Multivariate Control of Transcript to Protein Variability in Single Mammalian Cells

A long-standing question in quantitative biology is the relationship between mRNA and protein levels of the same gene. Here, we measured mRNA and protein abundance, the phenotypic state, and the population context in thousands of single human cells for 23 genes by combining a unique collection of cell lines with fluorescently tagged endogenous genomic loci and quantitative immunofluorescence with branched DNA single-molecule fluorescence in situ hybridization and computer vision. mRNA and protein abundance displayed a mean single-cell correlation of 0.732 at steady state. Single-cell outliers of linear correlations are in a specific phenotypic state or population context. This is particularly relevant for interpreting mRNA-protein relationships during acute gene induction and turnover, revealing a specific adaptation of gene expression at multiple steps in single cells. Together, we show that single-cell protein abundance can be predicted by multivariate information that integrates mRNA level with the phenotypic state and microenvironment of a particular cell.