To successfully perform goal-directed navigation, animals must know where they are and what they are doing—e.g., looking for water, bringing food back to the nest, or escaping from a predator. Hippocampal neurons code for these critical variables conjunctively, but little is known about how this where/what code is formed or flexibly routed to other brain regions. To address these questions, we performed intracellular whole-cell recordings in mouse CA1 during a cued, two-choice virtual navigation task. We demonstrate that plateau potentials in CA1 pyramidal neurons rapidly strengthen synaptic inputs carrying conjunctive information about position and choice. Plasticity-induced response fields were modulated by cues only in animals previously trained to collect rewards based on these cues. Thus, we reveal that gradual learning is required for the formation of a conjunctive population code, upstream of CA1, while plateau-potential-induced synaptic plasticity in CA1 enables flexible routing of the code to downstream brain regions.

To successfully perform goal-directed navigation, animals must know where they are and what they are doing-e.g., looking for water, bringing food back to the nest, or escaping from a predator. Hippocampal neurons code for these critical variables conjunctively, but little is known about how this "where/what" code is formed or flexibly routed to other brain regions. To address these questions, we performed intracellular whole-cell recordings in mouse CA1 during a cued, two-choice virtual navigation task. We demonstrate that plateau potentials in CA1 pyramidal neurons rapidly strengthen synaptic inputs carrying conjunctive information about position and choice. Plasticity-induced response fields were modulated by cues only in animals previously trained to collect rewards based on available cues. Thus, we reveal that gradual learning is required for the formation of a conjunctive population code, upstream of CA1, while plateau-potential-induced synaptic plasticity in CA1 enables flexible routing of the code to downstream brain regions.

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.

Action potentials are the end product of synaptic integration, a process influenced by resting and active neuronal membrane properties. Diversity in these properties contributes to specialized mechanisms of synaptic integration and action potential firing, which are likely to be of functional significance within neural circuits. In the hippocampus, the majority of subicular pyramidal neurons fire high-frequency bursts of action potentials, whereas CA1 pyramidal neurons exhibit regular spiking behavior when subjected to direct somatic current injection. Using patch-clamp recordings from morphologically identified neurons in hippocampal slices, we analyzed and compared the resting and active membrane properties of pyramidal neurons in the subiculum and CA1 regions of the hippocampus. In response to direct somatic current injection, three subicular firing types were identified (regular spiking, weak bursting, and strong bursting), while all CA1 neurons were regular spiking. Within subiculum strong bursting neurons were found preferentially further away from the CA1 subregion. Input resistance (R(N)), membrane time constant (tau(m)), and depolarizing "sag" in response to hyperpolarizing current pulses were similar in all subicular neurons, while R(N) and tau(m) were significantly larger in CA1 neurons. The first spike of all subicular neurons exhibited similar action potential properties; CA1 action potentials exhibited faster rising rates, greater amplitudes, and wider half-widths than subicular action potentials. Therefore both the resting and active properties of CA1 pyramidal neurons are distinct from those of subicular neurons, which form a related class of neurons, differing in their propensity to burst. We also found that both regular spiking subicular and CA1 neurons could be transformed into a burst firing mode by application of a low concentration of 4-aminopyridine, suggesting that in both hippocampal subfields, firing properties are regulated by a slowly inactivating, D-type potassium current. The ability of all subicular pyramidal neurons to burst strengthens the notion that they form a single neuronal class, sharing a burst generating mechanism that is stronger in some cells than others.

Neurons perform computations by integrating inputs from thousands of synapses-mostly in the dendritic tree-to drive action potential firing in the axon. One fruitful approach to studying this process is to record from neurons using patch-clamp electrodes, fill the recorded neurons with a substance that allows subsequent staining, reconstruct the three-dimensional architectures of the dendrites, and use the resulting functional and structural data to develop computer models of dendritic integration. Accurately producing quantitative reconstructions of dendrites is typically a tedious process taking many hours of manual inspection and measurement. Here we present ShuTu, a new software package that facilitates accurate and efficient reconstruction of dendrites imaged using bright-field microscopy. The program operates in two steps: (1) automated identification of dendritic processes, and (2) manual correction of errors in the automated reconstruction. This approach allows neurons with complex dendritic morphologies to be reconstructed rapidly and efficiently, thus facilitating the use of computer models to study dendritic structure-function relationships and the computations performed by single neurons.

CA1 pyramidal neurons are a major output of the hippocampus and encode features of experience that constitute episodic memories. Feature-selective firing of these neurons results from the dendritic integration of inputs from multiple brain regions. While it is known that synchronous activation of spatially clustered inputs can contribute to firing through the generation of dendritic spikes, there is no established mechanism for spatiotemporal synaptic clustering. Here we show that single presynaptic axons form multiple, spatially clustered inputs onto the distal, but not proximal, dendrites of CA1 pyramidal neurons. These compound connections exhibit ultrastructural features indicative of strong synapses and occur much more commonly in entorhinal than in thalamic afferents. Computational simulations revealed that compound connections depolarize dendrites in a biophysically efficient manner, owing to their inherent spatiotemporal clustering. Our results suggest that distinct afferent projections use different connectivity motifs that differentially contribute to dendritic integration.

The conventional view of neurons is that synaptic inputs are integrated on a timescale of milliseconds to seconds in the dendrites, with action potential initiation occurring in the axon initial segment. We found a much slower form of integration that leads to action potential initiation in the distal axon, well beyond the initial segment. In a subset of rodent hippocampal and neocortical interneurons, hundreds of spikes, evoked over minutes, resulted in persistent firing that lasted for a similar duration. Although axonal action potential firing was required to trigger persistent firing, somatic depolarization was not. In paired recordings, persistent firing was not restricted to the stimulated neuron; it could also be produced in the unstimulated cell. Thus, these interneurons can slowly integrate spiking, share the output across a coupled network of axons and respond with persistent firing even in the absence of input to the soma or dendrites.

Tissue and organ function has been conventionally understood in terms of the interactions among discrete and homogeneous cell types. This approach has proven difficult in neuroscience due to the marked diversity across different neuron classes, but it may be further hampered by prominent within-class variability. Here, we considered a well-defined canonical neuronal population-hippocampal CA1 pyramidal cells (CA1 PCs)-and systematically examined the extent and spatial rules of transcriptional heterogeneity. Using next-generation RNA sequencing, we identified striking variability in CA1 PCs, such that the differences within CA1 along the dorsal-ventral axis rivaled differences across distinct pyramidal neuron classes. This variability emerged from a spectrum of continuous gene-expression gradients, producing a transcriptional profile consistent with a multifarious continuum of cells. This work reveals an unexpected amount of variability within a canonical and narrowly defined neuronal population and suggests that continuous, within-class heterogeneity may be an important feature of neural circuits.

Because of their strategic position between the granule cell and pyramidal cell layers, neurons of the hilar region of the hippocampal formation are likely to play an important role in the information processing between the entorhinal cortex and the hippocampus proper. Here we present an electrophysiological characterization of anatomically identified neurons in the fascia dentata as studied using patch-pipette recordings and subsequent biocytin-staining of neurons in slices. The resting potential, input resistance (RN), membrane time constant (taum), "sag" in hyperpolarizing responses, maximum firing rate during a 1-s current pulse, spike width, and fast and slow afterhyperpolarizations (AHPs) were determined for several different types of hilar neurons. Basket cells had a dense axonal plexus almost exclusively within the granule cell layer and were distinguishable by their low RN, short taum, lack of sag, and rapid firing rates. Dentate granule cells also lacked sag and were identifiable by their higher RN, longer taum, and lower firing rates than basket cells. Mossy cells had extensive axon collaterals within the hilus and a few long-range collaterals to the inner molecular layer and CA3c and were characterized physiologically by small fast and slow AHPs. Spiny and aspiny hilar interneurons projected primarily either to the inner or outer segment of the molecular layer and had a dense intrahilar axonal plexus, terminating onto somata within the hilus and CA3c. Physiologically, spiny hilar interneurons generally had higher RN values than mossy cells and a smaller slow AHP than aspiny interneurons. The specialized physiological properties of different classes of hilar neurons are likely to be important determinants of their functional operation within the hippocampal circuitry.