Presented at The Bioprocessing Summit 2018, Boston, MA, USA
In the production of biopharmaceuticals, sequence variants (SVs) are protein species that contain unintended changes to the target amino acid sequence. The presence of SVs can adversely affect the safety and efficacy of biopharmaceutical preparations, and therefore an analytical method that enables sensitive detection of SVs would represent a significant step toward ensuring product quality. Mass spectrometric methods enable identification of SVs, but the minimization of false positives remains a challenge during data analysis. We present an automated processing workflow for LC-MS/MS data that provides comprehensive characterization and sensitive quantification of sequence variants and describe an effective strategy to minimize the number of false positive and false negative identifications.
Peptide mapping was performed using the NIST monoclonal antibody (mAb). After denaturation, mAb samples were reduced, alkylated, and trypsin-digested. The obtained peptide mixture was analyzed without pretreatment using reversed-phase HPLC separation and mass spectrometric detection.Raw LC-MS/MS data was analyzed using the dedicated software platform Genedata Expressionist® (Genedata AG, Basel, Switzerland).
Traditional approaches to SV identification rely solely on the efficacy of search engines. We developed a data analysis strategy that incorporated a systematic reduction of the space searched for SV candidates. Briefly, all peptides (both unmodified and with expected post-translational modifications) belonging to the target mAb sequence were identified with high confidence and excluded from the search space before SV analysis. This reduction of the search space dramatically reduced the number of false positive identifications.Subsequent steps in the automated workflow were created to search for specific groups of misincorporations; such as Phe®Tyr and those occurring within under-alkylated peptides. These targeted searches were key to identifying variants that might otherwise have been overseen (false negatives). Applied to an NIST antibody sample, our workflow generated only a few dozen SV candidates. Investigation of these potential SVs — through manual validation of MS/MS spectra — required significantly less time than would have been required to investigate the large number of candidates generated using traditional approaches.