Structure-Function Analysis of the Human Insulin-like Growth Factor Binding Protein-4*

To identify the molecular mechanism by which insulin-like growth factor binding protein-4 (IGFBP-4) exerts its inhibitory effects on insulin-like growth factor (IGF) actions, we localized and determined the role of the IGF binding domain in modulating IGF actions in human osteoblasts. Deletion analysis using IGFBP-4 expressed in bacteria revealed that the N-terminal sequence Leu72–Ser91 was essential for IGF binding. The C-terminal fragments (His121–Glu237 or Arg142–Glu237) did not bind to IGF but loss of these regions decreased IGF binding activity. Detailed deletion analysis identified the residues Cys205–Val214as the motif to facilitate IGF binding. Mitogenic studies revealed that an IGFBP-4 mutant (His74 replaced by Pro74) and an N-terminal peptide (N terminus to Thr71) with little IGF binding activity failed to inhibit IGF-II-induced human osteoblast proliferation. An N-terminal peptide (N terminus to Asn182) with reduced IGF binding activity inhibited IGF action but with lower potency. In contrast, an IGFBP-4 mutant (His74 replaced with Ala74) exhibited similar IGF binding activity and potency in inhibiting the activity of IGF-II compared with the wild type. Therefore, the N-terminal sequence (Leu72–Ser91) and the C-terminal sequence (Cys205–Val214) are necessary to form the high affinity IGF binding domain, which is the major structural determinant of the IGFBP-4 function.

Insulin-like growth factors (IGFs) 1 play a critical role in promoting the differentiation and proliferation of a variety of cell types including human osteoblasts (hOBs) (1)(2)(3)(4). The functions of IGFs depend not only on the amount of IGF produced but also on the level of IGF binding proteins (IGFBP) which modulate their actions (5)(6)(7)(8)(9) as well as the specific IGFBP proteases that regulate the availability of the IGFBPs (10 -14). Although hOBs produce multiple IGFBPs, IGFBP-4 is particularly important with regard to local reduction of IGF function based on its abundance (10,15) and biological potency (1,6). In vitro studies demonstrate that IGFBP-4 inhibits IGF-stimulated osteoblast cell proliferation (1,6). The importance of IGFBP-4 as a negative regulator of hOB cell proliferation is also attested to by studies on the regulation of IGFBP-4 production. In cultured osteoblasts, agents that inhibit osteoblast cell proliferation (cAMP, 1,25-dihydroxyvitamin D 3 , and parathyroid hormone) increase IGFBP-4 levels (7, 16 -18), while agents that increase cell proliferation (progesterone, IGFs, transforming growth factor-␤, and bone morphogenetic protein-7) decrease IGFBP-4 concentrations in the conditioned medium (10,19,20). In vivo studies on serum regulation of IGFBP-4 have shown that the serum IGFBP-4 levels are elevated with age and in type II osteoporosis patients (21)(22)(23). In addition to the role of IGFBP-4 in modulating IGF actions in bone, the inhibitory effect of IGFBP-4 on the mitogenic effect of IGFs has also been demonstrated in other cell types, such as human fibroblasts (24), neuronal cells (25), and colon carcinoma cells (26). These findings suggest that IGFBP-4 is a key component of the IGF system in a variety of tissues including bone.
The mechanisms by which IGFBPs stimulate or inhibit IGF actions have not been clearly defined and may vary among the IGFBPs. In this regard, recent evidence suggests that some of the IGFBPs may exert IGF-independent actions (6,(27)(28)(29), in addition to modulating IGF actions (6,8,30). Studies in our laboratory demonstrate that IGFBP-4 primarily acts to inhibit cell proliferation by an IGF-dependent mechanism (6). The findings which support this concept include: 1) IGFBP-4 has been shown to compete with IGF receptors for IGF binding in both cells in monolayer culture and in purified type I IGF receptor preparations (6), 2) IGFBP-4 has no effect on cell proliferation induced by IGF-I or -II analogs that exhibit reduced affinity for IGFBP-4 (6), and 3) IGFBP-4 proteolytic fragments with reduced IGF binding activity are unable to inhibit IGF-induced cell proliferation (10,30). Although these studies provide indirect evidence that binding of IGFBP-4 to an IGF is essential for inhibiting IGF action, this has not been verified by using IGFBP-4 analogs that do not bind the IGFs.
The amino acids that constitute the IGF binding domain have not been identified for any of the six known high affinity IGFBPs. A few reports have indicated that both the N-terminal and C-terminal regions are critical for IGF binding (31)(32)(33)(34)(35)(36). Interestingly, the sequences and residues important for IGF binding appear to differ among the IGFBPs studied to date (31)(32)(33)(34)(35)(36). Recombinant IGFBP-1 with a 15-residue deletion in the C-terminal region did not exhibit IGF binding activity (31). Chemical modification experiments of the bovine IGFBP-2 suggest that Tyr 60 in the N-terminal region is important for IGF binding (32). Studies of the IGF binding site in IGFBP-3 indicate that both N-terminal and C-terminal fragments of IG-FBP-3 bind to IGF (33). Analysis of the IGF binding activity of IGFBP-4 proteolytic fragments revealed that a N-terminal fragment binds to IGFs with a 15-fold reduced affinity compared with the intact form (37). Taken together, these findings suggest that the structure, location, and perhaps, the number of IGF binding sites in the IGFBPs may differ.
The purpose of this study was to characterize the hIGFBP-4 IGF binding domain and evaluate if the IGF binding domain is essential for the inhibitory effect of IGFBP-4 on hOB proliferation. To evaluate the critical regions in IGFBP-4 which are essential for IGF binding, we generated various IGFBP-4 analogs by protein engineering and used these analogs for evaluation of IGF binding and biological activity.

EXPERIMENTAL PROCEDURES
Materials-Human osteosarcoma MG63 cells were from American Type Culture Collection (Rockville, MD) (CRL 1427). Dulbecco's modified Eagle's medium was from Life Technologies, Inc. Calf serum was from Hyclone (Logan, UT). Bovine serum was purchased from Fluka. The CyQUANT cell proliferation kit was from Molecular Probes (Eugene, OR). Recombinant human IGF-I and IGF-II were from Bachem, Inc. (Torrace, CA). 125 I was from NEN Life Science Products. Nickelagarose resin and the pQE32 plasmid were from Qiagen (Chatsworth, CA). Glutathione-Sepharose 4B and pGEX 5x plasmid were from Amersham Pharmacia Biotech. The QuikChange site-directed mutagenesis kit was from Stratagene (La Jolla, CA). Erase-A-Base kit and Escherichia coli-competent cells were from Promega (Madison, WI). All other chemicals used were at least reagent grade and were from Sigma.
Preparation of hIGFBP-4 GST Fusion Expression Constructs-To prepare a wild type hIGFBP-4 expression construct, the hIGFBP-4 cDNA (7) encoding residues Gly Ϫ5 -Glu 237 was excised from a pGEM 3zf(Ϫ) plasmid with SerfI and XhoI. The first residue, Asp, in mature IGFBP-4 was designated residue 1 and the first residue, Met, in the signal peptide was designated residue Ϫ21. This cDNA fragment was cloned into the SmaI and XhoI sites of the pGEX 5x-2 expression vector. A 50-kDa GST-IGFBP-4 fusion protein, designated GST-BP-4 (Ϫ5/237), was synthesized by E. coli strain HB101 transformed with this vector as described previously (21).
PCR was conducted using the GST-BP-4(Ϫ5/237) plasmid as the template and DNA polymerase, Pfu, by standard methods under the following annealing and amplification conditions: one cycle of 98°C for 5 min (hot start), 32 cycles of 98°C for 0.5 min (denaturation), 65°C for 0.5 min (annealing), and 74°C for 2 min (extension). The PCR products were digested with BamHI and XhoI and cloned into the pGEX 5x-2 vector cut with the same enzymes. The recombinant proteins were expressed in E. coli strain, HB101 (21).
Preparation of His 6 -tagged hIGFBP-4 Expression Constructs-Since cleavage of the GST sequence from the IGFBP-4 fusion proteins by Factor X was not always successful, GST-IGFBP-4 fusion peptides were used in this study. Subsequently, we also prepared the His 6 -tagged IGFBP-4 expression constructs. Use of the pQE32 vector led to expression of IGFBP-4 peptides with 6 histidine residues attached to the N terminus. To prepare a wild type IGFBP-4 expression construct in the pQE32 vector, the GST-BP-4(Ϫ5/237) plasmid was first cut with XhoI and treated with the Klenow fragment of DNA polymerase I to prepare a blunt-ended product. The cDNA insert was released from this blunt ended DNA by further digestion with BamHI. The resulting IGFBP-4 cDNA fragment was cloned into BamHI and the filled PstI (with T4 DNA polymerase and dNTPs) sites of the pQE32 vector. The ligation products were transformed into E. coli XL-1 blue cells. The desired plasmid was cloned and the recombinant His 6 -BP-4(Ϫ5/237) was expressed in E. coli XL-1 blue cells.
To prepare the IGFBP-4 protein without the sequence His 121 -Pro 141 (His 6 -BP-4 (⌬121/141)), the N-terminal fragment of the IGFBP-4 cDNA was obtained by digesting the plasmid GST-BP-4(Ϫ5/237) with BamHI  (16). Data shown here are representative of two to three independent bacterial cell preparations.

FIG. 2. Localization of the N-terminal IGF binding sequence.
Each lane was loaded with approximately 3 g of GST-N-terminal IGFBP-4 peptides purified by glutathione-agarose affinity chromatography (21). Panel A, Coomassie Blue staining of the purified peptides. Panel B, 125 I-IGF-II ligand blot analysis of the purified IGFBP-4 peptides (16). Similar results were obtained using proteins purified from two to three different bacterial extracts.
and XmnI and the C-terminal fragment cDNA was obtained by digesting GST-BP-4(Ϫ5/237) plasmid with XmnI and XhoI. The N-terminal and C-terminal IGFBP-4 cDNAs were ligated into the pGEX 5x-2 vector that had been cut with BamHI and XhoI. The ligation products were transformed into E. coli XL-1 blue cells, and the desired construct was selected by restriction mapping (both XmnI and SmaI sites were disrupted in the desired clones). The engineered hGFBP-4 cDNA insert was released from pGEX 5x-2 and cloned into the pQE32 vector. To prepare the IGFBP-4 construct with deletion of the residues Pro 94 -Gln 119 , His 6 -BP-4(⌬94/119), the His 6 -BP-4(Ϫ5/237) plasmids were digested with PstI to remove an internal cDNA sequence encoding Pro 94 -Gln 119 . The deleted plasmid was self-ligated and cloned in E. coli XL-1 blue cells.
The His 6 -BP-4(Ϫ5/71), His 6 -BP-4(Ϫ5/91), His 6 -BP-4(Ϫ5/120), and His 6 -BP-4(Ϫ5/214) constructs were prepared by transferring the cDNA fragments from the pGEX 5x-2 constructs into the pQE32 vector using appropriate restriction enzymes. Recombinant plasmids were transformed into E. coli strain XL-1 blue cells for protein expression. The full-length IGFBP-4 mutant construct, His 6 -BP-4(H74P), was prepared by site-directed mutagenesis using the Quick-Change sitedirected mutagenesis kit. Briefly, two complementary 34-mer primers with a single nucleotide change (underlined) were designed. The forward primer, 5Ј-CTGCACACACTGATGCCCGGGCAAGGCGTGTG-C-3Ј and the reverse primer, 5Ј-GCACACGCCTTGCCCGGGCATCAG-TGTGTGCAG-3Ј were used for PCR amplification. PCR amplification was conducted using the His 6 -BP-4(Ϫ5/237) plasmid as a template and DNA polymerase, Pfu, under the following annealing and amplification conditions: one cycle of 98°C for 5 min (hot start), 18 cycles of 98°C for 30 s (denaturation), 65°C for 1 min (annealing), and 74°C for 16 min (extension). The newly synthesized plasmids were treated with DpnI to digest the methylated parental DNA template and transformed into E. coli XL-1 blue cells. The plasmid was isolated and checked for the desired mutation by SmaI restriction mapping (the mutant plasmid contains an extra SmaI site) and DNA sequencing. The full-length mutant construct, His 6 -BP-4(H74A), was prepared by using the His 6 -BP-4(H74P) plasmid as the template to conduct PCR under the conditions described above. Primers were identical to those used for preparation of the His 6 -BP-4(H74P) mutant except that the CCC codon for Pro 74 was replaced with the GCC codon for Ala 74 . The desired mutant  (16). Panel C, relative IGF-II binding activity of the IGFBP-4 peptides after adjusting for protein loading (mean Ϯ S.E.). The amount of protein in the band approximately equivalent to the expected size was estimated by analytical SDS-PAGE using the His 6 -BP-4(Ϫ5/237) as a standard. The radioactivity was assessed by gamma counting of the excised band approximately equivalent to the expected size. Similar results were obtained using proteins purified from two to three different bacterial extracts. was selected based on a mutation of Pro 74 to Ala that would disrupt the previously introduced SmaI restriction site. The desired mutation was also confirmed by DNA sequencing.
Overexpression and Purification of Recombinant IGFBP-4 Analogs-Recombinant GST-IGFBP-4 analogs were expressed in E. coli HB101 cells and were purified from bacterial extracts using glutathione-Sepharose 4B affinity chromatography (21). Recombinant His 6 -IGFBP-4 proteins were purified by sequential nickel-agarose and IGF-I affinity chromatography. To undertake purification, bacteria from 2 liters of IPTG-induced (incubated for 6 h at 37°C) cultures were resuspended in lysis buffer (300 mM NaCl, 50 mM phosphate, 8 M urea, pH 7.0) with 35 mM imidazole, and incubated on ice for 2 h with shaking (35 mM imidazole was added to reduce nonspecific binding). The 10,000 ϫ g supernatant was incubated with 5 ml of nickel-agarose resin on ice for 1 h. After washing with the lysis buffer containing 35 mM imidazole, bound proteins were eluted with lysis buffer containing 350 mM imidazole. To remove IGFBP-4 fragments degraded by bacterial proteases, dialyzed samples against phosphate-buffered saline were applied to an IGF-I-agarose column (an IGF-I affinity column was chosen because either the wild type or these IGFBP-4 mutants bound to IGF-I and IGF-II with similar activity as determined by ligand blotting using nickel-agarose affinity-purified IGFBP-4 peptides). After washing with phosphate-buffered saline, bound proteins were eluted with 4 M guanidine in 10 mM Tris, pH 7.4 (1). To remove guanidine, the bound fractions were passed through a nickel-agarose column, eluted with 350 mM imidazole in phosphate-buffered saline, and dialyzed against phosphate-buffered saline. To conduct cell proliferation assay, purified His 6 -BP-4(Ϫ5/237), His 6 -BP-4(H74A), and His 6 -BP-4(H74P), His 6 -BP-4(Ϫ5/ 182), and His 6 -BP-4(Ϫ5/71) were subjected to further purification by HPLC reverse phase chromatography on a C8 column using gradients of acetonitrile in 0.1% trifluoroacetic acid. Samples (2 ml/fraction) were evaporated under negative pressure. Proteins of interest were identified, resuspendend in water, and stored at Ϫ80°C prior to use.
Western 125 I-IGF Ligand Blot Analysis-Whole bacterial cell lysates or purified recombinant IGFBP-4 peptides were mixed with SDS-PAGE loading buffer (100 mM Tris, pH 6.8, 10% SDS, 0.01% phenol blue), boiled for 5 min, and separated by 10 or 12% SDS-PAGE gels. When indicated, proteins were reduced by adding ␤-mercapatoethanol in the samples to a final concentration of 10% prior to addition of SDS-PAGE loading buffer. Proteins were transferred to nitrocellulose filters and subjected to 125 I-IGF-I and/or IGF-II Western ligand blotting as described previously (16). The specific activity of 125 I-IGF-I and 125 I-IGF-II used for ligand blot analysis was 200 -300 Ci/g of protein. For quantitation of IGF binding activities of the IGFBP-4 peptides, the radioactivity in the bands was assessed by gamma counting the excised band of interest. Background radioactivity was determined in the filter without sample and subtracted from the total counts corresponding to IGFBP-4 peptides that bound to IGFs.
Cell Proliferation Assay-Human osteosarcoma MG63 cells were seeded into 96-well plates at 1,000 cells/well in 100 l of Dulbecco's modified Eagle's medium supplemented with 10% calf serum. The medium was replaced after 6 h with Dulbecco's modified Eagle's medium supplemented with 0.1% bovine serum albumin. IGF-II (10 ng/ml) and IGFBP-4 (0 -1,000 ng/ml) were added 20 h later. After an additional 48 h of incubation, nucleic acid (DNA/RNA) content was determined with a CytQUANT cell proliferation kit according to the manufacturer's instructions. Briefly, cells were frozen and thawed for three cycles before adding 200 l of a lysis buffer/dye mixture to each well. After incubation for 20 min at room temperature, the fluorescence was determined using a microplate fluorescence reader with filters for 480-nm excitation and 520-nm emission maxima.
Statistical Analysis-Statistical analysis of the data was performed by t test or ANOVA.

Localization of the N-terminal Sequence Critical for IGF
Binding-To localize the N-terminal sequence that is required for IGF binding, recombinant IGFBP-4 peptides with sequential deletion from the N terminus and C terminus were expressed in E. coli, and their IGF binding activities were analyzed. Fig. 1 shows the results of 125 I-IGF-I and IGF-II Western ligand blot analysis of the GST-BP-4 peptides in crude bacterial lysates. The bands of recombinant proteins with different molecular weights were arrayed according to their molecular weights in order to distinguish them from bacterial proteins (Fig. 1A). Except for the GST-BP-4(Ϫ5/237) and GST-BP-4(5/ 214) peptides, the majority of which were degraded by proteases of bacterial origin, the recombinant proteins were consistently expressed at high levels. The GST-BP-4(Ϫ5/237) and GST-BP4(Ϫ5/214) peptides (Ϸ50 kDa) demonstrated the highest IGF-I and -II binding activity despite the lower amount of intact recombinant protein present in the bacterial lysates (Fig.  1, B and C). A lower IGF-labeled band (Ϸ36 kDa), a proteolytic fragment of the GST-BP-4(Ϫ5/237) and GST-BP-4(Ϫ5/214) fusion proteins, also retained IGF-I and -II binding activity. Sequential deletion from the C terminus to residue Thr 71 abolished IGF-I and IGF-II binding activity. The C-terminal peptides, GST-BP-4(121/237) and GST-BP-4(142/237), bound to neither IGF-I nor IGF-II. However, IGFBP-4 peptides with these C-terminal sequences deleted showed reduced IGF-I and -II binding activity.
To further localize the N-terminal sequence that was critical for IGF binding, additional N-terminal IGFBP-4 peptides were prepared. As shown in Fig. 2, the GST-BP4(Ϫ5/91) peptide retained IGF-II binding activity. Further deletion of two more residues, Glu 90 and Ser 91 (GST-BP4(Ϫ5/89)) resulted in a significant reduction in IGF-II binding activity. Subsequent deletions to residues Glu 84 and Gly 75 did not cause a further significant reduction in IGF-II binding activity. As previously demonstrated using crude cell lysate (Fig. 1), purified GST-BP4(Ϫ5/71) failed to bind to IGF-II. Similar results were obtained using IGF-I tracer for ligand blot analysis (data not shown).
Localization of the C-terminal Sequence That Enhances IGF Binding-In this study, we found evidence that an IGFBP-4 C-terminal sequence facilitates IGF binding. To localize this sequence, peptides with progressive C-terminal deletions were expressed in E. coli, purified with nickel-agarose and IGF-I agarose affinity chromatography, and subjected to IGF-II ligand blot analysis. Evaluation of the purity of the His 6 -BP-4 analogs by SDS-PAGE followed by Coomassie Blue staining revealed that some of these preparations contained in addition to the band of interest, one or more smaller or larger bands (Fig. 3A). For accurate quantitation of the IGF binding activity, the truncated peptides were loaded in greater amounts than the wild type recombinant IGFBP-4 because of their reduced IGF binding activity. Fig. 3B shows the IGF-II ligand blot analysis of the nickel-agarose and IGF-I agarose affinity-purified preparations under nonreducing conditions. In this particular experiment, the migration of the His 6 -BP-4(Ϫ5/167) protein band was not faster than that of the His 6 -BP-4(Ϫ5/182) due to the high residual salt concentration in the His 6 -BP-4(Ϫ5/167) preparation after the sample volume was reduced under negative pressure. Expected migration patterns were observed when desalted samples were applied (data not shown). Fig. 3C shows the relative IGF-II binding activities of the IGFBP-4 peptides under nonreducing conditions after adjusting for protein loading. Consistent with the results obtained using crude bacterial extracts (Fig. 1), the purified peptides His 6 -BP-4(Ϫ5/237) and His 6 -BP-4(Ϫ5/214) exhibited similar IGF-II binding activity. However, shorter peptides with various deletions from the C terminus showed a significant reduction in IGF-II binding activity (8 -14% of the His 6 -BP-4(Ϫ5/237)). Similar results were obtained using an IGF-I tracer for ligand blot analysis (data not shown). Interestingly, the high molecular weight bands in some of these preparations also bound to IGF-I or IGF-II with considerable activity. To evaluate whether these high molecular weight bands represent dimers or multimers formed by interchain disulfide bonds, both the wild type, His 6 -BP-4(Ϫ5/237) and a representative truncated peptide preparation, His 6 -BP-4(Ϫ5/204) were subjected to immunoblot analysis after separation by SDS-PAGE under both reducing and nonreducing conditions (Fig. 4). Upon treatment of the sample with a reducing agent, high molecular weight protein bands in the His 6 -BP-4(Ϫ5/204) preparation disappeared with a concurrent increase in the size of a major His 6 -BP-4 protein band (Ϸ33 kDa). Immunoblot analysis indicated that these high molecular weight protein band under nonreducing conditions and the major 33-kDa band under re- ducing conditions from the His 6 -BP-4(Ϫ5/204) preparation reacted with IGFBP-4 antiserum. Under reducing conditions, the IGF-II binding activity of both the wild type and the His 6 -BP-4(Ϫ5/204) preparations was significantly reduced. Therefore, the high molecular weight protein bands capable of binding to IGF-II under nonreducing conditions may represent products formed by an inter-molecule disulfide bond between intact IG-FBP-4 or/and its proteolytic fragments.

Evaluation of the Inhibitory Effect of Recombinant Wild Type and Mutant IGFBP-4 on IGF-II-induced Cell Proliferation-
Our previous studies showed that IGFBP-4 inhibits IGF action by binding to IGFs and preventing the binding of IGFs to their receptors (6). To evaluate this further, we determined the inhibitory effect of wild type IGFBP-4 and IGFBP-4 mutants that exhibited little or no IGF binding. Based on the finding that the deletion of Gly 75 to Thr 71 results in a loss of IGF binding activity (Fig. 2) and based on the finding that the basic residue His 74 is conserved among the IGFBP-4s from different species (39), we prepared two single amino acid mutants in which His 74 was replaced with Ala or Pro, respectively. His 6 -BP-4(H74A) and the His 6 -BP-4(Ϫ5/237) exhibited similar IGF-II binding, whereas His 6 -BP-4(H74P) retained less than 3% of the IGF-I or -II binding activity (Fig. 6). The wild type His 6 -BP-4(Ϫ5/237) preparation and native IGFBP-4 purified from the hOB conditioned medium exhibited similar IGF-II binding affinity and potency in inhibiting IGF-II actions in MG63 cells (data not shown). As shown in Fig. 7, treatment with 10 ng/ml IGF-II for 48 h increased cell proliferation by approximately 70% over vehicle control (p Ͻ 0.001). At a dose of 50 ng/ml, His 6 -BP-4(Ϫ5/ 237) and His 6 -BP-4(H74A) inhibited the IGF-II-induced cell proliferation by 21 and 26%, respectively (p Ͼ 0.05). At the dose of 150 ng/ml, His 6 -BP-4(Ϫ5/237) and His 6 -BP-4(H74A) inhibited the IGF-II-induced cell proliferation by 94 and 85%, respectively (p Ͼ 0.05). The truncated peptide, His 6 -BP-4(Ϫ5/ 182) at doses of 100 ng/ml and 300 ng/ml inhibited IGF-II stimulated cell proliferation by 16 and 43%, respectively. In contrast, treatment with the mutant, His 6 -BP-4(H74P) at 300 ng/ml or the truncated peptide, His 6 -BP-4(Ϫ5/71) at 1,000 ng/ml did not inhibit IGF-II-induced cell proliferation. In this particular experiment, treatment with His 6 -BP-4(H74P) at 100 ng/ml resulted in a slight but statistically significant increase in cell proliferation. However, this result was not reproducible in additional experiments. DISCUSSION This is a first study that shows the results of the systematic analysis of the IGF binding activity of N-terminal-, mid-molecule-, and C-terminal-deleted recombinant IGFBP-4 analogs. Our findings demonstrate that although the N-terminal but not the C-terminal recombinant IGFBP-4 fragments bind radiolabeled IGF-I and IGF-II, structural elements in both the N-terminal and the C-terminal regions of IGFBP-4 are essential for high affinity IGF binding. The findings of this study also demonstrate that disruption of the IGF binding domain abolishes the inhibitory effect of IGFBP-4 on IGF-induced cell proliferation in serum-free cultures of human osteoblasts.
It is generally accepted that the IGF binding site for the various high affinity IGFBPs is in the N-terminal region, based on the findings that N-terminal IGFBP proteolytic fragments or recombinant N-terminal IGFBP peptides retain various IGF binding activities (6,27,33,37). As a first step toward identifying the critical sequences in IGFBP-4 for IGF-I and IGF-II binding, we generated a series of N-terminal and C-terminal fragments of IGFBP-4 by recombinant DNA technology and evaluated their IGF binding activity by Western ligand blot analysis using radiolabeled IGF-I and IGF-II as tracers. Our findings demonstrate that a number of N-terminal fragments encoded by exon 1 and 2 of IGFBP-4 retained IGF binding activity but none of the C-terminal IGFBP-4 fragments tested exhibited measurable IGF binding activity (Figs. 1-3). These findings are consistent with the hypothesis that the IGF binding site in IGFBP-4 is located in the N-terminal region. In contrast to our findings, Spencer and Chan (33) have recently reported that a C-terminal fragment of recombinant IGFBP-3 encoding residues 151-263, expressed in bacteria, exhibits IGF binding activity. Moreover, an IGFBP-2 C-terminal proteolytic fragment (148 -270) purified from Life BRL-3A conditioned medium also retained partial IGF-I and -II binding activity (34). Therefore, while hIGFBP-4 does not seem to contain an independent IGF binding site in the C-terminal region, other IGFBPs may contain additional IGF binding sites in this region. It is not known whether the structural elements that are involved in IGF binding are similar or different for the six high affinity IGFBPs. In this study, we found that an N-terminal fragment Gly Ϫ5 to Ser 91 but not Gly Ϫ5 to Thr 71 binds to the IGFs, suggesting that the region Leu 72 -Ser 91 is critical for IGFBP-4 binding to the IGFs. To further confirm the importance of this region in IGF binding, we replaced His 74 , a basic residue that is conserved in IGFBP-4 from different species, with Pro 74 in order to disrupt the IGFBP-4 structure in this region. This structural disruption led to a greater than 50-fold reduction in IGF-I and IGF-II binding activity (Fig. 6). Sequence comparison of IGFBPs derived from different species indicates that this region is highly conserved among IGFBP-4s derived from human, bovine, and rat, whereas some of other IGFBPs from different species demonstrate more variability in this region (39,40) (Fig. 8). Since the sequence, Leu 72 -Ser 91 resided in an area that contained both conserved and variable amino acid sequences among the six IGFBPs, it is speculated that the IGF binding domain in IGFBP-4 is different from that of other IGFBPs. Although deletion and initial mutagenesis studies suggest that amino acids, particularly Glu 90 and Ser 91 , in the region Leu 72 -Ser 91 are functionally significant in contributing to the formation of an IGF binding domain and that this region may represent the IGF binding domain, additional site-directed mutagenesis studies involving single and multiple amino acid substitutions will be required in order to determine the contribution of each residue to IGF binding.
Our findings also demonstrate that, although C-terminal fragments of IGFBP-4 do not exhibit measurable IGF binding, the C-terminal region is required for high affinity binding of IGFBP-4 to the IGFs. Based on the findings that the deletion of Val 214 -Cys 205 caused at least a 6-fold reduction in IGF binding activity (Fig. 3), we predict that this region in IGFBP-4 is critical for high affinity IGF binding. Consistent with these data, other studies have shown that the C-terminal domain in IGFBP-3 is essential for IGF binding (27,29,33). Furthermore, substitution of Cys 226 to Tyr 226 in IGFBP-1 led to dimerization and a loss of IGF binding activity (31). Comparison of the amino acid sequence of the six high affinity IGFBPs from different species revealed that a Cys-containing sequence, Cys 205 -Tyr 206 -Cys 207 -Val 208 -Asp 209 , is highly conserved among the high affinity IGFBPs (Fig. 8). Further studies are needed, however, to evaluate whether this conserved motif represents a critical domain in facilitating the binding of IGFs to the N-terminal motif in IGFBP-4.
IGFBP-4 is unique among the six known high affinity IG-FBPs by having two extra cysteine residues (Cys 110 and Cys 117 ) in the variable region encoded by exon 2 (7,39). In order to evaluate whether this region containing the two extra cysteine residues contributes to IGF binding activity, we prepared an IGFBP-4 analog in which the region Pro 94 -Gln 119 was deleted. Based on the finding that the deletion of this variable region had no effect on the IGF binding activity (Fig. 5), we concluded that the two unique Cys residues contained in the mid-region of hIGFBP-4 were not important in the formation of the IGFBP-4 ϩ IGF complex.
The findings of our study are also consistent with the concept that IGFBP-4 contains a single binding site for both IGF-I and IGF-II. Our findings that the deletion of Leu 72 -Ser 91 results in the loss of both IGF-I and IGF-II binding (Figs. 1 and 2) and that substitution of His 74 to Pro 74 essentially abolished both IGF-I and IGF-II binding (Fig. 6) support the above idea. In a Addition of the His 6 tag usually slows down the migration of the fused proteins by several kilodaltons. In this study, the apparent molecular mass of the His 6 -BP-4(Ϫ5/237) was approximately 31 kDa, 5 kDa larger than the calculated molecular mass of 26 kDa. Consequently, 5 kDa was added to the calculated molecular mass of each His 6 -tagged IGFBP-4 peptide. The predicted molecular mass of GST-IGFBP-4 fusion proteins shown in parentheses were the sum of the molecular mass of the GST peptide (25 kDa) and the calculated molecular mass of the IGFBP-4 peptide. The predicted sizes of most recombinant IGFBP-4 peptides roughly match their apparent masses on SDS-PAGE gel although minor inconsistency exists in certain analogs, possibly due to the structural modifications. b IGF binding activity was determined by Western 125 I-IGF-I and IGF-II ligand blot analysis. c Relative IGF-II binding activity was determined by IGF-II ligand blotting with purified His 6 -tagged proteins followed by quantitation of the radioactivity in the bands by scintillation counting. d NQ, not quantitated. addition, previous studies have shown that both unlabeled IGF-I and IGF-II are equally effective in displacing either of the IGF tracers bound to IGFBP-4 (1). In contrast to IGFBP-4, IGFBP-6 binds to IGF-II but not IGF-I with high affinity (40). In addition, IGFBP-2 and IGFBP-5 bind to IGF-II with higher affinity than does IGF-I (23). Thus, it is possible that IGFBP-2, IGFBP-5, and IGFBP-6 may contain separate IGF binding sites, one for IGF-I and the other for IGF-II. Future studies on identification of the IGF-I and IGF-II binding domain for those IGFBPs that bind IGF-I and IGF-II with different affinities are needed to evaluate whether separate binding sites are present for IGF-I and IGF-II in some of the IGFBPs. The findings of this study that the C-terminal region in IGFBP-4 is essential for high affinity IGF binding are consistent with the earlier reports that hIGFBP-4 proteolytic fragments bind IGFs with little or no affinity compared with the intact protein (10,14,24,37). These findings together with the observation that there is significant sequence conservation in the cysteine rich N-terminal and C-terminal regions of IGFBPs suggest that both the N-terminal and the C-terminal domains act in a cooperative manner to bind to the IGFs. However, the molecular mechanism by which the C-terminal motif in IG-FBP-4 enhances IGF binding can only be speculated. In this regard, it is possible that the N-terminal sequence Leu 72 -Ser 91 and the C-terminal sequence Cys 205 -Val 214 may contribute to the overall IGFBP-4 tertiary structure that is important for the high affinity association of this molecule with an IGF. Alternatively, the intact IGF binding domain, located in the N-terminal region, may become more accessible to the ligand in the presence of the C-terminal region. X-ray crystallographic studies and site-directed mutagenesis of IGFBP-4 are required to elucidate the molecular mechanism by which C-terminal region of IGFBP-4 interacts with the N-terminal region to form a high affinity IGF binding site.
In previous studies, we found that IGFBP-4 inhibited the binding of IGF tracer to purified IGF receptors (6). Based on these data and the data that IGFBP-4 had no effect on cell proliferation induced by those IGF analogs that exhibited Ͼ100-fold reduced affinity for binding to IGFBP-4, we proposed that IGFBP-4 inhibits IGF actions by preventing the binding of IGF to its receptor. Consistent with this hypothesis, we have now found that the IGFBP-4 analog in which Pro 74 has been substituted to His 74 resulted in greater than 50-fold reduction in IGF-I and -II binding activity ( Fig. 6 and Table I) and that this analog had a negligible effect on cell proliferation induced by IGF-II (Fig. 7). In contrast, another IGFBP-4 analog in which His 74 is replaced by Ala 74 resulted in no loss of either IGF binding activity (Fig. 6) or the ability to inhibit IGF-IIinduced osteoblast cell proliferation (Fig. 7). The truncated peptide, His 6 -BP-4(Ϫ5/182), which exhibited at least 5-fold reduction in IGF-II binding, inhibited 43% of the IGF-II-stimulated cell proliferation at 300 ng/ml. However, this potency was much lower compared with that of the wild type IGFBP-4, which inhibited 94% of the IGF-II-stimulated cell proliferation at 150 ng/ml. The truncated peptide, His 6 -BP-4(Ϫ5/71), which had no IGF binding activity, did not inhibit IGF-II-stimulated cell proliferation even at 1,000 ng/ml. These data demonstrate that the binding of IGF to IGFBP-4 is essential for IGFBP-4 to modulate its inhibitory effect on IGF-induced cell proliferation and that the IGF binding domain may represent the major structural determinant of IGFBP-4 biological activity.