Neurons use multiple molecules to target to specific layers

Understanding how the countless neurons of the central nervous system guide themselves through relatively large distances in order to find their specific synaptic targets is one of the most vital yet complex questions in neuroscience. We have used genetic and molecular manipulations at the single cell level to address how neurons in the fly connect the eye to the brain.

HFSP Long-Term Fellow Wael Tadros and colleagues
authored on Fri, 01 March 2013

The central nervous systems of many different species are often organized into stratified synaptic layers.  These layers comprise the axons and dendrites of many different neurons, which form specific synapses with each other.  Despite their predominance across the animal kingdom, little is known about how these layers are formed during development. 

Figure: L1 lamina neurons synapse with photoreceptors in the eye and project axons that normally terminate in the M5 layer of the medulla in the brain. Here we have made single L1’s labeled with a membrane bound GFP (green) that are doubly mutant for sema-1a and CadN. Due to a loss of repulsive and adhesive pathways from these two molecules, respectively, during development these mutant L1’s now terminate beyond the M5 layer and often end up in much deeper neuropils of the brain. Photoreceptor axons (blue) are used a layer reference. The asterisk (*) demarcates an L2 neuron which is sometimes labeled with this genetic strategy. The scale bar is 5 microns.

We have addressed this question in the visual system of Drosophila due to its genetic malleability, its characterization at the ultrastructural level and the availability of many cell-type specific markers. Specifically we have taken advantage of second-order lamina neurons, of which there exist five sub-types (L1-L5) each terminating in distinct layers (M1-M10) of the medulla neuropil.  We began by asking what genes are required for the targeting of L3 neurons, which terminate solely in the M3 layer, via an RNAi screen that knocked down messages for about 900 cell-surface/secreted molecules.  The knockdown of transcripts for two genes, semaphorin-1a (sema-1a) and N-Cadherin (CadN), was found to cause a significant number of L3 neurons to terminate beyond the M3 layer, in M6. Although CadN, a homophilic adhesive molecule, was previously shown to be required for L3 targeting, the involvement of a Sema-1a, a repulsive signaling molecule, was a novel finding.  Their identification led to us to ask, firstly,  whether both proteins were expressed in L3 neurons by employing a novel tagging method: We modified the endogenous gene to include a conditionally expressed exon that would both tag the protein and cause the neuron to be fluorescently labeled.  By incorporating this exon in single, sparsely distributed cells, we demonstrated that Sema-1a and CadN were present in all five lamina neuron subtypes during their targeting stages and were highly localized to their growth cones. Next, through the use of single-cell genetic clones, we showed that both sema-1a and CadN were required redundantly and cell-autonomously in L3’s in a two-step targeting program:  In the first step, L3 growth cones project to a broad domain, within which L1 and L5 growth cones project (see figure). Here, CadN acts to adhere L3 growth cones to surrounding CadN positive processes while Sema-1a acts to repel them from the Plexin A positive serpentine neurons lying immediately proximal.  When both sema-1a and CadN are mutated in L3s their axons project past this common domain and extend into the medulla.  In the second step, L3 terminals retract by several microns and restrict themselves to the narrow M3 layer through the repulsive activity of Sema-1a acting in a temporally distinct window from its earlier function in controlling initial targeting.  Further experiments show that both Sema-1a and CadN act in a similar fashion in the wiring of two other lamina neuron subtypes (L1 and L5) which also project to the same domain in their first step of targeting.

Our work demonstrates the importance of neurite interactions between different neurons expressing distinct cell surface proteins in setting up a complex structure such as the medulla neuropil. We show that the assembly of synaptic layers is a stepwise process that, in fact, does not end with the arrival of lamina neurons to their correct layer.  After L3 terminals are refined into the incipient M3 layer they produce Netrin which is necessary for the targeting of R8 photoreceptors to this layer. The existence of complementary domains of Semaphorin and Plexin expressing neurites in the developing mouse retina as means for proper layering of amacrine neurites suggests cross-species conservation of these pathways in layer assembly.


Multiple interactions control synaptic layer specificity in the Drosophila visual system. Matthew Y. Pecot*, Wael Tadros*, Aljoscha Nern, Maya Bader, Yi Chen, S. Lawrence Zipursky. Neuron (2013) 77:299-310. *These authors contributed equally to this work.

Pubmed link