Structural Determinants for Binary and Ternary Complex Formation ...

THEJOURNALOF BIOLOGICAL CHEMISTKY 1992 by T h e American Soclety for Biochemistry and Molerular B~ology, Inc

Val. 267. No. 1, Issue of January 5, pp. 60:65, 1992 Printed In U.S.A.

ic'

Structural Determinants for Binary and Ternary Complex Formation IGF Binding between Insulin-like Growth Factor-I (IGF-I) and Protein-3* (Received for publication, April 22, 1991)

Robert C. BaxterSg, Marvin L. Baynefl, and MargaretA. CascieriII From the $Department of Endocrinology, Royal Prince Alfred Hospital, Camperdown, New South Wales 2050, Australia and the lDepartment of Growth Biochemistryand IIBiochemical Endocrinology, Merck Sharp & Dohme Research Laboratories, Rahway, New Jersey07065

Structural analogs of recombinant human insulinlike growth factor-I (IGF-I), with alterations to each of the B, C, A, and D domains, have been tested for their ability to form binarycomplexes with IGF-binding protein-3 (IGFBP-3) and ternarycomplexes with IGFBP-3 and the acid-labile subunit (0-subunit). Two functionally distinct regions of IGF-I have been identified. The first,involving residues 3 and 4 and the ahelix between residues 8 and 18 of the B-domain, as well as residues 49-51 in the A-domain, appears important for IGFBP-3 binding, such that substitution of theseresiduesresultsindecreasedbinary complex available for a-subunit binding. The second region, distal to the IGFBP-3-binding epitope and primarily involving the D-domain and B-domain near residue 24, with some involvement of the C-domain, appears slightly inhibitory to binary complex formation, such that analogs with atruncated D-domain or with aGly, bridge substituted for the C-domain show enhanced binding to IGFBP-3. However, binary complexes formed from these analogs bind the a-subunit with reduced affinity, the effect being most marked when substitution of the C-domain, or replacement of TyrZ4, is superimposed on D-domain truncation. It is concluded that although the a-subunitdoes not itself bind IGF-I, its interaction with IGFBP-3 in the ternary complex is dependent on structural determinants on IGF-I distal to the IGFBP-3 binding domain.

observation that bound formsof IGFs have greatly extended circulating half-lives comparedwith free IGFs (9). In contrast, the a-subunit circulates in a 2-%fold molar excess over the other components of the ternary complex; this may help to explain why most of the IGFBP-3 is in the ternary form, despite the low affinity constant for a-subunit binding (5, 8, 10). The two binding sites on IGFBP-3 are functionally related. Whereasthe presence of a-subunithasno effect onthe interaction of IGF-I or IGF-I1 with the IGF binding site (5), the interaction of a-subunit with its binding siterequires the presence of an IGF (5, 6). Thus there are nounoccupied IGF binding sites in the circulating 140-kDa complex since a-fl dimers are unableto form. Little else is known of the factors affecting a-subunit binding although it clear is that thereis a major charge component to the interaction,which is strongly inhibited by high ionic strength andpolyanions such as heparin (11).The regulation of ternary complex formationis likely to be an important componentof the overall regulation of IGF action since circulating IGFs are thought to be unable to cross the capillary barrier when in the ternarycomplex but are able to reach the tissueswhen in binary complexes with IGFBPs (12, 13). Although the presence of IGF is requiredfor a-subunit binding to IGFBP-3, IGF binding site occupancy is not the only determinant of a-subunit binding. For example, whereas IGF-I1 binds to IGFBP-3 with a higher affinity than IGF-I (7), a-subunit binds with higher affinity to the P-7 complex containing IGF-I than to that containing IGF-I1 (5). To gain a more detailed understandingof the structural requirements for ternary complex formation, we have employed a series of IGF-I analogs (14-17) with defined mutations affecting each of the four domains (B, C, A, and D) of the IGF-I structure (18). These studies indicate that binding affinity o f the asubunit for P-7 complexes is strongly dependent on the structure of the IGF in thecomplex, with determinants in each of the IGF-I domains.

Peptides of the insulin-like growth factor (IGF)' family circulate predominantly as part of ternary complexes of approximately 140 kDa, comprising an acid-labile glycoprotein of -85 kDa (a-subunit), an acid-stableglycoprotein of 40-50 kDa, IGF-binding protein-3 (P-subunit), and either IGF-I or IGF-I1 (y-subunit) (1-6). IGF-binding protein-3 (IGFBP-3) appears central to thiscomplex, with a separate high affinity binding site ( K O= 2-3 X 10" M") for IGFs (7) and lower affinity binding site ( K O= 4-6 x loR M-') for the a-subunit MATERIALS AND METHODS (5).The total concentration of IGFs in serum is approximately Peptides and Proteins-Natural human IGF-I was isolated from equimolar with that of IGFBP-3 (8), consistent with the * This study was supported by a grant from the National Health and Medical Research Council, Australia (to R. C. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18U.S.C. Section 1734 solely to indicate this fact. To whom correspondence should be sent. Tel.: 61-2-516-6150; Fax: 61-2-516-1273. The abbreviations used are: IGF, insulin-like growth factor; IGFBP, insulin-like growth factor-binding protein.

60

Cohn fraction IV of human plasma (19) and iodinated with N a 9 as in previous studies (7). Recombinant human IGF-I and the following IGF-I analogs were prepared, expressed, purified, and characterized as described previously. B-domain variants were [Gln3,Ala']IGF-I, [Tyr15, Leu"]]IGF-I, [Gln3,Ala4,Tyr15,Leu16]IGF-I, Phe",Val',Asn2, Gln3,His4,SerR,Hisg,Glu1z,Tyr'5,Leu'"]IGF-I (B-chainmutant) (151, and [SerZ4]IGF-I(14). A-domain variants were [Ile41,G1~45,Gln46, Thr4Y,Ser5~,Ile",Ser",Tyr55,Gln56]IGF-I (A-chainmutant), [Thr4', Ser50,11e51]IGF-I, and [ T Y ~ ~ ~ , G ~ ~(16); ~ ~ ]CI Gand F -D-domain I variants were [1-62]IGF-I, [1-27,Gly4,38-70]IGF-I, [l-27, Gly4,38-62] IGF-I (17), and [LeuZ4,1-62]IGF-I.IGFBP-3 was isolated from Cohn

IGF-I Domains Involved

IGFBP-3 in

and a-Subunit Complexes

61

fraction IV of human plasma (7) and the a-subunit from human serum ( 3 , as in previous studies; 'Z51-labeleda-subunit was prepared and purified by ion-exchange chromatography as described previously (10). Binding Studies-Binding of IGF-I analogs to IGFBP-3 was measured by competition with lZ51-labeledIGF-I, essentially as described previously (5, 7). In brief, incubations in 0.3mlof 50 mM sodium phosphate, 2.5 g/liter bovine albumin (Sigma, radioimmunoassay grade), pH 6.5, contained 0.5 ng of pure IGFBP-3, lZ51-labeledIGF-I (approximately 10,000 cpm, 25 pg), and IGF-I analogs tested over the concentration range 0.01-100 ng/0.3 ml. After 2 h at 22 "C, bound tracer was separated from free by incubation for 4 h at 22 "C with 0.5 pl of anti-IGFBP-3 antiserum R-7, and precipitation in the presence of2.5 p1 of goat anti-rabbit immunoglobulin (Bioclone, Sydney, Australia) and 4.5% finalconcentration polyethylene glycol 6000 (Merck) as described previously ( 5 ) . Ternary complex formation in the presence of different IGF-I analogs was measured by the binding of lZ51-labeleda-subunit (approximately 10,000 cpm, 1 ng) to mixtures of IGFBP-3 (10 ng) and IGF-I analogs, exactly as described previously ( l l ) , except that the IGF-I analogs were tested over the concentration range 0.05-100 ng/ 0.3 ml. Bound tracer was separated from free by immunoprecipitation as described above, with 1 pl of antiserum R-7 and 5 pl of goat antirabbit immunoglobulin. To study the IGFBP-3 dose dependence of ternary complex formation, an identical protocol was used, except that the IGF-I analog concentration was held constant at 10 ng/0.3 ml, and theIGFBP-3 concentrationwas varied over the range 0.5-50 ng/0.3 ml. Because of the higher IGFBP-3 concentrations used, 2.5 pl of antiserum R-7 and 10 pl anti-rabbit immunoglobulin were required for immunoprecipitation. The association constant for a-subunit binding to P-7 (i.e. IGFBP. IGF) complexes was determined as described above, except that the concentrations of IGFBP-3 and theIGF-I analogs were held constant at 10 ng/0.3 ml. Unlabeled pure human a-subunit was generally added over the concentration range 10-400 ng/0.3 ml, although for some curves the 400-ng point was omitted. Bound and free tracer were separated with 1 p1 of antiserum R-7 and 5 pl of anti-rabbit immunoglobulin. Nonspecific binding, measured as radioactivity immunoprecipitated in the absence of IGFBP-3, was subtracted from total binding before Scatchard plots were constructed. Data were fitted by linear regression, and differences between binding parameters were evaluated by Scheffgs test after analysis of variance, performed using Statview SE+Graphics (Abacus Concepts, Berkeley, CAI.

[Gln3,Ala4,Tyr'5,Leu16]IGF-I was parallel to thatfor IGF-I; in three experiments, the relative activity was lower than that of IGF-I by a factor of 110 f 30 (S.D.). The B-chain mutant, tested at up to 100 ng/0.3 ml, was unable to compete with '2sI-labeled IGF-I for binding toIGFBP-3. Mutations of TyrZ4 were tested by comparing [SerZ4]IGF-Iwith intact IGF-I(Fig. l), and the carboxyl-terminally truncated mutant, [LeuZ4,162lIGF-I, with thecorresponding IGF-I analog, [l-62lIGF-I. Data for[LeuZ4,1-62]IGF-I are shown below, under"The Effect of C- and D-domain Mutations." For both IGF-I and [l-62lIGF-I, analogs substituted at residue 24 gave displacement curves that were approximately parallel to the correspondingunsubstitutedanalogand reduced in relative IGFBP-3 binding activityby about 2-fold. Fig. 2a shows IGF dose-response curves for ternary complex formation, i.e. a-subunit binding to binary complexes. In five experiments, half-maximal a-subunit binding was seen at an IGF-I concentration of0.26 f 0.15 ng/0.3 ml (S.D.). Thebinding of a-subunit to the @-y complex containing [Tyr's,Leu'6]IGF-I consistently occurred less readily than to the complex containing [Gln3,Ala4]IGF-Ialthough their binding to IGFBP-3 appeared similar. Interestingly, a-subunit binding in the presence of the latter peptide was virtually identical to that seen in the presence of IGF-I, over the entire range of IGF concentrations (Fig. 2a). For this reason and because of limited supplies of [Gln3,Ala4]IGF-I, no further ternary complex formation studies were performed with this peptide. [Gln3,Ala4,Tyr'5,Leu'6]IGF-I gave adose-response curve for a-subunit binding parallel to that for IGF-I, but 100-fold lower in apparent activity, consistent with the lower binding of this analog to IGFBP-3. In presence the of B-chain mutant little or no a-subunit binding was seen, as expected from its lack of binding to IGFBP-3. The dose-response curves for [SerZ4]IGF-I and [Le~*~,1-62]1GF-Ireached both aplateau at 30-40% of the a-subunitbinding seen with the corresponding peptides lacking the residue 24 substitution (not shown). IGFBP-3 dose-responsecurvesfor B-domainvariantsare shown in Fig. 2b. Inthreeexperiments, half-maximal aRESULTS subunit binding in the presence of IGF-I was seen at an The Effect of B-domain Mutations-Fig. 1 shows the com- IGFBP-3 concentration of 1.9 f 0.2 ng/0.3 ml (S.D.). The petition for l*sI-labeled IGF-I binding to IGFBP-3 by IGF-I curves obtained in the presence of [Tyr'S,Le~'6]IGF-I and [SerZ4]IGF-Iwere parallel to that seen with IGF-I butreduced and IGF-I variants altered in the B-domain. Similar curves for IGF-I were obtained whether the natural or recombinant in apparent activityby 2-5-fold. Similarly, a-subunit binding peptide was used, with half-maximal displacement by 0.19 in thepresence of [LeuZ4,1-62]IGF-Ishowed parallelism with a 5-fold decrease inapparent f 0.06 ng of IGF-I/0.3 ml reaction mixture (S.D., n = 5). that for [l-62lIGF-I,with gave a non[Gln3,Ala4]IGF-Iand [Tyr's,Leu'6]IGF-I gave similar compe- activity. In contrast, [Gln3,Ala4,Tyr's,Leu'"]IGF-I tition curves, nonparallel with the IGF-I curve and somewhat I lower inactivitythanIGF-I.Incontrast,the curvefor

-E