Illuminating the Dark Host Cell Proteome: A host cell protein coverage method for LC-MS impurity assays.

Host cell protein (HCP) analysis by liquid chromatography-mass spectrometry (LC-MS) enables identification and quantification of individual impurities in biopharmaceutical products, supporting risk-based impurity assessment. However, unlike enzyme-linked immunosorbent assay (ELISA) methods, systematic evaluation of method-specific HCP coverage for LC-MS assays remain poorly defined. In particular, limited attention has been given to proteins that are inherently undetectable under specific analytical conditions, potentially creating blind spots in impurity characterization. Here, we present DARK-COV, a transparent and generalizable in silico pipeline for assessing theoretical HCP coverage of LC-MS-based impurity assays. The pipeline combines physicochemical filtering of in silico-digested proteomes-based on peptide hydrophobicity, charge state, and mass-to-charge ratio-with assay-specific LC-MS parameters to predict which peptides and proteins can be detected. Proteins predicted to yield fewer than two detectable peptides were classified as Dark Host Cell Proteins. The pipeline was benchmarked against a large-scale experimental dataset derived from more than 15,000 LC-MS analyses curated in HCPedia™, a comprehensive database of empirically observed HCP peptides generated using the same LC-MS workflow. As a proof-of-concept demonstration, DARK-COV was applied to four industrially relevant expression systems-Chinese Hamster Ovary (CHO) cells, Escherichia coli, Human Embryonic Kidney (HEK) cells, and Spodoptera frugiperda (Sf9) cells-revealing that only 1.0-5.2% of proteins across these proteomes were predicted to be theoretically undetectable under idealized in silico conditions. These predictions present an upper bound on detectability and do not account for matrix effects, ion suppression, or concentration-dependent detectability in real samples. Comparison with experimental observations across > 15,000 LC-MS datasets demonstrated strong concordance with DARK-COV predictions, supporting the classification of proteins that are likely to fall outside the detectable range of the LC-MS assay. Functional annotation and comparison against known high-risk HCPs indicated that the majority of Dark Host Cell Proteins are low-molecular-weight proteins with limited relevance to biopharmaceutical impurity risk. Overall, this pipeline provides a reproducible, method-specific approach for quantifying theoretical coverage and comparing predictions with experimental observations in LC-MS assays. By identifying analytical blind spots, the strategy supports risk-based evaluation of undetected HCPs and offers a rational foundation for regulatory justification of the LC-MS-based HCP impurity analyses described in USP Chapter 1132.1.