Interactions of High Affinity Insulin-like Growth Factor-binding Proteins with the Type V Transforming Growth Factor-β Receptor in Mink Lung Epithelial Cells*

High affinity insulin-like growth factor-binding proteins (IGFBP-1 to -6) are a family of structurally homologous proteins that induce cellular responses by insulin-like growth factor (IGF)-dependent and -independent mechanisms. The IGFBP-3 receptor, which mediates the IGF-independent growth inhibitory response, has recently been identified as the type V transforming growth factor-β receptor (TβR-V) (Leal, S. M., Liu, Q. L., Huang, S. S., and Huang, J. S. (1997) J. Biol. Chem. 272, 20572–20576). To characterize the interactions of high affinity IGFBPs with TβR-V, mink lung epithelial cells (Mv1Lu cells) were incubated with 125I-labeled recombinant human IGFBPs (125I-IGFBP-1 to -6) in the presence of the cross-linking agent disuccinimidyl suberate and analyzed by 5% SDS-polyacrylamide gel electrophoresis and autoradiography.125I-IGFBP-3, -4, and -5 but not 125I-IGFBP-1, -2, and -6 bound to TβR-V as demonstrated by the detection of the ∼400-kDa 125I-IGFBP·TβR-V cross-linked complex in the cell lysates and immunoprecipitates. The analyses of125I-labeled ligand binding competition and DNA synthesis inhibition revealed that IGFBP-3 was a more potent ligand for TβR-V than IGFBP-4 or -5. Most of the high affinity 125I-IGFBPs formed dimers at the cell surface. The cell-surface dimer of125I-IGFBP-3 preferentially bound to and was cross-linked to TβR-V in the presence of disuccinimidyl suberate. IGFBP-3 did not stimulate the cellular phosphorylation of Smad2 and Smad3, key transducers of the transforming growth factor-β type I/type II receptor (TβR-I·TβR-II) heterocomplex-mediated signaling. These results suggest that IGFBP-3, -4, and -5 are specific ligands for TβR-V, which mediates the growth inhibitory response through a signaling pathway(s) distinct from that mediated by the TβR-I and TβR-II heterocomplex.

I-IGFBP⅐T␤R-V cross-linked complex in the cell lysates and immunoprecipitates. The analyses of 125 I-labeled ligand binding competition and DNA synthesis inhibition revealed that IGFBP-3 was a more potent ligand for T␤R-V than IGFBP-4 or -5. Most of the high affinity 125 I-IGFBPs formed dimers at the cell surface. The cell-surface dimer of 125 I-IGFBP-3 preferentially bound to and was cross-linked to T␤R-V in the presence of disuccinimidyl suberate. IGFBP-3 did not stimulate the cellular phosphorylation of Smad2 and Smad3, key transducers of the transforming growth factor-␤ type I/type II receptor (T␤R-I⅐T␤R-II) heterocomplex-mediated signaling. These results suggest that IGFBP-3, -4, and -5 are specific ligands for T␤R-V, which mediates the growth inhibitory response through a signaling pathway(s) distinct from that mediated by the T␤R-I and T␤R-II heterocomplex.
High affinity insulin-like growth factor-binding proteins 1-6 (IGFBP-1 to -6) 1 are a family of structurally homologous ϳ24 -43-kDa proteins composed of three defined domains including a nonconserved central domain flanked by conserved cysteine-rich N-and C-terminal domains (1)(2)(3). Recently, several low affinity IGFBPs with sequence homology to the N-terminal domains of the high affinity IGFBPs have been identified and referred to as IGFBP-7 to -10 (4).
The T␤R-V is a 400-kDa non-proteoglycan membrane glycoprotein (20). It is a Ser-specific protein kinase and co-expresses with type I, type II, and type III TGF-␤ receptors (T␤R-I, T␤R-II, and T␤R-III) in most cell types (21)(22)(23). The T␤R-V is a low affinity TGF-␤ receptor with K d of ϳ0.4 nM for TGF-␤ 1 and TGF-␤ 2 and ϳ5 nM for TGF-␤ 3 (23,24). Nevertheless, several lines of evidence suggest that T␤R-V is important in mediating TGF-␤-induced growth inhibitory responses. These include the following: 1) cells lacking T␤R-V but expressing T␤R-I, T␤R-II, and T␤R-III do not exhibit the growth inhibitory response to stimulation by exogenous TGF-␤, although exogenous TGF-␤ is able to induce transcriptional activation of plasminogen activator inhibitor 1 and fibronectin in these cells (24); 2) T␤R-V mediates the growth inhibitory response in the absence of T␤R-I or T␤R-II, but both T␤R-I and T␤R-II are required for maximal growth inhibition (24); and 3) the cells lacking T␤R-V have been found to be carcinoma cells, whereas all normal cell types studied express T␤R-V (19,21,24). This implies that the loss of T␤R-V, which mediates negative growth regulation, may contribute to malignancy of certain carcinoma cells (19,21,24).
To define the function of T␤R-V, we developed specific peptide antagonists that showed higher affinity to T␤R-V than to other TGF-␤ receptor types (25). The structural and functional analyses of these peptide antagonists revealed that a W/RXXD motif is essential for the antagonist activity. Multiple conjugation of the peptide antagonists to carrier proteins conferred TGF-␤ agonist activity in growth inhibition but not in transcriptional activation (25). These results prompted us to identify structurally unrelated TGF-␤ agonists that possess the W/RXXD motif. IGFBP-3 was the first TGF-␤ agonist identified (19). IGFBP-3 possesses a putative TGF-␤ active site motif (WCVD) near its C terminus (1)(2)(3).
Because IGFBP-3 is structurally homologous to other high affinity IGFBPs and because four of six high affinity IGFBPs (IGFBP-3 to -6) possess the putative TGF-␤ active site motif (WCVD) near their C termini (1-3), we hypothesized that at least some of these IGFBPs might bind to T␤R-V, which may mediate the IGF-independent activities of these IGFBPs. To test this hypothesis, we characterized the interactions of IGFBP-1 to -6 with T␤R-V in mink lung epithelial cells (Mv1Lu cells). In this communication, we show that IGFBP-3, -4, and -5 but not IGFBP-1, -2, or -6 bind to T␤R-V as demonstrated by 125 I-labeled ligand affinity labeling of T␤R-V in Mv1Lu cells. IGFBP-4 and -5 bind to T␤R-V with lower affinities than that of IGFBP-3. The cell surface-associated dimeric form of IGFBP-3 exhibits a preference for binding to T␤R-V. We also demonstrate that IGFBP-3-induced growth inhibition mediated by T␤R-V does not involve the stimulated phosphorylation of Smad2 and Smad3.
Biotinylation of IGFBP-3 and Detection of the 125 I-IGFBP-3-biotinylated IGFBP-3 Complex-The biotinylation of IGFBP-3 was carried out in a reaction mixture (50 l) containing IGFBP-3 and sulfo-NHS-biotin (1:25, mol/mol) in 50 mM NaHCO 3 , pH 8.5. After 30 min at room temperature, the reaction was terminated by the addition of glycine (10 mM). For dimer formation studies, Mv1Lu cells were incubated with a premix of 8 nM 125 I-IGFBP-3 and 2 nM biotinylated IGFBP-3 at 0°C for 4 h in the absence or presence of a 100-fold excess of unlabeled IGFBP-3. Cells were rinsed twice with 1 ml of binding medium, collected by scraping, and pelleted by centrifugation at 10,000 rpm for 5 min. The cell pellets were lysed in solubilization buffer (10 mM Tris-HCl, 1 mM EDTA, pH 7.4) containing 1% Triton X-100 and mixed for 1 h at 4°C followed by centrifugation at 12,000 rpm for 10 min. The supernatant was transferred to an Eppendorf tube containing 0.036 mg of avidin-agarose (10 l of suspension) and mixed at 4°C for 1 h in the absence or presence of D-biotin (20 g or ϳ1.6 mM) (to estimate nonspecific binding). The avidin-agarose beads bound by biotinylated IGFBP-3 and 125 I-IGFBP-3-biotinylated IGFBP-3 complexes were washed once with 1 ml of solubilization buffer containing Triton X-100 (0.2%) and 0.5 M NaCl followed by two more washes with salt-free solubilization buffer. Concentrated SDS sample buffer (2ϫ) containing ␤-mercaptoethanol was added to the agarose beads. The bead suspension was boiled for 5 min and vortexed vigorously to release biotinylated IGFBP-3 and 125 I-IGFBP-3 from the beads. The agarose beads were pelleted, and the supernatant was analyzed by 12% SDS-PAGE and autoradiography.
[ 32 P]Orthophosphate Metabolic Labeling and Immunoprecipitation by Anti-Smad2 and Anti-Smad3 IgGs-Mv1Lu cells were plated on 100-mm Petri dishes in DMEM containing 10% FCS at near confluency. After 16 -18 h at 37°C, cells (14 ϫ 10 6 total) were rinsed twice with phosphate-free DMEM and incubated with 5 ml of phosphate-free DMEM containing 0.2% dialyzed FCS for 1 h at 37°C. Cells were metabolically labeled with 1.0 mCi/ml [ 32 P]orthophosphate in phosphate-free DMEM containing 0.2% dialyzed FCS for 2 h at 37°C. The 32 P metabolically labeled cells were treated with TGF-␤ 1 (10 ng/ml or 0.4 nM) or IGFBP-3 (1 g/ml or ϳ33 nM) for an additional 3 h at 37°C, rinsed twice with 5 ml of cold phosphate-buffered saline, and lysed with lysis buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.5% Triton X-100, 50 mM NaF, 10 mM sodium pyrophosphate, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, and 20 g/ml aprotinin) for 10 min at 0°C followed by repeated aspiration through a 21-gauge needle. The cell lysates were centrifuged at 10,000 rpm for 15 min, and the supernatant was precleared with 1 g of goat IgG and protein G-Sepharose at 4°C. The precleared cell lysates were incubated with anti-Smad2 and anti-Smad3 IgGs (2 g) for 2.5 h at 4°C and incubated with protein G-Sepharose for an additional 1 h. The protein G-Sepharose beads were rinsed 4 times with 1 ml of lysis buffer, suspended in 40 l of SDS sample buffer containing ␤-mercaptoethanol, and boiled for 5 min. The immunoprecipitates were analyzed by 7.5% SDS-PAGE and autoradiography.
The K d of IGFBP-3 binding to T␤R-V was previously estimated to be ϳ6 nM (19). To determine the relative affinities of IGFBP-4 and -5 to T␤R-V in Mv1Lu cells, we performed competition experiments using 125 I-IGFBP-3 (1 nM) as the ligand and unlabeled IGFBP-3, -4, and -5 as competitors. As shown in Fig. 3A, increasing concentrations of unlabeled IGFBP-3 quantitatively inhibited 125 I-IGFBP-3 binding to T␤R-V as determined by 125 I-IGFBP-3 affinity labeling of T␤R-V. The quantitative analysis of this inhibition revealed that unlabeled IGFBP-3 blocked the 125 I-IGFBP-3 binding with an IC 50 of ϳ6 nM, which is identical with the estimated K d of IGFBP-3 binding to T␤R-V (19) (Fig. 3B). Unlabeled IGFBP-4 and -5 weakly inhibited 125 I-IGFBP-3 binding to T␤R-V with an IC 50 of Ն100 nM (Fig. 3B). We also determined the effects of various concentrations of unlabeled IGFBP-1, -2, and -6 on 125 I-IGFBP-3 binding to T␤R-V. Unlabeled IGFBP-1, -2, and -6 did not show any significant effect on the binding of 125 I-IGFBP-3 to T␤R-V at concentrations up to 100 nM (data not shown). These results suggest that IGFBP-3 binds to T␤R-V with higher affinity than IGFBP-4 and -5 and that IGFBP-1, -2, and -6 are not ligands for T␤R-V. IGFBP-3 was previously shown to inhibit growth of Mv1Lu cells as measured by DNA synthesis (19). This inhibition appeared to be mediated by T␤R-V because the IGFBP-3-induced growth inhibition was blocked in the presence of ␤ 1 -(41-65), a specific TGF-␤ peptide antagonist that blocked IGFBP-3 binding to T␤R-V (19). Because IGFBP-3, -4, and -5 bind to T␤R-V with different affinities, we determined the relative potencies of IGFBP-3, -4, and -5 for DNA synthesis inhibition in Mv1Lu cells. As shown in Fig. 4, at 1 g/ml (ϳ33 nM) IGFBP-3 inhibited ϳ50% of DNA synthesis of Mv1Lu cells, whereas IGFBP-4 and -5 produced ϳ15-20% inhibition of DNA synthesis at the same concentration. The potent DNA synthesis inhibitory activity of IGFBP-3 is consistent with its high affinity to T␤R-V.
We also determined the effects of IGFBP-1, -2, and -6 on DNA synthesis of Mv1Lu cells. IGFBP-1, -2, and -6 exhibited ϳ5-10% inhibition of DNA synthesis of Mv1Lu cells at 1 g/ml (data not shown). This inhibition may be because of scavenging endogenous IGFs from the IGF-1 receptor. IGF-1 is a weak growth factor or mitogen for Mv1Lu cells under these experimental conditions. Fig. 1, most high affinity 125 I-IGFBPs formed covalently linked dimers at the cell surface. This experiment was performed in the presence of the cross-linking agent DSS. Because 125 I-labeled proteins prepared by the chloramine-T procedure are known to acquire the properties of covalent linking during 125 I-labeling (27)(28)(29), the covalently linked dimers of 125 I-IGFBPs may be spontaneously produced in a DSS-independent manner during the incubation (3 h at 0°C) of cells with 125 I-IGFBPs that were also prepared using chloramine T. To test this possibility, we investigated the formation of the covalently linked dimer of 125 I-IGFBP-3 in aqueous solution and at the cell surface of Mv1Lu cells in the absence of added cross-linking agents. As shown in Fig. 5, less than 2% of 125 I-IGFBP-3 (5 nM) spontaneously formed co-valently linked dimers in binding medium (lane 1). However, approximately 20% of the 125 I-IGFBP-3 associated with the cell surface was found to be covalently linked dimers (lane 2). These results indicate that the formation of the covalently linked 125 I-IGFBP-3 dimer does not require the presence of crosslinking agents. These results also suggest that the 125 I-IGFBP-3 dimer formation may be enhanced at the cell surface. Alternatively, the 125 I-IGFBP-3 dimer may associate with the cell surface of Mv1Lu cells with an affinity higher than that of the 125 I-IGFBP-3 monomer.

IGFBP-3 Forms Dimers at the Cell Surface That Preferentially Bind to T␤R-V-As demonstrated in
As described above, we demonstrated the covalently linked dimer formation of 125 I-IGFBP-3 or other 125 I-IGFBPs by taking advantage of the properties of covalent linking of 125 I-IGFBPs prepared by the chloramine-T procedure. To prove that the dimer formation is an inherent property of IGFBP-3, we determined the formation of the IGFBP-3 dimer using an approach in which IGFBP-3 was tagged with 125 I or biotin. The formation of the IGFBP-3 dimer was detected by identifying the 125 I-IGFBP-3-biotinylated IGFBP-3 complex in the lysates of Mv1Lu cells that were incubated with a premix of 125 I-IGFBP-3 and biotinylated IGFBP-3 (4:1, mol/mol). After 2.5 h at 0°C, the cell lysates were incubated with avidin-agarose. After centrifugation, the avidin-agarose pellets were analyzed by 7.5% SDS-PAGE under reducing conditions and autoradiography. As shown in Fig. 6 (lane 1), 125 I-IGFBP-3 was detected  in the avidin-agarose pellets of lysates of cells incubated with a premix of 125 I-IGFBP-3 and biotinylated IGFBP-3. Very little 125 I-IGFBP-3 was detected in the avidin-agarose pellets of lysates of cells incubated with a premix of 125 I-IGFBP-3 and biotinylated IGFBP-3 in the presence of a 100-fold excess of unlabeled IGFBP-3 or ϳ1.6 mM biotin (Fig. 6, lanes 3 and 2). These results further verify the ability of IGFBP-3 to form dimers at the cell surface.
TGF-␤ is known to stimulate cellular responses by inducing hetero-oligomerization of TGF-␤ receptors through its covalent dimeric structure (30,31). We hypothesize that IGFBP-3 inhibits cellular growth by a similar mechanism in which the dimeric form of IGFBP-3 is required for activation of T␤R-V. To test this hypothesis, we determined the binding of the dimeric form of IGFBP-3 to T␤R-V in Mv1Lu cells in the presence and absence of DSS. As shown in Fig. 7, the covalently linked dimer of 125 I-IGFBP-3 was detected in the medium (lane 1) and lysates (lane 3) of cells incubated with 125 I-IGFBP-3 without added cross-linking agents. It is of importance to note that the exposure times for the autoradiograms of the medium and cell lysates were 16 and 2 h, respectively, to have comparable intensities of covalently linked 125 I-IGFBP-3 dimers. In the presence of the cross-linking agent DSS, most of the covalently linked dimer associated with the cell surface was cross-linked to T␤R-V (lane 4 versus 3). These results suggest that the cell surface-associated dimeric form of IGFBP-3 preferentially binds to T␤R-V. These results are also consistent with our previous observation that both T␤R-V and 125 I-IGFBP-3 dimers were immunoprecipitated by specific antiserum to T␤R-V after the 125 I-IGFBP-3-affinity labeling of T␤R-V in Mv1Lu cells (19).
IGFBP-3 Does Not Stimulate the Cellular Phosphorylation of Smad2 and Smad3-In Mv1Lu cells, Smad2 and Smad3 have been identified as key signal transducers within the signal trans-duction cascade initiated by the T␤R-I⅐T␤R-II heterocomplex following stimulation by TGF-␤ (32). The phosphorylation of Smad2 and Smad3 by T␤R-I is essential for their complex formation with Smad4 and subsequent translocation to the nucleus where they regulate transcriptional activities required for cell cycle arrest and other cellular responses (32,33). Because T␤R-V forms complexes with T␤R-I (24), the phosphorylation of Smad2 and Smad3 may also be involved in the signaling mediated by the T␤R-I⅐T␤R-V heterocomplex (24). To test this possibility, we investigated the effect of IGFBP-3 on the phosphorylation of Smad2 and Smad3 in Mv1Lu cells. As shown in Fig. 8, the phosphorylation of Smad2 and Smad3 was not affected by IGFBP-3 treatment (lane 4 versus 2), but TGF-␤ treatment enhanced the phosphorylation of Smad2 and Smad3 by ϳ7and ϳ2-fold, respectively (lane 3 versus 2). This result suggests that the IGFBP-3-induced growth inhibition, which is mediated by T␤R-V, may involve a signaling pathway that is distinct from that mediated by the T␤R-I⅐T␤R-II heterocomplex. DISCUSSION High affinity IGFBPs are important modulators of IGF actions (1)(2)(3). Accumulated evidence suggests that IGFBPs are also able to induce cellular responses in an IGF-independent manner (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15). The IGF-independent actions for IGFBPs are believed to be mediated by specific cell-surface receptors or membrane-binding proteins (1, 16 -18). Several membranebinding proteins for IGFBPs were identified, but none of these proteins were well characterized (6, 16 -18). We have recently identified the IGFBP-3 receptor as T␤R-V, which mediates the IGF-independent growth inhibitory response induced by IGFBP-3 (19). In this communication, we show that IGFBP-4 and -5 are also specific ligands for T␤R-V, although their affinities for T␤R-V are weaker than that of IGFBP-3. The T␤R-V is likely the same receptor for IGFBP-5, which has been recently identified in mouse osteoblasts (34). The T␤R-V and putative  1 and 2) is because of the influence from a large quantity of bovine serum albumin in the binding medium, which migrates closely with the 125 I-IGFBP-3 dimer on the SDS-polyacrylamide gel.
IGFBP-5 receptor in osteoblasts share similar properties including the following: 1) they have almost identical molecular weights (ϳ400,000) (19 -25, 34), 2) both show ligand (TGF-␤/ IGFBP-3 and IGFBP-5)-stimulated serine-specific autophosphorylation and kinase activity toward caseins 2 (23,24,34), and 3) the T␤R-V is expressed in most cell types including osteoblasts (21). 3 Among high affinity IGFBPs, four (IGFBP-3 to -6), which possess the putative TGF-␤ active site motif WCVD near their C termini, were initially predicted to bind to T␤R-V. However, although IGFBP-3, -4, and -5 were found to interact with T␤R-V in Mv1Lu cells, IGFBP-6 did not. The inability of IGFBP-6 to interact with T␤R-V may be because of its unique structure. IGFBP-6 contains 10 of 12 N-terminal cysteine residues conserved in other high affinity IGFBPs and possesses additional O-linked carbohydrate moieties in the central domain and possibly near the C-terminal end (1-3, 35, 36). These distinct structural features may yield a conformation that does not allow the WCVD motif in IGFBP-6 to interact with T␤R-V. It is also possible that the WCVD motif is not the only determinant required for the interactions of IGFBPs with T␤R-V. The WCVD motif is contained within the thyroglobulin type-1 repeat of IGFBP-3 (37). Thyroglobulin, which contains multiple WCVD motifs per monomer, has recently been shown to exhibit an authentic TGF-␤ antagonist/agonist activity after activation by acidic pH/denaturing agent treatments and chemical modifications (38). This implies that certain structural configura-tions of the WCVD motif are required for optimal interaction with T␤R-V.
Several polypeptide growth factors are known to stimulate the cytoplasmic kinase activities of their respective receptors by inducing receptor dimerization through their dimeric structures (39 -41). The covalent dimeric structure is also known to be required for TGF-␤ activities (42). Most 125 I-labeled IGFBPs form covalently linked dimers at the cell surface. Approximately 20% of cell surface-associated 125 I-IGFBP-3 is estimated to be in the form of covalently linked dimers, whereas less than 2% exists as the covalently linked dimer in binding medium. This suggests that the cell surface association enhances the dimer formation of IGFBP-3. Assuming that the efficiency of the spontaneous covalent linking of 125 I-IGFBP-3 is ϳ20%, it is estimated that almost 100% of the cell surfaceassociated 125 I-IGFBP-3 are dimers. The cell-surface dimeric form of IGFBP-3 appears to be the active form of IGFBP-3 for binding to T␤R-V.
The major cell-surface binding sites for IGFBP-3 dimers appear to be membrane proteins other than T␤R-V because cells lacking T␤R-V (human colorectal carcinoma cells) express these binding sites (19). Interestingly, the binding of 125 I-IGFBP-3 and 125 I-IGFBP-5 dimers to their major cell-surface binding sites is blocked by ␤ 1 -(41-65), a specific TGF-␤ peptide antagonist, whereas the binding of 125 I-IGFBP-1 and -4 dimers to their major binding sites is resistant to the blocking by the TGF-␤ peptide antagonist (Fig. 1). This suggests that the major cell-surface binding sites for IGFBP-3 and IGFBP-5 dimers are distinct from those for other IGFBPs dimers. This suggestion has been supported by the observation that heparin inhibits the binding of 125 I-IGFBP-3 and -5 dimers but not 125 I-IGFBP-1 and -4 dimers to cell-surface binding sites (18,19). 4 The functions of these major cell-surface binding sites are unknown. However, one function may involve presentation of IGFBPs to their respective cell-surface receptors. In the case of IGFBP-3, these binding sites may present IGFBP-3 to T␤R-V as demonstrated in Fig. 7. This would explain the observation that heparin inhibits the binding of 125 I-IGFBP-3 to both the major cell-surface binding sites and to the T␤R-V (18,19). 4 The signaling mediated by T␤R-V has been difficult to define because of the co-expression of T␤R-I, T␤R-II, T␤R-III, and T␤R-V in the same cells. The identification of IGFBP-3 as well as IGFBP-4 and -5 as specific ligands for T␤R-V has enabled us to investigate the signaling mediated by T␤R-V in cells containing other TGF-␤ receptors. In this communication, we show that IGFBP-3 does not stimulate the cellular phosphorylation of Smad2 and Smad3, both of which play key roles in the signaling mediated by the T␤R-I and T␤R-II heterocomplex (31,32). This result is consistent with the observation that IGFBP-3 induces growth inhibition but not transcriptional activation of plasminogen activator inhibitor-1 in Mv1Lu cells (19). The TGF-␤-induced expression of plasminogen activator inhibitor-1 is mainly mediated by the T␤R-I⅐T␤R-II complex (24). Furthermore, IGFBP-3 has been shown to inhibit the growth of mutant mink lung epithelial cells (DR26 and R-1B cells), which express T␤R-V but lack the expression of the functional T␤R-II or T␤R-I (24).