Interplay of inhibition and excitation shapes a premotor neural sequence

Abstract

In the zebra finch, singing behavior is driven by a sequence of bursts within premotor neurons located in the forebrain nucleus HVC (proper name). In addition to these excitatory projection neurons, HVC also contains inhibitory interneurons with a role in premotor patterning that is unclear. Here, we used a range of electrophysiological and behavioral observations to test previously described models suggesting discrete functional roles for inhibitory interneurons in song production. We show that single HVC premotor neuron bursts are sufficient to drive structured activity within the interneuron network because of pervasive and facilitating synaptic connections. We characterize interneuron activity during singing and describe reliable pauses in the firing of those neurons. We then demonstrate that these gaps in inhibition are likely to be necessary for driving normal bursting behavior in HVC premotor neurons and suggest that structured inhibition and excitation may be a general mechanism enabling sequence generation in other circuits.

Keywords: behavior; circuit; inhibition; intracellular; neural sequence; premotor.

Figure 1.

Models of synaptic conductances contributing to song-related activity in HVC. A, During singing, HVC_(RA) neurons typically exhibit one high-frequency burst of spikes as well as multiple depolarizing synaptic events. A dashed box is used to indicate the occurrence of a burst event across all panels. Several models have been proposed to explain this behavior. B, C, Activity may be exclusively driven by excitation, with inhibition either absent (B) or lacking the temporal structure required to contribute directly to motor patterning (C). D, E, Nonuniform inhibition may contribute directly to activity during singing. D, Bursts could result from excitatory events that are temporally aligned with local minima in inhibition, whereas excitatory events in the presence of inhibition remain subthreshold. E, Inhibition could act as a filter in the presence of global excitation such that short gaps in inhibition result in depolarizations and longer inhibitory pauses can be sufficient to trigger a burst.

Figure 2.

Inhibition is necessary for normal singing behavior. A, Representative sonogram (range: 500–7500 Hz) of a song motif after bilateral global infusion of PBS or gabazine into HVC. B, Global similarity scores after gabazine infusion for the bird in A. Shown are gabazine-control (red) and control-control (black) comparisons. C, The mean song similarity for five birds (filled circles representing the bird highlighted to the left).

Figure 3.

Coherent gaps in HVC interneuron spiking activity during singing. A, Example juxtacellular recording from an interneuron during singing along with a dot raster plot showing precise temporal structure across six song motifs. The mean firing rate is described in the plot below and gaps are designated with shaded rectangles in the raster plot. B, C, Histograms showing the number of gaps per song motif (B) as well as the duration of gaps (C) across a population of 19 HVC interneurons. D, Example intracellular recording from an interneuron during singing and a raster plot showing the spike times of that neuron across nine song motifs (in dark blue) as well as three additional interneurons (shades of blue) recorded from the same electrode track in one bird. E, Gaps in the firing of different interneurons during song production are indicated by lines (n = 13 interneurons in 4 birds). Cells from each bird are recorded within a single electrode track. Dots symbolize the end of the motif. F, Occurrence of inhibitory gaps (plotted as percentage of total motif duration) for the population of interneurons shown in E. A value of 0 indicates a lack of gaps across the population and a value of 1 indicates only one gap occurring at a particular time. Values of 2 or more indicate simultaneously occurring gaps.

Figure 4.

Classification of HVC cell types in vitro. A, Schematized parasagittal section through the zebra finch brain, showing a retrograde dye injection into nucleus RA and Area X to label HVC_(RA) and HVC_(X) neurons, respectively. B, Dodt contrast image of HVC in the slice (HVC perimeter represented with a dashed line). C, Individual, DiI-labeled HVC_(RA) neurons visualized with a 40× objective (red overlay of fluorescence image) and targeted with whole-cell patch pipettes. D, HVC interneurons frequently formed inhibitory synaptic connections onto nearby cells (see Fig. 5 for an example), confirming their identity (blue triangles, n = 22 pairs). Otherwise, putative HVC interneurons (gray triangles, n = 13 cells) were classified based on their intrinsic properties, namely spike waveform (overlay on left) and RMP, which were distinct from a set of retrogradely labeled HVC_(RA) neurons (n = 39 cells). E, Putative HVC interneurons were also separable from a set of retrogradely labeled HVC_(X) neurons (n = 17 cells). On average, the RMP of HVC_(RA) neurons, HVC_(X) neurons, and HVC interneurons were −70.37 ± 5.70 mV, −61.35 ± 3.27 mV, and −53.97 ± 2.80 mV, respectively. Spike half-width of these three major HVC cell classes were also different across these groups (HVC_(RA): 0.58 ± 0.15 ms, HVC_(X): 0.80 ± 0.13 ms, interneuron: 0.27 ± 0.06 ms).

Figure 5.

Synaptic interactions between HVC_(RA) neurons and interneurons. A, B, Reconstruction of an HVC_(RA) neuron (axon = red; A) and an HVC interneuron (axon = blue; B) recorded in vitro. Inset are responses of these cells to hyperpolarizing (−400 pA) and depolarizing current steps (HVC_(RA): 200 pA, interneuron: 400 pA). C, D, Reciprocal synaptic connectivity between the neurons shown in A and B (black trace = average response). E, F, Synaptic strength of all recorded HVC_(RA)→interneuron (E) and interneuron→HVC_(RA) (F) connections with respect to intersomatic distance (gray is nonconnected). G, Frequency dependence of the HVC_(RA)→interneuron connections showing that a 300 Hz HVC_(RA) burst activates a postsynaptic interneuron (RMP = −54.5 mV). H, An HVC_(RA) burst often produces a reliable spike in postsynaptic interneurons (7 pairs, average in black). I, A typical HVC_(RA)→interneuron pair in which facilitating EPSCs can be visualized using voltage-clamp. J, Triplet recording from one HVC_(RA) neuron connected to two interneurons, as shown by the schematic on top. A single 300 Hz burst (3 spikes) in the HVC_(RA) neuron (red trace) was able to reliably evoke synchronous spiking in two postsynaptic interneurons (blue = example traces, black = raster plots from 15 trials).

Figure 6.

Prominent disynaptic inhibition between HVC_(RA) neurons. A, Schematic and recording of two HVC_(RA) neurons and an intermediate interneuron. An evoked high-frequency burst (4 spikes at 300 Hz) in one HVC_(RA) neuron typically elicits an action potential in an interneuron (RMP = −54.4 mV) and subsequent inhibitory responses in another HVC_(RA) cell. B, HVC_(RA) neuron pairs often exhibit frequency-dependent disynaptic inhibition quantified across a population of 9 pairs (average in black). C, Distribution of delay times for all connections (disynaptic HVC_(RA)→HVC_(RA), black line; monosynaptic HVC_(RA)→interneuron, dotted line; monosynaptic interneuron→HVC_(RA) connections, gray line). D, HVC_(RA)→HVC_(RA) disynaptic inhibitory connections plotted with respect to their position in the parasagittal plane (red is presynaptic cell; size of black circles show strength of inhibitory connection; gray Xs are not connected). E, HVC_(RA) → HVC_(RA) disynaptic IPSC strengths (n = 140 pairs) also plotted with their relative directionality within HVC. We sampled uniformly from all angular positions in the parasagittal plane of HVC (p = 0.428, Rayleigh test). F, Smoothed average of IPSC strength values from C, including nonconnected as 0 pA strength (see Materials and Methods for details). Red line (34° ventral deviation from anterior → posterior axis of HVC) shows the largest bimodal bias of disynaptic IPSC strength coupling between pairs of HVC_(RA) neurons.

Figure 7.

Inhibition is necessary for sparse HVC_(RA) neuron spiking but not temporal precision. A, Example of a song motif after the local microinfusion of gabazine into HVC. B, Global similarity scores after gabazine infusion for the bird in A. Shown are gabazine-control (red) and control-control (black) comparisons. C, Mean song similarity for four birds (filled circles representing the bird highlighted to the left). D, HVC_(RA) neuron recorded during singing under normal conditions and 12 min after local infusion of gabazine (0.05 mm). E, Another intracellular recording of an HVC_(RA) neuron after local gabazine infusion (0.05 mm) aligned to singing behavior (example at top, 6 spike trains from that neuron shown below).

Figure 8.

Pauses in interneuron firing can shape HVC_(RA) neuron output. A, Stimulation of the HVC-RA fiber tract (50 μA) recruits disynaptic inhibition in the HVC_(RA) neuron network. Left, Simultaneous recording of an HVC_(RA) neuron (average trace = red) and an interneuron (average trace = blue). The peak of monosynaptic excitation evoked in the interneuron (dashed blue line) occurs before the onset of the disynaptic inhibitory current in the HVC_(RA) neuron. Right, An HVC_(RA) neuron in a similar configuration. The disynaptic inhibitory current is abolished at the chloride reversal potential (bottom trace), revealing a small EPSC (stimulation current: 70 μA). B, Population data from a set of HVC_(RA) neurons recorded in this fashion, with peak IPSC and EPSC strengths quantified at −46 mV. 3 of 9 cases exhibited evidence for an excitatory current (average EPSC amplitude: 15.7 ± 9.0 pA). C, D, In vitro recording of an interneuron (C) and an HVC_(RA) neuron (D) while stimulating the HVC-RA fiber tract with an extracellular electrode at 60 μA (stimulation artifacts clipped for clarity) to evoke interneuron spiking similar to the patterns seen during singing. Gaps in firing are indicated by gray shaded regions. Mean IPSP strength in the HVC_(RA) neuron (measured at location of red star) was 7.8 ± 2.3 mV (n = 11 cells). E–G, Configuration as above showing spiking responses in the HVC_(RA) neuron to a 35 ms somatic current pulse during an inhibitory gap (E) or ongoing inhibition (F, G). H, I, Quantification of the HVC_(RA) spiking responses of individual neurons to various current pulses during an inhibitory gap (H) and ongoing inhibition (I). Each line represents the spiking response of a single neuron to a current pulse normalized to the maximal response from that cell across conditions. All-or-none bursting was only observed during pauses of inhibition. The L-type calcium agonist BAY K8644 was bath-applied to facilitate HVC_(RA) neuron burst firing from somatic current injection in C–I.