Expression profiling of cells is a valuable tool in understanding gene regulatory networks and in identifying genes necessary for normal developmental processes. In the haematopoietic system, cell surface markers are well defined allowing the isolation of cells from various developmental stages using flow cytometry. However gene expression studies of solid tissues are often hampered by the difficulty in obtaining a homogenous cell population. Often appropriate cell markers are not well defined for given cell stages therefore making gene expression studies difficult. In addition studies of rare cell populations mean that the amount of material obtained is insufficient for analysis. To overcome these problems we have employed a system where a given cell population is GFP-labeled and then sorted by flow cytometry in order to obtain a pure homogenous population. RNA was then prepared from this material and amplified by a linear amplification method allowing large scale microarray gene expression analysis. This approach was successfully applied to gene expression studies of the midbrain-hindbrain organizer region of the mouse but can be applied to many other systems to isolate cell populations.
Using the mouse as a model organism, our group investigated the role of the paired domain transcription factor Pax2 in brain development (Bouchard et al., 2005). Pax2 is required for the formation of the isthmic organizer (IsO) at the midbrain-hindbrain boundary, where it initiates expression of the IsO signal Fgf8. To gain further insight into the role of Pax2 in mid-hindbrain patterning, our group searched for novel Pax2-regulated genes by cDNA microarray analysis of FACS-sorted GFP+ mid-hindbrain cells from wild-type and Pax2-/- E8.5 embryos carrying a Pax2GFP BAC transgene (Figure 1). Our sorting protocol yielded 5,000 to 10,000 GFP+ cells per embryo which was insufficient to generate enough RNA to hybridize to a microarray. Therefore RNA from the sorted cells was subjected to a linear amplification as described (Hoffmann et al., 2003) with some modifications. Following this, the RNA was used in a microarray screen as described (Cheung et al., 1999). Our unbiased chip approach identified the En2, Brn1, (Pou3f3), Sef, Tapp1 and non-coding Ncrms genes as genetic Pax2 targets that are totally dependent on Pax2 function for their expression in the mid-hindbrain region. Note that genes which showed only a 2 fold regulation on the microarray were confirmed as Pax2 target genes by in-situ hybridisation. This proves that the amplification protocol is linear and that genes with low induction ratios are indeed Pax2-regulated target genes.
Research Institute for Molecular Pathology
Dr. Bohrgasse 7
A1030, Vienna, Austria