Self-Association Properties of Monomeric Insulin Analogs Under Formulation Conditions

Purpose. The purpose of the current study was to investigate the effects of two important excipients, zinc and m-cresol, on the self-association properties of a series of monomeric insulin analogs. In this way, the effects on formulation behavior of individual amino acid substitutions in the C-terminal region of the insulin B-chain could be compared. Methods. The self-association of ten insulin analogs was monitored by equilibrium and velocity analytical ultracentrifugation under three different conditions: (i) in neutral buffer alone; (ii) in neutral buffer containing zinc ion; and (iii) in neutral buffer containing both zinc ion and phenolic preservative (a typical condition for insulin formulations). The self-association properties of these analogs were compared to those of human insulin and the rapid-acting insulin analog Lys^B28Pro^B29-human insulin. Results. The analogs in the current study exhibited a wide range of association properties when examined in neutral buffer alone or in neutral buffer containing zinc ion. However, all of these analogs had association properties similar to human insulin in the presence of both zinc and m-cresol. Under these formulation conditions each analog had an apparent sedimentation coefficient of s* = 2.9–3.1 S, which corresponds to the insulin hexamer. Conclusions. Analogs with changes in the B27–B29 region of human insulin form soluble hexamers in the presence of both zinc and m-cresol, and m-cresol binding overrides the otherwise destabilizing effects of these mutations on self assembly.     

Assembly and Dissociation of Human Insulin and LysB28ProB29-Insulin Hexamers: A Comparison Study

Purpose. Investigations into the kinetic assembly and dissociation of hexameric Lys^B28Pro^B29-human insulin (LysPro), a rapid-acting insulin analog produced by the sequence inversion of amino acids at positions B28 and B29, were designed to explain the impact that the sequence inversion has on the formulation and pharmacokinetics of the insulin analog. Methods. The kinetics of phenolic ligand binding to human insulin and LysPro were studied by stopped-flow spectroscopy. The kinetics of R₆ hexamer disruption were studied by extraction of Co(II) with EDTA. Results. Phenolic ligand binding to human insulin yielded rate constants for a fast and slow phase that increased with increasing ligand concentration and are attributed to the T₆ T₃R₃ and T₃R₃ R₆ transitions, respectively. However, the kinetics of phenolic ligand binding with LysPro was dominated by rates of hexamer assembly. The kinetic differences between the insulin species are attributed to alterations at the monomer-monomer interface in the dimer subunit of the LysPro analog. The extraction of Co(II) from both hexameric complexes by EDTA chelation is slow at pH 8.0 and highly dependent on ligand concentration. Cobalt extraction from LysPro was pH dependent. Of the various phenolic ligands tested, the relative affinities for binding to the human and LysPro hexamer are resorcinol > phenol > m-cresol. Conclusions. The extraction data support the formation of an R₆-type LysPro hexamer under formulation conditions, i.e., in the presence of divalent metal and phenolic ligand, that is similar in nature to that observed in insulin. However, the formation kinetics of LysPro identify a radically different monomeric assembly process that may help explain the more rapid pharmacokinetics observed with the hexameric formulation of LysPro insulin relative to human insulin.     

Preparation of a microcrystalline suspension formulation of Lys(B28)Pro(B29)-human insulin with ultralente properties.

The monomeric analogue, Lys(B28)Pro(B29)-human insulin (LysPro), has been crystallized using similar conditions employed to prepare extended-acting insulin ultralente formulations. In the presence of zinc ions, sodium acetate and sodium chloride, but without phenolic preservative, LysPro surprisingly forms small rhombohedral crystals with similar morphology to human insulin ultralente crystals with a mean particle size of 20 +/- 1 microm. X-ray powder diffraction studies on the LysPro crystals prior to dilution in ultralente vehicle ([NaCl] = 1.2 M) revealed the presence of T(3)R(3)(f) hexamers. Consistent with human insulin ultralente preparations, LysPro crystals formulated as an ultralente suspension ([NaCl] = 0. 12 M) contain T(6) hexamers indicating that a conformational change occurs in the hexamer units of the crystals upon dilution of the salt concentration. The pharmacological properties of subcutaneously administered ultralente LysPro (ULP) were compared to ultralente human insulin (UHI) using a conscious dog model (n = 5) with glucose levels clamped at basal. There were no statistically significant differences between the kinetic and dynamic responses of ULP compared to UHI [C(max) (ng/mL): 3.58 +/- 0.76, ULP and 3.61 +/- 0. 66, UHI; T(max) (min): 226 +/- 30, ULP and 185 +/- 42, UHI; R(max) (mg/kg min): 11.2 +/- 1.9, ULP and 13.3 +/- 2.0, UHI; and T(Rmax) (min): 336 +/- 11, ULP and 285 +/- 57, UHI]. Although the Pro to Lys sequence inversion destabilizes insulin self-assembly and greatly alters the time action of soluble LysPro preparations, this modification has now been found neither to prevent the formation of ultralente crystals in the absence of phenolics nor to compromise the protracted activity of the insulin analogue suspension.     

Insulin self-association and the relationship to pharmacokinetics and pharmacodynamics

The treatment of type 1 diabetes requires multiple, daily injections of insulin. While many improvements involving formulation adjustments have been made in an attempt to optimize therapy, clinical experience indicates that the commercially available insulin preparations used for treatment have significant limitations. One principal deficiency relates to poor simulation of the physiological insulin secretion pattern, making achieving normalization of blood glucose concentrations difficult. Endogenous insulin secretion in nondiabetic subjects is characterized by a pulsatile profile that displays multiple, meal-stimulated phases and low basal concentrations between meals and overnight. Optimal diabetes therapy, therefore, requires insulin preparations that display a rapid onset of action with corresponding rapid clearance to provide for meal ingestion as well as preparations that can maintain a sustained, peakless profile for basal requirements. Recent efforts in pharmaceutical research have used the concept of rational-based design of the insulin molecule in an attempt to produce preparations that display more ideal pharmacological profiles. Using detailed structural information obtained from X-ray crystallographic studies to guide design strategies and exploit the nonrestrictive synthetic capabilities of recombinant DNA technology, researchers have prepared a number of insulin analogs that display a reduced propensity towards self-association. Clinical evaluations have shown that these so called "monomeric" analogs better mimic the meal-stimulated pharmacokinetics of insulin secretion observed in nondiabetics. Two monomeric insulin analog preparations have successfully obtained regulatory approval and are now commercially available. Efforts to produce optimized basal-acting insulin analogs have lagged behind. While some of these analogs have been engineered using recombinant DNA technology, design strategies in many cases exploit physicochemical properties of insulin other than self-association. One basal insulin analog has recently received regulatory approval. This paper reviews insulin self-association and its relationship to pharmacokinetics and pharmacodynamics. Particular emphasis is placed on the approaches used to manipulate self-assembly resulting in meal-time insulin analogs that display optimal pharmacological properties. Other design strategies used to develop improved basal insulin preparations are also considered.     

A novel protein cross-linking reaction in stressed Neutral Protamine Hagedorn formulations of insulin

The covalent insulin-protamine product molecules formed by heat stress in Neutral Protamine Hagedorn formulations of insulin and the insulin analogue [LysB28,ProB29] were examined by mass spectrometry. The results demonstrated that the covalent cross-link between insulin and protamine was not caused by linkage through the protamine N-terminal amino group, as had been previously thought. Our results indicate that the linkage was formed between the side chain of a protamine arginine and a histidine in the insulin B chain, resulting in a net mass change of -5 Da, compared to the sum of the protamine and insulin molecular masses. A mechanism for this new type of covalent cross-linking reaction is proposed.