The Biochemical, Biological, and Pathological Kaleidoscope of Cell Surface Substrates Processed By Matrix Metalloproteinases

By Cauwe, Benedicte Van den Steen, Philippe E; Opdenakker, Ghislain

ABSTRACT Matrix metalloproteinases (MMPs) constitute a family of more than 20 endopeptidases. Identification of specific matrix and non-matrix components as MMP substrates showed that, aside from their initial role as extracellular matrix modifiers, MMPs play significant roles in highly complex processes such as the regulation of cell behavior, cell-cell communication, and tumor progression. Thanks to the comprehensive examination of the expanded MMP action radius, the initial view of proteases acting in the soluble phase has evolved into a kaleidoscope of proteolytic reactions connected to the cell surface. Important classes of cell surface molecules include adhesion molecules, mediators of apoptosis, receptors, chemokines, cytokines, growth factors, proteases, intercellular junction proteins, and structural molecules. Proteolysis of cell surface proteins by MMPs may have extremely diverse biological implications, ranging from maturation and activation, to inactivation or degradation of substrates. In this way, modification of membrane-associated proteins by MMPs is crucial for communication between cells and the extracellular milieu, and determines cell fate and the integrity of tissues. Hence, insights into the processing of cell surface proteins by MMPs and the concomitant effects on physiological processes as well as on disease onset and evolution, leads the way to innovative therapeutic approaches for cancer, as well as degenerative and inflammatory diseases. KEYWORDS ectodomain shedding, receptor, membrane, cancer, immunity, autoimmunity

ABBREVIATIONS

Abeta, beta-amyloid protein; ACE, angiotensin-converting enzyme; ADAM, a disentegrin and metalloproteinase; ADAMTS, a disentegrin and metalloproteinase with thrombospondin-like motif; ANF-R, atrial natriuretic factor receptor; AP, alkaline phosphatise; ApoE, apolipoprotein E; APP, amyloid precursor protein; AR, amphiregulin; ARF, acute renal failure; BBB, blood-brain barrier; BP, bullous pemphigoid; BP-180, BP antigen-2; BRB, blood-retinal barrier; C1qRp, complement component 1 q subcomponent receptor 1; C3b/4b, complement component 3b/4b; CD, cluster of differentiation of human (glyco)proteins; CD44ICD, intracellular domain of CD44; c-Met, met proto-oncogene TKR; CNS, central nervous system; CTL, cytotoxic T cell; EAE, experimental autoimmune encephalomyelitis; E-cadherin, epithelial cadherin; ECM, extracellular matrix; EDTA, ethylenediaminetetraacetic acid; EGF(R), epidermal growth factor (receptor); EGTA, ethylene glycol-bis(2-aminoethylether)-N,N,N’,N’- tetraacetic acid; EMMPRIN, extracellular matrix metalloproteinase inducer; FasL, Fas ligand; FcyRIII, low affinity Ig gamma Fc receptor III; FcepsilonRII, low affinity Ig epsilon Fc receptor II; FGF, fibroblast growth factor; FGFR-1, FGF receptor-1; gC1qR, receptor of complement component 1q; GHR, growth hormone receptor; GM-CSF-Ralpha, granulocyte-macrophage colony-stimulating factor receptor alpha chain; GnRH, gonadotropin-releasing hormone; GP VI, glycoprotein VI; GPCR, G protein-coupled receptor; GPI, glycosyl phosphatidylinositol; HB-EGF, heparin-binding epidermal growth factor-like growth factor; HER2/4, tyrosine kinase-type cell surface receptor HER2/4; ICAM-1, intercellular adhesion molecule-1; ICE, IL- 1beta-converting enzyme; Ig, immunoglobulin; IL- (R), interleukin- (receptor); IL-1beta, interleukin-1beta; KitL, Kit ligand; KS, keratoconjunctivitis sicca; LDL(-R), low density lipoprotein (receptor); LIGHT, homologous to lymphotoxins, exhibits inducible expression, and competes with herpes simplex virus glycoprotein D for herpes virus entry mediator, a receptor expressed by T lymphocytes; LPS, lipopolysaccharide; LR, 34/67 kDa laminin receptor; LRP, low-density lipoprotein receptor-related protein; L- selectin, leukocyte-selectin; m-, membrane-bound; MBP, myelin basic protein; M-CSF, macrophage-colony stimulating factor; MEVC, microvascular endothelial cell; MICA/B, MHC class I chain-related gene A/B; MMP, matrix metalloproteinase; MMR-1, macrophage mannose receptor-1; MP, metalloproteinase; MT-MMP, membrane-type MMP; MUC1, mucin-1; N-cadherin, neuronal cadherin; N-CAM L1, neural cell adhesion molecule L1; NE, neutrophil elastase; NGF-R, nerve growth factor receptor; NK cell, natural killer cell; NKR BY55, NK cell receptor BY55; PAR1, protease-activated receptor-1; PC, propretein convertase; PDGF, platelet-derived growth factor; PDR, proliferative diabetic retinopathy; PECAM-1, platelet endothelial cell adhesion molecule-1; PLAD, preligand assembly domain; PMA, phorbol 12- myristate 13-acetate; PSGL-1, P-selectin glycoprotein ligand-1; PVR, proliferative vitreoretinopathy; -R, receptor; RANK(L), receptor activator of nuclear factor kappaB (ligand); s-, soluble; SDF-1, stromal cell-derived factor-1; SorLA, sorting protein-related receptor containing LDL-R class A repeats; SP-D, surfactant protein- D; SRCR M130, scavenger receptor cysteine-rich type 1 protein M130; Ssc, systemic sclerosis; TACE, TNF-alpha-converting enzyme; TAPI, TNF-alpha protease inhibitor; TGF-alpha/beta, transforming growth factor-alpha/beta; Tie-1, tyrosine kinase receptor tie-1; TIL, tumor infiltrating lymphocyte; TIMP, tissue inhibitor of metalloproteinases; TKR, tyrosine kinase receptor; TMD, transmembrane domain; TMPS, triple membrane-passing signal mechanism; TNF-alpha, tumor necrosis factor-alpha; t-PA, tissue- type plasminogen activator; TSHR, thyrotropin receptor; tTG, tissue transglutaminase; ULBP2, UL16-binding protein 2; u-PA, urokinase- type plasminogen activator; uPAR, urokinase-type plasminogen activator receptor; VCAM-1, vascular cell adhesion molecule-1; VE- cadherin, vascular endothelial-cadherin; VEGF, vascular endothelial growth factor; VPR V2, V2 vasopressin receptor; X, Xenopus.

INTRODUCTION

Matrix metalloproteinases (MMPs) constitute a family of over 20 different endopeptidases characterized by a conserved Zn^sup 2+^- binding motif HEXXHXXGXXH in the catalytic domain and a number of conserved protein domains. The three histidines in this motif are responsible for binding the catalytic Zn^sup 2+^ ion (Figure 1) (Nagase and Woessner, 1999; Brinckerhoff and Matrisian, 2002). Expression of most MMPs is under transcriptional regulation, e.g., by growth factors, hormones, cytokines and oncogenic transformation. MMPs are neutral endopeptidases produced as secreted or membrane- bound pro-enzymes or zymogens, which become activated by removal of the NH^sub 2^-terminal propeptide. The interaction of a conserved cysteine in the propeptide with the catalytic Zn^sup 2+^ ion seals the catalytic site and results in the latency of the pro-enzyme (Figure 1) (Visse and Nagase, 2003). Removal of the propeptide, for example by proteolysis, alters this coordination with Zn^sup 2+^. Due to the subsequent conformational change, the Zn^sup 2+^ ion becomes available for the binding of a hydrolytic water molecule and of the substrate. Therefore, the MMP activation mechanism was named the ‘cysteine switch mechanism’ (Van Wart and Birkedal-Hansen, 1990). This activation can be mediated by proteases and other MMPs, or chemically by means of organomercurials, urea, some detergents and also by reactive oxygen species. In addition, some MMPs are activated intracellularly by furins (Nagase and Woessner, 1999).

Examples of ‘archetypical’ MMPs are the collagenases, which are active against fibrillar collagens, and the stromelysins, which cleave non-collagen components of the extracellular matrix (ECM) (Folgueras et al., 2004). They contain a COOH-terminal hemopexin domain (Figure 1), which contributes to substrate specificity, and to interactions with endogenous inhibitors and cargo receptors (Piccard et al., 2007). This hemopexin domain is absent in the smallest MMPs, the matrilysins. In addition, the gelatinases, active on denaturated collagens, incorporate three fibronectin type II repeats for the binding of gelatin, and MMP-9 is the only MMP to possess a Ser/Thr/Pro-rich O-glycosylated domain, which forms an attachment site for multiple O-linked sugars (Van den Steen et al., 2006). In addition to the secreted MMPs, six human membrane-bound MMPs (MT-MMPs) exist. These are linked to the cell surface through a COOH-terminal transmembrane domain (MT1-, MT2-, MT3- and MT5-MMP) or a glycosyl phosphatidylinositol (GPI) anchor (MT4- and MT6-MMP) (Visse and Nagase, 2003).

Once switched on, MMP proteolytic activity is under tight control by specific inhibitors, primarily the tissue inhibitors of metalloproteinases (TIMPs) (Brew et al., 2000). In plasma, the general protease inhibitor, alpha^sub 2^-macroglobulin, is the predominant MMP inhibitor (Baker et al., 2002), whereas the TIMPs are considered to be the key inhibitors in tissue. In humans, four different TIMPs (TIMP-1 to -4) have been characterized that form non- covalent 1:1 complexes with MMPs. TIMP-1, TIMP-2, and TIMP-4 are present in the extracellular environment in a soluble form, whereas TIMP-3 is insoluble, sequestered by the ECM (Gomez et al., 1997). Although TIMPs bind tightly to most MMPs, some important differences exist in the inhibition profile (Table 1). For instance, TIMP-1 is known to inhibit most soluble MMPs, but it is a poor inhibitor for MT1-MMP, MT2-MMP, MT3-MMP, MT5-MMP, and MMP-19. Binding affinity to MMP-9 is high for TIMP-1, but low for TIMP-2 and TIMP-3, whereas TIMP-2, TIMP-3, and TIMP-4 bind with high affinity to MMP-2 (Gomez et al., 1997; Nagase et al., 2006). In addition, TIMP-3 inhibits members of the ‘a disentegrin and metalloproteinase (ADAM)-family’, including ADAM-10/Kuzbanian, ADAM-12, ADAM-17/tumor necrosis factor- alpha (TNF-alpha)-converting enzyme (TACE) and ADAM-19, as well as the aggrecan-degrading enzymes ‘a disentegrin and metalloproteinase with thrombospondin-like motif’ (ADAMTS)-4 and ADAMTS-5, and TIMP-1 inhibits ADAM-10 (Baker et al., 2002). As a consequence, the TIMP inhibition profile of a particular cleavage can already shed some light on the identity of the protease in charge. FIGURE 1 Domain structure of the human MMPs. An archetypal MMP contains a signal peptide for secretion, a propeptide, a catalytic domain with a conserved Zn^sup 2+^-binding motif and a COOH-terminal domain. The hemopexin domain is absent in the smallest MMPs, the matrilysins, whereas the gelatinases incorporate three fibronectin type II repeats for the binding of gelatin, and MMP-9 is the only MMP to possess a Ser/Thr/Pro-rich O-glycosylated domain. Some MMPs are attached to the cell surface through a COOH-terminal transmembrane domain or a GPI anchor. The interaction of a conserved cysteine in the propeptide with the catalytic Zn^sup 2+^ ion seals the catalytic site and results in the latency of the pro-enzyme. MMPs are activated according to the ‘cysteine switch mechamism’ in which removal of the propeptide frees the catalytic Zn^sup 2+^ ion, allowing it to bind a hydrolytic water ion and the substrate. CA- MMP, cysteine array-MMP; GPI, glycosyl phosphatidylinositol; Ig, Immunoglobulin; RASI-1, rheumatoid arthritis synovial inflammation- 1. Based on (Van den Steen et al., 2002), (Folgueras et al., 2004) and (Nagase et al., 2006).

TABLE 1 TIMP characteristics and inhibition profiles

Thanks to their structural diversity and broad substrate range, MMPs play a part in multiple physiological and pathological processes (Sternlicht and Werb, 2001). As it is more challenging to study membrane-bound molecules, the initial identification of MMP substrates included mainly soluble proteins. However, due to the fast development and fine-tuning of powerful biochemical techniques, insights into MMP cleavage of membrane-bound substrates grew in parallel with the appreciation of its relative importance in physiology as well as pathology. This review contains a survey of known membrane-bound substrates of every MMP, or more specifically of all proteins with a transmembrane domain that have been shown to be the subject of modification by one or several MMPs. In addition, we will discuss the cleavage of some important cell surface proteins that do not contain a transmembrane domain but are always attached to the cell surface by other means of anchoring. The consequences of proteolysis of cell surface proteins by MMPs will be stressed to generate, from scattered information, a clear view on the importance of this process. Consequently, the substrates are grouped in functional classes in the context of the physiological or pathological roles they play with or without proteolysis. To facilitate comprehensive reading, all substrate molecules are grouped in a master table which includes the modifying MMPs, the context of cleavage, known cleavage sites, the biological effect of proteolysis and the physiopathological implications (Table 2). In many instances, various names, abbreviations and acronyms have been given to these substrates. For clarity, these are provided once at the beginning of each section (see also the list of abbreviations).

1. PROTEOLYTIC MODIFICATION OF CELL SURFACE PROTEINS IN CANCER DEVELOPMENT AND EVOLUTION

Tumorigenesis and cancer progression originate from at least seven fundamental alterations in cellular physiology: 1) production of autocrine growth signals; 2) unsensitivity to growth-inhibitory signals; 3) escape from apoptosis; 4) loss of senescence; 5) sustained angiogenesis; 6) tissue invasion; and 7) metastasis (Hanahan and Weinberg, 2000). MMPs have an impact on the microenvironment of tumors. Moreover, the expression and activity of specific MMPs is elevated in almost every kind of cancer. Some MMPs are synthesized principally by tumor cells (e.g., MMP-7), whereas other MMPs (e.g., MMP-2 and -9) are rather produced by stromal cells in the tumors, like fibroblasts, inflammatory cells, and endothelial cells (Egeblad and Werb, 2002). In earlier studies, the MMPs were claimed to be important in migration, invasion and metastasis mainly by the degradation of basement membranes and structural components in the ECM. However, more recent findings show that proteolysis of a panacea of essential secreted and membrane-bound molecules, like growth factor precursors and receptors, tyrosine kinase receptors (TKRs), cytokines and chemokines, cell adhesion molecules, and other proteases, modifies the tumor microenvironment. These modifications will affect tumor progression at all levels (Nelson et al., 2000; McCawley and Matrisian, 2001). In this first part we will discuss proteolytic modifications of cell surface proteins by MMPs and their concomitant effects on cancer evolution (see Figure 2). Insight into the paradoxical consequences of proteolysis of cell surface proteins by MMPs might be of paramount importance in the design of new cancer therapies based on MMP inhibition.

TABLE 2 Proteolysis of cell surface proteins by MMPs in phsychological and pathological processes

TABLE 2 Proteolysis of cell surface proteins by MMPs in phsychological and pathological processes

TABLE 2 Proteolysis of cell surface proteins by MMPs in phsychological and pathological processes

TABLE 2 Proteolysis of cell surface proteins by MMPs in phsychological and pathological processes

TABLE 2 Proteolysis of cell surface proteins by MMPs in phsychological and pathological processes

TABLE 2 Proteolysis of cell surface proteins by MMPs in phsychological and pathological processes

TABLE 2 Proteolysis of cell surface proteins by MMPs in phsychological and pathological processes

TABLE 2 Proteolysis of cell surface proteins by MMPs in phsychological and pathological processes

1.1 Promotion or Inhibition of Cancer Cell Proliferation

MMPs stimulate tumor cell proliferation by catalysing the release of growth factors from the ECM, by the activation of membrane-bound growth factor receptors or by cleavage of membrane-bound substrates, e.g., integrins, that assist in the induction of cell proliferation. Conversely, MMPs might also negatively regulate cancer-cell growth by releasing pro-apoptotic molecules like Fas ligand (FasL) and TNF- alpha, or by activating transforming growth factor-beta (TGF-beta), which has a tumor suppressing effect in early phases of oncogenesis (Egeblad and Werb, 2002).

7.1.1 Fibroblast Growth Factor Receptor-1 (FGFR-1)

The fibroblast growth factors (FGFs) constitute a family of twenty structurally related polypeptides with a common high affinity to heparin. Their biological functions are broad and range from the induction of cellular proliferation to tissue regeneration, neurite outgrowth, and angiogenesis (Powers et al., 2000). FGFs elicit their biological response by binding to four different cell surface TKRs, which are typically composed of an extracellular part with three immunoglobulin (Ig)-like domains, a single transmembrane domain and a bipartite TKR domain. Regulation of FGF biological activity may be achieved through several mechanisms, including binding to high and low affinity receptors on the cell surface, release of FGF from the ECM by heparanases or other proteases or binding to a carrier protein that can deliver FGFs to their receptors. In addition, FGF activity might be regulated by the release of the entire ectodomain of the FGF receptor-1 (FGFR-1, basic fibroblast growth factor receptor-1 (bFGF-R), fms-like tyrosine kinase 2, c-fgr, CD331) into the circulation and the ECM (Levi et al., 1996; Powers et al., 2000). MMP-2 is able to free the entire FGFR-1 ectodomain from an immobilized FGFR1 ectodomain-alkaline phosphatase (FRAP) fusion protein in vitro by hydrolyzing the Val368-Met369 bond, eight amino acids upstream of the transmembrane domain (Figure 3). By this truncation, the ligand binding site is released as a soluble ectodomain retaining its FGF binding ability (Levi et al., 1996). FGFs may stimulate tumor cell growth in at least one of the following ways: 1) by acting as mitogens for the tumor cells themselves; 2) by promoting angiogenesis for the growing tumor and 3) by inhibiting apoptosis and allowing tumor cells to keep growing beyond normal constraints (Powers et al., 2000). In this regard, release of the FGFR-1 ectodomain by MMP-2 would counter these mechanisms in two ways. On the one hand, the released FGFR-1 ectodomain might bind extracellular FGF and in this manner diminish the biological availability and growth promoting activity of FGF. On the other hand, treatment of FGFR-1 overexpressing cells with MMP-2 clearly reduces binding of the FGF mitogen (Levi et al., 1996). Thus, investigating the release of the FGFR-1 ectodomain by MMP-2 or by related MMPs in vivo may provide new insights in tumor cell growth inhibition.

1.1.2 Heparin-Binding Epidermal Growth Factor-Like Growth Factor (HB-EGF)

The heparin-binding epidermal growth factor (EGF)-like growth factor (HB-EGF, also called diphtheria toxin receptor [DT-R]) is a member of the EGF family of growth factors, which encompasses the receptor family of a number of structurally homologous mitogens like EGF, transforming growth factor-alpha (TGF-alpha) and amphiregulin (AR) (Raab and Klagsbrun, 1997; Iwamoto and Mekada, 2000). HB-EGF binds to the EGF receptor (EGFR/HER1/ErbB1) as well as to HER4/ ErbB4, inducing homo- or heterodimerisation of the monomeric receptors with consequent tyrosine phosphorylation in the cytoplasmic domains, resulting in the activation of several signal tranduction pathways (Zwick et al., 1999; Herbst, 2004). HB-EGF is synthesized as a trans-membrane protein (mHB-EGF) with the ectodomain containing a heparin-binding and an EGF-like domain. mHB- EGF can be cleaved at the plasma membrane to yield soluble HB-EGF (sHB-EGF) (Raab and Klagsbrun, 1997; Iwamoto and Mekada, 2000). mHB- EGF is not only a precursor for sHB-EGF but is a bioactive molecule itself, which inhibits growth of neighboring cells. As a contrast, sHB-EGF is a potent mitogen for a number of cells including smooth muscle cells, epithelial cells and keratinocytes. Therefore, release of HB-EGF causes the conversion of a juxtacrine growth inhibitor into a paracrine/autocrine growth factor, with an opposite impact on cell growth (Iwamoto and Mekada, 2000; Higashiyama, 2004). mHB-EGF proteolysis is probably mediated by metalloproteinases as it is prevented by the broad spectrum metalloproteinase inhibitor batimastat (Prenzel et al, 1999). In cell cultures, MMP-3 cleaves mHB-EGF in the juxtamembrane domain at the Glul51-Asn152 site (Suzuki et al., 1997; Wu et al, 2004), whereas MMP-7, MMP-2 and MMP- 9 and some proteases of the ADAM family [ADAM-9, ADAM-10, ADAM-12, and ADAM-17/TACE] mediate mHB-EGF cleavage in vivo under specific circumstances (Higashiyama, 2004; Hao et al., 2004). FIGURE 2 Proteolytic modification of cell surface proteins by MMPs in cancer development and evolution. All cell surface-associated molecules cleaved by MMPs during cancer development are presented on the cell membrane of a migrating tumor cell. These susbstrates are grouped according to the most affected cancer mechanism. Ectodomain cleavage of growth factor receptors may amplify the promotion of cell proliferation (HER2) or on the contrary release a soluble decoy receptor that binds soluble growth factors, thus decreasing their binding to intact growth-promoting receptors (FGFR-1). Proteolysis of FasL and Fas disturbs trimerization, which attenuates apoptosis induction, allowing the tumor cells to survive and proliferate. Formation of new vessels is crucial for the survival of a growing tumor. Cleavage of proteins on the endothelial cell membranes by MMPs may have a pro-angiogenic (VE-cadherin, PAR1) as well as an anti-angiogenic effect (uPAR, Betaglycan). Furthermore, alternation of adhesion and anti-adhesion is required as the cell migrates during the processes of invasion and metastasis. MMPs gromote metastasis by proteolysis of molecules (integrin precusors, tTG, CD44, E-cadherin), cytokines (RANKL), receptors (PAR1, EMMPRIN, LRP) and structural proteins (syndecan-1, APP). Many of these cleavages are mediated by MT1-MMP, which colocalizes with CD44 at the ruffling edge of migrating tumor cells. CD44 forms a platform to cluster MMPs, which stimulate migration by cleaving substrates on the cell surface and by degrading the ECM. Finally, cleavage of the trasmembrane proteins, IL-2Ralpha and ICAM-1 on transformed leukocytes, allows tumor cells to escape assualts from the immune system. Arrows indicate cleavages by MMPs. For details of the substrate acronyms, see the list of abbrevations.

FIGURE 3 Cleavage of the human FGFR-1 ectodomain-AP fusion protein by MMP-2 and comparison with murine FGFR-1. MMP-2 cleaves the FGFR-1 ectodomain from an immobilized FGFR1 ectodomain-alkaline phosphatase (FRAP) fusion protein in vitro at the Val368-Met369 bond, eight amino acids upstream of the transmembrane domain (A). This truncation releases a soluble ectodomain that retains FGF binding ability, thus decreasing FGF binding sites on the tumor cell and competing with the remaining intact FGFR-1. Murine FGFR-1 is shown for comparison (B). Arrow, cleavage site of MMP-2; AP, alkaline phosphatase; Ig, immunoglobulin domain; TK, tyrosine kinase domain; TM, transmembrane domain; R*, start of the AP. Adapted from (Levi et al., 1996). (Figure appears in color online at www.crbmb.com)

HB-EGF plays an important role in a multitude of biological processes and diseases. According to the concerned process, the mHB- EGF proteolysis is induced by varying stimuli and mediated by different metalloproteinases. Release of sHB-EGF allows cross- talking between G protein-coupled receptors (GPCRs) and TKRs in accordance with the triple membrane-passing signal mechanism (TMPS) for the transactivation of TKRs (Figure 4). In the TMPS, GPCRs are activated by an extracellular ligand and subsequently induce the activation of metalloproteinases through the initiation of several signaling cascades. Finally, metalloproteinase activity releases sHB- EGF that activates its TKR, the EGFR (Higashiyama, 2004; Shah and Catt, 2004a). Several studies showed enhanced HB-EGF gene expression in tumors compared to normal tissue (Raab and Klagsbrun, 1997). sHB- EGF expression increased tumor growth rate, colony-forming ability, and activation of the cyclin D1 promotor, as well as induction of vascular endothelial growth factor (VEGF) in vitro. In addition, sHB- EGF induced the expression and activity of MMP-3 and MMP-9, leading to enhanced cell migration. In vivo sHB-EGF enhanced tumorigenesis and angiogenesis. Hence, release of sHB-EGF by MMPs may be an interesting step for therapeutical intervention (Ongusaha et at, 2004; Miyamoto et al., 2004).

FIGURE 4 Release of sHB-EGF by MMPs allows cross-talking between G protein-coupled receptors (GPCRs) and tyrosine kinase receptors (TKRs) in accordance with the triple membrane-passing signal mechanism (TMPS). In the TMPS, GPCRs are first activated by an extracellular ligand (1) and subsequently induce the production of metalloproteinases through the initiation of several signaling cascades (2). Finally, metalloproteinase activity releases sHB-EGF that activates its TKR, the EGFR, which induces signal transduction to stimulate cell differentiation, growth and survival (3). Arrow, cleavage by an MMP. Adapted from (Shah and Call, 2004a). (Figure appears in color online at www.crbmb.com)

Besides stimulating tumor cell proliferation, sHB-EGF induces proliferation of Muller glial cells, which is a common feature of several diseases of the sensory retina. During proliferative vitreoretinopathy (PVR), Muller cells proliferate continuously, migrate onto retinal surfaces, and participate in the formation of periretinal cellular membranes (Rentsch, 1973; Bringmann and Reichenbach, 2001). Proliferation of Muller cells in PVR seems to be stimulated by binding of ATP to P2Y purine receptors (GPCRs) and requires the transactivation of two TKRs. Activation of P2Y receptors by ATP leads to an increase in intracellular Ca^sup 2+^ concentration and may cause a release of platelet-derived growth factor (PDGF) from the cells. Released PDGF activates the PDGF- alpha receptor (1st TKR), which causes MMP-9 production. MMP-9 then releases sHB-EGF, which in turn activates the EGFR (2nd TKR). Finally, the activated EGFRs and PDGF-alpha receptors induce proliferative activity in Muller cells by initiating several kinase signaling pathways. As a neutralizing antibody against MMP-9 reversed the mitogenic effect of ATP, inhibition of MMP-9 in the pathway may be a useful tool to suppress uncontrolled intraocular proliferation in PVR (Milenkovic et al., 2003; Shah and Catt, 2004a).

Besides the stimulating effects on tumor progression and PVR, proteolysis of mHB-EGF by MMPs plays a part in cardiovascular diseases (see Section 2.1.1) and reproductive endocrinology (see Section 5.2), which will be discussed in later chapters.

1.1.3 Tyrosine Kinase-Type Cell Surface Receptor HER2

The EGFR family comprises four TKRs of which tyrosine kinase- type cell surface receptor HER2 (HER2, receptor tyrosine-protein kinase erbB-2, p185erbB2, c-ErbB2, NEU proto-oncogene [neu], MLN 19, CD340) is the most oncogenic, as it is active in the absence of a stimulating ligand. In addition, HER2 in heterodimeric receptors is highly mitogenic because it decelerates growth factor dissociation from its partner receptor, prolonging the duration of intracellular signaling. In addition, endocytosis of HER2-containing complexes is relatively slow and these complexes tend to recycle back to the cell surface, rather than being degraded in the lysosomes. HER2, a 185 kDa transmembrane glycoprotein receptor, underlies many altered functions of tumor cells, including excessive growth, invasive behavior and attraction of blood vessels (Mosesson and Yarden, 2004). It is overexpressed in 25% to 30% of breast cancers, and it has been associated with high risks of relapse and death (Slamon et al., 1987), although the prognostic value of HER2 has been the matter of some controversy (Ross et al., 2003). Blockage of HER2 with the humanized anti-HER2 monoclonal antibody trastuzumab is an example of successful immunotherapy for HER-2-positive breast cancer patients, as it has been demonstrated to reduce the risk of recurrence by roughly 50% in five randomized clinical trials (PiccartGebhart, 2006).

The HER2 ectodomain is released and detected in the serum of cancer patients. High serum levels of HER2 ectodomain correlate with a poor prognosis and decreased responsiveness to therapy in patients with advanced breast cancer. This may be due to the enhanced signaling activity of the remaining cell-associated part of HER2. The cleavage is inhibited by the metalloproteinase inhibitors batimastat and TNF-alpha protease inhibitor (TAPI), as well as by TIMP-1, which inhibits soluble MMPs (Codony-Servat et al., 1999). In addition, trastuzumab inhibits basal and induced HER2 cleavage, preceding antibody-induced receptor downmodulation (Molina et al., 2001). Further identification of the metalloproteinase in charge could lead to new options for the therapy of patiens with breast cancer and high levels of HER2 shedding, for example by combining MMP inhibition with trastuzumab therapy. 1.2 Survival of Cancer Cells or Induction of Apoptosis

Evasion of apoptosis permits survival of tumor cells in spite of genetic instability. Low levels of oxygen and nutrients, host defence against the tumor by the immune system, anti-cancer treatments and local in vivo changes in the ECM with effects on invasion and metastasis alter the protease load in the tumor environment (Reed, 1999). Proteolysis by MMPs can have apoptotic as well as anti-apoptotic effects.

1.2.1 Fas Ligand (FasL)

An important effector in apoptosis is the Fas/FasL system, which is involved in three types of immune-associated killing: 1) elimination of virally infected cells and tumor cells by cytotoxic T cells (CTLs) and natural killer (NK) cells; 2) maintenance of immune privilege or survival of tumor cells; and 3) regulation of lymphocyte development and maintenance of peripheral immune homeostasis. Fas ligand (FasL, tumor necrosis factor ligand superfamily member 6, CD95 L, apoptosis antigen ligand (APTL), APO- 1 L, CD178) is a ~40 kDa type II transmembrane protein of the TNF family of death factors. It is expressed on activated T lymphocytes like CTLs and tumor infiltrating lymphocytes (TILs), on macrophages and NK cells and on cells within immune privileged tissues including the eye, testis, uterus, and placenta (Linkermann et al., 2003). After trimerization FasL induces apoptosis by binding to a trimer of its receptor Fas (tumor necrosis factor receptor superfamily member 6, apoptosis-mediating surface antigen FAS, APO-1, FASLG receptor, CD95). FasL can be released form the cell surface by MMP-3 and MMP- 7. However, the effects of soluble FasL (sFasL) on apoptosis and tumor progression seem to vary. Apoptosis may be induced in epithelial cells through cleavage of membrane-bound FasL to functional sFasL by MMP-7. Furthermore, this cleavage turns out to be crucial for apoptosis of prostate epithelium after castration, as prostate involution is significantly reduced in MMP-7 deficient mice (Powell et al., 1999). In contrast with apoptosis induction in epithelial cells, FasL shedding by MMP-7 has an anti-apoptotic effect in tumor cells, protecting them from chemotherapeutic drug toxicity (Mitsiades et al., 2001). This protection of tumor cells may be explained by the fact that the ability of sFasL to induce apoptosis is significantly lower than that of its cell surface precursor (Tanaka et al., 1998). A possible explanation for the paradoxical effects of FasL in both cell types might be the higher sensivity of epithelial cells to apoptotic signals, while FasL confers a kind of immune privilege to tumors by inducing apoptosis in infiltrating lymphocytes. Tumor cells indeed express lower levels of Fas and increased levels of FasL, which allows them to ‘counterattack’ Fas-bearing immune cells (Kim et al., 2004). Acute overexpression of MMP-7 in vitro as well as in vivo induces apoptosis, whereas chronic or repeated exposure to MMP-7 can select for tumor cells that are less sensitive to death-inducing stimuli (Fingleton et al., 2001; Vargo-Gogola et al., 2002b). Another potential explanation for the mentioned discrepancies in sFasL activity came up after the identification of novel MMP-7 cleavage sites in murine and human FasL. Mutational analysis showed that MMP- 7 cleaves human and murine FasL at an ELAELR sequence, close to the transmembrane domain (Figure 5), followed by a secondary cleavage at the COOH-terminally located SL sites. Strong indications exist that cleavage of human FasL also occurs at another ELR sequence in the trimerization domain. As trimerization of FasL has been shown to be important for its activity, cleavage within the trimerization domain may play a significant role in the regulation of sFasL function, releasing less active FasL. Human sFasL peptides generated after MMP- 7 cleavage at ELAELR contain 13 or 16 additional amino acids compared to the peptides formed after proteolysis at the SL site. The entire extracellular portion of FasL was shown to be pro- apoptotic in vitro, whereas sFasL produced by cleavage at the SL site was inactive in the same assay (Hohlbaum et al., 2000). This implies that the amino acid sequence between the SL cleavage site and the transmembrane domain contributes to the pro-apoptotic activity of sFasL. Therefore, truncation at the SL site might release inactive sFasL peptides (Vargo-Gogola et al., 2002a). Differential FasL proteolysis by MMP-7 might thus lead to the release of sFasL molecules with diverse biological activities.

1.2.2 Fas

MMP-7 expression also contributes to another strategy by which tumor cells can resist Fas-induced apoptosis. As a matter of fact, Fas too is cleaved by MMP-7, between Glu19-Leu20 and Asn32-Leu33 (Figure 6) (Strand et al., 2004). These cleavages remove 19 or 32 amino acids from the extracellular NH2-terminus of the Fas molecule and delete part of a domain for self-association termed ‘preligand assembly domain’ (PLAD). The PLAD domain has been shown to facilitate oligomerization of Fas receptors before ligand binding. Preassembly of Fas receptors might be crucial for the regulation of Fas signaling (Siegel et al., 2000). Fas proteolysis by MMP-7 indeed results in attenuated apoptosis induction (Strand et al, 2004). MMP- 7 is produced by the tumor cells themselves at early stages of tumor development. Therefore, the impact of this MMP on apoptosis resistance provides a potential target for new combination therapies in which MMP inhibition may significantly augment the efficacy of conventional chemotherapy (Poulaki et al., 2001).

1.3 Regulation of Angiogenesis

Oxygen and nutrients, provided by the vascular system, are crucial for cell function and survival, obligating virtually all cells in a tissue to reside within 100 [mu]m of a capillary blood vessel. Consequently, angiogenesis or the formation of new blood vessels, is a prerequisite for the survival of proliferating cells and is the result of subtle and often complex interactions between regulator and effector molecules (Hanahan and Weinberg, 2000; Pepper, 2001; Bergers and Benjamin, 2003).

1.3.1 Urokinase-Type Plasminogen Activator Receptor (uPAR)

The urokinase-type plasminogen activator receptor (uPAR, monocyte activation antigen Mo3, CD87) is a GPI-anchored receptor containing three homologous domains (D1, D2, and D3) (Blasi and Carmeliet, 2002). D1 is involved in the binding of the urokinase-type plasminogen activator (urokinase or u-PA) and enhances the interaction with the ECM by binding the ECM molecule vitronectin. D2 and D3 are also indispensable for high-affinity interactions (Hoyer- Hansen et al., 1997a; Oda et al., 1998). uPAR regulates u-PA activity-the activation of plasminogen to plasmin that degrades fibrin (Blasi and Carmeliet, 2002)-on the cell surface and is also important for the activation of signaling pathways through the interaction with several integrins (Reuning et al., 2003). Besides the membrane-anchored uPAR, a soluble receptor (suPAR) is released after proteolysis of the GPI anchor by cellular phospholipase D (Figure 7) (Wilhelm et al., 1999). Both uPAR and suPAR can be cleaved between the D1 and D2 domain, generating a D1-fragment and a D2D3-fragment (Blasi and Carmeliet, 2002). The D1 domain can be released by proteolytic activity of u-PA, directly or indirectly through activation of plasminogen (Hoyer-Hansen et al., 1997b), or by other proteases such as chymotrypsin and elastase (Ploug and Ellis, 1994). The specificty of the cleaving protease is crucial, as uPAR fragments may or may not be chemotactically active on different cell types in vitro, depending on the presence or absence of a specific NH^sub 2^-terminal peptide, SRSRY, on the D2D3-fragment (Fazioli et al., 1997). Inhibition of MMPs in a three-dimensional fibrin matrix, used as an experimental angiogenesis model, caused enhanced formation of capillary-like tubular structures and showed that the cleavage between D1 and D2 can also be performed by MMPs. The first MMP to be identified as uPAR sheddase was MMP-12 (Koolwijk et al., 2001), but also several other MMPs were able to release the D1 domain of suPAR in vitro: MMP-2, MMP-3, MMP-8, MMP-9, MMP-13, MMP- 19, MT1-MMP, and MT6MMP (Andolfo et al., 2002). However, the most efficient MMPs for uPAR cleavage were MMP-3, MMP-12, MMP-19, and MT6- MMP. In particular, MMP-3, MMP-12, and MT6-MMP show the same specificity of cleavage having the Thr86-Tyr87 peptide bond as major cleavage site, while MMP-19 cleaves suPAR predominantly at Tyr87- Ser88 (Figure 7). uPAR cleavage by MMPs results in the exposure of the chemotactic epitope SRSRY at the NH^sub 2^-terminus of the generated D2D3-fragment and may thus generate biologically active fragments. Murine uPAR is cleaved by MMP-12 as well (at the Pro89- Gln90 peptide bond), despite the limited sequence homology between the linker regions (Andolfo et al., 2002).

FIGURE 5 Cleavage sites of MMP-7 in the ectodomain of human and murine FasL. MMP-7 cleaves human and murine FasL at an ELAELR sequence, close to the transmembrane domain (TM), followed by a secondary cleavage at COOH-terminally located SL sites. Human sFasL peptides generated after MMP-7 cleavage at ELAELR were shown to be pro-apoptotic in vitro. However, the peptides formed after proteolysis at the S^sub 126^-L^sub 127^ site, which contain 13 or 16 amino acids less, were inactive in the same apoptosis assay. This might mean that the amino acid sequence between the SL cleavage site and the transmembrane domain contributes to the pro-apoptotic activity of sFasL. Cleavage of human FasL almost certainly occurs at an additional ELR sequence, E^sub 142^-L^sub 143^R, in the trimerization domain (TRIM). As trimerization of FasL has been shown to be important for its activity, cleavage within the trimerization domain may release less active FasL. Arrows indicate the cleavage sites of MMP-7. Adapted from (Vargo-Gogola et al., 2002a). (Figure appears in color online at www.crbmb.com) FIGURE 6 MMP-7 cleaves the ‘preligand assembly domain’ (PLAD) in the Fas ectodomain. Fas is cleaved by MMP-7 between Glu19-Leu20 and Asn32-Leu33. These cleavages remove 19 or 32 amino acids from the Fas NH^sub 2^- terminus and delete part of a domain for self-association termed ‘preligand assembly domain’ (PLAD). The PLAD domain facilitates oligomerization of Fas receptors before ligand binding. Preassembly of Fas receptors may be crucial for the regulation of Fas signaling and proteolysis of the PLAD domain by MMP-7 indeed results in decreased sensitivity of tumor cells to Fas-mediated apoptosis. Arrows indicate the cleavage sites of MMP-7; CRD, cysteine-rich domain; TM, transmembrane domain. Adapted from (Strand et al., 2004). (Figure appears in color online at www.crbmb.com)

FIGURE 7 Cleavage of uPAR and suPAR by MMPs releases the u-PA- binding domain D1. uPAR is a GPI-anchored receptor for u-PA containing 3 homologous domains (D1, D2, and D3). In addition to the membrane-anchored uPAR, a soluble receptor (suPAR) is released after cleavage of the GPI anchor by cellular phospholipase D. Both uPAR and suPAR can be cleaved between the D1 and D2 domains, generating a D1-fragment and a D2D3-fragment. Release of the D1 domain can be mediated by MMPs, but also by serine proteases such as u-PA, plasmin, chymotrypsin and elastase. The first MMP to be identified as uPAR sheddase was MMP-12 but also several other MMPs release the D1 domain of suPAR in vitro. However, the MMPs most efficient at uPAR cleavage are MMP-3, MMP-12, MMP-19 and MT6-MMP. In particular, MMP-3, MMP-12 and MT6-MMP show the same specificity of cleavage with the Thr86-Tyr87 peptide bond as major cleavage site, whereas MMP-19 cleaves suPAR predominantly at Tyr87-Ser88. Loss of the D1 domain results in less cellular u-PA binding and, consequently, in diminished formation of new capillary structures in an angiogenesis model. (Figure appears in color online at www.crbmb.com)

Unexpected enhanced angiogenesis through MMP inhibition might be caused by the higher availability of functional uPAR at the cell surface, resulting in increased u-PA binding and, subsequently, enhanced formation of new capillary structures. As plasmin activates several MMPs (for reviews, see: Collen, 2001; Pepper, 2001; Van den Steen et al., 2001), release of the D1 fragment by an MMP might be a kind of feedback regulation. Since angiogenesis in and growth of some tumors are inhibited by competitors of uPAR, and taking into account that some MMPs (such as MMP-9) exert proangiogenic effects, the influence of MMP inhibition on uPAR levels and on angiogenesis certainly requires further examination.

Cleavage of uPAR by MMP-12 may also play a role in the autoimmune disease systemic sclerosis. This will be detailed in a subsequent paragraph (see Section 3.4.4).

7.3.2 Betaglycan

Betaglycan, also known as the TGF-beta type III receptor, is a membrane-anchored proteoglycan whose glycosaminoglycan chains consist of heparan and chondroitin sulfate. Membrane-bound betaglycan is generally considered a positive regulator of TGF-beta because it increases the binding affinity of TGF-beta for its receptor II, enhancing cell responsiveness to TGF-beta (Lopez- Casillas et al., 1994). TGF-beta controls many physiological processes and has tumor-suppressing activity in the early phases of carcinogenesis. In subsequent stages of tumor progression, the increased secretion of TGF-beta by both tumor cells and stroma cells, is involved in the enhancement of tumor invasion and mestastasis, accompanied by immunosuppression (Kim et al., 2004). TGF-beta upregulates MMP-9 activity, which may also amplify angiogenesis and tumor growth. Additionally, MMP-9 is capable of activating latent TGF-beta (Yu and Stamenkovic, 2000; Bandyopadhyay et al., 2005), thanks to the anchoring of MMP-9 and TGF-beta to CD44 on the cell surface (see Section 1.4.7) (Yu and Stamenkovic, 1999).

Two soluble forms of betaglycan are relased by proteolytic cleavage. In some cell types these cleavages are induced by the tyrosine phosphatase inhibitor, pervanadate, and generate a bigger fragment of 120 kDa (sBG-120), which encompasses almost the entire extracellular domain, and a smaller 90 kDa fragment (sBG-90). The cleavage that generates sBG-90 is inhibited by TIMP-2, but not by TIMP-1, which points to an MT-MMP as the involved protease. Overexpression of MT1-MMP and MT3-MMP, but not of the other MT- MMPs, indeed releases the sBG-90 fragment. Surprisingly, MT2-MMP overexpression decreases the levels of betaglycan and of MT1-MMP (Velasco-Loyden et al., 2004). In contrast with membrane-bound betaglycan, recombinant soluble betaglycan has been shown to inhibit TGF-beta in vitro (Lopez-Casillas et al., 1994). Therefore, betaglycan might function as a dual modulator of TGF-beta activity: as a membrane-anchored protein it enhances TGF-beta activity, whereas its soluble form causes TGF-beta inhibition. TGF-beta- promoted tumor-host interactions leading to enhanced angiogenesis have been shown to be effectively attenuated by the systemic administration of soluble betaglycan in a xenograft model of prostate cancer. The inhibition of tumor angiogenesis and consequently of tumor growth appears at least in part due to the inhibition of TGF-beta-induced MMP-9 upregulation (Bandyopadhyay et al., 2005). Further investigation will clarify if sBG-90 and sBG- 120 have the same TGF-beta-inhibiting activities and if proteolysis of betaglycan by MT1-MMP or MT3-MMP can reduce angiogenesis and tumor growth.

7.3.3 Vascular Endothelial Cadherin (VE-Cadherin)

Vascular endothelial-cadherin (VE-cadherin, cadherin-5, 7B4 antigen, CD144) is a member of the large cadherin family that includes Ca^sup 2+^-dependent cell-cell adhesion molecules responsible for cell-to-cell recognition and adhesion in solid tissues. Cadherins dimerize through the extracellular domain with other cadherin molecules on adjacent cells (trans-interaction). In this homotypic interaction, the intracellular domain interacts with various catenin proteins to form the cytoplasmic cell-adhesion complex (CCC), which is crucial for strong cell-cell adhesion and potent suppression of invasion. Cadherins are expressed in several types of tissues with some specificity: Epithelial (E)-cadherin is mostly present in epithelial cells, Neuronal (N)-cadherin in the nervous system, smooth muscle cells, fibroblasts and endothelial cells, and VE-cadherin is specific for the endothelium (Cavallaro and Christofori, 2004; Cavallaro et al., 2006).

MMP-7 treatment of human umbilical endothelial cells (HUVECs) accelerates HUVEC proliferation and degrades VE-cadherin on the cell surface, with concomitant accumulation of beta-catenin in the nucleus and an increase of MMP-7 expression. These results suggest that MMP-7-mediated cleavage of VE-cadherin releases beta-catenin from the VE-cadherin/catenin complex, allowing it to translocate from the cytoplasm to the nucleus, where it can activate T-cell factor DNA binding protein, which accelerates cell proliferation and MMP-7 expression (Ichikawa et al., 2006). However, MMP-7 also has another important role in angiogenesis as it cleaves plasminogen and converts it to angiostatin (Patterson and Sang, 1997), which is one of the strongest inhibitors of angiogenesis. In light of this, it is crucial to further examine the sometimes paradoxical effects of MMP- 7 on angiogenesis before targetting its activity in anti-cancer therapy.

7.3.4 Semaphorin 4D

Semaphorins are secreted, transmembrane or GPI-linked proteins, defined by cysteine-rich semaphorin protein domains, that have essential roles in a variety of tissues. Functionally, semaphorins were initially characterized for their importance in the development of the nervous system and in axonal guidance. More recently, they have been found to play a role in a wide range of processes, including tissue organization during development, angiogenesis, immunoregulation, and tumor progression. A common theme in the mechanisms of semaphorin function is that they alter the cytoskeleton, i.e., the organization of actin filaments and the microtubular network, through binding with their receptors. The best characterized semaphorin receptors are members of the neurophilin and plexin families (Yazdani and Terman, 2006). Plexin-1B is highly expressed in endothelial cells and promotes migration and tubulogenesis in vitro as well as in vivo when bound by its ligand, semaphorin 4D (BB18, A8, GR3, CD100) (Basile et al, 2004). Semaphorin 4D is highly expressed in head and neck squamous cell carcinomas as well as in some of the most prevalent solid tumors, including breast, prostate, colon and lung cancer tissues (Basile et al., 2006).

In order to exert its pro-angiogenic functions, semaphorin 4D, a transmembrane protein, must be processed and released into a soluble form to act in a paracrine manner on endothelial cells. Semaphorin 4D is expressed on the cell surface as a homodimer (see Figure 2), which is a prerequisite for its proteolytic release from the cell surface (Elhabazi et al., 2001). The shedding process can be inhibited by the metalloproteinase inhibitors EDTA, EGTA and Ilomastat/GM6001, as well as by TIMP-2, but not by TIMP-1 (Elhabazi et al, 2001; Basile et al, 2007). The inhibitor profile, combined with the observation that MT1-MMP, while not expressed in non- tumorigenic epithelial cell lines, was present in several head and neck squamous carcinoma cell lines, pointed to MT1MMP as the semaphorin 4D sheddase. Basile and coworkers (2007) demonstrated that MT1-MMP was required for processing and release of semaphorin 4D from these cells, thereby inducing endothelial cell chemotaxis in vitro and blood vessel growth in vivo. As a consequence, MT1-MMP- dependent shedding of semaphorin 4D may play a critical role in tumor-induced angiogenesis, and therefore may represent new fronts of attack in the anti-angiogenic therapy of cancer. 1.4 Stimulation or Inhibition of Migration, Invasion, and Metastasis

Much like tumorigenesis, the processes of invasion and metastasis are highly complex. In its simplest form, metastasis requires the tumor cell to detach from its primary location, invade through stromal elements or existing junctions between normal cells, enter and leave blood vessels or lymphatics, and then establish a colony at the metastatic site. These steps require molecular processes at the cell surface in which contacts between the invading tumor cell and surrounding cells and stroma are repeatedly broken (anti- adhesion) and new contacts established as the tumor cell moves forward (adhesion). This alternation of adhesion and anti-adhesion can be achieved through coordinated expression of proteases and adhesion molecules (Chambers et al., 2002; Hollingsworth and Swanson, 2004).

1.4.1 Mucin-1 (MUC1)

Mucin-1 (MUC1, polymorphic epithelial mucin [PEM, PEMT], episialin, tumor-associated mucin, carcinoma-associated mucin, tumor- associated epithelial membrane antigen [EMA], H23AG, peanut- reactive urinary mucin [PUM], breast carcinoma-associated antigen DF3, CD227), a transmembrane mucin, plays a key role in the inhibition of embryo implantation, in the protection of mucosal surfaces against microbial and proteolytic degradation, and in some aspects of tumor progression. It is expressed on several epithelial surfaces like uterine, lung and intestinal epithelia as well as on tumor cells. The relatively short cytoplasmic tail associates with cytoskeletal elements, cytosolic adaptor proteins and/or participates in signal transduction. The extracellular domain can be released form the cell surface, which might serve practical functions, such as the facilitation of rapid clearance of mucosa surface-associated material, or cell mobility. The mechanism controling this ectodomain release has not yet been elucidated, even though it has been postulated that alterations in pH, ionic concentration or hydration might trigger proteolytic release by specific proteases (Hollingsworth and Swanson, 2004). The major protease in this cleavage process is TACE (Thathiah et al., 2003). However, according to the TIMP-inhibition profile (see Table 1), an additional MUC1 shedding activity belonging to the MT-MMP family was identified in TACE deficient cells. MT1-MMP is expressed on these cells and, in addition, MT1-MMP overexpression or deficiency causes increased or inhibited MUC1 shedding, respectively. Furthermore, MT1- MMP indeed cleaves MUC1 in vitro and is colocalized with MUC1 in vivo in human uterine epithelia (Thathiah and Carson, 2004). Cancer cells, especially from adenocarcinomas, express aberrant forms and levels of mucins, which have an impact on the biological properties of tumors in several ways (Hollingsworth and Swanson, 2004). Evidences are emerging that cell surface mucins contribute to the regulation of differentiation and proliferation of tumor cells, through ligand-receptor interactions and morphogenetic signal transduction. In addition, MUC1 expression on tumor cells causes anti-adhesion through steric hindrance by forming multiple exposed glycosylated rod-like structures and by the binding of receptors on the same cell (cis-interactions), preventing interaction of these receptors with other cells. This antiadhesion permits tumor cells to detach form the tumor mass and to invade the surrounding stroma. Invasion is subsequently enhanced, as MUC1 also has an adhesive action by binding adhesion molecules on stroma cells and endothelial cells (fram-interactions). Finally, MUC1 also contributes to immune evasion by forming a leukocyte-impermeable barrier around the tumor and through immunosuppressive effects on T-cell proliferative responses. Since MT1-MMP is often expressed by cancer cells, the cleavage of MUC1 may affect all these processes.

7.4.2 Epithelial Cadherin (E-Cadherin)

Epithelial cadherin (E-cadherin, uvomorulin, cadherin-1, CAM 120/ 80, CD324) is another member of the cadherin family and is expressed mostly in epithelial cells (see Section 1.3.3). Loss of E-cadherin- mediated cell-cell adhesion has been shown to be a prerequisite for tumor cell invasion and metastasis (Birchmeier and Behrens, 1994). Proteolytic degradation of E-cadherin by MMPs is one of the regulation mechanisms by which epithelial cell-cell adhesion can be ablated. MMP-3 and MMP-7 both cleave the 120 kDa transmembrane E- cadherin, releasing a 80 kDa soluble form (sE-cadherin) (Lochter et al., 1997; Noe et al., 2001). Treatment of prostate cancer cells with hepatocyte growth factor/scatter factor (HGF/SF) causes shedding of E-cadherin through the induction of MMP-7, resulting in cell scattering and a switch to a more invasive phenotype (Davies et al., 2001). Released sE-cadherin induces invasion in vitro and inhibits cell aggregation indicating that it disturbs cell-bound E- cadherin functions in a paracrine way (Noe et al., 2001). However, it is not clear whether sE-cadherin induces invasion by perturbation of cell-cell interaction and/or by engaging a signalling pathway in which free beta-catenin can act as a transcriptional coactivator (Hecht and Kemler, 2000). An induction of MMP-2, MMP-9, and MT1-MMP expression was observed both at the mRNA and protein levels in the presence of sE-cadherin. ECM degradation by these MMPs might be an extra mechanism by which E-cadherin ectodomain shedding contributes to tumor invasion and metastasis formation (Nawrocki-Raby et al., 2003). Besides the influence on invasion and metastasis, proteolysis of E-cadherin also plays a role in epithelialmesenchymal transition, a conversion to an altered cellular phenotype which is associated with aggressive malignant behaviour (Cavallaro and Christofori, 2004). Induction of MMP-3 expression in mammary epithelial cells results in E-cadherin cleavage and triggers a progressive phenotypic conversion cumulating in cells that are unable to undergo lactogenic differentiation and that become invasive (Lochter et al., 1997; Sternlicht et al., 1999). Finally, during apoptosis, the cytoplasmic tail of E-cadherin is truncated by caspase-3, whilst a 84 kDa ectodomain fragment is released by a metalloproteinase. This simultanous cleavage of intracellular and extracellular domains might be a highly efficient mechanism to disrupt E-cadherin- dependent cell-cell contacts in apoptotic cells or tumor cells, which is a prerequisite for cell rounding and exit from the epithelium (Steinhusen et al., 2001).

In acute renal failure, E-cadherin degradation by MT1MMP leads to disruption of epithelial integrity and epithelial cell shedding (see Section 2.2.3), whereas in normal physiology, MMP-7-mediated shedding of E-cadherin is required for the repair of injured lung epithelium (McGuire et al., 2003).

1.4.3 Integrin Subunit Precursors

Integrins are a diverse family of transmembrane glycoproteins that form heterodimeric receptors for ECM molecules and membrane- associated molecules of the Ig family. Every integrin is composed of a non-covalently coupled alpha- and beta-subunit. The 18 known alpha- subunits and 8 beta-subunits form at least 25 distinct heterodimers in human, with each pair being specific for a unique set of ligands. Integrins are crucial for cell adhesion, migration and invasion, not only through the direct physical adhesion to the ECM and to other cells, but also because they send and receive molecular signals that are essential for these processes (inside-out and outside-in signaling, respectively) (Hood and Cheresh, 2002).

The binding of integrins to ECM molecules is altered by changes in integrin expression and affinity when cancer cells become metastatic, or when endothelial cells enter the angiogenic state (Varner and Cheresh, 1996; Demetriou and Cress, 2004). Maturation of some integrin subunits requires a posttranslational cleavage of the precursor chain. The exact role of this endoproteolytical modification in integrin function is unclear, but its absence has important consequences for signal transduction pathways and leads to alterations in integrin functions such as cell adhesion to vitronectin (Berthet et al., 2000). Proteolysis of pro-integrin chains is performed by proprotein convertases (PC) of the subtilisin/ kexin-like family, but also by MT1-MMP. MT1-MMP cleaves the pro- alpha^sub v^ chain (integrin alpha-V precursor, vitronectin receptor subunit alpha, CD51) at two different positions between the disulfide-connected Cys852 and Cys904. The first cleavage, between Asp891 and Leu892, generates the 125 kDa heavy chain, disulfide- bound to a 25 kDa light chain that is one residu shorter at the N- terminus in comparison with the PC-processed chain (Figure 8). The putative second MT1-MMP cleavage site is situated downstream from the Cys852 and produces a 115 kDa heavy alpha-chain (Ratnikov et al., 2002). These proteolytic modifications do not affect ligand binding of the resulting alpha^sub v^beta^sub 3^ integrin but enhance outside-in signal transduction. As a result, cells co- expressing MT1-MMP and alpha^sub v^beta^sub 3^ integrin show more efficient adhesion and migration on vitronectin, the ECM ligand of alpha^sub v^beta^sub 3^ (Deryugina et al., 2002).

FIGURE 8 Endoproteolytic cleavage of pro-alpha^sub v^-integrin by MT1-MMP and by a proprotein convertase. Proteolysis of pro-integrin chains is performed by proprotein convertases, but also by MT1-MMP. MT1-MMP cleaves the pro-alpha^sub v^ chain at two different positions between the disulfide-connected Cys852 and Cys904. The first cleavage, between Asp891 and Leu892, generates a 125 KDa heavy alpha-chain, disulfide-bound to a 25 kDa light chain that is one residu shorter at the N-terminus in comparison with the convertase- processed chain. The second MT1-MMP cleavage removes an additional 10 kDa from the heavy alpha-chain and is situated downstream from the Cys852. These proteolytic modifications do not affect ligand binding of the resulting alpha^sub v^beta^sub 3^ integrin but enhance outside-in signal transduction, which results in more efficient adhesion and migration on vitronectin. Adapted from (Ratnikov et al., 2002). (Figure appears in color online at www.crbmb.com) In addition, the MT1-MMP-mediated cleavage seems to regulate cross-talk between alpha^sub v^beta^sub 3^ and alpha^sub 2^beta^sub 1^ integrins. In MT1-MMP deficient cells, the functional activity of the collagen-binding alpha^sub 2^beta^sub 1^-integrin, which is not cleaved by MT1-MMP, is suppressed by the presence of alpha^sub v^beta^sub 3^ integrin, resulting in diminished cell adhesion to collagen type I. Co-expression of MT1-MMP and alpha^sub v^beta^sub 3^ restores the alpha^sub 2^beta^sub 1^-mediated collagen binding. Expression of both alpha^sub v^beta^sub 3^ integrin and MT1- MMP is elevated in malignant tumor cells and is correlated with increased migration. Cross-talk between alpha^sub v^beta^sub 3^ and alpha^sub 2^beta^sub 1^ integrins through MT1-MMP-mediated pro- alpha^sub v^-proteolysis might thus contribute to efficient adhesion of aggressive tumor cells to type I collagen, an important substratum of the ECM (Baciu et al., 2003). In addition, MT1-MMP modifies the beta^sub 3^-subunit (integrin beta-3 precursor, platelet membrane glycoprotein IIIa (GPIIIa), CD61) of alpha^sub v^beta^sub 3^ integrin from a 95 kDa to a 90 kDa chain, which is also correlated with functional alpha^sub v^beta^sub 3^ integrin activation and increased adhesion on vitronectin (Deryugina et al., 2000). Moreover, after functional activation by MT1-MMP, alpha^sub v^beta^sub 3^ integrin shows a higher affinity for the MMP-2 hemopexin domain, and binding to alpha^sub v^beta^sub 3^ integrin concentrates MMP-2 at specific spots on the cell surface, which may again contribute to cell migration through the ECM. Hence, functional regulation of integrins by MT1-MMP plays an essential role in efficient adhesion and directional migration of tumor cells. As a consequence, this pathway could be another interesting target in therapeutic approaches aiming at suppressing tumor cell spreading.

MT1-MMP also mediates maturation of the pro-alpha^sub 3^ (integrin alpha-3 precursor, galactoprotein B3 (GAPB3), VLA-3 alpha chain, FRP-2, CD49c) and pro-alpha^sub 5^ (integrin alpha-5 precursor, fibronectin receptor subunit alpha, integrin alpha-F, VLA- 5, CD49e) chains to the respective mature disulfide-bound heavy and light alpha-chains (Baciu et al., 2003). However, pro-alpha^sub 5^ cleavage does not affect alpha^sub 5^beta^sub 1^ integrin function. The consequences of pro-alpha^sub 3^ cleavage have not been investigated.

MMP-7 is another tumor-derived MMP capable of integrin subunit proteolysis. MMP-7 releases a 90 kDa fragment of the 200 kDa beta^sub 4^-subunit precursor (integrin beta-4 precursor, GP150, CD104), possibly through cleavage of the Tyr106-Ile107 and Gly416- Leu417 peptide bounds. The first putative cleavage site is very close to the ligand-binding domain, and cleavage might thus interfere with the binding of the alpha^sub 6^beta^sub 3^ integrin to its ligand laminin (von Bredow et al., 1997). alpha^sub 6^beta^sub 4^ integrin has been most implicated in epithelial carcinogenesis (Watt, 2002). In epithelium-derived carcinoma, increased levels of beta^sub 4^ integrin and loss of its polarized distribution to the basolateral membrane site of the cell have been correlated with tumor aggressiveness (Rigot et al., 1999). Adhesion of breast cancer cells to endothelial cells through binding of alpha^sub 6^beta^sub 4^ integrin to a specific lung-endothelial cell adhesion molecule is critical for lung metastasis, but is totally abolished after cleavage of the beta^sub 4^ integrin ectodomain by MMP-7 (Abdel-Ghany et al., 2001). Hence, in this case, MMP-mediated modification of an integrin subunit reduces tumor cell adhesion and migration.

Furthermore, integrins are important players in tumor cell proliferation, apoptosis and angiogenesis, in leukocyte migration and in a whole array of pathologies (Wehrle-Haller and Imhof, 2003). As a consequence, insight in the functional regulation of these bidirectional signaling molecules by MMPs may generate new possibilities for therapeutic intervention.

1.4.4 Tissue Transglutaminase (tTG)

Another adhesion and signaling receptor being cleaved by MT-MMPs is the ubiquitously expressed cell surface-associated tissue transglutaminase (tTG, protein-glutamine gamma-glutamyltransferase 2, TGase C (TGC), transglutaminase-2, TGase-H). tTG catalyzes covalent cross-linking between reactive lysine and glutamine residues of proteins and protein polymers. In addition, tTG functions as a coreceptor for beta^sub 1^ and beta^sub 3^ integrins and promotes integrin-dependent adhesion and cell spreading on fibronectin. Good functioning of cell surface tTG is essential for regulation and maintenance of cell-matrix interactions, as well as for the mobility of tumor and host cells. On the contrary, deregulation of tTG activity is associated with multiple human diseases (Griffin et al., 2002). Overexpression of MT1-MMP by glioma and fibrosarcoma cells causes proteolytic degradation of tTG at the leading edge of motile cancer cells, leading to specific suppression of cell migration and adhesion on fibronectin (Belkin et al., 2001). 80 kDa tTG degradation is mediated in vitro by MT1-MMP, MT2-MMP, and MT3-MMP, but not MT4-MMP, and produces fragments of ~53 kDa, ~41 kDa and ~32 kDa. MT1-MMP cleavage at Arg458-Ala459 and His461-Leu462 generates the 53 and 32 kDa fragments, while cleavage at Pro375- Val376 splits the protein in half, providing the 41 kDa fragments. Cleavage at any of these three sites abolishes receptor and enzymatic activity by separating the NH^sub 2^-terminal fibronectin- binding domain and the COOH-terminal integrin-binding domain, as well as by inactivation of the catalytic domain. tTG proteolysis suppresses cell adhesion and migration on fibronectin. Reciprocally, fibronectin protects its surface r