Developmental retinal ganglion cell death and retinotopicity of the murine retinocollicular projection.

During mammalian visual system development, retinal ganglion cells (RGCs) undergo extensive apoptotic death. In mouse retina, approximately 50% of RGCs present at birth (postnatal day 0; P0) die by P5, at a time when axons innervate central targets such as the superior colliculus (SC). We examined whether RGCs that make short-range axonal targeting errors within the contralateral SC are more likely to be eliminated during the peak period of RGC death (P1-P5), compared with RGCs initially making more accurate retinotopic connections. A small volume (2.3 nL) of the retrograde nucleophilic dye Hoechst 33342 was injected into the superficial left SC of anesthetized neonatal C57Bl/6J mice at P1 (n = 5) or P4 (n = 8), and the contralateral retina wholemounted 12 hr later. Retrogradely labelled healthy and dying (pyknotic) RGCs were identified by morphological criteria and counted. The percentage of pyknotic RGCs was analyzed in relation to distance from the area of highest density RGC labelling, presumed to represent the most topographically accurate population. As expected, pyknotic RGC density at P1 was significantly greater than P4 (p < 0.05). At P4, the density of healthy RGCs 500-750 µm away from the central region was significantly less, although this was not reflected in altered pyknotic rates. However, at P1 there was a trend (p = 0.08) for an increased proportion of pyknotic RGCs, specifically in temporal parts of the retina outside the densely labelled center. Overall, the lack of consistent association between short-range targeting errors and cell death suggests that most postnatal RGC loss is not directly related to topographic accuracy.