Designing insulin for diabetes therapy by

protein engineering
Jens Brange, Guy G. Dodson and Bing Xiao
Novo Research Institute, Bagsvaerd, Denmark and University of York, York, UK

Although injected insulin has been used successfully in the control of

diabetes for 60 years, even the modern therapeutic preparations still have
deficiencies. Protein engineering has given biochemists a new tool with
which to modify the properties of insulin. This review concentrates on
recent protein-engineering experiments on insulin.

Current Opinion in Structural Biology 1991, 1:934-940

Introduction small blood vessels with age, a strong imperative exists

The discovery in 1921 by Banting and Best that insulin
could prevent death from diabetes by simple injections
was a medical miracle and a landmark in the therapeutic
use of hormones. With millions of diabetics in the world
needing insulin to stay alive, the hormone is of outstand-
ing medical and biochemical significance. It is no sur-
prise, therefore, that insulin was the first protein to be
sequenced and synthesized, and is now being manufac-
tured using recombinant methods.
Insulin's metabolic effects are diverse; its most widely
known action is the stimulation of the transport of sugar
from the blood into cells. The structural and chemical
events responsible for sugar transport and the subse-
quent metabolic activities in cells are initiated by the
binding of the insulin molecule to a cell-surface receptor
[ 1]. The potency of insulin can thus be determined in the
whole animal by measuring its effect on glucose level in
the blood, by measuring the binding and/or metabolic
effects of insulin on cells, or if necessary by measuring
the binding directly to the receptor itself.
When insulin is injected subcutaneously, it forms a de-
pot from which it leaks into the blood; there, it circu-
lates and binds to its receptor, stimulating transport of
sugars into cells [2°]. The solubilization and absorption
of insulin from the injection site is a complex process af-
fected by many factors, such as the injection site, the state
of the patient and the nature of the insulin preparation.
Even in the best circumstances, however, the arrival o f "
insulin in the bloodstream of a diabetic patient following
an injection is markedly different from that of the nor-
mal individual. Whereas pancreatic insulin is secreted into
the blood within minutes of eating, injected insulin will
take up 30 min or more to reach the maximum concen-
tration in the blood, depending on the state of the indi-
vidual. Consequently, whereas in healthy people the level
of sugar in the blood is tightly controlled, this is not the
case in insulin-treated diabetics. In addition, the contin-
uous low-level secretion of insulin that occurs in healthy
people is, of course, difficult to reproduce in diabetics.
As these deficiencies may be associated with the compli-
cations often observed in diabetics, such as the failure of
to improve the control of sugar levels in diabetics by im-
proving the insulin preparations that they use.
The origin of the sluggish absorption of insulin into
the bloodstream is thought to lie in its aggregation into
dimers and hexamers, illustrated in Fig. 1. These large
oligomers will migrate more slowly from the injection
site into the blood stream than dimers or monomers.
Once diluted in the blood stream, the hexamers and
dimers very rapidly dissociate into monomers, the active
species. Once in the circulation, the monomers can bind
to the insulin receptor, stimulating sugar transport into
the cell and the related metabolic events. Thus, from the
beginning of these experiments, the obvious target for
a rapidly acting preparation was a stable monomeric in-
sulin molecule.
Protein engineering offered a new route to the construc-
tion of such an insulin. The three-dimensional structure
of the insulin hexamer, dimer and monomer were all
known and the interactions responsible for self-associ-
ation of the hormone were understood [3",4°J. Further,
the surface on the monomer responsible for binding to
the receptor, though not precisely defined, had been well
enough characterized to permit the construction of se-
lected mutations that would not compromise the hor-
mone's potency [3-,4°].
A number of chemically produced insulin analogues,
such as the monomeric sulphated insulin [5"], desB1-
insulin [6] and proinsulin [7], have been used in therapy
but their impact has been fairly minor. However, chemi-
cal modifications of insulin have helped to identify the
importance of the B-chain carboxy-terminal residues in
receptor-binding activity [8o°,9.oj. By combining chemi-
cal and enzymatic techniques, insulin analogues with con-
siderably elevated effective potency have been synthe-
sized [10-].
Design of a fast-acting (monomeric) insulin
Examination of the insulin hexamer has shown that it
is assembled from three dimers which are held together
by coordination to centrally positioned Zn 2 + ions, with
the dimers making both non-polar and polar contacts

934 (~) Current Biology Ltd ISSN 0959-440X

 Designing insulin for diabetes therapy by protein engineering Brange, Dodson, Xiao
examer
Dimer Dimer
i-
~ ~ ~ _ ~ g l ~ imer
A21 A21
Fig. 1. The assembly of insulin from monomer to dimer and to the
zinc-containing hexamer, viewed along the threefold axis. The in-
sulin molecule A and B chains are shown as a Ca trace. The side
chains making interactions that affect dimer or hexamer forma-
tion are represented in full with their van der Waals radii drawn
as a Connolly (or dotted) surface.
[3°]. The monomers within the dimers are held together
by main-chain hydrogen bonds between B24 and B26,
arranged in an antiparallel 13-sheet, and by the closed
packed burial of an extensive non-polar surface. Figure 2
illustrates the dimer-forming surfaces, and the close con-
tacts that they make.
The obvious route to creating a fast-acting stable in-
sulin monomer is to prevent dimer or hexamer forma-
tion [11°.]. Two means of doing this are suggested by
the interactions illustrated in Figs 2 and 3. The first is to
introduce a bulky side chain into the dimer-forming sur-
face so as to interfere with the specific contacts made by
these surfaces. The second is to introduce a negatively
charged carboxylic acid which will oppose and repel a
naturally occurring carboxylic acid in the dimer (see Fig.
3). In either case, the residue selected for mutation must
935
[a
,~e£ ~. ,J/o
.o
h
- \
/
Fig. 2. Contacts made between the insulin monomer surfaces
within the dimer of the 2Zn insulin hexamer, viewed along the
local twofold axis. (a) The contacts made by ValB12 and TyrB26.
(b) The contacts made by the main-chain sequence B20-B28. The
close approach of ProB28 to B20-B23 can be seen.
be one that is not important to receptor binding. In this
paper, we shall discuss four selected mutations in detail.
ValB1211e
The introduction of one extra methyl group at each B12
residue requires considerable adjustment by the local
structure (Fig. 2). It was found, however, that this mu-
tation has a relatively small effect on dimer formation as,
in the presence of Zn 2 +, the mutant ValB12Ile insulin still
forms hexamers and rhombohedral crystals [11o.]. X-ray
analysis of the mutant insulin hexamer has revealed that
the two isoleucine groups are accommodated within the
dimer by small but extensive adjustments at the interact-
ing surfaces (X Bing, unpublished data). The resulting in-
crease in absorption rate caused by this mutation is 30 %
of that of the native hormone. The same mutation also
results in some reduction in dimer formation but, even
at 0.2 mM insulin concentration, the dimer is a significant
species.
936 Proteins
B9
Fig. 3. The insulin dimer from the Zn insulin hexamer viewed in
the direction of the local twofold axis. Only the Ca positions
are shown for the A and B chains, except for the side chains
of SerBg, HisB'10 and GluB13, which are highlighted. The close
hydrogen-bonding approach of GluB'I3 and its proximity to SerB9
and HisB'10can be seen.
As the receptor-binding affinity o f this mutant is reduced
to ~ 30 % of that of native insulin [ 11--], it suggests that
valine at B12 is not critical for bonding, even though it
is invariant and lies close to other residues, such as B24,
that are important.
SerBgAsp and ThrB27Glu
In this double mutant, the key residue for preventing
dimer formation is SerB9 which, in the monomer, is ad-
jacent to GluB13 in the a-helix of the B chain; in the
dimer, the two GluB13 side chains are paired together
and must therefore represent some kind of destabilizing
effect [11-.]. Conversion of SerB9 to aspartic acid would
yield four contiguous negatively charged side chains in
the dimer. The disposition of AspB9 and GIuB13 in the
insulin dimer are illustrated in Fig. 3; in the hexamer,
these repulsive contacts would be repeated three times
(see Fig. 1). This mutation effectively destabilizes the
dirner to a m o n o m e r at millimolar concentrations.
The NMR studies of Roy et al. [12] indicate that the
AspB9/GIuB27 insulin retains the major structural fea-
tures of the native molecule in the dimer and hexamer.
The effect of the AspB9 and GIuB27 side chains on ab-
sorption rate is striking; the rate of absorption into the
bloodstream is approximately three times that of the na-
tive hormone [2.,11..].
A proton NMR study on the single mutant AspB9 insulin
has also established that the mutant hormone's structure
in solution is essentially identical to that of the native
molecule in the crystal [14o]. The experiments were car-
ried out at low pH where the carboxylate groups would
be protonated where the molecule is dimeric.
The receptor affinity of this insulin is ,-~20 % of that of na-
tive insulin. This reduction suggests that the B9 side chain
lies near the receptor-binding site or affects the confor-
mational changes associated with binding to the receptor.
Surprisingly, the doubly mutated insulin has been crystal-
lized as a hexamer. In this structure, the repulsive effects
existing between six closely packed GlnB13 side chains
and the six adjacent AspB9 side chains are neutralized by
six Ca 2 + ions. This enables the dimers to come together
and coordinate to two central Zn 2 + ions. The hexamer
is afforded further stability by phenol molecules, which
are bound specifically between the dimers (J Turkenburg,
unpublished data) [14]. The phenols favour the forma-
tion of a helix at B1-B8 which covers the axial zinc sites.
The chemical and structural environments of the Ca 2 +
ions, Zn 2 + ion and phenol molecules, which together
cancel the effects of the AspB9 mutation, are illustrated
in Fig. 4. The capacity of chemical additives to produce
stable hexamers from a mutant monomeric molecule re-
vealed by this study opens up new possibilities for com-
bining the established tradition of pharmaceutical formu-
lation and the new opportunities of protein engineering.
HisB10Asp
This mutation introduces a negative charge ~ 1 0 A
from the naturally occurring GluB13 (Fig. 3). At this
distance, the electrostatic repulsion experienced by
AspB10 is less than that experienced by the mu-
tant AspB9 [11..]. Consequently, the AspB10 dimer
is more stable than the AspB9 dimer. On the other
hand, the loss of the histidine at B10 would pre-
vent any chance of zinc coordination and hexamer for-
mation. Osmometry experiments indicate that at milli-
molar concentrations this mutation still forms dimers to
a significant level [ 11., ]. The rate of absorption of insulin
from the tissue is increased by a factor of ,-, 2, suggest-
ing that the hexamer is the main cause of slow release
from the injection site. The potency of this mutant is ele-
vated to ,-, 200 % of that of native insulin, a finding that
suggests that, like B9, B10 lies near the binding site and
thus can influence the shape of the active surface or the
conformational behaviour required for binding.
Crystals of the AspB10 insulin dimer have been grown
but the structure has not yet been solved (J Turkenburg,
unpublished data). In the presence of Zn 2 + and Ca 2 +
ions, the AspB10 mutant forms large cubic crystals in
which the molecule proves to be organized as a dode-
career, as illustrated in Fig. 4. This large structure has lo-
cal 32 symmetry with two sets of threefold-related dimers
organized around a central threefold axis. Two Zn 2 + ions
on the axis each coordinate with three B5 histidines in
a structure reminiscent of the HisB10 zinc coordination
structure seen in the native insulin hexamer. Only one
HisB5 in each dimer binds to the Zn 2 +, and the structure
of the dimer in this organization is essentially identical to
that of the native insulin (Fig. 5).
ProB211Asp
In the native insulin dimer, ProB28 is largely buried
in a non-polar environment (Fig. 2b). Substitution with
charged aspartic acid reduces self-association almost as
 Designing insulin for diabetes therapy by protein engineering Brange, Dodson, Xiao
much as the AspB9 mutation [9"']. No loss of potency is
associated with this mutation, a characteristic of changes
and deletions at residues B26-B30 [3",8"',9"']. So far, no
crystals of this mutant as a monomer have been grown,
though there are crystals of a hexamer grown in the pres-
ence of Zn 2+ and Ca 2+ (B Xiao, unpublished data).
Design of slow-acting insulins
In order to prolong insulin release and to mimic better
the natural low-level continuous secretion provided by
the pancreas, mutations have been designed to stabilize
either the insulin hexamer or the crystals. The philosophy
behind these experiments is to increase the stabilizing in-
teractions made within the hexamer and/or the tendency
to form crystals at neutral pH. The strategy has therefore
been to develop an experimental basis for designing ei-
ther insulin preparations containing crystals that would
dissolve slowly, or insulin solutions in which the absorp-
tion from the subcutaneous layer is delayed, either by de-
creased dissociation of the hexamer or by the formation
of insulin microcrystals at the site of injection.
GluB13GIn
The centre of the insulin hexamer features six glutamic
acids packed closely together (Fig. 1). Their electrostatic
repulsion must represent a destabilizing factor in the hex-
amer. An obvious mutation is to convert the glutamic
acid residues to glutamines, which are neutral. Solution
937
Fig. 4. The AspB9/GluB27 insulin hex-
amer is constructed from three dimers
coordinated via HisB'10to two central
Zn 2+ ions (0). The six GluB13 and six
B9Asp side chains are shown; these are
arranged close together at the hexamer
centre and form Ca 2 + -binding sites (•).
The six phenol molecules packed be-
tween the dimers are also shown.
studies show that this mutated insulin forms discrete
and stable hexamers even in the absence of Zn 2 + ions
[15.,16,17"]. The crystal structure of a zinc-free hexamer
has been determined (X Bing, unpublished data). In this
structure, the six GlnB13 side chains are completely hy-
drated and make no contacts with protein a t o m s - - i n
contrast with GIuB13 in the 2Zn insulin hexamer [3"]
and with the arrangement of the GInB13 side chains pro-
posed by Markussen and colleagues [18]. Figure 6 il-
lustrates the conformation of the glutamic acid and glu-
tamine residues at the centre of the B13 Gin and B13 Gln
2Zn insulin hexamers.
Size exclusion chromatography experiments have con-
firmed that the GlnB13 substitution results in a reduced
tendency of the hexamer to dissociate in neutral solution
[19]. However, the effect on insulin absorption rate after
injection of a neutral solution of the GInB13 analogue
is not dramatic, the rate being reduced by only ,~ 25 %
(U Ribel, personal communication). This small effect is
in agreement with absorption experiments using a non-
dissociable cobalt-insulin hexamer [2%20].
More successful results have been achieved when the
substitution GluB13Gln is combined with the introduc-
tion of extra positive charge [18]. The resulting insulin
is less soluble at physiological pH; as a result, crystalliza-
tion takes place in the subcutis after injection of a slightly
acid solution of such modified insulin, and a substantially
protracted effect is observed.
938 Proteins

Fig. 5. The Aspgl0 insulin dodecamer,

viewed in the direction of the three-
fold axis. One set of three dimers is
drawn in thick lines, the other in thin
lines. The HisB5 and AspB10 side chains
are shown. Three HisB5 side chains, one
from each triad of dimers, coordinate
to the central Zn 2+ ions (A). There are
three local twofold axes perpendicular
to the threefold axis relating the dimers.
Each dimer also contains its own local
axis of symmetry.

O rl

n °v s

Fig. 6. The six B13 side chains in the insulin hexamer. (a) The orientation 6f GluB13 in 2Zn insulin; hydrogen bonds exist between pairs
of adjacent B13 glutamic acids and glutamates. (b) Glng13 modelled to optimize hydrogen-bond interactions (n, amide; o, oxygen) [15].
(c) GInB13 side chains in the zinc-free insulin hexamer; the glutamyl side chains are solvated and make no mutual hydrogen bonds.

ThrB27ArWB30 amide to an amide group by enzymatic techniques, further in-

The observation that insulins modified to contain ex-
tra positive charges were generally longer acting has led
to a programme of introducing arginines and lysines by
protein engineering [11-.,18]. In this case, the modifica-
tions have to be designed to protect the hormone from
tryptic proteolysis. An arginine located at B27, adjacent
to ProB28, was immune to cleavage. In addition, the
B chain carboxy-tenninal carboxylic acid was converted
creasing the positive charge. This doubly mutated insulin
exhibited some prolongation of action.
The largest increase in protraction with a single substitu-
tion has been obtained with the substitution of ThrB27
with arginine, but the B30 amide and the GInB13 sub-
stitutions also caused significant increases. These effects
appear to be additive, as the slowest disappearance was
 Designing insulin for diabetes therapy by protein engineering Brange, Dodson, Xiao 939
obtained with the analogue combining all three substitu-
tions, i.e. GInB13, ArgB27 and B30 amide [18].
The crystal structure of GInB13/ArgB27/B30 amide re-
veals that the B30 amide and ArgB27 make contacts in
the lattice that are quite different from those anticipated,
and which do not adequately explain the reduced solu-
bility of the mutant insulin crystal. The mutated arginyl
side chains make no lattice contacts, and the interactions
made by B30 amide do not appear to be more favourable
than those of the native carboxytate group (3 Turken-
burg, unpublished data).
Conclusion
The alteration of insulin's properties of self-association
but not its metabolic effects by informed mutagenesis is
an important first step in the application of protein en-
gineering to medicine. Surprises and difficulties still re-
main however. For example, the unexpected formation
of the hexamer by the AspB9/GIuB27 mutant insulin and
the dodecamer by the AspB10 mutant insulin are strik-
ing examples of how the complexity of protein surfaces
presents formidable problems in predicting the chemical
and structural consequences of residue mutation. In addi-
tion, the very complex environment in the tissues where
the hormone is absorbed means great care has to be
taken with clinical characterization of any new insulins.
Nonetheless, the progress made so far encourages opU-
mism and, despite the uncertainties, it is likely that pro-
tein engineering will play a vital role in the future devel-
opments of pharmaceutical preparations. In addition, we
can expect to see insulins available with a wide spectrum
of properties whose exploitation will lead to better man-
agement of the disease.
References and recommended reading
Papers of special interest, published within the annual period of review,
have been highlighted as:
• of interest
oo of outstanding interest
1. KAHNCR: The Molecular Mechanism of Insulin Action. A n n u
Rev Med 1985, 36:429-451.
2. BRANGEJ, OWENS DR, KANGS, VOLUNDPc Monomeric Insulins
• and Their Experimental and Clinical Implications. Diabetes
Care 1990, 13:923-954.
The authors describe the biological behaviour of a large selection of
insulin mutants and report on animal and human trials.
3. BAKEREN, BLUNDELLTL, CtrFFIELD JF, CUTFIELD SM, DODSON
• EJ, DODSON GG, HODGKIN DC, HUBBARD RE, ISAACS NWI,
REYNOLDS CO, SAKABE K, SAKABE N, VIJAYAN NM: T h e Struc-
ture of 2Zn Pig Insulin Crystals at 1.5 A. Phil T r a m R Soc
Lond [B] 1988, 319:369-456.
This paper contains a very comprehensive descripUon of the 2Zn in-
sulin crystal structure and its pattern of assembly into dimers and (in
the presence of zinc ions) hexamers. The hormone's active surface is
identified and the evidence for its involvement in receptor binding re-
viewed.
4. DEREWENDAU, DEREWENDAZ, DODSON GG, HUBBARD RE: In-
. sulin Structure. In Handbook of Expenmental Pharmacol-
ogy, Vol 92 edited by Cuatracasas P, Jacobs S [book]. Berlin:
Springer Verlag, 1990, pp 23-39.
This paper contains detailed descriptions of the different structures in-
sulin has in different crystals.
5. MOLONEYPJ, APmLEMA, WILSON S: Sulfated Insulin for Treat-
• m e n t of Insulin Resistant Diabetics. J N e w Drugs 1964,
4:4258-4263.
This paper describes the successful treatment of insulin resistant dia-
betes by using 'monomeric' sulphated insulin.
6. KERP L, STEINHILBERS, KASEMIRH, HAN J, HEINRICHS HR, GEIGER
R: Changes in Immunospecificity and Biologic Activity of
Insulin due to Subsequent Removal of t h e Amino Acids B1,
B2 and B3. Diabetes 1974, 23:651-656.
7. GLAUBERHS, RIVERS RR, HENRY R, SCHMEISER R, WALLACE P,
KOTrERMANN 0(3, COHEN RM, RUBENSTEIN AH, GALLOWAYJA,
FRANK BH ET AL: In ViVO Deactivation of Proinsulin Action
on Glucose Disposal and Hepatic Glucose Production in
Normal Man. D/abetes 1986, 35:311-317.
8. MIRM1RARA, NAKAGAWASH, TAGER H: Importance of the
•• Character and Configurations of Residues B24, B25 and
B26 in Insulin Receptor Interactions. J Biol Chem 1991,
266:1428-1436.
These experiments were carried out using trypsin-catalysed semi-
.synthesis, first exploited by Inouye et al. [21], in which synthetic frag-
ments are combined with desoctapepUde (B23-B30) insulin. Using this
approach, non-natural amino acids can be incorporated into an inves-
tigation of the hormone's structural and biological behaviour. These
studies have identified that the PheB24 side chain, criUcal in dimer for-
mation, is not critical for receptor binding and that conformational con-
straints in the sequence B23-B26 affect binding potency.
9. STUMPOLF, HARTMANNH, BRANDENBURGD, CRU'VZFELDW: In
•• vivo Metabolic Activity of Des(B26-B30)-Insulin B25 Amide
and Related Analogues in the Rat. Diabetes Res Clin Pract
1990, 9:257-264.
The effects that different modificaUons at the B-chain carboxyl terminus
have on insulin's biological behaviour are analyzed. The results high-
light the importance of residues B24 and B25 and the unimportance of
the sequence B26-B30 in receptor binding. This is an excellent exam
pie of the kind of information needed if the protein engineering is to
be successful.
10. BURKE GP, HU SQ, OHTA N, SCHWARTZ GP, ZONG L,
** KAT,~YANNIS PG: Superactive Insulins. Biochem Biophys Res
Comm 1991, 173:982-987.
This paper describes the synthesis of a series of mutated and modi-
fied insulins with elevated potency. The study reveals that the elevation
of potency and binding is not strictly additive the mutation AspB10,
however, always increases these properties significantly.
11. BRANGEJ, PdBELU, HANSENJ, DODSON G, HANSENM, HAVELUND
•• S, MELBERGS, NORRIS F, NORRIS K, SNEL L, ET AL.: Monomeric
Insulins Obtained by Protein Engineering and their Medical
Implications. Nature 1988, 333:679-682.
The authors discuss the design and preparation of a series of h u m a n
insulin mutations and describe their solution and absorption behaviour.
12. ROY M, LEE W-K, KAAP,SHOLM NC, THOGERSEN tt, BRANGE J,
DUNN MF: Sequence Specific 1H NMR Assignments for the
Aromatic Region of a Biologically Active Monomeric Insulin
Mutant. Biochim Biopbys Acta 1990, 1053:63-73.
13. KRISTENSENSM, JORGENSEN AMM, LED JJ, BAL~CHMIDTP, HANSEN
• FB: Proton Nuclear Magnetic Resonance Study of the
B9 (Asp) Mutant of H u m a n Insulin. J Mol Biol 1991,
218:221-231.
This study was carried out at low pH where the insulin was dimeric
and where the monomerizing properties of AspB9 would be negated.
NonetheLess, this mutant insulin gave a sharper and more interpretable
NMR spectrum than native insulin, indicating ?
14. I)ODSON GG: Realistic Protein Engineering In Proceedings
of the Society of General Microbiology Symposium, Vol 44
edited by Baumberg S, Hunter IS and Rhodes PM [book].
Cambridge: Cambridge University Press, 1989, pp 293-307.
940 Proteins

15. HANSENJF: T h e Self Association of ZInc Free H u m a n Insulin 19. BRANGEJ, DREJER K, HANSENJF, HAVELUNDS, KAARSHOLMNK,
• and Insulin Analogue G13-Glutamine. Biophys Chem 1991, MEmERG SG, SORENSON A_R: Design of Novel Insulins with
39:107-110. Changed Self-association and Ligand Binding Properties. In
In this paper, it is established by osmometric and spectroscopic meth- Advances in Protein Design, International Workshop 1988
ods that~GInB13 assembles to a specific hexamer in the absence of edited by Blooker H, Collins J, Schmid RD, Schomburg D
Zn 2 +. [book]. Weinheim: VCH Publishers, 1989, pp 139-144.
16. RoY M, BRADER ML, LEE RW-K, KAARSHOLMNC, HANSEN JF, 20. HILL CP, DAUTERZ, DODSON EJ, DODSON GG, DUNN MF: The
DUNN MF: Spectroscopic Signatures of t h e T to R Confor- X-ray Structure of an Unusual Ca 2+ Site and t h e Roles
mational Transition in t h e Insulin Hexamer. J Biol Chem of Zn 2+ and Ca 2+ in Assembly, Stability and Storage. B/~
1989, 264:19081-19085. chem/stry 1991, 304:917-924.
17. KRUGERP, GILGE G, CABUKY, WotJdVmR A: Cooperativity and
• Intermediate States in t h e T ~ R Structural Transition of 21. INOUYEK, WATANABEK, MORIHAR IC, TOCHINO Y, KANAYAT,
Insulin. Biol Chem HoppeSeyler 1990, 371:669-673. EMURA J, SAKIKABARAS: Enzyme Assisted Semi-Synthesis of
This paper describes circular dichroism studies on native and GlnB13 H u m a n Insulin. J A m Chem Soc 1979, 101:751-752.
insulin hexamers.
18. MARKUSSENJ, HOUGAARDP, RIBEL U, SORRENSONA, SORENSON J Brange, Novo Research Institute, Novo Alle, DK-2880 Bagsvaerd, Den-
E: Soluble, Prolonged-acting Insulin Derivatives (2) Degree
mark.
of Protraction and Crystallisability o f Insulins Substituted
in Positions A17, BS, B13, B27 and B30. Protein Eng 1987, GG Dodson and B Xiao, The Chemistry Department, University of York,
1:215-223. York YO1 5DD, UK.