In the current issue of Nature Methods, European researchers describe a technique three-dimensional imaging of macroscopic brain structures at cellular resolution. The new method allows for a level of detail that is unprecedented – it enables the vizualization of entire neuronal networks within intact brains.
To visualize the brain at the gross anatomical level, researchers use neuroimaging techniques such as functional magnetic resonance (fMRI). Presently, the resolution of these techniques is, at best, several millimetres; this corresponds to discrete regions of the brain containing several thousand neurons. The activity of these regions is measured indirectly, by visualizing the changes in blood flow around the brain, the assumption being that increased blood flow correlates to an increase in neuronal activity. To visualize the brain at the level of tissues or cells, techniques such as confocal or two photon microscopy can be used. But, because light is scattered as it enters biological tissues, it can penetrate no deeper than 1.5 mm, and this limits conventional microscopic techniques. Hence, conventional microscopy techniques can only be used to image thin slices of tissue, measuring several hundred micrometres thick, at most. Furthermore, reconstructing the data to generate three-dimensional images requires the “stacking” of data obtained from each slice; this is extremely laborious, and requires huge amounts of data.
The new technique, developed by Hans Ulrich-Dodt, of the Vienna University of Technology, Austria, and his colleagues, involves using an old but neglected technique called ultramicroscopy (also known as light sheet illumination), in combination with a novel way of making large pieces of tissue transparent. Tissue samples are first dehydrated and then immersed in a solution of benzyl-benzoate and benzyl-alcohol. It is this step in the procedure that makes the specimens transparent – the solution has the same refractive index as proteins. As it is absorbed by the sample, the solution enters the spaces between cells. Consequently, the extracellular and intracellular compartments in the sample have the same refractive index, so that light can enter the sample unhindered.
Once the tissue has been prepared in this way, it is scanned in a fluorescence microscope with sheets of blue laser light emanating from illumination sources on opposite sides of the sample. Light entering the sample is not refracted as it passes through the tissue, and therefore penetrates deeper, at the same angle at which it entered. As the light passes through the tissue unhindered, the sample appears to be transparent. In their experiments, Dodt’s group engineered the mice to express green fluorescent protein, so the light caused the cells in its path to emit fluorescence.
(a) Surface of an entire mouse brain reconstructed from 550 optical sections; scale bar = 1mm. (b) Reconstruction of an entire mouse hippocampus, in which individual cell bodies are visible; scale bar = 500 µm. (c) Reconstruction of part of the mouse hippocampus, showing the projections of cells in that region; scale bar = 200 µm. (d) Reconstruction of the dendritic spines of neurons in the CA1 region of the hippocampus; scale bar = 5 µm. (From Dodt, H.-U., et al., 2007.)
Using this technique, Dodt and his colleagues successfully obtained three-dimensional images of 2 cm-thick slices of brain tissue. This is several orders of magnitude thicker than the slices that can be visualized with conventional microscopy, but all the neurons within the specimen, as well as their processes. By introducing an additional dehydration step to the method, which involved rinsing the brains in 100% hexane solution for 1 hour, they were also able to image entire fruit flies and mouse embryos.
Currently, however, the method has limitations. For example, the 3D reconstructions are limited by computing power, and the size of specimens is also restricted – brains from mice older than 2 weeks of age cannot be imaged in their entirety. Finally, heavily myelinated structures, for example, cannot be made transparent, but this could be overcome when ways to remove fatty tissues from specimens are developed. Nevertheless, the technique can be used to obtain extremely fast reconstructions of large specimens. It could also be used to perform high throughput screening of mutant mice, and, with multiple colour microscopy, may enable the visualization of gene expression patterns at the cellular and network levels.
Dodt, H.-U. et al. (2007). Ultramicroscopy: three-dimensional visualization of neuronal networks in the whole mouse brain. Nat. Methods 4: 331 -336. [Abstract]