A prominent trend in single-cell transcriptomics is providing spatial context alongside a characterization of each cell's molecular state. This typically requires targeting an a priori selection of genes, often covering less than 1% of the genome, and a key question is how to optimally determine the small gene panel. We address this challenge by introducing a flexible deep learning framework, PERSIST, to identify informative gene targets for spatial transcriptomics studies by leveraging reference scRNA-seq data. Using datasets spanning different brain regions, species, and scRNA-seq technologies, we show that PERSIST reliably identifies panels that provide more accurate prediction of the genome-wide expression profile, thereby capturing more information with fewer genes. PERSIST can be adapted to specific biological goals, and we demonstrate that PERSIST's binarization of gene expression levels enables models trained on scRNA-seq data to generalize with to spatial transcriptomics data, despite the complex shift between these technologies.

Information processing in the neocortex is performed by GABAergic interneurons that are integrated with excitatory neurons into precisely structured circuits. To reveal how each neuron type shapes sensory representations, we measured spikes and membrane potential of specific types of neurons in the barrel cortex while mice performed an active, whisker-dependent object localization task. Whiskers were tracked with millisecond precision. Fast-spiking (FS) neurons were activated by touch with short latency and by whisking. FS neurons track thalamic input and provide feedforward inhibition. Somatostatin (SOM)-expressing neurons were also excited by touch, but with a delay (5 ms) compared to excitatory (E) and FS neurons. SOM neurons monitor local excitation and provide feedback inhibition. Vasoactive intestinal polypeptide (VIP)-expressing neurons were not driven by touch but elevated their spike rate during whisking, disinhibiting E and FS neurons. Our data reveal rules of recruitment for specific interneuron types, providing foundations for understanding cortical computations.

The mouse is an increasingly prominent model for the analysis of mammalian neuronal circuits. Neural circuits ultimately have to be probed during behaviors that engage the circuits. Linking circuit dynamics to behavior requires precise control of sensory stimuli and measurement of body movements. Head-fixation has been used for behavioral research, particularly in non-human primates, to facilitate precise stimulus control, behavioral monitoring and neural recording. However, choice-based, perceptual decision tasks by head-fixed mice have only recently been introduced. Training mice relies on motivating mice using water restriction. Here we describe procedures for head-fixation, water restriction and behavioral training for head-fixed mice, with a focus on active, whisker-based tactile behaviors. In these experiments mice had restricted access to water (typically 1 ml/day). After ten days of water restriction, body weight stabilized at approximately 80% of initial weight. At that point mice were trained to discriminate sensory stimuli using operant conditioning. Head-fixed mice reported stimuli by licking in go/no-go tasks and also using a forced choice paradigm using a dual lickport. In some cases mice learned to discriminate sensory stimuli in a few trials within the first behavioral session. Delay epochs lasting a second or more were used to separate sensation (e.g. tactile exploration) and action (i.e. licking). Mice performed a variety of perceptual decision tasks with high performance for hundreds of trials per behavioral session. Up to four months of continuous water restriction showed no adverse health effects. Behavioral performance correlated with the degree of water restriction, supporting the importance of controlling access to water. These behavioral paradigms can be combined with cellular resolution imaging, random access photostimulation, and whole cell recordings.

Neurons relevant to a particular behavior are often widely dispersed across the brain. To record activity in groups of individual neurons that might be distributed across large distances, neuroscientists and optical engineers have been developing a new type of microscope called a mesoscope. Mesoscopes have high spatial resolution and a large field of view. This Q&A will discuss this exciting new technology, highlighting a particular instrument, the two-photon random access mesoscope (2pRAM).

Spine growth and retraction with synapse formation and elimination plays an important role in shaping brain circuits during development and in the adult brain, yet the temporal relationship between spine morphogenesis and the formation of functional synapses remains poorly defined. We imaged hippocampal pyramidal neurons to identify spines of different ages. We then used two-photon glutamate uncaging, whole-cell recording, and Ca(2+) imaging to analyze the properties of nascent spines and their older neighbors. New spines expressed glutamate-sensitive currents that were indistinguishable from mature spines of comparable volumes. Some spines exhibited negligible AMPA receptor-mediated responses, but the occurrence of these "silent" spines was uncorrelated with spine age. In contrast, NMDA receptor-mediated Ca(2+) accumulations were significantly lower in new spines. New spines reconstructed using electron microscopy made synapses. Our data support a model in which outgrowth and enlargement of nascent spines is tightly coupled to formation and maturation of glutamatergic synapses.

Imaging neurons and neural circuits over large volumes at high speed and subcellular resolution is a difficult task. Incorporating a Bessel focus module into a two-photon fluorescence mesoscope, we achieved rapid volumetric imaging of neural activity over the mesoscale with synaptic resolution. We applied the technology to calcium imaging of entire dendritic spans of neurons as well as neural ensembles within multiple cortical regions over two hemispheres of the awake mouse brain.

Neuronal cell types are the nodes of neural circuits that determine the flow of information within the brain. Neuronal morphology, especially the shape of the axonal arbor, provides an essential descriptor of cell type and reveals how individual neurons route their output across the brain. Despite the importance of morphology, few projection neurons in the mouse brain have been reconstructed in their entirety. Here we present a robust and efficient platform for imaging and reconstructing complete neuronal morphologies, including axonal arbors that span substantial portions of the brain. We used this platform to reconstruct more than 1,000 projection neurons in the motor cortex, thalamus, subiculum, and hypothalamus. Together, the reconstructed neurons constitute more than 85 meters of axonal length and are available in a searchable online database. Axonal shapes revealed previously unknown subtypes of projection neurons and suggest organizational principles of long-range connectivity.

Neural computation involves diverse types of GABAergic inhibitory interneurons that are integrated with excitatory (E) neurons into precisely structured circuits. To understand how each neuron type shapes sensory representations, we measured firing patterns of defined types of neurons in the barrel cortex while mice performed an active, whisker-dependent object localization task. Touch excited fast-spiking (FS) interneurons at short latency, followed by activation of E neurons and somatostatin-expressing (SST) interneurons. Touch only weakly modulated vasoactive intestinal polypeptide-expressing (VIP) interneurons. Voluntary whisker movement activated FS neurons in the ventral posteromedial nucleus (VPM) target layers, a subset of SST neurons and a majority of VIP neurons. Together, FS neurons track thalamic input, mediating feedforward inhibition. SST neurons monitor local excitation, providing feedback inhibition. VIP neurons are activated by non-sensory inputs, disinhibiting E and FS neurons. Our data reveal rules of recruitment for interneuron types during behavior, providing foundations for understanding computation in cortical microcircuits.

Most cortical synapses are local and excitatory. Local recurrent circuits could implement amplification, allowing pattern completion and other computations. Cortical circuits contain subnetworks that consist of neurons with similar receptive fields and increased connectivity relative to the network average. Cortical neurons that encode different types of information are spatially intermingled and distributed over large brain volumes, and this complexity has hindered attempts to probe the function of these subnetworks by perturbing them individually. Here we use computational modelling, optical recordings and manipulations to probe the function of recurrent coupling in layer 2/3 of the mouse vibrissal somatosensory cortex during active tactile discrimination. A neural circuit model of layer 2/3 revealed that recurrent excitation enhances sensory signals by amplification, but only for subnetworks with increased connectivity. Model networks with high amplification were sensitive to damage: loss of a few members of the subnetwork degraded stimulus encoding. We tested this prediction by mapping neuronal selectivity and photoablating neurons with specific selectivity. Ablation of a small proportion of layer 2/3 neurons (10-20, less than 5% of the total) representing touch markedly reduced responses in the spared touch representation, but not in other representations. Ablations most strongly affected neurons with stimulus responses that were similar to those of the ablated population, which is also consistent with network models. Recurrence among cortical neurons with similar selectivity therefore drives input-specific amplification during behaviour.

Most functional plasticity studies in the cortex have focused on layers (L) II/III and IV, whereas relatively little is known of LV. Structural measurements of dendritic spines in vivo suggest some specialization among LV cell subtypes. We therefore studied experience-dependent plasticity in the barrel cortex using intracellular recordings to distinguish regular spiking (RS) and intrinsic bursting (IB) subtypes. Postsynaptic potentials and suprathreshold responses in vivo revealed a remarkable dichotomy in RS and IB cell plasticity; spared whisker potentiation occurred in IB but not RS cells while deprived whisker depression occurred in RS but not IB cells. Similar RS/IB differences were found in the LII/III to V connections in brain slices. Modeling studies showed that subthreshold changes predicted the suprathreshold changes. These studies demonstrate the major functional partition of plasticity within a single cortical layer and reveal the LII/III to LV connection as a major excitatory locus of cortical plasticity.