SummaryThe mammalian nervous system executes complex behaviors controlled by specialized, precisely positioned, and interacting cell types. Here, we used RNA sequencing of half a million single cells to create a detailed census of cell types in the mouse nervous system. We mapped cell types spatially and derived a hierarchical, data-driven taxonomy. Neurons were the most diverse and were grouped by developmental anatomical units and by the expression of neurotransmitters and neuropeptides. Neuronal diversity was driven by genes encoding cell identity, synaptic connectivity, neurotransmission, and membrane conductance. We discovered seven distinct, regionally restricted astrocyte types that obeyed developmental boundaries and correlated with the spatial distribution of key glutamate and glycine neurotransmitters. In contrast, oligodendrocytes showed a loss of regional identity followed by a secondary diversification. The resource presented here lays a solid foundation for understanding the molecular architecture of the mammalian nervous system and enables genetic manipulation of specific cell types.

Correlated spiking is often observed in cortical circuits, but its functional role is controversial. It is believed that correlations are a consequence of shared inputs between nearby neurons and could severely constrain information decoding. Here we show theoretically that recurrent neural networks can generate an asynchronous state characterized by arbitrarily low mean spiking correlations despite substantial amounts of shared input. In this state, spontaneous fluctuations in the activity of excitatory and inhibitory populations accurately track each other, generating negative correlations in synaptic currents which cancel the effect of shared input. Near-zero mean correlations were seen experimentally in recordings from rodent neocortex in vivo. Our results suggest a re-examination of the sources underlying observed correlations and their functional consequences for information processing.

Neuronal populations in sensory cortex produce variable responses to sensory stimuli, and exhibit intricate spontaneous activity even without external sensory input. Cortical variability and spontaneous activity have been variously proposed to represent random noise, recall of prior experience, or encoding of ongoing behavioral and cognitive variables. Recording over 10,000 neurons in mouse visual cortex, we observed that spontaneous activity reliably encoded a high-dimensional latent state, which was partially related to the mouse’s ongoing behavior and was represented not just in visual cortex but across the forebrain. Sensory inputs did not interrupt this ongoing signal, but added onto it a representation of external stimuli in orthogonal dimensions. Thus, visual cortical population activity, despite its apparently noisy structure, reliably encodes an orthogonal fusion of sensory and multidimensional behavioral information.

Similarities in neocortical circuit organization across areas and species suggest a common strategy to process diverse types of information, including sensation from diverse modalities, motor control, and higher cognitive processes. Cortical neurons belong to a small number of major classes. The properties of these classes are remarkably similar between areas, including their local and long-range connectivity, developmental history, gene expression, intrinsic physiology, and in vivo activity patterns. Each class contains multiple subclasses; for a rapidly growing number of these, conserved patterns of input and output connections are also becoming evident. The ensemble of circuit connections constitutes a basic circuit pattern that appears to be repeated across neocortical areas, with area- and species-specific modifications. Such “serially homologous” organization may adapt individual neocortical regions to the type of information each must process.

Simultaneous recording from large numbers of neurons is a prerequisite for understanding their cooperative behavior. Various recording techniques and spike separation methods are being used toward this goal. However, the error rates involved in spike separation have not yet been quantified. We studied the separation reliability of "tetrode" (4-wire electrode)-recorded spikes by monitoring simultaneously from the same cell intracellularly with a glass pipette and extracellularly with a tetrode. With manual spike sorting, we found a trade-off between Type I and Type II errors, with errors typically ranging from 0 to 30% depending on the amplitude and firing pattern of the cell, the similarity of the waveshapes of neighboring neurons, and the experience of the operator. Performance using only a single wire was markedly lower, indicating the advantages of multiple-site monitoring techniques over single-wire recordings. For tetrode recordings, error rates were increased by burst activity and during periods of cellular synchrony. The lowest possible separation error rates were estimated by a search for the best ellipsoidal cluster shape. Human operator performance was significantly below the estimated optimum. Investigation of error distributions indicated that suboptimal performance was caused by inability of the operators to mark cluster boundaries accurately in a high-dimensional feature space. We therefore hypothesized that automatic spike-sorting algorithms have the potential to significantly lower error rates. Implementation of a semi-automatic classification system confirms this suggestion, reducing errors close to the estimated optimum, in the range 0-8%.

Preface The brain continuously adapts its processing machinery to behavioural demands. To achieve this it rapidly modulates the operating mode of cortical circuits, controlling the way information is transformed and routed. This article will focus on two experimental approaches by which the control of cortical information processing has been investigated: the study of state-dependent cortical processing in rodents, and attention in the primate visual system. Both processes involve a modulation of low-frequency activity fluctuations and spiking correlation, and are mediated by common receptor systems. We suggest that selective attention involves processes similar to state change, operating at a local columnar level to enhance the representation of otherwise nonsalient features while suppressing internally generated activity patterns.

Most neuronal interactions in the cortex occur within local circuits. Because principal cells and GABAergic interneurons contribute differently to cortical operations, their experimental identification and separation is of utmost important. We used 64-site two-dimensional silicon probes for high-density recording of local neurons in layer 5 of the somatosensory and prefrontal cortices of the rat. Multiple-site monitoring of units allowed for the determination of their two-dimensional spatial position in the brain. Of the approximately 60,000 cell pairs recorded, 0.2% showed robust short-term interactions. Units with significant, short-latency (<3 ms) peaks following their action potentials in their cross-correlograms were characterized as putative excitatory (pyramidal) cells. Units with significant suppression of spiking of their partners were regarded as putative GABAergic interneurons. A portion of the putative interneurons was reciprocally connected with pyramidal cells. Neurons physiologically identified as inhibitory and excitatory cells were used as templates for classification of all recorded neurons. Of the several parameters tested, the duration of the unfiltered (1 Hz to 5 kHz) spike provided the most reliable clustering of the population. High-density parallel recordings of neuronal activity, determination of their physical location and their classification into pyramidal and interneuron classes provide the necessary tools for local circuit analysis.

Two-photon microscopy of calcium-dependent sensors has enabled unprecedented recordings from vast populations of neurons. While the sensors and microscopes have matured over several generations of development, computational methods to process the resulting movies remain inefficient and can give results that are hard to interpret. Here we introduce Suite2p: a fast, accurate and complete pipeline that registers raw movies, detects active cells, extracts their calcium traces and infers their spike times. Suite2p runs on standard workstations, operates faster than real time, and recovers ~2 times more cells than the previous state-of-the-art method. Its low computational load allows routine detection of ~10,000 cells simultaneously with standard twophoton resonant-scanning microscopes. Recordings at this scale promise to reveal the fine structure of activity in large populations of neurons or large populations of subcellular structures such as synaptic boutons.