Search

Neuronal computations underlying higher brain functions depend on synaptic interactions among specific neurons. A mechanistic understanding of such computations requires wiring diagrams of neuronal networks. In this study, we examined how the olfactory bulb (OB) performs ‘whitening’, a fundamental computation that decorrelates activity patterns and supports their classification by memory networks. We measured odor-evoked activity in the OB of a zebrafish larva and subsequently reconstructed the complete wiring diagram by volumetric electron microscopy. The resulting functional connectome revealed an over-representation of multisynaptic connectivity motifs that mediate reciprocal inhibition between neurons with similar tuning. This connectivity suppressed redundant responses and was necessary and sufficient to reproduce whitening in simulations. Whitening of odor representations is therefore mediated by higher-order structure in the wiring diagram that is adapted to natural input patterns. The authors measure evoked activity and perform dense reconstruction of the olfactory bulb wiring diagram in a zebrafish larva, uncovering a mechanism for whitening, a computation that decorrelates activity for pattern classification by memory networks.

Inference of action potentials (‘spikes’) from neuronal calcium signals is complicated by the scarcity of simultaneous measurements of action potentials and calcium signals (‘ground truth’). In this study, we compiled a large, diverse ground truth database from publicly available and newly performed recordings in zebrafish and mice covering a broad range of calcium indicators, cell types and signal-to-noise ratios, comprising a total of more than 35 recording hours from 298 neurons. We developed an algorithm for spike inference (termed CASCADE) that is based on supervised deep networks, takes advantage of the ground truth database, infers absolute spike rates and outperforms existing model-based algorithms. To optimize performance for unseen imaging data, CASCADE retrains itself by resampling ground truth data to match the respective sampling rate and noise level; therefore, no parameters need to be adjusted by the user. In addition, we developed systematic performance assessments for unseen data, openly released a resource toolbox and provide a user-friendly cloud-based implementation. Rupprecht et al. compiled a large database of simultaneous electrophysiological and calcium recordings from the same neurons. An algorithm (termed CASCADE) trained with this ground truth enables reliable spike inference without the need to tune parameters.

Summary Inhibitory neurons orchestrate the activity of excitatory neurons and play key roles in circuit function. Although individual interneurons have been studied extensively, little is known about their properties at the population level. Using random-access 3D two-photon microscopy, we imaged local populations of cerebellar Golgi cells (GoCs), which deliver inhibition to granule cells. We show that population activity is organized into multiple modes during spontaneous behaviors. A slow, network-wide common modulation of GoC activity correlates with the level of whisking and locomotion, while faster (, Graphical abstract, Highlights • Inhibitory circuit population activity is organized on multiple spatiotemporal scales • Slow circuit-wide Golgi cell activation reflects general level of behavioral activity • Multidimensional differential population activity encodes behavioral information • Electrically coupled Golgi cell circuit model reproduces population-level properties, Inhibitory interneurons orchestrate the activity of neural circuits, but little is known about their population dynamics. By using 3D random-access calcium imaging, Gurnani and Silver show that cerebellar Golgi cell circuits exhibit multidimensional activity with common and distributed modes. A biologically detailed circuit model implicates electrical coupling in shaping the population dynamics.

The zebrafish is an important model in systems neuroscience but a key limitation is the lack of viral tools to dissect the structure and function of neuronal circuitry. We developed methods for efficient gene transfer and retrograde tracing in adult and larval zebrafish by herpes simplex viruses (HSV1). HSV1 can be combined with the Gal4/UAS system to target cell types with high spatial, temporal and molecular specificity. We also established methods for efficient transneuronal tracing by modified rabies viruses in zebrafish. We demonstrate that HSV1 and rabies viruses can be used to visualize and manipulate genetically or anatomically identified neurons within and across different brain areas of adult and larval zebrafish. An expandable library of viruses is provided to express fluorescent proteins, calcium indicators, optogenetic probes, toxins and other molecular tools. This toolbox creates new opportunities to interrogate neuronal circuits in zebrafish through combinations of genetic and viral approaches.

The dense reconstruction of neuronal wiring diagrams from volumetric electron microscopy data has the potential to generate fundamentally new insights into mechanisms of information processing and storage in neuronal circuits. Zebrafish provide unique opportunities for dynamical connectomics approaches that combine reconstructions of wiring diagrams with measurements of neuronal population activity and behavior. Such approaches have the power to reveal higher-order structure in wiring diagrams that cannot be detected by sparse sampling of connectivity and that is essential for neuronal computations. In the brain stem, recurrently connected neuronal modules were identified that can account for slow, low-dimensional dynamics in an integrator circuit. In the spinal cord, connectivity specifies functional differences between premotor interneurons. In the olfactory bulb, tuning-dependent connectivity implements a whitening transformation that is based on the selective suppression of responses to overrepresented stimulus features. These findings illustrate the potential of dynamical connectomics in zebrafish to analyze the circuit mechanisms underlying higher-order neuronal computations.

Calcium imaging is a key method to record patterns of neuronal activity across populations of identified neurons. Inference of temporal patterns of action potentials (‘spikes’) from calcium signals is, however, challenging and often limited by the scarcity of ground truth data containing simultaneous measurements of action potentials and calcium signals. To overcome this problem, we compiled a large and diverse ground truth database from publicly available and newly performed recordings. This database covers various types of calcium indicators, cell types, and signal-to-noise ratios and comprises a total of >35 hours from 298 neurons. We then developed a novel algorithm for spike inference (CASCADE) that is based on supervised deep networks, takes advantage of the ground truth database, infers absolute spike rates, and outperforms existing model-based algorithms. To optimize performance for unseen imaging data, CASCADE retrains itself by resampling ground truth data to match the respective sampling rate and noise level. As a consequence, no parameters need to be adjusted by the user. To facilitate routine application of CASCADE we developed systematic performance assessments for unseen data, we openly release all resources, and we provide a user-friendly cloud-based implementation.

SummaryAnimal bodies are composed of hundreds of cell types that differ in location, morphology, cytoarchitecture, and physiology. This is reflected by cell type-specific transcription factors and downstream effector genes implementing functional specialisation. Here, we establish and explore the link between cell type-specific gene expression and subcellular morphology for the entire body of the marine annelidPlatynereis dumerilii. For this, we registered a whole-body cellular expression atlas to a high-resolution electron microscopy dataset, automatically segmented all cell somata and nuclei, and clustered the cells according to gene expression or morphological parameters. We show that collective gene expression most efficiently identifies spatially coherent groups of cells that match anatomical boundaries, which indicates that combinations of regionally expressed transcription factors specify tissue identity. We provide an integrated browser as a Fiji plugin to readily explore, analyse and visualise multimodal datasets with remote on-demand access to all available datasets.

SummaryIntelligent behavior requires a comparison between the predicted and the actual consequences of behavioral actions. According to the theory of predictive processing, this comparison relies on a neuronal error signal that reflects the mismatch between an internal prediction and sensory input. Inappropriate error signals may generate pathological experiences in neuropsychiatric conditions. To examine the processing of sensorimotor prediction errors across different telencephalic brain areas we optically measured neuronal activity in head-fixed, adult zebrafish in a virtual reality. Brief perturbations of visuomotor feedback triggered distinct changes in swimming behavior and different neuronal responses. Neuronal activity reflecting sensorimotor mismatch, rather than sensory input or motor output alone, was prominent throughout multiple forebrain areas. This activity preceded and predicted the transition in motor behavior. Error signals were altered in specific forebrain regions by a mutation in the autism-related gene shank3b. Predictive processing is therefore a widespread phenomenon that may contribute to disease phenotypes.

Neuronal computations underlying higher brain functions depend on synaptic interactions among specific neurons. A mechanistic understanding of such computations requires wiring diagrams of neuronal networks. We examined how the olfactory bulb (OB) performs ‘whitening’, a fundamental computation that decorrelates activity patterns and supports their classification by memory networks. We measured odor-evoked activity in the OB of a zebrafish larva and subsequently reconstructed the complete wiring diagram by volumetric electron microscopy. The resulting functional connectome revealed an overrepresentation of multisynaptic connectivity motifs that mediate reciprocal inhibition between neurons with similar tuning. This connectivity suppressed redundant responses and was necessary and sufficient to reproduce whitening in simulations. Whitening of odor representations is therefore mediated by higher-order structure in the wiring diagram that is adapted to natural input patterns.

Fixation and staining of large tissue samples are critical for the acquisition of volumetric electron microscopic image datasets and the subsequent reconstruction of neuronal circuits. Efficient protocols exist for the staining of small samples, but uniform contrast is often difficult to achieve when the sample diameter exceeds a few hundred micrometers. Recently, a protocol (BROPA, brain-wide reduced-osmium staining with pyrogallol-mediated amplification) was developed that achieves homogeneous staining of the entire mouse brain but requires very long sample preparation times. By exploring modifications of this protocol we developed a substantially faster procedure, fBROPA, that allows for reliable high-quality staining of tissue blocks on the millimeter scale. Modifications of the original BROPA protocol include drastically reduced incubation times and a lead aspartate incubation to increase sample conductivity. Using this procedure, whole brains from adult zebrafish were stained within 4 days. Homogenous high-contrast staining was achieved throughout the brain. High-quality image stacks with voxel sizes of 10 × 10 × 25 nm3 were obtained by serial block-face imaging using an electron dose of ~15 e-/nm2. No obvious reduction in staining quality was observed in comparison to smaller samples stained by other state-of-the-art procedures. Furthermore, high-quality images with minimal charging artifacts were obtained from non-neural tissues with low membrane density. fBROPA is therefore likely to be a versatile and efficient sample preparation protocol for a wide range of applications in volume electron microscopy.

We present SBEMimage, an open-source Python-based application to operate serial block-face electron microscopy (SBEM) systems. SBEMimage is designed for complex, challenging acquisition tasks, such as large-scale volume imaging of neuronal tissue or other biological ultrastructure. Advanced monitoring, process control, and error handling capabilities improve reliability, speed, and quality of acquisitions. Debris detection, autofocus, real-time image inspection, and various other quality control features minimize the risk of data loss during long-term acquisitions. Adaptive tile selection allows for efficient imaging of large tissue volumes of arbitrary shape. The software’s graphical user interface is optimized for remote operation. In its user-friendly viewport, tile grids covering the region of interest to be acquired are overlaid on previously acquired overview images of the sample surface. Images from other sources, e.g., light microscopes, can be imported and superimposed. SBEMimage complements existing DigitalMicrograph (Gatan Microscopy Suite) installations on 3View systems but permits higher acquisition rates by interacting directly with the microscope’s control software. Its modular architecture and the use of Python/PyQt make SBEMimage highly customizable and extensible, which allows for fast prototyping and will permit adaptation to a wide range of SBEM systems and applications.

Summary Neuronal computations critically depend on the connectivity rules that govern the convergence of excitatory and inhibitory synaptic signals onto individual neurons. To examine the functional synaptic organization of a distributed memory network, we performed voltage clamp recordings in telencephalic area Dp of adult zebrafish, the homolog of olfactory cortex. In neurons of posterior Dp, odor stimulation evoked large, recurrent excitatory and inhibitory inputs that established a transient state of high conductance and synaptic balance. Excitation and inhibition in individual neurons were co-tuned to different odors and correlated on slow and fast timescales. This precise synaptic balance implies specific connectivity among Dp neurons, despite the absence of an obvious topography. Precise synaptic balance stabilizes activity patterns in different directions of coding space and in time while preserving high bandwidth. The coordinated connectivity of excitatory and inhibitory subnetworks in Dp therefore supports fast recurrent memory operations.

Sensory systems balance stability and plasticity to optimize stimulus representations in dynamic environments. We studied these processes in the olfactory system of adult zebrafish. Activity patterns evoked by repeated odor stimulation were measured by multiphoton calcium imaging in the olfactory bulb (OB) and in telencephalic area Dp, the homolog of olfactory cortex. Whereas odor responses in the OB were highly reproducible, responses of Dp neurons adapted over trials and exhibited substantial variability that could be attributed to ongoing activity and to systematic changes in neuronal representations following each stimulus. An NMDA receptor antagonist did not affect the magnitude of odor responses but strongly reduced the variability and experience-dependent modification of odor responses in Dp. As a consequence, odor representations became stable over trials. These results demonstrate that odor representations in higher brain areas are continuously modified by experience, supporting the view that olfactory processing is inseparable from memory, even in the absence of reinforcement.

Neuronal circuits in the olfactory bulb transform odor-evoked activity patterns across the input channels, the olfactory glomeruli, into distributed activity patterns across the output neurons, the mitral cells. One computation associated with this transformation is a decorrelation of activity patterns representing similar odors. Such a decorrelation has various benefits for the classification and storage of information by associative networks in higher brain areas. Experimental results from adult zebrafish show that pattern decorrelation involves a redistribution of activity across the population of mitral cells. These observations imply that pattern decorrelation cannot be explained by a global scaling mechanism but that it depends on interactions between distinct subsets of neurons in the network. This article reviews insights into the network mechanism underlying pattern decorrelation and discusses recent results that link pattern decorrelation in the olfactory bulb to odor discrimination behavior.

Learning is mediated by experience-dependent plasticity in neuronal circuits. Activity in neuronal circuits is tightly regulated by different subtypes of inhibitory interneurons, yet their role in learning is poorly understood. Using a combination of in vivo single-unit recordings and optogenetic manipulations, we show that in the mouse basolateral amygdala, interneurons expressing parvalbumin (PV) and somatostatin (SOM) bidirectionally control the acquisition of fear conditioning--a simple form of associative learning--through two distinct disinhibitory mechanisms. During an auditory cue, PV(+) interneurons are excited and indirectly disinhibit the dendrites of basolateral amygdala principal neurons via SOM(+) interneurons, thereby enhancing auditory responses and promoting cue-shock associations. During an aversive footshock, however, both PV(+) and SOM(+) interneurons are inhibited, which boosts postsynaptic footshock responses and gates learning. These results demonstrate that associative learning is dynamically regulated by the stimulus-specific activation of distinct disinhibitory microcircuits through precise interactions between different subtypes of local interneurons.

Large-scale reconstructions of neuronal populations are critical for structural analyses of neuronal cell types and circuits. Dense reconstructions of neurons from image data require ultrastructural resolution throughout large volumes, which can be achieved by automated volumetric electron microscopy (EM) techniques. We used serial block face scanning EM (SBEM) and conductive sample embedding to acquire an image stack from an olfactory bulb (OB) of a zebrafish larva at a voxel resolution of 9.25×9.25×25 nm3. Skeletons of 1,022 neurons, 98% of all neurons in the OB, were reconstructed by manual tracing and efficient error correction procedures. An ergonomic software package, PyKNOSSOS, was created in Python for data browsing, neuron tracing, synapse annotation, and visualization. The reconstructions allow for detailed analyses of morphology, projections and subcellular features of different neuron types. The high density of reconstructions enables geometrical and topological analyses of the OB circuitry. Image data can be accessed and viewed through the neurodata web services ( http://www.neurodata.io ). Raw data and reconstructions can be visualized in PyKNOSSOS. Machine-accessible metadata file describing the reported data (ISA-Tab format)

This chapter describes three fast and straightforward methods to introduce nucleic acids, dyes, and other molecules into small numbers of cells of zebrafish embryos, larvae, and adults using electroporation. These reagents are delivered through a glass micropipette and electrical pulses are given through electrodes to permeabilize cell membranes and promote uptake of the reagent. This technique allows the experimenter to target cells of their choice at a particular time of development and at a particular location in the zebrafish with high precision and facilitates long-term noninvasive measurement of biological activities in vivo. Applications include cell fate mapping, neural circuit mapping, neuronal activity measurement, manipulation of activity, ectopic gene expression, and genetic knockdown experiments.

International audience; There is a high demand for 3D multiphoton imaging in neuroscience and other fields but scanning in axial direction presents technical challenges. We developed a focusing technique based on a remote movable mirror that is conjugate to the specimen plane and translated by a voice coil motor. We constructed cost-effective z-scanning modules from off-the-shelf components that can be mounted onto standard multiphoton laser scanning microscopes to extend scan patterns from 2D to 3D. Systems were designed for large objectives and provide high resolution, high speed and a large z-scan range (>300 μm). We used these systems for 3D multiphoton calcium imaging in the adult zebrafish brain and measured odor-evoked activity patterns across >1500 neurons with single-neuron resolution and high signal-to-noise ratio.

The dense reconstruction of neuronal circuits from volumetric electron microscopy (EM) data has the potential to uncover fundamental structure-function relationships in the brain. To address bottlenecks in the workflow of this emerging methodology, we developed a procedure for conductive sample embedding and a pipeline for neuron reconstruction. We reconstructed ∼98% of all neurons (>1,000) in the olfactory bulb of a zebrafish larva with high accuracy and annotated all synapses on subsets of neurons representing different types. The organization of the larval olfactory bulb showed marked differences from that of the adult but similarities to that of the insect antennal lobe. Interneurons comprised multiple types but granule cells were rare. Interglomerular projections of interneurons were complex and bidirectional. Projections were not random but biased toward glomerular groups receiving input from common types of sensory neurons. Hence, the interneuron network in the olfactory bulb exhibits a specific topological organization that is governed by glomerular identity.

This chapter describes three fast and straightforward methods to introduce nucleic acids, dyes, and other molecules into small numbers of cells of zebrafish embryos, larvae, and adults using electroporation. These reagents are delivered through a glass micropipette and electrical pulses are given through electrodes to permeabilize cell membranes and promote uptake of the reagent. This technique allows the experimenter to target cells of their choice at a particular time of development and at a particular location in the zebrafish with high precision and facilitates long-term noninvasive measurement of biological activities in vivo. Applications include cell fate mapping, neural circuit mapping, neuronal activity measurement, manipulation of activity, ectopic gene expression, and genetic knockdown experiments.

Optogenetic approaches allow the manipulation of neuronal activity patterns in space and time by light, particularly in small animals such as zebrafish. However, most techniques cannot control neuronal activity independently at different locations. Here we describe equipment and provide a protocol for single-photon patterned optical stimulation of neurons using a digital micromirror device (DMD). This method can create arbitrary spatiotemporal light patterns with spatial and temporal resolutions in the micrometer and submillisecond range, respectively. Different options to integrate a DMD into a multiphoton microscope are presented and compared. We also describe an ex vivo preparation of the adult zebrafish head that greatly facilitates optogenetic and other experiments. After assembly, the initial alignment takes about one day and the zebrafish preparation takes

SummaryThe ability to detect and avoid predators is essential to survival. Various animals, from sea urchins to damselfly larvae, use injury of conspecifics to infer the presence of predators [1–7]. In many fish [1, 8, 9], skin damage causes the release of chemicals that elicit escape and fear in members of the shoal. The chemical nature of the alarm substance (“Schreckstoff” in German) [1], the neural circuits mediating the complex response, and the evolutionary origins of a signal with little obvious benefit to the sender, are unresolved. To address these questions, we use biochemical fractionation to molecularly characterize Schreckstoff. Although hypoxanthine-3 N-oxide has been proposed to be the alarm substance [10, 11], it has not been reliably detected in the skin [12] and there may be other active components [13, 14]. We show that the alarm substance is a mixture that includes the glycosaminoglycan (GAG) chondroitin. Purified chondroitins trigger fear responses. Like skin extract, chondroitins activate the mediodorsal posterior olfactory bulb, a region innervated by crypt neurons [15] that has a unique projection to the habenula [16]. These findings establish GAGs as a new class of odorants in fish, which trigger alarm behavior possibly via a specialized circuit.Video Abstract

In the olfactory bulb, the modulatory neurotransmitter dopamine (DA) is coexpressed with GABA by local interneurons, but its role in odor processing remains obscure. We examined functions of DA mediated by D2-like receptors in the olfactory bulb of adult zebrafish by pharmacology, whole-cell recordings, calcium imaging, and optogenetics. Bath application of DA had no detectable effect on odorant-evoked sensory input. DA directly hyperpolarized mitral cells (MCs) via D2-like receptors and slightly increased their response gain. Consistent with this effect on input–output functions of MCs, small odorant responses were suppressed, whereas strong responses were enhanced in the presence of DA. These effects increased the root-mean-square contrast of population activity patterns but did not reduce their correlations. Optical stimulation of interneurons expressing channelrhodopsin-2 evoked fast GABAergic inhibitory currents in mitral cells but failed to activate D2receptor-mediated currents when stimuli were short. Prolonged stimulus trains, however, activated a slow hyperpolarizing current that was blocked by an antagonist of D2-like receptors. GABA and DA are therefore both released from interneurons by electrical activity and hyperpolarize MCs, but D2-dependent dopaminergic effects occur on slower timescales. Additional effects of DA may be mediated by D1-like receptors. These results indicate that DA acts on D2-like receptors via asynchronous release and/or volume transmission and implicate DA in the slow adaptation of circuit function. The shift of the membrane potential away from spike threshold could adapt mitral cells to background input without compromising their sensitivity.

Neuronal activity patterns contain information in their temporal structure, indicating that information transfer between neurons may be optimized by temporal filtering. In the zebrafish olfactory bulb, subsets of output neurons (mitral cells) engage in synchronized oscillations during odour responses, but information about odour identity is contained mostly in non-oscillatory firing rate patterns. Using optogenetic manipulations and odour stimulation, we found that firing rate responses of neurons in the posterior zone of the dorsal telencephalon (Dp), a target area homologous to olfactory cortex, were largely insensitive to oscillatory synchrony of mitral cells because passive membrane properties and synaptic currents act as low-pass filters. Nevertheless, synchrony influenced spike timing. Moreover, Dp neurons responded primarily during the decorrelated steady state of mitral cell activity patterns. Temporal filtering therefore tunes Dp neurons to components of mitral cell activity patterns that are particularly informative about precise odour identity. These results demonstrate how temporal filtering can extract specific information from multiplexed neuronal codes. Optogenetic stimulation in the zebrafish olfactory bulb and downstream read out of activity in the homologue of olfactory cortex demonstrate how temporal filtering can extract specific components of neuronal codes. Neurons in the olfactory bulb of zebrafish and other species respond to odours with complex firing that contains both oscillatory and non-oscillatory components, but the functional importance of these different components is unclear. Using optogenetic stimulation, Rainer Friedrich and colleagues examine directly the downstream readout of oscillatory bulb activity in the posterior zone of the dorsal telencephalon (Dp), a zebrafish homologue of the olfactory cortex. Dp neurons were largely insensitive to synchronized oscillatory activity, but respond to steady-state activity. These findings could be accounted for by low-pass filter-like biophysical properties of Dp neurons, and demonstrate how filtering can extract individual components of neuronal codes.

Although projections from the insect antennal lobe to the mushroom body are probabilistic, those to the lateral horn are stereotyped, suggesting an interplay of preconfigured and plastic circuits in olfactory processing.

Decorrelation is a fundamental computation that optimizes the format of neuronal activity patterns. Channel decorrelation by adaptive mechanisms results in efficient coding, whereas pattern decorrelation facilitates the readout and storage of information. Mechanisms achieving pattern decorrelation, however, remain unclear. We developed a theoretical framework that relates high-dimensional pattern decorrelation to neuronal and circuit properties in a mathematically stringent fashion. For a generic class of random neuronal networks, we proved that pattern decorrelation emerges from neuronal nonlinearities and is amplified by recurrent connectivity. This mechanism does not require adaptation of the network, is enhanced by sparse connectivity, depends on the baseline membrane potential and is robust. Connectivity measurements and computational modeling suggest that this mechanism is involved in pattern decorrelation in the zebrafish olfactory bulb. These results reveal a generic relationship between the structure and function of neuronal circuits that is probably relevant for pattern processing in various brain areas.

The categorial nature of sensory, cognitive and behavioural acts indicates that the brain classifies neuronal activity patterns into discrete representations. Pattern classification may be achieved by abrupt switching between discrete activity states of neuronal circuits, but few experimental studies have directly tested this. We gradually varied the concentration or molecular identity of odours and optically measured responses across output neurons of the olfactory bulb in zebrafish. Whereas population activity patterns were largely insensitive to changes in odour concentration, morphing of one odour into another resulted in abrupt transitions between odour representations. These transitions were mediated by coordinated response changes among small neuronal ensembles rather than by shifts in the global network state. The olfactory bulb therefore classifies odour-evoked input patterns into many discrete and defined output patterns, as proposed by attractor models. This computation is consistent with perceptual phenomena and may represent a general information processing strategy in the brain.

A central goal of modern neuroscience is to obtain a mechanistic understanding of higher brain functions under healthy and diseased conditions. Addressing this challenge requires rigorous experimental and theoretical analysis of neuronal circuits. Recent advances in optogenetics, high-resolution in vivo imaging, and reconstructions of synaptic wiring diagrams have created new opportunities to achieve this goal. To fully harness these methods, model organisms should allow for a combination of genetic and neurophysiological approaches in vivo. Moreover, the brain should be small in terms of neuron numbers and physical size. A promising vertebrate organism is the zebrafish because it is small, it is transparent at larval stages and it offers a wide range of genetic tools and advantages for neurophysiological approaches. Recent studies have highlighted the potential of zebrafish for exhaustive measurements of neuronal activity patterns, for manipulations of defined cell types in vivo and for studies of causal relationships between circuit function and behavior. In this article, we summarize background information on the zebrafish as a model in modern systems neuroscience and discuss recent results.

The early processing of sensory information by neuronal circuits often includes a reshaping of activity patterns that may facilitate further processing in the brain. For instance, in the olfactory system the activity patterns that related odors evoke at the input of the olfactory bulb can be highly similar. Nevertheless, the corresponding activity patterns of the mitral cells, which represent the output of the olfactory bulb, can differ significantly from each other due to strong inhibition by granule cells and peri-glomerular cells. Motivated by these results we study simple adaptive inhibitory networks that aim to separate or even orthogonalize activity patterns representing similar stimuli. Since the animal experiences the different stimuli at different times it is difficult for the network to learn the connectivity based on their similarity; biologically it is more plausible that learning is driven by simultaneous correlations between the input channels. We investigate the connection between pattern orthogonalization and channel decorrelation and demonstrate that networks can achieve effective pattern orthogonalization through channel decorrelation if they simultaneously equalize their output levels. In feedforward networks biophysically plausible learning mechanisms fail, however, for even moderately similar input patterns. Recurrent networks do not have that limitation; they can orthogonalize the representations of highly similar input patterns. Even when they are optimized for linear neuronal dynamics they perform very well when the dynamics are nonlinear. These results provide insights into fundamental features of simplified inhibitory networks that may be relevant for pattern orthogonalization by neuronal circuits in general.

The brain generates coherent perceptions of objects from elementary sensory inputs. To examine how higher-order representations of smells arise from the activation of discrete combinations of glomeruli, we analyzed transformations of activity patterns between the zebrafish olfactory bulb and two of its telencephalic targets, Vv and Dp. Vv is subpallial whereas Dp is the homolog of olfactory cortex. Both areas lack an obvious topographic organization but perform complementary computations. Responses to different odors and their mixtures indicate that Vv neurons pool convergent inputs, resulting in broadened tuning curves and overlapping odor representations. Neuronal circuits in Dp, in contrast, produce a mixture of excitatory and inhibitory synaptic inputs to each neuron that controls action potential firing in an odor-dependent manner. This mechanism can extract information about combinations of molecular features from ensembles of active and inactive mitral cells, suggesting that pattern processing in Dp establishes representations of odor objects.

gamma-Aminobutyric acid (GABA)ergic synapses are thought to play pivotal roles in the processing of activity patterns in the olfactory bulb (OB), but their functions have been difficult to study during odor responses in the intact system. We pharmacologically manipulated GABA(A) and GABA(B) receptors in the OB of zebrafish and analysed the effects on odor responses of the output neurons, the mitral cells (MCs), by electrophysiological recordings and temporally deconvolved two-photon Ca2+ imaging. The blockade of GABA(B) receptors enhanced presynaptic Ca2+ influx into afferent axon terminals, and changed the amplitude and time course of a subset of MC responses, indicating that GABA(B) receptors have a modulatory influence on OB output activity. The blockade of GABA(A) receptors induced epileptiform firing, enhanced excitatory responses and abolished fast oscillations in the local field potential. Moreover, the topological reorganization and decorrelation of MC activity patterns during the initial phase of the response was perturbed. These results indicate that GABA(A) receptor-containing circuits participate in the balance of excitation and inhibition, the regulation of total OB output activity, the synchronization of odor-dependent neuronal ensembles, and the reorganization of odor-encoding activity patterns. GABA(A) and GABA(B) receptors are therefore differentially involved in multiple functions of neuronal circuits in the OB.

Thrombin, the ultimate proteinase of the coagulation cascade, is an attractive target for the treatment of a variety of cardiovascular diseases. Previously, a series of novel thrombin inhibitors, discovered by employing a powerful and new computer-assisted multiparameter optimization process (CADDIS), have been synthesized. We have now crystallized the complex of human alpha-thrombin with the most potent of these inhibitors, 8-5 (K(i)=3 nM), and have determined its 2.3A X-ray crystal structure. The Fourier map displayed clear electron density for the inhibitor. The central part of the inhibitor binds in an improved melagatran-like mode, while the structure identifies a d-tyrosine as P1 residue which forms a charged hydrogen bond with Asp 189 of thrombin. This is the first crystal structure of a thrombin-inhibitor complex, where an uncharged inhibitor residue makes hydrogen bonds within the thrombin S1 pocket. Additionally, novel favourable intermolecular hydrogen bonds of the inhibitor with the thrombin backbone become possible due to the d-configuration of the P1 residue. Two flanking voluminous side chains increase the strength of the subjacent hydrogen bonding system by shielding it from the bulk solvent.

In the adult olfactory bulb (OB) of vertebrates, local GABAergic interneurons (INs) mediate recurrent and lateral inhibition between the principal neurons of the OB, the mitral cells (MCs), and play pivotal roles in the processing of odor-evoked activity patterns. The properties and functions of INs in the developing OB are, however, not well understood. We studied the functional development of INs in the OB of living zebrafish larvae 3-6 days postfertilization using anatomical techniques and in-vivo two-photon Ca2+ imaging. We identified MCs and INs by cell-type-specific expression of transgenic fluorescent markers and found that the IN:MC ratio was lower than in the adult fish. Moreover, the fraction of INs responding with Ca2+ signals to a set of natural odors was substantially lower than in adults. Odors of different chemical classes evoked overlapping patterns of Ca2+ signals that were concentrated in the center of the IN layer. The GABA(A) receptor agonists GABA and muscimol strongly suppressed odor responses, whereas a GABA(A) receptor antagonist enhanced responses and altered the spatial distribution of odor-evoked activity. These results indicate that IN odor responses at early developmental stages are sparse and exhibit no obvious chemotopic organization. Nevertheless, GABAergic signaling is already inhibitory at early stages of OB development and strongly influences odor-evoked activity patterns. Hence, INs already participate in the processing of odor information at very early stages of OB development even though the majority of INs emerge only at later stages.

Although all sensory circuits ascend to higher brain areas where stimuli are represented in sparse, stimulus-specific activity patterns, relatively little is known about sensory coding on the descending side of neural circuits, as a network converges. In insects, mushroom bodies (MBs) have been an important model system for studying sparse coding in the olfactory system1–3, where this format is important for accurate memory formation4–6. In Drosophila, it has recently been shown that the 2000 Kenyon cells (KCs) of the MB converge onto a population of only 35 MB output neurons (MBONs), that fall into 22 anatomically distinct cell types7,8. Here we provide the first comprehensive view of olfactory representations at the fourth layer of the circuit, where we find a clear transition in the principles of sensory coding. We show that MBON tuning curves are highly correlated with one another. This is in sharp contrast to the process of progressive decorrelation of tuning in the earlier layers of the circuit2,9. Instead, at the population level, odor representations are reformatted so that positive and negative correlations arise between representations of different odors. At the single-cell level, we show that uniquely identifiable MBONs display profoundly different tuning across different animals, but tuning of the same neuron across the two hemispheres of an individual fly was nearly identical. Thus, individualized coordination of tuning arises at this level of the olfactory circuit. Furthermore, we find that this individualization is an active process that requires a learning-related gene, rutabaga. Ultimately, neural circuits have to flexibly map highly stimulus-specific information in sparse layers onto a limited number of different motor outputs. The reformatting of sensory representations we observe here may mark the beginning of this sensory-motor transition in the olfactory system.

The molecular chaperone Hsp90 is required for the folding and activation of a large number of substrate proteins. These are involved in essential cellular processes ranging from signal transduction to viral replication. For the activation of its substrates, Hsp90 binds and hydrolyzes ATP, which is the key driving force for conformational conversions within the dimeric chaperone. Dimerization of Hsp90 is mediated by a C-terminal dimerization site. In addition, there is a transient ATP-induced dimerization of the two N-terminal ATP-binding domains. The resulting ring-like structure is thought to be the ATPase-active conformation. Hsp90 is a slow ATPase with a turnover number of 1 ATP/min for the yeast protein. A key question for understanding the molecular mechanism of Hsp90 is how ATP hydrolysis is regulated and linked to conformational changes. In this study, we analyzed the activation process structurally and biochemically with a view to identify the conformational limitations of the ATPase reaction cycle. We showed that the first 24 amino acids stabilize the N-terminal domain in a rigid state. Their removal confers flexibility specifically to the region between amino acids 98 and 120. Most surprisingly, the deletion of this structure results in the complete loss of ATPase activity and in increased N-terminal dimerization. Complementation assays using heterodimeric Hsp90 show that this rigid lid acts as an intrinsic kinetic inhibitor of the Hsp90 ATPase cycle preventing N-terminal dimerization in the ground state. On the other hand, this structure acts, in concert with the 24 N-terminal amino acids of the other N-terminal domain, to form an activated ATPase and thus regulates the turnover number of Hsp90.

Thrombin generation and fibrinogen (Fbg) clotting are the ultimate proteolytic reactions in the blood coagulation pathway. Staphylocoagulase (SC), a protein secreted by the human pathogen Staphylococcus aureus, activates prothrombin (ProT) without proteolysis. The SC.(pro)thrombin complex recognizes Fbg as a specific substrate, converting it directly into fibrin. The crystal structure of a fully active SC fragment containing residues 1-325 (SC-(1-325)) bound to human prethrombin 2 showed previously that SC inserts its Ile(1)-Val(2) N terminus into the Ile(16) pocket of prethrombin 2, inducing a functional active site in the cognate zymogen conformationally. Exosite I of alpha-thrombin, the Fbg recognition site, and proexosite I on ProT are blocked by domain 2 of SC-(1-325). In the present studies, active site-labeled fluorescent ProT analogs were used to quantitate Fbg binding to the SC-(1-325).ProT complex. Fbg binding and cleavage are mediated by expression of a new Fbg-binding exosite on the SC-(1-325).ProT complex, resulting in formation of an (SC-(1-325).ProT)(2).Fbg pentameric complex with a dissociation constant of 8-34 nm. In both crystal structures, the SC-(1-325).(pre)thrombin complexes form dimers, with both proteinases/zymogens facing each other over a large U-shaped cleft, through which the Fbg substrate could thread. On this basis, a molecular model of the pentameric (SC-(1-325).thrombin)(2).Fbg encounter complex was generated, which explains the coagulant properties and efficient Fbg conversion. The results provide new insight into the mechanism that mediates high affinity Fbg binding and cleavage as a substrate of SC.(pro)thrombin complexes, a process that is central to the molecular pathology of S. aureus endocarditis.

Staphylocoagulase (SC) is a protein secreted by the human pathogen, Staphylococcus aureus, that activates human prothrombin (ProT) by inducing a conformational change. SC-bound ProT efficiently clots fibrinogen, thus bypassing the physiological blood coagulation pathway. The crystal structure of a fully active SC fragment, SC-(1-325), bound to human prethrombin 2 showed that the SC-(1-325) N terminus inserts into the Ile(16) pocket of prethrombin 2, thereby inducing expression of a functional catalytic site in the cognate zymogen without peptide bond cleavage. As shown here, SC-(1-325) binds to bovine and human ProT with similar affinity but activates the bovine zymogen only very poorly. By contrast to the approximately 2-fold difference in chromogenic substrate kinetic constants between human thrombin and the SC-(1-325).human (pro)thrombin complexes, SC-(1-325).bovine ProT shows a 3,500-fold lower k(cat)/K(m) compared with free bovine thrombin, because of a 47-fold increase in K(m) and a 67-fold decrease in k(cat). The SC-(1-325).bovine ProT complex is approximately 5,800-fold less active compared with its human counterpart. Comparison of human and bovine fibrinogen as substrates of human and bovine thrombin and the SC-(1-325).(pro)thrombin complexes indicates that the species specificity of SC-(1-325) cofactor activity is determined primarily by differences in conformational activation of bound ProT. These results suggest that the catalytic site in the SC-(1-325).bovine ProT complex is incompletely formed. The current crystal structure of SC-(1-325).bovine thrombin reveals that SC would dock similarly to the bovine proenzyme, whereas the bovine (pro)thrombin-characteristic residues Arg(144) and Arg(145) would likely interfere with insertion of the SC N terminus, thus explaining the greatly reduced activation of bovine ProT.

In the adult olfactory bulb (OB), particular chemical classes of odorants preferentially activate glomeruli within loosely defined regions, resulting in a coarse and fractured “chemotopic” map. In zebrafish, amino acids and bile acids predominantly stimulate glomeruli in the lateral and medial OB, respectively. We studied the development of these spatial response maps in zebrafish. At 3 d postfertilization (dpf), the OB contained protoglomerular structures that became refined and more numerous during subsequent days. In a transgenic zebrafish line expressing the Ca(2+) indicator protein inverse pericam, mainly in mitral cells, odor responses in the OB were first detected at 2.5-3 dpf. Already at this stage, amino acids and bile acids evoked activity predominantly in the lateral and medial OB, respectively. Two-photon Ca(2+) imaging using a synthetic indicator was used to reconstruct activity patterns at higher resolution in three dimensions. Responses to amino acids and bile acids were detected predominantly in the lateral and medial OB, respectively, with little overlap. Between 2.5 and 6 dpf, the number of odor-responsive units increased, but the overall spatial organization of activity persisted. Hence, a coarse spatial organization of functional activity maps is established very early during OB development when glomeruli are not yet differentiated. This spatial organization is maintained during development and growth of neuronal circuits and may have important functions for odor processing in larvae, for the differentiation of glomeruli, and for the refinement of activity maps at later developmental stages.

Summary The recent Cell Symposium ‘Genes, Circuits and Behavior' brought together researchers working on neural circuits in vertebrate and invertebrate species. In the interest of fostering communication across the ‘backbone-divide', we asked a number of neuroscientists from both camps for their views on the extent to which insights obtained from studies on neural circuits in one type of species are transferable to the other.

Staphylocoagulase (SC) secreted by Staphylococcus aureus is a potent non-proteolytic activator of the blood coagulation zymogen prothrombin and the prototype of a newly established zymogen activator and adhesion protein (ZAAP) family. The conformationally activated SC.prothrombin complex specifically cleaves fibrinogen to fibrin, which propagates the growth of bacteria-fibrin-platelet vegetations in acute bacterial endocarditis. Our recent 2.2 A X-ray crystal structures of an active SC fragment [SC(1-325)] bound to the prothrombin zymogen catalytic domain, prethrombin 2, demonstrated that SC(1-325) represents a new type of non-proteolytic activator with a unique fold. The observed insertion of the SC(1-325) N-terminus into the 'Ile 16' cleft of prethrombin 2, which triggers the activating conformational change, provided the first unambiguous structural evidence for the 'molecular sexuality' mechanism of non-proteolytic zymogen activation. Based on the SC(1-325) fold, a new family of bifunctional zymogen activator and adhesion proteins was identified that possess N-terminal domains homologous to SC(1-325) and C-terminal domains that mediate adhesion to plasma or extracellular matrix proteins. Further investigation of the ZAAP family may lead to new insights into the mechanisms of bacterial factors that hijack zymogens of the human blood coagulation and fibrinolytic systems to promote and disseminate endocarditis and other infectious diseases.