Understanding exactly which proteins are present in the cells lining and supporting brain blood vessels matters enormously for neurological disease research — from stroke to Alzheimer's to blood-brain barrier dysfunction. For years, scientists have relied on gene expression data as a proxy for protein content, an assumption this work puts under serious scrutiny.
Researchers constructed a comprehensive reference protein atlas of the adult mouse brain vasculature, cataloguing protein expression across the distinct cell types that compose cerebral blood vessels — including endothelial cells, pericytes, smooth muscle cells, and astrocytic endfeet. Using high-sensitivity mass spectrometry approaches capable of detecting low-abundance proteins at the single-cell level, the team demonstrated that transcriptomic data derived from single-cell RNA sequencing frequently fails to accurately predict actual protein abundance. The correlation between messenger RNA levels and final protein expression was poor across many functionally critical genes, meaning that cell identity and function inferred solely from RNA sequencing may be systematically mischaracterized.
This finding strikes at a foundational assumption in modern neurovascular biology. The field has invested heavily in scRNA-seq atlases as definitive maps of cellular identity, yet protein — not RNA — is what executes cellular function. The mismatch documented here is not trivial: in contexts like drug transport across the blood-brain barrier or signaling between pericytes and neurons, protein-level accuracy is essential for therapeutic targeting. The atlas itself represents a valuable resource for researchers modeling neurological diseases or screening compounds for CNS delivery. However, notable limitations apply: this is a mouse model dataset, and translational relevance to human brain vasculature requires dedicated human tissue studies. As a reference dataset rather than an interventional study, it is hypothesis-generating rather than causal. Still, as a methodological corrective and resource tool, this contribution is more than incremental — it challenges how the field interprets its own foundational data.