Apoptotic Cues from the Extracellular Matrix: Regulators of Angiogenesis

Apoptotic cues from the extracellular matrix: regulators of angiogenesis

Dwayne G Stupack1 and David A Cheresh*,1

1Department of Immunology, The Scripps Research Institute, 10550 N Torrey Pines Road., La Jolla, CA 92037, USA

A variety of factors cooperate to regulate neovessel giogenic factors initially favor the initiation of endothe- formation and persistence. Proangiogenic growth factors lial proliferation and invasion; however, these signals have remained an area of intense interest due to their are attenuated by the subsequent generation of anti- capacity to promote endothelial cell (EC) proliferation angiogenic factors (many produced by the endothelial and to initiate the angiogenic program. These growth cells themselves).This is best exemplified during the final factors are associated with increased cell survival, yet phase of angiogenesis, a process termed ‘vascular paradoxically, angiogenic ECs are more susceptible to pruning’, in which endothelial cells (ECs) that have apoptosis than quiescent ECs. Survival is regulated by failed to mature are eliminated (Keshet and Ben-Sasson, cooperation between growth factor receptors and integ- 1999). rins, which are in turn governed by the composition of the Angiogenic ECs are more susceptible to apoptosis local extracellular matrix (ECM). Integrin-mediated than their quiescent counterparts (Brooks et al., 1994; signaling is altered or disrupted by the presence of soluble, Kerbel et al., 2000). The continued survival of rather than matrix-bound ligands, thus providing a means angiogenic ECs is intimately tied to their local micro- by which ECM remodeling can influence both integrin- environment and, in particular, the presence of survival- and growth factor-mediated events. Ultimately, the promoting growth factors and extracellular matrix collaboration of these factors determines whether ECs (ECM) proteins. De novo protein synthesis/secretion, survive or die, thereby regulating neovascularization. deposition of ECM from plasma, and liberation of Oncogene (2003) 22, 9022–9029. doi:10.1038/sj.onc.1207110 protein fragments and growth factors from pre-existing ECM by proteases contribute to the dynamic restructur- Keywords: angiogenesis; apoptosis; endothelial cell; ing of the vascular microenvironment during angiogen- extracellular matrix; integrin esis.Blockade of these events can be sufficient to induce apoptosis and block an angiogenic response.For example, agents which disregulate growth factor receptor- or adhesion receptor-mediated signaling events can induce apoptosis in angiogenic ECs (Brooks et al., 1994; Angiogenesis is regulated by apoptosis in vivo Witte et al., 1998; Klement et al., 2002). It is worth noting that these interventions actively induce apoptosis The growth of new blood vessels plays a critical role in in the ECs, rather than simply suppressing the induction physiological (and developmental) processes, but is also (or execution) of proliferative or invasive events central to the progression of many diseases.Disorders required for angiogenesis.Thus, it appears that apop- such as cancer and inflammatory disease exploit tosis naturally regulates angiogenesis in vivo, acting as angiogenesis as a means to support expanding tumor the principle mechanism for ‘vascular pruning’ (Keshet and aberrant tissue growth, respectively.Angiogenesis and Ben-Sasson, 1999).Pruning can occur in a fraction contributes to other pathologies, including diabetic of neovessels, or all of them, since some neovasculariza- retinopathy, macular degeneration, and endometriosis. tion events are transient in nature.A good example of However, many endogenous antiangiogenic factors have this is neovascularization induced very rapidly by the now been isolated, suggesting that angiogenesis is a vascular endothelial growth factor (VEGF) during tightly controlled process regulated by the relative follicle-associated angiogenesis in the ovary.Concomi- quantities of proangiogenic and antiangiogenic factors tant ECM alterations and ovum release coincide with an (Folkman, 1995).Although pathological angiogenesis is abrupt halt to VEGF production, the termination of a rare event, the degree of neovascularization associated angiogenesis, and complete vascular regression via with disease is often a diagnostic and/or prognostic apoptosis (Dickson et al., 2001). Developmental angio- indicator of disease progression (Craft and Harris, 1994; genesis occurring during lens formation in the mamma- Gasparini et al., 1998). lian eye proceeds in a similar manner (Mitchell et al., During physiological forms of angiogenesis, the 1998).In contrast, many angiogenic events are perma- balance of proangiogenic and antiangiogenic factor nent and, in these cases, many vessels remain unaffected production changes as angiogenesis progresses.Proan- by ‘vascular pruning’.The persistence of these neovessels at the ‘conclusion’ of angiogenesis is related to the *Correspondence: DA Cheresh; E-mail: [email protected] duration of stimulus as well as other cues that govern Apoptotic cues from the ECM DG Stupack and DA Cheresh 9023 the maturation (or differentiation) of the vessels (Hirschi (Hox) master genes that are essential in activating and D’Amore, 1997; Abramovitch et al., 1998). secondary endothelial gene transcription events.Stimu- Observations in these physiological systems, where lation with angiogenic growth factors results in the transient angiogenic events occur in response to brief induction of HoxD3, which subsequently upregulates stimulation with cytokines, offer insight into the design the expression of a panel of genes promoting cell of angiogenesis-based therapies.For example, they may invasion, including matrix metalloproteinases (MMPs), explain why collateral vessel growth in the heart, liver or integrin avb3, and UPA (Boudreau et al., 1997) tumor models initiated by VEGF require the sustained (Figure 1).The formation of ‘immature vessels’ from presence of VEGF.Although transient exposure to invasive vascular sprouts requires a second Hox gene, VEGF will rapidly induce the formation of an immature HoxB3, which promotes lumen formation within the vessel network, these neovessels will regress upon sprouts via transcription of differentiation factors such growth factor withdrawal.By contrast, sustained VEGF as ephrin A1 (Myers et al., 2000). The proliferative ECs stimulation can promote vessel maturation, allowing that form the sprouts and neovessels are sensitive to vessels to persist if VEGF is eventually withdrawn (Dor apoptotic cues, such as disruption of the cell–ECM et al., 2002). These results suggest that, during contact or elimination of growth factors. physiological angiogenesis, the duration of the stimulus In contrast with EC, maturation and the restoration can act to influence the permanency of the response. of quiescence are associated with re-expression of Hox This may occur directly, if the continued presence of D10, a master gene present in pre-existing quiescent growth factors is a prerequisite for EC maturation. vessels but absent from the ‘activated’ angiogenic However, it is perhaps more likely that these growth endothelium.HoxD10 expression leads to quiescence factors act to promote survival while secondary through several mechanisms, including downregulation maturation factors, such as angiopoietins and/or peri- of invasive gene expression, competition with HoxD3 cytes, accumulate. for ‘common’ transcription cofactors (leading to

Hox genes and the angiogenic program

Why do angiogenic cells suddenly become susceptible to apoptotic stimuli, while nearby quiescent cells are not? The answer is most likely linked to changes in the gene expression proﬁles of the cells involved.Angiogenesis initiated by growth factors (or by proangiogenic ECM components) results in the activation of homeobox

Figure 2 Interacting factors influence EC survival during angiogenesis.The capacity of ECs to survive during tissue invasion is regulated by factors acting locally during angiogenesis.Remodeling of the local ECM required for invasion involves the production of soluble protein such as thrombospondins, as well as proteolytic cleavage of matrix proteins.The protein fragments generated Figure 1 Responsiveness to apoptotic cues among angiogenic during this process can promote EC apoptosis.Opposing this, new ECs.The stimulation of Ecs with angiogenic growth factors results provisional ECM is produced locally, or deposited from plasma. in alterations in EC biology, including increased proliferation, The deposited ECM bound by integrins acts as a survival factor, altered integrin expression, and the transcription of gene products both by limiting the impact of proapoptotic agents (such as death that act to remodel the local ECM.Invasive ECs (upper left) as well ligands) as well as potentiating EC responses to growth factors. as immature neovessels (upper right) are sensitive to stimuli which Growth factors themselves have somewhat differing effects on EC induce apoptosis (brown fill).In contrast, pre-existing blood vessels resistance to apoptosis, and cooperate with ECM-bound integrins (lower left) as well as neovessels which complete maturation at the to promote cell survival.In addition, cell–cell contacts appear to conclusion of angiogenesis (lower right) become quiescent (yellow act similarly to ECM ligation, promoting resistance to apoptotic fill) and display increased resistance to proapoptotic stimuli stimuli

Oncogene Apoptotic cues from the ECM DG Stupack and DA Cheresh 9024 downregulation of avb3), and transcription of gene stimulation with proangiogenic growth factors, suggest- products that block invasive cell behavior, such as the ing that they play an integral role in the angiogenic MMP inhibitors (Myers et al., 2002). These quiesced program (Brooks et al., 1994). Integrin ligation is ECs appear to be most likely to survive vascular pruning required for cellular responses to most angiogenic at the conclusion of angiogenesis. growth factors (Eliceiri, 2001), suggesting that the local ECM composition/remodeling will orchestrate EC responses, including apoptosis (Jones et al., 1993). Multiple factors influence apoptosis in angiogenic ECs Perhaps, the most dramatic upregulation of any integrin during angiogenesis occurs with the integrin It is not yet clear whether EC apoptosis during the avb3, which is not expressed on quiescent blood vessels conclusion of angiogenesis results from the induction of (Brooks et al., 1994). Integrin avb3 is not commonly death by a proapoptotic factor, or by the absence of a expressed in other tissues, and, with the exception of protective (survival) factor.The situation becomes more some hematopoietic (or hematopoietic-derived) cells, is complicated when one considers that the action of generally absent in adult organisms.Integrin avb3has proapoptotic factors is balanced by the presence of been linked to cell migration and invasion (Felding- survival factors (Figure 2).Therefore, it is perhaps not Habermann and Cheresh, 1993).However, integrin avb3 surprising to learn that EC death or survival is also appears to play a critical role in mediating cell influenced by diverse factors, including angiogenic survival.In particular, integrin avb3 ligation has been cytokines, local remodeling of the ECM, inflammation, linked to signaling via the NF-kB survival pathway and exposure to local stresses such as sustained hypoxia (Malyankar et al., 2000), and agents that block avb3 or chemotherapeutic agents. binding to the ECM will also promote apoptosis among The fact that so many factors balance the decision angiognic ECs in vitro and in vivo (Brooks et al., 1994; between life and death in angiogenic ECs has given rise Brassard et al., 1999). These results suggest that avb3 to the theory that interventions which impact any single may be a critical regulator of EC survival. factor strongly enough will be sufficient to generally ‘tip Interestingly, a population of Iraqi Jews lacking the balance’ towards EC apoptosis (Folkman, 1995). functional avb3 demonstrate a platelet defect but no However, despite dramatic results in a variety of vascular phenotype (Coller et al., 1991), while mice preclinical models, antiangiogenic agents have so far lacking integrin avb3 expression have apparently normal met with limited success in the clinic.In part, this may angiogenesis (Hodivala-Dilke et al., 1999). These results be due to the fact that human tumors arise sponta- suggest that either avb3 is not required as a survival neously, and thus are ‘selected’ for their ability to factor during angiogenesis, or, in its absence, a establish equilibrium between ECM remodeling, growth compensating or redundant mechanism is initiated. factor production, and survival of the local vasculature. Moreover, during chronic pathologies such as tumor However, it is also likely that the multiple proangiogenic growth, increased angiogenesis is observed in mice growth/survival factors produced by human tumors lacking expression of avb3 (Reynolds et al., 2002), may provide a level of redundancy.Thus, therapies that suggesting that avb3 expression may not promote target a single growth factor pathway or receptor may angiogenesis, but rather may act in the negative demonstrate initial effect, but may become less effective regulation of angiogenesis, especially during reiterative as a tumor ‘switches’ (i.e. is selected for the capacity to (i.e., chronic) angiogenic processes. produce different) arrays of angiogenic cytokines produced.This selection process may be augmented, at least in part, by deposition of survival factors such as Integrins as dependence receptors ECM proteins, which can sustain cellular viability in response to a wide array of antiangiogenic insults Recent findings suggest that integrins can act as (Meredith et al., 1993; Isik et al., 1998; Malyankar dependence receptors (Stupack et al., 2001). For et al., 2000; Aoudjit and Vuori, 2001). Thus, in the example, integrin avb3 (among others) can promote cell temporary absence of appropriate growth factors, the death when unligated, or ‘ligated’ by soluble (in contrast ECM may act as an interim survival factor, allowing to ECM-immobilized) antagonists, both in vitro and in ECs to persist until the next wave of angiogenic factors vivo (Brooks et al., 1994; Brassard et al., 1999; Storgard is produced.The presence of an inappropriate ECM et al., 1999). Integrin avb3 binds the widest array of does not promote EC survival under these stresses, and, ligands of any integrin (Cheresh, 1991), and it is perhaps in fact, may even promote apoptosis (see below). not surprising that a broad array of endogenous, soluble ‘ligands’ for avb3 are produced during angiogenesis (Figure 2).While some of these ligands are soluble ECM as a principle regulator of apoptosis native proteins such as thrombospondins (Chen et al., 2000), others are protein fragments generated from pre- Receptors for the ECM, in particular the integrins, act existing ECM components such as collagens (endosta- not only to provide anchorage for ECs, but also provide tin, tumstatin) (O’Reilly et al., 1997; Maeshima et al., information about the ECs’ local microenvironment 2001) or the noncatalytic domains of the proteases that facilitates the cell’s decision to proliferate, migrate themselves (angiostatin, PEX) (O’Reilly et al., 1994; or die.Integrin expression on ECs changes upon Brooks et al., 1998). However, the induction of

Oncogene Apoptotic cues from the ECM DG Stupack and DA Cheresh 9025 apoptosis by soluble ligands (antagonists) is not a unique feature of integrin avb3.Antagonists of other integrins, such a5b1 and a2b1, can also efficiently promote EC apoptosis (Senger et al., 1997; Kim et al., 2000).The relative impact of antagonism of these integrins on EC survival will relate to the level of expression of the integrin antagonized as well as the soluble ligands which impact that integrin.As noted, integrin avb3 is the target of a wide array of endogenous antagonists.This, together with the fact that integrin avb3 is selectively expressed only on angiogenic ECs, explains to a large degree the central role in angiogenesis attributed to integrin avb3 function. In this respect, it is of significant interest that soluble integrin ligands induce apoptosis, while the same ligand, if immobilized within the ECM, promotes survival and even acts to preserve the viability of ECs induced to undergo apoptosis (Petitclerc et al., 2000; Rehn et al., 2001).In this respect, integrins differ from most other classes of receptor, as they require cytosolic engagement of mechanical elements (i.e. actin assembly and con- tractility) for the so-called ‘productive’ signaling (In- gber, 1992).Mechanical stimulation alone is sufficient to Figure 3 Influence of integrin ligation on caspase-mediated death maintain integrin signaling (Ingber, 2002), and integrin- pathways in ECs.ECs can undergo apoptosis via either the extrinsic death pathway, initiated by surface-receptor-mediated mediated signaling is maximized in cells undertaking activation of caspase 8 (such as that by death receptors or mechanical activities, such as spreading or migrating on unligated/antagonized integrins – left side), or via the intrinsic ECM (Schwartz and Ingber, 1994).Conversely, disrup- pathway mediated by mitochondrial release of elements that tion of the mechanical linkage can likewise compromise activate caspase 9 in response to cell stress (right side).Caspase 8 signaling from integrins (Meredith et al., 1993), and activation can also trigger mitochondrial release of effectors, leading to caspase 9 activation and apoptosis via the ‘type II’ ablation of integrin-mediated cell attachment triggers an extrinsic pathway (not shown).All pathways result in the down- array of proapoptotic mechanisms (reviewed in Frisch stream activation of executioner caspases (caspase 3 is shown) and Screaton, 2001).However, questions remain about leading to cell death, and each is influenced by ECM ligation.The the exact roles played by antagonized (or unligated) intrinsic pathway is suppressed by integrin-mediated adhesion that leads to alterations in the abundance and activity of regulatory integrins that promote apoptosis. proteins such as the bcl-2 ‘gatekeeper’ proteins, c-FLIP (an inhibitor of caspase 8 activation), and several of the IAP proteins that block apoptosis by direct binding of caspase 3 and/or caspase 9 Apoptotic pathways used by ECs in vivo

It is not yet clear which cell death pathways may be used 2002).Importantly, antagonists of integrins avb3and by ECs (and under what circumstances), in vivo.Two a5b1 appear to function principally by triggering principle caspase-dependent pathways exist in ECs; one apoptosis (and not by blocking cell migration), since triggered in response to death receptor ligation, termed pharmacological agents that block apoptosis rescue the extrinsic apoptosis pathway, and a second triggered neovascularization in the presence of integrin antago- in response to mitochondrial changes that occur during nists.Considering these results, it becomes clear that stress, termed the intrinsic pathway.Unlike tumor cells, mice lacking avb3 behave as one might expect, given which may preferentially inactivate one or more that the ECs in these mice are unresponsive to avb3 apoptotic mechanisms, both pathways are available to antagonists (and therefore avb3-transmitted apoptotic mediate apoptosis in angiogenic ECs (Figure 3). cues).Moreover, observations in these mice suggest that The intrinsic pathway of apoptosis, induced by the action of endogenous integrin antagonists may play growth factor withdrawal or stresses, is blocked by a particularly important role during pathological forms integrin ligation, possibly by NkB activation (Scatena of angiogenesis (Reynolds et al., 2002). et al., 1998) and suppression of p53 activation (Strom- In contrast with the results implicating caspase 8, blad et al., 1996). Conversely, when EC apoptosis is caspase 9 does not appear to be essential for vascular initiated by soluble integrin antagonists (or ECM pruning, at least during developmental angiogenesis, as fragments), inhibitors of the extrinsic (or caspase 8/10- no general vascular phenotype is reported in the caspase mediated) apoptotic pathway block death, while in- 9 KO mice (Kuida et al., 1998). Nevertheless, suppres- hibitors of the intrinsic apoptotic pathway, or caspase 9- sion of the caspase 9 death pathway could still be mediated pathway, are ineffective.In this case, integrin important in pathological forms of angiogenesis, and antagonism promotes the activation of caspase 8 (and correlations have been made between EC survival and apoptosis) among ECM-attached ECs in a PKA- expression/activity of p53 (Stromblad et al., 1996), dependent manner (Stupack et al., 2001; Kim et al., mTor (Maeshima et al., 2000), bcl-2 family proteins

Oncogene Apoptotic cues from the ECM DG Stupack and DA Cheresh 9026 (Matter and Ruoslahti, 2001), and cell cycle inhibitors including Ras, Raf-1, Mek1, and the MAP kinases Erk1 (Plath et al., 2000), all of which function to block and Erk2.Signaling to Erk1/Erk2 is critical for caspase 9 activation.Similarly, antagonists of integrin angiogenesis, and inhibitors of MEK-1 block neovascu- avb3 suppress retinal neovascularization in p53wt larization (Eliceiri et al., 1998). Erk activation facilitates animals, yet have little antiangiogenic activity in proliferation and cellular migration; however, it also p53À/À animals (Stromblad et al., 2002). As p53 is protects cells from many forms of apoptosis (Cho and generally assumed to be involved selectively in the Klemke, 2000; Howe et al., 2002). regulation of the intrinsic apoptosis cascade, these data Integrins can also signal to the phosphoinositide 30 implicate caspase 9, rather than caspase 8, in EC kinase (PI3K) pathway, leading to activation of Akt and apoptosis in the retina. Rho family GTPases (Giancotti and Ruoslahti, 1999). The role of p53 in EC survival remains somewhat In particular, activation of Akt (by mutation of loss of confusing.For example, p53 activity opposes the the regulatory phosphatase PTEN) allows some trans- induction of apoptosis by rapamycin via mTOR (Huang formed cell types to remain viable in an anchorage- et al., 2001), and mTOR has been suggested to be a independent manner (Frisch and Screaton, 2001), likely downstream target of integrin antagonists in ECs in due to its capacity to block both pathways of apoptosis, vitro.(Maeshima et al., 2000) Thus, if mTOR is a bona described above.Integrin ligation also leads to the fide target of integrin antagonists, one might expect that activation of FAK (Schlaepfer et al., 1999). FAK deletion of p53 should increase sensitivity to integrin activation occurs upstream of Erk and PI3K, and is antagonists.Conversely, angiogenesis in p53-deficient likely to account for at least a portion of the ability of mice is resistant to the effects of integrin antagonists integrins to suppress p53-mediated apoptosis.FAK- (Stromblad et al., 2002). Similarly, cells expressing deficient mouse embryos die early (Ilic et al., 1995), and activated protein kinase Akt, which also opposes p53 cells derived from FAKÀ/À mice can only be cultured in function (Mayo and Donner, 2002), demonstrate a p53-deficient background, despite the fact that these increased sensitivity to mTOR inhibitors (Neshat et al., cells still activate Erk and PI3K pathways (Schlaepfer 2001; Yu et al., 2001), while ‘active’ Akt rescues EC et al., 1999). Interestingly, dominant-negative forms of survival in the presence of integrin antagonists (Mae- FAK, such as FRNK, can also either induce apoptosis shima et al., 2002). However, active Akt blocks the or predispose cells to apoptosis initiated by other means. apoptosis initiated by caspase 8 (Suhara et al., 2001; Li Interestingly, many of the pathways activated by et al., 2002) or caspase 9 (Frisch and Screaton, 2001) integrins (Src, PI3K, Erk) are also activated by pathways, and these studies with Akt may therefore be proangiogenic cytokines such as bFGF and VEGF in difficult to interpret.An additional ‘complication’ to the ECs (Eliceiri, 2001).It has become apparent that ligated interpretation of the role of p53 in EC survival during integrins may establish signaling partnerships with angiogenesis is the finding that p53 may bind within a specific growth factor receptors.Thus, integrins and 600 kDa þ macrocomplex that includes caspase 8 (Ding growth factor receptors likely play complementary roles et al., 2000). Since p53 has been reported to have in initiating the angiogenic program and in cell survival. proapoptotic activities independent of its transcriptional For example, although antagonism of integrin avb3, or activities (which activate the intrinsic apoptosis path- a5b1, can induce apoptosis during angiogenesis induced way), it will be interesting to determine whether p53 also by bFGF, they have dramatically reduced the effect on influences the extrinsic apoptosis pathway during angiogenesis induced by VEGF (Friedlander et al., angiogenesis. 1995).Interestingly, the protection afforded by VEGF signaling is itself dependent upon ligation of another integrin, in this case avb5.Blockade of integrin avb5, an Integration of death and survival signaling events during integrin which does not appear to be proapoptotic angiogenesis (Stupack et al., 2001), nevertheless disrupts VEGF- mediated signaling to Src and Erk (Eliceiri et al., 2002). The observation that the genetic deletion of any single The capacity of VEGF to protect against integrin EC integrin does not prevent angiogenesis (with the avb3 antagonism, which bFGF lacks, is interesting given exception of integrins a5 and b1, which are required for that these growth factors can activate common down- viability of the embryo) (Bouvard et al., 2001) fits well stream signaling cascades.In fact, both activate Src, with our knowledge of how integrins signal (Giancotti Ras, and the PI3K and Erk cascades (Ferrara, 2002). and Ruoslahti, 1999).ECs express eight or nine different Among these pathways, the activation of Erk1/Erk2 is integrin heterodimers, which engage a broad range of critical for angiogenesis, since inhibition of MEK ECM components. Loss of a single integrin (e.g., avb3) (Eliceiri et al., 1998) or Raf (Hood et al., 2002) blocks would not be expected to compromise endothelial cell agreement angiogenesis induced by either growth factor. adhesion, per se.Nor is it necessary to suggest Therefore, it is likely that protection from antagonists ‘compensation,’ or positive signaling by other integrins will be attributed to upstream signaling events parallel in the ‘absence’ of b3, since many integrins are capable to, or required for, the activation of the Erk pathway, of independently activating common signaling pathways which are unique to VEGF signaling. (Giancotti and Ruoslahti, 1999).Most integrin hetero- For example, although Src is activated by both VEGF dimers are capable of activating focal adhesion kinase and bFGF, only VEGF-induced angiogenesis is depen- (FAK) or Shc (or both), as well as downstream elements dent upon Src kinase activity (Eliceiri et al., 2002). In

Oncogene Apoptotic cues from the ECM DG Stupack and DA Cheresh 9027 this case, it appears that Src is critical for formation of be critical for targeting and activating these cells, at the integrin avb5/FAK complex necessary for signaling which point it is likely that they become dependent upon downstream of VEGFR2 (Eliceiri et al., 2002). Mice the underlying ECM as a survival factor (just as other lacking avb5 have impaired responses to VEGF (Eliceiri angiogenic ECs are) (Salven et al., 2003). It is also et al., 2002), although angiogenesis appears to proceed possible that sensitivity to apoptosis is gained as part of normally.It is not yet clear whether VEGF provides the ‘precursor to EC’ differentiation process, since a protection against apoptosis induced by avb3 integrin requirement for ECM interaction during this process antagonists in these mice. has been proposed to occur during this process in vitro and in vivo (Moldovan, 2003). ECM inﬂuences cell responses to other apoptotic triggers

Current data underscore the importance of the ECM as Conclusions and therapeutic outlook a critical survival factor for ECs in vivo.The data are less clear on the role of death receptors and their ligands. Angiogenesis involves the programmed dedifferentiation While it appears that Fas and FasL regulate angiogen- of pre-existing ECs (and possibly the recruitment of esis in the retina of adult animals (Kaplan et al., 1999), it CEPs) allowing proliferation and tissue invasion.How- is clear that soluble FasL can also induce an inflamma- ever, as a part of this process, ECs become susceptible to tory angiogenesis in Matrigel plugs in vivo (Biancone apoptosis, requiring growth factor stimulation and et al., 1997). Similarly, TNFa induces angiogenesis in integrin-mediated engagement of immobilized ECM vivo (Friedlander et al., 1995), and may contribute to components to maintain viability.Different integrins inflammatory angiogenesis, such as that seen in rheu- appear to be tied to signaling through distinct growth matoid arthritis (Walsh and Pearson, 2001).Activated factor receptors, suggesting that interventions which ECs express greater levels of death receptors than impact either the adhesion receptor or the growth factor quiescent ECs (Volpert et al., 2002), but are generally receptor can impact angiogenesis.This appears to be a resistant to apoptosis mediated by death ligands in vitro. natural process, as a variety of different proteins and This may be due to the production of survival- protein fragments produced during angiogenesis milieu promoting cytokines, such as VEGF, the presence of act as endogenous soluble ligands (antagonists) of an immobilized ECM, or both.Accordingly, agents integrins.Since apoptosis occurs in these vessels during which modulate ECM-binding properties of ECs also physiological forms of angiogenesis, it is tempting to render them susceptible to death ligand signaling speculate that these soluble integrin antagonists may (Ruegg et al., 1998; Aoudjit and Vuori, 2001; Volpert represent a natural homeostatic mechanism that governs et al., 2002). The actions of both ECM and death the behavior of invasive ECs, providing a ‘self-limiting’ ligands may be somewhat different in vivo.In contrast mechanism to prevent excessive or aberrant neovascu- to soluble integrin ligands, soluble death ligands, larization (perhaps even acting as an innate mechanism produced by proteolysis, are far less efficient at inducing to limit pathological forms of angiogenesis).In fact, apoptosis than immobilized death ligands (Mitsiades tumor vessels exhibit spontaneous regression, although et al., 2001). The actions of proteases present during this effect is generally offset by concomitant induction angiogenesis may therefore be protective against of new angiogenic processes. death ligand signaling.However, the generation of The initial promise of antiangiogenic therapies may soluble ECM fragments may compensate by compro- ultimately become a reality.Since ECs are genetically mising integrin-mediated prosurvival functions, or stable relative to tumor cells, they are far less likely to even inducing apoptosis directly through integrins acquire resistance to endogenous or chemotherapeutic themselves. agents.In contrast to tumor cells, which often inactivate one (or both) major apoptotic pathway, ECs have both intrinsic and extrinsic apoptotic pathways intact.More- EC precursors over, the architecture of the vasculature actually promotes collateral effect, as EC death and vessel Recent studies suggest that a significant fraction of ECs, collapse in turn promote the apoptosis of ‘downstream’ both in physiological and pathological forms of ECs and dependent tumor and stromal cells angiogenesis, are derived from circulating endothelial (Folkman and Hanahan, 1991).However, we have only precursor cells (CEPs).While it is not yet clear how the now begun to understand the molecular interactions survival of CEPs is regulated during angiogenesis, it is that regulate EC apoptosis, and an increased under- assumed that these cells do not require adhesion during standing will be necessary before it can be manipulated circulation for survival.Integrin-mediated events may with certainty.

References

Abramovitch R, Neeman M, Reich R, Stein I, Keshet E, Biancone L, Martino AD, Orlandi V, Conaldi PG, Toniolo A Abraham J, Solomon A and Marikovsky M.(1998). FEBS and Camussi G.(1997). J. Exp. Med., 186, 147–152. Lett., 425, 441–447. Boudreau N, Andrews C, Srebrow A, Ravanpay A and Aoudjit F and Vuori K.(2001). J. Cell Biol., 152, 633–643. Cheresh DA.(1997). J. Cell Biol., 139, 257–264.

Oncogene Apoptotic cues from the ECM DG Stupack and DA Cheresh 9028 Bouvard D, Brakebusch C, Gustafsson E, Aszodi A, Kaplan HJ, Leibole MA, Tezel T and Ferguson TA.(1999). Bengtsson T, Berna A and Fassler R.(2001). Circ. Res., Nat. Med., 5, 292–297. 89, 211–223. Kerbel RS, Viloria-Petit A, Klement G and Rak J.(2000). Eur. Brassard DL, Maxwell E, Malkowski M, Nagabhushan TL, J. Cancer, 36, 1248–1257. Kumar CC and Armstrong L.(1999). Exp. Cell. Res., 251, Keshet E and Ben-Sasson SA.(1999). J. Clin. Invest., 104, 33–45. 1497–1501. Brooks PC, Montgomery AM, Rosenfeld M, Reisfeld RA, Hu Kim S, Bakre M, Yin H and Varner JA.(2002). J. Clin. T, Klier G and Cheresh DA.(1994). Cell, 79, 1157–1164. Investig., 110, 933–941. Brooks PC, Silletti S, von Schalscha TL, Friedlander M and Kim S, Bell K, Mousa SA and Varner JA.(2000). Am. J. Cheresh DA.(1998). Cell, 92, 391–400. Pathol., 156, 1345–1362. Chen H, Herndon ME and Lawler J.(2000). Matrix Biol., 19, Klement G, Huang P, Mayer B, Green SK, Man S, Bohlen P, 597–614. Hicklin D and Kerbel RS.(2002). Clin. Cancer Res., 8, Cheresh DA.(1991). Cancer Metast. Rev., 10, 3–10. 221–232. Cho SY and Klemke RL.(2000). J. Cell Biol., 149, 223–236. Kuida K, Haydar TF, Kuan CY, Gu Y, Taya C, Karasuyama Coller BS, Cheresh DA, Asch E and Seligsohn U.(1991). H, Su MS, Rakic P and Flavell RA.(1998). Cell, 94, Blood, 77, 75–83. 325–337. Craft PS and Harris AL.(1994). Ann. Oncol., 5, 305–311. Li L, Okura M and Imamoto A.(2002). Mol. Cell. Biol., 22, Dickson SE, Bicknell R and Fraser HM.(2001). J. Endocrinol., 1203–1217. 168, 409–416. Maeshima Y, Colorado PC and Kalluri R.(2000). J. Biol. Ding HF, Lin YL, McGill G, Juo P, Zhu H, Blenis J, Yuan J Chem., 275, 23745–23750. and Fisher DE.(2000). J. Biol. Chem., 275, 38905–38911. Maeshima Y, Manfredi M, Reimer C, Holthaus KA, Hopfer Dor Y, Djonov V, Abramovitch R, Itin A, Fishman GI, H, Chandamuri BR, Kharbanda S and Kalluri R.(2001). Carmeliet P, Goelman G and Keshet E.(2002). Cancer Cell, J. Biol. Chem., 276, 15240–15248. 21, 1939–1947. Maeshima Y, Sudhakar A, Lively JC, Ueki K, Kharbanda S, Eliceiri BP.(2001). Circ. Res., 89, 1104–1110. Kahn CR, Sonenberg N, Hynes RO and Kalluri R.(2002). Eliceiri BP, Klemke R, Stromblad S and Cheresh DA.(1998). Science, 295, 140–143. J. Cell Biol., 140, 1255–1263. Malyankar UM, Scatena M, Suchland KL, Yun TJ, Eliceiri BP, Puente XS, Hood JD, Stupack DG, Schlaepfer Clark EA and Giachelli CM.(2000). J. Biol. Chem., 275, DD, Huang XZ, Sheppard D and Cheresh DA.(2002). 20959–20962. J. Cell Biol., 157, 149–160. Matter ML and Ruoslahti E.(2001). J. Biol. Chem., 276, Felding-Habermann B and Cheresh DA.(1993). Curr. Opin. 27757–27763. Cell Biol., 5, 864–868. Mayo LD and Donner DB.(2002). Trends Biochem. Sci., 27, Ferrara N.(2002). Nat. Rev. Cancer, 2, 795–803. 462–467. Folkman J.(1995). Nat. Med., 1, 27–31. Meredith Jr JE, Fazeli B and Schwartz MA.(1993). Mol. Biol. Folkman J and Hanahan D.(1991). Princess Takamatsu Cell, 4, 953–961. Symp., 22, 339–347. Mitchell CA, Risau W and Drexler HC.(1998). Dev. Dyn., Friedlander M, Brooks PC, Shaffer RW, Kincaid CM, Varner 213, 322–333. JA and Cheresh DA.(1995). Science, 270, 1500–1502. Mitsiades N, Poulaki V, Mitsiades CS and Anderson KC. Frisch SM and Screaton RA.(2001). Curr. Opin. Cell Biol., 13, (2001). Expert Opin. Investig. Drugs, 10, 1075–1084. 555–562. Moldovan NI.(2003). Adv. Exp. Med. Biol., 522, 99–113. Gasparini G, Brooks PC, Biganzoli E, Vermeulen PB, Bonoldi Myers C, Charboneau A and Boudreau N.(2000). J. Cell Biol., E, Dirix LY, Ranieri G, Miceli R and Cheresh DA.(1998). 148, 343–351. Clin. Cancer Res., 4, 2625–2634. Myers C, Charboneau A, Cheung I, Hanks D and Boudreau Giancotti FG and Ruoslahti E.(1999). Science, 285, N.(2002). Am. J. Pathol., 161, 2099–2109. 1028–1032. Neshat MS, Mellinghoff IK, Tran C, Stiles B, Thomas G, Hirschi KK and D’Amore PA.(1997). Exs, 79, 419–428. Petersen R, Frost P, Gibbons JJ, Wu H and Sawyers CL. Hodivala-Dilke KM, McHugh KP, Tsakiris DA, Rayburn H, (2001). Proc. Natl. Acad. Sci. USA, 98, 10314–10319. Crowley D, Ullman-Cullere M, Ross FP, Coller BS, O’Reilly MS, Boehm T, Shing Y, Fukai N, Vasios G, Lane Teitelbaum S and Hynes RO.(1999). J. Clin. Invest., 103, WS, Flynn E, Birkhead JR, Olsen BR and Folkman J. 229–238. (1997). Cell, 88, 277–285. Hood JD, Bednarski M, Frausto R, Guccione S, Reisfeld RA, O’Reilly MS, Holmgren L, Shing Y, Chen C, Rosenthal RA, Xiang R and Cheresh DA.(2002). Science, 296, Cao Y, Moses M, Lane WS, Sage EH and Folkman J. 2404–2407. (1994). Cold Spring Harb. Symp. Quant. Biol., 59, Howe AK, Aplin AE and Juliano RL.(2002). Curr. Opin. 471–482. Genet. Dev., 12, 30–35. Petitclerc E, Boutaud A, Prestayko A, Xu J, Sado Y, Huang S, Liu LN, Hosoi H, Dilling MB, Shikata T and Ninomiya Y, Sarras Jr MP, Hudson BG and Brooks PC. Houghton PJ.(2001). Cancer Res., 61, 3373–3381. (2000). J. Biol. Chem., 275, 8051–8061. Ilic D, Furuta Y, Kanazawa S, Takeda N, Sobue K, Nakatsuji Plath T, Detjen K, Welzel M, von Marschall Z, Murphy D, N, Nomura S, Fujimoto J, Okada M and Yamamoto T. Schirner M, Wiedenmann B and Rosewicz S.(2000). J. Cell (1995). Nature, 377, 539–544. Biol., 150, 1467–1478. Ingber DE.(1992). Semin. Cancer Biol., 3, 57–63. Rehn M, Veikkola T, Kukk-Valdre E, Nakamura H, Ilmonen Ingber DE.(2002). Circ. Res., 91, 877–887. M, Lombardo C, Pihlajaniemi T, Alitalo K and Vuori K. Isik FF, Gibran NS, Jang YC, Sandell L and Schwartz SM. (2001). Proc. Natl. Acad. Sci. USA, 98, 1024–1029. (1998). J. Cell. Physiol., 175, 149–155. Reynolds LE, Wyder L, Lively JC, Taverna D, Robinson SD, Jones PL, Schmidhauser C and Bissell MJ.(1993). Crit. Rev. Huang X, Sheppard D, Hynes RO and Hodivala-Dilke KM. Eukaryot. Gene Expr., 3, 137–154. (2002). Nat. Med., 8, 27–34.

Oncogene Apoptotic cues from the ECM DG Stupack and DA Cheresh 9029 Ruegg C, Yilmaz A, Bieler G, Bamat J, Chaubert P and Stromblad S, Fotedar A, Brickner H, Theesfeld C, Aguilar de Lejeune FJ.(1998). Nat. Med., 4, 408–414. Diaz E, Friedlander M and Cheresh DA.(2002). J. Biol. Salven P, Mustjoki S, Alitalo R, Alitalo K and Raﬁi S.(2003). Chem., 277, 13371–13374. Blood, 101, 168–172. Stupack DG, Puente XS, Boutsaboualoy S, Storgard CM and Scatena M, Almeida M, Chaisson ML, Fausto N, Nicosia RF Cheresh DA.(2001). J. Cell. Biol., 155, 459–470. and Giachelli CM.(1998). J. Cell Biol., 141, 1083–1093. Suhara T, Mano T, Oliveira BE and Walsh K.(2001). Circ. Schlaepfer DD, Hauck CR and Sieg DJ.(1999). Prog. Biophys. Res., 89, 13–19. Mol. Biol., 71, 435–478. Volpert OV, Zaichuk T, Zhou W, Reiher F, Ferguson TA, Schwartz MA and Ingber DE.(1994). Mol. Biol. Cell, 5, Stuart PM, Amin M and Bouck NP.(2002). Nat. Med., 8, 389–393. 349–357. Senger DR, Claffey KP, Benes JE, Perruzzi CA, Sergiou AP Walsh DA and Pearson CI.(2001). Arthritis Res., 3, 147–153. and Detmar M.(1997). Proc. Natl. Acad. Sci. USA, 94, Witte L, Hicklin DJ, Zhu Z, Pytowski B, Kotanides H, 13612–13617. Rockwell P and Bohlen P.(1998). Cancer Metast. Rev., 17, Storgard CM, Stupack DG, Jonczyk A, Goodman SL, Fox RI 155–161. and Cheresh DA.(1999). J. Clin. Investig., 103, 47–54. Yu K, Toral-Barza L, Discafani C, Zhang WG, Skotnicki J, Stromblad S, Becker JC, Yebra M, Brooks PC and Cheresh Frost P and Gibbons JJ.(2001). Endocr. Relat. Cancer, 8, DA.(1996). J. Clin. Investig., 98, 426–433. 249–258.

Oncogene