Pergamon

Progress in Growth Factor Research, Vol. 6. Nos. 2-4, pp. 215-222, 1995 Copyright © 1996 Elsevier Science Ltd. All rights reserved

Printed in Great Britain 0955--2235/95 $29.00 + .00

0955-2235(95)011111}4--6

MODULATION OF HUMAN IGF BINDING PROTEIN-3 ACTIVITY BY STRUCTURAL

MODIFICATION

Robert C. Baxter* and Sue M. Firth

Kolling Institute of Medical Research, Royal North Shore Hospital, Sydney, NSW 2065, Australia

To delinate regions of IGFBP-3 involved in ligand and cell-surface binding, DNAs encoding human IGFBP-311-264] and several variants were transfected into ClIO cells. Of three deletion (A) mutants, IGFBP-311-88], [1-184], and [A 89-184], none bound IGF-I tracer by ligand blotting, although all were detectable by immunoblotting. No ALS binding was detectable, as predicted by the lack of IGF binding. Normal- sequence IGFBP-3 associated with the CHO cells and was partly displaceable by IGF- I. Whereas IGFBP-311-88] and [1-184]failed to cell-associate, the non-IGF-binding central deletion variant [ A 89-184] did associate with CH O cells but was not displaced by IGF-L To further examine the role of the carboxy-terminal domain in cell-asso- ciation, the basic sequence IGFBP-31228-232] (KGRKR) was altered to the corre- sponding IGFBP-I residues MDGEA, a major charge reversal. This variant showed reduced IGF-I binding, and bound ALS with decreased affinity as determined by Scatchard analysis. It showed no cell binding, implicating the basic domain in cell- association. We conclude that, whereas the central and carboxy-terminal domain dele- tions fail to bind IGF-I, the ability to cell associate requires the carboxy-terminal but not the central domain. Specifically, the basic region [228-232] is essential for cell binding, and also affects IGF-I binding, and independently, ALS affinity.

Keywords: IGFBP-3, ALS, mutagenesis, cell association, complex formation.

INTRODUCTION

The insulin-like growth factor binding protein family comprises six proteins, named IGFBP-1 to IGFBP-6 in the approximate order of the characterization of their cDNA and protein sequences [1, 2]. Their highly conserved primary structures may be divided into three regions of approximately equal length, each protein having cysteine-rich amino-terminal and carboxy-terminal domains which are well conserved throughout the family, and a central domain unique to each protein.

*Correspondence to: R. C, Baxter. Acknowledgments--This study was supported by the National Health and Medical Research Council,

Australia.

215

216 R. C Baxter and S. M. Firth

Two members of the family, IGFBP-3 and IGFBP-5, are distinguished from the others by the highly basic regions located in the carboxy-terminal portion of their sequences. As shown in Fig. 1, these basic regions consist of almost identical 18- residue sequences (IGFBP-31215-232] and IGFBP-51201-218]) of which 10 amino acids are Lys or Arg, and none are acidic. In contrast, the corresponding region IGFBP-l[171-188] contains two basic and three acidic residues [1].

In addition to binding IGF-I and IGF-II, both IGFBP-3 and IGFBP-5 are known to associate with cell surface and matrix components [3-5]. All of these interactions appear to involve negative charges on the ligands (IGFs or cellular binding elements). For example, Glu 3 of IGF-I, and Glu 6 of IGF-II, are well estab- lished to be essential residues for interaction with IGFBPs, such that amino- terminal truncation (in des[1-3]IGF-I and des[1-6]IGF-II), or charge-reversal (in [Arg3]IGF-I or [Arg6]IGF-II) greatly decreases the binding affinity of the peptides [6]. Similarly, the association of both IGFBP-3 and IGFBP-5 with skin fibroblasts is reversed by the polyanion heparin, suggestive of the interaction of the binding proteins with sulfated glycosaminoglycans or other negatively charged elements [4, 7].

Uniquely among the IGFBPs, IGFBP-3 interacts with a third type of ligand, the leucine-rich glycoprotein known as the acid-labile subunit (ALS). Despite the marked hydrophobicity of this protein--125 of its 578 residues are leucines--its interaction with IGFBP-3 clearly involves charged residues, since ALS binding affinity is greatly decreased by quite modest increases in ionic strength, and by heparin and other polyanions [8, 9]. That ALS bears a strong negative charge at pH values near neutrality is illustrated by its tight binding to diethylaminoethyl and ethyleneimine anion-exchange columns, requiring high salt concentrations for displacement. Although no studies to date have identified specific domains of func- tional importance in the ALS sequence, the markedly acidic sequence hALS[23-28], -Asp-Asp-Asp--Ala-Asp-Glu-, clearly warrants attention in this regard [9].

Cysteine.dch, conserved Non-conserved Cysteine-dch, conserved ' '" ' ' " ' " ~ ~ C

IGFB~I

~ F B ~ 3

88 18& . . - " ~

P~GFYHSRQCRT~--------~LCWUVYPWNGKRIPGSPEIRGDPNCQIYFNV~N 211 ~ 26& P N C D X X G ~ Q C I ~ FCWCVDrZGQ P L PGYTTKGIaD~'HCYSMQ 8 K

197 252 IGFBP-5 PNCDRKGFYID~C 1 ~ 8RGRKRG l C~ZVDrZGNKLP GNrZVDGDFQCHTFD 8 ~

FIGURE 1. The domain organization of the IGF binding proteins, with residues marking the end of the amino-terminal and central domains of human IGFBP-3 (residues 88 and 184, respectively) indicated. Parts of the carboxy-terminal sequences of IGFBP-I, -3 and -5 are also shown, with the basic regions of IGFBP-3 and -5, and the corresponding sequence of IGFBP-I, underlined. Boxes indicate the basic 5- residue sequence of IGFBP-3 which was mutated to the corresponding acidic sequence of IGFBP-1. Derived from data in Drop et al. [1]

Characterization of IGFBP-3 Mutants 217

As part of this laboratory's continuing studies into the regulation and actions of IGFBP-3, we have investigated the regions of this protein involved in its various protein-protein interactions by introducing mutations into the cDNA for hlGFBP- 3 and expressing the corresponding proteins in mammalian cells.

METHODS

Full details will be presented elsewhere (Firth SM, Baxter RC, in preparation). In brief, constructs for transfection were based on a 1080 bp EcoRI-Pvu l l fragment of the IGFBP-3 cDNA, ibp.118 [10], inserted into pSELECT (Promega). Truncation and site-specific mutations were generated by polymerase chain reaction, and cloned into the corticosteroid-inducible mammalian expression vector, pMSG (Pharmacia). Transfection into CHO cells was performed using polybrene, and after a 3-week selection period to establish stable transfectants, cells were grown in serum-free a-MEM containing 0.1% BSA and 10 /~M dexamethasone.

Production of IGFBP-3 was established by specific RIA for human IGFBP-3 [11], and by ligand- and immunoblotting after SDS-PAGE. Cell-association of IGFBP-3 variants was determined immunologically [4], and ternary complex forma- tion was measured using 125I-labeled ALS [12].

RESULTS AND DISCUSSION

Deletion Mutants

Figure 1 shows the three structural domains of IGFBP-3 (which are common to all of the IGFBPs) and details part of the sequence of the carboxy-terminal domain of IGFBP-1, -3 and -5. Deletion mutants of IGFBP-3 were produced containing only the amino-terminal domain (IGFBP-3[1-88]), the amino-terminal and central domains (IGFBP-3[1-184]), and the amino-terminal and carboxy-terminal domains, with the central domain deleted (IGFBP-3[A89-184]). As determined by immunoblotting after SDS-PAGE, wild-type recombinant IGFBP-3 had a similar electrophoretic mobility to serum-derived IGFBP-3, with major bands at apparent molecular weights 40 and 45 kDa. The three potential N-glycosylation sites of IGFBP-3 are all found in the central domain, and we have shown elsewhere that Asn 89 and Asn 1°9 are usually glycosylated, whereas the doublet seen on SDS-PAGE is due to optional glycosylation of Asn 172 [13].

Of the deletion mutants, a single major band was detected by immunoblotting at ~14 kDa for [1-88], slightly larger than its predicted size, but consistent with its expected lack of glycosylation. The central domain deletion [A89-184], also showed a single band, of ~16 kDa, slightly smaller than its predicted size. In contrast, a doublet of ~30 and 35 kDa was seen for [1-184], suggesting that its glycosylation sites are intact, and that Asn t72 is again optionally glycoslyated. Although the epitopes recognized by the polyclonal IGFBP-3 antiserum R-100, used in this study, have not been identified, it is interesting to note that it reacted with all of the dele- tion mutants, suggesting that a major determinant must reside in residues 1-88, the only sequence common to all mutants.

218 R. C. Baxter and S. M. Firth

When analysed by ligand blotting, none of the deletion mutants showed any binding of ~25I-labeled IGF-I. This observation contrasts with the report by Sommer et al. [14] that the truncation mutants IGFBP-3 [1-88] and [1-161], and the deletion mutants [A92-184] and [A92-223], all retained some IGF-binding, as demonstrated by affinity labeling. While the explanation for these differing results is not known, it is possible that the cross-linking technique was able to stabilize some verY low- affinity interactions between these variants and IGF-I.

Cell Association

The consequences of the association of IGFBP-3 with cell-surface or matrix components are poorly understood. The ability of IGFBPs to be affinity cross- linked to cells was demonstrated almost a decade ago, although the IGFBPs involved were not identified [15]. More recently, cell-association of IGFBP-3 has been postulated to be an important step in the ability of this protein to potentiate IGF action in certain circumstances [16]. In this laboratory, heparin was shown to displace IGFBP-3 from the cell-surface or matrix of skin fibroblasts, after which the protein could be detected by RIA in the culture medium [4]. By analogy with other heparin-displaceable proteins, it was suggested that the cell-surface components involved in binding IGFBP-3 might be sulfated glycosaminoglycans. The recent demonstration that treatment of endothelial cells, glial cells and fibroblasts with heparinase, or growing endothelial cells in sodium chlorate, does not reduce their ability to bind IGFBP-3 would tend to exclude the involvement of heparan sulfate proteoglycans, at least in these cell types [7, 17]. In a more direct approach to the identification of the IGFBP-3 binding components, Oh et al. [18] demonstrated proteins of 20, 26 and 50 kDa on human breast cancer cell membranes, that could be covalently cross-linked to nonglycoslyated E. coli hlGFBP-3. However, the structure, and possible signal transducing properties of these proteins have not been established.

The binding of the IGFBP-3 deletion mutants [1-88], [1-184] and [A89-184] to the CHO cells that produced them is shown in Fig. 2. Binding was detected as the specific cell association of an IGFBP-3 antiserum, indicating the presence of binding protein [4]. Intact IGFBP-3 bound well to CHO cells, and was displaced by the addition of IGF-I, as previously demonstrated for fibroblast-bound IGFBP-3. This displacement is not an IGF receptor-mediated action of IGF-I, since IGF-I analogues with minimal affinity for the type I receptor are as effective as normal IGF-I [4]. The two truncated forms of IGFBP-3 lacking the carboxy-terminal domain, [1-88] and [1-184], failed to bind to CHO cells. In contrast, the deletion mutant which retained the carboxy-terminal domain but lacked the central domain, [A89-184], did associate with CHO cells. However, this mutant, which does not bind IGF-I, was not displaced from CHO cells by the addition of IGF-I, consistent with the observation that the displacement is a consequence of IGF-I binding to the binding protein.

Since the carboxy-terminal domain of IGFBP-3 was implicated in its cell-asso- ciation, we investigated the involvement of the highly basic region, IGFBP- 31215-232], in this binding. Selected for mutagenesis was the 5-residue sequence IGFBP-3 [228-232], Lys-Gly-Arg-Lys-Arg, since a substitution of these residues for the corresponding residues of IGFBP-1, Met-Asp-Gly-Glu-Ala, involves a

Characterization of lGFBP-3 Mutants -219

z 1 o i t..

"U"~C "1

"0 <. ~.=_

=ii2 • ~

J FIGURE 2 Association of IGFBP-3 and variants, as indicated, with CHO cells. Cell-bound IGFBPs were detected by incubation with IGFBP-3 antiserum and reaction with radiolabeled Protein-A. Cells were incubated without (open bars) or with (solid bars) 50 ng/ml IGF-I.

charge-reversal from +4 to -2, presumably without a major disturbance to the overall IGFBP conformation. This mutant was detectable by immunoblot and by ligand blot, although its IGF-binding appeared relatively weak compared to wild- type IGFBP;3 (not shown). However, it failed to show any binding to CHO cells (Fig. 2), suggesting that the basic residues at IGFBP-31228-232] are indeed important in cell-association.

ALS Binding

Finally, the ability of the various mutant forms of IGFBP-3 to form a ternary complex with ALS was determined. Although the concept that ALS binding requires a binary IGF-IGFBP-3 complex rather than free IGFBP-3 has been chal- lenged recently, on the basis of some studies with rat ALS, and others with non- glycoslyated E. coli hlGFBP-3 [19, 20], data from the authors' laboratory have consistently shown that, whether analysed by gel chromatography, affinity labeling and SDS-PAGE, or immunoprecipitation using antibodies against IGFBP-3, there is little or no binding of natural human ALS to natural human IGFBP-3 in the absence of IGF-I or IGF-II. Recently we have confirmed this by measuring the binding of radioiodinated natural human IGFBP-3 to ALS, with precipitation of the complex using anti-ALS antibodies. In the absence of IGF-I, no IGFBP-3 tracer is immunoprecipitated.

Indeed, a series of IGF-I analogues with reduced affinity for IGFBP-3 show correspondingly decreased ternary complex formation with ALS, emphasizing the importance of the binary complex as the 'binding site' for ALS [12]. The exact

220 R. C. Baxter and S. M. Firth

involvement of IGFs in the ternary complex is not fully understood, but it appears that key residues in the IGFs must stabilize the ALS interaction. In the presence of IGF-I analogues lacking the C- or D-domains, or Tyr 24, all of which have high affinity for IGFBP-3, ALS binding affinity to IGFBP-3 is very low, implicating these domains of IGF-I in the interaction with ALS [12].

As noted previously, the interaction between IGFBP-3 and ALS appears to have a major ionic component, despite the extremely hydrophobic nature of ALS. Therefore, it was of interest to determine regions of IGFBP-3 that might participate in this interaction, and particularly to investigate whether the strongly basic region in the carboxy-terminal domain might be involved. Figure 3a shows the binding of ~25I-labeled ALS to normal-sequence recombinant IGFBP-3, in the presence of IGF-I. None of the deletion mutants described above showed any ALS binding (not shown), as predicted by their lack of IGF binding. The 'basic region' mutant, however, retained the ability to bind ALS, though less than that of normal IGFBP-3 (Fig. 3a).

When analysed by Scatchard plot (Fig. 3b), normal IGFBP-3, in the presence of IGF-I bound ALS with an association constant of 1.55 +0.36 × 10 ~° 1/mol (mean + SD, n = 4), when measured at 22°C, low ionic strength, and pH 6.5. Due to a modification in the analytical method used previously, this estimate is approxi- mately I0 times higher than the values of ~10 9 1/mol which we reported in earlier studies [12]. The affinity for ALS is, however, reduced by two orders of magnitude to a value of 2.5 x 108 1/mol when measured at 37°C, physiological ionic strength and pH 7.4, with temperature, salt concentration and pH each having an indepen- dent effect [21]. Scatchard analysis of the 'basic region' mutant (Fig. 3b) showed an approximate 50% reduction in the affinity of ALS binding to 8.1 + 0.9 × 10 ~° 1/mol (SD, n = 4; P < 0.05), together with a marked decrease in the apparent binding site concentration, from 0.95 + 0.015 to 0.11 + 0.03 mol ALS bound per mol IGFBP-3

b U

so

40

o~

h5 30 .E m ~ 20

~'1o

0

~ rhlGFBP-3 / a 22SMDG~ Q)

i i . - . . . . . i t - . . . . . , J , . . . . . .

0.1 1 10 100 rhlGFBP-3 (ng/tube)

1.2

1.0

0.8

0.6

0.4

0.2

oN. 0 0.02

• rhlGFBP-3 MDGEA

0.04 0.06 0.08 0.10 ALS Bound (nmol/liter)

FIGURE 3. Binding of ALS to IGFBP-3 and the 'basic region' variant. (a) Increasing concentrations of binding proteins were incubated with [12Sl]ALS in the presence of 50 ng IGF-I. (b) Scatchard plots derived from binding curves in which increasing concentrations of unlabeled ALS competed for [1zSI]ALS binding to a fixed concentration of IGFBP-3 (1 ng for wild-type, 2.5 ng for the 'basic region' variant), in the presence of 50 ng IGF-I. Complexes were precipitated with IGFBP-3 antiserum. Derived binding constants are presented in the text.

Characterization of IGFBP-3 Mutants 221

(P < 0.0001). As we have reported earlier, a reduction in apparent binding site concentration is a reflection of decreased binary (IGF-IGFBP-3) complex [12]. Thus, we conclude from this analysis that the 'basic region' mutant has a much lower binding of IGF-I than normal IGFBP-3 consistent with the much weaker binding seen by ligand blotting.

SUMMARY AND CONCLUSIONS

In this study we have used deletion and site-specific mutants of human IGFBP- 3, expressed in its glycosylated form by transfected CHO cells, to help delineate some regions of the protein involved in its interactions with IGFs, ALS and cell- surface components. Of the major deletion mutants tested, lacking the central non- conserved domain, the carboxy-terminal domain, or both, none retained IGF binding activity as detected by ligand blotting. As predicted from this laboratory's observation that a binary IGF-IGFBP-3 complex is required for normal ALs binding, none of the deletion mutants could bind ALS. A variant lacking the central domain but retaining the carboxy-terminal domain was shown to associate with CHO cells, whereas those lacking the carboxy-terminal domain did not.

The highly basic sequence IGFBP-31228-232] was found to be a region of the carboxy-terminal domain specifically involved in cell-association, since its mutation to the acidic sequence found in the corresponding region of IGFBP- 1 totally blocked cell association. This mutation also had low IGF binding, as seen by ligand blot and confirmed by the low concentration of ALS binding sites (i.e. binary complex) determined by Scatchard analysis of ALS binding. In addition to its effect on IGF- I binding, an independent effect was seen on the binding of ALS, the affinity of which was reduced by half, compared to its binding to natural-sequence IGFBP-3.

We have therefore demonstrated that elements essential to the structure and/or conformation of the IGF binding site reside in both the central and carboxy-termi- nal domains of IGFBP-3, and that the central domain is not required for cell-asso- ciation. Basic residues 228-232 of the carboxy-terminal domain appear to play a role both in IGF binding and, independently, in ALS binding, and are essential for cell-association. The availability of an IGFBP-3 variant that retains at least some IGF binding, while failing to bind the cell surface, provides a useful tool to help delineate the role of cell-association in the potentiation of IGF action by IGFBP-3, and to investigate its possible IGF-independent actions.

REFERENCES

I. Drop SLS, Schuller AGP, Lindenbergh-Kortleve D J, Groggen C, Brinkman A, Zwarthoff EC. Structural aspects of the IGFBP family. Growth Regul. 1992; 2: 69-79.

2. Zapf J. Physiological role of the insulin-like growth factor binding proteins. Eur J Endocrinol. 1995; 132: 645~54.

3. Conover CA. Glycosylation of insulin-like growth factor binding protein-3 (IGFBP-3) is not required for potentiation of IGF-I action: evidence for processing of cell-bound IGFBP-3 Endocrinology 1991; 129: 3259-3268.

4. Martin JL, Ballesteros M, Baxter RC. Insulin-like growth factor-I (IGF-I) and transforming growth factor-ill release IGF-binding protein-3 from fibroblasts by different mechanisms. Endocrinology 1992; 131: 1703-1710.

5. Jones JI, Gockerman A, Busby WH, Camacho-Hubner C, Clemmons DR. Extracellular

222 R. C. Baxter and S. M. Firth

matrix contains insulin-like growth factor binding protein-5: potentiation of the effects of IGF-I. J Cell Biol. 1992; 121: 679-687.

6. Francis GL, Aplin S, McNeil KA, Milner S J, Forbes BE, Ross M, Ballard F J, Wallace JC. Molecular interactions of IGF-I and IGF-II with their binding proteins and receptors. In: Baxter RC, Gluckman PD, Rosenfeld RG, eds. The insulin-like growth factors and their regulatory proteins. Amsterdam: Elsevier; 1994: 57-66.

7. Booth BA, Boes M, Andress DL, Dake BL, Kiefer MC, Maack C, Linhardt RJ, Bar K, Caldwell EEO, Weiler J, Bar RS. IGFBP-3 and IGFBP-5 association with endothelial cells: role of C-terminal binding domain. Growth Regul. 1995; 5: 1-17.

8. Baxter RC. Glycosaminoglycans inhibit formation of the 140 kDa insulin-like growth factor- binding protein complex. Biochem J. 1990; 271: 773-777.

9. Leong SR, Baxter RC, Camerato T, Dai J, Wood WI. Structure and functional expression of the acid-labile subunit of the insulin-like growth factor-binding protein complex. Mol Endocrinol. 1992; 6: 870-876.

10. Wood WI, Cachianes G, Henzel WJ, Winslow GA, Spencer SA, Hellmiss R, Martin JL, Baxter RC. Cloning and expression of the growth hormone-dependent insulin-like growth factor-binding, protein. Mol Endocrinol. 1988; 2:1176-1185.

11. Baxter RC, Martin JL. Radioimmunoassay of growth hormone-dependent insulinlike growth factor binding protein in human plasma. J Clin Invest. 1986; 78: 1504-1512.

12. Baxter RC, Bayne ML, Cascieri MA. Structural determinants of binary and ternary complex formation between insulin-like growth factor-I (IGF-I) and IGF binding protein-3. J Biol Chem. 1992; 267: 60-65.

13. Firth SM, Baxter RC. The role of glycosylation in the action of IGFBP-3. Prog Growth Factor Res. In press.

14. Sommer A, Maack CA, Spratt SK, Mascarenhas D, Tressel TJ, Rhodes ET, Lee R, Roumas M, Tatsuno GP, Flynn JA, Gerber N, Taylor J, Cudny H, Nanney L, Hunt TK, Spencer EM. Molecular genetics and actions of recombinant insulin-like growth factor binding protein-3. In: Spencer EM, ed. Modern concepts of insulin-like growth factors. New York: Elsevier; 1991: 715-728.

15. Clemmons DR, Han VKM, Elgin RG, D'Ercole AJ. Alterations in the synthesis of a fibroblast surface associated 35K protein modulates the binding of somatomedin-C/insulin-like growth factor I. Mol Endocrinol. 1987; 1: 339-347.

16. Conover CA. Potentiation of insulin-like growth factor (IGF) action by IGF-binding protein-3: studies of underlying mechanism. Endocrinology 1992; 130: 3191-3199.

17. Yang YW-H, Yanagishita M, Rechler MM. Heparin-inhibitable binding of insulin-like growth factor binding protein-3 (IGFBP-3) to rat glial cells and human fibroblasts is not to heparan sulfate glycosaminoglycans. Proceedings, 77th Annual Meeting, Washington, DC: The Endocrine Society; 1995: 164.

18. Oh Y, Muller HL, Pham H, Rosenfeld RG. Demonstration of receptors for insulin-like growth factor binding protein-3 on Hs578t human breast cancer cells. J Biol Chem. 1993; 268: 26,045-26,048.

19. Lee CY, Rechler MM. A major portion of the 150-kilodalton insulin-like growth factor-binding protein (IGFBP) comialex in adult rat serum contains unoccupied, proteolytically nicked IGFBP-3 that binds IGF-II preferentially. Endocrinology 1995; 136: 668-678.

20. Barreca A, Ponzani P, Arvigo M, Giordano G, Minuto F. Effect of the acid-labile subunit on the binding of insulin-like growth factor (IGF)-binding protein-3 to [125I]IGF-I. J Clin Endocrinol Metab. 1995; 80: 1318-1324.

21. Holman SR, Baxter RC. Insulin like-growth factor binding protein-3: factors affecting binary and ternary complex formation. Growth Regul. 1996; 6: 42-47.