Gene expression microarray analysis of small, purified cell subsets

Gene expression microarray technology is potentially capable of examining all of the cellular processes at the mRNA level at a given moment. One major challenge in its applicability is that most biological samples are cell mixtures and the cell type of interest is often a minor cell subset. Using well-defined mixtures of model cell types with different cell ratios we found that the overall gene expression profile (GEP) of mixed cell populations was the weighted average of the GEPs for each cell subpopulation in the cell mixture. Thus, without applying any cell separation the cell type in majority dominated the overall GEP of the sample while the GEPs of minor cell subsets were diminished. We showed that the functional threshold for the necessary purity of a cell type in a sample to produce virtually identical overall GEP to a pure sample was 75% and this could be achieved by conventional cell sorting methods without altering the overall GEP in the process. For the purification of small, biohazardous samples, we tested the applicability of multistage magnetic sorting (Magsort) and laser enabled analysis and processing (LEAP). We developed optimized sample labeling and sorting protocols for both technologies and demonstrated that while the maximum purity we could achieve with Magsort was 75-80%, with the LEAP instrument we could purify fluorescently labeled cell subsets to 80-100%. The purified cells from biological samples often do not provide enough RNA for direct microarray studies without RNA amplification. We found that both linear and exponential amplification was capable of producing enough RNA for microarray analysis even from a single cell. Both methods distorted the GEP, however, with linear amplification much fewer genes were affected and only this method preserved the GEP differences between samples. Further studies are needed to analyze and possibly eliminate all GEP distortion. In conclusion, the purification of minor cell subsets from biological samples prior to microarray analysis is not only necessary, but also achievable without GEP distortion. Using linear RNA amplification of small purified samples, meaningful microarray data can be produced about the GEP of even a few cells.