Mouse primary somatosensory barrel cortex (wS1) processes whisker sensory information, receiving input from two distinct thalamic nuclei. The first-order ventral posterior medial (VPM) somatosensory thalamic nucleus most densely innervates layer 4 (L4) barrels, whereas the higher-order posterior thalamic nucleus (medial part, POm) most densely innervates L1 and L5A. We optogenetically stimulated VPM or POm axons, and recorded evoked excitatory postsynaptic potentials (EPSPs) in different cell-types across cortical layers in wS1. We found that excitatory neurons and parvalbumin-expressing inhibitory neurons received the largest EPSPs, dominated by VPM input to L4 and POm input to L5A. In contrast, somatostatin-expressing inhibitory neurons received very little input from either pathway in any layer. Vasoactive intestinal peptide-expressing inhibitory neurons received an intermediate level of excitatory input with less apparent layer-specificity. Our data help understand how wS1 neocortical microcircuits might process and integrate sensory and higher-order inputs.

Subdivisions of mouse whisker somatosensory thalamus project to cortex in a region-specific and layer-specific manner. However, a clear anatomical dissection of these pathways and their functional properties during whisker sensation is lacking. Here, we use anterograde trans-synaptic viral vectors to identify three specific thalamic subpopulations based on their connectivity with brainstem. The principal trigeminal nucleus innervates ventral posterior medial thalamus, which conveys whisker-selective tactile information to layer 4 primary somatosensory cortex that is highly sensitive to self-initiated movements. The spinal trigeminal nucleus innervates a rostral part of the posterior medial (POm) thalamus, signaling whisker-selective sensory information, as well as decision-related information during a goaldirected behavior, to layer 4 secondary somatosensory cortex. A caudal part of the POm, which apparently does not receive brainstem input, innervates layer 1 and 5A, responding with little whisker selectivity, but showing decision-related modulation. Our results suggest the existence of complementary segregated information streams to somatosensory cortices.

SummarySensory processing in neocortex is primarily driven by glutamatergic excitation, which is counterbalanced by GABAergic inhibition, mediated by a diversity of largely local inhibitory interneurons. Here, we trained mice to lick a reward spout in response to whisker deflection, and we recorded from genetically defined GABAergic inhibitory neurons in layer 2/3 of the primary somatosensory barrel cortex. Parvalbumin-expressing (PV), vasoactive intestinal peptide-expressing (VIP), and somatostatin-expressing (SST) neurons displayed distinct action potential firing dynamics during task performance. Whereas SST neurons fired at low rates, both PV and VIP neurons fired at high rates both spontaneously and in response to whisker stimulation. After an initial outcome-invariant early sensory response, PV neurons had lower firing rates in hit trials compared to miss trials. Optogenetic inhibition of PV neurons during this time period enhanced behavioral performance. Hence, PV neuron activity might contribute causally to gating the sensorimotor transformation of a whisker sensory stimulus into licking motor output.

Fluorescent calcium indicators are an indispensable tools for monitoring the spiking activity of large neuronal populations in animal models. However, despite the plethora of algorithms developed over the last decades, accurate spike time inference methods for rates greater than 20 Hz are lacking. More importantly, little attention has been devoted to the quantification the statistical uncertainties in spike time estimation, which is essential for assigning a confidence in the inference for a particular recording. To address these challenges, we introduce an auto-regressive generative model that accounts for bursting neuronal activity and baseline fluorescence modulation, and it also applies recent sequential Monte Carlo approaches to obtain joint posterior distributions of static and dynamic model parameters. We show that our inference method is competitive with state-of-the-art algorithms by analysing the CASCADE benchmark datasets. We also show that spike time intervals as short as five milliseconds can be inferred from fluorescence transients recorded using a state-of-the-art genetically encoded indicator. Overall, our study describes a Bayesian inference method to detect neuronal spiking patterns and their uncertainty. The use of particle Gibbs samplers allows for unbiased estimates of all model parameters and it provides a statistical framework to test more specific models of calcium indicators.

No abstract