0013-7227/01/$03.00/0
Printed in U.S.A.
The Journal of Clinical Endocrinology & Metabolism 86(10):4943– 4950
Copyright © 2001 by The Endocrine Society
Mutation of Three Critical Amino Acids of the
N-Terminal Domain of IGF-Binding Protein-3
Essential for High Affinity IGF Binding
C. K. BUCKWAY, E. M. WILSON, M. AHLSÉN, P. BANG, Y. OH,
AND
R. G. ROSENFELD
Department of Pediatrics, Oregon Health Sciences University (C.K.B., E.M.W., Y.O., R.G.R.), Portland, Oregon 97201; and
Department of Woman and Child Health, Karolinska Institute and Hospital (M.A., P.B.), Stockholm, Sweden
The N-terminal domain is conserved in all members of the
IGF-binding protein superfamily. Most recently, studies have
demonstrated the importance of an IGF-binding protein Nterminal hydrophobic pocket for IGF binding. To examine
more critically the amino acids important for IGF binding
within the full-length IGF-binding protein-3 protein while
minimizing changes in the tertiary structure, we targeted residues I56, L80, and L81 within the proposed hydrophobic
pocket for mutation. With a single change at these sites to the
nonconserved glycine there was a notable decrease in binding. A greater reduction was seen when both L80 and L81 were
substituted with glycine, and complete loss of affinity for
O
VER THE LAST decade, the role of the IGF system in
growth has been extensively studied. IGF-I and -II are
potent mitogens that are tightly regulated in vivo by the six
IGF-binding proteins (IGFBPs), for which they have high
affinity (1, 2). IGFBP-3 is the most abundant binding protein
in human serum and the major carrier of IGF-I in the circulation (3). Although IGFBP-3 can attenuate the effects of IGF-I
by sequestering it from its receptor (4), IGFBP-3 may also act
independently of IGF-I via a putative receptor (5), inhibiting
cellular proliferation (6 –11). Using synthetic IGFBP-3 fragments, Yamanaka et al. (12) presented evidence that the nonconserved midregion of IGFBP-3 most likely contains the
IGFBP-3 receptor-binding site and, presumably, the ability to
inhibit cellular growth in an IGF-independent manner.
Structural analysis of IGFBP-1 to -6 reveals similar modular construction, with the greatest similarity (58%) in the
N-terminus, including 12 cysteines in IGFBP-1 to -5 and 10
cysteines in IGFBP-6 (13). The mid region is highly variable,
but the C-terminus has 34% similarity, containing another
cysteine-rich region. Much of our knowledge of the function
of these regions has been learned through the examination of
proteolytic and synthetic IGFBP fragments and through directed mutational analysis. Several laboratories have shown
from fragment analysis that the N-terminus contains an IGFbinding site (14 –20), but a few others have also suggested
another binding region in the C-terminus (21–23).
The three-dimensional structures of IGF-I (24) and IGF-II
(25) have been known since the early 1990s, but other than
Abbreviations: GST, Glutathione-S-transferase; HBS-EP buffer, 0.01 m
HEPES, 0.15 m NaCl, 3 mm EDTA, and 0.005% surfactant P20, pH 7.4;
IGFBP, IGF-binding protein; IRMA, immunoradiometric assay; IRS-1,
insulin receptor substrate-1; NMR, nuclear magnetic resonance; rhIGF,
recombinant human IGF; TBS-T, Tris-buffered saline-0.1% Tween 20.
IGF-I and IGF-II occurred when all three targeted amino acids
were changed to glycine. Furthermore, the ability of the IGFbinding protein-3 mutants to inhibit IGF-I-stimulated phosphorylation of its receptor was a reflection of their affinity for
IGF, with the lowest affinity mutants having the least inhibitory effect.
These studies, thus, support the hypothesis that an Nterminal hydrophobic pocket is the primary site of high
affinity binding of IGF to IGF-binding protein-3. The
mutants provide a tool for future studies directed at IGFdependent and IGF-independent actions of IGF-binding
protein-3. (J Clin Endocrinol Metab 86: 4943– 4950, 2001)
the regional domains very little is known of the threedimensional structures of the IGFBPs, with the exception of
recent work by Kalus et al. (26). These investigators proposed
an N-terminal hydrophobic patch (amino acids V49, Y50,
P62, and K68 to L74) in IGFBP-5 critical for the binding of
IGF-II based on analysis of an IGFBP-5 fragment (A40 to I92)
by solution nuclear magnetic resonance (NMR) spectroscopy. Recent work after this study validated this hypothesis
in full-length IGFBP-5 and also the conserved region in
IGFBP-3 by substituting five amino acids in this pocket with
neutral or nonhydrophobic residues, thereby reducing IGF-I
affinity by more than 1000-fold (27).
Preliminary evidence from our laboratory of analysis of
small fragments of the IGFBP-3 N-terminus also led us to
believe it to be the primary site of IGF binding. Based on the
work by Kalus (26) as well as our preliminary IGFBP-3 deletional analysis, we targeted three amino acids within the
hydrophobic pocket (I56, L80, and L81) likely to have significant effects on IGF affinity. We compared both conserved
and nonconserved amino acid changes, mutating one, two,
and three residues at a time. Our goal was to disrupt the
binding site, but with the fewest number of residue alterations so that the least degree of disruption to the tertiary
structure of the protein would occur. Our data support the
existence of a high affinity hydrophobic IGF-binding site in
the N-terminus of IGFBP-3, probably conserved throughout
the IGFBPs, which, when disrupted, can affect the IGFdependent biological functions of IGFBP-3.
Materials and Methods
Materials
[125I]IGF-I and -II were provided by Diagnostics Systems Laboratories, Inc. (Webster, TX). IGF-I and -II were purchased from Austral
4943
4944
J Clin Endocrinol Metab, October 2001, 86(10):4943– 4950
Biologicals (Santa Clara, CA). Reagents for SDS-PAGE were purchased
from Bio-Rad Laboratories, Inc. (Hercules, CA). The BIAcore X instrument, sensor chip CM5 (research grade), HBS-EP buffer (0.01 m HEPES,
0.15 m NaCl, 3 mm EDTA, and 0.005% surfactant P20, pH 7.4), and the
amine coupling kit containing N-hydroxysuccinimide, N-ethyl-N⬘-(3diethylaminopropyl)carbodiimide, and ethanolamine hydrochloride
were purchased from BIAcore AB (Uppsala, Sweden). For use in the
BIAcore analysis, recombinant human IGF-I (rhIGF-I) was a gift from
Genentech, Inc. (South San Francisco, CA), and rhIGF-II was a gift from
Pharmacia Biotech (Uppsala, Sweden).
Cell culture
COS-7 cells were obtained from American Type Culture Collection
(Manassas, VA) and grown in DMEM with 10% FCS at 37 C in 5% CO2.
NIH-3T3 cells were a gift from Dr. C. T. Roberts, Jr. (Oregon Health
Sciences University, Portland, OR), and were grown in DMEM with 10%
FCS and 500 ␮g/ml G418 at 37 C in 5% CO2. All tissue culture media
and components were purchased from Life Technologies, Inc. (Grand
Island, NY), except FCS, which was obtained from HyClone Laboratories, Inc. (Logan, UT).
Generation, purification, and quantitation of recombinant
IGFBP-3 deletion fragments, wild-type and mutants
The cDNAs of IGFBP-3-(1– 46), -(1–75), -(1– 80), and -(1– 87) FLAG
epitope-tagged fragments were generated by PCR amplification from
the human IGFBP-3 cDNA and a C-terminal FLAG epitope sequence
(DYKDDDDK). After sequencing sense and antisense strands, the fragments were subcloned into pGEX4T-1 (Amersham Pharmacia Biotech,
Piscataway, NJ) and transformed into BL21DE3 Escherichia coli cells,
cultured overnight in Luria-Bertoni broth/ampicillin, and induced with
2 mm isopropylthio-␤-d-galactoside. Cell lysates were then harvested,
analyzed by SDS-PAGE staining with Coomassie blue and by Western
immunoblot with M2 anti-FLAG antibody (Eastman Kodak Co., Rochester, NY).
The preparation of expression vector pBSSK:IGFBP-3, containing a
full-length human IGFBP-3 cDNA with a C-terminal FLAG epitope
sequence (DYKDDDDK), by PCR amplification has been described previously (28). Single stranded phagemid DNA was generated from pBSSK:IGFBP-3FLAG, and mutations were introduced using synthetic degenerate oligonucleotides as substrates for antisense DNA synthesis.
The following complementary oligonucleotide was used to mutate I56
to V56 or G56 with an AclI site: agc gag ggc cag ccg tgc ggc rkc tac acc
gaa cgt tgt ggc tcc ggc ctt cgc (r ⫽ a or c; k ⫽ t or g). The following
complementary oligonucleotide was used to mutate L80 and L81 to V80
or G80 and/or V81 or G81 without a PstI site: gag gcg cga ccg ctg caa
gcg skg skg gac ggc cgc ggg ctc tgc gt (s ⫽ c or g; k ⫽ t or g). Sense and
antisense strands were sequenced, and preparations were subcloned
into pCMV6 for transient transfections of COS-7 cells, into pGEX4T-1 for
generation of E. coli glutathione-S-transferase (GST) fusion cell lysates
as described above, and into pFASTBAC1 (Life Technologies, Inc.,
Grand Island, NY) for baculovirus-generated proteins as described below. The full-length IGFBP-3 triple G mutant was constructed by subcloning a G80G81 fragment into the pCMV6:G56 mutant with BamHI
and BstAPI.
pFASTBAC1 preps were transformed into DH10Bac E. coli cells. The
amplified DNA was transfected into Sf9 insect cells (American Type
Culture Collection). HIGH-5 insect cells (Invitrogen, Carlsbad, CA) were
infected with P2 virus. The media were harvested on the third day and
incubated with an anti-M2 antibody affinity column overnight at 4 C.
The FLAG-tagged protein was then eluted by using FLAG peptide as
described previously (29). Eluted fractions were analyzed on 12% SDSPAGE under nonreducing conditions, followed by staining with Coomassie blue. Fractions were pooled and quantitated by two methods: 1)
comparison with known quantities of baculovirus IGFBP-3 by silver
staining (Bio-Rad Laboratories, Inc.), and 2) IGFBP-3 immunoradiometric assay (IRMA; Diagnostics Systems Laboratories, Inc.).
Transient transfections of COS-7 cells
Cells were plated in six-well plates, grown to 50 –70% confluence, and
transfected with a 1:2 ratio of cDNA and Mirrus Transit LT-1 (PanVera,
Buckway et al. • IGFBP-3 Mutants and Loss of IGF Affinity
Madison, WI). Medium was changed to serum-free medium after 16 h,
then collected 48 h later, and cellular debris was removed by centrifugation. COS-7 cells have some endogenous IGFBP-3 secretion, so the
conditioned medium was immunoprecipitated with M2 anti-FLAG antibody to pull out only the IGFBP-3 that had resulted from transient
transfections.
Western ligand blot analysis
Samples of E. coli-generated GST fusion cell lysates, purified baculovirus-expressed proteins, or conditioned medium at the concentrations indicated in the figure legends were mixed with Laemmli sample
buffer without a reducing agent and heated at 95 C for 5 min, then
electrophoresed on a 12% SDS gel and electroblotted onto nitrocellulose.
For dot blots, 2.5 ␮l sample were dotted directly onto the nitrocellulose
membrane. Membranes were then blocked for 1 h at 21 C in 1% BSA/
TBS-T (Tris-buffered saline-0.1% Tween 20), then incubated at 4 C overnight with 1 ⫻ 106 cpm [125I]IGF-I, [125I]IGF-II, or a mixture of the two.
The membranes were then washed, dried, and exposed to Biomax MS
film (Eastman Kodak Co.). The same membranes were then probed with
antibodies as described below. Bands were quantified using an image
analyzer (GS-700) equipped with MultiAnalyst version 1.0.2 software
(Bio-Rad Laboratories, Inc.).
Western immunoblot analysis
Samples of E. coli-generated GST fusion cell lysates, purified baculovirus-expressed proteins, whole cell lysates, or conditioned medium
at the concentrations indicated in the figure legends were mixed with
Laemmli sample buffer with or without a reducing agent and heated at
95 C for 5 min, then subjected to SDS-PAGE (8% or 12% gels) and
electroblotted onto nitrocellulose membranes. For dot blots, 2.5 ␮l sample were dotted directly onto the nitrocellulose membrane. The membranes were then blocked for 1 h at 21 C in 4% milk/TBS-T, followed by
overnight incubation at 4 C with anti-IGFBP-3 monoclonal antibody
(Diagnostics Systems Laboratories, Inc.), anti-GST polyclonal antibody
(Amersham Pharmacia Biotech, Piscataway, NJ), anti-PY20 monoclonal
antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), or antiinsulin receptor substrate-1 (anti-IRS-1) polyclonal antibody (Upstate
Biotechnology, Inc., Lake Placid, NY), all at 1:3000 dilutions. Membranes
were washed with TBS-T and incubated for 1 h at 21 C with a 1:3000
dilution of horseradish peroxidase-linked antirabbit or antimouse IgG
secondary antibodies (Amersham Pharmacia Biotech, Piscataway, NJ).
Proteins were detected by enhanced chemiluminescence reagents, according to the manufacturer’s protocol (NEN Life Science Products,
Boston, MA).
Affinity cross-linking
E. coli-generated GST fusion cell lysates of full-length IGFBP-3 or
mutants were incubated with 50,000 cpm [125I]IGF-I in the presence or
absence of unlabeled IGF-I (100 nm) overnight at 4 C and then crosslinked with 0.5 mm disuccinimidyl suberate for 15 min at 4 C. The
FIG. 1. Ligand dot blot of IGFBP-3 deletion fragments with [125I]IGFI and -II and corresponding immunoblot with anti-GST antibody. GST
fusion E. coli-generated cell lysate (2.5 ␮l) was dotted directly onto
nitrocellulose membrane and incubated overnight with 1 ⫻ 106 cpm
[125I]IGF-I and –II (upper panel). Blots representative of at least two
separate experiments were also probed with anti-GST antibody (lower
panel), showing the presence of approximately equal amounts of protein. Lane 1, IGFBP-3-(1– 46); lane 2, IGFBP-3-(1–75); lane 3, IGFBP3-(1– 80); lane 4, IGFBP-3-(1– 87); lane 5, full-length IGFBP-3; lane 6,
pGEX4T-1 vector alone.
Buckway et al. • IGFBP-3 Mutants and Loss of IGF Affinity
J Clin Endocrinol Metab, October 2001, 86(10):4943– 4950 4945
FIG. 2. Consensus sequences of the N-terminus of IGFBP-3 and IGFBP-5. The 10 amino acids predominantly forming the hydrophobic pocket
predicted to interact with IGFs based on solution NMR studies of IGFBP-5 are marked with black dots. The amino acids targeted for mutation
in our studies are marked with stars. Disulfide bonds are bracketed. Conserved residues are shaded.
samples were quenched with 100 mm Tris, pH 7.4, and were then
subjected to 12% SDS-PAGE and autoradiography on Biomax MS film
(Eastman Kodak Co.).
protein assay (Bio-Rad Laboratories, Inc.) and were separated on 8%
SDS-PAGE under reducing conditions. Anti-PY20 or anti-IRS-1 antibodies were used for Western immunoblot.
Solution binding assay
Results
Expression and analysis of E. coli-generated GST fusion
IGFBP-3 deletion fragments
Increasing amounts (0 –100 ng/ml) of purified baculovirus IGFBP-3
or mutant proteins in duplicate were incubated in 500 ␮l buffer (50 mm
Tris, pH 7.4, and 0.5% BSA) with 10,000 cpm [125I]IGF-I at 4 C overnight.
One milliliter of activated charcoal solution (0.5% activated charcoal, 0.2
mg/ml protamine sulfate, and 1% BSA in PBS) was added for 10 min
and then centrifuged for 10 min at 4000 rpm at 4 C to separate bound
and free IGF-I. A ␥-counter was used to measure the radioactivity of the
supernatants.
BIAcore analysis
Equal volumes of N-hydroxysuccinimide and N-ethyl-N⬘-(3-diethylaminopropyl)carbodiimide were mixed, and 35 ␮l of the mixture were
injected over the surface of the sensor chip to activate the carboxymethylated dextran. Eighty-three microliters of purified baculovirus-generated
wild-type or mutant IGFBP-3 solution (15 ␮g/ml in 10 mm sodium acetate,
pH 4.5) were injected over the activated surface, followed by 35 ␮l ethanolamine to deactivate remaining active carboxyl groups. The immobilization procedure was carried out at 25 C at a constant flow rate of 5 ␮l/min.
The first of the two flow cells of each chip was used as an in-line blank
reference cell. The carboxymethylated dextran in the reference cell was
activated and deactivated as described above, but without any ligand
bound. All experiments were carried out at 25 C at a constant flow rate of
10 ␮l/min HBS-EP buffer. Thirty-five microliters of the analyte (IGF-I or
IGF-II) diluted in HBS-EP buffer were injected over the immobilized wildtype or mutant IGFBP-3, followed by a 5-min period when buffer was
passed over the surface. Six concentrations of IGF-I and IGF-II were passed
over each chip (3.13, 6.25, 12.5, 25, 50, and 100 nm). All kinetic assays were
followed by an injection of 15 ␮l 0.1 m HCl to dissociate the remaining
ligand from the binding protein. The experiment was performed a total of
three times. BIAevaluation 3.0 software and SigmaStat were used for data
analysis, and a 1:1 mass transfer curve-fitting model was used in the
evaluation.
IGF-I-induced IGF-I receptor autophosphorylation assay
Confluent monolayers of NIH-3T3 IGF-I receptor cells were incubated in serum-free medium overnight. Purified baculovirus IGFBP-3
and mutant proteins (250 ng) were incubated with IGF-I (100 ng) in 1 ml
DMEM and 0.05% BSA for 30 min at 21 C and then added to the cells
for 10 min. The reaction was quenched with solubilization buffer [50 mm
Tris (pH 7.5), 2.5 mg/ml sodium deoxycholate, 150 mm sodium chloride,
1 mm sodium orthovanadate, 20 mm sodium fluoride, and 1% Nonidet
P-40]. Samples were normalized for protein concentration using a DC
Fragments of the N-terminus of IGFBP-3 were constructed
by PCR amplification and expressed as GST fusion cell lysates in E. coli. Amplification of the peptides was confirmed
by SDS-PAGE and Coomassie staining. Binding to [125I]IGF-I
was screened by dot-blot analysis (Fig. 1, upper panel) as
explained in Materials and Methods. Binding was strongly
detectable for both full-length IGFBP-3 and for IGFBP-3-(1–
87) GST fusion fragment. No binding was detectable for the
smaller fragments (1– 80, 1–75, and 1– 46).
Screening binding studies of IGFBP-3 mutant proteins
The studies with deletion fragments indicated that Nterminal residues were crucial to the binding of IGF, consistent with our prior studies and with the NMR solution
binding, which predicted a binding pocket in the N-terminus
of IGFBP-5. In IGFBP-3, therefore, we mutated three of the
amino acids (I56, L80, and L81) in this region in various
combinations either to a conserved residue, valine, or to a
nonconserved residue, glycine (Fig. 2). Initial binding studies
were performed by dot-blot analysis (Fig. 3A, upper panel) of
E. coli GST fusion proteins as described above and confirmed
by Western ligand blots on 12% SDS-PAGE (Fig. 3B, upper
panel). Both methods showed minimal change in binding
when the large, nonpolar, hydrophobic residue, valine, was
substituted for I56 or for L80L81. However, a clear reduction
in binding was observed with substitution of I56 or L80L81
by glycine, a small polar amino acid.
As predicted, the valine mutants showed relatively little
change in IGF affinity, and they were not further tested. The
glycine mutants, however, were also tested in an affinity
cross-linking study in which they were incubated with
[125I]IGF-I and cross-linked before gel electrophoresis. Again,
a clear reduction in binding was evident with the G56 mu-
4946
J Clin Endocrinol Metab, October 2001, 86(10):4943– 4950
Buckway et al. • IGFBP-3 Mutants and Loss of IGF Affinity
FIG. 4. Binding studies of immunoprecipitated conditioned medium
from COS-7 cells transiently transfected with full-length wild-type
and mutant IGFBP-3 cDNA. Western ligand blot (upper panel) of
conditioned medium collected 48 h after transfection of cells, immunoprecipitated with M2 anti-FLAG antibody to distinguish them from
endogenous IGFBP-3 of COS-7 cells, subjected to 12% SDS-PAGE,
and incubated overnight 4 C with 1 ⫻ 106 cpm [125I]IGF-I. The same
membrane probed with anti-IGFBP-3 antibody to assess the quantity
of protein loaded (lower panel). Lane 1, IGFBP-3; lane 2, G81 mutant;
lane 3, pCMV6 vector alone (vt); lane 4, G56 mutant; lane 5, G80
mutant; lane 6, G80G81 mutant; lane 7, G56G80G81 mutant; lane 8,
empty; lane 9, rhIGFBP-3. Blots shown are representative of at least
three different experiments.
FIG. 3. Binding studies of full-length mutant IGFBP-3 E. coligenerated GST-fusion cell lysates. A, Ligand dot blot (upper panel) of
2.5 ␮l cell lysate dotted directly onto nitrocellulose membrane and
incubated overnight with 1 ⫻ 106 cpm [125I]IGF-I and -II, and corresponding immunoblot with anti-GST antibody (lower panel). Lanes
from left to right, IGFBP-3, pGEX4T-1 vector, G56 mutant, V56
mutant, G80G81 mutant, and V80V81 mutant. B, Western ligand blot
(upper panel) using 20 ␮l protein/lane on a 12% SDS-PAGE gel under
nonreducing conditions, incubated overnight with 1 ⫻ 106 cpm
[125I]IGF-I and -II; the same membrane was probed with anti-GST
antibody (lower panel) for comparison of cell lysate concentrations.
Lanes are described in A. C, Affinity cross-linking of 20 ␮l protein with
50,000 cpm [125I]IGF-I after overnight incubation at 4 C, run on 12%
SDS-PAGE, and exposed to film. Each set of three lanes shows
duplicate cross-linking and competition by unlabeled IGF-I (lanes
marked with asterisk). Lanes 1 and 2, IGFBP-3; lane 3, IGFBP-3 with
competition by unlabeled IGF-I; lanes 4 and 5, pGEX4T-1 vector
alone; lane 6, pGEX4T-1 with unlabeled IGF-I; lanes 7 and 8, G56;
lane 9, G56 with unlabeled IGF-I; lanes 10 and 11, G80G81; lane 12,
G80G81 with unlabeled IGF-I. All blots shown are representative of
at least two separate experiments.
tant, whereas virtually no binding was detectable with the
G80G81 mutant (Fig. 3C). The binding of [125I]IGF-I to the
G56 mutant was specific, as it could be competed with unlabeled IGF-I.
To examine any differences between E. coli and mammalian expressed proteins, we tested conditioned medium collected from COS-7 cells that had been transiently transfected
with the cDNAs of our glycine mutants. The medium was
immunoprecipitated with the M2 anti-FLAG antibody before
analysis to remove any endogenous IGFBP-3 that may have
been produced by the COS-7 cells. On Western ligand blot,
reductions in binding were evident for all of the glycinesubstituted mutants, with G81 the least affected and with IGF
binding by G80G81 (double G) and G56G80G81 (triple G)
abolished (Fig. 4).
FIG. 5. Western ligand blot of purified baculovirus-generated, fulllength, human IGFBP-3, wild-type and mutants. Fifty nanograms of
purified protein were loaded and run on 12% SDS-PAGE, then incubated overnight with 1 ⫻ 106 cpm [125I]IGF-I (upper panel). A corresponding immunoblot was probed with anti-IGFBP-3 antibody
(lower panel). Lanes from left to right: IGFBP-3, G56 mutant, G80
mutant, G80G81 mutant (double G), and G56G80G81 mutant (triple
G). The graph shows densitometric analysis of bands representing the
mean percent binding, assuming wild-type IGFBP-3 to be 100%. At
least three separate experiments were performed, with error bars
representing ⫾1 SD. No binding was detectable for double G or triple
G, even when membranes were exposed to film up to 7 d.
Characterization of baculovirus-expressed IGFBP-3
mutant proteins
Based upon the results of the above-described screening
binding studies, we chose to produce the following four
FLAG-tagged, mutant human IGFBP-3 proteins in a baculovirus system: G56, G80, G80G81 (double G), and
G56G80G81 (triple G). The proteins were purified over an M2
FLAG antibody affinity column. The purity of the pooled
fractions was verified by silver staining of protein subjected
to SDS-PAGE and was quantitated relative to known quantities of baculovirus-generated human IGFBP-3. Quantitation was confirmed subsequently by IRMAs. The mutations
introduced did not interfere with the ability of the peptide to
be recognized by the anti-IGFBP-3 antibodies used in the
assay, and parallel curves of the mutants were generated
Buckway et al. • IGFBP-3 Mutants and Loss of IGF Affinity
J Clin Endocrinol Metab, October 2001, 86(10):4943– 4950 4947
Differences in affinity for IGF-I and -II were more apparent
by this method; specifically, the affinity for IGF-II was less
affected than that for IGF-I in mutants G56 and G80. However, the double G and the triple G mutants still showed little,
if any, binding by either ligand blot or solution binding assay
(Fig. 6).
BIAcore biosensor measurements confirmed the results
from the previous binding studies, providing kinetic affinity
data for the various IGFBP-3 mutants. Native and the four
glycine mutant IGFBP-3 proteins were each covalently
bound to a gold biosensor chip, and increasing concentrations of IGF-I or -II were used in the buffer flow. Kinetic
parameters are shown in Table 1. By this methodology, wildtype IGFBP-3 had a Kd of 0.79 ⫻ 10⫺9 for IGF-I and 0.69 ⫻
10⫺9 for IGF-II. The G56 mutant had a 1.7-fold lower affinity
for IGF-I, whereas the G80 mutant had a 4.3-fold lower affinity for IGF-I, with statistically significant increases in the
dissociation rates; both single mutants preserved normal
affinity for IGF-II. The double G mutant had only minimal
binding to IGF-I and -II, at least 3– 4 orders of magnitude
lower affinity. The triple G mutant had no detectable binding
of either IGF-I or IGF-II.
IGF-dependent actions of IGFBP-3 and mutants mirror
IGF affinity
FIG. 6. Specific binding of [125I]IGF-I and [125I]IGF-II to purified
baculovirus-generated IGFBP-3 compared with mutants in solution
binding assay. Zero to 100 ng/ml IGFBP-3, wild-type or mutants, were
incubated overnight at 4 C in 0.5 ml solution with 10,000 cpm
[125I]IGF-I (top graph) or [125I]IGF-II (bottom graph). Activated charcoal solution was then added to precipitate free IGF-I or -II, and the
radioactivity of supernatants was counted. Experiments were performed in duplicate at least two or three times. Graph represents the
mean percent specific binding, and error bars show ⫾ SD.
with each assay run, suggesting minimal disruption of the
tertiary structure and the epitopes recognized by the
antibodies.
When quantified by densitometry (Fig. 5, graph), bands
detected by Western ligand analysis (Fig. 5, upper panel)
showed a 60% reduction in binding of IGF-I for G56, a 70%
reduction for G80, and a reduction to background level for
the double G and triple G mutants. No binding was seen for
the double G or triple G mutants, even with exposures of up
to 1 wk. Similar levels of binding were seen when the ligand
used was IGF-II (data not shown). Verification of the quantity
of protein loaded (50 ng/lane) was made by immunoblot
with anti-IGFBP-3 monoclonal antibody (Fig. 5, lower panel).
Similar quantitation results for mutant proteins were observed on immunoblots using an anti-IGFBP-3 polyclonal
antibody produced in our laboratory (data not shown).
Solution binding assays were more sensitive than ligand
blots for determining subtle differences in binding of IGF-I
and -II by the single amino acid mutants (G56 and G80).
IGFBP-3 is known to inhibit IGF-I-stimulated phosphorylation of the type I IGF receptor (IGF-IR) when added to
culture medium (4). As the mutants have varying affinities
for IGF, we hypothesized they may well have different abilities to modulate this IGF-dependent action. Tyrosine phosphorylation of the ␤-subunit of the IGF-IR was seen when
COS-7 cells were treated with 100 ng/ml IGF-I, but was
blocked when IGF-I and 250 ng/ml wild-type IGFBP-3 were
preincubated and then used to treat the cells. Likewise, the
G56 mutant still bound IGF-I with sufficient affinity to inhibit
phosphorylation. Less inhibition of IGF-I receptor phosphorylation was seen with the other three mutants relative to the
degree of loss of IGF affinity; using the triple G mutant, no
significant inhibition was observed (Fig. 7).
Discussion
Several studies of fragments of IGFBP-3 have provided
evidence that IGF binds to both the N-terminus and the
C-terminus, but there are no definitive three-dimensional
structural analyses of the IGFBPs, including their interactions with the IGFs. The IGF affinity for the N-terminus is
clearly less than that of intact IGFBP-3, but is present, as
shown in our laboratory, both for 1– 87 and 1–97 fragments
(20, 30) and for the C-terminal 98 –264 IGFBP-3 fragment (23).
In general, most investigators agree that both the N-terminus
and the C-terminus of the IGFBPs are required to be in the
appropriate conformation, stabilized by disulfide bonds, for
high affinity IGF binding to occur.
Previous work by Kalus et al. (26) predicted a hydrophobic
patch on the N-terminus of IGFBP-5 (residues V49, Y50, P62,
and K68 to L74) to be the primary IGF-binding site, based
upon studies with solution NMR spectroscopy, using a fragment of IGFBP-5 and IGF-II as the ligand. This work was
validated by the finding of a full-length IGFBP-5 mutant with
4948
J Clin Endocrinol Metab, October 2001, 86(10):4943– 4950
Buckway et al. • IGFBP-3 Mutants and Loss of IGF Affinity
TABLE 1. BIAcore analysis: summary of kinetic data
Analyte
Ligand
Mean kon
(X105 1/ms)
95% CI kon
Mean koff
(X10⫺4 1/s)
95% CI koff
Ka
(X109 1/M
Kd
(nM)
IGFBP-3
IGF-I
IGF-II
6.81
7.91
1.25–12.37
4.62–11.20
5.36a
5.50
4.56– 6.16
5.13–5.87
1.27
1.44
0.79
0.69
G56
IGF-I
IGF-II
6.96
11.80
4.33–9.59
7.44–16.16
9.35a
3.76
8.81–9.89
2.25–5.27
0.75
3.15
1.34
0.32
G80
IGF-I
IGF-II
3.68
13.30
0.43– 6.93
2.9–23.7
8.92–16.08
0.65–7.05
0.29
3.45
3.40
0.29
GG
IGF-I, -II
NA
GGG
IGF-I, -II
NA
12.5a
3.85
Native and mutant IGFBP-3 were immobilized on the biosensor chips. Six concentrations of IGF-I and -II were passed over each chip. A 1:1
mass transfer curve-fitting model was used in the evaluation. The affinity measurements were performed in triplicate, and the mean kon and
koff with 95% confidence intervals (CI) were calculated. The Ka value was obtained by dividing kon with koff, and the Kd value by taking the
inverse of Ka. IGF-I and IGF-II only minimally bound to the double G mutant, with at least 3– 4 orders of magnitude lower affinity. The IGFs
did not bind to the triple G mutant at all. Therefore, it was not possible to calculate affinity constants for the double G or the triple G mutants
(NA, not applicable). The off-ratesa of IGF-I bound to the native IGFBP-3, G56, and G80 chips were the only significant changes, as calculated
by one-way ANOVA performed in the SigmaStat program. NA, Not applicable.
FIG. 7. Ability of IGFBP-3 to inhibit IGF-I-induced phosphorylation
of its receptor mirrors the affinity of IGFBP-3 wild-type and mutants.
A, Confluent monolayers of NIH-3T3 fibroblasts were serum-starved
overnight and treated for 10 min with 100 ng/ml IGF-I alone or after
preincubation with 250 ng/ml IGFBP-3 wild-type or mutant protein
for 30 min at 21 C. Cell lysates were collected, and equal amounts of
protein were run under reducing conditions on 8% SDS-PAGE. Phosphorylation of the ␤-subunit of the type I IGF-I receptor was probed
with anti-PY20 antibody (upper panel), and equal protein loading was
confirmed by immunoblot for anti-IRS-1 antibody (lower panel). Lane
1, No treatment; lane 2, IGFBP-3 alone; lane 3, IGF-I alone; lane 4,
IGFBP-3 and IGF-I; lane 5, G56 and IGF-I; lane 6, G80 and IGF-I;
lane 7, double G and IGF-I; lane 8, triple G and IGF-I. B, The graph
represents the mean densitometric values of three separate experiments ⫾1 SD of the percent IGF-I-stimulated phosphorylation of the
␤-subunit of the type I IGF-R, quantified from the immunoblots,
assuming that IGF-I alone stimulates 100%. The affinity of the mutants for IGF is reflected in the degree of inhibition of IGF-stimulated
phosphorylation of the receptor.
five amino acids changed within this pocket and with reduced affinity for IGF (27); similar results were shown for a
full-length IGFBP-3 mutant with the same five residues mutated. Our studies have more definitively defined the crucial
amino acids and substantiate the theory of a primary Nterminal IGF-binding site within the IGFBPs, but do not
address the necessity of the C-terminus. Based upon Nterminal conservation among the six high affinity binding
proteins, we chose to examine three of the amino acids of
IGFBP-3 (I56, L80, and L81) corresponding to V49, L73, and
L74 within the hydrophobic patch of IGFBP-5 proposed by
Kalus et al. (26).
The choice of these mutations was based on several factors.
First, the dot-blot screen of the smaller N-terminal fragments
showed binding to the IGFBP-3-(1– 87) fragment, but not to
the IGFBP-3-(1– 80) fragment. Deletion of residues 81– 87
thus appeared to disrupt a critical region for IGF binding,
and amino acid 81 is one of the proposed residues within the
hydrophobic pocket. Second, several other reports suggest
that this region of the N-terminus has significance. For example, Hashimoto et al. (19) found amino acids E52 to A92
of IGFBP-3 important for binding to IGF-II. In IGFBP4, L72
to S91 were essential for binding in studies by Qin et al. (31),
whereas residues C205 to V214 were facilitators of binding
rather than primary sites. Third, on the IGFBP-5 fragment
described by Kalus et al. (26), side-chains extend into solution
from residues V49, L70, and L74, creating a hydrophobic
surface corresponding to I56, L77, and L81 in IGFBP-3. In
addition, IGFBP-5 residues V49, L70, L73, and L74 had several specific intermolecular interactions with IGF-II by NMR
spectra. These studies provided a rationale for speculation
concerning which amino acids within the N-terminus have
singularly or in combination the greatest effect on IGF affinity without altering the disulfide bonds of these cysteinerich binding proteins.
Our initial screening studies support the importance of
hydrophobic residues for IGF affinity. Valine is a large nonpolar amino acid, as are isoleucine and leucine. When valine
was substituted for either isoleucine or leucine, it did not
dramatically affect binding. However, when glycine, a small
polar residue, was used instead of valine, it produced very
Buckway et al. • IGFBP-3 Mutants and Loss of IGF Affinity
different results, with more marked reductions in binding,
suggesting that a change from hydrophobic to nonhydrophobic residues was more relevant than the amino acid itself.
Binding studies using purified baculovirus-generated protein showed that the substitution of glycine for leucine at
position 80 led to the single greatest reduction in affinity and,
furthermore, only two mutations (double G mutant) were
necessary to reduce binding markedly to both IGF-I and -II
by Western ligand blot. On more sensitive solution binding
assays, binding was still not detected for the double G or the
triple G mutant. BIAcore analysis further validated these
findings and, more specifically, provided kinetics data supporting a loss of affinity by the mutants. The lower affinities
of the G56 and G80 mutants for IGF-I were statistically significant for increased rates of dissociation, indicating that
although IGF-I is able to bind these mutants, the complexes
are possibly not as stable as wild-type IGFBP-3. Interestingly,
IGF-II binding was not affected for the single mutations.
However, for both IGF-I and -II there was absolutely no
binding detectable for the triple G mutant and only very
minimal binding found for the double G mutant.
In the absence of x-ray crystallography or two-dimensional NMR spectroscopy, it is impossible to exclude the
possibility that the observed loss of binding may be secondary to an alteration of the tertiary structure of the protein.
However, there are several reasons to believe the disruption
of structure to be minimal. Throughout our studies the mutant proteins were easily detectable on immunoblot by both
a rabbit anti-IGFBP-3 polyclonal antibody and a monoclonal
anti-IGFBP-3 antibody, suggesting that no disruption of the
epitopes recognized by these antibodies occurred. Similarly,
the quantity of the mutants was accurately measured by an
IRMA that used a goat anti-IGFBP-3 polyclonal antibody,
and with each assay run, curves of dilutions of mutants and
native IGFBP-3 remained parallel. Furthermore, the IGFBP-3
and -5 mutants constructed by Imai et al. (27) that substituted
five amino acids with neutral or nonhydrophobic ones both
had similar decreases in affinity. They were able to show that
IGF-I could not inhibit proteolysis in the IGFBP-5 mutant,
whereas it did so in the native form, suggesting that the
five-amino acid change did not affect the proteolytic cleavage
sites, for which accurate conformation of the protein would
be crucial.
Functionally, our results also support the conclusion that
IGF-dependent actions of IGFBP-3 are via the sequestration
of IGF, thereby inhibiting stimulation of the receptor. This
appears to be primarily a function of the N-terminus, rather
than the mid region or C-terminus, as previously shown
using proteolytic fragments (23). Moreover, our data relate
the degree of affinity for IGF with the degree of inhibition of
phosphorylation. The lower the affinity a mutant IGFBP-3
has for IGF, the more IGF is able to access the receptor and
activate phosphorylation. Very recent studies (32) have suggested that an IGF-independent pathway may also exist in
MCF-7 breast cancer cells in which IGFBP-3 inhibits type I
IGF-R phosphorylation. Increasing amounts of IGFBP-3, -1,
and -5 were used to treat the cells, followed by stimulation
with wild-type and mutant forms of IGF-I that have less
affinity for the IGFBPs. Only IGFBP-3 inhibited phosphorylation of the receptor by all types of IGF-I tested. In the
J Clin Endocrinol Metab, October 2001, 86(10):4943– 4950 4949
future similar studies using mutant IGFBP-3 should prove
insightful regarding mechanisms and function for both IGFdependent actions and IGF-independent actions of IGFBP-3.
In conclusion, these data serve to clarify the importance of
three amino acids within the N-terminus of IGFBP-3 for IGF
affinity and delineate the individual and additive effects of
these residues. These results further support the existence of
an N-terminal hydrophobic patch as the primary binding site
of IGF to IGFBP-3 and suggest that this is similar in the other
high affinity binding proteins, given the significant conservation of amino acids within this region.
Acknowledgments
We acknowledge the assistance of Donna Graham and Allison Watts
in producing the purified baculovirus proteins, Katherine Pratt for performing the IGFBP-3 IRMAs, Eric Kofoed for technical assistance with
the phosphorylation studies, Christine Carlsson-Skwirut for assistance
with the BIAcore studies, and Dr. Vivian Hwa for helpful discussion.
Received February 5, 2001. Accepted July 2, 2001.
Address all correspondence and requests for reprints to: Caroline K.
Buckway, M.D., Department of Pediatrics, NRC-5, Oregon Health Sciences University, 3181 SW Sam Jackson Park Road, Portland, Oregon
97201. E-mail: [email protected].
This work was supported by NIH grants CA-58110 and DK-51513 (to
R.G.R.) and 5T32-HD-07497 (to C.K.B.), Department of Defense Grants
17-96-1-6304 and 17-97-1-7204 (to Y.O.), and grants from the Swedish
Medical Research Council (no. 11364), the Karolinska Institute Research
Foundation, the Swedish Society of Medicine, the Wera Ekström’s Foundation, the Swedish Freemason’s Foundation, the Märta and Gunnar V.
Philipson’s Foundation, the Milk Drop Foundation in Stockholm, and
HRH Crown Princess Lovisa’s Society of Pediatric Health Care (to P.B.).
References
1. Jones JI, Clemmons DR 1995 Insulin-like growth factors and their binding
proteins: biological actions. Endocr Rev 16:3–34
2. Stewart CE, Rotwein P 1996 Growth, differentiation, and survival: multiple
physiological functions for insulin-like growth factors. Physiol Rev 76:
1005–1026
3. Rajaram S, Baylink DJ, Mohan S 1997 Insulin-like growth factor-binding
proteins in serum and other biological fluids: regulation and functions. Endocr
Rev 18:801– 831
4. Clemmons DR 1997 Insulin-like growth factor binding proteins and their role
in controlling IGF actions. Cytokine Growth Factor Rev 8:45– 62
5. Oh Y, Muller HL, Lamson G, Rosenfeld RG 1993 Insulin-like growth factor
(IGF)-independent action of IGF-binding protein-3 in Hs578T human breast
cancer cells. Cell surface binding and growth inhibition. J Biol Chem 268:
14964 –14971
6. Cohen P, Lamson G, Okajima T, Rosenfeld RG 1993 Transfection of the
human IGFBP-3 gene into BALB/c fibroblasts: a model for the cellular functions of IGFBPs. Growth Regul 3:23–26
7. Valentinis B, Bhala A, DeAngelis T, Baserga R, Cohen P 1995 The human
insulin-like growth factor (IGF) binding protein-3 inhibits the growth of fibroblasts with a targeted disruption of the IGF-I receptor gene. Mol Endocrinol
9:361–367
8. Rajah R, Valentinis B, Cohen P 1997 Insulin-like growth factor (IGF)-binding
protein-3 induces apoptosis and mediates the effects of transforming growth
factor-␤1 on programmed cell death through a p53- and IGF-independent
mechanism. J Biol Chem 272:12181–12188
9. Nickerson T, Huynh H, Pollak M 1997 Insulin-like growth factor binding
protein-3 induces apoptosis in MCF7 breast cancer cells [published erratum
appears in Biochem Biophys Res Commun 1997 Nov 7;240(1):246]. Biochem
Biophys Res Commun 237:690 – 693
10. Oh Y 1998 IGF-independent regulation of breast cancer growth by IGF binding
proteins. Breast Cancer Res Treat 47:283–293
11. Butt AJ, Firth SM, Baxter RC 1999 The IGF axis and programmed cell death.
Immunol Cell Biol 77:256 –262
12. Yamanaka Y, Fowlkes JL, Wilson EM, Rosenfeld RG, Oh Y 1999 Characterization of insulin-like growth factor binding protein-3 (IGFBP- 3) binding to human
breast cancer cells: kinetics of IGFBP-3 binding and identification of receptor
binding domain on the IGFBP-3 molecule. Endocrinology 140:1319 –1328
13. Hwa V, Oh Y, Rosenfeld RG 1999 The insulin-like growth factor-binding
protein (IGFBP) superfamily. Endocr Rev 20:761–787
4950
J Clin Endocrinol Metab, October 2001, 86(10):4943– 4950
14. Zapf J, Kiefer M, Merryweather J, et al. 1990 Isolation from adult human
serum of four insulin-like growth factor (IGF) binding proteins and molecular
cloning of one of them that is increased by IGF I administration and in
extrapancreatic tumor hypoglycemia. J Biol Chem 265:14892–14898
15. Andress DL, Loop SM, Zapf J, Kiefer MC 1993 Carboxy-truncated insulin-like
growth factor binding protein-5 stimulates mitogenesis in osteoblast-like cells.
Biochem Biophys Res Commun 195:25–30
16. Chernausek SD, Smith CE, Duffin KL, Busby WH, Wright G, Clemmons DR
1995 Proteolytic cleavage of insulin-like growth factor binding protein 4 (IGFBP-4).
Localization of cleavage site to non-homologous region of native IGFBP-4. J Biol
Chem 270:11377–11382
17. Fowlkes JL, Serra DM, Rosenberg CK, Thrailkill KM 1995 Insulin-like
growth factor (IGF)-binding protein-3 (IGFBP-3) functions as an IGF-reversible
inhibitor of IGFBP-4 proteolysis. J Biol Chem 270:27481–27488
18. Durham SK, Mohan S, Liu F, et al. 1997 Bioactivity of a 29-kilodalton insulinlike growth factor binding protein-3 fragment present in excess in chronic renal
failure serum. Pediatr Res 42:335–341
19. Hashimoto R, Ono M, Fujiwara H, et al. 1997 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. J Biol Chem 272:
27936 –27942
20. Vorwerk P, Yamanaka Y, Spagnoli A, Oh Y, Rosenfeld RG 1998 Insulin and
IGF binding by IGFBP-3 fragments derived from proteolysis, baculovirus
expression and normal human urine. J Clin Endocrinol Metab 83:1392–1395
21. Spencer EM, Chan K 1995 A 3-dimensional model for the insulin-like growth
factor binding proteins (IGFBPs); supporting evidence using the structural
determinants of the IGF binding site on IGFBP-3. Prog Growth Factor Res
6:209 –214
22. Ho PJ, Baxter RC 1997 Characterization of truncated insulin-like growth
factor-binding protein-2 in human milk. Endocrinology 138:3811–3818
23. Devi GR, Yang DH, Rosenfeld RG, Oh Y 2000 Differential effects of insulinlike growth factor (IGF)-binding protein-3 and its proteolytic fragments on
Buckway et al. • IGFBP-3 Mutants and Loss of IGF Affinity
24.
25.
26.
27.
28.
29.
30.
31.
32.
ligand binding, cell surface association, and IGF-I receptor signaling. Endocrinology 141:4171– 4179
Cooke RM, Harvey TS, Campbell ID 1991 Solution structure of human
insulin-like growth factor 1: a nuclear magnetic resonance and restrained
molecular dynamics study. Biochemistry 30:5484 –5491
Terasawa H, Kohda D, Hatanaka H, et al. 1994 Solution structure of human
insulin-like growth factor II; recognition sites for receptors and binding proteins. EMBO J 13:5590 –7
Kalus W, Zweckstetter M, Renner C, et al. 1998 Structure of the IGF-binding
domain of the insulin-like growth factor- binding protein-5 (IGFBP-5): implications for IGF and IGF-I receptor interactions. EMBO J 17:6558 –72
Imai Y, Moralez A, Andag U, Clarke JB, Busby Jr WH, Clemmons DR 2000
Substitutions for hydrophobic amino acids in the N-terminal domains of
IGFBP-3 and -5 markedly reduce IGF-I binding and alter their biologic
actions. J Biol Chem 275:18188 –18194
Vorwerk P, Oh Y, Lee PD, Khare A, Rosenfeld RG 1997 Synthesis of IGFBP-3
fragments in a baculovirus system and characterization of monoclonal antiIGFBP-3 antibodies. J Clin Endocrinol Metab 82:2368 –2370
Oh Y, Nagalla SR, Yamanaka Y, Kim HS, Wilson E, Rosenfeld RG 1996
Synthesis and characterization of insulin-like growth factor-binding protein
(IGFBP)-7. Recombinant human mac25 protein specifically binds IGF-I and -II.
J Biol Chem 271:30322–30325
Yamanaka Y, Wilson EM, Rosenfeld RG, Oh Y 1997 Inhibition of insulin
receptor activation by insulin-like growth factor binding proteins. J Biol Chem
272:30729 –30734
Qin X, Strong DD, Baylink DJ, Mohan S 1998 Structure-function analysis of
the human insulin-like growth factor binding protein-4. J Biol Chem 273:
23509 –23516
Ricort JM, Binoux M 2001 Insulin-like growth factor (IGF) binding protein-3
inhibits type 1 IGF receptor activation independently of its IGF binding affinity. Endocrinology 142:108 –113
JOURNÉES INTERNATIONALES D’ENDOCRINOLOGIE CLINIQUE
Henri-Pierre Klotz
Société Française d’Endocrinologie
First Announcement
The 45th Journées Internationales d’Endocrinologie Clinique will be held in Paris on May 23–24, 2002, and
will be devoted to: “A decade of advances in thyroidology.”
The program will include 20 state-of-the-art lectures and a limited number of selected free communications
for oral or poster presentation. The deadline for submission of abstracts is January 15, 2002.
For more information, contact Dr. G. Copinschi, Laboratory of Experimental Medicine, Brussels Free University-CP 618, 808 Route de Lennik, B-1070 Brussels, Belgium. Fax: 32-2-5556239; E-mail: [email protected].