Whisker-Mediated Touch System in Rodents: From Neuron to Behavior

Abstract

A key question in systems neuroscience is to identify how sensory stimuli are represented in neuronal activity, and how the activity of sensory neurons in turn is "read out" by downstream neurons and give rise to behavior. The choice of a proper model system to address these questions, is therefore a crucial step. Over the past decade, the increasingly powerful array of experimental approaches that has become available in non-primate models (e.g., optogenetics and two-photon imaging) has spurred a renewed interest for the use of rodent models in systems neuroscience research. Here, I introduce the rodent whisker-mediated touch system as a structurally well-established and well-organized model system which, despite its simplicity, gives rise to complex behaviors. This system serves as a behaviorally efficient model system; known as nocturnal animals, along with their olfaction, rodents rely on their whisker-mediated touch system to collect information about their surrounding environment. Moreover, this system represents a well-studied circuitry with a somatotopic organization. At every stage of processing, one can identify anatomical and functional topographic maps of whiskers; "barrelettes" in the brainstem nuclei, "barreloids" in the sensory thalamus, and "barrels" in the cortex. This article provides a brief review on the basic anatomy and function of the whisker system in rodents.

Keywords: barrel field; rodents; somatosensory; thalamic barreloids; vibrissae; vibrissal system; whisker system.

Figure 1

Schematic representation of whisker-barrel system. Each whisker is identified by a unique letter-number combination corresponding to its row (A to E from dorsal to ventral) and arc (identified by numbers 1, 2 and etcetera from caudal to rostral), with α, β, γ, and δ straddlers between rows. Colors indicate rows. Barrel, barreloid and barrelets are redrawn from Durham and Woolsey (1984). PMBSF, posterior-medial barrel sub-field; PO, posterior thalamic nucleus; PrV, principal trigeminal nucleus; SC, superior colliculus; SpVi, spinal trigeminal nuclei pars interpolaris; SpVo, pars oralis; SpVc, pars caudalis; TRN, thalamic reticular nucleus; VPM, ventro-posterior medial nucleus; vMI, vibrissal primary motor cortex; vSI, vibrissal primary somatosensory cortex; vSII, secondary somatosensory cortex with the somatotopic map from Benison et al. (2007); DLS, dorsolateral striatum; ZIv, ventral zona inserta. The evidence for somatotopic map in vM1 is provided in Ferezou et al. (2007) and Sreenivasan et al. (2016).

Figure 2

The laminar organization of SI. A coronal section of the somatosensory cortex with cresyl violet Nissl Staining (left panel). The white arrowheads indicate barrels in layer IV. Excitatory neurons in layer II/III are GFP labeled with their terminals in Layer Va. The laminar pathway containing glutamatergic excitatory projections from VPM to layer IV and sparsely to layers Vb and VI (labeled red). The paralaminar pathway containing the projections from POm to layer Va and I (labeled cyan). The pink boxes represent the barrels, and the light purple boxes represent infrabarrels. Adopted from Petersen (2007) and modified. Immunohistology and confocal microscopy image by Ehsan Kheradpezouh and Mehdi Adibi.

Figure 3

Neural activity in somatosensory cortex. (A) The population activity increases with the magnitude of whisker deflection stimulation (single-cycle sine-wave at 80 Hz). Error bars represent the standard error of means across populations with more than five simultaneously recorded units (n = 8). (B) Trial-to-trial variations in neuronal response (in terms of Fano factor) as a function of stimulus intensity for single neurons (n = 64). The inset depicts the histogram of the linear regression slope of the Fano factor with respect to the z-scored neuronal activity for individual neurons. The dark bars correspond to recordings with a significant linear regression (p<0.05). (C) Color indicates the proportion of joint spike counts for a pair of simultaneously recorded neurons. White circles indicate mean spike counts for each stimulus. The Pearson's correlation coefficient of the spike counts is indicated by ρ for each panel. (D) The mean Pearson's correlation coefficient across all possible pairs of neurons (n = 245) as a function of stimulus intensity. Error bars indicate standard error of means. The inset depicts the histogram of regression slopes of noise correlation against average firing rate for pairs of neurons. Dark bars indicate the cases with significant linear regression (p<0.05). (E) The noise correlation index (Adibi et al., 2013b) as a function of stimulus magnitude averaged across populations containing at least five simultaneously recorded neurons (data from A). Error bars are standard error of mean across populations (n = 8). Most of the neurons exhibit a negative slope indicating Fano factor (B) and noise correlations decrease with firing rate. (F) The strength of correlation, denoted by h: the peak of the cross correlation of a pair of electrodes relative to the chance level (denoted by C). Electrodes were divided into two groups of “Responsive” and “Nonresponsive” based on the median of the mutual information between neuronal responses and whisker stimulation. The distribution of h values for Responsive pairs (where both electrodes were from the Responsive group; cyan) and Nonresponsive pairs (where both electrodes in a pair were from the Nonresponsive group; gray). The inset depicts the average and standard error of means of strength of correlation, h, across electrode pairs as a function of their distance for each category. (G) The histogram shows the joint distribution of h values and noise correlations. r represents the correlation coefficient. (H) Same as (G), but for signal correlation. Inset depicts the histogram of r value calculated for groups of electrode pairs with identical distance. The distribution of r values is positive with a mean of 0.3 indicating that the positive correlation between h and signal correlation is independent of the distance between electrodes and is present across all distances. (A–E) are based on Adibi et al. (2013b), and (F–H) are from Sabri et al. (2016).

Figure 4

The two-alternative-choice behavioral tasks in rodents. (A) Schematic representation of the comparative discrimination paradigm. On every trial, two vibrations S_i and S_j were presented. (B) Four rats were trained in the detection/discrimination task to identify the vibration with the higher amplitude. The neuronal performance is the average performance (based on the area under ROC) across single-units (n = 35) and multiunit clusters (n = 58) from Adibi and Arabzadeh (2011). For each neuron, the stimulus intensity whose detection performance was closest to 60% was chosen as detection threshold (Th). The stimuli corresponding to $\frac{1}{2}$−, $1\frac{1}{2}$−, and 2-fold Th were then selected for estimating the discrimination performances. The same threshold of 60% defined as detection threshold for rats. The rats performed the comparison task between 0−Th, $\frac{1}{2}-1\frac{1}{2}$ and Th − 2Th. Error bars indicate standard error of means across rats or neurons. (C) Schematic representation of the categorical discrimination paradigm. Stimuli were defined as either S+ or S−. In each trial, one of the two vibrations was S+ and the other was S−. Having identified the S+ vibration, the rodent expressed its choice by turning toward the corresponding drinking spout. (D) (Left) Stimulus space. Each circle represents the frequency–amplitude combination of one stimulus. Two groups of rats were trained in the task. For one group (top-left), two frequencies (f = 80 Hz and 2f = 160 Hz) and three amplitudes ($\frac{1}{2}A$ = 8 μm, A = 16μm, and 2A = 32 μm) were used to generate five vibrations, and for second group (bottom-left) three frequencies ($\frac{1}{2}f$ = 40 Hz, f = 80 Hz and 2f = 160 Hz) and two amplitudes (A = 16 μm and 2A = 32 μm) were used to generate five vibrations. Stimuli that were presented together and had to be discriminated (paired stimuli) are connected by lines. The right panel shows the proportion of correct trials (performance) for the corresponding four stimulus-pairs averaged across rats. Error bars are s.e.m. across rats. Re-plotted from (Adibi et al., 2012). (E) The schematic representation of the categorization paradigm. The stimuli are divided into two categories of S_L and S_R, corresponding to left and right choices, respectively. A stimulus S was presented on every trial. The rat identifies the category which stimulus S belongs to. (F) Rats were trained to categorize the orientation of a 9.8 cm-diameter disc with alternating ridges and grooves by licking at one of the two reward spouts. Psychometric functions correspond to two rats trained to categorize orientations 0–45° as horizontal, and 45–90° as vertical (green), and another two rats trained to categorize orientations 0–22.5° as horizontal, and 22.5–90° as vertical (blue). The curves correspond to a Gaussian cumulative function fitted to data. The dots on each curve represent the perceptual decision boundary of each rat. The blue and green vertical dashed lines represent the categorization boundaries of 22.5° and 45°, respectively.