Binding and degradation of 125I-insulin by rat hepatocytes.

The binding and the velocity of degradation of 125I-insulin in the absence or presence of varying concentrations of native procline insulin were studied using isolated rat hepatocytes. At insulin concentrations ranging from 5 X 10(-11) to 10(-6) M, insulin degradation velocity showed a first order dependence on the total concentration of insulin bound at steady state. The overall reaction had an apparent rate constant of 0.030 +/- 0.011 min-1. Furthermore, the degradation of a given amount of 125I-insulin bound to cells was more rapid and extensive than the degradation of the same amount of insulin which had been newly exposed to fresh cells. Mid pretreatment of isolated hepatocytes with trypsin or chymotrypsin at concentrations of 5 to 20 mug/ml depressed to the same degree the amount of 125-I-insulin bound at steady state and the 125I-insulin degradation velocity. Peptide or protein hormones unrelated to insulin, including the oxidized A and B chains of insulin, failed to depress the amount of insulin bound or the velocity of insulin degradation when present at concentrations of 10-5 or 10-6 M. Over a wide range of concentrations, various synthetic insulin analogues and naturally occurring insulins depressed to the same degree the amount of 125I-insulin bound at steady state and the 125I-insulin degradation velocity. These observations suggest that insulin bound to hepatocyte plasma membranes is the substrate for insulin degradation by the liver.

The binding and the velocity of degradation of "Y-insulin in the absence or presence of varying concentrations of native porcine insulin were studied using isolated rat hepatocytes. At insulin concentrations ranging from 5 x 10-l' to lOME M, insulin degradation velocity showed a first order dependence on the total concentration of insulin bound at steady state. The overall reaction had an apparent rate constant of 0.030 h 0.011 min-'. Furthermore, the degradation of a given amount of '%insulin bound to cells was more rapid and extensive than the degradation of the same amount of insulin which had been newly exposed to fresh cells. Mild pretreatment of isolated hepatocytes with trypsin or chymotrypsin at concentrations of 5 to 20 pg/ml depressed to the same degree the amount of '251-insulin bound at steady state and the lZ51-insulin degradation velocity. Peptide or protein hormones unrelated to insulin. including the oxidized A and B chains of insulin, failed to depress the amount of insulin bound or the velocity of insulin degradation when present at concentrations of lo-' or 10eB M. Over a wide range of concentrations, various synthetic insulin analogues and naturally occurring insulins depressed to the same degree the amount of '261-insulin bound at steady state and the Y-insulin degradation velocity. These observations suggest that insulin bound to hepatocyte plasma membranes is the substrate for insulin degradation by the liver.
The initial interaction of insulin with insulin-sensitive cells probably occurs on the outer face of the plasma membrane. Kono and Barham (1) and Cuatrecasas (2) have shown that trypsin treatment of fat cells or the plasma membranes of fat and liver cells diminishes the binding of 1251-insulin, and Cuatrecasas (3) has reported that intact '*"I-insulin can be readily recovered following treatment of the bound material with acid. Furthermore, various lines of evidence suggest that this interaction triggers the characteristic biological response. Several workers have reported that trypsin-treated fat cells show diminished insulin responsiveness (2. 4-6). and Freychet et al. (7. 8), Gavin et al. (9), Gliemann and Gammeltoft (lo), and Simon et al. (11) have shown that the relative abilities of insulin analogues to compete with labeled insulin for insulin binding sites in various systems correspond well to their relative biological potencies. Also, Crofford has demonstrated that the stimulatory effect of insulin on fat cells can be reversed by extensively washing the cells or by the addition of anti-insulin serum (12). Many insulin-sensitive tissues, notably the liver, also degrade the hormone, and several studies have suggested that insulin degradation occurs intracellularly.
The data of Rubenstein et al. (13) and of Mortimore and Tietze (14) suggest that liver homogenates are more active at degrading insulin than is the isolated cyclically perfused liver, and several insulin- is diminished in extracts of plasma membranes obtained from trypsin-treated fat cells (19). Although the data of Freychet et al. (20) suggest that the insulin receptor site of rat liver plasma membranes is not itself a site of insulin degradation, they do not preclude the possibility that binding of insulin to a receptor may be the initial step of the degrading process in the intact cell. These considerations prompted us to investigate the relationship between the binding and degradation of insulin in isolated hepatocytes. A preliminary report of portions of this investigation has been presented (21 To rule out leakage of insulin-degrading material from the cells, 1 ml of cell suspension, incubated without insulin in parallel with the experimental cells. was centrifuged, and the supernatant was reincubated as above with the appropriate amount of "'I-insulin for an additional 60 min at 30". Degradation by the medium alone never exceeded 1.5'7' of the total insulin present or 3.5 to 4'2 of the insulin degradation occurring in the presence of cells. Cells added to the control degradation medium during extraction of radioactivity were without effect. Sample Analysis-The 100. to 200.~1 aliquots of cell suspensions taken for determination of total degradation were frozen and thawed three times, then extracted two or three times with 0.1% Triton X-100 containing 3 M acetic acid and 6 M urea until 99% of the radioactivity was in cell-free extract.
When appropriate, the cell pellets which had been used to measure the amount of '%insulin bound were suspended in the same extraction solvent, frozen immediately, and similarly extracted.
Each extract was then gel filtered over a column (1 x 50 cm) of Sephadex G-50F using a solution containing 6 M urea, 1 M acetic acid, and 0.15 M NaCl. Over 99% of the radioactivity applied to the column was recovered in the eluate. More than 96% of the radioactive material in the extracts eluted either with native insulin or in the internal volume of the column (see "Results"). The latter was considered to represent degradation products.
The amounts ofradioactivity eluting in the insulin peak (intact insulin) and in the internal volume (degradation products) were estimated by digital computer (PDP8) as the percentage of the total radioactivity applied to the column and were reproducible within 0.5% of applied column counts. To further test whether the iodinated material bound to cells was intact insulin, the iodinated material eluting with native insulin on gel filtration was pooled and desalted by elution through Sephadex G-50F in 1 M acetic acid. The appropriate fractions were dried in uacuo, and the material was subjected to polyacrylamide gel electrophoresis at pH 4.5, at which iodination of the tyrosine residues has no effect on mobility.
IJnder these conditions, over 98? of this iodinated material co-migrated with native insulin. Over 90% of the hormone judged to be intact by gel filtration in these experiments was bound by fresh cells. Data Analysis-Incubation of 1 x lo6 cells/ml with 0.1 PCi of [hydroxymethyl-'"C]inulin, a 5000.dalton polysaccharide which distributes only extracellularly, followed by isolation of the cell pellet as described above, revealed that no more than 0.1%' of the total radioactivity was trapped in the cell pellet. Therefore, the amount of "'I-insulin bound at any time during the incubation was measured as the radioactivity bound to the cells less 0.1'7r of the total. That is, total binding was measured without correction for the amount of "'I-insulin bound in the presence of 1Om6 M native insulin (7,11,20,27). Since binding attained a steady state by 30 min under the conditions of cell density and of insulin concentrations used, the amount bound was taken to be the amount bound at 30 min less the correction mentioned above. Time courses of binding were measured in each experiment. Depending upon the particular experiment, binding was expressed either as the amount of '2SI-insulin bound or as the total amount of insulin (native insulin plus "'I-insulin) bound. These were determined as: (radioactivity bound at 30 min less radioactivity trapped in the pellet/radioactivity in the same volume of cell suspension) x (concentration of "'I-insulin or of total insulin present in the incubation medium).
The concentration of insulin degraded in each cell suspension sample was calculated, after gel filtration, as: (radioactivity eluting in the internal volume/total radioactivity applied to the column) x (insulin concentration in the incubation medium).
In the experiments in which iodinated material bound to cells was analyzed, the amounts of intact insulin or of degraded iodinated material associated with cells was determined as: (per cent of intact insulin, or per cent of degradation products, in the cell pellet extract) x (total amount of iodinated material bound). Degradation velocity was calculated directly as the regression coefficient of a plot of concentration of degradation products in the cell suspension versus time using no fewer than four time points. The mean deviation of ordinate determinations from the regression line was never more than 5% of the range of ordinate values.

Concentration
Dependencies of Binding and Degradation - Fig.  1 shows the time course of binding (upper) and of degradation (lower) of '251-insulin at concentrations of native porcine insulin between 3.3 x 10-l' and 9.9 x lo-' M. The binding of Y-insulin was depressed with increasing concentrations of native insulin. In all of the binding curves, binding reached a steady state by 20 to 30 min at 30". At cell densities greater than 1 x 10' cells/ml and at total insulin concentrations less than or equal to 9.0 x lo-", binding began to decline after 45 min. Therefore, although all data were taken from time courses, steady state binding was measured at 30 min. More than 90% of the iodinated material associated with cells at all times during the incubation period was intact insulin, as judged both by gel filtration and by polyacrylamide gel electrophoresis at pH 4.5. The velocity of degradation of "'I-insulin was depressed by the addition of native insulin at concentrations between 3.3 x 10-l' M and 9.9 x lo-' M (F' 1~. 1, lower (2 x lo6 cells/ml) were incubated at 30" with '251-insulin (5.6 x lo-" M) in the absence or the presence of native insulin (3.a x lO-" to 9.9 x 10m7 M) for 60 min in a total volume of 1 ml. A small aliquot of cell suspension was removed for determination of degradation, and the amount of 'Z51-insulin bound was determined by analysis of the remainder (see "Experimental Procedure"). Data are expressed as the percentage of '251-insulin bound (upper) or degraded (loluer) at the appropriate time per 2 x lo6 cells/ml. Each symbol represents a different concentration of native insulin. '1, no native insulin; 0, 3.3 x lo-" M; 0, 9.9 X lo-" M; A, 33 x I()-' M; ., 9.9 x 10m9 M; n , 9.9 X 1Om8 M; and A, 9.9 x 10~ 7 M.

8391
and 10 min at 30". This lag was not shortened by the addition of native insulin. At total insulin concentrations less than or equal to 5 x lo-" M, the slight depression in '2SI-insulin binding after 45 min was accompanied bq a commensurate depression of '*'I-insulin degradation velocity. Therefore, at all concentrations tested, the degradation velocity was measured over a time period when degradation velocity was linear (15 to 45 or 60 min). The amount of 1251-insulin bound at steady state as well as the degradation velocities were linearly related to cell density between 0.3 and 2.0 x lo8 cells/ml. Fig. 2  medium at 15, 30, and 45 min after the addition of "'I-insulin alone. Intact '251-insulin eluted with native insulin; degradation products eluted with Na'*'I. The small amount (2 to 3%) of the total applied radioactivity, eluting in the void volume. is attributable to aggregate present in the 'Y-insulin preparation. This material was not retained by cells. Iodinated material eluting between native insulin and the salt peak, including the A chain which eluted immediately after insulin, was never seen.
Competition with Insulin, Insulin Analogues, and Other Peptides-In order to measure the affinities of the binding and the degradative processes for native insulin, various insulin analogues and other naturally occurring insulins. competitive binding experiments were performed. The results, shown in Fig. 3 (upper), suggested that both the binding and the degradation velocity of 'Z51-insulin was inhibited by 50% at native insulin concentrations of 3.5 x lo-'M. Only about 8% of the iodinated insulin was bound in the presence of lOme M native insulin. The experimental points deviate, particularly at high and low concentrations of native insulin, from a curve expected on the basis of a simple reversible interaction between insulin and a homogeneous group of insulin binding sites with a Kd for insulin of 3.5 x 10-O M. In addition, Scatchard plots of these data were nonl@ear (not shown).
The concentrations of insulin analogues necessary for 50% inhibition of binding and degradation velocity of ""I-insulin (apparent K,) were read directly from the same type of competitive binding curve (Fig. 3, upper). In order to express the degree of deviation of degradation velocity from the concentration of Y-insulin bound at different analogue concentrations, the degradation velocity in the presence of different concentrations of analogue was plotted against the corresponding amount of '*Yinsulin bound at steady state as a percentage of the control. These results are shown in Fig. 3 (lower) for native insulin. The dispersion of the data was expressed as the mean deviation of the per cent degradation velocity from the regression line along the ordinate. The data with several insulin analogues are summarized in Table I. The  competitive binding curves with analogues from which these data are derived are available on request.'The correspondence between the concentrations of analogue necessary for 50% inhibition of the binding and the degradation of "'I-insulin by all analogues tested, including desalanyl-desasparaginyl insulin, suggests that these relative affinities of the binding and degradative systems for lz51-insulin and for each analogue are the same within our margin of error.
Other peptide hormones had no effect on the binding and degradation velocity of "%insulin, when present at concentrations of lo-' to lo-' M ( Table II) 'This insulin will be referred to as "prehound insulin." 'This insulin will he referred to as "non-prehound insulin."  Fig. 3 (lowerpanel). 'Mean of five experiments +S.D. between the two experimental conditions at the beginning of the final incubation is that in the one case, all of the Y-insulin is bound to cells, whereas in the other case, none of the "Y-insulin is bound to cells. Fig. 4 (lower) shows the time course of degradation in each case. A comparison of the areas under the two degradation curves shows that the total concentration of degradation products formed from prebound insulin is greater than that formed from a comparable concentration of non-prebound insulin. Furthermore, there is no appreciable time lag in the degradation of prebound insulin at the intervals measured, in contrast to the lag in the degradation of nonprebound insulin (Fig. 4, lower; see also Fig. 1, lower). Furthermore, over the time interval when the amount of prebound insulin is greater than the amount of nonprebound insulin (0 to 30 min; Fig. 4 less than the amount of non-prebound insulin (upper panel), the degradation velocity of prebound insulin is lower than that of the non-prebound insulin. Two aspects of the above experiment should be noted. First, when the cell-free medium from the cell suspension which had been incubated without insulin for the duration of the entire experiment was incubated with "'I-insulin for an additional hour at 30", less than 1.3% of the insulin was degraded; that is, at least 98% of the insulin degradation was attributable to the cells and not to degradative enzymes which had leaked from the cells. Second, more than 95% of the total prebound radioactivity at the beginning of the experimental incubation period eluted with intact insulin on gel filtration. Therefore, the final degradation products were newly formed during the experimental incubation period and do not represent release of preformed degradation products from the cells. A closer examination of the handling of prebound insulin showed that at least 90% of the iodinated material retained by the cells was intact insulin throughout the final 60-min incubation period (Fig. 5, upper panel). The finding that most of the bound insulin is intact is comparable to the findings of others (20, 27) and suggests that final degradation products are not retained by the cells. The lower panel shows that the early loss of iodinated material from the cells (0 to 15 min) represents release of degradation products as well as dissociation of intact 1z61-insulin. By 15 min, the amount of intact '2'I-insulin dissociated from the cells has reached a plateau. However, a steady state is not maintained due to the continuing binding and degradation of "'I-insulin by the cells, as reflected in the slow decline of intact lZsI-insulin in the medium after 30 min. Thus, the degradation products appearing before 20 min arise primarily from "'I-insulin bound at the beginning of the incubation (prebound insulin), whereas the degradation products appearing after 30 min arise from Yinsulin which has dissociated and rebound. The slight lag in the initial appearance of degradation products in the supernatant contrasts with the immediate release of degradation products from prebound "'I-insulin shown in Fig. 4 attributable primarily to the short preincubation period (5 min) used.
A specific dissociation rate constant of insulin, calculated on the basis of the early (up to 10 min) loss of intact '261-insulin from the cells (as in Fig. 5, upper panel) was estimated to be 0.0385 f 0.0075 min' (mean f S.D. from eight experiments). However, analysis of the cell-free supernatant (Fig. 5, lolver panel) suggests that a significant fraction of the radioactive material dissociating at early times from the cells is iodinated degradation products. In view of this observation, the physical significance of the specific rate constant of dissociation is unclear. It should be emphasized that the cell-free medium in these experiments did not degrade more than 1.5Y of the added iodinated insulin, so that more than 98% of the iodinated degradation products formed resulted from degradation by the cells and not from degradation of released intact insulin by the medium.
Quantitative Relationship between Insulin Binding and Degradation Velocity--In order to express our results over a wide range of insulin concentrations, we plotted the binding and degradation data in log-log form. Plots of the log total insulin bound or of the log degradation velocity uersus log total insulin concentration in the medium were linear over total medium insulin concentrations of 5 x 10-l' to 10-O M (Fig. 6, left and right panels, respectively).
Dilutions of "'I-insulin were used to measure binding and degradation velocity at insulin concentrations lower than lo-' M, whereas final insulin concentrations greater than lo-' M were achieved by the addition of appropriate concentrations of native porcine insulin. The continuity of these plots strongly suggests that iodinated and native insulin behave similarly with respect both to binding and to degradation in this system. We have found that 92.52 h 2.02%~ (mean * S.D. of 14 experiments) of the 'Y-insulin bound to cells at concentrations between 5.5 x 10-l' and 9.9 x 10-l' M is inhibited by the presence of lo-@ native insulin. Therefore, the binding at concentrations lower than lo-' M in Fig. 6 (left) represents primarily binding which is displaceable by lOmE M native insulin, whereas the binding of "'I-insulin at concentrations greater than lo-* M represents primarily binding which is not displaceable by lOmE M native insulin.
The fact that plots of insulin binding and degradation velocity uersus total insulin concentration are parallel (Fig. 6) suggests that degradation velocity and the concentration of bound insulin are related to one another in a simple way. Accordingly, the log of the degradation velocity of total insulin was plotted against the log of the total insulin bound at steady state in the presence of insulin concentrations of 7 x lo-"' to lo-'M. As seen in Fig. 7, this plot is linear and therefore can be expressed in the following form: log dP/dt = m log (1R) + log k,, or, taking antilogs, where P is the instantaneous total concentration of insulin degradation product; dP/dt, the instantaneous degradation velocity; (ZR), the concentration of total insulin bound at steady state corrected only for the "%insulin trapped in the pellet fluid; k,,, the apparent rate constant relating degradation velocity to the amount of insulin bound; and 0, the reaction order of degradation with respect to total insulin bound at steady state. In a plot of log dP/dt uersus log (IR) (equation l), 0 is evaluated directly from the slope of the line and was found to be 0.994 + 0.082 (mean + S.D. from five experiments), suggesting that degradation velocity is first order with respect to total amount of insulin bound. The log k,, is equal to the log dP/dt when log (ZR) is 0, and is 0.030 * 0.011 min-' (mean k S.D. from five experiments).
The relationship between insulin degradation velocity and the amount of insulin bound at steady state persisted even at low '*%insulin concentrations (5 x lo-" to 1.6 x IO-' M), at which only about 7.5% of the bound '*'I-insulin was not displaced by the addition of high concentrations of native insulin. At these low concentrations, a plot of insulin degradation velocity uersus the amount of insulin bound is linear and has a slope, equal to the k,,, of 0.0385 min", which is within the error of the k,, mentioned above (Fig. 8). The ordinate intercept is -0.0026, not significantly different from 0. A plot of these data as log dP/dt uersus log (ZR) (not shown here) was linear with slope, equal to 1.08 (correlation coefficient of 0.99), implying a reaction order of 1.

DISCUSSION
We have measured total insulin binding in order to avoid a priori assumptions about the class or classes of insulin binding proteins on the cell surface which might be involved in the degradative process. The linearity of the log-log plot of insulin bound UBFSUS total insulin concentration (Fig. 6, left) is in part a reflection of the inclusion of two different types of insulin binding sites which can be distinguished operationally by their interaction with '251-insulin in the presence and absence of lo-@ M native insulin. At total insulin concentrations between 5 x lo-" and 9 x lo-"' M, over 90% of the binding of "'I-insulin is inhibited by lOme M native insulin; at concentrations greater than lo-' M, most of the binding of '2sI-insulin is not inhibited. However, the observed linearity ( Fig. 6) must in addition reflect deviation of the saturable component of insulin binding from the behavior expected on the basis of a simple reversible association between insulin and a binding protein (Fig. 3). The peculiar properties of this association have been implicitly recognized in the literature in the form of attempts to rationalize insulin binding in terms of multiple orders of binding sites (9, 28, 29), insulin-insulin interactions (30), and negative homotropic interactions among binding sites (31). Linear plots of log total insulin bound uersus log total insulin concentrations (as in Fig. 6, left) are obtained upon similar analysis of much of the actual data appearing in the literature from a variety of systems (7,12,27,29,(32)(33)(34)(35).
In agreement with the data of many others (1, 7, 9, 27-29, 3%36), treatment of our binding data in the conventional fashion reveals both saturable and nonsaturable components (Fig. 3). The saturable component has an apparent K, for insulin ( Fig. 3 and Table I) comparable to reported values (1,33); its relative affinities for insulin analogues, including desalanyl-desasparaginyl insulin (Table I and Appendix3), are similar to reported values for apparent K, and for the relative biological potencies of these analogues (7, 8, 10, 11, 20. 24, 33, 37-42); peptides unrelated to insulin, including A and B chains of insulin, do not inhibit the binding of '2sI-insulin to this site (Table II); and the site occurs on the cell surface, as suggested by the protease experiments (Table III and Refs. 1,2,446). The properties of the nonsaturable site, measured only at high concentrations of native insulin or of insulin analogues, are more difficult to define.
Several aspects of our data suggest that Y-insulin bound to saturable sites on the plasma membrane is the substrate for insulin degradation.
First, at all concentrations tested, native insulin and various insulin analogues inhibit degradation velocity to the same extent that they inhibit '*'I-insulin binding at steady state ( Fig. 3 and Table I). Second, high concentrations of peptides unrelated to insulin affect neither the steady state binding nor the degradation velocity of '*'I-insulin (Table II). Third, trypsin treatment, which has been shown to depress both 1261-insulin binding as well as the biological response to insulin in various systems (1,2,(4)(5)(6), depresses binding and degradation of '*'I-insulin to the same extent (Table III). Furthermore, prebound insulin is degraded more rapidly and extensively than non-prebound insulin (Fig.  4).
The fact that a plot of the log of degradation velocity uersus log (IR) has a slope of 1 even at high total insulin concentrations (Figs. 7 and 8) suggests that the relationship between insulin binding and degradation should be extended to include what have been called nonsaturable binding sites as well. For if degradation velocity were dependent only upon binding either to saturable sites or to nonsaturable sites, this plot would not be linear over the entire range of insulin concentrations measured, and its slope would be less than 1 at concentrations greater than lOma or less than lOme M, respectively. Thus, while there may be a structural or functional differentiation of sites with respect to insulin binding, all of these sites appear to be homogeneous with respect to insulin degradation. Interestingly, Goldstein and Brown have shown that degradation of low density lipoprotein is related to its binding to both saturable and nonsaturable sites on the plasma membranes of cultured fibroblasts (43).
The more rapid degradation of prebound insulin with respect to non-prebound insulin (Fig. 4) is incompatible with a model in which insulin reversibly associates with a receptor protein in an interaction which does not modify the insulin molecule and then is degraded by the cell independently of the receptor. If this were the case, the degradation of prebound insulin would be retarded relative to that of non-prebound insulin because two steps-dissociation and reassociation-would be necessary for the former. Considered alone, the more rapid degradation of prebound insulin is compatible with a model in which prebound insulin is released as a slightly modified, presumably less active, species which is then rapidly degraded by the cells in a manner independent of the insulin receptor. Such a scheme would require two recognition sites with similar susceptibilities to protease treatment but with different specificities. While this scheme is consistent with our protease data (Table  III), it is inconsistent with our data on the specificities of binding and degradation (Table I and Appendix3).
The simplest model which is compatible with all of our data is that insulin binds to an apparently heterogeneous set of binding sites on the cell surface, some of which are readily saturable and have an apparent K, of 3.5 x lo-' M. The bound insulin is then either released intact or is subsequently degraded without dissociating from the cell.
Our observations are consonant with the compartmentalization of the binding and degradation of insulin in intact hepatocytes.
First, the lag in the initial appearance of iodinated degradation products (Fig. 1) has also been noted by Mortimore and Tietze (14) in cyclic rat liver perfusions with '251-insulin as well as by the authors following 1-min infusions of "'I-insulin into noncyclically perfused rat livers (44).' Our observation that increasing concentrations of insulin do not shorten this lag either in hepatocytes or in perfused liver is consistent with an obligatory translocation of insulin from a binding site to an intramembranous or intracellular degrading site. In contrast, studies with various broken cell preparations (13,(45)(46)(47)(48)(49) and isolated enzymes (17, 50-54) do not show a lag. Second, with intact cells, degradation velocity is negligible at temperatures of O-20", but rises rapidly between 20 and 30".' In agreement with this, Mortimore has shown that although degradation of insulin by cyclically perfused liver is negligible at O", low temperature does not completely inhibit degradation of insulin by liver homogenates (14). It is possible that the rapid rise of insulin degradation velocity between 20 and 30" is similar to the temperature dependence of membrane transport functions in alveolar macrophage and reflects a temperaturedependent membrane phase transition (55). Third, the difference between the apparent affinity of the over-all degradation process for insulin which we report (Fig. 3, Table I) and the K, values reported for degradation by isolated plasma membranes (20) and by various isolated insulin-degrading enzymes (17,48,54) suggests that an insulin-degrading step, per se, is not rate-limiting in the over-all degradation process. And fourth, the results of the protease experiments (Table III) suggest, but "Susan Terris and Donald F. Steiner, manuscript in preparation.
do not directly prove, that an insulin-degrading enzyme is not exposed to proteases at the cell surface. If it were, degradation velocity would probably be more depressed than insulin binding.
Our data do not suggest a specific mechanism whereby insulin binding is related to its degradation, nor do they suggest a particular mode of enzymatic degradation. Barrnett and Ball (56) found that insulin stimulates pinocytosis in fat cells. A similar endocytotic mechanism, associated with lysoso-ma1 degradation, might account for the dependence of degradation velocity on the total amount of insulin bound by isolated hepatocytes. However, Crofford et al. have (44), ' and Izzo et al. have made similar observations on the in. uiuo degradation of '*'I-insulin by the liver (57). Our finding that degradation velocity is not depressed even at high total insulin concentrations ( Fig. 6. right) suggests that the capacity of the hepatocyte for insulin degradation greatly exceeds its binding capacity. In accordance with this possibility. several workers have reported that liver homogenates degrade insulin to a greater extent than do either liver slices or perfused liver (13, 14. 58).
Our findings differ from those of Gammeltoft and Gliemann (33) and of Freychet et al. (20). Gammeltoft and Gliemann (33) reported that the K, of degradation for insulin differed from the apparent K, of binding for insulin in isolated fat cells. Although this may be indicative of significant differences in tissue binding mechanisms, there are also important methodological differences in that we have considered total bound insulin while they considered only that portion of the bound insulin that was inhibited by the presence of 1Om6 M native insulin. In addition, they assumed that degradation of insulin by intact cells proceeds in accordance with Michaelis-Menten kinetics, i.e. that an insulin-degrading step is rate-limiting for the overall degradation process in intact cells and also that the substrate (insulin) concentration exceeds, by at least a factor of lo', the concentration of the enzyme(s) (Ref. 59,p. 31). It is possible that these assumptions are not applicable to degradation of insulin by intact cells. Freychet et al. (20) have suggested that insulin degradation and binding by isolated rat liver plasma membranes are unrelated because, whereas the insulin binding site has a low affinity for desalanyl-desasparaginyl insulin and a high affinity for insulin, degradation of both iodinated proteins by plasma membranes was comparable. In addition, desalanyl-desasparaginyl insulin inhibited the degradation of "'I-insulin as well as did native insulin (cf. our Table I). Several methodological differences may account for our different results. First, Freychet et al. also measured only a portion of.the total binding of Y-insulin and 1Z61-desalany1-desasparaginyl insulin. Second, they did not directly assess the degradation of '*'I-insulin. Third, there may be functional differences between the intact 8397 cell and isolated plasma membranes with respect to insulin degradation.
Separate binding and degradation sites occurring in different cell compartments could have different specificities, which become manifest only in broken cell preparations which expose the degrading enzyme(s) directly to unbound substrate. Furthermore, the enzyme(s) whose degrading activity was measured may or may not be that normally involved in insulin degradation.
The fact that Freychet et al. observed that degrading activity declined with increasing specific activity of insulin binding, an observation which may be compatible with solubilization of a physiological insulin-degrading enzyme loosely associated with the plasma membrane, is also consistent with the removal of proteases adsorbed during plasmamembrane isolation. More work will be necessary to resolve these differences.
The data presented here indicate that the binding of insulin to cells is necessary for its subsequent degradation.
However, not all the insulin bound is degraded (Fig. 5, lower).
If our estimate of the specific rate constant of dissociation of bound insulin (k_,), a value which is in general agreement with those reported in the literature (3,27,28,33,60) is approximately correct, a rough estimate of the relative amount of bound insulin which is subsequently degraded can be made. Assuming, in accordance with the considerations discussed above, that bound insulin is either released intact or degraded to final small degradation products, the fraction of bound insulin which is degraded should be given by the expression: k,,/(k-, plus k,,), or 43.8%. Indeed, we have found that about 40% of the total amount of the insulin retained following pulse infusion of '2sI-insulin into noncyclically perfused liver is degraded (44).' Although more work is necessary to define in detail the relation between insulin binding and degradation, two immediate implications are clear. First, at physiological insulin concentrations, the specificity of the over-all degradation process in hepatocytes may be governed primarily by the specificity of binding. And, second, as suggested by many workers (19,(61)(62)(63). the functional.
possibly molecular, link between insulin binding and degradation may perform a regulatory function in terminating the insulin signal.