(Received for publication, November 4, 1996, and in revised form, January 27, 1997)
From the Department of Vascular Biology, The Scripps Research Institute, La Jolla, California 92037
The ligand binding functions of vitronectin (Vn) are regulated by its conformational state/degree of multimerization. In the native plasma form of Vn, the C-terminal glycosaminoglycan (GAG) binding domain is believed to be cryptic. Here, evidence is provided that the addition of fucoidan or dextran sulfate to unfractionated plasma results in the formation of covalently and non-covalently stabilized Vn multimers. These multimers express conformationally sensitive antibody epitopes and ligand binding sites located in the N terminus of the Vn molecule. While heparin forms complexes with monomeric plasma Vn and induces conformational changes, a reduction in ionic strength is required for induction of multimerization. In addition, heparin serves as a template for the assembly of type 1 plasminogen activator inhibitor-induced disulfide-linked Vn multimers. These results support a new model for the structure of native Vn. The C-terminal GAG binding domain is predicted to be exposed in the native conformation, whereas the N terminus is cryptic. Ligand binding to the GAG binding site unfolds the N terminus, thereby exposing cryptic ligand binding sites.
Vitronectin (Vn),1 an adhesive
glycoprotein present in the circulatory system and in a variety of
tissues, belongs to a group of molecules that play key roles in the
attachment of cells to their surrounding matrix (1, 2). In addition, Vn
also regulates the complement, coagulation, and fibrinolytic system.
During the last few years, it has been increasingly appreciated that Vn
is a conformationally labile molecule. For example, only 2% of the Vn
in plasma is in a conformation capable of binding to heparin-Sepharose, whereas the relative amount increases to 7% by the generation of serum
(3). In addition, the formation of complexes of Vn with type 1 plasminogen activator inhibitor (PAI-1) (4), thrombin-antithrombin III
(5), and complement C5b-C9 (6) induces conformational changes and
heparin binding. Moreover, conformational changes in the Vn molecule
can also be induced by denaturation with chaotropic agents, heat
treatment, and acidification (1, 2). These changes are accompanied by
the spontaneous formation of disulfide-linked Vn multimers (7). Plasma
Vn and Vn purified under non-denaturing conditions (i.e.
native Vn) are monomeric and lack exposure of conformationally
sensitive epitopes (8). In contrast, Vn purified from serum by heparin
affinity chromatography in the presence of 8 M urea
(i.e. denatured Vn) is highly multimerized and
conformationally altered (8, 9). A second circulatory pool of Vn is
contained in platelet -granules (10). Platelet Vn is present in both monomeric and multimeric forms, and the latter form expresses epitopes
for conformationally sensitive monoclonal antibodies (mAbs, see Ref.
8). Thus, platelets provide a physiological correlate for the
experimentally observed multimerization of Vn.
Defining the biological functions of Vn has been complicated, because a
number of its ligand binding domains are differentially expressed
according to the conformation of this molecule. A number of ligands
were identified that interact preferentially with the conformationally
altered, denatured form of Vn. These ligands include collagen (11),
GAGs (12-14), -endorphin (15), PAI-1 (16), urokinase
receptor-urokinase complex (17), and cells (18). In contrast,
thrombin-antithrombin III complexes appear to preferentially interact
with native Vn (19), and PAI-1 appears to bind to both forms of Vn
(20).
The observations that binding of plasma Vn to heparin is increased upon ligand binding or treatment with protein denaturants led to the concept that the GAG binding domain (located between amino acids 341 and 379) is, at least partially, cryptic in the native conformation (1, 2). However, conformational changes in the Vn molecule upon denaturation are apparently not limited to the GAG binding domain, but also occurred in the somatomedin B (SMB) domain (amino acids 1-51) and the connecting region/first hemopexin-like repeat (amino acids 52-239) (9). Moreover, the epitope for a mAb that blocks the high affinity interaction of PAI-1 with the SMB domain of Vn (i.e. mAb 153) was not expressed in plasma Vn (8). These observations raise the possibility that both the GAG binding domain and N-terminal ligand binding sites are cryptic in the native Vn molecule. However, most ligand binding events in Vn have been mapped to either of these domains. Despite these observations, plasma Vn binds ligands that are reported to interact with the N terminus. How could these apparently conflicting observations be explained considering that both binding sites were not available in the native form? Here, we provide evidence that sulfated GAGs bind to native plasma Vn, resulting in Vn multimerization and conformational changes, thereby exposing cryptic N-terminal ligand binding sites. Taken together, these observations imply that plasma Vn is GAG binding-competent.
Denatured Vn was purified by
heparin affinity chromatography in the presence of 8 M urea
(21). PAI-1 was isolated and activated with guanidinium hydrochloride
(22), dialyzed against 5 mM sodium phosphate, 1 mM EDTA, 0.345 M NaCl, pH 6.5, to stabilize the
biological activity of PAI-1 (23), and the specific activity was
determined by titration against urinary-type plasminogen activator
(Calbiochem) (22). Prior to each experiment, PAI-1 was dialyzed into
ice-cold phosphate-buffered saline (PBS). The Vn polypeptides
encompassing amino acids 1-51 and 52-239 were expressed in
Escherichia coli strain BL 21 and purified by nickel
affinity chromatography from sonicated cell lysates according to the
manufacturer (Novagen). Briefly, fragments of the human Vn cDNA
were polymerase chain reaction-amplified using the following polymerase
chain reaction primer containing BamHI sites at both the 5
and 3
end and ligated into the BamHI-digested expression
plasmid using standard cloning procedures (24). 5
primer 1-51: 5
-TGC
ATC ATG GAT CCC GAC CAA GAG TCA TGC AAG-3
; 3
primer 1-51: 5
-GTT TCA
GGA TCC TTA CAT AGT GAA CAC ATC CCC-3
; 5
primer 52-239: 5
-TGC ATC
ATG GAT CCG GAG GAT GAG TAC ACG-3
; 3
primer 52-239: 5
-AAC GTT TCA
GGA TCC TTA GTC CAG GAC ACC ATC CTC-3
. The orientations of the inserts were determined by polymerase chain reaction using two oligonucleotide primers based in the insert and the parental plasmid, respectively, and
the sequences were confirmed by double-stranded DNA sequencing. mAbs
153 and 1244 were obtained using standard hybridoma technology (22),
and IgG was produced in mice as ascites fluid and purified by using
protein A-Sepharose. Protein concentrations were determined by the
bicinchoninic acid method (Pierce). Dextran sulfate (average Mr 8,000), fucoidan (average
Mr 50,000), heparin sodium salt from porcine
intestinal mucosa (170 USP units/mg), low molecular weight heparin from
porcine intestinal mucosa (average Mr 3,000),
and chondroitin sulfate A were obtained from Sigma, dissolved at 50 mg/ml in PBS, and stored at
80 °C. Protamine sulfate was obtained from Calbiochem, dissolved at 25 mg/ml in PBS, and stored at
80 °C. Platelet-poor plasma (PPP) was prepared under conditions
aimed at preventing platelet activation as described (8). The Vn concentration in PPP was estimated by semiquantitative immunoblotting (8).
PPP (2 µl) was incubated (37 °C) with the indicated concentration of GAGs and/or PAI-1, and the volume was adjusted to 50 µl with PBS. At the end of the incubation period (2 h), the samples were immediately fractionated by gel electrophoresis and analyzed by immunoblotting using mAb 1244 (22). For some experiments, the GAG-treated PPP was analyzed by gel filtration chromatography using a Sephacryl S-300 superfine column (125 × 1.5 cm). The Vn concentration in the column fractions was determined by sandwich ELISA as described (4). The mAb used in the ELISA is conformationally sensitive. The fractions were heat-treated (55 °C, 1 h) to completely denature Vn, thereby avoiding differential recognition of Vn conformers by the ELISA (4).
Immunological and Functional Characterization of Vn MultimersConformational changes in the Vn molecule were quantified by competitive ELISA essentially as described (9). As a modification, microtiter wells were coated (5 µg/ml) with recombinant Vn fragments encompassing the immunoepitope of the mAbs (i.e. Vn amino acids 1-51 for mAb 153, and Vn amino acids 52-239 for mAb 1244). Results are expressed as percentage binding of antibodies in the presence of PPP (10 µg/ml) in the absence of PAI-1 or GAGs. The binding of Vn to PAI-1 was determined by competitive ELISA (22). Briefly, SMB polypeptides coated (see above) and blocked wells were co-incubated (1 h, 37 °C) with activated PAI-1 (0.6 nM) and PPP (10 µg/ml Vn; preincubated (1 h, 37 °C) with the indicated concentration of GAGS). Bound PAI-1 was detected using rabbit anti-human PAI-1 IgG (22). This assay system detects only the binding of active forms of PAI-1, whereas latent PAI-1 or plasminogen activator·PAI-1 complexes fail to interact with the immobilized SMB polypeptide (not shown). Control experiments revealed that in the concentration range used, GAGs did not interfere with mAb or PAI-1 binding to immobilized ligands or remove the immobilized ligand from the microtiter wells.
The
effects of highly sulfated GAGs on the structure of plasma Vn was
characterized. Under non-reducing conditions, plasma Vn migrated as a
broad band between Mr 75,000 and 65,000, and heparin fails to alter the mobility of plasma Vn on SDS-PAGE (Fig. 1, lane 1; compare with Fig. 3, lane
1). However, analysis by native Vn reveals that heparin shifts the
mobility on a native PAGE in a dose-dependent manner. The
first shift in the mobility is observed at 1 µg/ml heparin, and at
100 µg/ml, all plasma Vn is shifted to the cathode (Fig. 1,
lanes 2-5). Analysis by gel filtration chromatography
reveals that plasma Vn incubated with 1 mg/ml heparin elutes close to
albumin, indicating that heparin fails to induce higher molecular
weight Vn forms (not shown). It should be noted that the resolution of
the S-300 column is not sufficient to discriminate between Vn monomers
(Mr 75,000) and Vn monomer-heparin complexes
(expected average size 95,000). Analysis of the Vn elution fractions by
native PAGE (Fig. 1, lane 6) reveals that the
electrophoretic mobility of the Vn is similar to heparin-treated plasma
(compare Fig. 1, lane 5). This observation raised the
possibility that a heparin-Vn monomer complex is formed that is stable
during size exclusion chromatography. Protamine sulfate dissociates
heparin-protein complexes. To test whether the shift of Vn mobility on
native PAGE was due to non-dissociated Vn monomer-heparin complexes,
PPP was incubated with heparin alone (Fig. 1, lane 7), or
with heparin followed by the addition of protamine sulfate (Fig. 1,
lane 8). Protamine sulfate reverses the heparin-induced
shift in Vn mobility. This observation suggested that Vn
monomer-heparin complexes are formed that are sensitive to protamine
sulfate. The shift in the Vn mobility on native PAGE is not limited to
standard heparin, but is also observed in the presence of LMW heparin
(Fig. 1, lane 9).
Heparin Induces Vn Multimerization at Low Ionic Strength
We
recently reported (25) that increases in ionic strength prevented
heat-induced multimerization of native Vn. Here, we tested whether Vn
is more prone to multimerize upon reduction of the ionic strength.
Plasma Vn incubated in the absence of heparin remains monomeric (Fig.
2, lane 1). However, in the presence of heparin, reduction of the NaCl concentration results in the formation of higher molecular weight Vn forms (Fig. 2, lane 2). The
higher molecular weight forms of Vn are sensitive to reducing agents, indicative of disulfide-linked Vn multimers (not shown). Heparin also
induces the formation of higher Mr Vn forms that
fail to enter the separating gel on native PAGE (not shown). To confirm and extend these observations by an independent assay system, PPP was
incubated in the presence of heparin under reduced ionic strength
(i.e. 75 mM NaCl) and fractionated by size
exclusion chromatography (Fig. 2B). In initial experiments,
the possibility that the combination of heparin and reduction of ionic
strength shifts the overall protein elution profile was tested. The
protein elution profile (A280) was unchanged
under reduced ionic strength (Fig. 2B) or at physiological
NaCl concentrations (compare Fig. 3). These initial
experiments excluded a generalized effect of the experimental
conditions on the mobility of plasma protein on size exclusion
chromatography. In the following experiments, the effect of heparin and
reduction in ionic strength on the elution of Vn was evaluated. The
majority of the Vn immunoreactivity is shifted to higher
Mr fractions (Fig. 2C, closed
squares), whereas Vn co-elutes as a single symmetric peak with
albumin in the absence of heparin (open squares). Selected
gel filtration fractions were also analyzed by SDS-PAGE under
non-reducing conditions. While the starting material contains both
multimeric and monomeric Vn (Fig. 2A, lane 3),
the high molecular weight fractions consist of SDS-stable Vn multimers
(Fig. 2A, lane 4), whereas the lower molecular
weight fractions contain Vn monomers (Fig. 2A, lane 5). Thus, the gel filtration experiments confirmed the gel
electrophoretic results, indicating that in the presence of heparin and
reduced ionic strength, higher molecular weight Vn forms are
generated.
Dextran Sulfate and Fucoidan Induce Vn Multimerization
The results obtained with heparin were compared with fucoidan and dextran sulfate, two GAGs which have an increased sulfate content in comparison to heparin. Increasing amounts of fucoidan were added to a constant amount of plasma at physiological ionic strength, and after incubation for 2 h at 37 °C, the resulting samples were analyzed by SDS-PAGE under non-reducing (Fig. 3A, lanes 1-3) or reducing (Fig. 3A, lanes 4 and 5) conditions, followed by immunoblotting using mAb 1244. Addition of fucoidan to plasma (Fig. 3A, lane 1) results in a dose-dependent appearance of high molecular weight Vn immunoreactive bands (Fig. 3A, lanes 2 and 3) when analyzed by SDS-PAGE under non-reducing conditions. The molecular mass of the predominant high molecular weight Vn form was estimated to be 150 kDa. The size of this band is consistent with being a Vn dimer. The importance of covalent interactions for the formation of the high molecular weight Vn forms was evaluated. Under reducing conditions, plasma Vn migrates as a doublet of Mr 75,000 and 65,000 on SDS-PAGE (Fig. 3A, lane 4). The addition of fucoidan has no effect on the electrophoretic mobility of plasma Vn under reducing conditions (Fig. 3A, lane 5). Thus, fucoidan-induced high molecular weight Vn forms are sensitive to reducing agents. Electrophoresis under non-denaturing conditions was employed to understand whether the Mr 75,000-65,000 monomers detected in non-reducing SDS gels are actually Vn monomers or non-covalently associated SDS-sensitive Vn multimers. While plasma Vn is detected as a single major band that migrated to the anode (Fig. 3A, lane 6), the addition of fucoidan results in a dose-dependent disappearance of monomeric Vn and a shift of the immunoreactivity toward the cathode (Fig. 3A, lanes 7 and 8). Vn is completely shifted to the cathode on the native PAGE, but only a relatively small portion is present in SDS-stable disulfide linked forms even at the highest concentration of fucoidan employed. These observations raise the possibility that fucoidan induces the formation of non-covalently associated Vn multimers that are only partially stabilized by disulfide-exchange between Vn molecules. Alternatively, the change in electrophoretic mobility on native PAGE may be due to non-dissociated Vn-fucoidan complexes.
Molecular sieving chromatography was employed to discriminate between these two possibilities. Plasma was incubated for 2 h at 37 °C in the absence or presence of 1 mg/ml fucoidan and applied to a gel filtration column. The protein elution profiles (i.e. A280) are very similar in the presence or absence of fucoidan (Fig. 3B). The Vn content in the column fractions was determined by specific ELISA (Fig. 3C). The elution of plasma Vn is practically completely shifted to higher molecular weight fractions, and extended to the void volume (Mr 2 × 106) of the Sephacryl S-300 column. In contrast, plasma incubated in the absence of fucoidan eluted with the expected molecular weight of approximately 67,000, indicative of monomeric Vn (Fig. 3C). The size of the fucoidan-induced Vn multimers on size exclusion chromatography is higher than that expected for monomeric Vn-fucoidan complexes (i.e. Mr 120,000). Thus, fucoidan apparently induces non-covalently and covalently stabilized Vn multimers.
The induction of Vn multimerization under physiological ionic strength
is not limited to fucoidan. Dextran sulfate induces the formation of
higher molecular weight Vn forms on non-reducing SDS-PAGE in a
dose-dependent manner (Fig. 4A,
lanes 2 and 3). Again, these multimers are
stabilized by disulfide bonds as revealed by sensitivity to reducing
agents (not shown). Analysis by native PAGE reveals that dextran
sulfate shifts in a dose-dependent manner the majority of
the Vn to the cathode (Fig. 4A, lanes 4-6).
Again, the majority of Vn is shifted to the cathode on native PAGE,
whereas only limited amounts of SDS-stable Vn multimers were
detectable. Analysis by gel filtration chromatography revealed that the
overall protein elution profile is unchanged in the presence or absence of dextran sulfate (Fig. 4B). In contrast, the Vn elutes in
two pools from the gel filtration column in the presence of dextran sulfate. The high molecular weight pool extends to the void volume, whereas the lower molecular weight peak co-incited with the albumin elution peak (Fig. 4C). Analysis of the high molecular
weight peak by native PAGE reveals the presence of Vn multimers that failed to enter the separating gel (Fig. 4A, lane
7), whereas the lower molecular weight peak contains traces of Vn
monomers and higher molecular weight Vn forms that enter the separating gel (Fig. 4A, lane 8).
Sulfated GAGs Induce the Expression of Conformationally Sensitive Antibody Epitopes in Plasma Vn
Multimerization of Vn, as well as
complex formation with molecules of the coagulation, complement, and
fibrinolytic cascade, results in the exposure of immunoepitopes in the
Vn molecule that are not present in the monomeric plasma form (8). To
determine whether the Vn multimers induced by GAGs contain these
neoepitopes, plasma Vn was compared with Vn incubated with GAGs by
competitive ELISA using mAbs that are derived to these conformationally
sensitive sites (Fig. 5). It should be noted that the
competitive binding assays were performed with immobilized Vn fragments
containing the respective antibody epitope. This modified procedure
prevents interference of the assay by GAGs (presumably due to the
binding of GAG·Vn complexes to the C-terminal GAG binding domain of
intact immobilized Vn; not shown). Fucoidan (closed circles)
and dextran sulfate (open circles) induce the exposure of
the mAb 153 epitope (located between amino acids 1 and 51 (22)) in a
dose-dependent manner (Fig. 5A). At higher
concentrations, heparin (closed squares) and LMW heparin
(closed triangles) also induce the exposure of this
neoepitope, whereas chondroitin sulfate has little effects. In
comparison to the mAb 153 epitope, GAGs have rather limited effects on
the exposure of the mAb 1244 epitope (located between amino acids 52 and 239 (22)). For example, heparin (not shown) and LMW heparin (not
shown) do not stimulate the exposure of the mAb 1244 epitope, and
fucoidan (closed circles) and dextran sulfate (open
circles) have a rather limited effect in comparison to the mAb 153 epitope (Fig. 5B; note the different scale on the
y-axis in comparison to panel A). Thus, highly
sulfated GAGs induce the exposure of N-terminal immunoepitopes in Vn
that are not present in the native plasma form of this molecule.
Sulfated GAGs Induce High Affinity PAI-1 Binding to the SMB Domain of Vn
To test whether the observation that highly sulfated GAGs
induce the exposure of cryptic antibody epitopes in the SMB domain can
be extended to physiological Vn ligands, the effects of different GAGs
on the high affinity interaction of active PAI-1 and the isolated SMB
domain was tested in competitive binding assays. Plasma was
preincubated with a dose response of GAGs and tested for the
competition of the binding of active PAI-1 to the immobilized SMB
domain (Fig. 6A). Sulfated GAGs induce the
exposure of the high affinity PAI-1 binding domain in plasma Vn in a
dose-dependent manner, although to a different extent.
While fucoidan (closed circles) and dextran sulfate
(open circles) are approximately equally potent, heparin
(closed squares) and LMW heparin (open squares)
are significantly less effective in exposing the PAI-1 binding site in
the SMB domain (Fig. 6A). In contrast, chondroitin sulfate
(closed triangles) has no significant effect. Moreover, heparin (open triangle), or fucoidan, dextran sulfate, and
LMW heparin (not shown) have no effect on the binding of denatured Vn
to PAI-1 (Fig. 6A). These observations suggest that ligand binding to the C-terminal GAG binding domain regulates N-terminal ligand binding events.
The interaction of heparin with proteins is mainly due to electrostatic interactions and thus is reversible by an increase in ionic strength (26). This consideration suggests that NaCl should reverse the heparin-induced conformational changes in Vn. To test this hypothesis, unfractionated plasma was first incubated with heparin, followed by a dose response of NaCl, and the competition for the binding of active PAI-1 to immobilized SMB polypeptide was determined (Fig. 6B). Increases in ionic strength reverses heparin-induced PAI-1 binding in a dose-dependent manner, and at the highest concentration of NaCl, approaches that of untreated plasma. Control experiments reveal that the addition of GAGs or NaCl in the concentration range employed in this study has no effect on the binding of PAI-1 or mAbs to the immobilized polypeptides (not shown).
Heparin Functions as a Template for the Assembly of Covalently Linked PAI-1-induced Vn DimersWe recently reported that the
addition of active PAI-1 to unfractionated PPP induces the formation of
disulfide-linked Vn dimers (4). Moreover, heparin also induces
conformational changes in Vn, thereby promoting high affinity binding
to active PAI-1 (Fig. 6). In some biological interactions
(e.g. formation of thrombin-antithrombin III complexes),
heparin not only induces conformational changes, but also serves as a
template for the assembly of these complexes. The latter approximation
mechanism due to the heparin binding function of both antithrombin III
and thrombin is characterized by a biphasic heparin dose response.
While at low and intermediate concentrations of heparin complex
formation is stimulated, high concentrations result in reduced
association between the molecules. The possibility that heparin may
also serve as a template for the assembly of PAI-1-induced
disulfide-linked Vn dimers was tested. Active PAI-1 was incubated with
native Vn present in plasma and an extended dose response of heparin
(Fig. 7). PAI-1 in the absence of heparin induces the
formation of disulfide-linked Vn dimers (Fig. 7, lane
2) as previously reported (4). At high concentrations of heparin,
the PAI-1 induced Vn dimer formation is practically completely blocked
(Fig. 7, lane 3). Intermediate concentrations of heparin
increase the formation of disulfide-linked Vn dimers (Fig. 7,
lanes 4-7), whereas low concentrations of heparin are ineffective (Fig. 7, lane 8). This biphasic dose-response
curve is typical for a template mechanism of heparin and thus is
compatible with the idea that heparin functions as a template for the
assembly of disulfide-linked Vn dimers.
The results presented in this study suggest a new model for the
folding of native plasma Vn and its multimerization upon ligand binding
(summarized as a minimal model in Fig. 8). The first
issue in this model is related to the exposure of the GAG binding
domain in native Vn. Currently, it is believed that this positively
charged domain is cryptic in the native conformation (reviewed in Refs. 1 and 2). However, this concept is already questioned by the published
observations that heparin increases the mAb 8E6 epitope expression
(located between amino acids 52 and 239 (22) in preparations of
purified native Vn (5), that partially purified plasma Vn binds to
heparin Sepharose under low ionic strength (7), and that plasma Vn
functions as a non-competitive inhibitor of the heparin-stimulated
reactions of antithrombin III (27). Furthermore, this study provides
evidence that highly sulfated GAGs induce Vn multimerization at
physiological ionic strength and that the electrophoretic mobility of
plasma Vn on native PAGE is quantitatively shifted by heparin in a
reversible fashion. Moreover, GAGs stimulates the expression of the mAb
153 epitope and high-affinity ligand binding events (i.e.
PAI-1) to the SMB domain, again suggesting that plasma Vn is capable of
interacting with GAGs. Taken together, these observations suggest that
native plasma Vn has the potential to bind to GAGs.
Four considerations may account at least in part for the lack of GAG binding of plasma Vn observed in a number of previous studies. First, heparin affinity chromatography experiments of unfractionated plasma could have been performed under conditions when heparin was the limiting factor. Under these conditions, the experiment determines the competition of plasma Vn with other heparin-binding proteins, like antithrombin III and histidine-rich glycoprotein, for heparin binding, rather than the affinity of heparin for Vn itself. Consistent with this assumption, reports vary between 2 and 20% binding of Vn present in unfractionated plasma to heparin-Sepharose (3, 12). Secondly, increased binding of Vn present in unfractionated urea-treated serum to heparin-Sepharose has been viewed as evidence that urea-dependent unfolding and multimerization of Vn is required for heparin binding. However, the apparent increased binding may be due to decreased or destroyed heparin affinity of many plasma proteins, thereby reducing the extent of competition for limited heparin binding sites on the column. Thirdly, urea treatment of serum induces formation of Vn multimers. Even under the assumption that the affinity of a single GAG binding domain within the Vn multimer remains unchanged in comparison to the monomer, the avidity of the multimer for a single heparin molecule, which is believed to contain more than one GAG binding site (28), should be drastically increased in comparison to the Vn monomer. This would result in selective retainment of Vn on the heparin column. Fourthly, Vn contains two consensus heparin binding sequences in the extended GAG binding domain located between residues 341 and 379. More specifically, heparin binding consensus sequences have been localized between amino acids 347-352 and 354-362 (26). Molecular modeling studies revealed that both elements in this domain form distinct areas of contact with heparin that wrap around and fold over a heparin octasaccharide (26). Thus, a single heparin molecule is expected to interact with more than one binding site in monomeric Vn, again increasing the avidity of the interaction. It is easily foreseeable that parts of the binding sites in heparin immobilized to a solid-phase support are blocked due to stearic hindrance or reduced flexibility, thus effectively reducing the avidity for Vn. It should be noted that GAG binding to native Vn apparently does not require a plasma cofactor, since similar shifts in the electrophoretic mobility of native Vn in the presence of heparin and fucoidan were also observed in a purified system (not shown).
The second issue in this model is related to the exposure of the SMB
domain in plasma Vn. A number of studies suggest that this site is
indeed cryptic in native Vn. Firstly, we recently reported that the mAb
153 binding site (epitope located between amino acids 1 and 51) is only
weakly or not expressed in plasma Vn (8), but strongly expressed in
urea-treated native Vn (9), denatured Vn (8), PAI-1-induced Vn
multimers (4), or after GAG binding (this study). Secondly, kinetic
studies revealed that the interaction of native Vn with active PAI-1 is
at least one order of magnitude lower than that with denatured Vn (16).
This observation suggests that the high affinity PAI-1 binding site in
the SMB domain of Vn is not readily available in plasma Vn. Thirdly,
the principal site for cell attachment in Vn (RGD sequence, amino acids
45-47) is located in the SMB domain, and multimeric conformationally
altered Vn is more potent in promoting cell adhesion (18). Moreover, we
provided evidence that native Vn fails to bind to purified glycoprotein
IIb/IIIa and v
3, whereas multimerization of native Vn, induced either by heat treatment or binding to PAI-1, induces integrin binding competency.2 Taken
together, these observations support the model depicted in Fig.
8A suggesting that the SMB domain is cryptic in native plasma Vn, whereas the GAG binding domain is readily available to
solution-phase ligands.
The addition of sulfated GAGs to unfractionated plasma increased the expression of an immunoepitope (i.e. mAb 153) and high affinity active PAI-1 ligand binding events to the N-terminal SMB domain. Vn contains a single GAG binding domain (29). Thus, these observations suggest that ligand binding to the C-terminal GAG binding domain induces conformational changes in the native Vn molecule, resulting in the expression of the N-terminal SMB domain. Thus, interactions with C-terminal ligand binding sites modulate N-terminal ligand binding events. These observations raise the possibility that an unfolded intermediate is formed upon GAG interaction with the GAG binding domain. The predicted unfolded intermediate is depicted as a monomer in the cartoon (Fig. 8, panel B), but it remains unclear at present whether this intermediate is a conformationally altered monomer or reversibly associated multimer. The formation of this unfolded intermediate appears to be, at least in the case of heparin, reversible. This conclusion is based on the observations that increases in ionic strength after heparin binding to native plasma Vn resulted in reduced N-terminal high affinity PAI-1 ligand events. Also, neutralization of heparin by protamine sulfate reversed the shift in the electrophoretic mobility of Vn on native PAGE. In addition, Vn purified by heparin affinity chromatography under low ionic strength is present in the native conformation (7).
The third issue in this model is related to the temporal events of disulfide-linked versus non-covalently associated multimerization. We previously reported (4) that the non-covalently stabilized PAI-1-induced Vn multimerization is complete within 1 h of incubation, whereas the formation of disulfide bonds increased up to 24 h. Moreover, PAI-1 (4) and dextran sulfate and fucoidan (this study) completely shifted the mobility of Vn on the native PAGE, whereas the extent of disulfide-linked Vn multimers was rather limited. In addition, fucoidan completely shifted Vn immunoreactivity to higher molecular weight fractions on the size-exclusion chromatography, whereas again Vn was only partially present in disulfide-linked multimers. Taken together, these observations suggest that initially non-covalently associated Vn multimers are formed, whereas the covalent stabilization of Vn multimers is a relatively slow secondary event. However, the formation of disulfide bonds requires a close approximation between two Vn molecules within non-covalently associated Vn multimers. The easiest explanation for this observation is that Vn multimers present a higher order assembly of non-covalently associated Vn dimers that are then secondarily stabilized by intermolecular disulfide bonds. This model is depicted in Fig. 8, panel C. It remains unclear whether Vn dimers are formed by parallel or anti-parallel assembly of Vn monomers.
The fourth issue in this model is related to the question of how
ligands with a high affinity for the SMB domain (e.g. PAI-1) can interact with native plasma Vn. At least two possibilities should
be considered to account for this apparent discrepancy. Firstly, plasma
Vn may be in a dynamic equilibrium between an N-terminal-folded and
unfolded conformation. If this model is correct, mAb 153 should bind to
the SMB domain in the unfolded state. The affinity of mAb 153 for Vn is
relatively high (kd 109 mol/liter
(22)), and thus the binding should shift the equilibrium between the
folded and unfolded forms. This would predict that plasma Vn should
compete to an appreciable extent with the binding of mAb 153 to
immobilized denatured Vn. However, this was not observed experimentally
(8) suggesting that this model is not valid. Secondly, a two-site
interaction with Vn could account for the high affinity ligand binding
to plasma Vn. For example, the GAG binding domain has been identified
as a low affinity PAI-1 binding site (30). This observation raises the
possibility that low affinity interactions of PAI-1 with the GAG
binding domain can replace heparin, thereby inducing an unfolded
intermediate and facilitating high affinity PAI-1 binding events to the
N-terminal SMB domain. Further support for this concept is derived from
the competition experiment using overlapping synthetic peptides derived from the GAG binding domain of Vn (30). These experiments indicated that the structural requirements in the GAG binding domain for binding
to heparin and PAI-1 are very similar (30), pointing to the possibility
that PAI-1 can replace heparin in this function.
Heparin is known to exert dual actions on the thrombin/antithrombin III system (28). First, it binds to antithrombin III and by that induces conformational changes resulting in increased reactivity with thrombin. Secondly, heparin serves as a template for the assembly of thrombin-antithrombin III complexes and also PAI-1·thrombin complexes (20). Here, evidence is provided that heparin exerts similar actions on the PAI-1/Vn interaction. It induces conformational changes in the Vn molecule, leading to the exposure of the N-terminal high affinity PAI-1 binding site. Moreover, heparin also serves as a template for the PAI-1-induced assembly of disulfide-linked Vn dimers. It should be noted that the template mechanism was already observed at pharmacological concentrations of heparin, suggesting that heparin therapy can modulate the PAI-1/Vn interaction in vivo.