Two ligands for Eph‐related receptor tyrosine kinases, RAGS and ELF‐1, have been implicated in the control of development of the retinotectal projection. Both molecules are expressed in overlapping gradients in the tectum, the target area of retinal ganglion cell axons. In two in vitro assays ELF‐1 is shown to have a repellent axon guidance function for temporal, but apparently not for nasal axons. RAGS on the other hand is repellent for both types of axons, though to different degrees. Thus, RAGS and ELF‐1 share some and differ in other properties. The biological activities of these molecules correlate with the strength of interaction with their receptors expressed on RGC axons. The meaning of these findings for guidance of retinal axons in the tectum is discussed.

These results are consistent with a role for En in determining rostrocaudal polarity of the developing tectum. The demonstration that both RAGS and ELF-1 are upregulated following En misexpression provides a molecular basis for understanding the previous observation, also based on retrovirus-mediated En misexpression, that nasal axons form ectopic connections in rostral tectum, from which temporal axons are excluded.

Previous results of an in vitro guidance test, the stripe assay, have demonstrated the presence of a repulsive axon guidance activity for temporal retinal axons in the posterior part of the vertebrate optic tectum. Ephrin-A5 and Ephrin-A2 are ligands for the EphA subfamily of Eph receptor tyrosine kinases, which are expressed in overlapping gradients in the posterior part of the tectum. When recombinantly expressed, both proteins have been shown to guide retinal ganglion cell axons in the stripe assay. While these results suggest that Ephrin-A5 and Ephrin-A2 form part of the posterior repulsive guidance activity, they do not elucidate whether they are necessary components. Here we report that soluble forms of the ligands at nanomolar concentrations completely abolish this repulsive activity. Similar results were obtained with the soluble extracellular domain of EphA3, which is a receptor for Ephrin-A2 and Ephrin-A5, but not with the corresponding domain of EphB3, a receptor for the transmembrane class of Eph ligands. These experiments show that the repulsive axon guidance activity seen in the stripe assay is mediated by Ephrin-A ligands.

The isolation and characterisation of two zebrafish Eph receptor ligand cDNAs which we have called zfEphL3 and zfEphL4 is described. These genes are expressed in the presumptive midbrain of developing embryos from 6 somites. By 24 hours L3 is expressed throughout the midbrain including the region of the presumptive tectum whereas L4 is strongly expressed in the midbrain caudal to the presumptive tectum. At later stages of development L3 is expressed in a graded fashion throughout the tectum and L4 is maintained at its posterior margin. Growth cone collapse and pathway selection assays demonstrate that both these proteins have a collapse activity for retinal ganglion cells. When faced with a choice of substrate on which to grow, temporal axons from chick retinal ganglion cells selectively avoided membranes from Cos cells transfected with L3, whereas nasal axons did not. Both temporal and nasal axons avoided membranes from Cos cells transfected with L4. The expression patterns together with the functional data suggest that although both ligands may be able to guide retinal ganglion cells axons in vitro, they have different roles in the guidance of retinotectal projections in vivo. The expression of L3 is consistent with a role in the guidance of retinal ganglion cells to their targets on the tectum whereas that of L4 suggests a role in delineating the posterior boundary of the optic tectum.