Image courtesy of Vladislav Petyuk, Pacific Northwest National Laboratory, Richland, Washington.

The whole is equal to the sum of its parts, but the heterogeneous composition of the brain makes high-throughput protein localization difficult. Like most problems, however, the brain is easier to analyze in small components. Petyuk et al. generate spatial maps of protein localization by summing protein quantification data obtained from small, uniform blocks of brain in a recent article in Genome Research.

The authors removed a coronal section containing the basal ganglia and internal capsules from each of two mouse brains. They physically divided one of the sections into 71 one-millimeter³ cubes called 'voxels' with a lattice of blades. Then they isolated and trypsinated protein from each of the cubes and the whole brain section. To generate an internal isotopically labeled reference, they labeled the protein sample from the whole section with oxygen-18 (¹⁸O) and divided it equally among the 71 remaining protein samples.

Highly sensitive techniques identified and quantified proteins in the 71 samples. The authors used capillary liquid chromatography coupled with Fourier transform ion cyclotron resonance mass spectrometry, which is more sensitive than traditional mass spectrometry. They identified proteins by comparing their mass and elution times to a database of recorded mass and elution times. To estimate the relative abundance of a given protein per voxel, they calculated the ratio of endogenous ¹⁶O-containing protein to ¹⁸O-labeled protein.

In all, the authors detected 1028 proteins at least twice. Almost half of the proteins localized to more than 65 voxels. Each voxel contained on average 600-700 proteins, and no one voxel contained all 1028 proteins. However, many regions showed enriched expression of particular proteins. The striatum, cortex, ventromedial region (containing the diencephalon) and medial region (containing the corpus callosum, fornix and anterior commissure) showed enriched expression of 17, 59, 29 and 28 proteins, respectively. In contrast, 150 proteins showed uniform distribution.

Protein localization generally agreed with gene localization, as shown by comparisons between this dataset, the Allen Brain Atlas and the Gene Expression Nervous System Atlas (GENSAT). Genes encoding most of the proteins that were enriched in specific brain regions also showed predominant expression in the same brain regions. For example, ProSAAS, which processes neuropeptide hormones, and the GABA-4 transporter, which clears GABA from synapses, predominantly localized to the diencephalon, consistent with the localization of Pcsk1n and Slc6a11 in the Allen Brain Atlas.  DARPP-32, which is known to localize to striatal neurons, and the guanine nucleotide binding protein gamma 7 subunit, predominantly localized to the striatum, as did the genes that encode them. Like their ubiquitous gene expression patterns, cytoplasmic actin (Actg1) and mitochondrial NADH-ubiquinone oxireductase (Ndufs3) showed uniform patterns of protein localization.

Unlike both the Allen Brain Atlas and GENSAT databases, which required brain sections from thousands of mice, the authors used only two mice in this study. Therefore, this highly sensitive, high-throughput proteomics technique might be preferable to existing genomics techniques to identify proteins misexpressed in mouse models of neurological disease.

1. Petyuk, V. A. et al. Spatial mapping of protein abundances in the mouse brain by voxelation integrated with high-throughput liquid chromatography–mass spectrometry . Genome Research(2007).