Voxelation—an exciting marriage of biomedical imaging and microarray technology—promises to reveal new molecular insights into brain biochemostry and neurological disorders, all in captivating 3-D. Kevin Davies reports.
April 7, 2002 | There can be no question that microarray technology holds abundant potential for analyzing the parallel activity of thousands of genes and producing detailed molecular profiles of normal and diseased cells (see "Moving Chips to the Clinic"). But until now, most microarray studies, chiefly in mutation and expression analysis, have been of the ex vivo variety. In a study published in the February 2002 issue of Genome Research, University of California at Los Angeles researcher Desmond J. Smith and colleagues introduce a highly promising technique called voxelation, which they have used to acquire 3-D gene expression patterns in the brain "analogous to the images reconstructed in biomedical imaging systems, such as CT and PET." With further enhancements, voxelation could lead to important insights into the pathogenesis of Alzheimer's disease, Parkinson's disease and many other common disorders of the central nervous system.
A physicist by training, the British-born Smith is a member of the UCLA Department of Molecular
| Brain waves: Expression in the cerebral cortex of a subset of 30 genes that show statistically significant differences in activity in Alzheimer's disease. The hemisection is reflected along the midline to produce the symmetrical image.
and Medical Pharmacology chaired by Michael Phelps, the inventor of PET scanning. It was while listening to an early Sunday morning talk on PET at a departmental retreat, seriously hung-over he sheepishly recalls, that Smith began to imagine the possibilities of marrying imaging and microarrays to study what Woody Allen affectionately terms his "second favorite organ" — the brain. Smith's major research interest was in genetically engineering mouse models for Down syndrome, with the aim of teasing apart the contribution of individual genes to match specific aspects of the disorder.
Smith began collaborating with computational biologist Richard Leahy, a fellow ex-pat at the University of Southern California. His team began by taking relatively thick 8-mm slices from the brains of an unaffected and an Alzheimer's subject, then marking each hemisection into a grid of 24 precise 3-D cubes, dubbed voxels. The next step was to prepare and label RNA from each voxel and quantify the expression level of 2,000 genes in each voxel using conventional microarray hybridization methods (Smith's group uses a Cartesian arraying robot and an Axon Instruments scanner). This allowed Smith's team to compare expression patterns in neighboring regions of the brain, as well as the same regions from different specimens. Using Monte Carlo simulations, they identified several clusters of genes whose expression was either positively or negatively co-regulated. They also noted some intriguing, if preliminary, expression differences between healthy and Alzheimer's samples. Importantly, the gene expression patterns from a specific voxel appear consistent across unrelated samples.
In several instances, the analysis was enhanced with IT applications. BLAST searches of the genome sequence flanking the expressed genes revealed information on the critical DNA sequence motifs that
|Brown, V.M.; Ossadtchi, A.; Khan, A.H.; Cherry, S.R.; Leahy, R.M. & Smith, D.J. "High-Throughput Imaging of Brain Gene Expression." Genome Research 12, 244-254 (2002).
control gene activity, revealing some highly conserved regulatory sequences in otherwise distantly related genes. "The genetic wiring diagram should aid the neuronal wiring diagram," says Smith. And a method called singular value decomposition (SVD), used in biomedical imaging, was used to produce striking gene activity images that revealed groups of genes expressed in significantly different quantities in various parts of the brain, including the hippocampus, cortex and caudate.
With further refinements in voxel density and RNA labeling, the number of voxels in the human brain could grow by an order of magnitude. For example, using a more sensitive probe-labeling technique such as dendrimer technology, the number of voxel images could grow to more than 300,000, about twice as many as a typical PET scan. By expanding the number of genes analyzed on the microarrays to 30,000, researchers could construct a high-resolution 3-D atlas of human brain gene expression consisting of some 10 billion data points. "Computational analysis will become an important issue," Smith concedes without irony. Not that he's unduly worried: "From the point of view of computing," he says, "we're not scared. Moore's Law is holding in genomics, and we suspect gene expression databases will also follow a law of inexorable expansion."
Smith intends to produce comparative maps of several human brains for improved statistical analysis of variations in expression levels. "Voxelation may ultimately reveal the molecular ontology of the brain," he writes, "demonstrating which parts of the brain are most closely related in terms of gene expression patterns to other parts." These data should prove particularly valuable for studies of schizophrenia, manic depression, and other neuropsychiatric disorders. Whereas conventional and electron microscopy studies have revealed few reliable neuroanatomical clues in the pathogenesis of such disorders, Smith believes voxelation could lead to "keener insights into the real locus of abnormality." Smith's group is also developing a related technique called gene expression tomography, as featured in the March issue of Physiological Genomics. This method reconstructs 3-D gene expression images from cryostat brain slices taken around different rotational axes.
As important as these human brain studies are, an immediate goal is to adapt voxelation for studies of the mouse brain — no trivial task, given that the mouse brain is a mere one-fiftieth the size of the human brain, roughly the difference between an almond and a coconut. But given the immense power of mouse genetic manipulation for studying behavior and the nervous system, Smith's team is undeterred and embarking on such studies. They hope to report their findings this summer.
PHOTO BY D. SMITH