Gene expression measurements are typically performed on a fixed-weight aliquot of RNA, which assumes that the total number of transcripts per cell stays nearly constant across all conditions. In cases where this assumption does not hold (e.g., when comparing cell types with different cell sizes) the expression data provide a distorted view of cellular events. Assuming constant numbers of total transcripts, increases in expression of some RNAs must be compensated for by decreases in expression of others.
Therefore, University of Regensburg researchers propose calibrating gene expression data to an external reference point, the number of cells in the sample, using whole-cell spike-ins. In a systematic dilution experiment, they mixed varying numbers of human cells with fixed numbers of Drosophila melanogaster cells and scaled the expression levels of the human genes relative to those of the Drosophila genes. This approach restored the original gene expression ratios generated by the dilutions. They then used Drosophila whole-cell spike-ins to uncover non-symmetric gene expression changes, in this case much larger numbers of induced than repressed genes, under perturbations of the human cell line P493-6. Drosophila whole-cell spike-ins are an experimentally and computationally easy and low-priced method to derive mRNA fold changes of absolute abundances from RNA sequencing (RNA-Seq) and quantitative real-time PCR (qPCR) data.
Isolation of RNA from different numbers of human cells and constant numbers of Drosophila melanogaster spike-in cells simulates changes in total RNA amounts per cell
(A) Experimental setup. Condition A: 500,000 cells from human B-cell line P493–6; Condition B: 1,000,000 P493–6 cells; and Condition C: 2,000,000 P493–6 cells. Samples were spiked with 100,000 S2 D. melanogaster cells, mixed, lysed, and total RNA was extracted. RNA was subjected to RNA sequencing (RNA-Seq) and quantitative real-time PCR (qPCR). Each condition (A, B, and C) consisted of three replicate samples. (B) Different cell numbers spiked with constant numbers of D. melanogaster cells simulate non-symmetric gene expression changes. The global ratios were 2:1 between Conditions B and A as well as between Conditions C and B, and 4:1 between Conditions C and A. (C) After taking a fixed-weight aliquot of the total RNA consisting of human and D. melanogaster transcripts for sequencing, the fraction of reads originating from D. melanogaster transcripts inversely correlated with the amount of human total RNA. (D) After calibration of human gene expression levels to D. melanogaster gene expression levels, the original ratios between Conditions A, B, and C were restored.