A virtual rodent predicts the structure of neural activity across behaviors.

Animals have exquisite control of their bodies, allowing them to perform a diverse range of behaviors. How such control is implemented by the brain, however, remains unclear. Advancing our understanding requires models that can relate principles of control to the structure of neural activity in behaving animals. To facilitate this, we built a 'virtual rodent', in which an artificial neural network actuates a biomechanically realistic model of the rat 1 in a physics simulator 2. We used deep reinforcement learning 3-5 to train the virtual agent to imitate the behavior of freely-moving rats, thus allowing us to compare neural activity recorded in real rats to the network activity of a virtual rodent mimicking their behavior. We found that neural activity in the sensorimotor striatum and motor cortex was better predicted by the virtual rodent's network activity than by any features of the real rat's movements, consistent with both regions implementing inverse dynamics 6. Furthermore, the network's latent variability predicted the structure of neural variability across behaviors and afforded robustness in a way consistent with the minimal intervention principle of optimal feedback control 7. These results demonstrate how physical simulation of biomechanically realistic virtual animals can help interpret the structure of neural activity across behavior and relate it to theoretical principles of motor control.

Leaving flatland: Advances in 3D behavioral measurement.

Animals move in three dimensions (3D). Thus, 3D measurement is necessary to report the true kinematics of animal movement. Existing 3D measurement techniques draw on specialized hardware, such as motion capture or depth cameras, as well as deep multi-view and monocular computer vision. Continued advances at the intersection of deep learning and computer vision will facilitate 3D tracking across more anatomical features, with less training data, in additional species, and within more natural, occlusive environments. 3D behavioral measurement enables unique applications in phenotyping, investigating the neural basis of behavior, and designing artificial agents capable of imitating animal behavior.

Geometric deep learning enables 3D kinematic profiling across species and environments.

Comprehensive descriptions of animal behavior require precise three-dimensional (3D) measurements of whole-body movements. Although two-dimensional approaches can track visible landmarks in restrictive environments, performance drops in freely moving animals, due to occlusions and appearance changes. Therefore, we designed DANNCE to robustly track anatomical landmarks in 3D across species and behaviors. DANNCE uses projective geometry to construct inputs to a convolutional neural network that leverages learned 3D geometric reasoning. We trained and benchmarked DANNCE using a dataset of nearly seven million frames that relates color videos and rodent 3D poses. In rats and mice, DANNCE robustly tracked dozens of landmarks on the head, trunk, and limbs of freely moving animals in naturalistic settings. We extended DANNCE to datasets from rat pups, marmosets, and chickadees, and demonstrate quantitative profiling of behavioral lineage during development.

Continuous Whole-Body 3D Kinematic Recordings across the Rodent Behavioral Repertoire.

In mammalian animal models, high-resolution kinematic tracking is restricted to brief sessions in constrained environments, limiting our ability to probe naturalistic behaviors and their neural underpinnings. To address this, we developed CAPTURE (Continuous Appendicular and Postural Tracking Using Retroreflector Embedding), a behavioral monitoring system that combines motion capture and deep learning to continuously track the 3D kinematics of a rat's head, trunk, and limbs for week-long timescales in freely behaving animals. CAPTURE realizes 10- to 100-fold gains in precision and robustness compared with existing convolutional network approaches to behavioral tracking. We demonstrate CAPTURE's ability to comprehensively profile the kinematics and sequential organization of natural rodent behavior, its variation across individuals, and its perturbation by drugs and disease, including identifying perseverative grooming states in a rat model of fragile X syndrome. CAPTURE significantly expands the range of behaviors and contexts that can be quantitatively investigated, opening the door to a new understanding of natural behavior and its neural basis.

Cross-hemispheric gamma synchrony between prefrontal parvalbumin interneurons supports behavioral adaptation during rule shift learning.

Organisms must learn new strategies to adapt to changing environments. Activity in different neurons often exhibits synchronization that can dynamically enhance their communication and might create flexible brain states that facilitate changes in behavior. We studied the role of gamma-frequency (~40 Hz) synchrony between prefrontal parvalbumin (PV) interneurons in mice learning multiple new cue-reward associations. Voltage indicators revealed cell-type-specific increases of cross-hemispheric gamma synchrony between PV interneurons when mice received feedback that previously learned associations were no longer valid. Disrupting this synchronization by delivering out-of-phase optogenetic stimulation caused mice to perseverate on outdated associations, an effect not reproduced by in-phase stimulation or out-of-phase stimulation at other frequencies. Gamma synchrony was specifically required when new associations used familiar cues that were previously irrelevant to behavioral outcomes, not when associations involved new cues or for reversing previously learned associations. Thus, gamma synchrony is indispensable for reappraising the behavioral salience of external cues.

Diametric neural ensemble dynamics in parkinsonian and dyskinetic states.

Loss of dopamine in Parkinson's disease is hypothesized to impede movement by inducing hypo- and hyperactivity in striatal spiny projection neurons (SPNs) of the direct (dSPNs) and indirect (iSPNs) pathways in the basal ganglia, respectively. The opposite imbalance might underlie hyperkinetic abnormalities, such as dyskinesia caused by treatment of Parkinson's disease with the dopamine precursor L-DOPA. Here we monitored thousands of SPNs in behaving mice, before and after dopamine depletion and during L-DOPA-induced dyskinesia. Normally, intermingled clusters of dSPNs and iSPNs coactivated before movement. Dopamine depletion unbalanced SPN activity rates and disrupted the movement-encoding iSPN clusters. Matching their clinical efficacy, L-DOPA or agonism of the D2 dopamine receptor reversed these abnormalities more effectively than agonism of the D1 dopamine receptor. The opposite pathophysiology arose in L-DOPA-induced dyskinesia, during which iSPNs showed hypoactivity and dSPNs showed unclustered hyperactivity. Therefore, both the spatiotemporal profiles and rates of SPN activity appear crucial to striatal function, and next-generation treatments for basal ganglia disorders should target both facets of striatal activity.

Neural ensemble dynamics underlying a long-term associative memory.

The brain's ability to associate different stimuli is vital for long-term memory, but how neural ensembles encode associative memories is unknown. Here we studied how cell ensembles in the basal and lateral amygdala encode associations between conditioned and unconditioned stimuli (CS and US, respectively). Using a miniature fluorescence microscope, we tracked the Ca2+ dynamics of ensembles of amygdalar neurons during fear learning and extinction over 6 days in behaving mice. Fear conditioning induced both up- and down-regulation of individual cells' CS-evoked responses. This bi-directional plasticity mainly occurred after conditioning, and reshaped the neural ensemble representation of the CS to become more similar to the US representation. During extinction training with repetitive CS presentations, the CS representation became more distinctive without reverting to its original form. Throughout the experiments, the strength of the ensemble-encoded CS-US association predicted the level of behavioural conditioning in each mouse. These findings support a supervised learning model in which activation of the US representation guides the transformation of the CS representation.

Cell-Type-Specific Optical Recording of Membrane Voltage Dynamics in Freely Moving Mice.

Electrophysiological field potential dynamics are of fundamental interest in basic and clinical neuroscience, but how specific cell types shape these dynamics in the live brain is poorly understood. To empower mechanistic studies, we created an optical technique, TEMPO, that records the aggregate trans-membrane voltage dynamics of genetically specified neurons in freely behaving mice. TEMPO has >10-fold greater sensitivity than prior fiber-optic techniques and attains the noise minimum set by quantum mechanical photon shot noise. After validating TEMPO's capacity to track established oscillations in the delta, theta, and gamma frequency bands, we compared the D1- and D2-dopamine-receptor-expressing striatal medium spiny neurons (MSNs), which are interspersed and electrically indistinguishable. Unexpectedly, MSN population dynamics exhibited two distinct coherent states that were commonly indiscernible in electrical recordings and involved synchronized hyperpolarizations across both MSN subtypes. Overall, TEMPO allows the deconstruction of normal and pathologic neurophysiological states into trans-membrane voltage activity patterns of specific cell types.

High-fidelity optical reporting of neuronal electrical activity with an ultrafast fluorescent voltage sensor.

Accurate optical reporting of electrical activity in genetically defined neuronal populations is a long-standing goal in neuroscience. We developed Accelerated Sensor of Action Potentials 1 (ASAP1), a voltage sensor design in which a circularly permuted green fluorescent protein is inserted in an extracellular loop of a voltage-sensing domain, rendering fluorescence responsive to membrane potential. ASAP1 demonstrated on and off kinetics of ∼ 2 ms, reliably detected single action potentials and subthreshold potential changes, and tracked trains of action potential waveforms up to 200 Hz in single trials. With a favorable combination of brightness, dynamic range and speed, ASAP1 enables continuous monitoring of membrane potential in neurons at kilohertz frame rates using standard epifluorescence microscopy.

Optical strategies for sensing neuronal voltage using quantum dots and other semiconductor nanocrystals.

Biophysicists have long sought optical methods capable of reporting the electrophysiological dynamics of large-scale neural networks with millisecond-scale temporal resolution. Existing fluorescent sensors of cell membrane voltage can report action potentials in individual cultured neurons, but limitations in brightness and dynamic range of both synthetic organic and genetically encoded voltage sensors have prevented concurrent monitoring of spiking activity across large populations of individual neurons. Here we propose a novel, inorganic class of fluorescent voltage sensors: semiconductor nanoparticles, such as ultrabright quantum dots (qdots). Our calculations revealed that transmembrane electric fields characteristic of neuronal spiking (~10 mV/nm) modulate a qdot's electronic structure and can induce ~5% changes in its fluorescence intensity and ~1 nm shifts in its emission wavelength, depending on the qdot's size, composition, and dielectric environment. Moreover, tailored qdot sensors composed of two different materials can exhibit substantial (~30%) changes in fluorescence intensity during neuronal spiking. Using signal detection theory, we show that conventional qdots should be capable of reporting voltage dynamics with millisecond precision across several tens or more individual neurons over a range of optical and neurophysiological conditions. These results unveil promising avenues for imaging spiking dynamics in neural networks and merit in-depth experimental investigation.

Improving FRET dynamic range with bright green and red fluorescent proteins.

A variety of genetically encoded reporters use changes in fluorescence (or Förster) resonance energy transfer (FRET) to report on biochemical processes in living cells. The standard genetically encoded FRET pair consists of CFPs and YFPs, but many CFP-YFP reporters suffer from low FRET dynamic range, phototoxicity from the CFP excitation light and complex photokinetic events such as reversible photobleaching and photoconversion. We engineered two fluorescent proteins, Clover and mRuby2, which are the brightest green and red fluorescent proteins to date and have the highest Förster radius of any ratiometric FRET pair yet described. Replacement of CFP and YFP with these two proteins in reporters of kinase activity, small GTPase activity and transmembrane voltage significantly improves photostability, FRET dynamic range and emission ratio changes. These improvements enhance detection of transient biochemical events such as neuronal action-potential firing and RhoA activation in growth cones.