|Titel:||Development of robust workflows for quantitative therapeutic protein characterization using multi-attribute methodology (MAM)||Sprache:||Englisch||Autor*in:||Widgren Sandberg, Magdalena||Schlagwörter:||multi-attribute methodology; post-translational modifications; protein characterization; data independent acquisition; therapeutic proteins; peptide mapping; mass spectrometry||Erscheinungsdatum:||2021-06-24||Tag der mündlichen Prüfung:||2021-10-15||Zusammenfassung:||
During therapeutic protein production, proteoforms are formed that can affect the efficacy and safety of the drug. It is therefore of high importance to characterize, quantify and remove such critical proteoforms so that their levels may be maintained low. Quantitative bottom-up proteomics using liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) is a powerful method to characterize and quantify such proteoforms with high sensitivity. This enables connecting specific modifications located at defined amino acids within the sequence of a proteoform (attributes) to functional effects on the therapeutic protein. Although such methods, commonly called multi-attribute methods (MAMs) in the pharmaceutical industry, yield a high amount of information about the therapeutic protein, there are still some aspects that are not covered by conventional MAMs. Those are for example measurements in non-purified samples where signal interference from host cell proteins is an issue and the structural aspects of some proteoforms. The main aim of this thesis was to improve current bottom-up based proteomics approaches for quantification of low abundant proteoforms in therapeutic proteins with regard to the assay performance in non-purified samples and the kind of protein modifications that are covered. A MAM based on data-independent acquisition (DIA) quantification mode was developed to circumvent signal interference from host cell proteins and to assure identification and quantification of specific amino acid modifications located to single defined amino acid sites. The method was optimized for the quantification of proteoforms of a biosimilar of the therapeutic monoclonal antibody (IgG) adalimumab including a DIA-spectral library consisting of 15 out of 20 asparagine deamidation sites, 2 out of 29 glutamine deamidation sites, 5 out of 5 methionine oxidation sites and 12 N-glycan variants. A cell culture filtrate of the IgG was analyzed by LC-MS/MS and quantification of the modification degrees was compared using the DIA based method or a data-dependent acquisition (DDA) based method, the latter being the conventionally used MAM approach. The quantitative results obtained with the DIA based method did not differ from those obtained by DDA by more than 20% for modification degrees measuring > 2%. The two approaches gave similar precision for quantification of the modification degree in technical as well as biological replicates, applying the same cutoff level. Signal interference in the precursor spectra was identified in three deamidated peptides measuring under 2%. This led to overestimation of the deamidation degree of one site and possibly to the high CV values measured for the other two sites. It was demonstrated that the signal interference could be avoided in all three peptides by basing the quantification on extracted ion chromatograms (XICs) from fragment ions from the DIA based MS method instead of the precursor ions from a DDA based MS method. When the two quantification approaches were applied on low-abundance deamidated peptides in a purified IgG sample which had been subjected to forced degradation at pH 9 and 37 ֯C for two weeks, the DDA based approach did, on the contrary, show a higher precision than DIA. This could be attributed to the precursor ions giving a higher assay sensitivity than the fragment ions, when signal interference was not an issue. Further, the linearity of the DIA based MAM was demonstrated for quantification of methionine oxidation and was found to be high with R2 values > 0.99 when the fragments used for quantification had been optimized. In this study, in addition, a novel modification of +70 Da identified in samples of rhGM-CSF expressed in E. coli was characterized by chemical assays in combination with MS. The modification could be located at the protein N-terminal and at lysine residues. Further, the modification likely contained a carbonyl group since its mass increased by 2 Da by reaction with borane pyridine complex, and it reacted with 2,4-dinitrophyenylhydrazine. C4H6O was suggested as elemental composition of the modification, and the reaction product of crotonaldehyde through Michael addition as chemical structure. The presence of a carbonyl moiety indicated that the modification may be critical for the protein’s therapeutic effect since protein carbonylation has been connected to various diseases and protein degradation. Furthermore, a MAM method was optimized for quantification of disulfide scrambling in rhGM-CSF by selection of proteolytic enzymes generating disulfide linked peptides with few disulfide bonds and with appropriate peptide sizes for bottom-up proteomics. By combining the enzymes GluC and Chymotrypsin, mainly disulfide linked peptides with one disulfide bond were generated with peptide sizes of 9-15 amino acids. With the MAM method, disulfide scrambling was quantified in samples subjected to forced degradation, in a refolding sample stored under reducing conditions and in two production process development samples. Protein deamidation was followed in samples subjected to forced degradation over a 3-week period. The degree of +70 Da modified protein sites was quantified in the process development samples, together with amino acid substitution from isoleucine to valine and methionine oxidation. The quantitative accuracy of the disulfide scrambling assay may be affected by differences in ionization efficacies between the different disulfide linked peptides. In summary, several MAMs were developed in this work covering aspects which are not covered by conventional quantitative bottom-up approaches and which has the potential to improve the quality of therapeutic protein production processes. Further, a novel modification was identified in rhGM-CSF produced in E. coli which may be critical for the protein’s therapeutic effect.
|Enthalten in den Sammlungen:||Elektronische Dissertationen und Habilitationen|
geprüft am 06.12.2021
geprüft am 06.12.2021