Open Access Highly Accessed Methodology article

Single cell transcriptomic analysis of prostate cancer cells

Christopher J Welty1, Ilsa Coleman2, Roger Coleman2, Bryce Lakely1, Jing Xia2, Shu Chen2, Roman Gulati2, Sandy R Larson1, Paul H Lange13, Bruce Montgomery4, Peter S Nelson24, Robert L Vessella13 and Colm Morrissey15*

Author Affiliations

1 Department of Urology, University of Washington, Seattle, WA, USA

2 Fred Hutchinson Cancer Research Center, Seattle, WA, USA

3 Department of Veterans Affairs Medical Center, Seattle, WA, USA

4 Department of Medicine, University of Washington, Seattle, WA, USA

5 Genitourinary Cancer Research Laboratory, Department of Urology, University of Washington, Box 356510, Seattle, WA, 98195, USA

For all author emails, please log on.

BMC Molecular Biology 2013, 14:6  doi:10.1186/1471-2199-14-6

Published: 16 February 2013



The ability to interrogate circulating tumor cells (CTC) and disseminated tumor cells (DTC) is restricted by the small number detected and isolated (typically <10). To determine if a commercially available technology could provide a transcriptomic profile of a single prostate cancer (PCa) cell, we clonally selected and cultured a single passage of cell cycle synchronized C4-2B PCa cells. Ten sets of single, 5-, or 10-cells were isolated using a micromanipulator under direct visualization with an inverted microscope. Additionally, two groups of 10 individual DTC, each isolated from bone marrow of 2 patients with metastatic PCa were obtained. RNA was amplified using the WT-Ovation™ One-Direct Amplification System. The amplified material was hybridized on a 44K Whole Human Gene Expression Microarray. A high stringency threshold, a mean Alexa Fluor® 3 signal intensity above 300, was used for gene detection. Relative expression levels were validated for select genes using real-time PCR (RT-qPCR).


Using this approach, 22,410, 20,423, and 17,009 probes were positive on the arrays from 10-cell pools, 5-cell pools, and single-cells, respectively. The sensitivity and specificity of gene detection on the single-cell analyses were 0.739 and 0.972 respectively when compared to 10-cell pools, and 0.814 and 0.979 respectively when compared to 5-cell pools, demonstrating a low false positive rate. Among 10,000 randomly selected pairs of genes, the Pearson correlation coefficient was 0.875 between the single-cell and 5-cell pools and 0.783 between the single-cell and 10-cell pools. As expected, abundant transcripts in the 5- and 10-cell samples were detected by RT-qPCR in the single-cell isolates, while lower abundance messages were not. Using the same stringency, 16,039 probes were positive on the patient single-cell arrays. Cluster analysis showed that all 10 DTC grouped together within each patient.


A transcriptomic profile can be reliably obtained from a single cell using commercially available technology. As expected, fewer amplified genes are detected from a single-cell sample than from pooled-cell samples, however this method can be used to reliably obtain a transcriptomic profile from DTC isolated from the bone marrow of patients with PCa.

Prostate cancer; Single-cell; Transcriptome; Disseminated tumor cells