The auditory midbrain mediates tactile vibration sensing

Abstract

Vibrations are ubiquitous in nature, shaping behavior across the animal kingdom. For mammals, mechanical vibrations acting on the body are detected by mechanoreceptors of the skin and deep tissues and processed by the somatosensory system, while sound waves traveling through air are captured by the cochlea and encoded in the auditory system. Here, we report that mechanical vibrations detected by the body's Pacinian corpuscle neurons, which are distinguished by their ability to entrain to high-frequency (40-1,000 Hz) environmental vibrations, are prominently encoded by neurons in the lateral cortex of the inferior colliculus (LCIC) of the midbrain. Remarkably, most LCIC neurons receive convergent Pacinian and auditory input and respond more strongly to coincident tactile-auditory stimulation than to either modality alone. Moreover, the LCIC is required for behavioral responses to high-frequency mechanical vibrations. Thus, environmental vibrations captured by Pacinian corpuscles are encoded in the auditory midbrain to mediate behavior.

Keywords: in vivo physiology; mechanotransduction; multi-sensory integration; somatosensory neurons; vibration.

Figure 1.. LCIC and VPL neurons are differentially tuned to mechanical vibratory stimuli.

A) The brainstem dorsal column nuclei (DCN) receive inputs from dorsal root ganglia (DRG) primary sensory neurons and spinal cord projection neurons. Non-overlapping subsets of DCN projection neurons target the ventroposterolateral thalamus (VPL; blue) and the external nucleus of the inferior colliculus (LCIC; red). B and C) Representative VPL (B) and LCIC (C) neuron spike raster and histogram in response to 50 Hz and 500 Hz vibratory stimuli delivered at a 10 mN force to the hindlimb. Top: spike raster; bottom: average histogram in 10 ms bins. D) Percentage of neurons (mean ± SD) with hindbody brush receptive fields (tested by brushing the hindpaw, thigh, and trunk) that respond to 500 Hz stimulation of the limb at a 10 mN force in the VPL (n= 182 neurons from 4 animals) and LCIC (n= 76 neurons from 4 animals). Each data point corresponds to measurements in one animal. See also Supplementary Figure 3. E and F) Example VPL (E) and LCIC (F) neuron receptive field organization for 50 Hz and 500 Hz stimulation to multiple regions across the hindlimb delivered at a 10 mN intensity. Stimulus locations are superimposed on images of the hindpaw.

Figure 2.. LCIC and VPL neurons exhibit distinct force-frequency tuning curves.

A and B) Representative VPL (A) and LCIC (B) neuron responses to vibrational frequencies delivered at different forces to the hindlimb. Heatmap intensity corresponds to the neuron’s average change in firing to the sustained portion of vibratory stimuli (200 ms window beginning 50 ms after stimulus onset). C and D) Force-frequency threshold curves obtained from heatmaps in A and B, where the force threshold for a vibrational frequency is defined as the minimum force required to elicit 20% of the neuron’s maximum firing during the sustained portion of the stimulus. Frequencies in which the highest force tested (50 mN) failed to reach threshold are designated NR (non-responsive). E) Average threshold curves (mean ± 95% confidence interval) for neurons responsive to hindlimb vibration recorded in the VPL (n=15 neurons from 6 animals) and LCIC (n=22 neurons from 6 animals). F and G) Force-frequency threshold curves for representative Meissner corpuscle innervating Aβ RA1-LTMR afferent (F) and Pacinian corpuscle innervating Aβ RA2-LTMR afferent (G). Threshold is defined as the minimum force necessary for the neuron to reach 50% vibratory entrainment. Frequencies in which the highest force tested (50 mN) failed to reach threshold are designated NR (non-responsive). H) Average threshold curves (mean ± 95% confidence interval) for hindlimb Meissner Aβ RA1-LTMRs (n=4 neurons from 4 animals) and Pacinian Aβ RA2-LMTRs (n=6 neurons from 5 animals).

Figure 3.. LCIC vibration responses are mediated by Pacinian corpuscle Aβ RA2-LTMRs.

A) Meissner corpuscles, located in the plantar pad and digit glabrous skin, are absent in Avil^Cre; TrkB^flox/flox (TrkB cKO) mice, which lack TrkB in DRG sensory neurons. Images of skin histology from a TrkB cKO and littermate control, stained for anti-NFH (labels myelinated axons, green) and anti-S100 (labels glial cells, magenta). Scale bar = 50 μm. See also Supplementary Figure 4 for quantification. B) Pacinian corpuscles, located in the periosteum of the fibula, are absent in Avil^Cre; Ret^flox/flox (Ret cKO) mice, which lack Ret in DRG sensory neurons. Images of whole mount histology of the limb of a Ret cKO animal and littermate control, stained for anti-NFH and anti-S100 to visualize Pacinian corpuscles. White dashed line indicates position of the fibula. Scale bar = 100 μm. See also Supplementary Figure 4 for quantification. C and D) Force-frequency heat maps for representative VPL (C) and LCIC (D) neurons in control, TrkB cKO and Ret cKO animals. Heatmap intensity corresponds to the neuron’s average change in firing to the sustained portion of vibratory stimuli. E) Maximum firing rate for VPL and LCIC (mean ± 95% confidence interval) neurons to vibratory stimuli in control, TrkB cKO and Ret cKO mice. Firing rates of LCIC neurons in Ret cKO and control mice were significantly different (****p<0.0001 Mann-Whitney U test). Firing rates of VPL neurons in TrkB cKO, Ret cKO, and controls were not significantly different. Firing rates of LCIC neurons in TrkB cKO and controls were not significantly different (Mann-Whitney U test). F) Average threshold curves of VPL neurons (mean ± 95% confidence interval) recorded in control (n=15 neurons from 6 animals), TrkB cKO (n=10 neurons from 3 animals), and Ret cKO (n=14 neurons from 4 animals) mice. Thresholds between Ret cKO, TrkB cKO and controls were not significantly different (Mann-Whitney U Test with Bonferroni correction for multiple comparisons). G) Average threshold curves of LCIC neurons (mean ± 95% confidence interval) recorded in control (n=22 neurons from 6 animals), TrkB cKO (n=9 neurons from 4 animals), and Ret cKO (n=18 neurons from 4 animals) mice. (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, Mann-Whitney U test with Bonferroni correction for multiple comparisons). While thresholds between Ret cKO and controls were significantly different, thresholds between TrkB cKO and controls were not (Mann-Whitney U Test with Bonferroni correction for multiple comparisons).

Figure 4.. Audio-tactile multimodality of LCIC neurons in awake mice.

A) Recordings of LCIC neurons were conducted in awake head-fixed mice as they stood on an acrylic platform, which could be stimulated at different vibratory frequencies. An accelerometer was attached underneath the acrylic platform to measure vibration frequency and amplitude. An example motor command and accelerometer signal are shown for 500 Hz vibration of the platform. A speaker is positioned on the contralateral side of the head from the recording site. B) Peristimulus time histogram (PSTH) for a representative LCIC neuron, aligned to 50 Hz (left) and 500 Hz (right) vibration stimulation of the acrylic platform at a vibration amplitude of ~0.2 g (gravitational acceleration). PSTHs are baseline subtracted and reported in 10 ms bins. C) Percentage of hindlimb sensitive LCIC neurons that responded to 50 Hz and 500 Hz vibratory stimuli (n=24 neurons from 5 animals). D) 500 Hz substrate vibration (amplitude 0.2 g) and auditory white noise (frequency 0–50 kHz, 100 ms pulses, 65 dB SPL) were delivered in isolation and coincidentally to awake head-fixed mice. E-G) Example PSTHs from LCIC neurons that respond to auditory stimuli (E), vibration (F), or both (G), aligned to vibration (v), sound (s) and convergent (vs) stimuli. PSTHs are baseline subtracted and in 10 ms bins. Right: The example neuron’s average firing (during 0–0.1 s and 0.5–0.6 s windows) for each stimulus type. H) Heatmaps of baseline subtracted PSTHs from LCIC neurons aligned to vibration (v), sound (s) and convergent (vs) stimuli. Each row of the heatmap corresponds to an individual neuron (n=29 neurons from 5 animals), and neurons are ranked by their responsiveness to vibration. I) Percentage of neurons in control animals that respond to 500 Hz vibration only, sound only, or convergent stimuli (n=29 neurons from 5 animals). J) Scatter plot of firing rates in response to combined vibration and sound (vs) plotted against best unimodal response (vibration only or sound only). Unitary line denotes equal firing between best unimodal stimulus and combined stimulus conditions (n=21 neurons from 5 animals). K) Quantification of multisensory enhancement (corresponding to I), calculated using the following equation: $\frac{\text{vs}-(\text{Max}(\text{v}\text{or}\text{s}))}{\text{Max}(\text{v}\text{or}\text{s})}*100$. Neurons above 0 display multisensory enhancement (n=21 neurons from 5 animals, *p<0.05 one sample t-test)^,.

Figure 5.. LCIC neuron responses in animals that lack Pacinian corpuscles or cochlear function.

A) Schematic of Ret cKO mouse which lacks Pacinian corpuscles, but has intact hearing. B) Schematic of GJB2 cKO mouse which lacks cochlear function, but has an intact somatosensory system. C) 500 Hz substrate vibration (amplitude 0.2 g) and auditory white noise (frequency 0–50 kHz, 100 ms pulses, 65 dB SPL) were delivered in isolation and coincidentally to awake head-fixed mice. D) Example PSTHs from an LCIC neuron of a wildtype mouse aligned to vibration (v), sound (s) and convergent (vs) stimuli. PSTHs are baseline subtracted and in 10 ms bins. Right: The example neuron’s average firing (during 0–0.1 s and 0.5–0.6 s windows) for each stimulus type. E and F) As in (D), example LCIC neurons in a Ret cKO animal, which lacks Pacinian corpuscles (E), and a GJB2 cKO animal, which is unable to hear (F). G) Scatter plot of each neuron’s average change in firing rate to vibration and sound stimuli in control (n=29 neurons from 5 animals), Ret cKO (n=14 neurons from 3 animals), and GJB2 cKO (n=15 neurons from 3 mice) animals. H) Average change in firing rate to vibration, sound, or combined stimuli (mean ± SD) in control (n=29 neurons from 5 animals), Ret cKO (n=14 neurons from 3 animals), and GJB2 cKO (n=15 neurons from 3 mice) animals (**p<.01, ***p<0.001, ****p<0.0001, Mann-Whitney U test). I) Quantification of multisensory enhancement in control (n=21 neurons from 5 animals), Ret cKO (n=14 neurons from 3 animals), and GJB2 cKO (n=15 neurons from 3 mice) mice, calculated using the following equation: $\frac{\text{vs}-(\text{Max}(\text{v}\text{or}\text{s}))}{\text{Max}(\text{v}\text{or}\text{s})}*100$. Neurons above 0 display multisensory enhancement (*p<0.05, Mann-Whitney U test)^,.

Figure 6.. A behavioral paradigm to assess behavioral responsiveness to high frequency environmental vibration.

A) Animals were placed in a 2-chamber behavioral setup. The floors of each chamber are disconnected to prevent propagation of vibration between them, and an exciter speaker is attached to the floor of one chamber to generate mechanical vibrations. The chamber walls are partially open to prevent sound resonance from the exciter. B and C) The percentage of time wildtype CD-1 animals spend on each side of the chamber (mean ± SD) during the 60 s baseline explore period (B) and the vibration periods (C) of the trial. Each dot represents one animal (n=20 animals, ****p<0.0001 paired t-test). D-F) Representative activity traces of two example wildtype CD-1 animals (D), littermate control animals (E), and Ret cKO animals (F), tracking movement during the baseline and vibration periods. G and H) The speaker was detached from the floor of the chamber to test whether the sound of the exciter speaker alone can drive a behavioral preference. Time spent on each side of the chamber during the 60 s baseline explore period (G) and the sound on periods (H) (not statistically different, n=10 animals, paired t-test). I and J) The percentage of time Ret cKO mice (n=14 animals) and littermate controls (n=24 animals) spent on each side of the chamber during the 60 s baseline explore period (I) and the vibration periods (J) of the behavioral trial. (****p<0.0001, paired t-test within group, unpaired t-test between groups).

Figure 7.. The LCIC mediates behavioral avoidance of high frequency mechanical vibration.

A) Targeted injections of TMR-X conjugated to the GABA-A receptor agonist, muscimol, were used to silence neural activity in the inferior colliculus. Top-down view of a schematic (left) and representative muscimol injected brain. Dotted lines depict borders between the inferior colliculus and indicated neighboring brain regions. Scale bar = 1 mm. B) PSTHs from LCIC neurons in awake animals after saline (gray) or muscimol (red) injection, aligned to 500 Hz vibration of the surface platform or white noise pulses. Population averages are displayed for saline (black) and muscimol (dark red) conditions. PSTHs are baseline subtracted and reported in 10 ms bins. C) Reduction in LCIC firing rate in awake animals after saline or muscimol injection (mean ± SEM), measured by calculating the change in total number of spikes on the probe before and after injection (n=3 animals). D and E) The proportion of time muscimol injected mice (n=9 animals) and saline injected controls (n=8 animals) spent on each side of the chamber during the 60 s baseline explore period (D) and the vibration periods (E) of the trial. (**p<0.01, ***p<0.001, paired t-test within group, unpaired t-test between groups). F) Total distance traveled during the behavioral assay in saline injected controls (n=8) and muscimol injected animals (n=9 animals, not significant, unpaired t-test). G) Representative activity traces for 2 saline injected (control) and 2 muscimol injected animals during the baseline and vibration periods of the behavioral trial. H) Schematic of texture preference assay. I) Percentage of time spent on the smooth and rough sides of the chamber during the 10 min assay for saline controls (n=9 animals) and muscimol silenced animals (n=9 animals, ***p<0.001, paired t-test within group, unpaired t-test between groups). J) Schematic of temperature preference assay. K) Percentage of time spent on the control (30 °C) and cold (18°C) sides of the chamber during the 5 min assay for saline controls (n=8) and muscimol silenced animals (n=8, ***p<0.001, paired t-test within group, unpaired t-test between groups).