One possible explanation is that isolated rat RPCs used in the previous study were relatively late in retinogenesis and were already dominated by PD and DD division modes. However, the number of cell cycles of some in vitro rat RPC lineage trees is similar to that of zebrafish RPCs, suggesting that they might not be that late. Thus, it will be interesting for future research to compare side-by-side stochastic retinogenesis models between these two systems in a more stringent way and to look for both conserved features and dissimilarities. Although the great variation in individual

check details RPC lineages seems to contradict a deterministic programming model and instead favors the stochastic model, this does not mean that the regulation of RPCs and their progeny is completely without any deterministic elements in fate choice. For example, in the two progeny from DD divisions of zebrafish RPCs, the same cell-type combinations of BCs, HCs, and PRs are produced at much higher frequencies than predicted by pure unbiased stochastic choices (He et al., 2012). Similarly, in rat RPCs in vitro, certain cell-type choices in two

successive RPC divisions might not be completely independent (Gomes et al., 2011). Furthermore, a dedicated subpopulation of zebrafish RPCs has been shown to divide symmetrically to generate exclusively BCs (Godinho et al., 2007). These examples illustrate how much deterministic PI3K Inhibitor Library research buy inputs might bias the stochastic choices. Such inputs are probably from those genes differentially expressed in RPCs that regulate progeny cell fates. For example, as mentioned above, the expression of Vsx1, Vsx2, Foxn4, and Ath5 is important for restricting progeny fates of RPC subpopulations (Vitorino et al., 2009). Furthermore, mouse NeuroD6, a member of the atonal-like

family of bHLH transcription factors, is critical for AC fate choice as forced NeuroD6 expression leads to significant increase in ACs (Cherry et al., 2011). In mice, Olig2+ RPCs, which appear later in RPC lineages, usually divide in DD (terminal) mode but the fate of the progenies varies nearly over time: embryonic Olig2+ RPCs are biased toward generating cone PRs and HCs, while postnatal Olig2+ RPC progenies are enriched for rod PRs and ACs (Hafler et al., 2012). The high heterogeneity of RPC transcriptomes (Trimarchi et al., 2008) suggests that there are more examples of such genes waiting to be characterized. Future research will have to understand the mechanisms that regulate the expression of these transcription factors and whether their expression is strictly controlled by temporal and/or by spatial patterning. This would suggest a general deterministic control. If the regulations of these cell fate genes were to show typical stochastic features, this would provide further support for the stochastic model.