ABSTRACT: Thirty-five adult (8-16 week old) C57BL/6 mice (which had undergone previous behavioral protocols) were anesthetized with isoflurane, and the brains were removed quickly, embedded in OCT, and frozen in a dry ice-ethanol mixture. The interval between decapitation and complete freezing was always less than 3 minutes. The brain was warmed up to -20oC, and cut into 10-14 m sections with a cryostat (Leica). Serial coronal sections were kept on slides in 100% ethanol until sectioning was complete. Tissue was then processed with hematoxylin and eosin (H&E) staining for cell layer visualization, followed by dehydration with increasing concentrations of ethanol and then xylene. We focused on two cell types, the Purkinje (Pk) and granule cells (gc) of the flocculus. We used an Arcturus PixCell II laser capture microdissection (LCM) scope to locate and capture cells from the appropriate areas, obtaining ten samples per cell type per mouse. We estimate that each Pk sample contained about 50-100 Purkinje cells, and each gc sample contained a few thousand granule cells. Granule cell samples (dorsal flocculus) were captured before Purkinje cell samples, to reduce contamination of the Purkinje cell samples by granule cell material. Samples were removed from the LCM caps with RNEasy lysis buffer (Qiagen) containing 1% -mercaptoethanol, then frozen at -80oC. For each mouse, all the samples of a given cell type were pooled together; samples from different mice were not pooled. Total RNA samples were isolated using RNEasy kits (Qiagen), then amplified in two rounds of in vitro transcription (IVT) using the Ambion MessageAmp aRNA kit. IVT-amplified samples were hybridized to microarrays if the end product after the second round of amplification was of concentration greater than or equal to 0.2 g/l, and at least 10 times the negative control signal (measured after two rounds of amplification of a blank sample). Due to the small starting sample sizes, samples often were rejected due to insufficient quantity for hybridization: thus, 18/35 granule cell samples and 23/35 Purkinje cell samples survived this quality control process, and were run on arrays (41 arrays total). Mus musculus spotted cDNA microarrays (MM arrays) containing ~42,000 spots were obtained from the Stanford Functional Genomics facility (complete information at http://www.microarray.org/), in order to analyze gene expression in each cell type. We used a type II experimental design, where all experimental samples were hybridized against a common reference sample for multi-way comparison. The reference sample comprised mRNA extracted from neonatal and adult brain and liver, and amplified twice by IVT. We used standard protocols for cDNA labeling, as well as array hybridization, washing, scanning, and data analysis (http://cmgm.stanford.edu/pbrown/protocols/index.html).
