Insulin Interactions with Liver Plasma Membranes

Insulin interactions with purified plasma membranes of rat liver were studied with respect to insulin degradation and specific binding to receptors. W-insulin was rapidly degraded upon exposure to liver membranes. After only 5 min of incubation at 30” of 1251-insulin (0.3 nlvr) and liver membranes (1 to 2 mg of protein per ml), 40 to 60% of the labeled hormone was degraded as measured by its ability to specifically bind to a second aliquot of membranes. After 90 min of exposure, less than 10% of the lz51-insulin was intact when measured by its ability to bind specifically to membranes. Binding by anti-insulin antibody, precipitation by trichloroacetic acid, and adsorption by talc were less sensitive methods of measuring degradation. Degradation of 1251-insulin was dramatically reduced at 1”. No significant deiodination was associated with the degradation process. Gel filtration patterns suggested that lzsIinsulin degradation products are composed mainly of small peptide fragments that loosely adsorb to the gel and are eluted after the salt peak. The independence of binding to receptors and degradation is strongly suggested by the following findings. (a) ‘=Idesalanine-desaparagine insulin, which has an affinity for receptors that is only 2% that of insulin, is degraded to the same extent as 1251-insulin. (b) There is no relationship between the bioactivity of an insulin analogue and its ability to prevent the degradation of 1251-insulin. (c) The apparent K, for insulin degradation is 1.7 X 1O-7 M, which is 40-fold more than that concentration of insulin that produces halfmaximal inhibition of specific binding of 1251-insulin to receptors in the liver membrane. (d) Insulin that is recovered from membranes upon dissociation of the hormone-receptor complex is undegraded. Proinsulin is very slowly degraded by the liver membranes, but may act as a competitive inhibitor of insulin degradation. This is similar to the findings by others of the insulin-specific protease of muscle and liver. These studies provide evidence that interaction of insulin with liver plasma membranes is a complex phenomenon

that involves at least two processes, degradation and binding to receptors. These two processes are largely independent and unrelated under a variety of conditions. Degradation of insulin must be accounted for in making precise quantitative measurements of the interaction of insulin with its receptors in the plasma membrane.
The liver is the major organ that removes insulin from the circulation (I, 2). It is also clear that insulin directly alters various metabolic pathways in the liver (34, while at the same time it is being inactivated by the liver. This degradative phenomenon has been attributed to proteolysis (7,8) or reduct,ive cleavage (9, 10). Until now it has been impossible to separate the interaction of insulin with its biologically active site and its degradation.
In previous studies (ll), we described and characterized a site of specific binding of 1251-insulin to purified liver plasma membranes. The data presented strongly suggested that this site is of specific biological importance for several reasons. (a) The insulin concentration for binding and displacement was in the range of hepatic portal blood concentrations; (b) the binding was inhibited only by insulins that are biologically active; and (c) in all cases inhibition of 12GI-insulin binding was proportional t,o the biological activity of the insulin preparation.
We have furthermore shown that monoiodoinsulin retains its full biological activity and that it is an appropriate tracer for studying t,he reactions between insulin and liver plasma membranes (12).
In the present work we have studied another aspect of the interaction of insulin with liver, the degradation or inactivation of hormone upon exposure to purified liver plasma membranes. The purpose of this paper is to describe some pertinent features of this degrading system and to establish that specific binding of the hormone to its biologically important receptors and degradative processes are largely independent phenomena. This is demonstrated by comparing the degradation of insulin analogues which have a wide variety in their affinity for the biologically important receptor. Furthermore, it will become apparent that the measured degree of inactivation of the insulin molecule Porcine insulin (P. J. 5589) and proinsulin (615-984B-99-C) were gifts of Eli Lilly.
Carboxymethylated A and B chains were purchased from Mann Chemical Company.
Synthetic oxytocin was a gift of Sandos, glucagon a gift of Eli Lilly, and bovine growth hormone a gift of Dr. R. Bates .
Guinea pig anti-porcine insulin serum was kindly supplied by Dr. P. Wright.
Other chemicals were of reagent grade.
Iodinations were performed with the following modifications of the method previously described (12). (a) The proportion of hormone: iodine :chloramine T was 1: 1:0.5 to 1, in molar equivalents; (b) the iodination volume was kept small (up to 50 ~1) ; (c) 20 to 30 s after exposure of hormone and lzfiI to chloramine T, sodium metabisulphite was added in 3-fold excess, and bovine serum albumin was then added to give a final albumin concentration of 1.57,; (d) the iodination mixt.ure was then immediately purified by chromatography on cellulose according to the procedure of Yalow and Berson (13); (e) after elution from the cellulose, the labeled hormone was diluted in Krebs-Ringer phosphate buffer (pH 7.5) t,hat contained 1% bovine serum albumin and was then stored at -20".
These modifications were introduced in order to shorten the chromatographic purification of ['2"I]monoiodoindin previously described (12) while retaining the stoichiometric conditions required to minimize the introduction of more than 1 atom of iodiue per insulin molecule and the possible deleterious effects of the oxidizing agent. "SI-insulin that had been prepared following this modified method was 97y0 precipitated by 55~; trichIoroacet,ic acid, 95% adsorbed to talc, and 91 to 95% bindable to anti-insulin antibody. It exhibited the same ability to bind specifically to liver plasma membranes as the preparation previously described (12). Labeled hormones were stored in small aliquots at -20" and thawed as needed. There was some gradual damage to hormones during storage with time, but the effect of this in each experiment was minimized by use of appropriate cont,rols for each experiment as described later.

Liver Cell Memhanes
Plasma membranes were prepared from rat livers as previously described (15). The fully purified plasma membrane fract,ion (Step 15 of Reference 15) was used in all studies except where specifically stated.
Protein concentrations were determined by the method of Lowry (16) using bovine serum albumin as the standard.

Incubation Procedure
Labeled hormone and liver membranes were mixed in the Krebs-Ringer ph0sphat.e buffer to give a final volume of 0.5 or 1 ml with final concentrations of 40 mM NaCl, 1.7 ml* KCl, 0.4 mM WgSO,, 0.4 mM KH#Oa, pH 7.5. The incubation mixture also cont.ained 1% bovine serum albumin and concentrations of hormone Lund membrane as indicated in legends to the figures. The reaction was stopped by cooling the mixture and immcdiatel\ centrifuging for 5 min at 10,000 X g in a Sorvall cent,rifuge or l3eckman-Spinco microfuge in a 5" refrigerated room. The supernates were rapidly transferred to chilled tubes and kept at lo. Analyticu,l procedures were then immediately performed.

Methods of A!leasuring Degradation and Inactivation of Hormone
Precipitation by 5% Trichloroocetie Acid-Aliquots (10 to 20 ~1) of the supernatantx were transferred to 1 ml of chilled medium containing 0.257, hutnan serum albumin in 0.05 M Verona1 buffer, pH 8.6. One milliliter of 10% trichloroacetic acid was immediately added, and the tubes were centrifuged for 5 min at 2500 rpm. Radioactivity was then counted in each precipitate and supernate.
Adsorption to Talc-The procedure was as above except that one 50-mg talc tablet was added to the medium containing each supernabe aliquot, vigorously mixed on a Vortex mixer, then centrifuged for 5 min at 2500 rpm. Under these conditions, the radioactivity adsorbed to talc is a measure of the undamaged hormone (17).
Paper Chroma~oeZ&rophoresti-*Xliquots (50 to 100 ~1) of the supernates were mised with 20 ~1 of plasma containing bromphenol blue and 0.05 DI KI and immediately applied to Whatman No. 3MM paper strips in a chromatoelectrophoresis apparatus. Undamaged hormone is estimated by its adsorption to the point of application, whereas damaged component)s migrate with serum proteins and free iodide migrst,es furt.her toward t,he anode (18).
Binding to Anti-Insulin AnfiOody-Aliquots (10 to 20 ~1) of each supernate were transferred to 0.5 ml of chilled medium that contained 0.25% human serum albumin, 0.01% rabbit Fraction II, 5 mM EDT+ 0.25% guinea pig serum, and guinea pig antiinsulin serum in a final dilution of 1 :lOOO in 0.05 M Verona1 buffer, pH 8.6. After 4 hours at 4", the antibody-bound hormone was precipitated by addition of a rabbit anti-guinea pig ant,iserum (14). After an addit.ional 15 hours at 4", 0.5 ml of cold medium was added and the mixture was cent'rifuged for 20 min at 2800 rpm in a 4" refrigerated centrifuge.
The radioactivity of both the supernate and the precipitate was counted.
Specific Binding to Liver fiiembranes-The binding activity of the '*"I-insulin remaining at the end of the experiment was determined by mixing duplicate 100.~1 aliquots of the superna-1'. I+eychet, R. Kahn, J. Roth, mu1 D. M. Neuille, Jr. tant containing the lz"I-iiisuliil with 100~~1 aliquots of 1 mnl KHCOI containing 200 to 300 pg of liver plasma membranes in the absence and presence of 10 pM uulnbeled insulin.
For each pair, the specific binding was calculat,ed by subtracting the percentage of 1251 radioactivity bound t,o the membrane pellet in the presence of 10 pM unlabeled insulin from the percentage bound in the absence of unlabeled insulin. This procedure is useful in eliminating the nonspecific adsorpt,ion or trapping since the specific binding sites of insulin are primarily occupied by unlabeled insulin.
The presence of 10 /.L~I unlabeled insulin, and the residual radioactivity bound by the membranes is accounted for by these nonspecific factors (II).
Expression of Results-In each experiment., appropriate controls were prepared which were identical with respect IO temperature, time, $1, and composition of the buffer, escept that membranes were omitted.
These controls represented 100% of the substrate available for degradation.
For most procedures used to evaluate hormone degradation, results were calculated as percentage of control rerttaining i l S follows.
$& Hormone remainirlg y0 Intact hormone ill experimental tube = y0 Intact hormone in control tube In all experiments there was less than 55, degradation control hormone.

RESULTS
x 100 of the Time and Tejnperature Dependents of Degradation-After only 10 min of exposure to purified liver membranes at 30' (Fig. l), approximately 40% of the '%insulin was degraded as measured by binding to anti-insulin antibodies or binding to fresh liver membranes.
This percentage increased to SOY0 by 90 min when measured by binding to liver membranes.
The parameters of talc adsorption and trichloroacetic acid precipitation also revealed degradation but were significantly less sensitive in all experiments.
When the same experiment was performed at 1" (Fig. I) using the same membrane and insulin preparations and concentrations, the degradation of l*"I-insulin was dramatically reduced to less than 15% after 90 min. This latter finding was in sharp contrast to the persistence of significant specific binding of insulin to liver membranes at this temperature (11). Gel Filtration Studies-The nature of the degradation products was studied by gel filtrutiou on a Sephades G-50 (fine) column which was carefully calibrated wit,h '"'I-albumin, '"II-proinsulin, 'Winsulin, and [13*I]NaI. Aft,er exposure of 'Winsulin to liver membranes for 90 min at 30", the majority of the radioactivity was recovered in a peak which elutes slight,ly after the lrlI marker ( Fig. 2, left). The fraction is probably composed of iodotyrosyl peptide fragments which absorb to the Sephades, accounting for their delayed elution (19). The fact t,hat this was not just deiodination and free 12sI was confirmed by chromatoelectrophoresis on Whatman No. 3x131 paper and by lack of adsorption to Dowes 1 x 10 anion exchange resin at pH 1. The possibility that the degradation results in A and B chains is excluded by the elution volume of the products.
The estimate of intact insulin by gel filtration closely approximated t,hat obtained by the ant.ibod\-binding method (Table I). 1251-insuliil at, 3.6 X lo-lo hl was csposcd to liver membranes (1.7 mg per ml) for the indicated times, as described under "Methods." At each time, supermttar~ts isol:rtcd by centrifugation were analyzed with respect to t)heir ability to adsorb to t,alc, be precipitated by trichloroacetic acid, bouud IO anti-insulin antibody and bound specifically to fresh liver menbranes (1 mg per ml). Analytical conditions are described undr~ "Methods." Data are expressed as percentage of controls (see "Methods").
Data (not shown) obtained by measuring adsorption of radioactivity to 1 X 10 Dowes (see "Methods") did not reveal deiodination of the labeled hormone. Results obtained with 1251-insulin at the same concentrntion (3.6 X lP" M) exposed to the same preparation of membranes (1.7 mg per ml) at 1" are shown by the dashed lines.
In this experiment, the intact insulin, as determined 1)~ billtlillg to fresh membranes, was again less than t'he value obt8nined 1)~ all other methods, suggesting the sensitivit,y of specific rcceptol binding to subtle changes in t,he iusulin molecule.
Note that :I significant portion of the radio&icky not recovered in t.he insulin peak was adsorbed to talc and precipitatc~d by t richloroacetic acid (Table I).
Dependence on Membrane atld Substrate Concerlt/,ccliotl-~ Degradation of insulin is a function of rnembr:~nr prot till CVXcentration (Fig. 3). Using fresh membranes at coucent r:itions as low as 0.125 mg per ml, a concentration at wltich only 4 to 6~~ of the labeled hormone was specifically bound to the memblaurs, as much as 42T0 of the 12"1-insuliu was degraded within 20 nlin at 30" as measured by binding to fresh membranes. 'lhre Tv:ls increased degradation with increasing membrane concentrations when measured at both 5 and 20 min (Fig. 3). The lack of linear dependence on membrane concentrations suggests that, inhibitors of enzyme action are also increased as membrane concentration is increased (20). It should be pointed out that most binding studies are conducted at lower membrane concentrations (0.25 to 0.50 mg per ml) and for only 30 min of incubation. Under these conditions degradation usually does not exceed 15 cr as measured by the antibody-binding method. To study this question, the degradation of 1251-insulin at 2.5 x lo+ M was measured in the absence and in the presence of unlabeled insulin, proinsulin, and DAA-insulin at concentrations of 2.5 X 1O-7 M. At this concentration insulin, proinsulin, and DAA-insulin inhibit the specific binding of l%insulin to liver membranes by 100, 70, and 40'& respectively (11) UnIabeled proinsulin which has an intermediate affinity for the liver receptor, was slightly less effective in inhibiting degradation than insulin and DAA-insulin.
Degradation of Proinsulin and DAA-Insulin-To further substantiate t,his difference betwee specific binding to insulin receptors and degradation by liver membranes, the degradation of 1251-proinsulin and 1251-l)hA-insulia was studied directly.
When lWproinsulin at 7 X lo-10 M was exposed to liver membranes at 30", very little degradation occurred (Fig. 7, left). Even aft,er 90 min of exposure, 95% of the radioactivity was still precipitated by trichloroacetic acid and adsorbed to talc, 90% was bindable to anti-insulin antibody, and SOY0 was capable of binding to fresh liver membranes. The degradation of Winsulin was determined simultaneously with t,he same membrane preparation for comparison (Fig. 7, left). The degradation of '2Qroinsulin was also studied by gel filtration (Fig. 2, right, and Table I). There was a good agreement between this method and other methods of measuring intact '251-proinsulin. It is of particular interest that there was no conversion of proinsulin to insulin.
On the other hand, 1251-DAA-insulin was degraded to almost the same extent as 12"1-insulin when exposed to the same membrane preparation in parallel experiments (Fig. 7, Tight). Thus, insulin and DAA-insulin, which differ by almost two orders of  Fig. 1 using liver membranes at 1.2 mg per ml in a total volume of 6 ml at 30" with 10d9 M 12%insulin.
Aft,er 30 min the mixture was centrifuged at 12,500 X g in a Sorvall centrifuge at 4" for 5 min.
The supernatant was removed and the membrane pellet was washed twice in 10 ml of buffer. The membranes were then resuspended in 0.1 in HCl containing 2y0 bovine serum albumin, allowed to incubate for 10 min, and then centrifuged at 12,500 X g at 4".
The To investigate the possibility of degradation of insulin bound t,o the membrane, 12"I-insulin which had been bound was eluted with acid and tested for degradation.
As can be seen in Table II, even after 30 min exposure to membranes, t,here was no degradation of the 1261-insulin bound to and eluted from the membranes, and, in fact, there was actually a purificatiou of the insulin toward the characteristics which it had on the day of iodination. This is in contrast to the insulin in the supernat,ant of the same experiment which was 75% degraded as measured by binding to fresh membrane (Table II).
Degradation in Other Liver Fractions and Fat Cells--We attempted to study the various fractions of liver and fat cells to on degradation The initial incubations were performed in t,he usual manner using a 600-y volume containing membrane protein (0.6 mg per ml), 1251-insnlin (5 X lo+ d, and the inhibitors in the concentration noted. After 30 min at 30" the mixtures were centrifuged and the supernst,ants were collected and tested for intact insulin by the four standard methods. The data were then calculated giving t,he control without inhibitor an arbitrary value of 100. The activity is expressed as percentage of control f S. determine whether the activity of the insulin-degrading system increases with purification of the insulin-specific receptor. As can be seen in Table III, there is an over-all decrease in the insulin-degrading activity with purification of the membranes and insulin receptors.
We also found insulin-degrading activity in fat cells and fat cell fractions which is of about the same order of magnitude as the purified liver membrane when amounts which have comparable specific insulin binding are used.

Inhibitors of Degradation-The degradation
of insulin by liver membranes will obviously affect any quantitative calculations of binding sit'es and affinity constants.
For subsequent studies,2 it was therefore important to attempt to find a suit,able inhibitor of degradation.
N-ethylmaleimide and p-chloromercuribenzoic acid appear to markedly inhibit insulin degradation, while Trasylol, zinc, and high sodium chloride concentrations are somewhat less effective (Table IV).
Concentrations of albumin from 0.1 to 4% have no effect on degradation.
Further studies are being conducted to determine the effects of these inhibitors on binding of insulin to liver plasma membranes.
In contrast to these findings, other investigators (33)(34)(35) have shown that insulin binding may occur to intact fat cells and to particulate fractions prepared from fat cells and rat liver under conditions where no * 12. Kahn, P. Freychet, J. Roth, and D. M. Neville, Jr., manuscript in preparation. degradation can be demonstrated. These latter studies, however, were performed at 2@' and with low concentrations of receptor preparation, both conditions which minimize degradation. We have previously described a site of specific binding of 1251-insulin in highly purified rat liver plasma membranes which has the characteristics of a biologically important recept,or (11). In the studies presented here we have demonstrated insulin degradation by these purified plasma membranes and characterized its specificity in contrast to that of the site of specific binding.
Studies of insulin degradation are complex. Not only is degradation a function of time, temperature, and membrane concentration, it is also a function of the criteria used t.o evaluate the degradation.
In this respect, the ability of insulin to bind specifically t,o fresh membrane receptors appears to be t.he most sensitive measure of structural integrity since this correlates well with bioactivity.
Binding by ant.i-insulin antibody provides useful information on the general integrity of the hormone molecule, but, as shown by our data in the present and other studies (11)) antibody binding often poorly discriminates changes that affect bioactivity of specific binding to receptors. Gel filtration, chromatoelectrophoresis, precipitation by trichloroacetic acid, and adsorption to t'alc consistently underestimate the actual extent of degradation of inactivation of the hormone (Table I).
We have not attempted to identify the products of the degradation process. There is, however, some evidence that insulin undergoes a proteolytic cleavage upon exposure to the membranes, as shown by the gel filtration data, as well as by similarities (see infra) between our dat.a and t.hose obtained by Brush with a soluble enzyme from rat muscle and liver (21, 26). Proteolytic degradation of insulin by rat adipose tissue has also been established (36). Using the assay described by Katzen and Stetten (37) we3 were unable to det,ect in purified liver membranes a significant glutathione-insulin transhydrogenase activity which would result in a reductive cleavage of insulin such as suggested by Tomizawa (9). Degradation of insulin upon exposure to the purified liver membranes and specific binding of the hormone to its recept,ors in the same membranes appear to be largely independent phenomena. This is strongly suggested by several findings. 1. '251-DAA-insulin whose affinity for receptors is only 2% that of insulin (11) is degraded to the same extent as 1251~insulin. Conversely, 1251-proinsulin whose affinity for insulin receptors is 20% that of insulin on a molar basis (11) is degraded to a much smaller extent by the membranes than either insulin or DAAinsulin.
2. There is no relationship between the bioactivity of insulin analogue and its ability to prevent the degradation of 1251-insulin. Thus, DAA-insulin is as effective as insulin in inhibiting 1251insulin degradation but has only 2% of the binding affinit,y and bioactivity of insulin (Fig. 6).
3. The apparent K, (half-maximal concentration) of insulin for the degradative process (1.7 X 10d7 M) is about 50 times higher than the concentration of insulin that produces 50% inhibition of the 1251~insulin specific binding to the membranes (11). This difference might be even higher if the latter value were corrected by accounting for insulin degradation during exposure to menlbranes.4 Vol. 247,No. 12 4. Finally, serial purification of the plasma membrane and specific insulin binding site is associated with a decrease, rather than increase, in insulin degrading activity.

Independence of Insulin
There is, therefore, subst,antial evidence that specific binding to receptors and degradation are separate phenomena whose occurrences appear to be unrelated under a variety of conditions.
The labeled hormone that is actually bound to the membranes appears t,o be protected from degradation (Table II). Similar findings have been reported with glucagon (32) and insulin (33-35).
This does not exclude the possibility that insulin which has bound to the receptor and activated cellular processes is inactivated and rapidly released into the medium.
Further studies are required to investigate this hypothesis.
It is of interest that no significant conversion of proinsulin to insuliu occurs during exposure of proinsulin to the liver membranes. No conversion of proinsulin to insulin has also been observed in viva (38) or with the isolated perfused rat liver or with rat hemidiaphragm (38). Although gel filtration data do not allow us to exclude the conversion of intact single chain proinsulin to two chain intermediate forms between proinsulin and insulin, this possibility appears unlikely.
The absence of significant conversion of proinsulin to insulin complements other observations which strongly suggest that proinsulin has intrinsic biological activity (38-40) that may be entirely accounted for by its lower affinity for receptors (11).
The insulin degradation process is not restricted to liver. Degradation of insulin has also been observed with kidney (26, 41), muscle (21), adipose tissue (42), isolated fat cells (42,43) and both homogenates and particulate fractions of fat cells (36,44). Our data on fat cells and their fractions (Table III) confirm these findings.
Serial fractions obtained during the purification procedure of the liver plasma membranes show a progressive decrease in degrading activity with a progressive increase in specific binding.
Another interesting aspect of the work presented here is the remarkable similarity, if not identity, between the insulin degrading system of the purified liver membranes and the partially purified soluble enzymes from rat muscle (21) and liver (25). Both systems show a relative specificity for insulin when compared to proinsulin, whereas the latter behaves as a competitive inhibitor of insulin degradation. The K, for insulin is remarkably close and both systems show inhibit.ion by sulfhydryl inhibitors and stimulation by reduced glutathione. Other polypeptide hormones do not significantly affect the degrading activity.
A discrepancy between the degradation of insulin and proinsulin has been observed with isolated perfused rat liver (31), kidney (41), and rat adipose tissue (21,42).
The possible physiological significance of these findings deserves comment.
Significant insulin degradation was observed at concentrations close to those in hepatic portal blood (45). The possibility that, an "insulin-specific enzyme" is located in the plasma membrane is suggested by the similarity between the liver perfusion studies and the studies presented here. Hemmington and Dunn (46) have separated a discrete peak of insulin immunoreactivity from liver homogenates which appears to be bound to a 300,000 molecular weight fraction.
This fraction elutes coincident with the insulinase activity. The liver plasma membranes which we use have been shown to be rich in protease activity (47). These findings all suggest that the insulin degradation by the liver plays a major role in allowing more of the secreted proinsulin to reach the peripheral circulation, thus accounting for the longer half-life of proinsulin (48) and the reduced anteriovenous difference (49) when compared to insulin.
In summary, the interaction of insulin with purified plasma membranes of rat liver involves at least two largely independent processes: specific binding to receptors and degradation.
Insulin is not degraded when bound to the membranes.
In contrast to insulin and DAA-insulin which, despite the wide difference in their affinity for receptors, are degraded to the same extent, proinsulin is much more slowly degraded and is not converted to insulin.
However, proinsulin is a competitive inhibitor of insulin degradation.
Further studies are in progress to quantitate the specific interaction of insulin with its receptors by accounting for the insulin degradation process.