CONTROLLING INSULIN BIOAVAILABILITY BY CRYSTAL CONTACT ENGINEERING
Rolf Hilgenfeld1 ,Tom Sicker1, M. Dörschug2,3, R. Obermeier2, K. Geisen2, G. Seipke2, Harald Berchtold2
1Institute of Molecular Biotechnology,
Department of Structural Biology and Crystallography,
Beutenbergstr. 11, D-07745 Jena, Germany, e-mail: hilgenfd@imb-jena.de
2 Hoechst Marion Roussel, Metabolism Group,
D-65926 Frankfurt, Germany
3 present affiliation: Bayer AG, Wuppertal,
Germany
Keywords: insulin, long-acting insulin, crystal contact
engineering
Currently available insulins are unable to simulate the pattern of endogeneous insulin secretion (1). Recombinant DNA technology offers the opportunity to engineer the hormone with the aim of achieving either a faster onset of blood glucose-lowering activity or a prolongued action. In all commercially available insulin preparations the hormone exists predominantly as the 2Zn hexamer, while the monomer is the active species circulating in the blood stream and binding to the receptor. Destabilization of the hexamer by mutagenesis yields derivatives with an accelerated onset of action, the most successful representative of this class so far being Insulin Lispro® (B28-Lys,B29-Pro-Insulin) (2).
In addition to fast-acting insulins, there is also a strong clinical need for long-acting analogues simulating basal insulin secretion. We have reduced the solubility of the hexamer by increasing the stability of insulin crystals through protein engineering. The B-chain has been extended at its carboxy terminus by attachment of basic or hydrophobic amino acid residues. We could show by high-resolution X-ray crystallographic analysis that these extra residues make additional interactions with other hexamers in the crystals. As a result, the packing density of the crystals is increased, i.e. their water content is reduced. Consequently, the crystals are stabilized against dissolution, thus acting as an efficient depot form for the hormone. These considerations are relevant even in the case of the application of acidic solutions of the insulin analogues, because crystals will form out of these in the subcutaneous depot.
During the design process, it became clear that too large an increase of the packing density, caused by maximization of crystal contacts, leads to an unfavourable overstabilization of the crystals and, therefore, to almost complete inactivity of the respective insulin analogue. What turned out to be necessary was a fine-tuned crystal contact engineering process that involved creation of individual interhexamer contacts and deletion of others. This approach yielded an insulin analogue exhibiting a flat activity profile extending over up to 24 hours in humans, which is a dramatic prolongation compared to the 6-hour activity of unmodified human insulin. The derivative, named Hoe 901, is now in phase III clinical trials and expected to be on the market early in the year 2000.