Endocrinology and Metabolism

Impaired glucose homeostasis in insulin-like growth factor-binding protein-3-transgenic mice

Josef V. Silha, Yaoting Gui, Liam J. Murphy


Glucose homeostasis was examined in male transgenic (Tg) mice that overexpressed the human insulin-like growth factor (IGF)-binding protein (IGFBP)-3 cDNA, driven by either the cytomegalovirus (CMV) or the phosphoglycerate kinase (PGK) promoter. The Tg mice of both lineages demonstrated increased serum levels of human (h) IGFBP-3 and total IGF-I compared with wild-type (Wt) mice. Fasting blood glucose levels were significantly elevated in 8-wk-old CMV-binding protein (CMVBP)-3- and PGK binding protein (PGKBP)-3-Tg mice compared with Wt mice: 6.35 ± 0.22 and 5.22 ± 0.39 vs. 3.99 ± 0.26 mmol/l, respectively. Plasma insulin was significantly elevated only in CMVBP-3-Tg mice. The responses to a glucose challenge were significantly increased in both Tg strains: area under the glucose curve = 1,824 ± 65 and 1,910 ± 115 vs. 1,590 ± 67 mmol · l−1 · min for CMVBP-3, PGKBP-3, and Wt mice, respectively. The hypoglycemic effects of insulin and IGF-I were significantly attenuated in Tg mice compared with Wt mice. There were no differences in adipose tissue resistin, retinoid X receptor-α, or peroxisome proliferator-activated receptor-γ mRNA levels between Tg and Wt mice. Uptake of 2-deoxyglucose was reduced in muscle and adipose tissue from Tg mice compared with Wt mice. These data demonstrate that overexpression of hIGFBP-3 results in fasting hyperglycemia, impaired glucose tolerance, and insulin resistance.

  • insulin resistance
  • diabetes
  • resistin
  • peroxisome proliferator-activated receptor-γ

although the insulin-like growth factors (IGFs) can exert hypoglycemic effects when administered to rodents or human subjects (2, 22), the role of endogenous IGF-I and -II in glucose homeostasis remains unclear. Although IGF-I and -II are present in much higher concentrations in the circulation than insulin, the presence of high-affinity binding proteins limits the free concentration of the IGFs in plasma to levels comparable with insulin (3).

In experiments with rodents and human subjects, the IGFs are ∼5–15% as potent as insulin in terms of their acute hypoglycemic effects (2, 22). Thus the 2- to 8-ng/ml plasma concentrations of free IGF-I and -II (8) are unlikely to have a major hypoglycemic effect compared with the 1- to 4-ng/ml of insulin normally present in human plasma. However, hyperglycemia has been convincingly demonstrated after the injection of IGF-binding protein (IGFBP)-1 in rats (13) and in transgenic (Tg) mice that overexpress IGFBP-1 (20). In these situations, the hyperglycemia could be attributable to a reduction in free IGF-I levels, although this was not documented in either study (13, 20). Furthermore, in mice carrying a liver-specific null mutation in the IGF-I gene, there is an increase in serum insulin levels despite normal blood glucose levels (24), possibly indicating that the reduction in the tonic hypoglycemic effect of free IGF-I resulted in a compensatory increase in insulin levels.

Conversely, hypoglycemia is observed when there are increased levels of free IGF present in the circulation. The pathological condition of nonislet tumor-associated hypoglycemia is due to synthesis and secretion of a 15-kDa IGF-II variant (7). In this condition, elevated levels of “big IGF-II” are associated with suppressed growth hormone (GH) levels and, as a consequence, reduced levels of IGF-I, IGFBP-3, and the acid-labile subunit (32). This results in a decreased proportion of the IGFs present in the ternary complex and presumably increased levels of free IGFs, particularly free IGF-II, in the blood (32). Taken together, these data suggest that the free IGF-I and -II may have a tonic hypoglycemic effect and that modulation of the IGFBPs by regulation of the proportion of free and bound IGF may be important in glucose homeostasis.

In addition to modulating the availability of IGF-I and -II, IGFBP-3 has been shown to have IGF-independent effects on cellular proliferation and apoptosis in a variety of cell lines (18, 19, 28). These effects may be mediated either via cell surface binding proteins or nuclear binding sites (19,22). Thus the potential exists for IGFBP-3 to have both IGF-dependent and IGF-independent effects on glucose homeostasis. Of note in this regard is the recent observation that IGFBP-3 interacts with the retinoid X receptor-α (RXRα; Ref. 14). The latter is an important binding partner for the peroxisome proliferator-activated receptor-γ (PPARγ), a nuclear protein that is involved in transcriptional regulation of a variety of enzymes involved in glucose and lipid metabolism (11).

In this report, we investigated the effects of overexpression of IGFBP-3 in Tg mice on glucose homeostasis. Expression of PPARγ, RXRα, and resistin, a potential mediator of the insulin-sensitizing PPARγ agonists (25), was also assessed in these Tg mice.


Tg mice.

The generation of the Tg mice and the characterization of their phenotype have been described in detail elsewhere (15). A human (h) IGFBP-3 cDNA containing the entire coding region was subcloned downstream of either the mouse phosphoglycerate (PGK) promoter (11) or the cytomegalovirus (CMV) promoter (9). Tg mice were generated by microinjection of the transgene into pronuclei of fertilized CD-1 zygotes. The founders were bred with CD-1 mice. CD-1 mice from the same colony, bred in a similar fashion, provided wild-type (Wt), non-Tg control mice of the same genetic background. The Tg mice of both lineages demonstrated ubiquitous expression of hIGFBP-3 mRNA and serum levels of hIGFBP-3 (∼5 μg/ml) compared with Wt mice (15).

Homozygous Tg mice and Wt male mice of 8 wk of age were used for all experiments. Studies on fasting mice were performed starting at 10 AM after an overnight fast. For all experiments, mice were anesthetized with 2.4% Avertin, and blood samples were collected from the retroorbital sinus using heparinized capillaries unless otherwise stated. Samples for insulin, GH, and leptin assays were collected between 9 and 11 AM. Because GH levels were suppressed by avertin-induced anesthesia, blood for GH determinations was collected by cardiac puncture after cervical dislocation. All experiments were performed in accordance with protocols approved by the Animal Care Committee of the Faculty of Medicine, University of Manitoba.

Blood glucose determinations and glucose tolerance test.

Glucose was measured in whole blood with the use of a glucose analyzer (YSI 2300; Yellow Springs, OH). For the intraperitoneal glucose tolerance test, mice were fasted overnight. Blood samples were collected before and 15, 30, 60, 120, and 180 min after administration of an intraperitoneal injection of glucose (1 mg/g body wt in sterile 0.45% saline). The area under the glucose curve (AUCglucose) was calculated by the trapezoid method. In a separate experiment, the insulin concentration in fasting mice was measured before and 15, 30, and 60 min after an intraperitoneal glucose injection (1 mg/g body wt).

Insulin and IGF-I tolerance tests.

Recombinant human insulin and IGF-I were purchased from Boehringer (Mannheim, Germany) and GroPep (Adelaide, Australia), respectively. The insulin and IGF-I tolerance tests were performed on overnight-fasted 8-wk-old male mice. Blood glucose was measured before and 20, 40, 60, 80, 100, and 120 min after the subcutaneous administration of insulin (10 μg/kg body wt) or IGF-I (50 μg/kg body wt).

RNA extraction and RNase protection assays.

Total RNA was isolated from skeletal muscle and hepatic and adipose tissue with TRIzol reagent (GIBCO-BRL Life Technology, Burlington, ON, Canada). The concentration of RNA was determined spectrophotometrically, and the integrity of the RNA in all samples was documented by visualization of the 18S and 28S ribosomal bands after electrophoresis through a 0.8% formaldehyde-agarose gel. RNase protection assays (RPA) were used to measure resistin, RXRα, and PPARγ mRNA abundance in adipose tissue. To generate radiolabeled cRNA for mouse resistin, RT-PCR was used to generated a 267-bp fragment of mouse resistin corresponding to nucleotides 110–376 of the published sequence (25). This fragment was subcloned into pCR II vector (Invitrogen, San Diego, CA). Similarly, fragments of mouse PPARγ corresponding to nucleotides 489–749 (6) and mouse RXRα corresponding to nucleotides 155–363 (5) were generated by RT-PCR and subcloned into pCR II. The linearized plasmids were used as templates for RNA probe synthesis. A human cRNA probe (15) was used to compare transgene expression in skeletal muscle of 8-wk-old PGK-binding protein (PGKBP)-3- and CMV binding protein (CMVBP)-3-Tg mice.

The in vitro transcription and labeling reaction was carried out with the use of 80 μCi of [32P]UTP (NEN Life Science, Boston, MA) and the Maxiscript SP6/T7 polymerase kit (Ambion, Austin, TX). A riboprobe for mouse cyclophilin (Ambion) was used as the internal standard, and century RNA markers from Ambion were used as the molecular size indicators. Total RNA (20 μg) from mouse tissues was hybridized with ∼3 × 105 counts/min (cpm) of32P-labeled cRNA by incubation overnight at 45°C. After hybridization, single-stranded RNA was digested with RNase A/T at 37°C for 30 min. The remaining RNA duplexes were separated on 5% polyacrylamide-8 M urea gel. The hybridization signal was detected by autoradiography. The protected sizes for resistin, PPARγ, RXRα, β-actin, hIGFBP-3, and cyclophilin fragments were 267, 164, 87, 245, 267, and 103 bp, respectively. The intensities of the bands were quantified by densitometry. In all cases, samples from at least six mice per group were analyzed.


The fasting insulin concentrations in plasma were measured by radioimmunoassay (RIA; Pharmacia Upjohn Diagnostics, Uppsala, Sweden). The sensitivity of the assay was <2 μU/ml. Plasma leptin concentration in fasted and nonfasted animals was measured by an immunoradiometric assay specific for mouse leptin, purchased from Linco Research (St. Charles, MO). The sensitivity of the assay was 0.2 ng/ml. GH was measured by RIA with a sensitivity of 1.5 ng/ml by using reagents obtained from Amersham Pharmacia Biotech (Baie d'Urfe, QC, Canada). Resistin levels were measured in plasma from overnight-fasted mice by use of reagents obtained from Phoenix Pharmaceuticals (Belmont, CA). The sensitivity of the assay was 4 ng/ml. IGF-I was determined by using an IGF-I RIA kit purchased from Nichols Institute Diagnostics (San Juan Capistrano, CA). Sensitivity of the assay is 13.5 ng/ml, and intra-assay coefficient of variation was 6.3%. For total IGF-I measurement, IGF-I was separated from its binding proteins by acid-ethanol extraction. For separation of free IGF-I from serum, centrifugal ultrafiltration, under conditions approaching those in vivo, was used as described elsewhere (4). Briefly, 400 μl of serum were applied to Amicon YMT 30 membrane (Millipore, Bedford, MA), incubated for 30 min at 37°C at 5% CO2atmosphere, and centrifuged (1,500 g at 30°C, 25 min). Serum-free IGF-I was determined directly in the ultrafiltrates. To minimize binding of IGF-I to plastic surfaces, filtrate cups were preincubated with 1 mg/ml RIA-grade bovine serum albumin (Sigma, St. Louis, MO) and washed twice with PBS before centrifugation.

Western blotting.

Epididymal fat tissues were homogenized in RIPA buffer (10 mM Tris · HCl, pH 8.0, 10 mM EDTA, pH 8.0, 0.15 M NaCl, 1% NP-40, 0.5% SDS, 1 μg/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride). The homogenate was clarified by centrifugation at 15,000g for 15 min at 4°C and separated by electrophoresis through an 8% SDS-polyacrylamide gel. The separated proteins were transferred to nitrocellulose membrane (MSI, Westborough, MA) at 200 mA for 2 h. The blots were incubated overnight at 4°C with a 1:500 dilution of rabbit anti-PPARγ or rabbit anti-RXRα antibody (Santa Cruz Biotechnology, Santa Cruz, CA). After a washing with TBST buffer (5 mM Tris · HCl, pH 7.4, 136 mM NaCl, and 0.05% Tween 20), the blots were incubated for 2 h at room temperature with horseradish peroxidase-linked goat anti-rabbit IgG (Santa Cruz Biotechnology) at a dilution of 1:10,000. Detection of immune complexes was achieved with the use of an enhanced chemiluminescence (ECL) Western blotting kit (Amersham Pharmacia Biotech).

2-Deoxyglucose uptake.

Overnight-fasted mice were anesthetized with Avertin. 2-Deoxy-[3H]glucose (2-DG; NEN Life Science Products) was administered as a bolus via a tail vein. To achieve the same specific activity immediately after the injection, individual mice received 0.1 μCi 2-DG · mmol−1 · l−1 of blood glucose. Blood was collected from retroorbital sinus before and 1, 5, 10, 20, 30, and 40 min after 2-DG injection for measurement of blood glucose and determination of blood glucose specific activity. After 40 min, animals were exsanguinated and tissues were immediately frozen and stored at −70°C. For determination of tissue 2-DG uptake, tissue samples were homogenized in distilled water and centrifuged for 10 min at 10,000 g. These and subsequent steps were carried out at 4°C. Tissue supernatants were deproteinized with an equal volume of 7% perchloric acid, and, after 30 min on ice, samples were centrifuged at 10,000 g for 10 min. Supernatant (600 μl) was then neutralized with 2.2 M KHCO3 (150 μl). An aliquot was taken for determination of radioactivity. For measurement of plasma 2-DG radioactivity, 10 μl of plasma were deproteinized as described above.

The specific activity of the circulating glucose pool was determined as a total radioactivity [disintegrations/min (dpm)] per millimole per liter blood glucose at each time point. To compare 2-DG uptake into muscle and fat, the glucose utilization index (GUI) was then calculated as described previously (20).

Statistical analysis.

Data are expressed as means ± SE. Student's t-test was used for single comparisons between Tg and Wt mice. For determining statistical differences between multiple groups, an analysis of variance with repeated measures followed by Dunnett'st-test was used.


Fasting blood glucose levels were significantly elevated in 8-wk-old CMVBP-3- and PGKBP-3-Tg mice compared with Wt mice (Fig.1). Plasma insulin levels were significantly elevated in CMVBP-3-Tg mice compared with Wt mice, but the insulin levels were similar in Wt and PGKBP-3-Tg mice (Fig. 1).

Fig. 1.

Fasting blood glucose and plasma insulin levels in 8-wk-old male wild-type (Wt) and transgenic (Tg) mice. The data represent means ± SE for 13–23 mice/group. The significant differences between the Tg and Wt mice are indicated. CMVBP-3, cytomegalovirus-binding protein-3; PGKBP-3, phosphoglycerate kinase-binding protein-3.

When the blood glucose response to an intraperitoneal glucose challenge was assessed, more marked glycemic excursions were apparent in blood from CMVBP-3- and PGKBP-3-Tg mice than from Wt mice (Fig.2). PGKBP-3-Tg mice showed the largest glycemic excursions. When the AUCglucose was quantified, it was significantly greater in both CMVBP-3- and PGKBP-3-Tg mice than in Wt mice (Fig. 2). The insulin response to the glucose challenge, as measured as the area under the insulin curve (AUCinsulin), was similar for PGKBP-3-Tg and Wt mice: 546 ± 23 vs. 608 ± 43, respectively (P = 0.34). In contrast, the AUCinsulin for CMVBP-3-Tg mice was significantly increased compared with Wt mice: 760 ± 57 vs. 608 ± 43 (P = 0.035).

Fig. 2.

Glucose tolerance tests in 8-wk-old male Wt and Tg mice. The data represent means ± SE for 9–16 mice/group.B: area under the glucose curve (AUC) for each group of mice. The significant differences between Tg and Wt mice are indicated.

The effects of insulin and IGF-I on blood glucose were also examined in Tg and Wt mice. The hypoglycemic effects of insulin were markedly attenuated in CMVBP-3 mice and modestly reduced in PGKBP-3 mice (Fig.3). The difference between the insulin tolerance curve for CMVBP-3-Tg mice and Wt mice was significant (P < 0.001, analysis of variance with repeated measures), whereas the differences between curves for PGKBP-3-Tg and Wt mice did not achieve statistical significance (P = 0.15). However, when the first 80 min were analyzed separately, there was a statistical difference between the curves for PGKBP-3-Tg and Wt mice (P = 0.04). Because the starting basal blood glucose levels were lower in Wt mice than in PGKBP-3 mice, a plateau in blood glucose was reached in Wt mice after ∼1 h, possibly as a result of counterregulatory mechanisms. As a consequence, the Wt and PGKBP-3 insulin tolerance curves overlapped in the later time points.

Fig. 3.

Hypoglycemic response to insulin and insulin-like growth factor (IGF)-I in Wt and Tg mice. Insulin (10 μg/kg body wt) or IGF-I (50 μg/kg) was administered subcutaneously. The data represent means ± SE for 9 or 10 mice/group.

The effects of IGF-I on blood glucose were markedly attenuated in both CMVBP-3- and PGKBP-3-Tg mice compared with Wt mice (Fig. 3). The differences between CMVBP-3 and Wt mice and PGKBP-3 and Wt mice were significant (P = 0.002 and P = 0.021, respectively). However, there were no significant differences in the blood glucose response to IGF-I between CMVBP-3- and PGKBP-3-Tg mice.

Leptin levels were measured in both fasting and nonfasting 8-wk-old male mice (Table 1). Leptin levels were significantly increased in CMVBP-3 mice compared with Wt mice but were similar in PGKBP-3-Tg mice and Wt mice. In all groups of mice, food deprivation reduced plasma leptin levels by ∼50%; however, leptin levels remained significantly higher in the serum from fasted CMVBP-3-Tg mice than fasted Wt mice.

View this table:
Table 1.

Serum hormone concentrations in Wt and Tg mice

Serum GH levels were undetectable in fasted mice, consistent with previous reports of fasting-induced suppression of GH secretion in rodents (26). In nonfasted mice, GH levels were similar in CMVBP-3-Tg and Wt mice. In contrast, GH levels were significantly increased twofold in PGKBP-3 mice (Table 1).

Total and free IGF-I levels are shown in Table 1. Total IGF-I and free IGF-I levels were significantly elevated in both CMVBP-3- and PGKBP-3-Tg mice compared with Wt mice. The ratio of free to total IGF-I was not significantly different in any of the groups of mice.

Tritiated 2-DG uptake was assessed under basal conditions in Tg and Wt mice. The disappearance curves for Tg and Wt mice are shown in Fig.4. The clearance of 2-DG from the circulation was significantly reduced in CMVBP-3-Tg mice compared with Wt mice, and the uptake into skeletal muscle and adipose tissue was also reduced in these Tg mice. The clearance of 2-DG from the circulation of PGKBP-3-Tg mice was also significantly slower than that of Wt mice, and there was a significant reduction in uptake of 2-DG in the quadriceps muscle in PGKBP-3-Tg mice. There was no significant difference between Wt and Tg mice in tritiated 2-DG uptake in the liver, spleen, or kidney (Table 2).

Fig. 4.

Disappearance of tritiated 2-deoxyglucose (2-DG) from the circulation. The disappearance of radioactivity from the circulation after administration of tritiated 2-DG from the circulation is plotted. The data represent means ± SE for 15 or 16 mice/group. The differences between the disappearance curves for the Tg mice and Wt mice were significant at P = 0.001 andP = 0.024 for the CMVBP-3- and PGKBP-3-Tg mice, respectively.

View this table:
Table 2.

GUI in tissues from Wt and Tg mice

Previously, we have demonstrated that adipose tissue from CMVBP-3- and PGKBP-3-Tg mice had similar levels of transgene expression (15). However, we had not previously examined the levels of expression of the transgene in skeletal muscle from these mice. Human IGFBP-3 mRNA was significantly increased (∼14-fold) in skeletal muscle from CMVBP-3 compared with PGKBP-3-Tg mice (Fig.5). Interestingly, the absolute weight of calf muscles of both PGKBP-3- and CMVBP-3-Tg mice was significantly reduced compared with Wt mice: 0.320 ± 0.011 and 0.361 ± 0.012 vs. 0.395 ± 0.006 g (P < 0.0001 and P < 0.02, respectively). The relative weight of the calf muscles was also significantly decreased in the CMVBP-3-Tg mice compared with Wt mice (0.994 ± 0.026 vs. 1.148 ± 0.023; P < 0.0001) but not significantly different in the PGKBP-3 mice, which were proportionately smaller and lighter than that of Wt mice.

Fig. 5.

Transgene expression in skeletal muscle. Transgene-derived human (h) IGF-binding protein (IGFBP)-3 mRNA abundance was measured with the use of an RNase protection assay (RPA) in quadriceps muscle of Tg and Wt mice. Top: autoradiographic results. Bottom: data derived from densitometric analysis have been plotted as means ± SE for 6 Tg mice of each strain. The optical density (OD) area units were derived by dividing the densitometric signal for the hIGFBP-3 mRNA by the signal obtained with the internal standard cyclophilin. Note that mouse IGFBP-3 mRNA is not detected in the muscle from Wt mice by using this assay.

Because IGFBP-3 has been shown to interact with the RXRα transcriptional regulator (14), which can form heterodimers with PPARγ (11), we investigated the hypothesis that the effects of IGFBP-3 overexpression may have been mediated via perturbation in the PPARγ/RXRα/resistin pathway. There was no significant difference in the abundance of resistin mRNA in white adipose tissues in any of the groups of mice (Fig.6). Fasting plasma resistin levels were similar in all groups of mice (Table 1). Similarly, PPARγ mRNA abundance did not differ significantly among the three groups of mice (Fig. 7). Immunoblotting was used to evaluate the levels of PPARγ and RXRα protein in adipose tissue. The levels of both PPARγ and RXRα proteins were similar in all groups of mice (Fig. 8).

Fig. 6.

Resistin mRNA abundance in epididemal adipose tissue from Tg and Wt mice. Top: representative results from an RPA assay.Bottom: data derived from densitometric analysis have been plotted as means ± SE for 4 mice/group. The OD area units were derived by dividing the densitometric signal for the resistin mRNA by the signal obtained with the internal standard cyclophilin.

Fig. 7.

Peroxisome proliferator-activated receptor-γ (PPARγ) mRNA abundance in epididemal adipose tissue from Tg and Wt mice. Top: representative results from an RPA assay. Bottom: data derived from densitometric analysis have been plotted as means ± SE for 6 mice/group. The OD area units were derived by dividing the densitometric signal for the PPARγ mRNA by the signal obtained with the internal standard β-actin.

Fig. 8.

Western blot analysis of PPARγ and retinoid X receptor (RXR)-α protein in adipose tissue from Tg and Wt mice. Representative data for 3 mice from each group are shown; however, a total of 6 mice/group were analyzed.


Insulin is the central hormone in regulating blood glucose levels; however, other hormonal and nonhormonal factors are clearly important in blood glucose homeostasis. The majority of the insulin-like activity present in serum is due to the IGFs rather than insulin itself (4). Because most of IGF-I and -II present in the circulation is bound to the IGFBPs, predominantly IGFBP-3, insulin and the opposing counterregulatory hormones are able to acutely regulate the blood glucose in response to glucose influx in the prandial state. However, free unbound and/or easily dissociable IGF-I and -II, which constitute a minor component of the total IGF present in plasma (8), may have some role in short-term glucose regulation. Several lines of evidence support this notion. These include the acute changes in IGFBP-1 expression that accompany food intake (17), the hyperglycemia that accompanies infusion of IGFBP-1 (13) and IGFBP-3 (29), the glucose intolerance observed in liver-specific IGF-I knockout mice (30), and the hyperglycemia observed in Tg mice overexpressing IGFBP-1 (20) and IGFBP-3 (this study).

Partitioning of IGF among the free unbound, the easily dissociable, and the tightly bound, slowly turning over IGF pools in the circulation may also be important in longer-term control of glucose homeostasis. However, the free or the more appropriately termed easily dissociable IGF-I was elevated in the PGKBP-3-Tg and CMVBP-3-Tg mice compared with Wt mice. This assay may not measure true in vivo free IGF-I levels, particularly under abnormal conditions such as the elevated IGFBP-3 levels observed in the Tg mice. The ratio of the measured “free IGF-I” to total IGF-I was similar in Tg and Wt mice. However, in PGKBP-3-Tg mice, GH levels were significantly increased compared with CMVBP-3-Tg and Wt mice, possibly indicating that in the PGKBP-3-Tg mice, free IGF-I levels, at least in the pituitary gland, were reduced.

The majority of circulating IGF is present in the ternary complex bound to IGFBP-3 (22). Unlike IGF present in binary complexes, IGF present in the ternary complex has a much slower turnover (10). Binary complexes of IGFBP-3-IGF-I or -II may exist, but because acid-labile subunit (ALS) is usually present in molar excess of IGFBP-3, this type of binary complex is not abundant under normal circumstances (12). In both strains of IGFBP-3-Tg mice, the majority of IGF-I is present in the ternary complex, indicating that human IGFBP-3 is able to bind to mouse ALS (Ref.15 and unpublished observations). Unlike IGFBP-1, IGFBP-3 has a long half-life in the circulation (1) and does not show acute changes in response to nutrition. Thus IGFBP-3-associated IGF-I or IGF-II represents a relatively slowly turning over circulating reservoir of IGF that constitutes the majority of the IGF present in the circulation.

The functional significance of the ternary complex IGF reservoir has recently been questioned with the demonstration that normal postnatal growth can occur in mice with conditional nullification of hepatic IGF-I expression (23, 31). These mice have low circulating total IGF-I levels, although free IGF-I levels appear to be normal (30). Interestingly, these mice demonstrate mild glucose intolerance and skeletal muscle insulin resistance, although fasting glycemia is normal. The insulin resistance in these mice may be due in part to elevated GH levels (30).

The data reported here, using two different strains of Tg mice generated with different transgene constructs, demonstrate that overexpression of IGFBP-3 is associated with disturbances in glucose homeostasis and reduced insulin sensitivity. In both of these Tg mouse strains, there is a significant increase in the circulating concentrations of IGFBP-3, IGF-I, and ternary complex (15). Serum hIGFBP-3 levels were ∼5 μg/ml, with no apparent reduction in endogenous mouse IGFBP-3 expression (15). Total circulating IGF-I, the majority of which was present as ternary complex, was increased ∼1.5-fold compared with Wt mice. In contrast to the findings of Yakar et al. (30) in liver-specific IGF-I knockout mice, GH was not elevated in the CMVBP-3-Tg mice, which demonstrated a more marked degree of insulin resistance than the PGKBP-3 mice. The GH levels were elevated in PGKBP-3-Tg mice. This is consistent with the greater growth retardation observed in PGKBP-3-Tg mice compared with CMVBP-3 mice (15).

The actual free IGF-I level measured in Wt mice (∼30 ng/ml) is consistent with that reported by Frystyk et al. (8) in fasted rats but is higher than that reported by Yakar et al. (30) in normal mice (∼5 ng/ml). Furthermore, the actual percentage of the total IGF-I measured as free IGF-I in our hands (∼12%) was higher than that reported for normal mice (∼2%) or rats (∼5%) (8, 30). The reason for this discrepancy is not immediately apparent, since a similar technique was used to measure free IGF-I in all reports.

In CMVBP-3- and PGKBP-3-Tg mice, both the fasting blood glucose and the glycemic response to a glucose challenge were significantly increased compared with Wt mice. CMVBP-3-Tg mice, unlike PGKBP-3-Tg mice, are characterized by increased adiposity. In the former but not the latter Tg mice, the fasting plasma insulin levels were significantly increased. However, although the fasting insulin levels were not elevated in the PGKBP-3-Tg mice, insulin resistance was demonstrable in these mice during the insulin tolerance test. In both strains of IGFBP-3-Tg mice, the hypoglycemic response to insulin was attenuated, although this was more marked in CMVBP-3-Tg mice. In contrast, the hypoglycemic response to IGF-I was equally reduced in both Tg mouse strains. Because circulating free IGF-I levels and the ratio of free to total IGF-I were similar in PGKBP-3- and CMVBP-3-Tg mice despite higher insulin levels in CMVBP-3-Tg mice, these parameters do not appear to correlate with the insulin resistance observed in the CMVBP-3-Tg mice. It is possible that the measurement of these parameters in the circulation does not reflect free IGF-I levels in skeletal muscle or other tissue beds important in insulin-sensitive glucose uptake, where abnormal production of IGFBP-3 has been generated by transgene expression. In this regard, it is of interest that expression of the transgene was markedly elevated in skeletal muscle from CMVBP-3-Tg mice compared with PGKBP-3-Tg mice. However, 2-DG uptake into skeletal muscle was also reduced in PGKBP-3-Tg mice.

In PGKBP-3 mice, GH levels were increased, whereas this was not the case in CMVBP-3 mice. Elevated GH levels, insulin resistance, and a “lean” phenotype have been documented in liver-specific conditional IGF-I knockout mice (24). However, in the CMVBP-3-Tg mice, insulin resistance is present despite normal GH levels.

In both CMVBP-3- and PGKBP-3-Tg mice, 2-DG clearance from the circulation was delayed compared with Wt mice. We also calculated the decline in specific activity of the circulating glucose pool during the disappearance of 2-DG from the circulation. Because 2-DG is not recirculated from nonhepatic tissues, the decline in specific activity reflects the appearance of new glucose in the circulation, that is, gluconeogenesis. The decline in specific activity was not enhanced in CMVBP-3- or PGKBP-3-Tg mice, indicating that, in these Tg mice models, unlike the PGKBP-1-Tg mice (20), hepatic insulin resistance is unlikely. However, reduced 2-DG uptake was observed in both skeletal muscle and adipose tissue of the CMVBP-3- and PGKBP-3-Tg mice, consistent with some degree of insulin resistance in these tissues.

Recently, RXRα has been identified as a binding partner for IGFBP-3 in a yeast two-hybrid screen. This observation, together with the previous reports of nuclear localization of IGFBP-3 (14,22), indicates that IGFBP-3 may have a role in modulating nuclear transcription of various genes involved in growth and metabolism. In this regard, it is of note that the nuclear transcription factor PPARγ is also a binding partner for RXRα (11). PPARγ is involved in the regulation of genes that control differentiation of preadipocytes (16). In terms of adipose tissue mass, CMVBP-3- and PGKBP-3-Tg mice have markedly different phenotypes (30). The CMVBP-3-Tg mice have marked obesity, whereas in PGKBP-3-Tg mice adipose tissue mass is similar to that in Wt mice. We have attributed this difference to subtle differences in the timing or level of expression of the transgene in the two mouse strains. In this regard, the increased expression of the transgene in skeletal muscle from CMVBP-3-Tg mice is of interest. Increased expression of the transgene in CMVBP-3-Tg mice could result in enhanced insulin resistance and, as a consequence, increased shunting of nutrients to adipose tissue. Despite differences in adiposity and insulin sensitivity, expression of resistin, RXRα, and PPARγ were similar in both strains of IGFBP-3-Tg mice and were not significantly different from Wt mice. Thus direct interaction of IGFBP-3 with RXRα, if this indeed occurs in vivo, does not appear to result in a disturbance in expression of RXRα, its binding partner PPARγ, or the downstream effector resistin.

The data reported here clearly demonstrate that overexpression of IGFBP-3 results in hyperglycemia, glucose intolerance, and insulin resistance. These effects cannot be explained by disturbances in either GH secretion, adiposity, or circulating “free” IGF-I levels.


This research was supported by a grant from the Canadian Institutes for Health Research. L. J. Murphy is a recipient of an endowed Research Professorship in Metabolic Diseases.


  • Address for reprint requests and other correspondence: L. J. Murphy, Dept. of Physiology, Univ. of Manitoba, Winnipeg R3E 0W3, Canada (E-mail:ljmurph{at}cc.umanitoba.ca).

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • 10.1152/ajpendo.00014.2002


View Abstract