Platelet Factor 4 Binds to Vascular Proteoglycans and Controls Both Growth Factor Activities and Platelet Activation*

Platelet factor 4 (PF4) is produced by platelets with roles in both inflammation and wound healing. PF4 is stored in platelet α-granules bound to the glycosaminoglycan (GAG) chains of serglycin. This study revealed that platelet serglycin is decorated with chondroitin/dermatan sulfate and that PF4 binds to these GAG chains. Additionally, PF4 had a higher affinity for endothelial-derived perlecan heparan sulfate chains than serglycin GAG chains. The binding of PF4 to perlecan was found to inhibit both FGF2 signaling and platelet activation. This study revealed additional insight into the ways in which PF4 interacts with components of the vasculature to modulate cellular events.

Platelet factor 4 (PF4), 3 or chemokine (CXC motif) ligand 4 (CXCL4), is a 7.8-kDa protein synthesized by megakaryocytes and packaged in the ␣-granules of platelets as a tetrameric complex bound to serglycin GAG chains (1). Upon platelet activation PF4 is released into the circulation where it has roles in both inflammation and wound healing, and these activities are thought to be modulated through interactions with the GAG chains of proteoglycans.
PF4 has been found in the blood vessel wall within minutes following removal of the endothelium and platelet attachment to the denuded basement membrane (2). It binds to endothelial cells through integrins ␣v␤3, ␣v␤5, and ␣5␤1 (3) and also to thrombomodulin on the endothelial cell surface through its chain (4). PF4 inhibits heparin binding growth factors, including FGF2 and VEGF165, binding to their receptors through both heparin-dependent and -independent mechanisms (5-7) with downstream effects on endothelial cell migration and proliferation (5,8). PF4 promotes pro-coagulant activities by preventing the formation of a stable heparin⅐anti-thrombin III⅐thrombin ternary complex (9).
The presence of heparin⅐PF4 immune complexes suggests that PF4 has a high affinity for heparin and that it is not a normally occurring complex in the blood stream. Heparin-induced thrombocytopenia can occur via both non-immune and immune mediated complications of heparin administration in some patients (10). Immune-mediated heparin-induced thrombocytopenia is caused by the formation of heparin⅐PF4 complexes that are recognized as foreign, resulting in the formation of antibodies against the complex. These antibodies bind to the heparin⅐PF4 complex and platelets themselves, increasing coagulation and reducing levels of circulating platelets.
Serglycin is an intracellular proteoglycan produced by hematopoietic cells including neutrophils, mast cells, macrophages, and platelets (11), as well as cells of non-hematopoietic origin including smooth muscle and endothelial cells (12,13). The name serglycin was derived from the serine-glycine repeat structure where up to eight GAG chains can decorate the protein core of ϳ18 kDa. The type and sulfation of GAG chains that decorate serglycin are cell type-dependent mechanisms in which connective tissue mast cells decorate serglycin with heparin (14), whereas mucosal tissue mast cells decorate serglycin with CS (15), neutrophils and platelets decorate serglycin with CS, and macrophages decorate serglycin with CS and HS (16).
Serglycin is located in the ␣-granules of platelets where it is bound to many cytokines and proteases, including PF4, until they become activated (15). Serglycin knock-out mice exhibit a reduced platelet aggregation response and reduced levels of platelet ␣-granule proteins including PF4, ␤-thromboglobulin, and platelet-derived growth factor that are thought to be related to defects in packaging and secretion of ␣-granule proteins, including proteases, that are required for platelet activation (17).
Perlecan is an HS proteoglycan present in the basement membrane that binds and supports the signaling of heparin binding growth factors including FGFs and VEGFs (18,19). In addition, perlecan supports platelet adhesion via the integrin ␣2␤1 (20, 21) but does not activate platelets (22).
This study demonstrated that platelet serglycin is decorated with CS and DS and that PF4 binds to the CS/DS chains. Additionally, PF4 has a higher affinity for perlecan HS chains than serglycin GAG chains and can be competed from complexes with serglycin in the presence of perlecan. The binding of PF4 to perlecan inhibits both FGF signaling and platelet activation.

Results
Characterization of Platelet-derived Serglycin-Peptide identification of proteins present in the proteoglycan-enriched platelet extract by mass spectrometry identified two cell surface proteoglycans, betaglycan and syndecan-1, as well as the intracellular proteoglycan serglycin (Table 1). This fraction also contained PF4, indicating that it may be closely associated with the proteoglycans in this fraction.
Platelet-derived serglycin was analyzed by Western blotting and detected as a smear between apparent relative molecular mass (M r ) of 75,000 and more than 250,000, which corresponded to the expected size of serglycin decorated with GAGs ( Fig. 1A, first lane). It should be noted that the protein standards migrate faster than glycosylated proteins; therefore the M r reported are estimates. Treatment of the serglycin with chondroitinase (C'ase) ABC to digest GAG chains containing both CS and DS disaccharides resulted in a reduction of the M r of the smear to ϳ25,000, corresponding to the protein core indicating that platelet-derived serglycin contained ϳ45,000 -225,000 CS and/or DS (Fig. 1A, second lane). Treatment of the serglycin with C'ase ACII that digests only terminal CS disaccharides resulted in a shift in M r of the platelet-derived serglycin in the range of 70,000 to more than 250,000, but not the generation of the protein core (Fig. 1A, third lane). Similarly C'ase B only removes terminal DS disaccharides and resulted in a shift in M r of the platelet derived serglycin to 65,000 to more than 250,000, but not the generation of the protein core (Fig.  1B, fourth lane). Because C'ases ACII and B are exoglycosidases, it indicated that the GAG chains that decorated this form of serglycin contained both CS and DS disaccharides (Fig. 1B). This would account for the small shift in M r when treated with either C'ase ACII or B, because once a terminal disaccharide was encountered that the enzyme could not digest, then digestion would cease. Treatment of the sample with heparinase III (HepIII) indicated that the platelet-derived serglycin was not decorated with HS because there was no change in M r compared with the undigested sample (Fig. 1A, first and fifth lanes).
Proteoglycan-enriched platelet extract was also analyzed by ELISA for the presence of serglycin (Fig. 1C). Low levels of serglycin were detected in the undigested sample; however, upon digesytion with either C'ase ABC or C'ase B, significantly higher (p Ͻ 0.05) levels of serglycin were detected, indicating that CS and DS disaccharides decorated the protein core of serglycin and that these GAG chains hindered the access of the antibody to the epitopes present on the protein core. Treatment of the proteoglycan-enriched platelet extract with HepIII did not alter the level of serglycin detected compared with the undigested sample, confirming that platelet-derived serglycin was not decorated with HS (Fig. 1C).
Serglycin was found to contain the 4-sulfated stub structure after C'ase ABC digestion, but not C'ase B digestion, indicating that only CS disaccharides were present close to the protein core and were decorated with 4-sulfated CS stubs (Fig. 1D). Unand 6-sulfated CS stubs were not detected after digestion with either C'ase ABC or B (data not shown). The CS chain structures present in the proteoglycan-enriched platelet extract were reactive with the CS antibody clone CS-56 (Fig. 1E), indicating the presence of CS disaccharides with the sequence GlcUA-GalNAc(4S)-GlcUA(2S)-GalNAc(6S) (23). There was a background level of reactivity with the CS chain antibody clones LY-111 and MO-225 (data not shown).
PF4 Binds to CS/DS and HS-The types of GAG chains attached to serglycin that were involved in the binding of PF4 were investigated using the proteoglycan-enriched platelet extract. Proteoglycan-enriched platelet extract was probed for the presence of PF4 and found to contain immunoreactive bands as a smear between 70,000 and more than 250,000, which corresponded to the expected M r of serglycin ( Fig. 2A, first lane 1). Additionally, PF4 was detected at M r 8,000, corresponding to its monomer form. PF4 was present at M r 8,000 in the extract following either C'ase ABC, ACII, or B digestion, indicating that PF4 was released from the proteoglycans in the extract following digestion of the CS/DS chains ( Fig. 2A, second, third, and fourth lanes). Together with the data presented in Fig. 1A, which demonstrated a minimal level of GAG digestion when either C'ase ACII or B was used alone, this indicated that the majority of PF4 bound to the non-reducing end of the GAG chains.
To confirm that PF4 was bound to serglycin via its CS/DS chains, a sandwich ELISA was performed where a serglycin polyclonal antibody was used to immobilize serglycin from the proteoglycan-enriched platelet extract and a monoclonal PF4 antibody was used to detect the presence of PF4 in the fraction or in the fraction following C'ase ABC, ACII, or B digestion (Fig.  2B). PF4 bound to serglycin in the undigested extract, whereas the levels of PF4 bound to serglycin were significantly (p Ͻ 0.05) reduced following digestion with either C'ase ABC, ACII or B, indicating that PF4 was bound to platelet serglycin via CS/DS chains.
To further explore the interaction between PF4 and serglycin, recombinantly expressed serglycin decorated with CS/DS, HS/heparin (24) was used. Recombinant serglycin did not contain PF4 (Fig. 3A, lane 2) when compared with a positive control of PF4 (Fig. 3A, lane 1). Incubation of recombinant serglycin with PF4 resulted in complexes with M r of 100,000 to greater than 250,000, as well as PF4 at M r of 8,000 when analyzed by Western blotting with a PF4 antibody (Fig. 3B, lane 1), indicating similar-sized complexes were formed as for platelet-derived serglycin with PF4. Incubation of recombinant serglycin with PF4 followed by digestion with either C'ase ABC or B resulted in a MOWSE score as determined by Mascot query. This is the value (P) that is a measure of the probability that the match is a random event expressed as Ϫ10log(P).
The higher the score, the more confidence that the match is not due to a random event.
a reduction in level of complexes with M r of 100,000 to greater than 250,000 and the presence of PF4 monomers at M r 8,000 ( Fig. 3B, lanes 2 and 4). Incubation of recombinant serglycin with PF4 followed by digestion with either C'ase ACII or HepIII resulted in a broader range of complexes with M r of 50,000 to greater than 250,000 and the presence of PF4 monomers at M r 8,000 (Fig. 3B, lanes 3 and 5). Incubation of recombinant serglycin with PF4 followed by digestion with both C'ase ABC and HepIII resulted in the presence of PF4 monomers at M r 8,000 (Fig. 3B, lane 6). These data indicated that PF4 bound to the recombinant serglycin predominantly via by CS, DS, and HS/heparin. These data also indicated differences in the binding of PF4 to recombinant serglycin compared with platelet serglycin because of the presence of the HS/heparin. A sandwich ELISA was also performed following incubation of recombinant serglycin with PF4 and GAG digestion. This assay revealed that PF4 bound to the recombinant serglycin via both CS and DS, with DS involved to a significantly greater (p Ͻ 0.05) extent than CS (Fig. 3C). Additionally, PF4 was found to bind to the HS/heparin chains attached to the recombinant serglycin.
PF4 Binds to HS Proteoglycans with a Higher Affinity than CS Proteoglycans-The ability of PF4 to bind to HS was further explored using endothelial-derived perlecan that was exclusively decorated with HS and as perlecan is abundant in the vascular basement membrane. PF4 bound to endothelial-de-rived perlecan to a significantly greater (p Ͻ 0.05) extent than recombinant serglycin when coated at the same concentration ( Fig. 4A). PF4 did not bind to endothelial perlecan that had been treated with HepIII to remove its HS chains (Fig. 4A), indicating that PF4 only bound to the HS that decorated the protein core. PF4 bound to recombinantly expressed perlecan domain V via both its CS and HS chains (Fig. 4B).
Surface plasmon resonance studies indicated that PF4 bound to immobilized heparin, perlecan, and serglycin in a dose-dependent fashion (Fig. 5). The association rate was similar for each condition with equilibrium reached rapidly followed by a slow dissociation (Fig. 5, A-C). The equilibrium constants were 25, 100, and 50 nM for heparin, serglycin, and perlecan, respectively, indicating that PF4 had the highest affinity for heparin followed by perlecan and serglycin (Fig. 5D). Control experiments performed with perlecan predigested with Hep III to remove HS and serglycin predigested with both C'ase ABC and HepIII to remove CS/DS and HS gave baseline sensorgrams indicating that growth factor binding was only to the GAG chains (data not shown). Analysis of the level of PF4 binding to heparin (Fig. 5A) indicated that one PF4 tetramer bound per heparin chain. This analysis is not possible for the proteoglycans because the relative contribution of the protein and GAG chains to the response units measured by the instrument cannot be determined. The relative affinity of PF4 for perlecan and serglycin was analyzed in an ELISA assay where serglycin was immobilized on the ELISA plate and incubated with PF4 followed by incubation with perlecan. Perlecan was able to release PF4 from serglycin as detected by a significant (p Ͻ 0.05) reduction in the level of PF4 bound to serglycin after incubation with perlecan compared with without soluble perlecan treatment (Fig. 6). However, when the PF4⅐serglycin complex was incubated with perlecan that had been digested with HepIII to remove its HS chains, there was no change in the level of PF4 bound to serglycin, indicating that the protein core of perlecan did not release PF4 from serglycin. Incubation of the PF4⅐serglycin complex with perlecan domain V was also able to release PF4 from serglycin, and this interaction was found to be dependent on the presence of the HS chains. Treatment of the PF4⅐serglycin complex with GAG digestion enzymes alone under the same conditions had no effect on the level of PF4 bound to serglycin because of the short incubation times used in this assay. This indicated that PF4 had a higher affinity for perlecan HS than serglycin GAG chains.
PF4 Inhibits the Activity of FGF2 Bound to Heparin and Perlecan-The ability of ternary complexes to form between heparin, FGF2, and FGF receptor 1c (FGFR1c) and signal in the presence of PF4 was analyzed in BaF-32 cells transfected with FGFR1c (Fig. 7). Heparin and FGF2 were used as a positive control for the assay, whereas cells in the presence of medium or growth factor alone were used as negative controls. BaF-32 cells expressing FGFR1c responded to FGF2 in the presence of heparin to a level significantly above (p Ͻ 0.05) the medium only control; however, in the presence of PF4 at concentrations of 0.64 -2.56 M, the level of signaling was significantly reduced (p Ͻ 0.05) compared with the positive control (Fig. 7A). Perlecan in the presence of FGF2 was able to signal to a similar extent   2). B, Western blot for the presence of PF4 in fractions of recombinant serglycin incubated with PF4 followed by digestion with either C'ase ABC, ACII, or B. C, sandwich ELISA for PF4 bound to serglycin using a rabbit polyclonal serglycin antibody capture and detected with a mouse monoclonal PF4 antibody. Samples were analyzed without and with C'ase ABC, ACII, or B or HepIII digestion. Individual data points are shown (n ϭ 3); bars indicate the means. * indicates significant differences (p Ͻ 0.05) compared with undigested samples analyzed by oneway ANOVA. ** indicates significant differences (p Ͻ 0.05) for C'ase B-treated samples compared with C'ase ACII-treated samples analyzed by one-way ANOVA.
as the positive control; however, in the presence of PF4, the level of signaling was significantly reduced (p Ͻ 0.05) and at a level comparable to the negative control as well as perlecan in the absence of growth factor (Fig. 7B). These data indicated that PF4 could inhibit the binding and signaling of FGF2 through both heparin and perlecan HS.
PF4 Activates Platelets, Whereas GAG Bound PF4 Does Not Activate Platelets-Platelet activation was determined by measuring P-selectin expression. Type I collagen was used as a positive control for the assay, whereas freshly isolated platelets were used as the negative control (Fig. 8A). Exposure of platelets to PF4 was found to increase P-selectin expression, but not to the same extent as exposure to collagen type I. Exposure of platelets to heparin, endothelial perlecan, or recombinant serglycin did not activate platelets (Fig. 8B). Preincubation of PF4 with perlecan, serglycin, or heparin before exposure to the platelets inhibited the PF4 mediated activation of platelets (Fig.   8C), suggesting that the GAG chains modulated the activity of PF4.

Discussion
This study demonstrated that platelets contain serglycin decorated with GAG chains that consist of both CS and DS disaccharides with 4-sulfated CS closer to the protein core and part of the linkage tetrasaccharide structure. This finding extends previous reports that platelet serglycin is decorated with CS containing 4-sulfated disaccharides (17). . PF4 binds to perlecan via HS. A, ELISA for PF4 bound to recombinant serglycin, endothelial perlecan, or endothelial perlecan predigested with HepIII. B, ELISA for PF4 bound to recombinant perlecan domain V either without treatment or predigested with C'ase ABC, HepIII, or both C'ase ABC and HepIII. Individual data points are shown (n ϭ 3); bars indicate the means. * indicates significant differences (p Ͻ 0.05) compared with undigested samples analyzed by one-way ANOVA. One representative curve (of triplicates) at each concentration is shown. For analysis, all curves were utilized for each concentration. D, equilibrium dissociation constant, K D , for PF4 interacting with heparin, perlecan, and serglycin. FIGURE 6. PF4 has a higher affinity for perlecan and serglycin. ELISA for PF4 that was removed from serglycin after exposure to perlecan or perlecan domain V either untreated or treated with C'ase ABC, HepIII, or both. As controls, the amount of PF4 removed from serglycin without additional treatment or treatment with either C'ase ABC, HepIII, or both was determined. The data are presented as percentages of PF4 removed from serglycin by treatment (individual data points are shown (n ϭ 3); bars indicate the means). * indicates significant differences (p Ͻ 0.05) compared with no treatment analyzed by one-way ANOVA.
PF4 was found to bind to platelet serglycin through both CS and DS as well as perlecan through HS chains. Additionally, PF4 bound to recombinant serglycin via CS, DS, and HS/heparin chains further supporting the affinity of PF4 for a variety of GAG chains. Previous analyses of the affinity of PF4 for various GAG chains revealed that PF4 had the highest affinity for heparin followed by CSD or E, HS, and CSA, B, or C (25,26) and that this binding involved lysine residues in PF4 binding to the sulfate groups on the GAG chains (1). Heparin had the highest affinity for PF4 in this study with a K D of 30 nM that was similar to previous reports in the range of 16 -60 nM (1,27,28). The affinity of PF4 for proteoglycans was explored for the time in this study and indicated that PF4 had a higher affinity for perle-can decorated with HS than serglycin predominantly decorated with CS/DS. PF4 has been established to be a tetrameric complex at physiological pH and ionic strength via NMR and neutron scattering (29,30). Models of these studies support the hypothesis that the heparin chain wraps around the PF4 tetramer (30). Although this arrangement is likely with exposure of PF4 tetramers to single GAG chains, physiologically it is likely that the PF4 will be exposed to serglycin with eight GAG chains and perlecan with three GAG chains. Given the close proximity of the GAG chains on both serglycin and perlecan, there is limited flexibility for the chains to wrap around the PF4 tetramers as is possible with an isolated GAG chain. Thus there is the possibility of multiple GAG chains binding the PF4 tetramers as depicted in Fig. 9; however, this model remains to be experimentally verified.
Following vascular injury, platelets are known to adhere to the denuded vascular basement membrane, particularly to collagen, inducing platelet aggregation, activation, and the release of PF4. PF4 is released from platelets in complex with serglycin (31). Thus the release of PF4 from serglycin GAG chains by perlecan HS, as demonstrated in this study, is a likely mechanism by which PF4 is transferred to the vascular basement  13-2.56 M, A; 2.56 M, B). Cells exposed to medium or the growth factor only were used as negative controls. Individual data points are shown (n ϭ 3); bars indicate the means. * denotes significant differences compared with the negative control analyzed by one-way ANOVA. ** indicates significant differences compared with the heparin ϩ FGF2 condition. membrane as observed in vivo (2). Additionally, under physiological conditions, endothelial cell surface GAG chains bind basal levels of platelet-derived PF4 (32).
PF4 binds to the surface of resting and activated platelets and supports platelet aggregation (31). This study also demonstrated that PF4 promotes platelet activation; however, platelets did not become activated when exposed to PF4 bound to either serglycin or perlecan GAG chains, suggesting that the binding of PF4 to proteoglycans can modulate platelet activation events.
PF4 is reported to be angiostatic as demonstrated in a chicken chorioallantoic membrane assay (8). The angiostatic effect of PF4 has been ascribed to its binding to the same regions of HS as growth factors involved in angiogenesis including FGF2 and VEGF (7). Additionally, PF4 can inhibit the binding of FGF2 to both fibroblast cell ECM and cell surface receptors (5) through binding to HS. The present study supported these findings using the BaF32 cell assay where the addition of PF4 could dose-dependently inhibit the binding and signaling of FGF2 that require the formation of ternary complexes with FGF2, the FGFR1c cell surface receptor, and either heparin or perlecan HS. FGF2 is produced by vascular endothelial cells and acts as a both a mitogenic and migratory signal for these cells. Indeed the addition of PF4 to cultures of endothelial cells inhibited their migration (5). Together these data suggest that both PF4 and FGF2 bind to similar regions of HS to inhibit the HS-dependent signaling of FGF2 via FGF receptors.
PF4 also exerts its angiostatic activity via HS-independent mechanisms as it was able to inhibit VEGF 121 induced cell proliferation but did not interfere with its binding to the VEGF receptor, flk-1 (6). HS-independent mechanisms include binding integrins on the endothelial cell surface (3), inhibiting growth factor binding to receptors and binding directly to FGF2 and preventing dimerization (7).
PF4 preferentially binds to newly formed blood vessels providing a mechanism to control their growth (33). Thus one of the physiological roles of PF4 may be to control normal and neoplastic vascularization through controlling the activities of growth factors. Thus it is hypothesized that the binding of PF4 to various GAG chains that decorate proteoglycans in the vasculature controls angiogenesis in wound healing through modulating the activity of growth factors and platelet activation.
Isolation of Proteoglycans from Platelets-Platelet lysate was prepared by incubating platelets with 1% (w/v) Triton X-100 for 1 h at 4°C followed by filtration with a 0.45-m syringe filter. Anion exchange chromatography using a diethylaminoethyl column (1 ml of DEAE-Sepharose Fast Flow; GE Healthcare) attached to a FPLC (AKTA purifier; GE Healthcare) was used to enrich samples for proteoglycans. Briefly, the DEAE column was equilibrated at 1 ml/min with running buffer (250 mM NaCl, 20 mM Tris, 10 mM EDTA, pH 7.5) before the addition of platelet lysate and baseline absorbance re-established with running buffer. Proteoglycans were eluted using an eluting buffer (1 M NaCl, 20 mM Tris, 10 mM EDTA, pH 7.5). Proteoglycanenriched fractions were subsequently concentrated and analyzed for protein concentration using a Coomassie Blue protein assay (Thermo Scientific, Scoresby, Australia).
Mass Spectrometry-DEAE-enriched samples (50 g/ml) were prepared for LC-MS 2 analysis by in-solution digestion, which involved reduction with 10 mM DTT for 10 min at 95°C and alkylation with 25 mM iodoacetamide for 20 min at 25°C. Samples were then incubated with 20 g/ml sequencing grade Trypsin in 50 mM NH 4 HCO 3 at 30°C for 16 h and then subjected to peptide analysis by LC-MS 2 . Samples were analyzed by LC-MS 2 using an LTQ mass spectrometer (Thermo Fisher Scientific). The results were analyzed with Xcalibur TM software (Bioworks version 3.1; Thermo Fisher Scientific) and the MASCOT database with a National Center for Biotechnology Information protein (homosapian) database.
Glycosaminoglycan Digestion-Samples of proteoglycan-enriched material were digested with 50 milliunits/ml chondroitinase (C'ase) ABC, ACII, or B in 0.1 M Tris acetate, pH 8, at 37°C for 16 h to confirm the presence and structure of the CS. Samples of proteoglycan-enriched material were digested with 10 milliunits/ml HepIII in PBS, pH 7.2, for 16 h at 37°C to confirm the presence and structure of the HS.
Western Blotting Analysis-Proteoglycan-enriched samples (200 g/ml based on Coomassie protein assay), with and without glycosaminoglycan digestion, were electrophoresed in 4 -12% (w/v) BisTris gels (Life Technologies, Sydney, Australia) under non-reducing conditions using MES buffer (50 mM MES, 50 mM Tris, 0.1% (w/v) SDS, 1 mM EDTA, pH 7.3) at 200 V for 45 min. A series of molecular mass markers (Precision Plus All Blue; Bio-Rad) were electrophoresed on each gel. Samples were then transferred to PVDF membrane using transfer buffer (5 FIGURE 9. Schematic of proposed mechanism of PF4 binding to proteoglycan GAG chains. The proposed model involves multiple GAG chains on proteoglycans, including serglycin and perlecan, binding PF4. mM Bicine, 5 mM BisTris, 0.2 mM EDTA, 50 g/ml SDS, 10% (v/v) methanol, pH 7.2) in a semi-dry blotter at 300 mA and 20 V for 60 min. The membrane was blocked with 1% (w/v) BSA in TBS (20 mM Tris base, 136 mM NaCl, pH 7.6) with 0.1% (v/v) Tween 20 (TBST) for 2 h at 25°C followed by incubation with primary antibody diluted in 1% (w/v) BSA/TBST for 2 h at 25°C. Membranes were subsequently rinsed with TBST, incubated with secondary HRP conjugated antibodies (1:50,000) for 45 min at 25°C, and rinsed with TBST and TBS before being imaged using chemiluminescence reagent (Femto reagent kit; Pierce) and X-ray film.
ELISA-Proteoglycan-enriched samples (10 g/ml based on Coomassie Blue protein assay), with and without glycosaminoglycan digestion, were coated onto high binding 96-well ELISA plates (Greiner) for 2 h at 25°C. Wells were rinsed twice with Dulbecco's PBS, pH 7.4 (DPBS), followed by blocking with 0.1% (w/v) casein in DPBS for 1 h at 25°C. Wells were rinsed twice with DPBS with 1% (v/v) Tween 20 (PBST) followed by incubation with primary antibodies diluted in 0.1% (w/v) casein in DPBS for 2 h at 25°C. Wells were rinsed twice with PBST followed by incubation with biotinylated secondary antibodies (1:1000) diluted in 0.1% (w/v) casein in DPBS for 1 h at 25°C, rinsed again twice with PBST, and then incubated with streptavidin-HRP (1:500) for 30 min at 25°C. Binding of the antibodies to the samples was detected using the colorimetric substrate, 2,2Ј-azino-di-(3-ethylbenzthiazoline sulfonic acid), and absorbance was measured at 405 nm.
A sandwich ELISA was performed by coating ELISA plates with a rabbit polyclonal anti-serglycin capture antibody in 0.1 M sodium carbonate buffer, pH 9.6, for 16 h at 4°C. Wells were rinsed twice with DPBS followed by blocking with 0.1% (w/v) casein in DPBS for 2 h at 25°C. Wells were rinsed with PBST followed by incubation with the proteoglycan-enriched samples (10 g/ml based on Coomassie Blue protein assay) for 2 h at 25°C and subsequent detection with primary and secondary antibodies as for the standard ELISA. Data for both the ELISA and sandwich ELISA were corrected for background absorbance.
PF4 Binding Assays-The binding of PF4 to recombinant serglycin prepared as described previously (24), immunopurified endothelial derived perlecan prepared as described previously (35,36), and recombinant perlecan domain V prepared as described previously (37) was assessed by Western blotting, ELISA, and surface plasmon resonance. For Western blotting, PF4 (5 g/ml) was incubated with serglycin (2 g/ml) in solution for 16 h at 37°C and either analyzed without further modification or treated with glycosaminoglycan digestion enzymes prior to analysis.
For ELISA, plates were coated with either serglycin, endothelial perlecan, or perlecan domain V at a concentration of 10 g/ml for 16 h at 4°C prior to blocking with 1% (w/v) casein in PBS for 2 h at 25°C. This was followed by incubation with 1 g/ml PF4 for 2 h at 37°C and then detection with the mouse monoclonal anti-PF4 antibody. The binding of PF4 to the proteoglycans was also analyzed after glycosaminoglycan digestion prior to immobilizing on the ELISA plates. The background level of PF4 binding to casein was also measured, and absorbance values were subtracted from sample absorbance values.
For the displacement assay, ELISA plates were coated with 10 g/ml serglycin that was either untreated or treated with C'ase ABC, HepIII, or both C'ase ABC and HepIII for 16 h at 4°C. The plates were then blocked with 1% (w/v) casein in PBS for 2 h at 25°C followed by incubation with 8 g/ml PF4 for 30 min at 37°C. Plates were then incubated with either 10 g/ml perlecan or 10 g/ml perlecan domain V that was untreated or after glycosaminoglycan digestion for 1 h at 37°C. As a control, selected wells were incubated with 0.05 unit/ml glycosaminoglycan digestion enzymes alone for 1 h at 37°C. The plates were then analyzed for bound PF4 using the mouse monoclonal anti-PF4 antibody by the ELISA technique.
Surface Plasmon Resonance-Heparin (Sigma H3393; 17-19 kDa), recombinant serglycin, and endothelial perlecan were biotinylated as described previously (38,39). The interaction between PF4 and either serglycin or endothelial perlecan or heparin was analyzed with a BiaCore 2000 (GE Healthcare). Streptavidin-coated sensor surfaces (GE Healthcare) were coated by injection of 100 l of 10 g/ml biotinylated heparin, endothelial perlecan, or recombinant serglycin at 5 l/min (diluted in DPBS, pH 7.4). The surfaces were then blocked by injection of 150 l of 1% (w/v) BSA in DPBS at 5 l/min. The binding of PF4 (0.1, 0.5, and 1 M) was conducted in DPBS, pH 7.4, at a flow rate of 20 l/min at 25°C using an injection volume of 50 l. Sensorgrams were analyzed using BIAcore 2000 evaluation software 3.0. Sensorgrams were fitted with differential rate equations for the association and dissociation rates. Control experiments were performed with samples that had been predigested with either C'ase ABC or HepIII and PF4 binding to blocked streptavidin chips.
BaF32 Cell Proliferation Assays-Baf32 cells are from an IL-3-dependent and HSPG-deficient myeloid B cell line that has been stably transfected with FGFR1c (40,41). Baf32 cells are a model system developed to identify heparan sulfate and heparin structures that interact with FGFs and their receptors. The readout of this assay is cell proliferation, which indicates the formation of ternary complexes in situ. Baf32 cells were maintained in RPMI 1640 medium containing 10% (v/v) FBS, 10% (v/v) WEHI-3BD conditioned medium, 100 units/ml penicillin, and 100 g/ml streptomycin. WEHI-3BD cells were maintained in RPMI 1640 medium supplemented with 2 g/liter sodium bicarbonate, 10% (v/v) FBS, 100 units/ml penicillin, and 100 g/ml streptomycin, and the conditioned medium was collected three times per week and stored at Ϫ20°C until required. For the mitogenic assays, the BaF32 cells were transferred into IL-3 depleted medium for 24 h prior to experimentation and seeded into 96-well plates at a density of 2 ϫ 10 4 cells/well in the presence of FGF2 (0.03 nM), heparin (30 nM), or immunopurified endothelial perlecan (4 pM) and PF4 (0.125-2.5 M). The cells were incubated for 96 h in 5% CO 2 at 37°C, and the amount of cells present was assessed using the MTS reagent (Promega, Madison, WI) by adding to the cell cultures for 6 h prior to measuring the absorbance at 490 nm.
Platelet Activation-Platelets were harvested from human donors under ethics approval from the University of New South Wales. Blood was collected in acid citrate dextrose anticoagulant-treated vacutainers. Platelet-rich plasma was prepared by centrifugation of the blood at 350 ϫ g for 20 min at 25°C fol-lowed by careful removal of the upper platelet-rich layer. Platelet-rich plasma was then centrifuged at 1200 ϫ g for 10 min to yield a platelet pellet with platelet poor plasma as the supernatant. The platelet pellet was resuspended in Tyrode's buffer (1.8 mM CaCl 2 , 1 mM MgCl 2 , 2.7 mM KCl, 136.9 mM NaCl, 0.4 mM NaH 2 PO 4 , 11.9 mM NaHCO 3 , 5.6 mM D-glucose, and 0.1 unit/ml apyrase) and centrifuged again at 1200 ϫ g for 10 min. The supernatant was discarded, and the platelets were resuspended in Tyrode's buffer to a concentration of 2 ϫ 10 7 platelets/ml. Platelet suspensions (2 ϫ 10 7 platelets/ml) in Tyrode's buffer were exposed to either PF4 (4 g/ml), collagen type I (30 g/ml), 10 g/ml recombinant serglycin, 10 g/ml endothelial perlecan, 10 g/ml heparin, or mixtures of 10 g/ml serglycin and 4 g/ml PF4, 10 g/ml perlecan and 4 g/ml PF4, or 10 g/ml heparin and 4 g/ml PF4 for 5 min at 37°C in the presence of FITC-labeled anti-CD62P (1 g/ml, clone 9E1; Bio-Scientific Pty. Ltd., Gymea, Australia) detects P-selectin on the surface of activated platelets. The reaction was stopped by adding paraformaldehyde to a final concentration of 1% (w/v). The fluorescence intensity of 10,000 platelets was analyzed using flow cytometry (FACScan; Becton Dickinson).
Statistical Analysis-A one-way analysis of variance (ANOVA) was performed, and the results of p Ͻ 0.05 were considered significant. The experiments were performed in triplicate and repeated.