Vertebrate behavioral thermoregulation: knowledge and future directions.

Thermoregulation is critical for survival across species. In animals, the nervous system detects external and internal temperatures, integrates this information with internal states, and ultimately forms a decision on appropriate thermoregulatory actions. Recent work has identified critical molecules and sensory and motor pathways controlling thermoregulation. However, especially with regard to behavioral thermoregulation, many open questions remain. Here, we aim to both summarize the current state of research, the "knowledge," as well as what in our mind is still largely missing, the "future directions." Given the host of circuit entry points that have been discovered, we specifically see that the time is ripe for a neuro-computational perspective on thermoregulation. Such a perspective is largely lacking but is increasingly fueled and made possible by the development of advanced tools and modeling strategies.

The preoptic area and dorsal habenula jointly support homeostatic navigation in larval zebrafish.

Animals must maintain physiological processes within an optimal temperature range despite changes in their environment. Through behavioral assays, whole-brain functional imaging, and neural ablations, we show that larval zebrafish, an ectothermic vertebrate, achieves thermoregulation through homeostatic navigation-non-directional and directional movements toward the temperature closest to its physiological setpoint. A brain-wide circuit encompassing several brain regions enables this behavior. We identified the preoptic area of the hypothalamus (PoA) as a key brain structure in triggering non-directional reorientation when thermal conditions are worsening. This result shows an evolutionary conserved role of the PoA as principal thermoregulator of the brain also in ectotherms. We further show that the habenula (Hb)-interpeduncular nucleus (IPN) circuit retains a short-term memory of the sensory history to support the generation of coherent directed movements even in the absence of continuous sensory cues. We finally provide evidence that this circuit may not be exclusive for temperature but may convey a more abstract representation of relative valence of physiologically meaningful stimuli regardless of their specific identity to enable homeostatic navigation.

Functional and pharmacological analyses of visual habituation learning in larval zebrafish.

Habituation allows animals to learn to ignore persistent but inconsequential stimuli. Despite being the most basic form of learning, a consensus model on the underlying mechanisms has yet to emerge. To probe relevant mechanisms, we took advantage of a visual habituation paradigm in larval zebrafish, where larvae reduce their reactions to abrupt global dimming (a dark flash). We used Ca2+ imaging during repeated dark flashes and identified 12 functional classes of neurons that differ based on their rate of adaptation, stimulus response shape, and anatomical location. While most classes of neurons depressed their responses to repeated stimuli, we identified populations that did not adapt or that potentiated their response. These neurons were distributed across brain areas, consistent with a distributed learning process. Using a small-molecule screening approach, we confirmed that habituation manifests from multiple distinct molecular mechanisms, and we have implicated molecular pathways in habituation, including melatonin, oestrogen, and GABA signalling. However, by combining anatomical analyses and pharmacological manipulations with Ca2+ imaging, we failed to identify a simple relationship between pharmacology, altered activity patterns, and habituation behaviour. Collectively, our work indicates that habituation occurs via a complex and distributed plasticity processes that cannot be captured by a simple model. Therefore, untangling the mechanisms of habituation will likely require dedicated approaches aimed at sub-component mechanisms underlying this multidimensional learning process.

Model discovery to link neural activity to behavioral tasks.

Brains are not engineered solutions to a well-defined problem but arose through selective pressure acting on random variation. It is therefore unclear how well a model chosen by an experimenter can relate neural activity to experimental conditions. Here, we developed 'model identification of neural encoding (MINE).' MINE is an accessible framework using convolutional neural networks (CNNs) to discover and characterize a model that relates aspects of tasks to neural activity. Although flexible, CNNs are difficult to interpret. We use Taylor decomposition approaches to understand the discovered model and how it maps task features to activity. We apply MINE to a published cortical dataset as well as experiments designed to probe thermoregulatory circuits in zebrafish. Here, MINE allowed us to characterize neurons according to their receptive field and computational complexity, features that anatomically segregate in the brain. We also identified a new class of neurons that integrate thermosensory and behavioral information that eluded us previously when using traditional clustering and regression-based approaches.

Functional and ultrastructural analysis of reafferent mechanosensation in larval zebrafish.

All animals need to differentiate between exafferent stimuli, which are caused by the environment, and reafferent stimuli, which are caused by their own movement. In the case of mechanosensation in aquatic animals, the exafferent inputs are water vibrations in the animal's proximity, which need to be distinguishable from the reafferent inputs arising from fluid drag due to locomotion. Both of these inputs are detected by the lateral line, a collection of mechanosensory organs distributed along the surface of the body. In this study, we characterize in detail how hair cells-the receptor cells of the lateral line-in zebrafish larvae discriminate between such reafferent and exafferent signals. Using dye labeling of the lateral line nerve, we visualize two parallel descending inputs that can influence lateral line sensitivity. We combine functional imaging with ultra-structural EM circuit reconstruction to show that cholinergic signals originating from the hindbrain transmit efference copies (copies of the motor command that cancel out self-generated reafferent stimulation during locomotion) and that dopaminergic signals from the hypothalamus may have a role in threshold modulation, both in response to locomotion and salient stimuli. We further gain direct mechanistic insight into the core components of this circuit by loss-of-function perturbations using targeted ablations and gene knockouts. We propose that this simple circuit is the core implementation of mechanosensory reafferent suppression in these young animals and that it might form the first instantiation of state-dependent modulation found at later stages in development.

Thermoregulation in fish.

Thermoregulation is critical for survival and animals therefore employ strategies to keep their body temperature within a physiological range. As ectotherms, fish exclusively rely on behavioral strategies for thermoregulation. Different species of fish seek out their specific optimal temperatures through thermal navigation by biasing behavioral output based on experienced environmental temperatures. Like other vertebrates, fish sense water temperature using thermoreceptors in trigeminal and dorsal root ganglia neurons that innervate the skin. Recent research in larval zebrafish has revealed how neural circuits subsequently transform this sensation of temperature into thermoregulatory behaviors. Across fish species, thermoregulatory strategies rely on a modulation of swim vigor based on current temperature and a modulation of turning based on temperature change. Interestingly, temperature preferences are not fixed but depend on other environmental cues and internal states. The following review is intended as an overview on the current knowledge as well as open questions in fish thermoregulation.

Elements of a stochastic 3D prediction engine in larval zebrafish prey capture.

The computational principles underlying predictive capabilities in animals are poorly understood. Here, we wondered whether predictive models mediating prey capture could be reduced to a simple set of sensorimotor rules performed by a primitive organism. For this task, we chose the larval zebrafish, a tractable vertebrate that pursues and captures swimming microbes. Using a novel naturalistic 3D setup, we show that the zebrafish combines position and velocity perception to construct a future positional estimate of its prey, indicating an ability to project trajectories forward in time. Importantly, the stochasticity in the fish's sensorimotor transformations provides a considerable advantage over equivalent noise-free strategies. This surprising result coalesces with recent findings that illustrate the benefits of biological stochasticity to adaptive behavior. In sum, our study reveals that zebrafish are equipped with a recursive prey capture algorithm, built up from simple stochastic rules, that embodies an implicit predictive model of the world.

Convergent Temperature Representations in Artificial and Biological Neural Networks.

Discoveries in biological neural networks (BNNs) shaped artificial neural networks (ANNs) and computational parallels between ANNs and BNNs have recently been discovered. However, it is unclear to what extent discoveries in ANNs can give insight into BNN function. Here, we designed and trained an ANN to perform heat gradient navigation and found striking similarities in computation and heat representation to a known zebrafish BNN. This included shared ON- and OFF-type representations of absolute temperature and rates of change. Importantly, ANN function critically relied on zebrafish-like units. We furthermore used the accessibility of the ANN to discover a new temperature-responsive cell type in the zebrafish cerebellum. Finally, constraining the ANN by the C. elegans motor repertoire retuned sensory representations indicating that our approach generalizes. Together, these results emphasize convergence of ANNs and BNNs on stereotypical representations and that ANNs form a powerful tool to understand their biological counterparts.

Distributed Plasticity Drives Visual Habituation Learning in Larval Zebrafish.

Habituation is a simple form of learning where animals learn to reduce their responses to repeated innocuous stimuli [1]. Habituation is thought to occur via at least two temporally and molecularly distinct mechanisms, which lead to short-term memories that last for seconds to minutes and long-term memories that last for hours or longer [1, 2]. Here, we focus on long-term habituation, which, due to the extended time course, necessitates stable alterations to circuit properties [2-4]. In its simplest form, long-term habituation could result from a plasticity event at a single point in a circuit, and many studies have focused on identifying the site and underlying mechanism of plasticity [5-10]. However, it is possible that these individual sites are only one of many points in the circuit where plasticity is occurring. Indeed, studies of short-term habituation in C. elegans indicate that in this paradigm, multiple genetically separable mechanisms operate to adapt specific aspects of behavior [11-13]. Here, we use a visual assay in which larval zebrafish habituate their response to sudden reductions in illumination (dark flashes) [14, 15]. Through behavioral analyses, we find that multiple components of the dark-flash response habituate independently of one another using different molecular mechanisms. This is consistent with a modular model in which habituation originates from multiple independent processes, each adapting specific components of behavior. This may allow animals to more specifically or flexibly habituate based on stimulus context or internal states.

A Brain-wide Circuit Model of Heat-Evoked Swimming Behavior in Larval Zebrafish.

Thermosensation provides crucial information, but how temperature representation is transformed from sensation to behavior is poorly understood. Here, we report a preparation that allows control of heat delivery to zebrafish larvae while monitoring motor output and imaging whole-brain calcium signals, thereby uncovering algorithmic and computational rules that couple dynamics of heat modulation, neural activity and swimming behavior. This approach identifies a critical step in the transformation of temperature representation between the sensory trigeminal ganglia and the hindbrain: A simple sustained trigeminal stimulus representation is transformed into a representation of absolute temperature as well as temperature changes in the hindbrain that explains the observed motor output. An activity constrained dynamic circuit model captures the most prominent aspects of these sensori-motor transformations and predicts both behavior and neural activity in response to novel heat stimuli. These findings provide the first algorithmic description of heat processing from sensory input to behavioral output.

The structure and timescales of heat perception in larval zebrafish.

Avoiding temperatures outside the physiological range is critical for animal survival, but how temperature dynamics are transformed into behavioral output is largely not understood. Here, we used an infrared laser to challenge freely swimming larval zebrafish with "white-noise" heat stimuli and built quantitative models relating external sensory information and internal state to behavioral output. These models revealed that larval zebrafish integrate temperature information over a time-window of 400 ms preceding a swimbout and that swimming is suppressed right after the end of a bout. Our results suggest that larval zebrafish compute both an integral and a derivative across heat in time to guide their next movement. Our models put important constraints on the type of computations that occur in the nervous system and reveal principles of how somatosensory temperature information is processed to guide behavioral decisions such as sensitivity to both absolute levels and changes in stimulation.

A convergent and essential interneuron pathway for Mauthner-cell-mediated escapes.

The Mauthner cell (M-cell) is a command-like neuron in teleost fish whose firing in response to aversive stimuli is correlated with short-latency escapes [1-3]. M-cells have been proposed as evolutionary ancestors of startle response neurons of the mammalian reticular formation [4], and studies of this circuit have uncovered important principles in neurobiology that generalize to more complex vertebrate models [3]. The main excitatory input was thought to originate from multisensory afferents synapsing directly onto the M-cell dendrites [3]. Here, we describe an additional, convergent pathway that is essential for the M-cell-mediated startle behavior in larval zebrafish. It is composed of excitatory interneurons called spiral fiber neurons, which project to the M-cell axon hillock. By in vivo calcium imaging, we found that spiral fiber neurons are active in response to aversive stimuli capable of eliciting escapes. Like M-cell ablations, bilateral ablations of spiral fiber neurons largely eliminate short-latency escapes. Unilateral spiral fiber neuron ablations shift the directionality of escapes and indicate that spiral fiber neurons excite the M-cell in a lateralized manner. Their optogenetic activation increases the probability of short-latency escapes, supporting the notion that spiral fiber neurons help activate M-cell-mediated startle behavior. These results reveal that spiral fiber neurons are essential for the function of the M-cell in response to sensory cues and suggest that convergent excitatory inputs that differ in their input location and timing ensure reliable activation of the M-cell, a feedforward excitatory motif that may extend to other neural circuits.

Whole-field visual motion drives swimming in larval zebrafish via a stochastic process.

Caudo-rostral whole-field visual motion elicits forward locomotion in many organisms, including larval zebrafish. Here, we investigate the dependence on the latency to initiate this forward swimming as a function of the speed of the visual motion. We show that latency is highly dependent on speed for slow speeds (<10 mm s(-1)) and then plateaus for higher values. Typical latencies are >1.5 s, which is much longer than neuronal transduction processes. What mechanisms underlie these long latencies? We propose two alternative, biologically inspired models that could account for this latency to initiate swimming: an integrate and fire model, which is history dependent, and a stochastic Poisson model, which has no history dependence. We use these models to predict the behavior of larvae when presented with whole-field motion of varying speed and find that the stochastic process shows better agreement with the experimental data. Finally, we discuss possible neuronal implementations of these models.

The study of psychiatric disease genes and drugs in zebrafish.

Mutations associated with psychiatric disease are being identified, but it remains unclear how the affected genes contribute to disease. Zebrafish is an emerging model to study psychiatric disease genes with a rich repertoire of phenotyping tools. Recent zebrafish research has uncovered potential developmental phenotypes for genes associated with psychiatric disorders, while drug screens have behaviorally characterized small molecules and identified new classes of drugs. Behavioral studies have led to promising models for endophenotypes of psychiatric diseases. While further research is needed to firmly link these models to psychiatric disorders, they are valuable tools for phenotyping genetic mutations and drugs. Recently developed tools in genome editing and in vivo imaging promise additional insights into the processes disrupted by mutations in psychiatric disease genes.

RhoL controls invasion and Rap1 localization during immune cell transmigration in Drosophila.

Human immune cells have to penetrate an endothelial barrier during their beneficial pursuit of infection and their destructive infiltration of tissues in autoimmune diseases. This transmigration requires Rap1 GTPase to activate integrin affinity. We define a new model system for this process by demonstrating, with live imaging and genetics, that during embryonic development Drosophila melanogaster immune cells penetrate an epithelial, Drosophila E-cadherin (DE-cadherin)-based tissue barrier. A mutant in RhoL, a GTPase homologue that is specifically expressed in haemocytes, blocks this invasive step but not other aspects of guided migration. RhoL mediates integrin adhesion caused by Drosophila Rap1 overexpression and moves Rap1 away from a concentration in the cytoplasm to the leading edge during invasive migration. These findings indicate that a programmed migratory step during Drosophila development bears striking molecular similarities to vertebrate immune cell transmigration during inflammation, and identify RhoL as a new regulator of invasion, adhesion and Rap1 localization. Our work establishes the utility of Drosophila for identifying novel components of immune cell transmigration and for understanding the in vivo interplay of immune cells with the barriers they penetrate.