Differential Roles of the Insulin and Insulin-like Growth Factor-I (IGF-I) Receptors in Response to Insulin and IGF-I*[boxs]

Insulin and insulin-like growth factor-I (IGF-I) receptors are highly homologous tyrosine kinase receptors that share many common steps in their signaling pathways and have ligands that can bind to either receptor with differing affinities. To define precisely the signaling specific to the insulin receptor (IR) or the IGF-I receptor, we have generated brown preadipocyte cell lines that lack either receptor (insulin receptor knockout (IRKO) or insulin-like growth factor receptor knockout (IGFRKO)). Control preadipocytes expressed fewer insulin receptors than IGF-I receptors (20,000 versus 60,000), but during differentiation, insulin receptor levels increased so that mature adipocytes expressed slightly more insulin receptors than IGF-I receptors (120,000 versus 100,000). In these cells, insulin stimulated IR homodimer phosphorylation, whereas IGF-I activated both IGF-I receptor homodimers and hybrid receptors. Insulin-stimulated IRS-1 phosphorylation was significantly impaired in IRKO cells but was surprisingly elevated in IGFRKO cells. IRS-2 phosphorylation was unchanged in either cell line upon insulin stimulation. IGF-I-dependent phosphorylation of IRS-1 and IRS-2 was ablated in IGFRKO cells but not in IRKO cells. In control cells, both insulin and IGF-I produced a dose-dependent increase in phosphorylated Akt and MAPK, although IGF-I elicited a stronger response at an equivalent dose. In IRKO cells, the insulin-dependent increase in phospho-Akt was completely abolished at the lowest dose and reached only 20% of the control stimulation at 10 nm. Most interestingly, the response to IGF-I was also impaired at low doses, suggesting that IR is required for both insulin- and IGF-I-dependent phosphorylation of Akt. Most surprisingly, insulin- or IGF-I-dependent phosphorylation of MAPK was unaltered in either receptor-deficient cell line. Taken together, these results indicate that the insulin and IGF-I receptors contribute distinct signals to common downstream components in response to both insulin and IGF-I.

Insulin and insulin-like growth factor-I (IGF-I) receptors are highly homologous tyrosine kinase receptors that share many common steps in their signaling pathways and have ligands that can bind to either receptor with differing affinities. To define precisely the signaling specific to the insulin receptor (IR) or the IGF-I receptor, we have generated brown preadipocyte cell lines that lack either receptor (insulin receptor knockout (IRKO) or insulin-like growth factor receptor knockout (IGFRKO)). Control preadipocytes expressed fewer insulin receptors than IGF-I receptors (20,000 versus 60,000), but during differentiation, insulin receptor levels increased so that mature adipocytes expressed slightly more insulin receptors than IGF-I receptors (120,000 versus 100,000).

In these cells, insulin stimulated IR homodimer phosphorylation, whereas IGF-I activated both IGF-I receptor homodimers and hybrid receptors. Insulin-stimulated IRS-1 phosphorylation was significantly impaired in IRKO cells but was surprisingly elevated in IGFRKO cells. IRS-2 phosphorylation was unchanged in either cell line upon insulin stimulation. IGF-I-dependent phosphorylation of IRS-1 and IRS-2 was ablated in IGFRKO cells but not in IRKO cells.
In control cells, both insulin and IGF-I produced a dosedependent increase in phosphorylated Akt and MAPK, although IGF-I elicited a stronger response at an equivalent dose. In IRKO cells, the insulin-dependent increase in phospho-Akt was completely abolished at the lowest dose and reached only 20% of the control stimulation at 10 nM. Most interestingly, the response to IGF-I was also impaired at low doses, suggesting that IR is required for both insulin-and IGF-I-dependent phosphorylation of Akt. Most surprisingly, insulin-or IGF-Idependent phosphorylation of MAPK was unaltered in either receptor-deficient cell line. Taken together, these results indicate that the insulin and IGF-I receptors contribute distinct signals to common downstream components in response to both insulin and IGF-I.
Insulin and insulin-like growth factors (IGF-I and IGF-II) 1 lead to an assortment of biological effects in insulin target cells such as adipocytes, hepatocytes, and myocytes. Insulin and IGF-I action is propagated by the insulin receptor (IR) and the IGF-I receptor (IGFR), which have similar heterodimeric ␣ 2 ␤ 2 structures and belong to the family of receptor tyrosine kinases (1). The analysis of the metabolic and mitogenic effects elicited by IR and IGFR in vivo is complicated by many factors. The two receptors are expressed on the surface of most cells, but their relative proportions vary in different tissues. Insulin and IGF-I are capable of binding to each other's receptors, although with a 100-fold lower affinity than that of its own cognate receptor (2,3). In addition, individual ␣␤ heterodimers from IR and IGF-IR can also combine to form disulfide-linked hybrid receptors, which can bind both insulin and IGF-I (4 -6). To complicate matters further, using similar mechanisms, the two receptors activate common intracellular pathways. Both receptors phosphorylate insulin receptor substrate (IRS) proteins on the same tyrosine residues (7)(8)(9)(10)(11). These IRS proteins then act as adaptor molecules to recruit and activate downstream signaling cascades such as the phosphatidylinositol 3-kinase and mitogen-activated protein kinase (MAPK) pathways (12)(13)(14)(15).
Despite this apparent overlap in receptor function, IR and IGFR are not functionally redundant molecules, as illustrated by the distinct phenotypes of IR-and IGFR-deficient mice. Mice lacking IR are born with very slight growth retardation, rapidly develop diabetic ketoacidosis, and die within a few days after birth (16,17). On the other hand, IGFR-deficient mice are severely growth-retarded, exhibit multiple growth-and differentiation-dependent abnormalities, and die shortly after birth, probably due to respiratory failure (18,19). These experiments suggest that IR is more important in fuel metabolism, whereas IGFR mediates growth and that one receptor cannot functionally compensate when the other receptor is absent.
Previous studies performed to distinguish the signaling properties of the receptors have focused on the overexpression of wild type or chimeric receptors in transfected cells (2,(21)(22)(23). The aberrant level of expression may distort the specificity of the receptors that would be seen with more physiological levels of receptor expression. Additionally, the cultured fibroblasts used in these studies normally do not express high levels of insulin or IGF-I receptors and, therefore, display a limited range of responses to insulin or IGF-I compared with physiologically relevant target tissues. Brown fat is a specialized form of adipose tissue involved in thermogenic energy regulation. Similar to white adipose tissue, brown fat is an insulin-sensitive endocrine organ that expresses both insulin and IGF-I receptors (24). Recently, a novel technique has evolved to generate brown adipose cell lines from genetically modified mice (25), allowing us to examine the role of insulin/IGF-I signaling molecules in adipocyte biology. In this study we have generated brown preadipocyte cell lines that lack either IR or IGFR and have compared insulin-versus IGF-I-induced adipogenesis and signaling in these receptor-deficient cells. We find that IR is required for brown adipocyte differentiation whereas IGFR is not, and each receptor activates downstream signaling molecules in distinct ways.
Cell Isolation and Culture-Cells that were homozygous for a floxed exon 4 of the insulin receptor (IR lox ) were used as controls for all studies. Cells were also derived from mice homozygous for a floxed allele of exon 3 of the IGF-I receptor (IGFR lox ). Brown preadipocytes were isolated from newborn control IR lox and IGFR lox mice by collagenase digestion as described previously (26). Preadipocytes were immortalized by infection with a pBABE retrovirus encoding SV40 T-antigen and selected with 2 g/ml puromycin. Cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% v/v fetal calf serum in a 5% CO 2 environment. For in vitro recombination of the insulin or IGF-I receptor, preadipocytes harboring a floxed allele of the insulin receptor (IR lox ) or IGF-I receptor (IGFR lox ) were first plated at a subconfluent density. After 24 h, cells were infected with an adenovirus encoding cre recombinase at a titer of 500 multiplicity of infection. After 1 h the viral supernatant was replaced with culture medium. Individual colonies were selected for IR or IGFR recombination using PCR with genomic DNA (27,28).
Adipocyte Differentiation and Oil Red O Staining-To differentiate cells, preadipocytes were first grown to confluence in culture medium containing 20 nM insulin and 1 nM triiodothyronine (differentiation medium). Adipogenesis was induced by treating confluent cultures (day 0) with differentiation medium supplemented with 0.5 mM isobutylmethylxanthine, 0.5 M dexamethasone, and 0.125 mM indomethacin (induction medium) for 48 h. After this induction phase (day 2), cells were returned to differentiation medium, which was replenished every 2 days until day 6 when cells were considered mature adipocytes. On day 6, representative dishes were washed once with phosphate-buffered saline and fixed with 10% buffered formalin for at least 16 h at 4°C. Cells were then stained for 4 h at room temperature with oil red O (5 g/liter in isopropyl alcohol), washed three times with water, and visualized. Differentiation studies were performed in four nonclonal control populations, one mixed population and nine clones of IRKO cells, and three IGFRKO clonal cell lines. Brown preadipocytes in culture have the capacity to differentiate beyond passage 30 without losing their adipogenic characteristics or showing signs of transformation (25). All cell lines in this study were used prior to passage 25 and have similar growth curves (data not shown), indicating that the mitogenic capacity of these cells was also unaltered.
Plasmids and Retroviral Infection-Generation of IR lox or IRKO cells overexpressing the insulin receptor was described previously (29). The coding sequence of human IGFR (provided by Eva Surmacz, Thomas Jefferson University, Philadelphia) was cloned into pBABE-bleo retroviral vector. Subconfluent NX packaging cells were transfected by LipofectAMINE TM Reagent (Invitrogen) with 10 g of retroviral vector, and viral supernatants were collected 48 h after transfection. Control or IGFRKO preadipocytes were infected with Polybrene (4 g/ml)-supplemented virus for 48 h and then placed in selection medium containing the bleomycin analog Zeocin (250 g/ml).
Ligand Binding and Competition Assays-125 I-Insulin or 125 I-IGF-I binding studies have been described previously (30) with the following modifications. Cell lines were grown to confluence and serum-starved overnight. Cells were washed three times with phosphate-buffered saline and incubated in 2 ml of HEPES binding buffer (HBB: 100 mM HEPES, pH 7.4, 118 mM NaCl, 5 mM KCl, 1.2 mM MgSO 4 , 8.8 mM dextrose, 1% bovine serum albumin) containing 0.01 nM 125 I-insulin or 125 I-IGF-I with increasing concentrations (0.1-1000 nM) of the respective unlabeled ligand for 2 h at 22°C. Binding was performed in triplicate for each dose. Cells were washed once with phosphate-buffered saline containing 1% bovine serum albumin, followed by two washes of phosphate-buffered saline. Cells were solubilized with 0.1 M NaOH containing 0.1% SDS, and radioactivity was measured using a ␥-counter. A duplicate set of plates was trypsinized, and cells were counted using a hemocytometer. The data were plotted as the ratio of bound ligand to free ligand versus the amount of ligand bound per total cell number. Linear regression of the curves (SigmaPlot) was performed to estimate the average binding affinity of the receptor-ligand interaction over the dose range examined. Competition for 125 I-insulin or 125 I-IGF-I was performed in cells overexpressing IR (IRKO ϩ IR) or IGFR (IG-FRKO ϩ IGFR), respectively. The data were plotted as % maximal binding of the uncompeted ligand.
Cell Lysis, Immunoblotting, and Immunoprecipitation-Cells were lysed in extraction buffer containing 50 mM HEPES, 137 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 10 mM NaP 2 O 7 , 10 mM NaF, 2 mM EDTA, 10% glycerol, 1% Triton X-100, 2 mM sodium orthovanadate, 10 g/ml trypsin inhibitor, 0.5 g/ml leupeptin, and 0.2 mM phenylmethylsulfonyl fluoride. Extracts were sonicated and clarified by centrifugation for 20 min, and protein concentrations were determined by the Bradford method (31). For immunoblotting of whole cell lysates, equal amounts of protein were directly solubilized in Laemmli sample buffer and resolved on SDS-containing polyacrylamide gels. Protein was transferred to polyvinylidene difluoride membranes and immunoblotted using the indicated primary antibodies. Blots were incubated with the appropriate peroxidase-conjugated secondary antibody and developed with enhanced chemiluminescence. For immunoprecipitations, 500 g of protein was incubated with the indicated antibodies and protein A-Sepharose for 16 h at 4°C. Antibodies directed against the ␤ subunit of IR or IGFR were used for their respective immunoprecipitations. Immune complexes were collected by centrifugation, washed in extraction buffer, solubilized in Laemmli sample buffer, and analyzed as above. Immunoblots for phospho-Akt and -MAPK were scanned on a Hewlett-Packard scanjet, and bands were quantified using ImageQuant software (Molecular Dynamics).

Insulin and IGF-I Receptors Are
Required for Brown Adipocyte Differentiation-To determine the role of IR and IGFR in adipogenesis, we have examined the differentiation capacity of brown preadipocyte cell lines that lack or overexpress each of these receptors created by homologous recombinant gene targeting (27,28). Cell lines containing the floxed allele of IR (IR lox ) or IGFR (IGFR lox ) served as controls. IR-deficient (IRKO) and IGFR-deficient (IGFRKO) cell lines were created by cre recombination of floxed cells in culture as described under "Experimental Procedures." Receptors were re-introduced into control and receptor-deficient cell lines by retroviral infection to create cell lines that overexpress either IR (IR lox ϩ IR; IRKOϩIR) or IGFR (IR lox ϩ IGFR; IGFRKO ϩ IGFR). We have shown that the insulin receptor that lacks exon 11 (IR A ) is the predominant isoform expressed in brown preadipocytes (29), and we have used this isoform for the following overexpression studies. Cells were induced to differentiate by the addition of isobutylmethylxanthine, dexamethasone, and indomethacin and were identified as mature adipocytes on day 6 of the differentiation program. As seen in Fig. 1, both control cell lines differentiated into adipocytes as indicated by oil red O staining of accumulated lipid. All 10 clonal populations of IRdeficient cells failed to differentiate indicating that this receptor was required for adipogenesis. Of the three clonal cell lines deficient for IGFR, one clone failed to differentiate, whereas the remaining two clones accumulated lipids normally (Fig. 2). We confirmed the absence of IGFR in the clones by immunoblotting with an IGFR-specific antibody on protein extracted during the days of differentiation. During adipogenesis, IGFR levels increased slightly in control cells at day 6 ( Fig. 2, top panel). As expected IGFR expression was not detectable in any of the IGFRKO clones. Similar to control cells, expression of peroxi-some proliferator-activated receptor ␥ and Glut4, two markers of adipocyte differentiation, were induced in IGFRKO clones 2 and 3 ( Fig. 2, middle panels). The fact that two IGFR-deficient clonal cells were capable of differentiation indicates that IGFR was not essential for this process. As we have shown previously (29), overexpression of IR failed to restore differentiation of IRKO cells, and overexpression of IR in control cells inhibited differentiation. In contrast, overexpression of IGFR in control cells did not inhibit adipogenesis (Fig. 1). This suggests that both the exact content and balance between these two receptors may be important in the adipogenic program.
Insulin and IGF-I Receptors Are Differentially Expressed in Brown Preadipocytes and Adipocytes-To understand the discordant effects of receptor overexpression, we examined the levels of IR and IGFR in differentiating adipocytes and in the various established cell lines. In control cells at the onset of differentiation (day 0), preadipocytes expressed very low levels of insulin receptor as determined by immunoblotting with an antibody specific to the ␤ subunit (Fig. 3A). During the course of differentiation, total IR level was increased and attained its highest level at day 6. Changes in IGFR expression, assessed by immunoblotting for its ␤ subunit, were less dramatic, although there was a slight increase at the later days of differentiation (Figs. 2 and 3A). Fig. 3B shows the relative protein levels of IR and IGFR in the genetically modified cell lines. Because the knockout cell lines cannot differentiate, we have focused on cells in the preadipocyte stage for further analyses. As expected, IR expression was absent in IRKO cells and easily detected in cells overexpressing IR. Similarly, IGFR was absent in IGFRKO cells but was highly elevated in cells overexpressing IGFR. There was no change in insulin receptor levels in IGFRKO cells, and IGFR levels in IRKO cells were similar to that seen in control cells. Thus, there was no receptor compensation when either receptor was absent.
To examine whether overexpression of either receptor would alter hybrid receptor formation, protein extracts from preadipocyte cell lines were immunoprecipitated with antibodies directed against the ␤ subunit of IR or IGFR, and immune complexes were immunoblotted with both IR and IGFR antibodies. Immunoprecipitation of IR followed by immunoblotting for IGFR revealed that the same amount of IGFR was associated with IR in control cells and cells overexpressing IR or IGFR (Fig. 3C, upper panels). Conversely, similar amounts of IR were co-precipitated with IGFR in control cells and cells overexpressing either IR or IGFR (Fig. 3C, lower panels). There were no hybrid receptors formed in IRKO or IGFRKO cells. These data suggest that a set population of hybrid receptors is produced, and overexpression of IR or IGFR does not significantly alter the formation of these hybrid receptors. Therefore, the signaling events observed in cells overexpressing these receptors can be attributed primarily to activation of IR or IGFR homoreceptors (see below). To determine more precisely the number of receptors on the cell surface, we performed insulin and IGF-I ligand binding assays on control preadipocytes and mature adipocytes, as well as on cells overexpressing IR or IGFR. Scatchard plots of ligand binding showed that insulin binding was low in preadipocytes ( Fig. 4A), which parallels the low expression of IR seen in immunoblots (see Fig. 3A). By extrapolation of the curve, we estimated that preadipocytes express about 20,000 insulin receptors on their cell surface (Table I). Following differentiation, adipocytes showed a dramatic increase in insulin binding (Fig.  4A), and this correlates with an approximate 6-fold increase in IR number to 120,000 receptors as estimated by Scatchard plot (Table I). In IRKO preadipocytes overexpressing insulin receptors (IRKO ϩ IR), insulin binding was similar to that seen in control adipocytes with a calculated receptor number of 100,000.
IGF-I binding was higher than insulin binding in control preadipocytes (compare the scale in Fig. 4B with Fig. 4A). This was reflected in a higher number of IGF receptors, 60,000 as estimated by Scatchard analysis, expressed on these cells (Table I). Following differentiation, IGF-I binding was slightly increased (Fig. 4B) with an increase in receptor number to 100,000 receptors per cell (Table I). This modest increase in IGFR number reflected the changes in expression observed by immunoblotting (see Fig. 3A). IGFRKO ϩ IGFR preadipocytes expressed 120,000 receptors, which was slightly greater than levels seen in control differentiated adipocytes. In both preadipocytes and adipocytes the average affinity of the IGF-I receptor was higher than that of the insulin receptor (1.26 Ϯ 0.20 versus 6.69 Ϯ 1.25 nM; Table I).
We next performed competition binding assays to assess the ability of each ligand to interact with the alternate receptor. In binding to the insulin receptor, the competition for binding of 125 I-insulin by insulin and IGF-I was similar, although IGF-I competition was displaced by an order of magnitude to the right (ED 50 ϭ 43.7 Ϯ 2.6 versus 3.4 Ϯ 0.1 nM; Fig. 5A). IGF-I was able to inhibit 125 I-IGF-I binding with an ED 50 of 0.55 Ϯ 0.11 nM, indicating high affinity receptors (Fig. 5B, open circles). In contrast, only high concentrations of insulin were able to displace IGF-I from IGFR (Fig. 5B, closed circles). However, there was a low, but detectable, level of binding of 125 I-insulin to IRKO cells; this represents the ability of insulin to bind IGF-I receptors (supplemental Fig. 1). Based on the low affinity of insulin for the IGFR, the activity of insulin to stimulate signaling was greater than one would predict based on binding (see below).

Insulin Stimulates IR Homodimer Phosphorylation, whereas IGF-I Stimulates IGFR Homodimers and Hybrid
Receptors-To analyze receptor activation by the different ligands, cells were stimulated with varying concentrations of insulin or IGF-I for 10 min, and protein extracts were immunoprecipi-tated with antibodies directed against the ␤ subunit of IR or IGFR and immunoblotted with a phosphotyrosine antibody. In control IR lox cells, insulin stimulated phosphorylation of IR at 10 and 100 nM (Fig. 6A). IGF-I induced phosphorylation of two bands that precipitated with anti-IR antiserum, one corresponding to the IR ␤ subunit (the faster migrating species) and the other corresponding to the IGFR ␤ subunit (the slower migrating species). The intensities of the two bands were equivalent, reflecting an equal molar ratio of receptor ␤ subunits, which indicates that this doublet results from the presence of hybrid receptors. As expected, neither insulin nor IGF-I could induce tyrosine phosphorylation of protein precipitated by an anti-IR antibody from IRKO cells. On the other hand, insulinstimulated phosphorylation of IR was slightly increased in IGFRKO cells, and in these cells, 100 nM IGF-I was able to weakly phosphorylate IR. In cells that overexpress IR, insulin greatly enhanced phosphotyrosine levels of IR because of the increased IR levels. In these same cells, IGF-I was also able to induce phosphorylation of IR and hybrid receptors as indicated by the doublet. In IGFRKO cells re-expressing IGFR, insulinand IGF-I-stimulated phosphorylation of IR or hybrids, respectively, were similar to that seen in control cells.
In control cells, insulin was only able to stimulate tyrosine phosphorylation of IGFR present in anti-IGFR precipitates at 100 nM (Fig. 6B), reflecting that some binding of insulin to the IGF-I receptors occurred at this high concentration of ligand. IGF-I was able to induce phosphorylation of both IGFR and IR in the complex; however, phosphorylation of IGFR was more intense than phosphorylation of IR, which indicates the presence of IGFR homoreceptors as well as the presence of hybrid receptors. In the absence of insulin receptors, only the high concentration of insulin was able to induce IGFR phosphorylation. IGF-I was only able to phosphorylate IGFR in IRKO cells, and the level of phosphorylation was similar to that seen in control cells for IGFR homodimers. Insulin and IGF-I failed to stimulate phosphorylation of IGFR immune complexes in IGFRKO cells. In cells overexpressing IR, insulin was able to weakly phosphorylate hybrid receptors. IGF-I was also able to stimulate phosphorylation of hybrids and IGFR homodimers in IRKO ϩ IR cells, although to a lesser extent than in control cells. In IGFRKO cells re-expressing IGFR, only 100 nM insulin was able to induce IGFR phosphorylation, and no phosphorylation of IR was detected. Because of the abundance of IGF receptors in these cells, IGF-I greatly enhanced phosphorylation of IGFR homodimers at all concentrations.

Insulin-and IGF-I-stimulated Phosphorylation of IRS-1 and IRS-2 Is Partially Dependent on Their Respective Receptors-
Insulin and IGF-I both phosphorylate and activate IRS proteins. To determine whether either ligand preferentially phosphorylates specific IRS proteins, we stimulated the receptordeficient and receptor overexpressing cell lines with insulin or IGF-I and examined tyrosine phosphorylation of IRS-1 and IRS-2, which are the most abundant IRS proteins in these cells (32). At the protein level, IRS-1 was elevated in IRKO cells relative to control cells (Fig. 7A). This increased expression was partially reversed in cells overexpressing IR. The total levels of IRS-1, on the other hand, were unaltered in IGFRKO cells either in the absence or presence of overexpressed IGFR. IRS-2 levels were similar in all cell lines.
In control cells, both insulin and IGF-I elicited a dose-dependent increase in tyrosine phosphorylation of IRS-1, although IGF-I was more potent than insulin in this effect (Fig.  7B). This may reflect the relative levels of IR and IGFR expressed on these cells (see Fig. 4), as well as the higher affinity of IGF-I as compared with insulin for their cognate receptors (see Table I). The tyrosine phosphorylation of IRS-1 at 1 nM IGF-I was similar to that seen with 10 nM insulin, reflecting these differences. Despite elevated IRS-1 protein levels, insulin-stimulated IRS-1 phosphorylation was reduced in IRKO cells, although the highest concentration was able to induce some phosphorylation of IRS-1, presumably by activating IGFR (see Fig. 6B). IGF-I-stimulated phosphorylation of IRS-1 was unaltered in IRKO cells.
In IGFRKO cells, insulin-induced phosphorylation of IRS-1 was enhanced, suggesting that IGFR may actually inhibit IR signaling to some extent. As expected, phosphorylation of IRS-1 by IGF-I was impaired in IGFRKO cells, although 100 nM IGF-I induced modest phosphorylation of IRS-1, possibly through cross-reactivity with IR. In cells overexpressing either IR or IGFR, both insulin and IGF-I were able to enhance phosphorylation of IRS-1. Increased IRS-1 phosphorylation in IRKO ϩ IR cells may be partially explained by the elevated IRS-1 levels.
In control cells, equivalent concentrations of insulin and IGF-I induced similar levels of IRS-2 phosphorylation (Fig. 7C). This ligand-induced phosphorylation was unaltered in IRKO cells, suggesting that insulin can signal fully through the IGFR to stimulate IRS-2. In IGFRKO cells, although insulin-stimulated phosphorylation of IRS-2 was not altered, IGF-I-induced phosphorylation of IRS-2 was dramatically reduced, indicating that IGFR was required for this response. Insulin-stimulated IRS-2 phosphorylation was elevated in cells overexpressing IR. However, the response to IGF-I was similar to control cells. This lends further evidence that IGF-I-dependent phosphorylation of IRS-2 required IGFR. Most surprisingly, phosphorylation of IRS-2 by insulin or IGF-I was inhibited in cells overexpressing IGFR, again suggesting some negative effect created by expression of IGFR.

IR and IGFR Deficiency Alters Phosphorylation of Akt but
Not MAPK-IRS proteins recruit additional adaptor proteins that initiate multiple signaling cascades. Akt and MAPK are two divergent kinases, which are downstream of IRS and are activated by both insulin and IGF-I. Therefore, we assayed the phosphorylation state of Akt and MAPK by using phosphospecific antibodies to these kinases. Total levels of Akt and MAPK protein were similar in all cell lines examined (data not shown). Stimulation with insulin resulted in a dose-dependent increase in Akt phosphorylation in control cells (Fig. 8A). This response to insulin was abolished in IRKO cells at 10 nM and significantly reduced at 100 nM. In IGFRKO cells, insulin-stimulated Akt phosphorylation was enhanced, although this did not reach statistical significance. However, these data suggest that IGFR may impair IR activation of Akt. Overexpression of IR greatly elevated insulin-induced Akt phosphorylation with a maximal stimulation at 10 nM, which was ϳ2-fold higher than that seen at 100 nM in control cells. Insulin-stimulated phosphorylation of Akt was also increased in cells overexpressing IGFR, although the dose response was shifted to the right compared with cells overexpressing IR.
IGF-I also induced a dose-dependent increase in Akt phosphorylation in control cells, with a maximal response at 10 nM (Fig. 8B). Note that IGF-I induced a greater phosphorylation of Akt than insulin, and this may reflect the fact that there are more IGF-I receptors in these cells compared with insulin receptors (see Table I). Most surprisingly IGF-I-stimulated Akt phosphorylation was impaired in IRKO cells at 1 and 10 nM, although at the latter concentration this was not statistically significant. This occurred despite the fact that IGFR levels are unaltered in these cells. In IGFRKO cells, phosphorylation of Akt by IGF-I was reduced, although this was only significant at 1 nM. Because IGF-I-dependent phosphorylation of Akt at 1 nM required both IGF-I and insulin receptors, we presume that this response occurred through a hybrid receptor. IGF-I induced Akt phosphorylation was unaltered in cells overexpressing IR. However, in cells overexpressing IGFR, Akt phosphorylation was increased with 1 nM IGF-I but unaltered at 10 -100 nM, presumably because the response at 10 nM had already reached maximal stimulation.
Phosphorylation of MAPK by insulin or IGF-I followed a different pattern than Akt. In control cells insulin induced a dose-dependent increase in MAPK phosphorylation (Fig. 8C). In IRKO cells, this response was slightly, although not significantly, blunted, indicating that insulin could signal through IGFR to stimulate MAPK. The dose response in IGFRKO cells was unaltered. As would be expected, insulin-stimulated MAPK phosphorylation was elevated in cells overexpressing IR with all doses eliciting an ϳ2-fold higher response. Most surprisingly, overexpression of IGFR impaired insulin-stimulated phosphorylation of MAPK, providing additional evidence that IGFR may inhibit IR signaling.
IGF-I also elicited MAPK phosphorylation in a dose-dependent manner in control cells (Fig. 8D). This response was only slightly elevated compared with insulin stimulation. IGF-Istimulated MAPK phosphorylation was unchanged in either IRKO or IGFRKO cells, suggesting that the presence of either receptor was adequate for this signal. Similarly, overexpression of either IR or IGFR did not effect MAPK phosphorylation by IGF-I. DISCUSSION Insulin and IGF-I receptors are structurally similar and activate many of the same signaling cascades. When IR and IGFR are expressed in the same cells, it is difficult to separate the actions of insulin and IGF-I because these ligands can bind either receptor, although with a lower affinity than their cognate receptor. To complicate matters further, individual ␣␤ heterodimers from IR and IGFR can combine to form hybrid receptors, which mostly act as IGF-I receptors but can also bind insulin (4,6). To isolate the individual contributions of IR and IGFR to signaling and eliminate hybrid receptor formation, we have established brown preadipocyte cell lines lacking either receptor, IRKO or IGFRKO. Previous reports have focused on the actions of one ligand on one receptor-deficient cell line, i.e. IGF-I on IR-deficient fibroblasts (33) or insulin on IGFR-deficient cells (34,35). This current study has examined the actions of both insulin and IGF-I on IR-and IGFR-deficient cell lines that were established by similar methods.
Previously, our laboratory has developed brown preadipocyte FIG. 7. Insulin-and IGF-I-stimulated phosphorylation of IRS-1 and IRS-2. A, total cell lysates from unstimulated preadipocyte cell lines were analyzed for IRS-1 and IRS-2 protein expression. Alternatively, preadipocyte cell lines were grown to confluence, serum-starved overnight, and stimulated with the indicated doses of insulin or IGF-I for 10 min. Lysates were subjected to immunoprecipitation (IP) with antibodies against IRS-1 (B) or IRS-2 (C). Immune complexes were then immunoblotted (IB) with an antibody directed against phosphotyrosine (PY). cell lines from mice with targeted disruptions of genes involved in the insulin/IGF-I signaling system to help define functional differences among structurally similar molecules. For example, brown preadipocytes lacking IRS-1 failed to differentiate into mature adipocytes and had impaired insulin-stimulated lipid synthesis (36,37). IRS-2-deficient adipocytes showed only modest defects in adipogenesis, but Glut4 translocation and glucose uptake were reduced (38). Recently (29), we have shown that the insulin receptor was required for differentiation of brown preadipocytes in vitro. In this current study, we have shown that IGFR is not required for adipogenesis of brown preadipocytes, which is unexpected considering that preadipocytes express more IGFR than IR. However, insulin-stimulated expression of various adipogenic markers is reduced in brown adipocytes derived from IGFR-null mice (35). These data suggest that both IR and IGFR have independent roles in adipogenesis that cannot be compensated for by signaling through the other receptor. Most surprisingly, when overexpressed to similar levels in control preadipocytes (ϳ100,000 receptors/ cell), the IR inhibits differentiation whereas IGFR does not, providing additional evidence that different signals may emanate from the receptors to effect adipogenesis.
A similar example of the individual requirements for IR and IGFR in differentiation can also be seen in skin development. Primary keratinocytes normally express both insulin and IGF-I receptors and respond to either ligand during differentiation (39). However, keratinocytes from IRKO mice fail to differen-tiate as measured by the absence of keratin markers K1 and K10 (40). This occurs despite elevated autophosphorylation of IGFR. Additionally, there appear to be defects in keratinocyte development in IGFR-null mice as evidenced by increased transparency and reduced cellularity of the skin (18).
In preadipocytes, there are roughly 3-fold more IGFR than IR, making IGFR the quantitatively predominant receptor in these cells. Therefore, it is not surprising that IGF-I is more potent than insulin in stimulating downstream targets such as IRS-1 and Akt. Even so, IGF-I receptors alone are not sufficient for brown fat differentiation or the activation of some biological targets. Because the interactions of these receptors with their ligands differ with the concentration of the ligand due to crossreceptor binding at high doses, we have designed a simplified model of IR and IGFR signaling by focusing on events elicited by physiological levels of ligand (i.e. 1-10 nM). One caveat to the interpretation of our results is that other factors not assessed here could effect signaling. Such possible differences include the time course of stimulation in different cell types or with different ligands, differences in the association or dissociation rates of the ligands, or the internalization kinetics of the receptor. Because of these variables, we have focused on the ability of insulin or IGF-I to activate signals in the receptordeficient cell lines instead of the intensity of the signals evoked.
In these cells, insulin primarily binds to and activates the insulin receptor but does not appear to activate hybrid receptors (Fig. 9A, thick arrows). The phosphorylation of both IRS-1 and Akt is dependent on the interaction of insulin with its cognate receptor. On the other hand, insulin-stimulated phosphorylation of IRS-2 and MAPK occurs in the absence of insulin receptors, indicating that insulin binding to IGFR can support the generation of these signals (Fig. 9A, dashed arrows).
IGF-I binds to and activates IGFR homodimers (Fig. 9B, thick arrows), as well as hybrid receptors (thin arrows). Unlike insulin, IGF-I stimulation of both IRS-1 and IRS-2 requires the presence of IGFR, although our data do not preclude the possibility that hybrid receptors may also be involved in these events. However, it is clear that hybrid receptors are essential for IGF-I activation of Akt because this phosphorylation does not occur when either IGFR or IR is absent (Fig. 9B, thin  arrows). IGF-I-stimulated MAPK phosphorylation in either receptor-deficient cell line is equivalent to that seen in wild type preadipocytes, suggesting that IGF-I may use either receptor with equal efficacy to activate this pathway.
It is difficult to ascertain if Akt or MAPK is directly linked to one specific IRS protein because both insulin and IGF-I are capable of phosphorylating IRS-1 and IRS-2. One inference that can be drawn, however, is that in IRKO cells physiological concentrations of insulin are capable of stimulating phosphorylation of IRS-2 and MAPK but not IRS-1 or Akt. Additionally, IGF-I, even at high concentrations, is not capable of stimulating phosphorylation of IRS-2 in IGFRKO cells, although stimulation of Akt phosphorylation does occur. These data suggest that stimulation of Akt by either ligand is more tightly linked to IRS-1 phosphorylation, whereas MAPK is downstream of either IRS protein or perhaps another adaptor molecule not examined here.
By dissecting signaling patterns in the receptor-deficient cell lines, we can assess the level of cross-receptor activation that occurs in these biologically responsive cells. At high concentrations, IGF-I is able to act through the IR in IGFRKO preadipocytes, and these signaling responses parallel the competitive binding capacity of IGF-I for the insulin receptor. By contrast, when insulin acts through the IGFR to stimulate downstream signals in IRKO cells, the responses are greater than one would predict based on its low binding affinity. Thus, insulin acting through the IGFR has a higher intrinsic activity than IGF-I itself bound to this receptor. This has also been observed in L6 myotubes where insulin stimulation of IGFR phosphorylation can be detected at concentrations of insulin where IGFR occupancy is not sufficient to inhibit labeled IGF-I binding (41). Additionally, Lamothe et al. (33) have shown that IR-deficient fibroblasts are still responsive to insulin at high doses acting through IGFR. Therefore, in the absence of one receptor, the other can act as an alternate receptor to mediate the biological effects of either ligand.
In the IRKO preadipocytes in which we have re-expressed IR, insulin receptor expression reaches levels normally seen in mature adipocytes, i.e. roughly 5-fold higher than levels in preadipocyte cells. We have previously hypothesized that this high level of IR expression in the preadipocyte stage may disrupt the balance of IR and IGFR signaling that may be required for normal adipogenesis (29). In these cells, hybrid receptor formation is unaltered, indicating that the essential signals emanate from IR or IGFR homoreceptors. Compared with preadipocytes that overexpress IR, the increase in IGFR number is modest in IGFRKO ϩ IGFR cells, but the IGF-I receptor level is still twice that seen in control preadipocytes. This moderate increase in IGFR levels may also upset the equilibrium of IR and IGFR signaling. Overexpression of IR in IRKO cells inhibits phosphorylation of IGFR homodimers, whereas overexpression of IGFR in IGFRKO cells enhances IGFR homodimer signaling to the detriment of hybrid phosphorylation. Although insulin and IGF-I may activate both IR and IGFR under certain conditions, it appears that cells must maintain a tightly coordinated balance of receptors to facilitate biological responses such as differentiation.
In IRKO cells IRS-1 levels are elevated relative to control cells, and this may reflect the role of insulin in maintaining normal IRS-1 levels and allow for some compensation of signaling in these cells. In support of this, re-expression of IR in IRKO cells partially reduces IRS-1 levels. By contrast, we did not detect any change in IRS-1 levels in IGFRKO cells. This directly contradicts reports by Mur et al. (34,35) where IRS-1 levels were reduced in their model of IGFR deficiency. One explanation for this difference may reside in the cell models used in these studies. We have focused on preadipocytes, while the previous studies were performed in adipocytes, and the differentiation stage of the cells may influence IRS-1 abundance. In fact during differentiation of wild type brown adipocytes, IRS-1 levels are high at the onset but decline as cells become mature adipocytes (32). Although there is no difference in IRS-1 levels in IGFRKO cells, insulin-stimulated IRS-1 tyrosine phosphorylation is elevated, indicating that IGFR may inhibit IR signaling. In addition, insulin-stimulated Akt phosphorylation is also elevated in IGFRKO cells, although this did not reach statistical significance. MAPK phosphorylation by insulin is inhibited in cells overexpressing IGFR. Therefore, the presence of IGFR may negatively correlate with insulin sensitivity in these cells.
Although brown fat is essential for thermoregulation and heat production, the role for brown adipose tissue in whole animal glucose homeostasis has been defined recently. Transgenic ablation of brown fat in mice results in obesity with concurrent glucose and insulin intolerance (42,43). Despite the localized tissue ablation, these mice also develop insulin resistance in both muscle and white adipose tissue. A similar phenotype is seen in mice with a knockout of the insulin receptor specifically in brown adipose tissue (20). Whereas these mice are initially born with brown fat, there is an accelerated agedependent decrease in brown fat mass, which results from a reduction in both cell size and lipid accumulation. A majority of these mice develop diabetes with fasting hyperglycemia as a result of impaired insulin secretion. Therefore, the role of the insulin receptor in an isolated tissue such as brown fat can directly impact whole animal physiology.
In summary, although insulin and IGF-I have been shown to activate similar signaling cascades, IR-and IGFR-mediated signalings are not functionally redundant. In the current example of brown adipocyte differentiation, deficiency of IR, which is expressed at low levels, limits adipogenesis, whereas deletion of the IGFR, which is relatively abundant, does not, suggesting that these homologous receptors have different roles in adipocyte biology. Similarly overexpressing one receptor would presumably increase the overall level of tyrosine kinase activity, which in turn would lead to an elevated signaling capacity. It appears that total receptor number is not as important as the ratio of IR and IGFR levels in a given cell. In the fat cell, mechanisms have evolved to regulate the number and ratio of insulin and IGF-I receptors and to facilitate proper signal output.