A cortico-collicular circuit for orienting to shelter during escape.

When faced with predatory threats, escape towards shelter is an adaptive action that offers long-term protection against the attacker. Animals rely on knowledge of safe locations in the environment to instinctively execute rapid shelter-directed escape actions1,2. Although previous work has identified neural mechanisms of escape initiation3,4, it is not known how the escape circuit incorporates spatial information to execute rapid flights along the most efficient route to shelter. Here we show that the mouse retrosplenial cortex (RSP) and superior colliculus (SC) form a circuit that encodes the shelter-direction vector and is specifically required for accurately orienting to shelter during escape. Shelter direction is encoded in RSP and SC neurons in egocentric coordinates and SC shelter-direction tuning depends on RSP activity. Inactivation of the RSP-SC pathway disrupts the orientation to shelter and causes escapes away from the optimal shelter-directed route, but does not lead to generic deficits in orientation or spatial navigation. We find that the RSP and SC are monosynaptically connected and form a feedforward lateral inhibition microcircuit that strongly drives the inhibitory collicular network because of higher RSP input convergence and synaptic integration efficiency in inhibitory SC neurons. This results in broad shelter-direction tuning in inhibitory SC neurons and sharply tuned excitatory SC neurons. These findings are recapitulated by a biologically constrained spiking network model in which RSP input to the local SC recurrent ring architecture generates a circular shelter-direction map. We propose that this RSP-SC circuit might be specialized for generating collicular representations of memorized spatial goals that are readily accessible to the motor system during escape, or more broadly, during navigation when the goal must be reached as fast as possible.

Differential Geometry Methods for Constructing Manifold-Targeted Recurrent Neural Networks.

Neural computations can be framed as dynamical processes, whereby the structure of the dynamics within a neural network is a direct reflection of the computations that the network performs. A key step in generating mechanistic interpretations within this computation through dynamics framework is to establish the link among network connectivity, dynamics, and computation. This link is only partly understood. Recent work has focused on producing algorithms for engineering artificial recurrent neural networks (RNN) with dynamics targeted to a specific goal manifold. Some of these algorithms require only a set of vectors tangent to the target manifold to be computed and thus provide a general method that can be applied to a diverse set of problems. Nevertheless, computing such vectors for an arbitrary manifold in a high-dimensional state space remains highly challenging, which in practice limits the applicability of this approach. Here we demonstrate how topology and differential geometry can be leveraged to simplify this task by first computing tangent vectors on a low-dimensional topological manifold and then embedding these in state space. The simplicity of this procedure greatly facilitates the creation of manifold-targeted RNNs, as well as the process of designing task-solving, on-manifold dynamics. This new method should enable the application of network engineering-based approaches to a wide set of problems in neuroscience and machine learning. Our description of how fundamental concepts from differential geometry can be mapped onto different aspects of neural dynamics is a further demonstration of how the language of differential geometry can enrich the conceptual framework for describing neural dynamics and computation.

Tonically active GABAergic neurons in the dorsal periaqueductal gray control instinctive escape in mice.

Escape behavior is a set of locomotor actions that move an animal away from threat. While these actions can be stereotyped, it is advantageous for survival that they are flexible.1,2,3 For example, escape probability depends on predation risk and competing motivations,4,5,6,7,8,9,10,11 and flight to safety requires continuous adjustments of trajectory and must terminate at the appropriate place and time.12,13,14,15,16 This degree of flexibility suggests that modulatory components, like inhibitory networks, act on the neural circuits controlling instinctive escape.17,18,19,20,21,22 In mice, the decision to escape from imminent threats is implemented by a feedforward circuit in the midbrain, where excitatory vesicular glutamate transporter 2-positive (VGluT2+) neurons in the dorsal periaqueductal gray (dPAG) compute escape initiation and escape vigor.23,24,25 Here we tested the hypothesis that local GABAergic neurons within the dPAG control escape behavior by setting the excitability of the dPAG escape network. Using in vitro patch-clamp and in vivo neural activity recordings, we found that vesicular GABA transporter-positive (VGAT+) dPAG neurons fire action potentials tonically in the absence of synaptic inputs and are a major source of inhibition to VGluT2+ dPAG neurons. Activity in VGAT+ dPAG cells transiently decreases at escape onset and increases during escape, peaking at escape termination. Optogenetically increasing or decreasing VGAT+ dPAG activity changes the probability of escape when the stimulation is delivered at threat onset and the duration of escape when delivered after escape initiation. We conclude that the activity of tonically firing VGAT+ dPAG neurons sets a threshold for escape initiation and controls the execution of the flight action.

Innate heuristics and fast learning support escape route selection in mice.

When faced with imminent danger, animals must rapidly take defensive actions to reach safety. Mice can react to threatening stimuli in ∼250 milliseconds1 and, in simple environments, use spatial memory to quickly escape to shelter.2,3 Natural habitats, however, often offer multiple routes to safety that animals must identify and choose from.4 This is challenging because although rodents can learn to navigate complex mazes,5,6 learning the value of different routes through trial and error during escape could be deadly. Here, we investigated how mice learn to choose between different escape routes. Using environments with paths to shelter of varying length and geometry, we find that mice prefer options that minimize path distance and angle relative to the shelter. This strategy is already present during the first threat encounter and after only ∼10 minutes of exploration in a novel environment, indicating that route selection does not require experience of escaping. Instead, an innate heuristic assigns survival value to each path after rapidly learning the spatial environment. This route selection process is flexible and allows quick adaptation to arenas with dynamic geometries. Computational modeling shows that model-based reinforcement learning agents replicate the observed behavior in environments where the shelter location is rewarding during exploration. These results show that mice combine fast spatial learning with innate heuristics to choose escape routes with the highest survival value. The results further suggest that integrating prior knowledge acquired through evolution with knowledge learned from experience supports adaptation to changing environments and minimizes the need for trial and error when the errors are costly.

Accurate determination of marker location within whole-brain microscopy images.

High-resolution whole-brain microscopy provides a means for post hoc determination of the location of implanted devices and labelled cell populations that are necessary to interpret in vivo experiments designed to understand brain function. Here we have developed two plugins (brainreg and brainreg-segment) for the Python-based image viewer napari, to accurately map any object in a common coordinate space. We analysed the position of dye-labelled electrode tracks and two-photon imaged cell populations expressing fluorescent proteins. The precise location of probes and cells were physiologically interrogated and revealed accurate segmentation with near-cellular resolution.

Visualizing anatomically registered data with brainrender.

Three-dimensional (3D) digital brain atlases and high-throughput brain-wide imaging techniques generate large multidimensional datasets that can be registered to a common reference frame. Generating insights from such datasets depends critically on visualization and interactive data exploration, but this a challenging task. Currently available software is dedicated to single atlases, model species or data types, and generating 3D renderings that merge anatomically registered data from diverse sources requires extensive development and programming skills. Here, we present brainrender: an open-source Python package for interactive visualization of multidimensional datasets registered to brain atlases. Brainrender facilitates the creation of complex renderings with different data types in the same visualization and enables seamless use of different atlas sources. High-quality visualizations can be used interactively and exported as high-resolution figures and animated videos. By facilitating the visualization of anatomically registered data, brainrender should accelerate the analysis, interpretation, and dissemination of brain-wide multidimensional data.

A cerebellar-thalamocortical pathway drives behavioral context-dependent movement initiation.

Executing learned motor behaviors often requires the transformation of sensory cues into patterns of motor commands that generate appropriately timed actions. The cerebellum and thalamus are two key areas involved in shaping cortical output and movement, but the contribution of a cerebellar-thalamocortical pathway to voluntary movement initiation remains poorly understood. Here, we investigated how an auditory "go cue" transforms thalamocortical activity patterns and how these changes relate to movement initiation. Population responses in dentate/interpositus-recipient regions of motor thalamus reflect a time-locked increase in activity immediately prior to movement initiation that is temporally uncoupled from the go cue, indicative of a fixed-latency feedforward motor timing signal. Blocking cerebellar or motor thalamic output suppresses movement initiation, while stimulation triggers movements in a behavioral context-dependent manner. Our findings show how cerebellar output, via the thalamus, shapes cortical activity patterns necessary for learned context-dependent movement initiation.

CXCL4 and CXCL4L1 Differentially Affect Monocyte Survival and Dendritic Cell Differentiation and Phagocytosis.

Upon inflammation, circulating monocytes leave the bloodstream and migrate into the tissues, where they differentiate after exposure to various growth factors, cytokines or infectious agents. The best defined macrophage polarization types are M1 and M2. However, the platelet-derived CXC chemokine CXCL4 induces the polarization of macrophages into a unique phenotype. In this study, we compared the effect of CXCL4 and its variant CXCL4L1 on the differentiation of monocytes into macrophages and into immature monocyte-derived dendritic cells (iMDDC). Differently to M-CSF and CXCL4, CXCL4L1 is not a survival factor for monocytes. Moreover, the expression of the chemokine receptors CCR2, CCR5 and CXCR3 was significantly higher on CXCL4L1-treated monocytes compared to M-CSF- and CXCL4-stimulated monocytes. IL-1 receptor antagonist (IL-1RN) expression was upregulated by CXCL4 and downregulated by CXCL4L1, respectively, whereas both chemokines reduced the expression of the mannose receptor (MRC). Furthermore, through activation of CXCR3, CXCL4L1-stimulated monocytes released significantly higher amounts of CCL2 and CXCL8 compared to CXCL4-treated monocytes, indicating more pronounced inflammatory traits for CXCL4L1. In contrast, in CXCL4L1-treated monocytes, the production of CCL22 was lower. Compared to iMDDC generated in the presence of CXCL4L1, CXCL4-treated iMDDC showed an enhanced phagocytic capacity and downregulation of expression of certain surface markers (e.g. CD1a) and specific enzymes (e.g. MMP-9 and MMP-12). CXCL4 and CXCL4L1 did not affect the chemokine receptor expression on iMDDC and cytokine production (CCL2, CCL18, CCL22, CXCL8, IL-10) by CXCL4- or CXCL4L1-differentiated iMDDC was similar. We can conclude that both CXCL4 and CXCL4L1 exert a direct effect on monocytes and iMDDC. However, the resulting phenotypes are different, which suggests a unique role for the two CXCL4 variants in physiology and/or pathology.