low to induce podosomes. To investigate further the existence of podosomes in tissues, we examined the effects of biologically active TGFβ concentrations on the aortic endothelium. Live aortic vessel segments were cultured ex vivo, in the presence or absence of exogenously added recombinant TGFβ1 (TGFβ) under conditions similar to those used to induce podosome rosettes in cultured BAE cells (Varon et al., 2006) (Fig. 2A-C). Aortic explants were then fixed, stained for F-actin and cortactin and prepared for en face viewing. In untreated samples, whereas some fields of view showed marker distribution that was similar to that obtained with aortic vessel segments fixed in vivo (supplementary material Fig. S1A-D), other fields of view showed less prominent cortactin and F-actin staining at cell-cell junctions, a pattern which persisted throughout the organ culture (supplementary material Fig. S1E-H). A similar staining pattern was observed in endothelia of TGFβ-treated samples (supplementary material Fig. S1I-L). However, careful examination revealed zones where F-actin and cortactin distinctly colocalized in ring-like structures within the endothelium of TGFβ-treated vessels (Fig. 2D-F), that were not detected in untreated specimens (supplementary material Fig. S1E-G). To assess endothelial cell responsiveness to TGFβ stimulation, aortic vessel segments were stained for canonical effectors of TGFβ (Lebrin et al., 2005). A twofold increase in phosphorylated Smad2 was detected in the cell nuclei of TGFβ-treated endothelial cells upon TGFβ exposure, attesting to efficient TGFβ signaling in this experimental setting (Fig. 3).
Molecular characterization of the structures formed in endothelium exposed to TGFβ
Further characterization of the ring structures revealed the presence of structural components such as vinculin and paxillin, signaling components such as FAK, phosphotyrosine-containing proteins and β3 integrin (which is an `activation' marker for endothelial cells), as well as genuine podosome markers (Linder and Aepfelbacher, 2003) Arp3, cortactin and dynamin 2 (Fig. 4A-C and F). We also detected MT1-MMP (Fig. 4D), MMP2 and MMP9 (supplementary material Fig. S3A,B) at the rosettes. N-WASp was distributed all along the structure, including its tip (Fig. 4E). This situation resembles that encountered in invadopodia, podosome-related structures assembled in cultured cells transformed by oncogenes (Baldassarre et al., 2006; Lorenz et al., 2004; Stylli et al., 2008). Interestingly, β1 integrins were distributed along the inner and outer rim of the podosome rosettes in vitro. With respect to integrin localization, the resemblance between ring-like actin structures in aortic vessel segments (Fig. 4G) and podosome rosettes in cultured BAE cells (Fig. 4H) is a convincing reason for considering that they are one and the same kind of structure.
Endothelial actin-rich VCAM+ docking structures that mediate adhesion of leukocytes to the endothelium have been described in inflammatory contexts (Barreiro et al., 2002). Here, the actin rings formed in response to TGFβ stained negative for endothelial (VCAM) or leukocyte (CD11b) adhesion molecules (data not shown), indicating that such endothelial figures are not generated by extravasating blood leukocytes undergoing transcellular diapedesis (Carman and Springer, 2004). Therefore, based on these criteria, we conclude that TGFβ induces genuine podosome rosettes in aortic vessel segments. Similar results were found with murine carotid artery segments processed in the same way (data not shown). Podosome rosettes were detected in less than 5% of arterial endothelial cells exposed to TGFβ for 24 hours.
Similarities and differences of podosomes in in vitro and ex vivo models
Interestingly, the endothelium that lines blood vessels represents a bi-dimensional space, suggesting that podosome rosettes formed in vivo will be found under the same sort of organization as those observed in vitro. Fig. 5A shows a rosette located on the basal side of the endothelial cell, in close apposition with the underlying substratum of the endothelium. Along the z-axis, the podosomal structure was seen to extend downwards, as if pushing in into the vSMC layer (Fig. 5B). The rosette formed in the endothelium displayed prominent cortactin staining at the distal extremity of the rosette, protruding through the smooth muscle cells (Fig. 5A,B). This organization of the podosomal components fits with the model of actin polymerization of podosomes, which is thought to proceed similarly to extending lamellipodia, but along the vertical axis (Linder and Aepfelbacher, 2003; Vignjevic and Montagnac, 2008). We conclude that these endothelial integrin-rich structures adhere to the underlying substratum. These findings confirm that the podosome rosettes depicted here constitute entities distinct from previously described actin structures that interact with circulating cells on the apical surface of the endothelium (Barreiro et al., 2002).
We next explored the spatial organization of podosomal proteins in endothelial rosettes from TGFβ-treated aortic vessel segments (Fig. 5C), cultured BAE cells (Fig. 5G) and in cultured BAE cells seeded onto a vessel where the endothelial monolayer, but not the basement membrane, had been removed (Fig. 5K). Rosettes formed in the intact endothelium or in BAE cells seeded onto the denuded aortic vessel segments were both found to be more compact and thicker along the z-axis than those formed on glass coverslips [outer diameter, 7.53±2.43 μm, n=11 (ex vivo); 9.17±2.67 μm, n=15 (BAE cells on denuded vessels); 16.13±2.39 μm, n=10 (in vitro)]. These findings suggest that the molecular composition of the underlying substratum rather than cell-cell interactions in the endothelium account for the differences observed between the in vitro and ex vivo characteristics of podosome rosettes.
Endothelial podosome rosettes formed in arterial vessel segments degrade the basement membrane collagen-IV
A specific feature of podosomes is their ability to degrade gelatin. Podosome rosettes formed in cultured TGFβ-treated endothelial cells fulfil this criterion, as measured by an in vitro gelatin-degradation assay (Varon et al., 2006). In vessels, endothelial cells rest on the basement membrane, which provides physical support for the cells and exerts a multitude of regulatory functions on their growth and differentiation state (Adams and Watt, 1993). Basement membrane consists primarily of laminin and type IV collagen bridged by non-covalent interactions with nidogens and perlecan (Kalluri, 2003; Rowe and Weiss, 2008; Yurchenco et al., 2004). To explore the degradative potential of endothelial podosome rosettes present in vessel segments, samples were stained with antibodies recognizing matrix proteins. Non-fluorescent patches were detected directly beneath the podosome rosettes in the endothelium of TGFβ-treated arterial vessel segments, when staining was performed with anti-collagen-IV antibodies (Fig. 6A-C). This pattern strongly suggests matrix protein degradation by proteolytic activities at podosome rosettes, in a manner analogous to that seen in podosome rosettes in BAE cells cultured on a collagen-IV matrix in vitro (Fig. 6F-H) or on the basement membrane of denuded vessel segments (supplementary material Fig. S2). Quantitative analysis revealed efficient matrix degradation beneath the rosettes formed in the endothelium exposed to TGFβ (Fig. 6D). Focal collagen-IV degradation at podosome sites was inhibited by the broad-spectrum MMP inhibitor GM6001 (Fig. 6E). Under similar conditions, laminin staining remained intact beneath the rosettes (Fig. 6I-K). This finding was confirmed in vitro in the BAE model. TGFβ-induced endothelial podosomes were unable to detectably alter the derived laminin coating (Fig. 6L,M). Likewise, thorough inspection of the elastic lamina did not reveal any alteration in its integrity at this stage (data not shown).
Podosomes have been extensively described in the artificial environment of cultured cells. Our findings demonstrate that podosome rosettes form in the living endothelium of large arterial vessel segments exposed to physiological levels of biologically active TGFβ. We also reveal their capacity to degrade the basement membrane. Our data provide the first evidence for podosomes in native tissues and support the concept that podosomes are physiological structures involved in biological processes in vivo. We believe that these findings represent a major advance in the field and will support researchers aiming to demonstrate the presence of invadopodia in cancer cells in vivo.
Imaging podosome rosettes in their native environment enabled us to disclose their distinctive features. Similarities and differences between podosomes and invadopodia, collectively called invadosomes (Machesky et al., 2008) are not yet definitive (Weaver, 2008). A major difference often reported between podosomes and invadopodia is the overall architecture of the microdomain. Whereas podosomes self-organize within the boundaries of the cell on a soft gelatin matrix, protrusive organization is the hallmark of invadopodia (Stylli et al., 2008). Our studies reveal the spatial organization of the components and the protruding architecture of endothelial podosome rosettes, providing evidence for similarities between the extensions formed by invadopodia of transformed cells (Baldassarre et al., 2006) and aggregated podosomes in rosettes of endothelial cells. The arrangement of the podosomal components in the structure is dependent on the substratum, suggesting a misguided organization on glass coverslips in in vitro settings. Our study thus describes a situation where the protrusive organization of invadopodia, rather than the non-protrusive organization of podosomes as described in vitro, vanishes in this in vivo context. Our results establish that the cell microenvironment and the experimental settings impact invadosome architecture.
A common characteristic feature of podosomes and invadopodia is their ability to degrade proteins of the extracellular matrix. We have defined the degradation capacities of podosome rosettes towards basement membrane proteins in their native environment. Matrix degradation measurements recorded upon 24 hours of exposure to TGFβ revealed collagen-directed matrix-degrading activities, which is consistent with the recruitment of MT1-MMP, MMP2 and MMP9 at the structures (Guegan et al., 2008; Tatin et al., 2006; Varon et al., 2006). Under these conditions, no alteration in the laminin pattern was detected. However, unlike the collagen-IV network, which is stabilized through covalent links, laminin polymers are not covalently associated, but are rather assembled together by the collagen-IV network (Yurchenco et al., 2004). It is the highly ordered and crosslinked nature of the type IV collagen network that confers basement membranes with their structural integrity. Therefore, disruption of the collagen bonds naturally disrupts the laminin network, and from there, the laminin-collagen-IV scaffold (Rowe and Weiss, 2008). In our experimental set-up, there is no laminin degradation, and the loosening of the laminin meshwork structure is not detectable. Since perforation of the basement membrane invadopodia takes several days (Hotary et al., 2006) and collagen-IV is the most abundant constituent of the basement membrane (Kalluri, 2003), we assume that our experimental setting enables us to detect the early events of a TGFβ-induced remodeling programme where collagen-IV degradation occurs at an early stage, loosening the laminin network and resulting in basement membrane splitting. Our results reveal, for the first time, the presence of podosome rosettes in the native endothelium and open new avenues to study their role in vascular pathophysiology.
Examination of the aortic and carotid arteries specimen fixed in situ revealed that, in the absence of stimulus, the endothelium of large vessels is normally devoid of podosome rosettes. These observations in vivo are consistent with the situation in vitro, where resting aortic endothelial cells show no podosomes. Thus, podosomes are expected to come into play in specific situations associated with local release of active TGFβ, such as those induced by disturbed blood flow or ischemia (Jones et al., 2009; Khalil, 1999; Sho et al., 2002; van Royen et al., 2002). Conditions associated with elevated levels of circulating TGFβ (Derhaschnig et al., 2002; Flores et al., 2004; Matt et al., 2009; Peterson, 2005) might also represent situations favouring podosome formation in the arterial vasculature. Testing for the physiological function of the structure is arduous and further studies are needed to place endothelial podosomes in their relevant context. Nevertheless, the proteolytic activity at sites of podosome rosettes targeting matrix components in the vessel provides some clues to their function. In our experimental setting, the number of podosome rosettes detected in each sample never exceeded 5%. This rare occurrence of podosomes might reflect a specific function played by a subset of endothelial cells, a situation encountered in the microvascular system where specialized endothelial tip cells lead the outgrowth of blood-vessel sprouts (Hellstrom et al., 2007). Alternatively, if endothelial podosomes do have a role in tissue remodeling, then application of active TGFβ on a normal resting endothelium might not represent a vascular context promoting podosome formation. Thus, the incidence of podosome rosettes remains to be explored in physiologically induced experimental settings, for instance, in a true inflammatory context associated with a pathological matrix, where intense vascular remodeling takes place, or in vascular diseases associated with increased TGFβ signaling in the vessel wall (ten Dijke and Hill, 2004). Podosomes could also be involved in vascular pathogenesis. For example, under non-resolved inflammatory conditions, exacerbation of the phenomenon might lead to tissue destruction. Probable consequences of the action of metalloproteases could be the emergence of sites prone to the initiation of arterial aneurysms or atheromatous plaque formation through deterioration of endothelium integrity.
Materials and Methods
Experiments were approved by the local ethics committee on the use of laboratory animals, and procedures were in accordance with INSERM institutional guidelines. Arterial vessel segments were derived from 6- to 8-week-old C57BL/6j mice (Charles River), which were anesthetized by subcutaneous administration of a mixture of ketamine (15 mg/kg, Merial) and xylazine (0.3 mg/kg, Bayer) in 100 μl in a sterile saline solution. After euthanasia, the carotid or aorta arteries were isolated. Examination of native tissues was performed after mice had received intracardiac injection (ivc) of a 4% paraformaldehyde solution continued by a 1 hour fixation step in vitro.
Vessel segments of aorta or common carotid arteries (length ∼6-8 mm) were freed of adipose and connective tissue and carefully handled, only at their outer ends without stretching them, to keep them viable. To avoid contact with air, they were kept moist during the whole preparation procedure (maximum 30 minutes) in culture medium (EGM-MV; Promocell, Heidelberg, Germany). Arterial vessel segments were longitudinally opened from the distal end then exposed to TGFβ for 24 hours, in EGM-MV medium at 37°C in a 5% CO2 atmosphere. Fixation was achieved by immersion (1 hour, RT) with 4% paraformaldehyde in PBS, permeabilized with 0.5% Triton X-100 for 1 hour, and blocked with 3% fetal bovine serum for 30 minutes (RT). The specimens were incubated with primary antibodies (overnight, RT), then with the appropriate secondary antibodies conjugated to Alexa Fluor-488, -546 or -633 (3 hours, RT). F-actin and nuclei were stained at this step with phalloidin conjugated to Alexa Fluor-546 or -488 and Hoechst 33342, respectively. Arterial vessel segments were spread then mounted on microscope slides with Fluoromount mounting medium. Denuded basement membrane arterial segments obtained by endothelial cell lysis upon exposure to hypotonic shock.
BAE cells (Lonza) were maintained in complete endothelial cell growth medium (EGM-MV; Promocell) at 37°C in a 5% CO2 humidified atmosphere and used between passages three and six (Varon et al., 2006).
Recombinant human TGFβ (TGFβ1, used at 5 ng/ml in all experiments) was obtained from R&D Systems and type IV collagen and laminin were obtained from Sigma. GM6001 was purchased from Calbiochem and Fluoromount mounting medium was from Clinisciences. Antibodies against the following proteins were obtained as indicated: cortactin (clone 4F11), Arp3, phosphorylated Smad2 (Ser465/467) and collagen-IV were from Millipore; Cdc42, paxillin, FAK, N-WASp from Cell Signaling; vinculin and phosphotyrosine (clone 4G10) from Sigma; laminin from Abcam; β3 integrin was from Emfret and β1 integrin from BD Biosciences; VE-cadherin was from MedSystems. Polyclonal antibodies directed against dynamin 2 were from Mark McNiven (Mayo Clinic, Rochester, MA). Anti-MT1-MMP is a monoclonal antibody (mAb-2) obtained from the same fusion as that described recently (Ingvarsen et al., 2008). MMP9 antibodies were from BioMol and those against MMP2 were from R&D systems. Alexa Fluor 546-phalloidin, Alexa Fluor 488-labeled secondary antibodies, fluorescein isothiocyanate (FITC) and Hoechst 33342 were purchased from Invitrogen.
Subconfluent cells grown on glass coverslips were prepared for immunofluorescence as previously described (Varon et al., 2006). The coverslips were washed in water and mounted on microscope slides with Fluoromount mounting medium. For matrix degradation assay, BAE cells were seeded on coverslips coated with collagen-IV (0.2 mg/ml) or laminin (20 μg/ml), and dark areas and podosome rosettes were visualized after staining with the relevant antibodies and Alexa Fluor 546-phalloidin, respectively.
Microscopy and image analysis
Cells and arterial vessel segments were analyzed by confocal imaging using a Zeiss LSM 510 inverted laser-scanning fluorescence microscope equipped with acquisition software (LSM 510 acquisition software; Zeiss) and a ×63 NA 1.4 oil-immersion objective. Triple-color imaging using Hoechst 33342, Alexa Fluor 488- or Alexa Fluor 633-labeled secondary antibodies, and Alexa Fluor 546-phalloidin was obtained using selective laser excitation at 350 nm, 488 nm, 633 nm and 543 nm, respectively. Each channel was imaged sequentially using the multitrack recording module before merging. Fluorescent images were processed with Adobe Photoshop 7.0. To make 3D images, serial optical sections of tissue samples or cells were captured with the software's automatic scanning mode. Z-stack digital images were collected optically at every 0.4 μm depth, saved and used for 3D reconstruction using Imaris software (Bitplane).
Measurement of matrix degradation underneath podosome rosettes in arterial vessel segments
Specimens were double stained for F-actin and for collagen-IV or laminin to detect podosome rosettes and basement membrane, respectively. Z-stack digital images were acquired, running through the rosette and first smooth muscle cell layer. Images were used for 3D reconstruction analysis using the LSM 510 acquisition software projection module. 3D images were saved and analyzed using ImageJ software after images had been converted to an eight-bit gray-scale image. For each rosette, the rosette area was delimited manually in the 3D F-actin image and this region was pasted on the same location on the 3D basement membrane image. In this 3D image, intensity profiles of regions of interest (rosette area) were measured and compared with intensity profile of regions in the rosette vicinity (control area). This measurement was repeated on 16 (collagen-IV) or 23 (laminin) podosome rosettes, and an average graph was drawn. For quantification of degradation with GM6001, `rosette area' and `control area' values were compared for each rosette. s.d. values of intensity profiles between rosette area and CT area determine the degradation potential of the rosette.
Data were compared between groups using Student's t-test. Differences were considered to be statistically significant at P<0.05.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/122/23/4311/DC1
We are grateful to Signe Ingvarsen and Lars Engelholm who provided the anti-MT1-MMP antibodies, and to Mark McNiven for those recognizing dynamin 2. We thank Steve Weiss, Bob Mecham, Brant Isakson, Rosemary Akhurst, Ellen Van Obberghen-Schilling, Ziad Mallat, Gareth Jones and Wolfgang Schaper for helpful discussions. We also wish to knowledge Alain-Pierre Gadeau and Claude Desgranges for expert help in the initial experiments and IJsbrand Kramer for critical comments on the manuscript. C.B. is a recipient of grant ANR-06-BLAN-0362 and her research is funded through this route. P.R. is supported by a predoctoral fellowship from the Region Aquitaine and INSERM. This work was supported by INSERM, ARC (3549), La Ligue Nationale contre le Cancer and Fondation de France and EU grant PITN-GA-2009-237946.
↵* Present address: Laboratory of Immunology, National Institute of Allergy and Infectious Diseases, NIH, Bethesda, Maryland, USA
- Accepted August 20, 2009.
- © The Company of Biologists Limited 2009