Amplification of input differences by dynamic heterogeneity in the spiral ganglion

Abstract

A defining feature of type I primary auditory afferents that compose ∼95% of the spiral ganglion is their intrinsic electrophysiological heterogeneity. This diversity is evident both between and within unitary, rapid, and slow adaptation (UA, RA, and SA) classes indicative of specializations designed to shape sensory receptor input. But to what end? Our initial impulse is to expect the opposite: that auditory afferents fire uniformly to represent acoustic stimuli with accuracy and high fidelity. Yet this is clearly not the case. One explanation for this neural signaling strategy is to coordinate a system in which differences between input stimuli are amplified. If this is correct, then stimulus disparity enhancements within the primary afferents should be transmitted seamlessly into auditory processing pathways that utilize population coding for difference detection. Using sound localization as an example, one would expect to observe separately regulated differences in intensity level compared with timing or spectral cues within a graded tonotopic distribution. This possibility was evaluated by examining the neuromodulatory effects of cAMP on immature neurons with high excitability and slow membrane kinetics. We found that electrophysiological correlates of intensity and timing were indeed independently regulated and tonotopically distributed, depending on intracellular cAMP signaling level. These observations, therefore, are indicative of a system in which differences between signaling elements of individual stimulus attributes are systematically amplified according to auditory processing constraints. Thus, dynamic heterogeneity mediated by cAMP in the spiral ganglion has the potential to enhance the representations of stimulus input disparities transmitted into higher level difference detection circuitry.

NEW & NOTEWORTHY Can changes in intracellular second messenger signaling within primary auditory afferents shift our perception of sound? Results presented herein lead to this conclusion. We found that intracellular cAMP signaling level systematically altered the kinetics and excitability of primary auditory afferents, exemplifying how dynamic heterogeneity can enhance differences between electrophysiological correlates of timing and intensity.

INTRODUCTION

The electrophysiological signature of primary auditory afferents is complex because it results from intrinsic heterogeneity (1–6). Consequently, neurons that compose the spiral ganglion have the endogenous capacity to respond differentially to uniform input. At first glance, this seems to be an unlikely phenotype for auditory afferents that receive kinetically precise and highly sensitive representations of acoustic stimuli from sensory receptors (7, 8). Yet, based upon what is known about processing in other sensory afferents where relative, rather than absolute stimulus level is encoded to achieve improved detection, exemplified by contrast enhancement in the retina (9), it is probable that spiral ganglion neurons are designed to carry out similar transformations. Recent evidence supports this view as dynamic regulation of primary auditory afferents by voltage and intracellular cAMP signaling shapes response properties across the neuron population (10). This process in the spiral ganglion is powerful because the resultant primary afferent excitability and kinetics can be shaped differentially, ostensibly in response to efferent modulation (11–14) and acoustic input characteristics (15, 16). Moreover, sculpting neuronal firing patterns by modulating the underlying ion channels permits a high degree of plasticity in different cell classes and locations. This underscores that unlike neural mechanisms limited by local circuitry such as lateral inhibition (17), dynamic heterogeneity in the spiral ganglion can have an extensive impact due to the potential interactions between myriad types and combinations of efferent receptors, neurotransmitters, and endogenous ion channels (11–16, 18).

The mechanisms that regulate spiral ganglion neuron firing, while complex, are highly regulated. Increased cAMP signaling, for example, that acts through G_s α subunits (19) shortened the action potential durations at threshold levels that remained unchanged (10). This observation is significant because it demonstrates how individual electrophysiological parameters can be altered independently from related ones. Thus, this mechanism provides a way to enhance differences between restricted stimulus attributes of an input signal when compared across conditions. Although stimulus attributes of intensity and timing are interrelated (20, 21), this aspect of population coding is consistent with separate higher order auditory pathways that process binaural intensity level differences separately from timing and spectral cues (22–24).

The importance of an encoding process in the primary auditory afferents that can potentially enhance stimulus difference detection warranted a more comprehensive evaluation. Therefore, we investigated both increased and decreased levels of cAMP signaling on immature P6 spiral ganglion neurons that display overall slower kinetics and higher levels of excitability than mature neurons (1). We reasoned that this approach would magnify the faster cAMP-modulated kinetics to permit a higher resolution examination of cAMP effects tonotopically and between adaptation classes. The higher excitability of these neurons also allows us to more stringently test whether threshold responses that remained constant in mature neurons remained so in more excitable cells. Moreover, we expect that what we learn from this investigation will likely have relevance to adult, functional processing, as it appears that voltage-gated ion channel density, rather than their distribution patterns contributes the major difference between mature and immature response profiles (3, 25, 26).

To examine the possibility that neuromodulation of spiral ganglion heterogeneity by cAMP could serve as a difference detection amplifier that feeds into higher levels of processing, we first evaluated whether both up- and downregulation of cAMP altered specific neuroelectric parameters, leaving others unchanged. We also determined whether these potential difference detections were maintained within a tonotopic framework as organized at higher processing levels (27–29). And finally, we assessed whether the electrophysiological responses were consistent with that expected for the physical attributes of acoustic stimuli. Our results show that neuromodulation of spiral ganglion neurons by cAMP did indeed independently modulate tonotopically organized differences for electrophysiological correlates having the appropriate relationships to physical features of the input signal. Intracellular cAMP signaling, therefore, does have the capacity to shape our perceptions of acoustic input by amplifying differences between specific electrophysiological correlates of stimulus attributes, consistent with transiently enhancing stimulus disparities.

METHODS

Tissue Culture

Procedures performed using CBA/CaJ mice were approved by the Rutgers University Institutional Review Board for the Use and Care of Animals, protocol 90-073. Both cochleae were removed from postnatal age 6 (P6) CBA/CaJ mice (Jackson Laboratory, Bar Harbor, ME) to examine spiral ganglion neurons with slower kinetics and higher excitability levels than their mature counterparts. This is within a time period of active development (30), before cochlea maturation is largely completed by P8 (31), hearing in the mid-frequency range is first detected at P11 (32) and the electrophysiological properties of basal innervation neurons plateau at P14–P15 (1). Subsequent refinements, including synaptic interactions, spontaneous rate, myelination, and some gene expression profiles, continue into the 4th postnatal wk (3, 33–36). Thus, although P6 spiral ganglion neurons do not possess adult ion channel distributions or densities, they do produce overshooting action potentials that display stable waveform profiles. Moreover, key channel proteins that play a predominant role in shaping the action potentials in the first postnatal week demonstrate distribution patterns throughout the ganglion that are maintained into adulthood (25, 37).

As previously described (25), the most basal, middle, and apical fifths of the spiral ganglion were isolated for tissue culture and plated as explants without exposure to enzymes. These regions of the ganglion correspond to the high-, middle-, and low-frequency regions of the cochlea that will ultimately form a frequency-specific (tonotopic) map. Thus, the tonotopic position of a neuron refers to its general frequency-specific region of innervation along the cochlear contour. Explant tissues were positioned onto the center of poly-l-lysine-coated dishes and maintained in growth medium (Dulbecco’s modified Eagle’s medium, supplemented with 10% fetal bovine serum, 4 mM l-glutamine, and 0.1% penicillin-streptomycin) for 5–11 days in vitro (DIV) at 37°C in a humidified incubator with 95%/5% O₂/CO₂.

Electrophysiology

Filamented borosilicate glass capillary tubes (Sutter Instruments, Novato, CA, Catalog No. BF150-110-10) were pulled on a two-stage vertical puller (Narishige, PP-83). Electrodes were coated with silicone-elastomer (Sylgard, Corning) and fire polished (Narishige, MF-83) just before use. Electrode resistances ranged from 3 to 6 MΩ in the bath solution used for this study. Pipette offset current was zeroed just before contacting the cell membrane. Current-clamp measurements were made using the I_fast circuitry of the Axopatch 200 A amplifier to reduce improper error currents (38, 39). The internal solution was as follows (in mM): 112 KCl, 2 MgCl₂, 0.1 CaCl₂, 11 EGTA, and 10 HEPES-KOH, pH 7.5. The bath solution (in mM) was 137 NaCl, 5 KCl, 1.7 CaCl₂, 1 MgCl₂, 17 glucose, 13 sucrose, and 10 HEPES, pH 7.5; ∼330 osmol/kgH₂O. A liquid junction potential of −5 mV was not corrected. Recordings were made from neuronal somata at room temperature (19°C–22°C). The absolute kinetic values that are temperature-dependent (40), did not alter conclusions drawn from our assessments, as we evaluated the relative latency and duration differences between neurons and action potential waveforms.

Neurons were electrophysiologically distinguished from surrounding supporting cells by the presence of a large, transient inward sodium current in whole cell voltage clamp mode with a depolarizing voltage step from −60 mV to −30 mV that has been shown in a previous study to be blocked by 0.2 µM tetrodotoxin (41). For current clamp recordings, current steps of 240-ms duration were elicited every 5 s in 10 pA increments. Parameters assessed include action potential latency and duration, voltage threshold, and accommodation (also referred to as APmax, measured as the maximum number of action potentials fired in response to suprathreshold stimulation). Any changes to these parameters during the course of an experiment rendered subsequent activity unacceptable for analysis. Voltage threshold was measured as the peak amplitude of the just-subthreshold response using 1 pA current injection increments. Neuronal resting membrane potential was measured before any current had been injected at the outset of each experiment in current clamp mode. This protocol was evaluated in a previous study to generate robust data that, while slightly depolarized compared with noninvasive single-channel assessments, reproduced the relative differences between cells (41). A holding potential (HP) of −80 mV was chosen to assess responses with minimal voltage-dependent ion channel inactivation while permitting assessments of hyperpolarization-activated (I_h) currents, which have relatively hyperpolarized voltage dependence of activation in spiral ganglion neurons (6). For Fig. 2, the HP was also depolarized to −60 mV to compare accommodation levels with −80 mV HP recordings. Data were digitized at 10 kHz with a CED Power 1401 interface using a personal computer and filtered at 2 kHz; the programs for data acquisition and analysis were written by MRP. Acceptable current clamp recordings met the following criteria: low noise levels (<4 mV peak-to-peak; typically, <1 mV peak-to-peak), stable holding potentials (<1 mV change) with minimal constant current injection (average of ≤12 pA), discernible membrane time constant on step current injection, and overshooting action potentials (magnitudes of ≥80 mV from baseline).

Figure 2.

Differential effects of H-89 and 8-bromo-cAMP (8-Br-cAMP) on accommodation. A: scatter plots of maximum number of action potentials (APmax) values obtained at −80 and −60 mV holding potentials for H-89, vehicle, and cAMP conditions. The stack of sweeps to the right highlight the accommodation categories and differences in action potential firing (red circles denote the corresponding neuron). The percentages are the numbers of cells that fall into unitary (APmax = 1, bottom percentages below horizontal dotted line), rapid (APmax = 2–8), and slow accommodation (APmax = ≥11) categories defined previously (1, 5). Note the shift toward unitary and rapid accommodation after H-89 treatment. Exposure to 8-Br-cAMP produced a slight increase in the number of slowly accommodating neurons and resulted in neurons that fired the most action potentials. Note the gap in APmax values, particularly in the H-89 and 8-Br-cAMP groups, that occurs between the rapidly and slowly accommodating categories. B: overlay of all APmax sweeps from H-89 (left stack), vehicle (middle stack), and 8-Br-cAMP (right stack) highlights the suppressive effect on firing by H-89. The numbers of experiments are shown below each column label. C: effect of H-89, vehicle, and 8-Br-cAMP, on APmax from a holding potential of −80 mV. Values for APmax (H-89: Base, 2.1 ± 0.3 AP; Middle, 3.6 ± 1.0 AP; Apex, 6.5 ± 1.9 AP; Vehicle: Base, 4.2 ± 1.8 AP; Middle, 10.7 ± 2.2 AP; Apex, 12.1 ± 2.4 AP; 8-Br-cAMP: Base, 7.5 ± 2.2 AP; Middle, 12.4 ± 3.0 AP; Apex, 14.6 ± 2.7 AP). Note the tonotopic distribution is relatively unchanged by 8-Br-cAMP exposure but H-89 causes a shift from a nonmonotonic to a graded distribution. D: same recorded cells as C but the cells were held at −60 mV to assess APmax. The overall effect on and distribution of APmax is the same as −80 mV. Values for APmax (H-89: Base, 3.5 ± 1.2 AP; Middle, 4.7 ± 1.4 AP; Apex, 7.8 ± 2.2 AP; Vehicle: Base, 8.2 ± 2.3 AP; Middle, 18.9 ± 2.1 AP; Apex, 17.4 ± 2.1 AP; 8-Br-cAMP: Base, 16.2 ± 3.2 AP; Middle, 20.2 ± 2.8 AP; Apex, 19.6 ± 2.5 AP). Scale bars: 20 mV, 50 ms (A) and the sweeps in B are normalized so only the time domain scale bar (50 ms) is displayed. Significant differences between vehicle control and each treatment group in bar charts (C and D) are indicated by ** (P > 0.01).

H-89 (10 µM) and 8-bromo-cAMP, sodium salt (8-Br-cAMP, 100 µM) (Tocris, Bristol, UK) were prepared as ×1,000 stock solutions using sterile water. Cultures were pretreated with either H-89, 8-Br-cAMP, or vehicle control (2 µL of sterile distilled water) added to the media 1 h before recording to assure that cAMP levels stabilized uniformly in the electrophysiologically heterogeneous neurons and to maintain consistency with previous studies from our laboratory (6, 10), without evoking longer-term effects (42, 43). For recordings, the same concentration of each reagent was added to the external recording solution and the internal solution (8-Br-cAMP) at the same concentration as the pretreatment condition. All recordings were performed from multiple platings and animals.

Statistical Analyses

Statistical comparisons between heterogeneous electrophysiological parameters were evaluated across three treatment groups and three gangliotopic locations with a two-way ANOVA (Table 1) followed by a Tukey–Kramer post hoc pairwise analysis (MATLAB, MathWorks). P values <0.05, shown with an asterisk (*) and P values <0.01, shown with a double asterisk (**) denote significant differences between vehicle control and each treatment group in bar charts. The data are displayed as means ± SE (standard error), as indicated.

Table 1.

Two-way ANOVA summary table for each electrophysiological parameter

Parameter	Sources of Variation	df	MS	F	P
Ih sag magnitude, mV	Location	2	194.0419	2.6480	0.0738
	Treatment	2	3.7191e+04	507.5354	5.6354e-71
	Location-treatment	4	163.8266	2.2357	0.0674
	Within groups	164	73.2785
Hyperpolarization following APmax, mV	Location	2	1.0218e+03	9.3140	1.4585e-04
	Treatment	2	3.9685e+03	36.1742	8.6161e-14
	Location-treatment	4	185.5578	1.6914	0.1543
	Within groups	168	109.7056
Resting potential, mV	Location	2	70.0717	11.1500	2.9168e-05
	Treatment	2	435.4597	69.2916	1.9491e-22
	Location-treatment	4	30.8119	4.9029	9.3342e-04
	Within groups	161	6.2845
APmax (−80 mV)	Location	2	634.5656	7.2247	9.7569e-04
	Treatment	2	824.7753	9.3902	1.3582e-04
	Location-treatment	4	35.4085	0.4031	0.8062
	Within groups	169	87.8332
APmax (−60 mV)	Location	2	556.2043	6.3278	0.0023
	Treatment	2	2.4249e+03	27.5881	6.2709e-11
	Location-treatment	4	110.4730	1.2568	0.2896
	Within groups	150	87.8982
Voltage threshold, mV	Location	2	497.1885	28.7366	1.8093e-11
	Treatment	2	343.4451	19.8505	1.8056e-08
	Location-treatment	4	21.8011	1.2601	0.2877
	Within groups	169	17.3016
Current excitability, pA	Location	2	8.4188e+04	26.4937	9.8104e-11
	Treatment	2	2.0054e+05	63.1102	3.3812e-21
	Location-treatment	4	5.4345e+03	1.7102	0.1500
	Within groups	169	3.1776e+03
Onset τ fast, ms	Location	2	0.5085	1.3644	0.2584
	Treatment	2	6.1537	16.5112	2.8847e-07
	Location-treatment	4	0.3481	0.9340	0.4457
	Within groups	166	0.3727
Onset τ slow, ms	Location	2	132.6791	4.9233	0.0084
	Treatment	2	332.7634	12.3478	1.0016e-05
	Location-treatment	4	26.3961	0.9795	0.4203
	Within groups	166	26.9493
Latency, ms	Location	2	158.7024	17.3071	1.4526e-07
	Treatment	2	256.6206	27.9854	3.1751e-11
	Location-treatment	4	26.5074	2.8907	0.0239
	Within groups	169	9.1698
Rise rate, mV/ms	Location	2	5.6172e+03	3.1399	0.0458
	Treatment	2	2.4067e+04	13.4531	3.7874e-06
	Location-treatment	4	6.0320e+03	3.3718	0.0110
	Within groups	169	1.7890e+03
Rapid repolarization, mV/ms	Location	2	83.6040	0.2791	0.7568
	Treatment	2	2.0538e+03	6.8572	0.0014
	Location-treatment	4	1.3110e+03	4.3771	0.0022
	Within groups	169	299.5075
Slow repolarization, mV/ms	Location	2	1.9942e+03	8.0616	4.6370e-04
	Treatment	2	1.5805e+03	6.3891	0.0021
	Location-treatment	4	736.7407	2.9783	0.0210
	Within groups	158	247.3687
Duration, ms	Location	2	0.2189	4.0134	0.0198
	Treatment	2	0.5699	10.4517	5.2529e-05
	Location-treatment	4	0.1850	3.3927	0.0107
	Within groups	169	0.0545

APmax, maximum number of action potentials; df, degrees of freedom; F, F-statistic; I_h, hyperpolarization-activated; MS, mean squares; P, P value.

RESULTS

Experiments were carried out on 178 spiral ganglion neurons isolated from 35 P6 CBA/CaJ mice, separated into three groups (61 vehicle control, 59 8-Bromo cAMP, and 58 H-89). To place our results into context, we consider excitability and membrane kinetics separately, as these features are independently regulated in the primary auditory afferents (1, 10).

Neuromodulation of Spiral Ganglion Excitability

Neuron excitability has multiple controlling parameters. The resting membrane potential (RMP) in combination with the voltage at which threshold is reached determines neuronal sensitivity, whereas the current required to reach threshold is a measure of neuronal responsiveness. The roles of RMP and threshold voltage in determining the sensitivity of a neuron work in opposition. As threshold becomes more depolarized the cell becomes less sensitive, whereas as RMP becomes more depolarized the cell becomes more sensitive. In combination, these two parameters shift neuronal sensitivity heterogeneously in spiral ganglion neurons (41).

Major regulators of RMP in spiral ganglion neurons are the hyperpolarization-activated cyclic nucleotide-gated (HCN) channels, which are modulated by increased [cAMP]_i (6, 41, 44–46). We first confirmed the effect of 8-Br-cAMP on HCN channels with current clamp recordings. Since cAMP acts to shift the voltage dependence of activation of HCN channels to more depolarized levels (47, 48), we evaluated neuronal voltage responses to hyperpolarizing constant current injections that highlight HCN contributions. As shown in Fig. 1A, under vehicle control conditions, a hyperpolarizing current injection to −185 mV from a holding potential of −80 mV results in a peak hyperpolarization that then sags back toward baseline, which is typical for I_h currents (6, 45, 46). Exposure to 8-Br-cAMP (100 µM) resulted in a marked reduction in the I_h sag magnitude (Fig. 1A, dark gray sweep) likely due to the expected depolarizing shift in activation by cAMP causing HCN channels to be already robustly activated. Thus, the hyperpolarizing current injection did not result in significantly more channels being activated since a large proportion were likely already open. Exposure to H-89 (10 µM) resulted in a slight increase in I_h sag magnitudes (Fig. 1A, light gray sweep) consistent with a minor indirect effect mediated by PKA (49), while most of the effect is through direct action of cAMP (48). H-89 had no effect on basal neurons (−51.5 ± 1.4 mV; P = 0.9696) and a significant effect on middle (−58.8 ± 2.8 mV; P = 0.0267) and apical neurons (−60.7 ± 2.7 mV; P = 0.0008), versus the respective vehicle control, which could be due to a negative shift in the half-maximal voltage (6). The average values for all three treatment conditions are summarized in Fig. 1B by location (H-89: −56.7 ± 1.4 mV; P = 0.0002; vehicle: −50.1 ± 1.3 mV; 8-Br-cAMP: −0.8 ± 0.6 mV; P < 0.001, H-89 vs. vehicle and 8-Br-cAMP vs. vehicle, respectively) and Table 2 by adaptation class (1).

Figure 1.

Assessment of 8-bromo-cAMP (8-Br-cAMP) and H-89 effects on whole cell currents shows the expected changes in response properties likely due to altered hyperpolarization-activated (I_h) currents. A: whole cell current clamp recordings of I_h sag magnitude demonstrate the effects of 8-Br-cAMP (dark-gray sweep) and H-89 (light-gray sweep) relative to vehicle control (black sweep). B: bar graphs of the average response of all spiral ganglion neurons for each treatment condition and individual scatter plots for each cochlear location. H-89 (base): −51.5 ± 1.4 mV; H-89 (middle): −56.2 ± 0.7 mV; H-89 (apex): −58.8 ± 0.7 mV. Vehicle (base): −60.7 ± 0.5 mV; vehicle (middle): −50.3 ± 2.5 mV; vehicle (apex): −48.9 ± 2.2 mV. 8-Br-cAMP (base): −8.3 ± 1.1 mV; 8-Br-cAMP (middle): −10.1 ± 1.1 mV; 8-Br-cAMP (apex): −11.4 ± 1.3 mV. C: overlapping sweeps at both threshold and APmax (the maximum number of action potentials that could be elicited) illustrate the effect of 8-Br-cAMP on the magnitude of the afterhyperpolarization, which manifests when the strong depolarizing current injection used for testing APmax is returned to baseline levels. The shaded rectangle and arrows highlight this current, which is present to a lesser degree in vehicle and H-89 conditions. D: bar graphs of the average hyperpolarization following APmax of all spiral ganglion neurons for each treatment condition and individual scatter plots for each cochlear location. H-89 (base): 27.4 ± 2.1 mV; H-89 (middle): 22.7 ± 2.8 mV; H-89 (apex): 33.3 ± 2.1 mV. Vehicle (base): 22.9 ± 2.4 mV; vehicle (middle): 22.4 ± 2.1 mV; vehicle (apex): 34.9 ± 1.9 mV. 8-Br-cAMP (base): 39.1 ± 2.3 mV; 8-Br-cAMP (middle): 41.8 ± 2.2 mV; 8-Br-cAMP (apex): 43.1 ± 3.1 mV. E: bar graphs of the average resting membrane potential of all spiral ganglion neurons for each treatment condition and individual scatter plots for each cochlear location. H-89 (base): −60.3 ± 0.5 mV; H-89 (middle): −56.2 ± 0.7 mV; H-89 (apex): −58.8 ± 0.7 mV. Vehicle (base): −60.7 ± 0.5 mV; vehicle (middle): −58.4 ± 0.8 mV; vehicle (apex): −59.8 ± 0.4 mV. 8-Br-cAMP (base): −55.1 ± 0.4 mV; 8-Br-cAMP (middle): −54.9 ± 0.6 mV; 8-Br-cAMP (apex): −52.9 ± 0.5 mV. Scale bars: 50 mV, 50 ms (A) and 20 mV, 50 ms (C). cAMP labeled in this and subsequent figures refers to 8-Br-cAMP. Significant differences between vehicle control and each treatment group in bar charts (B–E) are indicated by * (P<0.05) or ** (P>0.01).

Table 2.

Distribution of action potential profiles by adaptation within each experimental group

Parameter	H-89			Vehicle			cAMP
	UA (n = 22)	RA (n = 30)	SA (n = 6)	UA (n = 16)	RA (n = 24)	SA (n = 21)	UA (n = 15)	RA (n = 21)	SA (n = 23)
Ih sag magnitude, mV	52.72 ± 1.58	58.11 ± 2.11	64.06 ± 5.17	50.46 ± 2.54	52.06 ± 1.69	47.53 ± 2.53	6.40 ± 0.46	11.35 ± 1.17	10.62 ± 1.12
Hyperpolarization after APmax, mV	29.21 ± 2.49	24.63 ± 1.80	35.53 ± 5.02	24.31 ± 3.10	24.66 ± 2.10	30.28 ± 2.26	45.62 ± 2.99	37.44 ± 2.45	41.60 ± 2.04
Resting potential, mV	−59.63 ± 0.42	−57.65 ± 0.68	−58.47 ± 1.42	−61.11 ± 0.53	−59.64 ± 0.63	−58.41 ± 0.55	−55.50 ± 0.48	−54.07 ± 0.43	−54.13 ± 0.61
APmax (−80 mV)	1.00 ± 0.00	2.93 ± 0.33	19.67 ± 1.38	1.00 ± 0.00	2.96 ± 0.22	22.43 ± 1.20	1.00 ± 0.00	3.57 ± 0.36	24.87 ± 1.36
APmax (−60 mV)	1.3 ± 0.13	4.96 ± 0.90	21.80 ± 1.83	5.63 ± 2.13	13.29 ± 1.62	25.44 ± 0.95	2.67 ± 0.38	20.13 ± 2.01	26.62 ± 1.43
Voltage threshold, mV	−37.79 ± 1.16	−42.15 ± 0.61	−45.60 ± 1.43	−40.37 ± 0.83	−46.20 ± 0.76	−47.10 ± 0.48	−41.00 ± 1.53	−46.51 ± 0.77	−47.42 ± 0.88
Current threshold, pA	186.86 ± 14.59	101.40 ± 7.59	104.50 ± 12.88	187.25 ± 8.72	122.88 ± 9.26	109.62 ± 5.84	297.27 ± 23.57	235.62 ± 13.52	201.83 ± 10.69
Onset τ fast, ms	0.91 ± 0.10	1.15 ± 0.12	1.44 ± 0.16	1.42 ± 0.34	1.14 ± 0.11	1.49 ± 0.14	0.50 ± 0.05	0.75 ± 0.06	0.73 ± 0.06
Onset τ slow, ms	5.70 ± 0.55	10.58 ± 1.59	10.49 ± 1.89	4.66 ± 0.29	9.76 ± 1.21	10.77 ± 1.23	3.12 ± 0.25	5.15 ± 0.62	4.79 ± 0.48
Latency, ms	10.23 ± 0.64	13.81 ± 0.68	16.45 ± 1.73	8.48 ± 0.42	12.03 ± 0.52	13.81 ± 0.56	6.61 ± 0.49	8.89 ± 0.70	9.63 ± 0.46
Rise rate, mV/ms	181.97 ± 9.71	205.60 ± 8.72	232.69 ± 22.04	209.21 ± 10.94	247.58 ± 9.68	245.02 ± 6.07	223.75 ± 8.10	237.59 ± 9.52	244.09 ± 7.76
Rapid repolarization, mV/ms	−71.84 ± 3.70	−78.77 ± 3.81	−86.12 ± 4.36	−87.62 ± 5.01	−88.99 ± 2.74	−89.71 ± 3.85	−84.07 ± 4.35	−78.58 ± 4.39	−82.67 ± 3.59
Slow repolarization, mV/ms	−45.07 ± 2.05	−42.03 ± 1.87	−49.78 ± 2.25	−51.32 ± 3.10	−47.22 ± 1.53	−46.52 ± 1.59	−60.04 ± 4.05	−49.34 ± 3.06	−52.93 ± 2.53
Duration, ms	1.27 ± 0.05	1.31 ± 0.06	1.13 ± 0.08	1.04 ± 0.04	1.13 ± 0.03	1.13 ± 0.03	0.99 ± 0.06	1.14 ± 0.07	1.09 ± 0.05

All measurements are presented as means ± SE. n indicates number of neurons. APmax, maximum number of action potentials; RA, rapidly adapting; SA, slowly adapting; UA, unitary adapting.

Because the hyperpolarizing sag is not evident when HCN channels are open, one would expect a greater voltage change at the offset of applied step depolarizing constant current injections. This is what we observed. In neurons exposed to 8-Br-cAMP, there was a larger hyperpolarization following a prolonged step depolarization to baseline (Fig. 1C; right traces, shaded regions with arrows) consistent with a greater number of open HCN channels. When quantified at the current injection level that evoked the maximum number of action potentials fired during a 240-ms depolarizing current injection (APmax), we noted that neurons exposed to 8-Br-cAMP displayed significantly larger hyperpolarizations (Fig. 1C). This effect was significant for neurons exposed to 8-Br-cAMP but not H-89, relative to vehicle control (8-Br-cAMP: 41.2 ± 1.4 mV; vehicle: 27.3 ± 1.4 mV; H-89: 27.1 ± 1.5 mV; Fig. 1D; P < 0.001, 8-Br-cAMP vs. vehicle and P = 0.8769, H-89 vs. vehicle).

Both observations described earlier are consistent with contributions of HCN channels. Moreover, the effect of these channels on RMP are similar to previous observations where blockade of HCN channels with Cs⁺ resulted in ∼−4 mV hyperpolarization of the RMP (41). Here, we found that exposure of spiral ganglion neurons to 8-Br-cAMP resulted in ∼5 mV depolarization of the resting membrane potential relative to vehicle control (Fig. 1E; 8-Br-cAMP: −54.4 ± 0.3 mV; vehicle: −59.6 ± 0.4 mV, P < 0.001). H-89 exposure did not affect the RMP under these conditions (Fig. 1E; H-89: −58.5 ± 0.4 mV P = 0.0963). Thus, these data showing the robust contribution of I_h to the RMP by current clamp are compatible with analysis by voltage clamp (6).

Having replicated previous studies showing a pronounced effect of cAMP on HCN channels, we explored additional electrophysiological parameters under these same conditions since other voltage-gated ion channels found in spiral ganglion neurons (11, 50) are likely modulated by cAMP via PKA (19). The most obvious firing feature of these neurons is adaptation, which we characterized with measurements of APmax obtained for H-89, vehicle, and 8-Br-cAMP treatment groups. Holding potentials of −80 and −60 mV were combined in the initial part of the analysis to increase the number of observations in each adaptation category. Under vehicle control conditions, neurons that fired a single action potential (unitary adapting; UA) represented 19.8% of the population and neurons that fired 2–8 action potentials (APs) were rapidly adapting (RA), which represented 32.8% of the population (Fig. 2A, gray circles). The third group consisted of slowly adapting (SA) neurons, which fired ≥11 APs and were 47.4% of the population. Exposure to H-89 decreased APmax and shifted the adaptation categories such that the great majority of neurons were either UA (33.9%) or RA (53.2%), whereas only 12.9% were SA (Fig. 2A, black circles). Conversely, 8-Br-cAMP exposure tended to enhance firing and shifted the adaptation categories such that the majority was SA (52.3%, gray triangles). The 8-Br-cAMP-exposed neurons also fired the most action potentials. The three adaptation categories and heterogeneity of firing are displayed in the column of sweeps to the right of the scatter plot in Fig. 2. The displayed neurons are identified as red circles within the scatter plot. To highlight the effect of 8-Br-cAMP and H-89 on spiral ganglion neuron firing, we overlaid all APmax traces from each cochlear region (Fig. 2B, holding potential = −80 mV). The traces were aligned to the plateau region or, if the cell was SA and fired throughout the test pulse, the plateau was estimated (5). The plots show that compared with vehicle, H-89 exposure reduced firing both at −80 mV (Fig. 2C; P = 0.0089) and at −60 mV (sweeps not shown, Fig. 2D; P < 0.001) holding potentials [−80 mV: (veh) 9.1 ± 1.3 AP, (H-89): 3.9 ± 0.7 AP; −60 mV: (veh): 14.9 ± 0.7 AP, (H-89): 5.2 ± 0.9 AP]. Conversely, 8-Br-cAMP exposure tended to increase APmax yet was not statistically significant (−80 mV: 11.2 ± 1.5 AP, −60 mV: 18.4 ± 1.7 AP; P = 0.4663 and P = 0.1276, vs. vehicle, respectively). These data suggest that PKA inhibition suppresses whereas cAMP tends to promote firing in spiral ganglion neurons.

A single potassium channel class, K_V1, which can be modulated by PKA activation (51), regulates both adaptation and threshold in spiral ganglion neurons (41), which prompted our investigation of threshold level in 8-Br-cAMP and H-89 experimental conditions. This parameter was assessed by measuring the voltage of the just-subthreshold response (Fig. 3A, dotted line) from a holding potential of −80 mV. Threshold voltages remained stable between control and 8-Br-cAMP conditions when compared between adaptation classes (Table 2), similar to that observed in neurons isolated at P14–P15 (10). Moreover, we found that the cochleotopic U-shaped pattern of voltage thresholds (41, 52), where basal neurons have significantly higher levels relative to middle/apical neurons, was evident in the vehicle control condition and reiterated in the 8-Br-cAMP treatment group (Fig. 3B, bottom example traces).

Figure 3.

The tonotopic distribution of voltage thresholds is re-patterned by H-89 but not 8-Br-cAMP. A: representative sweeps showing the measurement of the threshold response (black trace) and the corresponding action potential (gray trace). B: neurons from the base have elevated voltage thresholds and achieve threshold faster relative to middle and apex neurons. Application of 8-bromo-cAMP (8-Br-cAMP) (bottom sweeps) had essentially no effect on voltage thresholds but did affect the timing of basal and apical neurons (see Figs. 4–7). However, H-89 significantly elevated voltage thresholds (top 3 sweeps) in all three cochlear regions. Sweeps from each region are represented by color as shown by the key next to the 8-Br-cAMP sweeps. C: bar charts show the average for each treatment condition and the scatter plots show the averages for each cochlear location. The individual values for each cochlear location show the nonmonotonic distribution of voltage thresholds in vehicle and 8-Br-cAMP conditions and the re-patterning of thresholds into a linear gradient following H-89 application. The values for voltage threshold are: H-89: Base, −37.7 ± 1.3 mV; Middle: −41.7 ± 0.7 mV; Apex: −43.8 ± 0.8 mV; Vehicle: Base, −40.8 ± 0.7 mV; Middle, −47.2 ± 0.6 mV; Apex −46.5 ± 0.7 mV; 8-Br-cAMP: Base, −43.1 ± 1.4 mV; Middle: −47.1 ± 0.7 mV; Apex, −46.5 ± 1.0 mV. D: bar charts of each treatment condition and scatter plots showing averages for each cochlear location for threshold current. The values for current threshold are: H-89: Base, 190.7 ± 15 pA; Middle, 89.8 ± 8.3 pA; Apex, 116.4 ± 8.6 pA; Vehicle: Base, 171 ± 11.1 pA; Middle, 110 ± 8.3 pA; Apex, 128 ± 9.2 pA; 8-Br-cAMP: Base, 270.5 ± 18.5 pA; Middle. 226.6 ± 12 pA; Apex, 209.8 ± 17.4 pA. Significant differences between vehicle control and each treatment group in bar charts (C and D) are indicated by ** (P > 0.01).

In contrast to the nonmonotonic pattern shaped by the lowest thresholds in middle neurons, H-89 established a monotonic pattern of thresholds from base to apex (Fig. 3B, top example traces). Quantification of these observations showed the overall nonmonotonic cochleotopic distribution with slightly lower voltage thresholds observed in neurons from the middle region in the vehicle condition (gray circles; these region-specific values for this and subsequent figures are located in the figure legend) with an average voltage threshold at −44.9 ± 0.5 mV (Fig. 3C, middle bar; n = 61). Similar to APmax, application of 8-Br-cAMP had its largest effect on basal neurons, although the change was not statistically significant (Fig. 3C, right bar; −45.4 ± 0.6 mV, n = 59; P = 0.3553). Conversely, exposure to H-89 significantly depolarized voltage thresholds in all locations (Fig. 3C, left bar; −40.8 ± 0.6 mV, n = 58; P < 0.001). Thus, H-89 exposure shifted the nonmonotonic distribution of heterogeneous voltage thresholds to a graded pattern—a result similar to the tonotopic redistribution produced by H-89 on APmax.

These results suggest that spiral ganglion excitability is affected by threshold and RMP across a wide range of cAMP levels. In the presence of H-89, low levels of PKA activation shift thresholds to higher voltages making the cells less apt to fire. As the levels of PKA activation shift to control values, thresholds are reduced, which enhance the responsiveness of the cells to current injections. Further increases in cAMP result in depolarized RMP levels that reduce the gap between RMP and threshold potentially enhancing neuronal sensitivity. The neuronal responsiveness, however, could be separately modified should the current required to reach threshold was altered. This was in fact the case for the cAMP condition, as a significantly larger current input was required to reach the threshold (Fig. 3D). Overall, the higher the levels of cAMP signaling, the greater the spiral ganglion neuron sensitivity, yet in conjunction with this, greater current input levels were required to reach threshold.

To examine the V-I relationship for the population, we plotted threshold voltage as a function of current injection level for the population of neurons within each of the three conditions (Fig. 4A). Compared with H-89, the relationship was shifted downward and to the right for the cAMP treatment group, indicating that higher current levels are required to initiate action potentials at overall lower thresholds. Although the data showed the expected diversity, fitted lines displayed similar slopes between the high and low cAMP conditions [Fig. 4A, solid (m = 0.479, R² = 0.4124) and dotted (m = 0.438, R² = 0.4166) black lines, respectively]. Overall, the greater sensitivity evident by comparing threshold level to RMP in neurons with elevated cAMP (data points closest to the dashed line; Fig. 4B) was accompanied by lower responsivity due to the shift in the V-I function. In contrast, the higher thresholds evident in the H-89 condition showed even lower sensitivity, given the shift away from a relatively stable RMP level (Fig. 4B; black circles). Moreover, the shift of the V-I function was not limited to threshold level, as it was also reflected in the suprathreshold measurement of APmax (Fig. 4C).

Figure 4.

The relationship between key excitability parameters were altered depending on intracellular cAMP signaling level. A: plot of voltage to current levels measured at threshold for the population of neurons in each of the three conditions. Linear fit R² values for H-89 (black solid line), Vehicle (red dashed line), and 8-bromo-cAMP (8-Br-cAMP) (gray dotted line) data were 0.412, 0.45, and 0.417, respectively. B: action potential threshold plotted as a function of resting membrane potential (RMP) for neurons within each of the three conditions. Although threshold levels shift relative to holding potential (HP), measurements at −80 mV showed that the lowest threshold neurons in the cAMP condition were closest to the line of unity (dashed line). C: the suprathreshold measurement of maximum number of action potentials (APmax) also shifted along the current axis for data obtained from the 8-Br-cAMP condition relative to vehicle and H-89 conditions. Legend in B applies to both B and C.

Overall, the individual components of spiral ganglion neuron excitability were separately regulated by cAMP neuromodulation. Neurons in the H-89 treatment group became less sensitive as a result of significantly lowered threshold sensitivity, whereas neurons in the 8-Br-cAMP treatment group became more responsive due to elevated RMPs. Thus, input signals would be differentially amplified when compared across neurons having high versus low cAMP signaling levels. These differences also extend to their distribution profiles such that altered parameters, including APmax, and threshold voltage and current, became reorganized into a tonotopically graded pattern. Thus, spiral ganglion sensitivity and responsivity are amplified and re-ordered by cAMP neuromodulation before downstream transmission.

Neuromodulation of Spiral Ganglion Kinetics

Because the endogenous properties of excitability and kinetics are controlled by distinct voltage-gated ion channel types (50) that are independently regulated (1, 10), we wondered whether the same principles observed for neuron excitability would also apply to neuronal kinetics. Two aspects of spiral ganglion neuron kinetics that we evaluated were action potential latency and duration at threshold. Both of these kinetic features result from the contributions of multiple components that likely fine-tune the endogenous neuronal firing profile. For example, there are two major components that impact the time course (τ) of the onset response at threshold, which influence the overall action potential latency. In contrast, action potential duration is composed of at least three separately regulated phases (fast depolarization, fast hyperpolarization, and slow hyperpolarization) all of which contribute to the overall action potential width. Thus, in the auditory system in which timing is a critical component of neuronal encoding, neurons possess the requisite fine controls over multiple kinetic parameters.

Action potential latency.

The first step toward examining action potential latency at threshold is to measure onset tau (τ). This parameter, evaluated at the just subthreshold voltage response, was best fitted with a double-exponential (Fig. 5A1, dashed line). As shown by the representative sweeps (Fig. 5A2), 8-Br-cAMP exposure quickened the voltage rise time (light gray trace, highlighted by black arrowhead) relative to vehicle control (gray middle trace) and H-89 exposure (black trace). These individual experiments are supported by an analysis of each treatment group of the fast (Fig. 5B) and slow (Fig. 5C) components. Recordings from H-89 treated neurons had average values for the fast (1.1 ± 0.1 ms) and slow (8.0 ± 0.7 ms) components that were similar to the values of vehicle control neurons (fast: 1.3 ± 0.1 ms; slow: 8.8 ± 0.7 ms; P = 0.0851 and P = 0.9958, compared with vehicle control, respectively). When the effect of 8-Br-cAMP on onset τ was quantified, we found that both components were significantly more rapid (fast: 0.6 ± 0.03 ms; slow: 4.4 ± 0.3 ms; P < 0.001 relative to vehicle control for both parameters). These data show that elevating intracellular cAMP signaling level results in speeding of the voltage response, and the accentuation of graded tonotopic differences for this timing feature.

Figure 5.

Differential effects of 8-bromo-cAMP (8-Br-cAMP) and H-89 on the tonotopic distributions of onset τ. A1: representative subthreshold and threshold sweeps with a double exponential fit (dotted line). A2: the vehicle sweep (middle) is bookended by H-89 (right) and 8-Br-cAMP (left). The arrowhead in the 8-Br-cAMP sweep points to the more rapid onset τ of threshold responses in 8-Br-cAMP-treated neurons. The representative sweeps reflect both the timing and threshold effects of 8-Br-cAMP and H-89 for each region. B: bar graphs and individual scatter plots of the fast component of the onset τ for H-89, vehicle, and 8-Br-cAMP. The values for the fast component of onset τ are: Base (H-89), 0.89 ± 0.10 ms; Middle (H-89): 1.28 ± 0.17 ms; Apex (H-89): 1.11 ± 0.11 ms; Base (vehicle): 1.39 ± 0.27 ms; Middle (vehicle): 1.36 ± 0.17 ms; Apex (vehicle): 1.24 ± 0.13; Base (8-Br-cAMP): 0.54 ± 0.04 ms; Middle (8-Br-cAMP): 0.73 ± 0.06 ms; Apex (8-Br-cAMP): 0.79 ± 0.08 ms. C: bar graphs, individual scatter plots and statistical comparisons of the slow component of the onset τ. Symbols are the same as B. The values for the slow component of onset τ are: Base (H-89): 7.45 ± 1.40 ms; Middle (H-89): 9.91 ± 1.34 ms; Apex (H-89): 8.79 ± 0.98 ms; Base (vehicle) 7.45 ± 1.40 ms; Middle (vehicle): 9.9 ± 1.33 ms; and Apex (vehicle): 8.79 ± 0.98 ms; Base (8-Br-cAMP): 3.67 ± 0.40 ms; Middle (8-Br-cAMP): 4.61 ± 0.33 ms; and Apex (8-Br-cAMP): 5.46 ± 0.86 ms.

The different onset time constants produced by 8-Br-cAMP and H-89 application, not unexpectedly, result in significant differences in action potential latency. As shown in the representative traces, action potential latencies were fastest in basal neurons and nonlinear across the tonotopic axis for H-89 treated neurons in the overlaid voltage sweeps (Fig. 6A) and group data (Fig. 6B). In contrast, neurons exposed to 8-Br-cAMP exhibited a tighter clustering of action potential latencies. In the group data, vehicle control neurons exhibited an average action potential latency of 11.7 ± 0.4 ms (Fig. 6B, middle bar). H-89 treatment did not affect action potential latencies when compared with vehicle control (12.7 ± 0.5 ms; left bar; P = 0.2255). Conversely, exposure to 8-Br-cAMP significantly shortened the action potential latencies (8.6 ± 0.4 ms; right bar; P < 0.001). In addition, the representative sweeps (Fig. 6A, right stack) exhibit a graded shift in latencies from base (top) to apex (bottom), which was confirmed in the scatter plots (Fig. 6B), revealing that cAMP exposure produced a linear gradient that was nonmonotonic under vehicle control and H-89 conditions. Thus, these latency data contrast with the effects of 8-Br-cAMP and H-89 on APmax and voltage threshold (Figs. 2 and 3, respectively), and interestingly, they show the opposite distribution pattern.

Figure 6.

Action potential latencies are tonotopically redistributed by 8-bromo-cAMP (8-Br-cAMP) but not H-89 in spiral ganglion neurons. A: overlay of 5–8 sweeps for H-89 (left column) and cAMP (right column) shows the range and heterogeneity of action potential latencies for basal (top row), middle (middle row), and apical (bottom row) neurons. Exposure to 8-Br-cAMP (right stack of sweeps) significantly reduced action potential latencies for basal and middle neurons. H-89 did not alter action potential latencies. Sweeps were normalized to the amplitude of the action potential. B: bar graphs and individual scatter plots for H-89, vehicle, and 8-Br-cAMP treatment conditions. Exposure to 8-Br-cAMP significantly shortened action potential latencies and produced a linear tonotopic gradient whereas H-89 did not significantly affect latencies or alter the nonmonotonic tonotopic gradient. Scatter plot values are as follows: H-89: Base, 10.0 ± 0.6 ms; Middle, 14.9 ± 0.9 ms; and Apex: 13.5 ± 0.8 ms; Vehicle: Base, 10.2 ± 0.7 ms; Middle: 12.7 ± 0.7 ms; Apex: 12.0 ± 0.5 ms; 8-Br-cAMP: Base, 7.3 ± 0.5 ms; Middle, 8.3 ± 0.3 ms; and Apex, 10.5 ± 0.9 ms. These latency data contrast with the effects of 8-Br-cAMP and H-89 on maximum number of action potential (APmax) and voltage threshold (Figs. 2 and 3, respectively), which show the opposite distribution pattern. The significant difference between vehicle control and the cAMP treatment group is indicated by ** (P > 0.01).

Action potential duration.

We also systematically examined another aspect of spiral ganglion neuron kinetics: action potential duration. This is a complex parameter, resulting from multiple components in the rising and falling phases of action potentials, which are captured in Fig. 7A by observing the differences between overlapping representative traces normalized for action potential amplitude. From the fastest to the slowest trace this compilation shows that at least three aspects of the spike are shaped differently: the action potential rise rate (Fig. 7C), the initial, rapid repolarization phase (Fig. 7D), and a later, slow repolarization phase (Fig. 7E). Although the effects on these action potential parameters at first appear somewhat complex due to the selective effects of each treatment condition, the overall kinetics are effectively faster with 8-Br-cAMP and slower with H-89. The action potential components were quantified using dV/dt [mV/ms vs. time (ms)] and phase plots (dV/dt vs. V; Fig. 7B). We assessed the rising phase (Vm Max—the maximum rate of rise), the rapid repolarization phase (Vm Min), and the slow repolarization phase. Under control conditions, the slope of the action potential rise was 236.6 ± 5.5 mV/ms (Fig. 7C). The results for neurons exposed to cAMP were statistically similar (240.5 ± 5.4 mV/ms; P = 1.00). However, H-89 exposure significantly slowed the action potential rise rate (199.4 ± 6.4 mV/ms; P < 0.001; Fig. 7C). The data for the initial, rapid repolarization phase of the action potential were similar. Exposure to H-89 significantly slowed the repolarization relative to vehicle control neurons (vehicle: −88.9 ± 2.1 mV/ms; H-89: −76.9 ± 2.5 mV/ms; P = 0.0008; Fig. 7D). Exposure to 8-Br-cAMP did not significantly alter the rapid repolarization (−82.1 ± 2.4 mV/ms; P < 0.0654; Fig. 7D). On the other hand, 8-Br-cAMP significantly increased the action potential slow repolarization relative to vehicle whereas H-89 did not [vehicle: −44.3 ± 2.7 mV/ms; cAMP: −53.6 ± 1.9 mV/ms (P = 0.0085); H-89: −42.7 ± 1.9 mV/ms (P = 0.8670); Fig. 7E]. These data show differential effects on the rapid phases (rise and fall), which are modified by PKA inhibition, and the slow repolarization phase of the action potential, which is affected by increased cAMP signaling.

Figure 7.

Component analyses reveals differential effects on action potential kinetics. A: overlay of scaled sweeps highlights the wide differences in action potential rising and falling phases of action potentials recorded during this study. B: representative phase plot used for the analyses in C–E which shows dV/dt (mV/ms) as a function of time (ms). The labels on the sweep refer to the points used for analysis. The inset shows the same dV/dt but as a function of voltage. The different phases of the action potential are indicated. C: bar graphs of averaged action potential rise rates by treatment condition are significantly different for H-89 versus vehicle but not 8-bromo-cAMP (8-Br-cAMP) vs. vehicle. Individual scatter plots for each cochlear location. H-89: Base, 189.6 ± 12.7 mV/ms; Middle, 193.4 ± 9.4 mV/ms; and Apex, 218.7 ± 10.3 mV/ms. Vehicle: Base, 213.6 ± 9.9 mV/ms; Middle, 253.2 ± 9.4 mV/ms; and Apex, 240.9 ± 7.7 mV/ms. 8-Br-cAMP: Base, 248.2 ± 10.2 mV/ms; Middle, 250.7 ± 8.7 mV/ms; and Apex, 218.6 ± 7.9 mV/ms. D: same analyses and basic finding as C but for the rapid repolarization phase of the AP. H-89: Base, −74.5 ± 4.4 mV/ms; Middle, −73.3 ± 3.9 mV/ms; and Apex, −84.1 ± 4.4 mV/ms. Vehicle: Base, −85.4 ± 3.3 mV/ms; Middle, −94.5 ± 4.5 mV/ms; and Apex, −86.5 ± 2.6 mV/ms. 8-Br-cAMP: Base, −92.0 ± 3.2 mV/ms; Middle, −79.4 ± 3.9 mV/ms; and Apex, −72.6 ± 4.3 mV/ms. E: same analyses as C but for the slow repolarization phase of the action potential (AP), which revealed a significant effect of 8-Br-cAMP but not H-89. H-89: Base, −46.6 ± 2.2 mV/ms; Middle, −35.4 ± 4.8 mV/ms; and Apex, −45.4 ± 5.1 mV/ms. Vehicle: Base, −49.8 ± 1.7 mV/ms; Middle, −47.9 ± 2.8 mV/ms; and Apex, −35.6 ± 7.0 mV/ms. 8-Br-cAMP: Base, −63.3 ± 2.3 mV/ms; Middle, −49.7 ± 1.7 mV/ms; and Apex, −45.0 ± 1.7 mV/ms. Significant differences between vehicle control and each treatment group in bar charts (C–E) are indicated by ** (P > 0.01).

The intrinsic components evaluated earlier all contribute to the action potential shape and, ultimately, its duration, which is critical for controlling the endogenous capacity of spike timing and number. To evaluate action potential durations for each treatment group, we measured the width of the action potential at the midpoint between the action potential peak and afterhyperpolarization nadir (Fig. 8A). As expected from the slower rates of the rapid depolarization and repolarization, H-89 exposure produced a significant increase in the average action potential duration across all spiral ganglion neurons (1.28 ± 0.04 ms; Fig. 8B, left bar; P = 0.0006) compared with the vehicle control neurons (1.11 ± 0.02 ms). A more abbreviated action potential duration in the 8-Br-cAMP treatment condition due to the significant difference noted for the slow repolarization might also have been expected, but this was not found (1.08 ± 0.04 ms; Fig. 8B, right bar; P = 0.7983). Thus, the longer action potential duration of neurons exposed to H-89 is likely mediated by voltage-gated ion channels that control the fast rising and falling phases of the AP. In addition, and distinct from other parameters, we noted that although the 8-Br-cAMP exposed neurons did not display a significant difference in action potential duration, the range and gain of this parameter simultaneously increased. This was achieved by differential trends across the tonotopic contour. Although not significant, basal neurons tended to become faster than vehicle control (Fig. 8B, base vehicle: 1.07 ± 0.03 ms; base cAMP: 0.93 ± 0.04 ms; P = 0.0907). Middle neurons were unchanged (middle vehicle: 1.11 ± 0.03 ms; middle cAMP: 1.10 ± 0.04 ms; P = 0.9807). Finally, apical neurons showed a trend toward longer action potential durations (Fig. 8, B and C; apex vehicle: 1.14 ± 0.02 ms; apex cAMP: 1.25 ± 0.09 ms; P = 0.3876). Thus, neurons exposed to 8-Br-cAMP not only exhibited a graded tonotopic distribution but also neurons from each region were differentially affected. The divergence in action potential durations caused by 8-Br-cAMP exposure is highlighted in Fig. 8C. As shown, action potentials from the base (black sweeps) were systematically faster than cAMP-exposed neurons from the apex (gray sweeps), which also exhibited substantial heterogeneity. This is a noteworthy finding because it shows that the enhanced range due to region-specific sensitivities to cAMP exposure also lends itself to an increase in gain for this parameter across the population of spiral ganglion neurons.

Figure 8.

Complex tonotopic effects of 8-bromo-cAMP (8-Br-cAMP) and H-89 on action potential durations. A: action potential duration was measured at ½ the distance between the action potential peak and the afterhyperpolarization nadir. B: bar graphs from all neurons per treatment group and scatter plots for H-89, vehicle, and 8-Br-cAMP. Strikingly, 8-Br-cAMP exposure significantly reduced durations in basal neurons, produced no change in middle neurons, and tended to prolong apical neuron durations. These differential effects of 8-Br-cAMP created a steep linear tonotopic gradient. Conversely, H-89 significantly prolonged durations relative to vehicle controls. Scatter plot values are as follows: H-89 (base): 1.26 ± 0.6 ms; H-89 (middle): 1.36 ± 0.8 ms; H-89 (apex): 1.21 ± 0.05 ms; vehicle (base): 1.70 ± 0.03 ms, vehicle (middle): 1.11 ± 0.03 ms, vehicle (apex) 1.14 ± 0.02 ms; 8-Br-cAMP (base): 0.93 ± 0.04 ms; 8-Br-cAMP (middle): 1.10 ± 0.04 ms; 8-Br-cAMP (apex): 1.25 ± 0.09 ms. The significant difference between vehicle control and the H-89 treatment group is indicated by ** (P > 0.01). C: overlay of all 8-Br-cAMP-exposed sweeps from the base (black traces) and apex (gray traces) highlight the heterogeneity and marked differences in the response to 8-Br-cAMP exposure from these two regions.

Yet, how do changes in dynamic range and gangliotopic reorganization affect the relationship between kinetic parameters across the neuronal population? To examine this issue, we plotted latency as a function of duration for each experimental condition (Fig. 9, A–C). We found that the strongest relationship between the two parameters was observed in neurons with elevated cAMP where the increased dynamic range of the duration measurements expanded the linear relationship (Fig. 9C, R² = 0.5901). When neurons were evaluated regionally within this condition, the latency versus duration functions were not equally linear. The strongest relationship was observed for the apical neurons (R² = 0.6468, Fig. 9D), compared with those from middle and basal regions (R² = 0.2383 and R² = 0.4436, respectively; not shown). We next evaluated whether the latency versus duration functions differed between adaptation groups in the apex (Fig. 9E). Interestingly, we found that UA/RA neurons formed a linear function having a high correlation (R² = 0.9922), whereas the SA neurons were more scattered. However, when the SA neurons were separated into three data sets, high correlations were noted in two groups that formed linear functions at the shorter and mid-duration levels (Fig. 9E; SA1, R² = 0.8053; SA2, R² = 0.9995, respectively). A single data point measured at a longer duration was classified as a third group (SA3, Fig. 9E, square). Overall, separating neurons by experimental condition, region, and adaptation class was revealing because the data suggest that in elevated cAMP, latency versus duration functions have the capability to encode action potential kinetics in multiple ways. Compared with the apical SA neurons that were parsed into multiple linear functions, apical UA/RA neurons formed a single function with a shallower slope (Fig. 9E). In conjunction with this, we did note that the UA neurons of the base and middle with elevated cAMP also formed linear functions (R² = 0.7893; middle R² = 0.9616, respectively; Fig. 9F), of which the slopes were most similar between the basal UA neurons and apical UA/RA neurons (m = 8.1 and m = 10.1, respectively; Fig. 9F). The results indicate that differential gangliotopic reorganization of adaptation groups in response to high cAMP signaling levels have the potential to coordinate dual coding strategies across the ganglion. This organization is also consistent with the overall distribution of neuron adaptation categories such that SA neurons are found in higher percentages in the apical and middle regions, whereas base neurons are composed predominately of UA neurons.

Figure 9.

Latency, a key electrophysiological correlate of stimulus timing, is precisely regulated according to location, adaptation class, voltage level, and current input. A–C: action potential latency at threshold plotted as a function of duration for H-89, Vehicle, and 8-bromo-cAMP (8-Br-cAMP) conditions, respectively. D: action potential latency plotted as a function of duration for neurons within the 8-Br-cAMP condition isolated from the apex show a higher R² value of the linear fit (R² = 0.6468) compared with the full population (R² = 0.5901). E: data plotted in D were separately fitted to adaptation classes. Fits to rapidly adapting (RA)/unitary adapting (UA) neurons showed that the R² value further increased (R² = 0.9922). Slowly adapting (SA) neurons were separated into three groups. Two groups formed linear distributions, which were separately fitted (R² = 0.8053, dashed line; R² = 0.9995, dotted line). F: linear fits to RA/UA adaptation classes separated into apex, middle, and basal groups (red, green, and blue, respectively) from C (R² = 0.9922, 0.9616, and 0.7893, respectively). G–I: action potential latency at threshold plotted as a function of voltage for H-89, Vehicle, and 8-Br-cAMP conditions, respectively. Data were fitted with linear functions. J–L: action potential latency at threshold plotted as a function of input current for H-89, Vehicle, and 8-Br-cAMP conditions, respectively. Data were best fitted with exponential functions. M: linear fits from G and I show the lower sensitively of neurons exposed to H-89 compared with those supplemented with 8-Br-cAMP. N: exponential fits from J and L show the reduced responsiveness of neurons supplemented with 8-Br-cAMP compared with those exposed to H-89. O: direct comparison of the latency vs. duration relationship fits for RA/UA neurons isolated from the apex and base (red and blue solid lines, respectively) compared with SA neurons isolated from the apex (red dashed lines) from E and F.

We evaluated action potential latency further because it is a clear electrophysiological correlate of acoustic timing, and it shows unique regulation within spiral ganglion neurons. Like action potential duration, the heterogeneity of latency is retained in axonally transmitted action potentials, but in distinction to duration, latency can be amplified by an order of magnitude in action potentials generated at the somata (53). Thus, it is important to keep in mind that while the levels presented herein represent those generated at the soma, they are highly correlated to onset profiles of axonally transmitted action potentials. The relationship between latency and neuron excitability was further evaluated by plotting latency against threshold voltage and current levels for the full population of neurons in each experimental condition (Fig. 9, G–L). Despite the diversity in individual values, the progressively lowered latencies were observed over threshold voltages that spanned similar ranges (Fig. 9, G–I). These same shifts in latency, however, spanned different current levels between conditions with higher current input required in the 8-Br-cAMP condition to achieve the faster latency responses (Fig. 9, J–L). When the lowest cAMP signaling levels in the H-89 condition were compared directly with the highest ones in the 8-Br-cAMP condition for threshold voltage (Fig. 9M) and current (Fig. 9N), an interesting picture emerged. Latency difference was represented by a clear step-wise downward threshold voltage shift with higher intracellular cAMP levels (Fig. 9M). This electrophysiological profile was consistent with a representation of distinct differences in action potential delay when comparisons are made at the same voltage level (Fig. 9M, vertical double arrow). In contrast, the latency difference was organized into a smooth function shifting to the right with higher cAMP levels (Fig. 9N, horizontal double arrow). This electrophysiological profile was distinct, as it retained the inverse relationship between action potential latency and input responsiveness while extending the dynamic range across conditions. It is of interest that higher current levels were required for increased intracellular cAMP levels because it could potentially counterbalance lower stimulus intensity levels associated with more rapid responses. Although this type of mechanism supports the concept of efferent modulation of binaural auditory nerve fiber activity levels (54), the electrophysiological analysis separating out different adaptation class responses showed that the process is more complex. There are likely at least two distinct mechanisms processed in parallel based on the analysis of latency versus duration for the cAMP condition. One, potentially at high resolution, restricted to the apical region (Fig. 9O, red dashed lines); the other at lower resolution, encompassing the full extent of the ganglion from the base to the apex (Fig. 9O, blue and red solid lines, respectively).

Analysis of spiral ganglion kinetics, show adaptation-dependent fine control over spiral ganglion firing patterns. Despite this complexity, however, there are two clear overarching conclusions that can be drawn when comparisons are made to the analysis of neuronal excitability. The first is that kinetics, like excitability, are oppositely controlled by high versus low cAMP levels. Neuron action potentials become faster with higher levels of cAMP, in contrast to slower action potentials that could possibly fail (55, 56) with lower levels of cAMP. The second is that similar to neuron excitability, altered kinetic parameters are also reconfigured into a graded tonotopic distribution. Thus, whether neuronal excitability or kinetics are altered, they are likely transmitted downstream in a tonotopically distributed organization.

DISCUSSION

Action potentials in the spiral ganglion are special because their intrinsically heterogeneous and precisely shaped analog waveforms are retained when conducted through myelinated distal processes (53). The transmission of these multiplexed signals indicates that differential kinetics in the primary auditory afferents are a paramount signaling element in the earliest stages of acoustic coding, likely contributing to presynaptic kinetic differences (57). Similarly, afferent thresholds are key to acoustic coding as they shape the responsiveness, sensitivity, and resolution of the system. To do so, primary auditory afferents possess endogenous threshold diversity, which can be dynamically modulated and regulated separately from kinetics. Herein, we tested the limits of this unique electrophysiological design by examining the neuroelectric effects of increased or decreased cAMP signaling on immature neuronal somata that display greater excitability and slower kinetics than their mature counterparts. Our rationale was to investigate neurons that display enhanced electrophysiological dynamic ranges to better detect differences and patterns of individual action potential parameters resulting from neuromodulation. Interestingly, we found that the system itself is designed to utilize a similar strategy, leading to the conclusion that in addition to orchestrating adaptable population encoding profiles (10), dynamic heterogeneity within the spiral ganglion can also enhance input differences.

The Spiral Ganglion as a Stimulus Difference Amplifier

Endogenous electrophysiological heterogeneity yields diverse neuroelectric profiles throughout the spiral ganglion providing a mechanism for neurons to respond differently to identical stimuli. Although this mechanism underlies processes such as intensity population coding, it can also be utilized for comparative mechanisms. If, for example, neurophysiological parameters were coordinated such that subtle differences between stimulus inputs were amplified, this same coding profile could instead enhance differences between specific acoustic attributes to improve neural detection beyond the physical constraints of the signal (58–60).

Neural coding aspects that are logically required to carry out such comparisons first depend on the basic neural substrate and its regulators. The neural substrate, represented by electrophysiological heterogeneity, is a prominent feature of the spiral ganglion both between- and within-adaptation classes and in developing and adult neurons (1–4, 6, 25, 37, 41, 44, 61–63). The prominent regulators are the profuse efferent neuromodulators and their specific receptors present at the postsynaptic membrane (11–13, 64). Relevant to the present studies, up- and downregulation of cAMP signaling levels represents the separate D1- and D2-like dopaminergic receptors in type I neurons (65), that act through G_s and G_i α subunits to either increase or decrease cAMP signaling, respectively (19).

In combination, the up- and downregulation of cAMP signaling precisely controlled each aspect of the action potential waveform resulting in a yin-yang relationship between and within the parameters of neuronal excitability and action potential kinetics (Fig. 10A). For example, neuron excitability can be directly increased (↑) or decreased (↓) resulting in an apparent endogenous sliding scale, yet through different mechanisms. Increased cAMP, for example, has a twofold effect. It places the RMP closer to an unchanged threshold voltage level (Fig. 10A) yet requires enhanced current to reach the threshold (Fig. 4A). This combination of properties, consistent with altered R_in (10), broadens the dynamic range of responsiveness without altering average threshold sensitivity. Decreased cAMP signaling, on the other hand, simply increases voltage threshold levels causing the population of neurons to be less sensitive (Fig. 10A), which is likely attributable, in part, to increased activity of Kv1.1 and Kv1.2 ion channel subunits (41, 66).

Figure 10.

Spiral ganglion neuron timing and excitability features are mechanistically separable. A: the shape of the action potential is finely controlled in 8-bromo-cAMP (8-Br-cAMP) and H-89 treatment groups. B: two different neuronal recordings in which action potential timing was identical and yet the voltage thresholds were different. We found that voltage thresholds were significantly depolarized in H-89 versus vehicle control neurons (−41.8 ± 0.7 mV vs. −45.8 ± 0.6 mV, respectively; P < 0.001, Student’s t test), whereas action potential latency was constant (vehicle: 12.7 ± 0.4 ms, H-89: 13.0 ± 0.4 ms; P = 0.6). C: the opposite scenario as B. Example recordings of two neurons that possessed identical voltage thresholds, yet their action potential latencies were very different. Because the voltage thresholds of spiral ganglion neurons did not change after exposure to 8-Br-cAMP, we grouped neurons by holding the threshold level constant for both vehicle and 8-Br-cAMP treated neurons (−43.5 ± 0.4 mV and −43.4 ± 0.6 mV, respectively; P = 0.9) and then measured the action potential latencies. A significant difference was noted between the action potential latencies (vehicle: 11.8 ± 0.5 ms, cAMP: 8.4 ± 0.5 ms; P < 0.001). D: neuronal sensitivity is regulated by both threshold voltage level and resting membrane potential (RMP), each have differential responses to 8-Br-cAMP and H-89. E: shifted latencies and expanded ranges of action potential duration are both orchestrated by 8-Br-cAMP. The color spectrum is used to denote the fastest neurons (purple) to the slowest neurons (red).

The yin-yang relationship extends to action potential fine structure. The action potential waveform became faster (←) or slower (→), suggestive of a sliding scale, but this time for kinetics, which also acts through diverse mechanisms. Increased cAMP signaling, for example, had a twofold effect. It reduced action potential latency (←), while expanding the dynamic range of action potential duration (Fig. 10A), likely due to increased activity of low-voltage activated Ca²⁺ channels and interactions between high-voltage activated Ca²⁺ and BK channels, respectively (25, 63). This combination of effects linearized the UA/RA relationship between action potential latency and duration across the ganglion (Fig. 9F), while retaining the local relationships between SA neurons in the apex at a higher apparent resolution (Fig. 9E). Decreased cAMP signaling, on the other hand, prolongs action potential duration by slowing the rapid rising phase of the action potential (→), consistent with reduced Na⁺ channel activation (55). This mechanism likely contributes to inhibiting action potential firing at the bouton (56), which could ultimately prolong action potential latency by an order of magnitude should the resulting excitatory postsynaptic potentials (EPSPs) reach threshold at the somatic membrane or a distant node of Ranvier (53, 67, 68). Thus, action potential latency also has the capacity to display quite a wide dynamic range in the spiral ganglion.

The yin-yang relationships orchestrated by cAMP signaling supports previous observations that neuron excitability and kinetics are separately regulated (1, 10). In the context of neuromodulation, however, differential regulation can be particularly impactful as it permits the system to change one intrinsic parameter at a time. Thus, a defined aspect of the action potential waveform can be altered while related parameters remain constant. This means that the expected electrophysiological relationships measured intracellularly can be altered when compared intercellularly. For example, the threshold voltage levels that significantly increased due to reduced cAMP signaling occurred without affecting average action potential latencies (Fig. 10B). Conversely, reductions in latency due to enhanced cAMP signaling occurred without a commensurate shift in threshold voltage level (Fig. 10C). Consequently, the flexibility of cAMP-mediated shifts in voltage level relative to kinetics augurs against standard input/output population coding. Rather, it is more consistent with a comparator function where changes in a single attribute can be detected against an electrophysiological background that is otherwise similar or identical.

Interestingly, we found that parameters modified by up- or downregulation of cAMP were also predictably reorganized tonotopically (Fig. 10A, asterisks). For example, elevated threshold levels attributable to decreased cAMP signaling shifted the nonmonotonic, U-shaped function that typifies the vehicle condition, in which neurons isolated from the mid-frequency region display the greatest sensitivity (52), to a tonotopically graded profile (Fig. 10D, threshold). Conversely, the nonmonotonic, inverted U-shaped RMP function, in which mid-frequency neurons have RMPs closer to threshold (52) shifted this region apically (Fig. 10D, RMP). Furthermore, the subtle tonotopically graded profiles of action potential latency and duration (62) became more pronounced with cAMP upregulation (Fig. 10E).

The tonotopic reorganization of action potential duration is noteworthy, as its average value remained relatively constant, yet its dynamic range greatly increased. This observation is consistent with immunocytochemical analysis of Ca²⁺ and BK ion channel densities in postnatal cultures (62, 63). cAMP actions to activate excitable voltage-gated ion channels can increase Ca²⁺ influx leading to prolonged action potential durations. In the basal neurons, however, this same Ca²⁺ influx can activate the higher density of BK channels causing a faster repolarization instead. Despite having a similar distribution pattern of BK channels in adult spiral ganglion tissue sections (25), recordings from more mature neurons showed uniformly faster profiles with reduced variance restricted to RA neurons (10). The kinetic differences between the two age groups are likely due to increased ion channel densities during maturation (1, 3). Because there is a substantial calcium contribution to spiral ganglion action potential profiles, multiple factors can alter their kinetics beyond channel density, including phosphorylation of BK by PKC to reduce channel activity (69, 70). In this regard, metabotropic receptors associated with G_q α subunits that activate PKC and inositol (1,4,5)-trisphosphate (IP3), which also affect intracellular calcium levels, are indeed present along with dopaminergic receptors on the spiral ganglion postsynaptic membrane (11). Thus, co-activation of metabotropic receptors linked to G_s and G_q α subunits could potentially transform the properties of mature neurons. Nevertheless, whether averaged levels shift significantly or the dynamic range increases, each parameter change is accompanied by cochleotopic reorganization, thus conforming to a configuration where patterns are established relative to the sensory end organ (71, 72).

Summary

We found that by shifting the intracellular signaling levels of cAMP, endogenous heterogeneity is altered to produce parameter-specific changes in spiral ganglion action potential waveforms. But these changes are not random. When considering action potential kinetics and threshold sensitivity as electrophysiological correlates of the acoustic stimulus parameters timing and intensity, the yin-yang organization makes sense. When cAMP levels are reduced, for example, the electrophysiological correlates of signal delays are paired with reduced intensity, whereas when cAMP levels are increased, the same electrophysiological correlates are reversed. Moreover, the neural modulation and flexibility inherent in regulating neuroelectric responses in the primary afferents indicate that these mechanisms are transient, as observed at higher levels (23, 58), and therefore activated by the system when required. In total, our observations show how endogenous plasticity may underlie amplification of signal differences, revealing another encoding design feature within the spiral ganglion attributable to dynamic heterogeneity.

GRANTS

This study was supported by National Institutes of Health (NIH) Grant NIDCD RO1 DC01856 (to R. L. Davis).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.A.C. and R.L.D. conceived and designed research; R.A.C. and Z.Q.W. performed experiments; R.A.C., Z.Q.W., J.P.-M., and R.L.D. analyzed data; R.A.C., Z.Q.W., M.R.P., and R.L.D. interpreted results of experiments; R.A.C., J.P.-M., and R.L.D. prepared figures; R.A.C. and R.L.D. drafted manuscript; R.A.C., M.R.P., and R.L.D. edited and revised manuscript; R.A.C., Z.Q.W., J.P.-M., M.R.P., and R.L.D. approved final version of manuscript.