Binding Sites and Binding Properties of Binary and Ternary Complexes of Insulin-like Growth Factor-II (IGF-II), IGF-binding Protein-3, and Acid-labile Subunit*

We have examined regions of rat IGF-binding protein-3 (IGFBP-3) important for complex formations using two kinds of deletion mutants, three kinds of chimera molecules between rat IGFBP-3 and rat IGFBP-2, and a synthetic peptide (41 residues, Glu52-Ala92) derived from rat IGFBP-3. Solid-phase binding assays using 96-well microtiter plates were designed to quantitate the relative binding affinities. It was found that not only the IGFBP-3 derivatives with the amino-terminal, cysteine-rich domain (N domain) but also the synthetic peptide maintained affinity for IGF-II. Ternary complex formation was observed with full-length IGFBP-3 and chimera IGFBP, the carboxyl-terminal cysteine-rich domain (C domain) of which was derived from IGFBP-3, unlike the mutants lacking the C domain and the chimera IGFBPs, the C domain of which was derived from IGFBP-2. These results were confirmed by affinity cross-linking experiments. Furthermore, the IGFBP-3 derivatives that possessed the C domain of IGFBP-3 bound to the acid-labile subunit, even in the absence of IGFs. Finally, we observed sites in IGF-II important for the ternary complex formation using various IGF-II mutants. These IGF-II mutants, which contained a substitution of Tyr27 for Leu, had extremely reduced activity. These results strongly suggest that: 1) the N domain, containing at least Glu52-Ala92, of rat IGFBP-3 is important for binding to IGF-II; 2) the C domain of IGFBP-3 is essential for binding to the acid-labile subunit both in the presence and absence of IGF-II; and 3) Tyr27 of IGF-II is important for the ternary complex formation.

ported that the mac25 gene encodes a preprotein (named IG-FBP-7) of 277 amino acids, which contain the common IGFBP motif in the amino-terminal domain. In this domain, 11 of the usual 12 cysteines are conserved (8), and IGFBP-7 binds IGFs with specificity (9). Although IGFBP-7 contains a total of 18 cysteines, consistent with the IGFBP-1 to IGFBP-5 molecules, the carboxyl-terminal domain contains only one homologous cysteine position.
Through binding analyses using various IGFs mutants, it was reported that Glu 6 , Phe 26 , Phe 48 , Arg 49 , and Ser 50 of human IGF-II were important for binding to IGFBPs (14 -16).
There is also information about binding regions of IGFBPs to IGFs derived from fragments of IGFBPs that exist in human and rat serum (6). For example, amino-terminal 30-kDa fragments of IGFBP-3 obtained from human (17) and rat (18,19) serum were able to bind IGFs. Furthermore, an amino-terminal 15-kDa fragment derived from human plasma IGFBP-3 was detected by ligand blotting using 125 I-labeled IGF-I (20). These natural fragments corresponding to the amino-terminal domain of IGFBP-3 were isolated without disulfide reduction, suggesting that these domains are not linked by disulfide bonds (6). In the case of the IGFs/IGFBP-3/ALS ternary complex formation, Tyr 24 and the D-domain (residues 63-70) of human IGF-I are thought to be important (21). However, the properties of IGFBP-3 and ALS for ternary complex formation are not understood well.
We previously characterized IGFs binding to their receptors and biological activities of IGFs using various recombinant IGF-II mutants (22,23) and confirmed the three-dimensional structure of recombinant IGF-II (4). In this study, we describe new solid-phase binding assays that can be performed easily using 96-well microtiter plates. Using these assays, we evaluated the binding properties of various mutants, chimeras, and a synthetic peptide derived from rat IGFBP-3 to analyze IGFs/IGFBP-3 and IGFs/IGFBP-3/ALS structure-function relationships.

EXPERIMENTAL PROCEDURES
Materials-A rat pancreas cDNA (gt 11) library of adult Sprague-Dawley females was purchased from CLONTECH Laboratories, Inc. (Palo Alto, CA). The prokaryotic expression vector pTrc99A, pGEX-2T, DEAE-Sepharose CL-6B resin, SP-Sepharose F.F. resin, and NHSactivated HiTrap affinity column were from Pharmacia Biotech, Inc. (Uppsala, Sweden). Protein A affinity resin (PROSEP-A) was from Bioprocessing, Ltd. (Durham, United Kingdom) and DEAE-5PW 75 ϫ 7.5 mm inside diameter was from Tosoh (Tokyo, Japan). A YMC-Pack Protein-RP column (250 ϫ 4.6 mm inside diameter) was from YMC Co., * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Preparations of Rat IGFBP-3 and Its Mutants Expressed in E. coli or Synthesized Chemically-Cloning of IGFBP-3 and IGFBP-2 cDNA was performed by polymerase chain reaction with reference to known cDNA sequences (26,27) using rat pancreas cDNA library (gt 11) as a template. Considering the significant homology among all rat IGFBPs, IGFBP-3 contained the three major domains: an amino-terminal (Gly 1 -Cys 89 ) and carboxyl-terminal domain (Cys 187 -Gln 265 ), rich in cysteine residues; and an linkage-domain (Ala 90 -Pro 186 ) with no significant homology. IGFBP-3 mutants designated NϩL domain (Gly 1 -Pro 186 ) and N domain (Gly 1 -Ser 93 ) were constructed by polymerase chain reaction methods using the rat IGFBP-3 cDNA as a template. A chimera protein whose N, L, and C domain were derived from IGFBP-3, IGFBP-2, and IGFBP-2, respectively, was named as chimera-322. In this way, chimera proteins designated chimera-322, chimera-323, and chimera-332 were constructed by mixing polymerase chain reaction-derived fragments from the N, L, and C domains using appropriate rat IGFBP-3 and rat IGFBP-2 cDNA clones as templates. All IGFBP-3 derivatives are illustrated in Fig. 1.
For the expression of IGFBP-3, NϩL domain, chimera-322, chimera-323, and chimera-332 in E. coli, the phosphatase A (PhoA) signal sequence was chemically synthesized and introduced into the NcoI-HindIII site of pTrc99A, upstream of these clones, which were devoid of their own signal sequences. These expression vectors were transformed into E. coli UT5600. The cells were grown at 37°C in an LB medium supplemented with 50 g/ml ampicillin. After isopropyl-1-thio-␤-D-galactopyranoside induction (final concentration, 0.3 mM), the cells were cultured an additional 1 h and then collected by centrifugation at 6000 rpm for 20 min. The periplasm prepared by osmotic shock (28) from the cells was adjusted to pH 7.2 with phosphate-buffered saline and applied to an IGF-II coupled affinity resin column prepared from an NHSactivated HiTrap affinity column according to the manufacturer's recommendations. Each IGFBP-3 derivative was eluted with 0.5 M acetic acid and lyophilized, dissolved in 0.1% trifluoroacetic acid, and then further purified by reverse-phase high-performance liquid chromatography (YMC-Pack Protein-RP 250 ϫ 10 mm inside dimater) with a linear gradient from 25 to 55% acetonitrile in 0.1% trifluoroacetic acid.
The mutant constructed of only the IGFBP-3 N domain was expressed in E. coli BL-21 as a fusion protein with glutathione S-transferase. The fusion protein, into which a cyanogen bromide-cleavable methionine was introduced, was constructed using expression vector pGEX-2T. The fusion protein was purified from cell lysate on a glutathione Sepharose-4B affinity resin using a batchwise method according to the manufacturer's recommendations (Pharmacia Biotech). The fraction, eluted with 25 mM reduced glutathione, was dialyzed against water and lyophilized. The N domain was released from the glutathione S-transferase fusion protein by treatment with cyanogen bromide in 70% (v/v) formic acid and lyophilized. The obtained N domain was refolded as described previously (29) and further purified using the IGF-II coupled affinity column and reverse-phase high-performance liquid chromatography as described above.
Two of the six disulfide bond pairs included in the N domain of IGFBP-3 were determined by peptide mapping procedures that involved trypsin digestion, reverse-phase high-performance liquid chromatography, and amino-terminal amino acid sequence analysis of isolated peptide fragments (data not shown). By referring to the identified disulfide bonds of Cys 56 -Cys 69 and Cys 63 -Cys 89 , a 41-residue peptide derived from positions Glu 52 -Ala 92 was synthesized by an Applied Biosystems model 431A peptide synthesizer using the selective S-S formation procedures (30).
Purification of Rat Serum ALS-Eighty ml of rat serum were dialyzed against 50 mM Tris-HCl, pH 8.2, and chromatographed on a DEAE-Sepharose CL-6B column and an IGF-II/IGFBP-3 complex affinity column essentially as described previously with some modifications (11,31). For this study, IGF-II was conjugated to an NHS-activated HiTrap affinity column, and recombinant IGFBP-3 was used instead of serum-derived IGFBP-3. Finally, the ALS was purified on a DEAE-5PW column (75 ϫ 7.5 mm inside dimater) using a linear gradient from 0.1 to 0.5 M NaCl in 10 mM sodium phosphate buffer, pH 8.0. Fractions containing ALS were collected and dialyzed against 50 mM sodium phosphate buffer, pH 6.5, by ultrafiltration.
Determinations and Analyses of Purified Proteins-The purity of each purified IGFBP-3, NϩL domain, N domain, chimera-322, chimera-323, chimera-332, the 41-residue peptide, and ALS was confirmed by SDS-polyacrylamide gel electrophoresis in the Laemmli buffer system (32) under reducing conditions followed by Coomassie Brilliant Blue staining. All of the protein concentrations were determined using an AccQ-Tag amino acid composition analysis column according to the manufacturer's instructions (Waters, Milford, MA). Amino acid sequences of all purified proteins were also confirmed by sequencing using an Applied Biosystems 476A gas phase sequencer.
Modifications of IGF-II and ALS for Binding Assay-HRP-IGF-II was prepared using a maleimide-activated HRP kit according to the manufacturer's recommendations (Pierce). Briefly, IGF-II was reacted with N-succinimidyl-S-acetylthioacetate to introduce a free sulfhydryl into the primary amines and was then conjugated with maleimideactivated HRP. ALS was biotinylated on primary amines using a biotinylation kit according to the manufacturer's recommendations (Amersham Co.).
Preparations of Rabbit Anti-IGFBP-3 and Anti-41 Residue Peptide Polyclonal Antibodies-Five injections (0.4 mg each) of IGFBP-3 or the 41-residue peptide were administered s.c. to five male rabbits at 2-week intervals. Each polyclonal antibody (anti-BP-3 pAb and anti-Pep pAb, respectively) was purified from the antisera using a PROSEP-A affinity column according to the manufacturer's recommendations and dialyzed against 50 mM sodium phosphate buffer, pH 6.5.
Characterization of Binding and Complex Formation Properties of IGFBP-3 Mutants, 41-residue Peptide, and IGF-II Mutants Using Solidphase Binding Assays-All solid-phase binding assays described were carried out with 96-well microtiter plates using binding components as illustrated in Fig. 2. For all assays, the plates were coated with proteins in 50 mM sodium phosphate buffer, pH 6.5, and blocked with the same buffer containing 1% (w/v) bovine serum albumin. All washes were done with the same phosphate buffer containing 0.03% (v/v) Tween 20. Binding proteins were diluted in the phosphate buffer containing 0.25% bovine serum albumin and 0.03% (v/v) Tween 20. After coating, all steps were performed at room temperature.
For the HRP-IGF-II/IGFBP-3 competitive binding assay ( Fig. 2A), 50 l of 150 ng/ml IGFBP-3 were immobilized at 4°C overnight. The wells were washed and blocked for 2 h at room temperature. The wells were washed, and 25 l of a 1:6000 dilution of HRP-IGF-II and 25 l of each competitor were added simultaneously to each well. The plates were incubated for 2 h and washed, and 100 l of 2,2Ј-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) substrate (Kirkegaard & Perry Laboratories) were added. After 20 min, the absorbance of the reaction product at 405 nm was read using a V max microplate reader (Molecular Devices, Menlo Park, CA).
To study complex formation with ALS, indirect immobilization procedures using biotinylated ALS were performed. Wells were precoated with 50 l of 1 g/ml streptavidin, and unbound sites were blocked. After washing, 50 l of 200 or 50 ng/ml biotinylated ALS were added and incubated for 2 h to immobilize ALS indirectly. To observe the relative affinities of the IGFBP-3 derivatives to form ternary complexes ( Fig. 2B; the HRP-IGF-II/IGFBP-3/ALS solid-phase binding assay), 50 l of 200 ng/ml biotinylated ALS were used. After indirect immobilization of ALS, 25 l of a 1:1000 dilution of HRP-IGF-II and 25 l of various concentrations of IGFBP-3 derivatives were incubated for 2 h to form the ternary complex. The quantity of the formed complex in the well was measured as described above. To detect the binding of IG-FBP-3 derivatives to ALS in the absence of IGFs (Fig. 2C; the IGFBP-3/ALS solid-phase binding assay), various concentrations of IGFBP-3 derivatives were incubated with the immobilized biotinylated ALS. After washing the wells, the quantity of the formed complex was detected by incubation with 50 l of 3 g/ml anti-BP-3 pAb for 2 h, followed by washing and 50 l of a 1:1000 dilution of anti-rabbit IgG-HRP for 2 h. To observe the ability of the IGF-II mutants to form the ternary complexes ( Fig. 2D; the IGFs/IGFBP-3/ALS solid-phase binding assay), 50 l of 50 ng/ml biotinylated ALS were immobilized indirectly to avoid detection of the IGFBP-3/ALS complex in the absence of IGFs. After indirect immobilization of ALS, 25 l of various concentrations of IGF-II mutants and 25 l of 50 ng/ml IGFBP-3 were incubated for 2 h. The formed ternary complexes were detected using the anti-BP-3 pAb and anti-rabbit IgG-HRP system as described above.
Characterizations of Complex Formation between 125 I-Labeled IGF-II, the 41-Residue Peptide, and the IGFBP-3 Mutants by Affinity Crosslinking-Each IGFBP-3 mutant or the 41-residue peptide was incubated at 4°C overnight in 270 l of 25 mM sodium phosphate buffer, pH 6.5, containing 0.05% Tween 20, 0.1% sodium azide, and 60,000 cpm of 125 I-labeled IGF-II in the absence (for the binary complex) or in the presence (for the ternary complex) of 0.6 g of ALS. Following an overnight incubation, 30 l of 1.5 mM disuccinimidyl suberate were added, and the reaction mixture was further incubated on ice for 15 min. The cross-linking reaction was terminated by adding 60 l of 1 M Tris-HCl buffer, pH 8.5. Proteins were precipitated by trichloroacetic acid at a final concentration of 20% and applied to a 12% SDS-polyacrylamide gel electrophoresis gel under reducing conditions, followed by autoradiography.

Preparation of Recombinant IGFBP-3 Mutants, the 41-residue Peptide, and Rat
Serum ALS-IGFBP-3, two kinds of deletion mutants, three kinds of chimera IGFBPs, the 41-residue peptide, and rat serum ALS were prepared as described under "Experimental Procedures," and their purity was confirmed by reverse-phase high-performance liquid chromatography (data not shown) and by SDS-polyacrylamide gel electrophoresis (Fig. 3). The chimera IGFBPs were expressed using the same expression vector as the IGFBP-3, but all chimera IGFBPs showed doublet bands. The doublet bands were also detected by Western blotting using anti-BP-3 pAb and by far Western ligand blotting using HRP-IGF-II (data not shown). From the results of amino-terminal amino acid sequence analyses, it was found that all chimera IGFBPs contained the expected aminoterminal sequence of Gly-Ala-Gly-Ala-Val-Gly-Ala as the primary sequence and an unexpected minor Gly-Ala-Val-Gly-Ala sequence. The minor sequence is probably due to different processing of the PhoA signal sequence, resulting in the doublet bands. Recombinant rat IGFBP-3 expressed in E. coli did not show such doublet bands nor the minor sequence, however, it is possible that IGFBP-3 may have contained the minor sequence at undetectable levels.

Characterizations of IGF-II/IGFBP-3 Binary Complex Formation Using IGFBP-3 Mutants and the 41-residue Peptide-
Using the HRP-IGF-II/IGFBP-3 solid-phase competitive binding assay, the relative affinities of five IGFBP-3 mutants and the 41-residue peptide to IGF-II were calculated from their displacement curves (Fig. 4) and compared with that of IG-FBP-3, which was set to 100% (Table I). The chimera IGFBPs retained over 40% of the relative affinity, but the isolated NϩL domain or N domain showed much lower affinity. A three-fold difference in affinity was observed between NϩL domain and N domain. In the case of the 41-residue peptide, although it showed only 0.008% affinity relative to IGFBP-3, an anti-Pep pAb inhibited HRP-IGF-II binding to immobilized IGFBP-3. To verify the IGFBP-3 mutants and the 41-residue peptide binding to IGF-II, affinity cross-linking procedure was performed using 125 I-labeled IGF-II. The IGFBP-3, NϩL domain, N do-  main, and the 41-residue peptide all showed bands consistent with cross-linking specifically to 125 I-labeled IGF-II (Fig. 5). The negative control using only ALS did not form a complex band. These results are consistent with the conclusion that the N domain of IGFBP-3, and at least a portion of the 41-residue peptide (position Glu 52 -Ala 92 ), is important for binding to IGF-II.
Characterizations of Binding Activities of IGFBP-3 Derivatives to ALS in the Presence or Absence of IGFs-Using the HRP-IGF-II/IGFBP-3/ALS solid-phase binding assay, activities for the ternary complex formation of IGFBP-3 mutants and the 41-residue peptide could be compared. In this assay, indirect immobilization of ALS was adopted because direct immobilization of ALS or biotinylated ALS did not give a measurable absorbance using 25 ng/ml of IGFBP-3 (data not shown). IG-FBP-3 showed concentration-dependent ternary complex formation between 0.1 and 10 nM, the same range as the competitive binding assay shown in Fig. 4. However, neither the NϩL domain, N domain, nor the 41-residue peptide, which all had low affinity for IGF-II, were observed to form ternary complexes (Fig. 6A). Of the chimeras, only chimera-323 showed ternary complex formation (Fig. 6B). The activities of the chimera IGFBPs for the ternary complex formation were also examined by affinity cross-linking experiments using 125 I-labeled IGF-II and ALS (Fig. 7). The bands derived from the binary complex with 125 I-labeled IGF-II were detected with IGFBP-3 and all chimera IGFBPs, whereas bands derived from the ternary complex were detected only with IGFBP-3 and chimera-323. Because chimera-332, which contains only the C domain from IGFBP-2, did not show activity, this suggests that the L domain is not involved for ternary complex formation.
To detect the IGFBP-3/ALS binary complex, the indirect immobilization procedure was adopted, similar to the HRP-IGF-II/IGFBP-3/ALS solid-phase binding assay. Various concentrations of IGFBP-3 were added to the wells, and the IG-FBP-3/ALS binary complex formed was detected by anti-BP-3 pAb followed by anti-rabbit IgG-HRP. Preliminary experiments confirmed that the anti-BP-3 pAb could bind to all chimera IGFBPs equally by Western blot analysis (data not shown). As shown in Fig. 8, IGFBP-3 and chimera-323 bound to ALS in a concentration-dependent manner. The other chimera IGFBPs (chimera-322 and chimera-332) had essentially no binding to ALS. These results strongly suggested that the C domain of IGFBP-3 plays an important role for the binding of IGFBP-3 to ALS in the presence and absence of IGFs.
Characterizations of Ternary Complex Formation Mediated by IGF-II Mutants-The relative affinities of four IGF-II mutants to IGFBP-3 were determined by the HRP-IGF-II/IG-FBP-3 solid-phase competitive binding assay from their displacement curves shown in Fig. 9A and compared with that of IGF-II, which was set to 100%. The affinities of [Leu 27 ]rIGF-II, [Leu 43 ]rIGF-II, [Arg 54 , Arg 55 ]rIGF-II, and [Leu 27 , Leu 43 ]rIGF-II for IGFBP-3 were 100, 108, 25, and 100%, respectively, relative to IGF-II. The values from the first three mutants are similar to those reported previously, i.e., 70, 100, and 37%, respectively (16). The ability to form ternary complexes with these IGF-II mutants were examined with the IGFs/IGFBP-3/ ALS solid-phase binding assay (Fig. 9B). The [Leu 43 ]rIGF-II and [Arg 54 , Arg 55 ]rIGF-II showed almost the same activities relative to IGF-II. However, the two mutants with substitution  a Relative affinities were determined from the competitive binding inhibition data shown in Fig. 4. of Tyr 27 to Leu showed markedly less ternary complex formation, suggesting that Tyr 27 of IGF-II contributes to the binding energy in the ternary complex.

DISCUSSION
To ensure optimal folding of the recombinant IGFBP-3 and the mutant, the expression system with PhoA signal sequence was used with the exception of the N domain molecule. All of the expressed proteins were purified by IGF-II coupled affinity column chromatography, a binding property that requires correct folding and permitted comparison of relative affinities. The IGFBP-3 derivatives used in this study were highly purified except for chimera IGFBPs. Each chimera IGFBP contained a mature form and a minor percentage with two amino acids deleted from the amino terminus. Although the reason is unclear, it is possible that the Ala 2 of each chimera IGFBP was recognized as an alternative processing site because the PhoA signal sequence is processed after Ala. On the other hand, IGFBP-3 and the NϩL domain were also expressed with the same PhoA signal sequence, and they showed only Gly-Ala-Gly-Ala-Val-Gly-Ala sequence. We cannot rule out the possibility that IGFBP-3 and the NϩL domain may also be processed at the secondary site, albeit at undetectable levels.
From the HRP-IGF-II/IGFBP-3 solid-phase binding assay and affinity cross-linking experiments, the deletion mutants of IGFBP-3, the NϩL domain, and the N domain showed 12.5 and 4.2% affinities for IGF-II, respectively, relative to full-length IGFBP-3. Previous studies showed that a 30-kDa IGFBP-3 in pregnancy plasma has 20-fold lower affinity for IGF-I than normal plasma IGFBP-3 (33), and a carboxyl-terminal truncated 31-kDa IGFBP-3 from rat serum also has lower affinity for IGF-I (34). The truncated IGFBP-3 resulted from proteolysis by IGFBP-3 proteases and reduced its affinity for IGFs, thereby facilitating dissociation of the complexes and hence increasing the bioavailability of the IGFs (33). Results obtained from the binding of IGFBP-3 deletion mutants to IGF-II are in good agreement with that of proteolyzed and truncated form of IGFBP-3. Furthermore, we found that the 41-residue peptide (position Glu 52 -Ala 92 ) could bind to IGF-II. In a previous study, the amino-terminal 88 amino acids of human IGFBP-3 showed low affinity for IGF-I (20). The 41-residue peptide is the smallest fragment of the IGFBP-3 derivatives found to bind to IGFs. This peptide was designed to obtain two S-S bonds, which were identified by peptide mapping procedures. The remaining four S-S bonds in the N domain could not be determined because of the cysteine-rich region at positions Cys 42 -Gly-Cys 44 -Cys 45 - , and chimera-332 (Ⅺ) were added to the well. The formed complex was detected by reaction first with anti-BP-3 pAb and then with anti-rabbit IgG-HRP, as described under "Experimental Procedures." After subtraction of nonspecific binding obtained in the absence of IGFBP-3, the percentages of bound/total were plotted, where Bound was the quantity of immunoreactive IGFBP-3 derivatives, and Total was obtained in the presence of 100 nM IGFBP-3. The data are expressed as the mean (bars, S.D.) of three determinations.
Leu-Thr-Cys 48 . Therefore, we prepared one peptide that contained the S-S bonds. Although the 41-residue peptide showed 1/500th of the affinity for IGF-II relative to the N domain, the 41-residue peptide whose disulfide bonds were reduced and alkylated did not show any binding affinity (data not shown). Thus, the structure maintained by the S-S bonds are thought to be important for the binding affinity. However, compared to the relative affinities of chimera-332 (100%) and the NϩL domain (12.5%) for IGF-II, the presence of only the N domain is important but not sufficient for full binding. For example, the L and C domains possibly play important roles for conformational stabilization of the N domain.
For ternary complex formation, it was found that only the chimera IGFBP, which possessed the C domain derived from IGFBP-3 (chimera-323), showed ternary complex formation activity in the HRP-IGF-II/IGFBP-3/ALS solid-phase binding assay. Thus, the mutants lacking the C domains, such as the NϩL domain, the N domain, and the 41-residue peptide, were not active in the affinity cross-linking experiment, and it was found that the bands derived from the ternary complex were observed using only IGFBP-3 and chimera-323. Although complexes between 125 I-labeled IGF-II and ALS were not observed (Fig. 5, lane 5), bands derived from the binary complex of 125 I-labeled IGF-II and ALS were detected in the presence of IGFBP-3 or chimera-323 as shown in Fig. 7. A previous study also showed that such a complex was detected by affinity crosslinking in the presence of IGFBP-3 (10). It is speculated that IGF-II does not have a significant binding affinity for ALS but may be positioned near ALS in the ternary complex such that the cross-linker can link IGF-II and ALS in the presence of IGFBP-3. From the observation that IGFBP-3 and chimera-323 can form the ternary complex but that the NϩL domain and chimera-332 cannot, it is likely that the C domain of IGFBP-3 is necessary for binding to ALS. The results of the IGFBP-3/ ALS solid-phase binding assay also suggest that the C domain of IGFBP-3 is important for binding to ALS, even in the absence of IGF-II. In this study, we did not perform an affinity cross-linking experiment to observe chimera IGFBPs binding to ALS. To clarify the binding of IGFBP-3 to ALS, only the C domain expressed in E. coli should be used in the solid-phase binding assay and affinity cross-linking experiments.
Previously, we confirmed the three-dimensional structure of recombinant IGF-II (4) and identified the binding sites of IGF-II for their receptors (22,23) and for IGFBPs (16) using various IGF-II mutants. We also observed IGF-II sites important for the ternary complex formation using IGF-II mutants. Previous studies revealed that Tyr 27 of IGF-II is important for binding to insulin and IGF-I receptors (22) and not for binding to the IGF-II/cation-independent mannose 6-phosphate receptor, and that the Tyr 27 residue is structurally positioned at the opposite side from a binding region for IGFBPs, which include Glu 6 , Phe 48 , Arg 49 , and Ser 50 (4). In this study, the results of the IGFs/IGFBP-3/ALS solid-phase binding assay suggest that mutation of Tyr 27 to Leu does not support ternary complex formation. Similar results have been obtained by other groups using IGF-I mutants in which Tyr 24 substituted with Leu ([Leu 24 ]rIGF-I) reduced the ternary complex formation activity (21). Tyr 24 of IGF-I is also important for binding to insulin and IGF-I receptors (35). Therefore, the Tyr 27 residue of IGF-II, which is important for binding to insulin and IGF-I receptors, interacts only weakly with ALS and is on a side opposite to the Glu 6 , Phe 48 , Arg 49 , and Ser 50 residues, which bind the aminoterminal domain of IGFBP-3, within positions Glu 52 -Ala 92 . Additionally, the carboxyl-terminal domain of IGFBP-3 binds to ALS, even in the absence of IGF-II.