MECHANISTIC FEATURES OF ECTODOMAIN SHEDDING
Ectodomain shedding affects a diverse list of molecules, including cytokines, growth factors, cell adhesion molecules, enzymes, and many more (Table 1). There are several common mechanistic features in the ectodomain shedding of cell surface molecules (Fig. 1). Most show little basal shedding, but are dramatically induced upon cellular activation. Phorbol esters induce shedding of the majority of protein ectodomains, suggesting that protein kinase C (PKC) is one of the key regulators. Indeed, chemically mutagenized CHO cells that are defective in PMA-induced TGFα shedding are also defective in phorbol 12-myristate 13-acetate (PMA)-induced shedding of several other transmembrane proteins, such as TNFα, amyloid precursor protein (APP), and L-selectin (Arribas and Massague,1995; Arribas et al.,1996). Furthermore, inhibitors of protein tyrosine kinases (PTKs; Fitzgerald et al.,2000; Gutwein et al.,2000) and MAP kinases inhibit shedding (Fan and Derynck,1999; Fitzgerald et al.,2000), whereas agonists of G-protein coupled receptors (GPCRs), calcium ionophores, and ceramide induce shedding of a number of molecules (Porteu and Nathan,1990; Madge et al.,1999; Fitzgerald et al.,2000; Lemjabbar and Basbaum,2002; Matthews et al.,2003; Mortier et al.,2004). These data illustrate that several intracellular signaling pathways regulate ectodomain shedding and, consistent with these observations, agonists that activate these signaling pathways, such as growth factors, cytokines, and bacterial toxins, have been shown to enhance shedding (Subramanian et al.,1997; Jones et al.,1999; Yabkowitz et al.,1999; Fitzgerald et al.,2000; Park et al.,2000,2004; Chen et al.,2007). In addition, peptide hydroxamate inhibitors of metalloproteinases inhibit the shedding of most cell surface proteins, with the exception of glycosylphosphatidylinositol (GPI)-anchored proteins, which are primarily shed by phospholipases. Collectively, these data indicate that ectodomain shedding is generally regulated by outside-in signaling, which in turn modulates substrate cleavage by metalloproteinase sheddases at the cell surface.
|Cell adhesion molecules||Growth factors and receptors||Immunomodulators and receptors|
|Collagen XIII (Vaisanen et al.,2004)||Amphiregulin (Sahin et al.,2004)||Betaglycan (Velasco-Loyden et al.,2004)|
|Collagen XVII (Franzke et al.,2002)||Betacellulin (Sahin et al.,2004)||CD23 (Letellier et al.,1990)|
|Collagen XXIII (Veit et al.,2007)||CHL1 (Naus et al.,2004)||CD30 (Hansen et al.,2004)|
|E-Cadherin (McGuire et al.,2003)||c-Met (Nath et al.,2001)||CD40 (Contin et al.,2003)|
|N-Cadherin (Reiss et al.,2005)||EGF (Sahin et al.,2004)||CD93 (Bohlson et al.,2005)|
|VE-Cadherin (Herren et al.,1998)||EGFRs (Lin and Clinton,1991)||CD223 (Li et al.,2007)|
|CD44 (Bazil and Strominger,1994)||Ephrins (Hattori et al.,2000)||CSF-1 (Tuck et al.,1994)|
|Glycoprotein V (Rabie et al.,2005)||Epiregulin (Sahin et al.,2004)||CXCL16 (Matloubian et al.,2000)|
|Glycoprotein VI (Bergmeier et al.,2004)||FGFR1 (Hanneken et al.,1994)||Fas (Cheng et al.,1994)|
|ICAM1 (Becker et al.,1991)||GHR (Alele et al.,1998)||FasL (Tanaka et al.,1998)|
|L1 (Beer et al.,1999)||HB-EGF (Suzuki et al.,1997)||Fractalkine (Garton et al.,2001)|
|NCAM (Hubschmann et al.,2005)||HGF (Mizuno et al.,1994)||GM-CSFR (Prevost et al.,2002)|
|Nectin-1α (Tanaka et al.,2002)||KL-1 (Pandiella et al.,1992)||IL-1RII (Cui et al.,2003b)|
|Nectin-4 (Fabre-Lafay et al.,2005)||Neuregulin (Sahin et al.,2004)||IL-2R (Junghans and Waldmann,1996)|
|Neuroglycan C (Shuo et al.,2007)||NgR (Ahmed et al.,2006)||IL-6R (Croucher et al.,1999)|
|NG2 (Asher et al.,2005)||NTR(p75) (Weskamp et al.,2004)||IL-15R (Mortier et al.,2004)|
|PECAM-1 (Ilan et al.,2001)||Osteoactivin (Furochi et al.,2007)||LAR (Streuli et al.,1992)|
|E-selectin (Wyble et al.,1997)||TGFα (Wong et al.,1989)||LOX-1 (Hayashida et al.,2005)|
|L-selectin (Migaki et al.,1995)||TIE1 (Marron et al.,2007)||RANKL (Hikita and Tanaka,2007)|
|P-selectin (Semenov et al.,1999)||TSHR (Couet et al.,1996)||TNFα (Black et al.,1997)|
|SHPS-1 (Ohnishi et al.,2004)||TNFRI, TNFRII (Porteu and Nathan,1990)|
|Syndecan-1 (Subramanian et al.,1997)||TRANCE (Lum et al.,1999)|
|Syndecan-2 (Fears et al.,2006)||TREM1 (Gomez-Pina et al.,2007)|
|Syndecan-3 (Reizes et al.,2001)|
|Syndecan-4 (Subramanian et al.,1997)||Enzymes||Others|
|VCAM-1 (Singh et al.,2005)||ACE (Sadhukhan et al.,1999)||APP (Esch et al.,1990)|
|BACE1 (Hussain et al.,2003)||Delta (LaVoie and Selkoe,2003)|
|MT1-MMP (Osenkowski et al.,2004)||Jagged (LaVoie and Selkoe,2003)|
|Leptin receptor (Ge et al.,2002)|
|MUC-1 (Thathiah et al.,2003)|
|Notch (Pan and Rubin,1997)|
|Transferrin receptor (Chitambar et al.,1991)|
Extracellular and Intracellular Regulators of Shedding
Ectodomain shedding can be induced by various stimuli, such as phorbol esters (Arribas and Massague,1995; Arribas et al.,1996), cytokines (Becker et al.,1991; Wyble et al.,1997; Jones et al.,1999; Yabkowitz et al.,1999; Singh et al.,2005), growth factors (Subramanian et al.,1997; Fitzgerald et al.,2000; Mortier et al.,2004), pervanadate (Reiland et al.,1996; Gutwein et al.,2000; Kaup et al.,2002; Le Gall et al.,2003; Velasco-Loyden et al.,2004), ceramide (Fitzgerald et al.,2000; Walev et al.,2000; Matthews et al.,2003), cellular stress (Fitzgerald et al.,2000), calcium ionophores (Dethlefsen et al.,1998; Reiss et al.,2005), and bacterial toxins (Walev et al.,1996; Park et al.,2000,2004; Chen et al.,2007). Some of these globally activate shedding of many cell surface proteins, whereas others specifically stimulate shedding of certain protein ectodomains. Rapid induction of shedding by phorbol esters is one of the hallmark features of ectodomain shedding and, accordingly, several PKC isozymes regulate shedding. For example, PKCϵ is required for TNFα shedding (Wheeler et al.,2003). PKCδ regulates heparin binding (HB) epidermal growth factor (EGF)-like shedding (Izumi et al.,1998), whereas PKCδ and PKCη are involved in IL-6 receptor (IL-6R) shedding (Thabard et al.,2001). These findings indicate that different substrates require distinct PKC isozymes for its shedding regulation, but the underlying basis of this specificity is not understood.
Several studies indicate that PTKs are also positive regulators of ectodomain shedding (Subramanian et al.,1997; Fitzgerald et al.,2000; Gutwein et al.,2000; Phong et al.,2003). Pervanadate, a general and potent inhibitor of protein tyrosine phosphatases, enhances the shedding of the adhesion molecule L1, APP, syndecan-1 and -4, and angiotensin converting enzyme (ACE), among others, and PTK inhibitors inhibit the shedding of these surface molecules (Subramanian et al.,1997; Fitzgerald et al.,2000; Gutwein et al.,2000; Park et al.,2000; Phong et al.,2003). Precisely how kinases modulate shedding has yet to be clearly defined, but current data suggest that both PTKs and PKCs do not enhance shedding by phosphorylating the substrate. For example, pervanadate-induced ACE shedding does not require the cytoplasmic domain of ACE (Santhamma et al.,2004). Similarly, deletion of the cytoplasmic tails of IL-6R (Mullberg et al.,1994) and TNFα receptor II (TNFRII) (Crowe et al.,1993) does not affect shedding of their ectodomains by PMA, despite the fact that the cytoplasmic domains of these cytokine receptors are phosphorylated by PMA stimulation. Antibody crosslinking induces the shedding of L-selectin by sequestering L-selectin to a lipid raft membrane compartment where L-selectin is highly Tyr phosphorylated (Phong et al.,2003), but there is no evidence that Tyr phosphorylation of the L1 cytoplasmic domain is required for shedding. Consistent with these data, the cytoplasmic tail of L1 is not Tyr phosphorylated in PMA-induced shedding (Gutwein et al.,2000), and Tyr residues of the L1 cytoplasmic domain are dispensable in hepatocyte growth factor (HGF)-induced shedding (Heiz et al.,2004). Shedding of APP induced by pervanadate and PMA stimulation is blocked by PTK inhibitors, but the cytoplasmic domain of APP is not Tyr phosphorylated by pervanadate or PMA stimulation (Slack et al.,1995). Further, PMA and several other inducers of HB-EGF shedding induce Ser phosphorylation of the HB-EGF cytoplasmic domain, but mutation of the phosphorylated Ser residues has no effect on HB-EGF shedding (Wang et al.,2006). Collectively, these data indicate that PKCs and PTKs phosphorylate components other than the substrate in enhancing ectodomain shedding.
The cytoplasmic domain of substrates, however, can affect ectodomain shedding through interactions with intracellular modifiers. Calmodulin binds to the cytoplasmic domain of L-selectin and ACE in a constitutive manner, and its dissociation induced by calmodulin kinase inhibitors or PMA activates shedding, indicating that calmodulin is a negative regulator of L-selectin and ACE shedding (Kahn et al.,1998; Matala et al.,2001). In fact, calmodulin kinase inhibitors and calcium ionophores also induce the shedding of CD44, EGF, betacellulin, N-cadherin, and IL-6R (Nagano et al.,2004; Reiss et al.,2005; Sanderson et al.,2005). These data suggest that regulation of shedding by calmodulin and calcium is a common mechanism. The APP homolog APLP1 binds to APP and interferes with endocytosis of APP through a conserved NPTY motif in the cytoplasmic domain and increases APP shedding (Neumann et al.,2006). The carboxyl terminus Val in the cytoplasmic domain of TGFα is required for its shedding induced by various stimuli (Bosenberg et al.,1992), suggesting that structural features of the TGFα cytoplasmic tail is essential for its interaction with intracellular regulators of shedding. Available data also implicate actin-binding proteins of the ERM family in the regulation of shedding. Ezrin is bound to the cytoplasmic tail of L-selectin in both resting and PMA-stimulated lymphocytes, but only moesin binding to the L-selectin cytoplasmic tail is associated with the activation of shedding (Ivetic et al.,2002). Importantly, mutation of the Ezrin/Radixin/Moesin (ERM) binding site interferes with L-selectin shedding, suggesting that ERM interactions are critical (Ivetic et al.,2004). Furthermore, retention of substrates in intracellular compartments can negatively influence shedding at the cell surface. Santhamma et al. showed that the molecular chaperone, immunoglobulin binding protein (BiP), associates with ACE, and that BiP overexpression blocks ACE shedding by retaining ACE in the endoplasmic reticulum (ER) (Santhamma and Sen,2000).
In addition to PKC and PTK signaling, other signaling pathways involving ATP (Hubschmann et al.,2005; Sengstake et al.,2006), MAP kinases (Fan and Derynck,1999; Fitzgerald et al.,2000), and GPCRs (Prenzel et al.,1999; Fitzgerald et al.,2000) have been shown to influence ectodomain shedding. Transactivation of the EGF receptor through GPCR stimulation by lysophosphatidic acid, endothelin, thrombin, bombesin, or carbachol has been shown to enhance HB-EGF shedding (Prenzel et al.,1999), and GPCR activation by thrombin-derived peptides induces the shedding of syndecan-1 and -4 (Fitzgerald et al.,2000). Inhibitor studies have shown that MAP kinases are required for the shedding of (i) syndecan-1 and -4 by EGF, thrombin, ceramide, and osmotic stress (Fitzgerald et al.,2000); (ii) IL-6R by PMA (Thabard et al.,2001); (iii) TGFα, TNFα, and L-selectin by growth factors (Fan and Derynck,1999); and (iv) HB-EGF by PMA (Gechtman et al.,1999). In L-selectin shedding, nonsteroidal anti-inflammatory drugs increase the intracellular concentration of ATP, which in turn activates PKC and MAP kinase (Budagian et al.,2003; Sengstake et al.,2006). Altogether, these data suggest that extracellular agonists activate several distinct intracellular signaling pathways that converge to regulate a metalloproteinase-mediated cleavage mechanism at the cell surface.
The first clue to identify the sheddase was provided by the finding that peptide hydroxamate inhibitors of metalloproteinases block the shedding of membrane bound TNFα (Mohler et al.,1994). Subsequently, shedding of the majority of proteins was also found to be inhibited by peptide hydroxamates, and TNFα converting enzyme (TACE) was identified as the TNFα sheddase in 1997 (Black et al.,1997; Moss et al.,1997). TACE is a member of the disintegrin and metalloproteinase (ADAM) family (Schlondorff and Blobel,1999), and is also referred to as ADAM17. TACE/ADAM17 is also a sheddase for other surface molecules, such as TNFRs (Peschon et al.,1998), L-selectin (Peschon et al.,1998), vascular cell adhesion molecule-1 (VCAM-1; Garton et al.,2003), fractalkine (Garton et al.,2001), ErbB-4 (Rio et al.,2000), colony stimulating factor-1 (CSF-1) (Horiuchi et al.,2007b), APP (Buxbaum et al.,1998; Huovila et al.,2005), and the EGF family ligands TGFα (Peschon et al.,1998), HB-EGF (Sunnarborg et al.,2002; Jackson et al.,2003), amphiregulin (Sunnarborg et al.,2002; Sternlicht et al.,2005), and epiregulin (Sahin et al.,2004).
Although TACE/ADAM17 can shed many cell surface proteins, it is not the only sheddase and several proteins that are not shed by TACE/ADAM17 (e.g., ACE, TRANCE, syndecan-1) (Sadhukhan et al.,1999; Fitzgerald et al.,2000; Schlondorff et al.,2001). ADAM9 and ADAM12 shed HB-EGF (Izumi et al.,1998; Asakura et al.,2002), ADAM10 sheds E-cadherin (Maretzky et al.,2005), N-cadherin (Reiss et al.,2005), HB-EGF (Lemjabbar and Basbaum,2002), betacellulin (Horiuchi et al.,2007a), Notch (Pan and Rubin,1997), L1 (Mechtersheimer et al.,2001), APP (Huovila et al.,2005) and CXCL16 (Abel et al.,2004), and ADAM8, ADAM15, and ADAM28 shed CD23, a low affinity IgE receptor (Fourie et al.,2003).
Furthermore, metalloproteinases belonging to the family of matrix metalloproteinases (MMPs) can shed several protein ectodomains. MMP-2 sheds syndecan-2 (Fears et al.,2006), MMP-3 sheds Fas ligand (FasL) (Matsuno et al.,2001) and HB-EGF, MMP7 sheds TNFα (Gearing et al.,1994,1995; Haro et al.,2000), FasL (Powell et al.,1999), HB-EGF (Yu et al.,2002), E-cadherin (Noe et al.,2001; McGuire et al.,2003), β4 integrin (von Bredow et al.,1997), and syndecan-1 (Li et al.,2002), and MMP-9 sheds E-cadherin (Symowicz et al.,2007), ICAM-1 (Fiore et al.,2002), c-kit ligand (Heissig et al.,2002), and syndecan-1 and -4 (Brule et al.,2006). The transmembrane MMP, MMP-14, can mediate the shedding of TRANCE (Schlondorff et al.,2001), CD44, and syndecan-1. In addition, many bacterial pathogens, including Pseudomonas aeruginosa, E. coli, Bacteroides fragilis, Staphylococcus aureus, Streptococcus pyogenes (Group A Streptococcus, GAS), Listeria monocytogenes, Bacillus anthracis, and Streptococcus pneumoniae, express metalloproteinases that shed various inflammatory factors from the host cell surface (Vollmer et al.,1996; Lathem et al.,2003; Grys et al.,2005; Chung et al.,2006; Chen et al.,2007; Leduc et al.,2007; Wu et al.,2007). Thus, many metalloproteinases possess the capacity to function as sheddases. However, ectodomain shedding is likely specific in vivo because the expression of the substrate, sheddase, and extracellular and intracellular regulatory factors are tightly controlled in a spatial and temporal manner.
Examination of the cleavage site sequence of several substrates shed by TACE/ADAM17 indicates that the distance from the plasma membrane and structure of the cleavage site region are more important than the specific sequence in ectodomain shedding. Most mutations in the cleavage site or adjacent residues do not affect the shedding of TGFα (Wong et al.,1989), TNFα (Tang et al.,1996), APP (Sisodia,1992), L-selectin (Migaki et al.,1995), and TNFRI (Brakebusch et al.,1994) by TACE/ADAM17 as long as the overall distance is maintained. However, insertion of amino acids that disrupt the secondary structure (e.g., Pro, Gly) in the juxtamembrane region blocks shedding of L-selectin (Zhao et al.,2001c), TGFα (Brachmann et al.,1989; Wong et al.,1989), APP (Sisodia,1992), TNFRI (Brakebusch et al.,1994), and TNFRII (Herman and Chernajovsky,1998) by TACE/ADAM17. Of the two naturally occurring isoforms of ErbB-4, the isoform with 23 amino acids inserted in the juxtamembrane region is easily shed, but the other one with 13 amino acids is not shed efficiently (Elenius et al.,1997). Further, a mutant construct of APP with deletion of the cleavage site is still cleaved at a different -P1-P′ 1- sequence at exactly the same distance from the membrane as ™-type APP (Maruyama et al.,1991). However, a 14-amino acid sequence of the juxtamembrane domain of APP or TGFα is sufficient to confer shedding susceptibility to betaglycan (Arribas et al.,1997), whereas an 11 amino acid deletion within the juxtamembrane region of APP inhibited its shedding by α secretase (Sisodia et al.,1990). Thus, although the cleavage site is predominantly specified by the distance from the plasma membrane and structure of the cleavage site region, these latter findings suggest that there is some degree of sequence specificity in the shedding of certain substrates.
Trafficking, Modification, and Activation of Sheddases
ADAMs and MMPs are produced as zymogens, and proteolytic removal of the prodomain activates these metalloproteinases. ADAMs typically have a cleavage site in their prodomain for the serine endopeptidases, furin and furin-like proprotein convertase (PC), and several studies have shown that ADAM zymogens are activated through cleavage of the prodomain by furin-like PCs in the trans-Golgi/endosomal compartment (Lum et al.,1998; Loechel et al.,1999; Roghani et al.,1999). Although proteolytic processing by furins is generally constitutive and occurs in intracellular compartments, several studies have shown that furin can process proteins in a specific cellular compartment formed by the fusion of endocytic furin-containing compartments and substrate-containing compartments (Band et al.,2001; Chen et al.,2001). Further, furin can also function at the cell surface as exemplified by its capacity to activate bacterial toxins at the cell surface (Klimpel et al.,1992; Schiavo and van der Goot,2001). More importantly, shedding of several proteins, such as TNFα, Kit ligand-1 (KL-1), KL-2, CD44, and LOX-1, is inhibited by serine proteinase inhibitors (Scuderi,1989; Pandiella et al.,1992; Bazil and Strominger,1994; Robache-Gallea et al.,1995; Murase et al.,2000), and furin-deficient cells are impaired in PKC-dependent shedding of APP (Lopez-Perez et al.,2001). In addition, furin can activate several MMPs (e.g., MMP-14, MMP-15), suggesting that furins may also regulate ectodomain shedding through MMP activation. Alternatively, cells can activate sheddases through specific modifying reactions. Reactive oxygen species induced by PMA activate TACE/ADAM17 through an oxidative attack of the thiol group in the Cys switch, which disrupts the inhibitory effect of the prodomain on the enzymatic activity (Zhang et al.,2001).
Some studies have also shown that intracellular trafficking of sheddases influences ectodomain shedding. For example, PMA treatment induces the translocation of ADAM12 to the cell surface in a PKCϵ-dependent manner (Sundberg et al.,2004). Another study showed that in polarized epithelial cells, ADAM10 is sorted exclusively to the basolateral compartment and this depends on specific Pro residues in the Src homology 3 binding domain (SH3) in the cytoplasmic domain of ADAM10 (Wild-Bode et al.,2006). Interestingly, this SH3 motif is also required for ADAM10-mediated shedding of E-cadherin (Wild-Bode et al.,2006). In addition, ethyl-methane sulfonate-induced CHO mutant cells defective in TACE/ADAM17-dependent ectodomain shedding have impaired trafficking of TACE/ADAM17 from the ER, which interferes with the activation of TACE/ADAM17 by furins (Borroto et al.,2003).
Other means of posttranslationally regulating the expression of sheddases include sequestering the sheddase by internalization to a location where it is either isolated from or in close vicinity to its potential substrates (Jiang et al.,2001; Lafleur et al.,2006) and targeting the sheddase by polarized secretion to specific membrane domains (Galvez et al.,2004; Labrecque et al.,2004). For instance, in Madine Darby canine kidney (MDCK) epithelial cells, EGF is targeted to both the basal and apical membrane compartments, but EGF is predominantly expressed on the apical surface because its sheddase is targeted to the basal surface, and EGF is preferentially cleaved at this site to achieve the polarized expression at the apical surface (Dempsey et al.,1997).
Recruitment and Modification of Substrates
Cells also regulate shedding by targeting the substrate to specific cellular compartments. Cholesterol depletion from cells, which displaces substrates from lipid raft microdomains, induces the shedding of CD30 by TACE/ADAM17 and IL-6 receptor by ADAM10 and TACE/ADAM17 (Matthews et al.,2003; von Tresckow et al.,2004), and antibody-induced clustering induces shedding of CD30 (Hansen et al.,2000), CD44 (Shi et al.,2001), and L-selectin (Phong et al.,2003). Protein–protein interactions between the substrate and modifying proteins also influence shedding. Substrate interactions with signaling molecules, such as PKC, activate shedding by potentially inducing a conformational change in the substrate to facilitate the substrate-sheddase encounter. Binding of ARTS-1 (aminopeptidase regulator of TNFRI shedding) to TNFRI and IL-6R promotes shedding of these cytokine receptors (Cui et al.,2002,2003a). However, although ARTS-1 is an active enzyme, it has no sheddase activity, suggesting that it binds to the substrate and induces a conformational change. In addition, Nichols et al. recently reported that binding of Delta/Serrate/Lag-2 (DSL) ligands to heterodimeric Notch on an adjacent cell induces ADAM10-dependent shedding to activate the Notch signaling pathway. Endocytosis of Notch-DSL ligands by DSL ligand expressing cells provides a mechanical force to physically dissociate heterodimeric Notch to induce Notch shedding and signaling (Nichols et al.,2007). Altogether, these findings indicate that various molecular interactions, both inside and outside the cell, regulate ectodomain shedding to specify how and where the protein ectodomains function. However, precisely how cells sense the extracellular environment to regulate ectodomain shedding is not fully understood.