The Scientist

This week?s advance online publication of __Nature Neuroscience__ linkurl:details a neat new technique;http://www.nature.com/neuro/journal/vaop/ncurrent/abs/nn1654.html called FACS-array profiling, which should be of interest to anyone studying central nervous system development. X. William Yang and colleagues at the University of California, Los Angeles, used transgenic mice from the linkurl:GENSAT;http://www.gensat.org (gene expression nervous system atlas) project to compare gene expression profiles in two distinct cell subpopulations of the basal ganglia. Each transgenic GENSAT mouse expresses the enhanced green fluorescent (EGFP) protein under the control of a particular central nervous system-restricted locus. In this case Yang?s team compared gene expression in mice containing EGFP insertions that specifically labeled either the striatopallidal or striatonigral medium spiny neurons (MSNs), two cell types that are "morphologically indistinguishable and mosaically distributed." The team harvested and enzymatically dissociated striata from mice at postnatal day 20, labeled the cells with propidium iodide, and sorted by FACS to isolate live (PI-negative), GFP-stained cells, from which they isolated RNA. They then used that RNA to probe mouse developmental microarrays from Agilent Technologies, identifying nine genes selectively enriched in striatonigral neurons and 32 in striatopallidal neurons. Nineteen of those genes were differentially expressed in adult (two-month old) mouse brains, as well. What?s really nice about this paper, in light of the never-ending debate over the quality and value of microarray data, is the evident reproducibility and accuracy Yang?s group observes. Homotypic comparisons (comparing independent biological replicates on different arrays) had an average correlation coefficient of 0.98; heterotypic comparisons (comparisons between mice with different labeled MSN subpopulations) had coefficients of 0.97. Eighteen of 23 differentially expressed genes were independently validated using RT-PCR. Most impressively, according to supplementary material, the team obtained strong reproducibility, even using as little as 30 pg RNA, an amount that corresponds to 30 or so cells. That, the authors note, should make the technique sufficiently robust for even very rare neuronal populations.