Development of an SOP for performing arrays on low-yield flow-sorted cells
A large number of studies in NHPs involve sorting rare subsets of cells. These samples pose a particular challenge because of the relatively low number of cells recovered. Despite amplification technologies that can work with ultra-low amounts of input RNA, most laboratories will have a high degree of sample loss due to improper handling of samples before the RNA stage. In recent years, we developed an SOP that starts with sorted cells from the flow cytometer and instructs user labs on how to properly harvest the cells and purify the RNA with minimal loss of samples. As a start, we initially pooled data on the yield of tetramer positive CD8 T cells from a large in vivo macaque project, and observed that while that while range of cells was from 500-80,000, 97% of the samples had 2000 cells or more (Figure 1A). After testing several different RNA purification techniques, we developed a SOP to purify RNA from samples as low as 500 cells, in which very few samples failed, and which we could use Bioanalyzer traces to predict samples likely to fail SPIA amplifcation(Figure aB). We then conducted a pilot experiment in collaboration with Dr. Michael Betts at the University of Pennsylvania to determine the reliability (i.e. sample failure rate), reproducibility and biological validity of low yield samples processed using our SOP. We sorted multiple replicate aliquots of naïve (CD27+CD45RO-) and non-naïve (CD27+CD45RO+;CD27-CD45RO+;CD27- CD45RO-) CD8+ T cells from a single human donor at cell yields of 500, 2500 and 10,000 cells and hybridized them to Affymetrix arrays (pilot overview shown in Figure 1C). Using our protocol, were able to successfully generate high quality arrays from 6/6 samples at the 10,000 cell mark, 5/6 samples at 2500 cells, and 5/6 samples at 500 samples. The resulting data show a high degree of reproducibility at the individual array level (Figure 1D). Lastly, to gauge the ability of our assay to accurately reflect the transcriptomic changes underlying the biological phenotype, we used Gene Set Enrichment Analysis (GSEA) to query our transcriptomes using data from a similar experiment in the literature(Haining et al J Immunol 2008). The analysis showed a remarkable degree of concordance between the new data and the dataset, demonstrating that our protocol maintains the underlying biology within samples.
Figure 1 – Protocol development to accurately assess transcriptomic changes in low input cells. (A) The frequencies of sorted CD8+Tet+ lymphocytes from a large vaccine study at various yields were estimated to establish a target amount that would allow optimal sample recovery while maintaining the maximal amount of cell input.(B) we developed an SOP to reliably harvest flow sorted cells and identify samples that would succeed or fail at yielding high quality RNA amplification (C) to assess reproducibility and biological accuracy, we conducted a pilot study on flow sorted CD8+ naïve and CD8+ non-naïve lymphocytes at varying frequencies from a single donor. This established that the SOP yielded (D) highly reproducible array data with few failed chips, and that (E) the measured gene expression was accurate.