Self-Supervised Keypoint Discovery in Behavioral Videos.

We propose a method for learning the posture and structure of agents from unlabelled behavioral videos. Starting from the observation that behaving agents are generally the main sources of movement in behavioral videos, our method, Behavioral Keypoint Discovery (B-KinD), uses an encoder-decoder architecture with a geometric bottleneck to reconstruct the spatiotemporal difference between video frames. By focusing only on regions of movement, our approach works directly on input videos without requiring manual annotations. Experiments on a variety of agent types (mouse, fly, human, jellyfish, and trees) demonstrate the generality of our approach and reveal that our discovered keypoints represent semantically meaningful body parts, which achieve state-of-the-art performance on keypoint regression among self-supervised methods. Additionally, B-KinD achieve comparable performance to supervised keypoints on downstream tasks, such as behavior classification, suggesting that our method can dramatically reduce model training costs vis-a-vis supervised methods.

A genetically tractable jellyfish model for systems and evolutionary neuroscience.

Jellyfish are radially symmetric organisms without a brain that arose more than 500 million years ago. They achieve organismal behaviors through coordinated interactions between autonomously functioning body parts. Jellyfish neurons have been studied electrophysiologically, but not at the systems level. We introduce Clytia hemisphaerica as a transparent, genetically tractable jellyfish model for systems and evolutionary neuroscience. We generate stable F1 transgenic lines for cell-type-specific conditional ablation and whole-organism GCaMP imaging. Using these tools and computational analyses, we find that an apparently diffuse network of RFamide-expressing umbrellar neurons is functionally subdivided into a series of spatially localized subassemblies whose synchronous activation controls directional food transfer from the tentacles to the mouth. These data reveal an unanticipated degree of structured neural organization in this species. Clytia affords a platform for systems-level studies of neural function, behavior, and evolution within a clade of marine organisms with growing ecological and economic importance.

Whole-animal multiplexed single-cell RNA-seq reveals transcriptional shifts across Clytia medusa cell types.

We present an organism-wide, transcriptomic cell atlas of the hydrozoan medusa Clytia hemisphaerica and describe how its component cell types respond to perturbation. Using multiplexed single-cell RNA sequencing, in which individual animals were indexed and pooled from control and perturbation conditions into a single sequencing run, we avoid artifacts from batch effects and are able to discern shifts in cell state in response to organismal perturbations. This work serves as a foundation for future studies of development, function, and regeneration in a genetically tractable jellyfish species. Moreover, we introduce a powerful workflow for high-resolution, whole-animal, multiplexed single-cell genomics that is readily adaptable to other traditional or nontraditional model organisms.

An improved whole life cycle culture protocol for the hydrozoan genetic model Clytia hemisphaerica.

The jellyfish species Clytia hemisphaerica (Cnidaria, Hydrozoa) has emerged as a new experimental model animal in the last decade. Favorable characteristics include a fully transparent body suitable for microscopy, daily gamete production and a relatively short life cycle. Furthermore, whole genome sequence assembly and efficient gene editing techniques using CRISPR/Cas9 have opened new possibilities for genetic studies. The quasi-immortal vegetatively-growing polyp colony stage provides a practical means to maintain mutant strains. In the context of developing Clytia as a genetic model, we report here an improved whole life cycle culture method including an aquarium tank system designed for culture of the tiny jellyfish form. We have compared different feeding regimes using Artemia larvae as food and demonstrate that the stage-dependent feeding control is the key for rapid and reliable medusa and polyp rearing. Metamorphosis of the planula larvae into a polyp colony can be induced efficiently using a new synthetic peptide. The optimized procedures detailed here make it practical to generate genetically modified Clytia strains and to maintain their whole life cycle in the laboratory.This article has an associated First Person interview with the two first authors of the paper.

Connectional architecture of a mouse hypothalamic circuit node controlling social behavior.

Type 1 estrogen receptor-expressing neurons in the ventrolateral subdivision of the ventromedial hypothalamus (VMHvlEsr1) play a causal role in the control of social behaviors, including aggression. Here we use six different viral-genetic tracing methods to systematically map the connectional architecture of VMHvlEsr1 neurons. These data reveal a high level of input convergence and output divergence ("fan-in/fan-out") from and to over 30 distinct brain regions, with a high degree (∼90%) of bidirectionality, including both direct as well as indirect feedback. Unbiased collateralization mapping experiments indicate that VMHvlEsr1 neurons project to multiple targets. However, we identify two anatomically distinct subpopulations with anterior vs. posterior biases in their collateralization targets. Nevertheless, these two subpopulations receive indistinguishable inputs. These studies suggest an overall system architecture in which an anatomically feed-forward sensory-to-motor processing stream is integrated with a dense, highly recurrent central processing circuit. This architecture differs from the "brain-inspired," hierarchical feed-forward circuits used in certain types of artificial intelligence networks.

Anatomically Defined and Functionally Distinct Dorsal Raphe Serotonin Sub-systems.

The dorsal raphe (DR) constitutes a major serotonergic input to the forebrain and modulates diverse functions and brain states, including mood, anxiety, and sensory and motor functions. Most functional studies to date have treated DR serotonin neurons as a single population. Using viral-genetic methods, we found that subcortical- and cortical-projecting serotonin neurons have distinct cell-body distributions within the DR and differentially co-express a vesicular glutamate transporter. Further, amygdala- and frontal-cortex-projecting DR serotonin neurons have largely complementary whole-brain collateralization patterns, receive biased inputs from presynaptic partners, and exhibit opposite responses to aversive stimuli. Gain- and loss-of-function experiments suggest that amygdala-projecting DR serotonin neurons promote anxiety-like behavior, whereas frontal-cortex-projecting neurons promote active coping in the face of challenge. These results provide compelling evidence that the DR serotonin system contains parallel sub-systems that differ in input and output connectivity, physiological response properties, and behavioral functions.

Basal forebrain circuit for sleep-wake control.

The mammalian basal forebrain (BF) has important roles in controlling sleep and wakefulness, but the underlying neural circuit remains poorly understood. We examined the BF circuit by recording and optogenetically perturbing the activity of four genetically defined cell types across sleep-wake cycles and by comprehensively mapping their synaptic connections. Recordings from channelrhodopsin-2 (ChR2)-tagged neurons revealed that three BF cell types, cholinergic, glutamatergic and parvalbumin-positive (PV+) GABAergic neurons, were more active during wakefulness and rapid eye movement (REM) sleep (wake/REM active) than during non-REM (NREM) sleep, and activation of each cell type rapidly induced wakefulness. By contrast, activation of somatostatin-positive (SOM+) GABAergic neurons promoted NREM sleep, although only some of them were NREM active. Synaptically, the wake-promoting neurons were organized hierarchically by glutamatergic→cholinergic→PV+ neuron excitatory connections, and they all received inhibition from SOM+ neurons. Together, these findings reveal the basic organization of the BF circuit for sleep-wake control.

Viral-genetic tracing of the input-output organization of a central noradrenaline circuit.

Deciphering how neural circuits are anatomically organized with regard to input and output is instrumental in understanding how the brain processes information. For example, locus coeruleus noradrenaline (also known as norepinephrine) (LC-NE) neurons receive input from and send output to broad regions of the brain and spinal cord, and regulate diverse functions including arousal, attention, mood and sensory gating. However, it is unclear how LC-NE neurons divide up their brain-wide projection patterns and whether different LC-NE neurons receive differential input. Here we developed a set of viral-genetic tools to quantitatively analyse the input-output relationship of neural circuits, and applied these tools to dissect the LC-NE circuit in mice. Rabies-virus-based input mapping indicated that LC-NE neurons receive convergent synaptic input from many regions previously identified as sending axons to the locus coeruleus, as well as from newly identified presynaptic partners, including cerebellar Purkinje cells. The 'tracing the relationship between input and output' method (or TRIO method) enables trans-synaptic input tracing from specific subsets of neurons based on their projection and cell type. We found that LC-NE neurons projecting to diverse output regions receive mostly similar input. Projection-based viral labelling revealed that LC-NE neurons projecting to one output region also project to all brain regions we examined. Thus, the LC-NE circuit overall integrates information from, and broadcasts to, many brain regions, consistent with its primary role in regulating brain states. At the same time, we uncovered several levels of specificity in certain LC-NE sub-circuits. These tools for mapping output architecture and input-output relationship are applicable to other neuronal circuits and organisms. More broadly, our viral-genetic approaches provide an efficient intersectional means to target neuronal populations based on cell type and projection pattern.

Presynaptic partners of dorsal raphe serotonergic and GABAergic neurons.

The serotonin system powerfully modulates physiology and behavior in health and disease, yet the circuit mechanisms underlying serotonin neuron activity are poorly understood. The major source of forebrain serotonergic innervation is from the dorsal raphe nucleus (DR), which contains both serotonin and GABA neurons. Using viral tracing combined with electrophysiology, we found that GABA and serotonin neurons in the DR receive excitatory, inhibitory, and peptidergic inputs from the same specific brain regions. Embedded in this overall similarity are important differences. Serotonin neurons are more likely to receive synaptic inputs from anterior neocortex while GABA neurons receive disproportionally higher input from the central amygdala. Local input mapping revealed extensive serotonin-serotonin as well as GABA-serotonin connectivity with a distinct spatial organization. Covariance analysis suggests heterogeneity of both serotonin and GABA neurons with respect to the inputs they receive. These analyses provide a foundation for further functional dissection of the serotonin system.

Dissecting local circuits: parvalbumin interneurons underlie broad feedback control of olfactory bulb output.

In the mouse olfactory bulb, information from sensory neurons is extensively processed by local interneurons before being transmitted to the olfactory cortex by mitral and tufted (M/T) cells. The precise function of these local networks remains elusive because of the vast heterogeneity of interneurons, their diverse physiological properties, and their complex synaptic connectivity. Here we identified the parvalbumin interneurons (PVNs) as a prominent component of the M/T presynaptic landscape by using an improved rabies-based transsynaptic tracing method for local circuits. In vivo two-photon-targeted patch recording revealed that PVNs have exceptionally broad olfactory receptive fields and exhibit largely excitatory and persistent odor responses. Transsynaptic tracing indicated that PVNs receive direct input from widely distributed M/T cells. Both the anatomical and functional extent of this M/T→PVN→M/T circuit contrasts with the narrowly confined M/T→granule cell→M/T circuit, suggesting that olfactory information is processed by multiple local circuits operating at distinct spatial scales.

High-resolution analysis of parent-of-origin allelic expression in the mouse brain.

Genomic imprinting results in preferential expression of the paternal or maternal allele of certain genes. We have performed a genome-wide characterization of imprinting in the mouse embryonic and adult brain. This approach uncovered parent-of-origin allelic effects of more than 1300 loci. We identified parental bias in the expression of individual genes and of specific transcript isoforms, with differences between brain regions. Many imprinted genes are expressed in neural systems associated with feeding and motivated behaviors, and parental biases preferentially target genetic pathways governing metabolism and cell adhesion. We observed a preferential maternal contribution to gene expression in the developing brain and a major paternal contribution in the adult brain. Thus, parental expression bias emerges as a major mode of epigenetic regulation in the brain.