doi: 10.1161/01.RES.0000191547.64391.e3
© 2005 American Heart Association, Inc.
Review |
Endothelial Extracellular Matrix
Biosynthesis, Remodeling, and Functions During Vascular Morphogenesis and Neovessel Stabilization
From the Department of Pathology (G.E.D.), Texas A&M University System Health Science Center, College Station; and Department of Pathology (D.R.S.), Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Mass.
Correspondence to Donald R. Senger, Department of Pathology, Beth Israel Deaconess Medical Center and Harvard Medical School, Research North, 99 Brookline Ave, Boston, MA 02215. E-mail dsenger{at}bidmc.harvard.edu
Guest Editor: This Review is part of a thematic series on Vascular Cell Diversity, which includes the following articles:
Heart Valve Development: Endothelial Cell Signaling and Differentiation
Molecular Determinants of Vascular Smooth Muscle Cell Diversity
Endothelial/Pericyte Interactions
Endothelial Extracellular Matrix: Biosynthesis, Remodeling, and Functions During Vascular Morphogenesis and Neovessel Stabilization
Joyce Bischoff
 |
Abstract |
|---|
 Top Abstract IntroductionECM Function in EC...ECM Function in EC...Endothelial ECM: RemodelingEndothelial ECM Biosynthesis,...References |
|---|
The extracellular matrix (ECM) is critical for all aspects of vascular biology. In concert with supporting cells, endothelial cells (ECs) assemble a laminin-rich basement membrane matrix that provides structural and organizational stability. During the onset of angiogenesis, this basement membrane matrix is degraded by proteinases, among which membrane-type matrix metalloproteinases (MT-MMPs) are particularly significant. As angiogenesis proceeds, ECM serves essential functions in supporting key signaling events involved in regulating EC migration, invasion, proliferation, and survival. Moreover, the provisional ECM serves as a pliable scaffold wherein mechanical guidance forces are established among distal ECs, thereby providing organizational cues in the absence of cell–cell contact. Finally, through specific integrin-dependent signal transduction pathways, ECM controls the EC cytoskeleton to orchestrate the complex process of vascular morphogenesis by which proliferating ECs organize into multicellular tubes with functional lumens. Thus, the composition of ECM and therefore the regulation of ECM degradation and remodeling serves pivotally in the control of lumen and tube formation and, finally, neovessel stability and maturation.
Key Words: extracellular matrix • endothelial cells • angiogenesis • vascular morphogenesis • vessel stabilization
|
Introduction |
|---|
|
TopAbstractIntroductionECM Function in EC...ECM Function in EC...Endothelial ECM: RemodelingEndothelial ECM Biosynthesis,...References |
|---|
The extracellular matrix (ECM) provides critical support for vascular endothelium. Primarily through adhesive interactions with integrins on the endothelial cell (EC) surface, ECM provides a scaffold essential for maintaining the organization of vascular ECs into blood vessels. In addition, EC adhesion to ECM is required for EC proliferation, migration, morphogenesis, survival, and ultimately blood vessel stabilization, all of which are critical for neovascularization. The specific mechanisms through which ECM supports EC functions are complex and involve both external structural support and regulation of multiple signaling pathways within the cell, including signaling pathways that control apoptosis, proliferation, the cytoskeleton, and cell shape. Thus, through both mechanical and signaling functions, the ECM affects many fundamental aspects of EC biology. Moreover, the diversity of ECM components in the EC microenvironment and the diversity of mechanisms for controlling the synthesis and degradation of ECM have suggested an intricate level of complexity sufficient for ECM to exert significant and precise control over many aspects of neovascularization and blood vessel maturation.
The purpose of this review is to provide an overview of the importance of ECM for the biology of vascular ECs, and it is divided into 4 parts: (1) ECM Function in Cellular Morphogenesis and Signaling; (2) ECM Function in EC Lumen Formation and the Switch to Vessel Maturation; (3) Endothelial ECM: Remodeling; and (4) Endothelial ECM Biosynthesis, Assembly, and Structural Functions: Critical Role for EC Basement Membrane Matrix in Vessel Stabilization. Because of our own specific research interests, we have focused particularly on the importance of ECM for vascular morphogenesis, ie, the complex process through which proliferating ECs organize into new blood vessels with functional lumens.
|
ECM Function in EC Signaling and Morphogenesis |
|---|
|
TopAbstractIntroductionECM Function in EC...ECM Function in EC...Endothelial ECM: RemodelingEndothelial ECM Biosynthesis,...References |
|---|
Although much emphasis has been placed on the role of angiogenic cytokines such as vascular endothelial growth factor (VEGF) in EC migration, survival, and proliferation, considerable evidence indicates that ECM is equally or more important. Moreover, in most cases, cytokine function is entirely dependent on EC adhesion to ECM. There is a growing body of evidence that ECM also drives capillary morphogenesis through sustained signaling, resulting in persistent EC cytoskeletal reorganization and changes in cell shape. Also, ECM scaffolds may function during angiogenesis to immobilize angiogenic cytokines and thereby coordinate signals transduced through both growth factor and ECM receptors, as addressed in the section Endothelial ECM Biosynthesis, Assembly, and Structural Functions: Critical Role for EC Basement Membrane Matrix in Vessel Stabilization. In this section, we focus attention on how ECM regulates signaling pathways to control EC proliferation, survival, and capillary morphogenesis.
Adhesion to ECM Regulates EC Migration, Proliferation, and Survival
Vascular ECs require adhesion to ECM for migration, and EC migration is important for angiogenesis, particularly during sprouting of new blood vessels from the existing vasculature.1 Although gradients of cytokines or other agonists are required to drive chemotactic migration, such directed motility is absolutely dependent on EC adhesion to ECM. Moreover, evidence from in vitro experiments indicate that many of the interstitial and provisional ECM components that are encountered during sprouting angiogenesis, including interstitial fibrin and collagen I, are capable of supporting chemotactic migration.2,3 In addition, gradients of immobilized ECM components can by themselves drive haptotactic migration in vitro, and this type of motility is not dependent on cytokines.3,4 Although the significance of such haptotactic migration has not been established in vivo, it seems plausible that the high concentrations of interstitial collagen encountered by ECs during the sprouting phase of angiogenesis may drive outward migration, in part, through haptotaxis. Indeed, interstitial collagen is highly effective at promoting haptotactic migration in vitro.3 Thus, sprouting ECs may migrate in response to both chemotactic gradients of angiogenic cytokine and haptotactic gradients of ECM. Regardless, any and all such motility-promoting stimuli are dependent on EC adhesion to ECM.
One of the most fundamental functions provided by ECM involves support of EC proliferation and survival. Both proliferation and survival are highly dependent on adhesion to ECM through cell-surface integrins.5–8 In particular, activation of the p44/p42 (Erk1/Erk2) mitogen-activated protein kinase (MAPK) signal transduction pathway in ECs is critical for EC proliferation and angiogenesis.9–12 EC anchorage to ECM through integrins is necessary for efficient MAPK activation by cytokines.13,14 Also, the expression and activities of cyclin-dependent kinases, which are required for cell cycle progression and therefore for cell proliferation, are dependent on EC adhesion to ECM.15–17 Thus, failure of ECs to attach to ECM results in cessation of cell proliferation through multiple mechanisms. Adhesion to ECM is similarly important for EC survival.6 In particular, adhesion-dependent activation of the MAPK pathway, which, as noted above, is required for EC proliferation, functions critically in EC survival.18–20 By controlling the transcription of key signaling molecules that regulate caspase-mediated apoptosis, the MAPK pathway suppresses cell death.19 Thus, apoptosis is induced and proliferation ceases when ECs are prevented from attaching to ECM, indicating that ECM is pivotal for the most fundamental aspects of EC biology.
Although the importance of ECM for EC migration, proliferation, and survival is well established, the relative importance of specific ECM components in supporting these processes is less understood and often difficult to ascertain because of functional overlap. For example, cytokine activation of the MAPK pathway in microvascular ECs and proliferation of microvascular ECs are similarly supported by attachment to either collagen I or vitronectin.20 Moreover, a variety of ECM components provide sufficient support for EC migration,2,3,21 although not with equal potency.21 There is also evidence that components of ECM exhibit maximal activity in combination with each other and thereby may function cooperatively.20
ECM Provides Guidance Cues That Support Capillary Morphogenesis
During angiogenesis, proliferating and migrating ECs organize to form new 3D capillary networks. This process has been studied extensively in the embryo, thus establishing that an early stage of capillary morphogenesis involves transition of endothelial precursor cells to a spindle-shaped morphology.22 Coincident with this spindle-shape transition, EC precursors align and connect into solid, multicellular, precapillary cord-like structures that form an integrated polygonal network.23,24 Solid precapillary cord-like structures have also been identified during angiogenesis in the adult.25 During maturation, solid vascular cords form hollow lumens for the transport of blood, and ECs are sequestered from the interstitial matrix through establishment of a continuous basal lamina.22,25
During vascular morphogenesis, the ECM serves as a 3D malleable scaffold in which individual ECs and clusters of ECs can transduce mechanical forces to other ECs at a considerable distance. Thus, by generating mechanical, contractile forces within ECM, ECs are able to establish tension-based guidance pathways that allow them to form interconnected cords. These guidance pathways provide a mechanism for ECs to organize into large multicellular structures at a distance without the initial requirement of cell:cell contact.24,26
ECM Signaling Controls EC Morphogenesis
In addition to serving as an adhesive support in which EC mechanical forces can be transduced to promote multicellular organization, ECM exerts important signaling functions directly on ECs to regulate cell shape and contractility. It has long been recognized that 3D interstitial collagen type I provokes ECs in culture to undergo marked shape changes that closely imitate precapillary cord formation observed during embryonic vasculogenesis and adult angiogenesis. Within hours after addition of collagen I to confluent EC cultures, the cells partially retract and exhibit a spindle-shaped morphology, coincident with realignment to form solid cords organized in a polygonal pattern27–31 (see Figure 1). Subsequently, over the course of several days, these structures mature to form tubes with hollow lumens through a process involving development and coalescence of intracellular vacuoles.26 In sharp contrast to ECs, fibroblasts do not respond to collagen I with changes in cell shape or alignment into cords, suggesting a special connection between collagen I signaling and EC morphogenesis.32
|
In support of the importance of interstitial collagens in driving ECs to form precapillary cords, there is much evidence that interactions between interstitial collagens and ECs are highly relevant in vivo. During sprouting angiogenesis, ECs within existing blood vessels degrade basement membrane33 and migrate and proliferate within connective tissue abundant in interstitial collagens.34 Furthermore, the key angiogenic cytokine VEGF, which induces sprouting angiogenesis also induces microvascular ECs to express integrins 
1ß1 and 2ß1,35 which are the key interstitial collagen receptors on microvascular ECs. Moreover, antagonism of these integrins inhibits dermal and tumor angiogenesis in vivo.3,35 Also, analyses of genes expressed in human tumor endothelium have demonstrated that ECs isolated from tumors express >10-fold more transcripts encoding interstitial collagens type I and III than ECs isolated from corresponding control tissue, indicating that tumor ECs express their own interstitial collagens,36 thus suggesting that interstitial collagen expression by tumor ECs may be conducive for angiogenesis. In support of this hypothesis, expression of collagen I by isolated EC clones in vitro correlates with spontaneous multicellular organization of these ECs into cords.37,38 Finally, neovascularization can be inhibited in animal models both by proline analogues that interfere with collagen triple-helix assembly and by ß-aminopropionitrile, which inhibits collagen cross-linking,39 indicating that collagens play a crucial role in angiogenesis.
Recently, the mechanisms through which collagen I selectively provokes EC formation of cord-like structures that closely imitate precapillary cords in vivo have begun to be identified. First, collagen I-ligation of integrins 1ß1 and 2ß1 in microvascular ECs in vitro suppresses cAMP and thereby suppresses the activity of cAMP-dependent protein kinase A (PKA). Suppression of PKA activity leads to a marked induction of actin polymerization that contributes to the formation of prominent stress fibers and EC contractility. Suppression of PKA and induction of actin polymerization closely parallel cord formation in vitro, consistent with functional significance. In contrast to collagen I, laminin-1 neither suppresses cAMP nor PKA activity nor induces actin polymerization, indicating specificity among matrix proteins regarding this signaling pathway.32 Consistent with the importance of this pathway for formation of precapillary cords, laminin-1 also failed to induce changes in cell shape. Moreover, in contrast to ECs, fibroblasts did not respond to collagen I with changes in cAMP or actin polymerization, consistent with the failure of collagen I to provoke cord formation by fibroblasts. Pharmacological elevation of cAMP blocks collagen-induced actin polymerization and formation of cords by ECs; conversely, pharmacological suppression of either cAMP or PKA induces actin polymerization. Thus, collagen I–induced suppression of cAMP-dependent PKA and induction of actin polymerization is an important mechanism through which collagen I drives EC organization into multicellular precapillary cords.32
In addition to suppressing cAMP-dependent PKA, collagen I stimulation of microvascular ECs in vitro induces activation of Src kinase and the GTPase Rho, also through interactions with ß1 integrins.40 Similar to suppression of PKA, activation of Src and Rho parallels cord formation. The Src inhibitor PP2, dominant-negative Src, the Rho inhibitor exoenzyme C3 transferase, and dominant-negative RhoA each individually inhibit collagen I induction of actin stress fibers that mediate cell retraction, and each inhibit capillary morphogenesis.40,41 Thus, collagen I induces actin stress fibers through activation of Src and Rho,40 and such activation likely complements collagen I–induced actin polymerization, as mediated by PKA suppression (see above), to achieve the robust cytoskeletal contractility that drives cord assembly. Importantly, the significance of actin stress fibers and EC contractility for angiogenesis in vivo has been established in a mouse model of VEGF-driven neovascularization. Specifically, retrovirus-mediated transduction of ECs with a dominant-negative RhoA mutant that blocks EC contractility and cord formation in vitro markedly suppressed organization of proliferating ECs into new blood vessels in vivo.41
Activation of Src by collagen I not only contributes to the formation of actin stress fibers but also disrupts vascular endothelial (VE)-cadherin from intercellular junctions. Both the Src inhibitor PP-2 and dominant-negative Src preserve VE-cadherin localization to regions of cell–cell contact; in the absence of such inhibitors, VE-cadherin is disrupted by collagen I. Also, an active Src mutant disrupted VE-cadherin and cell–cell contacts similarly to collagen I. In sharp contrast to collagen I, laminin-1, which, as noted above, does not induce capillary morphogenesis, also does not induce activation of Src or Rho. Rather, laminin-1 induces persistent activation of the GTPase Rac.40 Thus, in addition to suppression of PKA activity, activation of Src and Rho is also a key mechanism through which collagen I provokes capillary morphogenesis of microvascular ECs. Moreover, activation of p38 MAPK also has been implicated as important to the mechanism by which collagen I drives ECs to form cords,42 although the precise contribution of such activation to morphogenesis is unclear.
Finally, an intriguing aspect of differences observed between collagen I and laminin-1 signaling in microvascular ECs is that they suggest a model whereby interstitial collagen and basement membrane laminin differentially regulate various stages of angiogenesis. As summarized in Figure 1 (top), interstitial collagen I activation of both Src and Rho and suppression of PKA promotes formation of prominent actin stress fibers that mediate EC retraction and capillary morphogenesis. In addition, collagen I activation of Src disrupts VE-cadherin from cell junctions and promotes disruption of cell–cell contacts, thus facilitating multicellular reorganization. In sharp contrast to collagen I, basement membrane laminin-1 does not provoke EC morphogenesis or induce activation of Src or Rho or suppression of PKA. Rather, laminin-1 induces persistent activation of the GTPase Rac, whereas collagen I suppresses Rac activity after a transient increase.40 The potential significance of such distinctions in ECM signaling is summarized in Figure 1 (top). During the sprouting and proliferative stages of angiogenesis the laminin-rich basal lamina is degraded, resulting in reduction of EC–laminin interactions. Thus, the available in vitro data predict that loss of laminin substratum results in reduced Rac activity and therefore a loss of Rac function in supporting integrity of cell–cell junctions during this phase. Moreover, on degradation of basement membrane, sprouting ECs are exposed to underlying interstitial collagens and begin to invade it, resulting in activation of Src and Rho, suppression of PKA, and initiation of capillary morphogenesis. Subsequently, as the newly formed capillary sprouts mature into new vessels with mature lumens, the intact basement membrane is reestablished. The continuous basement membrane sequesters ECs from interstitial collagens and thereby reestablishes normal activation levels for Rac, Rho, Src, and PKA. Thus, the in vitro evidence supports a model in which the laminin-rich basement membrane serves not only to maintain the integrity of the mature endothelium but also to sequester and thereby insulate ECs from interstitial collagens. In contrast, degradation of basement membrane exposes ECs to interstitial collagens and activates signaling pathways that drive cytoskeletal reorganization and sprouting morphogenesis.
|
ECM Function in EC Lumen Formation and the Switch to Vessel Maturation |
|---|
|
TopAbstractIntroductionECM Function in EC...ECM Function in EC...Endothelial ECM: RemodelingEndothelial ECM Biosynthesis,...References |
|---|
ECM Stimulates EC Lumen Formation and Tubular Morphogenesis Through Integrins
As discussed above, considerable data suggest that EC integrin interactions with interstitial matrix proteins (ie, interstitial collagens and fibrin/fibronectin) are key receptors in stimulating EC tubular morphogenesis as well as concurrent EC activation (Figures 1 and 2
).32,40,43–45 The major integrins responsible for these interactions are 2ß1, 1ß1, and vß3, 5ß1, which are collagen and fibrin/fibronectin receptors, respectively3,26,35,44–46 (Figure 2). EC tubular morphogenesis is initiated by the matrix-integrin-cytoskeletal signaling axis,44 which causes simultaneous sprouting (ie, cord formation) and vacuole and lumen formation events that eventually lead to interconnecting networks of EC-lined tubes (Figure 3). These events depend on integrin interactions with collagen or fibrin matrices through their respective integrins, as indicated above.
|
|
Much work indicates that lumen formation during capillary morphogenesis in 3D collagen matrices depends on the formation and coalescence of integrin-dependent pinocytic intracellular vacuoles.26,44–47 These intracellular vacuoles fuse together to form intracellular lumens.44,45,47–52 These compartments then fuse through exocytic events to create interconnections between adjacent ECs to form continuous luminal structures.44,45 This intracellular lumen formation process is a highly efficient mechanism to rapidly create ECM-free space26,44,45 and begin the establishment of apical-basal polarization. Intracellular vacuole and lumen formation requires the activity of MT-MMPs on the EC surface (see the section Endothelial ECM: Remodeling).
The integrin-dependent pinocytic process that regulates intracellular vacuole and lumen formation, which resembles in many respects the process of macropinocytosis, also depends on Rho GTPases, which are downstream of integrin–ECM interactions.45,47 In particular, Cdc42 and Rac1 are required for intracellular vacuole and lumen formation to occur. We observed that dominant-negative forms of Cdc42 and Rac1 markedly block these events,47 whereas RhoA blockade did not regulate lumen formation in our 3D collagen morphogenesis model.47 Furthermore, we demonstrated that green fluorescent protein fusion proteins with Cdc42 or Rac1 allowed for the targeting of these constructs to intracellular vacuole membranes during the lumen formation process.45,47 Thus, integrin–ECM interactions regulate Cdc42 and Rac1 activation and vacuole membrane targeting to control the proper cytoskeletal rearrangements and vacuole targeting mechanisms to properly orient the position of the developing lumen.
Potential Key Roles for Integrin Signaling to Control the Switch Between EC Tubular Morphogenesis Versus EC Tube Stabilization
In contrast to the provisional matrix-binding integrins discussed above, basement membrane laminin-binding integrins such as 6ß1 and 3ß1 may be more important for the process of tube stabilization (see Figure 2). As illustrated in Figure 1, the marked distinctions in signaling by collagen type I ("fire") and laminin-1 ("ice") suggest a mechanism through which degradation of basement membrane and exposure of activated and proliferating ECs to collagen type I initiates morphogenesis of new capillary sprouts. Conversely, they suggest that laminins serve a maturation function. Integrin 6ß1 has been reported to regulate cord formation in planar angiogenesis systems in which ECs are placed on the surface of basement membrane matrix gels.24,53,54 Therefore, it is important to consider that multiple integrins may act together in regulating vascular morphogenesis (ie, 2ß1 and 1ß1 in collagen matrices or 5ß1 or vß3 in fibrin matrices)44,45 and vascular stabilization (possibly 6ß1 or 3ß1 in combination with 2ß1 or 1ß1) to thereby create unique combinations of signals (see Figures 1 and 2
). The ability of 3ß1 and 6ß1 to interact with components of the tetraspanin web55 makes them particularly attractive as key regulators of the switch from EC activation/morphogenesis to EC stabilization/quiescence. Also, 3ß1 is intriguing because of its known affinity for tissue inhibitor of metalloproteinase (TIMP)-2, an angiogenic inhibitor,56 as well as other molecules that may regulate the EC proteolytic machinery (eg, extracellular MMP inducer and urokinase-type plasminogen activator receptor)57–60 that control morphogenesis or regression (see the section Endothelial ECM: Remodeling). Furthermore, 3ß1 may participate in early EC tubular morphogenic events along with integrins such as 2ß1 and 1ß1 (Figure 2). For example, the tetraspanin, CD151, which interacts strongly with 3ß1, has been shown to stimulate cell motility and invasion,55 which are critical for tubular morphogenesis. Although 3ß1 shows very low affinity for laminin-1, it shows strong affinity for the epithelial laminin, laminin-5,61 and laminin-10, with detectable but variable binding to laminin-8, depending on the cell type.62,63 Integrin 6ß1 appears to have a broader specificity for different laminin isoforms.62,64 Thus, these receptors interact with the key laminins in EC basement membrane matrices and signals derived from these individual integrins or in combination with other laminin-binding receptors (ie, 1ß1, 2ß1, dystroglycan, sulfatides)64,65 may act to suppress the EC tubular morphogenic program by blocking sprouting and vacuole and lumen-formation events. In addition, these receptors may suppress EC proliferation and signaling pathways that regulate EC activation such as the Ras-MAPK and nuclear factor 
B pathways,66,67 thereby promoting EC quiescence. Following establishment of the EC basement membrane, signals from 6ß1 and 3ß1 may also control EC gene expression, which regulates EC quiescence.
Thus, signaling events downstream of laminin-binding integrins are likely crucial to the establishment of EC tube stabilization through their interactions with the EC basement membrane. The components of this basement membrane matrix including laminin-10 and laminin-8 are likely to interact with both the EC as well as its supporting cells such as pericytes, which are known to be required for vascular stabilization. These integrin–laminin interactions appear to tip the balance toward stabilization and, furthermore, probably actively inhibit sprouting morphogenesis (Figures 1 and 2). Further work is required to define the key molecular interactions between EC or pericyte integrins and ECM, together with elucidation of key downstream signaling pathways. Such interactions and signaling pathways very likely regulate the critical balance between sprouting morphogenesis and vessel stabilization. A better understanding of these fundamental issues will hopefully lead to more sophisticated approaches for regulating neovascularization in clinical settings.
|
Endothelial ECM: Remodeling |
|---|
|
TopAbstractIntroductionECM Function in EC...ECM Function in EC...Endothelial ECM: RemodelingEndothelial ECM Biosynthesis,...References |
|---|
Critical Role for Membrane-Type Matrix Metalloproteinases in EC Tubular Morphogenesis in Three-Dimensional Extracellular Matrices
The ability of ECs and their supporting cellular elements to directly modify (ie, remodel) their immediate ECM environment is critically important for forming tubular networks, stabilizing these networks, and inducing regression of these networks. Many studies indicate that ECs are highly proteolytic during the formation or regression of vascular networks.44,68,69 A fundamental goal in vascular biology is to understand which proteinases are relevant and to identify their substrate targets at various stages of neovascularization. Steps in which proteolysis appears likely to be critical during EC morphogenesis include (1) the initiation of angiogenesis whereby basement membrane breakdown occurs, (2) invasion into collagen type I or fibrin matrices, and (3) lumen formation.44,45 Considerable data now suggest that MT-MMPs are the critical regulators of EC invasion into 3D collagen or fibrin matrices.44,68,70–77 In addition, we have shown that that MT-MMPs are important regulators of lumen formation in 3D extracellular matrices.70 Moreover, the proteinase inhibitors TIMP-2, TIMP-3, and TIMP-4 all inhibit EC vacuole and lumen formation, whereas TIMP-1 does not (W.B. Saunders and G.E. Davis, manuscript in preparation). MT-MMPs are blocked by TIMP-2, -3, and -4 but not by TIMP-1.78 In support of these conclusions, small-interfering RNA knockdown of MT1-MMP also markedly inhibits EC vacuole and lumen formation (W.B. Saunders and G.E. Davis, manuscript in preparation).
In contrast to MT-MMPs, many soluble MMPs have not been found to be involved in these invasive or morphogenic events, despite the fact that some of these enzymes are induced during EC tubular morphogenesis (ie, MMP-1 or MMP-9).44,71,79,80 A few studies indicate a role for MMP-9 during retinal angiogenic responses in vivo,81 and MMP-9 has been reported to liberate VEGF from ECM sites.82 In contrast, several studies using in vitro approaches have failed to reveal a role for MMP-9 in tubular morphogenesis in 3D collagen matrices.71,80 Thus, the induction of an MMP during morphogenesis does not necessarily mean that it plays a role during the tube-formation process. However, it remains possible that the influence of a particular MMP may depend on the vascular bed analyzed or on a particular type of EC. In other studies, we have recently shown that induction of both MMP-1 and MMP-10 during tubular morphogenesis appears to be primarily associated with regression of tubular networks80 and not the formation of these networks (see below). Thus, some soluble MMPs appear to be produced to control the process of tube regression rather than morphogenesis (see below).44,79,80,83 Also, some of the complexity and confusion related to the influence of MMP inhibitors on wound- or tumor-induced angiogenesis84,85 may directly relate to the different functional roles of specific MMPs in controlling tubular morphogenesis versus regression. An MMP inhibitor that blocks regression may actually promote angiogenesis by facilitating the maintenance and survival of neovessels.
ECM Degradation As a Regulatory Signal for Initiating Vascular Morphogenesis and Controlling Vessel Stabilization
The enzymes that control basement membrane degradation and the initiation of angiogenesis have yet to be defined, including the relative contributions of membrane-associated versus soluble MMPs or other classes of enzymes such as serine proteinases. Various MMPs are known to degrade basement membrane matrix components, and these include MT-MMPs, MMP-3, MMP-10, and MMP-9.44,68,69,75 Perhaps a combination of these enzymes controls the invasion of ECs through basement membrane during the sprouting phase of angiogenesis. In addition to the function of MMPs or other proteases in degrading basement membrane matrix, another possible function of proteinases during sprouting angiogenesis may be to inactivate a constitutive inhibitor of EC sprouting morphogenesis. Such an inhibitor may be present within basement membranes to prevent sprouting in a quiescent vessel and may be produced by supporting cells such as pericytes.44,86 One such inhibitor may be the heparan sulfate–binding proteinase inhibitor TIMP-378 (ie, allowing it to interact directly with the basement membrane proteoglycan, perlecan), which is highly expressed by pericytes87 and induced by EC-pericyte interactions (W.B. Saunders et al, manuscript in preparation ). We have previously shown that TIMP-3 is able to markedly inhibit EC invasion and tube formation in 3D collagen or fibrin matrices.70 It has been reported that serine proteinases can inactivate TIMP-1,88 suggesting the possibility that a function of serine or other proteinases may be to inactivate TIMPs that control vascular stabilization. Another function of TIMPs and TIMP-3, in particular, is to decrease stimuli that activate ECs. Delivery of TIMP-3 is known to markedly suppress various membrane-associated MMPs including MT-MMPs as well as various disintegrins and metalloproteinases such as ADAM17, ADAM10, and ADAM15.78,89–92 This class of proteinases controls ECM degradation as well as the release of EC activators such as membrane-associated growth factors that interact with epidermal growth factor receptors (eg, heparin-binding epidermal growth factor).89–91
The integrity of EC–EC junctions93–96 is also key for maintaining a stabilized state. Membrane-associated proteinases that are known to facilitate disassembly of junctional contacts include ADAM-15 (which is concentrated in junctions of both endothelial and epithelial cells) and ADAM-10,96,97 MT-MMPs,98 and soluble proteinases such as MMP-3 or MMP-7.99 Because ADAMs and MT-MMPs are known to control proteolysis of cell–cell junctional adhesion molecules, blockade of these proteinases with TIMP-3 (which, uniquely among the TIMPs, blocks both ADAMs proteinases and MT-MMPs)78 may strongly promote cell–cell junction formation and stability. Junctional stability may also enhance basement membrane matrix stability by facilitating the continued deposition of basement membrane components in a polarized basal direction.100,101 The combination of EC–EC junctional stability (with appropriate apical-basal polarity), and continuous basement membrane matrix deposition is likely required for EC tube stabilization and a quiescent phenotype.
Soluble MMPs Such As MMP-1 and MMP-10 Play a Role As Vascular Regression Factors
Whereas considerable data now suggests that MT-MMPs control EC tube morphogenesis in 3D extracellular matrices,44,70–73,102 other MMPs such as MMP-1 and MMP-10 (stromelysin-2) appear to control the process of regression rather than morphogenesis.79,80 Blockade of these MMPs with proteinase inhibitors such as TIMP-1 and 2-macroglobulin, blocking antibodies, and small-interfering RNA knockdown does not influence tube formation. In contrast, these reagents markedly inhibit EC tube regression events in 3D collagen matrices.79 We have found that multiple serine proteases such as plasmin and plasma kallikrein induce activation of both MMP-1 and MMP-10 proenzymes to control regression.80 Plasmin and MMP-10 (as well as previous reports with MMP-3) synergistically activate pro–MMP-1, creating a superactive enzyme that is approximately 10-fold more active than MMP-1 activated with plasmin alone.103,104 Thus, the ability of ECs to coinduce MMP-1 and MMP-10 during tube morphogenesis establishes the conditions necessary to regulate tube regression when a serine protease–activating stimulus is supplied.44,79,80
|
Endothelial ECM Biosynthesis, Assembly, and Structural Functions: Critical Role for EC Basement Membrane Matrix in Vessel Stabilization |
|---|
|
TopAbstractIntroductionECM Function in EC...ECM Function in EC...Endothelial ECM: RemodelingEndothelial ECM Biosynthesis,...References |
|---|
EC Synthesis and Deposition of Basement Membrane Matrix
ECs exposed to blood flow are typically associated with continuous basement membrane matrix structures. It has largely been assumed that ECs are primarily responsible for the synthesis and deposition of these ECM components. However, little is known about how basement membrane assembly is regulated by ECs or whether the ECs require supporting cells such as pericytes105–109 or astrocytes110 to properly assemble these structures. The mechanisms that regulate basement membrane matrix assembly by ECs and its supporting elements remain a critical question in vascular biology. Nevertheless, much has been learned over the past decade or so concerning the molecular control of basement membrane matrix assembly during development with a primary focus on how epithelia and muscle cells assemble basement membranes.64,65,101,111–114 A major message from these studies is that laminins are the primary determinants of basement membrane assembly and that other basement membrane components such as collagen type IV variants,115 perlecan (basement membrane heparan sulfate proteoglycan-2), nidogens, and collagen type XVIII are accessory components.64,101,114 Collagen type IV deletion results in embryonic lethality when defects arise in cells in contact with basement membranes that are placed under mechanical stress such as in the heart and blood vessels.116 EC basement membranes are thought to contain predominantly the collagen type IV isoform [
1]2ß1.111 Consistent with the data from collagen type IV knockout mice, knockout of the collagen-specific chaperone heat shock protein (Hsp) 47 also resulted in a similar lethal vascular phenotype.117,118 Hsp47 controls the assembly of triple-helical collagens115 and without this chaperone,117,118 the triple-helical regions of collagen type IV as well as other collagens are unstable and rapidly degraded by enzymes such as serine proteinases (which are normally unable to proteolyze these triple helices). Thus, in Hsp47 knockout mouse, marked abnormalities were observed in both interstitial collagen matrices and basement membrane matrices. Recent studies reveal an accumulation of collagen type IV in the endoplasmic reticulum of Hsp47 knockout cells that leads to cellular apoptosis.117 Similar vascular defects, but of lesser severity to those of the collagen type IV null mice, were observed in perlecan-null mice, suggesting that perlecan also plays a key role in basement membrane function in mechanically stressed cells.119 Thus, both collagen type IV and perlecan, although not absolutely required for basement membrane formation, are necessary for stability of basement membranes under conditions of mechanical stress, and both genes are required for embryonic survival during development.116,119 In mice lacking these genes, the continuous laminin-containing basement membrane matrix forms normally but then over time becomes unstable and begins to develop discontinuities and physical breaks.116,119 Although the mechanisms underlying such instability of basement membrane laminins are not entirely clear, the absence of collagen type IV or perlecan may render laminins more susceptible to proteolysis. A summary of the vascular phenotypes of mice with null mutations for various basement membrane matrix components is presented in Table I in the online data supplement available at http://circres.ahajournals.org.
Role for Laminin-8 and Laminin-10 in EC Tubular Morphogenesis and Stabilization
Laminins are a family of proteins containing three subunits arranged in different combinations. These molecules are discussed in detail in several excellent recent reviews.64,101,114 Some laminin isoforms are thought to be particularly adept at assembling through polymerization reactions to form basement membranes containing laminin-1 (1ß1
1)112,113,120,121 (embryonic basement membranes), laminin-2 (2ß11) (skeletal muscle basement membranes),122,123 and laminin-10 (adult and vascular basement membranes) (5ß11).123–125 EC basement membranes are thought to primarily consist of the laminin-8123,126 and laminin-10 isoforms,123,127,128 with some additional contribution of laminin-9 (4ß21) or laminin-11 (5ß21). The ß2 subunit of laminin creates isoforms containing forms of laminin (s-laminin) found to be concentrated in synaptic contacts at neuromuscular junctions.123,129 Laminin-10 is an abundant and broadly expressed laminin in adult tissues.64,123 This latter laminin variant is a major laminin synthesized by ECs and appears to be the major self-polymerizing laminin that is likely responsible for EC basement membrane assembly,64,101,114,127,130 particularly in adult tissues. During embryonic development, it appears that laminin-8 (4ß11) is the major laminin produced by ECs, and it predominates in vascular basement membranes over laminin-10 at this time.124,126,127,131 However, laminin-10 does appear to overlap with laminin-8 expression in some vascular beds during development.123,127
Laminin-8 lacks globular domains on its short arms, which are thought to be critical for self-polymerization reactions to form basement membranes. Although laminin-8 can interact with other basement membrane components, it may not be capable of itself assembling the actual basement membrane structure.64,124 Knockout of the laminin 4 chain results in the appearance of hemorrhages around many vessels in embryos, but, despite this, many animals survive beyond embryonic development and eventually normalize because of the appearance of laminin-10 in vascular basement membranes within the first 2 to 3 weeks of postnatal age (see online Table I).126 The 4 laminin chain knockout animals show poorly formed vascular basement membranes during development and shortly after birth show observable discontinuities in basement membrane structure, and, despite this, most animals survive with a relatively normal appearing vasculature. These knockout animals show increased angiogenic responses following application of angiogenic cytokines or the implantation of tumor cells.132 This response may reveal that the EC/pericyte-derived basement membrane matrix is a signal for vascular quiescence, and stabilization and it may be that a decrease in this quiescence signal (perhaps laminin-8, -10, or both) can facilitate the activation of ECs or initiation of neovascularization during angiogenic responses (see section ECM Function in EC Lumen Formation and the section Switch to Vessel Maturation and section ECM Function in EC Lumen Formation and the Switch to Vessel Maturation). Similar results have recently been obtained using the aortic ring EC sprouting assay from collagen type XVIII knockout mice, whereby increased sprouting is observed compared with wild-type mice.133 However, these mice are, for the most part, normal with respect to their vasculature (except in the developing eye).134 These results may be attributable to the inhibitory activity of the NC1 domain fragment of collagen type XVIII, endostatin, which, like many NC1 domain fragments, is able to block angiogenic responses.135–137
The knockout of the laminin 5 subunit results in an embryonic lethal phenotype that does not progress beyond embryonic day 16.5.127 The lethality appears to relate to abnormalities of the placental as well as fetal vasculature (online Table I). In these animals, reduced complexity in the branching patterns of these vascular beds was observed, and existing vessels had larger lumen diameters.127 Similarly, angiogenic vessels from laminin 4 knockout vessels also had markedly increased lumen diameters.126 Thus, a function of basement membrane matrix through laminins may be to inhibit tubular morphogenesis by interfering with lumen expansion. Normally, some fetal vessels are juxtaposed with trophoblasts (eg, in the placental labyrinth) that adhere to the EC basement membrane, and these were clearly abnormal in the laminin 5 knockout mice.127 A similar defect occurs in the developing kidney glomerulus in these mice such that there are marked abnormalities of mesangial cell–EC basement membrane interactions causing an absence of capillary loop formation130 (a key morphological change required for glomerular development). The related nature of mesangial–EC interactions with pericyte–EC interactions138,139 are such that molecules such as laminin-10 may represent key regulators of this interaction that when absent may cause the observed microvascular abnormalities and instability in the knockout animals. Thus, laminins mediate interactions of ECs with basement membranes as well as adjacent supporting cells such as pericytes to facilitate vessel stabilization.
An interesting concept is that there may be intermediate stages of basement membrane assembly (ie, provisional basement membranes) and that these may be critical during processes such as EC sprouting and tubular morphogenesis. Early work suggested the concept that fibronectin may provide such provisional matrix signals for vascular morphogenesis while laminins are deposited later for stabilization.140 Other work suggests the possibility that laminin-8 and collagen type IV might participate in providing a provisional ECM scaffold during tubular morphogenesis before the development of a mature basement membrane containing laminin-10 (ie, leading to tube stabilization).64,126 In support of these concepts, antagonism of the 4 laminin subunit with blocking antibodies resulted in inhibition of EC morphogenic responses and EC survival in various in vitro models,141,142 and, in our model of EC tubular morphogenesis in 3D collagen matrices, the 1 and 4 laminin subunit genes were induced, whereas the 5 laminin subunit gene was expressed in a stable fashion.143 In addition, a laminin switch is known to occur during kidney development where the laminin-1 isoform (embryonic development) is eventually replaced with laminin-10 (adult).130 Thus, the EC basement membrane matrix may undergo regulated changes in laminins during the transition from morphogenesis to stabilization.
A related question involves the extent to which ECs are able to assemble basement membrane matrices in a provisional matrix such as fibrin.46,144–147 Fibrin and fibronectin are recruited together from plasma through increased vascular permeability following tissue injury to modify the local ECM that regulates vascular morphogenesis.43 Recent studies have revealed that fibrin-induced EC tube morphogenesis is markedly enhanced by the presence of fibroblasts145 that similarly may contribute interstitial matrix proteins or missing components of basement membrane matrix such as nidogen-1.148
Potential Role for Endothelial Cell Supporting Cells in the Regulation of Basement Membrane Matrix Assembly and Stabilization
Another important question is the extent to which ECs themselves can produce functional basement membranes versus the necessity of these cells to interact with supporting cells such as pericytes to regulate basement membrane matrix synthesis or assembly.149,150 Supporting cells such as pericytes may contribute molecules such as nidogen-1, which, along with perlecan, can facilitate the assembly of a stabilized basement membrane containing laminin-10 (which self-assembles), collagen type IV, perlecan, and nidogens 1 and 2.64 Nidogen-2 has been found to be present in higher amounts in EC basement membranes,64,151,152 and it is induced during EC tubular morphogenesis,153 although it is expressed at much lower levels compared with nidogen-1 basement membranes throughout most tissues.151,154
It is clear in many instances that epithelial–mesenchymal interactions regulate basement membrane assembly by cells such as epithelium.148,155–160 This is particularly apparent in the skin whereby keratinocytes are not able to assemble a basement membrane structure (as determined with electron microscopy and immunofluorescent staining) unless underlying fibroblasts are present. Fibroblasts have been shown to be a producer of nidogens148 that control interactions of laminins with collagen type IV isoforms in different basement membranes.64 In our system of EC tube morphogenesis in 3D type I collagen matrices over 3 to 5 days,26,143 we have never observed the development of complete basement membrane structures with EC cultures alone (electron microscopic analysis). If pericytes are required for EC basement membrane assembly, this would provide explanation for why these supporting cells are required for vascular stabilization.86,105,107,108,149,150,161 As such, EC basement membrane matrix assembly may closely resemble the parallel process of epithelial basement membrane assembly where mesenchymal cells are necessary. Further support for this possibility comes from findings that tubular morphogenesis and stabilization of endothelial versus epithelial structures appear highly related.44,48,162–164 Furthermore, in some situations, ECs themselves have been shown to produce ECM proteins that are typically classified as interstitial matrix proteins such as collagen type I, collagen type VI, and fibronectin.36,38,165 A major step in embryonic development involves epithelial–mesenchymal transition, and ECs can undergo a similar transition, for example, during cardiac valve development.166 In contrast, many differential gene array experiments using ECs undergoing morphogenesis in 3D matrices have revealed a much more prominent induction of basement membrane matrix–related genes (ie, epithelial-like genes).143,153 Yet another question relates to EC production of collagens that bridge EC basement membranes to the collagen type I interstitial matrix. Such proteins include collagen type VI, which through its NC1 domains can bind both collagen type IV and type I,115,167,168 as well as possibly collagen type VIII.169 ECs are capable of synthesizing collagen type VI, although it is possible that pericytes or other supporting cells such as fibroblasts may contribute collagen type VI to facilitate the linkage of EC basement membranes to interstitial matrices.
Limited information is available concerning the ability of perivascular cells such as pericytes to produce ECM components. It is clear that pericytes synthesize basement membrane matrix proteins, proteoglycans, such as decorin, biglycan, versican, and aggrecan, and fibronectin and various collagens.105,170–175 A key regulator of ECM synthesis in this and other contexts is transforming growth factor-ß, particularly in concert with connective tissue growth factor.176–178 Pericyte–EC interactions are known to markedly induce transforming growth factor-ß activation,179–182 which ultimately results in increased ECM synthesis, deposition, and decreased ECM turnover through secretion of proteinase inhibitors (ie, plasminogen activator inhibitor-1 and TIMP-1). Major questions remaining include whether pericytes are required for EC basement membrane synthesis and assembly or whether they are particularly involved in stabilization of these structures.
Immobilization of Angiogenic Cytokines by ECM Scaffolds
Finally, an interesting concept in EC/ECM biology involves immobilization of angiogenic cytokines on ECM scaffolds. Such immobilization may provide important cues for directional growth of EC sprouts during early morphogenesis. In addition, such growth factor–rich ECM scaffolds may provide the means for ECs to respond simultaneously to growth factor receptor and integrin cosignaling pathways (eg, through clustering of receptors). Most angiogenic cytokines have affinity for heparin and heparan sulfate and thereby become anchored to heparan sulfate proteoglycans (ie, syndecans, perlecan, versican) either on the surface of ECs or within the surrounding ECM.86,111,183 Recent work suggests that MMPs and serine proteinases such as plasmin can liberate functional VEGF-165 from heparan sulfate–containing ECM.184,185 Also, relevant angiogenic cytokines may be capable of directly binding to angiogenesis-promoting ECM scaffolds such as collagen type I and fibrin/fibronectin matrices.186 In support of this possibility are studies showing that VEGF can bind directly to fibronectin187 and that hepatocyte growth factor and keratinocyte growth factor each can bind directly to collagen type I and remain functional.188,189 Also, the ECM protein thrombospondin-1,190 which in some cases inhibits angiogenesis,191 binds various angiogenic cytokines such as fibroblast growth factor-2, VEGF, and hepatocyte growth factor192 and thereby may prevent their binding to proangiogenic ECM. Thus, angiogenic cytokines show direct binding affinities for angiogenesis-promoting ECM scaffolds that stimulate the process of vascular morphogenesis.
|
Acknowledgments |
|---|
This work was supported by NIH grants HL59373 and HL79460 (to G.E.D.), NIH grant CA77357 (to D.R.S.), and grants from the V. Kann Rasmussen Foundation (to D.R.S.).
|
Footnotes |
|---|
Original received June 23, 2005; revision received October 3, 2005; accepted October 11, 2005.
|
References |
|---|
|
TopAbstractIntroductionECM Function in EC...ECM Function in EC...Endothelial ECM: RemodelingEndothelial ECM Biosynthesis,...References |
|---|
- Ausprunk DH, Folkman J. Migration and proliferation of endothelial cells in preformed and newly formed blood vessels during tumor angiogenesis. Microvasc Res. 1977; 14: 53–65.[CrossRef][Medline] [Order article via Infotrieve]
- Dejana E, Languino LR, Polentarutti N, Balconi G, Ryckewaert JJ, Larrieu MJ, Donati MB, Mantovani A, Marguerie G. Interaction between fibrinogen and cultured endothelial cells. Induction of migration and specific binding. J Clin Invest. 1985; 75: 11–18.[Medline] [Order article via Infotrieve]
- Senger DR, Perruzzi CA, Streit M, Koteliansky VE, de Fougerolles AR, Detmar M. The alpha1beta1 and alpha2beta1 integrins provide critical support for vascular endothelial growth factor signaling, endothelial cell migration, and tumor angiogenesis. Am J Pathol. 2002; 160: 195–204.
[Abstract/Free Full Text] - Senger DR, Ledbetter SR, Claffey KP, Papadopoulos-Sergiou A, Peruzzi CA, Detmar M. Stimulation of endothelial cell migration by vascular permeability factor/vascular endothelial growth factor through cooperative mechanisms involving the alphavbeta3 integrin, osteopontin, and thrombin. Am J Pathol. 1996; 149: 293–305.[Abstract]
- Akiyama SK, Yamada SS, Chen WT, Yamada KM. Analysis of fibronectin receptor function with monoclonal antibodies: roles in cell adhesion, migration, matrix assembly, and cytoskeletal organization. J Cell Biol. 1989; 109: 863–875.
[Abstract/Free Full Text] - Meredith JE, Schwartz MA. Integrins, adhesion and apoptosis. Trends Cell Biol. 1997; 7: 146–150.[CrossRef][Medline] [Order article via Infotrieve]
- Giancotti FG, Ruoslahti E. Integrin signaling. Science. 1999; 285: 1028–1032.
[Abstract/Free Full Text] - Wary KK, Mainiero F, Isakoff SJ, Marcantonio EE, Giancotti FG. The adaptor protein Shc couples a class of integrins to the control of cell cycle progression. Cell. 1996; 87: 733–743.[CrossRef][Medline] [Order article via Infotrieve]
- Roovers K, Assoian RK. Integrating the MAP kinase signal into the G1 phase cell cycle machinery. Bioessays. 2000; 22: 818–826.[CrossRef][Medline] [Order article via Infotrieve]
- Seger R, Krebs EG. The MAPK signaling cascade. FASEB J. 1995; 9: 726–735.[Abstract]
- Vinals F, Pouyssegur J. Confluence of vascular endothelial cells induces cell cycle exit by inhibiting p42/p44 mitogen-activated protein kinase activity. Mol Cell Biol. 1999; 19: 2763–2772.
[Abstract/Free Full Text] - Assoian RK, Schwartz MA. Coordinate signaling by integrins and receptor tyrosine kinases in the regulation of G1 phase cell-cycle progression. Curr Opin Genet Dev. 2001; 11: 48–53.[CrossRef][Medline] [Order article via Infotrieve]
- Short SM, Talbott GA, Juliano RL. Integrin-mediated signaling events in human endothelial cells. Molecular Biology of the Cell. 1998; 9: 1969–1980.
[Abstract/Free Full Text] - Aplin AE, Short SM, Juliano RL. Anchorage-dependent regulation of the mitogen-activated protein kinase cascade by growth factors is supported by a variety of integrin alpha chains. J Biol Chem. 1999; 274: 31223–31228.
[Abstract/Free Full Text] - Fang F, Orend G, Watanabe N, Hunter T, Ruoslahti E. Dependence of cyclin E-CDK2 kinase activity on cell anchorage. Science. 1996; 271: 499–502.[Abstract]
- Assoian RK. Anchorage-dependent cell cycle progression. J Cell Biol. 1997; 136: 1–4.
[Free Full Text] - Zhu X, Ohtsubo M, Bohmer RM, Roberts JM, Assoian RK. Adhesion-dependent cell cycle progression linked to the expression of cyclin D1, activation of cyclin E-cdk2, and phosphorylation of the retinoblastoma protein. J Cell Biol. 1996; 133: 391–403.
[Abstract/Free Full Text] - Ilan N, Mahooti S, Madri JA. Distinct signal transduction pathways are utilized during the tube formation and survival phases of in vitro angiogenesis. J Cell Sci. 1998; 111 (Pt 24): 3621–3631.[Medline] [Order article via Infotrieve]
- Aoudjit F, Vuori K. Matrix attachment regulates Fas-induced apoptosis in endothelial cells: a role for c-flip and implications for anoikis. J Cell Biol. 2001; 152: 633–643.
[Abstract/Free Full Text] - Perruzzi CA, de Fougerolles AR, Koteliansky VE, Whelan MC, Westlin WF, Senger DR. Functional overlap and cooperativity among alphav and beta1 integrin subfamilies during skin angiogenesis. J Invest Dermatol. 2003; 120: 1100–1109.[CrossRef][Medline] [Order article via Infotrieve]
- Senger DR, Perruzzi CA. Cell