The Subunit Structures of Two Distinct Receptors for Insulin-like Growth Factors I and I1 and Their Relationship to the Insulin Receptor*

Intaul cells and membranes isolated from several human and rodent tissues have been affinity-cross- linked to human ‘261-insulin-like growth factor I (IGF-I) and ‘261-insulin-like growth factor I1 (IGF-11). Dodecyl sulfate-polyacryalmide gel electrophoresis of the affin-ity-labeled material resolves two types (type I and type II) of labeled membrane components that fulfill the properties expected for high affinity growth factor receptors. Type I receptors consist of three disulfide- linked forms (&& = 350,000, 320,000 and 290,000) structurally similar to the insulin receptor forms in mem- brane preparations from various tissues (Massague, J., Pilch, P. F., and Czech, M. P. (1980) h c . NutZ. A c d Sei U. S. A. 77,7137-7141). The proposed subunit stoichiometries of these type I IGF receptor forms are (p-S-S-a)-S-S-(a-S-S-b), (pl-S-S-a)-S-S-(a-S-S-b), and (pl-S- S-a)-S-S-(a-S-S-bl), respectively, based on two-dimen-sional electrophoretic analysis. The a and /3 subunits migrate with apparent M, = 130,000 and 98,000, respectively, and the b1 subunit is a proteolytic fragment of the fl subunit. The disulfide-linked M, = 350,000 receptor species probably represents

ssague, J., Guillette, B. J., and Czech, M. P. ( 1 9 8 1 ) J Biol. Chem 256, 2122-2125). The ability of IGF-I rihd IGF-I1 to compete with '251-insulin for affinity labeling the high affinity insulin receptor in all tissues examined is lower than that of insulin. In conclusion, we have identified a specific growth factor receptor with high affinity for IGF-I that consists of a heterotetrameric disulfide-linked subunit composition virtually identical with the insulin receptor structure. A second growth factor receptor with high affinity for IGF-I1 is structurally distinct from these disulfide-linked receptor types. Insulin-like growth factors I and I1 from human serum (1) are two closely related peptides with amino acid sequences considerably homologous to that of insulin (2, 3). IGF-I' and IGF-I1 stimulate sugar and amino acid transport, sulfation of rat costal cartilage, and macromolecule synthesis in several tissues (4). They are also potent mitogens for certain cell types (4). Many biological properties of IGF-I and IGF-I1 are also shared by somatomedin C (5, 6), somatomedin A (7, 8), and rat multiplication-stimulating activity (9). Recent amino acidsequencing studies have indicated that rat somatomedin C 2 and multiplication-stimulating activity (10) are the rat counterparts of human IGF-I and IGF-11, respectively.
Like insulin, the primary event in the action of IGF-I and IGF-I1 on target cells appears to be their binding to specific receptors on the cell surface. A basic question concerning the mechanism of action of these three structurally related polypeptide hormones is whether their respective receptors share also common structural and biological properties. Radioligand binding studies in several laboratories (4, 7, 11) have shown that the binding of IGFs to various cell and membrane systems is rather complex and suggests a considerable heterogeneity in IGF receptor forms and distribution among tissues. Studies of this kind have also indicated that the IGF-binding sites in some tissues exhibit a certain degree of affinity for insulin, and the high affinity insulin receptor can specifically interact with IGF-I and IGF-I1 ( 4 , l l ) . Until recently, direct structural information clarifying the nature and distribution of the receptor(s) for IGF-I and IGF-I1 had been lacking. In a preliminary report, we have described the characteristics of a receptor structure affinity-labeled by cross-linking to 1251-multiplication-stimulating activity (12). We report here the structural characteristics of two types of affinity-labeled membrane components that exhibit high affinity and specificity for IGF-I and IGF-11. The relative affinities of these two putative receptors for IGF-I and IGF-11, their subunit composition, tissue distribution, and relationship to other known receptor structures are also described.

Insulin-like Growth
with a tight fitting Dounce homogenizer. The homogenate was centrifuged a t 3,000 X g for 10 min and the resulting supernatant was centrifuged a t 30,000 X g for 20 min. The material pelleted in the latter centrifugation was resuspended in 0.25 M sucrose, 10 mM Tris, 1 mM EDTA, pH 7.4, layered on top of a 20-50% (w/w) sucrose gradient and centrifuged at 100,000 X g for 60 min. The membranous fraction that equilibrated a t 35-37% (w/w) sucrose was pelleted a t 30,000 x g for 30 min after dilution with 10 mM Tris, 1 mM EDTA, pH 7.4, and used in the affinity-labeling experiments. Protease inhibitors 1 mM phenlymethylsulfonyl fluoride and 0.1 mg/ml soybean trypsin inhibitor were occasionally used during preparation of membranes from these tissues with no significant effect on the pattern of specifically aftiiity-labeled membrane components. Rat H-35 hepatoma cells (gift of Dr. Gerald Litwack, Temple University) were grown a t 37 "C in Dulbecco's modified Eagle's medium supplemented with 10% calf serum. A membrane fraction was obtained from H-35 hepatoma cells by homogenization of mechanically detached cells in the presence of 0.25 M sucrose, 10 mM Tris, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, pH 7.2. The homogenate was centrifuged a t 1,800 X g for 10 min and the resulting supernatant was centrifuged at 30,000 X g for 30 min. The resulting pellet was resuspended in homogenization medium, layered on top of a 20-40% (w/w) sucrose gradient and centrifuged a t 100,000 X g for 60 min. The fraction equilibrating at 31-32% sucrose was collected and used in the affinity-labeling experiments. Mouse 3T3-LI fibroblasts (American Type Culture Collection) were grown a t 37 "C in Dulbecco's modified Eagle's medium, supplemented with 10% fetal calf serum. Confluent 3T3-Ll fibroblasts were differentiated to adipocytes (16) by treatment with 0.5 m~ I-methyl-3-isobutylxanthine and 0.40 CM dexamethasone as described before (17). At least 70% of the cells expressed the adipocyte phenotype when they were used 7 days after treatment with drugs. A membrane preparation was obtained from 3T3-LI cells as described for H-35 cells.
Human IM-9 lymphocytes (gift of Dr. Jerrold Olefsky, University of Colorado) were grown a t 37 "C in RPMI 1640 tissue culture medium supplemented with 5% fetal bovine calf serum. Human RPMI 6666 and RPMI 7666 lymphoblasts (American Type Culture Collection) were grown a t 37 "C in RPMI 1640 medium supplemented with 20% fetal calf serum. Human osteogenic Sarcoma cells (TE85, clone F-5) (American Type Culture Collection) were grown a t 37 "C in Eagle's minimal essential medium and Earle's balanced salt solution supplemented with 10% fetal calf serum. Human A875 melanoma cells (gift of Dr. Gary Johnson, University of Massachusetts) were grown a t 37 "C in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Rat lymphocytes were obtained by trimming of mesenteric lymph nodes isolated from male 150-200-g animals. The nodes were gently teased between ground glass slides to yield a singlecell suspension. Rat lymphocytes were incubated a t 37 "C for 20 min in the presence of 30% (w/v) glycerol, pelleted at 2,000 X g for 10 min and lysed at 4 "C with 10 mM Tris, 1 mM CaC12, 1 mM MgCI?, pH 7.4. The cell lysate was centrifuged at 2,000 X g for 10 min and a crude membrane fraction was obtained by centrifugation of the resulting supernatant at 30,000 X g for 30 min. A plasma membrane-enriched microsomal fraction was prepared from rabbit superior cervical ganglia (Pel-Freez, Rogers, AR) by the method of Banerjee et al. (18).
Affinity-labeling Protocols-Membranes (150-200pg of membrane protein) in 200 pl of Krebs-Ringer phosphate buffer, pH 7.4, were incubated at 10 "C for 90 min with '"I-labeled hormones in the presence or absence of nonradioactive hormones at the indicated final concentrations. At the end of this incubation, the excess unbound hormone was washed out by dilution with ice-cold Krebs-Ringer phosphate buffer, pH 7.4, and centrifugation a t 12,000 X g for 3 min. The membrane samples were resuspended in the latter medium, and disuccinimidyl suberate freshly dissolved in dimethyl sulfoxide was added a t final 0.20 mM concentration under the conditions previously described (19). The cross-linking reaction was allowed to proceed for 15 min a t 0 "C before it was terminated by quenching the unreacted disuccinimidyl suberate with an excess Tris buffer, pH 7.4. The incubation mixture was Centrifuged at 12,000 X g for 5 min and the pellet was collected for electrophoresis.
Intact cells resuspended in Krebs-Ringer phosphate buffer, pH 7.4, with 4% bovine serum albumin were affinity-labeled using a protocol similar to that used for affinity-labeling isolated membranes. Incubation with '251-labeled and unlabeled hormones was conducted at 23 OC for 60 min, and the cross-linking reaction was done at 15 "C for 15 min in the presence of 0.40 m~ disuccinimidyl suberate. The treatment with the cross-linking agent was terminated by diluting and washing the cell sample with 0.25 M sucrose, 10 mM Tris, 1 mM EDTA, pH 7.4. After this treatment, adipocytes, fibroblasts, Sarcoma, melanoma, and hepatoma cells were homogenized in the latter buffer and a crude membrane fraction was obtained by centrifugation of the resulting homogenate a t 30,000 X g for 20 min, after removing most of the nuclei, mitochondria, and other heavy fractions by centrifugation at 3,000 X g for 20 min. Following affinity-labeling, the lymphocyte-type cells were directly solubilized in the presence of 2% sodium dodecyl sulfate, 50 m~ Tris, pH 6.8, and centrifuged a t 100,000 X g for 20 min, and the supernatant was kept for gel electrophoresis.
Electrophoresis a n d Autoradiography-The cross-linked samples were boiled for 1 rnin in the presence of I% sodium dodecyl sulfate, 50 mM Tris, pH 6.8, with or without 50 mM dithiothreitol, and subjected to electrophoresis on polyacrylamide gels according to Laemmli (20). Gels were 5% polyacrylamide (100:l acrylamide/bisacrylamide ratio) unless otherwise indicated. After electrophoresis, gels were stained for protein, dried, and subjected to autoradiography as described before (12).  IGF-I1 (b) at 10 "C for 60 min in Krebs-Ringer phosphate buffer, pH 7.4, containing 1% bovine serum albumin. After washing out the unbound hormone, membranes were incubated with 0.2 mM disuccinimidyl suberate at 0 "C for 15 min, and the cross-linking reaction was stopped by addition of excess Tris buffer, pH 7.4. Samples (100 pg of membrane protein) of the cross-linked membranes were electrophoresed on 5% polyacrylamide slab gels according to Laemmli (20), in the absence (lanes 1 to 4) or presence (lanes 5 to 8) of 50 mM dithiothreitol (DTT). Autoradiograms from the fixed, dried gels are shown. Two distinct types (types I and 11) of affinity-labeled bands were revealed on the autoradiograms as described in the text.

Insulin-like Growth Factor Receptor Structures
Mr = 350,000,320,000, and 290,000, respectively, in the absence of dithiothreitol. This group of affinity-labeled species was present in human placenta and skin fibroblast membranes, but not in rat liver or adipocyte membranes. The M, = 350,000, 320,000, and 290,000 membrane components affinity-labeled by '251-IGF-I were not observed on the gels when electrophoresis was performed in the presence of the disulfide reducing agent, dithiothreitol ( Fig. la, lanes 7 and 8). Instead, an intensely labeled band appeared in the M, = 130,000 region of these gels. This observation indicates that the M , = 350,000, 320,000, and 290,000 membrane components affinity-labeled by lZ51-IGF-I consist of disulfide-linked complexes that apparently contain a M, = 130,000 subunit which carries most of the cross-linked label.
The characteristics of these type I membrane species affnity-labeled by 1251-IGF-I are strikingly similar to those of the insulin receptors forms identified in membranes from various tissues and animal species by affinity-labeling with '251-insulin (14,21). Insulin receptors are also found in plasma membrane preparations as three distinct M, = 350,000, 320,00G, and 290,000 receptor forms. The native M, = 350,000 insulin receptor form consists of two a (MI = 125,000) receptor subunits and two fl (M, = 90,000) receptor subunits, all disulfide-linked in a proposed (p-S-S-a)-S-S-(a-S-S-~) symmetrical structure (21). A unique site at about the center of the j3 receptor subunit amino acid sequence is particularly sensitive to lysosomal protease and to elastase (14). Proteolysis at this site converts the p subunit to a PI fragment that remains disulfide- or PI receptor subunits (14,21). This dissociation process is concomitant with an increase of the apparent M , of the receptor subunits, probably by reduction of intrachain disulfides (21). These structural features are apparently shared by the type I membrane components affinity-labeled by Iz5I-IGF-I ( Fig. 1 and see below). The second type (type 11) of membrane components affnity-labeled by Iz5I-IGF-I consists of a single species in membranes from all tissues examined in this experiment ( Fig. l a ,   lanes 1 to 4). The apparent Mr of this labeled species varied somewhat among tissues. It migrated at apparent M, = 234,000 in rat adipocyte membranes, M, = 222,000 in rat liver membranes, and M, = 214,000 in membranes from human skin fibroblasts and human placenta. During electrophoresis in the presence of dithiothreitol, this labeled species migrated with an apparent M, that ranged between 268,000 and 258,000 depending on the tissues (Fig. la, lanes 5-8). This increase in apparent M, upon addition of dithiothreitol may reflect the presence of intrapeptide disulfide bonds that compact the structure of this membrane component. This effect of dithiothreitol has also been observed on bovine serum albumin known to contain several intrachain disulfide bonds (22), and on membrane receptors for insulin (21) and nerve growth factor (23). Fig. l b shows an autoradiogram obtained after dodecyl sulfate-polyacrylamide gel electrophoresis of membrane samples cross-linked to Iz5I-IGF-II. The conditions used for this experiment were the same as for the experiment in Fig. la. Apparently, 1251-IGF-II was cross-linked to the same two types of membrane components affinity-labeled by lZ5I-IGF-I. The tissue distribution, apparent M,, and sensitivity to dithiothreitol of the membrane components labeled by 1251-IGF-II were indistinguishable from those labeled by lZ5I-IGF-I (Fig. 1). However, the relative amount of radioactivity associated with the type I labeled species was higher when membranes were affinity-labeled with lZ5I-IGF-I than with 1251-IGF-II (Fig. la,  lanes 3 and 4 versus Fig. lb, lanes 3 and 4). Conversely, the labeling of type I1 species was more intense when membranes were cross-linked with '251-IGF-II than with '*'I-IGF-I.
The pattern of labeled bands shown in Fig. 1 remained the same when different membrane preparations from a given tissue or different preparations of '251-labeled peptides were used during the course of this study. Preliminary experiments indicated that the concentration of disuccinimidyl suberate used in the affhity-labeling protocol could be varied from 0.05 to 0.5 mM without any apparent change in the molecular weight of the affinity-labeled species or in the protein staining pattern of the gels. Therefore, it seems unlikely that any of the type I or type I1 affhity-labeled species results from the artifactual cross-linking of two or more adjacent membrane components. Electrophoresis of affinity-labeled membraneL from various tissues under conditions resolving peptides in the M, = 40,000-400,000 range did not show any specifically labeled peptide other than those described above or their respective reduction counterparts (not illustrated).
The electrophoretic characteristics of the type 11 lZ5I-IGF-I labeled species are similar to those of the putative receptor for multiplication-stimulating activity affinity-labeled in membranes from several rat and human tissues (12). Furthermore, both unlabeled IGF-I and MSA compete with '251-IGF-I, 1251-IGF-II, or Iz5I-MSA for the affinity labeling of these species on rat liver and adipocyte membranes, suggesting that both kinds of ligand share this putative growth factor receptor? Insulin present at M during incubation of membranes with lZ5I-IGF-I or Iz5I-MSA does not compete for the labeling of the type I1 labeled species Human placenta membranes affinity-labeled with Iz5I-IGF-I were subjected to a bidimensional electrophoresis analysis to further document the subunit composition of the disulfidelinked type I labeled species, as well as to establish more rigorously their structural relationship with the known insulin receptor complexes. The affinity-labeled placenta membranes were fist electrophoresed on a dodecyl sulfate-polyacrylamide gel in the absence of dithiothreitol. This fist dimension electrophoresis resolved the type I and type I1 labeled bands depicted in Fig. la, lane 3, as well as two very minor labeled species with M, = 205,000 and 155,000, respectively (Fig. 2). This fist gel was subjected to a second dimension electrophoresis in the presence of dithiothreitol. The characteristics of the labeled species separated after this treatment support the hypothesis that the M, = 350,000,320,000, and 290,000 membrane components labeled by '251-IGF-I have a subunit stoichiometry equivalent to their respective (a/%, (a/3)(a/31), and (a/3,)* insulin receptor counterparts. Thus, the main radioactive species released by dithiothreitol from all three M, = 350,000,320,000, and 290,000 putative IGF receptors was a Mr = 130,000 species (Fig. 2). We denote this species as the a IGF receptor subunit. A minor radioactive species (Mr = 98,000) deriving from the M, = 350,000 labeled complex (Fig. 2) is (M, = 225,000 to 250,000) deriving from the affinity-labeled insulin receptor forms are also consistently observed after treatment of membranes with hydroxysuccinimide esters (21, 24). A M, = 170,000 minor labeled species (Fig. 2, species 11 and 12) is predicted to consist of (ab,) partially reduced receptor fragments derived from the putative (ap)(apI) (Mr = 320,000) and (a/31)2 (Mr = 290,000) type I growth factor receptor forms.
Small amounts of (ap) and (apI) insulin receptor half-fragments that are nat disulfide-linked as complete receptor complexes are present in native membranes from rat liver (25) and human placenta.' The apparent M, of these native receptor fragments is increased by low concentration of dithiothreitol, probably due to reduction of intrapeptide disulfide bonds (20). At high concentrations, dithiothreitol promotes the complete dissociation of these (ap) and (apI) insulin receptor fragments into their respective a and p or PI receptor subunits. Native membranes also appear to contain unassembled (up) and (apl) IGF receptor fragments. Supportive evidence for this postulate is provided by the detection of reduced a IGF receptor subunit in the second electrophoresis dimension (Fig.  2, species 13 and 14) at positions expected for the precursor unassembled The affinity-labeling of the p subunit from type I IGF receptors was more clearly observed when individual receptor complexes were isolated from a first gel and re-electrophoresed in the presence of dithiothreitol (Fig. 3, inset). Under these conditions, the proposed (a/?)2 (Mr = 350,000) receptor form, but not the (ap1)2 (M, = 290,000) receptor form yielded labeled / 3 receptor subunit. Compared to the labeling of the a subunit, the affinity-labeling of the p subunit of the type I-IGF receptor using the present methodology is apparently rather low (Fig. 2). A similar result is obtained when the insulin receptor is cross-linked to "'I-insulin by disuccinimidyl suberate (14, 19, 21, 24). In the case of the insulin receptor, the a and p receptor subunits present in the native (a/3)2 form of this receptor can be purified using insulin-agarose affinitychromatography (26, 27).3 The results obtained using this methodology directly indicate the presence of equimolecular amounts of both subunits in the receptor complex (27). A similar conclusion can be drawn from the immunoprecipitation of biosynthetically or chemically labeled insulin receptors using anti-insulin receptor autoantibodies from patients with severe insulin resistance and acanthosis nigricans (28, 29). Whereas the structural characteristics of the insulin receptor have been documented by these several independent methodologies, the subunit stoichiometry of the ( (~p )~ IGF receptor complex proposed here is based only in the strict analogy between the electrophoretic properties of this affinity-labeled IGF receptor species and the insulin receptor. Confirmation of this proposal must await direct chemical characterization of the type I IGF receptor.
The affinity-labeling of all type I and type I1 IGF receptor forms was abolished by the presence of 1 PM IGF-I during incubation of human skin fibroblast membranes with "'I-IGF-I (Fig. 3, lanes 2 and 6). The labeling of type I IGF receptors was only slightly decreased by the presence of 1 PM insulin during incubation with '2sI-IGF-I (Fig. 3, lanes 3 and 7). An insulin concentration of 10 PM displaced most of the "'1-IGF bound to this receptor type (Fig. 3, lanes 4 and 8). However, 1 or 10 p insulin did not substantially affect the labeling of type I1 IGF receptors by "'I-IGF-I (Fig. 3, lanes 3, 4, 7, and  8). A similar pattern of 1251-IGF-I displacement by IGF-I and insulin was observed with IGF receptors in human placenta membranes (IGF receptor type I and type 11) and on rat liver and rat adipocyte membranes (IGF receptor type 11) (data not shown).
The high affinity insulin receptor present on human skin fibroblasts can be affinity-labeled by cross-linking to "'I-insulin with disuccinimidyl suberate (Fig. 4, lanes 1 and 5). Like the insulin receptor on membranes from other tissues, the affinity-labeled fibroblast insulin receptor appears as (a/3)2,  (lanes I to 4). Membrane samples on lanes 5 to 8 received identical treatment as samples on lanes I to 4, respectively, except that they were added with 50 x" dithiothreitol (DTT) during solubilization in sodium dodecyl sulfate. An autoradiogram from a fixed, dried gel is shown. The subunit stoichiometry of the insulin receptor complexes is indicated. dissociated into free receptor subunits by incubation with dithiothreitol in the presence of sodium dodecyl sulfate. The presence of 1 PM unlabeled insulin during incubation of skin fibroblast membranes with "'I-insulin prevents the radioactive labeling of the insulin receptor species (Fig. 4, lanes 2 and  6). The affinity-labeling of the human skin fibroblast insulin receptor is also prevented by IGF-I added in excess during incubation of membranes with "'I-insulin (Fig. 4, lanes 3, 4,  7, and 8). Under similar conditions, IGF-I also prevents the affinity-labeling of insulin receptor on rat adipocyte (Fig. 5) and human placenta membranes but not on rat liver membranes (not illustrated). The relative potency of IGF-I to prevent the labeling by "'I-insulin of the insulin receptor in rat adipocyte membranes is about 100-500 times lower than t,hat, of unlabeled insulin (Fig. 5). These data correlate well with the relative affinity of IGF-I for the insulin receptor in adipocytes as determined by binding studies (4).

Insulin-like Growth Factor Receptor Structures
These competition studies described above indicate thot (a) the type I IGF receptor and the insulin receptor show structural similarities but are two distinct species, ( b ) the type I IGF receptor exhibits a high affinity for IGF-I and a low affinity for insulin, ( c ) the type I1 IGF receptor exhibits a high affinity for IGF-I1 and no significant affinity for insulin, and (d) human skin fibroblasts, human placenta, and rat adipocyte insulin receptors exhibit a low affinity for IGF-I while the rat liver insulin receptor does not have any apparent affinity for IGF-I. Interestingly, this apparent heterogeneity among insulin receptors from different tissues is not detected when their affinity for insulin itself is examined (30).
The relative intensity of labeling of type I IGF receptor and type I1 IGF receptor by "'I-IGF-I was different from their relative intensity of labeling by '"I-IGF-11 (compare Fig. 1, a  and 6 ) . This observation suggested a different affinity of each IGF receptor type for IGF-I and IGF-11. A more complete documentation of the affiity of both IGF receptor types for these two ligands is provided in Figs. 6 and 7. In these experiments, membranes from human skin fibroblasts and rat liver were incubated with 1251-IGF-I or "'I-IGF-11 in the presence of various concentrations of unlabeled IGF-I or IGF-11. The labeling of skin fibroblast type I IGF receptors by 0.5 nM 12sI-IGF-I was substantially decreased by 1 nM IGF-I but not by 1 nM IGF-I1 (Fig. 6). At 10 nM, IGF-I abolished the labeling of type I IGF receptor by '"I-IGF-I whereas the same concentration of IGF-I1 only partially decreased the labeling of this receptor type (Fig. 6). These observations suggest that the highest affinity ligand for type I IGF receptors is IGF-I.
On membranes from human skin fibroblast or rat liver, IGF-I1 is about 10 times more potent that IGF-I in inhibiting the affinity-labeling of type I1 IGF receptor (Fig. 7). The higher ability of IGF-I1 to compete for the labeling of the type I1 IGF receptor is observed regardless of whether membranes are affinity-labeled with 12sII-IGF-I (Fig. 7, series 3 and 4) or with '251-IGF-II (Fig. 7, series 1 and 2 or series 5 and 6). These data suggest that the highest affinity ligand for type I1 IGF receptors is IGF-11. Interestingly, the heterogeneity among type I1 IGF receptors from different tissues as suggested by differences between their respective molecular weights (see Fig. 1) can be accompanied by differences in the affinity of these receptors for IGF-I1 and IGF-I. Thus, the affinities of human skin fibroblast type I1 IGF receptor for IGF-I1 and IGF-I are about 10 times lower than the respective affinities of rat liver type I1 IGF receptor for the same two ligands (Fig.  7, compare series 1 to series 3 and 5, and series 2 to series 4   and 6).
The type I and type I1 IGF receptor structures affinitylabeled in membranes from rat liver and adipocytes, and from human placenta and skin fibroblasts, have also been observed in other tissues and cultured cell lines of human and rodent origin screened for the presence of these receptor structures. Table I summarizes the relative amount of insulin receptor and of the two types of IGF receptors present in some of these systems as estimated by affinity-labeling of the corresponding whole cell or isolated membrane preparations. The human RPMI 6666 and RPMI 7666 lymphoblasts exhibit insulin receptors, but not IGF receptors, on their surface. Mouse 3T3-L1 fibroblasts exhibit a low number of affinity-labeled insulin and type I IGF receptors, but the intensity of labeling of the type I1 IGF receptor in these cells is similar to that in rat adipocytes. 3T3-Ll fibroblasts are characterized by their ability to differentiate to an adipocyte phenotype (16) sensitive to short term metabolic effects of insulin (17). It has been shown (17,32) that the number of high affinity insulin-binding sites, but not that of epidermal growth factor-binding sites, increases upon differentiation of 3T3-Ll fibroblasts into adipocytes. Consistent with these observations, we have observed a more intense labeling of the insulin receptor in 3T3-Ll adipocytes than in 3T3-Ll fibroblasts (Table I). This apparent increase in insulin-binding capacity was accompanied by an equivalent increase in the amount of labeling associated with the type I IGF receptor (Table I). These observations suggest that differentiation of 3T3-Ll cells includes the induction of

TABLE I Tissue distribution and relative amounts of types I and II IGF receptors and insulin receptor Intact cells ( a ) or isolated membranes ( b )
were screened for their contents in IGF receptors and insulin receptors under standard affinity-labeling conditions. Cell suspension aliquots (2 X 10" cells) were incubated with 5 nM "'I-labeled hormones for 60-90 min at 22 "C, and with 0.40 mM disuccinimidyl suberate for 15 min a t 10-15 "C. Membrane samples (150-200 pg of membrane protein) were affinitylabeled by sequential incubation with 5 nM "'I-labeled hormones for 60 min a t 10 "C and with 0.20 mM disuccinimidyl suberate for 15 min a t 0 "C. Other details of the affinity-labeling protocol, and the subsequent dodecyl sulfate-polyacrylamide gel electrophoresis and autoradiography are described under "Materials and Methods." Affinity-labeled human normal skin fibroblasts and rat white fat cells and their respective isolated membrane preparations were taken as standards for quantitative comparison with other affinity-labeled cells or membranes. "High" or "Low" receptor contents in a given cell or membrane type are rough estimates based on the amount of I2'Ihormone bound to these cells or membranes, and on the relative intensity of receptor labeling as compared with the above mentioned standards. -, affinity-labeled receptors were not detected even after extensive autoradiography exposure of the electrophoresis gels. IGF

Insulin-Like Growth Factor Receptor Structures
(a& insulin and IGF receptor structures possibly through a mechanism common for both receptor types. Little or no increase of the amount of labeling associated with the type I1 IGF receptor was observed upon differentiation of 3T3-Ll fibroblasts to adipocytes. Like rat and mouse liver membranes, H-35 rat hepatoma cells do not exhibit labeling of type I IGF receptors. The limited number of insulin receptors in H-35 hepatoma cells as determined by radioligand binding studies (31) results in a low intensity of insulin receptor labeling in these cells (Table I). Interestingly, H-35 hepatoma cells present a large number of type I1 IGF receptors both on the cell surface and in isolated membrane preparations. Types I and I1 of IGF receptors and insulin receptor are found in variable proportions in A875 melanoma cells, TE85 sarcoma cells, IM-9 lymphocytes, and rat lymphocytes, as well as in membrane preparations from rabbit superior cervical ganglia ( Table I).
The two types of membrane components affinity-labeled by '*'I-IGF-I and '251-IGF-II have the properties expected for physiologically relevant IGF receptors. They exhibit high  N,N,N',N'-tetraacetic acid, pH 7.2, and Polytron homogenization of the cells. A crude membrane fraction was obtained from the homogenate by centrifugation at 30,000 X g for 20 min after separating most of the nuclei, mitochondria, and unbroken cells by centrifugation at 3,000 X g for 10 min. This membrane fraction was solubilized in the presence of 1% sodium dodecyl sulfate and subjected to electrophoresis on 5% polyacrylamide gels. The autoradiogram from a fixed, dried gel is shown. affinity and specificity for IGF-I and IGF-11. They are present on the surface of intact target cells (Fig. 8). Yet, these two receptor types differ from each other in subunit composition, highest affinity ligand, and tissue distribution. The IGF binding kinetics observed in other laboratories for tissues containing different relative amounts of type I and type I1 IGF receptors is readily predicted from the results presented here. For example, the finding that IGF-I1 is more potent than IGF-I in displacing '"I-IGFs bound to rat liver membranes (11) is consistent with our demonstration that this tissue is devoid of the type I receptor. Also, IGF-I is more potent than IGF-I1 in displacing 12'II-IGF-I bound to human skin fibroblast membranes, consistent with the presence of some type I receptor in those cells (11). High concentrations of insulin compete for the binding of 12'I-IGF~ to membranes from human skin fibroblasts (33) but not from rat liver (11,34). This point agrees with our results showing that the type I growth factor receptor has some affinity for insulin but the type I1 receptor does not have any significant affinity for his hormone.
The use of affinity-labeling methodology to link '"I-IGF-I to components of BRL 3A2 rat liver cells and human placenta membranes has been recently reported (35,36). These data are consistent with the results presented in this study on the type I growth factor receptor structure. The structural characteristics of the receptors for IGF-I, IGF-11, and insulin revealed by the present affinity-labeling methodology raise several basic questions concerning the evolution, genetic expression, and biological functions of these receptors. The close structural similarity between the type I IGF receptor and the high affinity insulin receptor, as well as their apparent ability to undergo a similar unique proteolytic transformation, suggest that these two kinds of receptors may have similar amino acid sequences and may have evolved from a common ancestor molecule. This hypothesis would also imply that the genetic expression of the type I IGF receptor and the insulin receptor on a given cell type may involve specific rearrangements of common DNA sequences. The direct assessment of the presence of type I and type I1 IGF receptors on different cellular systems by the affinity-labeling methodology used in the present study may prove useful for the clarification of these issues.