Reversible insulin self-assembly under carbohydrate control

Article abstract of DOI:10.1021/ja051038k

Insulin with built-in pairs of boronates and polyols can produce soluble high molecular weight self-assemblies under control by carbohydrates. The illustrated principle has potential utility for general protein and peptide protraction and controlled drug

Full text of DOI:10.1021/ja051038k

Published on Web 04/12/2005
Reversible Insulin Self-Assembly under Carbohydrate Control
Thomas Hoeg-Jensen,* Svend Havelund, Peter K. Nielsen, and Jan Markussen
NoVo Nordisk, NoVo Alle 6B2.54, DK-2880 BagsVaerd, Denmark
Received February 17, 2005; E-mail: tshj@novonordisk.com
The biological properties of peptides and proteins are under
significant influence by their molecular weights, which impact,
among other things, diffusion rates and kidney-mediated clearances.
The rate of diffusion, in turn, is proportional to the half-life of
subcutaneously injected depots of peptide and protein drugs. Insulin
provides a good illustration: Monomeric insulins are fast acting
(t1/2 1 h),1 and hexameric insulins are intermediary acting (t1/2 2-4
h); on the other hand, insulins that bind to high molecular weight
proteins2 or form soluble high molecular complexes3 are long acting
(t1/2 > 10 h).
Peptide and protein protraction via delayed diffusion has often
been pursued by drug entrapment in polymers.4 Furthermore,
stimuli-controlled drug release from polymer systems has been
demonstrated with polymers whose swelling is, for example,
glucose-dependent.5 However, the practical use of polymers faces
a number of difficulties such as biocompatibility, drug instability,
or denaturation and increase in drug volumes.
We report here a novel concept for peptide and protein stimuli-
controlled protraction by reversible formation of soluble high
molecular weight self-assemblies, as exemplified by insulin self-
assembly under control by D-sorbitol and D-glucose. Specifically,
by equipping insulin with pairs of boronates and carbohydrates or
other polyols, sufficient driving force can be provided for formation
of soluble high molecular weight insulin hexamer-hexamer self-
assemblies. The use of boronate-carbohydrate binding for the
peptide self-assembly enables carbohydrate-mediated displacement
of the binding and, hence, carbohydrate-controlled peptide release,
Figure 1.
Figure 1. Illustration of insulin self-assembly under carbohydrate control.
withdrawing groups,7 whereby the pKa of aryl boronic acids can
be tuned to 7.4 or lower.
Insulin 1 was prepared as described in Supporting Information.
Insulins 2 and 3 were similarly prepared and used as controls. The
required building blocks were attached to insulin LysB29 by
acylation with succinimidyl esters at pH 10.5.8 Importantly, to
prevent the boronate-polyol pair of insulin 1 from binding
internally within the insulin 1 monomer, the boronate-carbohydrate
motifs were positioned on insulin via a common scaffold
(L-glutamic acid).
Insulin in formulation with zinc(II) yields insulin hexamers,
which constitute the stored form of insulin in both the pancreas
and drug formulations. When given subcutaneously to diabetes
patients, hexameric insulin is absorbed mainly via dissociation to
dimer and monomer, and the resulting drug half-life is ap-
proximately 2 h. This timing fits fairly well with the meal-related
requirement for insulin but is insufficient in the basal need for
insulin (24 h/day). Long-acting insulin preparations have been
obtained by insulin coprecipitation with protamine or by precipita-
tion via tuning of the isoelectric point of the peptide. However,
protractions from soluble depots generally yield better day-to-day
reproducibility of drug absorptions. Soluble insulin depots have been
obtained by insulin acylation with lipophilic groups that either
enable binding to high molecular-weight proteins2 or impose
formation of soluble high molecular weight insulin self-assemblies.3
However, it would be desirable if the insulin rate of absorption
could be controlled by outside factors such as glucose or other small
molecules that could be applied as meal-related release factors.
Boronates are known to bind to 1,2-diols such as carbohydrates
in aqueous solution with affinities in the mM to µM range (Kd).6
However, because simple boronic acids possess pKa values of
approximately 8.5, the binding is weak at the physiological pH of
7.4. Luckily, this problem can be corrected by use of electron-
The insulin receptor affinities of 1-3 were tested as previously
described,9 and the affinities were in the expected range relative to
native insulin: 69, 122, and 85%, respectively. The LysB29 position
is generally tolerant with respect to insulin receptor affinity and
preservation of the overall properties of insulin.
The molecular weights of insulins 1-3 in zinc(II)/phenol
formulations were analyzed by size-exclusion chromatography
9
6158
J. AM. CHEM. SOC. 2005, 127, 6158-6159
10.1021/ja051038k CCC: $30.25 © 2005 American Chemical Society

C O M M U N I C A T I O N S
is weakly bound and removed during the SEC analysis. However,
when phenol is present in the SEC eluent, the self-assembly is
retained during the SEC analysis and can be detected. Light
scattering measurements confirmed the SEC data, including the
effect of phenol (see Supporting Information). Notably, phenol at
low mM concentration is generally included in insulin formulations
for human use, acting as both a preservative and a physical
stabilizer, as the R-form is more stable than the T-form.12
The carbohydrate-controlled reversibility of the self-assembly
of 1 was tested with D-sorbitol and D-glucose, which were added
either to the formulation or to the SEC eluent. D-Sorbitol, which
binds to boronates with a Kd of approximately 0.1 mM,13 was found
to disassemble the complex in a dose-dependent manner with an
EC50 of 50 mM (Figure 2d). Equilibrium was reached as fast as
the samples could be analyzed by SEC (t1/2 is <5 min). Glucose,
which binds to boronates with Kd of approximately 10 mM,6 was
not able to effect the disassembly of insulin 1. However, by
formulating 1 as a 1:1 mixture with DesB30 insulin, it was possible
to observe an effect with glucose. Not surprisingly, the extent of
self-assembly of the 1:1 insulin 1/DesB30 mixture was less
pronounced than with neat 1 (60 vs >95%). When glucose was
included in the samples, there was a dose-dependent erosion of
the self-assembly, and, for example, 100 mM glucose yielded 53%
self-assembly.
In conclusion, the reported work illustrates a novel concept for
peptide or protein protraction by soluble reversible self-assembly
via polar interaction. Furthermore, the extent of self-assembly can
be controlled by the addition of small molecules such as carbohy-
drates. The use of external polymers as diffusion modulators in
drug protraction can thus be circumvented. The concept has been
demonstrated for insulin 1 using boronates-polyols as recognition
pairs and SEC-based detection of the soluble high molecular weight
self-assemblies under control by D-sorbitol or D-glucose. Notably,
the geometry of protein-protein packing appears to be critical in
obtaining self-assembly. In the present case, insulin R-folding and
hence the presence of phenol is required. The overall concept should
be applicable to other peptides or proteins, although there will be
a requirement for inherent formation of peptide or protein oligomers.
Oligomer formation is, however, quite common for peptides and
proteins or may be accessible by protein engineering.14
Figure 2. SEC analysis of insulin 1 with phenol and D-sorbitol in PBS
buffer, pH 7.4, 37 °C.
(SEC) on a non-carbohydrate BioSep column. The standards being
used were ferritin (500 kDa), Co(III) insulin hexamer (36 kDa),
and AspB9-GluB27 insulin monomer (6 kDa), Figure 2a. Insulins
1-3 were dissolved at the standard conditions for clinical use of
insulin: 600 µM insulin with 3 Zn(II)/hexamer, 16 mM phenol,
16 mM m-cresol, 7 mM phosphate, pH 7.4. The standard formula-
tion contains glycerol as an isotonic agent, but due to the potential
interference of this diol with boronate-carbohydrate binding,
glycerol was replaced with 100 mM sodium chloride.
SEC analysis of insulin 1 initially showed only hexameric insulin
peaks, Figure 2b. However, upon inclusion of 4 mM phenol in the
SEC eluent, insulin 1 appeared as the desired high molecular weight
complex, Figure 2c. Importantly, since the pH values of all
formulations and buffers were adjusted to 7.4, the observation is
not an effect of pH. Notably, since the control compounds, insulins
2 and 3, appeared as hexamers in all SEC analysis (see Supporting
Information), the boronate-glycol interaction is required for
obtaining the observed self-assembly with insulin 1.
Supporting Information Available: Procedures for synthesis of
insulin 1 and SEC analysis of insulins 1-3 (PDF). This material is
available free of charge via the Internet at http://pubs.acs.org.
References
(1) Chapman, T. M.; Noble, S.; Goa, K. L. Drugs 2002, 62, 1945-1981.
(2) Havelund, S.; Plum, A.; Ribel, U.; Jonassen, I.; Vølund, A.; Markussen
J.; Kurtzhals, P. Pharm. Res. 2004, 21, 1498.
(3) Whittingham, J. L.; Jonassen, I.; Havelund, S.; Roberts, S. M.; Dodson,
E. J.; Verma, C. S.; Wilkinson, A. J.; Dodson, G. G. Biochemistry 2004,
43, 5987.
(4) Allen, T. M.; Cullis, P. R. Science 2004, 303, 1818.
(5) Miyata, T.; Uragami, T.; Nakamae, K. AdV. Drug DeliV. ReV. 2002, 54,
79.
The role of phenol in the self-assembly of 1 can be explained
by considering the well-described phenol-dependent folding of
insulin.10 In formulation without phenol, the N-terminal of the
B-chain of insulin is coiled (T-form),11 whereas in phenol-containing
formulation the same segment is R-helical (R-form, [phenol] > 2
mM).11 Since self-assembly of insulin 1 is observed only in the
presence of phenol, insulin 1 must apparently be in the R-form to
allow the hexamers to pack in such geometry that the boronate-
polyol groups can pair and hence effect formation of high molecular
weight self-assembly. When phenol is present in the formulation
only, the self-assembly cannot be detected by SEC, because phenol
(6) James, T. D.; Shinkai, S. Host-Guest Chem. 2002, 218, 159.
(7) Vedsø, P.; Olesen, P. H.; Hoeg-Jensen, T. Synlett 2004, 892.
(8) Myers, S. R.; Yakubumadus, F. E.; Johnson, W. T.; Baker, J. E.; Cusick,
T. S.; Williams, V. K.; Tinsley, F. C.; Kriauciunas, A.; Manetta, J.; Chen,
V. J. Diabetes 1997, 46, 637.
(9) Schaffer, L. et al. Proc. Nat. Acad. Sci. 2003, 100, 4435.
(10) Brzovic, P. S.; Choi, W. E.; Borchardt, D.; Kaarsholm, N. C.; Dunn, M.
F. Biochemistry 1994, 33, 13057.
(11) Pittman, I.; Tager, H. Biochemistry 1995, 34, 10578.
(12) Berchtold, H.; Hilgenfeld, R. Biopolymers 1999, 51, 165.
(13) Shiino, D.; Murata, Y.; Kataoka, K.; Koyama, Y.; Yokoyama, M.; Okano,
T.; Sakurai, Y. Biomaterials 1994, 15, 121.
(14) Clackson, T. et al. Science 2000, 287, 826.
JA051038K
9
J. AM. CHEM. SOC. VOL. 127, NO. 17, 2005 6159