Desplan Lab
New York University. 100 Washington Square East.
1009 Silver Center. New York, NY. 10003.

Optic Lobe Project



The neural network in the medulla for color vision (Initiated by Javier Morante): The Drosophila optic lobe can be divided in four neuropils: lamina, medulla, lobula and lobula plate. The lamina receives innervation from R1-R6. R7-R8 projections target two distinct medulla layers. The medulla is formed by ~40,000 cells. Associated with each set of R7/R8 projections, there are ~800 'columns', defined as fixed cassettes of cells that repetitively contact every R7-8 fascicule. Columns likely represent the functional units in the medulla. Some of these medulla 'columnar' cells contact one R7 and one R8 of the same ommatidium to compare their output (vertical integration, e.g. UV vs. blue in p ommatidia). Other non-columnar medulla cells compare information between several p and y ommatidia (horizontal integration, e.g. blue vs. green). A variety of local interneurons perform a first level of processing. We have identified the 70 cell types present in the medulla, some of which process color information, and we are analyzing the logic behind development of the medulla. [Morante Current Biol 2008].

Temporal patterning of neural progenitors in the medulla: (Xin Li) We performed an antibody screen looking for transcription factors (TF) expressed in the developing medulla, and have identified a large set of TFs expressed in subsets of neural progenitors (neuroblasts: NBs) and neurons. We identified a temporal gene sequence, composed of five TFs: Homothorax, Eyeless, Sloppy-paired, Dichaete and Tailless, that are sequentially and transiently expressed in medulla NBs as they transit through different temporal stages. Loss of one TF blocks further progression of the temporal sequence and causes the NBs to be stuck at the preceding TF stage. Notch-dependent asymmetric division of the intermediate progeny of NBs, the Ganglion Mother Cell (GMC), causes most neuronal TF markers to only be expressed in one of the two daughters of a certain GMC. We are studying how the temporal TF sequence in NBs and the Notch-dependent binary fate choice control the combinatorial expression of different TF markers in their neuronal progeny, and thus determine the sequential generation of different neuron types. For example, we found that the Non progeny of Hth+ GMCs express Apterous and Bsh, and become Mi1 local neurons; while the Noff progeny either commit apoptosis or adopt different neuronal fates depending on the spatial identity of the NB (see below). We are currently examining the fate of the progeny from NBs expressing other TFs.

A second (dorso-ventral) axis for the generation of neuronal diversity: (Ted Erclik) We have identified 35 TF genes that are expressed in subsets of medulla progenitors and neurons. By mapping these TFs onto the medulla neuroepithelium, we found that neuronal specification in the medulla is the product of the intersection of two spatial axes: (1) The temporal switch of NBs (see above); (2) in the dorsal-ventral axis, the medulla crescent–shaped neuroepithelium is sub-divided into spatial subdomains by Vsx1 (central domain), Hh (ventral), Optix (lateral) or Dpp or Wg (in two distinct regions at the tips of the crescent). While the sequential progression of neuroblasts is identical in each region, the types of neurons that are generated by a given NB are region-specific. We have determined how these two axes intersect to generate diversity in the progeny of NBs. For instance, Hth NBs generate ~800 columnar Mi1 neurons throughout the larval crescent while non-columnar Pm3 neurons, which are much fewer, are only produced by Hth NBs in the center Vsx1 region. In these central NBs, co-expression of Vsx1 and Hth leads to the neuronal expression of Seven-up and Prospero to direct the formation of ~70 non-columnar Pm3 neurons. Thus, Pm3 specification is the product of combinatorial input from both the NB (Hth) and dorsal-ventral (Vsx1) axes. We have extended this analysis to the other NBs and have mapped 7 additional neuronal-types to distinct regions in the larval neuroepithelium. We are now investigating the lineage of medulla cells and their pattern of migration in order to define how retinotopy is established: we find that columnar neurons (e.g. Mi1) are born locally throughout the larval crescent whereas non-columnar neurons (e.g. Pm3) are generated by specialized regional NBs and then migrate to take up their final position in the adult medulla cortex.

Special neurogenesis in the tips of the medulla neuroepithelium: (Claire Bertet) The two tip regions of the medulla neuroepithelium are defined by Wingless expression and exhibit a very different type of neurogenesis than the main region of the medulla: NBs have a different temporal sequence of TFs expression and, unlike the main region, only one temporal factor seems to be important for neuronal specification. In addition, NBs in the tip regions change their mode of division over time, starting as type 0 NBs (where GMCs do not divide) before later switching to typical type 1 NBs. While the main region produces numerous medulla neuronal subtypes, the tip regions generate only a few types of medulla interneurons, mainly transmedullary and Pm neurons. This region also generates neurons of the lobula complex that are located in the proximity.

Neurogenesis of the lobula plate and medulla rim: (Nathalie Neriec) The central region of the optic lobe called the 'lobula plug', which is independent from the medulla neuroepithelium exhibits yet another type of neurogenesis. We are investigating how these cells are generated from a special class of migrating NBs. As for the medulla, the NBs undergo a (shorter) temporal progression by expressing specific TFs but then exhibit different modes of division (from type I to type II transit amplifying NBs). As they age, these NBs generate two distinct neuronal populations, one superficial that gives rise to the medulla rim, and a central population that gives rise to the lobula plate. We are analyzing how these differences lead to the generation of neurons with higher stoichiometry (more than one neuron per R7/R8 column).

Cell lineage of the optic lobe: (Alberto del Valle) We are systematically generating adult cell clones by using MARCM, Twin Spot MARCM, G-TRACE, Flybow, etc. This indicates that neuronal diversity is generated in the optic lobes through different strategies. In the lamina, specialized neuronal precursors that are induced to divide and differentiate by PRs appear to generate single populations of the same cell type. In the medulla, most NBs produce a many cell types. In the 'lobula plug', specialized NB gives rise to clusters of 4 neuron types with different ratios. We are correlating the TF code identified in larval NBs with the adult cell clones to provide a clear picture of how the optic lobe develops.

Profiling of medulla neurons: (Dominic Didiano) This project aims to decode the genetic basis of neural diversity with the long-term goal of manipulating gene expression to generate neurons with specific properties. We first identify lines that are expressed in one neuronal type and use them to GFP-label specific cells and sort them through FACS. We then use RNA-seq to determine the specific transcriptome of each cell type. We aim to elucidate the mechanisms by which specific TFs control the specific features of each neuron from arborization to neurotransmitter and neurotransmitter receptor expression.

Lamina neuron subtype specification: (Zhenqing Chen) This project aims to uncover the neurogenesis of lamina structure in the Drosophila optic lobe. There are five major types of lamina neurons, which are organized in cylindrical groups called cartridges and maintain contact with R1-R6 projections from the retina. During Drosophila development, the lamina structure is derived from one edge of the outer proliferative center (OPC) in the larval optic lobe under the tight control of Hedgehog (HH) and Epidermal Growth Factor (EGF) signaling from the innervating photoreceptor axons. We are now focusing on how the cell fates of different types of lamina neurons are specified during development.

Behavioral analysis of color vision: (Nina Vogt) To correlate the anatomy of medulla neurons with their function, we develop a behavioral assay for color vision. Previously, a choice paradigm based on the innate preference for shorter wavelengths has been employed to study the role of different photoreceptors or medulla neurons. However, this behavior is strongly dependent on intensity. Alternatively, we use an operant learning paradigm. We train flies in a LED arena using heat punishment to distinguish between two colors, independently of their respective intensities. Silencing of different sets of medulla neurons using our collection of cell type specific GAL4 lines allows us to address their function in true color vision. [Yamaguchi PNAS 2008; Yamaguchi PNAS 2010]

Electrophysiology of medulla neurons: (Rudy Behnia) In an effort to understand the function of specific optic lobe neurons in the processing of visual information, we have developed an in vivo system in which the activity of medulla neurons is measured in response to specific patterns of light using GFP-targeted whole-cell patch-clamp recording. We are currently focusing on three columnar cell types: Mi1, Tm1 and Tm2 neurons, whose anatomy suggests involvement in motion detection. Indeed, Mi1 is postsynaptic to lamina neuron L1 while Tm1 and Tm2 are both postsynaptic to lamina neuron L2. L1 and L2 are the first steps in the motion detection pathway. We have determined the response of Mi1, Tm1 and Tm2 to brightness increments and decrements, as well as a other visual stimuli. Further experiments examining the spatial and temporal properties of these cells will uncover the neural substrate for motion detection in flies. These recordings are now being extended to candidate color sensitive neurons in the medulla.