Endocrinology and Metabolism

Plasminogen binds the heparin-binding domain of insulin-like growth factor-binding protein-3

Phil G. Campbell, Susan K. Durham, Adisak Suwanichkul, James D. Hayes, David R. Powell

Abstract

Limited proteolysis lowers affinity of insulin-like growth factor (IGF)-binding protein (IGFBP)-3 for bound IGFs, resulting in greater IGF bioavailability. Plasmin is one of many proteases that cleave IGFBP-3, and the plasmin system may regulate IGFBP-3 proteolysis and IGF bioavailability in cultured cells in vitro. A role for the plasmin system in IGFBP-3 proteolysis in vivo is suggested by data presented here showing that IGFBP-3 binds plasminogen (Pg; Glu-Pg) with a dissociation constant (K d) ranging from 1.43 to 3.12 nM. IGF-I and Glu-Pg do not compete for IGFBP-3 binding; instead, the binary IGFBP-3/Glu-Pg complex binds IGF-I with high affinity (K d= 0.47 nM) to form a ternary complex. Competitive binding studies suggest that the kringle 1, 4, and 5 domains of Glu-Pg and the heparin-binding domain of IGFBP-3 participate in forming the IGFBP-3/Glu-Pg complex, and other studies show that Glu-Pg in this complex is activated at a normal rate by tissue Pg activator. Importantly, IGFBP-3/Glu-Pg complexes were detected in both human citrate plasma and serum, indicating that these complexes exist in vivo. Binding of IGFBP-3 to Glu-Pg in vivo suggests how Glu-Pg activation can specifically lead to IGFBP-3 proteolysis with subsequent release of IGFs to local target tissues.

  • plasmin
  • insulin-like growth factors
  • proteolysis
  • kringle domains
  • tissue plasminogen activator

insulin-like growth factor (IGF)-binding protein (IGFBP)-3 is one of at least six structurally related IGFBPs that have in common the ability to bind IGF peptides with high affinity (30). IGFBP-3 is a 29-kDa protein found in serum and other body fluids as multiple ∼40- to 50-kDa forms due to extensive glycosylation (16,27, 30). IGFBP-3 is the most common IGFBP in the bloodstream during extrauterine life, where it circulates at ∼150 kDa complexed to other proteins; most of this IGFBP-3 appears to be bound to one IGF peptide and a single acid-labile subunit (ALS) in an ∼150-kDa ternary complex (3, 24, 30).

The ability to bind IGF peptides with high affinity allows IGFBP-3 to modulate the insulin-like mitogenic, metabolic, and differentiative effects of IGF-I and -II. One important role for IGFBP-3 is to bind IGF peptides in the ∼150-kDa serum complex. However, IGFBP-3 expression by many tissues and cell lines suggests that this protein also functions locally to modulate the autocrine/paracrine actions of IGF peptides (30). Because intact IGFBP-3 has a higher affinity for IGF peptides than does the type I IGF receptor, intact IGFBP-3 inhibits IGF action in serum and at the local level (13, 30). However, proteolysis of IGFBP-3 has been described in many in vivo and in vitro situations (2, 4, 8, 12, 13, 17-19, 22, 23). The resulting fragments, in particular, a major ∼30-kDa form, have a reduced affinity for IGFs that is lower than that of the type I receptor; in these situations, IGFs are released from IGFBP-3, and increased biological effects of IGFs may be observed (2, 4, 8, 13, 16, 19, 22).

A number of IGFBP-3 proteases have been identified, including plasmin, matrix metalloproteases, kallikreins, prostate-specific antigen, and cathepsin D (2, 4, 8, 12, 17-19, 22, 23), and the roles of each in various clinical and in vitro conditions are under investigation. The plasmin system has been implicated in the regulation of IGFBP-3 proteolysis and IGF bioavailability in cultured osteosarcoma and prostate carcinoma cells (2, 22). The recent observation that intact native plasminogen (Pg; Glu-Pg) binds directly to the 18-amino acid heparin-binding domain (HBD) of IGFBP-5 (7) suggests that Glu-Pg might also bind directly to IGFBP-3, since IGFBP-3 has an 18-amino acid HBD homologous to that of IGFBP-5 (5). This would suggest an important and generalized role for plasmin in IGFBP-3 proteolysis and would also raise the possibility that other preproteases might have preferential access to IGFBP-3 by binding to its HBD. The present studies confirm that Glu-Pg does indeed bind to IGFBP-3 and that the IGFBP-3 HBD participates in Glu-Pg binding.

MATERIALS AND METHODS

Materials.

Recombinant human IGF-I was purchased from GroPep (Adelaide, Australia). Recombinant human IGFBP-5 was produced in a baculovirus expression system and purified by affinity chromatography and reverse-phase (RP)-HPLC (1). IGFBP-1 was purified from human amniotic fluid by affinity chromatography and RP-HPLC (6); quantitation was by RIA using a kit from Diagnostic Systems Laboratories (DSL; Webster, TX). Glu-Pg was purified from fresh-frozen human plasma with the use of Lys-Sepharose in the presence of 1 mM benzamidine, 50 μg/ml trypsin inhibitor, and 20 KIU/ml aprotinin (9). Commercial vendors provided Lys-Pg (Enzyme Research Laboratories, South Bend, IN), mini-Pg (which consists of Glu-Pg kringle domain 5 and the B chain containing the catalytic portion of Glu-Pg) and PgK-4 (which consists of Glu-Pg kringle domain 4) (American Diagnostics, Greenwich, CT), PgK-1–3 (which consists of Glu-Pg kringle domains 1–3), α2-antiplasmin and fibrinogen (Sigma, St. Louis, MO), plasmin substrate S2251 (Chromogenics-Pharmacia), and tissue Pg activator (tPA; Genentech, South San Francisco, CA). IGFBP-5 peptides IGFBP-5-(138—152), IGFBP-5-(130—142), and IGFBP-5 HBD (Table 1) were kindly provided by Dr. Dennis Andress (Univ. of Washington, Seattle, WA). The IGFBP-3 HBD peptide (Table 1) was synthesized by Genemed Synthesis (South San Francisco, CA). Human serum and plasma were obtained from outdated blood bank supplies or from laboratory personnel.

View this table:
Table 1.

Protein and peptide sequences

IGFBP-3 mutagenesis.

A full-length IGFBP-3 cDNA was isolated from a human placental cDNA library (14); it spanned from −47 to +1,239 bp of the human IGFBP-3 cDNA sequence reported by Wood et al. (40) and had the same nucleotide sequence as the published cDNA. This IGFBP-3 cDNA was inserted into the M13-based vector M13mp18 at theEcoR I site and was mutated by the Kunkel method with the use of synthetic oligonucleotides and the Muta-Gene kit (Bio-Rad) (36). The initial mutation used oligonucleotide 5′-CAGAGCAAGTAGAAGCTTGCCGCAAGGTTA-3′ to introduce aHind III site at base pairs +877 to +882, immediately 3′ to the IGFBP-3 translation stop codon. The oligonucleotide 5′-GCAGTGTCGCCCTTCCATGGACGGGGAGGCGGGCTTCTGCTGGTGTGT-3′ was then used to create IGFBP-3mHBD (Table 1). The sequence of each mutation was confirmed by DNA sequence analysis using Sequenase (US Biochemicals, Cleveland, OH) in the dideoxy chain termination method (36).

IGFBP-3 expression and purification.

The expression vector pKG3226 contains the human β-actin promoter, SV40 polyadenylation signal and neomycin phosphotransferase resistance gene (34). IGFBP-3 and IGFBP-3mHBD cDNAs were introduced into pKG3226 at the EcoR I (5′) andHind III (3′) sites. The resulting expression vectors pKG-BP3 and pKG-BP3mHBD were stably transfected into Chinese hamster ovary (CHO)-K1 cells. Briefly, 3 × 105 CHO-K1 cells/60-mm plate were maintained for 24 h in McCoy’s 5A medium (GIBCO BRL, Gaithersburg, MD) containing 2.5% FCS and 7.5% calf serum (complete medium). Five micrograms of pKG-BP3 or pKG-BP3mHBD plasmid were mixed with opti-MEM medium (GIBCO BRL) and 8 μl Lipofectin (GIBCO BRL), and the mixture was then incubated with the CHO-K1 cells for 6 h at 37°C. The DNA/Lipofectin mixture was then removed from the CHO-K1 cells; after incubation overnight in complete medium, these cells were then incubated in complete medium containing 1 mg geneticin (G-418)/ml. Surviving cells formed individual clones; ∼60 pKG-BP3 and pKG-BP3mHBD clones were transferred into individual wells of 96-well plates. Cells were incubated in serum-free McCoy’s 5A medium and then screened for IGFBP-3 expression by immunoblot (seeImmunoblotting) and by immunoradiometric assay (IRMA) using a commercially available kit (DSL). Clones expressing the highest amounts of intact IGFBP-3 or IGFBP-3mHBD were expanded into 162-cm2 tissue culture flasks (Costar, Cambridge, MA). Confluent cells were washed and then incubated for 72 h in 20 ml of McCoy’s 5A medium. CHO-K1-conditioned medium was then collected, centrifuged at 2,000 gfor 30 min at 4°C, and stored frozen at −80°C.

IGFBP-3 and IGFBP-3mHBD were purified by passing ∼200-ml aliquots of CHO-K1-conditioned media containing 2.5 M NaCl across an affinity column consisting of 1 mg IGF-I covalently linked to 2 ml Affi-Gel 10 beads (Bio-Rad, Richmond, CA) (6). The column was washed with 500 ml of 10 mM sodium phosphate, pH 7.5, and 2 M NaCl. IGFBPs were eluted from the column in 15 ml of 1% acetic acid. Three acetic acid eluates (45 ml) from the IGF-I affinity column were pooled, made 50 mM with sodium phosphate, and titrated to pH 7.5 with NaOH. This solution was recirculated overnight across an affinity column containing 10 mg of anti-human IGFBP-3 monoclonal antibody DSL-03056 (a kind gift from DSL) that had been covalently linked to 2 ml of Affi-Gel 10 beads. This column was washed with 125 ml of 20 mM ammonium acetate, pH 6.5, and then eluted with 12 ml of 1% acetic acid.

Final purification was achieved by RP-HPLC using a Vydac C4 column (25 cm × 4.6 mm ID, 5 μm particle size; Rainin, Woburn, MA) and a Beckman HPLC system described previously (6). The entire acetic acid eluate from the anti-human IGFBP-3 monoclonal antibody affinity column was loaded onto the Vydac column using 0.1% trifluoracetic acid (Sigma) as aqueous mobile phase. The column was eluted with an increasing gradient of acetonitrile. Column eluant was monitored at 280 nm; ultraviolet peaks were collected individually and assayed for IGFBP-3 by IRMA (DSL). Peaks containing IGFBP-3 or IGFBP-3mHBD by IRMA were separated by SDS-PAGE and transferred to nitrocellulose membranes; these membranes were then stained for proteins by India ink (13), immunoblotted using a human IGFBP-3 antibody (seeImmunoblotting), or blotted with125I-labeled IGF-I (6). Pure preparations of intact, normally glycosylated IGFBP-3 and IGFBP-3mHBD that bound 125I-IGF-I were identified; these preparations were lyophilized and stored at room temperature until use in the present studies.

Solid-phase plate-binding assay.

Binding of IGFBP-3 to Glu-Pg was characterized using an immobilized Glu-Pg-based assay system (7). Optimal binding conditions, Glu-Pg coating concentration, plate type, and incubation parameters were determined and are presented in the following assay protocol. Ninety-six-well immunologic plates (Polysorb, NUNC, Fisher Scientific, Pittsburgh, PA) are coated with 5 μg/ml Glu-Pg in 0.1 M Na2CO3, pH 9.8, overnight at 4°C. The plates are rinsed with 200 μl of 10 mM sodium phosphate, pH 7.5, and 150 mM NaCl and blocked with 200 μl of 10 mM Tris ⋅ HCl, 150 mM NaCl, 0.05% Tween 80, 1% BSA, and 0.02% NaN3, pH 7.5, for 1 h at 37°C. Plates are rinsed twice with 200 μl of 10 mM sodium phosphate, pH 7.5, and 150 mM NaCl and once with 200 μl of 30 mM Tris acetate, pH 7.4, 10 mM sodium phosphate, 0.1% Tween 20, and 0.2% NaN3 (assay buffer). IGFBP-3 and IGF-I were iodinated by the chloramine T method to a specific activity of ∼150 μCi/μg protein (8).125I-labeled IGFBP-3 [50,000 counts/min (cpm)] or125I-IGF-I (50,000 cpm) was incubated with various concentrations of IGFBP-3, IGFBP-3mHBD, IGFBP-1, IGFBP-5, IGFBP-5-(138—152), IGFBP-5-(130—142), IGFBP-3 HBD peptide, IGFBP-5 HBD peptide, IGF-I, Glu-Pg, mini-Pg, Lys-Pg, PgK-1–3, PgK-4, Arg, ε-aminocaproic acid (ε-ACA), α2-antiplasmin, or heparin in 100 μl of assay buffer; unless otherwise stated, incubations were for 1 h at 37°C. Unbound radioactivity was removed by rinsing the wells twice with 200 μl of ice-cold assay buffer. Bound radioactivity was solubilized with 200 μl of 1 N NaOH, transferred to 12 × 75 mm glass test tubes, and counted for radioactivity.

Fluid-phase plate-binding assay.

Ninety-six-well immunologic plates (MaxiSorb, NUNC, Fisher Scientific) were coated with 5 μg/ml rabbit anti-human Glu-Pg antibody (Boehringer Mannheim Biochemicals, Indianapolis, IN) overnight at 4°C or for 3 h at 23°C in 100 μl of 0.1 M Na2CO3, 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 Glu-Pg, IGFBP-3,125I-IGFBP-3, and/or human plasma or serum preparations were incubated in a total volume of 200 μl 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 μl of reactants were transferred to the blocked and rinsed 96 wells that were coated with the anti-Glu-Pg antibody. After 30 min at 23°C, unbound radioactivity was removed by rinsing wells twice with assay buffer. Bound radioactivity, representing IGFBP-3/Glu-Pg complexes, was released by incubating wells with 200 μl of 1 M acetic acid for 10 min at 23°C. Acid washes were transferred to 12 × 75 mm glass test tubes and counted for radioactivity. Radioactivity bound in the absence of added Glu-Pg served as an estimate of nonspecific binding.

125I-IGFBP-3 ligand blotting.

Pure Glu-Pg (10 μg) or IGFBP-1 (10 μg) was electrophoresed on a 12% SDS-polyacrylamide gel under nonreducing conditions. These proteins were then transferred to a nitrocellulose membrane as described previously (6). 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, and then incubated overnight with ∼7 × 106 cpm125I-IGFBP-3 in 0.1% Tween 20-TBS. After being washed and dried, film was exposed to filters overnight.

Copurification of IGFBP-3, ALS, and Glu-Pg from human plasma.

Six milligrams of either goat anti-human IGFBP-3 antibody (a kind gift from DSL) or nonimmune goat IgG (Jackson ImmunoResearch Laboratories, West Grove, PA) were covalently linked to 2 ml of Affi-Gel 10 beads (Bio-Rad) and then poured into separate columns. Citrate human plasma (150 ml) mixed with 150 ml of 10 mM sodium phosphate, pH 7.5, and 150 mM NaCl was then run across each column. After being washed with 600 ml of 10 mM sodium phosphate, pH 7.5, and 150 mM NaCl, each column was step-wise eluted with 100 ml of 50 mM sodium phosphate, pH 7.5, containing either 0.3, 0.6, or 1.0 M NaCl, followed by a final elution using 20 ml of 1% acetic acid. Aliquots (100 μl) of each elution fraction from the goat anti-human IGFBP-3 antibody column or goat nonimmune IgG column were electrophoresed on a 7.5–15% gradient SDS-polyacrylamide gel under reducing conditions (100 mM dithiothreitol). Separated proteins were then transferred to a nitrocellulose membrane as described previously (6). Fractions from each column were tested for the presence of proteins by staining of the nitrocellulose membrane with India ink (6). Fractions from each column were also tested for the presence of IGFBP-3, ALS, and Glu-Pg by immunoblotting.

Immunoblotting.

Nitrocellulose membranes containing proteins eluted from the anti-human IGFBP-3 antibody column and nonimmune goat IgG column were immunoblotted using specific antibodies against human IGFBP-3, ALS, and Glu-Pg. IGFBP-3 and Glu-Pg immunoblot analyses were performed using a rabbit anti-human IGFBP-3 antibody (DSL) at a 1:5,000 dilution or a rabbit anti-human Glu-Pg antibody (Boehringer Mannheim Biochemicals) at a 1:10,000 dilution; in each case, a 1:10,000 dilution of the peroxidase Affinipure goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories) served as second antibody. ALS immunoblot analyses were performed using a goat anti-human ALS antibody (DSL) at a 1:7,500 dilution (26) and the Affinipure rabbit anti-goat IgG (Jackson ImmunoResearch Laboratories) at a 1:10,000 dilution. In each case, detection was by enhanced chemiluminescence (Amersham Life Sciences, Buckinghamshire, UK) following the recommendations of the manufacturer.

Double-antibody sandwich RIA for IGFBP-3/Glu-Pg complexes.

In some studies, 100 ng of IGFBP-3 were complexed with increasing amounts of Glu-Pg at 37°C in a total volume of 200 μl of assay buffer in a 96-well test plate. After 1 h, 100 μl were transferred to MaxiSorb plates previously coated overnight at 4°C with 5 μg/ml goat anti-human IGFBP-3 antibody (DSL). After 30 min at 23°C, 100 μl of assay buffer containing 100,000 cpm of125I-labeled rabbit anti-Glu-Pg antibody (Boehringer Mannheim Biochemicals) were added. After an additional 30 min of incubation at 23°C, plates were rinsed with assay buffer and bound radioactivity was released by incubating plates for 10 min at 23°C with 200 μl of 1 M acetic acid. In other studies, increasing volumes of citrate plasma or serum were placed directly into plates precoated with the anti-human IGFBP-3 antibody, and IGFBP-3/Glu-Pg complexes were determined as described above using purified IGFBP-3 and Glu-Pg. In all studies, the binding of125I-labeled anti-Glu-Pg antibody to plates precoated with the anti-human IGFBP-3 antibody served to establish nonspecific binding.

Activation of Glu-Pg by tPA.

Glu-Pg (400 nM) was incubated alone, with 100 nM IGFBP-3, or with 100 nM fibrinogen for 1 h at 37°C in 200 μl of assay buffer. After plasmin substrate S2251 was added (0.2 mM final concentration), the Glu-Pg activation reaction was initiated by adding 0.77 nM tPA. Substrate hydrolysis was monitored at 37°C over a 10-min period at an absorbance of 405 nM. Proteolysis of IGFBP-3 via the activation of Glu-Pg to plasmin was confirmed by incubation of 200,000 cpm of125I-IGFBP-3 with physiologically relevant concentrations of IGFBP-3 (4.5 μg/ml), IGF-I (200 ng/ml), IGF-II (600 ng/ml), and various combinations of Glu-Pg (10 μg/ml) and tPA (5 or 50 ng/ml). Incubations, performed for 1 h at 37°C in 50 μl of assay buffer, were terminated by adding 25 μl of 3× SDS-PAGE nonreducing sample buffer and boiling the samples for 3 min. Samples were separated by 17.5% SDS-PAGE, and gels were dried and autoradiographed overnight.

RESULTS

125I-IGFBP-3 binds immobilized Glu-Pg.

125I-IGFBP-3 bound specifically to Glu-Pg immobilized on immunocapture plates. Figure1 Apresents a time course which shows that, for the assay conditions employed, specific 125I-IGFBP-3 binding to Glu-Pg plateaued at ∼1 h. This binding competed with IGFBP-5 as well as IGFBP-3 but not IGFBP-1 (Fig.1 B). Scatchard analysis of125I-IGFBP-3 binding to Glu-Pg (Fig. 1 C) revealed that IGFBP-3 bound Glu-Pg with a dissociation constant (K d) = 1.43 ± 0.12 nM (mean ± SE of 3 independent experiments).

Fig. 1.

125I-labeled insulin-like growth factor (IGF)-binding protein (IGFBP)-3 binding to plasminogen (Pg; Glu-Pg) immobilized on immunocapture plates. Plates were coated overnight at 4°C with 5 μg/ml Glu-Pg and then incubated with ∼50,000 counts/min (cpm)125I-IGFBP-3 at 37°C in assay buffer. Nonspecific binding was determined by adding unlabeled IGFBP-3; this value was subtracted from total binding to give specific binding. Each data point is mean of duplicate determinations.A: time course showing equilibration of 125I-IGFBP-3. Nonspecific binding was determined by adding 1 μg/ml unlabeled IGFBP-3.B: competition curve for125I-IGFBP-3 binding to Glu-Pg using increasing concentrations of IGFBP-1, IGFBP-3, and IGFBP-5. Highest concentration of unlabeled IGFBP-3 (4 μg/ml) was used to estimate nonspecific binding and was subtracted from total binding to derive specific binding (B). Bo represents125I-IGFBP-3 bound in the absence of any competing ligand.C: Scatchard plot of125I-IGFBP-3 binding to Glu-Pg using data presented in B; B/F, ratio of bound to free 125I-IGFBP-3 in each sample.

125I-IGFBP-3 also bound Glu-Pg by ligand blot. Glu-Pg and IGFBP-1 were electrophoresed by SDS-PAGE, transferred to a nitrocellulose membrane, and then incubated with 125I-IGFBP-3. As shown in Fig.2, 125I-IGFBP-3 clearly bound Glu-Pg but not IGFBP-1.

Fig. 2.

125I-IGFBP-3 binding to Glu-Pg by ligand blot. Glu-Pg and IGFBP-1 (10 μg each) were separated by SDS-PAGE on a 12% gel and then transferred to a nitrocellulose membrane. Membrane was incubated overnight at 4°C with ∼7 × 106 cpm125I-IGFBP-3, washed, and then autoradiographed. Molecular mass (M r; in kDa) of protein markers is shown atleft.

IGF-I/IGFBP-3 complex binds immobilized Glu-Pg.

Coincubation of 125I-IGF-I and increasing amounts of unlabeled IGFBP-3 in plates coated with Glu-Pg resulted in a steady increase in125I-IGF-I binding to these plates (Fig.3 A). Similar results were noted when increasing amounts of unlabeled IGFBP-3 were incubated with immobilized Glu-Pg before the addition of125I-IGF-I (Fig.3 B). In each of these experiments, very little 125I-IGF-I bound to the Glu-Pg-coated plates in the absence of IGFBP-3, and coincubation of125I-IGF-I with 100 ng/ml IGFBP-3 did not result in specific125I-IGF-I binding if plates were not coated with Glu-Pg; these results suggest that125I-IGF-I was binding to Glu-Pg-bound IGFBP-3. To determine the affinity of IGF for IGFBP-3 bound to immobilized Glu-Pg, increasing amounts of unlabeled IGF-I were coincubated with 125I-IGF-I in the presence of the IGFBP-3/Glu-Pg complex (Fig.3 C); Scatchard analysis of these data shows that IGF-I bound the immobilized IGFBP-3/Glu-Pg complex withK d = 0.47 ± 0.12 nM (mean ± SE of 2 independent experiments). Finally, coincubation of unlabeled IGF-I with125I-IGFBP-3 did not affect the ability of 125I-IGFBP-3 to bind immobilized Glu-Pg (Fig. 3 D); this provides indirect evidence that Glu-Pg binds equally well to IGFBP-3 and the IGFBP-3/IGF-I complex.

Fig. 3.

IGFBP-3 binds Glu-Pg in presence of IGF-I. In each study, incubations were performed in Glu-Pg-coated immunocapture plates.A: ∼50,000 cpm125I-labeled IGF-I were coincubated with increasing amounts of unlabeled IGFBP-3. Bound125I-IGF-I was measured. Nonspecific binding was monitored by determining binding of125I-IGF-I in absence of IGFBP-3 (<0.4% of total counts added) and was subtracted from all groups. Each data point is mean ± SE of triplicate determinations.B: increasing amounts of unlabeled IGFBP-3 were incubated with immobilized Glu-Pg for 1 h at 37°C. Unbound IGFBP-3 was removed, and ∼50,000 cpm125I-IGF-I were added for another 1 h at 37°C. Bound 125I-IGF-I was measured. Nonspecific binding (<0.4% of total counts added) was monitored as in A and was subtracted from all groups. Each data point is mean ± SE of duplicate determinations. C: IGFBP-3 (0.5 μg/ml) was incubated with immobilized Glu-Pg for 1 h at 37°C. Unbound IGFBP-3 was removed, and ∼50,000 cpm125I-IGF-I were added with increasing amounts of unlabeled IGF-I for an additional 1 h at 37°C. Bound 125I-IGF-I was measured. Each data point is mean ± SE of duplicate determinations.D: increasing amounts of unlabeled IGF-I were coincubated with ∼50,000 cpm125I-IGFBP-3 in Glu-Pg-coated plates. Bound IGFBP-3 was measured. Each data point is mean ± SE of 2 separate experiments performed in duplicate.

IGFBP-3 HBD participates in Glu-Pg binding.

Because 1) the IGFBP-5 HBD participates in binding to Glu-Pg (7),2) IGFBP-3 has a homologous HBD (5), and 3) IGFBP-5 competes efficiently with 125I-IGFBP-3 for Glu-Pg binding, it seemed likely that the IGFBP-3 HBD participates in Glu-Pg binding. To test this hypothesis, heparin and synthetic 18-amino acid peptides encoding the IGFBP-3 and IGFBP-5 HBDs were evaluated for their ability to compete with125I-IGFBP-3 for Glu-Pg binding. As shown in Fig.4 A, heparin clearly interfered in a dose-dependent manner with binding of125I-IGFBP-3 to immobilized Glu-Pg. As shown in Fig. 4 B, both the IGFBP-3 and IGFBP-5 HBD peptides competed efficiently with125I-IGFBP-3 for Glu-Pg binding, but the IGFBP-5-(138—152) peptide and the Lys- and Arg-rich IGFBP-5-(130—142) peptide did not; the inhibitory effect of increasing amounts of the IGFBP-3 HBD peptide on125I-IGFBP-3 binding is shown in Fig. 4 C. In addition, IGFBP-3mHBD, a full-length IGFBP-3 protein mutated in the HBD region (see Table 1), had an approximately fivefold lower affinity for Glu-Pg than did native IGFBP-3 (Fig. 4 D).

Fig. 4.

IGFBP-3 heparin-binding domain (HBD) participates in binding to Glu-Pg. In each study, ∼50,000 cpm125I-IGFBP-3 were incubated with or without competitors in Glu-Pg-coated immunocapture plates. Nonspecific binding was determined in presence of 100 μg heparin, which was subtracted from total binding to derive specific binding. Bo was determined in absence of any competing ligand.A: increasing amounts of heparin were coincubated with 125I-IGFBP-3. Each data point is mean of duplicate determinations.B: 1 μg/ml IGFBP-3 HBD peptide (BP3HBD), IGFBP-5 HBD peptide (BP5HBD), IGFBP-5-(138—152), or IGFBP-5-(130—142) was coincubated with125I-IGFBP-3. Data are presented as percentages of nontreatment (no added competitor) control values; each data point is mean ± SE of triplicate determinations.C: increasing amounts of IGFBP-3 HBD peptide were coincubated with125I-IGFBP-3. Each data point is mean ± SE of duplicate determinations.D: increasing amounts of native IGFBP-3 or IGFBP-3mHBD (mutation) were coincubated with125I-IGFBP-3. Each data point is mean ± SE of duplicate determinations.

Multiple Pg domains participate in IGFBP-3 binding.

The compound ε-ACA binds with high affinity to Lys binding sites present in Glu-Pg kringle domains 1, 4, and 5 (10), whereas Arg and the Arg analog p-aminobenzamidine bind specifically to Glu-Pg kringle domain 5 (38). The ability of ε-ACA (Fig.5 A), Arg (Fig. 5 A), andp-aminobenzamidine (data not shown) to compete with 125I-IGFBP-3 for Glu-Pg binding suggests that Glu-Pg kringle domain 5 and possibly Glu-Pg kringle domains 1 and 4 participate in IGFBP-3 binding.

Fig. 5.

Determination of IGFBP-3 binding sites on Glu-Pg. In each study, ∼50,000 cpm 125I-IGFBP-3 were incubated with or without competitors in Glu-Pg-coated immunocapture plates. Nonspecific binding was determined in presence of 100 μg heparin. A: increasing concentrations of either Arg or ε-aminocaproic acid (ε-ACA) were coincubated with125I-IGFBP-3. Data are presented as percentages of nontreatment (no added competitor) control values; each data point is mean ± SE of 3 independent determinations performed in duplicate. B: increasing concentrations of Glu-Pg, Lys-Pg, PgK-1–3, PgK-4, or mini-Pg were coincubated with 125I-IGFBP-3 (seematerials and methods for explanation of PgK-1–3, PgK-4, and mini-Pg). Data are presented as percentages of nontreatment (no added competitor) control values; each data point is mean ± SE of duplicate determinations.

Figure 5 B examines the ability of Glu-Pg and a variety of Glu-Pg fragments to compete with immobilized Glu-Pg for 125I-IGFBP-3 binding. Lys-Pg (the plasmin-cleaved form of Glu-Pg, which contains kringle domains 1–5) competed more efficiently than Glu-Pg for125I-IGFBP-3 binding. All other fragments tested, including PgK-1–3 (which consists of Glu-Pg kringle domains 1–3), PgK-4 (which consists of Glu-Pg kringle domain 4), and mini-Pg (which consists of Glu-Pg kringle domain 5 and the B chain containing the catalytic portion of Glu-Pg) also competed with immobilized Glu-Pg for125I-IGFBP-3 binding in a dose-dependent manner. The PgK-1 domain was further implicated by studies showing that α2-antiplasmin, which specifically binds Glu-Pg kringle domain 1 (39), competed with125I-IGFBP-3 for binding to Glu-Pg (data not shown).

IGFBP-3, Pg, and ALS forms copurify from human citrate plasma.

Human citrate plasma was run over columns containing Affi-Gel 10 beads covalently bound to either goat anti-human IGFBP-3 antibody or to a nonimmune (control) goat IgG; after being washed, the columns were eluted with a step-wise NaCl gradient followed by a final elution using 1% acetic acid. India ink staining of the elution fractions showed that multiple proteins bound to the anti-IGFBP-3 antibody column in the presence of high NaCl concentrations and that most of these proteins were eluted from the column only in the presence of acetic acid; in contrast, very little protein bound to the goat IgG column (Fig.6 A). Immunoblotting showed that Glu-Pg, ALS, and IGFBP-3 forms each bound to the anti-IGFBP-3 antibody column but not to the nonimmune (control) goat IgG column (Fig. 6, B–D). The Glu-Pg immunoblot (Fig. 6 B) suggested that most of the proteins eluting in the acetic acid fractions were Pg forms. The largest form migrated at a molecular mass (M r) slightly less than ∼97 kDa; in other experiments, this form comigrated with purified Glu-Pg, which has anM r of 92 kDa (data not shown). A second form, which migrated slightly more slowly than ALS, has anM r consistent with that of Lys-Pg (84 kDa). The anti-Glu-Pg antibody also recognized third and fourth forms, which migrated at ∼60 and ∼25 kDa, consistent with theM r of plasmin heavy and light chains, respectively. The characteristic ALS doublet at ∼85 kDa was released from the anti-IGFBP-3 antibody column primarily in the NaCl elution fractions but also to a lesser extent in the acetic acid fractions (Fig. 6 C). The IGFBP-3 immunoblot (Fig. 6 D) showed that all IGFBP-3 forms remained on the anti-IGFBP-3 antibody column in the presence of 1 M NaCl and were removed from the column only in the presence of acetic acid. The largest IGFBP-3 forms were the characteristic doublet of intact, differentially glycosylated IGFBP-3, which migrated at ∼50 kDa (Fig.6 D); in other experiments, this doublet comigrated with intact, glycosylated human IGFBP-3 expressed in CHO-K1 cells (data not shown). In addition, an abundant ∼29-kDa IGFBP-3 form was present in the acetic acid fractions; this is almost certainly an IGFBP-3 fragment commonly found in the circulation (16). Control immunoblots on fresh human plasma, which was treated immediately with protease inhibitors (3 mM EDTA, 100 μg/ml soybean trypsin inhibitor, 3 mMp-benzamidine, and 100 U/ml aprotinin), showed that Pg and IGFBP-3 fragmentation occurred before plasma was passed over the affinity column and may well reflect the distribution of Pg and IGFBP-3 forms in the circulation (data not shown).

Fig. 6.

Copurification of IGFBP-3, acid-labile subunit (ALS), and Pg forms from citrated human plasma. Two columns were packed with Affi-Gel 10 beads covalently linked to either goat anti-human IGFBP-3 antibody (Ab) or to a comparable amount of nonimmune (control) goat IgG. Citrated human plasma (150 ml) was run over each column. After being washed, each column was stepwise eluted with 100 ml of 50 mM sodium phosphate, pH 7.5, containing either 0.3, 0.6, or 1.0 M NaCl, followed by a final elution with 20 ml of 1% acetic acid (HAc). An amount of 100 μl of each elution fraction from each column was then separated by SDS-PAGE and transferred to nitrocellulose membranes.M r (in kDa) of protein markers is shown at left.A: India ink protein stain.B: Pg immunoblot.C: ALS immunoblot.D: IGFBP-3 immunoblot.

125I-IGFBP-3 binds Pg forms in the fluid phase.

As shown in Fig.7 A, wells coated with anti-Glu-Pg antibodies were able to bind125I-IGFBP-3 that had been preincubated with Glu-Pg in the fluid phase. In these studies, the binding of 125I-IGFBP-3 to Glu-Pg was dose dependent and saturable (Fig.7 B), withK d = 3.12 ± 0.65 nM (mean ± SE of 2 independent experiments). In additional studies, 125I-IGFBP-3 also bound Lys-Pg, PgK-1–3, PgK-4, and mini-Pg in the fluid phase (data not shown). 125I-IGFBP-3 was also found to complex with Pg forms present in serum or citrate plasma but not in heparin plasma (Fig. 7 C); this suggested that IGFBP-3/Pg complexes may well form in the fluid phase in vivo.

Fig. 7.

Fluid-phase binding of125I-IGFBP-3 to Pg forms.A: increasing amounts of Glu-Pg were incubated at 37°C with ∼100,000 cpm125I-IGFBP-3 in a total volume of 200 μl of assay buffer in uncoated 96-well test plates;125I-IGFBP-3 incubated in absence of Glu-Pg served to estimate nonspecific binding. After 1 h, 100 μl of each sample were transferred to MaxiSorb 96-well plates previously coated with 5 μg/ml anti-Glu-Pg antibody. After 30 min at 23°C, wells were rinsed and then incubated for 10 min at 23°C with 200 μl of 1 M acetic acid to release bound125I-IGFBP-3; acid washes were then collected, and their radioactivity was counted. In absence of Glu-Pg, <0.5% of added125I-IGFBP-3 bound to plates coated with anti-Glu-Pg antibody. Each data point is mean ± SE of duplicate determinations. B: Glu-Pg (500 ng/ml) was incubated with125I-IGFBP-3 and increasing amounts of unlabeled IGFBP-3 under same conditions as inA. Each data point is mean ± SE of 2 independent experiments performed in duplicate.C: increasing volumes of human serum, heparin plasma, or citrate plasma were incubated with 100,000 cpm125I-IGFBP-3 under same conditions as in A. Each data point is mean ± SE of 2 independent experiments performed in duplicate.

IGFBP-3/Pg complexes are present in human plasma and serum.

Figure 8 Ashows that Glu-Pg/IGFBP-3 complexes formed in vitro can be detected by a double-antibody sandwich RIA using antibodies to Glu-Pg and IGFBP-3. This assay was then used to demonstrate the presence of endogenous IGFBP-3/Pg complexes in human citrate plasma and, to a lesser extent, in human serum (Fig. 8 B).

Fig. 8.

Detection of IGFBP-3/Pg complexes. Values are means ± SE of duplicate determinations. A: detection of IGFBP-3/Glu-Pg complexes formed in vitro. IGFBP-3 (100 ng) was complexed with increasing amounts of Glu-Pg at 37°C in a total volume of 200 μl in a 96-well test plate. After 1 h, 100 μl of each sample were transferred to plates precoated with anti-IGFBP-3 antibody. After 30 min at 23°C, plates were rinsed with assay buffer, and 100 μl of assay buffer containing ∼100,000 cpm125I-labeled anti-Glu-Pg antibody were added. After 30 min, plates were rinsed with assay buffer and incubated for 10 min at 23°C with 200 μl of 1 M acetic acid to release bound radioactivity. Binding of125I-anti-Glu-Pg antibody to plates precoated with anti-IGFBP-3 antibody served as an estimate of nonspecific binding. B: detection of endogenous IGFBP-3/Pg complexes present in human plasma and serum. Increasing volumes of either citrate plasma or serum were incubated overnight at 4°C in plates precoated with anti-IGFBP-3 antibody. IGFBP-3/Pg complexes were then determined as inA except that125I-anti-Glu-Pg antibody was incubated with each sample at 23°C for 30 min.

IGFBP-3 does not affect the tPA-catalyzed activation of Glu-Pg.

As shown in Table 2, the ability of tPA to activate Glu-Pg was enhanced by fibrinogen. In contrast, IGFBP-3 appeared to have no effect on the tPA-catalyzed conversion of Glu-Pg to plasmin.

View this table:
Table 2.

Initial rate of tPA-catalyzed activation of Glu-Pg in presence of IGFBP-3

Figure 9 confirms that in the presence of physiological plasma levels of IGFBP-3, IGFs, Glu-Pg, and tPA, IGFBP-3 is proteolyzed when Glu-Pg is activated to plasmin. Significant but incomplete proteolysis occurred with 5 ng/ml tPA (circulating concentration during normal, resting state), and complete proteolysis occurred with 50 ng/ml tPA (circulating concentration during pathological conditions such as sepsis).

Fig. 9.

Activation of Glu-Pg by tissue Pg activator (tPA) results in IGFBP-3 proteolysis. Unlabeled IGFBP-3 (4.5 μg/ml) was incubated with 200,000 cpm 125IGFBP-3, IGF-I (200 ng/ml), IGF-II (600 ng/ml), and various combinations of Glu-Pg (10 μg/ml) and tPA (5 or 50 ng/ml) for 1 h at 37°C. IGFBP-3 proteolysis was determined by separating samples on a 17.5% nonreducing SDS-PAGE gel; gels were autoradiographed after drying.M r (in kDa) of protein markers is shown at right.

DISCUSSION

Complex formation between IGFBP-3 and Pg forms was shown by several experimental approaches, including solid- and fluid-phase plate-binding assays, ligand blotting, copurification of IGFBP-3 and Pg forms from human citrate plasma using an anti-IGFBP-3 antibody affinity column, and detection of IGFBP-3/Pg complexes in human serum and citrate plasma using anti-Glu-Pg and anti-IGFBP-3 antibodies in a double-antibody sandwich RIA. These last two approaches strongly suggest that IGFBP-3/Pg complexes are present in the circulation in vivo. Similar affinity techniques have been used to identify other proteins that bind Pg (11, 25).

The binding affinity of IGFBP-3 for Glu-Pg is comparable to or higher than the binding affinity of other extracellular and plasma proteins known to interact with Pg. IGFBP-3 binds immobilized Glu-Pg with aK d = 1.43 nM and binds to Glu-Pg in fluid phase with aK d = 3.12 nM. These values are comparable to the binding affinities of fibrinogen and collagen type IV for Glu-Pg (35); are up to 100-fold greater than the binding affinities of fibronectin, osteonectin, thrombospondin, and laminin for Glu-Pg (15, 21, 31, 32); and are between 100- and 1,000-fold greater than the binding affinities of procarboxypeptidase B, vitronectin, and histidine-rich glycoprotein for Glu-Pg (25, 29,37). In addition, the binding affinity of Glu-Pg for IGFBP-3 is 2,800- and 30,000-fold greater than the affinity of Glu-Pg for α2-antiplasmin and fibrin, respectively (39).

IGFBP-3 appears to interact with Lys binding sites within Glu-Pg kringle domains. These same Lys binding sites mediate the interaction of Glu-Pg with a number of other plasma proteins, including fibrin, α2-antiplasmin, histidine-rich glycoprotein, tetranectin, osteonectin, and procarboxypeptidase B (10,11, 21, 25, 37, 39). The ability of Glu-Pg kringle domains 1–3 (PgK-1–3), PgK-4, mini-Pg, α2-antiplasmin, Arg, and the Lys analog ε-ACA to each inhibit IGFBP-3 binding to Glu-Pg suggests that IGFBP-3 interacts with the PgK-1, PgK-4, and PgK-5 regions of Glu-Pg (10, 11, 15, 21, 25, 37, 39). The high affinity of IGFBP-3 for Glu-Pg may be explained by the interaction of IGFBP-3 with multiple Glu-Pg kringle domains.

Glu-Pg appears to bind to a specific HBD of IGFBP-3 that is rich in Lys groups. Another Lys-rich peptide, IGFBP-5-(130—142), fails to inhibit Glu-Pg binding to IGFBP-3, suggesting that the particular spatial orientation of Lys groups in the HBD is required for binding. The sequence and location of the IGFBP-3 HBD is conserved in two of the other five IGFBPs, IGFBP-5 and IGFBP-6 (5), and IGFBP-5 has been reported previously to bind Glu-Pg (7). Studies presented here show that IGFBP-5 has an approximately fivefold greater affinity for Glu-Pg than does IGFBP-3 and suggest that this binding requires the IGFBP-5 HBD. In contrast, the IGFBP-3 HBD sequence is not conserved in IGFBP-1, IGFBP-1 does not bind appreciably to Glu-Pg, and partial replacement of the IGFBP-3 HBD sequence with homologous sequence from IGFBP-1 (creating IGFBP-3mHBD) results in a fivefold loss of affinity for Glu-Pg. Thus the HBD may play an important role in IGFBP-3 cleavage by providing preferential access to Glu-Pg. The possibility that this may be a more generalized mechanism is suggested by the observation that prekallikrein also binds the IGFBP-3 HBD (data not shown).

Formation of IGFBP-3 and Glu-Pg into a binary complex does not prevent the additional binding of IGF-I to IGFBP-3, thus forming a ternary complex. The binding affinity (K d value) estimated for IGF-I binding to IGFBP-3 in the IGFBP-3/Glu-Pg binary complex is ∼0.47 nM. ThisK d value is in the range reported previously for IGF-I/IGFBP-3 binding (0.03 to 0.5 nM) and is close to the value of 0.23 nM recently reported by biosensor measurement (20). Thus binding of Glu-Pg to IGFBP-3 does not appear to have a major inhibitory effect on the ability of IGFBP-3 to bind IGF-I.

The binding affinity of IGFBP-3 for Glu-Pg is comparable to theK d = 2.0 nM binding affinity of the IGFBP-3/IGF-I binary complex for ALS (24). These results suggest that IGFBP-3/Glu-Pg complexes should form in the circulation, a hypothesis confirmed by data presented in this report. However, the affinity of Glu-Pg for numerous other proteins in the circulation and on the endothelial cell surface may limit the Glu-Pg pool available to interact with IGFBP-3. For example, plasma levels of ∼1.5 μM for histidine-rich glycoprotein and ∼1 μM for α2-antiplasmin suggest that these proteins may bind up to ∼50 and ∼15%, respectively, of circulating Glu-Pg (25). In contrast, ALS has not been reported to bind other proteins and may bind exclusively to IGFBP-3. Evidence that IGFBP-3 is present in a ternary serum complex with IGFs and ALS is based in part on gel filtration studies performed at neutral pH; here, IGFBP-3 is found almost exclusively in the ∼150-kDa serum fractions. However, Glu-Pg has a similarM r to ALS; thus a portion of the IGFBP-3 circulating in the ∼150-kDa fractions could be complexed to Glu-Pg. Although the binary IGFBP-3/Glu-Pg complex binds IGF-I, IGFBP-3/Glu-Pg should not form a ternary complex with ALS, since both ALS (3) and Glu-Pg bind IGFBP-3 through the HBD.

Although citrate plasma and serum were used to demonstrate that IGFBP-3/Glu-Pg complexes exist in vivo, it is unclear how much IGFBP-3/Glu-Pg complex is present in the circulation. More of this complex was detected in citrate plasma, which contains essentially all circulating proteins, than in serum, which is depleted of proteins involved in clot formation. Nevertheless, IGFBP-3/Glu-Pg complexes were clearly present in serum. In addition, others have attempted to purify ALS from rat serum using an affinity column in which covalently bound IGF-II was complexed with purified IGFBP-3 after the method of Martin and Baxter (27). NH2-terminal sequence analysis of the major purified protein band revealed the presence of two proteins; most was ALS, but Pg was also present in detectable amounts, suggesting that Pg in serum bound specifically to the IGFBP-3 on the affinity column (G. Ooi, C. Y. Lee, and M. M. Rechler, personal communication).

Situations in which Glu-Pg/IGFBP-3 complexes play a physiological role probably depend on Glu-Pg activation, which is important to many biological processes. Under normal physiological conditions, it is unlikely that Glu-Pg activation to plasmin represents a significant contribution to IGFBP-3 proteolysis in the plasma phase of blood, because, under normal conditions, Glu-Pg activation in the circulation is restricted to cellular and fibrin surfaces (28). However, an important biological process that does appear to involve Glu-Pg/IGFBP-3 complexes is wound healing. At the wound site, Pg activators convert Glu-Pg to plasmin within the fibrin clot. Plasmin is known to proteolyze IGFBP-3 to fragments that, because of lowered affinity for IGF peptides, can release IGFs to target tissues where they may stimulate mitogenic and metabolic pathways (2, 4, 8, 13, 19, 22); in this instance, IGFs released from the fibrin clot would stimulate the wound healing process (33). Data presented here suggest that physiological plasma levels of tPA can activate Glu-Pg complexed with IGFBP-3, resulting in IGFBP-3 proteolysis. In addition, recent studies show that Glu-Pg/IGFBP-3/IGF complexes localize to fibrin clots and that addition of tPA leads to IGFBP-3 proteolysis and release of IGFs from these clots (Campbell, unpublished observations), suggesting that the mechanism outlined above is operative in wound healing. It is likely then that in wounds or in any setting in which Pg activators are present along with Glu-Pg/IGFBP-3/IGF complexes, activated Glu-Pg specifically cleaves bound IGFBP-3; this lowers the affinity of IGFBP-3 for IGFs, effectively allowing IGFs to confer their mitogenic and metabolic effects on local target tissues.

Acknowledgments

We thank Heide Eash and Richard Ting for assistance in conducting the solid- and fluid-phase plate-binding assays.

Footnotes

  • Address for reprint requests: D. R. Powell, Texas Children’s Hospital, Feigin Center, MC# 3-2482, 6621 Fannin, Houston, TX 77030.

  • This work was supported in part by National Institute of Diabetes and Digestive and Kidney Diseases Award RO1-DK-38773 (to D. R. Powell).

REFERENCES

View Abstract