Genentech, Inc.
The purpose of this workshop was to highlight some of the unique species produced during the biosynthesis of recombinant DNA-derived proteins. Bernard Violand from Monsanto, Hsieng Lu from Amgen, and Eleanor Canova-Davis from Genentech discussed experiences in their respective laboratories. The first two speakers emphasized protein heterogeneity that was a result of mistranslational events, especially in Escherichia coli systems. The last talk described the isolation and identification of a novel cross-link involving cysteine residues and summarized some of the variant proteins found to date.
An understanding of the modifications which may be present in recombinant proteins is essential for the production of therapeutically useful proteins. With this knowledge a better control of product quality including process validity can be attained. Fermentation and purification schemes can then be devised to remove these variants which usually are present in small amounts. Of course, the realization of these goals is dependent upon the utilization of state of the art analytical techniques. This workshop clearly illustrated the role that peptide mapping procedures, mass spectrometry, high performance liquid chromatography (HPLC), and electrophoretic methods play in this area.
Mistranslations can be a result of aberrant initiation, a frame shift, a missense error, tRNA hopping, or a termination bypass (Santos and Tuite, 1993). The detection and identification of specific translational errors was discussed. It is known that norleucine can be charged onto Met-tRNA and become incorporated into proteins biosynthesized in E. coli. Norleucine is a product of the leucine pathway, and hence its synthesis can be reduced by inhibiting the pathway. Alternatively, methionine can be supplied in sufficient quantities to successfully compete during tRNA charging. Detection of norleucine incorporation is usually via reversed-phase HPLC since its incorporation results in a more hydrophobic molecule. Identification is accomplished by mass spectrometry (since the variant has a mass 18 units less than expected) and amino-terminal sequence analysis of the modified peptides generated by peptide mapping procedures. Quantification is best accomplished by amino acid analysis (Bogosion et al., 1989; Lu et al., 1988).
A number of mistranslations owing to codon mismatch during protein synthesis have been reported. For example, the stop codon UGA has been recognized by both suppressor and normal tRNAs with the subsequent incorporation of tryptophan and the continuation of protein synthesis to the next stop codon resulting in a larger molecular weight species which can be easily distinguished by mass spectrometry. This phenomenon was observed in the E. coli expression of both neurotrophin-3 (Hui et al., 1995) and platelet-derived growth factor B (unpublished observations). In both cases an odd tryptophan-containing peptide was isolated from the peptide map. A similar occurrence has been observed at the stop codon UAG with the subsequent incorporation of glutamine (codons CAA or CAG) as seen in the biosynthesis of bovine growth hormone in E. coli (Bogosian et al., 1991). These read-through events can be eliminated by the use of two stop codons, one of which is UAA, a preferred E. coli stop codon. Glutamine has also been incorporated instead of histidine (codon CAU) in granulocyte colony stimulating factor leading to a species of lesser charge which can be detected by ion-exchange HPLC (Lu et al., 1994). In this particular case the substitution resulted in peptide fragments from an endoproteinase Glu-C digestion that were 9 mass units less than expected. Amino-terminal sequence analyses confirmed the substitutions. These missense translations may be difficult to isolate and characterize depending upon the specific amino acid substitution.
A novel translational hop which generated a bovine placental lactogen molecule missing two consecutive amino acids from its internal sequence was described (Kane et al., 1992). An abnormal peptide was isolated in place of two normal peptides from a tryptic digest. It was determined that the structural gene had an arginine codon (AGG), which is rare for E. coli, leading to this two amino acid deletion. Replacement of the rare arginine codon to a preferred E. coli codon corrected the problem. No abnormal tryptic peptide was observed and no mass indicative of the deletions was seen in the intact molecule. As an added benefit the expression level increased significantly.
The identification of the unusual amino acid (E)-N- acetyllysine was also presented (Violand et al., 1994). This is a rare amino acid which abolishes the side chain positive charge of lysine and is present in several eukaryotic proteins, for example, in histones. It has not been observed in prokaryotic proteins. Since it does lead to a charge difference, techniques such as anion-exchange HPLC and immobilized pH gradient electrophoresis were effective in its detection. Electrospray mass spectrometry revealed the presence of a protein variant of porcine growth hormone produced in E. coli with 42 mass units greater than expected. Amino acid analysis after complete enzymatic digestion, necessary since acid hydrolysis would destroy the modification, confirmed the assignment as (E)-N-acetyllysine.
Thirty years ago studies concerning the chemistry of sulfur-transferring enzyme systems suggested that cystine trisulfide existed in biological systems. A recombinant human growth hormone variant containing a trisulfide cross- link has been isolated from E. coli fermentations (Jespersen et al.). The increased hydrophobicity of this variant contributed to its isolation by reversed-phase HPLC. A combination of tryptic mapping procedures, tandem mass spectrometry, and elemental composition of the modified peptide led to the assignment of the observed plus 32 mass units as one sulfur rather than two oxygen atoms (unpublished observations).
Other curious discoveries alluded to included an arginine to lysine substitution in E. coli expression of IGF-I owing to arginine codons which are rare for the bacteria (Seetharam et al., 1988), concatenated dimers in bovine growth hormone (Tou et al., 1993), formylation of amino- terminal methionine residues (Clogston et al., 1992), glutathione adduct formation with the free sulfhydryl groups in F(ab) antibody fragments biosynthesized in E. coli (unpublished observations), hydroxamate formation at asparagine and glutamine residues during hydroxylamine cleavage of fusion proteins (Canova-Davis et al., 1992), lysine modification to hydroxylysine in non-collagenous proteins (Molony et al., 1995), lanthionine formation at disulfide bonds (Toren et al., 1988), serine to glycine conversion at the NH2-terminus of human nerve growth factor (Canova-Davis et al., 1994), urea exposure during processing leading to carbamylation of amino groups (Stark et al., 1960), and finally, a tyrosine to glutamine substitution via DNA mutations in Chinese hamster ovary cells (Harris et al., 1993).
References
Return to the The ABRF Home Page