Properties of the Insulin Receptor Isolated from Liver and Fat Cell Membranes*

SUMMARY The insulin-binding proteins isolated from membranes of rat liver and of fat cells with the nonionic detergent, Triton X-100, appear to have similar or identical physicochemical and kinetic properties. The denaturation of the soluble receptor by reagents, such as sodium dodecyl sulfate, urea, and guanidine .HCl, is reversible if moderately low concentrations of the reagents are used. The binding protein is relatively stable to storage at 4 or -20”. Digestion of the water-soluble receptor with phospholipase C and neuraminidase does not affect its capacity to bind i2jI-insulin. High concentrations of NaCl (2 M) are virtually without effect on the insulin-receptor-binding interaction. The specific insulin-binding activity of the receptor is completely destroyed by digesting with concentrations of trypsin which are too low to modify the native receptor in the intact membrane but which are equally effective in destroying the receptor of phospholipase-treated membranes. experiments on columns of Sepharose calibrated with several standard [r4C]acetyl

In recent studies the insulin-binding molecules of liver and fat cell membranes have been quantitatively extracted in n-atersoluble form by using the nonionic detergents, Triton X-100 and Lubrol-WX (I).
The water-soluble insulin-binding material obtained in this way does not sediment significantly upon centrifugation for 6 hours at 300,000 x g, and it passes unhindered through membrane filters having pores with a diameter of 0.2 pm.
The specific binding of insulin to this soluble protein is a reversible and saturable process which is unaffected by a variety of unrelated peptide hormones. Chemical derivatives of insulin, however, compete with native insulin for binding in direct proportion to their biological potency. Several features of the extraction procedures, and some of the properties of the binding interaction, indicate (I) that this solubilized material represents those structures which have previously been identified in liver and fat cell membranes as being the biologically significant insulin receptors (2)(3)(4)(5)(6).
The present studies present more detailed properties of the insulin-binding protein solubilized from liver and fat cell membranes with Triton X-100.
The results further strengthen the bovine albumin was used as the standard.
Trypsin-agarose was prepared by the cyanogen bromide procedures described earlier (3) ; the derivative contained 5.9 mg of trypsin per ml of agarose.
The procedures used to measure the specific binding of lZZIinsulin to membranes and particulate materials have been described (3)(4)(5)(6).
1251-Insulin was prepared and purified as described elsewhere (2).
The assay used to detect water-soluble insulin-receptor complexes has recently been described in detail (1  The Triton X-100 membrane extracts were dialyzed for 16 hours at 4" against the buffer to be used in preparing the sucrose solutions. Samples (0.1 or 0.2 ml) were incubated at 24" with 1251-insulin (1 to 8 x lo4 cpm) for 30 min and cooled in ice before layering on top of the gradients. In every centrifugal run control samples were included in which native insulin (20 Fg) was added to the membrane-extracted solution before addition of 9-insulin. A sample was also included which contained 9-insulin but no tissue extract.
Centrifugation was performed at 1" at 34,000 rpm for 16 hours in a Spinco SW 50.1 rotor with the Spinco model L2-65B ultracentrifuge.
Fractions (20 drops each) were collected b) puncturing the bottom of the tube with a Ruchler piercing unit. In all cases the bottom part of the empty tubes was cut, placed in 0.5 ml of 107, (w/v) sodium dodecyl sulfate, and counted for radioactivity with a liquid scintillation counter in the presence of 10 ml of TLX toluene fluoroalloy and 2 ml of Isio-Solv Solubilizer BBS-3 (Beckman).
Portions (50 to 100 ~1) of the fractions were examined for radioactivity with a well-type y counter (for 1251) or with a liquid scintillation spectrometer (for 14C) using the counting phosphors described above.
No significant qucnching was detected with sucrose, sodium bromide, cesium chloride, or Triton X-100 under these conditions. Sedimentation coefficients were determined by comparison to the behavior of ['"clacetyl apoferritin, albumin, y-globulin, thyroglobulin, and insulin under the same conditions. Virtually all of the studies to be described were performed on extracts from both liver and fat cell membranes to determine possible differences in the receptor from these two tissues.
Xlthough the data for both types of membrane extracts are not presented in every instance, in no case have significant differences in their properties been detected.

EJect of Protein
Denaturants---It has been shown previously (1) that the nonionic detergents, Triton X-100 and Lubrol-WX, in concentrations greater than 0.59; (v/v) and 0.2yc (w/v), respectively, decrease the binding of insulin to the water-soluble insulin receptor if they are present in the incubation mixture used in the binding assay. However, simple exposure of the membrane proteins to high concentrations (3%) of these detergents does not adversely affect the insulin-binding protein since removal of the detergent by dialysis or dilution fully restores the capacity of this protein to bind insulin specifically.
It is of interest that dialysis of the Triton X-100 extracts against detergent-free buffers results in the formation of precipit'ates which contain all of the insulin-binding activity originally present in the intact membrane or in the detergent-solubilized proteins. Commonly used protein-denaturing reagents, such as sodium dodecyl sulfate, urea, and guanidine, caused a marked loss in binding activity (Table I).
Glycerol in concentrations greater than 20yo (v/v) similarly interferes with the insulin-receptor interaction.
Some of the adverse effects of these reagents on the insulin-binding protein appear to be reversible by simple dilution  TABLE III  TABLE IV   Stability  of water-soluble  insulin  receptor  to storage Liver and fat cell membranes were extracted with 1% (v/v) Triton X-100 as described in the text.
The high speed supernatants of liver (8.2 mg of protein per ml) and fat (3.1 mg of protein per ml) extracts were stored at 4 and -20" in the presence and absence of 20 The insulin-binding capacity was tested periodically with the polyethylene glycol assay described in the text., with 1.4 X 10P1o M 1251-insulin.
Effect of NaCl and EDTA on speciJic binding of insulin to soluble receptors from liver ancl fat cell membranes The specific binding of 1251-insulin (2.8 X lo-'0 M) to Triton X-100 extracts of liver (6.3 mg of protein per ml) and fat cell (4.1 mg of protein per ml) membranes was determined with the polyethylene glycol assay described in the text except that NaCl or sodium EDTA were added to the assay mixture as indicated.  (Table II). By comparing the data in Tables  I and II it is apparent that although 0.16% sodium dodecyl sulfate, 3 M urea, and 20y0 glycerol cause severe or nearly total suppression of binding, the effects are almost completely reversed by decreasing the concentration of the reagent IO-fold. Exposure of the insulin receptor to higher concentrations of these reagents results in apparent irreversible denaturation.
E$ect of digesting "Triton"-extracted water-soluble insulin receptor wilh neuraminiclase, trypsin, and phospholipases The high speed supernatants of Triton extracts of liver (6.3 Stability of Soluble Receptor to Conditions of Storage---The insulin receptor extracted from liver and fat cell membranes with Triton X-100 is reasonably stable to storage for at least 4 weeks at 4 or -20" as long as the concentration of the detergent is greater than 0.50/', (Table  III). The lower temperature is preferable since greater stability is maintained and fewer problems are encountered with bacterial contamination. During storage, particularly with freezing, a fine precipitate develops which contains no insulin-binding activity and which can be removed readily by centrifugation.
It has been shown that repeated (four times) freezing and thawing by itself is not injurious to the receptor.
Storage in 20% (v/v) glycerol does not provide protection against the loss of binding activity (Table  III). Dithiothreitol at concentrations of 0.1 and 1 m&f is also not helpful in maintaining stability during storage at 4 or -20". It is also of interest, furthermore, that this reagent does not decrease the insulin-binding activity of the receptor under these conditions. mg of protein per ml) and fat cell (4.1 mg of protein per ml) membranes were incubated with trypsin at 37" for 20 min (3), neuraminidase from C. perjringens purified (12) by affinity chromatography (37", 35 min), phospholipase C (4) from C. perjringens (37", 30 min), or phospholipase A from V. russelli (37", 30 min). The incubations containing phospholipases A and C also contained 5 and 3 mM CaC12, respectively.
The tryptic digestion was stopped by adding soybean trypsin inhibitor (50 rg per ml) but the other enzymes were not removed or inhibited prior to assay of insulin binding.
Specific 1251-insulin binding was tested with the polyethylene glycol assay (see text) with 2.8 X lo-10 ,\I 12SI-insulin. Eflect of NaCZ and EDNA-Exposure of liver (5) or fat cell (4) membranes to increasing concentrations of NaCl (up to 2 M) results in a 3-to 6-fold enhancement in the specific binding of insulin to these membranes, presumably as a result of displacement of membrane phospholipids and consequent exposure of normally masked receptors (4). In contrast, only small effects on the binding of lZ51-insulin to the soluble receptor are observed by varying the concentration of NaCl in the assay mixture from 0.4 to 3 M (Table IV).
With 2 M NaCl the specific binding of insulin to the Triton-solubilized membrane proteins is increased by 10 to 200/c, and with 3 M NaCl the binding is decreased.
In harmony with the evidence that heavy metals are not required for the formation of specific insulin-receptor complexes in intact membranes (4), no significant effects on the binding of insulin to  Table VI. membranes (3), the decreased binding of insulin to soluble proteins treated xvith low concentrations of trypsin appears to be complete and is not accompanied by detect'able changes in the affinit\-of the binding protein for insulin.
I digestion of the solubilized membrane proteins with high concentrations of purified neuraminidase from C. perfringens (12) does not modify the specific insulin binding to these proteins (Table  V). Similarly, digestion with high concentrations of phospholipases h and C are without effect (Table V).

Gel Filtration and Determination
of Stokes Radius-The agarose filtration chromatography pattern of the solubilized insulinbinding protein (Fig. 1) suggests that it is a protein of large size and that its affinity for insulin must be very great.
The nearly total prevention of binding of 12%insulin by native insulin is easily demonstrable in these experiments. The reversibility of this insulin-protein complex has been shown earlier by gel filtration experiments on Sephadex G-50 (I). Similar experiments with Triton X-100 extracts of red blood cell ghosts do not result in significant radioactivity which is displaceable by native insulin. Water-soluble Triton X-100 extracts (0.5 ml) of liver and fat cell membranes were chromatographed on the Sepharose 6B column described in Fig. 1 after incubating with lz51-insulin (5 X lo5 cpm) for 30 min at 24".
The conditions of chromatography are as described in Fig. 1 except that the column-equilibrating buffer was altered as indicated in the table.
The chromatographic emergence of the insulin-receptor complex, expressed as the distribution coefficient (10) refers to the fast moving peak of Fig. 1  The elution position (Knv) of the insulin-receptor complex in these agarose chromatography experiments is highly reproducible, and no difference is detected in the behavior of the proteins solubilized from liver or fat cell membranes (Table VI). Furthermore, the chromatographic pattern of the insulin-receptor complex does not vary with the concentration of Triton X-100 in the buffer unless the detergent is omitted entirely or its concentration is decreased below 0.05$& (v/v). Under these circumstances highly aggregated forms of the receptor occur (Table VI). No detectable changes occur in the distribution coefficient of the insulin-binding protein by ~~erforming the chromatography in buffers containing 1.5 M NaCl or 10% (w/v) sucrose (Table VI). The binding protein obtained by Triton X-100 cstraction of phospholipase C-treated or delipidated membranes is indistinguishable by gel filtration from the protein which is similarl) extracted from native membranes (Table VI). It is not possible to determine the molecular weight of the insulin-binding protein by gel filtration chromatography alone since it is apparent that the elution behavior of macromolecules correlates wit,h the Stokes radius of the protein rather than with its molecular weight (9). Accordingly, the Stokes radius of the Sedimentation Behavior-On sucrose density gradient centrifugation the 1251-insulin-binding protein of membrane extracts separates readily from free '251-insulin, and prior addition of native insulin to the extract completely blocks formation of the labeled macromolecule (Fig. 3). Recentrifugation of the labeled insulinreceptor complex isolated from such experiments (Fig. 3C) does not change its sedimentation behavior, although a small peak appears in the position of free insulin, indicating some dissociation of the complex.
These experiments again point to the very tight nature of the complex.
If the 1251-insulin-receptor complex is heated at 37" for 65 min in the presence of native insulin (50 pg per ml) there is nearly complete exchange of the radioactivity, consistent with the reversibility of the binding process. Triton X-100 extracts of red blood cell ghosts do not have significant Fat cell membranes were extracted with 1% (v/v) Triton X-100 and centrifuged for 70 min at 44,000 rpm as described in the text.
The supernatant was dialyzed for 16 hours at 4" against 0.1 M sodium phosphate buffer containing 0.27, (v/v) Triton X-100. The supernatant (0.2 ml, containing 0.5 mg of protein) was incubated at 24" for 20 min with lz51-insulin (3.5 X lo* cpm) before being subjected to gradient centrifugation for 16 hours (11) under the conditions described in the text (C). Another sample of supernatant was processed identically except that native insulin (20 pg) was added 5 min before incubating with lz51-insulin (B With knowledge of the independently determined Stokes radius (a) and sedimentation coefficient (s) it is possible to calculate the molecular weight (M) and the frictional ratio (f/f") of the protein according to the classical equations (9, 15) (3) where p is the solvent density in grams per cm3, N is Avogadro's number, and 6 is the solvation factor in grams of solvent per g of protein.
The values calculated by assuming a partial specific volume (0) of 0.734 cm3 per g and a solvation factor (6) of 0.2 g of solvent per g of protein, which are typical for proteins (15, 16) are described in Table VIII.
These data indicate that the insulin receptor is a highly asymmetrical protein with a molecular weight of about 300,000.   (14) is given in parentheses f Calculated (14) for prolate ellipsoids; 8 is assumed to be 0.2 g per g of protein.
Highly unusual solvation factors almost certainly are not involved since it can be calculated that a value of 6 greater than 2 g of solvent per g of protein would be necessary to account for the Stokes radius if a frictional ratio in the normal range (1.2) is assumed.
This value is an order of magnitude larger than is observed for undenatured proteins (15). AZ very high value for the partial specific volume (v) of the insulill-binding protein, indicative of significant lipid content, must be considered seriously.
The fact that rather vigorous digestion of the membranes with phospholipases and strong delipidat,ion procedures do not change Stokes radius or distribution coefficient of the binding protein, suggests that the receptor is not a lipoprotein.
However, it is difficult to exclude by these procedures the presence of lipid material which may be bound to the protein with extraordinary avidity or in a manner not accessible to enzymes or solvent molecules.
For this reason studies were performed to determine directly the density of the insulinbinding protein.
Such determinations can be performed (9, 17) on crude protein mixtures by measuring the sedimentation of the 1Winsulin-protein complex in solutions of varying density, provided the salts required to achieve these densities do not interfere seriously with the integrity of the complex. It was not possible to perform these experiments in sucrose solutions since the specific binding of insulin was severely affected by solutions having a density greater than 1.1 (Table IX).
Sodium bromide solutions at concentrations greater than 20% (w/v) also interfere with binding.
Very high concentrations of cesium chloride, however, do not significantly impair the specific binding of insulin (Table  IX).
The sedimentation behavior of the insulin-binding protein in high density solutions of cesium chloride reveal good migration into a solution with a density of 1.228 and very small but definite sedimentation with a density of 1.298 (Table X). These experiments argue against a lipoprotein nature of the insulin-binding  protein since even the most dense of lipoproteins (p = 1.21 g per cm3 (18)) would not have sedimented significantly in these solutions.
The density of the hydrated form of the binding protein is therefore at least 1.25 g per cm3, which corresponds to a maximal hydrated specific volume of 0.80 cm3 per g (9,19). It is of interest that in these experiments free 1251-insulin (0 = 0.735) sedimented faster than the receptor protein (Table X). Although these results are consistent with the absence of lipid in the insulin-binding protein, caution must be exercised in placing too much emphasis on the density experiments alone.
Since the density of glycoproteins is known to be higher than that of simple when more than three experiments were performed, the data are presented as the mesn value & standard error. b Calculated from data, such as those presented in Fig. 6, by substitution in a second order equation (2,5,6). c Calculated from semilog plots, as illustrated in Fig. 6. proteins, since the carbohydrate content of the insulin-binding proteins is not known, and since little is known about the preferential interaction terms in CsCl solvents, it is possible that the high density observed reflects compensating contributions of lipid and carbohydrate moieties. The combined data indicate that the molecular parameters (Table VIII) are probably reasonably valid, and that the insulin receptor is indeed a rather large, asymmetrical protein.
The unique combination of a large Stokes radius and a low sedimentation constant similar to that presented in this report has recently been described for the progesterone-binding components of t'he chick oviduct (20), for the estrogen-binding proteins of calf uterus (21), and for the glucocorticoid-binding component of mouse fibroblasts (22).  (Table  XI).
These values are similar to the values calculated for the insulin-receptor interaction in intact fat cells (2) and in liver and fat cell membranes (5,6). The formation and dissociat.ion of the insulin-receptor complex were studied separately as described in Fig. 6. The association process is a bimolecular process, as indicated by the fit of the data to a second order equation (1,5). Dissociation of the complex is a first order process which is highly temperaturedependent.
The rate constants calculated from such data for extracts of liver and fat cell membranes are summarized in Table  XI studies measure the same processes which are ordinarily being studied in the standard binding assays.
The nature of 1%insulin which tightly forms a complex with the soluble membrane receptor was esamined by precipitation of the complex with polyethylene glycol followed by dissociation of the separated complex with guanidine.HCl (Table XII). The insulin I\-hich is eluted in this way is indistinguishable from fresh insulin, indicating that binding to the soluble proteins is not associated with detectable degradation. Furthermore, complex formation is not accompanied by the formation of stable covalent bonds or disulfidc interchange reactions.

DISCUSS101
Konionic detergents such as Triton X-100 quantitat.ively extract in soluble form the insulin-binding structures of liver and fat cell membranes (1). The present studies provide further evidence that the detergent-solubilized molecules are very similar or identical with those high affinity binding structures which can be mea<ured and studied in biologically responsive intact fat cells (2) and in isolated membrane preparations from liver (6) and fat cells (5). There is very strong evidence that the specific insulinbinding structures of these particulate preparations are the biologically significant receptors for insulin (reviewed in References 24 and 25).
The present studies reveal striking similarities in the kinetic properties of the interaction of insulin with the soluble and with the particulate (2,5,6) structures. Thus, the interaction in both cases is saturable and reversible, and it does not involve inactivation of insulin. The rate constants of association and dissociation, and the dissociation constants (about 10-l" RI) determined independently from equilibrium data, are very similar in the soluble and particulate states of the receptor. In addition to the kinetically homogeneous nature of the binding interaction, the inability to detect heterogeneous patterns on gel filtration and sucrose gradient centrifugation further suggests that the insulin receptor may be a single and unique receptor class. Furthermore, the nearly identical properties of the solubilized receptors from liver and from fat cell membranes points to the possible identity of the insulin receptors from these two tissues. This possibility has been suggested earlier (6) on the basis of the similarities of the binding interactions observed in the particulate state.
It has been shown that digestion of liver or fat cell membranes with phospholipases, and that, extraction of these membranes with organic solvents, results in the exposure of a substantial (2-to B-fold) number of new insulin-binding structures of similar kinetic behavior to those which are normally exposed (4). These effects are believed to result from the hydrolysis or displacement of membrane phospholipids which normally may mask or shield these receptors from large macromolecules (such as insulin and trppsin) in the solvent. A similar unmasking of insulin-binding sites in these membranes by high salt concentrations (2 M NaCl) suggests that the phospholipid polar head groups are important in this shielding effect (5). The studies on the solubilization of the insulin receptor are in agreement with these observations. Extraction of the membranes with increasing concentrations of detergents causes a progressive loss of insulin-binding activity of the residual membranes and the concomitant appearance in the soluble fraction of binding structures in progressively higher yield than were originally present in the membranes (1). Delipidation of the membranes by phospholipase digestion or by estraction with organic solvents does not lead to increased yields of binding activity after subsequent detergent extraction. Also, as shown here, digestion of the solubilized proteins with phospholipase C does not appreciably increase insulin binding, and high concentrations of P\SaCl do not affect the binding of insulin to the solubilized proteins.
These observations suggest that solubilization of the receptor by detergents is accompanied by gross displacement of membrane phospholipids from the insulin-binding structures. This is clearly confirmed in the present studies which show that phospholipase digestion or organic solvent extraction do not alter the molecular properties of the solubilized receptor. Furthermore, the sedimentation behavior of the receptor is compatible with the absence of significant lipid material. It therefore would have been surprising if NaCl or phospholipase digestion had significantly modified the binding process. It is also of some interest that the isolated and the membrane-bound receptor are not apparently denatured and their molecular parameters are not altered by such high concentrations of salt. The extreme susceptibility of the solubilized receptor to tryptic digestion is also consistent with the removal of membrane phospholipids during the solubilization procedure. It has been shown (3, 4) that after digesting liver or fat cell membranes with phospholipases the insulin receptor is rapidly and completely destroyed by concentrations of trypsin which are too low to measurably alter the receptor of normal cells. Furthermore, the primary early effect of tryptic digestion of normal cells or membranes is to decrease the affinity of the receptor for insulin without changing the maximal binding c:tpacitJ-(3  with trypsin (0.5 mg per ml) and another similar fraction with trypsin-agarose (1 mg per ml, Reference 3) for 15 min at 37" at which time soybean trypsin inhibitor (2 mg per ml) was added.
The trypsin-agarose was then removed by centrifugation and washing with three changes of buffer (10 ml) ; the cells were suspended in 8 ml of buffer. Samples of these cells (intact cells) were assayed for specific '251-insulin binding (2). Fractions (5 ml) of these cells were homogenized with a Polytron PT-10 (Brinkmann) for 30 set at a setting of 3.0. Specific insulin binding (2, 5) was determined on samples of these whole homogenates. Other portions (3 ml) of the homogenates were centrifuged at 30,000 rpm for 30 min.