Array tomography is a technique pioneered in the laboratory of Stephen Smith at Stanford University. It consists of fluorescence imaging of thin (70-100nm) slices of embedded tissue with specific proteins labeled by antibodies. This allows the tracing of specifically labeled neurons (e.g. with a genetically encoded marker) and determining the location and abundance of proteins of interest within connected neurons.
We have implemented high-throughput, automated imaging tailored to array tomography samples, including sophisticated autofocus and inline image quality assessment. It has now run many samples with up to half a million images each. Bill Karsh developed the alignment software to accurately and rapidly assemble these large data sets into 3D volumes. The first application of these tools was an ambitious large-volume reconstruction in the mouse barrel cortex led by JC Rah, a visitor in the Svoboda lab, which has now been published (Front Neural Circuits. 2013 Nov 12;7:177). Erik Bloss, in Nelson Spruston’s lab, is currently using the system to explore the patterns of connectivity of neuronal subtypes in the mouse hippocampus; see some more great data from this instrument on their array tomography page.
Maximum intensity projection of 1000 slices in mouse barrel cortex; the full size of this volume is 1.3 mm x 1.4 mm x 0.2 mm; a typical barrel is about 0.3 mm in diameter. The resolution of the original data – which is downsampled for display – is 250x250x400 nm. The green somata are layer 5 pyramidal neurons, while the red labels projections from the thalamus. Note that only about 5% of the cells are labeled—the “empty space” is filled with unlabeled cells.
Zoomed in on apical dendrites. This volume is 30 x 60 x 20 microns.