Interactions of glucagon and glucagon analogs with isolated canine hepatocytes.

We have used glucagon and nine glucagon analogs to investigate the interactions of these ligands with glucagon-binding sites present on isolated canine hepatocytes. Curves reflecting the inhibition of 125I-labeled glucagon or 125I-labeled analog binding to cells by the 10 peptides spanned, overall, a 10(6)-fold range of hormone concentration, were consistent with hormone binding to two classes of binding sites in each case, and fell into two groups, one of which contained curves that were considerably more shallow than the other. Only conditions that emphasized prior binding to low affinity sites resulted in the rapid and extensive dissociation of receptor-bound ligand from isolated cells. Finally, all 10 peptides exhibited a concentration-dependent inhibition of the incorporation of [14C]fructose into hepatocyte glycogen that correlated best with dissociation constants for high affinity rather than for low affinity binding. We conclude that (a) the association of ligand with the high and low affinity glucagon-binding sites of isolated canine hepatocytes is a characteristic of analogs modified at diverse sites throughout the peptide hormone, (b) the different rates of dissociation of ligand from the two populations of binding sites most probably account for the biphasic dissociation of ligand from isolated cells and for the different affinities of the two receptor populations for ligand, and (c) the activity of glucagon and glucagon analogs to inhibit the incorporation of fructose into hepatocyte glycogen arises from the association of ligand with high affinity binding sites.

Interactions of Glucagon and Glucagon Analogs with Isolated Canine Hepatocytes* (Received for publication, September 15, 1986) William A. Hagopian4  We have used glucagon and nine glucagon analogs to investigate the interactions of these ligands with glucagon-binding sites present on isolated canine hepatocytes. Curves reflecting the inhibition of '2SI-labeled glucagon or '2SI-labeled analog binding to cells by the 10 peptides spanned, overall, a 1OB-fold range of hormone concentration, were consistent with hormone binding to two classes of binding sites in each case, and fell into two groups, one of which contained curves that were considerably more shallow than the other. Only conditions that emphasized prior binding to low affinity sites resulted in the rapid and extensive dissociation of receptor-bound ligand from isolated cells.
Finally, all 10 peptides exhibited a concentration-dependent inhibition of the incorporation of ['4C]fructose into hepatocyte glycogen that correlated best with dissociation constants for high affinity rather than for low affinity binding. We conclude that (a) the association of ligand with the high and low affinity glucagonbinding sites of isolated canine hepatocytes is a characteristic of analogs modified at diverse sites throughout the peptide hormone, (b) the different rates of dissociation of ligand from the two populations of binding sites most probably account for the biphasic dissociation of ligand from isolated cells and for the different affinities of the two receptor populations for ligand, and (e) the activity of glucagon and glucagon analogs to inhibit the incorporation of fructose into hepatocyte glycogen arises from the association of ligand with high affinity binding sites.
Although the importance of plasma membrane receptors in directing the specific interaction of peptide hormones with their target cells has been amply documented (1, 2), the question of receptor heterogeneity plays a crucial role in the analysis of ligand binding and in the assessment of receptor function. For low molecular weight and nonpeptide ligands 20595 (to H. S. T.) and DK-21085 (to V. J. H.) from the National * This work was supported in part by Grants DK-18347 and DK-Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. (such as those participating in the adrenergic ( 3 4 , opiate (6, 7), histaminergic (8,9) nicotinic (lo), and dopaminergic (11) systems), multiple receptor populations are readily demonstrated on a variety of target cells; and, in many cases, the separate activities of receptors in each population are known. Early studies on the binding of insulin to its target tissues showed that the interaction failed to correspond to that expected for a single binding equilibrium (12)(13)(14), and models involving two receptor populations (12)(13)(14)(15)(16) or negative homotropic interactions (17)(18)(19) have been proposed to account for the complexity of hormone association with receptor; recent experiments have in fact identified multiple molecular states of the insulin receptor on rat hepatoma cells (20) and adipocytes (21). Use of peptide hormone analogs and kinetic measurements in quite different systems has resulted in the identification of multiple populations of receptors for somatostatin (22), interleukin-2 (23), transforming growth factor 0 (24), and angiotensin I1 (25).
Complexity in the binding of glucagon to its target tissues (most notably the liver) has been recognized for some time, with the result that several investigators have modeled the interaction in terms of two separate receptor populations (26)(27)(28). Others have not observed deviations from apparently simple binding, however, and have proposed the existence of only a single population of glucagon receptors on hormonesensitive tissues (29)(30)(31)53 '3] glucagon is used as the radiolabeled probe (32). It is clear that the quality of radiolabeled probe, the concentration and extent of binding of radiolabeled ligand which apply to a given experiment, the species used, and the cell or membrane preparation employed (cf. Refs. 28,33,and 34) all have the potential for affecting the shape of relevant binding competition curves and the assignment of simple or complex binding kinetics.
By use of isolated canine hepatocytes and derived plasma membrane vesicles (28), canine hepatic plasma membranes (35), and isolated rat hepatocytes (36), we have modeled glucagon interactions with its target tissues in terms of two noninteracting populations of binding sites having different affinities for ligand. Additional evidence supporting the existence of high and low affinity binding sites on isolated hepatocytes arises from studies of a synthetic analog that appears to interact with only one of the two classes of binding sites (albeit with a very low affinity) (37) and analysis of glucagonreceptor complexes that are solubilized from plasma mem-branes by treatment with digitonin (35). In order to approach a better understanding of glucagon association with membrane receptors, we have examined both the interactions of glucagon and nine glucagon analogs with isolated canine hepatocytes and the abilities of these peptides to alter hepatocyte glycogen metabolism. Our results show that (a) in all cases, glucagon binding to isolated hepatocytes is best described by the association of ligand with two populations of binding sites; ( b ) whereas some structural modifications of glucagon decrease in parallel its affinity for high and low affinity binding sites on isolated hepatocytes, most preferentially decrease its affinity for the high affinity population of binding sites; ( c ) ligand retention by high affinity binding sites accounts for the biphasic curve defining the rate of dissociation of radiolabeled hormone from isolated cells; and (d) the action of glucagon to alter hepatocyte glycogen metabolism in our system correlates best with the association of hormone with high affinity binding sites. The accompanying paper (38) identifies both the absence of an easy correlation between the stimulation of hepatocyte CAMP accumulation by glucagon and the occupancy of high or low affinity binding sites and the effect of glucagon to cause homologous and heterologous inhibition of the hepatocyte adenylyl cyclase system.

MATERIALS AND METHODS
Glucagon and Glucagon Analogs-Glucagon was obtained from ELANCO (Indianapolis, IN). Table I identifies glucagon analogs used in this study. Glucagon, the product of total, solid-phase peptide synthesis, was the gift of Dr. David H. Coy (Tulane University, New Orleans, LA) (39). Des-His1-glucagon was obtained from Novo Pharmaceuticals (Copenhagen, Denmark). Cyanogen bromidecleaved glucagon was prepared by published methods (40); it was purified by use of reverse-phase HPLCl under isocratic conditions as described (32, 46). Methods for the preparation of [Nu-Cbm-His',N" Cbm-Ly~~~]glucagon (41) and [N"-trinitrophenyl-His',horn~-Arg'~] glucagon (41) have been described before. All other glucagon analogs were DreDared by total, solid-phase peptide synthesis in the laboratory PeDtide Radioiodination-Glucagon. des-His'-glucagon, ID-Phe41 glucagon, and cyanogen bromide-cleaved glucagon were each subjected to radioiodination by use of Na'*'I and chloramine T (32) and were purified by isocratic, reverse-phase HPLC (32, 46) as has been described acetonitrile concentrations ranged from 28 to 29.5% for optimal separation of the different peptides. Analysis of the position of ['2SI]iodotyrosine present in each of the radiolabeled peptides was accomplished both by peptide mapping after digestion with trypsin and by automated radiosequence determination (by use of Polybrene, a Beckman 890 Sequencer, and a standard 0.1 M Quadrol program). Radiolabeled peptides containing ['251]iodotyrosine exclusively at position 10 (position 9 in the case of des-His1-glucagon) were chosen for further study (data not shown).
Cells and Cell Incubation-Hepatocytes were isolated from canine liver by collagenase digestion and were incubated at 30 "C in glass scintillation vials (2 X lo6 cells/ml) as described (32,36); cell viability equaled or exceeded 98% as assessed by exclusion of the dye trypan blue. For experiments assessing the dissociation of receptor-bound ligand, hepatocyte suspensions were incubated with the appropriate peptides at 30 "C in larger volumes (by use of Erlenmeyer flasks), after which the suspensions were diluted with ice-cold buffer and centrifuged; the cell pellets were resuspended to their original volumes with ice-cold incubation medium, and dissociation of receptor-bound ligand was determined after incubation at 30 "C by removing aliquots of the suspension and measuring the decrease in cell-associated radioactivity after diluting and pelleting the cells. Control experiments (in which 1251-labeled glucagon was added to fresh hepatocytes in an amount equivalent to that occurring bound to receptors during a typical experiment) demonstrated that rebinding of dissociated ligand could have accounted for no more than 10% of the cell--.
' The abbreviations used are: HPLC, high performance liquid chromatography; Cbm, carbamoyl.
V. J. Hruby, unpublished results. associated radioactivity initially recorded. Experiments assessing the biological activities of glucagon and glucagon analogs relied on the ability of the hormone to inhibit the incorporation of ['4C]fructose into hepatocyte glycogen; details of the assay have been described (28,36). Each cell experiment was performed on at least two separate occasions by incubation of samples in duplicate or triplicate; representative experiments are shown.

RESULTS
Initial experiments tested the interactions of glucagon and the glucagon analogs identified in Table I with isolated canine hepatocytes. Fig. 1 ( a and b) demonstrates the ability of each of the 10 peptides to inhibit the binding of [ [1251]iodo-Tyr'o] glucagon to hepatocyte receptors. Fig. 1 documents that (a) the binding inhibition in each case is complete; and ( b ) the peptides appear to fall into two classes, as evaluated by the range of peptide concentrations required to decrease binding from 90 to 10% of the value obtained in the absence of inhibitor. Peptides 1-3, and 5 each require about a 30,000fold increase in peptide concentration and yield a family of rather shallow curves; peptides 4 and 6-10 each require about a 500-fold increase in peptide concentration and yield a family of steeper, but still complex curves. Binding inhibition could be modeled only in terms of two binding equilibria for each of the peptides examined. Use of Equation 10 from Ref. 28 and a nonlinear Gauss-Newton method (47) permitted the determination of apparent dissociation constants for the high and low affinity binding of each of the peptides to hepatocyte receptors. (The model considers two non-interacting populations of glucagon-binding sites which do not change in number and which differ only in their affinities for ligand.) As documented in Table I, dissociation constants for high affinity binding ( K D~) ranged from 0.33 nM to 1.3 PM, whereas dissociation constants for low affinity binding (KD2) ranged from 63 nM to 3.6 PM. The more disparate values obtained for KO, and KD2 for peptides 1-3 and 5, relative to the values of KDl and KD2 obtained for the remaining peptides, account for the shallowness of the binding curves shown in Fig. la. In fact, structural alterations in the glucagon analogs in most cases (e.g. peptides 4 and 6-10; see Table I) decrease the affinity of ligand binding to high affinity sites to a greater extent than they do the affinity of ligand binding to low affinity sites.
As we wished to study in detail the direct interaction of glucagon analogs with hormone receptors of isolated hepatocytes, peptides 3, 5 , and 6 were subjected to direct radioiodination, and the products were purified by reverse-phase HPLC as described under "Materials and Methods." The respective peptides containing [ 1251]iodotyrosine at position 10 (position 9 in the case of des-His1-glucagon) were first examined by assessing the time courses of ligand binding to cell-surface receptors. As shown in Fig. 2 A , the extent to which the four radiolabeled probes bound to isolated hepatocytes decreased in the order glucagon > [D-Phe4]glucagon > des-His1-glucagon > cyanogen bromide-cleaved glucagon. Notwithstanding both the different extents to which the various ligands were bound and the shape of the binding curve for des-His'-glucagon (a curve that reproducibly rises rapidly and then declines somewhat at longer periods of incubation), Fig. 2B shows that the binding of all four radiolabeled probes approached steady state at approximately the same rate. Control studies assessing cell-mediated degradation of lZ5Ilabeled glucagon, [D-Phe4]glucagon, des-His1-glucagon, or cyanogen bromide-cleaved glucagon by gel filtration and tryptic peptide mapping (32,48) showed that greater than 93% of cell-associated radioactivity was intact hormone or analog in each case and that no preferential degradation of any of the four radiolabeled peptides had occurred in cell incubation  Fig. 7. Values of KD, and Km (apparent dissociation constants for high and low affinity binding sites, respectively, on isolated canine hepatocytes) were derived by mathematical modeling from the data shown in Fig. 1  e Dissociation constants were derived by analysis of binding inhibition data by use of a nonlinear Gauss-Newton method according to Equation 10 in Ref. 28. For our analysis, the value PI (the fraction of radiolabeled hormone that binds to high affinity receptors in the absence of competing peptide (a parameter independent of the nature of the competitor)) was set at 0.52. The value of Pp (the fraction of radiolabeled hormone that binds to low affinity receptors in the absence of competitor) was allowed to vary somewhat to account for the apparently non-zero value of binding inhibition at great dilution (see Ref. 28 Table I. Cell incubations proceeded for 30 min at 30 "C. Each vial contained radiolabeled peptide (about 20 fmol in 1 ml) and unlabeled peptide at the concentrations shown; the data represent the mean of duplicate or triplicate determinations. For ease in presentation only, binding inhibition curves (each derived by mathematical modeling) have been grouped into two separate panels and error bars are not shown; standard errors of the replicate determinations were always less than 5% of control binding. No correction has been made for socalled nonspecific binding. Control binding is defined as the amount of radiolabeled ligand becoming cell-associated in the absence of any added competitor (8-12% of that originally added). medium (data not shown). Furthermore, results indistinguishable from those reported in Fig. 1 were obtained when cells were incubated with radiolabeled glucagon at 20 "C, a temperature that does not permit hormone internalization (49). Thus, neither the accumulation of cell-mediated hormone degradation products nor the internalization of hormone can account for the greater complexity of binding data observed for some analogs when compared to the data obtained for others.
T o examine the effect of different radiolabeled probes on our ability t o assess glucagon-receptor interactions and to confirm the validity of our interpretations, we next compared  Table I) and therefore with the relative binding of the two radiolabeled probes being equivalent for high and low affinity binding sites in the absence of competitor. Fig. 3c shows the markedly different curves obtained for the inhibition of [ [1251] i~do-Tyr~~]glucagon binding by glucagon and by des-His'glucagon and demonstrates a case very different from the one presented above; a greater increase in the value of KD, relative to that of KD2 for the binding of des-His'-glucagon to hepatocytes (KD2/KD1 = 4.9) has caused the curve to steepen considerably. Fig. 3d shows the results of complementary experiments performed by use of lZ5I-labeled des-His'-gluca- Although the maximal binding of Iz5I-labeled glucagon cleaved by cyanogen bromide was only 0.9% of total added radioactivity, control experiments showed that, in all cases, less than 0.4% of added radioactivity could be accounted for as nonspecifically bound or trapped peptide. The data shown were corrected for this so-called nonspecific binding.  -glucagon (0). No correction in the data has been made for so-called nonspecific binding. Control binding is defined as the amount of radiolabeled ligand becoming cell-associated in the absence of any added competitor (13,8,13, and 6% of total radiolabeled peptide added in a-d, respectively). gon as the tracer; curves reflecting the inhibition of lZ5Ilabeled des-His1-glucagon binding by glucagon and by des-His1-glucagon are parallel, exhibit the steepness characteristic of the use of des-His'-glucagon as inhibitor (Fig. 3c), and appear to reflect the preferential association of this radiolabeled peptide at low concentrations with lower affinity binding sites.
Considerations of the fraction of total receptor-bound li-gand that is associated with high and low affinity binding sites are important for understanding the results of Fig. 3 (c  and d). Fig. 4a presents the relevant theoretical curves (derived from mathematical modeling (28)) for the association of glucagon and des-His1-glucagon with isolated canine hepatocytes. In the case of glucagon, high and low affinity binding sites account about equally for ligand association with cells at low hormone concentrations; ligand binding to low affinity sites (sites present in 100-fold greater relative number than those accounting for high affinity binding, cf. Ref.

28)
predominates only as the concentration of hormone is increased through the range corresponding to that saturating high affinity binding sites. In the case of des-His1-glucagon, low affinity binding sites account for the vast majority of ligand association with cells at any hormone concentration that might be chosen. The logarithmic reciprocal plots of Fig.  4 (b and c ) illustrate important differences between the use of lZ5I-labeled glucagon and lZ51-labeled des-His1-glucagon. Whereas (a) the complexity of the curve describing the inhibition of lz5I-1abeled glucagon binding by glucagon arises from the dissimilar dissociation constants that apply to its interaction with high and low affinity binding sites, and (b) the apparent simplicity of the curve describing the inhibition of lZ5I-labeled glucagon binding by des-His1-glucagon arises from the greater similarity of the applicable dissociation constants, the simplicity of the curves describing the inhibition of lZ5Ilabeled des-His1-glucagon binding by the two peptides arises from the selection of a probe that preferentially binds to low affinity sites and assesses primarily the low affinity binding of competing ligands. It is clear from Fig. 4a that lZ5I-labeled des-His1-glucagon or lZ5I-labeled glucagon plus unlabeled glucagon in total concentration exceeding about 20 nM would each bind in greatest fraction to low affinity binding sites of canine hepatocytes. In an attempt to provide a third condition under which ligand would be preferentially bound to low affinity sites, we rea-

Glucagon Interactions with
Canine Hepatocytes Log molar peptide concentration

FIG. 4. Theoretical curves depicting for canine hepatocytes the fraction of receptor-bound glucagon and des-Hid-glucagon that is bound to high and low affinity receptors as a function of peptide concentration and logarithmic reciprocal plots for glucagon and des-His'-glucagon binding to isolated
canine hepatocytes. a, the fraction of total receptor-bound ligand separately associated with high and low affinity receptors for glucagon (-) and for des-His'-glucagon (---) as a function of peptide concentration. The data were calculated from binding parameters (&(high), KDclOw), receptor numberhigh, receptor numberl,,) determined by mathematical modeling (28); for each peptide and for each receptor population, the number of occupied receptors was calculated as ([ligand]/([ligand] + K O ) ) X receptor number. b and c, logarithmic reciprocal plots according to the method of Loftfield and Eigner (54) expressing the inhibition of '251-labeled peptide binding shown in Fig.

b, inhibition of [[1251]iodo-Tyr'o]glucagon binding by glucagon (a) and des-His'-glucagon (0). c, inhibition of [ [1*61]iodo-Tyrg]des-His'-
glucagon binding by glucagon (e) and des-His'-glucagon (0). Data in these plots have been corrected for so-called nonspecific binding by subtracting the amount of radiolabeled ligand that remained in the pellet in the presence of M glucagon.
soned that, given the 100-fold greater number of low affinity binding sites relative to high affinity binding sites on canine hepatocytes (28), binding of [ ['25J]iodo-Tyr'o]gl~cag~n under pre-steady-state conditions might favor ligand association with low affinity sites; this suggestion is consistent with the data of Fig. 2 and would hold as long as the rate constant for ligand binding to high affinity sites were not much greater than that for ligand binding to low affinity sites. Fig. 5 compares the concentration dependence of the inhibition of lZ5I-labeled glucagon binding to hepatocytes by unlabeled glucagon at 0.5 and 30 min of incubation. Indeed, binding to lower affinity sites is favored during very short incubation periods. To test further the consequences of glucagon interactions with both high affinity and low affinity binding sites of canine hepatocytes, we examined in detail the dissociation of previously bound '251-labeled glucagon and glucagon analogs. As shown in Fig. 6a, the dissociation of receptor-bound [['TI i~do-Tyr'~]glucagon is clearly biphasic, a result that has been documented before (26,27,36) and that is consistent with dissociation of the ligand from two different populations of binding sites. The dissociation of [D-Phe4,['251]iodo-Tyr10] glucagon (a ligand that appears to bind to receptors in a manner very similar to that of '251-labeled glucagon (Fig. 3, a  and b) parallels that of [['251]iodo-Tyr10]glucagon. In contrast to these findings, the dissociation of [['251]iodo-Tyr'o]des-His1-glucagon (a ligand which has been shown to bind preferentially to low affinity binding sites) is rapid and is nearly complete by 30 min. Moreover, as is shown in Fig. 6c,  glucagon under conditions where the radiolabeled ligand would be expected to interact in greatest proportion with low affinity binding sites, that is, in the presence of 20 n M glucagon (see Fig. 4a) and after only 0.5 min of incubation (see Fig. 5). In each case, the extent of dissociation of the radiolabeled glucagon greatly exceeds that which was found under conditions of low ligand concentration and steady-state binding. Together, Fig. 6 (a-c) documents the rapid dissociation of glucagon that has been bound to low affinity binding sites and, concomitantly, the retention of hormone that has been bound to high affinity binding sites of the canine hepatocyte.
Last, we tested the activities of glucagon and glucagon analogs in altering hepatocyte glycogen metabolism. Fig. 7a illustrates the abilities of the peptides listed in Table I  and were then used in the dissociation experiment as described above. c, normalized data from a obtained from the dissociation of '251-labeled glucagon (0) and lZ5I-labeled des-His'-glucagon (0) after association for 30 min. Data were normalized as follows: ((radioactivity remaining cell-associated after each dissociation period) -(radioactivity remaining cell-associated after 60 min of dissociation))/((radioactivity remaining cellassociated at the close of the initial incubation) -(radioactivity remaining cell-associated after 60 min of dissociation)). Normalization in this way permits comparison of the relative rates of peptide dissociation from the readily dissociable component of receptor-bound ligand in each case. Data represent the mean f S.E. of triplicate or duplicate determinations. Control binding was defined as the amount of radiolabeled peptide bound to cells (as a percent of total radioactivity added) prior to resuspending cells at 30 "C to initiate ligand dissociation; values were 14, 13, and 5.7% for labeled glucagon, [~-Phe']glucagon, and des-His'-glucagon, respectively (a), and 14, 3.5, and 1.5% for the 30-min, 30-min with 20 nM glucagon, and 30-9 incubations, respectively (b).
peptide 9), and the 100,000-fold range of concentrations required for the summed half-maximal effects. Fig. 7b shows the relationship between the biological activities of these peptides and their dissociation constants for interactions with the high and low affinity binding sites previously identified on isolated cells. Correlation coefficients assessing the relationship between half-maximal biological activity and the two dissociation constants are 0.80 for the high affinity binding site and 0.49 for the low affinity binding site. Although peptide 9 has been omitted from these considerations, this analog represents an unusual case in that it does not appear to stimulate the accumulation of hepatocyte CAMP (38) and that its biological activity is obtained only at concentrations of peptide causing nearly total saturation of hepatocyte-binding sites (see Fig. 7a).

DISCUSSION
Our approach to examining the question of heterogeneity in the interactions of glucagon with hepatic parenchymal cells has relied on the use of isolated hepatocytes, preparations that exhibit more obvious complexity in glucagon binding than that observed for isolated hepatic plasma membranes (29-31, 50-53). Although we do not yet know if differences identified between glucagon binding to hepatocytes and to hepatic plasma membranes arise in greatest part from selective recovery of only the bile canalicular face of parenchymal plasma membranes in the latter case (33,34) or from changes in the interactions of glucagon with binding sites occurring in different membrane environments (28), it is clear that, under our conditions, glucagon binding to canine hepatocytes cannot be modeled in terms of ligand association with a single receptor population. We therefore undertook this study (a) to clarify the basis for assigning hormone interaction to both high and low affinity binding sites at steady state, ( b ) to examine the consequences of potential receptor heterogeneity under non-steady-state conditions, and (c) to determine the relationship between the association of glucagon with cellsurface receptors and the ability of the hormone to modify hepatocyte glycogen metabolism.
First, our analysis of the inhibition of binding of [ ['251]iodo-Tyr'o]glucagon to canine hepatocytes by glucagon and by nine glucagon analogs has revealed the importance of glucagon structure in determining ligand association with high and low affinity binding sites. Related findings identify (a) three hormone analogs that show shallow binding inhibition curves characteristic of those exhibited by glucagon itself, (b) the independence of the shape of these curves on the choice of radiolabeled ligand as long as that ligand is chosen from the same family of peptides, (c) six hormone analogs that show steeper binding inhibition curves which are still not consistent with the simple association of ligand with a single class of binding sites, and (d) the use of a radiolabeled ligand from the second group of peptides ([ ['251]iodo-Tyrg]des-His'-glucagon) as a probe for investigating hormone interaction with low affinity binding sites. The degree of divergence of the dissociation constants calculated for high and low affinity binding, rather than the absolute values of these constants, determines the shape of the corresponding binding inhibition curves; for cyanogen bromide-cleaved glucagon (the peptide exhibiting the most shallow curve), the ratio KD,/KD, is 220, whereas for [~-Phe~,Tyr~,Arg'~]glucagon (the peptide exhibiting the steepest curve), the ratio KDB/KD1 is only 3.
Whereas the structural modifications of glucagon identified in Table I undoubtedly have multiple and complex consequences in determining the association of hormone analogs with high and low affinity binding sites, several conclusions with regard to structure-function relationships can be made.  Table I. a, biological activity of glucagon and glucagon analogs. Hepatocytes were incubated with ['4C]fructose and with peptides at the indicated concentrations prior to the isolation and assay of glycogen. The data (mean of duplicate or triplicate determinations) are presented as the percentage of basal incorporation of fructose (that occurring in the absence of added peptide, approximately 7% of added radioactivity) uersus the logarithm of the molar peptide concentration. Standard errors were always less than 5% of control incorporation of fructose and are not shown for the sake of clarity. b, correlation of biological activity with dissociation constants for high and low affinity peptide binding. For each peptide, individual points plot the concentration of peptide causing half-maximal biological activity (abscissa) uer5u.s the calculated dissociation constants for high affinity (-) or low affinity (---) binding as determined by mathematical modeling and as presented in Table I (ordinate). Correlation coefficients assessing the statistical relationship between half-maximal biological activity and high and low affinity dissociation constants are 0.80 and 0.49, respectively. Peptide 9 has not been included in this analysis for the reasons stated in the text. association with high affinity binding sites to a greater degree than they do association with low affinity binding sites; the exceptions are peptides 1,3, and 5, which show nearly parallel decreases in affinity for high and low affinity binding. (b) The COOH-terminal few residues of glucagon appear to play only a small role in determining selectivity for high affinity binding (although they contribute significantly to the affinity of ligand-receptor interactions overall); additional studies will be needed to establish the differential requirements of the COOH-terminal residues for high and low affinity binding. ( c ) The NH,-terminal histidyl residue of glucagon appears to play an important role in determining hormone association with high affinity sites (although neither the histidyl residue itself nor a peptidyl a-amino group appears to be absolutely required). (d) The introduction of diiodo-Tyr" (along with Lys17, Lys" and G1uZ1, amino acid substitutions that affect ligand binding only slightly; see Table I) has a major consequence in reversing by 60-100-fold the low receptor binding potency exhibited by [~-Phe~,Tyr~,Arg'~]glucagon. Whether this effect arises from a specific side-chain contact between diiode-Tyr" and the glucagon receptor or from the correction of an unfavorable conformation caused by the introduction of D-Phe4 and Tyr5 remains to be determined. It is the case, however, that these structural changes have resulted in an unusual glucagon analog that apparently fails to stimulate hepatocyte adenylyl cyclase (38)3 and that stimulates hepatocyte glycogenolysis only a t very high ligand concentrations (this report).
Second, we have shown that radiolabeled glucagon and radiolabeled glucagon analogs associate with isolated canine hepatocytes to different degrees, but approach steady-state binding at essentially the same rate.
Because at low concentrations of ligand the respective extents of interaction of [ ['251]iodo-Tyr'o]gl~cag~n and [ [1251]iodo-Tyrg]des-His1-glucagon with high and low affinity binding sites differ considerably, it appears that receptor avidity depends (at least in these cases) more on the relevant rate constants for ligandreceptor dissociation than on the relevant rate constants for ligand binding. In fact, separate determination of the rates of dissociation of previously bound ligand is in agreement with this suggestion. Since (a) receptor-bound [ ['251]iodo-Tyrg]des-His1-glucagon dissociates from hepatocytes essentially completely and to an extent significantly greater than that found for either lZ5I-labeled glucagon or [D-Phe4]glucagon, (b) normalized curves describing the rapid phases of dissociation of [ ['251]-iodo-Tyrg]de~-Hi~1-glucagon and of [ [1251]iodo-Tyr10] glucagon are superimposable, and (c) conditions favoring the prior association of lZ5I-labeled glucagon with low affinity binding sites greatly enhance the extent of ligand dissociation, we can say that it is ligand bound to low affinity sites that undergoes rapid dissociation. Whereas related experiments have been interpreted in terms of the association of lZ5Ilabeled glucagon with two interconverting states of the glucagon receptor within a single receptor population (31), the two conclusions actually differ rather little. Thus, both steadystate and non-steady-state measurements identify the complexity of glucagon association with isolated hepatocytes as arising from ligand interactions with two kinetically distinct binding components.
Third, our use of glucagon analogs that have different affinities for association with high and low affinity binding sites has allowed us to assess the relationship between ligandreceptor interactions and the effects of ligand on hepatocyte glycogen metabolism. Whereas quantitative aspects of the correlation between receptor binding and biological activity are imperfect, our results indicate that occupancy of high affinity binding sites accounts for the metabolic effects of glucagon in our system. Thus, notwithstanding (a) the 100fold greater number of low affinity glucagon-binding sites present on canine hepatocytes (relative to the number of high affinity sites) (28), (b) the low but increased occupancy of those sites even through the range of ligand concentrations that cause inhibition of ['4C]fructose incorporation into hepatocyte glycogen, and ( c ) the participation of low affinity sites in the homologous and heterologous inhibition of the hepatocyte adenylyl cyclase system (38), high affinity hormone binding plays an apparently significant role in the expression of glucagon's biological effects on the liver. These conclusions are based on correlations, however, and must await eventual confirmation by use of analogs that interact specifically with one class of hepatocyte glucagon-binding sites or the other (the glucagon equivalent, in some ways, of a-and P-adrenergic agonists).
For most glucagon analogs, the degree of coupling between the occupancy of high affinity binding sites and the stimulation of biological response differs only slightly from one peptide to another; the fractional steady-state occupancy of high affinity sites required to elicit a half-maximal biological response in our assay (determined as [peptide for half-maximal response]/([peptide for half-maximal response] + K D J ) ranges from only 0.006 to 0.05 (mean = 0.02) for peptides 1-8. It is thus the case that glucagon appears to elicit these cellular actions with only a very low degree of receptor occupancy. Such divergence between effector concentrations that stimulate biological activity and those that cause receptor saturation (often attributed to so-called spare receptors) is in fact commonly found and can arise either from cascading events that amplify receptor-mediated intracellular signals or from systems that assess only a portion of the full range of potential effector actions. Whereas glucagon's effects on cellular metabolism have generally been thought to arise from the generation of intracellular CAMP, (a) the quite separate inositol phosphate pathway has recently been implicated in glucagon actions (55); and (b) peptide 9, a glucagon analog that stimulates glycogenolysis at concentrations nearing saturation of cell-surface binding sites, does not enhance in any way the accumulation of CAMP in isolated hepatocytes (38). It thus appears that glucagon may activate two distinct signal transducing systems in hepatic parenchymal cells. Important questions remaining to be answered include (a) whether glycogenolysis, gluconeogenesis, and additional actions of glucagon are mediated by one of the two signal transducing systems or the other (or perhaps by a mixture of the two); ( b ) whether all glucagon analogs will necessarily activate each of the two systems equivalently; and ( c ) whether the two classes of glucagon-binding sites identifed on isolated hepatocytes are linked to different signal transducing systems. Extensions of these studies will provide further clues as to the separate functions of high and low affinity hepatic binding sites for glucagon and the mechanisms by which glucagon-receptor interactions transmit the appropriate transmembrane signal to modulate hepatic fuel metabolism.