THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 251, No. 7, Issue of April 10, pp. 1877-1888, 1976

Printed in U.S.A.

Site-Site Interactions among Insulin Receptors

CHARACTERIZATION OF THE NEGATIVE COOPERATIVITY

(Received for publication, July 31, 1975)

PIERRE DE MEYTS,* A. RAFFAELE BIANCO,$ AND JESSE ROTH

From the Diabetes Branch, National Institute of Arthritis, Metabolism and Digestive Diseases, National Znstitutes of Health, Bethesda, Maryland 20014

By studying the dissociation of ‘Z”I-insulin from its receptors in the absence and presence of unlabeled insulin, we had demonstrated the presence of site-site interactions of the negatively cooperative type for the insulin receptors. In the present study we extend our experimental method and develop its theoretical basis. The insulin receptors in highly purified mouse and rat liver membranes as well as in human circulating monocytes and human cultured lymphocytes demonstrated negative cooperativity that was extraordinarily similar. The major difference noted was that 1261-insulin dissociates from its receptors on membranes more slowly than it does from its receptors on whole cells. The dissociation rate of 12sI-insulin was directly dependent on fractional saturation of the receptors by hormone. Concentrations of insulin that are commonly found in duo, which occupy only a small percentage of the receptor sites (1 to 5%), are sufficient to accelerate dissociation of hormone from receptor. At these insulin concentrations insulin is entirely monomeric, and in fact at higher concentrations of insulin (greater than lo-’ M) where insulin dimers predominate, the cooperativity effect is progressively lost.

The dissociation rate of ‘Z61-insulin alone (that is at very low fractional saturation of receptors) was markedly accelerated by dropping the pH from 8.0 to 5.0, whereas the dissociation of ‘Y-insulin at high receptor occupancy was only slightly accelerated by the fall in pH. The dissociation rate was directly related to the temperature, but the dissociation rate of ‘Z61-insulin at low receptor occupancy was much more affected by reduction in temperature and showed a sharp transition at 21”. Urea at concentrations as low as 1 M produced a marked acceleration of iZsI-insulin dissociation. Divalent cations (calcium and magnesium) appear to stabilize the insulin-receptor interaction, since higher degrees of receptor occupancy were required to achieve a given rate of dissociation of 1Z61-insulin. These data make it likely that the insulin receptors exist as oligomeric structures or clusters in the plasma membrane. Insulin recep- tor sites appear to switch from a “slow dissociating” state to a “fast dissociating” state when their occu- pancy increases; the proportion of sites in each state is a function of occupancy of the receptor sites by the insulin monomer as well as of the physicochemical environment. Other models which could explain apparent negative cooperativity besides site-site interactions, i.e. polymerization of the hormone, steric or electrostatic hindrance due to ligand-ligand interactions, or unstirred (Noyes-Whitney) layers are considered unlikely in the case of insulin receptors on both experimental and theoretical grounds.

With the introduction of labeling methods that preserve the biological properties of the native hormone, it has become possible to study and characterize the binding of polypeptide hormones to their receptors in the cell membrane (l-7). In general, the reaction of a hormone with its receptor is rapid, saturable, and reversible. A steady state is achieved in a reasonable time, and quantitative analysis can be performed

* Research fellow of the Fends National de la Recherche Scientifique, Belgium. Recipient of Public Health Service International Postdoctoral Fellowship F05-TW1918 and of the 1975 Solomon A. Berson Research and Development Award from the American Diabetes Association.

$ Recipient of Public Health Service International Postdoctoral Fellowship F05-TW1964. Present address, Institute of Clinical Medi- cine, Department of Medicine, University of Naples Medical School II, Naples, Italy.

according to methods established for protein-ligand reactions at equilibrium (8-21). After corrections for hormone and recep- tor degradation (22, 23) and “nonspecific” binding (24), the most common method of analyzing the data is to plot the bound/free ratio of the labeled hormone (B/F) as a function of the concentration of hormone that is bound to the receptors, i.e. a Scatchard plot (8, 9). The numerous assumptions to be fulfilled for the validity of this analysis have been stressed (5, 6, 10, 11).

In some studies, this plot was linear, e.g. for growth hormone (25), gonadotropins (26), calcitonin (27), and prolactin (28). In these cases, a single homogeneous class of independent binding sites is present, and the affinity constant, K,, can be derived from the slope of the line and the binding capacity (or total re-

1877

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from

1878 Negative Cooperatiuity of Insulin Receptors

ceptor site concentration) from the intercept at the abcissa. In most cases, however, curvilinear plots, concave upwards,

were obtained. This can be due to a variety of causes, the best

known being (a) the presence of multiple classes of binding

sites that have different but fixed affinities or (b) the existence

of site-site interactions of the type defined as “negative

cooperativity” (29, 30). In the latter case, the affinity of the

receptors is not fixed, but decreases as the occupancy of the receptors increases. In some cases, a mixture of a and b, i.e.

multiple classes of interacting sites, might be present. Steady

state data alone do not discriminate between these models (15,

21). The interaction of insulin with its receptors is a typical

example of curvilinear Scatchard plot. The quantitative analy-

sis of this system using a “multiple classes of sites” model has

been extensively discussed, and the results have been critically

reviewed by Kahn et al. (5, 11). Discrepancies were found

beween the results yielded by applying the above methods to

steady state data and the results obtained from kinetic data

(11). A cooperative model, however, was not considered, and

we decided to test it experimentally. Using an original method

based on the kinetics of dissociation of insulin from its

receptors, we demonstrated site-site interactions of a type

consistent with negative cooperativity upon binding of insulin

to its receptors in human cultured lymphocytes (31). Using the

same method, we showed the absence of cooperativity in the

binding of growth hormone to is receptors on the same cells

(31). In the present study, using the insulin receptors in several

human and animal tissues, we extend further our experimental methods and study the influence of various factors such as

temperature, pH, and ionic strength on the cooperative in- teractions. Furthermore, we present the general implications of

negative cooperativity in hormone binding and action and dis-

cuss alternative models of hormone-receptor interactions.

MATERIALS AND METHODS

Insulin

Porcine insulin (PJ 5589) was purchased from Eli Lilly and Co. It was iodinated with I*7 by a modification (3, 4) of the chloramine-T method (32) (“stoichiometric monoiodination”) at specific activities of 180 to 250 rCi/,ug, equivalent to 0.5 to 0.7 iodine atom/hormone. Monoiodoinsulin prepared at this specific activity by the stoichiomet- ric method has properties which are identical with the carrier-free monoiodoinsulin prepared by other gentle iodination methods (33, 34) and behaves in the binding assay in a way indistinguishable from unlabeled hormone (6).

Receptor Preparations

The characteristics and methods of preparation of the various receptors used in this study have been described in detail elsewhere. The incubation conditions are briefly summarized here. Human cultured lymphocytes, line IM-9 (6, 35, 36), are incubated at 15” in an assay buffer consisting of 100 mM iV-Z-hydroxyethylpiperazine-2- ethanesulfonic acid (Hepes), 120 mM NaCl, 1.2 mM magnesium sulfate, 1 mM EDTA, 10 mM glucose, 15 mM sodium acetate, and 1 mg/ml of bovine serum albumin, pH 7.6. Human circulating blood mononuclear cells prepared by Ficoll-Hypaque gradient according to Boytim (6, 37, 38) are incubated at 23” in the buffer described above, except that the pH is 8.0. A recent study by this group has demonstrated that the monocytes, which constitute about 20% of the cells, are the major insulin-binding cells in this preparation (39, 40). Highly purified liver plasma membranes from rat (41, 42) and mice (43, 44) are incu- bated at 23” in Krebs-Ringer phosphate buffer (NaCI, 118 mM; KCl, 5 mM; MgSO, 1.2 mM; KH,PO,, 1.2 mM; Na,HPO,, 10 mM), pH 7.5.

Experimental Design; Its Theoretical Basis and Controls

The basic experimental design assumes that in the case of negative cooperativity, the decreased affinity will result at least partially from

an increased dissociation rate. When this is the case, it is possible to measure the cooperative interactions by studying the dissociation of bound labeled hormone from the receptors following “infinite” dilution of the hormone .receptor complex and comparing it with dissociation of the same complex in the same medium but in the presence of an excess of unlabeled insulin, which increases the concentration of occupied receptors. Slight differences in methodology are required for whole cells and membrane preparations.

Cells at high concentration in a single batch are reacted with the labeled hormone at low concentration such that only a small minority of the receptor sites are occupied by tracer. (See results and legends of figures for specific conditions in the various experiments.) Association is monitored by centrifugation of aliquots of the incubation mixture in a Beckman microfuge. When a steady state of binding is achieved, the cells are centrifuged at 4”. The supernatant, which contains the un- bound ‘2SI-insulin. is discarded, and the pellet is resuspended im- mediately up to the initial volume with ice-cold buffer. At this point. only a small minority of receptors are filled with labeled hormone, and most receptors are free; the free hormone concentration in the medium at this time is effectively zero. Aliquots (100 ~1) are immediately dis- tributed in two sets of tubes; half contain 10 ml of hormone-free buffer (= “dilution”) and half contain 10 ml of buffer to which unlabeled hor- mone (2 x 10-l) has been added (= “dilution + cold hormone”). To monitor the dissociation, duplicate tubes from each set are centrifuged at 4O, 700 x R for 2 min at regular intervals, the supernatants are dis- carded, and the radioactivity in the cell pellets is counted. When mem- branes instead of whole cells are used, the method is identical, ex- cept that for studying dissociation the content of the dilution tubes is filtered through Millipore filters (0.45 pm). The membranes. which are retained on the filter, are washed with 10 ml of ice-cold buffqr and counted. The dissociation rate by dilution only and by dilution + cold hormone are then compared. Since the interpretation of the experi- ment is based on alterations in dissociation rate constants, an impor- tant condition to measure true dissociation rates in this experiment is that no measurable rebinding of’ the labeled hormone occurs during dis- sociation. When the receptor-bound labeled insulin is diluted in the presence of a large excess of unlabeled insulin (as in the dilution + cold insulin situation), no measurable rebinding of the label will occur due to the isotopic dilution of the dissociating ‘251-hormone. Even with dilu- tion in the absence of unlabeled insulin, no rebinding would occur if the dilution factor were truly “infinite, ” since the concentration of free “‘I- hormone would be maintained equal to zero. However, since the dilu- tion is not infinite, but n-fold, some rebinding will occur in the ab- sence of unlabeled insulin, and the system will approach a new equi- librium where the new B/F = l/n x the B/F at the time zero for dissoci- ation.

Indeed, if the reaction of hormone with receptor were a simple, reversible, bimolecular one that obeys the law of mass action, then

H + R z HR and K. = [HR]/(H] [RI,

where [H] = concentration of free hormone, [R] = concentration of free receptor sites, [HR] = concentration of bound hormone = concentra- tion of occupied receptor sites, K. = affinity constant, p] + [HR] = p.1 = the total concentration of receptor sites, and

B/F = [HRl/[Hl = K.[Rl = KAlRol - Ml).

When only a small minority of the receptor sites are occupied, i.e. [HR] << &I, then [RI, the concentration of free receptor sites, approaches’ the total concentration of receptor sites, [R.]. Hence,

[HRl/[Hl(orBP) -&FL!.

Thus, at the end of the association period of our experiment,

WLI/[H,I = B,/F, 1 K.[R,I.

When aliquots are then diluted n-fold, the reaction tends toward a new equilibrium where

‘This approximation is made for the sake of simplicity. This demonstration can be made more rigorously by developing the law of mass action into a function of R, and using the exact, but more complex, formula:

(HR] = FL1 + [%I + l/K, {([H,] + [R,] + l/K,)? - 4 [H,] [R,]JH 2

(De Meyts, unpublished results).

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from

Negative Cooperativity of Insulin Receptors 1879

TABLE I

Rebinding of hormone

B/F at new steady state

Dilution: Dilution: P n-fold

With unlabeled hormone (dilu- 0 0 tion + cold insulin)

Without unlabeled hormone (di- 0 (B/F), x l/n lution only)

[HR,I/[H,I = B,/F, = K.&l/n = [HR,II[H,l~lIn.

In other words, the B/F after dissociation will be l/n of the B/F at the end of association. If the free hormone is discarded before dilution, this new B/F will be even lower since it will re-equilibrate from only the hormone bound at the end of the association. The “rebinding” problem is summarized in Table I.

In practice, one wants to increase n, the dilution factor, until the residual B/F is negligible, i.e. smaller than the error in measuring B/F. If the reaction of [H] with [R] is not a simple bimolecular reaction, this remains true provided that the association rate constant does not increase when [HRV[R,] decreases during the dissociation. In any case, we always demonstrate experimentally the absence of significant reassociation of ‘261-insulin (see below),

In the dilution + cold situation, the unlabeled hormone will bind to free receptor sites. This will not affect the rate of dissociation of tracer from filled sites unless the filling of the empty sites by unlabeled hormone induces site-site interactions,’ or more exactly, the two rates of dissociation (dilution only and dilution + cold hormone) will not differ at any point by more than l/n x (B/F at t = 0). As we have reported before, this was the case for growth hormone and its receptors in human cultured lymphocytes, showing that the receptor sites were independent. This was consistent with the linear Scatchard plot of steady state binding. In contrast, insulin receptors exhibited the behavior expected from a negatively cooperative system, as illustrated in Fig. 1; the presence of a saturating concentration of insulin markedly speeds up the dissociation of labeled insulin, demonstrating the existence of site-site interactions between sites filled with unla- beled insulin and sites filled with labeled insulin.

Experimental Control of Absence of “Rebinding”

From Bulk Solution-We use two methods as controls for the absence of rebinding (31). (a) Dilution alone is compared with dilution plus unlabeled hormone over a wide range of dilutions: above a 50-fold dilution, the difference in dissociation between dilution only and dilution + cold is independent of the dilution factor (31). (b) The two sets of cells (dilution only and dilution + cold) are prepared in duplicate; in one of each, before distributing the aliquots for dissocia- tion, an additional equal aliquot of fresh cells that have never been exposed to hormone is added. If rebinding occurs during dissociation, it will occur to a more than doubled population of free receptors and should then be magnified. This did not happen in our experimental conditions (31).

From “Unstirred Layers”-Even in these situations, however, reassociation of labeled insulin could still occur inside a hypothetical unstirred layer (see “Discussion”) surrounding the cells; the effect of unlabeled hormone could be due to isotopic dilution of labeled insulin in the unstirred layer. This possibility can be reasonably excluded by the fact that the increased dissociation is not observed for the growth hormone receptor on the same cells, and by the high degree of structural specificity of the observed effect. Indeed, some insulin analogues, desalanine-desasparagine insulin and desoctapeptide in- sulin, which at high concentration saturate insulin receptors, totally fail to increase dissociation of labeled insulin:5 these results exclude a nonspecific diffusion barrier (45).

’ This is so unless cold hormone can dimerize or polymerize with the label, and the polymer has a different affinity, which we think unlikely in the case of insulin (see “Discussion”).

‘P. De Meyts, J. Roth, D. Brandenburg, and H. Gattner, manu- script in preparation.

1; ~c;F:;i~] 5ol y-y-y 0 30 60 0 30 60

MINUTES OF DISSOCIATION

FIG. 1. Effect of unlabeled insulin on the dissociation of ‘*4-insulin. Left “‘I-insulin (5 x lo-” M) was incubated with human cultured lymphocytes of the IM-9 line (2.5 x 10’lml) for 30 min at 15’ and pH 7.6 in a total volume of 2.5 ml, after which the cells were sedimented at 4”. The supernatant was discarded and replaced by an equal aliquot of chilled (4’) fresh medium, the cells were resuspended, an aliquot (100 ~1) was removed for measurement of bound radioactive hormone (“time zero”), and aliquots (100 ~1) were transferred to a series of tubes that contained 10 ml of medium in the presence and absence of un- labeled hormone (1.7 x 10-l M, 15’). At intervals, two tubes of each set were centrifuged, and the radioactivity in the cell pellet was counted. The radioactivity on the cells, expressed as a percentage of the radioac- tivity present at time (t) = 0 min, is plotted as a function of the time elapsed after the dilution of the system. Each point is a mean of dupli- cates, which differed by less than 5%. When the bound radioactivity was measured at the completion of the incubation before the sedimentation step, it differed by less than 5% from that measured at time zero. Typically with these conditions, IM-9 lymphocytes bound 70% of the added tracer, and about 5Yo of the receptor sites were occupied. Right, the experimental conditions are as in the left-hand figure, except as follows: ‘*sI-insulin was incubated with purified plasma membranes of rat liver (1 mg/ml) for 60 min at 20°, pH 7.8; dissociation was at 20”. At intervals, the contents of two tubes of each set were filtered through Millipore filters (0.45 nm). The filters were rinsed once with 10 ml of ice-cold buffer, and the radioactivity on the filter was counted. Typically with these conditions 12% of the added tracer is bound to membranes, and about 5% of the receptor sites were occupied.

Effect of “Nonspecific” Binding

To assess the participation of the nonspecific compartment in the observed dissociation rates, a tracer size of labeled insulin was associated in the presence of a large excess of unlabeled insulin (256 pg/ml or 4.25 x lo-’ M); in this case, most of the binding is to the nonspecific compartment. The cells or membranes were then submit- ted to the dissociation experiment in the absence and in the presence of cold hormone.

As seen in Fig. 2, the nonspecifically bound radioactivity, represent- ing only 2.5% of the total tracer, dissociated almost instantaneously up to a trivial residual (and irreversible) binding of 0.3Y~; this was not affected by the presence of unlabeled hormone in the dilution. Specific binding measured in the same conditions was 60% of the total tracer (data not shown). In these experiments, the dissociation rates observed are thus negligibly affected by the nonspecific binding. However, in systems where nonspecific binding represents an appreciable part of the total binding, it is necessary to include this control in each experiment and to correct the dissociation curves. The basic experi- mental design, being well controlled, was then applied to a variety oP experimental conditions.

RESULTS

Effect of Source of Insulin Receptors-Documented studies have shown that the properties of the insulin receptor including specificity toward a wide range of insulin analogues (46,47) are identical in a large variety of tissues and animal species (48). This was true also for the negative cooperativity. We observed accelerated dissociation of l*sI-insulin by unlabeled insulin independently of the tissue used as a source of insulin receptors as well as of the animal species used (Figs. 1 and 3).

We found results qualitatively similar for insulin receptors in human cultured lymphocytes, human blood peripheral mono-

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from

1880 Negative Cooperativity of Insulin Receptors

E H

s o.5 fk$o- . cDILUTION ONLY ~ ,

5 7

DILUTION + COLD INSULIN

” 0.2 1 ' b ' b ' c ' k 'I ' 0 10 20 30 40 50 60

MINUTES OF DISSOCIATION

FIG. 2. Dissociation of nonspecific binding [**‘I]Insulin (1.7 x 10-O M) and unlabeled insulin (4.25 x 10e6 M) were incubated with IM-9 lymph&ytes (2.5 x lO’/ml). Conditions are as described in the legend to Fig. 1 (left). The radioactivity bound to the cells at each time point is expressed as a percentage of the total radioactivity with which these cells had been incubated.

= m 1 HUMAN LYMPHOCYTES ( MOUSE LIVER I E '6 100 - E

z

5 80-

; - lp t; 60-

It! - 5 $J 40- F - 2 2 zo-

8 ou lo-" 10-9

INSULI~~~NCE%ATI~~ 10 -7 10-5M

FIG. 3. Effect of insulin concentration on dissociation of ‘Y-insulin. The binding and dissociation of ‘Y-insulin with human cultured lymphocytes (left) and mouse liver membranes (right) were studied as described in the legend to Fig. 1; 0.1 ml of the resuspended cells or membranes was transferred to 10 ml of medium that contained no unlabeled insulin or unlabeled hormone at lo-” M to 10.’ M. After 30 min of dissociation at 15” (lymphocytes) or 20” (liver membranes), the cells or membranes were separated from the medium, and the bound radioactivity was counted. The difference in the radioactivity present on the cells with dilution only and dilution + cold insulin was expressed in percentage of the maximal difference observed, and was plotted as a function of the concentration of unlabeled insulin. Note that circulating insulin levels in peripheral blood of normal man can reach 10es M, and go up to lo-* M in some severely insulin-resistant patients. Levels up to 10-O M are commonly found in obese mice.

cytes, and rat and mouse liver plasma men:branes. The only difference was that the overall kinetics of insulin dissociation from receptors was slower in membrane preparations than in whole cells. The reason for this difference is unknown. The ordinate scale in Fig. 1 (right) was adjusted to match the dissociation rates of Fig. 1 (left). In these conditions, it is clear that the relative increase in the presence of unlabeled insulin was comparable. One or more of these various preparations was used for subsequent experiments reported in this paper. Other workers in our group have also found accelerated dissociation

by dilution + cold insulin for insulin receptors in human cultured placental cells (49), human peripheral granulocytes (50), human cultured fibroblasts,’ and turkey erythrocytes. a Comparable experiments using soluble receptor preparations have also demonstrated negative cooperativity.’

Effect of Insulin Concentration and Fractional Saturation of Receptors-The decrease (due to unlabeled insulin) in the amount of radioactivity remaining bound after a given period of dissociation has been used to express the extent of site-site interactions as a function of the concentration of unlabeled insulin in the dilution medium; the dose-response curve is extremely sensitive (Fig. 3), with 20% of the total increase in dissociation observed with as low as 2 to 5 x lo-“’ M insulin, and a maximum effect between lo-* and lo-’ M. A further increase in insulin concentration results in a decrease in the cooperative effect, likely due to insulin dimerization (see below).

The sensitivity of the dose-response curve for increased dissociation suggested that significant cooperativity may al- ready be induced by a small fractional saturation of the receptors (1 to 5%) (20). The effect of the fractional saturation of the receptors on the dissociation rate of the complex was tested directly in incubating mixtures of labeled and unlabeled insulin at increasing concentrations and submitting the result- ing complex to dissociation by dilution and dilution + cold. In one experiment (Fig. 4), labeled insulin at two concentrations (2.2 x 10-l’ M and 1.7 x lOA1 M) was incubated up to a steady state of binding with cells at high concentration. After discard- ing the free hormone, aliquots of the resuspended cells were submitted to dissociation by 100x dilution in the absence and in the presence of 1.7 x lo-’ M unlabeled hormone. The initial binding of 2.2 x lo-” M labeled insulin was 60% of the tracer uersus 12% for 1.7 x lo-’ M tracer. The half-time of dissocia- tion for the smaller tracer by dilution only was more than 60 min uersus 5 min for the bigger tracer. With dilution + cold, the relative increase due to unlabeled hormone in the dilution medium was much smaller for the bigger tracer. Thus a higher occupancy by labeled hormone resulted in a faster dissociation by dilution only and reduced the observed difference in the presence of cold hormone.

In a second experiment (Fig. 5), mixtures of labeled and unlabeled insulin totaling lo- I0 to 10m5 M insulin were associ- ated. After discarding the free insulin, aliquots of each tube were allowed to dissociate in a 100x dilution in the absence and in the presence of unlabeled hormone for 30 min. The binding of tracer at time = 0 min was normalized to 100%. The amount of the initial radioactivity remaining bound was then determined and plotted as a function of tracer size. As the tracer size (and the occupancy of receptors) was increased, the rate of dissociation by dilution alone increased and reached a maximum at lo-’ to lOme M, at which no difference was observed between dilution and dilution + cold. At this point, most of the tracer was bound nonspecifically. In other words, the dissociation rate by dilution only of a tracer which occupies an appreciable fraction of the sites may already be a “coopera- tively” increased rate; it is thus important, if one wants to observe a more “basal” rate by dilution only, to optimize the experiment by reducing the size of the tracer as much as allowed by counting accuracy.

‘M. M. Rechler and J. Podskalny, submitted for publication. ‘B. Ginsberg, C. R. Kahn, and J. Roth, submitted for publication. ‘B. Ginsberg, R. Cohen, and C. R. Kahn, manuscript in preparation.

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from

Negative Cooperativity of Insulin Receptors

DILUTION ONLY

0

TRACER = 2.2 x lo-“M

DILUTION t COLD INSULIN

11 1 ’ ’ ’ 1 ’ I ’ ’ ’ ’ ’ 0 10 20 30 40 50 60

MINUTES OF DISSOCIATION FIG. 4. Effect of the fractional saturation of receptors by insulin on

dissociation of ‘211-insulin. ‘*‘I-insulin (2.2 x lo-” M) in the absence or presence of unlabeled insulin (1.7 x 10m7 M) was incubated with IM-9 lymphocytes. The experimental conditions are as described in the legend to Fig. 1 (left). The radioactivity bound to the cells at each time point is expressed as a percentage of the total radioactivity with which these cells had been incubated.

Effect of Other Compounds-Increased dissociation of la- beled insulin from the receptor was not obtained with glucagon, adrenocorticotropic hormone, thyrocalcitonin, and growth hor- mone. Prostaglandin E,, GTP, and theophylline also had no effect.

Effect of pH-The dissociation by dilution only was mark- edly dependent on the pH (Figs. 6 and 7) with a markedly increased rate of dissociation at more acid pH. For example, at pH 6.5, 83% of the label had dissociated after 10 min of incubation in contrast to only 23% at pH 7.6. The sharp sigmoid pH curve observed in the dissociation by dilution only has a midpoint at pH 7 in human cultured lymphocytes and at pH 7.5 in the peripheral monocytes. The dissociation in the presence of excess cold insulin appeared much less pH-depend- ent: whereas the tracer dissociated in 30 min by dilution only was increased by 600% from pH 9 to 6, the tracer dissociated in 30 min by dilution + cold insulin was only increased by, respectively, 35% (lymphocytes) and 11% (monocytes) from pH 9 to pH 6.

Effect of Temperature-Dissociation of the labeled insulin. receptor complex by a lOO-fold dilution only was also markedly affected by temperature (Fig. 8). The dissociation rate at 4” appeared to be first order (within the limits of time studied) and very slow (only 20% dissociated in 3 hours). When tempera-

ture was increased, the dissociation curve became multiexpo- nential with a marked acceleration of the overall dissociation

(95% dissociated in 3 hours at 37”).

Comparison of dissociation by dilution only and dilution + cold insulin at various temperatures is shown in Fig. 9. The

dissociation by dilution + cold insulin was also accelerated by

increasing temperature, but to a lesser extent than dilution

DILUTION ONLY

1881

M

TRACER ‘)I I-INSULIN CONCENTRATIDN FIG. 5. Effect of fractional saturation of the receptors by insulin on

the dissociation of ‘z’I-insulin. [‘Y]Insulin (1.7 x 10-i’ M) enriched with unlabeled insulin (0 to lo-’ M) was incubated with lymphocytes for 30 min, diluted 1:lOO in the absence (dilution only) or presence (dilution + cold insulin) of unlabeled insulin (1.7 x 10.’ M), and after 30 min of dissociation the cells were sedimented, and the radioactivity was counted. The experimental conditions are described in detail in Fig. 1 (left). The radioactivity present on the cells after 30 min of dissociation is expressed as a percentage of the radioactivity present at time (t) = 0 min and is plotted as a function of the concentration of insulin initially incubated with the cells.

[

DILUTION ONLY

I I I J 0 10 20 30 40 50 60

MINUTES OF DISSOCIATION

FIG. 6. Effect of pH on the dissociation of ‘Y-insulin. ‘“I-Insulin (lo-” M) was incubated with IM-9 lymphocytes at pH 7.6. Conditions are as described in the legend to Fig. 1 (left), except that dissociation was studied by dilution only, using insulin-free medium at pH 7.6 and 6.5.

only; as a result, the difference between dilution only and dilu- tion + cold decreased markedly with increased temperature.

The extent of dissociation was plotted by analogy with Arrhenius plots as a function of the reciprocal of the absolute

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from

1882 Negative Cooperativity of Insulin Receptors

90 CULTURED LYMPHOCYTES

8 4 F x

80 t

20 + COLD INSULIN

10

CIRCULATING

MONOCYTES

DILUTION ONLY

+ COLD INSULlh

I-II 5 6 7 8 9 5 6 7 8 9

PH

FIG. 7. Effect of pH on the dissociation of ‘zSI-insulin. Left, [Y]- insulin (5 x 10-l’ M) was incubated with cultured (IM-9) lymphocytes at pH 7.6. Conditions are those described in Fig. 1 (left), except that the pH of dilution medium ranged from 5 to 9, and the dissociation period was for 30 min. The data are expressed as in Fig. 5. R&t, same as left, except that circulating monocytes (7 x lO’/ml) were incubated with the labeled hormone for 180 min at 23”, pH 8.0, and the pH of the dilution medium ranged from 6 to 9.

100

50

10

5 0 60 120 180

MINUTES OF DISSOCIATION FIG. 8. Dissociation of ‘251-insulin at different temperatures. I*-%

Insulin (lo-” M) was incubated with human peripheral monocytes (7.2 x 10’ cells) at 23” for 180 min at pH 8.0. The dissociation was studied with insulin-free medium (i.e. dilution only) at 4’ to 37”.

temperature (Fig. 10). For dissociation by dilution + cold, a straight line was obtained; dilution only generated a biphasic plot with a break around 21°, and there was a lo-fold difference between the two slopes. The steeper slope corresponds to temperatures below 21”; the shallow slope, to temperature above 21’. The later slope is identical to the slope of the temperature dependency of the dissociation by dilution + cold at all temperatures.

The same plot was performed for the temperature depend- ency of the percentage of tracer dissociated after more pro- longed dissociation periods (data not shown). Dissociation by dilution + cold insulin yielded straight lines with comparable slopes for all times of dissociation. Dissociation by dilution only still yielded biphasic curves, but the sharp slope observed

(I 0 60 120 180 60 120 180 60 120 180 60 120 180 MINUTES OF DISSOCIATION

FIG. 9. Effect of temperature on dissociation of ‘zSI-insulin. The experiment is that in Fig. 8, except that dissociation was studied in the absence (dilution only) as well as in the presence of 1.7 x lo-’ M unla- beled insulin (dilution + cold insulin).

3.2 i.- 3.3

\

DILUTION ONLY

-I-I --I

3.4 3.5 3.6

1 - x 103 T

FIG. 10. Effect of temperature on dissociation of ‘211-insulin. The data from the experiment in Fig. 9 are replotted. The radioactivity dissociated from the cells by 30 min, expressed as a percentage of the radioactivity bound to the cells at time (t) = 0 min is plotted as a func- tion of the reciprocal of the absolute temperature.

below 21” decreased with time and became closer to the slope of the plot for dilution + cold as dissociation progressed.

Effect of Urea-Dissociation of the labeled insulin.receptor complex by dilution only was markedly increased by urea (Fig. 11). As little as 1 M urea had a very significant effect, with a resulting 4-fold decrease in the overall half-life of the insulin. receptor complex; 6 M urea produced dissociation of 95% of the complex by denaturation of insulin or the receptor, since of urea is interesting (Fig. 12): a smooth curvilinear plot was observed, which suggests that urea does not dissociate the complex by denaturation of insulin or the receptor, since denaturation usually results in a plateau with a sharp break at the urea concentration that unfolds either molecule (51). The half-maximal dissociating effect of urea was observed with 1.2 M urea, which corresponded roughly to the maximal dissocia; tion obtained in the same dissociation time with dilution + unlabeled insulin in the absence of urea.

Effect of Dioalent Cations-Divalent cations affected the dissociation rate of insulin from the receptors. Dissociation by dilution only was decreased by both Mg2+ and Cal+ (illus- trated for Mg2+ in Fig. 13); 1.2 mM Mg*+ resulted in a significant decrease, whereas 12 mM induced a marked de- crease.

The effect of divalent cations on the dissociation by dilution + cold insulin is shown in Fig. 14. The dose-response curve for

the enhanced dissociation of labeled insulin as a function of the concentration of unlabeled insulin was shifted by both Caz+

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from

Negative Cooperatiuity of Insulin Receptors 1883

DILUTION ONLY

1’ 1 5 10 30 60

MINUTES OF DISSOCIATION FIG. 11. Effect of urea on the dissociation of 124-insulin. ‘*4-Insulin

(IO-” M) was incubated with IM-9 lymphocytes at pH 7.6. Conditions were as described in the legend to Figure 1 (left), except that dissocia- tion was studied by dilution only, using insulin-free medium containing lMor6Murea.

DILUTION ONLY

0 1 2 3 4 5 6M UREA CONCENTRATION

FIG. 12. Dose-response curve for the effect of urea on dissociation of ‘24-insulin. The experimental design is identical with that in Fig. 11. Dilution was performed in insulin-free medium containing urea concen- trations from 0 to 6 M. The radioactivity on cells after 30 min of disso- ciation, in percentage of radioactivity bound at time (t) = 0 min, is plotted as a function of urea concentration. The arrow on the vertical axis indicates the extent of dissociation obtained by dilution and unlabeled insulin in the same dissociation time, in the absence of urea.

and Mg*+ in such a way that the accelerating effect of insulin concentrations below lOme M was decreased. The effect of

higher insulin concentrations, as well as the maximal effect, was not significantly modified.

DISCUSSION

Insulin binding to its receptors in several human and animal

tissues has been studied in detail at steady state and in kinetic

x 100

‘:: ; 90 DILUTION ONLY

4

60 120

MINUTES OF DISSOCIATION

FIG. 13. Effect of magnesium on the dissociation of ‘*?-insulin. ‘ZsI-Insulin (lo- ‘I M) was incubated with IM-9 lymphocytes at pH 7.6. Conditions were as described in the legend to Fig. 1 (left), except that dissociation was studied by dilution only in insulin-free medium con- taining 0, 1.2, and 12 mM MgCl,.

8 .-I x

h

20 1 I I I I I I I lo-” IO-‘0 10-g 10-8 10-7 10-11 lo-‘0 10.9 1o-8 10

L. -7 I INSULIN MOLARITY

FIG. 14. Effect of divalent cations on the insulin-induced accelera- tion in dissociation of LZKI-insulin. ‘2sI-Insulin (5 x lo-” M) was incu- bated with IM-9 lymphocytes under conditions identical with those in the legend to Fig. 1 (left). The cells were then diluted in sets of tubes containing 10 ml of either Mgz+-free and Caz+-free medium, or in medium containing 12 mM MgCl, or CaCl,, in the presence of unlabeled insulin at lo-” to lo-’ M. The radioactivity bound to cells after 30 min of dissociation, in percentage of the radioactivity bound at time (t) = 0 min, is plotted as a function of the unlabeled insulin concentration.

experiments. Steady-state data alone are consistent with

several models, the three most likely being: (a) insulin binds in a simple bimolecular reversible reaction to several classes of receptor sites, each with a discrete fixed affinity; (b) the receptor sites constitute a homogeneous population, but their occupancy with insulin induces site-site interactions which decrease their affinity, a phenomenon which would come under the more specific heading of negative cooperativity; (c) the receptor sites are heterogeneous and undergo negatively coop- erative interactions.

To avoid semantic confusion, we want to make precise that we, as well as many others (29, 30, 52, 53), use the term negative cooperativity in a purely descriptive sense for interac- tions resulting in a decrease in the apparent affinity of receptors for insulin when fractional saturation of the receptors increase, a binding isotherm that is shallower than the curve for “statistical” binding and with a Hill coefficient (54) smaller than 1.0. By using the term negative cooperativity, no assump- tions are being made as to the mechanistic or molecular nature of the events which cause the decrease in affinity (tertiary or

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from

1884 Negative Cooperativity of Insulin Receptors

quaternary conformational changes, receptor aggregation or insulin and its receptors, we consider this hypothesis unlikely dissociation, etc.). on the basis of the findings that follow.

Distinguishing between negative cooperativity occurring on ligand binding and pre-existing asymmetry or heterogeneity of the sites has been a long debate in the study of equilibria between macromolecules and ligands (see Matthews and Bernhardt for a review, 55). The kinetic method presented here could be of widespread applicability for making such distinc- tions.

A method analogous in principle to ours has in fact been applied to the study of the cooperative kinetics of hemoglobin (56, 57): dissociation of oxygen was studied in the presence of an excess dithionite or sodium hydrosulfite which will react with free 0, and thus maintain its concentration at zero (as does dilution only in our experiment), or in the presence of carbon monoxide, a competing ligand of high affinity which will rapidly fill empty hemoglobin binding sites and replace the dissociating 0, (as does dilution + cold in our experiment). Since hemoglobin is positively cooperative, the dissociation rate of O2 was slowed down by carbon monoxide.

1. We observe (Fig. 3) accelerated dissociation of ‘Y-insulin in the presence of concentrations of insulin as low as lo-lo M;

20% of the maximal increase is obtained with 5 x lO-‘O M, a concentration at which there is one dimer for 26,000 monomers (60). Blundell et al. (61) have calculated that with insulin at lo-* M, pH 8.0, conditions at which our observed cooperative effect is approaching maximal, the final mole fraction of dimers would be only 0.00214.

In the case of insulin receptors, we have demonstrated that filling empty receptor sites with unlabeled insulin affects the rate at which sites occupied with labeled insulin dissociate in an infinite dilution, demonstrating site-site interactions of a destabilizing (i.e. negatively cooperative) type. These data cannot be explained by site heterogeneity alone, but do not preclude its concomitant occurrence. Heterogeneity of sites alone would cause the labeled species to dissociate in a multiexponential way, but would not cause an increase in that dissociation when empty sites are filled with the unlabeled species. In many of the experiments presented here, the dissociation of labeled insulin by “dilution only” was multiex- ponential (Fig. 1). This could be caused by pre-existing heterogeneity of binding sites or by site-site interactions among the filled sites themselves. Two findings tend to favor the latter hypothesis and argue against pre-existing heterogeneity.

2. The cooperative effect in fact decreases progressively at insulin concentrations of >lO-’ M, where it is known that dimers become a significant proportion of the molecular species (31, 45), and is totally lost with lo-’ M insulin. The fall in cooperative effect superimposes with the theoretical curve for dimer formation, suggesting that dimers bind to the re- ceptor, but do not induce site-site interactions.3 In a separate study, dose-response curves for accelerated dissociation of labeled insulin in the presence of unlabeled insulin were per- formed with several dozen analogues of insulin, varying lOOO-fold in their affinity for the insulin receptor and, pro- portionally, in their biological potency. We showed that whereas most analogues exerted the negative cooperativity in direct proportion to their affinity for the insulin receptor, some chemical modifications of insulin, all localized in a well de- fined area at the surface of the monomer, destroyed the ability of insulin to induce the site-site interactions, but not the ability to occupy the receptors and induce biological effects. This led us to conclude that there exist separate bioactive and cooperative sites on both insulin and the re- ceptors (31, 45). The area corresponding to the cooperative site is buried in the dimer, which explains why the dimer would not induce the cooperativity.

1. At 4”, the dissociation by dilution only is first order; even at 15’, a sufficient reduction in the initial occupancy by decreasing the concentration of 1261-insulin results in an apparent first order dissociation rate (data not shown).

3. Nondimerizing insulin species, tetra(nitrotyrosine)- insulin (62) and guinea pig insulin (63-66) induce the acceler- ated dissociation in strict correlation with their relative ability to bind to the insulin receptor sites (6, 45). Since they do not dimerize, no fall in the cooperative effect is observed at high concentrations.

2. Concanavalin A, which does not bind to the insulin binding site but inhibits the site-site interactions and prevents the effect of unlabeled insulin or dissociation almost linearizes the Scatchard plot without changing the binding capacity (58, 59).’ When negative cooperativity is suppressed, concomitant site heterogeneity will cause residual nonlinearity in the Scatchard plot: this was small if present at all (data not shown).

4. Desalanine-desasparagine insulin dimerizes with an asso- ciation constant which is 100 times lower than insulin (67). However, even when present at concentrations 10,000 times higher, it does not accelerate the dissociation of ‘*‘I-insulin (45). Again, dimerization appears unrelated to the negative cooperativity.

In interpreting the results obtained with our dissociation method, the following models should also be critically exam- ined before attributing the negative cooperativity to site-site interactions.

Ligand-Ligand Interactions-If the unlabeled hormone could form dimers or polymers of a higher order with the labeled hormone bound to the receptor sites, and if the polymer had a rate of dissociation from the receptor that is fast relative to that for bound monomer, accelerated dissociation of the label could be observed in the presence of concentrations of unlabeled hormone at which polymerization becomes signifi- cant, especially if the rate of dimerization is fast. In the case of

‘P. De Meyts, J. R. Gavin, III, and J. Roth, manuscript in preparation.

In a recent article (68), Cuatrecasas and Hollenberg con- firmed our initial report that insulin accelerates its own dissociation, but challenged our interpretation by attributing this effect to insulin dimerization. However, they did not mention that we reported some of the above controls in our initial paper (31), and they further supported their view with inaccurate quotations from our paper and from the literature. We did not use insulin concentrations “which are all super- saturating with respect to receptor binding;” in fact, we showed that occupancy of a small percentage of the sites induces significant cooperativity (31). Zimmerman et al. have not “failed to demonstrate dimerization” of guinea pig insulin due to the low insulin concentration used in sedimentation equilibrium studies (64). In a complete study (65, 66) following their preliminary report (64) Zimmerman et al. indeed clearly demonstrated that guinea pig insulin remains monomeric even at concentrations of 3 mg/ml (5 x lo-’ M); Cuatrecasas and

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from

Negative Cooperatiuity of Insulin Receptors 1885

Hollenberg did not mention this work (68). Finally, Pekar and Frank (60) did not report a dissociation constant for dimeriza- tion of 7 x 10m7 M as quoted by Cuatrecasas and Hollenberg (68); they reported an equilibrium constant of 1.4 x lo6 liters/mol, which yields a dissociation constant of -7 x lo-'M.

This agrees well with the value reported by Goldman and Carpenter (67) and is consistent with the fall in cooperative effect observed in our studies. (In fact, this K, coincides almost exactly with the concentration at which half of the cooperativ- ity has disappeared: Fig. 3.)

In the same article (68) Cuatrecasas and Hollenberg have also shown that insulin accelerates the dissociation of labeled insulin from talc, and speculate that this could be due only to ligand-ligand interactions. This is likely but not necessarily true, since silica surfaces have a net electric charge, and it cannot be excluded, for example, that adsorption of insulin could modify the repartition of charges and induce site-site interactions by altering coulombic forces at the surface (69). Moreover, it should be noted that in contrast with the cooperative effect observed on cells at physiologic insulin concentrations, accelerated dissociation from talc was studied only at insulin concentrations ranging from 1 to 100 rg/ml, at which insulin dimerization is expected to occur. This is exactly the range of concentration at which the accelerating effect on cells decreases and finally disappears (Fig. 3). We thus agree that the effect observed by Cuatrecasas and Hollenberg on talc is more likely due to insulin dimerization, and we have likewise attributed to insulin dimerization the loss of cooperativity observed in our studies with cells at very high insulin concen- trations.

In summary, the interpretation that accelerated dissociation is due to the insulin monomer and, if anything, is lost with the dimer, is the most consistent with the presently known properties of the natural insulin monomer and dimer.

One may argue that the above argumentation, based on the known properties of the natural dimer, may not be relevant since the bound dimer need not necessarily have the same structure as the conventional dimer. However, to explain the acceleration of labeled insulin by formation on the receptor of a totally new species of dimer, one would have also to explain how three dozen analogs of insulin, ranging lOOO-fold in affinity for the receptor, exert the cooperativity in direct proportion to their affinity for the receptor (31, 45). This is understandable if the analogs bind to a receptor site and induce site-site interactions, but not if the analogs dimerize with a bound insulin monomer, unless the affinity of the analogues for bound insulin happens to be superimposable to their affinity for the receptor itself, which seems highly unlikely. This hypothesis would also leave unexplained the disappearance of the acceler- ated dissociation at higher insulin concentrations.

The negativity cooperativity observed with other hormones also appears to be unrelated to polymerization of the hormone. Growth hormone, which does dimerize (70), does not accelerate its own dissociation. The glycoprotein hormones all dimerize poorly, but luteinizing hormone and human chorionic gonado- tropin receptors do not behave cooperatively (23, 26), whereas thyroid-stimulating hormone, which has the worst dimeriza- tion, markedly accelerates thyroid-stimulating hormone dis- sociation (71). Catecholamines, which have been shown by NMR to be monomeric even at 0.1 M (72) accelerate the dissociation of the labeled antagonist [SH]alprenolol from @-adrenergic receptors (73).

Less specific types of ligand-ligand interactions, like steric or electrostatic hindrance, would explain curvilinear Scatchard plots, through a progressively decreasing association rate but not an accelerated dissociation rate of already bound labeled molecules when the complex is diluted in the presence of an excess of unlabeled molecule. Such a model seems also ex- cluded by the high degree of structural specificity of the cooperativity.

The binding of flexible ligands can also yield curvilinear Scatchard plots (74). This model seems inconsistent with the highly structured tertiary and quaternary conformation of insulin (but should be considered in the case of less rigid mole- cules like adrenocorticotropic hormone) and does not explain the accelerated dissociation of a ligand molecule by others.

Unstirred Layers (“Noyes-Whitney” Layers, Diffusion

Boundary Layers)-The role of liquid stationary films as diffu- sion barriers* has been considered in various areas of research, from batch adsorption of gases (76, 77) to membrane transport kinetics (78, 79). This model has been ruled out under “Mate- rials and Methods” section. Also, the observed effect occurs at hormone concentrations (lo-” to lo-’ M) well outside the range where the rate of mixing of reactants is the rate-limiting step (80).

Our data indicate that a highly specific mechanism regulates the affinity of the receptor sites according to the concentration of insulin available to the cell.

Although we have as yet no direct insight into the precise molecular nature of the site-site interactions, our data can be described on a purely phenomenological basis by the following model. The receptor sites can exist reversibly in at least two conformational states: a slow dissociating state and a fast dissociating state. The sites are able to undergo a transition from the slow dissociating state to a fast dissociating state when an increased fraction of them is occupied by insulin. This transition is facilitated by increased temperature, lowered PI-I, and by a low concentration of divalent cations (Table II).

The effect of temperature on dissociation rates is complex. Superficial examination of the data may lead to the conclusion that since the difference between the dissociation by dilution only and dilution + cold hormone decreases above 21”, there is less negative cooperativity at physiologic temperatures. How- ever, it should be kept in mind that the experiment is designed to detect the difference in cooperativity which exists between the situation in which few sites are occupied by the tracer only and the situation in which the receptors are, in addition, saturated to a variable extent with unlabeled insulin. In cases where cooperativity would already be maximal with the frac- tional saturation achieved by the tracer itself, no increase in the presence of unlabeled insulin would be found. Thus, increased temperature may very well decrease the threshold at which occupancy induces significant site-site interactions. The Arrhenius plots from dissociation curves for dilution + unlabeled insulin are single parallel lines at all time points; the part of the curve for dilution only above 21” is also par- allel to the former ones. This could be interpreted as indi- cating that a single activation energy governs dissociation

8Reassociation of labeled ligand inside the unstirred layer may be favored by “retention” effects exerted by high local concentrations of the binding protein, as recently theorized by Silhavy et ai. (i5). Such models may be valid in some cases, but in the case of insulin receptors, the same specificity arguments which make unlikely the “unstirred

layer” hypothesis also apply to the retention effect.

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from

1886 Negative Cooperativity of Insulin Receptors

TABLE II

“Two-state” model for insulin receptors

“Slow-dissociating state” favored by:

Low fractional saturation with insulin

Alkaline pH (S-9) Low temperature (4’) High concentration of Cal+/

Mg*+ Concanavalin A (20 rglml)

“Fast-dissociating state” favored by:

Increased fracional saturation with insulin

Acid pH (5-6) High temperature (37”) Low concentration of Cal+/

Mg2+

Urea”

“The fact that urea can promote almost complete dissociation of insulin from receptors within minutes indicates that no convalent bonds are involved in insulin binding or in the mechanisms regulating insulin dissociation. Direct effects of urea on the insulin-receptor interaction cannot, however, be distinguished from possible effects on intersubunit bonds in the receptor.

in these conditions, which would correspond to the fast dissociating conformation. The slope of dissociation under 21” for dilution only would then represent a IO-fold increase in the activation energy, possibly representing the energy necessary to shift from the slow dissociating state to the fast dissociating state when only a small minority of the sites is occupied with labeled insulin.

The nature of the site transition from one state to another

could be one of the following, by analogy with known coopera- tive phenomena in other macromolecules.

1. A change in the tertiary and/or quaternary configurations’ of individual oligomeric receptors (81-84). One such model is schematically illustrated in Fig. 15, by analogy with the struc-

ture and function of hemoglobin.

2. A reversible aggregation or dissociation of subunits (85, 86). Singer and Nicolson, in 1972 (87), suggested that the clus- tering of binding sites through ligand-induced movements in the fluid membrane would confer to these sites cooperative properties (see Ref. 88 for further discussion). Another possibil- ity is the liberation of an “effector” from a dissociating receptor which would “mediate” the site-site interactions (89).

3. A combination of both where conformational changes in individual oligomeric receptors are further propagated in other receptors when clustering occurs (90). The clustering model may well apply to insulin receptors. Some surface receptors are indeed able to undergo considerable redistribution in the fluid mosaic membrane (87, 88, 91) upon binding of ligands. A clus- tered configuration has recently been demonstrated in the binding of biologically active ferritin-insulin to fat cell or liver membranes by electron microscopy (92-94). We observed a temperature transition around 21” in the Arrhenius plots of dis- sociation by dilution only. A transition around this temperature has recently been shown to be consistent with formation of clus- ters in artificial lipid bilayers, rather than with classical melt- ing transitions (95). Concanavalin A, which inhibits cluster for- mation induced by ligands in certain conditions (91), inhibits the site-site interactions among insulin receptors in lympho-

s Such conformational transitions are classically divided into two kinds of models: the allosteric model where the ligand modifies the pre-existing equilibrium between two symmetrical conformational states in the macromolecule (the Monad-Wyman-Changeux model, Ref. 81)-model incompatible with negative cooperativity-or “in- duced fit” models as developed by Koshland et al. (83, particularly suitable for negative cooperativity.

HEMOGLOBIN

I Tertiary and quaternary constraints

declxy- OXY conformation conformation

POSITIVE COOPERATIVITY

-

INSULIN RECEPTOR

G ;:yg @ “empty” “filled”

NEGATIVE COOPERATIVITY

u = subunit in a htgh affinity k.low dissociating~ state

0= subunit in a low affinity (fast dissociating) state

FIG. 15. A plausible model for the negative cooperativity in insulin receptors is presented by analogy with the structure and function of hemoglobin. Isolated hemoglobin chains (a and (3), like myoglobin, exhibit Michaelian (noncooperative) saturation curves and high affin- ity for oxygen. The constitution of the tetramer involves constraints which bring the subunits in a state of lower affinity for oxygen; the binding of oxygen then releases the constraints and improves the affinity of neighboring subunits for oxygen (positive cooperativity). The teleologic rationale for this mechanism, in which affinity for oxygen is low at low partial pressure of oxygen, is that hemoglobin is functionally a carrier molecule and must release oxygen more easily when its tissue concentration is low. For a receptor. whose function is to bind insulin at low physiological concentration and transmit its biological effects; high affinity at low concentrations of insulin is favorable, which is the case in a negative cooperative binding. Hence the model on the right, the “mirror image” of the hemoglobin model, is one of many plausible ones for a receptor molecule. It is not implied that subunits change shape in a concerted way in the case of the insulin receptor: intermediate steps have not been illustrated. Also, in the case of negative cooperativity, the conformational change has to be ligand- induced.

cytes without binding to the biological receptor sites (58). How- ever, the fact that a detergent-solubilized preparation of insulin receptors still exhibits negative cooperativity, with a dose- response curve superimposable to the curve reported in this paper, seems to exclude a large membrane reorganization as a mechanism for the cooperativity.’

Negative cooperativity in a hormonal receptor provides a mechanism in which binding to receptor is favored at low concentrations of hormone, but becomes more difficult when concentration of hormone increases. The relationship between negative cooperativity and “spare receptors” has been recently analyzed for the negative cooperativity coupled to a “cluster- ing” mechanism (90, 96). The conformational change in receptors which “turns off” binding of the hormone could well be the molecular event triggering further steps in the chain of biological events induced by hormone binding. Negative coop- erativity in binding can coexist with apparent negative (97, 98) as well as positive (99, 100) cooperative effects in biological activation, depending on the coupling between binding and activation. Cooperative models of drug action in pharmacology have recently been discussed (101).

Analysis of other published data on hormone binding indi- cate that negative cooperativity in binding could well be the rule rather than the exception in hormonal receptors. Curvilin- ear Scatchard plots of steady state binding have been described for a number of hormonal and nonhormonal systems. In some of these, kinetic data add to the probability of true negative

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from

Negative Cooperativity of Insulin Receptors 1887

cooperativity, i.e. dissociation is faster when studied by 27. addition of unlabeled hormone than by dilution, e.g. thyroid- stimulating hormone (102) and thyroliberin (103). Likewise, dis- 28.

sociation of lasI-glucagon is slower in the presence of de-His 29. glucagon than in the presence of glucagon (104), which should not happen if the unlabeled species simply exchanges with the 30.

tracer. Using the kinetic methods described in our previous report (31) and developed in this paper, negative cooperativity

31.

was actually demonstrated in the receptors for nerve growth factor (105, 106), thyroid-stimulating hormone (70), and /3- 32.

adrenergic catecholamines (72). This approach should be 33.

helpful in clarifying the processes which regulate the binding of many hormones to their cell receptors, and, coupled with direct

:s4

morphological correlations, should improve our understanding 35. of the structure and function of biological receptors.

Marx, S. J., Aurbach, G. D., Gavin, J. R., III, and Buell, D. W. (1974) J. Eiol. Chem. 249,6812-6816

Shiu, R. P. C., and Friesen, H. G. (1974) J. Biol. Chem. 249, 7902-7911

Conway, A., and Koshland, D. E., Jr. (1968) Biochemistry 7, 4011-4023

Levitzki, A., and Koshland. D. E., Jr. (1969) Proc. Natl. Acad. Sci. U.S.A. 62, 1121-1128

De Meyts, P., Roth, J., Neville, D. M., Jr., Gavin, J. R.. III, and Lesniak, M. A. (1973) Biochem. Biophys. Res. Commun. 66, 154-161

Hunter, W. M., and Greenwood, F. C. (1962) Nature 194, 495-496 Freychet. P., Roth, J., and Neville, D. M., Jr. (1971) Biochem.

Biophys. Res. Commun. 43, 400-408 Sodoyez, J. C., Sodoyez-Goffaux, F., Goff, M. M.. Zimmerman,

A. E., and Arquilla, E. R. (1975) J. Biol. Chem. 250.4268-4277 Fahey, J. L., Buell. D. N., and Sax, H. C. (1971) Ann. N. Y. Acad.

Sci. 190, 221-234 36.

Acknowledgments-Enlightening discussions on this subject 37. with D. E. Koshland, Jr., A. Levitzki, D. M. Neville, Jr., H. A. 38. Saroff, D. L. Hunston, D. Rodbard, and K. C. Ingham are gratefully acknowledged. Mrs. Dorothy E. Beall provided 39.

excellent secretarial assistance. 40.

Gavin, J. R., III, Gorden, P., Roth, J., Archer, J. A., and Buell, D. N. (1973) J. Biol. Chem. 248, 2202-2207

Boyiim, A. (1968) Stand. J. Clin. Inuest. 21 (SuppI. 97). ‘ii-89 Archer, J. A., Gorden. P., Gavin, J. R., III, Lesniak. M. A., and

Roth, J. (1973) J. Clin. Endocrinol. Metab. 36, 627-633 Bianco, A. R., Schwartz, R. H., and Handwerger, B. S. (1974)

Diabetologia 10, 359 Schwartz, R. H., Bianco, A. R., Kahn, C. R., and Handwerger, B.

S. (1975) Proc. Nat. Acad. Sci. U.S. A. 72, 4744478 Neville, D. M.. Jr. (1968) Biochim. Biophys. Acta 154, 540-552 Freychet, P., Roth, J., and Neville, D. M.. Jr. (1971) Proc. Natl.

Acad. Sci. CJ. S. A. 68, 1833-1837 1.

2. 3. 4. 5.

6.

7. 8. 9.

10.

11.

12. 13. 14.

15.

16.

17. 18.

19. 20.

21.

REFERENCES

Lefkowitz, R. J., Pastan, I., and Roth, J. (1969) in The Role of Adenyl Cyclase and Cyclic 3’,5’-AMP in Biological Systems, NIH Fogarty International Center Proceedings (Rail, J. E., Rodbell, M., and Condliffe, P., eds) No. 4, pp. 88-95, National Institutes of Health, Bethesda, Maryland.

Goodfriend, T.. and Lin, S. Y. (1969) Clin. Res. 17,243 Roth, J. (1975) in Methods Enzymol. 37, 223-233 Roth, J. (1973) Metabolism 22, 1059-1973 Kahn, C. R. (1975) in Methods in Membrane Biology (Korn, E.

D., ed) Vol. 3, pp. 81-146, Plenum Press, New York De Meyts, P. (1975) in Methods.in Receptor Research (Blecher,

M., ed) Marcel Dekker, New York, in press Kahn, C. R. (1976) J. Cell Eiol., in press Scatchard, G. (1949) Ann. N. Y. Acad. Sci. 51, 660-672 Berson, S. A.. and Yalow. R. S. (1959) J. Clin. Znuest. 38.

1996-2016 Rodbard, D. (1973) in Receptors for Reproductive Hormones

(O’MalIey, B. W., and Means, A. R., eds) pp. 327-364, Plenum Publishing Co., New York

Kahn, C. R., Freychet, P., Roth, J., and Neville, D. M., Jr. (1974) J. Biol. Chem. 249.2249-2257

Hart, H. E. (1965) Bull. Math. Biophys. 27, 87798 Rosenthal, H. E. (1967) Anal. Biochem. 20, 525532 Miinck, A. (1975) in Receptors and Mechanism of Action of

Steroid Hormones, (Pasqualini, J. R., ed) Marcel Dekker, New York, in press

Fletcher, J. E., Spector, A. A., and Ashbrook, J. D. (1970) Biochemistry 9, 4580-4587

Klotz, I. M., and Hunston, D. L. (1971) Biochemistry 10, 30653069

Klotz, I. M. (1974) Act. Chem. Res. 7, 162-168 Klotz, I. M., and Hunston, D. L. (1975) J. Biol. Chem. 250,

300-3009 Hunston. D. L. (1975) Anal. Biochem. 63, 99-109 De Meyts, P.. and Roth, J. (1975) Abstracts of the 57th Annual

Meeting, The Endocrine Society, p. 51 De Meyts, P.. and Roth, J. (1975) Eiochem. Biophys. Res.

Commun. 66, 1118-1126 22. Freychet, P., Kahn, R., Roth, J., and Neville, D. M., Jr. (1972)

J. Biol. Chem. 247, 3953-3961 23. Ketelslegers, J. M., Knott, G. D., and Catt, K. J. (1975)

Biochemistry 14, 3075-3083 24. Neville, D. M.. Jr., Kahn, C. R., Soll, A., and Roth, J. (1973) in

Protides of the Biological Fluids, 21st Colloquium (Peeters, H., ed) pp. 269-273, Pergamon Press, New York

25. Lesniak, M. A., Gorden, P., Roth, J., and Gavin, J. R., III (1974) J. Biol. Chem. 249, 1661-1667

26. Catt, K. J., Tsuruhara, T., and Dufau, M. L. (1972) Biochim. Biophys. Acta 279, 194-201

41. 42.

43.

44.

45.

46.

47. 48.

49.

50.

51.

Neville, D. M., Jr., and Kahn, C. R. (1974) in Subcellular Particles, Structures and Organelles (Laskin. A. I., and Last. J. A., eds) pp. 57-85, Marcel Dekker, New York

Soll, A. H., Kahn, C. R., and Neville, D. M., Jr. (1975) J. Eiol. Chem. 250, 4702-4707

De Meyts, P., Roth, J., Neville, D. M., Jr., and Freychet, P. (1974) Abstracts of the 56th Annual Meeting of the Endocrine Society, p. 189

Frevchet. P.. Brandenburg. D.. and Wollmer. A. (1974) D;‘abet&ogia 10, l-5 ”

Gliemann, J., and Gammeltoft, S. (1974) Diabetologio 10, 105-113 De Meyts, P.. Kahn, C. R., Ginsberg, B. H., and Roth. J. (1975)

Diabetes 24 (Suppl. 2). 393 Podskalny, J., Chou, J., and Rechler, M. M. (1975) Arch.

Biochem. Biophys. 170, 504-513 Fussganger, R. D.,-Kahn, C. R., Roth, J., and De Meyts, P. (1975)

Diabetes 24, Suppl. 2, 393 Blecher, M., Giorgio, N. A., and Johnson, C. B. (1974) Ado.

Enzyme Regul. 12, 289-309 52. Saroff, H. A., and Minton, A. P. (1972) Science 175, 1253-1255 53. Saroff, H. A. (1973) Biopolymers 12, 599-610 54. Wyman, J. (1948) Adu. kitein Chem. 4, 407-531 55. Matthews, B. W., and Bernhardt, S. A. (1973) Annu. Reu.

Biophys. Bioeng. 2, 257-317 56. Gibson, Q. H., and Roughton, F. J. W. (1955) Proc. R. Sot.

London B Biol. Sci. 143,310-342 57. Olson, J. S., Gibson, Q. H., Nagel, R. L., and Hamilton, H. B.

(1972) J. Biol. Chem. 247, 7485-7493 58. De Meyts, P., Gavin, J. R., III, Roth, J., and Neville. D. M., Jr.

(1974) Diabetes 23 (Suppl. l), 355 59. De Meyts, P. (1975) J. Supramol. Struct., in press 60. Pekar, A. H., and Frank, B. H. (1972) Biochemistry 11,4013-4016 61. Blundell, T., Dodson, G., Hodgkin, D., and Mercola, D. (1972)

Ado. Protein Chem. 26, 279-402 62. Boesel, R. W., and Carpenter, F. H. (1972) Fed. Proc. 31, 255 63. Zimmerman, A. E., and Yip, C. C. (1972) Fed. Proc. 31, 255 64. Zimmerman, A. E., Kells, D. I. C., and Yip, C. C. (1972)

Biochem. Biophys. Res. Commun. 46, 2127-2133 65. Zimmerman, A. E., and Yip, C. C. (1974) J. Biol. Chem. 249,

4021-4025 66. Zimmerman, A. E., Moule, M. L., and Yip, C. C. (1974) J. Biol.

Chem. 249,402&4029 67. Goldman, J., and Carpenter, F. H. (1974) Biochemistry 13,

4566-4574 68. Cuatrecasas, P., and Hollenberg, M. (1975) Biochem. Biophys.

Res. Commun. 62, 31-41 69. Fripiat, J., Chaussidon, J.. and Jelli, A. (1971) in Chimie

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from

1888 Negative Cooperatiuity

physique des phknomknes de surface: application aux oxydes et aux silicates, pp. 229-260. Masson et Cie. Paris

70. Li, C. H. (19733. in Advances in Human Growth Hormone Research (Raiti. S., ed) Department of Health Education and Welfare Publication No. (NIH) 74.612, pp. 321-341, National Institutes of Health, Bethesda, Md.

71. Kohn, L., and Winand, R. (1975) in Molecular Aspects of Membrane Phenomena (Kaback, H. R., Radda, G., and Schwyzer, R., eds) Springer Verlag, Heidelberg, in press

72. Weiner. N.. and Jardetzkv. 0. (1964) Naunvn-Schmiedeberes

73 94. Orci, L., Rufener, C., Malaisse-Lagae, F., Bendel, B., Amherdt,

M., Bataille, D., Freychet, P., and Perrelet, A. (1975) IsraelJ. Med. Sci. 11, 639-655

74 75.

76.

Arch. Phkmakol. Enp. &hol. 248, 308-318” Limbird, L., De Meyts,.P., and Lefkowitz, R. J. (1975) Biochem.

Biophys. Res. Commun. 64, 1160-1168 Laiken, N., and Ni.methy, G. (1971) Biochemistry 10,2101-2106 Silhavy, T. J., Szmelcman, S., Boos, W., and Schwartz, M.

(1975) Proc. N&l. Acad. Sci. U.S.A. 72, 2120 Noyes. A. A., and Whitney, W. R. (1897) Z. Phys. Chem. 23.

95. Lee, A. G., Birdsall, N. J. M., Metcalfe, J. C., Toon, P. A., and Warren, G. B. (1974) Biochemistry 13, 3699-3705

96. Levitzki, A., Segel, L. A., and Steer, M. L. (1975) J. Mol. Biol. 91, 125-130

ti89-699 11. Davis, H. S., and Crandall, G. S. (1930) J. Am. Chem. Sot. 52,

3757-3785 78. 79.

Miller, D. M. (1968) Biophys. J. 8, 1339-1352 Regen, D. M., and Tarpley, H. L. (1974) Biochim. Biophys. Acta

339, 218-233

97. Pochet, R., Boeynaems, J. M., and Dumont, J. E. (1974) Biochem. Biophys. Res. Commun. 58,446-453

98. Roy, C., Barth, T., and Jard, S. (1975) J. Biol. Chem. 250, 3149-3156

99. Rodbard, D., Moyle, W., and Ramachandran, J. (1974) in

80. 81.

Wang, J. H. (1955) J. Phys. Chem. 59, 1115-1116 Monod, J., Wyman, J., and Changeux, J. P. (1965) J. Mol. Biol.

12,88-118

Hormone Binding and Target Cell Actiuation in the Testis (Dufau, M. C., and Means, A. R.. eds) po. 79-87

82.

83.

Koshland, D. E., Jr., NBmethy, G., and Filmer, D. (1966) Biochemistry 5, 365-385

Rubin. M. M., and Changeux, J. P. (1966) J. Mol. Biol. 21, 265-274

100. Rodbard, D. (1974) Endocrinology 94, 142711437 101. Colquhoun, D. (1973) in Drug Receptors (Rang, H. P., ed) pp.

149-182, University Park Press, Baltimore 102. Verrier, B., Fayet, G., and Lissitsky, S. (1974) Eur. J. Biochem.

42,355-365

84. 85.

Hertzfeld, J., and Stanley, H. E. (1974) J. Mol. Biol. 83,231-265 Frieden, C. (1968) in Regulation of Enzyme Activity and Allo-

steric Interactions, pp. 59-71, Academic Press, New York Ingham, K. C., Saroff, H. A., and Edelhoch. H. (1975) Biochemis-

103. Grant, G., Vale, W., and Guillemin, R. (1972) Endocrinology 92, 1629-1633

104. Birnbaumer, L., and Pohl, S. L. (1973) J. Biol. Chem. 248, 2056-2061

86. tly14,4745-4758

105. Frazier, W. A., Boyd, L. F., and Bradshaw, R. A. (1974) J. Biol. Chem. 249.5513-5519

87. Singer, S. J., and Nicolson, G. L. (1972) Science 175, 720-731 106. Frazier, W. A., Boyd, L. F., Pulliam, M. W., Szutowicz, A., and 88. Singer, S. J. (1976) in Surface Membrane Receptors; Interface Bradshaw, R. A. (1974) J. Biol. Chem. 249, 5918-5923

of Insulin Receptors

between Cells and Enuironment, Plenum Press, in press 89. Boeynaems, J. M., and Dumont, J. E. (1975) J. Cyclic Nucleotide

Res. 1,123-142 90. Levitzki, A. (1974) J. Theoret. Biol. 44,367-372 91. Edelman, G. M. (1974) in Control of Proliferation in Animal Cells

(Clarkson, B., and Baserga, R., ids) pp. 357-377, Cold Spring Harbor Conferences on cell proliferation. I

92. Jarett, L., and Smith, R. M. (1974) J. Biol. Chem. 249,7024-7031 93. Jarett, L., and Smith. R. M. (1975) Proc. Natl. Acad. Sci. U. S. A.

72.352&35ao

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from

P DeMeyts, A R Bainco and J Rothcooperativity.

Site-site interactions among insulin receptors. Characterization of the negative

1976, 251:1877-1888.J. Biol. Chem. 

  http://www.jbc.org/content/251/7/1877Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/251/7/1877.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at

by guest on February 12, 2018http://w

ww

.jbc.org/D

ownloaded from