Syndecans as receptors and organizers of the extracellular matrix

REVIEW

Syndecans as receptors and organizersof the extracellular matrix

Xiaojie Xian & Sandeep Gopal & John R. Couchman

Received: 20 April 2009 /Accepted: 17 June 2009 /Published online: 14 July 2009# Springer-Verlag 2009

Abstract Syndecans are type I transmembrane proteinshaving a core protein modified with glycosaminoglycanchains, most commonly heparan sulphate. They are anancient group of molecules, present in invertebrates andvertebrates. Among the plethora of molecules that caninteract with heparan sulphate, the collagens and glyco-proteins of the extracellular matrix are prominent. Fre-quently, they do so in conjunction with other receptors,most notably the integrins. For this reason, they are oftenreferred to as “co-receptors”. However, just as withintegrins, syndecans can interact with actin-associatedproteins and signalling molecules, such as protein kinases.Some aspects of syndecan signalling are understood butmuch remains to be learned. The functions of syndecans inregulating cell adhesion and extracellular matrix assemblyare described here. Evidence from null mice suggests thatsyndecans have roles in postnatal tissue repair, inflamma-tion and tumour progression. Developmental deficits inlower vertebrates in which syndecans are eliminated arealso informative and suggest that, in mammals, redundancyis a key issue.

Keywords Proteoglycan . Heparan sulphate . Signalling .

Extracellular matrix assembly

Introduction

Syndecans are type I transmembrane cell surface heparansulphate proteoglycans (HSPGs). Four syndecan familymembers occur in vertebrates: syndecan-1, -2, -3 and -4.Most cell types, with the exception of erythrocytes, expressat least one syndecan family member and a few may evenexpress all four. However, syndecan-1 is expressed mostlyin epithelial cells, syndecan-2 is present in cells ofmesenchymal origin, syndecan-3 is primarily found inneuronal tissues but is more widely in development,whereas syndecan-4 exhibits a much broader distribution.Syndecans are now considered to have important rolesduring development, wound healing, inflammation andtumour progression. However, no developmental phenotypeis evident in either syndecan-1 or -4 null mice, althoughneuronal deficits have been found in the syndecan-3 nullmouse (Hienola et al. 2006). Roles in development for thesyndecans are more apparent in lower vertebrates (Muñozet al 2006; Matthews et al 2008; Kuriyama and Mayor2009; Olivares et al. 2009). Syndecans can bind a widearray of ligands via their heparan sulphate chains andevidence is accumulating that they have roles in cell-matrixinteractions and, in some cases, matrix assembly. Inaddition, all syndecans have been shown to interact withactin-associated proteins and, at least in the case ofsyndecan-4, evidence has been presented for signalling tothe cytoskeleton (Couchman 2003; Morgan et al. 2007).Since a functional actin cytoskeleton is a requirement forextracellular matrix (ECM) assembly, this supports a rolefor the syndecans. In contrast, little evidence is available

Xiaojie Xian and Sandeep Gopal contributed equally to this work.

The authors are supported by the Danish National ResearchFoundation, the Danish Medical Research Council, Vilhelm PedersenFonden, Haensch Fonden, Mizutani Foundation for Glycoscience,Grosserer Ernst Fischers mindelegat and the Department ofBiomedical Sciences at the University of Copenhagen. S.G. issupported by the Faculty of Health Sciences and the MolecularMechanisms of Disease PhD programme at the University ofCopenhagen.

X. Xian : S. Gopal : J. R. Couchman (*)Department of Biomedical Sciences, University of Copenhagen,Biocenter,Ole Maaløes Vej 5,2200 Copenhagen, Denmarke-mail: [email protected]

Cell Tissue Res (2010) 339:31–46DOI 10.1007/s00441-009-0829-3

that glypicans, the other major family of cell surfaceproteoglycans, have such roles. These are anchored to theouter leaflet of the cell membrane through a glycosylphos-phatidylinositol anchor and are not, therefore, transmem-brane. Evidence from both invertebrates and vertebratessuggests that glypicans have essential roles in growth factorand morphogen responses. The glypicans will not, there-fore, be considered further.

Structural organisation of syndecans

Syndecan core proteins range from 20-40 kDa and aremodified with glycosaminoglycan (GAG) chains. Heparansulphate is the principal GAG present in all foursyndecans, although syndecan-1 and -3 may also besubstituted with chondroitin sulphate. The GAG chainsare present on the syndecan ectodomains (Fig. 1), oftenclose to the N-terminus, in contrast to the glypicans. Apartfrom GAG attachment sites, the extracellular domains ofsyndecans are sequence-divergent. This initially gave riseto the idea that the ectodomains were nothing more thanprotein tags for GAG attachment, but now it is clear thatthe ectodomains of at least syndecans -1, -2 and -4 haveother functions in cell adhesion (Whiteford et al. 2007;Beauvais et al. 2009). The GAGs consist of repeatingdisaccharides of which from 50 to 150 disaccharides

might occur on each chain. The repeating disaccharidesare either N-acetylglucosamine and uronic acid in heparansulphates, or N-acetylgalactosamine and uronic acid inchondroitin/dermatan sulphates. The repeating disaccharidesare attached to the core protein, at a serine- glycinesequence lying within consensus regions rich in acidicresidues (Esko and Zhang, 1996), through a tetrasaccharidelinker of xylose-galactose-galactose-uronic acid (Fig. 1).Once the polymer is formed, a series of modifications takesplace including sulphation and epimerisation. Details ofheparan sulphate synthesis have been recently reviewed(Bishop et al. 2007; Nadanaka and Kitagawa, 2008). Briefly,some of the glucosamine residues are N-deacetylated andthen sulphated. Adjacent to this modification, some of theuronic acid residues are then epimerised to iduronic acid,which can then be 2-O-sulphated. Further hexosaminesulphations can occur at the 6-O and more rarely 3-Opositions. However, the key element of heparan sulphatesynthesis is that these polymer modifications do not goto completion. The end result is a chain with low(or unsulphated) regions interspersed with regions ofhigh sulphation. Often, small zones of intermediatesulphation lie at the boundaries between the two regions(Gallagher 2001; Murphy et al. 2004). Frequently, thelow-sulphated and high-sulphated regions are around10-12 saccharides in length and are characteristic of eachcell type. This contrasts to heparin, where high sulphation

Fig. 1 a Syndecan structure. Syndecan ectodomains are substitutedwith heparan sulphate (HS) chains and, in the case of syndecans-1and -3, can also bear chondroitin/dermatan sulphate (CS) chains. Asingle transmembrane domain (TD) and a cytoplasmic domaincomposed of two highly conserved regions (C1, C2) and a variableregion (V) are present. b Fine structure of heparan sulphate. Only afew saccharides are shown but a mature chain can contain more than

50 disaccharides. The initial replacement of N-acetyl groups withN-sulphate on glucosamine residues is followed by uronic acidepimerisation and 2-O-sulphation, followed by 6-O and (to a muchless extent) 3-O sulphation of glucosamine. In heparan sulphate, thesemodifications do not go to completion, the final structure havingdomains of higher sulphation interspersed with regions of lower or nosulphation. Generally, the sulphated domains interact with ligands

32 Cell Tissue Res (2010) 339:31–46

is nearly continuous. Therefore, in broad terms, the finestructure of heparan sulphate is controlled by cells.Furthermore, the fine structure requirements of heparansulphate for ligand interactions vary with the ligand(Kreuger et al. 2006). The extreme case is a specificpentasaccharide required for heparin interaction withantithrombin III. For fibronectin interactions, on the otherhand, N-sulphation is required, whereas 2-O-sulphation isnot (Mahalingam et al, 2007). In many cases, the overallstructure of heparan sulphate domains is of importance,rather than fine structure (Kreuger et al. 2006).

In contrast to the ectodomain, the transmembrane region ofsyndecans is highly conserved between the four members ofthe syndecan family. It contains a GXXXGmotif that stronglypromotes dimerisation that is substantially SDS-resistant insolution (Choi et al. 2005; Dews and Mackenzie 2007).Multimerisation increases the valency of syndecans, therebyenhancing their interaction with the ECM ligands and mayregulate signalling through the cytoplasmic domain. Thesmall cytoplasmic domains of the syndecans are composedof two highly conserved regions (C1 and C2) and a variableregion (V), which is unique to each syndecan but showsacross-species sequence conservation. The C1 and C2regions are proximal and distal to the transmembrane domainrespectively, separated by the central V region (Fig. 1;Couchman et al. 2001; Couchman 2003). Current evidencesuggests that the C1 region can interact with ezrin-radixin-moesin proteins, which are cortical-membrane- and actin-associated (Granés et al. 2003). The functional role for thisinteraction remains, however, unclear. In contrast, wellcharacterised interactions occur between the C2 region anda number of PDZ-domain-containing proteins. Examples aresyntenin, synectin (GIPC), synbindin and CASK (Multhauptet al. 2009). Evidence is accumulating that these interactionsregulate the trafficking of syndecans to, and perhaps from,the membrane (Zimmermann et al. 2005). Since some PDZproteins can form submembranous networks, they might alsopromote the clustering of syndecans into functional entities.As the V regions are specific to each syndecan, butconserved across species, they are assumed to endow thesyndecan with interactional and functional specificity. How-ever, information regarding most syndecans is lacking, witha significant amount of data concerning interaction partnersand signalling capacity only being available for syndecan-4.Certainly, the syndecan cytoplasmic domains are too small tocontain intrinsic enzymatic activity and therefore requireinteractions to be able to signal.

Syndecans as receptors for ECM molecules

A wide array of extracellular ligands interact with heparinand with heparan sulphate, including polypeptide growth

factors, chemokines, morphogens, lipid-regulatingenzymes, ECM proteins, cell-cell adhesion molecules andblood-coagulation factors. Many ECM molecules have oneor more discrete domains that have heparin-bindingcharacteristics (Fig. 2) and, in general, these have beendemonstrated also to bind to the less sulphated heparansulphate. However, where examined, usually the highlysulphated domains (S domains) of the heparan sulphatehave the ligand-binding potential (Esko and Selleck 2002;Kreuger et al. 2006). A major, but unresolved, issue iswhether all heparan sulphates from a cell type, derived fromseveral distinct glypicans and syndecans, possess the abilityto interact with ECM heparin-binding domains. Certainly intissue culture, only some subtle differences have been notedbetween glypican and syndecan heparan sulphate in aparticular cell type (Tumova et al. 2000). Perhaps this isaccentuated in vivo, where, for example, syndecan andbasement membrane heparan sulphates might be synthes-ised in distinct pathways. Despite this, however, some ECMligands can trigger specific responses through syndecans.The situation is complicated by the finding that many ECMmolecules interact with receptors other than syndecans, theintegrins being among the most important. Again, with theexception of erythrocytes, one (but usually more than one)integrin type is present on all vertebrate cells. Syndecanshave often been described as co-receptors, since they canfrequently work alongside other receptors, such as high-affinity tyrosine kinase growth factor receptors (Alexopou-lou et al. 2007; Murakami et al. 2008). In the case of ECMmolecules, the syndecans can be co-receptors with integ-rins. Nevertheless, this should not be taken to imply thatthey only function to concentrate ligands at the cell surface(Lander 1998). Syndecans can act as signals, perhapsindependently, as indicated by various types of evidence.This is however a field in its infancy and, particularly withrespect to invertebrate syndecans, details on signalling arefew. The best understood example is syndecan-4 ofmammals, where elements of a signalling pathway havebeen elucidated.

Laminins

Laminins are heterotrimeric basement membrane andconnective tissue proteins comprised of three distinctsubunits (α, β and γ) forming cross- or rod-shapedmolecules (Fig. 2f). In mammals, the five α chains, four βchains and three γ chains can combine to give rise to 15laminin isoforms (Miner and Yurchenco 2004). Theirexpression is both spatially and temporally regulated. Severalstudies provide evidence that syndecans interact with lamininα chains through their heparan sulphate chains. The principalsite for interactions of laminins with cell surface receptors isthe large globular domain (LG domain) at the C-terminal of

Cell Tissue Res (2010) 339:31–46 33

the α subunits, which is divided into five homologoussubdomains (LG1-5). Whereas the LG1-3 domains interactpredominantly with integrins, the LG4-5 modules can bindheparin and sulphatides and to syndecan receptors(but perhaps not glypicans; Yamashita et al. 2004). This isa property of all five laminin α chains. Moreover, thereseems little distinction in the ability of the differentsyndecans to interact with the LG4-5 modules. Whereassyndecan-3 has not been scrutinised, syndecans-1, -2 and -4all interact with the LG4 module of the α1 and α3 chains(Okamoto et al. 2003; Matsuura et al. 2004; Hozumi et al.

2006). Indeed, this might be an ancient property of laminins,since evidence has been presented that Drosophila syndecancan interact with laminin LG4 modules and might be aregulator of haemocyte island formation (Narita et al. 2004).If so, this is a rare example of an invertebrate syndecanhaving a clear role in cell-matrix interactions.

The N-terminal regions that form the laminin short armsare mainly involved in the matrix assembly of laminins inbasement membranes. Laminin-111 (α1β1γ1) from theEngelbreth-Holm-Swarm tumour was the first isoformisolated; it is highly enriched in this tumour but is also

Fig. 2 a–f Models ofselected extracellular matrixglycoproteins, includingvitronectin, fibronectin,fibrillin-1, tenascin-C andthrombospondin 1, showingoverall structures and thelocation of major heparin-binding sites (Hep heparin, EGFepidermal growth factor, TBtransforming growth factorβ-binding, TA tenascinassembly, TSPthrombospondins, LG largeglobular domain, LN lamininN-terminal region)

34 Cell Tissue Res (2010) 339:31–46

expressed in the blastocyst, neuroectodermally derivedtissues and developing kidney of the early embryo (Minerand Yurchenco 2004). Peptide AG73 (RKRLQVQLSIRT)from the α1 chain LG4 globule is active in cell attachmentand is a major binding site for syndecans. Culture of humansubmandibular salivary gland cells on laminin-1 inducesacinar differentiation, whereas the addition of peptide AG73from the α1 chain or the homologous peptide MG73(KNRLTIELEVRT) from the α2 chain as competitorsdecreases acinar size (Hoffman et al 1998). From the samestudy, syndecan-1 has been proposed as a key receptor forlaminin-111 and laminin-211. Peptides corresponding toAG73 and homologous regions of the other four α chainshave been tested as competitors in mandibular gland culturesembedded in matrigel (which is rich in laminin-111). Onlythe α1-chain-derived peptide blocks branching morphogen-esis; the same principle has been demonstrated for networkformation by B16F10 cells in matrigel (Hoffman et al. 2001;

Suzuki et al. 2003). Whereas this might argue for specificity,A4G73 and A5G73 from the α4 and α5 chains, respectively,have a lower affinity for heparin (Hoffman et al. 2001).Moreover, the situation is complicated by findings thatdystroglycan binding to the α1LG4 module is required forsalivary gland branching morphogenesis (Durbeej et al.2001). Dystroglycan interacts with the LG4 regions of theα1, α2 and α5 chains (Suzuki et al. 2005) and the bindingsites for heparin and dystroglycan overlap in the α1LG4module (Andac et al. 1999). Further complexity has beenshown by Yokoyama et al. (2005). Fibroblasts spread wellon the α1LG4 module but poorly on the homologousmodules of α2, α3 and α4 chains. Cells on the α1LG4module even form focal adhesions and microfilamentbundles but this is, at least in part, a result of α2β1 integrininvolvement that may not pertain to attachment on other αchain LG4 modules (Yokoyama et al. 2005; Hozumi et al.2006).

Fig. 2 (continued)

Cell Tissue Res (2010) 339:31–46 35

Laminin-332 (α3β3γ2; laminin-5) is a major lamininisoform in the dermal-epidermal junction basement mem-brane of the skin (Sugawara et al. 2008). The LG4-5 moduleof the α3 chain interacts with syndecans, whereas the LG1-3domain binds α3β1, α6β1 and α6β4 integrins to supportseveral biological activities of epithelial basement mem-branes (Sugawara et al. 2008). Syndecans-1, -2 and -4 havebeen demonstrated as α3LG4-5 receptors, with syndecans-2and -4 (but not glypican-1) having been reported to interactwith a recombinantly expressed α3LG4 module to supportcell adhesion (Utani et al. 2001) and to induce the expressionof matrix metalloproteinase-1 (MMP-1) through themitogen-activated protein kinase signalling pathway (Utaniet al. 2003). Interactions between syndecan-1 and arecombinant LG4-5 fragment of the α3 chain have beenshown to participate in keratinocyte adhesion and spreading(Okamoto et al. 2003).

Upon secretion, human laminin-332 undergoes proteolyticprocessing from a precursor form consisting of a 190-kDaα3 chain, a 135-kDa β3 chain and a 150-kDa γ2 chain intoa mature form of a 160-kDa α3 chain and a 105-kDa γ2chain. The 150- kDa γ2 chain of laminin-332 is processed tothe 105-kDa form in many laminin-332-producing humancell lines and this can be mediated by MT-MMP1 (MMP14).However, other proteases such as tolloid and bone morpho-genetic protein 1 have been suggested as being morephysiologically relevant (Sugawara et al. 2008). The pro-cessed γ2 chain (γ2sa) is known to convert this lamininfrom being cell-adhesion-promoting to motility-promoting.The active site of γ2sa has been localised to the NH2-terminal epidermal growth factor (EGF)-like sequence(domain V or LEa). Syndecan-1 has been identified as amembrane receptor of γ2sa in EJ-1 cells, but not syndecan-2(Ogawa et al. 2007a, b).

Cleavage of the α3 chain occurs between the LG3 andLG4 modules, thereby releasing the heparin (and syndecan)-binding region. Bachy et al. (2008) have shown that thepresence of LG4/5 in unprocessed laminin-332 reducesintegrin-mediated adhesion and that keratinocyte migrationon laminin is syndecan-1-dependent. Keratinocytes assemblefascin-containing microspikes or protrusions throughsyndecan-1 signalling, reminiscent of a similar process onthrombospondin-1, also mediated by syndecan-1 signalling(Chakravati et al. 2005). In vivo cleavage of the α3 lamininchain may cause the LG4/5 module to be lost or it might beretained. Either way, laminin-332 processing is likely toinfluence cell adhesion and behaviour, an area for furtherexamination.

Collagen

Collagens are major insoluble fibrous and matrix proteinsand are ubiquitously expressed in all vertebrates. Of the at

least 24 collagen types, type I is the most abundant proteinin the vertebrate body. Collagen fibrils contribute to thetensile strength, integrity and function of many tissues inpart resulting from interactions with many other collagens,glycoproteins, proteoglycans and growth and differentiationfactors. In addition, collagens interact with cell surfacereceptors, most notably the I-domain containing integrins.Many collagens, both fibrillar and non- fibrillar, interactwith heparin and heparan sulphate. The site in type Icollagen has been mapped (Sweeney et al. 2008) and isdistinct from other integrin and proteoglycan-binding sites.In the triple helical fibril, the heparin-binding region liesbetween the “cell interaction domain” and the mineralisa-tion/gap zone (Sweeney et al. 2008). Type XVIII collagenis a non-fibrillar basement membrane collagen whosemonomeric C-terminal domain, known as endostatin, hasreceived great attention because of its potential to blockangiogenesis. This cryptic fragment is released by proteasessuch as MMPs and elastin. However, in its trimeric form,the NC1 domain is motogenic for endothelial cells, afunction that is inhibited through interactions with heparinand cell surface heparan sulphate (Clamp et al. 2006).

In early work with NMuMG mouse mammary epithelialcells, the ectodomain of cell surface HSPG (probablysyndecan-1)-bound collagen types I, III and V but nottypes II, IVor denatured type I. The proteoglycan interactedwith a heparin-binding site on collagen, since addition ofheparin blocked this interaction (Koda et al. 1985). Furtherevidence confirmed syndecan as a cell surface receptor forcollagen. In B lymphoid (MPC-11) cells, syndecan-1 boundto type I collagen in a dose-dependent manner. This bindingwas mediated by heparan sulphate chains on syndecansince pre-treatment of collagen with heparin or removal ofheparan sulphate from the cell surface before incubation ofcells with collagen inhibited cell binding. Moreover, thegrowth of cells in the presence of chlorate, a competitiveinhibitor of sulphation, also ablated cell adhesion tocollagen, again suggesting a heparan sulphate chainrequirement for the interaction of syndecan with collagen(Sanderson et al. 1992). A recent study showed thatsyndecan-1 supported α2β1 integrin-mediated adhesion tocollagen in MDA-MB-231 cells. This adhesion wasreduced by transfection of syndecan-1 short interferingRNA (siRNA). Using over-expression and knock-downexperiments, the enhanced binding of integrin α2β1 tocollagen was only observed with syndecan-1, but notsyndecan-2 or -4 (Vuoriluoto et al. 2008).

Fibronectin

Fibronectin is a body fluid and ECM glycoprotein that isessential not only for embryonic development, but also fortissue repair. It can readily be incorporated into transient

36 Cell Tissue Res (2010) 339:31–46

matrices formed from plasma clots together with fibrin(ogen). The soluble plasma forms are synthesised in the liverbut many cell types, particularly in cell culture, alsosynthesise variants of this glycoprotein, particularly formsthat readily polymerise (Mao and Schwarzbauer 2005; Leisset al, 2008). Its expression tends to be high in embryonic orrepairing tissue but decreases with maturity. Despite thesingle gene product, multiple isoforms are generated throughmRNA alternate splicing (Pankov and Yamada 2002). Threemajor heparin-binding sites have been found in fibronectin(Fig. 2b), one of low affinity at the N-terminus (Hep I), onelocated in the alternately spliced V region, whereas the bestcharacterised is the more C-terminal Hep II (Barkalow andSchwarzbauer 1991). The more cryptic Hep III domain(repeats III4-5) can block fibronectin fibrillogenesis(Maqueda et al. 2007). This region may also bind chondroitinsulphate proteoglycans that co-operate with integrin α4β1in cell adhesion. As has been recognised for a long time,heparin binding to fibronectin causes conformationalchanges (especially in Hep I) that may expose otherbinding sites (Khan et al. 1988). The Hep II domainconsists of type III repeats 12-14 and several studies haveshown that type III13 contains the major heparin-bindingsite. Once mutated, the entire Hep II domain losesconsiderable affinity for heparin. The III14 module binds toheparin with much lower affinity (Barkalow and Schwarz-bauer 1991; Ingham et al. 1993; Yoneda et al. 1995) but thecrystal structure of the Hep II domain suggests that theheparin-binding motifs of III13 and III14 align to form acontiguous heparin- or heparan-sulphate-binding site(Sharma et al. 1999). In turn, this explains why relativelylong heparin oligosaccharides are required for high-affinitybinding (Ingham et al 1990). A comparison of glypican andsyndecan has shown that all HS chains have comparableaffinity in binding the Hep II domain of fibronectin (Tumovaet al. 2000). Cell attachment studies with embryonicfibroblasts have demonstrated the absolute requirement forheparan sulphate N-sulphation, a result that mirrors in vitrobiochemical analysis of heparin-Hep II binding (Lyon et al.2000; Mahalingam et al. 2007). Iduronic acid sulphation isnot required but Hep-II-promoted focal adhesion assemblymay also require 6-O-sulphation of hexosamine residues(Mahalingam et al. 2007). In addition to the Hep II domain,fibronectin also possess a RGD sequence in III10 that canserve as a ligand for many integrins. As discussed below, thisenables complex combinatorial signalling through integrinand syndecan that can affect cell adhesion, migration andmatrix assembly.

Tenascin-C

Tenascin-C is an adhesion modulatory matricellular glyco-protein that can interact directly with syndecans (Fig. 2d).

Its expression is strongly regulated in developing orrepairing tissues and can be markedly upregulated in thestroma of several types of carcinoma (Tucker and Chiquet-Ehrismann 2009). It is one of four mammalian tenascins,although the others have received little attention withrespect to interactions with HSPGs. Tenascin-C is adisulphide-linked hexamer with the six polypeptide chainslinked at their amino termini via a tenascin assemblydomain (Chiquet-Ehrismann 2004). Each chain is com-posed of multiple EGF-like repeats, a series of fibronectintype III (FN-III) repeats and a terminal globular domain thatresembles the carboxyl-terminal portion of the β and γchains of fibrinogen (Orend and Chiquet-Ehrismann 2006).Two heparin-binding sites have been noted, one in aconstitutively expressed type III repeat, the other in the C-terminal globular domain. Syndecan-1 was shown to bindtenascin-C of embryonic tooth mesenchyme via its heparansulphate chains in solid phase assay (Salmivirta et al. 1991).In more detailed studies of this glycoprotein with fibroblasts,fibronectin-fibrinogen matrix contraction was reduced con-siderably in the presence of tenascin-C compared to itsabsence (Midwood et al. 2004). Fibroblast heparan sulphatechains on syndecan-4 bound to the tenascin-C, therebyinhibiting fibronectin-syndecan-4 interactions. Additionally,tenascin-C bound the Hep II domain of fibronectin, in turnleading to a block in the formation of syndecan-4-promotedfocal adhesions and stress fibers and reducing matrixcontraction. Consistent with its distribution at the edges ofthe wounds, tenascin-C may promote migration rather thanpremature contraction (Midwood et al. 2004). A similarmechanism of adhesion modulation has been proposed forthe unrelated protein, fibulin-1, suggesting that this may be acommon mechanism to fine-tune adhesion (Williams andSchwarzbauer 2009). Other research, again in fibroblasts, hasshown that the blocking of fibronectin-syndecan-4 interac-tion by tenascin-C inhibits cell anchorage on fibronectin,leading to the blockage of cell cycle progression but thestimulation of tumour cell proliferation (Orend et al 2003;,Huang et al. 2001). An HSPG-binding site on tenascin-C hasbeen identified by photoaffinity labelling of WI38VA13cells. TNIIIA2, a synthetic 22-mer peptide corresponding tothis site in tenascin-C, has been demonstrated to interact withsyndecan-4 by affinity chromatography by using TNIIIA2-immobilised beads. In this study, siRNA-based syndecan-4knock-down reduced TNIIIA2-induced β1 integrin activa-tion and consequent cell adhesion to fibronectin. On theother hand, over-expression of syndecan-4 core proteinenhances TNIIIA2-induced activation of β1 integrin(Saito et al. 2007). In total, these studies demonstratethe apparent complexity of tenascin-C/syndecan/integrininteractions but are consistent with data suggesting thatsyndecan core proteins can influence integrin- mediateadhesion (see below).

Cell Tissue Res (2010) 339:31–46 37

Thrombospondin

The thrombospondins are a family of multifunctional proteins.The five members can be divided into two subgroups: A,which contains homotrimeric thrombospondin-1 and -2, andB, which contains homopentameric thrombospondin-3, -4 and-5. Thrombospondin-1 has been implicated in cancer celladhesion, migration and invasion, the inhibition of angio-genesis and the activation of latent transforming growthfactor-β (TGF-β; Isenberg et al. 2008; Kazerounian et al.2008). A heparin-binding domain has been well charac-terised in the N-terminal domain of thrombospondin-1(Fig. 2e) and, since this region is immediately upstream ofa coiled-coil oligomerisation domain (Carlson et al. 2008),there is presumably a cluster of three closely spacedheparin-binding sites in the trimeric protein. TheN-terminal domain also interacts with a variety of othercell surface receptors, including integrins.

Syndecan-1 is required for the organisation of fascinspikes in response to thrombospondin-1-mediated celladhesion, and the over-expression of syndecan-1 in aheterologous cell type is sufficient to cause dramaticenhancement of cell spreading and the formation of fascinspikes in response to thrombospondin-1 (Adams et al.2001). In a second study, an 18-kDa recombinant proteinencompassing the N-terminal residues 1-174 of humanthrombospondin-1 (thrombospondin-18) induced tubulo-genesis by human umbilical vein endothelial cells (Nuneset al. 2008). Here, syndecan-4 was the candidate, since amonoclonal antibody against the core protein N-terminusblocked thrombospondin-18-induced tubulogenesis. TwoN-terminal motifs, thrombospondin Hep I and II, wereidentified that compete with the fibronectin Hep II domainfor binding to syndecan-4 on the endothelial cell surface,indicating that they exert their effects by interfering withthe recognition of fibronectin by syndecan-4 (Nunes et al.2008). This is one of a number of recent examples in whichan antibody against the core protein blocks a functionapparently mediated by heparan sulphate chains. Howmight this happen? The most obvious possibility is thatthe antibody clusters on the syndecan, which is thenendocytosed or cleaved from the surface. Such a possibilityhas not been examined but might shed light on therelationship between the core protein and GAG.

Vitronectin

Vitronectin is an abundant plasma glycoprotein that canalso be present in ECM. It strongly promotes cell adhesionand spreading through αV integrins. Vitronectin exists as asingle chain form but can multimerise, a process thought tobe aided by protein unfolding or denaturation. There hasbeen some controversy over the ability of native vitronectin

to bind heparin. Some reports suggest that the heparin-binding domain of vitronectin is cryptic in the solubleplasma form and can be revealed by mild denaturingconditions or perhaps cleavage. However, other evidencesuggests that soluble plasma vitronectin can bind heparin,although the extent of binding increases with milddenaturation or multimerisation (Underwood et al. 2002).Certainly, heparin appears capable of inducing conforma-tional changes. A 12-amino-acid basic motif within theheparin-binding domain has been identified that interactswith syndecans-1, -2 and -4 (Wilkins-Port et al. 2003). Thisinteraction is able to support adhesion to the vitronectinheparin-binding domain, a process that is not supported byglypican-1. However, the physiological relevance of thisparticular syndecan function may be endocytosis and theclearance of vitronectin.

Fibrillin-1

Fibrillin is a large multidomain glycoprotein and a majorconstituent of tissue microfibrils (Fig. 2c). These fibrils canserve as a template for elastin deposition and the formationof elastic fibers. The molecule is dominated by 47 EGFrepeats, of which 43 are calcium-binding consensussequences (cbEGF domains). Interspersed among thesedomains are TB modules that are homologous to TGF-β-binding 8-cysteine-containing motifs found in latent TGF-β-binding proteins. Like many other cbEGF domains, thoseof fibrillin form an extended rod-like conformation in thepresence of calcium (Cain et al. 2008a, b). So far, twoisoforms of fibrillin have been identified in the human,fibrillins 1 and 2, although they are structurally related tothe large TGF-β-binding proteins. Fibrillin-1 is an abun-dant glycoprotein and has been extensively studied, sincemutations are associated with Marfan syndrome (Kielty etal. 2002). Tiedemann et al. (2001) were the first tocharacterise the heparin-binding sites in fibrillin-1. Twocalcium-independent heparin-binding sites were identifiedat the N- and C-termini (Arg45–Thr450 and Asp1528–Arg2731) with one calcium-dependent central binding site(Asp1028–Thr1486; Tiedemann et al. 2001). Small specificheparin-binding motifs were not identified in this study.Ritty et al. (2003) narrowed down the C-terminal-bindingsite of fibrillin-1 to the last 17 residues of the mature proteinand also identified heparin-binding sites in fibrillin-2. Thefibrillin-1 site mediated adhesion of several cell typesthrough cell surface HSPGs. Later, four high-affinityheparin-binding sites in fibrillin-1 were identified by in vitrobinding assays in which the N-terminal site had particularlyrapid kinetics in binding to heparin (Cain et al. 2005). In thisstudy, the heparin-binding sites were more precisely local-ised. Hyaluronan and chondroitin sulphate did not interactsignificantly with fibrillin-1, indicating the specificity of

38 Cell Tissue Res (2010) 339:31–46

GAG interaction. Kinetic studies showed that the affinity ofthe fibrillin-1 heparin-binding fragments increased in parallelwith heparin saccharide chain length (Cain et al. 2005).

Fibrillin-1 has a single Arg-Gly-Asp (RGD) cell adhesionmotif within the fourth TB repeat (TB4). Integrin α5β1 andαvβ3 has been demonstrated to bind to fibrillin at this motifand further data suggest that α5β1 integrin is the majorfibrillin-1 receptor (Bax et al. 2003, 2007). The adjacentheparin-binding site in TB5 has also been demonstrated tosupport cell adhesion in mouse and rat embryonic fibro-blasts, since engagement of this site by cell surface HSPGenhances focal adhesion and stress fiber formation. Site-directed mutagenesis has identified two arginine residues thatare crucial for heparin binding (Bax et al. 2007). Syndecan-4null mouse fibroblasts are defective in focal adhesionformation on fibrillin-1 fragments that encompass the TB4and TB5 modules, indicating the similarity of responses tointegrin and heparin-binding sites to those seen in fibronectin(Bax et al. 2007). It is intriguing that, in both these matrixglycoproteins, an RGD integrin-binding site lies just up-stream of a heparin-binding site that can engage syndecan-4and promote focal adhesion assembly, even though themotifs involved are of quite a different structure.

To characterise the N- and C-terminal high-ffinityheparin-binding sites further, a series of recombinantfibrillin-1 fragments were used in surface plasmon reso-nance spectroscopy studies. The N-terminal cbEGF domainencoded by exon 7 and the three C-terminal fibrillin-1cbEGFs encoded by exons 59–62 were identified. Bothsites can induce focal adhesion formation and cytoskeletonorganisation in fibroblasts. In this case, syndecan-4 nullfibroblasts can form some adhesion plaques when plated onN-terminal heparin-binding fragments, indicating that morethan one cell surface HSPG can support cell adhesion (Cainet al. 2008a, b). The N- and C-terminal heparin-bindingmotifs might be more involved with fibrillin assembly intomicrofibrils and elastic fibres. Heparin and MAGP com-peted for binding to the N-terminal region, whereas heparinand elastin competed for C-terminal interactions (Cain et al.2005). Interestingly, recent evidence indicates that fibrillinassembly into microfibrils in tissue culture requires afibronectin matrix as a template (Kinsey et al. 2008).

Syndecans as organisers of ECM

As transmembrane proteoglycans, with known cytoskele-tal interactions in their cytoplasmic domains, the capacityof syndecans to regulate the ECM is suggested, althoughthe way that ligand interactions with heparan sulphatechains can transmit the appropriate signals through thecore proteins is not so clear. As with other receptors,syndecans can probably cluster in response to a ligand,

and this has been shown in some cases, although again themolecular basis is unclear. Additionally, more recentdevelopments now show the importance of the ectodo-main of the core protein. Evidence is fast accumulatingthat they may influence integrins, directly or indirectly,with further possibilities for affecting ECM assembly orturnover. Below, specific syndecan-mediated interactionswith ECM are considered, but with the omission ofsyndecan-3, for which interactions appear to be limitedto growth factors.

Syndecan-1

Syndecan-1 is expressed predominantly in epithelial cellsand is thought to play an important role in maintainingepithelial phenotype and morphology. Syndecan-1 is down-regulated in several types of carcinoma (Beauvais andRapraeger 2004) and loss of cell surface syndecan-1induces cells to acquire a fibroblast-like phenotype (Katoet al. 1995). This suggests a potential role in epithelial-mesenchymal transition.

On the other hand, syndecan-1 is recorded as beingupregulated in several malignant tumours such as pancreaticcancer. Thus, the role of syndecans in tumour developmentmay vary with tumour stage and type. Syndecan-1 cancooperate with several integrins, including αvβ3, αvβ5,α2β1, α3β1 and α6β4, to regulate adhesion-complexformation, cytoskeletal organisation and cell spreading anddirectional migration (Beauvais and Rapraeger 2003). Afunctional coupling between the αvβ3 integrin andsyndecan-1 has been revealed in a series of studies inMDA-MB-231 human breast cancer cells. This cooperationregulates the activation and signalling of the integrin duringcarcinoma cell spreading and migration on vitronectin. Theepithelial cells require syndecan-1 in order to activate αVβ3,and Rapraeger’s group have shown that the two moleculesinteract directly (Beauvais et al. 2009). The active site insyndecan-1 for integrin activation has been mapped to aregion of the extracellular core protein (amino acids 88-121);the GAG chains are apparently not involved. This peptide,now known as synstatin, can compete with syndecan-1 forintegrin activation and therefore is an inhibitor (Beauvais et al.2009). Moreover, in tumour models, synstatin can reducetumour mass and, in particular, seems to block angiogenesis.This has been confirmed in other angiogenesis assays inwhich the synstatin peptide can inhibit a process that isknown to be αVβ3 and αVβ5 integrin-mediated. However,the syndecan-1 null mouse develops normally, so thatembryonic angiogenesis cannot be absolutely dependentupon syndecan-1. Interestingly, in zebrafish in whichsyndecan-1 is absent (as in perhaps all bony fish),syndecan-2 has been reported as being required for angio-genesis in the embryo (Chen et al. 2004).

Cell Tissue Res (2010) 339:31–46 39

Other recent studies have connected syndecan-1 to MMPregulation. Adenoid cystic carcinoma (CAC2) and myoepi-thelioma (M1) cells grown inside three-dimensional laminin-111 matrices enriched with the laminin α1 AG73 peptideexhibit large spaces in the matrix. Increased MMP9 secretionhas been identified, and siRNA knockdown of syndecan-1and β1 integrin has shown that they are located downstreamof AG73 during the regulation of adhesion and MMPproduction by the tumour cells (Gama-de-Souza et al. 2008).Since MMP-9 is an important regulator of tumour angio-genesis and tumour ECM remodelling, this study has alsoprovided evidence for a role of syndecan-1 in the regulationof MMP-9 activity and tumorigenesis. Integrin α3β1 hasbeen shown to work closely with syndecan-1 to bindprecursor laminin-322 in keratinocytes. Their collaborationinduces the formation of fascin-containing protrusions; thisprocess is dependent upon activation of GTPase Rac andcdc42 (Bachy et al. 2008). Syndecan-1 is required formigration on pre-laminin-332, consistent with the reducedmigration of null keratinocytes in the knockout mouse.However, the knockout biology is complex. Null keratino-cytes appear to be defective in matrix assembly but migratenormally when provided with exogenous laminin-332matrices (Stepp et al. 2007). Consistent with data above,MMP regulation has been speculated to be disrupted in theabsence of syndecan-1. In addition, data from keratinocytesalso argue for an association of αV integrins with syndecan-1 (Stepp et al. 2007). Another connection to MMPs has beenseen, again with MD-MB-231 cells, but on collagen I, ratherthan vitronectin substrates (Vuoriluoto et al. 2008). In over-expression and knock-down experiments, syndecan-1, butnot syndecan-2 or -4, has been demonstrated to enhanceα2β1 integrin-mediated adhesion. Moreover, syndecan-1co-localises with α2β1 integrin and regulates actin organi-sation on collagen. In these experiments, heparan sulphatechains of HSPGs are required. Crosstalk between syndecan-1and α2β1 integrin induces the transcription of downstreamtarget MMP-1 in response to collagen binding (Vuoriluoto etal. 2008).

In contrast to keratinocytes, syndecan-1 null primarydermal fibroblasts migrate faster than wild-type cells andexpress significantly more αV and β1 integrin. Treatmentwith TGFβ1 increases both integrin expression and migra-tion differences between wild-type and null fibroblasts.When cells are replated on vitronectin and fibronectin-coated surfaces, syndecan-1 null cells show altered cellspreading and focal adhesion formation. Syndecan-1 nullcells all exhibit increased surface expression of α2β1 andα3β1. Here again, syndecan-1 is proposed to activate αV-containing integrins (Jurjus et al. 2008).

In total, syndecan-1 works closely with specific integrinreceptors in response to the ECM environment. Thiscooperation plays a major role in the regulation of matrix

remodelling, cell adhesion, migration and cytoskeletonrearrangement. The clearest evidence for synergisticcollaboration concerns the αV integrins, with the potentialto control signalling pathways that regulate MMPs, forexample. Determination of the molecular basis for syndecan-1 interaction with, and the regulation of, αV integrins shouldbe of interest. Such regulation seems important in postnatallife, as is often the case with mammalian syndecans. Perhapsredundancy in the mammalian embryo obscures the roles ofsyndecans in development and, hence, the value of lowervertebrate models for syndecan work is apparent.

Syndecan-2

Syndecan-2 was originally identified in lung fibroblasts and ishighly expressed in mesenchymal tissue surrounding bloodvessels, in organs such as the kidney, lung, liver and stomach,and in cells that form cartilage and bone (Zimmermann andDavid 1999; Essner et al. 2006). In zebrafish, syndecan-2 hasbeen implicated in developmental vascular angiogenesis(Chen et al. 2004). Neuronal tissue has also been shown asa source of syndecan-2, with possible roles in dendritic spinemorphogenesis (Ethell et al. 2001). Syndecan-2 expressionhas further been identified in several cancer types, includingLewis lung cancer and colon cancer (Park et al. 2002).Cooperation between syndecan-2 and integrin α5β1 in celladhesion to fibronectin and actin cytoskeletal organisationhas been demonstrated in strains of Lewis lung carcinoma-derived P29 cells (Munesue et al. 2002). High expressers ofsyndecan-2 form focal adhesions and microfilament bundles,whereas low expressers, e.g. LM66-H11 cells, only formcortical actin. Metastatic potential is inversely proportional tosyndecan-2 levels and therefore to actin filament organisa-tion. Enhanced cytoskeletal organisation can be promoted bythe ectopic expression of syndecan-2, with a correspondingreduction in metastatic potential. The result suggests thatsyndecan-2 regulates actin cytoskeleton organisation in anexpression-level-dependent manner (Munesue et al. 2002).However, this is an isolated example, since of the foursyndecans, syndecan-4 has been more frequently implicatedin focal adhesion assembly. Later work from Munesue et al.(2007) has shown that syndecan-2 is a possible suppressor ofMMP-2 activity in HT1080 cells. Indeed, MMP-2 has beenrevealed as a heparan-sulphate-binding enzyme. This raisesthe possibility that MMPs can regulate adhesion and matrixassembly in a proteoglycan-dependent manner.

Syndecan-2 can be pivotal in matrix assembly, as shown bytwo studies. Chinese hamster ovary cells have been stablytransfected with full-length (S2) or truncated syndecan-2lacking the C-terminal 14 amino acids of the cytoplasmicdomain (S2∆S). Cells expressing S2∆S cannot assemblelaminin or fibronectin into a fibrillar matrix. The matrixassembly defect is not attributable to reduced cell surface

40 Cell Tissue Res (2010) 339:31–46

Fig. 3 Elements of integrinand syndecan-4 signalling(PKCα protein kinase Cα).Integrins signal through tyrosinekinases and also interact withcytoskeletal elements such astalin and kindlins (not shown).Furthermore, regulation ofp190RhoGAP is consistent withprotrusive rather than contractilebehaviour. Syndecans may alsocontribute to this but, at othertimes, signal through a RhoA/Rho kinase (ROCK I especially)pathway. This leads to myosinlight chain and myosinphosphatase phosphorylation,microfilament contraction andfocal adhesion assembly

Fig. 4 Less-structured actin cytoskeleton of syndecan-4 null cells.Wild-type mouse fibroblasts (a-c) and corresponding syndecan-4 nullcells from knock-out mice (d-f) stained for paxillin (green, a, d) andF-actin (red, b, e) with overlays (c, f). Wild-type cells have abundant

focal adhesions (arrows) and long thick microfilament bundles.Adhesions of knock-out cells are fewer and generally small (arrows),with a more diffuse meshwork of F-actin. Bar 10µm

Cell Tissue Res (2010) 339:31–46 41

integrins or a lack of integrin activation but is perhaps relatedto a reduction in cytoskeletal organisation. These resultssuggest a regulatory role for syndecan-2 inmatrix assembly, inaddition to the previously suggested roles for activatedintegrins (Klass et al. 2000). In a second study of matrixassembly in HT1080 cells, specific knock-down ofsyndecan-2 by siRNA has been demonstrated to reducefibronectin matrix assembly; reduction of the other cellsurface HSPGs present in these cells (syndecans-1 and -4,glypican-1 and betaglycan) has no effect (Galante andSchwarzbauer 2007). This study has also highlighted thekey role of heparan sulphate in matrix assembly.

Syndecan-4

Syndecan-4 is unique in that it can become incorporated intofocal adhesions together with integrins (Woods and Couchman1994; Couchman and Woods 1999; Couchman 2003). So far,no other syndecan has been suggested to play a role injunction assembly. Studies from several groups have shownthat syndecan-4 enhances focal adhesion assembly, althoughthe essential role of integrins is also clear (Saoncella et al.1999; Woods et al. 2000; Mostafavi-Pour et al. 2003). Focaladhesions are the points of strong attachment of cells tomatrix and lie at the termini of microfilament bundles. Thenumber of focal adhesions, their size and their stability allinfluence cell migration and, in general, large focal adhesionsoppose migration, being points of stable anchorage (Pellegrinand Mellor 2007). The best characterised model system isfibroblast adhesion to fibronectin domains. Cell attachmentand spreading are readily promoted by the RGD-containingintegrin-binding domain, whereas the Hep II domain pro-motes syndecan-4 signalling and focal adhesion assembly(Woods et al. 2000; Couchman 2003; Morgan et al. 2007).The combinatorial interactions of syndecan-4, fibronectin andintegrins involve a series of downstream signalling events,including the activation of tyrosine kinases by the integrins.Syndecan-4 cytoplasmic domain, however, recruits theinositide PtdIns4,5P2, which initiates a conformationalchange that may provide a binding platform for proteinkinase Cα (PKCα; Shin et al. 2001; Lim et al. 2003). Thekinase is persistently activated (Oh et al. 1998; Horowitz etal. 1999; Keum et al. 2004) in the downstream pathwayleading to RhoA and its targets, the Rho kinases (Dovas et al.2006). These promote stress fibre and focal adhesionassembly through their phosphorylation of myosin II andmyosin phosphatase (Pellegrin and Mellor 2007). Our recentdata suggest a pathway that can contribute to enhanced GTP-RhoA levels. Rho guanine dissociation inhibitor (RhoGDI) isa cytoplasmic protein that can sequester members of the Rhofamily in their GDP-bound form. In this way, it can act as a“sink” for the GTPases but its role is probably more complex(Dovas and Couchman 2005; DerMardirossian and Bokoch

2005). RhoGDI may also target the GTPases to subcellularsites. Our findings have revealed that RhoGDIα can bephosphorylated by PKCα, leading to a selective release ofRhoA, but not other GTPases such as Rac1 and cdc42(Dovas et al. submitted). Once released, the RhoA can thenbecome GTP-loaded and activate the Rho kinases (Fig. 3) todrive focal adhesion assembly. Although focal adhesions areunderstood to be able to mature into matrix assembly points,roles for syndecan-4 in this transition are unclear. Therefore,direct roles of syndecan-4 in matrix assembly are also notfully understood. However, this syndecan can certainlyinfluence the cytoskeleton, since α-actinin binds the cyto-plasmic domain directly (Greene et al. 2003) and “knock-out”fibroblasts have decreased organisation of actin with fewermicrofilament bundles, perhaps indicative of decreasedtension exerted on the matrix (Fig. 4; for a review, see Okinaet al. 2009).

However, syndecan-4 and integrins do not seem tointeract directly. Recently, mesenchymal cells have beenshown to respond in a second syndecan-4-dependent celladhesion pathway. The syndecan-4 ectodomain withoutheparan sulphate chains can trigger focal adhesion assemblyin which roles for α5β1 integrin have been determined(Whiteford et al. 2007). A key NXIP motif within thesyndecan-4 ectodomain has been identified in the adhesion-promoting response (Whiteford and Couchman 2006).Finally, two interesting recent reports have revealed thepotential importance of syndecan-4 in wound healing andtissue repair. First, Telci et al. (2008) have shown that thefibronectin-associated tissue transglutaminase TG2 can bindsyndecan-4 heparan sulphate chains and trigger PKCα-mediated signalling, feeding downstream to RhoA and stressfibre formation. This also involved β1 integrin, andcombinatorial signalling leads to ERK1/2 mitogen-activatedsignalling and survival pathways. Second, the migration offibroblasts from collagen gels into a fibrin-rich matrixrequires fibronectin and, specifically, syndecan-4. Interac-tions between syndecan-4 and the fibronectin Hep II domainhave been described, as has a transcriptional sensitivity ofsyndecan-4 to platelet-derived growth factor (Lin et al.2005). This last study raises the issue of the roles ofsyndecan-4 not only in processes such as wound contractionin which stress fibres are important, but also cell migration.On this point, Bass et al. (2007) have shown that syndecan-4can promote persistence in fibroblast migration, throughspatial control of GTP-Rac. Although syndecan-4 nullfibroblasts have high GTP-Rac levels, their migration iscompromised since it is not targeted to a leading lamella.Data from endothelial cells (Tkachenko et al. 2006) alsosuggest that the clustering of syndecan-4 leads to Racactivation and cell migration, this time dependent oncytoplasmic interactions with the PDZ protein synectin(GIPC). In addition, Bass et al. (2008) have suggested that

42 Cell Tissue Res (2010) 339:31–46

syndecan-4 regulates the distribution of a Rho-GTPase-activating protein (p190RhoGAP) in a PKC-dependentmanner. Integrin-mediated signalling promotes GAP tyrosinephosphorylation, which leads to decreased GTP-RhoA levels(Fig. 3). The combined activities of integrin and syndecan-4therefore lead to targeted RhoGAP activity, consistent withmigratory activity. There remains much to discover, sincesyndecan-4 is now implicated not only in cell migration andRac activity, but also focal adhesion assembly, a Rho-drivenprocess. The spatial and temporal control of these activitiesmay be key to understanding the role of syndecan-4. As withsyndecan-1, the null mouse has wound healing (granulationtissue angiogenesis) problems that might be related to cellmigration but such defects do not manifest themselves indevelopment. Nonetheless, data from both Xenopus andDanio suggest roles for syndecan-4 in embryonic cellmotility (Muñoz et al. 2006; Matthews et al. 2008).

Concluding remarks

Taken together, vertebrate syndecans act as receptors formultiple ECM proteins and can contribute to ECM assembly.However, syndecans cannot be considered in isolation.Syndecans and integrins collaborate extensively in mediatingcell adhesion to a broad range of ECM molecules and cancombine synergistically to regulate adhesion-complex forma-tion, cell spreading and directional migration. Interactionsbetween syndecans and integrins can be direct or indirect,depending on the circumstances, and some growth factors andtheir high-affinity receptors can clearly influence cell-matrixrelationships. Downstream effects on the cytoskeleton are alsoaccompanied by the transcriptional control of, for example,MMPs. In turn, these enzymes can have profound effects onmatrix stability and cell behaviour. Much remains to beunderstood regarding syndecan signalling and the way that itis regulated. Apart from syndecan-4, our knowledge of thecytoplasmic interactions of these proteoglycans is stillfragmentary. In vivo techniques, built on a solid appreciationof structure and interactions, should shed light on where andwhen signalling occurs.

Acknowledgements The authors thank Dr. Pia Klausen (BiotechResearch and Innovation Center, University of Copenhagen) forpreparing Figs. 2, 3.

References

Adams JC, Kureishy N, Taylor AL (2001) A role for syndecan-1 incoupling fascin spike formation by thrombospondin-1. J CellBiol 152:1169–1182

Alexopoulou AN, Multhaupt HA, Couchman JR (2007) Syndecans inwound healing, inflammation and vascular biology. Int JBiochem Cell Biol 39:505–528

Andac Z, Sasaki T, Mann K, Brancaccio A, Deutzmann R, Timpl R(1999) Analysis of heparin, alpha-dystroglycan and sulfatidebinding to the G domain of the laminin alpha1 chain by site-directed mutagenesis. J Mol Biol 287:253–264

Bachy S, Letourneur F, Rousselle P (2008) Syndecan-1 interactionwith the LG4/5 domain in laminin-332 is essential for keratino-cyte migration. J Cell Physiol 214:238–249

Barkalow FJ, Schwarzbauer JE (1991) Localization of the majorheparin-binding site in fibronectin. J Biol Chem 266:7812–7818

Bass MD, Roach KA, Morgan MR, Mostafavi-Pour Z, Schoen T,Muramatsu T, Mayer U, Ballestrem C, Spatz JP, Humphries MJ(2007) Syndecan-4-dependent Rac1 regulation determines direc-tional migration in response to the extracellular matrix. J CellBiol 177:527–538

Bass MD, Morgan MR, Roadch KA, Settleman J, Goryachev AB,Humphries MJ (2008) p190RhoGAP is the convergence point ofadhesion signals from α5β1 integrin and syndecan-4. J Cell Biol181:1013–1026

Bax DV, Bernard SE, Lomas A, Morgan A, Humphries J,Shuttleworth CA, Humphries MJ, Kielty CM (2003) Celladhesion to fibrillin-1 molecules and microfibrils is mediatedby alpha 5 beta 1 and alpha v beta 3 integrins. J Biol Chem278:34605–34616

Bax DV, Mahalingam Y, Cain S, Mellody K, Freeman L, Younger K,Shuttleworth CA, Humphries MJ, Couchman JR, Kielty CM(2007) Cell adhesion to fibrillin-1: identification of an Arg-Gly-Asp-dependent synergy region and a heparin-binding site thatregulates focal adhesion formation. J Cell Sci 120:1383–1392

Beauvais DM, Rapraeger AC (2003) Syndecan-1-mediated cellspreading requires signaling by alphavbeta3 integrins in humanbreast carcinoma cells. Exp Cell Res 286:219–232

Beauvais DM, Rapraeger AC (2004) Syndecans in tumor celladhesion and signaling. Reprod Biol Endocrinol 2:3

Beauvais DM, Ell BJ, McWhorter AR, Rapraeger AC (2009)Syndecan-1 regulates αVβ3 and αVβ5 integrin activation duringangiogenesis and is blocked by synstatin, a novel peptideinhibitor. J Exp Med 206:691–705

Bishop JR, Schuksz M, Esko JD (2007) Heparan sulphate proteogly-cans fine-tune mammalian physiology. Nature 446:1030–1037

Cain SA, Baldock C, Gallagher J, Morgan A, Bax DV, Weiss AS,Shuttleworth CA, Kielty CM (2005) Fibrillin-1 interactions withheparin. Implications for microfibril and elastic fiber assembly. JBiol Chem 280:30526–30537

Cain SA, Baldwin AK, Mahalingam Y, Raynal B, Jowitt TA,Shuttleworth CA, Couchman JR, Kielty CM (2008a) Heparansulfate regulates fibrillin-1 N- and C-terminal interactions. J BiolChem 283:27017–27027

Cain SA, Raynal B, Hodson N, Shuttleworth A, Kielty CM (2008b)Biomolecular analysis of elastic fibre molecules. Methods45:42–52

Carlson CB, Lawler J, Mosher DF (2008) Structures of thrombo-spondins. Cell Mol Life Sci 65:672–686

Chakravati R, Sapountzi V, Adams JC (2005) Functional role ofsyndecan-1 cytoplasmic V region in lamellipodial spreading,actin bundling, and cell migration. Mol Biol Cell 16:3678–3691

Chen E, Hermanson S, Ekker SC (2004) Syndecan-2 is essential forangiogenic sprouting during zebrafish development. Blood103:1710–1719

Chiquet-Ehrismann R (2004) Tenascins. Int J Biochem Cell Biol36:986–990

Choi S, Lee E, Kwon S, Park H, Yi JY, Kim S, Han IO, Yun Y, Oh ES(2005) Transmembrane domain-induced oligomerization is cru-

Cell Tissue Res (2010) 339:31–46 43

cial for the functions of syndecan-2 and syndecan-4. J Biol Chem280:42573–42579

Clamp A, Blackhall FH, Henrioud A, Jayson GC, Javaherian K, EskoJ, Gallagher JT, Merry CLR (2006) The morphogenic propertiesof oligomeric endostatin are dependent on cell surface heparansulfate. J Biol Chem 281:14813–14822

Couchman JR (2003) Syndecans: proteoglycan regulators of cell-surface microdomains? Nat Rev Mol Cell Biol 4:926–937

Couchman JR, Woods A (1999) Syndecan-4 and integrins: combina-torial signaling in cell adhesion. J Cell Sci 112:3415–3420

Couchman JR, Chen L, Woods A (2001) Syndecans and cell adhesion.Int Rev Cytol 207:113–150

DerMardirossian C, Bokoch GM (2005) GDIs: central regulatorymolecules in RhoGTPase activation. Trends Cell Biol 15:609–613

Dews IC, Mackenzie KR (2007) Transmembrane domains of thesyndecan family of growth factor coreceptors display a hierarchyof homotypic and heterotypic interactions. Proc Natl Acad SciUSA 104:20782–20787

Dovas A, Couchman JR (2005) RhoGDI: multiple functions in theregulation of Rho family GTPases. Biochem J 390:1–9

Dovas A, Yoneda A, Couchman JR (2006) PKCalpha-dependentactivation of RhoA by syndecan-4 during focal adhesionformation. J Cell Sci 119:2837–2846

Durbeej M, Talts JF, Henry MD, Yurchenco PD, Campbell KP,Ekblom P (2001) Dystroglycan binding to laminin alpha1LG4module influences epithelial morphogenesis of salivary gland andlung in vitro. Differentiation 69:121–134

Esko JD, Selleck SB (2002) Order out of chaos: assembly ofligand binding sites in heparan sulfate. Annu Rev Biochem71:435–471

Esko JD, Zhang L (1996) Influence of core protein sequence onglycosaminoglycan assembly. Curr Opin Struct Biol 6:663–670

Essner JJ, Chen E, Ekker SC (2006) Syndecan-2. Int J Biochem CellBiol 38:152–156

Ethell IM, Irie F, Kalo MS, Couchman JR, Pasquale EB, YamaguchiY (2001) Eph B2/syndecan-2 signaling in dendritic spinemorphogenesis. Neuron 31:1001–1013

Galante LL, Schwarzbauer JE (2007) Requirements for sulfatetransport and the diastrophic dysplasia sulfate transporter infibronectin matrix assembly. J Cell Biol 179:999–1009

Gallagher J (2001) Heparan sulfate: growth control with a restrictedsequence menu. J Clin Invest 108:357–361

Gama-de-Souza LN, Cyreno-Oliveira E, Freitas VM, Melo ES, Vilas-Boas VF, Moriscot AS, Jaeger RG (2008) Adhesion and proteaseactivity in cell lines from human salivary gland tumors areregulated by the laminin-derived peptide AG73, syndecan-1 andbeta1 integrin. Matrix Biol 27:402–419

Granés F, Berndt C, Roy C, Mangeat P, Reina M, Vilaró S (2003)Identification of a novel ezrin-binding site in syndecan-2cytoplasmic domain. FEBS Lett 547:212–216

Greene DK, Tumova S, Couchman JR, Woods A (2003) Syndecan-4associates with α-actinin. J Biol Chem 278:7617–7623

Hienola A, Tumova S, Kulesskiy E, Rauvala H (2006) N-syndecandeficiency impairs neural migration in brain. J Cell Biol174:569–580

Hoffman MP, Nomizu M, Roque E, Lee S, Jung DW, Yamada Y,Kleinman HK (1998) Laminin-1 and laminin-2 G-domainsynthetic peptides bind syndecan-1 and are involved in acinarformation of a human submandibular gland cell line. J Biol Chem273:28633–28641

Hoffman MP, Engbring JA, Nielsen PK, Vargas J, Steinberg Z,Karmand AJ, Nomizu M, Yamada Y, Kleinman HK (2001) Celltype-specific differences in glycosaminoglycans modulate thebiological activity of a heparin-binding peptide (RKRLQVQLSIRT)from the G domain of the laminin α1 chain. J Biol Chem276:22077–22085

Horowitz A, Murakami M, Gao Y, Simons M (1999) Phosphatidyli-nositol 4, 5- bisphosphate mediates the interaction of syndecan-4with protein kinase C. Biochemistry 38:15871–15877

Hozumi K, Suzuki N, Nielsen PK, Nomizu M, Yamada Y (2006)Laminin alpha1 chain LG4 module promotes cell attachmentthrough syndecans and cell spreading through integrin alpha2-beta1. J Biol Chem 281:32929–32940

Huang W, Chiquet-Ehrismann R, Moyano JV, Garcia-Pardo A, OrendG (2001) Interference of tenascin-C with syndecan-4 binding tofibronectin blocks cell adhesion and stimulates tumor cellproliferation. Cancer Res 61:8586–8594

Ingham KC, Brew SA, Atha DH (1990) Interaction of heparin withfibronectin and isolated fibronectin domains. Biochem J272:605–611

Ingham KC, Brew SA, Migliorini MM, Busby TF (1993) Bindingof heparin by type III domains and peptides from thecarboxyterminal hepII region of fibronectin. Biochemistry32:12548–12553

Isenberg JS, Frazier WA, Roberts DD (2008) Thrombospondin-1: aphysiological regulator of nitric oxide signaling. Cell Mol LifeSci 65:728–742

Jurjus RA, Liu Y, Pal-Ghosh S, Tadvalkar G, Stepp MA (2008)Primary dermal fibroblasts derived from sdc-1 deficient micemigrate faster and have altered alpha v integrin function. WoundRepair Regen 16:649–660

Kato M, Saunders S, Nguyen H, Bernfield M (1995) Loss of cellsurface syndecan-1 causes epithelia to transform intoanchorage-independent mesenchyme-like cells. Mol Biol Cell6:559–576

Kazerounian S, Yee KO, Lawler J (2008) Thrombospondins in cancer.Cell Mol Life Sci 65:700–712

Keum E, Kim Y, Kim J, Kwon S, Lim Y, Han I, Oh ES (2004)Syndecan-4 regulates localisation, activity and stability of proteinkinase Cα. Biochem J 378:1007–1014

Khan MY, Jaikaria NS, Frenz DA, Villanueva G, Newman SA (1988)Structural changes in the NH2-terminal domain of fibronectinupon interaction with heparin. J Biol Chem 263:11314–11318

Kielty CM, Baldock C, Lee D, Rock MJ, Ashworth JL, ShuttleworthCA (2002) Fibrillin: from microfibril assembly to biomechanicalfunction. Phil Trans R Soc Lond Biol 357:207–217

Kinsey R, Williamson MR, Chaudry S, Mellody KT, McGovern A,Takahashi S, Shuttleworth CA, Kielty CM (2008) Fibrillin-1microfibril deposition is dependent on fibronectin assembly. JCell Sci 121:2696–2704

Klass CM, Couchman JR, Woods A (2000) Control of extracellularmatrix assembly by syndecan-2 proteoglycan. J Cell Sci113:493–506

Koda JE, Rapraeger A, Bernfield M (1985) Heparan sulfateproteoglycans from mouse mammary epithelial cells: cell surfaceproteoglycan as a receptor for interstitial collagens. J Biol Chem260:8157–8162

Kreuger J, Spillmann D, Li JP, Lindahl U (2006) Interactions betweenheparan sulfate and proteins: the concept of specificity. J CellBiol 174:323–327

Kuriyama S, Mayor R (2009) A role for syndecan-4 in neuralinducation involving ERK- and PKC-dependent pathways.Development 136:575–584

Lander AD (1998) Proteoglycans: master regulators of molecularencounter. Matrix Biol 17:465–472

Leiss M, Beckmann K, Giros A, Costell M, Fassler R (2008) The roleof integrin binding sites in fibronectin matrix assembly in vivo.Curr Opin Cell Biol 20:502–507

Lim ST, Longley RL, Couchman JR, Woods A (2003) Direct binding ofsyndecan-4 cytoplasmic domain to the catalytic domain of proteinkinase C alpha (PKCα) increases focal adhesion localisation ofPKCα. J Biol Chem 278:13795–13802

44 Cell Tissue Res (2010) 339:31–46

Lin F, Ren XD, Greiling D, Clark RA (2005) Three-dimensionalmigration of human adult dermal fibroblasts from collagenlattices into fibrin/fibronectin gels requires syndecan-4 proteo-glycan. J Invest Dermatol 124:906–913

Lyon M, Rushton G, Askari JA, Humphries MJ, Gallagher JT (2000)Elucidation of the structural features of heparan sulfate importantfor interaction with the Hep-2 domain of fibronectin. J BiolChem 275:4599–4606

Mahalingam Y, Gallagher JT, Couchman JR (2007) Cellular adhesionresponses to the heparin-binding (HepII) domain of fibronectinrequire heparan sulfate with specific properties. J Biol Chem282:3221–3230

Mao Y, Schwarzbauer JE (2005) Fibrinectin fibrillogenesis, a cell-mediated matrix assembly process. Matrix Biol 24:389–399

Maqueda A, Moyana JV, Hernández Del Cerro M, Peters DM, Garcia-Pardo A (2007) The heparin III-binding domain of fibronectin(III4–5 repeats) binds to fibronectin and inhibits fibronectinmatrix assembly. Matrix Biol 26:642–651

Matsuura Y, Momota Y, Murata K, Matsushima H, Suzuki N, NomizuM, Shinkai H, Utani A (2004) Localization of the laminin alpha4chain in the skin and identification of a heparin-dependent celladhesion site within the laminin α4 chain C-terminal LG4module. J Invest Dermatol 122:614–620

Matthews HK, Marchant L, Carmona-Fontaine C, Kuriyama S,Larraín J, Holt MR, Parsons M, Mayor R (2008) Directionalmigration of neural crest cells in vivo is regulated by syndecan-4/Rac1 and non-canonical Wnt signalling/RhoA. Development135:1771–1780

Midwood KS, Valenick LV, Hsia HC, Schwarzbauer JE (2004)Coregulation of fibronectin signaling and matrix contraction bytenascin-C and syndecan-4. Mol Biol Cell 15:5670–5677

Miner JH, Yurchenco PD (2004) Laminin functions in tissuemorphogenesis. Annu Rev Cell Dev Biol 20:255–284

Morgan MR, Humphries MJ, Bass MD (2007) Synergistic control ofcell adhesion by integrins and syndecans. Nat Rev Mol Cell Biol8:957–969

Mostafavi-Pour Z, Askari JA, Parkinson SJ, Parker PJ, Ng TT,Humphries MJ (2003) Integrin-specific signaling pathwayscontrolling focal adhesion formation and cell migration. J CellBiol 161:155–167

Multhaupt HAB, Yoneda A, Whiteford JR, Oh ES, Lee W, CouchmanJR (2009) Syndecan signaling: when, where and why. J PhysiolPharmacol (in press)

Munesue S, Kusano Y, Oguri K, Itano N, Yoshitomi Y,Nakanishi H, Yamashina I, Okayama M (2002) The role ofsyndecan-2 in regulation of actin-cytoskeletal organization ofLewis lung carcinoma-derived metastatic clones. Biochem J363:201–209

Munesue S, Yoshitomi Y, Koyama Y, Nishiyama A, Nakanishi H,Miyazaki K, Ishimaru T, Okayama M, Oguri K (2007) A novelfunction of syndecan-2, suppression of matrix metalloproteinase-2activation, which causes suppression of metastasis. J Biol Chem282:28164–28174

Muñoz R, Moreno M, Oliva C, Orbenes C, Larraín J (2006)Syndecan-4 regulates non-canonical Wnt signalling and isessential for convergence and extension movements in Xenopusembryos. Nat Cell Biol 8:492–500

Murakami M, Elfenbein A, Simons M (2008) Non-canonicalfibroblast growth factor signalling in angiogenesis. CardiovascRes 78:223–231

Murphy KJ, Merry CL, Lyon M, Thompson JE, Roberts IS, GallagherJT (2004) A new model for the domain structure of heparansulfate based on the novel specificity of K5 lyase. J Biol Chem279:27239–27245

Nadanaka S, Kitagawa H (2008) Heparan sulphate biosynthesis anddisease. J Biochem 144:7–14

Narita R, Yamashita H, Goto A, Imai H, Ichihara S, Mori H, KitagawaY (2004) Syndecan-dependent binding of Drosophila hemocytesto laminin α3/5 chain LG4–5 modules: potential role in sessilehemocyte islets formation. FEBS Lett 576:127–132

Nunes SS, Outeiro-Bernstein MA, Juliano L, Vardiero F, Nader HB,Woods A, Legrand C, Morandi V (2008) Syndecan-4 contributesto endothelial tubulogenesis through interactions with two motifsinside the pro-angiogenic N-terminal domain of thrombospondin-1. J Cell Physiol 214:828–837

Ogawa T, Tsubota Y, Hashimoto J, Kariya Y, Miyazaki K (2007a) Theshort arm of laminin γ2 chain of laminin-5 (laminin 332) bindssyndecan-1 and regulates cellular adhesion and migration bysuppressing phosphorylation of integrin β4 chain. Mol Biol Cell18:1621–1633

Ogawa T, Tsubota Y, Hashimoto J, Kariya Y, Miyazaki K (2007b) Theshort arm of laminin gamma2 chain of laminin-5 (laminin-332)binds syndecan-1 and regulates cellular adhesion and migrationby suppressing phosphorylation of integrin beta4 chain. Mol BiolCell 18:1621–1633

Oh ES, Woods A, Lim ST, Theibert AW, Couchman JR (1998)Syndecan-4 proteoglycan cytoplasmic domain and phosphatidy-linositol 4, 5-bisphosphate coordinately regulate protein kinase Cactivity. J Biol Chem 273:10624–10629

Okamoto O, Bachy S, Odenthal U, Rigal D, Lortat-Jacob H, Smyth N,Rousselle P (2003) Normal human keratinocytes bind to thealpha 3LG4/5 domain of unprocessed laminin-5 through thereceptor syndecan-1. J Biol Chem 278:44168–44177

Okina E, Jensen TM, Whiteford JR, Couchman JR (2009)Syndecan proteoglycan contributions to cytoskeletal organi-zation and contractility. Scand J Med Sci Sports doi:10.1111/j.1600-0838.2009.00941.x

Olivares GH, Carrasco H, Aroca F, Carvallo J, Segovia F, Larraín J(2009) Syndecan-1 regulates BMP signaling and dorso-ventralpatterning of the ectoderm during early Xenopus development.Dev Biol 329:338–349

Orend G, Chiquet-Ehrismann R (2006) Tenascin-C induced signalingin cancer. Cancer Lett 2:143–163

Orend G, Huang W, Olayioye MA, Hynes NE, Chiquet-Ehrismann R(2003) Tenascin-C blocks cell-cycle progression of anchorage-dependent fibroblasts on fibronectin through inhibition ofsyndecan-4. Oncogene 22:3917–3926

Pankov R, Yamada KM (2002) Fibronectin at a glance. J Cell Sci115:3861–3863

Park H, Kim Y, Lim Y, Han I, Oh ES (2002) Syndecan-2 mediatesadhesion and proliferation of colon carcinoma cells. J Biol Chem277:29730–29736

Pellegrin S, Mellor H (2007) Actin stress fibres. J Cell Sci 120:3491–3499

Ritty TM, Broekelmann TJ, Werneck CC, Mecham RP (2003)Fibrillin-1 and -2 contain heparin-binding sites important formatrix deposition and that support cell attachment. Biochem J375:425–432

Saito Y, Imazeki H, Miura S, Yoshimura T, Okutsu H, Harada Y,Ohwaki T, Nagao O, Kamiya S, Hayashi R, Kodama H, HandaH, Yoshida T, Fukai F (2007) A peptide from tenascin-C inducesbeta1 integrin activation through syndecan-4. J Biol Chem282:34929–34937

Salmivirta M, Elenius K, Vainio S, Hofer U, Chiquet-Ehrismann R,Thesleff I, Jalkanen M (1991) Syndecan from tooth mesenchymebinds tenascin. J Biol Chem 266:7733–7739

Sanderson RD, Sneed TB, Young LA, Sullivan GL, Lander AD(1992) Adhesion of B lymphoid (MPC-11) cells to type Icollagen is mediated by integral membrane proteoglycan,syndecan. J Immunol 148:3902–3911

Saoncella S, Echtermeyer F, Denhez F, Nowlen JK, Mosher DF,Robinson SD, Hynes RO, Goetinck PF (1999) Syndecan-4

Cell Tissue Res (2010) 339:31–46 45

signals cooperatively with integrins in a Rho-dependent mannerin the assembly of focal adhesions and actin stress fibers. ProcNatl Acad Sci USA 96:2805–2810

Sharma A, Askari JA, Humphries MJ, Jones EY, Stuart DI (1999)Crystal structure of a heparin- and integrin-binding segment ofhuman fibronectin. EMBO J 18:1468–1479

Shin J, Lee W, Lee D, Koo B-K, Han I, Lim Y, Woods A, CouchmanJR, Oh E-S (2001) Solution structure of the dimeric cytoplasmicdomain of syndecan-4. Biochemistry 40:8471–8478

Stepp MA, Liu Y, Pal-Ghosh S, Jurjus RA, Tadvalkar G, Sekaran A,LoSicco K, Jiang L, Larsen M, Li L, Yuspa SH (2007) Reducedmigration, altered matrix and enhanced TGF-β1 signaling aresignatures of mouse keratinocytes lacking Sdc1. J Cell Sci120:2851–2863

Sugawara K, Tsuruta D, Ishii M, Jones JCR, Kobayashi H (2008)Laminin-332 and 511 in skin. Exp Dermatol 17:473–480

Suzuki N, Ichikawa N, Kasai S, Yamada M, Nishi N, Morioka H,Yamashita H, Kitagawa Y, Utani A, Hoffman MP, Nomizu M(2003) Syndecan binding sites in the laminin α1 chain G domain.Biochemistry 42:12625–12633

Suzuki N, Yokoyama F, Nomizu M (2005) Functional sites in thelaminin alpha chains. Connect Tissue Res 46:142–152

Sweeney SM, Orgel JP, Fertala A, McAuliffe JD, Turner KR, Di LulloGA, Chen S, Antipova O, Perumal S, Ala-Kokko L, Forlino A,Cabral WA, Barnes AM, Marinin JC, San Antonio JD (2008)Candidate cell and matrix interaction domains on the collagenfibril, the predominant protein of vertebrates. J Biol Chem283:21187–21197

Telci D, Wang Z, Li X, Verderio EA, Humphries MJ, BaccariniM, Basaga H, Griffin M (2008) Fibronectin-tissue trans-glutaminase matrix rescues RGD-impaired cell adhesionthrough syndecan-4 and β1 integrin co-signaling. J BiolChem 283:20937–20947

Tiedemann K, Bätge B, Müller PK, Reinhardt DP (2001) Interactionsof fibrillin-1 with heparin/heparan sulfate, implications formicrofibrillar assembly. J Biol Chem 276:36035–36042

Tkachenko E, Elfenbein A, Tirziu D, Simons M (2006) Syndecan-4clustering induces cell migration in a PDZ-dependent manner.Circ Res 98:1398–1404

Tucker RP, Chiquet-Ehrismann R (2009) The regulation of tenascinexpression by tissue microenvironments. Biochim Biophys Actadoi:10.10.16/j.bbamcr.2008.12.012

Tumova S, Woods A, Couchman JR (2000) Heparan sulfate chainsfrom glypican and syndecans bind the HepII domain offibronectin similarly despite minor structural differences. J BiolChem 275:9410–9417

Underwood PA, Kirkpatrick A, Mitchell SM (2002) New insights intoheparin binding to vitronectin: studies with monoclonal anti-bodies. Biochem J 365:57–67

Utani A, Nomizu M, Matsuura H, Kato K, Kobayashi T, Takeda U,Aota S, Nielsen PK, Shinkai H (2001) A unique sequence of the

laminin α3 G domain binds to heparin and promotes celladhesion through syndecan-2 and -4. J Biol Chem 276:28779–28788

Utani A, Momota Y, Endo H, Kauya Y, Beck K, Suzuki N, NomizuM, Shinkai H (2003) Laminin α3 LG4 module induces matrixmetalloproteinase-1 through mitogen-activated protein kinasesignaling. J Biol Chem 278:34483–34490

Vuoriluoto K, Jokinen J, Kallio K, Salmivirta M, Heino J, Ivaska J(2008) Syndecan-1 supports integrin alpha2beta1-mediated ad-hesion to collagen. Exp Cell Res 314:3149–3181

Whiteford JR, Couchman JR (2006) A conserved NXIP motif isrequired for cell adhesion properties of the syndecan-4 ectodo-main. J Biol Chem 281:32156–32163

Whiteford JR, Behrends V, Kirby H, Kusche-Gullberg M, MuramatsuT, Couchman JR (2007) Syndecans promote integrin-mediatedadhesion of mesenchymal cells in two distinct pathways. ExpCell Res 313:3902–3913

Wilkins-Port CE, Sanderson RD, Tominna-Sebald E, McKeown-Longo PJ (2003) Vitronectin’s basic domain is a syndecan ligandwhich functions in trans to regulate vitronectin turnover. CellCommun Adhes 10:85–103

Williams S, Schwarzbauer JE (2009) A shared mechanism of adhesionmodulation for tenascin-C and fibulin-1. Mol Biol Cell 20:1141–1149

Woods A, Couchman JR (1994) Syndecan 4 heparan sulfateproteoglycan is a selectively enriched and widespread focaladhesion component. Mol Biol Cell 5:183–192

Woods A, Longley RL, Tumova S, Couchman JR (2000) Syndecan-4binding to the high affinity heparin-binding domain of fibronec-tin drives focal adhesion formation in fibroblasts. Arch BiochemBiophys 374:66–72

Yamashita H, Goto A, Kadowaki T, Kitagawa Y (2004)Mammalian and Drosophila cells adhere to the laminin alpha4LG4 domain through syndecans, but not glypicans. Biochem J382:933–943

Yoneda J, Saiki I, Igarashi Y, Kobayashi H, Fujii H, Ishizaki Y,Kimizuka F, Kato I, Azuma I (1995) Role of the heparin-binding domain of chimeric peptides derived from fibro-nectin in cell spreading and motility. Exp Cell Res217:169–179

Yokoyama F, Suzuki N, Kadoya Y, Utani A, Nakatsuka H, Nishi N,Haruki M, Kleinman HK, Nomizu M (2005) Bifunctionalpeptides derived from homologous loop regions in the lamininα chain LG4 modules interact with both α2β1 integrin andsyndecan-2. Biochemistry 44:9581–9589

Zimmermann P, David G (1999) The syndecans, tuners of transmem-brane signaling. FASEB J 13:S91–S100

Zimmermann P, Zhang Z, Degeest G, Mortier E, Leenaerts I,Coomans C, Schulz J, N’Kuli F, Courtoy PJ, David G (2005)Syndecan recycling is controlled by syntenin-PIP2 interactionsand Arf6. Dev Cell 9:377–388

46 Cell Tissue Res (2010) 339:31–46