Network Adaptation Improves Temporal Representation of Naturalistic Stimuli in Drosophila Eye: I Dynamics

Because of the limited processing capacity of eyes, retinal networks must adapt constantly to best present the ever changing visual world to the brain. However, we still know little about how adaptation in retinal networks shapes neural encoding of changing information. To study this question, we recorded voltage responses from photoreceptors (R1–R6) and their output neurons (LMCs) in the Drosophila eye to repeated patterns of contrast values, collected from natural scenes. By analyzing the continuous photoreceptor-to-LMC transformations of these graded-potential neurons, we show that the efficiency of coding is dynamically improved by adaptation. In particular, adaptation enhances both the frequency and amplitude distribution of LMC output by improving sensitivity to under-represented signals within seconds. Moreover, the signal-to-noise ratio of LMC output increases in the same time scale. We suggest that these coding properties can be used to study network adaptation using the genetic tools in Drosophila, as shown in a companion paper (Part II).     

Synaptic limitations to contrast coding in the retina of the blowfly Calliphora

We investigate the effects of synaptic transmission on early visual processing by examining the passage of signals from photoreceptors to second order neurons (LMCS). We concentrate on the roles played by three properties of synaptic transmission: (1) the shape of the characteristic curve, relating pre- and postsynaptic signal amplitudes, (2) the dynamics of synaptic transmission and (3) the noise introduced during transmission. The characteristic curve is sigmoidal and follows a simple model of synaptic transmission (Appendix) in which transmitter release rises exponentially with presynaptic potential. According to this model a presynaptic depolarization of 1.50-1.86 mV produces an e-fold increase in postsynaptic conductance. The characteristic curve generates a sigmoidal relation between postsynaptic (LMC) response amplitude and stimulus contrast. The shape and slope of the characteristic curve is unaffected by the state of light adaptation. Retinal antagonism adjusts the characteristic curve to keep it centred on the mean level of receptor response generated by the background. Thus the photoreceptor synapses operate in the mid-region of the curve, where the slope or gain is highest and equals approximately 6. The dynamics of transmission of a signal from photoreceptor to second-order neuron approximates to the sum of two processes with exponential time courses. A momentary receptor depolarization generates a postsynaptic hyperpolarization of time constant 0.5-1.0 ms, followed by a slower and weaker depolarization. Light adaptation increases the relative amplitude of the depolarizing process and reduces its time constant from 80 ms to 1.5 ms. The hyperpolarizing process is too rapid to bandlimit receptor signals. The noise introduced during the passage of the signal from receptor to second-order neuron is measured by comparing signal:noise ratios and noise power spectra in the two cell types. Under daylight conditions from 50 to 70% of the total noise power is generated by events associated with the transmission of photoreceptor signals and the generation of LMC responses. According to the exponential model of transmitter release, the effects of synaptic noise are minimized when synaptic gain is maximized. Moreover, both retinal antagonism and the sigmoidal shape of the characteristic curve promote synaptic gain. We conclude that retinal antagonism and nonlinear synaptic amplification act in concert to protect receptor signals from contamination by synaptic noise. This action may explain the widespread occurrence of these processes in early visual processing.     

Assessing the Role of Cell-Surface Molecules in Central Synaptogenesis in the Drosophila Visual System

A hallmark of the central nervous system is its spatial and functional organization in synaptic layers. During neuronal development, axons form transient contacts with potential post-synaptic elements and establish synapses with appropriate partners at specific layers. These processes are regulated by synaptic cell-adhesion molecules. In the Drosophila visual system, R7 and R8 photoreceptor subtypes target distinct layers and form en passant pre-synaptic terminals at stereotypic loci of the axonal shaft. A leucine-rich repeat transmembrane protein, Capricious (Caps), is known to be selectively expressed in R8 axons and their recipient layer, which led to the attractive hypothesis that Caps mediates R8 synaptic specificity by homophilic adhesion. Contradicting this assumption, our results indicate that Caps does not have a prominent role in synaptic-layer targeting and synapse formation in Drosophila photoreceptors, and that the specific recognition of the R8 target layer does not involve Caps homophilic axon-target interactions. We generated flies that express a tagged synaptic marker to evaluate the presence and localization of synapses in R7 and R8 photoreceptors. These genetic tools were used to assess how the synaptic profile is affected when axons are forced to target abnormal layers by expressing axon guidance molecules. When R7 axons were mistargeted to the R8-recipient layer, R7s either maintained an R7-like synaptic profile or acquired a similar profile to r8s depending on the overexpressed protein. When R7 axons were redirected to a more superficial medulla layer, the number of presynaptic terminals was reduced. These results indicate that cell-surface molecules are able to dictate synapse loci by changing the axon terminal identity in a partially cell-autonomous manner, but that presynapse formation at specific sites also requires complex interactions between pre- and post-synaptic elements.     

Generating oscillatory bursts from a network of regular spiking neurons without inhibition

Avian nucleus isthmi pars parvocellularis (Ipc) neurons are reciprocally connected with the layer 10 (L10) neurons in the optic tectum and respond with oscillatory bursts to visual stimulation. Our in vitro experiments show that both neuron types respond with regular spiking to somatic current injection and that the feedforward and feedback synaptic connections are excitatory, but of different strength and time course. To elucidate mechanisms of oscillatory bursting in this network of regularly spiking neurons, we investigated an experimentally constrained model of coupled leaky integrate-and-fire neurons with spike-rate adaptation. The model reproduces the observed Ipc oscillatory bursting in response to simulated visual stimulation. A scan through the model parameter volume reveals that Ipc oscillatory burst generation can be caused by strong and brief feedforward synaptic conductance changes. The mechanism is sensitive to the parameter values of spike-rate adaptation. In conclusion, we show that a network of regular-spiking neurons with feedforward excitation and spike-rate adaptation can generate oscillatory bursting in response to a constant input.     

Feedback network controls photoreceptor output at the layer of first visual synapses in Drosophila

At the layer of first visual synapses, information from photoreceptors is processed and transmitted towards the brain. In fly compound eye, output from photoreceptors (R1-R6) that share the same visual field is pooled and transmitted via histaminergic synapses to two classes of interneuron, large monopolar cells (LMCs) and amacrine cells (ACs). The interneurons also feed back to photoreceptor terminals via numerous ligand-gated synapses, yet the significance of these connections has remained a mystery. We investigated the role of feedback synapses by comparing intracellular responses of photoreceptors and LMCs in wild-type Drosophila and in synaptic mutants, to light and current pulses and to naturalistic light stimuli. The recordings were further subjected to rigorous statistical and information-theoretical analysis. We show that the feedback synapses form a negative feedback loop that controls the speed and amplitude of photoreceptor responses and hence the quality of the transmitted signals. These results highlight the benefits of feedback synapses for neural information processing, and suggest that similar coding strategies could be used in other nervous systems.     

Hallucinogen persisting perception disorder in neuronal networks with adaptation

We study the spatiotemporal dynamics of neuronal networks with spike frequency adaptation. In particular, we compare the effects of adaptation being either a linear or nonlinear function of neural activity. We find that altering parameters controlling the strength of synaptic connections in the network can lead to spatially structured activity suggestive of symptoms of hallucinogen persisting perception disorder (HPPD). First, we study how both networks track spatially homogeneous flickering stimuli, and find input is encoded as continuous at lower flicker frequencies when the network??s synapses exhibit more net excitation. Mainly, we study instabilities of stimulus-driven traveling pulse solutions, representative of visual trailing phenomena common to HPPD patients. Visual trails are reported as discrete afterimages in the wake of a moving input. Thus, we analyze several solutions arising in response to moving inputs in both networks: an ON state, stimulus-locked pulses, and traveling breathers. We find traveling breathers can arise in both networks when an input moves beyond a critical speed. These possible neural substrates of visual trails occur at slower speeds when the modulation of synaptic connectivity is increased.     

Neural field model of binocular rivalry waves

We present a neural field model of binocular rivalry waves in visual cortex. For each eye we consider a one-dimensional network of neurons that respond maximally to a particular feature of the corresponding image such as the orientation of a grating stimulus. Recurrent connections within each one-dimensional network are assumed to be excitatory, whereas connections between the two networks are inhibitory (cross-inhibition). Slow adaptation is incorporated into the model by taking the network connections to exhibit synaptic depression. We derive an analytical expression for the speed of a binocular rivalry wave as a function of various neurophysiological parameters, and show how properties of the wave are consistent with the wave-like propagation of perceptual dominance observed in recent psychophysical experiments. In addition to providing an analytical framework for studying binocular rivalry waves, we show how neural field methods provide insights into the mechanisms underlying the generation of the waves. In particular, we highlight the important role of slow adaptation in providing a “symmetry breaking mechanism” that allows waves to propagate.     

A Change in the Ion Selectivity of Ligand-Gated Ion Channels Provides a Mechanism to Switch Behavior

Behavioral output of neural networks depends on a delicate balance between excitatory and inhibitory synaptic connections. However, it is not known whether network formation and stability is constrained by the sign of synaptic connections between neurons within the network. Here we show that switching the sign of a synapse within a neural circuit can reverse the behavioral output. The inhibitory tyramine-gated chloride channel, LGC-55, induces head relaxation and inhibits forward locomotion during the Caenorhabditis elegans escape response. We switched the ion selectivity of an inhibitory LGC-55 anion channel to an excitatory LGC-55 cation channel. The engineered cation channel is properly trafficked in the native neural circuit and results in behavioral responses that are opposite to those produced by activation of the LGC-55 anion channel. Our findings indicate that switches in ion selectivity of ligand-gated ion channels (LGICs) do not affect network connectivity or stability and may provide an evolutionary and a synthetic mechanism to change behavior.     

A Model of Self-Organizing Head-Centered Visual Responses in Primate Parietal Areas

We present a hypothesis for how head-centered visual representations in primate parietal areas could self-organize through visually-guided learning, and test this hypothesis using a neural network model. The model consists of a competitive output layer of neurons that receives afferent synaptic connections from a population of input neurons with eye position gain modulated retinal receptive fields. The synaptic connections in the model are trained with an associative trace learning rule which has the effect of encouraging output neurons to learn to respond to subsets of input patterns that tend to occur close together in time. This network architecture and synaptic learning rule is hypothesized to promote the development of head-centered output neurons during periods of time when the head remains fixed while the eyes move. This hypothesis is demonstrated to be feasible, and each of the core model components described is tested and found to be individually necessary for successful self-organization.     

Analysis of growth cone extension in standardized coordinates highlights self-organization rules during wiring of the Drosophila visual system

A fascinating question in neuroscience is how ensembles of neurons, originating from different locations, extend to the proper place and by the right time to create precise circuits. Here, we investigate this question in the Drosophila visual system, where photoreceptors re-sort in the lamina to form the crystalline-like neural superposition circuit. The repeated nature of this circuit allowed us to establish a data-driven, standardized coordinate system for quantitative comparison of sparsely perturbed growth cones within and across specimens. Using this common frame of reference, we investigated the extension of the R3 and R4 photoreceptors, which is the only pair of symmetrically arranged photoreceptors with asymmetric target choices. Specifically, we found that extension speeds of the R3 and R4 growth cones are inherent to their cell identities. The ability to parameterize local regularity in tissue organization facilitated the characterization of ensemble cellular behaviors and dissection of mechanisms governing neural circuit formation.     

Multiple death pathways in retina-derived 661W cells following growth factor deprivation: crosstalk between caspases and calpains

During development of the mammalian retina, neurons that do not succeed in establishing functional synaptic connections are eliminated by apoptosis, allowing the formation of a finely tuned network. Growth factors play a crucial role in controlling the balance between apoptosis and survival signals not only at developmental stages but also in long-term preservation of retinal functions. In the present work, we explore the apoptotic mechanisms triggered by growth factor deprivation of retina-derived 661W cells. Under serum starvation conditions, these cone photoreceptors underwent cell death with participation of caspase-9, -3 and -12. Interestingly, inhibition of caspases did not prevent apoptosis but only resulted in a temporary delay. We show m-calpain activation in parallel with caspases, indicating that more than one execution pathway is available to cone photoreceptors. Moreover, crosstalk of the caspase and calpain pathways was detected, suggesting a loop that may act to amplify the apoptotic cascade.     

Emergence of Functional Specificity in Balanced Networks with Synaptic Plasticity

In rodent visual cortex, synaptic connections between orientation-selective neurons are unspecific at the time of eye opening, and become to some degree functionally specific only later during development. An explanation for this two-stage process was proposed in terms of Hebbian plasticity based on visual experience that would eventually enhance connections between neurons with similar response features. For this to work, however, two conditions must be satisfied: First, orientation selective neuronal responses must exist before specific recurrent synaptic connections can be established. Second, Hebbian learning must be compatible with the recurrent network dynamics contributing to orientation selectivity, and the resulting specific connectivity must remain stable for unspecific background activity. Previous studies have mainly focused on very simple models, where the receptive fields of neurons were essentially determined by feedforward mechanisms, and where the recurrent network was small, lacking the complex recurrent dynamics of large-scale networks of excitatory and inhibitory neurons. Here we studied the emergence of functionally specific connectivity in large-scale recurrent networks with synaptic plasticity. Our results show that balanced random networks, which already exhibit highly selective responses at eye opening, can develop feature-specific connectivity if appropriate rules of synaptic plasticity are invoked within and between excitatory and inhibitory populations. If these conditions are met, the initial orientation selectivity guides the process of Hebbian learning and, as a result, functionally specific and a surplus of bidirectional connections emerge. Our results thus demonstrate the cooperation of synaptic plasticity and recurrent dynamics in large-scale functional networks with realistic receptive fields, highlight the role of inhibition as a critical element in this process, and paves the road for further computational studies of sensory processing in neocortical network models equipped with synaptic plasticity.     

The cadherin Flamingo mediates level-dependent interactions that guide photoreceptor target choice in Drosophila

Quantitative differences in cadherin activity have been proposed to play important roles in patterning connections between pre- and postsynaptic neurons. However, no examples of such a function have yet been described, and the mechanisms that would allow such differences to direct growth cones to specific synaptic targets are unknown. In the Drosophila visual system, photoreceptors are genetically programmed to make a complex, stereotypic set of synaptic connections. Here we show that the atypical cadherin Flamingo functions as a short-range, homophilic signal, passing between specific R cell growth cones to influence their choice of postsynaptic partners. We find that individual growth cones are sensitive to differences in Flamingo activity through opposing interactions between neighboring cells and require these interactions to be balanced in order to extend along the appropriate trajectory.     

Shaping embodied neural networks for adaptive goal-directed behavior

The acts of learning and memory are thought to emerge from the modifications of synaptic connections between neurons, as guided by sensory feedback during behavior. However, much is unknown about how such synaptic processes can sculpt and are sculpted by neuronal population dynamics and an interaction with the environment. Here, we embodied a simulated network, inspired by dissociated cortical neuronal cultures, with an artificial animal (an animat) through a sensory-motor loop consisting of structured stimuli, detailed activity metrics incorporating spatial information, and an adaptive training algorithm that takes advantage of spike timing dependent plasticity. By using our design, we demonstrated that the network was capable of learning associations between multiple sensory inputs and motor outputs, and the animat was able to adapt to a new sensory mapping to restore its goal behavior: move toward and stay within a user-defined area. We further showed that successful learning required proper selections of stimuli to encode sensory inputs and a variety of training stimuli with adaptive selection contingent on the animat's behavior. We also found that an individual network had the flexibility to achieve different multi-task goals, and the same goal behavior could be exhibited with different sets of network synaptic strengths. While lacking the characteristic layered structure of in vivo cortical tissue, the biologically inspired simulated networks could tune their activity in behaviorally relevant manners, demonstrating that leaky integrate-and-fire neural networks have an innate ability to process information. This closed-loop hybrid system is a useful tool to study the network properties intermediating synaptic plasticity and behavioral adaptation. The training algorithm provides a stepping stone towards designing future control systems, whether with artificial neural networks or biological animats themselves.     

Adaptive synaptogenesis constructs networks that maintain information and reduce statistical dependence

This report demonstrates the effectiveness of two processes in constructing simple feedforward networks which perform good transformations on their inputs. Good transformations are characterized by the minimization of two information measures: the information loss incurred with the transformation and the statistical dependency of the output. The two processes build appropriate synaptic connections in initially unconnected networks. The first process, synaptogenesis, creates new synaptic connections; the second process, associative synaptic modification, adjusts the connection strength of existing synapses. Synaptogenesis produces additional innervation for each output neuron until each output neuron achieves a firing rate of approximately 0.50. Associative modification of existing synaptic connections lends robustness to network construction by adjusting suboptimal choices of initial synaptic weights. Networks constructed using synaptogenesis and synaptic modification successfully preserve the information content of a variety of inputs. By recording a high-dimensional input into an output of much smaller dimension, these networks drastically reduce the statistical dependence of neuronal representations. Networks constructed with synaptogenesis and associative modification perform good transformations over a wide range of neuron firing thresholds.     

Synaptic connections of calbindin-immunoreactive cone bipolar cells in the inner plexiform layer of rabbit retina

In the mammalian retina, information concerning various aspects of an image is transferred in parallel, and cone bipolar cells are thought to play a major role in this parallel processing. We have examined the synaptic connections of calbindin-immunoreactive (IR) ON cone bipolar cells in the inner plexiform layer (IPL) of rabbit retina and have compared these synaptic connections with those that we have previously described for neurokinin 1 (NK1) receptor-IR cone bipolar cells. A total of 325 synapses made by calbindin-IR bipolar axon terminals have been identified in sublamina b of the IPL. The axons of calbindin-IR bipolar cells receive synaptic inputs from amacrine cells through conventional synapses and are coupled to putative AII amacrine cells via gap junctions. The major output from calbindin-IR bipolar cells is to amacrine cell processes. These data resemble our findings for NK1 receptor-IR bipolar cells. However, the incidences of output synapses to ganglion cell dendrites of calbindin-IR bipolar cells are higher compared with the NK1-receptor-IR bipolar cells. On the basis of stratification level and synaptic connections, calbindin-IR ON cone bipolar cells might thus play an important role in the processing of various visual aspects, such as contrast, orientation, and approach sensing, and in transferring rod signals to the ON cone pathway.     

Different photoreceptors within the same retina express unique combinations of potassium channels

Single electrode current and voltage clamp recordings in Calliphora, and whole-cell voltage clamp recordings in Drosophila were used to characterise the voltage-gated K channels in both major classes of photoreceptors, R7/8 (long visual fibres, LVFs) and R1-6 (short visual fibres, SVFs). R7/8 were identified by their unique spectral properties, ca. 3–4 fold higher input resistances and 3–4 fold lower cell capacitance. In Calliphora SVFs possess both fast and slow activating delayed rectifier potassium conductances. Drosophila SVFs possess a slowly inactivating delayed rectifier (I_Ks), a very rapidly inactivating A channel encoded by the Shaker gene (I_A), and, in a minority of cells, a third K conductance with intermediate kinetics (I_Kf). In both specs the LVFs lack the slowest component, but exhibit the faster K conductance(s) with properties indistinguishable from those in the SVFs. These findings add to established evidence demonstrating the significant role played by potassium channels in tuning the photoreceptor membrane. The results also suggest that R1-6 photoreceptors and R7/8 form inputs to visual subsystems tuned to different temporal frequencies.Abbreviations LVF long visual fibre - SVF short visual fibre - R1-6 retinular cells 1 to 6 inclusive - R7/8 retinular cell 7 and 8 - I _A rapidly inactivating A type potassium conductance; channel coded by Shaker gene - I _Kf rapidly activating, slowly inactivating delayed rectifier-like potassium conductance - I _Ks slowly activating, slowly inactivating delayed rectifier-like potassium conductance - I _KDs slowly activating delayed rectifier potassium conductance - I _KDf rapidly activating delayed rectifier potassium conductance