Purpose:
The retinal ganglion cell(RGC) population in mice is thought to consist of 20-30 cell types. Identifying these circuit elements is crucial towards mastery of retinal circuitry. Genetically defined subsets of the RGCs can be imaged using genetic engineering tools. This paves the way to test the limits of stratification specificity in the retina. We have quantified the morphological stereotypy of genetically defined RGC types, focusing on the depth at which dendrites of each type stratify in the inner plexiform layer.
Methods:
Confocal image stacks of genetically defined RGCs are enhanced using a supervised machine learning algorithm to suppress the background and alleviate the differences in brightness levels on the dendritic arbor. These enhanced images are traced using the Simple Neurite Tracer in FIJI software package.Image stacks need to be warped and registered for a depth analysis, to account for the deformations occuring during preparation and imaging. We used the ChAT bands as landmarks in the retina and warped the stacks so that the bands form two parallel planes.Depth profiles are obtained from the trace points, which are low-pass filtered to suppress noise. Finally, they are normalized so that each neuron's profile has the same energy.
Results:
We show the depth profiles of 118 cells, including JAM-B, CB2, W3, W7, and KIAA-CRE cells, as well as neurons from GFP-M animals. The depth profiles are remarkably consistent. The alignment reveals that the peak stratification plane can be determined with submicron accuracy. For instance, the standard deviation of peak depth is 0.04 times the distance between ChAT bands for CB2 cells, which corresponds to ~0.5um. The KIAA-CRE cells are accompanied by a second, less strongly GFP-expressing, population. These are easily distinguished on the basis of stratification.
Conclusions:
RGCs target their synaptic partners in narrow through-plane regions. For homogeneous, genetically defined cell types they show submicron specificity. This precision will be useful for further studies classifying the cells.
Keywords: ganglion cells • gene/expression • image processing