Insulin-like growth factor (IGF)-binding protein-3 (IGFBP-3) functions as an IGF-reversible inhibitor of IGFBP-4 proteolysis.

Previous studies have shown that insulin-like growth factor (IGF)-binding protein-4 (IGFBP-4) is degraded only in the presence of exogenous IGFs; however, we found that cation-dependent proteinase activity present in conditioned medium of MC3T3-E1 osteoblasts degrades 125I-recombinant human (rh)IGFBP-4 in the absence of IGFs. Addition of IGF-I, IGF-II, or insulin to conditioned medium had little affect on 125I-rhIGFBP-4 proteolysis, while extraction of IGFs resulted in only a approximately 10% reduction in proteinase activity. Since factors other than IGFs appeared to be involved in regulating IGFBP-4 proteolysis, we hypothesized that IGFBP-3, an IGFBP produced by many cell lines, but not MC3T3-E1 cells, might function as an inhibitor of IGFBP-4 proteolysis. Addition of rhIGFBP-3 to conditioned media inhibited 125I-rhIGFBP-4 proteolysis by 90%, while IGF-I and IGF-II reversed the inhibitory effects of rhIGFBP-3 in a dose-dependent manner. 125I-rhIGFBP-4 proteolysis was not inhibited by N-terminal rhIGFBP-3 fragments that bind IGFs, but was inhibited by two synthetic peptides corresponding to sequences contained in the mid-region or C-terminal region of IGFBP-3. Both inhibitory peptides contain highly basic, putative heparin-binding domains and heparin partially reversed the inhibitory effects of rhIGFBP-3 on 125I-rhIGFBP-4 proteolysis. These data demonstrate that rhIGFBP-3 inhibits IGFBP-4-degrading proteinase activity and binding of IGFs or glycosaminoglycans to IGFBP-3 may induce conformational changes in the binding protein, causing disinhibition of the proteinase.

Previous studies have shown that insulin-like growth factor (IGF)-binding protein-4 (IGFBP-4) is degraded only in the presence of exogenous IGFs; however, we found that cation-dependent proteinase activity present in conditioned medium of MC3T3-E1 osteoblasts degrades 125 I-recombinant human (rh)IGFBP-4 in the absence of IGFs. Addition of IGF-I, IGF-II, or insulin to conditioned medium had little affect on 125 I-rhIGFBP-4 proteolysis, while extraction of IGFs resulted in only a ϳ10% reduction in proteinase activity. Since factors other than IGFs appeared to be involved in regulating IGFBP-4 proteolysis, we hypothesized that IGFBP-3, an IGFBP produced by many cell lines, but not MC3T3-E1

cells, might function as an inhibitor of IGFBP-4 proteolysis. Addition of rhIGFBP-3 to conditioned media inhibited 125 I-rhIGFBP-4 proteolysis by 90%, while IGF-I and IGF-II reversed the inhibitory effects of rhIGFBP-3 in a dose-dependent manner. 125 I-rhIGFBP-4 proteolysis was not inhibited by N-terminal rhIGFBP-3 fragments that bind
IGFs, but was inhibited by two synthetic peptides corresponding to sequences contained in the mid-region or C-terminal region of IGFBP-3. Both inhibitory peptides contain highly basic, putative heparin-binding domains and heparin partially reversed the inhibitory effects of rhIGFBP-3 on 125 I-rhIGFBP-4 proteolysis. These data demonstrate that rhIGFBP-3 inhibits IGFBP-4-degrading proteinase activity and binding of IGFs or glycosaminoglycans to IGFBP-3 may induce conformational changes in the binding protein, causing disinhibition of the proteinase.
Insulin-like growth factor (IGF) 1 -binding proteins (IGFBPs) are a group of six homologous, yet distinct proteins (IGFBPs 1-6) which bind both IGF-I and IGF-II with high affinity (for recent reviews, see Refs. [1][2][3][4]. Among these IGFBPs, IGFBP-4 has been shown to function as an inhibitor of IGF bioactivity (1)(2)(3)(4). For instance, IGFBP-4 was the only inhibitory IGFBP isolated from conditioned media from a colon adenocarcinoma cell line, although other IGFBPs were present in the condi-tioned medium (5). Furthermore, Malpe et al. (6) have demonstrated that treatment of human bone cells with antisense oligonucleotides directed against IGFBP-4 mRNA results in decreased production of IGFBP-4 and a striking increase in cellular proliferation, suggesting that IGFBP-4 plays a major role in regulating cellular proliferation. Since IGFBP-4 has been purified from a number of sources (1)(2)(3)(4) and since mRNA for IGFBP-4 has been detected in all tissues studied by Shimasaki et al. (7), it is likely that IGFBP-4 may serve to restrain IGF activity in many tissues.
Although IGFBP-4 functions as a potent inhibitor of IGF action, the factors controlling production, secretion, and turnover of IGFBP-4 have only recently been addressed. IGFs have been shown to decrease concentrations of IGFBP-4 in the conditioned media of human fibroblasts (8 -10), human breast cell carcinoma (11), and human decidual cells (12,13). However, IGF-I or IGF-II have little or no effect on IGFBP-4 mRNA in several cell lines (10,14) and the effect of IGF-I on IGFBP-4 levels in conditioned media of human fibroblasts (8,9) and human decidual cells (15) is not blocked by a monoclonal antibody to the type-1 IGF receptor. These observations suggest that IGFs might decrease IGFBP-4 concentrations via posttranslational mechanisms.
Consistent with this hypothesis, recent studies from our laboratory have demonstrated that in human and sheep fibroblasts the addition of IGFs induces IGFBP-4 proteolysis by a proteinase(s) not yet identified (16). Since this original report, IGF-dependent IGFBP-4 proteolysis has been confirmed in human fibroblasts (17), and similar IGF-dependent IGFBP-4 proteinase activity has been reported in human decidual cell cultures (15), vascular smooth muscle cells (18), and human bone cells (19). The action of the IGF-dependent IGFBP-4 proteinase(s) provides a novel mechanism through which IGFs can increase their own bioavailability and bioactivity.
In the current study, we use the murine MC3T3-E1 osteoblast cell line to explore the role of IGFs in the regulation and induction of IGFBP-4 proteolysis. Herein, we demonstrate that IGFBP-3 can function in a unique role as an inhibitor of IGFBP-4 proteolysis; however, its inhibitory effects can be reversed by the presence of IGFs.

EXPERIMENTAL PROCEDURES
Materials-Recombinant human (rh)IGFBP-3 produced in Escherichia coli (rhIGFBP-3) was kindly provided by Dr. Christopher Maack, Celtrix Pharmaceuticals, Santa Clara, CA (20). rhIGFBP-4 was purchased from Austral Biologicals, San Ramon, CA. rhIGF-I was kindly provided by Genentech Inc., South San Francisco, CA, and rhIGF-II was generously supplied by Lilly Research Laboratories, Indianapolis, IN. Recombinant human insulin was from Novo Nordisk Pharmaceuticals Inc., Princeton, NJ. Polyclonal antisera to IGF-I was a gift from Dr. Lewis Underwood, University of North Carolina, Chapel Hill, NC, and a monoclonal antibody to rat IGF-II was purchased from Amano Pharmaceutical Co., Ltd., Nagoya, Japan. Reagents used for SDS-polyacrylamide gel electrophoresis (SDS-PAGE) were purchased from Bio-Rad. Low molecular mass heparin (6 kDa) from porcine intestinal mucosa and all proteinase inhibitors were purchased from Sigma, with the exception of the proteinase inhibitors 3,4-dichloroisocoumarin (3, and L-trans-epoxysuccinyl-leucylamide-(4-guanidino)butane (E-64), which were purchased from Boehringer Mannheim. Na 125 I and Hyperfilm-ECL were obtained from Amersham Corp. Tissue culture plasticware was obtained from Corning Glass Works, Corning, NY. Growth media and cell culture reagents were obtained from Life Technologies, Inc.
To induce cellular differentiation, cells were plated at an initial density of 5 ϫ 10 4 cells/well in 35-mm diameter multiwell plastic culture dishes; cells were then grown in ␣-minimal essential medium/ 10% fetal bovine serum supplemented with 10 mM ␤-glycerol phosphate and 25 g/ml ascorbic acid. Differentiating cultures were refed every 3 days throughout a 30-day culture period. Under these conditions, these cells display an orderly developmental pattern in culture; replication of preosteoblasts (days 1-10) is followed by growth arrest and the sequential expression of mature osteoblastic characteristics, including increased alkaline phosphatase activity (Ͼday 10), matrix accumulation (days 14 -21), osteocalcin expression (Ͼday 21), and eventual mineralization (Ͼday 25) (22). Every 2-3 days during the 30-day culture period, selected plates were washed and incubated for 48 h in serum-free medium (Dulbecco's modified Eagle's medium/F-12 ϩ 0.1% bovine serum albumin). Cell-free conditioned media collected during these 48-h incubations were used in the subsequent experiments.
Preparation of IGF-extracted MC3T3-E1 Conditioned Media-Endogenous IGF-I and IGF-II present in MC3T3-E1 conditioned media were removed by immunoabsorption using anti-IGF-I and anti-IGF-II protein G-Sepharose. To prepare the immunoaffinity matrix, 18 l of anti-IGF-I polyclonal antisera and 600 ng of an anti-IGF-II monoclonal antibody were incubated with 150 l of protein G-Sepharose overnight at room temperature. The matrix was washed extensively with 20 mM Hepes, 1 M NaCl, pH 7.4, to remove excess serum. Immunoabsorption was performed by incubating 400-l samples of MC3T3-E1 conditioned media with 75 l of the immunoaffinity matrix overnight at 4°C in a microcentrifuge tube. The mixtures were centrifuged at 2,000 rpm for 5 min at 4°C, and the supernatant was removed and used for 125 I-rhIGFBP-4 proteinase assays as described below.
Degradation of 125 I-IGFBP-4 by MC3T3-E1 CM-125 I-rhIGFBP-4 proteinase assays using cell-free conditioned media were performed as described previously, with minor modifications (16). rhIGFBP-4 was labeled with Na 125 I using the chloramine-T method to a specific activity of ϳ50 Ci/g of protein.
Preparation of IGFBP-3 Fragments Produced by MMP-3-Sixty g of rhIGFBP-3 was digested by 200 ng of MMP-3 (kindly provided by Dr. Hideaki Nagase, University of Kansas Medical Center, Kansas City, KS) in a total volume of 60 l of 50 mM Tris, pH 7.5, 0.15 M NaCl, 10 mM CaCl 2 , 0.02% NaN 3 , 0.05% Brij 35 (TNC buffer) for 8 h at 37°C (24). The digestion was stopped by the addition of EDTA (final concentration: 10 mM), and the digestion products were analyzed on 15% SDS gels under reducing conditions. Seven rhIGFBP-3 fragments, designated "a-f," were identified as described elsewhere (24) (see Fig. 5). To separate rhIGFBP-3 fragments produced by MMP-3 digestion, the digestion mixture was incubated with 1 ml of heparin-Sepharose (Sigma) overnight at 4°C. The matrix was washed with 50 mM Tris-HCl, pH 7.4, and heparin-bound fragments were eluted in 50 mM Tris-HCl, pH 7.4, containing 1 M NaCl. All fractions were analyzed by SDS-PAGE and stained with Coomassie Blue. The wash fractions (i.e. 50 mM Tris-HCl, pH 7.4) contained only the smallest IGFBP-3 fragments (fragments e and f) produced by MMP-3, which correspond to the first 100 -110 N-terminal amino acids of IGFBP-3. Fragments a-d bound to the heparin column, and all four fragments were eluted with 50 mM Tris-HCl, pH 7.4, containing 1 M NaCl. The inability of fragments e and f to bind heparin-Sepharose was anticipated, since their sequences do not contain either of the two putative heparin-binding sites present in IGFBP-3 (2, 25). In covalent cross-linking studies, fragments e and f bound specifically 125 I-rhIGF-I and 125 I-rhIGF-II with IC 50 values of ϳ8.9 and ϳ3.0 nmol/liter, respectively. 2 These findings are consistent with a previous report demonstrating that an 88-amino acid N-terminal mutant of IGFBP-3 bound 125 I-IGF-I (20). Fragments a-f together or fragments e and f together were tested for their abilities to alter 125 I-rhIGFBP-4 degradation by MC3T3-E1 conditioned media as described above.
Preparation of Synthetic hIGFBP-3 Peptides-Peptides based on amino acid sequences contained in both the non-homologous mid-region of hIGFBP-3 (peptides I, II, and III) and in the highly conserved C terminus (peptide IV) (see Fig. 5 below) were produced by solid phase peptide synthesis using 9-fluorenylmethoxycarbonyl chemistry. The sequences are as follows: 94 SRLRAYLLPAPPAP 107 (peptide I), 149 KKGHAKDSQRYKVDYESQS 167 (peptide II), 168 TDTQNFSSESK-RETEY 183 (peptide III), and 214 DKKGFYKKKQLRPSKGR 230 (peptide IV). All peptides were purified on a Vydac C-8 HPLC column using a Gilson automated HPLC system, and were shown to be Ͼ98% pure. Sequence verification was performed by electrospray mass spectrometry. All peptides were synthesized with an additional N-terminal cysteine for use in thiol-coupling reactions. The internal cysteine in peptide IV was acetylmethylated because it is normally involved in a disulfide bond. All synthetic peptides were tested for their abilities to alter 125 I-rhIGFBP-4 degradation by MC3T3-E1 conditioned media as described above.
Statistical Analysis-All experiments were repeated using conditioned media from two to five different experiments. Relative concentrations of intact 125 I-rhIGFBP-4 and fragments of 125 I-rhIGFBP-4 were determined by scanning densitometry (Beckman). Graphic data were normalized to the proteolysis of 125 I-rhIGFBP-4 observed in unconditioned media (i.e. 100% inhibition) and the proteolysis of 125 I-rhIG-FBP-4 observed in cell-free conditioned media (i.e. 100% proteolysis). All data are expressed as the mean Ϯ S.E. Statistical significance between groups was determined by paired Student's t test. Curvefitting and IC 50 values were calculated using InPlot Software (Graph-Pad Software, San Diego, CA).

Characterization of IGFBP-4 Proteases in MC3T3-E1 Cultures-
To determine if MC3T3-E1 osteoblasts produce IGFBP-4-degrading proteinases, conditioned media from different time points during differentiation were examined for their abilities to degrade 125 I-rhIGFBP-4. When incubated with conditioned media from MC3T3-E1 osteoblasts, intact 125 I-IGFBP-4 (ϳ28 kDa) was degraded into ϳ20and ϳ14-kDa fragments (Fig. 1). By 24 h, Ͼ50% of the binding protein was degraded (p Ͻ 0.0001, n ϭ 4). IGFBP-4 protease activity present in MC3T3-E1 conditioned media increased progressively as osteoblasts matured in culture; maximal proteinase activity was detected in conditioned media from osteoblasts displaying a differentiated phenotype. Therefore, all subsequent studies were performed with conditioned media from Ͼ20 day cultures. To characterize further the IGFBP-4-degrading proteinase(s) present in MC3T3-E1 conditioned medium, samples of cell-free conditioned media were analyzed for their ability to degrade 125 I-rhIGFBP-4 in the presence or absence of various protease inhibitors. As shown in Table I, only the cation-dependent proteinase inhibitors (EDTA and 1,10-phenanthroline) significantly inhibited IGFBP-4-degrading proteinase activity. Serine (phenylmethylsulfonylfluoride, aprotinin, and 3,4-DCI) and cysteine (E-64) proteinase inhibitors had little or no effect on degradation of 125 I-rhIGFBP-4. These data suggested that the IGFBP-4-degrading proteinase in MC3T3-E1 conditioned medium is a cation-dependent proteinase.
Effects of IGFs on 125 I-rhIGFBP-4 Proteolysis-When IGF-I, IGF-II or insulin were added to MC3T3-E1 conditioned media, only high dose IGF-II (500 ng/ml) induced any significant, although modest, increase in 125 I-rhIGFBP-4 proteolysis (Table  II). Because exogenous IGFs were not required for 125 I-rhIG-FBP-4 degradation and exerted only marginal effects on 125 I-rhIGFBP-4 degradation by MC3T3-E1 conditioned medium, we postulated that endogenously produced IGFs by MC3T3-E1 osteoblasts might account for the constitutive degradation of 125 I-rhIGFBP-4. To test this hypothesis, IGF-I and IGF-II were first immunoabsorbed from cell-free MC3T3-E1 conditioned media and IGF-extracted conditioned media were then analyzed for their ability to degrade 125 I-rhIGFBP-4. Using this method, only a 10% (p ϭ 0.03) decrease in the degradation of 125 I-rhIGFBP-4 was observed. In other experiments, direct addition of anti-IGF-II monoclonal antibodies to conditioned media resulted in no change in 125 I-rhIGFBP-4 proteolysis (data not shown). These data suggested that endogenous IGFs contributed only minimally, if at all, to the constitutive degradation of 125 I-rhIGFBP-4 by MC3T3-E1 cells.
Effect of rhIGFBP-3 on 125 I-rhIGFBP-4 Proteolysis-In previous studies using other cell lines, we and others have demonstrated that the addition of IGFs to cells or conditioned media is essential for the induction of IGFBP-4 protease activity (reviewed in Ref. [1][2][3][4]; in contrast, our current studies failed to demonstrate an absolute requirement for IGFs in inducing IGFBP-4 proteolysis in MC3T3-E1 osteoblasts. To clarify this discrepancy, we searched for differences in the IGF/IGFBP systems between MC3T3-E1 osteoblasts and cell lines previously reported to display only IGF-induced IGFBP-4 proteolysis. Since MC3T3-E1 cells do not produce IGFBP-3 (26), while other cell lines exhibiting IGF-induced IGFBP-4 proteolysis do produce IGFBP-3 (15)(16)(17)(18)(19), we speculated that IGFBP-3 might function as an inhibitor of IGFBP-4 proteolysis.
Determination of the Epitope(s) in IGFBP-3 Involved in Inhibition of IGFBP-4 Proteolysis-To investigate the mechanism through which rhIGFBP-3 inhibits IGFBP-4-degrading proteinase activity, the effects of (a) intact rhIGFBP-3, (b) rhIG-FBP-3 fragments produced by MMP-3, and (c) IGF-binding, N-terminal rhIGFBP-3 fragments were compared for their inhibitory effects on 125 I-rhIGFBP-4 proteolysis by MC3T3-E1 conditioned media. Fig. 3 demonstrates that the addition of intact rhIGFBP-3 to MC3T3-E1 conditioned media readily inhibited the degradation of 125 I-rhIGFBP-4 (IC 50 ϭ 0.24 g/ml). When a mixture of all rhIGFBP-3 fragments produced by MMP-3 (fragments a-f) were added to MC3T3-E1 conditioned media, 125 I-rhIGFBP-4 degradation was also inhibited in a dose-dependent fashion, with ϳ2-fold less potency than intact rhIGFBP-3. In contrast, the addition of IGF-binding, N-terminal fragments of IGFBP-3 (fragments e and f) had little or no effect on 125 I-rhIGFBP-4 proteolysis (IC 50 Ͼ 40.0 g/ml), suggesting that inhibitory activity does not reside in the first 110 amino acids present in IGFBP-3 (see Fig. 4). Since N-terminal IGFBP-3 fragments bind IGF-I and IGF-II, sequestration of endogenous IGFs could not explain the inhibitory effect of IGFBP-3 on 125 I-rhIGFBP-4 proteolysis. These observations support earlier conclusions that endogenous IGFs contribute little to the constitutive IGFBP-4-degrading proteinase activity in MC3T3-E1 conditioned media. Together, these data suggested that the inhibitory domain(s) in IGFBP-3 resides in the non-homologous, mid-region of the molecule and/or in the Cterminal domain (Fig. 4) and that degradation of rhIGFBP-3 fails to destroy the inhibitory effects of rhIGFBP-3 on 125 I-rhIGFBP-4 proteolysis.
To identify the epitope(s) in the last ϳ150 amino acids of

IGFBP-3 Is an Inhibitor of IGFBP-4 Proteolysis
IGFBP-3 that inhibit IGFBP-4 degradation, four synthetic peptides were prepared. Three of these peptides correspond to epitopes present in the mid-region of the IGFBP-3 molecule (peptides I, II, and III), a region that has little or no homology with the other five IGFBPs (1); the fourth peptide (peptide IV), corresponds to a region in the C-terminal portion of the binding protein (Fig. 4). Table III demonstrates that both peptide II and peptide IV, when added to MC3T3-E1 conditioned media at 200 M/liter, inhibited significantly the degradation of 125 I-rhIG-FBP-4. In contrast, peptides I and III, used at the same concentrations, had no discernible inhibitory activities. Fig. 5 demonstrates that both peptides II and IV inhibited 125 I-rhIGFBP-4 degradation in a dose-dependent manner (Fig. 5, panels A and B, respectively), and each produced a displacement curve parallel to the other peptide (Fig. 5). However, peptide IV (IC 50 ϭ 25 m) was approximately 3-fold more potent than peptide II (IC 50 ϭ 74 M) in inhibiting IGFBP-4degrading proteinase activity. When used together, the peptides demonstrated no additive effect on inhibiting 125 I-rhIG-FBP-4 proteolysis (Fig. 5, panel C).
Because IGFs reversed the inhibitory effects of rhIGFBP-3 on 125 I-rhIGFBP-4 proteolysis, IGF-I was examined for its ability to reverse the inhibitory effects of peptides II and IV on 125 I-rhIGFBP-4 proteolysis by MC3T3-E1 conditioned media. As shown in Fig. 7, IGF-I had little or no effect on reversing the inhibitory effects of peptides II and IV on 125 I-rhIGFBP-4 degradation. In covalent cross-linking studies, it was demonstrated that while 125 I-IGF-I and 125 I-IGF-II both bound to intact rhIGFBP-3 or IGFBP-3 fragments e and f, neither radioligand bound to peptides II and IV (data not shown). These data suggest that IGFBP-3 inhibits IGFBP-4 proteolysis via epitopes present within the amino acid sequences contained in peptides II and IV and that IGFs must be able to bind to IGFBP-3 in order to reverse the inhibitory effects of IGFBP-3 on IGFBP-4 proteolysis. Fig. 5, peptides II and IV both contain putative heparin-binding domains (2,25). Peptide II contains the sequence 149 KKGHA 153 which resembles a short heparin-binding domain (BBXBX; B ϭ basic amino acid, and X ϭ non-basic amino acid) and peptide IV contains the sequence 219 YKKKQCRP 226 , which resembles a long heparin-binding motif (XBBBXXBX) (25). This suggested that both of these highly basic, putative heparin-binding domains present in IG-FBP-3 could inhibit IGFBP-4-degrading activity in MC3T3-E1

IGFBP-3 Is an Inhibitor of IGFBP-4 Proteolysis
conditioned media; therefore, we next examined the effect of heparin on modulating the inhibitory effect of rhIGFBP-3 on 125 I-rhIGFBP-4 proteolysis. For this purpose, a submaximal dose of rhIGFBP-3 (200 ng/ml) or of peptide IV (20 M) was added to MC3T3-E1 conditioned medium with or without heparin (100 g/ml). All solutions were incubated with 125 I-rhIG-FBP-4 and processed in parallel with conditioned media containing no additives or heparin alone. As Fig. 8 demonstrates, heparin alone (bar 4) had no significant effect on 125 I-rhIG-FBP-4 proteolysis. In contrast, when heparin was added to conditioned media containing rhIGFBP-3, heparin reversed the inhibitory effects of rhIGFBP-3 on 125 I-rhIGFBP-4 proteolysis by 33% (p Ͻ 0.01) (compare bars 2 and 3). Similarly, heparin almost entirely reversed the inhibitory effects of peptide IV on 125 I-rhIGFBP-4 proteolysis (data not shown). functions as a potent inhibitor of IGF action (for recent reviews, see Refs. [1][2][3][4]. Previously, we demonstrated that IGFs can induce the proteolytic degradation of IGFBP-4 into fragments that display little or no affinity for IGFs, providing a mechanism by which IGFs can directly increase their own bioavailability and/or bioactivity (16). Subsequently, Conover et al. (17) demonstrated that the degradation of IGFBP-4 increased IGF activity in human fibroblast cultures. How IGFs induce IG-FBP-4 proteolysis remains unclear; we (16) and others (15) have hypothesized that IGFs induce IGFBP-4 degradation by binding to IGFBP-4, thus making it more susceptible to proteolysis, while others have proposed that IGFs directly activate an IGFBP-4 proteinase (17). Insights from the current studies would suggest that neither of these hypotheses is entirely correct, and that a more complex mechanism is involved in IGF-induced IGFBP-4 degradation.
Since other studies have now demonstrated IGF-inducible IGFBP-4 degrading proteinase activity in a number of cell lines (1-4), including bone cells (19), we chose to explore IGFBP-4 proteolysis in a murine bone cell line, MC3T3-E1 osteoblasts. Initial studies demonstrated that the IGFBP-4-degrading proteinase(s) present in MC3T3-E1 conditioned media cleaved the protein into two major fragments and that the proteinase was cation-dependent, making it similar to IGFBP-4-degrading proteinase activity reported in other cell lines (15)(16)(17). Nevertheless, one major difference existed between the IGFBP-4-degrading proteinase activity observed in MC3T3-E1 conditioned media compared with that produced by other cell lines; no exogenous IGFs were required to induce IGFBP-4 proteolysis (1)(2)(3)(4)(15)(16)(17)(18)(19). In fact, addition of IGF-I to MC3T3-E1 conditioned media had no effect on 125 I-rhIGFBP-4 degradation, while a maximal dose of IGF-II (500 ng/ml) only modestly increased the degradation of the binding protein. This is in marked contrast to our previous report using human and sheep fibroblast condition media, where we demonstrated that little or no proteolysis of IGFBP-4 occurred during a prolonged incubation (i.e. 72 h), yet with the addition of IGF-I or IGF-II, almost complete proteolysis of IGFBP-4 was achieved (16). Since MC3T3-E1 osteoblasts secrete both IGF-I and IGF-II (26,27), it was possible that endogenous IGFs induced IGFBP-4 proteolysis in this cell line. In this regard, Durham et al. (28) have very recently demonstrated that in certain primary human bone cell lines, IGFBP-4 can be degraded without the addition of IGFs, while in other bone cell lines the addition of IGFs is necessary for IGFBP-4 proteolysis. They suggest that cell lines not requiring exogenous IGFs for IGFBP-4 proteolysis produced more IGFs than those cell lines requiring exogenous IGFs for IGFBP-4 proteolysis. Since immunoabsorption of IGFs from conditioned media or the addition of IGF-binding, N-terminal fragments of IGFBP-3 to conditioned media had little or no effect on inhibiting 125 I-rhIGFBP-4 degradation, endogenous IGFs do not appear to contribute significantly to the constitutive degradation of 125 I-rhIGFBP-4 observed in MC3T3-E1 cells. Furthermore, since many cell lines that display IGF-induced IGFBP-4 proteolysis also produce some IGF-I and/or IGF-II, it seemed unlikely that the presence of IGFs in MC3T3-E1 conditioned media could account for the constitutive nature of the IGFBP-4-degrading proteinase activity.
We have demonstrated that MC3T3-E1 cells secrete immunoreactive IGFBP-2, -4, and -5 (26), while IGFBP-3 is not produced by these cells (26) and the cells produce no IGFBP-3 mRNA (29). Based on the observation that MC3T3-E1 conditioned media contains little or no IGFBP-3, while other cell lines displaying IGF-dependent IGFBP-4 proteolysis do produce IGFBP-3 (15)(16)(17)(18)(19), we postulated that IGFBP-3 might function as an inhibitor of IGFBP-4 proteolysis. Indeed, the addition of rhIGFBP-3 to conditioned media significantly inhibited the constitutive degradation of 125 I-rhIGFBP-4 by conditioned media from MC3T3-E1 cells in a dose-dependent manner. This demonstrated that exogenous IGFBP-3 is an IGFBP-4 proteinase inhibitor and also suggests that endogenous IG-FBP-3 could function as a physiological inhibitor of IGFBP-4 proteolysis since it is present in a variety of biological fluids at concentrations ϳ5-20 times greater (30) than the IC 50 needed to inhibit IGFBP-4 degradation. IGFBP-3 may also be induced in certain cell systems and function as an IGFBP-4 proteinase inhibitor. For instance, Conover and colleagues (31) have demonstrated that treatment of human fibroblasts with phorbol esters induces an inhibitor(s) of IGFBP-4 proteolysis, while Albiston et al. (32) have shown that IGFBP-3 promotor activity is increased when transfected cells are treated with a phorbol ester. Therefore, it is possible that the phorbol ester-induced inhibitor of IGFBP-4 proteolysis is IGFBP-3.
Similar to the effects of intact IGFBP-3 and inhibitory IGFBP-3 fragments, two peptides inhibited IGFBP-4 proteolysis, while two other peptides did not. The feature common to both inhibitory peptides was that each contained one of the two highly basic, putative heparin-binding domains present in IGFBP-3 (2,25). Further analysis revealed that peptide IV, which contains a long heparin-binding motif, was ϳ3 times more potent in inhibiting IGFBP-4 proteolysis than was peptide II, which contains a short heparin-binding site, and there was no demonstrable synergy when both peptides were used together. This suggests that both peptides work through a similar inhibitory mechanism, although the specifics of the mechanism are currently unknown and are under investigation. Preliminary data from our laboratory demonstrate that other proteins, such as fibronectin and vitronectin, which are commonly found in cell cultures and which contain both long and short forms of heparin-binding motifs, do not inhibit IG-FBP-4 proteolysis by MC3T3-E1 conditioned media. 3 However, IGFBP-5, which contains almost the same stretch of basic residues that is present in IGFBP-3 and peptide IV (1), also inhibits IGFBP-4 proteolysis, 3 suggesting that these basic domains are specific in their ability to inhibit IGFBP-4 proteolysis. These highly basic regions may interact with IGFBP-4 itself, protecting it from proteolysis; however, this mechanism seems doubtful since heterodimerization of IGFBPs has not been reported to date. These regions may function as competitive substrates for the IGFBP-4 proteinase; however, this seems unlikely since IGFBP-4 itself contains no heparin-binding motifs (2) and since a recent study demonstrated that the cleavage site in human IGFBP-4 produced by the IGF-depend- ent IGFBP-4 proteinase is at Met 135 -Lys 136 (33), a bond that is not present in either of the inhibitory peptides. Interestingly, Nam et al. (34) have demonstrated recently that an IGFBP-5 synthetic peptide that is ϳ80% homologous with peptide IV significantly inhibited the proteolysis of IGFBP-5 by a partially purified IGFBP-5-degrading proteinase. Thus, together these data suggest that IGFBPs containing heparin-binding domains (i.e. all IGFBPs with the notable exception of IGFBP-4; Ref. 2) may function as direct inhibitors of the IGFBP-4 proteinase(s). Furthermore, the finding that IGF-I was unable to reverse the inhibitory effects of peptides containing heparin-binding domains suggests that fragments of IGFBP-3 that contain one or more of these domains, but do not contain IGF-binding sites, might function as inhibitors whose effects are not reversible by IGFs.
The mechanism by which IGFs reverse the inhibitory effects of IGFBP-3 on IGFBP-4 degradation is currently unknown; however, one explanation is that binding of IGFs to IGFBP-3 results in a conformational change in the binding protein, making the putative heparin-binding domains less available to inhibit IGFBP-4 proteinase activity. Several lines of investigation support this hypothesis. For instance, in vitro IGFBP-3 binds to cell surfaces and/or cell matrix via glycosaminoglycans and possibly through specific IGFBP-3 cell-surface receptors (see Ref. 4 for review). How IGFBP-3 binds to cell monolayers is currently unclear; however, Bar et al. (35) have shown that binding of IGFBP-3 to endothelial cells involves the putative heparin-binding motif contained in the C terminus of IGFBP-3 and in peptide IV used in our study. Despite its association with cell monolayers, IGFBP-3 can be released from cell surfaces when bound to IGFs (4). Together, these data suggest that unsaturated IGFBP-3 may present certain epitopes (i.e. the putative heparin-binding domains) on its surface, which interact with a variety of molecules such as glycosaminoglycans and cell-surface receptors; however, once bound to IGFs, IGFBP-3 may undergo a conformational change and lose its affinity for these molecules. In the same context, binding of IGFs to IGFBP-3 may alter its conformation in such a way that inhibitory epitopes present in the molecule are no longer available to inhibit the IGFBP-4 proteinase.
In contrast to recent reports by Gockerman and Clemmons (36) and Arai et al. (37) demonstrating that heparin inhibits IGFBP-2 proteinase activity produced by porcine aortic smooth muscle cells and IGFBP-5 proteinase activity produced by human fibroblasts, respectively, our data suggest that heparin has little or no effect on IGFBP-4 degradation by MC3T3-E1 conditioned media. Nevertheless, heparin significantly reversed the inhibitory effects of IGFBP-3 and peptide IV on 125 I-rhIGFBP-4 degradation by MC3T3-E1 conditioned media. The divergent effects of heparin on degradation of these various IGFBPs may simply reflect the differences in the proteinases that are involved. However, it is also possible that heparin alters the inhibition of these proteinases in divergent ways. For example, heparin has been shown to enhance inhibition of thrombin and factor Xa by antithrombin III (38), but decrease the rate of inhibition of neutrophil elastase by ␣ 1 -proteinase inhibitor (39), although both proteinase inhibitors are members of the serpin family. The mechanism by which heparin interferes with IGFBP-3's ability to inhibit IGFBP-4 proteolysis is unclear; however, association of heparin with heparin-binding sites present in IGFBP-3 may make them less available to inhibit the proteinase. Because heparin has been shown to interfere with binding of IGFs to IGFBP-5 and IGFBP-3 (25), an alternative hypothesis would be that heparin causes dissociation of IGFs from endogenous IGFBPs, making IGFs available for binding to exogenous rhIGFBP-3, thereby partially mitigating the inhibitory effect of IGFBP-3 on IGFBP-4 proteolysis. Regardless of the mechanism, heparin, and possibly other glycosaminoglycans, may have a role similar to the IGFs in inducing IGFBP-4 proteolysis.
IGFBP-3 has been shown to both enhance and inhibit IGF activity in vitro (reviewed in Refs. [1][2][3][4]. Our data would suggest that this dichotomy may be explained partially by the finding that IGFBP-3 inhibits IGFBP-4 proteolysis, thus decreasing IGF activity. However, when bound to IGFs, and possibly glycosaminoglycans, IGFBP-3 loses its capability to inhibit IGFBP-4 proteolysis, thus enhancing IGF activity by facilitating the degradation of IGFBP-4. Fig. 9 presents a summary of these studies and a hypothetical scheme of how IG-FBP-3 might function in vivo to regulate IGFBP-4 proteolysis. In the absence of IGFBP-3 (and possibly other IGFBPs containing heparin-binding motifs), IGFBP-4 is degraded constitutively (Fig. 9, A) into non-IGF binding fragments. In the presence of IGFBP-3, IGFBP-4 degradation is markedly attenuated (Fig. 9, B). However, the inhibitory effects of IGFBP-3 can be reversed via binding of IGFs to IGFBP-3 ( Fig. 9, C). Similarly, if IGFBP-3 is bound to glycosaminoglycans or cell surfaces, heparin-binding domains present in the molecule may be less available for interaction with the proteinase (Fig. 9, D). This scenario may be altered in the event that IGFBP-3 is degraded by proteinases such as MMPs, since these proteinases can cleave the N-terminal IGF-binding domain from the heparin- binding domains present in the mid-region and C-terminal tail of the molecule (24). In this instance, fragments containing the heparin-binding motif(s) could inhibit proteinase activity; yet since they bind little or no IGFs, their inhibitory effects may not be reversible by IGFs, making them IGF-resistant IGFBP-4 proteinase inhibitors (Fig. 9, E). Together these data suggest a new role for IGFBP-3 as an IGFBP-4 proteinase inhibitor and they exemplify the intricate complexities involved in the IGF/ IGFBP/IGFBP-proteinase system.