Transcriptom analysis

The characterization of the global gene expression profiles described in this Ph.D. work gives me and others a database to explore the basic biology of functionally distinct precursor populations. In the current study, I performed an oligonucleotide microarray analysis on the neurogenic precursors (GFP+) and the more non-neurogenic population (GFP++) from E14 hGFAP-eGFP mice and the neurosphere cells isolated from the E14 cortex undifferentiated and differentiated for 1 days. The undifferentiated neurospheres are a population enriched for multipotent precursors. The neurosphere cultures differentiated for 1 day are a population of mostly gliogenic precursors, and would therefore be most similar to the non-neurogenic population (GFP++) from E14 hGFAP- eGFP mice (Fig. 3).

The comparison of gene expression in the model of multipotent cells, the neurosphere cultures, to previous analysis revealed several similarities (Geschwind et al., 2001; Ivanova et al., 2002; Ramalho-Santos et al., 2002). Recent descriptions of gene expression patterns within stem-cell populations are starting to define their molecular profiles (Geschwind et al., 2001; Ivanova et al., 2002; Ramalho-Santos et al., 2002). Ivanova et al. generated neurospheres from the embryonic ganglionic eminence, while Ramolho-Santos derived their neurospheres from the adult mouse SEZ. As discussed

previously, there are at least for some important candidate genes no significant differences in gene expression between both regions. These authors then compared the gene expression in neurosphere cells to embryonic or hematopoietic stem cells to identify common genes determining “stemness” multipotent fate.

Notably I also observed upregulation of a number of genes in the undifferentiated neurosphere cultures that were identified as stem-cell specific in common by those groups (Ivanova et al., 2002; Ramalho-Santos et al., 2002). These genes include Fyn, Slit2, tenascin C, and insulin-like growth factor binding protein 2 or cyclin D.

Furthermore I looked for genes specifically associated with neural development. It is of interest that only few of the Wnt genes are expressed within the neurospheres (Wnt 5a, 7a, 7b). Wnt has recently been show to antagonize neural development in mouse ES cells (Aubert et al., 2002). In some cases, these findings may be very relevant to future studies looking at differentiation or proliferation of these cells. For example, the FGF receptors FGFR1 and FGFR3 were expressed but not FGFR2. The expression of FGF-2 and its receptor FGFR1 may imply paracrine or autocrine effects of this mitogen on the growth of these neurosphere cells. Similarly, platelet-derived growth factor receptor (PDGFR) alpha, but not beta, was expressed. However, a number of genes that I expected to be expressed at high levels were not expressed. For example, musashi1 has been associated with neural stem cell populations (Kaneko et al., 2000; Keyoung et al., 2001), but was absent in the mRNA of undifferentiated neurospheres. As Keyoung et al. examined musashi1 in short-term cultures only, it is possible that this RNA-binding protein is down-regulated after the longer growth periods used in this study.

Since I was able to isolate and to enrich three to four functionally distinct precursor populations (Fig. 3), I wanted to assess which genes might be linked to

different fate decisions. Several known genes were expressed in only one or two related populations of the four different precursor populations. Some of these genes have not yet been implicated with neural development. Some genes are specifically highly expressed in undifferentiated neurosphere cultures and would therefore be expected to play a role in multipotent cell fate. One of these was tweety is a novel maxi-Cl(-) channel (Suzuki and Mizuno, 2004) and FABP7 is the brain specific fatty acid binding protein (Hartfuss et al., 2003; Hartfuss et al., 2001; Veerkamp and Zimmerman, 2001). While the function of

tweety in multipotent precursors remains to be determined, FABP7 (BLBP) was shown to label a population of cells in the embryonic telencephalon enriched for multipotent precursors (Anthony et al., 2004). The expression levels of sirtuin2, a NAD-dependent deacetylase that regulates muscle gene expression and differentiation by possibly functioning as a redox sensor (Fulco et al., 2003), were also increased in undifferentiated neurosphere cultures.

The expression of Sox9 peaked in the populations enriched for gliogenic precursors (GFP++, dd1) and indeed when Sox9 is ablated in neural stem cells the specification of astrocytes and oligodendrocytes is strongly affected (Stolt et al., 2003). Finally Sox4 and LMO-1 (Lim domain only-1) were strongest expressed in the neurogenic precursors. Both of them are known to be expressed in the developing and adult CNS in specific precursor populations, but their specific function in cell fate specification still remains to be determined (Cheung et al., 2000; Hinks et al., 1997).

Clearly, the limitations of detection of the microarray system preclude the complete description of gene expression in neural stem cells and restricted precursors. However, given this caveat, this database displays highly reliable gene expression data and may be considered as at least a representative part of distinct telencephalic precursors.

Taken together this gene expression analysis revealed the differential gene expression of several interesting candidates that could be grouped to functionally distinct precursor subsets. Most importantly, the expression of Pax6 could be linked to neuronal- restricted precursors, whereas the expression of Olig2 could be restricted to multipotent neural stem cells. A further analysis of the microarray data may allow us to understand the transcriptional context in which these two fate determinants act.