The synthesis of triaminoacyl-insulins and the use of the t-butyloxycarbonyl group for the reversible blocking of the amino groups of insulin

Intrinsic Viscosity of Proteins and Platonic Solids by Boundary Element Methods

The boundary element (BE) method is used to implement a very precise computation of the intrinsic viscosity for rigid molecules of arbitrary shape. The formulation, included in our program BEST, is tested against the analytical Simha formula for ellipsoids of revolution, and the results are essentially numerically exact. Previously unavailable, very precise results for a series of Platonic solids are also presented. The formulation includes the optional determination of the center of viscosity; however, for globular proteins, the difference compared to the computation based on the centroid is insignificant. The main application is to a series of 30 proteins ranging in molecular weight from 12 to 465 kD. The computation starts from the crystal structure as obtained from the Protein Data Bank, and a hydration thickness of 1.1 Å obtained in previous work with BEST was used. The results (extrapolated to an infinite number of triangular boundary elements) for the proteins are separated into two groups:  monomeric and multimeric proteins. The agreement with experimental measurements of the intrinsic viscosity in the case of monomeric proteins is excellent and within experimental error of 5%, demonstrating that the solution and crystal structure are hydrodynamically equivalent. However, for some multimeric proteins, we observe strong systematic deviations around -20%, which we interpret as a systematic deviation of the solution structure from the crystal structure. A possible description of the structural change is deduced by using simple ellipsoid model parameters. A method to obtain intrinsic viscosity values for proteins to 1-2% accuracy (better than experimental error) on the basis of a single BE computation (avoiding the need for an extrapolation on the number of surface triangles) is also presented.

Site-specific fluorescein labeling of human insulin

Three fluorescein derivatives of human insulin (HI, 1) labeled at positions N(αA1) , N(αB1) and N(εB29) respectively, were synthesized using an N-trifluoroacetyl-based protecting group scheme. The Tfa protecting group introduced by reaction with ethyl trifluoroacetate was found to be stable in aqueous and organic media and efficiently removed under mild basic conditions.

Synthesis and purification of NB1-palmitoyl insulin

A procedure for synthesizing NB1-palmitoyl insulin for incorporation into liposomes for targeting to hepatocytes was developed. The amino group of the first amino acid phenylalanine on the B chain (B1) of insulin was selected for conjugation with palmitic acid in anticipation that its binding to the insulin receptor would be preserved. Two other free amino groups present in insulin, the first amino acid glycine on the A chain (A1) and the 29th amino acid lysine on the B chain (B29), were first protected with a t-butoxycarbonyloxy (t-Boc) group to yield NA1, B29-di-(t-Boc) insulin. The identity of this di-(t-Boc) insulin was confirmed by amino acid analysis as well as by enzyme hydrolysis coupled with matrix-assisted laser-desorption time of flight mass spectrometry (MALDI-TOF MS). NA1,B29-Di-(t-Boc) insulin was then reacted with the N-hydroxysuccinimide ester of palmitic acid, followed by deblocking the t-Boc groups, to yield NB1-palmitoyl insulin, the structure of which was further confirmed by MALDI-TOF MS analysis. NB1-palmitoyl insulin was found to interact with the insulin receptor on fat cells, thereby catalyzing the conversion of [14C]glucose into lipids, at reduced efficiency (30-40%).

Characterization of glucose-mediated insulin release from implantable polymers

We characterized a glucose-sensitive, controlled-release insulin delivery system. Insulin release rates increased when glucose was perfused in the release media surrounding the matrix. The system was composed of solid, particulate insulin, incorporated into an ethylene-vinyl acetate copolymer (EVAc) matrix. Feedback control was mediated by the glucose oxidase enzyme immobilized to Sepharose beads, which were incorporated along with insulin into the EVAc matrix. When glucose in solution entered the insulin delivery system, gluconic acid was produced, causing a drop in the microenvironmental pH of the matrix. This fall in pH resulted in a rise in insulin solubility and consequently a rise in the insulin release rate from the matrix. Insulin concentrations increased in vitro and in vivo in response to glucose infusion. The increased insulin release was shown to consist of a finite pulse of insulin that required an optimal recovery period of 1 h to achieve a maximal repeated response to a glucose stimulus. Repeated pulses were demonstrated over a 4 h period. An optimum enzyme ratio was also determined.

Enzymatically controlled drug delivery

An approach for providing feedback control for polypeptide drugs in a polymeric controlled-release system uses a trigger molecule and a polymer-bound enzyme that, in the presence of that trigger molecule, will cause an acid or a base to form. When the pH inside the polymer system changes, the solubility of the drug shifts dramatically, which changes the diffusion or dissolution driving force, and hence the release rate changes correspondingly. This concept was tested using a controlled-release system of ethylene/vinyl acetate copolymer containing insulin and immobilized glucose oxidase. The enzymatic reaction of glucose to gluconic acid reduces the pH in the polymer microenvironment. Since insulin solubility increases with decreasing pH (at physiologic pH, this is true for an insulin with an isoelectric point of 7.4 or higher), the release of insulin increases in response to glucose concentration. The feasibility of this concept has been shown using trilysyl insulin with an isoelectric point of 7.4. Multiple exposures to buffered glucose solutions over several weeks caused insulin release to reversibly increase during each exposure. Polymer-implanted diabetic rats infused with glucose solutions showed a significant increase in insulin concentration in 30 min-an effect not observed in three different sets of control rats.

Semisynthesis and biological properties of the [B24-leucine]-, [B25-leucine[- and [B24-leucine, B25-leucine]-analogues of human insulin

Trypsin-catalyzed coupling of porcine desoctapeptide-insulin with synthetic octapeptides produced the [LeuB24]- (I), [LeuB25]- (II) and [LeuB24, LeuB25]- (III)analogues of human insulin. I, II and III displayed respectively 20--30%, 1--2% and 0.5% of the receptor binding activity of the normal hormone. Biological activities of these analogues seemed to be proportional to their binding potencies when assayed in vitro, while in an in vivo assay analogue I was fully active and II exhibited 10--20% of normal activity. III was less active than II in all assays tested.

Semisynthesis and biological activity of porcine [LeuB24]insulin and [LeuB25]insulin

Two analogs of porcine insulin with substitutions of leucine for phenylalanine in the COOH-terminal region of the insulin B chain have been prepared by a combination of solid-phase synthesis and semisynthesis. Solid-phase synthesis of the substituted octapeptides B23-B30 bearing the trifluoracetyl group on lysine-B29, enzymatic coupling of the octapeptides to bis(tertiary-butyloxycarbonyl)desoctapeptide insulin by trypsin, and deprotection of the corresponding adducts in formic acid and piperidine resulted in two insulin derivatives, one with leucine at position B24 and the other with leucine at position B25. These analogs had only about 10% and 1%, respectively, of the activity of porcine insulin in competing for the binding of [125I]iodoinsulin to both rat adipocytes and human IM-9 lymphocytes. The relative potencies of the analogs in stimulating glucose oxidation by rat adipocytes decreased in the order porcine insulin > [LeuB24]insulin > [LeuB25]insulin. However, at high concentrations both analogs had full agonists activity. Experiments in which the semisynthetic insulins were mixed with the native hormone showed that [LeuB24]insulin, but not [LeuB25]insulin, was an active antagonist of insulin action. These results suggest that the antagonistic activity of a human insulin variant having leucine at position B24 or B25 can be assigned to the molecule with the sequence Gly-Leu-Phe-Tyr (residues B23-B26) in its active site.

Synthesis and properties of Clostridium acidi-urici (Leu2)-ferredoxin: a function of the peptide chain and evidence against the direct role of the aromatic residues in electron transfer

Tyrosyl or other aromatic residues generally occur in two conserved positions in the peptide chain of clostridial-type ferredoxins and have been implicated in the electron transfer function of these iron-sulfur proteins. We have prepared and determined some of the properties of a derivative of Clostridium acidi-urici ferredoxin, [Leu(2)]-ferredoxin, in which a leucyl residue has been substituted for the tyrosyl residue in position 2 from the amino terminus. [Leu(2)]-ferredoxin is fully active as an electron carrier in two biological assays, the phosphoroclastic enzyme system and the ferredoxin-dependent reduction of cytochrome c in the presence of ferredoxin-TPN reductase and TPNH. Quantitative electron paramagnetic resonance experiments indicate that [Leu(2)]-ferredoxin accepts nearly two electrons upon enzymatic reduction by pyruvate-ferredoxin oxidoreductase and an excess of pyruvate. If electron transfer to an iron-sulfur cluster is the rate-limiting step in the assays used, and if the rate of electron transfer through Tyr(30) is not much faster than through Tyr(2), these results indicate that the primary pathway of electron transfer in clostridial-type ferredoxins is not via Tyr or other aromatic amino-acid residues. The syntheses of other ferredoxin derivatives with amino-acid substitutions or deletions in positions 1 and 2 indicate that a large bulky residue, but not necessarily an aromatic residue, is needed in position 2 for the stability of this ferredoxin. The residue in position 2, therefore, appears to act as a hydrophobic shield for an iron-sulfur cluster.

Human insulin: facile synthesis by modification of porcine insulin

Human insulin differs from porcine insulin by a single amino acid- the carboxyl terminal residue of the B chain. By means of chemical and enzymatic treatment, it is possible to remove quantitatively and selectively the carboxyl terminal octapeptide from porcine insulin B chain. This fragment can be replaced by an analogous synthetic human octapeptide to give a protein which is identical to human insulin by a number of criteria. By this method, human insulin can be prepared on a large scale simply and inexpensively from porcine insulin. The method is also useful for preparing specifically labeled radioactive human insulin, as well as insulins with modified amino acid sequences, for research purposes.

The acetylation of insulin

The acetylation of the free amino groups of insulin was studied by reaction of the hormone with N-hydroxysuccinimide acetate at pH6.9 and 8.5. The products formed were separated by chromatography on DEAE-Sephadex and were characterized by isoelectric focusing, by end-group analysis, by the incorporation of [(3)H]acetyl groups in the molecule, and by treatment with trypsin that had been treated with 1-chloro-4-phenyl-3-toluene-p-sulphonamidobutan-2-one (;tosylphenylalanyl chloromethyl ketone'). Three monosubstituted products, two disubstituted products and one trisubstituted derivative were prepared. The alpha-amino groups of the terminal residues and the in-amino group of the lysine-B29 were the sites of reaction. Acetylation of any of the free amino groups did not affect the biological activity of insulin. It was demonstrated, however, that substitution at the glycine-A1 amino group by the larger residues, acetoacetyl or thiazolidinecarbonyl, produced a decrease in biological activity. Modification of the lysine-B29 or phenylalanine-B1 amino groups with these larger reagents did not affect the biological activity. Modification of the phenylalanine-B1 amino group by any of the three substituents resulted in a large decrease in the affinity of insulin for anti-insulin antibodies raised in the guinea pig. Modification of the other two amino groups did not affect the reaction with antibody. These observations are correlated with the tertiary structure of insulin.

Acetoacetylation of insulin

Insulin was treated with diketen at pH6.9. The reaction mixture was resolved into four components by DEAE-Sephadex chromatography. The first component was unchanged insulin. The second and third components were shown by end-group analysis to be substituted on phenylalanine B-1 and glycine A-1 respectively. The fourth component was disubstituted on both phenylalanine B-1 and glycine A-1. The in-amino group of lysine B-29 was not involved in the reaction at low reagent concentrations. The purity of these derivatives was checked by their electrophoretic behaviour and by measurement of the rate of their reaction with trinitrobenzenesulphonic acid. The hormonal activity of the derivatives was determined. The effect of the modifications on the hormonal activity and the tertiary structure of insulin is discussed.