Abstract
Purpose: :
To develop an image analysis program that can accurately quantify ganglion cell number and the spatial loss of ganglion cells in mice carrying the Rosa3 allele.
Methods: :
ROSA3 mice carry the ßGeo reporter gene in the first intron of Fem1c (Fem1cRosa3 – R3). Previously, we showed that this gene was a potent marker for ganglion cells in the mouse retina. The R3 allele originated in C57BL/6 mice and was congenically crossed into DBA/2J mice. DBA/2JR3/+ mice were generated by crossing R3/R3 parents with wild type DBA/2J mice. At specific ages, retinas from R3/+ animals were removed, fixed and stained for ßGeo activity using X–Gal. Retinas were wholemounted and digitally imaged. Staining density was assessed using a MatLab program written to recognize the distinct color pattern for processed X–Gal. Cell density was measured by counting representative 400X fields.
Results: :
A baseline retina density map was made using ten 3 month retinas harvested from DBA/2J mice. Positive ganglion cells were distributed in an inferonasal (high) to superotemporal (low) gradient similar to reports using conventional morphometry. Quantitative changes in stain intensity, caused by retinal ganglion cell loss resulting from acute optic nerve lesion (crush) or chronic glaucoma were measured against the baseline density image. Acute lesion results in uniform loss with kinetics comparable to cell count data. Cell loss in chronic glaucoma can also be uniform, but may also occur in large sectors of retina. Using cell count data from target regions, the color density map is currently being correlated to an estimate of cell number.
Conclusions: :
A computer recognition program has been developed that recognizes stained precipitate in genetically marked mouse retinal ganglion cells. This development will allow high throughput quantification of ganglion cells over the entire retinal surface, enabling both quantitative and spatial analyses of cell loss.
Keywords: ganglion cells • apoptosis/cell death • image processing