The repellent axon-axon interactions have been demonstrated in vivo only for ventronasal and ventrotemporal axons, and not, for example, dorsonasal or dorsotemporal axons. However, our in vitro data did not show any differential sensitivities along the DV axis making it likely that this new mapping principle is relevant for all nasal and temporal axons. Based on the analysis of solitary axons in the zebrafish retinotectal projection, Gosse et al. (2008) put forward the idea that axon-axon interactions are not required for topographic mapping; however, as the authors further specify, this argument holds true only for the distal part of TZs which mapped appropriately even in solitude,

while the proximal end of their TZs was in fact significantly extended rostrally. While Idelalisib molecular weight the authors argued for the existence of a second tectum-derived gradient necessary to restrict the proximal end of a TZ (Gosse et al., 2008), possibly repellent N→T axon-axon interactions might lead here to the same effect. We show here that peripheral temporal axons are largely unaffected by the deletion of ephrinA5 from the colliculus and/or retinal axons (Figures 6F–6H), PI3K Inhibitor Library high throughput or in the full ephrinA5 KO (Figure 6D), and even mostly map to their normal topographic position in the ephrinA2/ephrinA5 DKO (n = 4; data not

shown). Our data for the DKO resemble those of Pfeiffenberger et al. (2006), who found robust targeting defects only if additionally ephrinA3 was deleted, i.e., in the TKO (Pfeiffenberger et al., 2006). These astonishing findings suggest that targeting of peripheral temporal axons might involve other and/or additional activities, for example, engrailed (Brunet et al., 2005 and Wizenmann

et al., 2009) (see also Willshaw et al., 2014). Furthermore, uniform expression of ephrinA3 in the retina and no detectable expression in the retinorecipient layers of the SC adds another layer of complexity to the mapping process, but highlights the importance of retinal ephrinA expression. Retinal axons from the centronasal area of the retina (n-axons) are strongly affected in the collicular medroxyprogesterone KO of ephrinA5, where they form prominent eTZs in rostral locations and also a weak eTZ at the very caudal pole of the SC (Figure 5C) (Frisén et al., 1998). This phenotype is not enhanced in mice with an additional retinal ephrinA5 KO (Figure 5E), demonstrating that the mapping of n-axons is predominantly controlled by collicular, and not (or to a much lesser extent) by retinal, ephrinA5. Given the severity of phenotypes, there is a good possibility that the mapping defects of centronasal axons involve interference from mistargeted peripheral nasal axons. Conversely, it is highly unlikely that their mapping defects are a secondary consequence of the comparably weak overshooting and eTZ formation of t-axons within the caudal SC (Figure 4C).