(T0930-02-10) Improved Identification of Host Cell Proteins in Monoclonal Antibodies by Combining Filter-Aided Sample Preparation and Native Digestion

Purpose: Host cell proteins (HCPs) are process impurities that may compromise the safety and stability of monoclonal antibody (mAb) therapeutics, even at trace levels. While enzyme‑linked immunosorbent assay (ELISA) provides high-sensitivity quantification of total HCP content, it does not identify individual HCPs, limiting risk assessment. Liquid chromatography-mass spectrometry (LC-MS) enables HCP identification but is restricted by its dynamic range, often missing low-abundance HCPs in high-concentration mAb matrices. There is a significant need for improved sample preparation methods that enhance HCP enrichment and identification regardless of molecular weight or abundance, enabling a more comprehensive understanding and control of HCP risk in biopharmaceutical development.

Methods: We develop a strategy that combines filter-aided sample preparation (FASP) with native digestion for HCP analysis. In this workflow, mAb samples containing HCPs undergo buffer exchange using a 10 kDa molecular weight cutoff (MWCO) filter, and trypsin is added directly to the filter for overnight digestion under native conditions. After digestion, centrifugation separates intact mAbs from the resulting HCP peptides, which are then collected, reduced, dried, and prepared for LC-MS/MS analysis. Method performance is evaluated using an mAb product containing spiked proteins (12–470 kDa) at concentrations down to 0.5 ppm and by analyzing the NISTmAb reference standard. Comparative experiments are performed with standard native digestion in tube, and HCP identification and quantitation are performed using a high‑resolution Orbitrap mass spectrometer and database-assisted identification.

Results: The native digestion on the filter (NDF) workflow demonstrates superior depletion of mAbs and enables unbiased capture of HCP peptides, regardless of their molecular weight. Compared to tube-based native digestion, the NDF workflow eliminates the need for a heat-induced precipitation step, reducing the potential loss of heat-labile or co‑precipitating HCPs and streamlining sample preparation. In protein spike-in experiments, the NDF method sensitively detects all six standard proteins ranging from 12 kDa to 470 kDa at 1 ppm, and shows strong performance even at the 0.5 ppm level, without molecular weight bias. In contrast, standard native digestion in solution favors detection of only low-molecular-weight proteins and exhibits poorer sensitivity for larger proteins. For the NISTmAb standard, the NDF workflow identifies 155 more HCPs than standard native digestion, with quantitation down to 0.05 ppm. Overall, the NDF method improves both the number and sensitivity of HCP identifications, especially for low‑abundance and high-molecular-weight species.

Conclusion: The integration of filter-aided sample preparation with native digestion significantly enhances the comprehensiveness and sensitivity of HCP identification in monoclonal antibody therapeutics by LC-MS/MS. This workflow achieves unbiased enrichment of HCPs across a wide molecular weight range and can detect and quantify HCPs at sub‑ppm levels, overcoming limitations of standard native digestion method. By removing the need for heat-induced precipitation, it minimizes the risk of losing heat-labile HCPs and simplifies the process. The method holds promise as a robust platform for routine and deep HCP characterization, directly supporting process development and quality control efforts in biopharmaceutical manufacturing. Further integration with advanced enrichment and separation techniques may enable even deeper HCP profiling in the future.

The native digestion on the filter (NDF) experimental workflow for HCP identification includes filter-aided sample preparation and native digestion. During native digestion on the filter, mAbs remain intact due to disulfide bonds, while HCPs are cleaved into peptides. Subsequent centrifuge filtration separates the intact mAbs from the HCP peptides, facilitating HCP enrichment for comprehensive identification, regardless of peptide size.

The NDF experiment detects all six protein spikes, ranging from 12 kDa to 470 kDa at 1 ppm, with no bias towards smaller HCPs, allowing for an unbiased analysis. In contrast, ND only detects the Mb and Cyt c spikes, losing detection for larger proteins. At a 0.5 ppm spike level, NDF can still detect all spikes except the largest 470 kDa FN and provides nearly twice the unique peptide identification for Mb and Cyt c compared to ND, demonstrating a lower detection limit and offering greater identification confidence in the NDF method.

The Venn diagram of total HCPs identified in NISTmAb shows that NDF detects substantially more unique HCPs than ND.