Mechanisms of spectral and temporal integration in the mustached bat inferior colliculus

Abstract

This review describes mechanisms and circuitry underlying combination-sensitive response properties in the auditory brainstem and midbrain. Combination-sensitive neurons, performing a type of auditory spectro-temporal integration, respond to specific, properly timed combinations of spectral elements in vocal signals and other acoustic stimuli. While these neurons are known to occur in the auditory forebrain of many vertebrate species, the work described here establishes their origin in the auditory brainstem and midbrain. Focusing on the mustached bat, we review several major findings: (1) Combination-sensitive responses involve facilitatory interactions, inhibitory interactions, or both when activated by distinct spectral elements in complex sounds. (2) Combination-sensitive responses are created in distinct stages: inhibition arises mainly in lateral lemniscal nuclei of the auditory brainstem, while facilitation arises in the inferior colliculus (IC) of the midbrain. (3) Spectral integration underlying combination-sensitive responses requires a low-frequency input tuned well below a neuron's characteristic frequency (ChF). Low-ChF neurons in the auditory brainstem project to high-ChF regions in brainstem or IC to create combination sensitivity. (4) At their sites of origin, both facilitatory and inhibitory combination-sensitive interactions depend on glycinergic inputs and are eliminated by glycine receptor blockade. Surprisingly, facilitatory interactions in IC depend almost exclusively on glycinergic inputs and are largely independent of glutamatergic and GABAergic inputs. (5) The medial nucleus of the trapezoid body (MNTB), the lateral lemniscal nuclei, and the IC play critical roles in creating combination-sensitive responses. We propose that these mechanisms, based on work in the mustached bat, apply to a broad range of mammals and other vertebrates that depend on temporally sensitive integration of information across the audible spectrum.

Keywords: biosonar; combination sensitivity; combination-sensitive; echolocation; facilitation; glycinergic; lateral lemniscus; medial nucleus of trapezoid body.

Figure 1

Spectral and temporal features of vocalizations. Schematic sonogram of echolocation signal displays emitted pulse (in red) and a Doppler (frequency) shifted and time delayed echo (in blue). Each signal is composed of CF, constant frequency; FM, frequency modulated components, with several harmonic elements (e.g., FM₁, FM₂, etc.). Line thickness indicates relative intensity: the second harmonic of the emitted pulse, near 60 kHz, is the most intense while the fundamental is usually less intense than either the second or third harmonics. Ovals indicate sonar elements to which the neurons in Figure 2 are tuned. Social vocalizations span the range from approximately 5–100 kHz.

Figure 2

Spectral and temporal tuning of combination sensitivity in the mustached bat's IC. Figure shows responses of a facilitated neuron (A and C) and an inhibited neuron (B and D). (A) Facilitation frequency tuning curves for high-frequency (red) and low-frequency (blue) tone bursts. These curves were obtained by fixing the frequency and level of one tone burst (X) while varying the frequency and level of a second tone burst in the other frequency band, in order to obtain threshold facilitative responses. Facilitation was defined as a response to the combination stimulus that was 20% greater than the sum of responses to the two stimuli presented separately. The high-frequency tone burst was presented at a delay corresponding to the neuron's best delay of facilitation (shown in C). (B) Excitatory (red) and inhibitory (filled blue) tuning of neuron showing combination-sensitive inhibition. The low-frequency inhibitory tuning curve was obtained by presenting a characteristic frequency tone burst at a fixed level (X), then varying the frequency and level of a low-frequency tone burst to obtain threshold inhibitory responses. Inhibition was defined as a response to the combination stimulus that was 20% less than the sum of responses to the two stimuli presented separately. The two tones were presented at the neuron's best delay of inhibition (shown in D). Black bars at top in (A and B) indicate frequency ranges of fundamental (H1) and third (H3) harmonic elements of biosonar call. (C) Delay tuning of facilitation for neuron in (A). Neuron responded poorly to individual tone bursts, but strongly to the combination of facilitating tones when the high-frequency signal was delayed by 0–4 ms. Note inhibition of high-frequency response by low-frequency signal at delay of 10 ms. (D) Delay tuning of inhibition for neuron in (B). Neuron's response to the ChF tone was inhibited by low-frequency tones when the signals were presented simultaneously. Adapted from Portfors and Wenstrup (1999), with permission.

Figure 3

Spectral and temporal features of combination-sensitive neurons in mustached bat IC. (A,B) Spectral tuning of facilitation (A) and inhibition (B). Black rectangles indicate frequency combinations that are present in echolocation signals. (C,D) Delay tuning of facilitation and inhibition. Best delays of facilitation (C) were broadly distributed for FM–FM neurons but tightly distributed around 0 ms delay for other types of facilitated neurons. Delay tuning of combination-sensitive inhibition (D) was similar for neurons showing only inhibition and for those facilitated neurons showing early inhibition. Data from Portfors and Wenstrup (1999); Leroy and Wenstrup (2000); Nataraj and Wenstrup (2005, 2006).

Figure 4

Delay tuning of neuron showing both facilitation and inhibition. (A) This neuron's excitatory response to ChF tones (at 82 kHz) is inhibited by low frequency tones (24 kHz) at short and long delays, but is facilitated at intermediate delays. (B) Post-stimulus time histograms showing temporal features of the neuron's response to tone combinations. Adapted from Nataraj and Wenstrup (2005), with permission.

Figure 5

Facilitation in combination-sensitive neurons is activated by the onset of the low-frequency signal and is phasic. (A) Delay sensitivity of neuron was tested with two durations of low-frequency signal. (B) FAC_START in scatterplot refers to the shortest delay that evokes facilitation. Data points fall along the solid line, indicating that the change in low-frequency duration had no effect on this measure. Thus, facilitation is locked to the onset of the low-frequency signal. If the facilitation was locked to low-frequency offset, the delay curve is expected to shift to the right and data points in this scatter plot would fall along the dashed line. (C) FAC_END refers to the longest delay that evokes facilitation, a measure of the duration of the facilitating effect of the low-frequency signal. FAC_END is invariant with changes in low-frequency duration, indicating that the facilitating effect is phasic and independent of low-frequency signal duration. Adapted from Gans et al. (2009), with permission.

Figure 6

Effects of receptor blockade on low-frequency-evoked inhibition and suppression. (A) In INLL neuron, blockade of GABA_A receptor (GABA_AR) by bicuculline did not eliminate 28 kHz inhibition, but addition of GlyR blockade by strychnine completely eliminated this inhibition. (B) Effects of receptor blockade on 23–30 kHz inhibition on population of tested NLL neurons. While GABA_AR blockade alone (at left) did not eliminate combination-sensitive inhibition in any neuron, GlyR blockade (at right) always eliminated or greatly reduced inhibition evoked by 23–30 kHz signal. In (B, D, and F), interaction index expresses the degree of facilitation (positive values) or inhibition (negative values). The greyed area indicates no significant interaction. Green dashed lines indicate results from GlyR receptor blockade alone, compared to black lines that show combined GABA_AR and GlyR blockade. (C) In same INLL neuron as in (A), blockade of GABA_AR or both GABA_AR and GlyR failed to eliminate 18 kHz suppression. (D) Effects of receptor blockade on <23 kHz suppression among NLL neurons. Neither GABA_AR nor GlyR blockade eliminated suppression tuned to frequencies below 23 kHz. (E) In an IC neuron, blockade of GlyR did not eliminate 26 kHz inhibition, although the delay function was narrowed. Combination of GlyR and GABA_AR blockade failed to eliminate 26 kHz inhibition. (F) Effects of receptor blockade on 23–30 kHz inhibition among IC neurons. Data suggest that many IC neurons inherit combination sensitivity from auditory brainstem inputs, but that some inhibitory inputs tuned to 23–30 kHz terminate onto high-ChF neurons in IC. Adapted from Peterson et al. (2009) (A–D) and Nataraj and Wenstrup (2005, 2006) (E,F), with permission.

Figure 7

Inputs to INLL neurons that show combination-sensitive inhibition. (A) Left. Distribution of retrograde labeling after INLL deposits in five animals. Right. Distribution of double labeled cells (glycine-immunopositive and retrogradely labeled) after INLL tracer deposits. The ipsilateral MNTB provides the strongest glycinergic input to INLL neurons. (B) Comparison of retrograde label in MNTB after INLL deposits at combination-sensitive site (left) and low-frequency tuned site (right). MNTB labeling after combination-sensitive deposits is in both medial and lateral locations, indicating input from both low and high-frequency bands. MNTB label after low-frequency tuned deposits is located laterally, i.e., in the low-frequency representation. Adapted from Yavuzoglu et al. (2010), with permission.

Figure 8

Schematic diagram of circuitry underlying combination-sensitive inhibition in INLL and IC. The red and blue striped arrow indicates sensitivity to both low and high-frequency bands.

Figure 9

Glycine receptor blockade eliminates combination-sensitive facilitation in IC neurons. (A) In IC neuron, blockade of GlyRs eliminates 27 kHz facilitation of 86 kHz ChF response. (B) Effects of GlyR blockade on low-frequency facilitation among IC neurons. In all neurons, GlyR blockade eliminated or greatly reduced combination-sensitive facilitation. (C) In an IC neuron, blockade of GABA_ARs did not eliminate facilitation, but addition of GlyR blockade eliminated facilitation. (D) In most IC neurons, facilitation was not eliminated by GABA_AR blockade, but addition of GlyR blockade always eliminated the facilitation. Adapted from Nataraj and Wenstrup (2005), with permission.

Figure 10

Glutamate receptors (iGluRs) play no role in combination-sensitive facilitation. (A) Responses to ChF and low-frequency tones are eliminated by iGluR blockade. (B) Blockade of iGluRs eliminated excitatory responses to single tones, but facilitated combination-sensitive responses persisted. Only the application of a GlyR blocker eliminated facilitatory interactions. (C) Effects of receptor blockade on low-frequency facilitation among IC neurons. In all neurons, iGluR blockade failed to eliminate facilitation, but GlyR blockade always eliminated facilitation. Blockade of GABA_ARs generally did not eliminate facilitation. Adapted from Sanchez et al. (2008), with permission.

Figure 11

Primary role of GlyR receptors in the facilitated response of IC neurons. Graphs show number of spikes evoked by ChF and combination stimuli at best delay, averaged across the number of neurons in the sample. iGluR blockade eliminates spikes evoked by ChF tones, but does not eliminate facilitated spikes evoked by combination stimuli. The addition of GABA_AR blockade to the iGluR blockade (+GABA_AR Block) has little additional effect on the facilitated spikes. Facilitation spikes are only eliminated by addition of the GlyR blockade (+GlyR Block). Neither + GABA_AR Block nor +GlyR Block revealed additional excitatory response to the ChF response, suggesting that iGluR blockade successfully eliminated glutamatergic excitation to the neurons. Adapted from Sanchez et al. (2008), with permission.

Figure 12

Schematic diagram of mechanisms and circuitry underlying combination-sensitive facilitation in IC. Inset shows hypothesized mechanism of post-inhibitory rebound. Neuron receives a variety of high frequency inputs tuned to its ChF (upper right) that do not appear to interact with glycinergic inputs related to facilitation (lower left). Response enhancement refers to hypothesized boost in the glycine rebound potentials that allows the facilitation signal to reach the spike trigger zone.

Figure 13

Lateral lemniscal nuclei provide key glycinergic inputs to facilitated combination-sensitive neurons in IC. (A) Left. Average percentages of neurons double-labeled by tracer (FluoroGold, FG) deposited at IC combination-sensitive recording sites and by glycine immunohistochemistry. This represents the distribution of glycinergic inputs to IC regions containing facilitated combination-sensitive neurons. Right. Distribution of “proximity-labeled neurons”; these neurons are retrogradely labeled by the IC tracer deposits and within 50 μm of a labeled terminal resulting from deposit of a second tracer in a low frequency part of AVCN. This represents neurons that likely receive input tuned to 23–30 kHz and project to high-ChF, combination-sensitive recording sites in IC. (B) Locations of these “proximity labeled” cells in VNLL and INLL from one experiment. Adapted from Yavuzoglu et al. (2011), with permission.

Figure 14

Schematic diagram of circuitry underlying combination-sensitive facilitation and inhibition in IC neuron.