Introduction
 

MicroRNA  (miRNAs) are short (18–24 nucleotides), non-coding RNAs that regulate  gene expression by both disrupting messenger RNA (mRNA) stability and  inhibiting mRNA translation. The expression of miRNA species in cellular  populations is thought to drive downstream gene expression and protein  functionality. Our goal was to determine the variability in miRNA  expression at the single cell level using a microfluidic system which  automates single cell capture and miRNA pre-amplification for downstream  expression analysis. We have developed a simple, modular workflow for  streamlined analysis of cell populations down to the single-cell level  (Figure 1). The workflow is centered on two key components: the C1TM  Single Cell Auto Prep System (Figure 1a: Sample Prep, including cell  isolation and cDNA preparation from miRNA species) and the Dynamic  Array™ IFC and BiomarkTM HD System (Figure 1b: Read out, for highly  parallel expression analysis). The Specific Target Amplification (STA)  chemistry performed on each individual cell captured on the C1TM IFC  borrows components from the Single Cell-to-Ct™ kit (Life Technologies)  for the lysis and preamplification steps and components from the TaqMan®  MicroRNA Reverse Transcription Kit (Life Technologies) for the Reverse  Transcription step (Figure 2).

Using  the Dynamic Array IFCs and the Biomark HD System, up to 96 cDNA samples  preamplified from the 96 single cells are each analyzed in parallel  with up to 96 microRNA TaqMan expression assays. Principal Component  Analysis (PCA) of the data using Fluidigm’s SINGuLAR™ Analysis Toolset  v2.0 reveals significant variations in the expression of discrete miRNA  species in a population of single cells from a single phenotype (Figure  3, 4, and 5). Comparison of phenotypically distinct populations (human  embryonic fibroblasts, human induced Pluripotent Stem Cells (iPS), human  Neural Progenitor Cells (NPC) derived from the iPS, and fully  differentiated human neurons (HN)) demonstrate more dramatic differences  in addition to the heterogeneity of expression within each group.

Results

Figure 1: Integrated workflow for miRNA analysis in single cells

The  C1 Single-Cell Auto Prep System performs Specific Target Amplification  (STA) of miRNA transcripts from single cells using the reagents and a  protocol developed for this purpose by Life Technologies (protocol  “Single-cell MicroRNA expression analysis”). The whole process, from  loading the cell suspension on the C1 Integrated Fluidic Circuit (IFC)  to full data analysis of the data can be accomplished in less than 24  hours.

Figure 2. C1 MicroRNA STA experimental workflow

Figure 3. Analysis: Single iPS cells and their NPC progeny

A)  Unsupervised clustering of the data clearly distinguishes iPS cells  from their NPC progeny obtained using small molecules1. Subpopulations  are also revealed within each group of cells. B) PCA shows a clear  difference between the two phenotypically distinct cell populations. C)  Violin plots show differential expression of miRNAs in different  subpopulations and reveal the main contributors to Principal Components 1  and 2 (in order from top left to right). The variations in expression  of a set of five miRNAs between iPS and NPC shows the same trends as  microarray measurements obtained with Embryonic Stem cells (ES) and  their NPC progeny (unpublished data, courtesy of Yao Shuyuan).

Figure 4. Human Neurons, iPS and NPC cells

A)  Unsupervised clustering of the data obtained with iPS, NPC and mature  neurons (HN) clearly distinguishes HN cells from iPS and NPC cells and  also reveals subpopulations within each cell type. miR-9 is more  frequently and more highly expressed in mature neurons (HN). B) PCA  clearly distinguishes between the three cell types based on miRNA  expression.The expression of miR-20a, 19b, 17 & 106a is lower in HN,  as expected based on neural differentiation and aging data2,3 .

Figure 5. Embryonic fibroblasts at different passage number

A)  Unsupervised clustering of the data from two different cultures of BJ  embryonic fibroblasts obtained at difference passage numbers (P13 and  P24) is not able to distinguish the populations from one another, even  though it can reveal different miRNA expression patterns between  individual cells. B) PCA analysis of the miRNA expression data from P13  and P24 cells confirms that the two cell populations are  undistinguishable based on miRNA expression. C) When the passage numbers  are more distant (P7 vs. P24), PCA analysis of the miRNA data (heatmap  not shown) distinguishes passage number P7 from P24.

Conclusion

•We  have developed a streamlined protocol on the C1TSingle-Cell Auto Prep  System to analyze the expression patterns of miRNA species in up to 96  individual cells processed in parallel with minimum hands-on time, in  less than 24 hours.

•The C1 miRNA STA protocol uses reagents optimized by Life Technologies for miRNA analysis. In particular, the Megaplex™ pools  of RT and PreAmp primers allow to produce cDNA from up to 380 different  miRNA species in each cell processed in the C1 IFC. The expression  patterns are read out using the Biomark HD System on 96.96 GE Dynamic  Array IFCs.

•Unsupervised clustering analysis and PCA of the miRNA data from different cell types reveal different patterns of  miRNA expression between the different cell types (confirmed by  microarray data or the literature) and also within each cell type.