`
`The Conformation of Glucagon
`
`T.L. BLUNDELL
`
`A. Introduction
`
`The biological activity of glucagon is mediated through binding, with high affinity
`and specificity, to a membrane receptor, implying extensive and well-defined inter(cid:173)
`molecular interactions (see Chap. 13). However, in dilute aqueous solutions gluca(cid:173)
`gon has little defined secondary structure and almost certainly exists as a popula(cid:173)
`tion of conformers in equilibrium. The formation of the receptor-hormone com(cid:173)
`plex must involve either selection of one conformer from the population or induc(cid:173)
`tion of a conformer as the interaction takes place. Whatever the mechanism, the
`definition of the receptor-bound conformer, as well as the nature of the hormone(cid:173)
`receptor interactions must be a primary objective in the understanding of the biol(cid:173)
`ogy of glucagon and the design of glucagon agonists and inhibitors (BLUNDELL
`1979; BLUNDELL and HUMBEL 1980). Unfortunately, the receptor has yet to be de(cid:173)
`fined biochemically and so direct study of the receptor-hormone complex is not
`possible at present. Instead we must examine the conformation of glucagon in
`aqueous solution, in crystals, in lipid micelles and other environments in order to
`establish the nature of the conformational dependence on intermolecular inter(cid:173)
`actions. Here, I first describe recent developments, especially in the use of X-ray
`diffraction and proton nuclear magnetic resonance (NMR) spectroscopy which
`have allowed description of the conformation in great detail under varied con(cid:173)
`ditions. I then discuss the relevance of these conformations to the molecular biol(cid:173)
`ogy of glucagon.
`
`B. The Crystal Structure
`I. Crystals
`
`It is convenient to begin a description of the conformation of glucagon by review(cid:173)
`ing the crystal structure analysis, for in the crystals the glucagon molecule has less
`flexibility, as it is stabilised by numerous intermolecular interactions, some of
`which may be maintained in other environments. The relatively ordered state al(cid:173)
`lows a medium resolution analysis using the techniques of protein X-ray crystal(cid:173)
`lography.
`KING (1959) showed that rhombic dodecahedral crystals of glucagon contain
`12 molecules packed with cubic symmetry (space group P2 13). Preliminary results
`(KING 1965; BLANCHARD and KING 1966) suggested a packing of a-helical rods and
`
`P. J. Lefèbvre (ed.), Glucagon I
`© Springer-Verlag Berlin Heidelberg 1983
`
`Page 1
`
`NPS EX. 2050
`CFAD v. NPS
`IPR2015-01093
`
`
`
`38
`
`T. L. BLUNDELL
`
`b
`
`c
`
`Phe-22
`Phe-22
`Phe-22
`Fig.la-c. The electron density at ~ 3 A resolution for Phe-22. Orthogonal views (a, b) of
`the refined fit of the side chain to the density; rotation of 50° (c) around the p-y bond (viewed
`in the same direction as a) leaves the side chain in the density, indicating rotational disorder
`in the crystals
`
`this was detailed by SASAKI et al. (1975) who carried out a medium resolution X-ray
`analysis using the methods of isomorphous replacement and anomalous scattering.
`Crystals of glucagon are most easily obtained at high pH ( '" 9.2) where the glu(cid:173)
`cagon is more soluble than at physiological pH. However, SASAKI et al. (1975)
`showed that these crystals undergo a phase change on lowering the pH to around
`6.5, involving a retention of the cubic symmetry, but a shortening of the cubic cell
`parameter from 47.9 to 47.1 A. Similar crystals are obtained by careful crystallisa(cid:173)
`tion at pH 6.5. A further crystal form of the same symmetry, but of a different cell
`dimension (48.7 A) is found at lower pH ('" 3) (DOCKERILL 1978). These ob(cid:173)
`servations are a reflection of the flexibility of the glucagon conformation and espe(cid:173)
`cially its dependence on the pH which is also observed in solution. Various metal
`ion complexes cause similar changes in the crystals, giving rise to complications in
`the use of the method of isomorphous replacement.
`Even the best crystals are disordered relative to those of most globular proteins.
`The nominal resolution is '" 3 A, which implies that the general conformation can
`be defined although the details are often unclear. For instance, the precise orien(cid:173)
`tation of the carbonyls of the peptide groups and the side chains are not defined.
`Figures 1 and 2 show some typical side chain electron densities. Figure I shows
`that the phenyl ring ofPhe-22 may be rotated by ± 50° around the f3-y bond from
`the optimal position and still remain in the electron density. As the resolution of
`the crystals is only '" 3 A, this implies that even in the better ordered parts of the
`crystal structure there .is considerable disorder. Other groups such as Leu-26 have
`a poorly defined conformation and the crystal probably contains a population of
`each of the three conformers with a staggered arrangement around the f3-y bond.
`Other parts of the molecule are even more disordered and these include residues
`1-4 at the NHrterminus and 28 and 29 at the COOH-terminus as well as the side
`chains of Asp-9, Lys-12, and Arg-18.
`
`Page 2
`
`
`
`The Conformation of Glucagon
`
`39
`
`Tyr-10
`
`Tyr-10
`
`Tyr-13
`
`Trp-25
`
`Trp-25
`
`Leu-26
`Met-27
`Fig. 2. The electron densities for a selection of side chains at ~ 3 A resolution with the side
`chain positions from the least-squares refinement. Note that the ends of some side chains,
`e.g. Met-27 C' have no density, indicating complete disorder, while others such as Leu-26
`are not unambiguously defined, indicating several possible conformations
`
`Page 3
`
`
`
`40
`
`T. L. BLUNDELL
`
`3
`
`Fig. 3 a, b. Two orthogonal views (a, b) of the conformation of the glucagon molecule (pro(cid:173)
`tomer) in the crystals. Note the amphipathic nature of the helical conformer with two hy(cid:173)
`drophobic regions involving Phe-6, Tyr-lO, Tyr-13, and Leu-14 towards the NHz terminus
`and Ala-19, Phe-22, Val-23, Trp-25, Leu-26, and Met-27 at the COOH terminus
`
`The model defined by X-ray analysis and described in this chapter is an average
`of the conformers present. In fact, the poor resolution of the crystals poses prob(cid:173)
`lems in the refinement of the molecular structure using least-squares techniques.
`The model presented by SASAKI et al. (1975) was a best fit to the electron density
`at pH 6.5, obtained. by the method of isomorphous replacement with anomalous
`scattering using an optical comparator (see BLUNDELL and JOHNSON 1976 for a re(cid:173)
`view). The first set of coordinates deposited with the Brookhaven Protein Structure
`
`Page 4
`
`
`
`The Conformation of Glucagon
`
`41
`
`Data Bank were refined using the methods considered optimal in 1975: cycles of
`real space refinement and calculation of electron density. M6re recently,
`J. MOREIRA, U. TICKLE, and T.L. BLUNDELL (1981) unpublished work) have shown
`that restrained least-squares refinement gives improved coordinates and these are
`used in this discussion.
`
`II. Proto mer Conformation
`
`Figure 3 illustrates the conformation of the glucagon molecule (a protomer) in the
`cubic crystals. Table 1 gives the main chain torsion angles (cP, 1jJ, and co) which
`show that residues 6-28 are in an approximately a-helical conformation. Table 2
`gives the lengths and angles subtended at the peptide oxygen and hydrogen atoms
`of possible hydrogen bonds. The variations in lengths and angles are partly a re(cid:173)
`flection of the conformational disorder although the existence of a 310 helix in res(cid:173)
`idues 5-11 is probably real. At Gly-4 the chain becomes poorly defined, indicating
`
`Table 1. The torsion angles defining the main chain
`conformation of the glucagon molecule from the
`medium resolution X-ray analysis of crystals at
`pH~6.5
`
`Residue
`
`if>
`
`'P
`
`His-l
`Ser-2
`Gln-3
`Gly-4
`Thr-5
`Phe-6
`Thr-7
`Ser-8
`Asp-9
`Tyr-lO
`Ser-11
`Lys-12
`Tyr-13
`Leu-14
`Asp-15
`Ser-16
`Arg-17
`Arg-18
`Ala-19
`Gln-20
`Asp-21
`Phe-22
`Val-23
`Gln-24
`Trp-25
`Leu-26
`Met-27
`Asn-28
`Thr-29
`
`-
`
`-
`
`90
`-
`-109
`114
`-162
`53
`31
`39
`63
`71
`52
`73
`46
`68
`76
`62
`85
`49
`79
`66
`80
`77
`72
`76
`67
`51
`76
`35
`174
`
`-
`
`-
`
`-
`
`-
`
`-
`
`-
`
`117
`6
`69
`137
`8
`29
`49
`61
`14
`56
`28
`53
`39
`22
`42
`30
`46
`42
`44
`31
`26
`22
`16
`- 48
`55
`54
`62
`66
`-169
`
`-
`
`w
`
`-173
`-177
`-177
`-177
`178
`178
`-179
`-179
`174
`177
`178
`178
`-179
`180
`-179
`180
`179
`-176
`-180
`179
`179
`177
`178
`178
`-178
`-176
`-179
`-180
`
`Page 5
`
`
`
`42
`
`T. L. BLUNDELL
`
`Table 2. Possible hydrogen bonds between peptide N-H and C= 0 groups
`
`Donor
`
`Acceptor
`
`N ... O
`
`N ... O=O
`
`N ... H ... O
`
`H ... O
`
`8
`9
`10
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`29
`
`5
`6
`6
`7
`7
`8
`9
`10
`12
`12
`13
`14
`150
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`2.6
`2.6
`3.4
`2.6
`2.9
`2.9
`2.6
`2.7
`2.5
`3.3
`3.3
`3.2
`3.0
`3.2
`3.0
`3.5
`3.0
`3.4
`3.2
`2.6
`2.8
`3.2
`3.5
`
`129
`142
`150
`123
`147
`148
`136
`165
`123
`167
`152
`141
`154
`134
`156
`128
`147
`131
`144
`152
`129
`132
`148
`
`140
`122
`146
`122
`138
`138
`143
`155
`147
`155
`138
`162
`149
`150
`153
`127
`133
`115
`143
`174
`151
`133
`148
`
`1.8
`2.0
`2.5
`1.9
`2.1
`2.2
`1.7
`1.8
`1.6
`2.4
`2.5
`2.2
`2.1
`2.3
`2.0
`2.3
`2.3
`2.8
`2.3
`1.6
`1.9
`2.4
`2.6
`
`that the glycine endows the chain with flexibility; consequently the placing of the
`NH 2-tenninus is somewhat arbitrary.
`The a-helical confonnation has few stabilising interactions other than the hy(cid:173)
`drogen bonds of the main chain. However, certain hydroxyl groups, in particular
`those of Thr-29 and Thr-5, hydrogen bond to carbonyl groups, and the accessibil(cid:173)
`ity of other residues - Tyr-lO, Tyr-13, Leu-14, Arg-17, Asp-21, Phe-22, and Gln-24
`- to solvent is significantly decreased through van der Waals' contacts in the helical
`confonner (R. WEST and B. GELLATLY 1980, unpublished work). Table 3 gives the
`accessibility of side chains to a water molecule calculated by the methods of LEE
`and RICHARDS (1971) and FINNEY (1978). The helical conformer brings together the
`hydrophobic groups in two "patches" comprising Phe-6, Tyr-l0, Tyr-13, and Leu-
`14 towards the NHrterminus, and Ala-19, Phe-22, Val-23, Trp-25, Leu-26, and
`Met-27 towards the COOH-terminus. These hydrophobic patches are responsible
`for the intennolecular interactions which stabilise the crystal form.
`
`III. Trimer Conformation
`
`The cubic arrangement of mainly a-helical glucagon molecules involves three-fold
`rotation and two-fold screw axes of symmetry leading to a continuous oligomeric
`structure. This may be considered in tenns of a series oftrimers which are not mu(cid:173)
`tually exclusive and each of which contains a perfect three-fold axis. One such
`trimer (trimer 1) is shown in Fig.4. The three molecules are related by applying
`
`Page 6
`
`
`
`The Conformation of Glucagon
`
`43
`
`Fig.4. A trimeric arrangement (trimer I) found in glucagon crystals formed from heterol(cid:173)
`ogous interactions between the hydrophobic patches at the NH2 and COOH termini. The
`central hydrophilic region contains charged groups such as Asp-21, Arg-17, and Arg-18
`
`twice the symmetry operation: x,y,z-+ 1-y, Y2 +z, Y2 -x, where x, y, and z are
`fractional coordinates. In the trimer, parts of the hydrophobic region towards the
`NH 2-terminus involving Phe-6, Tyr-tO, Tyr-13, and Leu-14 are brought into con(cid:173)
`tact with the hydrophobic region at the COOH-terminus involving Phe-22, Trp-25,
`and Leu-26 (Fig. 5). The centre of the "triangular" trimer is hydrophilic and in(cid:173)
`cludes charged groups Asp-9, Arg-17, Arg-18, and Asp-21, most of which appear
`to be relatively disordered. There is evidence for a hydrogen bond interaction be(cid:173)
`tween the Tyr-tO side chain hydroxyl and the Arg-18 main chain. Solvent access(cid:173)
`ibilities are listed in Table 3 and side chains in van der Waals' contact in Table 4.
`The angle between the helices is '" 103°.
`A second trimer (trimer 2) is shown in Fig. 6, in which molecules are related by
`applying twice the symmetry operation x,y,z-+z,x,y. The contact region involving
`only the COOH-terminal hydrophobic region - residues Ala-19, Phe-22, Val-23,
`
`Page 7
`
`
`
`44
`
`T. L. BLUNDELL
`
`Lys-12
`Fig.5. The intermolecular, mainly hydrophobic interactions which stabilise trimer I, shown
`in Fig. 4
`
`Leu-26, and Met-27 - is shown in Fig. 7. There is a hydrogen bond between Gln-20
`side chain amide and Ser-16 side chain hydroxyl groups. Table 3 gives the change
`in accessibility of these residues to solvent compared with the isolated helical mol(cid:173)
`ecule. The angle between the helices in the trimer is 76°.
`In view of the possibility of either of these trimers existing in solution (see Sect.(cid:173)
`C.II) it is relevant to consider factors which may affect their stability. Clearly the
`major interactions are nonpolar and hydrophobic. Trimer 1 involves a decrease of
`accessibility to solvent of 18.4% of the groups compared with an isolated helical
`conformer. The decrease of accessibility for trimer 2 is 14%. If the free energy gain
`by hydrophobic interactions is dependent only on the surface area made inacces(cid:173)
`sible to solvent, then trimer 1 would be most stabilised. However, trimer 1 involves
`the necessary formation of helix between residues 6 and 26 involving unfavourable
`entropy terms, whereas the interactions between molecules in trimer 2 require helix
`formation only between residues 16 and 27 (see Sect. C.II for further discussion).
`Trimer 1 would appear to be much more pH dependent in its formation. In par(cid:173)
`ticular, protonation of Asp-9 or Asp-21 might disturb the balance of charge and
`may be expected to de stabilise this trimer in acidic conditions. Perhaps of greater
`importance would be the ionisation in alkaline solutions (above pH 10) of tyrosyl
`hydroxyl groups, especially that of Tyr-1O which is buried in trimer 1. These ob(cid:173)
`servations may be the cause of difficulty in forming crystals outside the range, say
`pH 3 to pH 11, when there would be few species of the correct charge.
`The dependence of the crystal cell dimensions on the pH in the range 7.5-8.5
`probably arises from rather different interactions within the crystal. The NH 2-ter(cid:173)
`minus occupies a complicated polar region including two COOH termini, in addi(cid:173)
`tion to Asp-9 and Asp-15 of adjacent molecules. A change of pH from ~9.0 to
`,.... 6.0 will lead to protonation of the N a and the Nt of His-I, leading to small rear(cid:173)
`rangements. A similar transition occurs in the crystals of adenylate kinase, in which
`
`Page 8
`
`
`
`The Confonnation of Glucagon
`
`45
`
`Table 3. The accessibility to water (assumed to be a sphere of 1.4 A radius) of amino acid
`residues in different conformations. The values for random coil have been calculated from
`those of a residue (X) in a tripeptide Ala-X-Ala in LEE and RICHARDS (1971). The helical
`conformer and the two trimers (1 and 2) are those defined by X-ray analysis of crystals at
`pH 6.5 before restrained refinement
`
`Residue
`
`Accessibility a
`
`LR
`F
`LR
`LR
`Trimer
`Helical
`Random Helical
`conformer conformer 1
`coil
`
`F
`Trimer
`1
`
`LR
`Trimer
`2
`
`F
`Trimer
`2
`
`His-l
`Ser-2
`Gln-3
`Gly-4
`Thr-5
`Phe-6
`Thr-7
`Ser-8
`Asp-9
`Tyr-lO
`Ser-ll
`Lys-12
`Tyr-13
`Leu-14
`Asp-15
`Ser-16
`Arg-17
`Arg-18
`Ala-19
`Gln-20
`Asp-21
`Phe-22
`Val-23
`Gln-24
`Trp-25
`Leu-26
`Met-27
`Asn-28
`Thr-29
`Totals
`
`186
`108
`162
`72
`123
`165
`123
`108
`133
`183
`108
`173
`183
`135
`133
`108
`205
`205
`93
`162
`133
`165
`125
`162
`202
`135
`162
`133
`153
`4,268
`
`198
`121
`160
`37
`117
`152
`107
`78
`103
`97 b
`81
`160
`122b
`61 b
`115
`74
`148 b
`196
`52
`130
`76 b
`100 b
`93
`96 b
`209
`140
`124
`96
`163
`3,406
`
`123
`73
`102
`43
`87
`119
`84
`67
`77
`92
`70
`107
`108
`79
`85
`55
`108
`128
`45
`92
`71
`93
`75
`71
`146
`98
`99
`72
`104
`2,573
`
`198
`121
`160
`37
`117
`99 c
`104
`78
`91
`4 c
`62
`160
`80 c
`12 c
`115
`74
`112 c
`123 c
`52
`130
`40c
`50 c
`93
`76
`98 c
`108 c
`124
`96
`163
`2,778
`
`123
`73
`102
`43
`87
`102
`84
`67
`74
`36
`67
`107
`92
`68
`85
`55
`108
`114
`45
`92
`71
`64
`75
`71
`105
`84
`99
`72
`104
`2,369
`
`198
`121
`160
`37
`117
`152
`107
`78
`103
`97
`81
`160
`122
`61
`115
`44 c
`148
`151 c
`13 c
`96 c
`76
`34 c
`14 c
`96
`209
`76 C
`30 c
`93
`140
`2,929
`
`123
`73
`102
`43
`87
`119
`84
`67
`77
`92
`70
`107
`108
`79
`85
`55
`108
`117
`30
`90
`77
`64
`45
`70
`146
`70
`58
`72
`101
`2,419
`
`a Accessibilities are calculated by A. WEST and B. GELLATLY (1979, unpublished work) using
`the methods of LEE and RICHARDS (1971) indicated by LR and by FINNEY (1978) indicated
`by F. The methods give very different results, indicating different assumptions in the
`computational approach
`b Significant accessibility decrease for side chain on formation of the helical conformer
`c Significant accessibility decrease on trimer formation
`
`an imidazole and a carboxylate group are also close together. In both glucagon and
`adenylate kinase, a similar transition is effected by metal ions. For example, bind-
`ing of Ag+ and Hg2+ between the imidazole of His-l and the sulphur of Met-27
`stabilises the lower pH form; the metal cations appear to be playing the role of a
`proton.
`
`Page 9
`
`
`
`46
`
`T. L. BLUNDELL
`
`Table 4. Residues and atoms in van der WaaI's contact in trimers 1 and 2
`(Fig. 4-7)
`
`Residue
`
`Atom(s)
`
`Residue
`
`Atom(s)
`
`Trimer 1
`
`Trimer 2
`
`Leu-26
`Trp-25
`
`Phe-22
`Arg-18
`Met-27
`Leu-26
`Val-23
`
`Gln-20
`
`COl
`CP, CY, C(2
`CY, COl, N"
`C(2, C'H2
`C'H2
`CO2
`0
`CP
`CO2
`CYI
`CYI
`CYI
`O"/N"
`
`Phe-6
`Tyr-lO
`Tyr-13
`Asp-9
`Phe-6
`Tyr-lO
`Tyr-lO
`Phe-22
`Leu-26
`Val-23
`Ala-19
`Phe-22
`Ser-16
`
`C'2, C(
`cn
`CIl,CY, CO2
`0202
`0
`CO2
`OH
`C", COl
`CO2
`CY2
`CIl
`COl
`OY
`
`C. The Solution Structure
`I. Monomer
`
`Although some earlier experiments favoured a globular structure for the monomer,
`most recent experimental data indicate a predominantly flexible structure with
`little defined secondary structure. Circular dichroism (PANIJPAN and GRATZER
`1974; SRERE and BROOKS 1969) and fluorescence (EDELHOCH and LIPPOLDT 1969)
`are consistent with 10%-15% of helix.
`Recent proton NMR studies on dilute aqueous solutions (BOESCH et al. 1978;
`WAGMAN et al. 1980; WAGMAN 1981) are consistent with a largely unstructured and
`flexible chain. This is indicated by the narrow line-widths and a single set of chemi(cid:173)
`cal shifts which, with the exception of the aromatic tyrosines, are independent of
`temperature and phosphate concentration. Those chemical shifts which are pH de(cid:173)
`pendent are attributable to protonation (WAGMAN 1981). A mainly flexible and
`nonspherical structure is also indicated by viscosity measurements (WAGMAN
`1981), although not by the earlier results of EPAND (1971); this intrinsic viscosity
`is found to be insensitive to denaturating agents such as urea (WAGMAN 1981).
`Most chemical shifts are not affected by 8 M urea and are similar to those expected
`for random coil, the exceptions being those of Val-23 methyl resonances, Val-23
`HiJ and Leu-26 HY. These have been attributed by BOESCH et al. (1978) to a proxim(cid:173)
`ity between the side chain ofVal-23 and the aromatic ring of Trp-25 in about 20%
`of the population of conformers, and a similar interaction occurs in a peptide frag(cid:173)
`ment corresponding to residues 22-26; these residues cannot have a helical confor(cid:173)
`mation. There may also be some structure in a proportion of the conformers in(cid:173)
`volving Tyr -10 and Tyr -13 which unfolds with increasing temperature (WAGMAN
`1981). In summary, it appears that in dilute aqueous solutions, the glucagon
`monomer exists either as a mainly flexible structure with a small section of ordered
`
`Page 10
`
`
`
`The Conformation of Glucagon
`
`47
`
`Fig. 6. A trimeric arrangement (trimer 2) found in glucagon crystals involving hydrophobic
`contacts between equivalent regions towards the COOH terminus (Phe-22, Val-23, Leu-26,
`and Met-27). These interactions may be maintained in the absence of helix in residues 6-18,
`and such a partly helical trimer probably exists in concentrated aqueous solutions
`
`secondary structure, or as a series of conformers in equilibrium, none of which
`dominates and some of which may involve structure in residues Tyr-lO to Tyr-13
`and Phe-22 to Leu-26.
`
`n. Trimers
`The concentration dependence of most spectroscopic probes - circular dichroism,
`optical rotary dispersion, optical detection of magnetic resonance and NMR - in-
`
`Page 11
`
`
`
`48
`
`T. L. BLUNDELL
`
`Met-27 /he-22
`(3)
`(2)
`
`Val-23
`(3)
`
`Leu-26
`(3)
`
`Phe-22
`(3)
`
`\Phe-22
`\(1)
`
`Fig. 7. The hydrophobic interactions which stabilise trimer 2 shown in Fig. 6
`
`dicate that glucagon self-associates with an accompanying change in secondary
`structure. Although the nature and structure of the oligomers has been the subject
`of much debate, it now appears that at least in the pH ranges 2-4.5 and 9-11, a
`well-defined trimer is formed which is partly helical and involves intermolecular in(cid:173)
`teractions similar to those of the crystalline trimer 2 (WAGMAN et al. 1980; W AG(cid:173)
`MAN 1981).
`Circular dichroism and optical rotary dispersion (SRERE and BROOKS 1969;
`GRATZER and BEAVEN 1969) indicate that conformers of 35% a-helical content are
`induced in concentrated solutions. Fluorescence studies of rhodamine 6 G dye
`bound to glucagon at pH 10.6 (FORMISANO et al. 1978 a) and optical detection of
`magnetic resonance (Ross et al. 1976, 1977) favour preferential formation of sec(cid:173)
`ondary structure in the COOH terminal region close to the tryptophan (Trp-25).
`Proton NMR studies in concentrated solutions at high pH also suggest that aggre(cid:173)
`gation involves COOH terminal residues and a detailed assignment of resonance
`(WAGMAN 1981; WAGMAN et al. 1980) has shown that the largest chemical shifts
`and line-width broadening occur in COOH terminal residues (22-29).
`BLANCHARD and KING (1966) postulated trimer formation in solution on the
`basis of their early X-ray studies of crystals, and this was confirmed by GRATZER
`et al. (1972) who cross-linked glucagon solutions using dimethyl suberimidate. Al(cid:173)
`though SWANN and HAMMES (1969) had favoured a monomer-dimer-hexamer
`equilibrium on the basis of sedimentation studies, the concentration dependence
`of the chemical shifts in the proton NMR can be fully accounted for by a two-state
`model involving a monomer-trimer equilibrium (WAGMAN et al. 1980).
`The constants for trimerisation (K~) at pH 10.6 for glucagon in D 20 with
`1.08( ± 0.42) x 106 M- 2
`0.2 M
`sodium phosphate
`are
`at 30°C,
`and
`5.86( ± 1.07) x 104 M- 2 at 50 °C (WAGMAN 1981). Similar values are obtained
`from the concentration dependence of the circular dichroism at different temper-
`
`Page 12
`
`
`
`The Conformation of Glucagon
`
`49
`
`atures (FORMISANO et al. 1978 b). Analysis of the temperature dependence of the
`association constants (WAGMAN 1981; FORMISANO et al. 1978 b) and calorimetry
`(JOHNSON et al. 1979) show that glucagon association is characterised by large neg(cid:173)
`ative values of both L1Jr> and L1So which are temperature dependent, decreasing
`with increasing temperature. L1 Cp ° has a large negative value, indicating hydro(cid:173)
`phobic interactions (FORMISANO et al. 1978b). These data are consistent with a
`dominant negative entropy of a coil-helix transition on association which is greater
`than the positive entropy change resulting from hydrophobic interactions (W AG(cid:173)
`MAN 1981). The negative enthalpy change makes the overall free energy change fa(cid:173)
`vourable. The formation of trimers is favoured by phosphate (FORMISANO et al.
`1978 b; WAGMAN 1981), is inhibited by urea (WAGMAN 1981) and is strongly pH
`dependent in the pH range 9.5-10.5 where Ka decreases by a factor of25. However,
`the association is similar at pH 2 and 10.6 (WAGMAN 1981).
`These data are consistent with a trimeric structure similar to trimer 2 involving
`hydrophobic interactions but with a well-defined structure only in the COOH ter(cid:173)
`minal region. This is supported by nuclear Overhauser enhancement evidence, in(cid:173)
`dicating no interactions between residues widely separated in the glucagon se(cid:173)
`quence of the kind found in trimer 1. To define further the interactions in the
`trimer, WAGMAN (1981) has calculated the secondary shifts mainly resulting from
`ring current effects and the transverse relaxation times which might result from in(cid:173)
`teractions observed in the crystal trimer 2, and in closely related structures ob(cid:173)
`tained by changing the conformation to minimise the energy. In general there is
`a reasonable correlation of the observed shifts with those calculated for trimer 2,
`especially for Ala-19, Val-23, Met-27, and Thr-29. Although some observed shifts
`such as those for the Leu-26 methyl carbons are not in good agreement with those
`calculated on the basis of trimer 2, an improved agreement is obtained by small ro(cid:173)
`tations around bonds such as the f3-y bond ofPhe-22 (WAGMAN 1981). However,
`these rotations may not represent a real difference between solution and crystal
`structure as the crystals are clearly poorly ordered (resolution ~ 3 A) and rotations
`of up to ± 50° may be made consistent with the electron density (see Fig. I).
`Although the evidence is clear that trimer I does not exist in acidic or alkaline
`solutions, a small fraction of such conformers may exist in neutral conditions. The
`increase in oligomer formation on decreasing the pH corresponds to the pK of the
`tyro sines and the same tyrosine involvement in self-association is observed at lower
`pH. However, WAGMAN et al. (1980) find no evidence for an abnormal pK for the
`tyro sines which would be expected of trimer I.
`In summary, in concentrated solutions there is an equilibrium between a largely
`unstructured polypeptide and a partly helical trimer. The helical structure is sta(cid:173)
`bilised by hydrophobic interactions and possibly some hydrogen bonds, such as
`that between Ser-16 and Gln-20. On increasing concentrations - and most easily
`at neutral solutions - these trimers become involved in further intermolecular in(cid:173)
`teractions leading to an increased IX-helical content and crystallisation.
`
`III. Fibrils
`
`On standing in acid solution, the viscosity of glucagon increases and a birefringent
`gel is formed (GRATZER et al. 1968; BEAVEN et al. 1969). Sedimentation studies in-
`
`Page 13
`
`
`
`50
`
`T. L. BLUNDELL
`
`dicate the formation of large aggregates. On further standing or warming, a pre(cid:173)
`cipitate appears which has the appearance of long fibrils in the electron micro(cid:173)
`scope. Infrared spectra of the gel, of solid films and of the precipitated material
`show that in all these states the glucagon is in the form of antiparallel f3-pleated
`sheets. The ability of glucagon to form both a-helical and f3-pleated sheet confor(cid:173)
`mers is reflected in the prediction of secondary structure by CHOU and F ASMAN
`(1975), who find that the sequence favours both a-helix and f3-pleated sheet forma(cid:173)
`tion, and suggest that the conformation is delicately balanced between these two
`conformations.
`
`D. Conformation of Micelle-Bound Glucagon
`
`GRATZER et al. (1968) showed that glucagon assumes a more ordered structure -
`possibly 90% a-helical- in chloroethanol. As glucagon is amphipathic in a helical
`conformation (BLANCHARD and KING 1966; SCHIFFER and EDMUNDSEN 1967), hy(cid:173)
`drophobic surfaces favour helix formation. Indeed, fluorescence and circular
`dichroism studies (SCHNEIDER and EDELHOCH 1972; EPAND et al. 1977) show that
`detergents and surfactant micelles bind glucagon with a concomitant formation of
`ordered secondary structure. Wu and YANG (1980) have estimated that in 25 mM
`sodium dodecylsulphate below pH 4, glucagon may be 50% a-helical. They have
`shown that sodium decylsulphate is equally good at inducing secondary structure,
`but that dodecylammonium chloride and dodecylheptaoxyethylene are less effec(cid:173)
`tive. They have also shown that NH2 terminal fragments of glucagon are not af(cid:173)
`fected, whereas structure is induced in COOH terminal fragments such as Ala-19
`to Thr-29, the part of the sequence that has the highest propensity for helical struc(cid:173)
`ture as shown by CHOU and FASMAN (1975). It is possible that this region may also
`be a-helical when the complete molecule binds to a surfactant.
`More recently, WUTHRICH and co-workers (BOESCH et al. 1982) have under(cid:173)
`taken a complete conformational analysis of glucagon bound to micelles using high
`resolution NMR techniques. They show that one glucagon molecule binds to
`40 detergent molecules (perdeuterated dodecylphosphocholine) with a well-de(cid:173)
`fined and extended conformation. Electron paramagnetic resonance (EPR) and
`NMR studies indicate that the molecule is parallel to the micelle surface (BROWN
`et al. 1981). In a beautifully designed study, they have combined the use of proton(cid:173)
`proton Overhauser enhancements and distance geometry algorithms using upper
`limits for selected proton-proton distances to define the conformation of the glu(cid:173)
`cagon molecules bound to the micelle (BRAUN et al. 1981). They present three struc(cid:173)
`tures (see Fig. 8) which meet the distance criteria for the segment Ala-19 to Met-27
`given in Table 5. Although certain distances are inconsistent with the existence of
`the helical conformer found in the crystals, it is possible to build a helical confor(cid:173)
`mer with these interatomic distances by moving the side chain of Met-27, which
`is restricted by intermolecular interactions in trimer 2, and by rotating some side
`chains through small angles which are not precluded at 3.0 A resolution by the X(cid:173)
`ray analysis. The details of the conformer bound to the micelle will be further de(cid:173)
`fined as the NMR analysis proceeds.
`
`Page 14
`
`
`
`The Conformation of Glucagon
`
`51
`
`Fig.8a-c. Computer drawings of the residues Phe-22, Trp-25, and Leu-26 in three struc(cid:173)
`tures (a,b,c) predicted from NMR and geometry algorithms. BRAUN et al. (1981)
`
`E. Conformation and Storage Granules
`Glucagon is stored in granules which are usually amorphous in character. How(cid:173)
`ever, the glucagon granules ofteleosts are crystalline rhombic dodecahedra (LANGE
`and KLEIN 1974; LANGE 1976, 1979; LANGE and KOBAYASHI 1980). Figure 9 is a re(cid:173)
`production of the results of LANGE and KLEIN using tilting stage microscopy while
`Fig. 10 ist a drawing of a rhombic dodecahedral crystal of glucagon in a general
`view and along [100], [111], and [110]. The central view of the electron micrograph
`appears to be along [100] while the other views [111] or equivalent axes (the angle
`of tilt should be 54.7°) and [110] or equivalent axes (the angle of tilt should be 45°).
`The angles are compressed in the thin sections. These electron micrographs show
`that the glucagon granule has 12 faces and that it is a rhombic dodecahedron (LAN(cid:173)
`GE and KLEIN 1974). LANGE (1976, 1979) has shown by optical transforms that the
`cell dimension varies between 41 and 48 A. Values smaller than that observed for
`porcine glucagon crystals bathed in their solvent of crystallisation at pH 6.5 are ex(cid:173)
`pected as a result of dehydration necessary for electron microscopy. These crystal(cid:173)
`line granules almost certainly have a structure similar to that of the crystals studied
`by X-ray analysis and contain glucagon trimers.
`Xanthydrol, a reagent specific for tryptophan residues, produces an unusual
`blue-grey colour in A-cell granules (BUSSOLATI et al. 1971) as distinct from the nor-
`
`Page 15
`
`
`
`52
`
`T. L. BLUNDELL
`
`Table 5. Upper bounds for the distances used in the distance geometry
`algorithm as obtained from truncated driven Overhauser effect mea(cid:173)
`surements. All observed Overhauser effects were attributed to proton(cid:173)
`proton distances of 5.0 A or less. Where individual protons could not be
`resolved in the 1 H NMR spectrum, e.g. the ring protons of Ph-22 or
`«51 CH3 and «52 CH3 of Leu-26, a proton-proton distance sufficiently large
`to include all indistinguishable protons was used (BRAUN et al. 1981)
`
`Observed Overhauser effects
`
`Between ...
`
`. .. and ...
`
`Maximum
`distance
`(A)
`
`Ala-19
`Phe-22
`
`flCH 3
`ring
`
`Val-23
`
`IXCH
`
`Val-23
`
`Y1 CH3
`
`Val-23
`
`Y2 CH3
`
`Val-23
`
`IXCHa
`
`Val-23
`
`flCHa
`
`Trp-25 C2H
`Trp-25 C4H
`
`Trp-25 C5H
`Trp-25 C6H
`61,62 CH3
`Leu-26
`
`ring
`Phe-22
`IXCH
`Val-23
`Val-23
`Y1 CH3
`Trp-25 C5H
`Trp-25
`C7H
`61, «52 CH3
`Leu-26
`yCH
`Leu-26
`«51,«52 CH3
`Leu-26
`«51, «52 CH3
`Leu-26
`Met-27
`8CH3
`yCH
`Leu-26
`61, «52 CH3
`Leu