Mapping of matrix metalloproteinase cleavage sites on syndecan-1 and syndecan-4 ectodomains

Authors


Correspondence

J. R. Couchman, Department of Biomedical Sciences, University of Copenhagen, Biocenter, Ole Maaløes Vej 5, 2200 Copenhagen N, Denmark

Fax: +45 353 25669

Tel: +45 353 25670

E-mail: john.couchman@bric.ku.dk

Abstract

Syndecans are transmembrane heparan sulfate proteoglycans with roles in cell proliferation, differentiation, adhesion, and migration. They have been associated with multiple functions in tumour progression, through their ability to interact with a wide range of ligands as well as other receptors, which makes them key effectors in the pericellular microenvironment. Extracellular shedding of syndecans by tumour-associated matrix metalloproteinases (MMPs) may have an important role in tumour progression. Such ectodomain shedding generates soluble ectodomains that may function as paracrine or autocrine effectors, or as competitive inhibitors of the intact proteoglycan. Tumour-associated MMPs are shown here to cleave the ectodomains of human syndecan-1 and syndecan-4. Two membrane proximal regions of both syndecan-1 and syndecan-4 are favoured MMP cleavage sites, six and 15 residues from the transmembrane domain. Other sites are 35–40 residues C-terminal from the heparan sulfate chain substitution sites in both syndecans. The MT1-MMP cleavage sites in syndecan-1 and syndecan-4 were confirmed by site-directed mutagenesis. These findings provide insights into the characteristics of syndecan shedding.

Abbreviations
Cm-Tf

carboxymethylated transferrin

CS

chondroitin sulfate

DS

dermatan sulfate

ECM

extracellular matrix

GAG

glycosaminoglycan

HS

heparan sulfate

MMP

matrix metalloproteinase

Introduction

Cellular communication with the extracellular matrix (ECM) underpins many aspects of development, tissue repair, and disease pathogenesis. Alongside the major ECM receptors of the integrin family, near universal components of the cell surface are the syndecans. Syndecans are a small family of type 1 transmembrane heparan sulfate (HS) proteoglycans with a long evolutionary history. The four mammalian members, syndecan-1 to syndecan-4, all comprise an ectodomain, a single transmembrane domain, and a short cytoplasmic domain [1-3]. Covalently attached to the ectodomain are glycosaminoglycan (GAG) chains, which are predominantly HS. Syndecan-1 and syndecan-3 can be substituted additionally with chondroitin sulfate (CS) or dermatan sulfate (DS) at sites near the transmembrane domain [4]. Almost all nucleated mammalian cells possess one or more syndecans on the cell surface. Many syndecan functions are attributed to their HS chains, which interact with a wide range of ECM molecules, chemokines, cytokines, growth factors, proteinases and their inhibitors, morphogens, and adhesion molecules. Presumably, these interactions are triggers of conserved signalling responses that influence developmental, wound healing and tumour progression processes [1, 5-8].

The ectodomains of syndecans are constitutively shed to a small degree under physiological conditions, but this may be dramatically increased in response to stimuli [9, 10]. Such ectodomain shedding generates soluble ectodomains that may function as paracrine or autocrine effectors, or competitive inhibitors of intact proteoglycans. Shed ectodomains, complete with their GAG complement, may form gradients across tissues that influence cell behaviour, e.g. migration in tissue repair [9, 11]. As syndecans are important coreceptors for adhesion and growth factor receptors, their loss from the cell surface may also lead to termination of signalling.

Sheddases such as matrix metalloproteinases (MMPs) and ADAMs release ectodomains, whereas heparanase [12] and endosulfatases [13] modify the structures and bioactivity of HS chains within the extracellular environment, including the ability of the HS chains to bind and sequester growth factors [9, 14]. Evidence indicates the involvement of several MMPs in syndecan cleavage in vitro and in vivo. Matrilysin (MMP7) can cleave syndecan-1 [11], the gelatinases MMP2 and MMP9 can cleave syndecan-1, syndecan-2, and syndecan-4 [15, 16], and the membrane-associated metalloproteinases MT1-MMP and MT3-MMP are known to cleave syndecan-1 [17]. Current knowledge of the precise cleavage sites on syndecan core proteins is sparse. Among the family of MMPs, only MT1-MMP cleavage sites on syndecan-1 core protein have been reported. Human syndecan-1 is cleaved at Gly82-Leu83 and Gly245-Leu246, whereas cleavage of murine syndecan-1 occurs at Ala243-Ser244 [17, 18].

There is certainly a need for a deeper understanding of this shedding phenomenon, because of the wide variety of cellular responses that syndecans regulate. Moreover, syndecan shedding may be upregulated in tumours, and represents a possible target [12, 19, 20]. In this study, we identified cleavage sites within the ectodomains of syndecan-1 and syndecan-4 by the tumour-associated MMP2, MMP3, MMP7, MMP9 and MT1-MMP. These findings provide a better understanding of syndecan shedding that may lead to new options for diagnosis, and prognosis of disease where MMPs are implicated.

Results

MMP cleavage of syndecan core proteins

To identify precise cleavage sites for MMPs in the syndecan-1 and syndecan-4 core ectodomains, a range of tumour-associated MMPs were incubated with syndecan fusion proteins encompassing the entire ectodomain of the syndecans fused C-terminally with the fibronectin type III 12–13 repeats (FNIII12–13) from the HepII domain of fibronectin. These recombinant proteins could be purified by heparin-affinity chromatography, and, unlike glutathione S-transferase, the heparin-binding domain of fibronectin was not susceptible to any of the enzymes used (Fig. 1).

Figure 1.

Syndecan-1 and syndecan-4 core ectodomains are cleaved by proteinases. The fusion proteins syndecan-1/FN12–13 (A, C) and syndecan-4/FN12–13 (B, C) were subjected to cleavage by MMP2, MMP3, MMP9, MT1-MMP, plasmin, and thrombin. Cleavage by MMP7 is shown separately (C). Polypeptides are marked with coloured dots where cleavage sites were identified by N-terminal sequencing, and the green dots represent those polypeptides whose cleavage site could not be determined with certainty. The last lane in (C) is enzyme only. Molecular mass standards are shown.

Syndecan-1 and syndecan-4 ectodomains were cleaved by all of the MMPs tested, including MT1-MMP, MMP2, MMP3, MMP7, and MMP9 (Fig. 1). The serine proteases thrombin and plasmin were included as positive controls, as their cleavage sites are already known for syndecan-4 [21]. After enzyme treatment and separation by SDS/PAGE, major polypeptide products were subjected to N-terminal sequencing. The first five N-terminal residues of each polypeptide were identified for the unambiguous assignment of cleavage site(s) within the ectodomains (Table 1).

Table 1. N-terminal sequencing of polypeptides. Syndecan-1/FN12–13 and syndecan-4/FN12–13 polypeptides (Fig. 1) were subjected to N-terminal sequencing after proteinase treatment to identify the N-terminal residues, for unequivocal identification of cleavage sites. Where several residues were observed at one position, they were ranked in descending order of abundance. ND, not determined; NDI, not definitively identified
  Protein sequence report: amino acids
Fusion proteins/proteinasesIdentified cleavage sites (matched to known sequence)12345
Syndecan-1/FN12–13 (~ 45 kDa)
MMP2High Gly82-Leu83 (LEATA)

L

S

E

T

ATA
Low Gly82-Leu83 (LEATA)

L

S

E

H

T

A

T

P

A
MMP3Asp236-Gln237 (QGATG)QGATIG
Gly245-Leu246 (LLDIP)LL

D

L

I

D

P

MMP9Gly82-Leu83 (LEATA)LEATA
MT1-MMPGly82-Leu83 (LEATA)LEATA
Gly245-Leu246 (LLDIP)LLDIP
PlasminArg230-Asn231 (NQSPV)NQSPV
ThrombinArg126-Glu127 (ETTQL)ETTQL
Syndecan-4/FN12–13 (~ 34 kDa)
MMP2Asp31-Leu32 (LLEGR)LLE R
Arg36-Tyr37 (YFSGA)YSSGA
Tyr37-Phe38 (FSGAL)FFGA 
Lys105-Leu106 (LEENE)LEENE
MMP3ND     
MMP9Lys105-Leu106 (LEENE)LEENE
MT1-MMPAsp31-Leu32 (LEENE)LLEGR
 Ser130-Met131 (MSSTV)MS

S

E

V

TV
Asn139-Ile140 (IFEIP)IFEIP
PlasminHigh Lys128-Val129 (VSMSS)VSMSS
Low Lys128-Val129 (VSMSS)VSMSS
ThrombinArg36-Tyr37 (YFSGA)

A

Y

F

S

F

M

S

T

G

G

A

L

Lys114-Arg115 (RISPV)NDII

S

M

T

P

G

V

The cleavage sites identified for syndecan-1 were located in close proximity to the transmembrane domain, with the exception of two more N-terminal cleavage sites (Fig. 2). Several of the metalloproteinases shared cleavage sites; for instance, MT1-MMP, MMP7 and MMP3 shared the cleavage site Gly245-Leu246 on syndecan-1 core protein. As seen with syndecan-1, most syndecan-4 cleavage sites were also located in close proximity to the transmembrane domain, and, again, some proteinases shared identical cleavage sites. Additionally, a proteinase-sensitive region just N-terminal from the GAG attachment sites for syndecan-4 was identified (Fig. 3). The two syndecan-4 polypeptides that contained MT1-MMP cleavage sites, Asn139-Ile140 or Ser130-Met131, although migrating as a closely separated dimer, in fact were reversed in relation to their predicted mass (Fig. 1). The Asn139-Ile140 polypeptide fused to FNIII12–13 represents the smaller of the two polypeptides. However, on SDS/PAGE, it resolved with a larger apparent molecular mass than the cleaved Ser130-Met131 fusion protein. This is suggestive of marked changes in conformation arising from membrane proximal cleavage of the syndecan. Furthermore, it was observed that syndecan-4/FN12–13 cleavage (see Experimental procedures) with plasmin yielded two polypeptides that migrated differently, but had exactly the same cleavage sites (Lys128-Val129), on the basis of N-terminal sequencing. This was also observed with MMP2 cleavage of syndecan-1/FN12–13, where two polypeptides that migrated differently but had the same cleavage site (Gly82-Leu83) were identified. In these cases, there may be additional more C-terminal cleavage sites, as is apparent from MMP2 digestion.

Figure 2.

Syndecan-1 species alignment with identified cleavages sites for the human ortholog. Full-length sequences of syndecan-1 are aligned with the subdivision of the ectodomain, transmembrane domain, and cytoplasmic domain. The GAG attachment sites (SG) are specified in bold. The identified cleavage sites for the metalloproteinases and serine proteinases in human syndecan-1 are marked with arrows. Five almost entirely conserved cleavage sites were found; three sites are localized in close proximity to the transmembrane, whereas the remaining two are more distal. The GenBank accession numbers are listed for each sequence.

Figure 3.

Syndecan-4 species alignment with identified cleavage sites for the human ortholog. Full-length sequences of syndecan-4 are aligned with the subdivision of the ectodomain, transmembrane domain, and cytoplasmic domain. The GAG attachment sites (SG) are specified in bold. The identified cleavage sites in human syndecan-4 for the metalloproteinases and serine proteinases are marked with arrows. Eight highly conserved cleavage sites were identified, and some were shared among the proteinases: three cleavage sites localized in close proximity to the putative transmembrane domain, two sites localized more distally from the transmembrane domain, and three sites near the GAG chain attachment sites. The GenBank accession numbers are listed for each sequence.

Cleavage sites conserved across species

To determine whether the identified cleavage sites are conserved across species, the cleavage sites of the human syndecans obtained here were aligned with selected mammalian syndecan-1 and syndecan-4 species sequences. Highly conserved cleavage sites were present in syndecan-4, apart from the Lys105-Leu106 cleavage site for MMP2, MMP7, and MMP9. This site is conserved between human and monkey, but, for pig, mouse and rat, the basic Lys is replaced by an acidic Glu (Fig. 3). The identified cleavage sites for human syndecan-1 are much less conserved, except for the entirely conserved plasmin cleavage site at Arg230-Asn231 (Fig. 2). In the case of cleavage by MT1-MMP, MMP7 and MMP3 at Gly245-Leu246, in pig and mouse there is a conservative change to Ser245. For thrombin cleavage at Arg126-Glu127 in human, rat, and mouse, the Arg126 is replaced by Ser or Thr in pig and monkey, respectively (Fig. 2). For the Gly82-Leu83 site, cleaved by four MMPs (MT1-MMP, MMP2, MMP7, and MMP9), there are considerably different sequences in other species, although the N-terminal residue 82 is always Gly or Ser. Similarly the MMP3 site at Asp236-Gln237 always has a conserved Asp, whereas the C-terminal side of the cleavage site varies.

For MT1-MMP, cleavage occurred between a hydrophilic and a hydrophobic residue in each case (for syndecan-1 at Gly245-Leu246 and Gly82-Leu83, and for syndecan-4 at Ser130-Met131 and Asn139-Ile140). In the case of syndecan-1, both MT1-MMP cleavage sites were between Gly-Leu dipeptides. However, these sites are not conserved across mammalian species (Fig. 2).

MMP cleavage sites were localized to distinct core protein regions

We observed that the membrane proximal cleavage sites for cleavage by MMPs were located exactly six and 15 residues from the transmembrane domain for both syndecan-1 and syndecan-4 (Fig. 4). The Leu246 (MMP3, MMP7, and MT1-MMP) and Ile140 (MT1-MMP) sites in syndecan-1 and syndecan-4, respectively, are located six residues from the putative transmembrane domain. Gln237 (MMP3) and Met131 (MT1-MMP) of syndecan-1 and syndecan-4, respectively, are located 15 residues from transmembrane domain. In addition, plasmin-sensitive sites were 17–21 residues from the transmembrane domain. Additional MMP2 and MMP9 cleavage sites were located 35–40 residues C-terminal from the GAG chain attachment sites in both syndecan-1 and syndecan-4 (Fig. 4). Therefore, the pattern of cleavage sites for MMPs, plasmin and thrombin appears to be common to both syndecan-1 and syndecan-4 ectodomains. This suggests that there may be structural properties of the core protein that similarly expose conserved proteinase-sensitive sites.

Figure 4.

Localization of cleavage sites on the syndecan ectodomains. A schematic overview of the proteinase cleavage sites on the syndecan ectodomains is shown. The indicated numbers represent the number of residues between cleavage sites, the putative transmembrane domain, or the GAG attachment site. The proteinases cleave the syndecan ectodomains in the same pattern. The membrane-proximal MMP cleavage sites for syndecan-1 and syndecan-4 are spaced identically, six and 15 residues from the transmembrane domain. Furthermore, MMP2 and MMP9 cleavage sites are located ~ 35–40 residues C-terminal from the GAG chain attachment sites for both syndecan-1 and syndecan-4.

Single mutations block MT1-MMP cleavage of syndecan-1 and syndecan-4

To confirm the MT1-MMP cleavage sites in vitro, site-specific mutagenesis of the hydrophobic residues on the C-terminal side of the cleavage sites was carried out. For the syndecan-1 cleavage site Gly245-Leu246, the hydrophobic Leu was mutated to Asn, and this mutated protein was termed syndecan-1-GN/FN12–13. This single-residue mutation in syndecan-1-GN/FN12–13 eliminated the ~ 24-kDa polypeptide after cleavage with MT1-MMP. Correspondingly, a new polypeptide was apparent, of ~ 45 kDa (Fig. 5). For a negative control, FNIII12–15 was used, as it is resistant to all of the proteinases used in this study (Fig. 5), and carboxymethylated transferrin (Cm-Tf) was used as a positive control for metalloproteinase activity [20]. The GN mutation also blocked MMP7 cleavage, and, similarly to what was found for MT1-MMP cleavage, a new polypeptide of similar size was generated (Fig. 5).

Figure 5.

Single mutations in syndecan ectodomains can render them resistant to MT1-MMP and MMP7.The syndecan-1/FN12–13 and the mutated syndecan-1-GN/FN12–13 (A, C) fusion proteins were incubated with (+) or without (–) MT1-MMP or MMP7, and visualized on a 12% SDS/PAGE gel with Coomassie Brilliant Blue R-250. Similarly, syndecan-4/FN12–13 (B, D) fusion proteins, the syndecan-4-NN/FN12–13 and syndecan-4SQ/FN12–13 single mutants and the double mutant syndecan-4-SQNN/FN12–13 were also treated with the two MMPs. FNIII12–15 and Cm-Tf were included as negative and positive controls, respectively. Polypeptides arising from defined cleavage sites are denoted by arrowheads and labelling. The red, purple and green arrowheads highlight the polypeptides from cleavage sites that were subject to mutation. When Gly245-Leu246 in syndecan-1 was mutated, the protein was resistant to cleavage by both enzymes. In syndecan-4, the two MT1-MMP cleavage sites could both be rendered resistant by a single mutation. Only the Asn139-Ile140 site was cleaved by MMP7, and this was also rendered resistant by a single mutation (D). The final lane on some gels contains the MMP only.

For the syndecan-4 cleavage sites, Ser113-Met114 and Asn122-Ile123, the Met and Ile residues were mutated to Gln and Asn, respectively. In the former case, Gln was selected to avoid generating an N-glycosylation motif. In total, three mutated syndecan-4 constructs were prepared, with a single mutation for each of the cleavage sites (syndecan-4-SQ/FN12–13 and syndecan-4-NN/FN12–13), as well as a double mutation (syndecan-4-SQNN/FN12–13). The SQ mutation blocked the MT1-MMP cleavage site at Ser113-Met114 for syndecan-4, whereas cleavage at Asn122-Ile123 was still present. The reverse pattern was seen for mutation at Asn122-Ile123. The double mutation blocked cleavage of syndecan at both sites (Fig. 5). The NN mutation also blocked the MMP7 cleavage site at Asn122-Ile123 for syndecan-4 (Fig. 5). Consistent with the lack of MMP7 cleavage at Ser113-Met114, the SQ mutation had no effect on its cleavage of the syndecan-4 fusion protein (Fig. 5). This last experiment also showed that mutation at a site close to an identified cleavage site had no impact on cleavage at that site. Similarly, cleavage by MMP7 at Lys105-Leu106 in syndecan-4 and Gly82-Leu83 in syndecan-1 was maintained when mutations at other sites had been introduced. Therefore, the conformation and organization of the syndecan fusion proteins were not compromised by the introduction of single or double mutations.

The catalytic activity of MT1-MMP is not influenced by GAG chains

The activity of several soluble MMPs has been reported to be altered by exogenous GAGs [22, 23], but this is not known to be the case for MT1-MMP. As syndecans usually bear multiple GAG chains, it was important to ascertain whether MT1-MMP activity could be modified by their presence. In vitro experiments were performed to evaluate the influence of the different GAG chains on the activity of the MT1-MMP catalytic domain. A Cm-Tf cleavage assay was used to assess the catalytic activity of MT1-MMP in the presence of heparin, HS, CS or DS. However, none of the GAGs had a detectable effect on the catalytic activity of MT1-MMP over a range of concentrations and times (Fig. S1).

Discussion

Although our knowledge of the mechanisms underlying syndecan function is still sparse, it is already clear that shed syndecan has implications for wound healing, cancer, and bacterial and viral pathogenesis [9, 24]. Increased levels of shed syndecan-1 are present in dermal wound fluid [25] and sera of of patients with some cancer types [26, 27] to a much greater extent than in healthy individuals [28]. Perhaps the best evidence for the importance of shedding in cancer has been shown for syndecan-1 in myeloma. Soluble syndecan-1 is present at high levels in the sera of myeloma patients, and has been demonstrated to be an indicator of poor prognosis [28-32]. These reservoirs of syndecan-1 may play a critical role in promoting myeloma pathogenesis, as soluble syndecan-1 has been shown to promote growth of myeloma tumours in vivo [30]. To investigate MMP cleavage sites in syndecan-1 and sydecan-4 in vitro, suitable fusion proteins were required. We utilized the heparin-binding domain of FNIII12–13, as it is protease-resistant, is stable, and can be purified with heparin affinity chromatography [33, 34]. All investigated MMPs (MMP2, MMP3, MMP7, MMP9, and MT1-MMP) and serine proteinases (thrombin and plasmin) in this study cleaved human syndecan-1 and syndecan-4 in vitro. It has not been reported previously that MMP3 can cleave syndecan-1 and syndecan-4, or that MT1-MMP can cleave syndecan-4. It seems likely that most MMPs will share an ability to cleave syndecans, and they are also known to be sensitive to ADAM and ADAMTS proteinases [35-37]. Moreover, most of these cleavage sites are conserved across mammalian species.

The syndecan core proteins have small clusters of proteinase-sensitive sites in their ectodomains. However, it appears that MMPs release the core protein with the GAG chains as an intact unit, as, for both syndecan-1 and syndecan-4, there was no cleavage between the Ser residues known to be sites of GAG substitution. Therefore, the soluble ectodomains could be effective competitors for membrane-associated syndecan. In addition, protein-binding motifs within the ectodomain of syndecans do not seem to be affected by shedding. The conserved NXIP(87–90) ectodomain motif of syndecan-4 is located between the two proteinase-sensitive regions that we identified. This motif mediates indirect association with β1-integrin to promote cell adhesion [38]. Similarly, the two recorded binding motifs in syndecan-1 are preserved upon MMP cleavage. Residues 82–130, corresponding to what is now known as synstatin [39], interact directly with both αVβ3 and αVβ5 integrins [39-41]. Another syndecan-1 motif, AVAAV(222–226), has also been suggested to mediate cell invasion [42]. As MMPs release the ectodomain with the GAG chains and ectodomain core protein-binding properties preserved, it may be that shed syndecan ectodomains can be retained at the cell surface. For example, Li et al. [11] showed that MMP7 shedding of syndecan-1 regulates chemokine mobilization and transepithelial efflux of neutrophils in acute lung injury. It will be interesting to determine whether other functional motifs in the syndecan ectodomains may be affected by proteinases. For example, Whiteford et al. [43] showed syndecan-2 to have an adhesion regulatory domain (Pro124–Phe141) located in close proximity to the putative transmembrane domain (Val145–Tyr169). This region was shown to bind the orphan protein tyrosine phosphatase receptor CD148, promoting downstream β1-integrin-mediated adhesion and cytoskeletal organization. This association might be affected by MMP cleavage, as it corresponds to the proteinase-sensitive region identified in both syndecan-1 and syndecan-4.

Prior to this study, only four cleavage sites had been documented for human syndecans. MT1-MMP cleaves human syndecan-1 at Gly82-Leu83, and Gly245-Leu246 [17]. Plasmin and thrombin cleave human syndecan-4 at Lys114-Arg115/Lys192-Val130 and Lys114-Arg115, respectively [21]. We confirmed these sites, with the exception of plasmin cleavage at Lys114-Arg115. A novel thrombin cleavage site for syndecan-4 was identified at Arg36-Tyr37. It had not been reported previously whether plasmin and thrombin cleave other members of the syndecan family. However, we identified plasmin and thrombin cleavage sites for syndecan-1, and novel cleavage sites in human syndecan-1 for MMP2, MMP3, and MMP9. For human syndecan-4, novel metalloproteinase sites were identified for MMP2, MMP9, and MT1-MMP.

Interestingly, for both syndecan-1 and syndecan-4, we identified two hotspots where cleavage sites clustered, six and 15 residues N-terminal from the transmembrane domain. Additionally, MMP2 and MMP9 cleaved at sites ~ 35–40 residues C-terminal from the GAG chain attachments site in both syndecan-1 and syndecan-4. It would be interesting to determine how MMPs are able to access and cleave syndecans at sites very close to the cell surface.

Several soluble MMPs bind HS chains, including MMP2, MMP7, and MMP9, and their proteinase activity can be modulated by this. The HS of syndecan inhibited the activity of MMP2 [22], which might mean that this enzyme has reduced ability to cleave at the three sites immediately N-terminal from the first HS substitution site in syndecan-4 (Ser39). On the other hand, heparin can increase MMP7, MMP9 and MMP13 activity [23]. Heparin can also inhibit the activity of heparanase, whose function is to selectively cleave the HS chains [44]. Here, we tested whether heparin, HS, DS or CS had the potential to inhibit MT1-MMP activity. Our in vitro experiments showed no inhibition over a range of concentrations and times. However, the MT1-MMP used in these assays comprised only the catalytic domain (Tyr112–Gly284). MT1-MMP also comprises a pre/propeptide (Met1–Arg111), a hinge/linker region (Glu285–Ile318), a haemopexin domain (Cis319–Cys508), a stalk region (Pro509–Ser538), a transmembrane domain (Ala539–Phe562), and a cytoplasmic tail (Arg563–Val582) [45]. MT1-MMP cleavage of type I collagen at the cell surface requires haemopexin-dependent dimerization [46, 47], and the effects of GAGs on this process were not investigated here. The implication is, however, that dimerization of the proteinase is not required for it to cleave syndecan ectodomains.

In conclusion, it appears that syndecan ectodomains are sensitive to a wide array of metzincin enzymes, and that shedding is therefore a major component of syndecan biology. Moreover, most proteinases cleave the core proteins at more than one site. The extent to which this process terminates or alters syndecan signalling remains largely unknown, but is highly relevant, given the many cases where shedding is implicated in disease pathogenesis. Future structural studies on the syndecans may reveal the nature of the foci for cleavage that are identified here. Where syndecan shedding may become a therapeutic target, it is clearly a complex issue, given the high sensitivity of syndecans to a range of proteinases. No single enzyme or cleavage site is likely to be an effective target.

Experimental procedures

Reagents

All chemicals were from Sigma-Aldrich, unless stated otherwise.

HS from bovine kidney, CS A from bovine trachea and heparin from bovine intestinal mucosa were all from Sigma. CS B (DS) from pig skin was purchased from Seikagaku Corporation (Tokyo, Japan). MMP3 was provided by J. Jacobsen (University of Copenhagen, Denmark), MMP7 was provided by L. Matrisian (Vanderbilt University), and MMP2, MMP9, thrombin and plasmin were all from CalbioChem. Recombinant MT1-MMP (MMP-14) catalytic domain (Tyr112–Gly284) was kindly provided by Y. Itoh (Kennedy Institute for Rheumatology, Oxford, UK). The pQE30 plasmid containing FNIII12–15 was from J. Schwarzbauer (Princeton University). Cm-Tf is a positive control substrate for MMPs, as the modified protein is susceptible to most of these proteinases [48]. Human transferrin was S-carboxymethylated under reducing conditions with iodoacetic acid, as described previously [49]. After trichloroacetic acid precipitation, the Cm-Tf was dissolved in distilled H2O and adjusted to pH 7.0–8.0 by titration with sodium hydroxide.

Recombinant fusion proteins

Syndecan recombinant fusion protein constructs were generated by overlap PCR, whereby the heparin-binding domain of rat fibronectin (NM_019143) type III repeats 12–13 (Ile1812–Thr1991) were directly fused to the C-terminus of human syndecan-1 (NM_002997.4, Gln23-Asp248) or syndecan-4 (NM_002999.3, Glu19–Glu142). These constructs are abbreviated as syndecan-1/FN12–13 and syndecan-4/FN12–13. Subsequent mutations in syndecan-1/FN12–13 and syndecan-4/FN12–13 constructs were generated, whereby the identified MT1-MMP cleavage sites on syndecan-1 and syndecan-4 core protein were altered by the use of overlap PCR approaches with primers incorporating mutations. For syndecan-1, the MT1-MMP cleavage at Gly245-Leu246 was removed by mutation of Leu246 to Asn (abbreviated as syndecan-1-GN/FN12–13). For syndecan-4, MT1-MMP cleavage sites at Ser130-Met131 and Asn139-Ile140 were deleted by mutation of Met131 and Ile140 to Gln and Asn, respectively. These are abbreviated as syndecan-4-SQ/FN12–13 and syndecan-4-NN/FN12–13, respectively. Double-mutated syndecan-4/FN12–13, including both cleavage site mutations, was also prepared (abbreviated as syndecan-4-SQNN/FN12–13). Constructs were inserted into the pET24a vector (Novagen, Merck KGaA, Darmstadt, Germany) for expression in BL21 Escherichia coli, and all DNA sequences were confirmed. Purification from bacterial lysates was performed by affinity chromatography on heparin Sepharose 6 fast flow resin (GE Healthcare, Piscataway, NJ, USA) equilibrated in 50 mm sodium phosphate (pH 6.0), with salt gradient elution from 0 to 1.0 m NaCl.

Enzyme treatments and N-terminal sequencing

MMP, thrombin and plasmin digestions of syndecan-1 ectodomain/FN12–13 and syndecan-4 ectodomain/FN12–13 were performed in TNC buffer (150 mm NaCl, 10 mm CaCl2, 0.05% Brij35, and 50 mm Tris/HCl, pH 7.5) for 4 h to overnight at 37 °C, and analysed by 12% SDS/PAGE and Coomassie Brilliant Blue R-250 detection. No significant differences in cleavage patterns were seen when short and long enzyme incubation periods were compared. Enzyme/substrate ratios of 1 : 10 to 1 : 20 were commonly used, e.g. where N-terminal sequencing was required. For N-terminal sequence analysis of enzyme-digested syndecan-1 and syndecan-4 ectodomains, the products were resolved by 12% SDS/PAGE, and transferred to a poly(vinylidene difluoride) membrane, with 10 mm Caps buffer (pH 11) containing 10% methanol. The membranes were stained for 30 s in 0.02% Coomassie Brilliant Blue R-250, 40% ethanol, and 5% acetic acid, and this was followed by 1 min of destaining in 40% ethanol and 5% acetic acid, and several rinses with pure water. Chosen polypeptides were excised from the membrane and subjected to N-terminal sequencing (Cambridge Peptides, UK).

MT1-MMP catalytic activity

To test whether GAGs affect MT1-MMP catalytic activity, increasing concentrations (3–100 μg/mL) of CS, DS, HS or heparin were premixed with 7 μg of Cm-Tf in TNC buffer containing 50 mm Tris/HCl, 150 mm NaCl, 10 mm CaCl2, and 0.05% Brij35, before addition of 7 μg/mL MT1-MMP. The mix was incubated at 37 °C from 1 h to overnight, and the entire volume was then resolved by 12% SDS/PAGE, and stained with Coomassie Brilliant Blue R-250.

Acknowledgements

Support from the Danish National Research Foundation (0522-77MN), Lundbeck Fonden (R44-A4407), Novo Nordisk Fonden (UrBA) and the Department of Biomedical Sciences, University of Copenhagen, is gratefully acknowledged. T. Manon-Jensen was supported by a co-financed PhD Fellowship from the Copenhagen Graduate School of Health Sciences. We thank Y. Itoh (Kennedy Institute for Rheumatology, Oxford, UK) for providing the MT1-MMP catalytic domain.

Ancillary