Fibronectin binds insulin-like growth factor-binding protein 5 and abolishes Its ligand-dependent action on cell migration.

Insulin-like growth factor-binding protein 5 (IGFBP-5) is a secreted protein that binds to insulin-like growth factors (IGFs) and modulates IGF actions on cell proliferation, differentiation, survival, and motility. IGFBP-5 also regulates these cellular events through IGF-independent mechanisms. To elucidate the molecular mechanisms governing these diverse actions of IGFBP-5, we screened a human cDNA library by a yeast two-hybrid system using IGFBP-5 as bait and identified fibronectin (FN) as a potential IGFBP-5-interacting partner. The complex formation of IGFBP-5 and FN was established by glutathione S-transferase pull-down, solution, and solid phase binding assays using glutathione S-transferase-IGFBP-5 and native IGFBP-5 in vitro and by co-immunoprecipitation in vivo. Binding assay using deletion mutants indicated that the IGFBP-5 C domain binds to the 10th and 11th type I repeats of FN. IGFBP-5 potentiated IGF-I-induced cell migration in FN-null, but not in wild-type, mouse embryonic cells. When FN was reintroduced either as an adhesive substrate or in solution to the FN-null cells, the potentiating effect of IGFBP-5 on IGF-I-induced cell migration was abolished. Binding of IGFBP-5 to FN had no effect on the ability of IGFBP-5 to bind IGF-I, but it increased the proteolytic degradation of IGFBP-5. Inhibition of IGFBP-5 proteolysis restored the potentiating effect of IGFBP-5. These results suggest that FN and IGFBP-5 bind to each other, and this binding negatively regulates the ligand-dependent action of IGFBP-5 by triggering IGFBP-5 proteolysis.

Insulin-like growth factor-binding protein 5 (IGFBP-5) is a secreted protein that binds to insulin-like growth factors (IGFs) and modulates IGF actions on cell proliferation, differentiation, survival, and motility. IGFBP-5 also regulates these cellular events through IGF-independent mechanisms. To elucidate the molecular mechanisms governing these diverse actions of IGFBP-5, we screened a human cDNA library by a yeast two-hybrid system using IGFBP-5 as bait and identified fibronectin (FN) as a potential IGFBP-5-interacting partner. The complex formation of IGFBP-5 and FN was established by glutathione S-transferase pull-down, solution, and solid phase binding assays using glutathione S-transferase-IGFBP-5 and native IGFBP-5 in vitro and by co-immunoprecipitation in vivo. Binding assay using deletion mutants indicated that the IGFBP-5 C domain binds to the 10 th and 11 th type I repeats of FN. IGFBP-5 potentiated IGF-I-induced cell migration in FN-null, but not in wildtype, mouse embryonic cells. When FN was reintroduced either as an adhesive substrate or in solution to the FN-null cells, the potentiating effect of IGFBP-5 on IGF-Iinduced cell migration was abolished. Binding of IGFBP-5 to FN had no effect on the ability of IGFBP-5 to bind IGF-I, but it increased the proteolytic degradation of IGFBP-5. Inhibition of IGFBP-5 proteolysis restored the potentiating effect of IGFBP-5. These results suggest that FN and IGFBP-5 bind to each other, and this binding negatively regulates the ligand-dependent action of IGFBP-5 by triggering IGFBP-5 proteolysis.
Insulin-like growth factor-binding proteins (IGFBPs) 1 are a family of secreted proteins that bind to IGFs and modulate their distribution, stability, and cellular actions (1,2). IGFBP-5 is the most evolutionarily conserved member in this gene family (3). Like other IGFBPs, IGFBP-5 binds to IGFs in the extracellular environment with high affinity (1). In the blood, IGFBP-5 forms a ternary complex with IGF and acid labile subunit that controls the efflux of IGFs from the vascular space and prolongs IGF half-lives (2). Locally produced IGFBP-5 provides a means of localizing IGFs on target cells and can alter the biological activities of IGFs by modulating their in-teraction with the cell surface IGF receptors. IGFBP-5 has been shown to inhibit the proliferative responses of skeletal muscle cells and breast cancer cells to IGF-I (4 -6). In fibroblasts, osteoblasts, and vascular smooth muscle cells (VSMCs), however, IGFBP-5 potentiates the effect of IGFs on cell growth (7)(8)(9)(10)(11)(12). More recent studies have shown that IGFBP-5 also stimulates bone cell growth and mesangial cell and VSMC migration through an IGF-independent mechanism(s) (13)(14)(15)(16).
Although studies describing various cellular actions of IG-FBP-5 have been reported for over a decade, the molecular mechanisms governing these actions are still not well understood. To identify proteins that bind and modulate IGFBP-5 actions, we screened a human aorta cDNA library by a yeast two-hybrid system using human IGFBP-5 as bait and identified fibronectin (FN) as a potential IGFBP-5-interacting partner. Further biochemical and functional studies show that FN and IGFBP-5 directly bind to each other, and this binding abolishes the ligand-dependent action of IGFBP-5 on cell migration.

EXPERIMENTAL PROCEDURES
Materials-All of the chemicals and reagents were obtained from Fisher unless noted otherwise. Recombinant human IGF-I, IGFBP-4, and IGFBP-5 were purchased from GroPep Ltd (Adelaide, Australia). Human plasma FN was obtained from Chemicon International Inc. (Temecula, CA). Fetal bovine serum was purchased from Invitrogen. The Matchmaker TM two-hybrid system 3 and the human aorta Gal4 activation domain-cDNA library were purchased from Clontech Laboratories (Palo Alto, CA).
Yeast Two-hybrid Assay-The Matchmaker TM two-hybrid system 3 (Clontech) was used to identify clones that interact with human IG-FBP-5. The bait, pGBKT7-IGFBP-5, generated by inserting full-length human IGFBP-5 cDNA into the NcoI and BamHI sites of the pGBKT7, was used to screen a human aorta cDNA library constructed in the pACT2 vector (Clontech). Positive clones were selected by growth on a drop-out minimal medium lacking tryptophan, leucine, histidine, and adenine and by activation of X-␣-galactosidase gene. All of the positive clones identified in the screen were retested twice under high stringency. The resulted positive cDNA clones were recovered from the host cells and sequenced at the University of Michigan DNA Sequencing Core. The IGFBP-5 and FN interaction was further studied by cotransforming yeasts with pGBKT7-IGFBP-5 and pGADT7-FN plasmids. Transformed cells were plated on synthetic drop-out minimal medium or the tryptophan-and leucine-deficient medium. The liquid cultures were diluted to an A 600 of 1.0, and 10-fold serial dilutions of each yeast culture were spotted onto drop-out minimal medium and grown at 30°C for 2-3 days.
Plasmid Construction-DNA fragments corresponding to the N (containing residues 1-80 of the mature protein), L (residues 81-169), C (residues 170 -252), NL (residues 1-169), and LC (residues 81-252) domains of human IGFBP-5 were generated by PCR amplification using human IGFBP-5 cDNA as template and the following primer sets: P1 and P2 (N domain), P3 and P4 (L domain), P5 and P6 (C domain), P1 and P4 (NL domains), and P3 and P6 (LC domains). The sequences of these primers are shown in Table I. The resulted PCR products were subcloned into the NcoI and BamHI restriction sites of the pGBKT7 vector to produce fusion proteins with the Gal4 DNA-binding domain.
Cell Culture-Wild-type (MECϩ/ϩ) and FN-null mouse embryonic cells (MECϪ/Ϫ) were a gift from Dr. Jane Sottile (University of Rochester). These cells were grown in a 1:1 mixture of Cellgro (Mediatech, Herndon, VA) and Aim V (Invitrogen). This defined medium is serumfree and contains no adhesion molecules or growth factors. Both wildtype and FN-null cells were grown in tissue culture dishes precoated with 50 g/ml type I collagen (Sigma). Porcine VSMC cells were grown in Dulbecco's minimum essential medium supplemented with 4 mM L-glutamine, penicillin (100 units/ml), streptomycin (100 g/ml), and 10% fetal bovine serum. For GST pull-down or co-immunoprecipitation assays, the cells were lysed in 1ϫ binding buffer (30 mM Tris acetate, 10 mM sodium phosphate, pH 7.4, 0.1% Tween 20, 1 mM EDTA, 2 g/ml leupeptin, 4 g/ml aprotinin, 1 g/ml pepstatin A, and 1 mM phenylmethylsulfonyl fluoride) and used immediately after lysis.
GST Pull-down Assay-The cDNA fragment encoding full-length human IGFBP-5 was cloned into the NcoI and SalI sites of the pGEX-KG vector to produce a fusion protein with GST at the C terminus of IGFBP-5. JM109 Escherichia coli cells were transformed with pGEX-KG-IGFBP5 and grown in LB broth at 37°C to an A 600 of 0.4 -0.6. Following the addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 0.1 mM, the cells were incubated at 30°C for 2 h. After harvesting, the cells were resuspended in French press buffer (50 mM Tris, pH 8.0, 0.1 M NaCl, 1 mM EDTA, 0.05% Tween 20, 1 mM EDTA, 2 g/ml leupeptin, 4 g/ml aprotinin, 1 g/ml pepstatin A, and 1 mM phenylmethylsulfonyl fluoride). The cells were broken by French press and centrifuged at 20,000 ϫ g for 20 min at 4°C, and the supernatant (S1) was collected. S1 was further centrifuged at 100,000 ϫ g for 45 min, and that supernatant (S2) was collected. 10 ml of S2 was loaded onto 0.6 ml glutathione-Sepharose beads (Amersham Biosciences) at 4°C overnight. After centrifugation at 500 ϫ g for 5 min, the beads were extensively washed, resuspended in 0.6 ml of French press buffer, and used for the pull-down assay. 500 l of cell lysates or serum or 10 g of pure plasma FN in 250 l of binding buffer were incubated with GST or GST-IGFBP-5 immobilized on glutathione-Sepharose beads (ϳ15 g/20 l) at 4°C for 3 h. After the incubation, the beads were separated by centrifugation, washed with binding buffer, and boiled in Laemmli loading buffer. The eluted proteins were resolved by SDS-PAGE and analyzed by immunoblot analysis as previously reported (12).
Co-immunoprecipitation-500 l of cell lysates were incubated and precipitated with 2 g of a monoclonal human FN antibody (Pierce) or 2 g of mouse IgG following previously published method (17). The co-precipitated proteins were separated by SDS-PAGE followed by immunoblot analysis using an IGFBP-5 polyclonal antibody (Diagnostic Systems Laboratories, Inc., Webster, TX). Similarly, equal amounts of the cell lysates were immunoprecipitated with the IGFBP-5 antibody and analyzed by immunoblot.
Migration Assay-The cell migration assays were performed using 24-well cell culture inserts (Falcon) coated with 0.1% gelatin. In some experiments, the gelatin-coated inserts were further incubated with human FN (15 g/ml) at room temperature for 2 h. The free FN was removed by washing. Wild-type or FN-null cells were washed with 1ϫ phosphate-buffered saline and trypsinized. After trypsinization, the cells were washed once in 1ϫ phosphate-buffered saline. 5.0 ϫ 10 4 cells in 200 l of the serum-free defined medium (SFM) containing IGFBP-5 (200 ng/ml) or SFM were loaded into the upper wells. In some experiments, FN and/or protease inhibitors were also added into the upper wells. 800 l of the defined medium, with or without IGF-I (50 ng/ml), was loaded into each bottom well. After 6 h of incubation at 37°C, the cells were removed from the upper sides of the inserts. The migrated cells on the underside were stained in toluidine blue, and the total number of migratory cells was counted. The results were expressed as percentages of the values from the SFM control group.
Binding Assays-For solid phase binding assays, 96-well plates (Falcon) were coated with FN (5 g/well) or the vehicle overnight at 4 C. After coating, the plates were washed three times with the binding buffer and blocked with 0.1% bovine serum albumin in 1ϫ phosphatebuffered saline for 2 h at room temperature. After washing, pure IGFBP-5 (10 ng/well in 100 l of 1ϫ phosphate-buffered saline) was added and incubated for 3 h at 4 C. After removing the unbound IGFBP-5 by washing, increasing amounts of 125 I-IGF-I (0 -100,000 cpm) were added and incubated for 1 h at 37°C. The bound 125 I-IGF-I were recovered by adding 200 l of 0.1 N NaOH followed with 200 l of 1% SDS and measured in an automatic Gamma Counter (ICN Biomedicals, Inc., Huntsville, AL).
Immobilized GST or GST-IGFBP-5 beads (15 g/20 l) was incubated with 125 I-IGF-I (50,000 cpm) in the presence and absence of human FN (10 g) in 1ϫ binding buffer at a final volume of 120 l. After incubation for 14 h at 4°C, the beads were pelleted by centrifugation at 500 ϫ g for 1 min and washed three times in binding buffer. The 125 I-IGF-I precipitated was measured in an automatic Gamma Counter. Alternatively, immobilized GST or GST-IGFBP-5 (15 g) was incubated with pure FN (10 g) in the presence or absence of 2 g of IGF-I for 14 h at 4°C. The level of bound FN in the pulled down complex was determined by immunoblot analysis. Solution binding experiments were also carried out using 300 ng of pure human IGFBP-5, 125 I-IGF-I (50,000 cpm), and human FN (8 g) in 1ϫ binding buffer at a final volume of 120 l. The bound 125 I-IGF-I in the complex was separated from free 125 I-IGF-I by immunoprecipitation using the FN antibody. Proteolysis Assay-To determine the possible effect of FN on the stability of IGFBP-5, MECϪ/Ϫ cells were grown to subconfluence in 6-cm cell culture dishes. After being washed three times with SFM, the cells were incubated in 400 l of SFM without or with IGFBP-5 (1.0 g/ml), FN (8.0 g/ml), or IGFBP-5 plus FN at 37°C overnight. The media were collected and concentrated through a Centricon-10 microconcentrator (Millipore, Bedford, MA). The samples were analyzed by immunoblot analysis.
Statistical Analysis-The differences among groups were analyzed by one-way analysis of variance followed by Fisher's protected least significance difference test using Statview (Abacus Concept, Inc.).

TABLE I Primer sequences
The underlined sequences were used to introduce a NcoI cloning site; the bold sequences were used to introduce a BamHI cloning site; and the italic sequences were used to introduce a XhoI cloning site.

Primer names
Primer sequences

Identification of FN as an IGFBP-5-interacting
Protein-A total of 4.3 million colonies from a human aorta library were screened with a chimeric bait containing the Gal4 DNA-BD and full-length human IGFBP-5. 28 of the 60 positive clones encoded polypeptides that corresponded to human FN (Fig. 1A). Restriction enzyme digestion and DNA sequencing analyses grouped these clones into two types (Fig. 1A). To verify this interaction, a plasmid from each group was selected and cotransformed into yeasts with the plasmid pGBKT7-IGFBP-5 or the empty pGBKT7 vector. All of the yeast transformants grew on leucine-and tryptophan-deficient plates (Fig. 1B, left panel), indicating successful transformation. When plated on the selective medium, only the cells co-transformed with both IGFBP-5 and FN were able to grow (Fig. 1B, right panel,  sections 5 and 6).
The IGFBP-5 and FN interaction was further analyzed by GST pull-down and solution binding assays. As shown in Fig.  2A (top panel), when incubated with pure plasma FN, GST-IGFBP-5, but not GST, pulled down FN. Likewise, GST-IGFBP-5, but not GST, was able to pull down FN from fetal bovine serum ( Fig. 2A, middle panel), suggesting that GST-IGFBP-5 binds to plasma FN. FN not only exists in a soluble protomeric form in the blood (i.e. plasma FN), but it is also present in a multimeric form in the ECM of local tissues (cellular FN). To determine whether IGFBP-5 is also capable of interacting with cellular FN, GST pull-down assays were performed using VSMCs, which are known to synthesize and secrete FN (18). The result showed that GST-IGFBP-5 was able to bind cellular FN secreted by these primarily cultured cells ( Fig. 2A, bottom panel). The IGFBP-5 and FN complex formation was also confirmed using purified native IGFBP-5 and FN (see Fig. 6). To test whether IGFBP-5 and FN interact with each other under physiological conditions, co-immunoprecipi-tation experiments were performed using primarily cultured VSMCs. These cells synthesize both IGFBP-5 and FN (18). Immunoprecipitation with a FN antibody resulted in the co-  precipitation of IGFBP-5 (Fig. 2B, upper panel). The association of IGFBP-5 and FN was also confirmed in a reciprocal co-immunoprecipitation experiment (Fig. 2B, lower panel).
The IGFBP-5 and FN Binding Interaction Is Mediated by the 10th and 11th Type I Repeats in FN and the IGFBP-5 C domain-Sequence alignments of the 28 FN clones revealed that they overlapped in a region ranging from residue 1931 to the C-terminal end (Fig. 1A). This region contains two type III repeats, three type I repeats, and the C-terminal tail. Indeed, cells co-transformed with IGFBP-5 and Fn-A, a truncated FN construct that includes the overlapping region, grew well on the selective medium (Fig. 3). To determine the specific structural motif(s) that mediates its binding to IGFBP-5, two truncated constructs, Fn-B and Fn-C, were tested. Fn-B covers the two type III repeats, and Fn-C spans the type I repeats and the tail region. The results indicated that Fn-C, but not Fn-B, was able to interact with IGFBP-5 (Fig. 3). Because Fn-C includes three type I repeats (I 10 , I 11 , and I 12 ) and the C-terminal tail, we examined further truncated forms of FN-C, i.e. Fn-C/ I 10ϩ11ϩ12 . Fn-C/I 10ϩ11ϩ12 interacted with IGFBP-5 as strong as FN-A, suggesting that the type I repeats, but not the C-terminal tail, are important. To determine the role of individual repeat, Fn-C/I 10 , Fn-C/I 11 , and Fn-C/I 12 , three constructs that correspond to each of the three type I repeats, were tested, but none was capable of IGFBP-5 binding. Next, Fn-C/I 10ϩ11 and Fn-C/I 11ϩ12 were generated and tested. The results showed that Fn-C/I 10ϩ11 , but not Fn-C/I 11ϩ12 , was capable of IGFBP-5 binding (Fig. 3). These results suggest that I 10 and I 11 are the IGFBP-5-binding domain in FN.
IGFBP-5 consists of three domains: a highly cysteine rich N-terminal domain, a cysteine-rich C-terminal domain, and a central (L) domain with no cysteine residues. To determine which domain(s) of IGFBP-5 is involved in its binding interac-tion with FN, five truncated IGFBP-5 DNA fragments corresponding to the N, L, C, NL, and LC domains of IGFBP-5 were generated and subcloned into the pGBKT7 vector to generate GAL4 DNA-BD fusion proteins. When the various pGBKT7-IGFBP-5 domain-specific plasmids were introduced into cells together with pGADT7-FN, cells co-transformed with IGFBP-5 C domain and LC domain were able to grow on the selective medium (Fig. 4). Their growth was comparable with those of the full-length IGFBP-5. The cells co-transformed with FN and IGFBP-5 N, L, and NL domains did not grow under the same conditions (Fig. 4). These data suggest that C domain of IGFBP-5 is both required and sufficient for its binding interaction with FN.

Binding with FN Abolishes the Action of IGFBP-5 in Potentiating IGF-I-induced Cell Migration-Because most, if not all, adherent cells constitutively produce FN and deposit it into the ECM, we used FN-null mouse embryonic cells (MECϪ/Ϫ) to determine the functional importance of the IGFBP-5 and FN
interaction. These cells do not produce any FN but are capable of assembling a FN matrix when cultured in the presence of exogenous FN (19). They can be cultured in defined medium without any growth factor or adhesion molecule. When subjected to an IGF-I gradient, the MECϪ/Ϫ and MECϩ/ϩ cells both showed elevated migration (Fig. 5A). At 50 ng/ml, IGF-I caused a 86 Ϯ 13% (p Ͻ 0.05) and a 92 Ϯ 15% (p Ͻ 0.05) increase over the control in MECϪ/Ϫ and MECϩ/ϩ cells, respectively. The addition of IGFBP-5 together with IGF-I at an equal molar concentration resulted in a 202 Ϯ 12% increase in the FN-null cells (Fig. 5A, left panel). This value was significantly higher than that of the IGF-I alone group (p ϭ 0.01). In contrast, exogenous IGFBP-5 had no effect on IGF-I-induced cell migration in MECϩ/ϩ cells (Fig. 5A, right panel). The addition of IGFBP-5 alone had no effect on cell migration in FIG. 3. FN binds to IGFBP-5 through the type I repeats in the C-terminal region. DNA encoding the indicated FN fragments were subcloned into the pGADT7 vector as described under "Experimental Procedures." Each of these pGADT7-truncated FN plasmids was cotransformed into yeasts with pGBKT7-IGFBP-5. The yeast transformants were grown in the leucine-and tryptophan-deficient medium and then dotted on the leucine-, tryptophan-, adenine-, and histidine-deficient plates at different concentrations. Their growth on the leucine-, tryptophan-, adenine-, and histidine-deficient plates are shown. aa., amino acid. either group. Therefore, IGFBP-5 potentiated IGF-I action in the FN-null cells but not in the wild-type cells.
To determine whether these different biological effects of IGFBP-5 were due to the presence or absence of FN, MECϪ/Ϫ cells were subjected to migration assay using inserts precoated with or without FN. The introduction of FN did not change the basal migration rate, nor did it alter the chemotactic responses to IGF-I, but it abolished the potentiating effect of IGFBP-5 on IGF-I-induced migration (Fig. 5B). Likewise, the addition of soluble FN abolished the potentiating effect of IGFBP-5 (Fig.  5C). The effect of FN appeared to be specific for IGFBP-5, because FN did not abolish the ability of IGFBP-4 to inhibit IGF-I-induced cell migration in MECϪ/Ϫ cells (Fig. 5D). In fact, the inhibitory effect of IGFBP-4 was even more pronounced in the presence of FN.
FN Binding Does Not Alter the Ability of IGFBP-5 to Bind IGF-I-Because the FN and IGFBP-5 binding attenuates the ability of IGFBP-5 to potentiate IGF-induced cell migration, we wondered whether binding of IGFBP-5 to FN would inhibit its binding to IGF-I and consequently abolish the potentiating effect of IGFBP-5. To test this idea, the effect of FN on the IGFBP-5 and IGF-I binding was examined in a solid phase binding assay. Pure native IGFBP-5 (10 ng/well) was immobilized onto 96-well plates in the presence or absence of FN (5 g/well). After blocking with bovine serum albumin, increasing amounts of 125 I-IGF-I (0 -100,000 cpm) were added and incubated. The free 125 I-IGF-I was removed by washing, and bound 125 I-IGF-I was measured. There was little 125 I-IGF-I detected in the absence of IGFBP-5, suggesting that IGF-I does not bind to FN. 125 I-IGF-I bound to the immobilized IGFBP-5 in a dosedependent manner, and a plateau was reached at concentration higher than 40,000 cpm. Scatchard analyses indicated that the K d value of IGFBP-5 for IGF-I binding was 3.8 ϫ 10 Ϫ9 M in the absence of FN. In the presence of FN, the K d value was 3.0 ϫ 10 Ϫ9 M. These results suggest that the presence of FN does not significantly affect the IGF-I and IGFBP-5 binding. To determine whether IGFBP-5 can bind to FN and IGF-I simultaneously, 125 I-IGF-I was incubated with GST-IGFBP-5 in the presence or absence of FN. GST was used as a control. A high level of the added 125 I-IGF-I (21,690 Ϯ 8147 cpm) was recovered from the GST-IGFBP-5 pull-down complex (Fig. 6A). In comparison, there was little 125 I-IGF-I in the GST group (1,298 Ϯ 684 cpm). Again, the presence of FN did not change the amount of 125 I-IGF-I bound to GST-IGFBP-5 (20,719 Ϯ 6756 cpm). Western immunoblotting indicated that the added FN was present in the GST-IGFBP-5 complex (Fig. 6A), indicating that the three proteins are present in the same complex. We further studied the complex formation using purified, native IGFBP-5, and FN. 125 I-IGF-I was incubated with FN in the presence or absence of pure IGFBP-5. After incubation, FN was precipitated using a FN antibody. As shown in Fig. 6B, there was little 125 I-IGF-I in the immunoprecipitation complex in the absence of IGFBP-5. When IGFBP-5 was added, a high level of the added 125 I-IGF-I (10,056 Ϯ 3611 cpm) was found in the immunoprecipitation complex, suggesting that IGFBP-5 binds to FN and IGF-I simultaneously. Finally, FN and GST-IGFBP-5 were incubated in the presence or absence of excess amount of IGF-I. The GST-IGFBP-5-bound FN was pulled down and determined by immunoblotting. There was no significant difference in the amount of FN bound to GST-IGFBP-5 in the presence and absence of IGF-I (Fig. 6C). These data suggest that IGFBP-5 can simultaneously bind to IGF-I and FN, and these bindings are independent from each other.
FN Binding Inhibits the Potentiation Effect of IGFBP-5 by Triggering IGFBP-5 Proteolysis-To determine whether binding with FN will alter IGFBP-5 stability, IGFBP-5 was added to MECϪ/Ϫ cells and incubated in the presence or absence of soluble FN. As shown in Fig. 7A (lanes 1 and 2), no endogenous IGFBP-5 was detected in these cells under serum-free control condition or in the FN group. In the absence of FN, the added IGFBP-5 remained intact throughout the experiment (Fig. 7A,  lane 3). In contrast, most of the added IGFBP-5 was degraded into smaller fragments in the presence of FN (Fig. 7A, lane 4). To rule out the possibility that FN itself or other trace protein(s) in the FN preparation degrades IGFBP-5, IGFBP-5 was incubated with FN in the serum-free medium. No degradation of IGFBP-5 was detected (Fig. 7A, lane 5). To determine whether the loss of the potentiation effect of IGFBP-5 is due to its accelerated proteolysis when associated with FN, migration assays were carried out in the presence of a protease inhibitor mixture. As shown in Fig. 7B, the addition of these protease inhibitors did not affect the migratory response to IGF-I, but it restored the potentiating effect of IGFBP-5 on IGF-I-induced migration.

DISCUSSION
In this study, we have unraveled a novel mechanism controlling the ligand-dependent action of IGFBP-5 in regulating cell migration. We show that IGFBP-5 and FN form a complex by direct protein-protein interaction. The binding of IGFBP-5 to FN is independent from the IGFBP-5 and IGF binding and is mediated by the 10 th and 11 th type I repeats of FN and the IGFBP-5 C domain. IGFBP-5 potentiated IGF-I-induced cell migration in the FN-null but not wild-type mouse embryonic cells. When FN was reintroduced to the MECϪ/Ϫ cells, either in adhesive form or added in solution, the potentiating effect of IGFBP-5 on IGF-I-induced cell migration was abolished. This effect of FN was specific to IGFBP-5 because the addition of FN did not disrupt the ability of IGFBP-4 to inhibit IGF-I-induced cell migration under the same experimental conditions. These FIG. 4. IGFBP-5 C domain is required and sufficient for its binding to FN. DNA encoding the IGFBP-5 N, L, C, NL, and LC domains were subcloned into the pGBKT7 vector. Each pGBKT7-IGFBP-5 truncated plasmid was co-transformed into yeasts with pGADT7-FN-A. The full-length pGBKT7-IGFBP-5 was used as control. Those that grew on the leucine-, tryptophan-, adenine-, and histidine-deficient plates are indicated by a plus sign. We have demonstrated that IGFBP-5 and FN directly bind to each other in vitro and are present in the same complex in vivo. Although this finding is consistent with a previous observation showing that IGFBP-5 preferentially binds to FN immobilized on microtiter plate in vitro (7), it contradicts with a recent study reporting that IGFBP-3, but not IGFBP-5 or -4, binds to FN in the pericellular/territorial matrix of human articular cartilage (20). In the study by Martin et al. (20), the authors did not perform any biochemical assays to examine direct FN and IGFBP-5 binding. The conclusion that IGFBP-5 was not associated with FN was primarily based on the low levels of colocalization of immunoreactive IGFBP-5 and FN in these tissue samples. Their co-localization analysis was focused in the transitional and deep zones, where IGFBP-3 is most abundantly expressed. Therefore, the quantification of lo-localization, which is determined by overlapping area of the IGFBP staining and FN staining over the total area, could bias toward the much highly expressed IGFBP-3. It is also possible that the interaction between IGFBP-5 and FN may be tissue-specific. IGFBP-5 has been reported to have high affinity for several extracellular matrix (ECM) components in fibroblasts, bone cells, and VSMCs (7,21,9,15). To this end, it is of interest to note that FN was found with the highest frequency in our yeast two-hybrid screen of a human aorta cDNA library. Of the proximately 60 positive clones obtained, 28 were FN. Because both IGFBP-5 and FN are synthesized by vascular smooth muscles and are present in blood walls (18), the recurring identification of FN is suggestive of a strong interaction between these two proteins and/or a high level of FN expression in blood walls.
The potentiation of IGF-I actions by IGFBP-5 has been documented in a variety of cell culture systems (1,2). This ability of IGFBP-5 has been attributed to its ECM association (7,9,15,21). Studies by Clemmons and co-workers (21)(22)(23) have shown that binding of IGFBP-5 to ECM or glycosaminoglycans reduces its binding affinity for IGFs. This and other observations have led to the proposal that binding of the IGF-I/IGFBP-5 complex to cell surface/ECM may lead to a reduction in the affinity of IGFBP-5 for IGF-I, resulting in the release of IGF-I to its receptors and hence potentiating the IGF actions. Another proposed mechanism whereby IGFBP-5 potentiates IGF-I action is also related to its ECM association. IGFBP-5 binds to a number of ECM proteins, including throm-bospondin-1, osteopontin, plasminogen activator inhibitor-1, and vitronectin, and these ECM proteins enhanced the cellular responsiveness to IGF-I (24,25). Because binding of IGFBP-5 to these ECM proteins is independent from its ability to bind to IGFs, it was postulated that the potentiation of the IGF-I effect by IGFBP-5 is due to the increased bioavailability of the IGF-I to the IGF-I receptor after sequestration and concentration of the IGF-I/IGFBP-5 to the cell surface/ECM through the binding of IGFBP-5 to these ECM proteins. Our finding that the binding of IGFBP-5 to FN abolishes the ability of IGFBP-5 to potentiate IGF-I-induced cell migration is distinct from the previously reported mechanisms. The regulation of IGFBP-5 actions by FN cannot be attributed to alteration in the ligand binding affinity because the binding of IGFBP-5 to FN is independent from its interaction with IGFs and vice versa. In fact, our data indicated that IGFBP-5 can simultaneously bind to IGF-I and FN to form a ternary complex. This finding is in good agreement with a recent study on IGFBP-3 (26) and is consistent with our structural studies showing that FN binds to the IGFBP-5 C domain. Although several conserved residues in the C domain are also involved in IGF binding (27,28), the primary high affinity binding site for IGF is located in the IGFBP-5 N domain (29,30).
The abolishment of the IGF-dependent action of IGFBP-5 by FN is likely due to the accelerated IGFBP-5 proteolysis when associated with FN. This conclusion is supported by the observation that co-incubation of IGFBP-5 and FN resulted in a dramatic decrease in the levels of the intact IGFBP-5 and the corresponding appearance of IGFBP-5 fragments. Furthermore, the addition of protease inhibitors restored the potentiation effect of IGFBP-5 on IGF-I-induced cell migration. In many cell culture systems, IGFBP-5 is partially or completely proteolyzed into fragments with reduced affinity for IGF-I (12,31). The IGFBP-5 proteolytic activity in fibroblast-conditioned medium has been attributed to C1s and/or C1r (32). In addition, PAPP-A (an IGFBP-4 protease), its related metalloproteinase PAPP-A2, plasmin, and others were reported to have IGFBP-5 proteolytic activity (33)(34)(35)(36)(37). The protease(s) responsible for the IGFBP-5 cleavage in MECϪ/Ϫ cells is not known at present, but it is evident that the IGFBP-5 proteolysis in these cells occurs only when FN is present. Because there was little degradation of IGFBP-5 in the absence of FN, we speculate that binding of IGFBP-5 with FN may cause a conformational change of the IGFBP-5 molecule and make it more susceptible to a pre-exiting IGFBP-5 protease. Such mechanisms have been documented for the IGF binding-dependent cleavage of IGFBP-4 by PAPP-A (33,35) and the IGFBP-3 degradation triggered by the E7 protein binding (38). It is also plausible that the presence of FN may activate a dormant enzyme that degrades IGFBP-5. Further studies are needed to identify the specific IGFBP-5 protease(s) and to elucidate the molecular mechanisms underlying the FN-induced IGFBP-5 proteolysis.
In summary, this study has demonstrated that FN and IG-FBP-5 directly bind to each other, and this protein-protein interaction negatively regulates the ligand-dependent action of IGFBP-5 by triggering IGFBP-5 proteolysis. The identification of FN as an IGFBP-5-interacting protein has not only unraveled a novel mechanism controlling the IGF-dependent actions of IGFBP-5, but it has also provided necessary information to warrant future interest. FN plays an important role in the maintenance of normal cell morphology, cell growth, cell motility, metastasis, and cell survival by interacting with ␣ 5 ␤ 1 and ␣ IIb ␤ 3 integrins (39 -42). A recent study reports that the attachment of human Hs578T breast cancer cells to FN was increased by IGFBP-3 but decreased by IGFBP-5 (43). More studies are needed to elucidate the potential effect(s) of IG-FBP-5 binding on these FN actions and to determine whether the regulation of FN actions by IGFBP-5 contributes to any of the well documented ligand-independent actions of IGFBP-5 on cell growth, migration, differentiation, and survival.