Unbiased transcriptomes of resting human CD4+CD45RO+ T lymphocytes

Gene expression profiling of cells of the immune system isolated ex vivo is a unique tool to assess gene expression in vivo. Exemplified for CD4+CD45RO+ effector/memory T helper (TE/M) lymphocytes of human peripheral blood, we have analyzed the influence of different isolation procedures and storage conditions on the gene expression profile [1], and describe a method for the generation of unbiased transcriptomes. TE/M lymphocytes were isolated from fresh human peripheral blood to >98% purity by FACS as CD3+CD4+CD45RO+ CD45RA− cells. Prior to this, cells were enriched either by MACS with whole-blood CD4 beads (CD4 MACS), density gradient centrifugation (FICOLL), alone or in combination with CD3 MACS, or by lysis of erythrocytes (LYSIS) followed by MACS depletion of CD15+ cells. A total of 7 different protocols were compared among each other and to T cells stimulated for 3 h after isolation with PMA and Ionomycin (Fig. 1A and Supporting Information File 1). Cell isolation procedures lasted for 45–120 min, and were carried out according to manufacturer’s instructions at 4°C, unless indicated otherwise. Preparationinduced de novo transcription was blocked in one of the protocols by the addition of 2 μg/mL Actinomycin D (ActD). The transcriptomes of TE/M cells from two unrelated healthy donors (D1, D2) were determined on Affymetrix HG-U133 Plus 2.0 arrays (Affymetrix, Inc) that contain 54 675 probe sets representing approximately 47 400 transcripts and variants including 38 500 genes [2]. The data were evaluated using the High Performance Chip Data Analysis (HPCDA) [3] and Genes@Work software [4]. The nonstimulated samples clustered by donor, irrespective of the isolation protocol used, demonstrating that the interindividual differences exceeded those induced by the isolation procedures (Fig. 1B). Natural genome-wide variation in human gene expression has been shown to be able to account for significant fold changes between individuals in a population [5, 6] and should be taken into account during experimental planning. Genes encoding CD8, CD14, CD16, CD19, CD20, or CD56, indicating contamination of the CD4+ T cells with CD8+ T cells, monocytes, granulocytes, B lymphocytes, and NK cells confirmed the purity of the preparations (Fig. 1D). An exception was the contamination of CD4 T cells isolated by LYSIS with residual erythrocytes, as indicated by expression of the globin genes HBA and HBB (Fig. 1D). Comparing the different protocols, up to 759 Affymetrix probe set IDs (AffyIDs) were differentially expressed, by a factor of 1.5 or more (Fig. 1C and Supporting Information File 2). Cells isolated by CD4 MACS in the presence of ActD showed the lowest expression of genes indicating activation or stress, such as JUN and FOS [7], very similar to cells isolated by CD4 MACS in the absence of ActD (Fig. 1D). Sixteen AffyIDs were differentially expressed between cells isolated by CD4 MACS in the presence versus absence of ActD. Of these, only the pseudogene POLR2J4 showed an absolute fold change of 1.5 (1.6). These subtle differences could also reflect technical variation between the same cells, RNA prepared, amplified, and hybridized in parallel [8]. Thus, in cells isolated by CD4 MACS and FACS neither the degradation of RNA, nor the de novo transcription ex vivo skewed the in vivo transcriptomes. In general terms, ActD might qualify as a tool to prevent de novo transcription ex vivo in any kind of isolation protocol, but it cannot prevent degradation of transcripts, and thus is no alternative to isolation technologies preserving the in vivo transcriptome. Compared with cells isolated by the CD4 MACS protocol, cells isolated by the FICOLL or LYSIS procedures showed differential expression of more than 100 AffyIDs at an absolute fold change of 1.5 (Fig. 1C). Among them were immediate response genes like JUN and FOS, the EIF1 gene encoding a translation factor [9], and the genes encoding the antiand proapoptotic proteins Bcl-2 and Bcl10 (Fig. 1D) [10]. In the LYSIS samples the differences in expression were most pronounced. The TE/M cells that were stimulated after isolation cluster on a completely different branch clearly separated from the branches of the nonstimulated data of the two donors (Fig. 1B). They are set apart from each of the two donors by over 11 000 differentially expressed AffyIDs (absolute fold change 1.5) (Fig. 1C), revealing that changes induced by stimulation outweigh technical and biological variability. Decisive factors for the conservation of in vivo transcriptomes are the time