Disruption of the Insulin-like Growth Factor Type 1 Receptor in Osteoblasts Enhances Insulin Signaling and Action*

Defective bone formation is common in patients with diabetes, suggesting that insulin normally exerts anabolic actions in bone. However, because insulin can cross-activate the insulin-like growth factor type 1 receptor (IGF-1R), which also functions in bone, it has been difficult to establish the direct (IGF-1-independent) actions of insulin in osteoblasts. To overcome this problem, we examined insulin signaling and action in primary osteoblasts engineered for conditional disruption of the IGF-1 receptor (ΔIGF-1R). Calvarial osteoblasts from mice carrying floxed IGF-1R alleles were infected with adenoviral vectors expressing the Cre recombinase (Ad-Cre) or green fluorescent protein (Ad-GFP) as control. Disruption of IGF-1R mRNA (>90%) eliminated IGF-1R without affecting insulin receptor (IR) mRNA and protein expression and eliminated IGF-1R/IR hybrids. In ΔIGF-1R osteoblasts, insulin signaling was markedly increased as evidenced by increased phosphorylation of insulin receptor substrate 1/2 and enhanced ERK/Akt activation. Microarray analysis of RNA samples from insulin-treated, ΔIGF-1R osteoblasts revealed striking changes in several genes known to be downstream of ERK including Glut-1 and c-fos. Treatment of osteoblasts with insulin induced Glut-1 mRNA, increased 2-[1,2-3H]-deoxy-d-glucose uptake, and enhanced proliferation. Moreover, insulin treatment rescued the defective differentiation and mineralization of ΔIGF-1R osteoblasts, suggesting that IR signaling can compensate, at least in part, for loss of IGF-1R signaling. We conclude that insulin exerts direct anabolic actions in osteoblasts by activation of its cognate receptor and that the strength of insulin-generated signals is tempered through interactions with the IGF-1R.

Insulin and insulin-like growth factor 1 (IGF-1) 2 are related signaling molecules that evolved from a common ancestor pathway originally involved in sensing and integrating signals arising from nutrient and growth factor availability. During evolution, this primitive receptor pathway diverged into two distinct hormonal systems that in mammals serve different but overlapping developmental and metabolic functions (1,2). These receptors belong to the family of ligand-activated receptor kinases and are abundantly expressed in osteoblasts (3,4). Unlike other receptor tyrosine kinases, these proteins exist at the cell surface as homodimers composed of two identical ␣/␤ (␣ 2 /␤ 2 ) monomers or as heterodimers (discussed below) composed of two different receptor monomers (e.g. IR ␣ ␤/IGF-1R ␣ ␤). Upon ligand binding, they undergo a conformational change that facilitates binding to ATP and autophosphorylation (5,6). Autophosphorylation increases the kinase activity of IR-type receptors by 3 orders of magnitude, enabling them to phosphorylate a number of substrate proteins and engender growth or metabolic responses (7). IRS proteins act as mediators of insulin, IGF, and cytokine signaling in a variety of cell types. These adaptor molecules recruit and activate downstream signaling cascades such as the phosphatidylinositol 3-kinase and mitogen-activated protein kinase pathways (8). In addition to forming homodimers, IR and IGF-1R can form heterodimers with each other (9,10). However, it is still controversial whether these "hybrid" receptors subserve specific functions, e.g. by recruiting different substrates, especially because the phenotypes of the different receptor knock-outs are distinct, i.e. diabetes in IR knock-outs and dwarfism in IGF-1R knock-outs.
Whereas IGF-1 is a well established bone anabolic agent (11,12), the role of insulin in bone is uncertain despite the fact that humans and animals with diabetes frequently have defective bone formation and repair (13). Type 1 diabetes mellitus is associated with decreased bone density, an increased risk for osteoporosis (14) and fragility fracture (15), as well as poor bone healing and regeneration characteristics (16); conditions that all rely, in part, upon intramembranous bone formation. Quantitative histomorphometry on bone from diabetic animal models reveals a decreased osteoblast surface, osteoid surface, and mineral apposition rates (17), and a marked decline in the expression of the transcription factors that control osteoblast differentiation (18). In addition, recent studies have shown that diabetic rodents had impaired bone formation following bone injury compared with nondiabetic controls and that infusion of insulin into the distraction gap normalized bone formation in these models (19,20). However, defining the mechanisms of action of insulin and IGF-1 in bone has been difficult because of * This work was supported by a Veterans Affairs Merit Review grant. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  the potential for cross-talk between these two signaling pathways as described above.
To examine the direct actions of insulin in osteoblasts and elucidate signaling pathways downstream of the IR, we developed a model for conditional removal of the IGF-1 receptor in vitro by adenoviral introduction of the Cre-recombinase to primary mouse osteoblasts derived from mice carrying floxed IGF-1R alleles. We show that osteoblasts lacking the IGF-1R are four to five times more sensitive to insulin than are cells expressing both receptors. These findings provide the first unequivocal evidence for direct actions of insulin on osteoblasts and suggest that the strength of insulin generated signals in this cell type is tempered through interactions with IGF-1R.
Osteoblast Isolation and Culture-Osteoblasts were isolated from calvaria of newborn IGF-1R flox/flox mice by serial digestion in 1.8 mg/ml collagenase type I (Worthington, Lakewood, NJ) solution. Calvaria were digested in 10 ml of digestion solution for 15 min at 37°C with constant agitation. The digestion solution was collected, and digestion was repeated with fresh digestion solution an additional four times. Digestions 3-5 (containing the osteoblasts) were pooled together, centrifuged, washed with ␣MEM containing 10% FBS, 1% penicillin/streptomycin, and plated overnight at 37°C in a humidified incubator supplied with 5% CO 2 . For in vitro deletion of the IGF-1R, osteoblasts containing floxed IGF-1R alleles were cultured to be 70% confluent and then, in a small volume of PBS, were infected with adenovirus encoding Cre recombinase (Ad-Cre) (Vector Biolabs, Philadelphia, PA) at a titer of 100 multiplicity of infection (MOI). Infection with 100 MOI of adenovirus encoding green fluorescent protein (Ad-GFP) (Vector Biolabs) was used as control. After 1 h, culture medium containing 10% FBS was added directly to the cells in the presence of virus, and the cells were allowed to recover for the next 48 h. Greater than 90% IGF-1R deletion was confirmed for every infection by real time PCR. The Ad-GFP-infected (hereon referred to as control cells) or Ad-Cre-infected (hereon referred to as ⌬IGF-1R cells) osteoblasts were then replated on 100-mm tissue culture plates for immunoprecipitation and Western blotting experiments or on 6-well plates for the proliferation and mineralization assays.
Cell Lysis, Immunoprecipitation, and Western Blot Analysis-For signaling experiments, ⌬IGF-1R and control cells were cultured in 10% FBS ␣MEM to 90% confluence and then serumstarved in 0.1% bovine serum albumin for 24 h to reduce cellular activity to quiescent levels. At the end of the study, the cells were washed twice with ice-cold PBS and resuspended in lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1 mM MgCl 2 , 1 mM EDTA, 1% Triton X-100, and 10% glycerol). Protease and phosphatase inhibitors (Sigma) were added to the lysis buffer. The cell lysates were homogenized by rotating at 4°C for 30 min and then centrifuged at 14,000 rpm for 20 min at 4°C. The supernatant was transferred to a new microcentrifuge tube, and protein concentration was measured by Bradford protein assay (Bio-Rad). For immunoprecipitations, 1 g of IR, IGF-1R, IRS-1, or IRS-2 primary antibody (Santa Cruz Biotechnology) was added into 1 ml of cell lysate (500 g of total protein) and rotated at 4°C overnight. The protein complexes were immunoprecipitated by protein A-Sepharose beads. The immunocomplexes were collected by centrifugation, washed with immunoprecipitation buffer (150 mM NaCl, 1% Triton X-100, 50 mM Tris-HCl), dissolved in Laemmli sample buffer, and then analyzed by immunoblotting as described next. For immuno-blotting of whole cell lysates, equal amounts of protein (10 or 20 g/lane) were solubilized in Laemmli sample buffer and loaded onto a mini-SDS-PAGE system. Following electrophoresis, the proteins were transferred to a polyvinylidene difluoride membrane using a Bio-Rad wet transfer system. Protein transfer efficiency was verified using prestained protein markers. The membranes were then blocked with 5% nonfat dry milk for 1 h at room temperature and subsequently incubated overnight at 4°C with antibodies directed against the protein of interest. Signals were detected using a horseradish peroxidase-conjugated secondary antibody, and bound antibodies were visualized using the Supersignal West Dura Substrate (Pierce). Western blot photographic results were scanned with a Canon flatbed scanner. The relative levels of proteins of interest were then determined by measuring the density of the corresponding bands with ImageQuant software (Molecular Dynamics). All of the values were averages of at least three experiments and were normalized to the protein expression of a normalization protein.
Microarray Analysis-The ⌬IGF-1R osteoblasts were cultured in 6-well plates in ␣MEM with 10% FBS till near confluent. The cells were serum-starved in 0.05% FBS for 48 h and then restimulated with 100 nM insulin for 24 h. Total RNA was extracted and quantified from cells using the TRIzol method as recommended by the manufacturer (Invitrogen). Microarray quality biotin-labeled cRNA was synthesized using 3 g of RNA and subsequently purified as per TrueLabeling-AMP TM protocol recommended by manufacturer (SuperArray Bioscience Corp., Frederick, MD). 2 g of cRNA was then hybridized overnight with an insulin signaling pathway-specific gene expression array, and chemiluminescence signal was detected on photosensitive film as per Oligo GEArray system protocol recommendation (SuperArray). The raw images were analyzed using web-based GEArray Expression Analysis Suite software. The expression value for each gene on the array was calculated with the software and was verified by visual comparison with the autoradiographs. The usable linear range of pixel intensity was established as 9,000 -65,000 as described by others (21).
Osteoblast Proliferation Assay-The ⌬IGF-1R and control osteoblasts were plated in 6-well plates at low cell density (9 ϫ 10 4 cells/well of a 6-well plate) and cultured in ␣MEM with 1% FBS for 24 h to arrest the cells in G 0 phase. The cells were then restimulated with 10 nM insulin, 13 nM IGF-1, or 10% serum for the next 24 h. For proliferation analysis of the cells, 10 M BrdUrd was added to the medium 12 h before harvesting cells. The cells were then stained for anti-BrdUrd-APC and 7-aminoactinomycin D for proliferation and analyzed by FACSCalibur (Becton Dickinson), 20,000 events were collected for each sample, and the results were analyzed with WinMDI version 2.8.
Alkaline Phosphatase and von Kossa Staining-The ⌬IGF-1R and control osteoblasts were cultured in 6-well plates with 2.0 ϫ 10 5 cells/well in ␣MEM with 10% FBS for 4 days until they were confluent. The medium was then changed to osteogenic medium supplemented with 50 g/ml L-ascorbic acid and 5 mM ␤-glycerol phosphate to ␣MEM for next 15 days. The medium was changed every other day, and 10 or 100 nM insulin or 13 nM IGF-1 was supplemented with each medium change. Alkaline phosphatase and von Kossa histochemical staining was performed for 0-, 3-, 7-, and 14-day samples. The cells were washed with cold PBS, fixed in 2% paraformaldehyde/PBS for 10 min, and stored at 4°C in 100 mM cacodylic acid buffer (pH 7.4). For alkaline phosphatase staining, the fixed cells were incubated at 37°C with freshly prepared alkaline phosphatase substrate solution (100 mM Tris-maleate buffer, pH 8.4, 2.8% N,N-dimethyl formamide (v/v), 1 mg/ml Fast Red TR, and 0.5 mg/ml naphthol AS-MX phosphate) until red color developed. For von Kossa staining, the fixed cells were exposed to UV light in the presence of 3% silver nitrate (AgNO 3 ). The reaction was terminated by removing AgNO 3 solution and washing the monolayer carefully with water. Quantitative Real Time PCR-Total RNA was extracted from cells using the TRIzol method as recommended by the manufacturer (Invitrogen). The RNA concentration was estimated spectrophotometrically, and only pure RNA (A 260 :A 280 ratio Ն 1.8) was used for further analysis. First strand cDNA was synthesized using the iScript cDNA Synthesis Kit (Bio-Rad). The cDNA was amplified in the Opticon Continuous Fluorescent Detector (MJ Research, Waltham, MA) using IQ TM SYBR Green supermix (Bio-Rad) and sequence specific primers. PCRs were performed in triplicate for each cDNA, averaged, and normalized to endogenous ␤-actin reference transcripts. Primer sequences used were as follows: IGF-1R, F5Ј-TTGTGTTGTTCGTCCGGTGTG-3Ј, R5Ј-ATGTGCCC-AAGTGTGTGCG-3Ј; IR, F5Ј-TCAAGACCAGACCCGAA-GATT-3Ј, R5Ј-GTGATACCAGAGGATAGGAG-3Ј; ␤-actin, Statistical Analysis-All of the statistical analyses were performed using the Microsoft Excel data analysis program for Student's t test analysis with an assigned significance level of 0.05 (␣). All of the experiments were repeated at least three times unless otherwise stated. The values are expressed as the means Ϯ S.E.

Cre-mediated Disruption of IGF-1R and Hybrid Receptors in Primary Mouse
Osteoblasts-To selectively delete the IGF-1R, primary calvarial osteoblasts from mice carrying floxed IGF-1R alleles were infected with either adeno-GFP (Ad-GFP) or adeno-Cre (Ad-Cre) virus as described under "Experimental Procedures." As shown in Fig. 1A, osteoblasts infected with increasing loads of Ad-Cre expressed progressively less IGF-1R mRNA as compared with those infected with Ad-GFP. Infection with 100 MOI Ad-Cre virus decreased IGF-1R mRNA and protein expression more than 90% compared with 100 MOI Ad-GFP infected cells (Fig. 1, B and D). In the same samples, IR mRNA and protein expression was unaffected in Ad-Cre-infected cells compared with Ad-GFP-infected cells (Fig. 1, C and  E). Primary calvarial osteoblasts infected with Ad-Cre virus (thus lacking the IGF-1R) are hereafter referred to as ⌬IGF-1R, whereas Ad-GFP-infected cells (thus expressing IGF-1R) are indicated as control cells. To detect hybrid receptors in control and ⌬IGF-1R osteoblasts, the cell lysates were immunoprecipitated with either IR (Fig. 1F) or IGF-1R (Fig. 1G) antibodies and immunoblotted with both antibodies. IR/IGF-1R hybrid receptor were clearly detected in cells expressing both IR and IGF-1R but were greatly diminished in ⌬IGF-1R osteoblasts (Fig. 1, F and G). The apparent increase in abundance of the immunoreactive IR in the immunoprecipitate (Fig. 1F, bottom right) might reflect increased amounts of IR homodimers or alternatively indicates that the immunoprecipitating IR antibody preferentially recognizes homodimeric receptor.
Loss of the IGF-1R Enhances Sensitivity to Insulin-induced IR and IRS Phosphorylation-To investigate the effects of IGF-1R disruption on immediate IR signaling, we examined IR autophosphorylation in these cells. The ⌬IGF-1R and control cells were serum-starved, treated with insulin for 5 min, lysed, and immunoprecipitated with the anti-IR␤ antibody. Western blot analysis with an anti-Tyr(P) antibody showed that insulin stimulated tyrosine phosphorylation of the insulin receptor in both ⌬IGF-1R and control cells (Fig. 2, A and C). However, in the ⌬IGF-1R cells, IR was more responsive to insulin, leading to IR autophosphorylation at 100-fold lower insulin concentrations (0.1 nM) when compared with equivalent activation in control cells at the concentration of 10 nM.
We next investigated the effect of disruption of the IGF-1R on insulin and IGF-1 stimulation of IRS-1 and IRS-2. Under basal conditions, deletion of IGF-1R decreases phosphorylation of both IRS-1 and IRS-2. In the presence of IGF-1R, both insulin and IGF-1 induced a significant increase in tyrosine phosphorylation of IRS-1. However, in ⌬IGF-1R cells insulin-induced tyrosine phosphorylation of IRS-1, normalized to the untreated group, was more robust compared with insulin treatment in control cells (Fig.  3, A and B). As expected, IGF-1 did not induce IRS-1 phosphorylation in the ⌬IGF-1R cells. Basal IRS-2 activation was slightly diminished in the ⌬IGF-1R osteoblasts but was not significantly altered in control or mutant osteoblasts with insulin treatment. This suggests that insulin may preferentially act through IRS-1 in osteoblasts and induce a robust IRS-1 phosphorylation in the absence of IGF-1R.
Loss of the IGF-1R Enhances Sensitivity to Insulin-induced Akt and ERK Phosphorylation-Ligand binding to IR and IGF-1R activates two major signaling pathways: Akt and ERK. As shown in Fig. 4, IGF-1 acutely stimulated Akt (Fig. 4, A and  B, left panels) and ERK phosphorylation (Fig. 4, C and D, left panels) in control cells, whereas these effects were nearly abolished in ⌬IGF-1R cells. Interestingly, insulin treatment resulted in significantly greater induction of Akt and ERK phosphorylation in ⌬IGF-1R cells compared with control cells (Fig. 4, right  panels). The enhanced insulin sensitivity in ⌬IGF-1R cells is opposite of that observed for growth hormone signaling where removal of the IGF-1R diminishes growth hormone induction of JAK/STAT (Janus kinase/signal transducer and activator of transcription) phosphorylation (22).
Chemical Inhibition of IGF-1R Autophosphorylation Does Not Effect IR Signaling-As mentioned above, IR and IGF-1R can form hybrid receptors when co-expressed in the same cells, and the formation of such hybrid receptors might normally dampen insulin signaling. As a first step in studying this phenomenon in osteoblasts, we investigated whether insulin signaling is influenced by the presence of an IGF-1R that is defective in signaling. To accomplish this, we used PPP to selectively inhibit IGF-1R tyrosine autophosphorylation and hence subsequent downstream signaling (23). Wild-type osteoblasts were serum-starved for 20 h, then treated with PPP (10 M) or solvent Me 2 SO only for 4 h, and finally restimulated with IGF-1 (13 nM) or insulin (10 nM) for 5, 15, or 30 min. As expected, IGF-1R tyrosine phosphorylation was markedly decreased in the presence of PPP (Fig. 5, A, top panel, and B, left panel); however, the abundance of the insulin-IGF-1 hybrid receptor remained unchanged (Fig. 5, A, middle panel, and B, right  panel). IGF-1-induced Akt phosphorylation was markedly attenuated by treatment with the inhibitor (Fig. 5, C and D). By contrast, insulin-induced Akt phosphorylation remained unchanged in the presence of PPP when compared with control untreated cells (Fig. 5, C and D). (Note that PPP does not directly interfere with phosphorylation of Akt (24).) Insulin Modulates Global Gene Expression in Osteoblasts-As a first step in identifying genes regulated by insulin signaling in osteoblasts, we conducted a limited gene array analysis following acute exposure of ⌬IGF-1R osteoblasts to insulin. ⌬IGF-1R osteoblasts were serum-starved for 48 h and then restimulated with insulin for 24 h. The cells were then harvested and processed for insulin signaling pathway-specific microarray according to the manufacturer's protocol. Two replicate cultures were used, and each one was subjected to independent RNA isolation, amplified, and hybridized with the microarray filter. Gene expression changes were recorded on autoradiographic film for untreated (Fig. 6A) and insulintreated samples (Fig. 6B). Quantitative gene expression data were obtained from autoradiographic images using GEASuite Superarray software and was normalized using internal control genes glyceraldehyde-3-phosphate dehydrogenase, heat shock protein 1, peptidylprolyl isomerase A, ␤-2 microglobulin, and artificial sequences. The expression value obtained for each gene using the software was verified by visual comparison with the autoradiographs. Of 128 genes on the filter, 19 were significantly up-regulated (boundary value of 1.5), and seven genes were repressed (boundary value of 1.5) ( Table 1). A simple clustergram showing qualitative correlation of the genes in two groups is shown in Fig. 6C. The genes that were most up-regu-lated by insulin treatment included: Igf2 (insulin-like growth factor-2), Nos2 (iNOS) (inducible nitric-oxide synthase), Serpine1 (PAI-1) (plasminogen activator inhibitor-1), c-fos (FBJ osteosarcoma oncogene), c-jun (Jun oncogene), Slc2a1 (Glut1) (glucose transporter 1), Hk2 (hexokinase 2), and Vegfa (vascular endothelial growth factor-␣). The genes that were most repressed by insulin treatment included: PPAR␥ (peroxisome proliferator-activated receptor ␥) and Frs2 (fibroblast growth factor receptor substrate 2).
Insulin Increases Glut-1 mRNA Expression and Glucose Uptake in ⌬IGF-1R Osteoblasts-The induction of Glut 1 expression suggested that insulin might impact glucose transport in osteoblasts as it does in a variety of other cell types (25). Therefore, we next determined the effect of insulin on Glut 1 expression and glucose uptake in ⌬IGF-1R and control cells. Insulin increased Glut 1 mRNA in control cells by 3-fold and in ⌬IGF-1R cells by 7-fold at 12 h (Fig. 7A). Glucose uptake was not significantly altered in control osteoblasts but was induced by 1.6-fold in ⌬IGF-1R osteoblasts (Fig. 7B).
Disruption of IGF-1R Sensitizes Osteoblasts to the Proliferative Effects of Insulin-Insulin and IGF-1 are potent mitogens and survival factors in a number of cell types (26,27). To deter- mine whether the increased strength of insulin-generated signals in ⌬IGF-1R osteoblasts manifested in a greater biologic response, we determined the effect of removal of the IGF-1R on osteoblast proliferation by flow cytometric analysis. Removal of the IGF-1R did not significantly alter the serum-induced proliferative response (Fig. 8, A and B) but eliminated the BrdUrd incorporation in response to IGF-1 (Fig. 8C). By contrast, ⌬IGF-1R osteoblasts showed a much more robust proliferative response to insulin compared with those expressing IGF-1R (Fig. 8D). These changes were accompanied by increases in cyclin D1 protein expression (Fig. 8E), the cell cycle control protein known to be regulated by insulin (28) and Akt activation (Fig. 4) (29).
Partial Rescue of the Differentiation and Mineralization Defect in ⌬IGF-1R by Insulin-Our previous studies have shown that osteoblasts lacking the IGF-1R exhibit defective dif-ferentiation and mineralization in vivo (12). To investigate the possibility that insulin can rescue the differentiation and mineralization deficit, ⌬IGF-1R and control cells were cultured in the presence of ␤-glycerol phosphate and ascorbate (mineralization media) for 15 days in the presence or absence of IGF-1 or insulin. As shown in Fig. 9A, osteoblasts deficient in IGF-1R showed decreased alkaline phosphatase and calcified nodule staining. Treating these cells with insulin (10 or 100 nM) partially rescued the differentiation and mineralization defect as seen by enhanced alkaline phosphatase and calcified nodule staining, respectively. Similarly, in insulintreated ⌬IGF-1R cells, expression of Runx2 (Fig. 9B) and osteocalcin (Fig. 9C), markers for the differentiated osteoblast phenotype, were restored to equal or greater levels seen in control osteoblasts but not necessarily to the levels of control osteoblasts following insulin treatment. Our data suggest that insulin can act as an anabolic hormone in bone and may substitute for IGF-1 in case of diminished IGF-1R expression.

DISCUSSION
A number of previous studies have shown that insulin receptors are expressed in osteoblast-like cells (3,4,30) and have documented specific actions including effects on collagen synthesis (3,31,32), alkaline phosphatase production (33), and glucose uptake (34,35). However, because IR and IGF-1R can activate each others receptors (36,37) that are co-expressed in osteoblasts, it has been virtually impossible to distinguish the effects of insulin from those that likely occur by virtue of cross-activation of the IGF-1R. To complicate the matter further, a significant proportion of IR and IGF-1R present on a cell surface can form hybrid IR/IGF-1R receptors by combining the individual ␣-␤ heterodimers (10,38). This hybrid receptor can bind to both insulin and IGF-1 and has been suggested to generate signals similar to those arising from activation of IGF-1R (39). Therefore, the current studies, which use the Cre-loxP technique to remove IGF-1R in osteoblasts in vitro, has enabled us, for the first time, to examine insulin signaling in primary osteoblasts exclusively through its cognate receptor. These data show that insulin exerts direct actions in primary mouse osteoblasts but that the strength of IR signaling depends on the presence of the IGF-1R. were cultured in 6-well plates in ␣MEM with 10% FBS till near confluent. RNA was extracted, tagged with biotin, amplified, and hybridized with insulin signaling pathway specific microarray as explained under "Experimental Procedures." The autoradiograph images of the microarray are shown here as no treatment (A) and 24-h insulin treatment (B). For qualitative assessment, we indicate with red arrows a few genes that were up-regulated (Fos, Igf2, Nos2, Glut1, PAI-1, and Vegf-␣) or down-regulated (Frs2). The autoradiograph images were analyzed using web-based GEArray Expression Analysis Suite software. C, qualitative analysis of gene expression changes comparing the no treatment control (left lanes in all three sets) and insulin treatment group (right lanes in all three sets). A color code key for the magnitude of gene expression is shown at the bottom.
A striking observation from our studies was the increase in insulin responsivity in osteoblasts following deletion of the IGF-1R. The apparent inhibition of IR by IGF-1R does not require IGF-1R signaling because inhibition with a chemical that blocks IGF-1R receptor tyrosine phosphorylation (PPP) did not influence IR signaling. It is more likely that the presence of the IGF-1R normally dampens IR signaling through the formation of IGF-1R/IR hybrids. Indeed, previous studies have indicated that such hybrid receptors are less sensitive to insulin activation compared with IR homodimers (40). Interestingly, preadipocytes missing the IGF-1R also show enhanced sensitivity to insulin, which was associated with the loss of IGF-1R/IR heterodimers (37).
Insulin signaling through its homodimeric receptor in osteoblasts lacking the IGF-1R appears to promote more efficient coupling of the receptor to its distal signal transduction machinery. Such a mechanism might manifest at the level of the IRS proteins. In the ⌬IGF-1R osteoblasts, insulin increased IRS-1 tyrosine phosphorylation, whereas basal IRS-2 tyrosine phosphorylation levels were decreased but remained unaffected by treatment with insulin. This indicates that IR preferentially acti-   vates IRS-1. In agreement with this idea, other studies in brown adipocytes have reported that insulin activated only IRS-1 in IGF-1R null brown adipocytes (37,41). Although IRS-1 and IRS-2 both use the pleckstrin homology domain and the phosphotyrosine-binding domains to interact with the IR, IRS-2 uses an additional kinase regulatory loop-binding domain that may generate different downstream signals (42,43). IRS-1 and IRS-2 also differ in their cellular compartmentalization, and the activation kinetics of IRS-1-mediated signaling is more sustained, whereas IRS-2-mediated signaling is transient (44,45). The enhanced insulin receptor signaling in osteoblasts deficient in IGF-1R promotes greater activation of its major signal transduction pathways, Akt and mitogen-activated protein kinase. Given the preferential activation of IRS-1 by insulin in ⌬IGF-1R cells, it is likely that the increased Akt and ERK activation are linked with increased IRS-1 phosphorylation. However, additional studies are needed to establish the precise mechanisms responsible for the differing signaling strength in osteoblasts expressing different complements of IR and IGF-1R. Several lines of evidence including the new data presented here support the conclusion that insulin exerts specific actions in osteoblasts that are independent of IGF-1R. Thus, insulin increased proliferation, Glut-1, and glucose uptake in osteo-blasts lacking IGF-1R, confirming that insulin-generated signals are indeed coupled to function at the cellular level. The ability of insulin to increase proliferation and glucose metabolism is conserved across a number of species. For example, in Drosophila, insulin increases glucose oxidation (46) and knock-out of the insulin receptor homolog (Drosophila insulin receptor) causes growth retardation (47). Assessment of insulin actions in the mammalian skeleton has been difficult to approach experimentally because mice globally lacking IR die shortly after birth. Mice created with mosaic deletion of IR alleles survive for several weeks and exhibit extreme postnatal growth retardation, lipoatrophy, and hypoglycemia, a clinical constellation that resembles the human syndrome of Leprechaunism because of functional mutations of IR (48). However, these changes were accompanied with 60-fold increase in IGF-binding protein 1, which is known to bind circulating IGF-1 and limit its bioavailability. Therefore, although this model suggests that insulin influences skeletal growth, these effects are likely to be mediated in part through regulation of IGF-1 bioavailability. A recent study (49) showed that bone density measured at a single post-natal time point (6 months) was normal in euglycemic IR knock-out mice genetically reconstituted with IR through transgenic re-expression of the human IR transgene in the pancreas, liver, and brain. However, this mouse exhibited significant growth retardation in the early postnatal period in association with up-regulation of the IGF-1 receptor mRNA levels in osteoblasts. This suggests that increased IGF-1 receptor signaling might have compensated for the loss of the IR in bone. Indeed, our studies showing that insulin can partially rescue the mineralization defect and normalize the expression of osteoblast-specific genes in the IGF-1R null osteoblasts suggests that insulin exerts skeletal anabolic actions that overlap with those of IGF-1. Additional support for insulin as a bone anabolic factor is provided by Gandhi et al. (20), who reported that the delayed fracture healing in diabetic BB Wistar rats could be ameliorated by insulin delivery at the fracture site without affecting the systemic parameters of blood glucose. Taken together, these studies indicate that insulin is capable of exerting direct actions in mammalian bone that are distinct from those that might result through cross-activation of the IGF-1R.
The specific sets of genes modulated by insulin in osteoblasts provide intriguing leads for further interrogation of the mech- anism responsible for the bone anabolic activity of insulin. For example, the marked up-regulation of Vegf and iNOS suggests that insulin might activate angiogenic gene programs that are known to be important in skeletal development and turnover. It should be noted that these genes are known to be induced by insulin in other cell types through activation of hypoxia-inducible factor 1-␣ and aryl hydrocarbon nuclear translocator complex (50). Insulin also up-regulated c-fos expression as has been shown previously in H4IIE rat hepatoma cells (51). Finally, the ability of insulin to induce Glut-1 and hexokinase-2 is typical of insulin effects on metabolic processes in other responsive cell types (52,53). It is therefore possible that insulin promotes increased glucose uptake as a means to increase the metabolic activity of the osteoblast.
In summary, we provide evidence that insulin exerts direct anabolic actions in osteoblasts by activation of its cognate receptor and that the strength of insulin-generated signals is tempered through interactions with IGF-1R. We predict that the ability of IR and IGF-1R to form hybrid receptors controls, in part, the signaling mode between these two related receptors. Resistance to the anabolic activity of insulin in diabetic subjects might explain, in part, the bone disorders that are common in these patients.