Natural peptides as building blocks for the synthesis of large protein-like molecules with hydrazone and oxime linkages.

Methods are known for the production of synthetic protein-like molecules of nonlinear architecture with molecular masses in the 10-20 kDa range. To synthesize such compounds of higher molecular mass and complexity, chemoselective ligation of natural (as opposed to synthetic) peptide building blocks was studied. In preliminary experiments with model peptides, conditions for the formation of peptide oximes were investigated, and their stability at alkaline pH was examined, to resolve a literature controversy. It was found that low pH (down to 2.1) was suitable for polyoxime formation and that the oxime bond was stable for up to 65 h at pH 8 and for more than 2 h at pH 9. Then, using natural peptides, it was found to be possible to synthesize, and characterize by mass spectrometry, nine-component species with molecular masses > 48 kDa. This is about twice the size of homogeneous artificial proteins previously described. Such complex molecules of defined structure are beginning to find applications as vaccine candidates, as radioimmunodiagnostic agents, and as nonviral gene therapy delivery vehicles.

Site-specific attachment of functionalized poly(ethylene glycol) to the amino terminus of proteins.

A convenient method for the construction of site-specifically modified poly(ethylene glycol)-protein conjugates is described. This method relies on the ability to generate a reactive carbonyl group in place of the terminal amino group. If the protein has N-terminal serine or threonine, this can be done by very mild periodate oxidation and generates a glyoxylyl group. A method less restricted by the nature of the N-terminal residue, but which requires somewhat harsher conditions, is metal-catalyzed transamination, which gives a keto group. The N-terminal-introduced reactive carbonyl group specifically reacts, under mild acidic conditions, with an aminooxy-functionalized poly(ethylene glycol) to form a stable oxime bond. Using polymers of different size and shape (linear or multibranched), various conjugates of IL-8, G-CSF, and IL-1ra were constructed and further characterized with respect to their biological activity and pharmacokinetic behavior in rats. Unlike most previous methods, this approach places a single PEG chain at a defined site on the protein. It should therefore be more likely to conserve biological activity when the latter depends on interaction with another macromolecule (unlike enzymic activity which often survives multiple PEGylation).

A fluorescent interleukin-8 receptor probe produced by targetted labelling at the amino terminus.

Interleukin-8 is the most extensively characterised member of the structurally related chemotactic and pro-inflammatory proteins collectively called chemokines. It binds to two closely related members of the seven transmembrane chemokine receptor family found on a variety of leukocyte cell types. In order to study the interaction of interleukin-8 with its receptors, and their distribution, we have produced a fluorescently labelled protein as an alternative to the radioactive 125I-interleukin-8 ligand. Interleukin-8 is naturally produced as two forms, a 72-residue polypeptide by monocytes and a 77-residue form produced by endothelial cells which has an extension of five amino acids at the amino terminal. Both forms are active at nanomolar concentrations, implying that chemical modification to the amino terminus of the 72-residue form will not destroy activity. The 72-residue interleukin-8 sequence starts with a serine residue, which can be oxidised under mild conditions to give a reactive glyoxylyl function which is then reacted with a nucleophilic fluorescein derivative. The site-specifically labelled protein was easily isolated by reverse-phase HPLC. The dissociation constant of the fluorescently labelled interleukin-8 from its receptors on neutrophils was measured by displacement of 125I-interleukin-8 and found to be 10 nM compared to 1 nM for the unmodified protein. The modified protein is highly active in in vitro bioassays using human neutrophils, giving an EC50 of 7 nM in chemotaxis and an EC50 of 0.62 nM for shape change. The binding of the fluorescent protein to neutrophils can also be measured by fluorescent automatic cell sorter (FACS) analysis, and can be competed by unlabelled interleukin-8. The amino-terminal modification of interleukin-8 has produced a reagent which is useful for the quantification of interleukin-8 receptor expression, and will also be useful in monitoring the fate of the ligand after receptor binding.

Site-specific religation of G-CSF fragments through a thioether bond.

A new approach is described for linking, through a thioether bond, the C-terminus of one unprotected polypeptide with the N-terminus of another. Homocysteine thiolactone is attached to the C-terminus of one polypeptide by reverse proteolysis and provides through hydroxylamine treatment a free sulfhydryl group. The alpha-amino group of a second polypeptide is selectively iodoacetylated by reaction with iodoacetic anhydride at pH 6.0 or the N-hydroxysuccinimide ester derivative at pH 7.0. Coupling of the two modified fragments occurs in a spontaneous alkylation reaction under mild conditions. After preliminary experiments with small peptides, this approach was extended to large protein fragments derived from recombinant analogs of G-CSF by enzymatic digestion. This approach provides a means of making head-to-tail protein chimeras or introducing noncoded structural elements into a protein.

Chemo-enzymic backbone engineering of proteins. Site-specific incorporation of synthetic peptides that mimic the 64-74 disulfide loop of granulocyte colony-stimulating factor.

We present the concept of chemo-enzymic backbone engineering of proteins. Recombinant DNA techniques are used to produce appropriate proteins that are enzymically fragmented to give the starting materials. These fragments are modified specifically at their chain termini either enzymically (coupling of a hydrazide to the C terminus) or chemically (periodate oxidation of N-terminal serine to a glyoxylyl function). The modified fragments, which need no side protection whatever, are mixed together and religate themselves spontaneously under mild conditions. The hydrazone bond thus formed can be reduced if desired, which stabilizes the linkage and enhances the flexibility of the local conformation. In this way biologically or chemically derived structures can be incorporated into the protein, and the choice of the chemical ones is free of all of the constraints of the genetic code. We believe that this combined approach gives access to constructions that could not be derived by either recombinant or chemical methods alone. We illustrate the particularity of this concept by the engineered modifications of the 64-74 disulfide loop region of human granulocyte colony-stimulating factor. Analogs constructed include one which, in spite of having a nonpeptide link in its backbone, has full biological activity.

Construction of protein analogues by site-specific condensation of unprotected fragments.

The extreme sensitivity to periodate of 1-amino, 2-hydroxy compounds permits the selective conversion of N-terminal serine and threonine to an aldehydic group. We have used this reaction to construct analogues of human granulocyte colony stimulating factor (G-CSF) by allowing such oxidized peptides to react with others that have had a hydrazide derivative attached to the C-terminus by reversed proteolysis. Two recombinant analogues of G-CSF were used as starting materials. Both had only a single lysine residue (at position 62 and 75, respectively) followed immediately by a serine. Digestion of each analogue by the lysine-specific protease from Achromobacter lyticus gave two fragments, one of which could be N-terminally oxidized and the other converted to the C-terminal hydrazide derivative by reversed proteolysis using the same enzyme. After preliminary studies with model peptides, we first reacted the corresponding peptide pairs together and then, in order to eliminate the 64-74 disulfide loop, fragment 1-62 from the first analogue with fragment 76-174 from the second. Reactions are efficient (up to 80% product based on the oxidized fragment) and take place under very mild conditions. The hydrazone bond can easily be stabilized by reduction with NaBH3CN. This method represents a new, reasonably general route for the construction of large protein chimeras of precisely controlled structure.

Increased activity and stability of poly(ethylene glycol)-modified trypsin.

The reaction of trypsin with activated monomethoxypoly(ethylene glycol) with various molecular masses led to the development of a series of poly(ethylene glycol)-modified trypsins (PEG-trypsins). On determining the catalytic properties of PEG-trypsin using N-benzoyl-L-arginine p-nitroanilide as a substrate, a three- to fourfold increase in the maximal velocity of hydrolysis was found to occur, whatever the size of the PEG moiety used. PEG-trypsin with higher molecular mass moieties showed lower Michaelis constant values. The activation of trypsin was neither reversed by nucleophiles such as hydroxylamine, nor prevented when modification was carried out in the presence of benzamidine or in the presence of the polypeptidic soybean trypsin inhibitor. Chemical modification of about 80% of the free amino groups with PEG chains significantly improved the resistance to heat and detergents. This might result from the formation of a highly hydrogen-bonded structure around the enzyme.

Protein-protein interaction in low water organic media.

Trypsin either modified with polyethylene glycol or as a suspended powder was used to catalyze digestion of protein substrates in benzene in order to get insight into protein-protein interactions in water-immiscible organic media. Depending on whether suspended or soluble trypsin was used, catalysis was found to proceed differently. In the first case, the amount of water in the reaction mixture (up to 1% v/v) appeared to be critical, and adsorption of water from the reaction medium by the protein substrate allowed it to behave as a hydrophilic support material comparable to that involved in immobilized enzymes. In the latter case, the presence of an additional nucleophile was a prerequisite for catalysis to proceed, and thus both water and nucleophile concentrations had some influence on trypsin activity. Phe-NH(2) was the most potent nucleophile for proteolysis catalyzed by polyethylene glycol-modified trypsin in organic media containing 1-2% water (v/v). The organic solvent-soluble enzyme was found to bind reversibly to the protein substrate as a function of both extent of hydration of the reaction medium and time of incubation. The overall results strongly suggested that modified trypsin catalyzed peptide bond hydrolysis at the protein substrate-organic solvent interface. Peptide mapping of bovine insulin digest by reversed-phase high-performance liquid chromatography definitely showed that enzyme-catalyzed proteolysis did occur in organic solvents with a concomitant and significant transpeptidation reaction.

Peptide synthesis catalyzed by polyethylene glycol-modified chymotrypsin in organic solvents.

Chymotrypsin modified with polyethylene glycol was successfully used for peptide synthesis in organic solvents. The benzene-soluble modified enzyme readily catalyzed both aminolysis of N-benzoyl-L-tyrosine p-nitroanilide and synthesis of N-benzoyl-L-tyrosine butylamide in the presence of trace amounts of water. A quantitative reaction was obtained when either hydrophobic or bulky amides of L- as well as D-amino acids were used as acceptor nucleophiles, while almost no reaction occurred with free amino acids or ester derivatives. The acceptor nucleophile specificity of modified chymotrypsin as a catalyst in the formation of both amide and peptide bonds in organic solvents was quite comparable to that in aqueous solution as well as to that of the leaving group in hydrolysis reactions. By contrast, the substrate specificity of modified chymotrypsin in organic solvents was different from that in water since arginine and lysine esters were found to be as effective as aromatic amino acids to form the acyl-enzyme with subsequent synthesis of a peptide bond.