Podosome rosettes precede vascular sprouts in tumour angiogenesis

Abstract

Expansion of a vascular network requires breaking through the basement membrane, a highly crosslinked barrier that tightly adheres to mature vessels. Angiogenic endothelial cells are now shown to form podosome rosettes that are able to focally degrade the extracellular matrix, prior to vascular sprouting in tumour angiogenesis.

Angiogenesis, the process by which new vessels are formed from the pre-existing vasculature, occurs during development and under pathological conditions in adults. Vascular sprouting involves the concerted action of migrating tip endothelial cells at the leading edge of a growing vessel and proliferating stalk endothelial cells, which comprise the body of the new branch¹. The distinctions between developmental and adult vascular sprouting, if any, are yet to be identified. However, vascular growth from mature vessels requires the initial step of overcoming the physical barrier imposed by the basement membrane. In this issue, Seano et al.² show that endothelial cells must form podosome rosettes as a precursor to vessel outgrowth, shedding light on the mechanism by which sprouting endothelial cells pass through the basement membrane.

Podosomes are specialized actin-based structures formed by motile cells that control degradation of the extracellular matrix and promote invasive cell migration³ (Fig. 1). They include a core rich in filamentous actin and a cohort of intracellular molecules that control actin dynamics, including Rho GTPases and cortactin⁴. The cytoskeletal core is anchored on adhesion molecules and scaffolding proteins that mediate tight engagement to the extracellular matrix in the periphery while the central area promotes invasion⁵. The high concentration of membrane-bound proteases, such as membrane type-1 matrix metalloprotease (MT1-MMP), give these structures their ability to focally degrade extracellular matrix proteins and protrude through the heavily crosslinked basement membrane³.

Figure 1.

Podosome rosette formation in vascular sprouting. Schematic diagram of the steps involved in the formation of podosome rosettes during vessel sprouting. A mature vessel is shown on the left, with the area of the sprouting endothelial cell magnified on the right. Endothelial cell sprouting requires perforation of the basement membrane (BM) surrounding the mature vessel. (1) First, focal adhesions (FA) must disassemble, releasing α₆β₁ integrin (green and purple) from its tight association with laminin (blue semi-circles). (2) Next, focal adhesion components are relocalized (arrows) to form functional podosome rosettes (indicated by the arrows). (3) Activation of metalloproteinases, such as MT1-MMP (pink), within podosome rosettes allows the cell to invade (arrowheads) through the basement membrane, promoting the emergence of a new sprout. Actin filaments are shown in red, basement membrane is represented in light grey with the area of MT1-MMP-mediated extracellular matrix degradation shown in lighter grey, red ovals depict red blood cells inside the vessel.

Podosomes have been described in fibroblasts transformed by the non-receptor tyrosine kinase Src, endothelial cells, vascular smooth muscle cells, osteoclasts, immune cells of the myeloid lineage⁴ and in cancer cells, where they are also termed invadopodia as a reference to their invasive function and potential role in metastasis³. These structures occur either individually or in clusters that form higher order structures, including podosome belts in osteoclasts, and podosome rosettes in vascular smooth muscle cells and endothelial cells⁴. In certain cell types, the induction of podosomes requires signalling by growth factors⁵, such as in endothelial cells by transforming growth factor β (TGFβ; ref. 6) or the protein kinase C (PKC) activator phorbol-12-myristate-13-acetate (PMA; ref. 7). Growth factor signalling and PKC activation converge on the Src kinase pathway, which plays a key role in podosome assembly and function⁵. Vascular endothelial growth factor (VEGF), which can activate Src (ref. 8), also induces podosome formation in endothelial cells^9,10.

Podosome rosettes have been visualized in cultures of sprouting aortic endothelial cells¹¹, but without definitive evidence of their existence in vivo. However, Seano et al. were able to detect bona fide podosome rosettes in sprouting vessels in vivo, using adult angiogenesis models including two VEGF-dependent mouse models of tumour angiogenesis, a mouse model of hindlimb ischaemia, and clinical biopsies from human lung tumours. Specifically, in a mouse model of pancreatic insulinoma, the appearance of podosome rosettes increased as tumours progressed from the hyperplastic to the malignant stage, with the peak in their numbers coinciding with the VEGF-driven angiogenic switch.

To study the contribution of podosome rosette formation to VEGF-induced angiogenic sprouting, the authors treated in vitro and ex vivo cultures of endothelial cells with VEGF to mimic angiogenic endothelial cells. They observed that VEGF-treated endothelial cells formed more PMA-induced podosome rosettes compared to their untreated counterparts. These podosome rosettes were functional, as they were accompanied by an increase in MT1-MMP activity as well as larger areas of gelatin degradation. Interestingly, the podosome rosettes in mouse aortic ring explants were always distal to tip cells and at the base of the sprouting vessel. Live imaging of cytoskeletal dynamics revealed that podosome rosettes preceded vascular branches indicating a previously unknown hierarchical order in blood vessel formation.

A key mediator of signalling between the actin cytoskeleton and the extracellular matrix is the integrin family of adhesion molecules. Several integrins have been tied to podosome formation and function. For example, α_vβ₃ integrin is found in the podosomes of osteoclasts, and its inhibition results in defects in podosome function¹². In Src-transformed fibroblasts, β₁, but not β₃, integrins are essential for invadosome and rosette formation and function¹³. Seano et al. found that although most integrins expressed by endothelial cells were recruited to VEGF-induced podosome rosettes, only inhibition of α₆ integrin completely blocked the formation of VEGF-induced functional rosettes. VEGF stimulation strongly upregulated expression of α₆ integrins in cultured endothelial cells, and these integrins were also concentrated in rosettes of VEGF-stimulated aortic explants, whereas their loss suppressed the formation of VEGF-induced rosettes.

The authors further showed that the localization of α₆β₁ integrin to podosome rosettes, and VEGF-induced rosette formation were impaired when this molecule was associated with its major ligand, laminin. Conversely, aortic explants from mice lacking laminin showed significantly increased VEGF-induced podosome rosette formation. The levels of α₆β₁ integrin proved to be the limiting factor in this process, as overexpression of α₆ integrins resulted in increased rosette formation even in the presence of elevated laminin concentrations and did not require VEGF. The effect of laminin on rosette formation was shown to be mediated through its influence on the location of α₆ integrins. In the absence of laminin, VEGF-induced rosette formation coincided with a reduction of α₆β₁ integrin at the plasma membrane and into podosome rosettes; whereas when laminin was present, α₆β₁ integrin remained localized to the plasma membrane in classical focal adhesions.

Focal adhesions and podosomes share similar components, as both connect the cell to the extracellular matrix through the actin cytoskeleton, yet the former adhere cells to the substrate while the latter promote invasion. Thus, it is not surprising that a reciprocal relationship between the two structures has been reported, as for example in vascular smooth muscle cells, where focal adhesions must disassemble before the formation of podosomes¹⁴. Using total internal fluorescence microscopy to visualize focal adhesion dynamics and to dissect the reorganization of matrix adhesion sites, Seano et al. observed that focal adhesions had to first disassemble before podosome rosettes became detectable in endothelial cells. Interestingly, forced disassembly of focal adhesions was able to rescue the inhibitory effect of laminin on rosette formation. These results support that α₆β₁ integrin must be mobilized from focal adhesions into podosome rosettes.

Finally, the authors assessed the role of α₆ integrins in podosome rosette formation with respect to vascular branching. Genetic ablation of α₆ integrins in endothelial cells not only reduced rosette formation, but also decreased the incidence of angiogenic sprouts in the mouse aortic ring assay. By contrast, branching was significantly enhanced in mouse aortic rings from mice lacking α₄ laminin, consistent with previous reports of excessive tip cell formation in the retinas of mice lacking endothelial laminin¹⁵. When mice with pancreatic tumours were injected with a blocking antibody against α₆ integrins, the antibody rapidly localized to podosome rosettes. Extending the treatment of these mice during the onset of the angiogenic switch, strongly reduced podosome rosette density and vascular branching. These findings suggest that suppression of podosome rosette formation through the inhibition of α₆ integrins could be explored as a potential therapeutic strategy in angiogenic tumours.

In conclusion, Seano et al. showed that endothelial cells form podosome rosettes in angiogenic vessels in vivo prior to the emergence of new sprouts. This occurs through a mechanism that requires α₆β₁ integrin upregulation and/or relocalization from focal adhesions to podosomes. Given the link between invadopodia and cancer cell invasion³, it would be interesting to determine whether α₆β₁ integrins play a similar role in invadopodia formation by metastatic cancers, particularly in vivo, as this would raise the possibility of hitting multiple steps of cancer progression by targeting this integrin. Additional questions stemming from the present work relate to the signalling links between vascular sprouting and tip/stalk cell specification. It would be interesting to know whether specification is coupled to, or dependent on, rosette formation in adult angiogenesis. In addition, it would be interesting to understand whether angiogenesis during development also exhibits the formation of such rosettes. Finally, it will be important to identify the mechanism(s) for cell-specific selection of podosome formation and protrusive behavior; in particular, what restricts the phenomenon to a single cell or subset of cells amongst the large group of cells that are able to respond. These open questions notwithstanding, the work of Seano et al. has added to our knowledge of vascular morphogenesis and will be a spring-board for future studies to advance our understanding of the molecular and cellular mechanisms involved in this process.