Tools for Single-Cell Analysis
Ken Livak, Ph.D., a distinguished scientific fellow at Fluidigm, described how to derive cell signatures using the company’s Dynamic Array™ chips and BioMark™ system. While it is still quite costly to do whole-transcriptome analysis in a discovery or exploratory mode, time- and cost-efficient methods are available to analyze gene expression in subsets of genes.
To get meaningful results, one needs to collect data from a sufficient number of single cells to overcome the relatively high level of biological noise in the system, and so throughput and miniaturization become important factors, explained Dr. Livak. Looking at 100 genes in 100 cells, for example, requires 10,000 experiments.
Working with single cells means working at microvolume scale, and he described an emerging trend toward developing entire workflows optimized for processing at microvolume scale—from cell isolation to sample preparation to gene-expression analysis.
Dr. Livak’s talk focused mainly on how to analyze the data generated from gene-expression profiling of single cells. This is an evolving field, he noted, with no standard approach to data analysis. The goal is to identify correlations in transcript levels across a set of genes.
Because there is a high level of variation in the expression of any one gene and transcript levels are best represented by a log-normal distribution, one cannot normalize the expression data to those of housekeeping genes, as might be done in a conventional qPCR-based study. Instead, explained Dr. Livak, transcript levels in individual cells are normalized to the median of the distribution, thereby taking into account the fact that different genes will represent the median in different cells.
In his presentation, Dr. Livak referred to a study conducted by Paul Robson and colleagues at the National University of Singapore and the Genome Institute of Singapore to identify regulatory networks and developmental mechanisms that control cell fate decisions published in Developmental Cell in 2010. In initial studies to explore the expression of cell-type-specific transcription factors in the 64-cell stage mouse blastocyst, the researchers generated data from pools of cells. “As cell fate decisions are made by individual cells, this averaged expression may mask interesting single-cell dynamics,” the authors stated.
In the subsequent study described in the paper, they analyzed gene expression at the single-cell level and looked for correlations in the expression of multiple genes to identify cell signatures that correlate to an embryo transitioning from a 1- to 64-cell state. They disrupted each embryo into 64 individual cells and surveyed 48 genes across more than 500 cells using the Fluidigm Dynamic Array 48.48 and the BioMark system. Their results included the identification of two genes, ID2 and SOX2, which are the earliest markers of outer and inner cell populations, respectively.
Richard Fekete, a senior manager at Life Technologies, described the use of the company’s Ambion® Single Cell-to-CT™ kit and OpenArray® real-time qPCR technologies for gene-expression analysis.
With traditional sample-prep workflows, “we noticed that when you purify RNA or DNA from a small number of cells, it tends to bind irreversibly to the matrix, or it does not bind at all,” he said. Life Technologies designed the Cell-to-CT kit to perform sample prep (including cell lysis and genomic DNA removal), reverse transcription, and preamplification in a single tube to avoid loss of material that can result from sample transfer and processing.
Tracing the history of single-cell analysis at the company, Fekete described early work with preamplification and qPCR techniques that allowed for the analysis of RNA from a small number of cells. “We saw quite variable results from single cells suggesting a lot of cell-to-cell differences,” remarked Fekete.