Insulin-like Growth Factor-binding Protein-3 Binds Fibrinogen and Fibrin*

Following tissue injury, a fibrin network formed at the wound site serves as a scaffold supporting the early migration of stromal cells needed for wound healing. Growth factors such as insulin-like growth factor-I (IGF-I) concentrate in wounds to stimulate stromal cell function and proliferation. The ability of IGF-binding proteins (IGFBPs) such as IGFBP-3 to reduce the rate of IGF-I clearance from wounds suggests that IGFBP-3 might bind directly to fibrinogen/fibrin. Studies presented here show that IGFBP-3 does indeed bind to fibrinogen and fibrin immobilized on immunocapture plates, withK d values = 0.67 and 0.70 nm, respectively, and competitive binding studies suggest that the IGFBP-3 heparin binding domain may participate in this binding. IGF-I does not compete for IGFBP-3 binding; instead, IGF-I binds immobilized IGFBP-3·fibrinogen and IGFBP-3·fibrin complexes with affinity similar to that of IGF-I for the type I IGF receptor. In the presence of plasminogen, most IGFBP-3 binds directly to fibrinogen, although 35–40% of the IGFBP-3 binds to fibrinogen-bound plasminogen. IGFBP-3 also binds specifically to native fibrin clots, and addition of exogenous IGFBP-3 increases IGF-I binding. These studies suggest that IGF-I can concentrate at wound sites by binding to fibrin-immobilized IGFBP-3, and that the lower IGF affinity of fibrin-bound IGFBP-3 allows IGF-I release to type I IGF receptors of stromal cells migrating into the fibrin clot.

Insulin-like growth factor-binding protein-3 (IGFBP-3), 1 one of 6 structurally related IGFBPs that bind IGF peptides with high affinity, is a 29-kDa protein found in body fluids as multiple ϳ40 -50-kDa forms due to differential glycosylation (1)(2)(3). IGFBP-3 is the most abundant serum IGFBP during extrauterine life, where it circulates with one IGF peptide and a single acid-labile subunit in a ϳ150-kDa ternary complex (1,4,5). Expression of IGFBP-3 by many tissues suggests that it is also available locally to modulate the autocrine/paracrine actions of the IGF peptides (1).
IGF-I is a 7.6-kDa protein with mitogenic, metabolic, differentiative, chemotactic, and anti-apoptotic effects (1,6,7). Since IGFBPs such as IGFBP-3 have higher affinity for IGF-I than does the type I IGF receptor, it is not surprising that IGFBP-3 can inhibit IGF action (1,8). However, IGFBPs bound to extracellular matrix may have lower IGF affinity (9 -11). By binding an 18-amino acid heparin binding domain (HBD), which is highly conserved in IGFBP-3 and the closely related protein IGFBP-5, heparin and certain other glycosaminoglycans apparently change the conformation of these IGFBPs, resulting in significantly lower affinity for IGF-I (9,12). In addition, IGFBP-3 proteolysis has been described in many in vivo and in vitro situations, resulting in IGFBP-3 fragments with low affinity for IGFs (1,(13)(14)(15). In these situations, IGF-I may be released from IGFBP-3 to type I receptors on the cell surface with subsequent induction of IGF-I effects (1,2,8,(13)(14)(15)).
The IGF system plays an important role in wound healing (16,17), and both IGF-I and IGFBP-3 are present in wound fluid in significant concentrations (16 -19). Recent studies show that plasminogen (Glu-Pg) binds IGFBP-3 and the binary IGFBP-3⅐IGF-I complex with high affinity by interacting directly with the IGFBP-3 HBD (20). This suggests that Glu-Pg, which plays a crucial role in wound healing and binds to the fibrin clot with high affinity (21)(22)(23), may localize IGF-I to the wound site by binding directly to both IGFBP-3⅐IGF-I complexes and the fibrin clot. However, another possibility is that the IGFBP-3⅐IGF-I complex binds directly to fibrin; this would be reminiscent of the ability of fibrinogen/fibrin to bind with high affinity to basic fibroblast growth factor, another mitogen that plays a role in the wound healing process (24). This paper describes studies which show that IGFBP-3 and IGFBP-3⅐IGF-I complexes do indeed bind with high affinity to fibrinogen/fibrin and to fibrin clots, and suggest that the IGFBP-3 HBD participates in this binding process.

EXPERIMENTAL PROCEDURES
Materials-Recombinant human IGF-I and LR 3 -IGF-I were purchased from GroPep, Ltd (Adelaide, Australia). Recombinant human IGFBP-5 was produced in a baculovirus expression system and purified by affinity chromatography and reverse phase-high performance liquid chromatography (25). Most studies used recombinant human IGFBP-3 expressed in Chinese hamster ovary cells and purified as described previously (20,26). Some experiments used recombinant human IGFBP-3 BV and IGFBP-3hbdBP1 BV produced in a baculovirus expression system and purified as described previously (20,26); IGFBP-3hbdBP1 BV is identical to native IGFBP-3, except that the heparin binding domain has been replaced by the homologous but non-heparin binding sequence of IGFBP-1 (26). IGFBP-1 was purified from human amniotic fluid by affinity chromatography and reverse phase-high performance liquid chromatography (20); quantitation was by radioimmunoassay using a kit from Diagnostic Systems Laboratories, Inc. (Webster, TX). Fibrinogen was obtained from Enzyme Research Laboratories, Inc. (South Bend, IN). Peptide IGFBP-3hbd (KKG-FYKKKQCRPSKGRKR), which encodes the heparin binding domain of IGFBP-3 (20), was synthesized by Genemed Synthesis, Inc. (South San Francisco, CA). Glu-Pg was purified as described previously (20). Plasmin (Pm) was obtained from American Diagnostics (Greenwich, CT). Human citrate plasma was obtained from outdated blood bank supplies or from laboratory personnel.
Fluid-phase Plate Binding Assay-96-well immunological plates (MaxiSorb, Nunc) were coated with 5 g/ml rabbit anti-human fibrinogen antibody (Enzyme Research Laboratories) overnight at 4°C or for 3 h at 23°C in 100 l of 0.1 M Na 2 CO 3 , pH 9.8. Plates were rinsed with PBS and blocked as described in the solid-phase plate assay protocol. In a separate, uncoated 96-well test plate (Sarstedt) various concentrations of fibrinogen, IGFBP-3 BV , 125 I-IGFBP-3, and/or human citrate plasma were incubated in a total volume of 200 l of assay buffer for 1 h at 37°C; this allows all reactants to associate in the fluid phase. At the end of the incubation period, 100 l of reactants were transferred to the blocked and rinsed 96-well plates that were coated with the antifibrinogen antibody. After 30 min at 23°C, unbound radioactivity was removed by rinsing wells twice with assay buffer. Bound radioactivity, representing IGFBP-3⅐fibrinogen complexes, was released by incubating wells with 200 l of 1 M acetic acid for 10 min at 23°C. Acid washes were transferred to 12 ϫ 75-mm glass test tubes and counted for radioactivity. Radioactivity bound in the absence of added fibrinogen served as an estimate of nonspecific binding.
FIG. 1. 125 I-IGFBP-3 binding to fibrinogen/fibrin immobilized on immunocapture plates. Except where noted, plates were coated overnight at 4°C with 5 g/ml fibrinogen. In some experiments, fibrinogen was converted to fibrin as described under "Experimental Procedures." Nonspecific binding was determined by adding unlabeled IGFBP-3 or was estimated by determining the binding of IGFBP-3 to wells in which buffer without fibrinogen/fibrin was absorbed; this value was subtracted from total binding to give specific binding. Each data point is the mean of triplicate determinations. A, 125 I-IGFBP-3 binds to fibrinogen/fibrin. ϳ100,000 cpm of 125 I-IGFBP-3 was equilibrated with either solid-phase fibrinogen or fibrin for 1 h at 37°C. B, competition curve for 125 I-IGFBP-3 binding to fibrinogen. ϳ50,000 cpm of 125 I-IGFBP-3 was equilibrated with solid-phase fibrinogen for 1 h at 37°C either without unlabeled IGFBPs or in the presence of increasing concentrations of IGFBP-1, IGFBP-3, or IGFBP-5. C, 125 I-IGFBP-3 binding to fibrinogen/fibrin is stable but reversible. ϳ50,000 cpm of 125 I-IGFBP-3 was equilibrated with either solid-phase fibrinogen or fibrin for 1 h at 37°C. After unbound 125 I-IGFBP-3 was removed by rinsing, either buffer alone or buffer containing 0.5 M arginine, pH 7.4, was added and the incubations continued at 37°C. At the indicated time points, specific wells were rinsed and bound radioactivity was measured.
Fibrin Clot Binding Assay-Fibrin clots were formed in 1.5-ml Eppendorf tubes by mixing citrate plasma with assay buffer containing CaCl 2 ; total reaction volume was 100 l, final CaCl 2 concentration was 5 mM and, unless stated otherwise, 25 l of citrate plasma were added. After adding ϳ100,000 cpm of 125 I-IGF-I or 125 I-IGFBP-3, either with or without IGFBP-3, the mixture was incubated for 90 min at 37°C. Nonspecific incorporation of trace into fibrin clots was determined in mixtures not containing CaCl 2 . To terminate incubations, fibrin clots were carefully suspended in 500 l of ice-cold assay buffer and then pelleted by centrifuging at 14,000 ϫ g for 2 min. After this wash step was repeated, the vial tips were cut off and then counted for radioactivity.
Biotinylation of IGFBP-3-IGFBP-3 BV (13 g) was diluted in 0.25 M NaHCO 3 , pH 9.2, to a total volume of 45 l. After 1.4 mg of NHS-LCbiotin (Pierce ) was dissolved in 40 l of H 2 O, 5 l of this solution were mixed with the IGFBP-3 BV solution. The mixture was placed on an orbital shaker for 1 h at 22°C, after which the reaction was terminated by adding 200 l of 1 M Tris-HCl, pH 7.4. Unincorporated biotin was separated from biotinylated IGFBP-3 BV (bn-IGFBP-3) using a Quickspin column as instructed by the manufacturer (Roche Molecular Biochemicals). By SDS-PAGE, IGFBP-3 BV in bn-IGFBP-3 was the size expected for the intact native protein. Preparations of bn-IGFBP-3 were stored at 4°C.
Western Ligand Blotting Using bn-IGFBP-3-10 g each of pure fibrinogen, Glu-Pg, and IGFBP-1 were electrophoresed on a 7.5-15% gradient SDS-polyacrylamide gel under nonreducing conditions and then transferred to a nitrocellulose membrane as described previously (20). The membrane was washed for 30 min with 3% Nonidet P-40 in Tris-buffered saline (TBS), blocked for 2 h at 4°C in 0.5% BSA in TBS, washed for 10 min with 0.1% Tween 20 in TBS (TBST), and then incubated overnight with 32 ng/ml bn-IGFBP-3 in TBST. After washing in TBST buffer, the membrane was incubated in streptavidin-horse radish peroxidase diluted in TBST buffer (1:500) for 45 min at 22°C. After several washes in TBST, the membrane was incubated with the chemiluminescent reagants provided in the ECL kit (Amersham Pharmacia Biotech) following the recommendations of the manufacturer. The membrane was then exposed to Kodak film for 5-30 min at 22°C. ECL reagants and streptavidin-horseradish peroxidase were from Amersham Pharmacia Biotech.
Immunohistochemistry-Normal human citrate plasma was clotted by addition of CaCl 2 to a final concentration of 5 mM. After 90 min, the resultant fibrin clot was rinsed and cryosectioned into 5-m sections. Sequential sections were incubated with normal rabbit serum or with rabbit antisera to plasminogen (Roche Molecular Biochemicals) or IG-FBP-3 (Diagnostic Systems Laboratories, Inc.). Immunofluorescent staining was performed using fluorescein isothiocyanate-labeled don-key anti-rabbit IgG as the second antibody (Jackson Immunoresearch Laboratories, West Grove, PA).
IGFBP-3 also bound fibrinogen by ligand blot. Fibrinogen, Glu-Pg, and IGFBP-1 were electrophoresed by SDS-PAGE,  transferred to a nitrocellulose membrane, and then incubated with bn-IGFBP-3. As shown in Fig. 2, bn-IGFBP-3 bound as expected to the ϳ86-kDa Glu-Pg protein but not to IGFBP-1. In addition, bn-IGFBP-3 bound to at least two proteins of mass Ͼ200 kDa in the fibrinogen lane; the size of these proteins is consistent with the expected size of the common forms of plasma fibrinogen (24,28).
Role of the IGFBP-3 HBD in Fibrinogen/Fibrin Binding-Since the IGFBP-3 HBD participates in binding to plasminogen, prekallikrein, and the IGF⅐acid-labile subunit binary complex (4,20,26), it seemed likely that the IGFBP-3 HBD participates in fibrinogen binding. To test this hypothesis, heparin and a synthetic 18-amino acid peptide encoding the IGFBP-3 HBD were evaluated for their ability to compete with 125 I-IGFBP-3 for binding to fibrinogen/fibrin. Both IGFBP-3hbd (Fig. 3A) and heparin (data not shown) interfered in a dose-dependent manner with binding of 125 I-IGFBP-3 to immobilized fibrinogen or fibrin. In addition, IGFBP-3hbdBP1 BV , a full-length IGFBP-3 protein mutated to replace the HBD with the homologous but non-heparin binding region of IGFBP-1, had a ϳ10-fold lower affinity for fibrin than did native IGFBP-3 BV (Fig. 3B).
IGF-I⅐IGFBP-3 Complex Binds Immobilized Fibrinogen/Fibrin-Coincubating 125 I-IGF-I and increasing amounts of unlabeled IGFBP-3 in plates coated with fibrinogen or fibrin resulted in a steady increase in 125 I-IGF-I binding to these plates (Fig. 4A).
Similar results were noted when increasing amounts of unlabeled IGFBP-3 were incubated with immobilized fibrinogen or fibrin prior to the addition of 125 I-IGF-I (data not shown). In each of these experiments, very little 125 I-IGF-I bound to the fibrinogen-or fibrin-coated plates in the absence of IGFBP-3, and coincubating 125 I-IGF-I with 100 ng/ml IGFBP-3 did not result in specific 125 I-IGF-I binding if plates were not coated with fibrinogen or fibrin; these results suggest that 125 I-IGF-I was binding to fibrinogen-or fibrin-bound IGFBP-3.
To determine the affinity of IGF for IGFBP-3 bound to immobilized fibrinogen/fibrin, increasing amounts of unlabeled IGF-I were coincubated with 125 I-IGF-I in the presence of IGFBP-3⅐fibrinogen complexes (70 fmol/well) or IGFBP-3⅐fibrin complexes (115 fmol/well). As shown in Fig. 4B, IGF-I bound the immobilized IGFBP-3⅐fibrinogen complex with K d ϭ 2.9 Ϯ 0.6 nM (mean Ϯ S.E. of three independent experiments) and bound the immobilized IGFBP-3⅐fibrin complex with K d ϭ 2.3 Ϯ 0.3 nM (mean Ϯ S.E. of three independent experiments); in addition, there was a 63 Ϯ 10% increase in binding of 125 I-IGF-I to IGFBP-3⅐fibrin compared with IGFBP-3⅐fibrinogen (data not shown). Further studies (Fig. 4C) showed that 125 I-IGF-I binding to the IGFBP-3⅐fibrin complex was efficiently competed by IGF-I but not the IGF-I mutant LR 3 -IGF-I, which has a decreased affinity for IGFBP-3 (29). Finally, coincubating unlabeled IGF-I with 125 I-IGFBP-3 did not affect the ability of 125 I-IGFBP-3 to bind immobilized fibrinogen or fibrin (Fig. 4D); this suggests that fibrinogen and fibrin bind equally well to IGFBP-3 and the IGFBP-3⅐IGF-I complex.
125 I-IGFBP-3 Binds Fibrinogen in the Fluid Phase-As shown in Fig. 5A, wells coated with anti-fibrinogen antibodies were able to bind 125 I-IGFBP-3 that had been preincubated with purified fibrinogen in the fluid phase; in these studies, the binding of 125 I-IGFBP-3 to fibrinogen was dose-dependent and saturable. As shown in Fig. 5B, soluble 125 I-IGFBP-3⅐fibrinogen complexes were also detected when citrate plasma was used as a source for native fibrinogen; binding specificity was demonstrated by the ability of excess unlabeled IGFBP-3 to completely inhibit 125 I-IGFBP-3 binding (data not shown). Nevertheless, as shown in Fig. 5C, fluid-phase fibrinogen competed only minimally (ϳ30%) for 125 I-IGFBP-3 binding to immobilized fibrinogen, suggesting that IGFBP-3 binding to a fibrin clot can occur in the presence of physiologic concentrations (ϳ3-4 mg/ml) of plasma fibrinogen (30).

IGFBP-3 Increases IGF-I Binding in Fibrin
Clots-Native fibrin clots, induced using calcium saturation of citrate plasma, specifically bound 125 I-IGFBP-3 (10,600 Ϯ 130 cpm specifically bound/100,000 cpm 125 I-IGFBP-3 added; mean Ϯ S.E. of three experiments performed in triplicate). This suggests that IG-FBP-3 can be bound during clot formation. Furthermore, the amount of 125 I-IGF-I specifically bound within these fibrin clots is directly related to the amount of plasma used to generate the clot (data not shown). As IGFBP-3 is the primary circulating IGFBP, it is likely that IGFBP-3 is primarily responsible for immobilizing IGF-I in the fibrin clot. When forming fibrin clots using 25 l of citrate plasma, 125 I-IGF-I was specifically bound (6,200 Ϯ 140 cpm specifically bound/100,000 cpm 125 I-IGF-I added), and 125 I-IGF-I specific binding increased by 62 Ϯ 11% upon the further addition of 100 ng of IGFBP-3 (mean Ϯ S.E. of two experiments performed in quadruplicate). These data suggest that IGF-I is immobilized in fibrin clots by associating with fibrin-bound IGFBP-3.
Fibrin Clots Contain Both IGFBP-3 and Glu-Pg-Immunofluorescent staining was also used to determine whether IGFBP-3 was incorporated into native fibrin clots. As shown in Fig. 6, specific antibodies to IGFBP-3 and Glu-Pg did indeed recognize each protein in cross-sections of fibrin clots. This is consistent with the ability of Glu-Pg to bind fibrin with high affinity (22,23). However, both fibrin and Glu-Pg bind with high affinity to IGFBP-3, and Fig. 6 does not discriminate which protein binds IGFBP-3 in fibrin clots. Fig. 7A, partial proteolysis of immobilized fibrinogen by 5 ng/ml Pm completely inhibits IGFBP-3 binding, presumably by destroying IGFBP-3 binding sites. However, coincubation of IGFBP-3 with Glu-Pg allows IGFBP-3 binding to fibrinogen partially proteolyzed by 5 ng/ml Pm. This increase is likely due to IGFBP-3 binding to fibrin-bound Glu-Pg, since partial digestion of fibrinogen may actually increase Glu-Pg binding by uncovering previously hidden Glu-Pg binding sites (22).

IGFBP-3 Binds Fibrinogen Both Directly and via Glu-Pg-As shown in
Previous studies found that 10 mM ⑀ACA completely blocks Glu-Pg binding to fibrinogen (31). However, as shown in Fig.  7B, 25 mM ⑀ACA only weakly inhibits fibrinogen binding to IGFBP-3; as expected, arginine is a potent inhibitor of fibrinogen binding to IGFBP-3. This suggests that ⑀ACA can be used to discriminate between IGFBP-3 binding to fibrinogen and to fibrinogen-bound Glu-Pg. Indeed, as shown in Fig. 7C, when solid-phase fibrinogen is pre-equilibrated with Glu-Pg, IG-FBP-3 binding is inhibited 35-40% in the presence of 10 or 25 mM ⑀ACA, suggesting that 35-40% of IGFBP-3 binds to fibrinogen-bound Glu-Pg. DISCUSSION IGF-I participates in wound healing by acting as a chemotactic agent supporting the early migration of stromal cells into ogen antibodies. After 1 h at 37°C, unbound reactants were removed. Bound radioactivity was released from wells using 1 M acetic acid for 10 min at 23°C. Nonspecific binding ( 125 I-IGFBP-3 binding in the absence of fibrinogen) was subtracted. A, 125  Reactants were then added to wells precoated with 5 g/ml anti-fibrin-the fibrin clot at the wound site, and by stimulating proliferation of fibroblasts and endothelial cells (16,17). Potential sources for IGF-I found at the wound site include plasma, platelets (32), monocytes (33), and macrophages (34). As with IGFs in other body fluids, IGF-I in wound fluid is complexed with IGF-binding proteins, primarily IGFBP-3, which is abundant in serum and in platelets (18,19,32). Studies presented in this report, which show that IGFBP-3⅐IGF-I binary complexes bind with high affinity to fibrinogen/fibrin, suggest that direct binding of IGFBP-3⅐IGF-I complexes to the fibrin clot is an important mechanism by which IGF-I levels may be maintained at the wound site.
Fibrinogen and fibrin appear to interact with a specific HBD of IGFBP-3, which is rich in lysine groups and which participates in IGFBP-3 binding to acid-labile subunit, Glu-Pg, and prekallikrein (4,20,26). The sequence and location of the IGFBP-3 HBD is conserved in IGFBP-5 (12). Studies presented here show that IGFBP-5 efficiently competes with IGFBP-3 binding to fibrinogen or fibrin, suggesting that IGFBP-5 also binds fibrinogen and fibrin with high affinity. In contrast, the IGFBP-3 HBD sequence is not conserved in IGFBP-1 (12), IGFBP-1 does not bind appreciably to fibrinogen, and replacing the IGFBP-3 HBD sequence with homologous sequence from IGFBP-1 results in a 10-fold loss of affinity for fibrin. The residual ability of IGFBP-3hbdBP1 BV to compete with native IGFBP-3 for binding to fibrin suggests that, in addition to the HBD, other IGFBP-3 regions may participate in binding to fibrinogen/fibrin.
Polymerization of fibrinogen into fibrin did not alter affinity for IGFBP-3, but it did increase the number of 125 I-IGFBP-3 binding sites by 59% and the number of 125 I-IGF-I⅐IGFBP-3 binding sites by 63%. These data suggest that cleavage of fibrinogen exposes new binding sites for IGFBP-3, just as it exposes new binding sites for Glu-Pg (22). These data also suggest that formation of binary complexes between IGFBP-3 and immobilized fibrin does not prevent the further binding of FIG. 6. Both Glu-Pg and IGFBP-3 are detectable in fibrin clots. Normal human citrate plasma was clotted by addition of CaCl 2 to a final concentration of 5 mM. After 90 min, the resultant fibrin clot was rinsed and cryosectioned into 5-m sections. Sequential sections were incubated with normal rabbit serum (NRS) or with rabbit antisera to plasminogen or IGFBP-3. Immunofluorescent staining was recorded using fluorescein isothiocyanate-labeled donkey anti-rabbit IgG as the second antibody.

FIG. 7. IGFBP-3 binds to fibrinogen both directly and via Glu-Pg.
A, fibrinogen immobilized on immunocapture plates was pretreated with increasing concentrations of Pm for 1 h at 37°C. Pm was removed by rinsing wells extensively with 200 mM ⑀ACA followed by buffer alone. ϳ50,000 cpm 125 I-IGFBP-3 were equilibrated with digested fibrinogen for at least 1 h at 37°C with and without 10 g/ml Glu-Pg. Unbound 125 I-IGFBP-3 was removed by rinsing. Bound radioactivity was determined by addition of 1 N NaOH to well contents for 10 min at 23°C. After nonspecific binding was subtracted out, specific binding was normalized to % control values (i.e. binding of IGFBP-3 to solid-phase fibrinogen not exposed to Pm). Each data point is the mean Ϯ S.E. of quadruplicate determinations. B, ϳ50,000 cpm 125 I-IGFBP-3 were equilibrated with solid-phase fibrinogen and with increasing concentrations of either arginine or ⑀ACA for at least 1 h at 37°C. Unbound 125 I-IGFBP-3 was removed by rinsing. Bound radioactivity was determined by addition of 1 N NaOH to well contents for 10 min at 23°C. Each data point is the mean Ϯ S.E. of duplicate determinations. C, ϳ50,000 cpm 125 I-IGFBP-3 were equilibrated with solid-phase fibrinogen and either with or without 10 g of Glu-Pg along with 0, 10, or 25 mM ⑀ACA for at least 1 h at 37°C. Unbound 125 I-IGFBP-3 was removed by rinsing. Bound radioactivity was determined by addition of 1 N NaOH to well contents for 10 min at 23°C. After nonspecific binding was subtracted out, specific binding was normalized to percentage of control values (i.e. 125 I-IGFBP-3 alone or 125 I-IGFBP-3 ϩ Glu-Pg). Each data point is the mean Ϯ S.E. of quadruplicate determinations.
IGF-I to IGFBP-3, thus forming a ternary complex. Additional studies confirmed that IGF-I binds to IGFBP-3 complexed with immobilized fibrin (K d ϭ 2.3 nM), and also found that IGF-I binds to IGFBP-3 complexed with immobilized fibrinogen (K d ϭ 2.9 nM). These K d values are higher than the range of values (0.03-0.5 nM) reported previously for IGF-I binding to IGFBP-3 (42), and are comparable to the affinity of IGF-I for the type I IGF receptor (43). This suggests that type I IGF receptors on fibroblasts and other cells that are migrating into the fibrin clot can remove IGF-I from fibrin-bound IGFBP-3, resulting in proliferation of these cells at the wound site.
In addition to the direct binding of IGFBP-3 to fibrinogen, studies presented here suggest that IGFBP-3 also binds indirectly, to fibrinogen-bound Glu-Pg. It is well known that Glu-Pg binds with high affinity to fibrinogen/fibrin and that, through this interaction, Glu-Pg plays an important role in the wound healing process (21)(22)(23). It is also known that IGFBP-3 and IGF-I⅐IGFBP-3 complexes bind Glu-Pg with high affinity, and that IGFBP-3 binds to multiple sites on Glu-Pg (20). Thus, a complex of fibrin⅐Glu-Pg⅐IGFBP-3⅐IGF-I is likely to form; in one possible arrangement, the kringle 1-3 regions of Glu-Pg bind to lysines in fibrinogen/fibrin, while the kringle 5/catalytic region of Glu-Pg binds to the HBD of IGFBP-3. If formed, this complex would provide another mechanism for releasing IGF-I at the wound site, since tissue plasminogen activator can activate immobilized Glu-Pg that is complexed with IGFBP-3⅐IGF-I, resulting in IGFBP-3 proteolysis and IGF-I release (20). Indeed, in tissue culture, activation of Glu-Pg to Pm results in IGFBP-3 proteolysis and release of bound IGF-I to cultured cells where mitogenic and metabolic pathways are stimulated (44,45); thus, activation of Glu-Pg that is complexed both with fibrinogen and with IGFBP-3⅐IGF-I should ultimately make IGF-I available to stromal cells present at the wound site. In support of this hypothesis, additional studies show that plasminogen activators can induce proteolysis of IGFBP-3 complexed with fibrin-bound Glu-Pg, resulting in release of IGF-I (manuscript in preparation).
The observations that 125 I-IGFBP-3 and 125 I-IGF-I localize to fibrin clots generated from plasma in vitro, and that addition of exogenous IGFBP-3 greatly increases the incorporation of 125 I-IGF-I into these clots, suggest that IGF-I⅐IGFBP-3 complexes bind to fibrin clots in vivo at wound sites. This possibility is supported by immunoblot detection of IGFBP-3 in these fibrin clots. The further detection of Glu-Pg in these fibrin clots by immunoblotting suggests that IGFBP-3 may bind these clots directly through fibrin and indirectly through fibrin-bound Glu-Pg. Thus, data obtained from the study of these fibrin clots support the hypothesis that IGFBP-3⅐IGF-I complexes bind to fibrin clots formed at wound sites in vivo, and are consistent with the possibility that at least two mechanisms exist for the ready release of bound IGF-I to fibroblasts and endothelial cells migrating into the wound site.