Golden goal regulates target specificity in the Drosophila Lamina

In the visual system of Drosophila, photoreceptor (R) neurons
elaborate a precise retinotopic map of visual space in the brain.
The retina consists of 750 ommatidia, each containing eight
photoreceptor subtypes (R1-R8). R1-6 axons terminate in the first
optic ganglion, the lamina. R7 and R8 axons extend through the
lamina to innervate the second optic ganglion, the medulla. To
maintain retinotopy in the lamina, R1-R6 photoreceptor axons have
to undergo a complex axonal sorting during development, a process
called neural superposition. The mechanisms responsible for the
establishment of the highly organized connection pattern needed for
retinotopy remain incompletely understood. The transmembrane
receptor Golden goal (Gogo) is a known regulator of the developing
Drosophila visual system. During R8 pathfinding, Gogo acts in two
distinct steps. In larvae Gogo mediates repulsive axon-axon
interactions between R8 axons in the medulla to maintain proper
spacing. During pupal development, Gogo is required in R8 axons for
afferent-target interactions to promote layer recognition. The aim
of this thesis is to study how Gogo regulates target selection of
R1-R6 axons in the lamina to increase our knowledge on how target
specificity is controlled in vivo. The present work shows that Gogo
is required for R1-R6 axon lamina targeting and target cartridge
selection in distinct developmental steps. To analyze the
consequences of loss of gogo function specifically in photoreceptor
cells, I generated genetic mosaic eyes using targeted mitotic
recombination. During larval and early pupal development loss of
gogo function in large clones of R axons results in a disruption of
R1-R6 fascicle pattern formation across the lamina plexus. Using
single photoreceptor type rescue, I provide evidence that the first
outgrowing axon R8 uses Gogo to identify its intermediate target in
the lamina and to function as a pioneer axon for all follower R1-R6
axons for their correct patterning along the lamina plexus.
Interestingly, small clones of gogo deficient R axons perfectly
integrate into a proper retinotopic map suggesting that surrounding
R axons of the same or neighboring fascicles provide complementary
spatial guidance. Thus, Gogo acts in a partially redundant fashion
with local guidance cues provided by neighboring axons.
Additionally, during pupal stages at the onset of photoreceptor
sorting, I further show that R1-R6 axons fail to choose correct
target cartridges in the lamina when Gogo is absent in a large
fraction of R cells. I show that gogo mutant R1-R6 axons target
correctly to wild-type areas, whereas wild-type R1-R6 axons fail to
project correctly to areas innervated by mutant R axons.
Interestingly, rescue of Gogo in R8 axons was not only sufficient
for fascicle patterning earlier in development but also for R1-R6
axons to select their proper target cartridges in the lamina during
neural superposition. Finally, in a third developmental step, Gogo
is required for the elongation of R1-R6 axons along lamina
cartridges within the neuropile. In the absence of Gogo axons fail
to elongate in parallel fashion and intermingle with mutant axons
of neighboring cartridges. This suggests that Gogo, similar to its
role in medulla targeting, permits photoreceptor axons to stay
separated from each other. Based on the results of this thesis, I
propose that Gogo contributes to retinotopic map formation in the
Drosophila lamina during three steps: initial target recognition of
R1-R6 fascicles, target cartridge selection and cartridge
elongation.