Competition and position-dependent targeting in the development of the Drosophila R7 visual projections

Visual circuit assembly in Drosophila

Both insect and vertebrate visual circuits are organized into orderly arrays of columnar and layered synaptic units that correspond to the array of photoreceptors in the eye. Recent genetic studies in Drosophila have yielded insights into the molecular and cellular mechanisms that pattern the layers and columns and establish specific connections within the synaptic units. A sequence of inductive events and complex cellular interactions coordinates the assembly of visual circuits. Photoreceptor-derived ligands, such as hedgehog and Jelly-Belly, induce target development and expression of specific adhesion molecules, which in turn serve as guidance cues for photoreceptor axons. Afferents are directed to specific layers by adhesive afferent-target interactions mediated by leucine-rich repeat proteins and cadherins, which are restricted spatially and/or modulated dynamically. Afferents are restricted to their topographically appropriate columns by repulsive interactions between afferents and by autocrine activin signaling. Finally, Dscam-mediated repulsive interactions between target neuron dendrites ensure appropriate combinations of postsynaptic elements at synapses. Essentially, all these Drosophila molecules have vertebrate homologs, some of which are known to carry out analogous functions. Thus, the studies of Drosophila visual circuit development would shed light on neural circuit assembly in general.

Optic lobe development

The optic lobes comprise approximately half of the fly's brain. In four major synaptic ganglia, or neuropils, the visual input from the compound eyes is received and processed for higher order visual functions like motion detection and color vision. A common characteristic of vertebrate and invertebrate visual systems is the point-to-point mapping of the visual world to synaptic layers in the brain, referred to as visuotopy. Vision requires the parallel extraction of numerous parameters in a visuotopic manner. Consequently, the optic neuropils are arranged in columns and perpendicularly oriented synaptic layers that allow for the selective establishment of synapses between columnar neurons. How this exquisite synaptic specificity is established during approximately 100 hours of brain development is still poorly understood. However, the optic lobe contains one of the best characterized brain structures in any organism-both anatomically and developmentally. Moreover, numerous molecules and their function illuminate some of the basic mechanisms involved in brain wiring. The emerging picture is that the development of the visual system of Drosophila is (epi-)genetically hard-wired; it supplies the emerging fly with vision without requiring neuronal activity for fine tuning of neuronal connectivity. Elucidating the genetic and cellular principles by which gene activity directs the assembly of the optic lobe is therefore a fascinating task and the focus of this chapter.

Synaptic patterning by morphogen signaling

Gradients of secreted small morphogenic molecules control cell proliferation and patterning throughout animal development. Recent years have seen the discovery of surprising roles for morphogens in later developmental processes, including axon pathfinding and synaptogenesis. The latest addition is a role for the TGF-beta superfamily morphogen Activin in synaptic patterning of the Drosophila visual system. In contrast to classical instructive and long-range morphogen gradients, Activin acts as a permissive and local motility restriction signal around several hundred individual photoreceptor axon terminals. Activin must therefore act in concert with other instructively attracting and repelling signals as part of a larger genetic program for brain wiring.

Tiling of r7 axons in the Drosophila visual system is mediated both by transduction of an activin signal to the nucleus and by mutual repulsion

The organization of neuronal wiring into layers and columns is a common feature of both vertebrate and invertebrate brains. In the Drosophila visual system, each R7 photoreceptor axon projects within a single column to a specific layer of the optic lobe. We refer to the restriction of terminals to single columns as tiling. In a genetic screen based on an R7-dependent behavior, we identified the Activin receptor Baboon and the nuclear import adaptor Importin-alpha3 as being required to prevent R7 axon terminals from overlapping with the terminals of R7s in neighboring columns. This tiling function requires the Baboon ligand, dActivin, the transcription factor, dSmad2, and retrograde transport from the growth cone to the R7 nucleus. We propose that dActivin is an autocrine signal that restricts R7 growth cone motility, and we demonstrate that it acts in parallel with a paracrine signal that mediates repulsion between R7 terminals.

The need for stochastic replication of ecological neural networks

Artificial neural networks are becoming increasingly popular as predictive statistical tools in ecosystem ecology and as models of signal processing in behavioural and evolutionary ecology. We demonstrate here that a commonly used network in ecology, the three-layer feed-forward network, trained with the backpropagation algorithm, can be extremely sensitive to the stochastic variation in training data that results from random sampling of the same underlying statistical distribution, with networks converging to several distinct predictive states. Using a random walk procedure to sample error-weight space, and Sammon dimensional reduction of weight arrays, we demonstrate that these different predictive states are not artefactual, due to local minima, but lie at the base of major error troughs in the error-weight surface. We further demonstrate that various gross weight compositions can produce the same predictive state, suggesting the analogy of weight space as a 'patchwork' of multiple predictive states. Our results argue for increased inclusion of stochastic training replication and analysis into ecological and behavioural applications of artificial neural networks.

Individuals from different-looking animal species may group together to confuse shared predators: simulations with artificial neural networks

Individuals of many quite distantly related animal species find each other attractive and stay together for long periods in groups. We present a mechanism for mixed-species grouping in which individuals from different-looking prey species come together because the appearance of the mixed-species group is visually confusing to shared predators. Using an artificial neural network model of retinotopic mapping in predators, we train networks on random projections of single- and mixed-species prey groups and then test the ability of networks to reconstruct individual prey items from mixed-species groups in a retinotopic map. Over the majority of parameter space, cryptic prey items benefit from association with conspicuous prey because this particular visual combination worsens predator targeting of cryptic individuals. However, this benefit is not mutual as conspicuous prey tends to be targeted most poorly when in same-species groups. Many real mixed-species groups show the asymmetry in willingness to initiate and maintain the relationship predicted by our study. The agreement of model predictions with published empirical work, the efficacy of our modelling approach in previous studies, and the taxonomic ubiquity of retinotopic maps indicate that we may have uncovered an important, generic selective agent in the evolution of mixed-species grouping.

Visual circuit development in Drosophila

Fly visual circuits are organized into lattice-like arrays and layers. Recent genetic studies have provided insights into how these reiterated structures are assembled through stepwise processes and how precise connections are established during development. Afferent-derived morphogens, such as Hedgehog, play a key role in organizing the overall structure by inducing and recruiting target neurons and glia. In turn, the target-derived ligand DWnt4 guides Frizzled2-expressing photoreceptor afferents to their proper destination. Photoreceptor afferents select specific synaptic targets by forming adhesive interactions and regulating actin cytoskeleton in growth cones. Target specificity is probably achieved by restricting the expression of adhesive molecules, such as Capricious, to appropriate presynaptic and postsynaptic partners, and by differentially regulating the function of broadly expressed adhesive molecules such as N-cadherin.

Artificial neural network properties associated with wiring patterns in the visual projections of vertebrates and arthropods

We model the functioning of different wiring schemes in visual projections using artificial neural networks and so speculate on selective factors underlying taxonomic variation in neural architecture. We model the high connective overlap of vertebrates (where networks have a dense mesh of connections) and the less overlapping, more modular architecture of arthropods. We also consider natural variation in these basic wiring schemes. Generally, arthropod networks are as efficient or more efficient in functioning compared to vertebrate networks. They do not show the confusion effect (decreasing targeting accuracy with increasing input group size), and they train as well or better. Arthropod networks are, however, generally poorer at reconstructing novel inputs. The ability of vertebrate networks to effectively process novel stimuli could promote behavioral sophistication and drive the evolution of vertebrate wiring schemes. Vertebrate networks with less connective overlap have, surprisingly, similar or superior properties compared to those with high connective overlap. Thus, the partial connective overlap seen in real vertebrate visual projections may be an optimal, evolved solution. Arthropod networks with and without whole-cell neural connections within neural layers have similar properties. This indicates that neural connections mediated by offshoots of single cells (dendrites) may be fundamental to generating the confusion effect.

The confusion effect in predatory neural networks

A simple artificial neural network model of image reconstruction in sensory maps is presented to explain the difficulty predators experience in targeting prey in large groups (the confusion effect). Networks are trained to reconstruct multiple randomly conformed "retinal" images of prey groups in an internal spatial map of their immediate environment. They are then used to simulate prey targeting by predators on groups of specific conformation. Networks trained with the biologically plausible associative reward-penalty method produce a more realistic model of the confusion effect than those trained with the popular but biologically implausible backpropagation method. The associative reward-penalty model makes the novel prediction that the accuracy-group size relationship is U shaped, and this prediction is confirmed by empirical data gathered from interactive computer simulation experiments with humans as "predators." The model further predicts all factors known from previous empirical work (and most factors suspected) to alleviate the confusion effect: increased relative intensity of the target object, heterogeneity of group composition, and isolation of the target. Interestingly, group compaction per se is not predicted to worsen predator confusion. This study indicates that the relatively simple, nonattentional mechanism of information degradation in the sensory mapping process is potentially important in generating the confusion effect.

Control of central synaptic specificity in insect sensory neurons

Synaptic specificity is the culmination of several processes, beginning with the establishment of neuronal subtype identity, followed by navigation of the axon to the correct subdivision of neuropil, and finally, the cell-cell recognition of appropriate synaptic partners. In this review we summarize the work on sensory neurons in crickets, cockroaches, moths, and fruit flies that establishes some of the principles and molecular mechanisms involved in the control of synaptic specificity. The identity of a sensory neuron is controlled by combinatorial expression of transcription factors, the products of patterning and proneural genes. In the nervous system, sensory axon projections are anatomically segregated according to modality, stimulus quality, and cell-body position. A variety of cell-surface and intracellular signaling molecules are used to achieve this. Synaptic target recognition is also controlled by transcription factors such as Engrailed and may be, in part, mediated by cadherin-like molecules.

Flamingo regulates R8 axon-axon and axon-target interactions in the Drosophila visual system

Photoreceptors (R cells) in the Drosophila retina connect to targets in three distinct layers of the optic lobe of the brain: R1-R6 connect to the lamina, and R7 and R8 connect to distinct layers in the medulla. In each of these layers, R axon termini are arranged in evenly spaced topographic arrays. In a genetic screen for mutants with abnormal R cell connectivity, we recovered mutations in flamingo (fmi). fmi encodes a seven-transmembrane cadherin, previously shown to function in planar cell polarity and in dendritic patterning. Here, we show that fmi has two specific functions in R8 axon targeting: it facilitates competitive interactions between adjacent R8 axons to ensure their correct spacing, and it promotes the formation of stable connections between R8 axons and their target cells in the medulla. The former suggests a general role for Fmi in establishing nonoverlapping dendritic and axonal target fields. The latter, together with the finding that N-Cadherin has an analogous role in R7 axon-target interactions, points to a cadherin-based system for target layer specificity in the Drosophila visual system.

Optic target regulation of NADPH-diaphorase by larval retinal axons in Drosophila

Development of the visual system in Drosophila requires the establishment of precise retinotopic connections between photoreceptors and their synaptic targets in the central nervous system. Nitric oxide (NO) has been implicated as a candidate signal involved in the establishment of retinal projection patterns. In this study the expression of NADPH-diaphorase in the lamina of Drosophila, and by implication nitric oxide synthase (NOS), was investigated in larvae with varying degrees of retinal innervation. NADPH-diaphorase expression was seen to increase in the lamina and eye disk following retinal neuronal death in eye specific pro-apoptotic larvae (pGMR-hid) compared to wild type larvae, and was lower in the lamina in absent or reduced retinal innervation mutants (eyes absent and sine oculis). Retinal innervation is seen to regulate the expression of NADPH-diaphorase expression in target structures.

Nitric oxide and cyclic GMP regulate retinal patterning in the optic lobe of Drosophila

The photoreceptors of Drosophila express a nitric oxide-sensitive guanylate cyclase during the first half of metamorphosis, when postsynaptic elements in the optic lobe are being selected. Throughout this period, the optic lobes show NADPH-diaphorase activity and stain with an antibody to nitric oxide synthase (NOS). The NOS inhibitor L-NAME, the NO scavenger PTIO, the sGC inhibitor ODQ, and methylene blue, which inhibits NOS and guanylate cyclase, each caused the disorganization of retinal projections and extension of photoreceptor axons beyond their normal synaptic layers in vitro. The disruptive effects of L-NAME were prevented with the addition of 8-bromo-cGMP. These results suggest NO and cGMP act to stabilize retinal growth cones at the start of synaptic assembly.

The genetics of visual system development in Drosophila: specification, connectivity and asymmetry

Encoding visual information requires a complex neuronal network. Recently, genes regulating early tissue specification, the growth of retinal target structures, the connectivity of photoreceptor axons, and mirror-image retinal symmetry in Drosophila have been identified. The insights gained from studying visual system development in flies promise to inform our understanding of similar processes in vertebrates.

four-jointed is required for intermediate growth in the proximal-distal axis in Drosophila

Genes capable of translating positional information into regulated growth lie at the heart of morphogenesis, yet few genes with this function have been identified. Mutants in the Drosophila four-jointed (fj) gene show reduced growth and altered differentiation only within restricted sectors of the proximal-distal (PD) axis in the leg and wing, thus fj is a candidate for a gene with this coordination function. Consistent with a position-sensitive role, we show that fj is expressed in a regional pattern in the developing leg, wing, eye and optic lobe. The fj gene encodes a novel type II membrane glycoprotein. When the cDNA is translated in an in vitro translation system in the presence of exogenous microsomal membranes, the intralumenal portion of some of the molecules is cleaved, yielding a secreted C-terminal fragment. We propose that fj encodes a secreted signal that functions as a positive regulator of regional growth and differentiation along the PD axis of the imaginal discs.

Restricted expression of the irreC-rst protein is required for normal axonal projections of columnar visual neurons

The 104 kDa irreC-rst protein, a member of the immunoglobulin superfamily, mediates homophilic adhesion in cell cultures. In larval optic chiasms, the protein is found on recently formed axon bundles, not on older ones. In developing visual neuropils, it is present in all columnar domains of specific layers. The number of irreC-rst-positive neuropil stratifications increases until the midpupal stage. Immunoreactivity fades thereafter. The functional importance of the restricted expression pattern is demonstrated by the severe projection errors of axons in the first and second optic chiasms in loss of function mutants and in transformants that express the irreC-rst protein globally. Epigenesis of the phenotypes can be explained partially on the bases of homophilic irreC-rst interactions.

Remodeling of the insect nervous system

Our nervous systems and behavior are shaped by hormonally driven developmental changes that continue beyond the embryonic period. Key insights into this process have emerged from studies of the insect nervous system. During insect metamorphosis, the nervous system is remodeled through postembryonic neurogenesis, programmed cell death and the modification of persistent neurons. These changes are regulated to a large degree by gene cascades that are triggered by steroid hormones, the ecdysteroids. Current studies are attempting to reveal the molecular mechanisms involved in regulating these dramatic examples of developmental plasticity.