Received on June 9, 2009
Accepted on September 9, 2009

Molecular Diagnostics and Genetics

Quantitative Transcription Factor Analysis of Undifferentiated Single Human Embryonic Stem Cells

¹ Lundberg Laboratory for Cancer, Department of Pathology, Sahlgrenska Academy at University of Gothenburg, Gothenburg, Sweden, and TATAA Biocenter, Gothenburg, Sweden
² Department of Clinical Sciences, Lund University, Clinical Research Centre, UMAS, Malmö, Sweden
³ Department of Bioengineering, Imperial College London, London, UK, and Department of Ophthalmology and Program in Neurobiology, Children's Hospital Boston, Harvard Medical School, Boston, MA
⁴ Stem Cell Center, Lund University, Lund, Sweden

^* To whom correspondence should be addressed. E-mail: anders.stahlberg{at}tataa.com.

BACKGROUND: Human embryonic stem cells (hESCs) require expression of transcription factor genes POU5F1 (POU class 5 homeobox 1), NANOG (Nanog homeobox), and SOX2 [SRY (sex determining region Y)-box 2] to maintain their capacity for self-renewal and pluripotency. Because of the heterogeneous nature of cell populations, it is desirable to study the gene regulation in single cells. Large and potentially important fluctuations in a few cells cannot be detected at the population scale with microarrays or sequencing technologies. We used single-cell gene expression profiling to study cell heterogeneity in hESCs.

METHODS: We collected 47 single hESCs from cell line SA121 manually by glass capillaries and 57 single hESCs from cell line HUES3 by flow cytometry. Single hESCs were lysed and reverse-transcribed. Reverse-transcription quantitative PCR was then used to measure the expression POU5F1, NANOG, SOX2, and the inhibitor of DNA binding genes ID1, ID2, and ID3. A quantitative noise model was used to remove measurement noise when pairwise correlations were estimated.

RESULTS: The numbers of transcripts per cell varied >100-fold between cells and showed lognormal features. POU5F1 expression positively correlated with ID1 and ID3 expression (P < 0.05) but not with NANOG or SOX2 expression. When we accounted for measurement noise, SOX2 expression was also correlated with ID1, ID2, and NANOG expression (P < 0.05).

CONCLUSIONS: We demonstrate an accurate method for transcription profiling of individual hESCs. Cell-to-cell variability is large and is at least partly nonrandom because we observed correlations between core transcription factors. High fluctuations in gene expression may explain why individual cells in a seemingly undifferentiated cell population have different susceptibilities for inductive cues.