Single-cell proteomics reveals proteome remodeling and cellular heterogeneity during NGF-induced PC12 neuronal differentiation

1 Abstract Single-cell proteomics (SCP) enables direct measurement of cellular heterogeneity during dynamic biological processes. Here, we applied an SCP workflow to investigate proteome diversity during nerve growth factor (NGF)–induced differentiation of PC12 cells. Differentiated PC12 cells are highly adherent and prone to aggregation, complicating single-cell sample preparation. To address this challenge, sample handling was optimized using gentle dissociation, anti-adhesive conditions, and rapid processing immediately prior to cell isolation. Individual cells were deposited using a refined thermal inkjet (TIJ) dispensing system, enabling accurate single-cell placement with minimal sample loss. Inclusion of the mild nonionic surfactant n-dodecyl-β-D-maltoside (DDM) improved recovery of membrane-associated and other low-solubility proteins. Coupled with high-sensitivity liquid chromatography–ion mobility–mass spectrometry, this workflow consistently quantified approximately 2,000–3,000 proteins per cell across differentiation stages. Single-cell proteomic profiles acquired over the differentiation time course revealed clear separation between undifferentiated and NGF-treated cells by Day 6. At later stages (Days 4–6), cells further partitioned into two distinct subpopulations with protein expression patterns not evident in bulk measurements. Dimensionality reduction and non-negative matrix factorization identified multiple proteomic states coexisting within the same differentiation stages, characterized by coordinated differences in pathways related to intracellular trafficking, protein translation, and neuronal structural organization. Together, these results show that while global proteome remodeling during PC12 differentiation is captured in both bulk and single-cell data, single-cell proteomics uniquely resolves functionally distinct cellular subpopulations that are masked in population-averaged analyses.