Impaired Blockade of Insulin-Like Growth Factor I ... - Semantic Scholar

0013-7227/02/$15.00/0 Printed in U.S.A.

Endocrinology 143(5):1669 –1676 Copyright © 2002 by The Endocrine Society

Impaired Blockade of Insulin-Like Growth Factor I (IGF-I)-Induced Hypoglycemia by IGF Binding Protein-3 Analog with Reduced Ternary Complex-Forming Ability SUE M. FIRTH, FIONA MCDOUGALL, ANDREW J. MCLACHLAN,

AND

ROBERT C. BAXTER

Kolling Institute of Medical Research, University of Sydney, Royal North Shore Hospital (S.M.F., F.M., R.C.B.), St. Leonards, New South Wales 2065, Australia; and Faculty of Pharmacy, University of Sydney (A.J.M.), New South Wales 2006, Australia The hypoglycemic potential of circulating IGFs is thought to be regulated through the formation of ternary complexes consisting of an IGF, either IGF binding protein-3 (IGFBP-3) or IGFBP-5, and the acid-labile subunit. These high molecular weight complexes are confined to the circulation and represent a reservoir of IGF with a prolonged half-life. In this study, we show that hypoglycemia, induced by a bolus injection of recombinant human IGF-I into rats, can be blocked by coadministering equimolar concentrations of either recombinant glycosylated IGFBP-3 or nonglycosylated IGFBP-3 (IGFBP3NG). In contrast, an IGFBP-3 mutant with reduced acidlabile subunit affinity (IGFBP-3MUT) only partially blocked

I

GF-I AND IGF-II are potent mitogens that circulate bound to one of the members of a family of high-affinity binding proteins, the IGF-binding proteins (IGFBPs). However, only a relatively small proportion of the IGFs is found in these binary complexes (1). The majority of the IGFs are sequestered into 150-kDa ternary complexes with IGFBP-3 and the liver-derived glycoprotein, the acid-labile subunit (ALS) (2). Recently, it was discovered that IGFBP-5 is also capable of forming similar ternary complexes with IGF and ALS (3). It is thought that the molecular size of the ternary complexes precludes the complexes, and hence IGFs, from leaving the vasculature. This results in suppression of insulin-like activity and thus the hypoglycemic potential of the IGFs. Previous in vivo studies in rats have shown that although endogenous ALS circulates in excess of the other components of the ternary complex, it is the determining factor of the stability of the ternary complex due to its slow clearance rate (4). We have previously shown that the major ALS-binding site on IGFBP-3 resides within a carboxyl-terminal region rich in basic amino acid residues. By in vitro assays we have determined that the recombinant mutant, IGFBP-3MUT, which has the K228GRKR3 MDGEA mutation, binds IGF-I without a significant loss of affinity, but has 10-fold reduced affinity for ALS (5). Although a third of the molecular mass of IGFBP-3 is contributed by carbohydrate moieties, we have

Abbreviations: ALS, Acid-labile subunit; AUC, area under the curve; IGFBP, IGF binding protein; IGFBP-3MUT, IGF-binding protein-3 mutant with reduced acid-labile subunit affinity; IGFBP-3NG, nonglycosylated IGF-binding protein-3.

the IGF-I hypoglycemic effect. IGFBP-3 and IGFBP-3NG significantly enhanced IGF-I retention in the circulation, whereas IGFBP-3MUT had a smaller effect. IGFBP-3MUT clearance was more rapid than that of the other IGFBP-3 forms, and the retention of all IGFBP-3 forms was greatly enhanced by coadministration of IGF-I. Characterization of the molecular mass distribution of the IGFBP-3 analogs indicated that 60% of IGFBP-3 and IGFBP-3NG was initially found in ternary complexes compared with 30% of IGFBP-3MUT. These data confirm the hypothesis that regulation of IGF-I bioactivity in vivo by IGFBP-3 depends on its ability to form ternary complexes. (Endocrinology 143: 1669 –1676, 2002)

shown by in vitro assays that the carbohydrates are not essential components of the ALS-binding site on IGFBP-3 (6). However, there is a 2- to 3-fold increase in cell binding by recombinant nonglycosylated IGFBP-3 (IGFBP-3NG) compared with that by the glycosylated form (7). Therefore, the extent of glycosylation on IGFBP-3 may influence the distribution of IGFBP-3 between the ternary complexes in circulation and the cell surfaces. However, IGFBP-3MUT, which has reduced affinity for ALS, is unable to bind to cell surfaces, thus implicating the same carboxyl-terminal basic motif in the IGFBP-3 cell-binding function (5). The aim of this study was to compare the relative abilities of IGFBP-3, IGFBP-3MUT, and IGFBP-3NG to form ternary complexes and to prevent hypoglycemia induced by coadministering IGF-I in vivo. The disappearance profiles of IGFBP-3 and its analogs when administered with and without IGF-I to rats are also examined. Materials and Methods Reagents Recombinant human IGFBP-3, IGFBP-3(K228M/G229D/R230G/ K231E/R232A) (designated as IGFBP-3MUT), and nonglycosylated IGFBP-3(N89A/N109A/N172A) (designated IGFBP-3NG) were expressed in human 911 retinoblastoma cells using an adenovirus-mediated system and were purified by IGF-I affinity chromatography followed by reverse phase HPLC, as described previously (7). Recombinant proteins were quantitated by RIA specific for human IGFBP-3 (8), and equimolar dilutions prepared on this basis were confirmed as equimolar by quantitative amino acid analysis. Recombinant IGF-I was donated by Genentech, Inc. (San Francisco, CA). ALS was purified from human serum as described previously (9). Radioiodinated IGF-I and ALS were prepared according to published procedures (8, 9).

1669

1670

Endocrinology, May 2002, 143(5):1669 –1676

Firth et al. • Clearance of IGFBP-3 Analogs in Rats

Binding assays The affinity of IGF-I and ALS binding to the recombinant IGFBP-3/ IGF-I binary complexes was measured essentially as described previously (10). Scatchard analysis was performed as described previously (10).

volume of distribution at steady state were estimated for IGF-I and IGFBP-3 after different treatments using the coefficients (Ai) and exponents of the triexponential model (12). The distribution half-lives (t1/2,␣1, t1/2,␣2, and t1/2,␣3) were calculated from the triexponential model exponents (␣1, ␣2, and ␣3) using the general equation ln2/␣i.

In vivo experiments

Results

These experiments were carried out with the approval of the animal care and ethics committee, Royal North Shore Hospital. Cannulas were inserted into the jugular veins of 6-wk-old male Wistar rats 48 h before experiments. The in-dwelling cannulas were filled with 10 U/ml heparin-saline to maintain patency. After an overnight fast, two baseline blood samples (200 ␮l) were drawn before an iv bolus of approximately 13 nmol/kg IGF-I, IGFBP-3, or their combination (Table 1) was administered through the indwelling cannula into the conscious animals. The peptides were preincubated in 0.15 m NaCl at 22 C for 2 h before injection. Blood samples of 200 ␮l were drawn at 2, 5, 10, 20, 30, 45, 60, 90, and 120 min after injection and collected in tubes containing 2 ␮l 500 U/ml heparin. The plasma was separated immediately and stored at ⫺20 C for later assays. The blood cells were resuspended in 0.15 m NaCl and reinfused after the next blood sampling.

Assays Glucose levels were determined by an Advantage blood glucose monitor (Roche, Castle Hill, Australia). Human IGFBP-3 was measured by a human-specific IGFBP-3 RIA, as described previously (8). Human IGF-I was measured using acid-ethanol-extracted plasma in an IGF-I RIA using a human-specific monoclonal antibody (11), except that 125Ilabeled des(1–3)-IGF-I was used instead of 125I-labeled IGF-I.

Superose chromatography The molecular mass distribution of human IGFBP-3 in 25 ␮l rat plasma was determined by size fractionation on a Superose-12 column (Amersham Pharmacia Biotech, Uppsala, Sweden) in 50 mm sodium phosphate, 0.15 m NaCl, and 0.02% sodium azide (pH 6.5). The flow rate was 30 ml/h, and 0.5-ml fractions were collected. Aliquots of 100 ␮l from fractions 20 –31 from each run were assayed in duplicate for human IGFBP-3.

Data analysis Comparisons among treatment groups were analyzed by ANOVA, using repeated measures where appropriate, with post hoc testing by Fisher’s protected least significant difference test (StatView 5.0, Abacus Concepts, Inc., Berkeley, CA). A significant difference was defined as P ⬍ 0.05. The IGF-I and IGFBP-3 clearance data from the in vivo experiments were analyzed by two methods. First, to compare profiles in a model-independent way, the area under IGF-I and IGFBP-3 concentration-time curves was determined for each animal and analyzed by ANOVA, followed by Fisher’s protected least significant difference test. Second, data were fitted to a triexponential equation using Scientist software version 3.0 (Micromath Scientific Software, Salt Lake City, UT). The area under the concentration-time curve (AUC), clearance, and the

Using recombinant IGFBP-3 and IGFBP-3MUT expressed in Chinese hamster ovary cells, we have previously shown that IGFBP-3MUT has slightly, although not significantly, reduced affinity for IGF-I compared with IGFBP-3 (5). The affinity of IGFBP-3NG for IGF-I, however, was only assessed by ligand blot analysis (6). To obtain the relative affinities of adenoviral-expressed IGFBP-3 and its analogs for IGF-I, we measured the displacement of [125I]IGF-I from binary complexes formed in the presence of each IGFBP-3, by increasing amounts of unlabeled IGF-I in a solution binding assay (Fig. 1A). Data from these displacement curves were analyzed by Scatchard plots (data not shown), and the derived association constants for IGFBP-3 [(2.1 ⫾ 0.8) ⫻ 1011 liters/mol], IGFBP3MUT [1.8 ⫾ 1.0) ⫻ 1011 liters/mol], and IGFBP-3NG [(0.9 ⫾ 0.4) ⫻ 1011 liters/mol] were not significantly different from each other (mean ⫾ sd; n ⫽ 3). To compare the relative affinities of ALS for the IGFBP-3 analogs, we determined the displacement of [125I]ALS from the ternary complexes, formed in the presence of IGF-I and the IGFBP-3 analogs, by increasing concentrations of unlabeled ALS (data not shown). Association constants obtained from Scatchard plots (Fig. 1B) derived from these displacement assays indicated that IGFBP-3 [(4.0 ⫾ 0.6) ⫻ 1010 liters/mol] and IGFBP-3NG [(3.6 ⫾ 0.5) ⫻ 1010 liters/mol] had similar affinities for ALS. In contrast, IGFBP-3MUT [(1.5 ⫾ 0.2) ⫻ 109 liters/mol] had a 95% reduction in ALS affinity compared with IGFBP-3 (P ⫽ 0.001), indicating that this analog is very poor at forming ternary complexes. The effect of administering an iv bolus of 100 ␮g/kg human IGF-I, with or without various human IGFBP-3 analogs, or various IGFBP-3 analogs alone on fasting plasma glucose levels are shown in Fig. 2. The mean fed and fasted glucose levels were 7.8 ⫾ 0.1 and 4.8 ⫾ 0.1 mmol/liter, respectively. The administration of IGF-I alone to fasted rats caused a 48% decrease in plasma glucose levels within 10 min (Fig. 2A). In contrast, when coadministered with IGF-I, IGFBP-3 or IGFBP-3NG fully blocked the hypoglycemic effect of IGF-I. Coadministration of IGF-I and the IGFBP-3 analog with reduced ALS binding ability, IGFBP-3MUT, only partially pre-

TABLE 1. Rat treatments and weights Treatment groupa

IGF-I IGF-I ⫹ IGFBP-3 IGFBP-3 IGF-I ⫹ IGFBP-3MUTb IGFBP-3MUT IGF-I ⫹ IGFBP-3NGc IGFBP-3NG a

Dose (␮g/kg) IGF-I

IGFBP-3

No. of rats

Presurgery wt (g)

Postfast wt (g)

100 100 0 100 0 100 0

0 600 600 600 600 600 600

4 4 4 5 4 5 3

266.7 ⫾ 1.5 258.5 ⫾ 6.4 267.1 ⫾ 6.3 270.9 ⫾ 6.5 251.0 ⫾ 12.4 248.6 ⫾ 5.0 243.4 ⫾ 3.8

245.8 ⫾ 2.1 240.4 ⫾ 5.0 245.5 ⫾ 6.8 248.1 ⫾ 6.7 232.3 ⫾ 12.9 230.8 ⫾ 6.5 219.9 ⫾ 7.0

All proteins were prepared in 0.15 M NaCl preincubated at 22 C for at least 2 h before administration. IGFBP-3MUT has the following mutation in the carboxyl-terminal region, K228M/G229D/R230G/K231E/R232A. c IGFBP-3NG has all three N-glycosylation sites mutated, N89A/N109A/ N172A. b


FIG. 1. A, Competition for the binding of [125I]IGF-I to 0.25 ng of each IGFBP-3 analog by increasing amounts of unlabeled IGF-I. B/Bo, The ratio of [125I]IGF-I bound to IGFBP-3 in the presence of unlabeled IGF-I to that bound in the absence of unlabeled IGF-I. Data shown are the mean ⫾ SE of three independent measurements. B, Scatchard plots of ALS binding to 50 ng/tube IGF-I in the presence of 0.25 ng/tube IGFBP-3 (left panel), 25 ng/tube IGFBP-3MUT (middle panel), or 0.25 ng/tube IGFBP-3NG (right panel). The plots shown are representatives of three independent measurements for each analog.

vented the fall in plasma glucose levels (⬃13% at 10 min). Although there was a clear trend toward alleviating the hypoglycemia, the effect of coinjecting IGF-I and IGFBP3MUT on the plasma glucose level was not significantly different from that of administering IGF-I alone when analyzed by repeated measures ANOVA (P ⫽ 0.14), possibly due to the small sample size. In contrast, coinjecting IGF-I with either IGFBP-3 or IGFBP-3NG was significantly different from that of administering IGF-I alone (P ⱕ 0.02). The coinjection of IGF-I with either IGFBP-3 or IGFBP3NG not only reversed the hypoglycemic effect, but it appeared to increase the plasma glucose levels above baseline in the early time points (Fig. 2A). The same phenomenon was observed when the IGFBP-3 analogs were injected alone (Fig. 2B), where plasma glucose levels were increased by approximately 5–16% of baseline values at 10 –20 min postadministration. However, the effect of IGF-I coinfusion on the IGFBP-3-induced increase in blood glucose levels was not significant for either IGFBP-3 (P ⬎ 0.05) or IGFBP-3NG (P ⬎ 0.5). In contrast, there was a significant effect on blood glucose levels when IGF-I was coinjected with IGFBP-3MUT compared with IGFBP-3MUT alone (P ⬍ 0.05). Figure 3 shows the disappearance of an iv bolus of human IGF-I from the rat circulation when administered alone or in conjunction with various human IGFBP-3 analogs. Only human IGF-I was detected in the species-specific assay used. At the dilutions assayed in this study, the limit of detection in the IGF-I RIA was 35 ng/ml. IGF-I disappearance curves were initially examined in a model-independent manner by calculating the AUC for each animal over the experimental period from 0 –120 min. When compared by one-factor ANOVA, the mean AUC (⫾sd) for IGF-I administered with

Endocrinology, May 2002, 143(5):1669 –1676 1671

IGFBP-3MUT (12.7 ⫾ 2.8 ␮g/ml䡠min) was not different from that of IGF-I administered alone (8.2 ⫾ 4.6 ␮g/ml䡠min; P ⫽ 0.55), whereas the mean AUC estimates for IGF-I administered with IGFBP-3 (25.4 ⫾ 7.1 ␮g/ml䡠min; P ⫽ 0.042) or IGFBP-3NG (36.1 ⫾ 18.7 ␮g/ml䡠min; P ⫽ 0.002) were significantly higher, but not significantly different from each other. Similarly, the mean IGF-I clearance for IGF-I administered with IGFBP-3MUT (1.18 ⫾ 0.27 ml/min) was not different from that of IGF-I administered alone (1.53 ⫾ 1.03 ml/min; P ⫽ 0.35), whereas the estimated clearance rates for IGF-I administered with IGFBP-3 (0.69 ⫾ 0.41 ml/min; P ⫽ 0.047) or IGFBP-3NG (0.37 ⫾ 0.22 ml/min; P ⫽ 0.007) were significantly lower, although not significantly different from each other. That is, IGFBP-3MUT did not significantly retard overall IGF-I disappearance by this analysis, whereas the other two IGFBP-3 forms did. The mean IGF-I data could be adequately fitted by a triexponential model, as shown by the lines in Fig. 3. Kinetic parameters estimated from this model are shown in Table 2. These data suggest that IGF-I clearance is retarded by complexing with all IGFBP-3 forms, with IGFBP-3NG having a greater effect than IGFBP-3, and IGFBP-3MUT having a smaller effect. The estimates of volume of distribution are likely to reflect the molecular sizes of the different complexes, with IGFBP-3 forming complexes similar to IGFBP-3NG, and the higher value for IGFBP-3MUT reflecting its lower propensity to form ternary complexes. For IGF-I alone, the distribution half-lives of the first and second components were similar, suggesting that they may represent the same compartment; in the presence of all IGFBP-3 forms, three components were evident, with average half-lives of approximately 1.5, 15, and 90 min. Figure 4 compares the disappearance of each IGFBP-3 analog from the rat circulation when administered alone or in association with IGF-I. When examined in a model-independent manner, by calculating the AUC for each animal over 0 –120 min and analyzing the data by two-factor ANOVA, it was evident that for each protein, coadministration with IGF-I greatly increased its retention (P ⬍ 0.0001). The mean AUC was significantly lower for IGFBP-3MUT than for the other IGFBP-3 forms (P ⬍ 0.001) and was slightly higher for IGFBP-3NG than for IGFBP-3 (P ⫽ 0.033). Similar differences were seen when the mean IGFBP-3 clearance over 0 –120 min was analyzed by this method, except that no difference between IGFBP-3 and IGFBP-3NG was evident (P ⫽ 0.90; data not shown). The mean IGFBP-3 data could be adequately fitted by a triexponential model, as shown by the lines in Fig. 4. Kinetic parameters estimated from this model are shown in Table 3. These data confirm that the clearance of all IGFBP-3 forms is retarded by complexing with IGF-I. For each of the complexes with IGF-I, the volume of distribution of the IGFBP-3 analog was smaller than that when administered alone, reflecting the more ready ternary complex formation by these preformed binary complexes. The distribution half-lives derived from the three-component model showed a surprisingly rapid component for each IGFBP-3 analog (t1/2, ⬃1–3 min), similar to the rapid phase of IGF-I disappearance, a middle component with a mean t1/2 of approximately 15–30

1672



FIG. 2. Glucose levels in the plasma of rats administered IGF-I or IGFBP-3 analogs in the presence and absence of IGF-I. Groups of rats were administered IGF-I, IGF-I plus IGFBP-3 analog, or IGFBP-3, as indicated in Table 1. Blood was sampled over 2 h, and glucose levels were determined. Treatment groups are indicated by the key in panels, and results shown are the mean ⫾ SE. A, Comparison of blood glucose levels of rats administered IGF-I in the presence and absence of IGFBP-3 analogs. When analyzed by repeated measures ANOVA, the groups administered IGFBP-3 or IGFBP-3NG in the presence of IGF-I was significantly different from those receiving IGF-I alone (P ⱕ 0.02). There was no significant difference between the groups receiving IGF-I with and without IGFBP-3MUT (P ⬎ 0.1). B, Comparison of blood glucose levels of rats administered IGFBP-3 analogs in the presence and absence of IGF-I. The results are shown as separate panels for IGFBP-3 (top), IGFBP-3MUT (middle), and IGFBP-3NG (bottom) for clarity. When analyzed by repeated measures ANOVA, there was no significant difference between the groups receiving either IGFBP-3 or IGFBP-3NG in the presence or absence of IGF-I (P ⬎ 0.5). There was a significant difference between the groups receiving IGF-I plus IGFBP-3MUT and IGFBP-3MUT alone (P ⬍ 0.05).

FIG. 3. Concentration vs. time profiles of IGF-I administered with and without IGFBP-3 analogs. The concentration of IGF-I in the plasma samples were measured by RIA. Treatment groups are indicated by the key in the panel, and data points represent the mean ⫾ SE for each group. Solid lines represent the triexponential fits to mean experimental values, with half-lives for each component shown in Table 2.

min, and a slow phase with a shorter half-life for IGFBP3MUT than the other analogs. Size fraction chromatography was used to determine the

molecular mass distribution of the various IGFBP-3 analogs in the rat circulation. The results shown in Fig. 5 represent the mean of the pooled results for all animals in each group, with the error bars omitted for clarity. Two peaks of immunoreactivity were observed corresponding to ternary complexes (around fraction 24) and free or binary complexes (around fraction 28). Due to the lack of baseline separation between the two peaks, fraction 26 was distributed equally between the binary and ternary complexes. Two minutes after administration, approximately 60% of IGFBP-3 and IGFBP-3NG coinjected with or without IGF-I were found in the high mol wt complex (Fig. 5, A and B, and E and F). In contrast, only 30% of IGFBP-3MUT was present in this complex formed at 2 min (Fig. 5, C and D). It is also notable that whereas ternary complexes were retained in the serum over time for IGFBP-3 and IGFBP-3NG, they disappeared rapidly for IGFBP-3MUT, suggesting more rapid dissociation. At 2 min after administration, approximately 2-fold more IGFBP-3 and IGFBP-3NG were detectable, when coinjected with IGF-I compared with when injected alone, in both the high and low mol wt complex forms. In contrast, there was only approximately 1.2-fold more IGFBP-3MUT in the high and low mol wt complex forms when it was coadministered with IGF-I than when it was given without IGF-I. The clearance of the two separate components of the complexes was estimated by calculating the areas under the two peaks of the chromatographic profiles of each time point for every animal. The time course of disappearance of each peak for each IGFBP-3 analog administered with and without



TABLE 2. Pharmacokinetic parameters of IGF after administration alone or with IGFBP-3 analogs Treatment

IGF-I IGF-I IGF-I IGF-I

alone ⫹ IGFBP-3 ⫹ IGFBP-3MUT ⫹ IGFBP-3NG

AUCa (␮g/ml䡠min)

CLb (liter/min䡠kg)

Vssc (ml/kg)

t1/2,␣1d (min)

t1/2,␣2d (min)

t1/2,␣3d (min)

3.8 24.2 14.3 43.2

2.63 0.41 0.70 0.23

2287 222 545 230

1.4 1.7 1.5 1.2

2.1 12.2 13.8 21.5

80.2 70.7 92.4 106.6

Parameter estimates have been derived from mean concentration-time data using a triexponential model, as shown in Fig. 3. a Area under the IGF-I concentration time curve. b Clearance. c Volume of distribution at steady state. d t1/2,␣i, Distribution half-lives derived from the triexponential model exponents (␣1, ␣2, and ␣3).

FIG. 4. Concentration vs. time profiles of IGFBP-3 analogs administered with and without IGF-I. The concentrations of IGFBP-3 in the plasma samples were determined by RIA. Treatment groups are indicated by the key in the panel, and data points represent the mean ⫾ SE for each group. Solid lines represent the triexponential fits to mean experimental values, with half-lives for each component shown in Table 3.

TABLE 3. Pharmacokinetic parameters of IGFBP-3 analogs administered alone or in complex with equimolar IGF-I Treatment

AUCa (␮g/liter䡠min)

CLb (ml/min䡠kg)

Vssc (liter/kg)

t1/2,␣1d (min)

t1/2,␣2d (min)

t1/2,␣3d (min)

IGF-I ⫹ IGFBP-3 IGFBP-3 alone IGF-I ⫹ IGFBP-3MUT IGFBP-3MUT alone IGF-I ⫹ IGFBP-3NG IGFBP-3NG alone

528 197 225 52 673 145

1.06 2.84 2.49 10.86 0.83 3.85

154 814 271 559 166 560

2.8 1.0 2.8 0.9 2.4 1.4

35.8 13.4 14.0 20.5 32.6 16.4

138.6 223.6 100.5 84.5 177.8 144.4

Parameter estimates have been derived from mean concentration-time data using a triexponential model, as shown in Fig. 4. a Area under the IGFBP-3 concentration time curve. b Clearance. c Volume of distribution at steady state. d t1/2,␣i, Distribution half-lives derived from the triexponential model exponents (␣1, ␣2, and ␣3).

IGF-I is shown in Fig. 6. A combined repeated measures ANOVA was performed on the clearance curves of either binary or ternary complexes for all three analogs, injected with or without IGF-I. For each analog, the clearance of IGFBP-3 found in the lower molecular weight, binary complexes was similar regardless of whether the IGFBP-3 was coadministered with or without IGF-I (P ⬎ 0.05). Similarly, the clearance of IGFBP-3MUT found in the high molecular weight, ternary complexes was not significantly different when IGFBP-3MUT was coadministered with or without IGF-I (P ⫽ 0.69; Fig. 6B). In contrast, the disappearance of either IGFBP-3 (Fig. 6A) or IGFBP-3NG (Fig. 6C) found in the high molecular weight, ternary complexes was significantly different when IGFBP-3 was coadministered with IGF-I compared with when the IGFBP-3 was administered alone (P ⬍ 0.05). Discussion

The insulin-like activity of IGFs has been estimated to be 5–10% that of insulin (13, 14). Based on the high circulating

IGF concentration, the insulin-like potential of IGFs far exceeds that of insulin itself and yet it does not cause hypoglycemia. At least 80% of the IGFs are sequestered into 150kDa complexes with either IGFBP-3 or IGFBP-5 and ALS, with most of the remainder bound in binary complexes with one of the six IGFBPs. The sequestration of IGFs into the high mol wt ternary complexes prolongs their half-lives in the circulation, thus providing a ready reservoir of potentially bioavailable IGFs. The availability of the circulating IGFs to target tissues depends on the release of the IGFs from the ternary complexes and ultimately from the binary complexes. In humans, it is estimated that less than 1% of IGFs are found in the free, unbound form in the circulation (1), although higher levels have been reported in rats (15). In previous published studies the amount of IGF-I required, as an iv bolus injection, to induce hypoglycemia ranged from approximately 765 ␮g/kg BW of the rat (⬃100 nmol/kg) (16) to approximately 100 ␮g/kg (17). In this study the administration of 100 ␮g/kg IGF-I to rats caused hypoglycemia, which could be blocked by coadministering

1674



FIG. 5. The molecular mass distribution of the various IGFBP-3 analogs in plasma samples. After size fractionation of 25 ␮l plasma from each sampled time point (indicated by the key in A) on a Superose-12 column, the amount of IGFBP-3 in fractions 20 –32 was determined by RIA. Fractions 22–26 represent the 150-kDa ternary complexes, and fractions 26 –29 represent the 50-kDa binary complexes; the amount of IGFBP-3 in fraction 26 was distributed between the two peaks. The samples analyzed were from groups of rats treated with IGF-I plus IGFBP-3 (A), IGFBP-3 alone (B), IGF-I plus IGFBP-3MUT (C), IGFBP-3 alone (D), IGF-I plus IGFBP-3NG (E), and IGFBP-3NG alone (F), as described in Table 1. Results shown are the mean pooled results in each group, with error bars omitted for clarity.

FIG. 6. Relative clearance profiles of IGFBP-3 (A), IGFBP-3MUT (B), and IGFBP-3NG (C) in various molecular forms from the circulation of rats treated with each IGFBP-3 analog alone (open symbols) or in combination with IGF-I (closed symbols). The AUCs representing the 150-kDa ternary and 50-kDa binary complexes were calculated for each animal and each time point from the raw data of Fig. 5. Results shown are the mean ⫾ SE of IGFBP-3 in the 150-kDa (squares) and 50-kDa (circles) forms.

equimolar concentrations of either IGFBP-3 or IGFBP-3NG. The equal potency of the glycosylated and nonglycosylated IGFBP-3 is supported by previous in vitro data that showed that the carbohydrate moieties on IGFBP-3 have little impact on ternary complex formation (6), in contrast to the carbohydrate moieties on ALS, which play an essential role in this process (18).

Distribution of IGFBP-3 among the different complex forms was estimated by gel permeation chromatography. Due to the lack of baseline separation between the binary and ternary complex peaks, there may be a degree of imprecision in the estimates of IGFBP-3 distribution. However, consistent with a previous study (4), approximately two thirds of IGFBP-3 and IGFBP-3NG were found in ternary complexes


within 2 min of administration regardless of whether it was injected with or without IGF-I, suggesting that it is the rapid formation of the ternary complexes that is blocking the hypoglycemic effect of IGF-I. In contrast, only one third of IGFBP-3MUT, which has reduced affinity for ALS and hence reduced ternary complex formation ability, was found in ternary complexes at 2 min, and this form disappeared rapidly, suggesting rapid dissociation. IGFBP-3MUT was unable to fully block the IGF-I hypoglycemic effect, demonstrating for the first time in vivo the importance of the carboxylterminal basic residues (K228GRKR) of IGFBP-3 in its ability to form ternary complexes and consequently its ability to counteract the insulin-like activity of IGFs. This clearly illustrates the role of ALS in interacting with IGFBP-3 as a key glucoregulatory influence. When administered alone, at least 50% of the IGF-I was cleared from the circulation within 5 min. This suggests that there is little free endogenous rat IGFBP-3 or IGFBP-5 to retain the IGF-I in the circulation as part of ternary complexes. The most rapid component derived from triexponential analysis of the IGF-I time-course data (t1/2, ⬃1.5 min) may represent uncomplexed IGF-I and is comparable to the first phase (t1/2, 2 min) determined from a similar triexponential analysis of 125I-labeled IGF-I clearance in rats (19). Determined by model-independent AUC analysis, the retention of IGF-I over the 120-min period, when injected as a complex with IGFBP-3MUT, was not significantly different from that seen when IGF-I was injected alone, although the three-phase exponential analysis indicated a component of intermediate half-life (t1/2, ⬃15 min) that was absent when IGF-I was administered alone. This may represent the binary complex between IGF-I and IGFBP-3MUT, as a component of similar half-life also contributed to the IGFBP-3 and IGFBP-3NG curves, and an IGF half-life of 20 –30 min has previously been estimated in the absence of ALS (i.e. in hypophysectomized rats) (20). However, the estimate of this middle component half-life is considerably lower than the value of 72 min derived from analysis of 125I-labeled IGF-I clearance (19). The slowest phase of IGF-I disappearance (t1/2, ⬃90 min) might be related to IGF-I in ternary complexes, although earlier studies suggested a longer half-life for this form (4, 20). The half-life of ALS administered to rats has been estimated to be approximately 120 min (4), and this may be a major determinant of the ternary complex half-life. A recent triexponential analysis of 125I-labeled IGF-I disappearance estimated the half-life of the third component as approximately 6 h; the higher value possibly is the result of the much longer sampling period used in that study (19). Whatever the true value, it is evident that ternary complexes make a relatively minor contribution to the disappearance of IGF-I administered with IGFBP-3MUT, because its clearance over 120 min appeared similar to that of IGF-I alone, consistent with a reduced ability to bind ALS. Similar to the clearance of IGF-I, the mean concentration curves for each of the IGFBP-3 analogs could also be fitted to a triexponential model. The slowest component (average t1/2, ⬃150 min) may predominantly reflect ternary complexes, rate-limited by ALS as discussed previously, and the intermediate component (average t1/2, ⬃20 min) may reflect


binary complexes. Interestingly, these three-phase analyses yielded a rapid component for each IGFBP-3 analog, with a half-life of 1–2 min. If this represents the initial phase of uncomplexed IGFBP-3 clearance, it may be more rapid than previously estimated. The coadministration of IGF-I with each of the analogs significantly retarded its clearance compared with that when the protein was administered alone, and there was relatively less IGFBP-3MUT retained in the circulation than either IGFBP-3 or IGFBP-3NG regardless of whether IGF-I was coadministered, again reflecting the importance of ALS binding in retaining both IGFs and IGFBP-3 in the circulation. The rise in plasma glucose levels when IGFBP-3 was administered without IGF-I suggests that there could be free IGF-I available in the circulation, the temporary depletion of which causes transient hyperglycemia. A similar observation was made when IGFBP-1 was administered as an iv bolus in the rat (17), where plasma glucose levels increased transiently by approximately 10% of baseline values. Frystyk et al. (15) detected approximately 30 ng/ml free IGF-I (⬃3% of total IGF-I) in rats fasted for 1 d. It would appear that this free IGF-I has a tonic insulin-like effect in regulating circulating glucose levels. A similar rise in plasma glucose was seen when IGFBP-3 was coadministered with IGF-I. This is hard to explain if the IGFBP-3 was fully occupied by exogenous IGF-I and unable to bind endogenous free IGF-I. The possibility therefore cannot be excluded that IGFBP-3 exerts this effect by an alternative mechanism, possibly eliciting a rapid stress or counterregulatory response that transiently elevates blood glucose. This transient hyperglycemia was not seen when IGFBP-1 was coadministered with IGF-I (17). In fact, the hypoglycemic effect of IGF-I was not fully reversed by IGFBP-1, where the glucose levels remained approximately 10% below baseline levels. Interestingly, statistical analysis of the model-independent IGFBP-3 AUC estimates for individual animals showed that IGFBP-3NG was retained in the circulation to a significantly greater extent than IGFBP-3, although this effect was not significant by other analyses. This is consistent with the trend, seen in the IGF-I disappearance curves, for IGF-I administered with IGFBP-3NG to have the most prolonged serum retention and suggests a possible role for carbohydrate on IGFBP-3 in regulating its retention in the circulation. Carbohydrate moieties are known to serve as clearance markers that determine the lifetime of glycoproteins in circulation (21, 22) and may also have this function in IGFBP-3. Although the binding of IGFBP-3 to cell surfaces has been described for several different cell systems (23, 24), the putative IGFBP-3 binding molecules identified are yet to be characterized (25, 26). We have previously shown that there is approximately 2- to 3-fold increased cell binding by IGFBP3NG compared with its glycosylated form, IGFBP-3 (6, 7), and that the same basic amino acids in the carboxyl-terminus of IGFBP-3 are responsible for both cell and ALS binding (5). This suggests that the competition between the putative IGFBP-3 binding molecules on the cell surface and ALS for the same domain on IGFBP-3 could lead to dissociation of the ternary complex. However, it may well be that the increased binding of IGFBP-3NG to cells would retain more of this protein in the vascular compartment than IGFBP-3, thus

1676


allowing reassociation to occur. To this end, it has been shown that IGF-I can displace IGFBP-3 from cell surfaces (23), thus allowing the reformation of binary complexes and, subsequently, ternary complexes resulting in the prolonged retention of IGFBP-3NG in the circulation. In conclusion, we have shown that IGFBP-3MUT injected iv in conscious rats formed ternary complexes poorly, was rapidly cleared from the circulation, and was significantly less effective than normal IGFBP-3 in preventing IGF-Iinduced hypoglycemia. In contrast, IGFBP-3NG formed ternary complexes as efficiently as glycosylated IGFBP-3, was retained normally or longer in the circulation, and was as effective as IGFBP-3 in blocking IGF-I activity. These data 1) demonstrate that IGFBP-3 glycosylation does not affect its glucoregulatory function, and 2) confirm the in vitro observation that the basic carboxyl-terminal domain of IGFBP-3 determines its ability to complex with ALS. Finally, the data show for the first time in vivo that the regulation of IGF-I activity by IGFBP-3 is dependent on its ability to form ternary complexes.


6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

Acknowledgments

17.

The expert support of Dr. S. E. Kong with the cannulation technique is gratefully acknowledged.

18.

Received August 20, 2001. Accepted January 7, 2002. Address all correspondence and requests for reprints to: Dr. Sue M. Firth, Kolling Institute of Medical Research, Royal North Shore Hospital, St. Leonards, New South Wales 2065, Australia. E-mail: sfirth@med. usyd.edu.au. This work was supported by the National Health and Medical Research Council, Australia (Grant 990005 to R.C.B. and S.M.F.).

19.

20. 21.

References 1. 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 2. Baxter RC, Martin JL 1989 Structure of the Mr 140,000 growth hormonedependent insulin-like growth factor binding protein complex: determination by reconstitution and affinity-labeling. Proc Natl Acad Sci USA 86:6898 – 6902 3. Twigg SM, Baxter RC 1998 Insulin-like growth factor (IGF) binding protein-5 forms an alternative ternary complex with IGFs and the acid-labile subunit. J Biol Chem 273:6074 – 6079 4. Lewitt MS, Saunders H, Phuyal JL, Baxter RC 1994 Complex formation by human insulin-like growth factor-binding protein-3 and human acid-labile subunit in growth hormone-deficient rats. Endocrinology 134:2404 –2409 5. Firth SM, Ganeshprasad U, Baxter RC 1998 Structural determinants of ligand

22. 23. 24. 25. 26.

and cell surface binding of insulin-like growth factor binding protein-3. J Biol Chem 273:2631–2638 Firth SM, Baxter RC 1999 Characterisation of recombinant glycosylation variants of insulin-like growth factor binding protein-3. J Endocrinol 160:379 –387 Firth SM, Ganeshprasad U, Poronnik P, Cook DI, Baxter RC 1999 Adenoviralmediated expression of human insulin-like growth factor-binding protein-3. Protein Exp Purif 16:202–211 Baxter RC, Martin JL 1986 Radioimmunoassay of growth hormone-dependent insulin-like growth factor binding protein in human plasma. J Clin Invest 78:1504 –1512 Baxter RC, Martin JL, Beniac VA 1989 High molecular weight insulin-like growth factor binding protein complex. Purification and properties of the acid-labile subunit from human serum. J Biol Chem 264:11843–11848 Baxter RC, Bayne ML, Cascieri MA 1992 Structural determinants for binary and ternary complex formation between insulin-like growth factor-I (IGF-I) and IGF binding protein-3. J Biol Chem 267:60 – 65 Baxter RC, Axiak S, Raison RL 1982 Monoclonal antibody against human somatomedin-C/insulin-like growth factor-I. J Clin Endocrinol Metab 54: 474 – 476 Gibaldi M, Perrier D 1982 Pharmacokinetics. 2nd ed. New York: Marcel Dekker Guler HP, Zapf J, Froesch ER 1987 Short-term metabolic effects of recombinant human insulin-like growth factor-I in healthy adults. N Engl J Med 317:137–140 Zapf J 1997 Total and free IGF serum levels. Eur J Endocrinol 136:146 –147 Frystyk J, Delhanty PJD, Skjaerbaek C, Baxter RC 1999 Changes in the circulating IGF system during short-term fasting and refeeding in rats. Am J Physiol 40:E245–E252 Schmitz F, Hartmann H, Stumpel F, Creutzfeldt W 1991 In vivo metabolic action of insulin-like growth factor I in adult rats. Diabetologia 34:144 –149 Lewitt MS, Denyer GS, Cooney GJ, Baxter RC 1991 Insulin-like growth factor-binding protein 1 modulates blood glucose levels. Endocrinology 129: 2254 –2256 Janosi JB, Firth SM, Bond JJ, Baxter RC, Delhanty PJ 1999 N-Linked glycosylation and sialylation of the acid-labile subunit. Role in complex formation with insulin-like growth factor (IGF)-binding protein-3 and the IGFs. J Biol Chem 274:5292–5298 Dubaquie´ Y, Mortensen DL, Intintoli A, Hogue DA, Nakamura G, Rancatore P, Lester P, Sadick MD, Filvaroff E, Fielder PJ, Lowman HB 2001 Binding protein-3-selective insulin-like growth factor I variants: engineering, biodistributions, and clearance. Endocrinology 142:165–173 Zapf J, Hauri C, Waldvogel M, Froesch ER 1986 Acute metabolic effects and half-lives of intravenously administered insulinlike growth factors I and II in normal and hypophysectomized rats. J Clin Invest 77:1768 –1775 Lis H, Sharon N 1993 Protein glycosylation. Structural and functional aspects. Eur J Biochem 218:1–27 Opdenakker G, Rudd PM, Ponting CP, Dwek RA 1993 Concepts and principles of glycobiology. FASEB J 7:1330 –1337 Martin JL, Ballesteros M, Baxter RC 1992 Insulin-like growth factor-I (IGF-I) and transforming growth factor-␤1 release IGF-binding protein-3 from human fibroblasts by different mechanisms. Endocrinology 131:1703–1710 Oh Y 1998 IGF-independent regulation of breast cancer growth by IGF binding proteins. Breast Cancer Res Treat 47:283–293 Oh Y, Muller HL, Pham H, Rosenfeld RG 1993 Demonstration of receptors for insulin-like growth factor binding protein-3 on HS578T human breast cancer cells. J Biol Chem 268:26045–26048 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