The Relation of Predicted Structure to Observed Conformation and Activity of Glucagon Analogs Containing Replacements at Positions 19,22, and 23”

Six new analogs of glucagon have been synthesized containing replacements at positions 19, 22, and 23. They were designed to study the correlation between predicted conformation in the 19-27 segment of the hormone and the conformation calculated from circular dichroism measurements and the observed activa- tion of adenylate cyclase in the liver membrane. The analogs were [VallO]glucagon, [Va122]glucagon, [G1uZ3] glucagon, [Va110,Glu23]glucagon, [Gl~~~,Glu~~]gluca- gon, and [A1a22,A1a23]glucagon. The structures predicted for the 19-27 segment ranged from strongly a helical to weakly B sheet. The observed conformations varied as functions of amino acid composition, solvent, concentration, pH, and temperature but did not correlate well with prediction. There was, however, a cor- relation between predicted structure and activation of adenylate cyclase in rat liver membranes.


To whom correspondence should be addressed The Rockefeller
University, 1230 York Avenue, New York, NY 10021.
the hormone in solution based on their scale of conformational potentials of amino acid residues to form an (Y helix, /3 sheet, or p turn (14). The latter data were derived from proteins of known crystal structure, and the assumption was made that amino acid residues in a small peptide like glucagon might behave in the same way. The analysis showed that glucagon should have a p turn at residues 2-5, a /3 sheet from 5-10, turns at 10-13 and 15-18, and a region from 19 to 27 that is balanced closely between a sheet and a helix (13). It was predicted that substitution of 1 or 2 residues in this region by residues strongly favoring one of these conformations would be enough to change the entire 19-27 segment into a single conformation and that circular dichroism measurements would be sensitive enough to detect such a large change. Thus, they predicted that [GluZ3]glucagon would favor the helical structure and that a double substitution of the normal PheZ2-Valz3 sequence by Glu*'-Gl~~~ would lock the 19-27 region into an a helix while the / 3 sheet structure would be very unlikely statistically. They also proposed that the p conformer would probably be required for receptor binding and that the G~u~~-G~u~~ analog would bind poorly and be inactive. In an effort to test this hypothesis and to understand the role of conformation in the binding of glucagon to its hepatocyte receptors and in the transduction of the hormonal signal, we have synthesized a series of analogs containing amino acid replacements at positions 19, 22, and 23. These were chosen to favor either a helices or /3 sheets in order to be able to relate predicted conformation with structure calculated from circular dichroism measurements. If the applicability of the Chou-Fasman predictive rules to glucagon could be established, we should be able to construct analogs which bind tightly to the hepatocyte receptors and function either as superagonists or antagonists of glucagon action in the liver cell. The latter should reduce the hyperglycemia of diabetes. This kind of inhibitor would differ from an inhibitor of the release of glucagon from the A cell, such as somatostatin, and should be of considerable value in unraveling the details of the mode of action of glucagon. mCi/mmol, were prepared from the radiolabeled amino acids (Du Pont-New England Nuclear) by acylation with di-tert-butyl pyrocarbonate (17). Boc-amino acids were from Peninsula Laboratories, Belmont, CA. The sources and purification of all other reagents for peptide synthesis were the same as described previously (18).
All components for the adenylate cyclase assay were obtained from Sigma. The ATP was prepared synthetically by phosphorylation of adenosine in order to minimize contamination by unknown levels of GTP present in isolated ATP (19). Commercial glucagon (Sigma) was purified on a C,, reverse-phase column before use. Cyclic AMP binding protein assay kits were from Amersham Corp.
Methods-Circular dichroism spectra were recorded on an Aviv 60DS CD spectropolarimeter at controlled temperature in 1-mm cells.
The estimated percent of a helix, 6 sheet, p turn, and random coil structure in the peptides was calculated with their Prosec program. The spectra were measured from 300 to 190 nm at 1-nm intervals, and three scans were averaged. e was calculated using the mean residue weight of each analog. Concentrations were calculated from the extinction coefficients, c~~~ = 8260 for glucagon, in a 50% mixture of 0.1 M sodium phosphate, pH 6.9, and trifluoroethanol (TFE), and confirmed bv the oreviouslv determined specific activity of 3H and "C in the peitides.
Isolation of Liver Plasma Membranes-Partially purified rat liver plasma membranes were isolated by the method of Neville through step 11 (19,20). After resuspension in an equal volume of ice-cold 1 mM NaHC03, membranes were divided into 100-p1 aliquots, which were frozen rapidly and stored in liquid nitrogen. Aliquots were thawed immediately before use and never refrozen. For the adenylate cyclase assay the protein concentration was adjusted to 0.7 mg/ml as determined by a modified Lowry procedure (21).
Adenylate Cyclase Assay-Stock solutions of purified natural glucagon and the synthetic peptides were prepared fresh by stirring 0.2-0.5 mg of peptide with 2 ml of 25 mM Tris'HCl, 1 mg/ml bovine serum albumin buffer, pH 7.75, at 24 "C for 1 h. After Millipore filtration, the peptide concentration was determined from absorbance at 276 nm and specific radioactivity. Less concentrated solutions were obtained by serial dilution in buffer. The assay solution was prepared according to the general procedure of Salomon et al. (19). Incubation media were prepared by mixing 40 pl of assay solution, 40 pl of peptide solution, and 20 pi of membrane suspension. After 5 min at 30"C, the reaction was quenched in a boiling water bath, rapidly cooled in ice, and centrifuged at 12,000 rpm.
The cyclic AMP in the supernatant was quantitated using a commercial assay kit from Amersham Corp., which is based on the competition between unlabeled CAMP and a fixed quantity of added cyclic [8-3H]AMP for a specific high affinity cyclic AMP-binding protein, with removal of unbound nucleotides by adsorption on charcoal (22). The cAMP levels to be measured were adjusted to fall in the 1-5-pmol range. The basal activity of different membrane preparations was in the range of 35-45 pmol/mg of protein/min and the response to excess glucagon gave increases in cAMP of 3-to 4-fold. To compensate for these variations, the responses were converted to percent of maximal activation above the basal level (23).
Receptor Binding Assay-The binding of analogs to receptors of the rat liver membrane was measured by displacement of '251-glucagon according to Wright and Rodbell (24).

RESULTS
The Choice of Analogs-To test the hypothesis that the statistical predictive rules of conformational preference of amino acids derived from crystal structures of proteins would be applicable t o a small flexible peptide like glucagon, we followed the suggestion of Chou and Fasman (13). The idea was t o make sequence replacements in the 19-27 region of glucagon that would be expected to make significant changes in the conformation of that peptide segment as measured by circular dichroism. Table I (26). Thus, the 1st residue was attached to oxymethylphenylacetylnorleucyl aminomethyl resin (OCH2-Pan-am resin) (Fig. 1). The attachment of COOH-terminal Boc-Thr(Bz1) to the latter support was carried out as follows (with quantitative ninhydrin values (27) given in parentheses): (a) 6.15 g of aminomethyl-resin, 0.25 mmol/g, swollen in 100 ml of CH2CI2 and coupled for 16 h with 2.4 eq each of Boc-norleucine and dicyclohexylcarbodiimide in CHzClz (0.0037 mmol/g of free NH2 groups unreacted); ( b ) acetylation of residual amine with AczO/pyridine, 1:1 (0.0001 mmol of NH2/g); ( e ) Boc deprotection with CF3COOH/CH,C1,, 1:l (ninhydrin, 0.25 mmol/g); ( d ) 3-h coupling with 1.5 eq each of Boc-Thr(Bz1)-oxymethylphenylacetic acid and dicyclohexylcarbodiimide in CH2Clz (0.0023 mmol of NH2/g); (e) 3-h second coupling with 0.3 eq preformed symmetric anhydride of Boc-Thr(Bz1)-oxymethylphenylacetic acid in CHzClz (0.0007 mmol of NH2/g); (f) acetylation (0.0001 mmol of NH,/g). The loading was 0.22 mmol/g by picrate titration (28) of a deprotected sample.
Arginine was coupled with dicyclohexylcarbodiimide in CH2Cl2, asparagine and glutamine were coupled a s N -h ydroxybenzotriazole esters made in situ with dicyclohexylcarbodiimide in DMF/CH2Clp (2:1), and all other residues were coupled as preformed symmetric anhydrides in CHzClz except when they were added t o glutamine, in which case the solvent was DMF. Three eq of activated derivative were used for each coupling and, after a wash with tertiary amine, a second coupling was routinely carried out. For the anhydride reactions, the second coupling was in DMF in all cases except as noted. For a few residues, a third coupling was needed. Radiolabeled amino acids were incorporated into all of the peptides to aid in purification. For that purpose [3H]Le~26 and [ 14C]Gly4 were used. Monitoring the Syntheses-All syntheses were monitored for completion of reaction by the quantitative ninhydrin reaction (27). Table I1 shows data for one of the analogs, [Val'9]glucagon, and the results are typical of all the syntheses. In most instances, no difficulty was encountered in reducing residual amino groups below 1 pmol/g of peptide resin. However, to achieve this level for Aoc-Arg(Tos), the coupling time was extended from 2 t o 5 hours. Special difficulty was observed in coupling Boc-glutamine a t positions 24, 20, and 3. For the first two, a third coupling in N-methylpyrollidinone solved the problem. In the early stages of this synthesis, the swelling of the resin was 30% greater in Nmethylpyrollidinone than in DMF or CHPClz and was probably responsible for the improved coupling. Beyond the undecapeptide stage, swelling in DMF and CH2Cl, began to increase significantly.
The deprotected resin was also monitored by the ninhydrin Pa, potential of residue to be in a helix; Po, potential of residue to be in a @ sheet; <Pa>, potential for the segment to be an (

10
to [3H]leucine at residue 26 at the beginning of the synthesis.

A r g ( T o s~-A r g ( T o s~-A l a -G l n -A s p~O B z i ) -P h e -~O B z l~-G~~-T~p~F~~~-
When the ninhydrin-monitoring data were corrected for such [

29
chain termination (Fig. 2), the calculated curve and observed data points agree reasonably well. Such chain termination at glutamine was not observed in previous glucagon syntheses, poly(styrene-1%-divinylbenzene) resin bead. cetylnorleucylaminomethyl linker was omitted. @ represents the co-Ckauage and Purification of the Ambgs-The two-step "low/high HF" procedure (30) was used for cleavage of all peptides. This process simultaneously removed all protecting reaction at each cycle of the synthesis (Fig. 2). The data were groups, including the formyl on tryptophan, and reduced any corrected for the gain in weight due to the growing peptide methionine sulfoxide produced during the synthesis back to chain and are reported as millimoles of NH2/g of polystyrene. methionine. After evaporation of HF and dimethyl sulfide, Significant drops in amine occurred following incorporation the residual scavengers and by-products were extracted with of Boc-glutamine at residues 24 and 20, indicating about 20% ether and then the crude peptides were extracted into 10% loss of growing chains. Since radiolabel data showed clearly aqueous acetic acid. The solution was dialyzed against several that there was no loss of peptide chains from the resin, the changes of 10% acetic acid and lyophilized. decrease in growing chains is attributed to chain termination.
Analytical HPLC of the crude [Val'g]glucagon revealed the This was correlated with a 34% decreased incorporation of peptide peaks shown in Fig. 3. Repetition of the HF cleavage Activity of Synthetic Glucagon Analogs "The final ninhydrin value is divided by the amount of growing peptide chain remaining at each cycle of the synthesis, which is determined by ninhydrin analysis of the deprotected peptide-resin. These values have been corrected for a background of 0.50 pmollg, which is always seen in these assays. Its origin is uncertain, but it is presumed to be due to ninhydrin-positive impurities in the resin. [ G l~~~, G l u~~] g l u c a g o n was also cleaved by the "high HF" procedure (HF-p-cresol-p-thiocresol, 90:5:5), 5 ml for 0.2 g of resin, 1 h, 0 "C. In this case, the HPLC pattern of the crude peptide was quite different. The main peak eluting near 21 min (Fig. 4) was almost absent, and a new major peak appeared at 16 min which was considered t o be the Met(0) derivative produced by air oxidation during the synthesis. To support this conclusion, reduction with mercaptoethanol at pH 8 converted the 16-min product to a peptide eluting at 21 min that was indistinguishable from the product from the low-high procedure containing underivatized methionine. The isolated synthetic peptides were essentially homogeneous based on a highly loaded analytical HPLC (>95%) (Figs. 3 and 4), and they yielded satisfactory amino acid analyses (Table 111).
Synthetic yields were relatively good. Based on amino acid analysis for unique residues incorporated near the end of the synthesis (glycine and histidine), the assembly of the fully protected peptide resins were approximately 60 -+ 5%. The data for the Val" analog indicated histidine at 60% and glycine at 64% of the loading of the internal norleucine standard. These results were corroborated by radioactive measurements on peptide resin hydrolysates. The amount of 14C incorporated late in the synthesis was only 65% of norleucine, whereas the tritium label introduced early was recovered in theoretical amount. The combined yield of the HF cleavage   and dialysis steps was 70 f 5%, measured by analysis of glycine, histidine, and 14C in hydrolysates of the peptide resins and the cleaved and dialyzed peptides. Analytical HPLC indicated that the crude unpurified cleaved and dialyzed peptide mixture contained 65% of the desired target peptide, giving an overall synthetic yield of 60% X 70% X 65% = 27%. The actual isolated yield of homogeneous analog obtained after conservative cuts of the HPLC fraction was about 10%; yields of the other analogs ranged up to 35%. Conformation of the Analogs-The circular dichroism (CD) of solutions of glucagon and the synthetic analogs was measured as a function of solvent, concentration, pH, temperature, and time of standing in solution before measurement. The proportions of a helix, p sheet, p turn, and random coil were calculated as previously described (31) using the Prosec program, which is based on the conformational standards of Chang et al. (32). In a few instances, the results were confirmed by applying other programs to the data. Most of the measurements on glucagon were made on purified natural material, but in the several cases tested, synthetic glucagon was indistinguishable. All of the peptides were purified on preparative HPLC columns, lyophilized and stored in the freezer before dissolving in the 50% TFE/buffer and diluting to final composition and concentration. It has been shown (8-11) that glucagon aggregates in concentrated solution, and this effect is illustrated in Table IV. In 0.01 N HCl, the proportion of helix increased with concentration and in 0.01 M sodium phosphate, pH 9.2, containing 50% TFE, helicity decreased and p turns increased with concentration. The p turn values as determined by the Chang et al. (32) analysis can now be questioned, however. Since it has been established recently that there are at least three CD curves for various types of @ turns (33)(34)(35), the values used as the basic spectra by Chang et al. (32) do not serve adequately in the deconvolution of CD curves. Under either condition the effect was minimal below 0.1 or 0.2 mg/ml and this concentration range was selected for further experiments. In vivo glucagon is found in the crystalline, helical structure in the storage granules, but it circulates in very dilute solution in the 10-9-10"0 M range, and is thought to exist as monomers with little secondary structure.

Effect of concentration on the Conformation of glucagon
It is also well known that peptide conformation is strongly influenced by the presence of organic solvents in aqueous solutions and these conditions are thought to mimic the cell membrane where the hormone functions. Fig. 5 shows the mean residue ellipticity between 190 and 260 nm of [Glu22,Glu23]glucagon in dilute solution at pH 6.9 (phosphate) in the presence of 10, 50, and 86% TFE. The shift in the shape of the CD curve is a clear indication of increased helicity and the calculations indicate changes from 5 to 50% in helix and 49 to 9% in B sheet as the organic solvent increases.
These changes were rapid and reversible.
The effect of pH between 2 and 9.2 in dilute solutions of glucagon over the range of 0-86% TFE is illustrated in Fig.  6. The response was very similar at pH 2 and 9.2, with a sharp and parallel increase for a helix and decrease for / 3 sheet at either pH extreme. At pH 6.9, the changes were qualitatively similar but more gradual. Addition of salt up to 0.2 M NaCl had no significant effect on the observed ellipticity. No significant differences were found between lyophilized samples measured within a few minutes after dissolution or after Synthetic Glucagon Analogs 17309

TABLE V Effect of t i m e on conformation of dilute solutions of glucagon
Commercial glucagon was freed of an impurity by reverse-phase chromatography on a CIS silica column with a 25-45% CH&N gradient. The CHsCN was evaporated, and the aqueous solution was lyophilized. Samples of the solid were dissolved in 0.01 N HCl containing 0-90% TFE (final concentration, 0.14 mg/ml). CD readings at 22 "C were made within 30 min. After storage for 7 days at 4 "C, the samules were warmed to 22 "C and read again.  (Table  V). Conformational  ]. However, it started with higher helicity than glucagon and remained relatively high (47%) even at 86% TFE. It began with somewhat less structure, not more, than glucagon. The two analogs predicted to be strongly helical and low or negligible in / 3 structure, [Ala22,Ala23]-and [ G 1~~~, G 1~~~] g l~~a g o n , did not show such a conformation in this system. They were actually less helical than glucagon at all concentrations of TFE and showed more ( 3 structure than the natural hormone at all TFE levels except at 10%.
Corresponding data on our analogs at acidic pH in 0.01 N HCl are shown in Table VII. In this case, the solubilities were sufficient to allow CD data to be measured at 0% TFE. In all analogs the helicity was very low in the aqueous solvent, but again rose with TFE concentration to values even higher than at pH 6.9. The p sheet was somewhat lower in 10% TFE than at pH 6.9 and dropped much more rapidly with rising TFE. The major difference seen at the two pH values were with analogs containing the additional glutamic acid residues at positions 22 or 23. At pH 6.9, [GluZ3]glucagon went from 12% helix in 10% TFE to only 26% helix in 86% TFE whereas, at pH 2, this analog rose from 0% helix at 0% TFE to 17% at 10% TFE and 71% at 86% TFE. [Va1'9,G1~23]-and [ G 1~~~, G 1~~~] g l~~a g o n behaved similarly. The acid medium suppresses the ionization of the y carboxyl which, in turn, may promote the helical structure.
The helix of glucagon was also stabilized at lower temperatures (Fig. 7), although the effect was not dramatic. For example, at 5% TFE in pH 6.9 phosphate buffer, the apparent helix increased from 7 to 13% on going from 22 to -2 "C, an 86% increase. However, at 20% TFE, the change was from 48 to 52%, or only 8% increase. The proportion of sheet decreased in a reciprocal manner and the calculated proportion of disordered, or random coil, peptide chain was essentially constant at all temperatures and solvent compositions. The temperature effect also applied in 0.01 N HC1. The curves for cy helix and p sheet at 4 "C are shown in Figs. 8 and 9 for glucagon and three of the synthetic position 22 and 23 analogs. The conformational increment between 22 and 4 "C was significant for glucagon, [ValZ2]glucagon, and [Ala22,Ala23]gl~cagon, but was largest for [ G 1~~~, G 1~~~] g l~~a g o n .
In the experiments described here, a calculated value of 10% helix would mean that only 3 residues out of the 29 residues of glucagon would be in a helix if all molecules were in the same conformation, and that is not enough to stabilize such a structure. If the 19-27 segment of every molecule of glucagon changed from p sheet, or a random coil, to an a helix as a result of a solvent, temperature, or structural change, we would expect an increase from 0 to 9/29 = 31% helix. In 0.01 N HC1 or pH 6.9 phosphate glucagon was 8% helix at 4 "C.  TFE  TFE  TFE  TFE  TFE  TFE  TFE  TFE  TFE  TFE    Addition of 13% TFE at 4 "C was needed (Fig. 8) to obtain the CD value equivalent to a full helix in the 19-27 segment, although the part of the molecule responsible for the helicity was not actually identified. This solvent may or may not correspond to the average polarity of a globular protein. The larger helicity detected at higher TFE means that other regions must also assume this structure.
Hormonal Actiuity of the Analogs-The purified glucagon analogs were assayed for biological activity by measuring the CAMP resulting from activation of adenylate cyclase in purified rat liver membranes. The response curves are shown in Fig. 10, and the results are in Table VIII. It can be seen that [Val'g]glucagon gave a maximum response equivalent to that caused by excess glucagon in the production of CAMP, but more peptide was required. To reach a half-maximal response required 26 times as much peptide. The relative activity, therefore, was only 3.9%. been shown to be a full agonist of 10% relative potency (31).
[Va122]glucagon was 2.2% active but not a full agonist. None of the other analogs gave a full response in the concentrations that could be tested but [Va119,G1~23]-, [GluZ3]-, and [Ala22,Ala23]glucagon were very weak partial agonists. [G1~~~,G1u~~]glucagon gave no detectable response even at 2 X M and was less than 0.001% active. Since [Gluz3]glucagon was such a weak partial agonist (0.1% relative potency), it could be demonstrated to be a competitive inhibitor of glucagon. A standard response curve of glucagon could be obtained in the presence of 5 x M analog, where the analog alone gave no significant response. It was approximately parallel to the curve with glucagon alone but was shifted to the right. Thus 5.8 X lo-' M glucagon was required for half-maximal response compared with 6.9 X lo-' M in the absence o f the inhibitor. The relative binding constants were approximately 91. Since this analog was a weak partial agonist, it could not be used in the concentration required for complete inhibition of glucagon and was not an effective antagonist.
The inhibition of glucagon by [Ala22,Ala23]glucagon was demonstrated by a different protocol (Fig. 11)   maximal response was mixed in a series of tubes with increasing concentrations of analog. Total inhibition could be achieved at about M and 50% inhibition was reached at 1.6 X M. The inhibition index, I/A~o, was 160. Membrane binding by [Ala'2,Ala23]glucagon was also measured by the '251-glucagon competitive displacement assay of Wright and Rodbell (24). Binding relative to glucagon was 0.6%.

DISCUSSION
The six new anaiogs of glucagon described here were prepared by solid-phase methods and purified to homogeneity. Aside from some chain termination at glutamine there were no special difficulties in syntheses. The composition and structure of the peptides were established, and we think they are of good quality and are suitable for the conformation and activity measurements undertaken.
The hypothesis we have been examining is that residue replacements in the 19-27 region of glucagon could be made which would cause predictable conformational changes observable by circular dichroism measurements. The predictions were based on the Chou-Fasman probability factors developed from examination of protein x-ray crystal structures (14). By increasing the a helix or ,# sheet potential of the replacements, several analogs were synthesized which were expected to cover a range of probabilities of being in one conformation or another and, in the extremes, were expected to cause the entire 9-residue 19-27 segment to change from a 0 sheet to an a helix or vice versa. Such a change should be easily detected by circular dichroism measurements on the whole 29-residue peptide. Percent changes from 0% a and 51% p to 31% a and 20% 0 were expected.
From the CD data, we find no clear correlation with the predicted a helical and p sheet potentials of the amino acids in this series of small 29-residue peptides. The predictions, of course, were derived from x-ray crystal structure data on globular proteins where the molecules are much larger and more rigid and have an inside and an outside which allows some residues to be exposed to solvent and some to be excluded from solvent. Even in solution such proteins maintain some structure. Glucagon has a largely helical structure in the crystal (7), but in solution it has much more conformational freedom and few if any buried residues. The structural flexi-Synthetic Glucagon Analogs bility has been seen in NMR studies (12). It seems quite reasonable that the rules that govern the probabilities of individual amino acid residues being in a preferred conformation in a protein may not all apply to small peptides of this type. It has been shown previously that small polypeptides, e.g. preproparathyroid hormone (30 amino acid residues), can be predicted to have nearly equal probabilities of having two conformations (36). Under different environmental conditions, both predicted conformations have been verified by CD studies. In addition, based on the ellipticity at 210 nm, it has been concluded (37,38) that the helicity of synthetic [Ly~~'*~~,Glu~~]glucagon was somewhat larger than that of glucagon itself in pH 9.2 phosphate. Fasman has reviewed this subject recently (39). The positions of glucagon studied here were residues 19,22, and 23. When Valz3 or PheZ2 and Valz3 were replaced by a glutamic acid residue the potential for a helix in the 19-27 segment was markedly increased. However, these changes not only affected the predicted conformational potential but also the local charge, which may have a profound effect on structure. It was for this reason that the AlaZ2-Alaz3 analog was also synthesized. In it the high helix potential was retained without a charge difference, although the hydrophobicity was less than the natural. The [Val'g,Glu23]glucagon derivative was selected because a Val replacement of Ala at 19 partly counteracted the high a potential increase due to replacement of Val23 by GluZ3. Nonetheless, this analog was predicted to favor the helical conformation.
The second objective of this study was to try to correlate predicted and observed conformation in dilute solution with biological response. It was suggested (13) that a @ structure in the COOH-terminal region of glucagon will be the active conformation while others (7,9) have suggested that an a helical structure, particularly an amphipathic helix, will be the conformation that binds to the receptor and be responsible for the hormonal effect. Our data show that the changes in the 19-27 segment that were predicted to contain more @ structure retained more ability to activate adenylate cyclase and those that were predicted to be more a helical in this region were poorly bound and of very low potency. The measured conformational data in dilute solution, however, do not show this correlation.
We suggest, therefore, that these conformational predictions do not apply to dilute aqueous solutions of free glucagon or its derivatives, but that they may apply to these same peptides when they are at the membrane receptor. Those that can be induced to assume a /3 sheet may be able to bind and transduce the hormonal signal, whereas those that would be expected to favor an a helix in the 19-27 region when in the more shielded environment of a macromolecule may not be induced to assume a shape suitable for binding to the receptor. NMR studies of the hormone-receptor complex may eventually clarify this question and allow the design and synthesis of more effective superagonists and antagonists of glucagon.