Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

During angiogenesis, the ECM dynamically evolves, changing and adapting to cellular processes that are taking place within its structure. As ECs differentiate into tubular structures containing lumens and associated pericytes, significant matrix remodeling occurs. Proteinases carve out areas to allow for invasion of the nascent tubular structures into the surrounding stroma, creating new vasculature in response to hypoxia. A new basement membrane is laid down, and supporting interstitial matrix is layered beneath the newly formed structures. This chapter will discuss angiogenesis as it relates to the surrounding ECM, the various proteinases that are able to modulate key steps of this process, and finally, the link between ECM mechanical properties and cellular remodeling during angiogenesis.

Vasculogenesis and angiogenesis are two distinctly different processes by which blood vessels form (Fig. 1). In embryonic development, angioblastic cells assemble into a primary capillary plexus to create nascent vasculature de novo via vasculogenesis. By contrast, angiogenesis refers to the formation of capillaries via branching from existing vasculature after initial embryonic development. This requires a complex series of events starting with basement membrane degradation of the existing vasculature, followed by endothelial cell activation, migration, and proliferation, organization into immature vessel sprouts with leading tip cells, maturation and vessel stabilization via mural cell association, and finally, basement membrane deposition and pruning of the new vessels in response to the physiologic demands of the tissue [1]. Each step in this process requires interaction between cells and their surrounding ECM.

Fig. 1
figure 1

Schematic depiction of the two different processes by which new blood vessels form in the body. Vasculogenesis typically occurs in development, with angiogenesis occurring throughout life. This chapter focuses on how the ECM is remodeled during the latter process. (Figure reproduced from [139] with permission from Elsevier.)

Full size image

A balance between pro- and anti-angiogenic proteins, known as the ‘angiogenic switch’, is crucial to the control of angiogenesis, with soluble factors and insoluble factors actin regulating this switch. When the scale is tipped in favor of molecules that inhibit angiogenesis, the switch is ‘off’ until the levels of activating (pro-angiogenic) molecules are increased and able to overcome the inhibiting molecules. In healthy adults, the switch is typically maintained in the ‘off’ position, unless a pathological state which requires the formation of new vasculature occurs, such as cancer, wound healing, or ischemic disease. (In cancer, tumor growth beyond a threshold is achieved in part by recruiting host vasculature; however, because a detailed discussion of tumor angiogenesis is beyond the scope of this chapter, readers should refer to other reviews on the topic instead [2].) Many signals can tip the switch in favor of angiogenesis, such as hypoxia, low environmental pH, mechanical stresses, tumor growth, or the presence of immune or inflammatory cells. Soluble growth factors, such as vascular endothelial growth factor (VEGF), platelet-derived growth factor (PDGF), basic fibroblast growth factor (FGF), and hepatocyte growth factor (HGF), are potent pro-angiogenic factors. Stabilizing immature vessels requires molecules such as angiopoietins-1 and -2, TGF-β, and sphyngosine-1 phosphate (S1-P), which are also considered pro-angiogenic. On the other side of the balance, suppressive signals from angiogenesis inhibitors include α-interferon, platelet factor-4, and thrombospondin-1, as well as other cryptic protein fragments [3, 4]. Matrix metalloproteinases (MMPs) and their endogenous inhibitors can also be considered in the context of the angiogenic switch, as they function as both pro- and anti-angiogenic molecules and are essential in each step of capillary formation and remodeling [5].

2 Changes in the ECM Accompany Each Stage of Angiogenesis

During the initial stages of angiogenesis, contact between ECs and the ECM is a key controller of angiogenic signaling. ECs must adhere to the ECM in order to properly migrate, a necessary requirement for initiating angiogenic sprouting [6]. The ECM immobilizes angiogenic cytokines, and thus coordinates signals transduced to ECs via both growth factor receptors and integrin cell adhesion receptors [6–8]. Integrins, in turn, can regulate EC proliferation, survival, and the formation of functional vessel lumens [8–11]. If the ECs fail to adhere to the ECM, proliferation ceases and angiogenesis thus also stops [7, 12–15].

At the earliest stages of angiogenesis, the basement membrane, consisting primarily of laminin-1 and type IV collagen, gets degraded to expose the ECs to the surrounding interstitial matrix [16]. In a quiescent state, the basement membrane helps insulate the ECs from this interstitial matrix and inhibits EC invasion and migration [7]. Following its degradation, a gradient of ECM components and the cytokines attached to them provide a set of cues to direct EC motility. In wound healing, for example, the interstitial matrix consists primarily of fibrin and type I collagen, and supports subsequent EC migration and sprouting [17]. VEGF plays a particularly critical role at this stage, as it is known to induce α1β1 and α2β1 integrin expression, both of which bind type I collagen. Type I collagen also helps to transform the leading ECs into a tip cell phenotype [18]. One mechanism known to disrupt angiogenesis for therapeutic or alternative purposes is to disrupt the formation of the collagen triple helix via alteration of the prolines. This results in a cessation of collagen recognition and binding by the ECs, effectively halting angiogenic invasion and tubule formation [19].

Once a nascent tubule escapes the basement membrane and begins to invade the interstitial matrix, extension of the capillary sprout begins. Type I collagen induces nascent cord formation and a migratory EC phenotype in part by suppressing cyclic AMP, which causes increased actin polymerization and stress fiber formation within the EC cytoskeleton [7]. This enables the ECs to generate substantial contractile forces and apply tension to the matrix over relatively large distances, which in turn supports capillary cord formation along the matrix fibers. Other cell types do not migrate and produce cords when implanted in fibrin or collagen gels in the same way [20]. Disruption of vascular endothelial cadherin (VE-cadherin) intercellular junctions via signaling mechanisms induced by collagen I binding also help to induce initial sprouting of ECs from a base vessel [16]. Disrupted of their quiescent cell–cell contacts, ECs begin to migrate and develop into nascent cords [19, 21].

The next steps in the angiogenic process include the formation of hollow lumens, followed by maturation of the nascent vessels. As mentioned previously mentioned, integrin-mediated interactions between ECs and collagen, fibrin, and fibronectin provide key instructive signals [22–24]. α2β1 and α1β1 are known collagen receptors, while αVβ3 and α5β1 are known fibronectin receptors that also permit EC interactions with fibrin. Lumen formation is dependent on the formation of these integrin-dependent intracellular vacuoles that are initially formed by the process of pinocytosis, or in which small vesicles formation form to create pockets within the cell. These vacuoles fuse together by exocytosis between adjacent ECs and start to direct an apical-basal organization and polarization [25, 26]. This polarization requires membrane type-MMP (MT-MMP) to interact with the ECM at the exterior of the newly formed lumens. The roles of these MT-MMPs will be discussed in greater detail later in this chapter.

The final step in the angiogenic process, tube stabilization, coincides with the production of laminin to form a new basement membrane along the basal surface of the nascent tubules. Integrins α6β1 and α3β1, which bind ECs to specific laminin isoforms, are known indicators of capillary maturation [27, 28]. The expression of these integrins suppresses several signaling pathways, and trigger EC quiescence [29, 30]. The laminin-rich basement membrane also provides an interface with which both ECs and stabilizing pericytes can interact [7].

3 Proteinases Involved in Angiogenesis

The ECM must be broken down and reformed for many processes throughout life, including embryonic development, various morphogenic processes, cellular reproduction, and tissue remodeling. The matrix itself serves as a platform for cell growth and support, but is also capable of controlling cellular attachment, proliferation, migration, and differentiation of cells via cell-ECM interactions. Many cytokines and growth factors can also be sequestered in the matrix and stored for later use. Several different types of angiogenic proteinases modulate the ECM in varying ways to influence angiogenesis. Some proteases indirectly promote EC proliferation, while others degrade the ECM to allow for tunneling ECs to invade and form tubules. Others control growth factor release from the matrix, altering cues that can direct or inhibit the angiogenic process. A final group of proteinases also controls cell adhesion to the matrix, inducing polarity within the blood vessels. These adhesions may direct cells to migrate, proliferate, or remain quiescent, based on the levels of varying integrins expressed, and the contents of the matrix to which they bind. MMPs are the main degradative enzymes responsible for modulating the ECM in a tissue. They are always contributing to the evolving matrix as it changes in different ways to support and encourage various cellular processes. In addition to both membrane-bound and soluble MMPs, ADAMs are another important group of proteins that influence ECM remodeling. A final grouping of players is the tissue inhibitors of metalloproteinases (TIMPs), which can control angiogenesis and subsequent matrix remodeling by maintaining vascular quiescence and halting angiogenic cues to maintain the angiogenic switch in the “off” position.

A. Soluble Matrix Metalloproteinases

Secreted matrix metalloproteinases are a family of zinc-containing endopeptidases that are able to degrade various ECM components. They are produced as pro-enzymes that are proteolytically processed to become activated. A sulfhydryl group in the pro-domain of all MMPs, known as a “cysteine switch,” is able to work in sync with the zinc ion of the catalytic site to maintain cell quiescence [31]. By disrupting this cysteine-zinc binding, the MMP takes the first step toward activation [32].

In general, the naming scheme follows a simple numerical order, starting with the first to be discovered, MMP-1, which acts on collagens. It was originally discovered when collagen gels were degraded by tadpole fin explants. The rest of the MMPs are divided into subgroups based on domain structure and substrate specificities: matrilysins (MMP-7 and MMP-26), collagenases (MMP-1, MMP-8, MMP-13, and MMP-18), stromelysins (MMP-3, MMP-10, and MMP-11), gelatinases (MMP-2 and MMP-9), enamelysins (MMP-20) and epilysins (MMP-28), and others that don’t fit into one of these subgroups (MMP-19, MMP-21, MMP-22, MMP-23, and MMP-27) [33–35]. MMPs can also be classified based on the basic domain groups that are included in their structures. All MMPs have three common structural domains: the “pre” or signal sequence, the pro-peptide domain, and the catalytically active domain (Fig. 2). MMPs-7 and -26 contain only these 3 domains, have a broad range of substrate specificity, and are able to degrade many ECM proteins [36–38]. Addition of a hemopexin domain connected to the core region via a proline-rich hinge region at the catalytic domain allows for enhanced substrate specificity over those containing only the three basic regions. This hinge region is also responsible for binding the family of specific MMP inhibitors (TIMPs). Some examples of hemopexin-domain containing MMPs are collagenases, which degrade the native helix of fibrillar collagens such as types I, II, and III, as well as stromelysin-1 and -2, enamelysin, metalloelastase, and MMPs-19, -22, and -27, which again don’t fit into a specific structural class of MMPs [33]. These stromelysins, with their hemopexin domains, still have a rather broad substrate specificity, degrading several groups of ECM proteins, including proteoglycans, fibronectin, and laminin [32]. Another similar grouping is the gelatinases (MMPs-2 and -9) containing three head-to-tail cysteine-rich fibronectin type II-like repeats within the catalytic domain [36, 39]. These MMPs also degrade types IV, V, VII, and X native collagens, as well as denatured collagen (gelatin), fibronectin, and laminin [39]. When the cysteine-rich repeats are instead furin-susceptible sites, stromelysin-3 and epilysin are classified as a grouping. Finally, there are two outlying MMPs, including MMP-21, which does not have a hinge region at all, and instead contains a vitronectin-like region within the propeptide region, and MMP-23, which also lacks the hemopexin domain, containing a cysteine and proline-rich region followed by an immunoglobulin-imitating region [40].

Fig. 2
figure 2

Schematic representation of the structures of MMPs and ADAMs. Both protease families contained conserved features, a “pre” or signal sequence, a propeptide domain (pro) (with either a cysteine switch or a furin-susceptible site), and a catalytic, Zn-binding domain. Additional sequences in some MMPs include a hinge region (H) and hemopexin domain, and other features not shown here. ADAMs contain a disintegrin domain, a cysteine-rich motif, an EGF repeat, a transmembrane domain, and a cytoplasmic tail

Full size image

Soluble MMPs can have both pro- and anti-angiogenic roles. Their pro-angiogenic capacity is perhaps more obvious, given their ability to degrade ECM components and jumpstart the path toward angiogenesis. Many growth factors and cytokines are known to upregulate EC basement membrane degradation, as well as EC proliferation, migration, and differentiation into a pro-angiogenic phenotype. Examples of these molecules are VEGF, bFGF, and several interleukins, which increase the amounts of inactive [41] and active[42] MMP-2 and MMP-9 [43]. However, because certain ECM cleavage products have anti-angiogenic properties, MMPs may also be considered anti-angiogenic as well [44].

1. Pro-angiogenic roles of soluble MMPs

During basement membrane degradation, a subgroup of ECs, known as “tip cells”, initiate sprouting. These cells possess high proteolytic activity, enabling them to successfully break down the matrix and tunnel through the interstitial ECM and hypoxic tissue [45]. Upon signaling to initiate sprouting, the tip cells must first proteolyze the capillary basement membrane, which is primarily comprised of laminin, collagen IV, heparin-sulfate proteoglycans, and entactin [46]. Multiple MMPs can degrade these ECM components: MMP-2, MMP-3, MMP-7, MMP-9, MMP-10, MMP-12, and MT1-MMP [33]. As mentioned previously, two important growth factors in the initiation of the angiogenic cascade, VEGF and bFGF, produce vesicles containing pro-MMP-2 and pro-MMP-9, as well as MT1-MMP. Upregulation of these MMPs is thus associated with increased basement membrane invasion abilities of ECs [47].

After the basement membrane has been broken down, ECs induce MMP expression from interstitial cells by secreting extracellular matrix metalloproteinase inducer (EMMPRIN) [48]. The majority of MMP production may be from these surrounding interstitial and inflammatory cells present in the matrix, rather than the ECs forming the actual new capillary sprouts. Interstitial flow from the vasculature to the lymphatics, which is enhanced following degradation of the basement membrane barrier, combined with this increased MMP production, creates chemotactic gradients that further encourage EC invasion into the ECM. This is perhaps achieved by the interaction of various ECM breakdown products with the surface of the ECs [49]. Despite significant differences in pathologic and physiologic microenvironments, only small changes in soluble MMP expression are observed [50]. For example, switching from a physiologic ECM containing mostly collagen I to a provisional ECM comprised of fibrin, fibronectin, and vitronectin typically found during wound healing or prolonged ischemic diseases, results in slight modulation of the MMP expression profile [51].

Recent work suggests that there may be no single proteolytic mechanism utilized by ECs to degrade the ECM. Instead several MMPs are likely used in complement, and the specific combination and ratio of MMPs expressed and utilized depends on the identity of the matrix as well as the identity of stromal cell types that interact with the ECs. For example, when adipose-derived stem cells or a generic fibroblast source were included as interstitial cells to induce capillary sprouting in a fibrin matrix, the plasminogen activator-plasmin axis was the preferred proteolytic mechanism utilized for capillary invasion into the matrix and tubule lengthening, while MMPs appeared to play a distinct role regulating capillary diameter and stabilization only. In contrast, when mesenchymal stem cells from bone marrow were used in place of the adipose-derived cells, MT-MMPs were the sole proteases for ECM invasion and sprouting [52].

Further work illustrating knockdown of either of the gelatinases, MMP-2 and MMP-9, suggests that these two proteinases may work in concert to remodel the ECM during angiogenic processes. When one of the two is targeted for gene knockdown, sprouting is still able to occur. MMP-9 is unable to degrade type I collagen alone, so thus it does not serve to encourage tunneling and sprouting of ECs during angiogenesis via matrix proteolysis directly. Instead, its pro-angiogenic capacity may lie in its ability to release bound VEGF (secreted by stromal cells) from the matrix to induce sprouting. It is also capable of activating TGF-β, resulting in promotion of tissue remodeling [53, 54]. During in vivo wound healing and hind limb ischemia studies, the peak activity levels of these MMPs coincide with granulation tissue formation, fibroblast migration into the tissue, and vascularization of the wound [55, 56]. Several research groups have now fabricated synthetic hydrogels with linkages sensitive to MMP-2 and MMP-9 so that cellular invasion can occur in much the same way as in hydrogels of natural composition (e.g., collagen or fibrin). In vitro studies using RGD-functionalized version of these MMP-sensitive gels have demonstrated EC adhesion and capillary sprouting by mimicking key elements found in natural matrix proteins [57, 58]. Furthermore, tethering growth factors such as VEGF to the matrix via proteolytically sensitive linkages recapitulates the growth factor sequestration capacity of physiologic ECMs [59].

3. Anti-angiogenic roles of soluble MMPs

As mentioned previously, MMPs can be considered to be both pro- and anti-angiogenic. MMP expression and activity can impede blood vessel formation via one of two possible mechanisms. First, overactive MMPs can compromise ECM stability, which may result in vascular regression. Second, MMP activity can generate matrix fragments with anti-angiogenic capabilities.

With respect to the first possibility, the process of angiogenesis typically culminates with vascular pruning. During this process, vessels that have not been stabilized regress as some of the interstitial collagen is broken down. Specifically, plasmin-mediated activation of MMPs-1, -10, and -13 has been shown to induce vascular regression as each of these MMPs (although predominantly MMP-1) can digest interstitial collagens [60]. This subset of MMPs, in conjunction with MMPs-2, -9, and MT1-MMP, may act together to digest multiple different ECM components based on these enzymes’ specificities for different substrates [60, 61]. The first subset breaks down native type I collagen, while the gelatinases more efficiently degrade denatured collagens. A direct correlation between the levels of active MMPs in maturing capillary beds and the levels of vascular regression has been reported [62, 63].

With respect to the second possibility, the proteolytic degradation products of MMPs are anti-angiogenic. Proteolytic degradation of collagen IV, one of the primary components of basement membrane, generates anti-angiogenic fragments that include arrestin, canstatin, tumstatin, and metastatin [34]. MMP-9 predominantly produces free tumstatin, as well as smaller amounts of arresten and canstatin. Other MMPs, including MMP-2, -3, and -13, are also able to liberate tumstatin, although not as efficiently as MMP-9 [64]. Tumstatin targets the αVβ3 integrin, which is not expressed at measurable levels in physiologic angiogenesis, but is seen at much higher levels in tumor angiogenesis. Studies have explored the possibility of using tumstatin to reduce pathologic angiogenesis [64]. Arrestin, another collagen IV breakdown fragment, binds the α1β1 integrin receptor for collagen I, and inhibits EC proliferation, as well as migration and further tube formation in vitro. Similarly, collagen XVIII is a component of the interstitial matrix beneath the basement membrane of the vasculature. Collagen XVIII breakdown products are endostatin and neostatins, which are small, varying molecular weight molecules that are the further breakdown products of endostatin. MMPs-3, -7, -9, and -13, as well as MT1-MMP, act on collagen XVIII to produce these fragments. Endostatin affects VEGF signaling, EC proliferation and migration as well, in part by acting on the α5β1 integrin [65–68]. Another unique function of endostatin is its ability to inhibit MT1-MMP and MMP-2 activities [69].

B. Membrane-Type Matrix Metalloproteinases (MT-MMPs)

The membrane-type MMPs (MT-MMPs) represent another grouping of MMPs, so named because they are bound to the cell’s plasma membrane via either a C-terminal transmembrane domain or a glycophosphatidyl inositol (GPI) anchor [39, 70]. Both classifications include a “pre” region, a propeptide region with a furin-susceptible site, a catalytic domain, a hinge region, and a hemopexin domain. After furin activation intracellularly, the proteinase gets processed and sent to the cell membrane, where the catalytic, hinge, and hemopexin domains lie extracellularly. These external regions are held at the cell surface by a transmembrane region attached to a short amino acid tail residing in the cell cytoplasm, or a GPI domain that is fixed in the cell membrane. These MT-MMPs provide spatial control of matrix breakdown directly at the cell membrane surface [33]. MT-MMPs degrade gelatin, fibronectin, and aggrecan, as well as several other ECM substrates [37, 38].

Once nascent capillaries have formed, a basement membrane, a hallmark of a more mature vessel network, is deposited. Pericytes play a key role in this process. These cells are recruited from the host stroma in part via the secretion of PDGF-β by ECs. PDGF receptor-β signaling then initiates a cascade of events that control pericyte-EC binding, migration to the site, and proliferation. Interestingly, all of these processes are in some manner controlled by MT1-MMP [71–73]. Co-cultures of ECs and stromal cells of various origins in 3-D gels yield stable, pericyte-invested networks of capillaries characterized by the presence of basement membrane subjacent to the ECs along with the periendothelial location of the stromal cells (Fig. 3) [74]. In the absence of stromal cells, ECs express MT1-MMP, with very little basement membrane production, even at later time points. If stromal cells are included with ECs in culture, MT1-MMP expression is undetectable in the capillary stalks. The ECs in the stalk instead produce basement membrane proteins, and the expression of MT1-MMP is restricted to the tip cells. Similar expression profiles have also been see in vivo [45].

Fig. 3
figure 3

Three-dimensional co-cultures of ECs and stromal cells generate stable, pericyte-invested capillary networks in vitro. In these experiments, bone marrow-dereived mesenchymal stem cells (MSCs) MSCs were distributed throughout a fibrin-based 3-D ECM in the presence of microcarrier beads coated with ECs. a After 7 days, cultures were fixed and IF stained at day 7 for F-actin (green) and collagen IV (red) to visualize basement membrane deposition (white arrows). Scale = 50 μm. b MSCs expressing GFP were interspersed with mCherry-transduced ECs. Physical association of the ECs and MSCs were observed (white arrows). Scale = 25 μm. c Cultures containing mCherry-transduced ECs and MSCs were fixed and IF stained for pericyte markers αSMA (aqua) and NG2 (white). DAPI-stained nuclei are visible in the blue channel. Scale = 50 μm. (Figure reproduced from [74] with permission from Elsevier.)

Full size image

Further examination of the newly deposited basement membrane production yields some other interesting observations. The EC TIE-2 receptor is preferentially expressed in the stalk portion of a maturing capillary sprout, but are notably absent in the tip ECs. This expression is thought to be controlled by the pericytes, which produce Ang-1 to signal to the ECs via the TIE-2 receptor [45, 75]. This Ang-1/TIE-2 interaction is one mechanism by which pericytes and SMCs can regulate MMP activity of ECs. An alternative, and much more direct, means to achieve this control is via TIMP-1 secretion. TIMP-1 inhibits soluble MMPs, but also appears to stabilize vessels by inducing basement membrane protein production [76]. Other TIMPs also work in conjunction with both cell types to inhibit angiogenic sprouting and stabilize nascent vessels. TIMP-3 is secreted by perivascular cells, and is then presented to the ECs via heparin sulfate proteoglycans in the basement membrane, which suppress MT1-MMP activity and encourage sprouting of the ECs (Fig. 4) [77, 78]. Furthermore, all EC interactions with associating pericytes via Ang-1 or PDGF serve to inactivate and inhibit MT1-MMP after the basement membrane has been reproduced and to ensure vascular quiescence.

Fig. 4
figure 4

Schematic diagram illustrating the contribution of TIMP-2 and -3 to pericyte-induced vascular tube stabilization, as proposed in [77]. TIMP-2 is derived from ECs, whereas TIMP-3 is produced by pericytes. Together, they contribute to vascular stabilization by inhibiting a variety of MMPs, ADAMs, and VEGFR-2. The initiation of tube stabilization requires the blockade of both EC tube formation and EC tube regression, which further leads to the cessation of EC activation and the development of EC quiescence. Pericytes are required for ECs to assemble basement membrane matrices, which may locally capture and present TIMP-3 to ECs through heparan sulfate proteoglycans. (Figure reproduced from [77] with permission from The Rockefeller University Press.)

Full size image

C. A Disintegrin and metalloproteinase (ADAM)

ADAMs are a family of secreted and transmembrane proteins that control cell adhesion, as well as proteolytic processing of the ectodomains of cell surface receptors and signaling molecules [79]. Like previously described MMPs, ADAMs have both pre- and propeptide domains, with the pro- domain acting as an intramolecular chaperone that controls protein folding [80] and enzyme latency via a cysteine-switch mechanism [81, 82]. At this point, the structures differ, with ADAMs having a disintegrin-like domain with a loop that is able to interact with neighboring cell integrins [83]. Following this region, a cysteine-rich domain, then an EGF-like domain, followed by a membrane-spanning region and a cytoplasmic tail. The cytoplasmic domain is able to interact with proteins of intracellular signaling importance, as well as to control trafficking of proteins (Fig. 2). ADAMs are named such due to their original structural homology to the small proteins of hemorrhagic snake venoms that were able to bind platelet integrin α2bβ3a to block platelet aggregation [84].

Because ADAMs are an active family of metalloproteinases, they are able to cleave ECM proteins and cause degradation of the bulk ECM in a locale. One example of this is ADAM-9, which is able to cleave laminin and promote invasion [85]. Several studies have shown a connection between ADAM-10 and cleavage of adhesion molecules such as VE-cadherin, where the ADAM is able to disassemble the junctional contacts that control permeability and assist with encouragement of EC migration and tubule sprouting [7, 86]. Other ADAMs that are known to control ECM degradation and release activators of ECs to a migratory phenotype that will start the angiogenic cascade, are ADAMS-15 and -17 [7]. ADAMs are also able to induce shedding of adhesion molecules such as PECAM-1, or their activity may mobilize growth factors, chemokines, or other soluble factors that can influence angiogenic processes [87].

D. Inhibitors of Matrix Metalloproteinases

Tissue inhibitors of metalloproteinases (TIMPs) are the main enodogenous inhibitors of MMPs. There are four mammalian TIMPs that have been identified and characterized within the literature. They are all known to regulate MMP activity during periods of tissue remodeling, with molecular weights between 20 and 29 kDa [88]. All TIMPs inhibit MMP substrates in a 1:1 stoichiometric ratio [89], with each TIMP binding the active site cleft in the catalytic domain of an MMP, in the same manner as an ECM substrate would bind the MMP [40]. Each TIMP has disulfide bonds of a three loop N-terminal domain, which is where interaction with the catalytic domains of MMPs occurs, and a complementary three loop C-subdomain [90].

All TIMPs are secreted proteins, but TIMPs-2, -3, and -4 can all be found near the surface of a cell, in association with different MMPs [91]. All four TIMPs are known to inhibit active forms of all MMPs; however, their inhibition abilities vary widely. The main exception to this rule is that TIMP-1 is a poor inhibitor of MMP-19, MT1-MMP, and MT3-MMP [92]. TIMP-3 easily inhibits many of the ADAMs [93]. Examples of preferential binding are the ability of TIMP-1 to preferentially bind with pro-MMP-9, and TIMP-2 preferentially binding with pro-MMP-2 to inhibit conversion to active forms of these MMPs [94, 95].

TIMP-2 has a special functional role in controlling the activation of pro-MMP-2, with MT1-MMP also acting as a modulator. This activation step takes place on the cell surface, thus the need for inclusion of MT1-MMP for activation [96]. According to Strongin et al., the increased activation of MMP-2 in the presence of TIMP-2 is the result of the N-terminal inhibitory domain of TIMP-2 binding to the active site of MT1-MMP, and the C-terminal domain of TIMP-2 interacting with the C-terminal hemopexin domain of pro-MMP-2 [97]. An additional unique feature of TIMP-2 is its ability to suppress angiogenesis by reducing EC proliferation cues from bFGF via its C-terminal region [98].

TIMP-3, as briefly mentioned earlier, can inhibit both MMPs and the ADAM family of proteinases [40]. TIMP-3 is also known to block VEGF to VEGFR-2 binding, which further contributes to the anti-angiogenic capabilities of TIMP-3 [99]. A third unique property of TIMP-3 is its restricted diffusion caused by its tight binding to heparin sulfate proteoglycans in the surrounding ECM. Because it does not readily diffuse, it is thought that TIMP-3 instead functions to regulate angiogenesis after the angiogenic switch has been flipped into an “on” position. As MMP degradation of the ECM liberates the matrix-bound TIMP-3, it can then inhibit any subsequent MMP activation at the cell-ECM level [100]. Normally, TIMP-3 functions to form complexes via its C-domain with MMPs-2 and -9, thus effectively slowing pro-angiogenic signaling [90].

TIMP-4 is found mainly in the human heart, with low levels of the inhibitor found in the kidney, colon, placenta, and testes [101]. Levels are dysregulated in various cardiovascular diseases. This TIMP functions mainly to reduce EC motility, as well as proliferation, and induce apoptosis as well. In in vivo models, addition of TIMP-2 results in suppression of angiogenesis, while addition of TIMP-4 is not able to have this same effect [102]. TIMP-4 deficient mice instead showed reduced cardiac function with aging, due to increased apoptosis of cells [103].

An additional inhibitor of MMPs, while not in the TIMP family, is the GPI-anchored glycoprotein, reversion-inducing cysteine-rich protein with kazal motifs (RECK). It is known to inhibit the release of pro-MMP-9 from the EC surface. It also effectively inhibits MT1-MMP, which will result in inhibition of MMP-2, as previously discussed [104, 105]. In RECK knockout mice, blood vessels cannot reach a mature stage, and mice die in utero. Overexpression of RECK in tumor models results in a reduction of new blood vessel sprouting to sufficiently nourish the tumor [88]. An additional proteinase inhibitor, α2-macroglobulin, is the primary MMP inhibitor found in blood plasma [106]. Finally, thrombospondin-1, a known inhibitor of angiogenesis, is also known to inhibit pro-MMP-2 and pro-MMP-9 from becoming activated. Thrombospondin-2 is also known to complex with MMP-2 to increase clearance via receptor-mediated endocytosis [106].

A schematic summarizing many of the effects of soluble and membrane-bound MMPs, as well as the TIMPs, is shown in Fig. 5.

Fig. 5
figure 5

A simplified schematic of the current concepts depicting the roles and regulation of MMPs as they pertain to matrix remodeling. Figure reproduced from [140] with permission from The American Physiological Society.)

Full size image

4 Effects of Stromal Cells on the ECM

A. Stromal cells influence ECM synthesis and degradation

It has been widely established in the literature that the presence of pericytes covering EC tubules results in stabilization of the vessels, a decrease in vascular pruning, and decreased permeability of the nascent vessels [107]. Pericytes are a source of angiopoietin-1, which acts on EC TIE-2 to stabilize these heterogeneous cell–cell junctions [108]. Recent findings have shown that EC-pericyte interactions occur after ECs carve out “vascular guidance tunnels” within the ECM, which provide physical space for the EC-pericyte interactions to take place. Stratman et al. showed that the pericytes are recruited to the ablumenal surface and are able to move along the EC tubules through these pre-formed spaces to regulate tubule maturation and vascular basement membrane assembly [109, 110]. Basement membrane is deposited between ECs and pericytes within these tunnel voids. Further work by this group showed that PDGF-BB and HB-EGF are necessary for pericytes to accumulate within these tunnels and along EC tubules, and also for proper basement membrane deposition. Without these growth factors directing the behavior of the pericyte-EC interactions, their data suggest that no basement membrane will be laid down [111]. This growth factor-pericyte interplay is known to be regulated by MT1-MMP. Active MT1-MMP directly controls the binding to PDGF receptor-B of pericytes after PDGF- β secretion by ECs. This cascade of events then controls pericyte migration to sites of need along the vasculature, as well as proliferation to induce greater vasculature stabilization and angiogenic quiescence [71, 73]. VEGF, an important initial cue for increased vessel permeability followed by basement membrane degradation and EC sprouting, also affects pericyte coverage. Treatment with VEGF antagonists results in increased pericyte coverage and improved microcirculatory function with lower permeability [112, 113].

The TIE-2 receptor of ECs on the stalks of sprouting capillaries is a signaling receptor that results in collagen IV expression throughout the stalk but not in the tip cells of capillary sprouts. It is thought to be regulated by Ang-1 in the presence of vascular smooth muscle cells (vSMCs) [45]. When vSMCs are absent, the distinction between stalk and tip cells is eliminated, and MT1-MMP is expressed throughout the structure [45]. Pericytes in co-cultures with ECs also express TIMP-1, which inhibits the activity of soluble MMPs and promotes basement membrane deposition by ECs [76].

As mentioned earlier, pericyte contact with EC tubules reportedly plays a role in basement membrane deposition. Basement membrane proteins such as laminins and collagen IV are produced with the assistance of pericytes. In quiescent vessels, these basement membrane proteins inhibit tube morphogenesis by blocking access to the interstitial collagens that encourage EC migration and invasion. When pericytes are not present to encourage laminin deposition, ECs remain migratory, and vessel stabilization does not occur. Further work has also shown that EC-pericyte interactions induce production of fibronectin, nidogen-1, nidogen-2, and perlecan, as well as the laminins and collagen IV on the ablumenal surfaces of nascent EC tubules, all of which are constituents of basement membrane [109]. Additional work in this area showed that ECs produced the elevated levels of fibronectin found during these EC-pericyte interactions, while nidogen-1 was produced by the pericytes. These two proteins are both known to be binding molecules for the basement membrane specifically [7, 114]. Furthermore, fibronectin binds collagen IV and perlecan, and nidogen-1 binds collagen IV and laminin. Thus, fibronectin and nidogen-1 appear to be essential for assembly of the full basement membrane [114]. Collagen IV is mainly responsible for basement membrane structural support, and binds predominantly fibronectin and nidogen-1, and self-assembly of fibronectin that is induced by pericytes has a direct effect on collagen IV assembly as well [109].

Most mature capillary networks within the body have approximately 20–25 % of their area covered by pericytes [115]. Despite the fact that pericytes cover only about one fourth of the total EC tubule area, they appear to control basement membrane production and maintenance along the full length of each tubule (see Fig. 4). This may be due to their ability to move along the tube surface [115]. Collectively, these results suggest that heterotypic cell–cell contacts between ECs and pericytes, along with the constant pericyte motility along nascent tubules, are required events for basement membrane deposition and ultimate vessel stabilization.

While conventional wisdom suggests that stromal cells that differentiate into pericytes promote vessel stabilization, basement membrane production, and maturation of the interstitial ECM, there are several cases where stromal cells influence ECM protein degradation and subsequent remodeling. The recruitment of mural cells to the stalks of nascent EC tubules anatomically distinguishes tip cells from stalk cells, and leads to the further production of MMPs by the tip cells to facilitate their ability to break down and invade the matrix ahead of them. In aging adults, pericyte coverage can decrease, causing ultimate ECM breakdown and angiogenic activity increases [116]. Similar activities take place in various ischemic diseases and pathologies such as diabetes.

B. Pericytes modulate integrin-mediated EC-to-ECM attachment

Integrins mediate EC attachment to the ECM, and cross-talk with several growth factor receptors, and thus are of critical importance in the process of angiogenesis. ECs express a wide range of integrins depending on the location and state of activation of the cells. The repertoire of integrins expressed by quiescent ECs enables adhesion to components of an intact basement membrane. On the other hand, angiogenic ECs upregulate the expression of a small subset of integrins, most notably the αvβ3 heterodimer, which has been reported to be required for angiogenesis [117]. It has been proposed that this integrin is upregulated in angiogenic ECs to permit the binding to a wide range of provisional matrix components, including fibrinogen, vitronectin, von Willebrand factor, and fibronectin. Consistent with a requirement for this integrin in angiogenesis, a landmark study showed that αvβ3 integrin antagonists promote tumor regression by inducing apoptosis of angiogenic blood vessels [118]. There is some evidence that integrin α5β1 expression is also upregulated at the initiation of angiogenesis, perhaps because it allows the newly migrating ECs that are sprouting from a pre-existing vessel to bind the aforementioned components of the provisional matrix [119, 120]. A complete discussion of integrins and their roles in both normal and pathologic angiogenesis is beyond the scope of this chapter. (For an excellent review, the reader is referred to a paper by Stupack and Cheresh [121].)

However, two facts regarding integrin expression are particularly relevant in the context of this chapter on ECM remodeling. The first is that the integrin expression profiles of ECs undergoing angiogenesis are different than those of quiescent ECs, as noted in the preceding paragraph. This is due to the fact that angiogenic ECs are exposed to an interstitial matrix whose composition is significantly different than a basement membrane, as noted in the preceding paragraph. The second is that pericytes appear to also be involved in the regulation of EC integrin expression patterns. ECs that are in contact with pericytes express α5β1 to attach to the basement membrane. When pericytes are not present, expression of this integrin is down-regulated, and the nascent vessels remain in an unstable immature state (perhaps more appropriately called an angiogenic state versus a quiescent state). Other integrin subunits with high affinities for basement membrane components (i.e. nidogens, laminins, and collagen IV), including the α3, α6, and α1 subunits, are also not expressed by the ECs in appreciable amounts when pericytes are absent. In contrast, α2 integrin, which recognizes collagen I (a component of the interstitial ECM, but not the basement membrane), is expressed at high levels in ECs cultured alone in 3-D collagen matrices, but is down-regulated in EC-pericyte co-cultures after pericytes have been recruited to initiate vessel maturation [109]. This study by Stratman et al. also showed that a strong induction of α3, α5, and α6 integrin mRNAs in ECs, combined with a dramatic increase in pericyte α1, α3, and α6 integrin mRNAs only when ECs and pericytes were cultured together and coincident with the deposition of basement membrane matrices [109]. Thus, there appears to be a complex and dynamic crosstalk between the ECs and pericytes that governs the integrin expression profiles of both cell types. These dynamic changes in integrin expression profiles permit EC adhesion to the interstitial matrix during the initial sprouting phases of angiogenesis, followed by the subsequent adhesion of both cell types to the EC-deposited basement membrane as the capillaries stabilize and mature.

5 Linking ECM Mechanical Properties and Remodeling

The ECM’s ability to regulate angiogenesis is complex and multivariate. Not only does it impact capillary morphogenesis via biochemical regulatory mechanisms, including through growth factor sequestration, integrin-mediated adhesion, and protease susceptibility, it also acts as an instructive structural framework to support sprouting and nascent capillary functionality. An additional feature of the ECM that has been proposed as a regulator of angiogenesis is its mechanical resistance to cell-generated tractional forces [122]. Evidence from several studies corroborates the idea that mechanical cues directly impact tubulogenesis [123–130]. Vailhe et al. demonstrated that human umbilical vein ECs seeded on top of 2-D fibrin gels varying in concentration from 0.5 to 8 mg/ml only formed capillary-like structures on the softest of the gels. The authors concluded that the ECs do not form capillary-like structures on the more rigid gels because the cells are unable to generate the necessary contractile forces to remodel the more rigid substrate [125]. Another study conducted by Deroanne et al. showed that ECs seeded on collagen-functionalized polyacrylamide gels of different stiffness change morphologies from a monolayer to a tube-like phenotype as substrate rigidity decreases [123]. In 3-D culture, Urech et al. investigated angiogenic process extension in 3-D fibrin gels and manipulated their mechanical properties by adding exogenous factor XIII to form additional cross-links [127]. Sieminski et al. also studied the 3-D formation of capillary-like structures by two different types of ECs in freely-floating versus mechanically-constrained (attached) collagen gels, and concluded that changing the collagen concentration modulates the formation of these structures by regulating the amount of traction force exerted by the cells [124]. Further evidence links EC-generated traction forces with branching [128], the formation of capillary-like structures [123, 124, 131], and the transcriptional control of soluble pro-angiogenic molecules [129]. A more detailed discussion of the role of ECM mechanics and EC-generated tractional forces is found elsewhere in this book.

In the scope of this chapter, it is important to recognize that the mechanical properties of the ECM are highly dynamic due to active remodeling induced both by the ECs and stromal cells. A study by Lee et al. used second harmonic generation and two-photon excited fluorescence to show that ECs induce quantifiable alterations in local collagen matrix density via a process that involved cell-generated forces [132]. Another study by Krishnan et al. tracked changes to the ECM during the process of angiogenesis using a 3-D culture model [133]. The authors reported an overall softening of the ECM as MMP activity increased during the initial sprouting phase, and then a slow stiffening as the MMP activity held steady and sprouts increased in length during tubulogenesis. Other studies have suggested that matrix stiffness may also indirectly modulate MMP activity in ECs [134–136]; however, the underlying mechanisms linking ECM mechanical properties and protease expression and/or activity remain to be elucidated.

In addition to EC-generated forces regulating angiogenesis, a recent study by Kilarski et al. reported that external forces generated by myofibroblasts pulling on the ECM during wound contraction mediated the formation of vascular loops by pulling on preexisting vascular beds [137]. This process, known as intussusceptive microvascular growth [138], demonstrates that stromal cells influence the process of angiogenesis in multiple ways. Not only do they secrete pro- and anti-angiogenic soluble factors, some of which influence the expression and activity of certain MMPs, and act as pericytes to stabilize the nascent vasculature, they also generate local forces that dynamically remodel the ECM and preexisting tissue structures. Understanding the complex interplay between ECs, stromal cells, and the ECM remains an ongoing challenge for the field.

6 Conclusions

The ECM continuously remodels in response to multiple instructive cues during the complex process of capillary morphogenesis. From early development of the capillary plexus in vasculogenesis, to the angiogenic sprouting of new vasculature in ischemic tissue in adults, many different proteinases work in concert with ECs and stromal cells to drive matrix breakdown, capillary sprouting, and subsequent maturation. Various soluble MMPs, membrane-bound MMPs, ADAMs, and inhibitors of each of these active players, play important roles to maintain the balance between pro- and anti-angiogenic cues in quiescent vessels, and to tip the scales to induce capillary morphogenesis and blood vessel development when needed. An increasing body of literature strongly suggests that pericytes are not only essential in promoting the stabilization and long-term functionality of capillary networks, but they can also exert their influence on vessel formation in a multitude of ways. As discussed here, pericytes also dynamically communicate with ECs to influence matrix proteolysis, synthesis, integrin expression profiles, and the mechanical properties of the interstitial matrix, all of which can influence angiogenesis. Further work is needed to dissect the exact roles of these various players on matrix remodeling and angiogenesis.