Skip to main content

Home > Unit on Neuronal Connectivity

Development and Function of Drosophila Visual Circuits

Chi-Hon Lee, MD, PhD
  • Chi-Hon Lee, MD, PhD, Head, Section on Neuronal Connectivity
  • Chun-Yuan Ting, PhD, Research Fellow
  • Tzu-Yang Lin, PhD, Postdoctoral Fellow
  • Krishna Melnattur, PhD, Postdoctoral Fellow
  • Thangavel Karuppudurai, PhD, Postdoctoral Fellow
  • Mingmin Zhou, PhD, Postdoctoral Fellow
  • Moyi Li, BA, Biological Laboratory Technician

Using the Drosophila visual system as a model, we study how neurons form complex yet stereotyped synaptic connections during development and how the assembled neural circuits extract various visual attributes (such as color and motion) to guide animal behaviors. The fly's vision is mediated by three types of photoreceptors, R1-6, R7, and R8, which each respond to a specific spectrum of light. The three types of photoreceptor project their axons to retinotopically appropriate columns at three distinct layers in the brain, where the cells form synaptic connections with different target neurons. To determine the molecular mechanisms that wire photoreceptors and their synaptic targets, we took a forward-genetic approach and identified various loci that are required for guiding photoreceptor axons to appropriate target columns and layers. We are currently focusing on TGF-beta/Activin signaling pathway, which is required for restricting R7 axons to retintopically appropriate columns and a novel locus, overshoot, which regulates layer-specific targeting and synaptogenesis. To study visual circuit functions, we combined structural and functional approaches to map visual circuits. Using both light and electron microscopy studies, we identified the medulla neurons that are post-synaptic to photoreceptors. We are determining the role of these medulla neurons in processing motion and color information.

Molecular mechanisms regulating synaptic target selection of R7 photoreceptor axons

The chromatic photoreceptor neurons R7 and R8 project their axons to the M3 and M6 layers of the medulla neuropil, respectively, and each pair of R7 and R8 axons innervates a single medulla column. Together, these axonal termini tile the entire medulla neuropil. Such organizing features, called layer-specific targeting and retinotopic mapping, are the hallmark of all complex visual systems. We showed previously that, during development, R7 and R8 axons project to specific layers and columns in two distinct stages. The sequential targeting of R7 and R8 axons to columns and layers reduces the number of potential synaptic partners (from over sixty medulla neurons to several), effectively reducing wiring complexity.

By combining forward-genetic screens and behavioral assays, we identified a number of mutants in which layer-specific targeting or retinotopic mapping of R7s is disrupted. We are currently focusing on two loci: overshoot (osh) and premature extension (pex). The hypomorphic osh mutants exhibit a novel phenotype: osh-mutant R7s overshoot the M6 target layer and form additional synapses in the deeper medulla layer. Null osh–mutant R7 has a much stronger phenotype than that of the hypomorphic mutants. The mutant R7s exhibit abnormally enlarged synaptic boutons with excessive levels of the presynaptic structural protein Bruchpilot. The overshoot and enlarged synaptic bouton phenotype appears during the synaptogenesis stage around 80 hours after pupal formation. Both developmental staging and phenotype suggest that osh plays a regulatory role in synaptogenesis. By positional cloning, we located the osh locus to a genomic region of approximately 50 Kb on the right arm of the second chromosome and, by complementation, identified a candidate gene. We are in the process of confirming the identity of the osh gene by genomic rescue approach.

Premature extension or pex mutations affect the refinement of the R7 retinotopic map. During the larval stage, a gross retinotopic map of R7/R8 termini first forms via a Wnt4-dependent mechanism. During the late pupal stage, when the R7 and R8 growth cones regain motility and proceed to their final destined layer, the R7 and R8 growth cones are confined to their retinotopically appropriate columns to form synaptic connections, a process called retinotopic map refinement. Pex is a temperature-sensitive allele of Baboon, which encodes a type I TGF-beta/Activin receptor. Baboon-mutant R7 growth cones invade their neighboring columns, forming inappropriate synapses, indicating that Activin signaling is required to confine R7 growth cones to their retinotopically appropriate columns. We further determined that other known components of the canonical Activin signaling pathway, such as the downstream transcription factor Smad2, are required cell-autonomously in R7s for retinotopic map refinement. Immunohistochemistry analysis revealed that Smad2 is physically located at the R7 growth cones for receiving Activin signaling, which raises the question as to how the phosphorylated and hence activated Smad2 is transported from growth cones back to the nucleus to activate transcription. We found that the transport of Smad2 requires the Dynein/Dynactin-dependent retrograde axonal transport system. Disrupting retrograde transport by expressing a dominant negative mutant of p150glued phenocopies baboon. In addition, we found that Importin-alpha3, a known component of the nuclear import machinery, likely serves an unexpected role as a carrier to transport activated Smad2 from growth cones to the nucleus. Importin-alpha3–mutant R7s exhibit retinotopic map defects as seen in baboon mutants. Biochemical analysis further revealed that Importin-alpha3 co-localizes with Smad2 in the R7 growth cones, forming a physical complex. Finally, the nuclear accumulation of Smad2 is disrupted in the absence of Importin-alpha or baboon, indicating that the carrier Importin-alpha3 and Actin-signaling, and hence the phosphorylation of Smad2, are both required for transporting Smad2 from growth cones to the nucleus. We are now examining the expression profile of baboon and Importin-alpha3 mutants to identify the transcription targets of Activin signaling in R7s.

Mapping color-vision circuits

Previous primate electrophysiological and human psychophysical studies have provided great insights into the mechanisms of color vision. However, few hypotheses have been validated directly because of the complexity and variability of vertebrate neural circuits as well as the technical difficulty of establishing causality. To circumvent these problems, we use the Drosophila visual system as a model to study color vision. True color vision and high-order color vision functions have been demonstrated in many insects including bees and flies. We use a combination of molecular, genetic, histological, and behavioral approaches to determine the synaptic circuits involved in color vision and to identify the critical neurons that process color information.

To determine the synaptic circuits of color vision, we combined molecular-genetic and histological approaches. We are currently focusing on the medulla neuropil, which is analogous to the inner plexiform layer of the vertebrate retina. The medulla is innervated by the chromatic photoreceptors R7 and R8, as well the first-order interneurons (L1-5) of the achromatic photoreceptor R1-R6. Approximately sixty different types of medulla neurons process visual information carried by these afferents. To overcome the cellular complexity, we devised a "divide and conquer" strategy and subdivided the medulla neurons into several subclasses based on their use of neurotransmitters and receptors. We took advantage of the finding that fly photoreceptor neurons use histamine as neurotransmitter. Therefore, the first-order interneurons must express the histamine receptor (histamine-gated chloride channel or HisCl) in order to respond to histamine signal. Based on HisCl expression, we identified an amacrine neuronal type, Dm8, which relays R7 signal to projection neurons, and three types of projection neurons, Tm5, Tm9, and Tm20, that relay photoreceptor signal to the higher visual center—the lobula. Using the promoters of various transcription factors, neurotransmitter transporters, and synthesis enzymes, we further divided these neurons into subgroups. We found that the Tm5 can be further divided into three subtypes, Tm5a, Tm5b, and Tm5c, each of which has a unique axonal projection and dendritic arborization pattern correlated with its distinct gene expression profile.

Figure 1. Dual-View Imaging

Figure 1. Dual-View Imaging

Conventional confocal images suffer from anisotropic distortion, resulting in low resolution along the optic axis. The Z-distortion is especially severe when imaging thick samples. We developed a series of computer programs for combining two image stacks acquired orthogonally (frontal and horizontal view) to a higher resolution image. The improved images allow unambiguous tracing of fine dendritic processes of optic lobe neurons (Tm2, shown in the picture).

To characterize the dendritic branching pattern of different projection neurons, we developed an imaging technique called dual-view imaging, which generates high-resolution 3D images by combining two confocal image stacks collected in orthogonal orientations (Figure 1). Unlike typical confocal images, which have low axial resolution, the dual view images are isotropic. Using this technique, we characterized the dendritic branching pattern of Tm9 neurons as well as Tm2 neurons—the third-order interneurons involved in motion detection. We found that while different Tm neurons have stereotyped dendritic branching patterns, the detailed branching topology varies greatly from one neuron to another within a single neuronal type. We are now combining this technique with serial EM reconstruction to determine whether the connectivity is type invariant.

To determine the functions of distinct types of medulla neurons, we used a genetic technique to inactivate or restore their synaptic functions and examined the behavioral consequences—a procedure that allows us to assign specific functions to distinct neuronal subtypes, therefore establishing causality. The specific types of neurons amenable to this approach are limited by the specificity and diversity of genetic drivers available. The split-Gal4 system developed recently combines two promoters to enhance specificity, but the number of available drivers is rather limited. To overcome this problem, we developed a concatenated expression system, which is based on our split-LexA system and is compatible with existing Gal4 drivers abundantly available. Using this new expression system as well as the original split-Gal4 system, we have generated many genetic drivers that targeted specific medulla neurons. With these cell-specific drivers, we have begun to examine systematically whether a specific neuron subtype is "required" or "sufficient" for color and motion detection. We found that the amacrine neuron Dm8 is specifically required and sufficient for an animal's preference for UV light over green light but not for motion detection. Conversely, the lamina neurons L1 and L2 are required only for motion detection, not for color vision. In summary, our study validates the validity of our approach and reveals that different neuron subtypes subserve distinct visual functions.


  • Gao S, Takemura S-Y, Ting C-Y Huang S, Lu Z, Luan H, Rister J, Thum AS, Yang M, Hong S-T, Wang JW, Odenwald WF, White BH, Meinertzhagen IA, Lee C-H. The neural substrate of spectral preference in Drosophila. Neuron 2008; 60:328-342.
  • Root CM, Masuyama K, Green DS, Enell LE, Nassel DR, Lee C-H, Wang JW. A presynaptic gaincontrol mechanism fine-tunes olfactory behavior. Neuron 2008; 59:311-321.
  • Meinertzhagen IA, Takemura SY, Lu Z, Huang S, Gao S, Ting C-Y, Lee C-H. From form to function: the ways to know a neuron. J Neurogenet. 200; 23:68-77.
  • Hsu S-N, Yonekura S, Ting C-Y, Robertson HM, Iwai Y, Uemura T, Lee C-H, Chiba A. Conserved alternative splicing and expression patterns of arthropod N-cadherin. PLoS Genet 2009; 5:e1000441.


  • Ian Meinertzhagen, PhD, DSc, Dalhousie University, Halifax, Canada
  • Thomas Pohida, MSEE, Division of Computational Bioscience, CIT, NIH, Bethesda, MD
  • Randy Pursley, MSEE, Division of Computational Bioscience, CIT, NIH, Bethesda, MD
  • Mihaela Serpe, PhD, Program in Cellular Regulation and Metabolism, NICHD, Bethesda, MD
  • Mark Stopfer, PhD, Program in Developmental Neuroscience, NICHD, Bethesda, MD
  • Matthew McAuliffe, PhD, Division of Biomedical Imaging Research Services Section, CIT, NIH, Bethesda, MD
  • Nishith Pandya, BA, Division of Biomedical Imaging Research Services Section, CIT, NIH, Bethesda, MD
  • Benjamin White, PhD, Laboratory of Molecular Biology, NIMH, Bethesda, MD
  • Paul Smith, PhD, Laboratory of Bioengineering and Physical Science, NIBIB, NIH, Bethesda, MD

Top of Page