Intrinsic properties of avian interaural level difference sound localizing neurons

Abstract

Intrinsic properties of neurons are one major determinant for how neurons respond to their synaptic inputs and shape their outputs in neural circuits. Here, we studied the intrinsic properties of neurons in the chicken posterior portion of the dorsal nucleus of the lateral lemniscus (LLDp), the first interaural level difference (ILD) encoder of the avian auditory pathway. Using whole-cell recordings in brain slices, we revealed that the LLDp is composed of a heterogeneous neuron population based on their firing patterns. LLDp neurons were broadly classified as either phasic or tonic firing neurons, with further classification applied to tonic firing neurons, such as regular (most dominant, n=82 out of 125 cells, 65.6%), pauser, or adaptive firing. Neurons with different firing patterns were distributed about evenly across the dorsoventral as well as mediolateral axis of LLDp. Phasic firing neurons were of faster membrane time constant, and lower excitability than tonic firing neurons. The action potentials (APs) elicited at the current thresholds displayed significant differences in first spike latency, AP peak amplitude, half-width, and maximal rising and falling rates. Interestingly, for APs elicited at suprathreshold currents (400 pA above thresholds), some of the differences diminished while a few others emerged. Remarkably, most parameters of the APs at thresholds were significantly different from those of APs at suprathresholds. Combined with our previous study (Curry and Lu, 2016), the results lend support to the two-cell type model for ILD coding in the avian system.

1. Introduction

Computation of sound localization information using interaural level difference (ILD) as a cue in mammals takes place first in the lateral superior olive (LSO). In birds, the first nucleus that encodes ILD is the posterior portion of the dorsal nucleus of the lateral lemniscus (LLDp, formerly known as VLVp, the nucleus ventralis lemnisci lateralis pars posterior) (Mogdans and Knudsen, 1994). LLDp neurons receive excitatory input from the contralateral cochlear nucleus angularis (NA) (Conlee and Parks, 1986; Takahashi and Konishi, 1988), and receive inhibitory input primarily from the other LLDp, which is driven by excitatory input originating from the ipsilateral NA (Manley et al., 1988; Takahashi and Keller, 1992). A number of in vivo studies have demonstrated the roles of LLDp neurons in ILD processing (Adolphs, 1993; Sato et al., 2010; Takahashi and Keller, 1992). Our previous study has reported the mechanisms of synaptic inhibition underlying ILD coding in the LLDp, and proposed a two-cell type model (Curry and Lu, 2016). However, the report on the intrinsic properties of LLDp neurons is lacking, impeding our understanding of how these ILD coding neurons may respond to their synaptic inputs in ILD processing.

Knowledge of the mammalian system provides insight to potential strategies in the construction of intrinsic properties that may be favorable for ILD processing. For example, the principal ILD encoding LSO neurons exhibit a highly regular firing pattern in vivo (Tsuchitani, 1982; but also see Franken et al., 2018) and a tonic firing pattern in response to prolonged depolarizing current injections in vitro (Adam et al., 1999; Wu and Kelly, 1995), allowing for modulation of spike firing rate to represent the interaction of synaptic excitation and inhibition. This differs from the intrinsic properties of the neurons in the medial nucleus of trapezoid body (MNTB), which exhibit a characteristic single action potential (AP) in response to sustained depolarization, and maintain high temporal fidelity (Banks and Smith, 1992; Dodson et al., 2002; Klug and Trussell, 2006). Consequently, the MNTB provides fast and temporally precise and reliable inhibition to various binaural neurons including those in the LSO (reviewed by Grothe, 2003; Grothe et al., 2010). Based on the known circuitry, neurons in the LLDp on one side need to perform the tasks that are performed by both LSO in order to encode ILDs (Mogdans and Knudsen, 1994) and MNTB in order to provide direct inhibitory input to the LLDp on the other side of the brain (Curry and Lu, 2016; Takahashi and Keller, 1992). These two tasks are unlikely to be performed by the same population of cells displaying similar neuronal properties. Therefore, we hypothesized that LLDp neurons are heterogeneous in intrinsic properties, such that one population would exhibit a highly regular firing pattern, like principle LSO neurons, which could encode increases in sound levels by an increase in firing frequency, and another population would exhibit a precise phasic firing pattern, like MNTB neurons, which could provide temporally modulated inhibition to the opposite LLDp. To test this hypothesis, we investigated the intrinsic properties of LLDp neurons using whole-cell patch clamp to compare the passive properties (e.g. input resistance and membrane time constant) and active properties (e.g. AP parameters) to determine if LLDp neurons could be categorized into distinct cell-types that may relate to specific circuit function.

2. Results

In acutely obtained coronal brainstem slices, the LLD was recognized as a heavily myelinated ovoid region medial and ventral to the semilunar nucleus, a landmark structure with distinct shape and contrast (Fig. 1A). After biocytin staining processing, both the LLD and the semilunar nucleus showed distinct borders (Fig. 1B), clearly separated from surrounding tissue. The border of the LLDp was distinguished by a medial lamina, such that cells were sampled from the lateral portion of the LLD, which corresponds to the LLDp (Heil and Scheich, 1986). A total of 125 LLDp cells were recorded under whole-cell current clamp configuration. These cells were sampled across the whole LLDp nucleus in terms of location across the dorsoventral and the mediolateral dimensions, without intentional bias.

Figure 1. Identification of the LLD in brain slices, and the current clamp protocol.

(A) In a freshly obtained coronal brainstem slice, the LLD is clearly recognized as a dark pear-shaped nucleus ventral and medial to another distinct nucleus, the semilunar nucleus (SLu). (B) After biocytin staining processing, both the LLD and the SLu can be encircled with distinct border. (C) A current clamp protocol was used throughout the study to give prolonged (100 ms in duration) somatic current injections (typically −300 pA up to 2500 pA, increment of 100 pA) and to record voltage responses. The traces color in red represent the voltage response at the current threshold (300 pA for this cell).

2.1. Classification of LLDp neurons based on intrinsic firing patterns

Given that electrical or chemical stimulation of the contralateral LLDp gives rise to evoked inhibitory postsynaptic currents in the presence of antagonists for ionotropic glutamate receptors, two principal cell types could be the most advantageous model for encoding ILD in chicken (Curry and Lu, 2016). However, the contralateral stimulation experiments could not discern whether or not the LLDp is comprised of multiple principal cell types. LLDp neurons were therefore subjected to current injections under whole-cell current clamp to identify the intrinsic neuronal firing patterns. The regular recording protocol consisted of prolonged (100 ms in duration) current pulses with stepwise increasing intensity, ranging from typically −0.3 nA to 2.5 nA, with an increment of 100 pA (Fig. 1C). Three main patterns of action potential (AP) firing were observed and categorized as follows: phasic, tonic 1 (T1), and tonic 2 (T2) (Fig. 2). Phasic firing neurons were defined as neurons firing only one or few APs at the beginning of suprathreshold current injections (Fig. 2A). T1 firing neurons were defined as firing multiple APs in a regular fashion (with relatively constant inter-spike-interval) throughout the duration of suprathreshold current injections (Fig. 2B). T2 neurons also displayed multiple APs throughout the duration of suprathreshold current injections, with some irregularity in the pattern, such as a distinct gap between the first and second AP (termed tonic 2 pauser, T2p), or an adaptive decrease in number of APs and changes in AP waveform with increasing intensity of current injection (termed tonic 2 adapter, T2a) (Fig. 2C). Most of the recorded LLDp neurons displayed a T1 firing pattern (65.6%, n=82 out of 125 neurons), followed by phasic (n=18, 14.4%), T2p (n=13, 10.4%), and T2a (n=12, 9.6%) neurons (Fig. 2D).

Figure 2. LLDp neurons exhibit multiple firing patterns, with the regular tonic firing pattern being dominant.

(A–C) Representative membrane potential traces (top 3 rows) of 4 LLDp neurons with distinct firing patterns in response to depolarizing current injections. The action potentials (APs) elicited at the current thresholds vary dramatically in their onset timing, waveform, and number. The APs elicited at suprathreshold currents are used to classify the firing pattern. In response to suprathreshold current injections, phasic firing neurons displayed one or few APs at the onset of the current injections. Tonic 1 (T1) firing neurons displayed regular repetitive firing (with relatively constant inter-spike-interval) and constant AP amplitude. Tonic 2 (T2) neurons displayed irregular repetitive firing (with varying inter-spike-interval and/or varying amplitude), which were further sub-grouped as tonic 2 pauser (T2p, with an obvious gap between the first AP and subsequent APs) and tonic 2 adapter (T2a, with an adaptive substantial decrease in the number and amplitude of APs during the prolonged current injections). (D) Most recorded LLDp neurons displayed T1 firing (n = 82), and the rest displayed phasic firing (n = 18) or T2 firing (n = 25, with 13 T2p and 12 T2a cells). (E) The number of neurons recorded from the most caudal (slice #1) to most rostral (slice #5) brain slice, grouped by firing pattern. Neurons did not appear to cluster by firing pattern along the dorsoventral axis (F) or mediolateral axis (G).

Because of the small population, T2 cells were treated as a single group in the analysis in this figure for the distribution of neurons in LLDp (Fig. 1E-G). The majority of recorded cells were sampled from the third collected brain slice, with slice 1 being the most caudal and slice 5 being the most rostral (Fig. 2E). In general, no strong association between firing pattern type and neuron location was observed. Both tonic and phasic firing neurons were observed across the rostrocaudal axis. In addition, neurons with different firing patterns were relatively evenly distributed across the dorsoventral axis and mediolateral axis (Fig. 2F, G). Therefore, phasic firing neurons are dispersed throughout the LLDp and at a relatively lower density of distribution than tonic firing neurons. No clustering of cells was observed based on firing pattern.

2.2. Basic intrinsic properties of LLDp neurons

The basic intrinsic neuronal properties are critical determinants shaping up the neuron’s output in response to synaptic inputs. We analyzed five important membrane properties of LLDp neurons based on their firing patterns (Fig. 3A). Significant differences were detected in resting membrane potentials (RMPs) (Fig. 3B; ANOVA p = 0.0021, F(3, 121) = 5.174). The RMPs of phasic firing neurons (−63.6 ± 2.2 mV, n = 18) did not differ from those of tonic firing neurons. Among the tonic firing cells, the RMPs of T2a neurons (−57.1 ± 2.6 mV, n = 12) were more depolarized than T1 neurons (−65.8 ± 0.7 mV, n = 82; Tukey’s test p = 0.0011) and T2p neurons (−65.9 ± 1.9 mV, n = 13; Tukey’s test p = 0.0161). No significant differences were detected in I_h sag (likely due to the presence of hyperpolarization-activated cation current) between any two types of firing patterns (Fig. 3C; P: 3.2 ± 1.3 mV; T1: 6.8 ± 0.7 mV; T2p: 2.7 ± 1.3 mV; T2a: 4.5 ± 1.6 mV). Similarly, there were no significant differences in input resistance (R_in) ( Fig. 3D; P: 109.9 ± 16.1 MΩ; 147.1 ± 8.3 MΩ; T2a: 112.6 ± 14.6 MΩ; T2p: 153.8 ± 22.9 MΩ). Significant differences were detected in membrane τ among neurons with different firing patterns (Fig. 3E; ANOVA p = 0.0105, F(3, 121) = 3.909). Neurons with phasic firing pattern displayed faster membrane τ (6.39 ± 1.18 ms, n = 18) than T1 neurons (11.60 ± 0.70 ms, n = 82; Tukey’s test p = 0.0049), while no differences were detected among other groups. The fast membrane τ of phasic firing neurons is consistent with their potential timing coding function of this type of cells in the auditory system. Conversely, the membrane capacitance (C_m) of phasic firing neurons (57.6 ± 7.9 pF, n = 18) was significantly lower than T1 cells (82.1 ± 4.0 pF, n = 82; Tukey’s test p = 0.0488) (Fig. 3F). Supporting this observation, a biocytin-stained tonic firing cell showed more extensive dendritic arborization than a phasic firing neuron.

Figure 3. Basic neuronal properties of LLDp neurons.

(A) Sample traces indicating the measurements of membrane time constant (τ, top trace, red line indicates single exponential fitting to the voltage trace in response to a 100-pA hyperpolarizing current injection), and the size of the I_h sag (middle trace, double arrow). (B) The resting membrane potentials (RMPs) of T2a neurons were significantly more positive than those of T1 and T2p neurons. (C, D) No significant differences in I_h sag and input resistance (R_in) were detected among neurons with different firing patterns. (E) Neurons with phasic firing pattern displayed faster membrane τ than neurons with T1 firing pattern, while no differences were detected among other groups. (F) Membrane capacitance (C_m), calculated based on R_in and membrane τ, showed significantly larger values for T1 neurons than phasic neurons. Consistently, associated biocytin staining in the sampled cells showed that a tonic firing neuron (lower image) had more extensive dendritic arborization than the phasic firing neuron (upper image). For this and subsequent figures, mean ± SEM are shown, with the dots exhibiting individual data values. Cell number n = 18, 82, 13, and 12 for phasic, T1, T2p, and T2a neurons, respectively. *, **, and ***: ANOVA post hoc Tukey’s test p < 0.05, 0.01, and 0.001, respectively.

To characterize the input-output functions, we plotted the number of APs against the current intensity relative to the current thresholds (Fig. 4A). By definition, phasic firing neurons displayed an almost flat input-output function, due to the fact that they fired only one or a few APs in response to prolonged suprathreshold current injections, regardless of the current intensity. The tonic firing neurons, in contrast, increased the number of APs with increasing current intensities, reaching a plateau at currents of about 1 nA above the thresholds. Each of the tonic firing types also showed a slight reduction at the highest current intensities. This seemingly non-monotonic response or spike firing adaptation could be caused by depolarization block. In order to obtain voltage/current (V/I) curves, we plotted the averaged subthreshold membrane potentials against the amplitudes of the injected currents (Fig. 4B). Once the voltage response reached threshold and fired spikes, we did not obtain the membrane potential for the V/I plot. Because the current thresholds were different for different neurons, the number of data points used to calculate the averages of membrane potentials at each current injection gradually decreased with increasing current amplitudes (inset in Fig. 4B). Unexpectedly, the V/I curves were in parallel among neurons with different firing patterns (Fig. 4B). Typically, phasic firing neurons in the auditory brainstem possess stronger outward rectification, because of the presence of strong voltage-gated K⁺ (K_V) conductances (e.g. Dodson et al., 2002; Johnston et al., 2010). It is likely that the phasic firing neurons in LLDp do not express as strong K_V channels as other auditory timing coding neurons.

Figure 4. Input-output functions (number of APs against current intensity), and V/I curves for LLDp neurons with different firing patterns.

(A) Neurons with non-phasic firing pattern increased their firing rates in response to suprathreshold current injections with increasing amplitude. In contrast, neurons with phasic firing pattern, by definition, maintained a relatively flat input-output function, with minimal increase in firing rates over increasing current amplitudes. (B) V/I curves of neurons with different firing patterns are approximately parallel to each other, suggesting similar membrane rectification. The inset shows the number of data points used in the V/I curves for each different firing pattern in response to current injections ranging from zero to 1.0 nA (with a step of 0.1 nA). The number of data points reduced with increasing amplitude of current injections due to varying current thresholds for different neurons.

2.3. Properties of APs elicited at the current thresholds

Traditionally, the characterization of intrinsic firing properties is performed by analyzing the parameters of the first AP acquired at the current threshold (reviewed by Bean, 2007). Thus the following parameters of the APs elicited at the current thresholds were obtained in order to make comparisons among LLDp neurons with different firing patterns: voltage threshold, first spike latency, AP peak amplitude, AP height, AP half-width, the maximal AP rising rate, and the maximal AP falling rate (Fig. 5A; Table 1). The current thresholds (an indicator for cellular excitability) varied significantly among neurons with different firing patterns (Fig. 5B; ANOVA p < 0.0001, F(3, 121) = 9.650). T1 neurons had lower threshold currents (350 ± 26 pA, n = 82) than phasic firing neurons (603 ± 121, n = 18; Tukey’s test p = 0.0130), indicating that T1 neurons are generally more excitable. Among the tonic firing cells, T2p neurons (785 ± 101 pA, n = 13) were less excitable than T1 (p < 0.0001) and T2a neurons (312 ± 112 pA, n = 12; Tukey’s test p = 0.0015). The numerical values of other AP parameters are shown in Table 1, and therefore only statistical results are reported here. No significant differences were detected in voltage thresholds between any two groups of neurons although a significant difference was detected in ANOVA analysis (Fig. 5C; ANOVA p = 0.0476, F(3, 121) = 2.719), reflecting similar properties of voltage-gated Na⁺ (Na_V) channels underlying the voltage threshold in LLDp neurons. The first spike latency was significantly different among neurons with different firing patterns (Fig. 5D; ANOVA p = 0.0061, F(3, 121) = 4.344). Because phasic firing neurons generated their spikes at the onset of the current injections and the first spikes of tonic firing neurons varied their timing dramatically, the first spike latencies of phasic firing neurons were significantly shorter than those of T1 neurons (Tukey’s test p = 0.0053). The AP peak amplitude was significantly different among different groups (Fig. 5E; ANOVA p = 0.0135, F(3, 121) = 3.713). T2p neurons displayed higher peak amplitude than T1 (Tukey’s test p = 0.0367) and T2a neurons (Tukey’s test p = 0.0088). This feature did not consequently generate higher AP height in T2a neurons; the AP height was similar among neurons with different firing patterns (Fig. 5F; ANOVA p = 0.2090, F(3, 121) = 1.535). This is possibly due to the fact that T2p neurons had the most positive voltage thresholds (Fig. 5C). Significant differences were detected for AP half-width (Fig. 5G; ANOVA p = 0.00002, F(3, 121 = 7.237). The differences existed between phasic firing neurons (0.88 ± 0.19 ms, n = 18) and T1 neurons (0.50 ± 0.02 ms, n = 82; Tukey’s test p = 0.0001) and T2p neurons (0.43 ± 0.03, n= 13; Tukey’s p = 0.0018). Significant differences were detected in the AP maximal rising rates (Fig. 5H; ANOVA p = 0.0324, F(3, 121) = 3.022), and the post hoc analysis revealed differences between T2p neurons (225.5 ± 23.5 V/s, n = 13) and T2a neurons (139.7 ± 19.6 V/s, n = 12; Tukey’s test p = 0.0477). For the AP maximal falling rates (Fig. 5I; ANOVA p = 0.0006, F(3, 121) = 6.139), significant differences were seen between T2p neurons (−180.5 ± 15.4 V/s) and each of the other three different firing patterns (phasic: −107.2 ± 14.5 V/s; T1: −128.3 ± 6.1 V/s; T2a: −96.3 ± 11.5 Vs; Tukey’s test p = 0.0021, 0.0099, and 0.0012, respectively). Overall, T2a neurons seem to have APs with fast rising and falling phases.

Figure 5. Properties of APs elicited at the current thresholds.

(A) Sample traces of one AP and its first derivative indicating the measurements (voltage threshold, first spike latency, AP peak amplitude, height, half-width, maximal rising rate, and maximal falling rate) reported in this and the next figure. (B) The current thresholds vary with different firing patterns. T1 neurons were more excitable than phasic and T2p neurons, and T2a neurons were more excitable than T2p neurons. (C) No significant differences were detected in voltage thresholds among neurons with different firing patterns. (D) Firing mostly at the onset of the current injections, phasic firing neurons had significantly shorter first spike latencies than T1 neurons. (E) T2p neurons displayed higher AP peak amplitude than T1 and T2a neurons. (F) No significant differences were detected in AP height. (G) Interestingly, the AP half-widths of phasic neurons were wider than those of T1 and T2p neurons. (H-I) The maximal rising rates of T2p neurons were higher than those of T2a neurons, whereas the maximal falling rates of T2p neurons were higher than those of the other neurons with different firing patterns.

Table 1.

Comparisons of parameters between the APs elicited at the threshold current and at a suprathreshold level (400 pA above the threshold current).

Parameters	#cells	at threshold	suprathreshold	t (n-1)	t-test p
V threshold (mV)
P	18	−34.9 ± 2.5	−29.9 ± 2.8	3.226	0.0050**
T1	82	−31.0 ± 0.8	−28.1 ± 1.3	3.056	0.0030**
T2p	13	−28.4 ± 2.0	−32.7 ± 2.0	2.345	0.0371*
T2a	12	−35.2 ± 1.9	−33.5 ± 2.1	1.045	0.3183
Latency (ms)
P	18	9.4 ± 2.8	4.2 ± 1.2	3.090	0.0067**
T1	82	37.6 ± 4.0	10.0 ± 1.7	9.446	<0.0001***
T2p	13	28.8 ± 8.9	11.9 ± 5.4	3.216	0.0074**
T2a	12	20.3 ± 6.5	4.5 ± 1.4	3.002	0.0120*
Peak (mV)
P	18	11.4 ± 3.3	18.9 ± 3.1	5.324	<0.0001***
T1	82	11.5 ± 1.4	18.0 ± 1.4	7.861	<0.0001***
T2p	13	21.6 ± 3.1	26.2 ±3.1	5.122	0.0003***
T2a	12	5.5 ± 2.6	12.9 ± 2.8	6.810	<0.0001***
Height (mV)
P	18	46.5 ± 3.7	48.8 ± 3.8	1.719	0.1038
T1	82	42.3 ± 1.6	46.1 ± 1.7	4.917	<0.0001***
T2p	13	50.0 ± 3.5	58.9 ± 2.6	3.989	0.0018**
T2a	12	40.7 ± 3.1	46.4 ± 3.2	4.904	0.0005***
Half-width (ms)
P	18	0.88 ± 0.19	0.80 ± 0.13	1.199	0.2471
T1	82	0.50 ± 0.02	0.51 ± 0.02	0.916	0.3626
T2p	13	0.43 ± 0.03	0.44 ± 0.03	0.4751	0.6433
T2a	12	0.62 ± 0.06	0.63 ± 0.05	0.5767	0.5758
Rise rate (V/s)
P	18	149.5 ± 22.5	166.3 ± 22.9	4.149	0.0007***
T1	82	177.9 ± 8.8	201.6 ± 10.0	6.479	<0.0001***
T2p	13	225.5 ± 23.5	272.3 ± 19.3	3.752	0.0028**
T2a	12	139.7 ± 19.6	166.7 ± 21.4	4.614	0.0007***
Fall rate (−V/s)
P	18	107.2 ± 14.5	113.1 ± 13.4	2.492	0.0233*
T1	82	128.3 ± 6.1	135.9 ± 6.9	4.704	<0.0001***
T2p	13	180.5 ± 15.4	200.9 ± 12.8	3.214	0.0074**
T2a	12	96.3 ± 11.5	102.8 ± 11.8	2.925	0.0138*

Paired t-test

*, **, and ***: p < 0.05, 0.01, and 0.001, respectively.

2.4. Properties of APs elicited at suprathreshold currents

Under native external stimuli, neurons are rarely activated just at the threshold, so we analyzed the APs at a suprathreshold current level (400 pA above the current threshold) (Fig. 6A). The same X- and Y-axis range values are used as those in Figure 5, and the changes in significance level compared to Figure 5 are indicated by color (Fig. 6; black for persisting differences, and red for diminishing or newly emerging differences). Similar to APs at threshold currents, APs at suprathreshold currents displayed no significant differences in voltage thresholds among neurons with different firing patterns (Fig. 6B; ANOVA p = 0.2632, F(3, 121) = 1.344). For first spike latency, the significant differences between phasic firing and T1 neurons at threshold currents diminished (Fig. 6C; ANOVA p = 0.2538, F(3, 121) = 1.375), due to the fact that with increasing current injections T1 neurons initiated their firing more and more quickly. Significant differences in AP peak amplitude also largely diminished (Fig. 6D; ANOVA p = 0.0630, F(3, 121) = 2.496), with the difference between T2p neurons (26.2 ± 3.1 mV, n = 13) and T2a neurons (12.9 ± 2.8 mV, n = 12) remained (Tukey’s p = 0.0436). Interestingly, a significant difference in AP height emerged (Fig. 6E; ANOVA p = 0.0339, F(3, 121) = 2.986), between T1 neurons (46.1 ± 1.7 mV, n = 82) and T2p neurons (58.9 ± 2.6 mV, n = 13; Tukey’s test p = 0.0192). The large AP heights of T2p neurons are consistent with their high AP peak amplitudes. The AP half-width displayed significant differences among neurons with different firing patterns (Fig. 6F; ANOVA p < 0.0001, F(3, 121) = 8.505). Phasic firing neurons (0.80 ± 0.13 ms) had longer half-width than T1 (0.51 ± 0.02 ms; Tukey’s test p < 0.0001) and T2 p neurons (0.44 ± 0.03 ms; Tukey’s test p = 0.0005), similar to those for APs at threshold currents. Newly emerged significant differences were seen in AP maximal rising rates (Fig. 6G; ANOVA p = 0.0061, F(f3, 121) = 4.343). The AP maximal rising rates of T2p neurons (272.3 ± 19.3 V/s) were not only higher than those of T2a neurons (166.7 ± 21.4 V/s; Tukey’s test p = 0.0179), but also higher than those of phasic firing neurons (166.3 ± 22.9 V/s; Tukey’s test p = 0.0070) and T1 neurons (201.6 ± 10.0 V/s; Tukey’s test p = 0.0414). The differences in AP maximal falling rates remained (ANOVA p = 0.0001, F(3, 121) = 7.432), between T2p neurons (−200.9 ± 12.8 V/s) and each of the other firing patterns, including phasic firing neurons (−113.1 ± 13.4 V/s; Tukey’s test p = 0.0004), T1 neurons (−135.9 ± 6.9 V/s; Tukey’s test p = 0.0019), and T2a neurons (−102.8 ± 11.8 V/s; Tukey’s test p = 0.0003).

Figure 6. Properties of APs elicited at suprathreshold currents (400 pA above the current thresholds).

(A) Sample traces of one AP elicited at the current threshold (blue) and another AP at 400 pA above the current threshold (red). For readily comparisons, the X- and Y-axis for the next plots are of the same range of values for the same parameter as in Figure 5. Additionally, when there are changes in statistical significances (from being significant in Fig. 5 to non-significant in Fig. 6, or vice versa), the symbols are highlighted with red color. (B) No significant differences were detected in voltage thresholds among neurons with different firing patterns. (C) The differences in first spike latencies for APs elicited at current thresholds between phasic and T1 neurons diminished, because most neurons fired the first AP at the onset of the current injection in response to suprathreshold current injections, regardless of firing patterns. (D) T2p neurons displayed higher AP peak amplitude than T2a neurons. (E) T2p neurons displayed higher AP height than T1 neurons. (F) Consistent with APs at threshold, the AP half-widths of phasic neurons were wider than those of T1 and T2p neurons. (G, H) Both the maximal rising and falling rates of T2p neurons were higher than those of neurons with different firing patterns.

When compared between APs at threshold and APs at suprathreshold, most AP parameters showed significant differences, except for AP half-width (Table 1). The voltage thresholds were significantly more positive at suprathreshold currents for phasic and T1 firing neurons. The first spike latency became significantly shorter at suprathresholds for all four different firing patterns. The AP peak amplitudes were higher at suprathresholds, so were the AP heights except for phasic firing neurons. The APs at suprathresholds had both higher maximal rising rates and higher absolute maximal falling rates at suprathresholds. Therefore, APs at suprathreshold currents are different in their parameters when compared to APs at threshold, pointing to the necessity of analyzing them in order to understand neural functions under natural suprathreshold stimulation.

2.5. Further classification of T1 firing LLDp neurons based on I_h sag

The electrophysiological properties of T1 neurons were further analyzed, as a large degree of variation in a subset of subthreshold properties was seen within this group, which could further differentiate classes of neurons. Notably, many T1 neurons displayed an I_h sag (55 out of 82, 67%). In other auditory regions, the presence of I_h has been linked to temporal aspects of auditory processing, such as synchronized AP timing (Kim and Holt, 2013). We compared the neuronal properties between T1 neurons with a I_h sag and those without (Table 2). Most neuronal properties displayed similar values between these two subgroups, but a few important parameters showed differences. Neurons without I_h sags had faster membrane τ, and they were less excitable (higher current thresholds) than neurons with I_h sags. For APs elicited at the current threshold, neurons without I_h sags had higher AP peak, as well as longer half-width than neurons with I_h sags. For APs elicited at the suprathreshold currents, the significant difference in AP peak remained, whereas the difference in AP half-width diminished. Overall, subclasses of T1 neurons may exist based on these differences especially on current threshold (indication of cellular excitability), one critical parameter that differentiates neurons with different firing patterns.

Table 2.

Comparisons of neuronal properties between T1 neurons with I_h sags and T1 neurons without I_h sags.

Parameters	with sag (n = 55)	without sag (n = 27)	t (80)	t-test p
RMP (mV)	−65.9 ± 0.9	−65.6 ± 1.4	1.665	0.8682
Rin (MΩ)	157.9 ± 9.0	125.1 ± 16.7	1.893	0.0619
Membrane τ (ms)	12.6 ± 0.7	9.5 ± 1.4	2.176	0.0325*
Cm (pF)	85.4 ± 5.1	75.2 ± 6.3	1.195	0.2355
I threshold (pA)	306 ± 24	439 ± 60	2.445	0.0167*
AP at threshold
V threshold (mV)	−31.7 ± 0.9	−30.0 ± 1.8	1.149	0.2540
Latency (ms)	40.7 ± 5.2	31.3 ± 5.7	1.114	0.2685
Peak (mV)	9.3 ± 1.8	15.9 ± 1.8	2.238	0.0280*
Height (mV)	41.0 ± 1.9	45.5 ± 2.7	1.366	0.1758
Half-width (ms)	0.47 ± 0.02	0.56 ± 0.03	2.301	0.0240*
Rise rate (V/s)	179.3 ± 10.3	175.1 ± 16.8	0.2254	0.8222
Fall rate (−V/s)	126.9 ± 7.5	131.0 ± 10.7	0.3149	0.7537
AP at suprathreshold
V threshold (mV)	−29.0 ± 1.5	−26.3 ± 2.3	0.9720	0.3340
Latency (ms)	11.4 ± 2.5	7.2 ± 1.6	1.130	0.2618
Peak (mV)	15.6 ± 1.8	23.1 ± 2.1	2.560	0.0124*
Height (mV)	44.5 ± 1.9	49.4 ± 3.2	1.368	0.1753
Half-width (ms)	0.48 ± 0.02	0.55 ± 0.03	1.879	0.0639
Rise rate (V/s)	202.6 ± 11.9	199.5 ± 18.7	0.1462	0.8841
Fall rate (−V/s)	133.6 ± 8.4	140.5 ± 12.5	0.4643	0.6437

Un-paired t-test

*, **, and ***: p < 0.05, 0.01, and 0.001, respectively.

It is worth providing the justification for further dividing T2 neurons into T2p and T2a subclasses in data analyses, but leaving the most numerous T1 neurons as a single group. The standard criteria used to classify neurons are based on the spike firing patterns in response to suprathreshold somatic current injections. This is how we broadly defined phasic vs tonic firing neurons. Although the total number of T2 neurons is small (n=25 out of 125 cells), there existed two very different firing patterns, one with a “pause” after the initial spike(s) and the other being clearly “adapting”, and hence T2p and T2a subclasses for data analyses. In contrast, the spike firing patterns of all T1 neurons were similar, so they remained as a single group for data analyses. The T1 neurons did display differences in subthreshold responses (such as I_h sag). Therefore, we performed further analyses of T1 neurons based on the presence and absence of I_h sags, even though the T1 neurons did not differ within the group based on their spike firing patterns. Besides its involvement in pacemaking activities, I_h current participates in regulation of the RMP, counteracting with the low threshold K_V current. This is particularly important for binaural coincidence detector neurons in the medial superior olive (Khurana et al., 2011). While the T1 neurons with I_h sags had about the same RMP as the T1 neurons without I_h sags, the latter group tended to have a larger variation in RMP (Table 2). In addition, there exists differential expressions of I_h currents among different auditory brainstem nuclei that perform different functions (Leao et al., 2006), suggesting that the two subgroups of T1 neurons in the LLDp may play different functions in ILD coding.

3. Discussion

3.1. Heterogeneous LLDp neurons based on spike firing patterns

The results of this study reveal that the LLDp is composed of a heterogeneous neuronal population, which may allow for a segregation of ILD processing functions among distinct cell types. The majority of LLDp neurons display a tonic regular firing pattern (T1), however, there is evidence for different firing patterns, including phasic and tonic irregular firing types. Based on the sampled neurons, there does not appear to be a distinct spatial segregation of firing types in the LLDp, such that neurons with various firing patterns are found dispersed throughout the LLDp. For the passive neuronal properties, the most distinct difference among the LLDp neurons with different firing patterns is the membrane τ. Phasic firing LLDp neurons display faster membrane τ than T1 neurons, suggesting a potential temporal processing function for the phasic firing neurons (reviewed by Oertel, 1999; Trussell, 1999).

For active spike firing properties, as expected, the input-output function of phasic firing neurons was almost a flat curve, similar to many auditory brainstem timing-coding neurons. In contrast, tonic firing neurons displayed an almost monotonic relationship in their input-output functions, similar to auditory brainstem neurons encoding intensity and other aspects of sound features. The V/I curves, however, displayed somewhat surprising features. Compared to tonic firing neurons, phasic firing neurons typically possess strong K_V channels (reviewed by Gan and Kaczmarek, 1998; Johnston et al., 2010). However, the phasic firing neurons in LLDp did not exhibit a stronger outward rectification than the tonic firing neurons, suggesting little or no differences in K_V channel conductances among different firing patterns. Furthermore, based on the results that no significant differences were detected in voltage threshold and AP height (Fig. 5C, F), Na_V channel properties are unlikely to be different among LLDp neurons with different firing patterns. There may exist one exception. For T2p neurons, they tend to have the highest rising and falling AP rates, for both APs elicited at the threshold (Fig. 5H, I) and at suprathreshold (Fig. 6G, H). The differences were especially prominent for the suprathreshold APs because T2p neurons had significantly higher rising rates as well as falling rates than neurons with any other firing patterns, strongly suggesting higher conductance in both Na_V and K_V channels in T2p neurons. Being small in population (10.4% of recorded cells), T2p neurons behave like the delay neurons in mammalian LSO (Adam et al., 1999), which may be involved in functions other than ILD coding. Furthermore, the fact that most parameters were significantly different between APs at threshold and APs at suprathreshold (Table 1) points to the necessity examining APs at both levels in order to fully characterize firing properties and investigate functions of neurons.

Finally, because the T1 neurons constitute the majority (65.6%) of LLDp neurons, further classification of them into possible subtypes based on certain neuronal properties could be important in interpreting functional divergence in this multi-task nucleus. The two subpopulations of neurons based on the presence of an I_h sag displayed differences in membrane τ, current threshold, AP peak and half-width (Table 2), all of which are important factors determining neuronal output (reviewed by Bean, 2007).

3.2. Comparisons with the mammalian system

If LLDp neurons assume the combined functionality of LSO and MNTB neurons as proposed in Curry and Lu (2016), the intrinsic properties of LLDp neurons would tend to share similarity in cellular mechanisms with both nuclei in the mammalian system. LSO principal neurons exhibit in vivo a highly regular firing pattern, in which the first spike is short and has the lowest variability in time and subsequent APs occur at precisely timed intervals from the stimulus onset (Goldberg and Brown, 1969; Tsuchitani, 1982). When recorded in vitro in slice preparations, however, most LSO neurons are tonic firing (Wu and Kelly, 1995), and two types of neurons based on firing patterns exist, chopper neurons and delay neurons (Adam et al., 1999). Besides these multiple firing neurons, a less commonly encountered population of LSO neurons exhibit single spike phasic firing (Sanes, 1990; Barnes-Davies et al., 2004). Proportionally, more multiple firing neurons are seen when Ca_V1.3 is knocked out (Hirtz et al., 2011). In contrast, MNTB principal neurons exhibit a characteristic single AP in response to sustained depolarization, which can be attributed to the strong expression of low threshold K_V channels (Klug and Trussell, 2006). It is conceivable that in the avian ILD circuit, the subpopulation of LLDp neurons that are tonic firing may be more LSO-like, while the other subpopulation that are phasic firing more MNTB-like. However, even the phasic firing LLDp neurons are somehow different from the single spike MNTB neurons, as they did not show a strong outward rectification. Therefore, confirmation of such a loose match between the avian and the mammalian systems needs to be tested with functional systems level approaches. One of such approaches may be in vivo patch clamp recording, which can yield informative insights on the relationship between intrinsic neuronal properties and their functions under physiological conditions. For example, in contrast to the previous reports that LSO neurons mostly fired sustained spikes in response to prolonged inputs (Adam et al., 1999; Tsuchitani, 1982; Wu and Kelly, 1995), by using in vivo patch clamp recordings, Franken et al. (2018) reveals that most principal LSO neurons indeed assume a phasic intrinsic firing pattern. This discovery reinforces the idea that timing coding in LSO neurons may play a pivotal role in ILD processing.

3.3. Functional implications

While multiple cell types in LLDp are identified based on their firing patterns, however, the physiological roles of these different categories of neurons remain to be determined. Typically, in vitro firing patterns of auditory neurons are related to their functions. It is a hallmark for timing-coding neurons in the auditory system to fire in vitro a single or few APs at the onset of a prolonged current injection (reviewed by Oertel, 1999; Trussell, 1999), and for intensity-coding neurons to fire tonically (Dean et al., 2005). Such distinct firing patterns are well established in the AVCN during development in the first postnatal week (prior to hearing onset) (Wu and Oertel, 1987). Stellate cells in the AVCN with long-branched dendrites usually assume a tonic firing pattern, which encodes the strength of auditory inputs, while bushy cells with few short bushy-like dendrites have phasic firing pattern and encode precise timing information (reviewed by Oertel and Young, 2004). In the avian cochlear nucleus, timing-coding cochlear nucleus magnocellularis neurons in the middle and high frequency regions have phasic firing pattern (Hong et al., 2016; Reyes et al., 1994) while intensity-coding or multi-functional NA neurons mostly have tonic firing (Fukui and Ohmori, 2003; Shi and Lu, 2017; Soares et al., 2002). Therefore, the differences in firing patterns in the LLDp could represent a separation of function, such that one cell type (tonic firing neurons) primarily encodes the ILD, while the other cell type (phasic firing neurons) is responsible for sending the reciprocal inhibition to the other LLDp. However, LLDp neurons receive input from all four morphological types of NA neurons (Soares and Carr, 2001), three of which exhibit tonic firing (vertical, radiate, planar) and one of which exhibits phasic firing (stubby) (Soares et al., 2002), and therefore intrinsic and morphological differences in LLDp neurons may represent different aspects of auditory processing outside of direct ILD encoding.

The functionality of LLDp neurons with different intrinsic firing patterns has to be also considered in the context of their neurotransmitter phenotypes. GABA is one of the predominant neurotransmitters in the LLDp (Curry and Lu, 2016). The current data cannot determine whether the GABAergic neurons are tonic or phasic firing neurons. It is worth nothing that while GABAergic transmission dominates the reciprocal interactions between the two LLDp nuclei, functional glycinergic receptors with faster kinetics do exist almost as strongly as GABA_A receptors (Curry and Lu, 2016), consistent with the common presence of the glycinergic signaling in the avian auditory brainstem (Code and Rubel, 1989; Fischl and Burger, 2014; Fischl et al., 2014). It would be extremely interesting for the future work to determine whether the phasic firing neurons in the LLDp are glycinergic. If so, the phasic neurons may play the equivalent roles as MNTB neurons in the mammalian system. If not, then the fast inhibitory inputs to LLDp neurons that is necessary for the ILD processing may come from sources outside of the LLDp. Based on intrinsic neuronal properties, the LLDp is composed of a heterogeneous population of cells dispersed in different areas of the nucleus without distinct clustering, which may suggest a mixed distribution of neurons encoding ILD and neurons providing reciprocal inhibition, or a mixed distribution of neurons specialized for distinct ILD processing tasks. A combined approach in which synaptic and intrinsic physiological experiments are accompanied by determination of neurotransmitter phenotypes at the single cell level resolution seems imperative for future investigations, in order to establish the functional ILD circuit in the avian system.

4. Experimental procedure

4.1. Animals

All animal procedures were approved by the Institutional Animal Care and Use Committee at the Northeast Ohio Medical University (NEOMED) and were performed in accordance with the National Institutes of Health policies on animal use. Fertilized white leghorn chicken eggs were purchased from Charles River and incubated in our laboratory at NEOMED.

4.2. Slice preparation and in vitro whole-cell recordings

Brainstem slices (300 μm in thickness) were prepared from white leghorn chick embryos (E17–E19), as described previously (Curry and Lu, 2016). While the majority of LLDp studies have historically used barn owls as the avian model, the ILD circuitry appears to be conserved anatomically between avian species (Kubke and Carr, 2000; Wild et al., 2010). The chicken LLDp is able to encode ILD (Sato et al., 2010) and neurons in the chicken auditory midbrain are sensitive to both interaural time difference and ILD cues (Aralla et al., 2020), so we used the chick as our animal model. Developmentally equivalent to postnatal day 17–19 rodents, the selected age ranges (E17-E19) represent relative maturation of cellular properties in avian auditory brainstem neurons (Gao and Lu, 2008; Sanchez et al., 2010), although further refinements of neuronal properties take place after hatching (Carroll et al., 2018; Kuba et al., 2002). The warm (35 °C) artificial cerebrospinal fluid (ACSF) used for dissecting and slicing the brain tissue contained the following (in mM): 250 glycerol, 3 KCl, 1.2 KH₂PO₄, 20 NaHCO₃, 3 HEPES, 1.2 CaCl₂, 5 MgCl₂, and 10 glucose, pH 7.4 (when gassed with 95% O₂ and 5% CO₂). Slices were incubated in an interface chamber at 34–36°C for >1 h in normal ACSF containing the following (in mM): 130 NaCl, 26 NaHCO₃, 3 KCl, 3 CaCl₂, 1 MgCl₂, 1.25 NaH₂PO₄, and 10 glucose, pH 7.4. For recording, slices were transferred to a 0.5 ml chamber mounted on a Zeiss Axioskop 2 FS Plus microscope with a 40× water-immersion objective and infrared differential interference contrast optics. The chamber was continuously superfused with ACSF (2–5 mL/min) by gravity.

Patch pipettes were drawn on an Electrode Puller PP-830 (Narishige) to 1–2 μm tip diameter using borosilicate glass micropipettes (inner diameter of 0.84 mm, outer diameter 1.5 mm) (World Precision Instruments). The electrodes had resistances between 3 and 6 MΩ when filled with a solution containing the following (in mM): 125 K-gluconate, 5 Na-gluconate, 10 NaCl, 5 EGTA, 10 HEPES(K), 1 CaCl₂, 1 MgCl₂, 4 ATP-Mg, 0.48 GTP-Na, pH 7.3 (adjusted with KOH and osmolarity between 280 and 290 mOsm/L). The liquid junction potential was 13 mV, and data were corrected accordingly. Current-clamp experiments were performed with either an AxoPatch 200B or an AxoClamp 2B amplifier (Molecular Devices). Recordings were performed at 34–36°C. Current-clamp recordings were obtained at the resting membrane potential (RMP). Data were low-pass filtered at 3–10 kHz and digitized with a Data Acquisition Interface ITC-18 (Instrutech) at 50 kHz. To reveal cell morphology and location, biocytin (0.1%) was added to the internal solution for some experiments. After physiological recordings, brain slices were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer overnight and were processed with Vectastain ABC Elite Kit (Vector Labs), as described previously (Hamam and Kennedy, 2003). Recording protocols were written and run using the acquisition and analysis software AxoGraph X (AxoGraph Scientific). All chemicals were purchased from Sigma-Aldrich.

4.3. Data analysis

The RMP was determined immediately after whole-cell mode was established. The input resistance (R_in) was calculated with Ohm’s Law, by dividing the membrane potential change in response to −100 pA current injection by the current amplitude. The membrane time constant (τ) was obtained by fitting a single exponential curve to the same voltage trace as used for R_in calculation. Spontaneous spike activity of LLDp neurons in brain slice preparations was rarely noticed. All analyses of AP properties were based on spikes elicited by current injections. Multiple specific AP parameters were measured. The current threshold is defined as the minimal current amplitude that elicits APs, the voltage threshold as the membrane potential at the onset of the rising phase of the AP, first spike latency as the time period between the onset of the current injection and the occurrence of voltage threshold, height as the difference between the voltage threshold and the peak amplitude, half-width as the duration measured at the AP half height. The maximal rising and falling rates were measured as the most positive and negative values from the first derivative of the AP waveform, respectively.

Statistical analysis was performed, and graphs plotted using GraphPad Prism. Means and SEM are reported. Statistical differences were determined by a one-way ANOVA, or t-test. When significant differences were observed in an ANOVA, Tukey’s post hoc analysis was conducted to determine individual group differences. Significant differences were defined as a value of p < 0.05.

Highlights:

• LLDp neurons are heterogeneous in their intrinsic firing patterns.

• Tonic regular firing neurons dominate.

• Neurons with different firing patterns are about evenly distributed in LLDp.

• Action potentials differ in their parameters among different firing patterns.

• Phasic and tonic firing neurons may assume different functions in LLDp.

Abbreviations:

ACSF

artificial cerebrospinal fluid

AP

action potential

C_m

membrane capacitance

ILD

interaural level difference

LLDp

posterior portion of the dorsal nucleus of the lateral lemniscus

LSO

lateral superior olive

mGluR

metabotropic glutamate receptor

MNTB

medial nucleus of trapezoid body

NA

nucleus angularis

RMP

resting membrane potential

R_in

input resistance

V/I

voltage/current