Electronic Sorting of Immune Cell Subpopulations Based on Highly Plastic Genes

Pingzhang Wang,Wenling Han,Dalong Ma

DOI: https://doi.org/10.4049/jimmunol.196.supp.209.10

2016-01-01

The Journal of Immunology

Abstract:The publicly available omic big data provide a great and valuable resource to drive big-data-based discoveries. Previously, we used immunologically relevant transcriptome data from microarrays to analyze gene co-expression, gene plasticity (GPL) and biomarker evaluation for human and mouse immune cells, and have built two immunological databases, ImmuCo (http://immuco.bjmu.edu.cn/) and ImmuSort (http://immusort.bjmu.edu.cn/), for gene co-expression and gene plasticity analyses, respectively. In the current study, we addressed a brand new method for the first time, electronic sorting, which surveys big data samples to do virtual analysis and to sort immune cell subpopulations based on highly plastic genes at mRNA level. We also called this process either ‘virtual sorting’ or ‘in silico sorting’, because no actual immune cells were sorted as they are in the fluorescence-activated cell sorting (FACS) process. Through virtual sorting of human Tregs and cytokine-producing T helper cells, our results strongly supported the feasibility, effectiveness and predictability of this method. For example, we found that human CD4+FOXP3hi T cells were highly expressive of certain known molecules such as CD25 and CTLA4, and that this process of investigation did not require isolating or inducing these immune cells in vitro. We made a systematic virtual analysis of all of the potential human CD4+ T cell subpopulations, and potential signature genes or marker molecules for these subpopulations were identified. Besides CD4+ T cells, the methods in this study are also suitable for other immune cells, non-immune cells and even for cancers for studies on gene expression and regulation.