Conformational Properties of Cyanogen Bromide-cleaved Glucagon*

The circular dichroism of the cyanogen bromide peptide of glucagon, measured in dilute aqueous solution, in chloroethanol, and in urea is similar to that of glucagon. However, in more concentrated aqueous solutions the cyanogen bromide peptide does not rapidly aggregate to structures of higher helical content as does glucagon. Thus this property of glucagon is dependent on the presence of the two COOH-terminal amino acids. Intramolecular quenching of tryptophan fluorescence by tyrosinate is observed in the cyanogen bromide peptide. From this quenching the approximate distance between aromatic residues is calculated and it is concluded that the peptide is in a compact conformation. unmodified glucagon interpreted as the fixed angle between the aromatic residues. Nuclear magnetic studies MHz the chemical of several the protons water random suggest that the conformational change between a randomly coiled peptide and the glucagon monomer in several sections the

The circular dichroism of the cyanogen bromide peptide of glucagon, measured in dilute aqueous solution, in chloroethanol, and in urea is similar to that of glucagon. However, in more concentrated aqueous solutions the cyanogen bromide peptide does not rapidly aggregate to structures of higher helical content as does glucagon. Thus this property of glucagon is dependent on the presence of the two COOHterminal amino acids.
Intramolecular quenching of tryptophan fluorescence by tyrosinate is observed in the cyanogen bromide peptide. From this quenching the approximate distance between aromatic residues is calculated and it is concluded that the peptide is in a compact conformation. The lack of quenching in unmodified glucagon is interpreted as due to the fixed angle between the aromatic residues.
Nuclear magnetic resonance studies at 250 MHz reveal small differences in the chemical shift of several of the protons between the cyanogen bromide peptide in water and the random coil spectrum. These differences suggest that the conformational change between a randomly coiled peptide and the glucagon monomer in water involves several sections of the molecule.
The frictional coefficient calculated from sedimentation velocity results along with the fluorescence and nuclear magnetic resonance data suggests that the cyanogen bromide peptide, like glucagon, folds into a compact, globular structure in dilute aqueous solution.
Glucagon is a polypeptide hormone of 29 amino acids which is known to exist in various conformations.
A form containing CY helix occurs in the crystal (l), in concentrated aqueous solutions (2-4), and in 2-chloroethanol (4,5). The formation of structures of higher helical content in concentrated aqueous solution is accompanied by a self-association process (6, 7). Nuclear magnetic resonance studies have shown that the COOHterminal portion of the glucagon molecule is involved in the association of glucagon in acidic solution (8). The last two * This work was supported by Grant A6044 from the National Research Council of Canada and the Banting Research Foundation. amino acids in the COOH-terminal portion can be cleaved by reaction of the sole methionine at position 27 with cyanogen bromide (9). We have compared the conformational properties of the resulting 27 amino acid peptide with those of glucagon.

EXPERIMENTAL PROCEDURE
Materials-The amino acid analysis of crystalline glucagon (Sigma, Lot SOC-2900) agreed to within 5% of the reported value (10, 11) for each amino acid. Urea (Matheson, Coleman and Bell, reagent grade) was purified by passing an aqueous solution over a mixed bed deionizing resin (Bio-Rad resin AG501-X8(D)) followed by crystallization in the cold. Tris buffer was made from the free base (Sigma, Trizma grade) and HCI. 2Chloroethanol (Eastman) was distilled over anhydrous potassium carbonate and the fraction distilling at 127-128" was collected and stored in the dark at 4". Aqueous solutions of chloroethanol were made up immediately before use. Urea-da was purchased from Merck, Sharp and Dohme.
Ultra pure grade (Schwarz-Mann) guanidine hydrochloride was used.
Amino Acid Analyses-The analyses were determined with a Beckman 120 C analyzer.
Samples were hydrolyzed at 110" for 22 hours in sealed, evacuated ampoules containing 6 N HCl. The solution was then taken to dryness under vacuum on a rotary evaporator; the residue dissolved in water and a portion taken for determination of homoserine. The pH of this portion was adjusted to 9 with a dilute solution of sodium hydroxide and after 15 min at room temperature again evaporated to dryness. Homoserine, produced from homoserine lactone, was eluted between serine and glutamic acid and the area of the peak was compared with that of a standard (Sigma, nn-homoserine, Lot 10813-2250).
Amino acid analyses are calculated to give the expected total number of amino acids. Ammonia was determined from analysis of the fraction not treated with base. The ammonia peak overlapped with that of homoserine lactone.
For the estimation of asparagine and glutamine in the carboxypeptidase-hydrolyzed material, the amino acid analyzer was operated at 35" instead of the usual 55". Studies-These studies were performed at 20" with a Spinco model E ultracentrifuge using sapphire windows. A double sector, capillary-type, synthetic boundary cell was used for the sedimsntation velocity runs with the schlieren optical system and a speed of 60,000 rpm. The peak positions were calculated by the method of second moments.
The sedimentation equilibrium runs were performed with a 12.mm cell at a speed of 39,500 rpm, and a peptide concentration of 0.2% making use of the combined Rayleigh interference and schlieren optical systems (13). Results calculated solely from the schlieren system by the method of Lamm (14) gave very good agreement with the method using the combined optics.
These procedures circumvent the use of a synthetic boundary run with the undialyzable peptide.
The partial specific volume of the cyanogen bromide peptide was estimated to be 0.71 ml per g from the amino acid composition by the method of Cohn and Edsall (15), treating homoserine as threonine.
Attainment of equilibrium was verified by comparison of photographs taken at intervals of several hours.
Fluorescence Emission Spectra-These spectra were measured on Mark I spectrofluorometer (Farrand Optical Co.) at an ambient temperature of 26" in 0.1 M KCl, the pH being adjusted with KOH.
The sample was excited at 270 nm and the emission spectra were recorded.
The excitation wave length increases the tyrosyl t'o tryptophenyl emission ratio and is about at the isobestic point for the ionization of N-acetyltyrosinamide so that the fraction of light absorbed by tryptophan does not change with pH (16). Emission spectra were measured with l-cm2 cells for solutions having optical densities of less than 0.05 at 280 nm and are proportional to concentration. Proton Magnetic Resonance Spectra-These spectra were obtained with a modified Varian high resolution spectrometer which operates at a frequency of 250 MHz.
The proton signal of the DzO solvent was used as the internal lock. In separate experiments the position of this solvent line was measured with respect to 2,2-dimethyl-2-silapentane-5-sulfonate and spectra are reported with respect to this latter reference.
All spectra were taken with a single scan and repeated at least once.
Concentration of Protein Solutions-The concentration of protein solutions was determined from amino acid analysis using an internal standard of norleucine.
Reaction of Glucagon with Cyarwgen Bromide-This reaction was done by two methods.
In the first method a 0.9 y. solution of glucagon was reacted with a 35 molar excess of cyanogen bromide in 0.1 N HCI containing 5% dioxane for 24 hours at room temperature.
Dioxane inhibits aggregation in acidic solution (17). At the end of the reaction the material containing some precipitate was lyophilized.
The reaction was also carried out in 70% formic acid (18), USing 0.1 y0 glucagon with a 200 molar excess of cyanogen bromide. The reaction was allowed to proceed for 24 hours at room temperature at the end of which time the clear solution was diluted with 3 volumes of cold water and lyophilized.
Separation of Cyanogen Bromide Peptides-Separation of the cyanogen bromide peptides was done on a column (1.5 x 100 cm) packed with Sephadex G-10 (Lot 3411) which had been equilibrated with an alcohol-water solvent containing 2 volumes of ethanol per volume of water. Before application of the peptide several void volumes of solvent were passed through the column until the eluent was transparent in the ultraviolet.
The hydrostatic pressure was adjusted to give a flow rate of 10 ml per hour. Glucagon is eluted in the void volume at about 60 ml. Because of its moderate solubility in the water-alcohol mixture this is a convenient procedure to separate glucagon from low molecular weight substances.
The lyophilized products of the cyanogen bromide reaction were partially dissolved in the water-alcohol mixture.
A few drops of 0.1 N NaOH were added to increase solubility of the 70% formic acid product and then immediately placed on the Sephadex column.
In both cases only one peak absorbing at 278 nm was eluted at the void volume.
The cyanogen bromide peptide was lyophilized after dilution with 10 volumes of water. Ninhydrin analysis revealed a second peak at about twice the void volume whose amino acid analysis showed only aspartic acid and threonine in equimolar amounts.
Carboxypeptidase Hydrolysis-One milliliter of a suspension containing 4 mg of diisopropyl phosphorofluoridate-treated carboxypeptidase A (Nutritional Biochemicals) was centrifuged and the precipitate washed twice with 1 ml of water; 0.1 ml of 2 M NHJHCOg was added, and the resulting solution was clarified by centrifugation.
Ten microliters of this solution were added to 0.1 ml of a 0.5% solution of peptide in 4 M urea, lOA M Tris buffer, pH 8. The solution stood for 3 hours at room temperature after which time the reaction was stopped by acidification, and the material was immediately used for amino acid analysis. Under these conditions all of the amino acids up to aspartate (residue 21)) a poor substrate for carboxypeptidase (19), are completely released (Table II).
Blank runs, containing only the carboxypeptidase, released only traces of amino acids.

AND DISCUSSION
Treatment of glucagon with cyanogen bromide will convert the methionine residue at position 27 to homoserine and liberate the COOH-terminal dipeptide, asparaginylthreonine.
We have separated the two products of the cyanogen bromide cleavage by gel filtration.
This procedure would not be expected to separate the 27 amino acid cyanogen bromide peptide from unreacted glucagon.
We conclude that the peptides produced by cleavage in 0.1 N HCl contain only a few percent of unreacted glucagon and that the reaction in 70% formic acid is virtually quantitative because of the loss of the sole methionine residue, the recovery of 1 residue of homoserine, the loss of approximat'ely 1 residue of aspartic acid and threonine (Table I), the recovery of these two amino acids in another fraction of the column eluent and the enhanced solubility of the final product.
The product of the reaction in 70 y0 formic acid was further analyzed by digestion of the COOH-terminal amino acids with carboxypeptidase (Table II).
The amino acids released from glucagon correspond to the 8 COOH-terminal amino acids and show that the digestion proceeded up to the slowly hydrolyzed amino acid residue 21, aspartic acid. The remaining peptide containing residues 1 to 21 was isolated by Sephadex chromatography and its identity confirmed by amino acid analysis.
The amino acids released from the cyanogen bromide peptide also correspond to digestion up to residue 21 with the liberation of 6 amino acids; homoserine replacing methionine. The absence of unreacted glucagon in the cyanogen bromide peptide is most clearly demonstrated by the release of only 0.02 residue of threonine, the COOH-   Glucagon also exhibits cotton effects with magnitudes of about 40 degree cm2 per decimole in the spectral region of the aromatic amino acid chromophores which are enhanced in more concentrated solutions (7). These cotton effects were not detected for the cyanogen bromide peptide whose mean molar residue ellipticity was found to be 0 & 15 degree cm2 per decimole at peptide concentrations of 0.03 and 0.05% in Tris buffer. This suggests that the aromatic residues of glucagon have a fixed orientation, with respect to the asymmetric centers, which is lost in the cyanogen bromide peptide.
b Asparagine and glutamine cochromatograph.
terminal amino acid of glucagon, which along with asparagine is absent in the cyanogen bromide peptide.
The small amount of threonine released could have originated from the dipeptide asparaginylthreonine which was not completely separated on the Sephadex column, from autolysis of carboxypeptidase A and from a maximum of about 2% unreacted glucagon.
Unless otherwise stated, the cyanogen bromide peptide prepared in 70% formic acid was used for further study.
The lowered ability of the cyanogen bromide peptide to form aggregates of higher helical content may also be correlated with the loss in orientation of the aromatic chromaphores. The importance of the aromatic side chains to the aggregation process in glucagon was shown by difference spectra (2) and by circular dichroism (7). The aromatic residues have not been chemically altered in the production of the cyanogen bromide peptide since its ultraviolet absorption spectra (from 215 to 330 nm at 0.03% peptide in Tris buffer at pH 8) and fluorescence emission (from 280 to 400 nm at 5 PM peptide in water) spectra are coincident with that of glucagon. The circular dichroism of the cyanogen bromide peptide shows The cyanogen bromide peptide has a higher solubility at pH 8 a small negative ellipticity in the spectral region of the amide than does glucagon. This can be correlated with its lowered chromophore which is diminished in the presence of urea and tendency to aggregate and to form structures of higher helix enhanced by chloroethanol (Fig. 1). This is very similar to the content. We cannot rule out the possibility that some hydrolybehavior of glucagon (4, 7). However, the enhanced negative sis of the side chain amide bonds occurred during preparation of ellipticity exhibited by more concentrated aqueous solutions of the cyanogen bromide peptide. However, if it were an imporglucagon does not occur with the cyanogen bromide peptide, the tant factor we should expect to observe more hydrolysis in the In addition to rapid formation of aggregates of higher helical content, glucagon has also been shown to slowly form fl structures at high peptide concentrations, both in acidic (3, 17) and in basic (20) aqueous solutions.
A 0.5% solution of the cyanogen bromide peptide in Tris buffer, pH 8, on standing overnight at room temperature forms a gel-like substance. The gel has been identified as a p structure on the basis of its infrared spectra when suspended in DzO. The observed peak of the amide I band at 1618 cm-' in DzO is typical of a p structure and lies between 1613 cm-l and 1626 cm-r, the absorption maximum found for the gel formed from acidic (5) a,nd basic (20) DSO solutions.
Sedimentation velocity studies confirm that the cyanogen bromide peptide has a diminished ability to rapidly aggregate in 0.1 r~f NaCl, 10e2 M Tris buffer, pH 8, compared with that of glucagon (6). The observed a.&, = 0.77 S is virtually independent of peptide concentration between 0.1 and 0.4%. The lowered ability of the cyanogen bromide peptide to aggregate is concomitant with a lowered tendency to form structures of higher helical content in water.
These two phenomenon are linked equilibria (7). Molecular weights calculated from the sedimentation equilibrium runs in 0.1 M NaCl; 1OP M Tris buffer, pH 8, gave values which varied with distance from the center of the rotor.
The average slope, calculated by least squares, gave a molecular weight of 4500 while near the meniscus and bottom of the cell the molecular weight is 3500 and 6600, respectively. These values are for the z average molecular weight.
The centrifuge studies are done in the presence of 0.1 M NaCl to minimize charge effects. The addition of 0.1 M NaCl causes little change in the circular dichroism of the peptide, although its magnitude seems to decrease about lo%, only slightly more than the experimental error. The molecular weight calculated in the region of the cell near the meniscus (3500) is close to that expected on the basis of the amino acid composition of the monomer (3238).
The cause for the apparent increase in molecular weight near the bottom of the centrifuge cell is not known but could arise from the formation of aggregates in the form of a p structure.
We have shown that this type of aggregation occurs with the cyanogen bromide peptide in concentrated solutions over a period of many hours.
These conditions are fulfilled during the sedimentation equilibrium run where 24 hours are needed for the apparent attainment of equilibrium and the concentration at the bottom of the cell is greater than the initial concentration.
This effect does not occur in the sedimentation velocity runs where only an hour is required for its completion and the concentration at the solution side of the observed boundary is never greater than the initial concentration.
It is possible that some of the poor reproducibility between runs in the sedimentation equilibrium studies of glucagon (6) can be explained by the formation of fl structure aggregates. This could also explain why the association constants calculated by Swann and Hammes (6) from this data are higher than those calculated by Gratzer and Beaven (7) from circular dichroism studies. This effect due to the formation of p structure is more important for the cyanogen bromide peptide than for glucagon since the lower solubility of glucagon causes some of it to precipitate before fl structure is formed.
The observed sedimentation coefficient of the cyanogen bromide peptide reflects its hydrodynamic properties thus giving information about the shape of the particle in solution (21). From the observed sedimentation coefficient, .&,, = 0.77 S, we calculated the ratio of the observed frictional coefficient to the minimum possible frictional coefficient for an unhydrated sphere to be 1.10. This low value is similar to those of other globular proteins and is indicative of a compact spherical molecule.
Long range radiationless transfer of singlet excitation energy has been used to measure the distance between the transition moments of two interacting chromophores.
The efficiency of energy transfer is determined by the distance between the chromophores and also by the relative orientation of their transition dipoles (22). Singlet excitation transfer in proteins can be measured by the quenching of tryptophan fluorescence in alkaline pH due to radiationless transfer to tyrosinate.
No such quenching was observed with glucagon (16) suggesting either that the single tryptophan was further than 16 A from either of the two tyrosines or that the relative orientations of the transition dipoles of donor and acceptor were such as to preclude energy transfer.
The fluorescence emission spectra for glucagon and for the cyanogen bromide peptide, at neutral pH, were identical.
Tryptophan fluorescence was monitored at 350 nm while fluorescence due largely to tyrosine was recorded at 300 nm. Above pH 11.5 a small quenching by hydroxide occurs.
The pH change does not cause a conformational change in glucagon as shown by the invariance of the peptide cotton effects between pH 2 and 11.3. Our results with glucagon are in very good agreement with those of Edelhoch and Lippoldt (16) and show very little, if any, quenching of the tryptophan fluorescence by the ionization of tyrosine (Fig. 2). However, the cyanogen bromide peptide shows a decrease of 20 Y0 in the tryptophan fluorescence at higher pH and this decrease is proportional to the decrease in tyrosine fluorescence (Fig. 3) demonstrating that it is caused by energy transfer to tyrosinate.
The spectral red shift of the absorption spectrum of tyrosine upon ionization, to the region of tryptophan fluorescence emission, allows for intramolecular singlet excitation energy transfer between these two residues in basic solution. The NHs+ group of lysine at residue 12 also ionizes near this pH region, but it would be expected to quench the fluorescence of the two tyrosines at residues 10 and 13 much more than the tryptophan at residue 25.
Eisinger et al. (23) have tabulated the Forster distance, Ro, for the singlet excitation transfer between pairs of aromatic amino acids. For energy transfer from tryptophan to tyrosinate this distance is about 10 A, assuming a typical quantum yield.
The Forster distance is quite insensitive to the choice of quantum yield varying as the sixth root of this quantity.
From the efficiency of energy transfer we can calculate the distance between the transition moments of the interacting dipoles in the cyanogen bromide peptide to be 12 A, assuming random orientation of the side chains (23). This distance is comparable to the calculated radius of 9.7 A for a spherical model of the cyanogen bromide peptide (molecular weight 3238; 0 = 0.71 ml per g). The sole tryptophan residue in this peptide occurs at position 25 and the two tyrosines at positions 10 and 13. Using the method of Brant,Miller,and Flory (24) we can calculate that the distance between residues 13 and 25 in a randomly coiled polypeptide  chain would be 29 A which is much larger than our calculated value of 12 A. We have used their values of 5 for the ratio <r2 > /nB, (25) and 3.8 A for l,, where T is the end to end distance of a polypeptide containing n virtual bonds of average lengths I,.
Although singlet excitation transfer is precluded for a polypeptide chain with the average end to end distance of a random coil, the distance distribution function must also be considered for a PH PH in 0.10 M KCl.
FIG . 4 (right). The pH dependence of the tryptophan and tyrosine fluorescence of the cyanogen bromide peptide of glucagon in 6.0 M guanidine hydrochloride.
flexible system (26). The observed energy transfer in the cyanogen bromide peptide could come from a randomly coiled chain only if the distribution of end to end distances were broad. The absence of tyrosinate quenching of tryptophan fluorescence in the presence of guanidine hydrochloride (Fig. 4) shows that energy transfer is precluded when the peptide is unfolded. Guanidine hydrochloride causes only relatively minor changes in the emission spectra of tyrosine and tryptophan (16) and thus should not greatly affect the Forster distance, Ro. This experiment demonstrates the conformational dependence of the observed energy transfer.
The distance between aromatic residues is too great in the random conformation to allow energy transfer, while the distance between the residues in water, 12 A, is comparable to the radius of the cyanogen bromide peptide as a compact sphere, 9.7 A.
The increased quenching at alkaline pH of the cyanogen bromide peptide compared with glucagon would result either from a folding of the molecule on cleavage of the two COOHterminal amino acids or from a change in orientation of the transition dipoles of the aromatic amino acids. The former explanation seems unlikely and is contrary to the circular dichroism evidence which shows identical intrinsic cotton effects for the two peptides and to hydrodynamic evidence indicating compact structures for the peptides.
On the other hand, a change in orientation of the aromatic residues is supported by the loss of the extrinsic cotton effects on cleavage of glucagon with cyanogen bromide.
It thus appears that monomeric glucagon is in the form of a compact molecule and that no energy transfer occurs between tryptophan and tyrosine because of the fixed orientation of their transition dipoles. Glucagon thus possesses at least two properties in common wit,h larger globular proteins, its compact structure and the fixed position of its aromatic side chains.
The proton magnetic resonance spectra calculated for glucagon by the procedure of McDonald and Phillips (27) agrees fairly well with that measured in 8 M urea (Figs. 5 and 6). The measured spectra shows somewhat greater resolution of peaks because of the small size of glucagon and the improved instrumentation at the slightly higher fields of 250 instead of 220 MHz.
There are however some differences between the calculated and observed spectra (Table III).
The methyl peaks of threonine, FIG. 5. High field proton magnetic resonance spectra at 250 MHz of the cyanogen bromide peptide (2 mM, DzO, pH reading 6.9) ; glucagon (4 mM, 8 M urea-d4, pH reading 7.2) and a schematic representation of the glucagon random coil.
leucine, and valine and the 6CHz peak of arginine are shifted to higher field in the observed spectra. Peak 1 is especially shifted and is also seen in the spectra of Pate1 (8). Most of the peaks of the glucagon spectra in 8 M urea, however, do agree very well Ath the calculated value and their assignments are confirmed by measurement of the peak areas. Because of the overlap of Peaks 8 and 9 the measured total number of protons for these two peaks agrees well with the calculated value while each individual peak does not. In addition the measured area of Peak 11 is somewhat high because of overlap with other resonance lines.
There are several discernible differences between the spectra of glucagon in 8 M urea and that of the cyanogen bromide peptide in water.
The peaks of the cyanogen bromide peptide are  6. Low field proton magnetic resonance spectra at 250 MHz of the cyanogen bromide peptide (2 mM, DzO, pH 6.9) ; glucagon (4 mM, 8 M urea-d+ pH 7.2) and a schematic representation of the glucagon random coil. slightly broader and more complex having several overlapping peaks, indicating protein folding; especially for Peak B in Fig. 6.
In addition, there is clear indication of a change in chemical shift for Peaks 3, 8, and 11 between glucagon in 8 M urea and the cyanogen bromide peptide in water.
This shift is less for other peaks and Peak 9 and part of Peak 3 do not shift at all indicating that the shift is not caused by an unspecific soIvent effect. The two remaining threonine residues in the cyanogen bromide peptide produce resonance lines at 1.06 and 1.13 ppm. The resonance at 1.13 ppm occurs at the same position as Peak 3 in 8 M urea while that at 1.06 ppm occurs in a region where no signal is found in the 8 M urea spectrum.
This suggests that one of the threonine residues is still exposed to the solvent while the other is in a folded region of the peptide.
The amino acids which are implicated as being in the structured regions of the peptide are a threonine residue at position 5 or 7 and the valine and arginine residues at positions 23, 17, and 18. Thus the conformational by guest on March 21, 2020 http://www.jbc.org/ Downloaded from