Dendritic HCN Channels Shape Excitatory Postsynaptic Potentials at the Inner Hair Cell Afferent Synapse in the Mammalian Cochlea

Abstract

Synaptic transmission at the inner hair cell (IHC) afferent synapse, the first synapse in the auditory pathway, is specialized for rapid and reliable signaling. Here we investigated the properties of a hyperpolarization-activated current (I_h), expressed in the afferent dendrite of auditory nerve fibers, and its role in shaping postsynaptic activity. We used whole cell patch-clamp recordings from afferent dendrites directly where they contact the IHC in excised postnatal rat cochlear turns. Excitatory postsynaptic potentials (EPSPs) of variable amplitude (1–35 mV) were found with 10–90% rise times of about 1 ms and time constants of decay of about 5 ms at room temperature. Current–voltage relations recorded in afferent dendrites revealed I_h. The pharmacological profile and reversal potential (−45 mV) indicated that I_h is mediated by hyperpolarization-activated cyclic nucleotide-gated cation (HCN) channels. The HCN channel subunits HCN1, HCN2, and HCN4 were found to be expressed in afferent dendrites using immunolabeling. Raising intracellular cAMP levels sped up the activation kinetics, increased the magnitude of I_h and shifted the half activation voltage (V_half) to more positive values (−104 ± 3 to −91 ± 2 mV). Blocking I_h with 50 μM ZD7288 resulted in hyperpolarization of the resting membrane potential (∼4 mV) and slowing the decay of the EPSP by 47%, suggesting that I_h is active at rest and shortens EPSPs, thereby potentially improving rapid and reliable signaling at this first synapse in the auditory pathway.

INTRODUCTION

To perform tasks such as the localization of sound in space, neurons in the auditory pathway are specialized to accurately preserve timing information within sound signals (Oertel 1999; Trussell 1999). Several pre- and postsynaptic mechanisms enabling rapid and reliable transmission at auditory synapses have been described. Presynaptically, large calyceal structures release vesicles from many release sites synchronously, resulting in large excitatory postsynaptic potentials (EPSPs) that reliably activate action potentials (APs) (Schneggenburger and Forsythe 2006). Postsynaptically, rapid kinetics of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors result in brief excitatory postsynaptic currents (EPSCs) (Gardner et al. 1999, 2001; Parks 2000; Raman et al. 1994). Voltage-gated ion channels active near the resting membrane potential decrease the membrane resistance of the postsynaptic neurons and shorten the membrane time constant (Magee and Johnston 2005). This mechanism keeps EPSPs brief and prevents temporal summation of synaptic events. A hyperpolarization-activated cation channel (I_h) has been described in auditory brain stem neurons as one of the ion channels serving this role (Banks and Smith 1992; Golding et al. 1995; Rothman and Manis 2003b). To further shape EPSPs, auditory neurons can receive modulatory inputs that either open receptor-coupled ion channels (Funabiki et al. 1998; Smith et al. 2000) or regulate I_h via G protein coupled signaling pathways (Banks et al. 1993; Yamada et al. 2005).

The first synapse in the auditory pathway, the synapse between the inner hair cell (IHC) and auditory nerve fiber, also uses highly specialized mechanisms to preserve timing information (Fuchs 2005; Glowatzki et al. 2008; Moser et al. 2006). The auditory nerve fiber receives input from only one IHC via a single dendrite and large EPSCs are activated by multivesicular release at this ribbon-type synapse (Glowatzki and Fuchs 2002; Goutman and Glowatzki 2007; Grant et al. 2010; Keen and Hudspeth 2006; Li et al. 2009). Similar to EPSCs recorded from auditory brain stem synapses, AMPA-mediated EPSCs at this synapse are brief (Glowatzki and Fuchs 2002; Grant et al. 2010). However, not much is known regarding the expression pattern of voltage-gated ion channels in afferent dendrites and their involvement in shaping postsynaptic activity. In a first survey, using voltage-clamp recordings from afferent dendrites of the postnatal rat cochlea, we identified several voltage-gated conductances in afferent dendrites (Na⁺, Ca ²⁺, K⁺ conductances) including a hyperpolarization-activated conductance.

Here we focus on the characterization of I_h in IHC afferent dendrites. We find that I_h is mediated by HCN channels. I_h is active at rest and is modulated by intracellular levels of cyclic adenosine monophosphate (cAMP). I_h shortens the EPSP and is thus a good candidate for enabling rapid and reliable signaling at this first synapse in the auditory pathway.

METHODS

Animal protocols were approved by the Johns Hopkins University Animal Care and Use Committee. Rats (Sprague–Dawley; Charles River, Wilmington, MA) were anesthetized (pentobarbital 0.045 mg·g⁻¹, administered intraperitoneally or by isoflurane inhalation) and decapitated and cochleae were quickly removed from temporal bones.

Electrophysiological recordings

Excised apical cochlear turns of 7- to 14-day-old rats were placed into a chamber under an upright microscope (Axioskop2 FS plus, Zeiss, Oberkochen, Germany) and superfused with external solution at 1–3 ml/min (chamber volume ∼2 ml). IHCs and contacting afferent dendrites were visualized on a monitor via a ×40 water immersion objective, ×4 magnification, differential interference contrast optics using a green filter, and a NC 70 Newvicon camera (Dage, MTI, Michigan City, IN). The pipette solution consisted of (in mM): 135 KCl, 3.5 MgCl₂, 0.1 CaCl₂, 5 EGTA, 5 HEPES, and 0–2.5 Na₂ATP; or 135 KCl, 3.5 MgCl₂, 0.1 CaCl₂, 5 EGTA, 5 HEPES, 4 Na₂ATP, and 0.2 Na₂GTP; 290 mOsm, pH 7.2 (KOH). In some recordings in which the effect of cAMP was tested, the pipette solution contained (in mM):131 KCl, 1 MgCl₂, 5 EGTA, 5 HEPES, 5 Na₂ATP, and 10 Na₂-phosphocreatine. In addition, a small number of recordings (n = 6) was performed using a pipette solution containing (in mM): 110 K-methanesulfonate, 20 KCl, 5 EGTA, 5 HEPES, 0.1 CaCl₂, 5 Na₂-phosphocreatine, 4 MgATP, and 0.3 Tris-GTP. No significant differences were found in basic membrane properties such as input resistance and membrane time constant of the afferent fiber between recordings with KCl- or K-methanesulfonate-based pipette solutions. The external solution consisted of (in mM): 5.8 KCl, 155 NaCl, 1.3 CaCl₂, 0.9 MgCl₂, 0.7 NaH₂PO₄, 5.6 glucose, and 10 HEPES; 300 mOsm, pH 7.4 (NaOH). Drugs were dissolved daily in the external solution to their final concentrations from frozen stocks. Application of drug solutions was performed using a gravity-driven flow pipette (100 μm diameter) placed near the row of IHCs, connected with a VC-6 channel valve controller (Warner Instruments, Hamden, CT). 4-Ethylphenylamino-1,2-dimethyl-6-methylaminopyrimidinium chloride (ZD7288), 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), and 4-aminopyridine (4-AP) were purchased from Tocris Bioscience (Ellisville, MO) and tetrodotoxin (TTX) either from Alomone (Jerusalem, Israel) or Sigma (St. Louis, MO). All other chemicals were purchased from Sigma.

Recording pipettes were fabricated from 1 mm borosilicate glass (WPI, Sarasota, FL). Pipettes were pulled with a multistep horizontal puller (Sutter Instrument, San Rafael, CA) and fire-polished (10–15 MΩ). Pipettes were coated with Sylgard (Dow Corning, Midland, MI). Experiments were done at 22–25°C. Recordings were performed with a Multiclamp 700A or 700B amplifier (Molecular Devices, Sunnyvale, CA), pClamp version 9.2, and a Digidata 1322A board, digitized at 50 kHz, and filtered at 10 kHz.

In voltage-clamp mode, series resistance (R_s) was calculated from capacitative current responses to a 10 mV voltage step (−84 to −94 mV). The capacitative current responses were fitted with a sum of two or more exponential equations. The fastest component of the fit was considered to represent the capacitative current for the membrane area near the recording electrode (C_fast) and the slower components for current spreading further along the afferent nerve fiber (C_slow) (Llano et al. 1991). From the fastest component, the voltage-clamp time constant (36 ± 12 μs, n = 38), membrane capacitance C_fast (1.32 ± 0.43 pF), and R_s (30 ± 12 MΩ) were derived. Voltage-clamp data were discarded if R_s was >50 MΩ. Most I_h currents were <300 pA and, with R_s at about 30 MΩ, the estimated voltage error was <9 mV.

Assuming that the dendrite is formed like a cylinder and with a specific capacitance of 1 μF/cm² and a diameter of 1 μm (most likely an overestimation), the first component corresponds to the dendrite at a length of about 40 μm. This distance covers the extent of the HCN channel expression along the terminal; the HCN specific labeling ceases in the region of the first heminode of the peripheral dendrite, about 50 μm away from the afferent contacts with the IHC. Synaptic currents are generated within <3 μm of the pipette tip as the tip is directly positioned on the bouton ending. Therefore voltage-clamp conditions for recording synaptic currents and HCN channels should be sufficient. The input resistance (R_in) of afferent dendrites (394 ± 253 MΩ, n = 173) was determined in voltage clamp, with voltage steps from −64 to −84 mV.

In current-clamp mode, errors due to R_s were compensated using bridge balance and pipette capacitance neutralization. Membrane voltage responses to −10 pA current steps were fitted with a monoexponential equation and provided a membrane time constant (τ_m) of 3.97 ± 1.81 ms (n = 10). τ_m measured in current-clamp mode is larger than τ_m estimated from voltage-clamp data (0.5 ms) when only C_fast is used to estimate cell capacitance. This happens most likely because in current-clamp mode the effective cell capacitance is not limited to C_fast but also includes C_slow.

Liquid junction potentials (4 mV for KCl-based and 9 mV for K-methanesulfonate-based pipette solution) were corrected off-line. Data were analyzed off-line using pClamp version 9.2 (Axon Instruments, Union City, CA), Minianalysis (Synaptosoft, Decatur, GA), and Origin 7.5 (OriginLab, Northampton, MA). For statistical comparisons Sigmastat 3.5 (Systat Software, San Jose, CA) was used. Statistical significances of irreversible drug effects (ZD7288) were tested using a paired t-test. Effects of ZD7288 on the EPSP waveform (measurements were taken at three different conditions) were tested using one-way repeated measures ANOVA followed by Student–Newman–Keuls test. Effects of reversible drugs (CsCl and BaCl₂) were tested using one-way repeated measures ANOVA followed by Student–Newman–Keuls test. Effects of cAMP analogs on I_h amplitude and activation kinetics were tested using two-way repeated measures ANOVA one factor repetition. Effects of cAMP on the V_half and slope factor of I_h activation curves were tested using Student's t-test. Values are presented as means ± SD.

Immunolabeling

Cochleae from 9- to 10- and 20- to 21-day-old rats (P9–P10, P20–P21) were perfused through the round and oval windows with cold 4% paraformaldehyde prepared in phosphate buffered saline (PBS), pH 7.4, and then postfixed for 1 h at 4°C under agitation. Additionally, for HCN1 immunodetection, cochleae were rinsed three times in PBS and incubated 15 min in methanol at −20°C. Thereafter, preparations were washed three times in PBS and the cochleae were microdissected to facilitate access of the antibodies to the tissue. Whole-mount preparations were incubated for 1 h at room temperature in a blocking and permeabilizing solution (PBS with 20% of either normal goat serum or donkey serum and 0.3% Triton X-100) and were then incubated overnight at 4°C with the primary antibodies diluted in the same solution. After three 15 min washes in PBS, samples were incubated for 1 h at room temperature with fluorescently labeled secondary antibodies diluted at 1:800 in PBS with 10% of either normal goat serum or donkey serum and 0.15% Triton X-100. Samples were then rinsed once in PBS with 10% of either normal goat serum or donkey serum and 0.15% Triton X-100 and twice with PBS (15 min each, at room temperature) before the organs of Corti were mounted on slides using FluorSave mounting medium (Calbiochem, San Diego, CA). Specific labeling was initially examined with an Axio Observer inverted microscope (Zeiss) and further detailed images were obtained using a confocal laser scanning microscope (LSM 510 META, Zeiss) with ×20 air and ×100 oil objectives (optical section steps of 0.25 and 0.20 microns, respectively). Analysis and reconstruction were carried out using LSM Image Examiner (Zeiss) and Volocity 4.2.1 software (Improvision, Waltham, MA). No labeling was observed when the primary antibodies were omitted. Again, no labeling was observed when the primary antibodies were preabsorbed onto target peptides, except in the stereocilia of the sensory hair cells, which displayed staining using HCN1 and HCN4 antibodies, when the target peptides were used at the recommended concentration or at a 25-fold (HCN1) or 5-fold (HCN4) higher concentration. For HCN2, no test for preabsorption with a peptide was performed. For HCN3 a shorter fixation (15 min) was also tested.

Antibodies

Rabbit polyclonal antibodies against HCN1 and HCN4 (Alomone) were used at dilutions of 1:200 and 1:400, respectively. Rabbit polyclonal antibodies against HCN3 (Alpha Diagnostic International, San Antonio, TX) were used at dilutions of 1:200 to 1:20. Contrary to the antibodies against HCN3 from Chemicon (Temecula, CA) and Alomone, this antibody does not show cross reactivity for hHCN1, hHCN2, or hHCN4 (Kouranova et al. 2008). Monoclonal antibodies against HCN2 and HCN3 from UC Davis/National Institute of Neurological Disorders and Stroke (NINDS)/National Institute of Mental Health (NIMH) NeuroMab Facility (Davis, CA) were used at 1:25. Mouse monoclonal and rabbit polyclonal antibodies against recombinant rat calretinin (Chemicon) were diluted at 1:1,000. Guinea-pig serum against VGLUT3 was kindly provided by Dr. Robert H. Edwards' laboratory (Department of Physiology, School of Medicine, University of California, San Francisco) and used at a 1:1,000 dilution. Alexa Fluor 488 F(ab′)₂ fragment of goat anti-rabbit and Alexa Fluor 488 donkey anti-mouse IgG, Alexa Fluor 555 goat anti-mouse and Alexa Fluor 594 donkey anti-rabbit IgG, and Alexa Fluor 633 goat anti-guinea pig IgG (Molecular Probes, Eugene, OR) were used as secondary antibodies.

RESULTS

Voltage-gated ion channels in IHC afferent dendrites

To characterize voltage-gated conductances in IHC afferent dendrites, we used whole cell recordings from afferent dendrites directly where they contact the IHC. Recordings were performed in acutely excised apical cochlear turns from 7- to 14-day-old rats at room temperature. Resting membrane potentials of afferent dendrites were typically about −64 mV. We assume that the peripheral neurites of the recorded auditory nerve fibers were intact and connected with their spiral ganglion somata for the following reasons: First, in preparations where we separated the spiral ganglion from the cochlear coil, afferent dendrites were severely swollen and no recordings could be achieved. Second, in nine of nine experiments, where a fluorescent dye, Alexa Fluor 488 hydrazide salt (10 μM), had been included in the pipette solution and recordings had lasted >10 min, the unbranched afferent fibers could be traced back from the IHC toward the spiral ganglion for 200–500 μm (Supplemental Fig. S1)¹ and, in two cases, the fluorescent marker had reached the spiral ganglion somata at a distance of 450–500 μm from the IHC.

Our study focused on the characterization of I_h. However, because voltage-gated conductances have not been described for IHC afferent dendrites, in the following paragraph we will briefly summarize the different conductances that were observed in response to voltage step protocols. From a holding potential of −84 mV, voltage steps between −104 and −4 mV were applied (Fig. 1A). Voltage-gated sodium currents were found in 114 of 155 afferent dendrite recordings. Sodium currents often “escaped” the voltage clamp (Fig. 1A, inset). This is not surprising because the recording site is at the very tip of the unmyelinated afferent fiber ending and the AP initiation site is most likely located further away along the myelinated peripheral process (Hossain et al. 2005; Lacas-Gervais et al. 2004; McLean et al. 2009). For this study sodium channels were not further investigated, but rather blocked with 1–2 μM TTX (Fig. 1B). Additionally, small calcium currents that could be blocked with 200 μM CdCl₂ were detected in some recordings (data not shown). Because these small currents did not interfere with questions asked in this study, no effort was made to block them.

Fig. 1.

Current–voltage (I–V) relation recorded in an inner hair cell (IHC) afferent dendrite. Current responses to voltage steps in the absence (A) and the presence (B) of 1 μM tetrodotoxin (TTX). Voltage step protocol (inset in A): 200 ms voltage steps from −104 to −4 mV in 10 mV increments, from a holding potential of −84 mV. A: rapidly activating and inactivating sodium currents that sometimes escaped the voltage clamp (expanded trace shown in inset) were blocked by 1 μM TTX (B). C: current responses at 20 ms (solid circle) and 200 ms (open triangle) into the voltage steps after leak subtraction. A slowly activating inward current (i.e., a hyperpolarization-activated current [I_h]) was found at voltage steps to −94 mV or more negative potentials. Fast activating outward currents (within 5 ms) were activated at −64 mV and more positive voltages.

During depolarizing voltage steps, outward currents were observed in all 161 afferent dendrites recorded (Fig. 1). Outward currents reached their maximum within about 5 ms. The current–voltage (I–V) relations showed that outward currents activated at potentials as low as −64 mV (Fig. 1, B and C). We tested the effects of 4-AP (2–4 mM) and tetraethylammonium (TEA, 10–30 mM), drugs previously shown to inhibit the low-voltage activating potassium current (I_KL) and the high-voltage activating potassium current (I_KH), respectively, in spiral ganglion neurons (Szabo et al. 2002) and auditory brain stem neurons (Bal and Oertel 2001; Brew and Forsythe 1995; Cao et al. 2007; Manis and Marx 1991; Rathouz and Trussell 1998; Reyes et al. 1994; Rothman and Manis 2003a). 4-AP-sensitive and TEA-sensitive currents were observed in afferent dendrites and exhibited similar voltage dependent profiles to I_KL and I_KH, respectively. These conductances are still under investigation.

During hyperpolarizing voltage steps, a slowly developing inward current was found in 69 of 79 afferent recordings (Fig. 1, B and C). Voltage dependence and activation kinetics of this inward current were reminiscent of I_h recorded in dissociated spiral ganglion somata (Chen 1997; Mo and Davis 1997b).

I_h in afferent dendrites is mediated by HCN channels

To test whether I_h currents are mediated by HCN channels, we monitored I_h during the application of different blockers. In response to repeatedly applied negative voltage steps from −84 to −124 mV, an instantaneous inward current (I_inst) was followed by a slowly developing inward current (I_0.5s) (Fig. 2C). I_inst is partially blocked by I_h blockers and consists of a mixture of I_h and other conductances (see Fig. 2, A–C) (Bal and Oertel 2000; Rodrigues and Oertel 2006). We therefore report only on the amplitude of the slowly developing component of I_h (I_0.5s − I_inst). I_h currents showed some run-down during whole cell recordings and test protocols were designed accordingly. CsCl (2 mM) reversibly inhibited inward currents by 84 ± 11% (n = 8), whereas 2 mM BaCl₂ caused a reversible inhibition by 14 ± 8% (n = 4) (Fig. 2, A, B, D, E, and F). This combination of a strong Cs⁺ block and a weak Ba²⁺ block has been described for HCN channels, whereas inward rectifier potassium channels typically show a substantial Ba²⁺ block (Kubo et al. 2005; Robinson and Siegelbaum 2003; van Welie et al. 2005). Additionally, 50 μM ZD7288, an antagonist of HCN channels (Shin et al. 2001), irreversibly inhibited I_h by 93 ± 13% (n = 4) (Fig. 2, C, E, and F).

Fig. 2.

I_h currents in afferent dendrites are mediated by hyperpolarization-activated cyclic nucleotide-gated cation (HCN) channels. A–C: in afferent dendrite recordings, hyperpolarizing voltage steps from −84 to −124 mV were applied every 10 s for 0.5 s (in 1 μM TTX). Current responses consisted of an instantaneous inward current (I_inst) (most obvious in C) and a slowly developing inward current (I_0.5). A–C: representative traces before, during, and after application of 2 mM CsCl, 2 mM BaCl₂, or 50 μM ZD7288 (4-ethylphenylamino-1,2-dimethyl-6-methylaminopyrimidinium chloride). Fast inward deflections during the recordings in A to B represent excitatory postsynaptic currents (EPSCs). D and E: diary plots of the I_h amplitude during application of drugs. I_h amplitude was measured as I_0.5 − I_inst. The effects of CsCl and BaCl₂ were mostly reversible; the effect of ZD7288 was irreversible. F: percentage reduction in I_h amplitude during CsCl, BaCl₂, and ZD7288 application. The I_h amplitudes were determined from mean values of 5 consecutive traces in each condition. To compensate for rundown of I_h, current amplitudes in 2 mM CsCl or 2 mM BaCl₂ were compared with mean value of respective control and recovery. The I_h amplitude in 50 μM ZD7288 was compared with control.

The reversal potential of I_h was estimated using the following protocol: different conditioning voltages were applied for 3 s (−124, −104, and −84 mV), followed by 10 ms voltage ramps (from −144 to −74 mV) (Fig. 3F). The initial current response (∼3 ms) during the voltage ramp was discarded due to contamination with uncompensated capacitance currents. The linear portions of the current responses (−120 to −74 mV) were extrapolated to the region where the responses intersect, corresponding to an estimated reversal potential of I_h at −45 ± 2 mV (n = 4, Fig. 3F). This reversal potential is consistent with a mixed cation channel permeable for sodium and potassium and is in the range of reversal potentials described for HCN channels (Bal and Oertel 2000; Banks et al. 1993; Cao et al. 2007; Chen 1997; Cuttle et al. 2001; Mo and Davis 1997a; Moosmang et al. 2001; Rodrigues and Oertel 2006; Santoro et al. 2000; van Welie et al. 2005). In summary, the pharmacological profile and reversal potential suggest that I_h in afferent dendrites is mediated by HCN channels.

Fig. 3.

Properties of I_h in afferent dendrites. A and B: I_h currents in response to voltage steps (every 10 s for 3 s, from a holding potential of −64 mV to voltages between −144 and −54 mV in 10 mV increments; see inset). External solution with: TTX (1–2 μM), 4-aminopyridine (4-AP, 2 mM), tetraethylammonium (TEA, 10–30 mM), and 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX, 10 μM). A: in control solution. B: with 200 μM cyclic adenosine monophosphate (cAMP) intracellularly and additionally 200 μM 8-bromoadenosine-3′,5′ cyclic monophosphate (8-Br-cAMP) extracellularly. C: I–V relations in control (n = 3, black) and with cAMP analogs (n = 4, red). D: voltage dependence of I_h measured from tail currents. Tail current amplitudes were normalized and fit with a Boltzmann equation. V_half and slope factors were −104 ± 3 mV, 11 ± 1 in control (n = 3, black), and −91 ± 2 mV, 11 ± 1 in cAMP analogs (n = 4, red). E: activation kinetics of I_h currents. Current responses to voltage steps from −134 to −104 mV were fit with 2 exponentials, providing 2 time constants (τ_fast, τ_slow). Both time constants were significantly faster for currents recorded with cAMP analogs (n = 4, red) compared with control (n = 3, black). F: reversal potential of I_h. Conditioning voltages were applied for 3 s (to −124, −104, or −84 mV), followed by 10 ms voltage ramps (from −144 to −74 mV) (top traces: voltage commands; bottom traces: current responses to the commanding voltage ramps). The reversal potential was −45.5 mV for this recording.

HCN channels in afferent dendrites are modulated by cAMP

The I–V relation of I_h was recorded while blocking voltage-gated sodium channels (1–2 μM TTX), potassium channels (2 mM 4-AP and 10–30 mM TEA), and AMPA receptors (10 μM CNQX) (Fig. 3, A–E). From a holding potential of −64 mV, voltage steps were applied for 3 s, from −144 to −54 mV in 10 mV increments, and were followed by a voltage step to −74 mV, during which tail currents were recorded (Fig. 3A). I_h was measured as I_3s − I_inst. The I–V relation showed greater activation toward more negative voltages, with an I_h amplitude of 174 ± 48 pA at −144 mV (n = 3) (Fig. 3C, black trace). The voltage dependence of I_h was measured from tail currents and fitted by a Boltzmann equation (Fig. 3D, black traces). Half-maximum activation voltage (V_half) was −104 ± 3 mV and the slope factor was 11 ± 1 (n = 3). The activation curve shows that about 3% of I_h channels would be open at the resting membrane potential of the afferent dendrites (−65 mV). The activation range measured here corresponds to ranges measured under similar conditions for HCN channels in heterologous expression systems (Moosmang et al. 2001; Santoro et al. 2000) and I_h recorded in other neurons and spiral ganglion somata (Bal and Oertel 2000; Banks et al. 1993; Cao et al. 2007; Chen 1997; Cuttle et al. 2001; Mo and Davis 1997b; Rodrigues and Oertel 2006; van Welie et al. 2005).

Different HCN channel subunits activate on different timescales, with activation time constants ranging from milliseconds to seconds (Moosmang et al. 2001; Santoro et al. 2000). We measured the activation kinetics of I_h for current responses to voltage steps between −134 and −104 mV from a holding potential of −64 mV (Fig. 3E, black data points). The time course of activation was best fitted with two exponentials. At −134 mV, the time constants of the fast (τ_fast) and the slow (τ_slow) component were 66 ± 13 ms (n = 3) and 929 ± 40 ms, respectively [A_fast/(A_fast + A_slow) = 0.83 ± 0.09]. At more positive potentials, activation slowed down for both components.

It is well known that intracellular cyclic nucleotides can modulate HCN channel activity (Moosmang et al. 2001; Santoro et al. 2000; Wainger et al. 2001). Typically, with increased cAMP levels, the activation curve is shifted to more positive values and the activation kinetics speeds up. We therefore measured the I–V relation in the presence of cAMP analogs (Fig. 3B). cAMP (200 μM) was added to the pipette solution and, additionally, 200 μM of the membrane permeable 8-Br-cAMP (8-bromoadenosine-3′,5′ cyclic monophosphate) was added to the external solution. In the presence of cAMP analogs, the current amplitude of I_h increased significantly compared with control (from 174 ± 48 pA in control to 332 ± 91 pA in cAMP; at −144 mV, n = 4, P < 0.05 for all voltages tested) (Fig. 3, B and C, red traces). The activation curve shifted to more positive values, by about 12 mV (Fig. 3D, red traces), with V_half at −91 ± 2 mV (n = 4, P < 0.05) and no significant change in the slope factor (11 ± 1, n = 4, P = 0.611). Under these conditions, at −65 mV, about 9% of I_h would be active and have an estimated conductance of 1.4 nS (332 pA × 0.085/20 mV; reversal potential −45 mV). In the presence of cAMP analogs, activation time constants were significantly faster with a τ_fast of 23 ± 6 ms and a τ_slow of 175 ± 76 ms at a holding potential of −134 mV [n = 4, P < 0.05 for all voltages tested: A_fast/(A_fast + A_slow) = 0.85 ± 0.05] (Fig. 3E, red data points).

HCN channel subunit expression in IHC afferent dendrites

To investigate the expression pattern of HCN channel subunits in IHC afferent dendrites, we performed immunolabeling experiments with antibodies raised against the four known HCN channel subunits HCN1–HCN4 (Moosmang et al. 2001; Robinson and Siegelbaum 2003). We examined HCN distribution in the apical part of the cochlea in prehearing animals, at P9–P10, to match our recordings from afferent dendrites, and at P20–P21, to analyze HCN subunit expression in hearing animals. No immunolabeling of HCN3 above background was found and therefore the result could not be interpreted. Figure 4 shows overviews of apical cochlear turns at P9–P10 labeled for HCN1 or HCN4. At this age, we did not find HCN2 labeling. For better orientation, IHCs were labeled for the vesicular glutamate transporter VGLUT3 (Seal et al. 2008) (Fig. 4B). HCN1 and HCN4 specific labeling in the organ of Corti was found in the area directly below the IHCs, suggestive of a labeling of the unmyelinated peripheral processes of the afferent neurons. Additionally, HCN1 and HCN4 immunoreactivity was found in most of the spiral ganglion somata (Fig. 4 and Supplemental Fig. S2). The labeling was concentrated at the plasma membrane of the somata and in their processes within the ganglion close to the somata (Supplemental Fig. S2). At P20–P21, similar to the labeling pattern at P9–P10, labeling for HCN1, HCN4, and additionally for HCN2 was found directly below the IHCs (Fig. 5) and in the spiral ganglion somata (Supplemental Fig. S2).

Fig. 4.

HCN subunit expression pattern in the rat cochlea at P9. Three-dimensional reconstruction of confocal images from cochlear whole-mount preparations, apical turns. A: HCN1 labeling (green). B: HCN4 labeling (green). Vesicular glutamate transporter VGLUT3 (red) was used as a marker for IHCs. HCN1 and HCN4 labeling was found in the inner spiral plexus (isp) under the row of IHCs (ihc, arrowhead) as well as in the somata of spiral ganglion neurons (sgn). Scale bars: 50 μm.

Fig. 5.

HCN subunits are localized in afferent dendrites. A: 3-dimensional reconstruction of calretinin and HCN1-labeled whole-mount rat organ of Corti preparation, apical turn at P9. HCN1 labeling (green) is concentrated in the basolateral region of the IHCs. Calretinin (red) labels IHCs (ihc) and afferent dendrites. B–D: close-up view showing single confocal laser-scanning micrographs. As seen in the merged view (D), HCN1 (B) and calretinin (C) immunolabeling overlap in some afferent dendrites (arrowheads). E–M: single confocal laser-scanning micrographs of whole-mount organs of Corti preparations, apical turns at P21. Preparations were colabeled for HCN1, HCN2, or HCN4 (green) and calretinin (red). Arrowheads indicate examples of double-labeled afferent dendrites. Note examples of ringlike HCN labeling surrounding calretinin labeling (open arrowheads). Some fibers were labeled for HCN but not for calretinin (asterisks). Scale bars: 5 μm.

IHCs are surrounded by different cell types, including afferent fibers, efferent fibers, and supporting cells. To confirm that the afferent dendrites in the IHC area express the different HCN subunits, as indicated by our electrophysiological recordings, we performed double immunolabeling experiments for calretinin and HCN channel subunits. Calretinin is a calcium-binding protein involved in calcium buffering and transport. In the rat cochlea, calretinin has been detected both in the cytoplasm of IHCs and in most spiral ganglion neurons including their afferent dendrites (Dechesne et al. 1991, 1993). High resolution confocal imaging confirmed the coexpression of the HCN subunits and calretinin in the same afferent dendrites at P9–P10 and P20–P21 (Fig. 5). When assessing the labeling close to the IHCs at P9–P10, some afferent dendrites could be identified by their strong labeling with calretinin and were also found positive for HCN1 (Fig. 5, A–D, arrowheads) or HCN4 (data not shown). At P20–P21, HCN immunolabeling was more sharply defined, most likely reflecting a higher concentration or more intense clustering of the HCN channel subunits in afferent dendrites in the more mature organ of Corti. Most afferent dendrites positive for HCN1, HCN2, and HCN4 were also positive for calretinin labeling, indicating that HCN1, HCN2, and HCN4 are indeed localized in the afferent dendrites (Fig. 5, E–M, arrowheads). In some dendrites, HCN labeling appeared to surround the calretinin labeling in a ringlike fashion (open arrowheads), consistent with a clustering of the HCN channels in the plasma membrane. Some HCN positive dendrites were not calretinin positive (Fig. 5, asterisks). Similarly, not all HCN positive spiral ganglion somata were positive for calretinin labeling (data not shown). We conclude that the HCN subunits 1, 2, and 4 are expressed in afferent dendrites at the IHC afferent synapse.

EPSP waveforms in IHC afferent dendrites

In the CNS, dendritic I_h has been shown to shape EPSPs and to modulate excitability in neurons (Magee 2000). To investigate whether I_h affects the EPSP waveform in IHC afferent dendrites, we first characterized the EPSP waveform at rest. To isolate synaptic activity and block the generation of APs, experiments were performed in 1 μM TTX. Resting membrane potentials of afferent dendrites were −64 ± 8 mV (n = 34) when recorded in our standard extracellular solution containing 5.8 mM K⁺ and 1.3 mM Ca²⁺. Synaptic events occurred at a rate of 0.7 ± 0.9/s (n = 47; 13,575 synaptic events analyzed).

As shown before for EPSCs (Glowatzki and Fuchs 2002), some EPSPs appeared “multiphasic,” composed of multiple overlapping events. Other EPSPs appeared “monophasic,” presenting a monoexponential decay (Fig. 6F). To investigate rise and decay times, only monophasic events were analyzed. In three recordings, for direct comparison, both EPSCs and EPSPs were characterized in the same recording (Fig. 6, A–K). EPSCs were recorded at a holding potential of −94 mV and EPSPs at the resting membrane potential of the afferent dendrite in 5.8 mM K⁺. EPSPs were about three to four times slower than EPSCs, with a 10–90% rise time of 1.14 ± 0.17 ms compared with 0.36 ± 0.09 ms and a time constant of decay of 5.13 ± 1.69 ms compared with 1.16 ± 0.13 ms at room temperature (n = 3, 381 EPSPs; n = 8, 1,807 EPSCs analyzed) (Fig. 6, C, D, G, and H). For EPSPs, we found a wide range of amplitudes in every recording, varying from 1 to 35 mV; the shape of the amplitude distributions also varied widely. The amplitude distributions of two fibers were highly skewed, with median EPSP amplitudes of 2.3 and 2.4 mV, and most of the EPSPs were <10 mV (Fig. 6, I and J). For a third fiber, EPSP amplitudes spread more evenly within a range between 1 and 35 mV, with a median EPSP amplitude of 13.8 mV (Fig. 6K). However, for both EPSCs and EPSPs, rise and decay times did not change much over this wide range of amplitudes (Fig. 6, C, D, G, and H).

Fig. 6.

EPSCs and excitatory postsynaptic potentials (EPSPs) recorded at the IHC afferent synapse. A–K: whole cell recording from an afferent dendrite in the presence of 1 μM TTX showing EPSCs (A, B) (holding potential −94 mV) and EPSPs (E, F). B and F: overlaid representative traces of monophasic EPSCs and EPSPs on an expanded timescale. C, D, G, and H: 10–90% rise time (rise) or decay time constants (τ_decay) plotted against the EPSC or EPSP amplitude. Rise and τ_decay for EPSCs were 0.33 ± 0.14 and 1.24 ± 0.20 ms (324 EPSCs analyzed). Rise and τ_decay for EPSPs were 0.96 ± 0.12 and 3.81 ± 0.36 ms (241 EPSPs analyzed). EPSP waveforms remained relatively invariable over the wide range of EPSP amplitudes. I–K: EPSP amplitude distributions (bin size 1 mV) from 3 afferent dendrite recordings. Median EPSP amplitudes were 2.3, 2.4, and 13.8 mV and resting membrane potentials were −75, −56, and −68 mV, respectively. The number of events analyzed is indicated in each panel. L–N: whole cell current-clamp recording in the absence of TTX. A mixture of EPSPs and spikes was observed. M: overlaid representative traces of spikes on an expanded timescale. Spike threshold (arrow) was −47 mV for this recording. N: amplitude distribution (bin size 1 mV). A wide gap in amplitude histogram distinguishes spikes from EPSPs.

In auditory nerve fibers, the spike initiation zone is believed to be located close to the IHC afferent synapse, on the peripheral process (Hossain et al. 2005; Robertson 1976). A heminode with a high expression level of sodium channels, is <50 μm away from the afferent synapse (Hossain et al. 2005; McLean et al. 2009). The expression of sodium channels was also reported on the unmyelinated ending of the afferent fiber, peripheral to the heminode (Hossain et al. 2005). Indeed, when recording in the absence of TTX, at a resting membrane potential of −67 ± 5 mV (n = 7), 18 ± 22% of the events were spikes rather than EPSPs (range 0.5–50.5%, n = 4 afferent dendrites with >100 analyzable events each; total number of events analyzed: 829) (Fig. 6, L–N). Spikes were discriminated from EPSPs by their large and uniform amplitudes (39 ± 7 mV, n = 7, 171 spikes) that appeared as a separate group of events in the amplitude histograms (Fig. 6N). Additionally, about half of the spikes exhibited sudden slope changes during their rise (Fig. 6M, arrow), suggestive of a spike threshold at −50 ± 4 mV (n = 7, 97 spikes). The 10–90% rise time of the spikes (including the rise of the initiating EPSPs) was 2.29 ± 0.75 ms and the spike half-width was 2.83 ± 0.86 ms (n = 7, 171 spikes).

HCN channel activity shortens the EPSP waveform in IHC afferent dendrites

Next, we tested whether I_h is involved in shaping the EPSP waveform. To block I_h, we first applied CsCl, to allow for recovery of a possible effect. Without any cAMP analog added, 2 mM CsCl significantly increased τ_decay of the EPSP by 27 ± 17% (from 6.37 ± 0.64 to 8.04 ± 1.02 ms and, after washout, to 6.98 ± 0.53 ms; n = 4 fibers, 617 EPSPs analyzed, P < 0.05). Neither mean EPSP amplitude (control vs. CsCl: 13 ± 6 vs. 14 ± 8 mV, P = 0.839) nor 10–90% rise time (1.31 ± 0.20 vs. 1.48 ± 0.20 ms, P = 0.146) changed significantly. In the presence of cAMP analogs (200 μM cAMP intracellularly and, additionally, 200 μM 8-Br-cAMP extracellularly), application of CsCl increased τ_decay by 37 ± 31% (from 5.25 ± 1.37 to 7.15 ± 1.29 ms, n = 7, 878 EPSPs analyzed, P < 0.05) (Fig. 7, A–C). Again, EPSP mean amplitude (control vs. CsCl: 10 ± 4 vs. 11 ± 6 mV, P = 0.337) and 10–90% rise time (1.23 ± 0.20 vs. 1.31 ± 0.30 ms, P = 0.376) did not significantly change. Application of CsCl hyperpolarized the resting membrane potential of the afferent dendrite by about 1 mV (from −64 ± 3 to −65 ± 3 mV, n = 11, P = 0.002).

Fig. 7.

I_h shortens EPSPs in afferent dendrites. A–H: whole cell current-clamp recording from afferent dendrites. Recordings were done in the presence of cAMP analogs (200 μM 8-Br-cAMP extracellularly and additionally 200 μM cAMP intracellularly). A–C: EPSP waveform before and while blocking I_h with 2 mM CsCl. A: average EPSP waveform before (black) and during application of 2 mM CsCl (red). B: EPSP decay time constants (τ_decay) plotted against EPSP amplitudes before and while blocking I_h with 2 mM CsCl; control: τ_decay = 4.97 ms (black, 34 EPSPs); in 2 mM CsCl: τ_decay = 7.18 ms (red, 41 EPSPs). C: summarized results from 7 recordings. D–F: EPSP waveform before and during application of 50 μM ZD7288. D: average EPSP waveform before (black) and during application of 50 μM ZD7288 (magenta). E: EPSP decay time constants (τ_decay) plotted against EPSP amplitudes; control: τ_decay = 5.38 ms (black, 22 EPSPs), in 50 μM ZD7288: τ_decay = 8.30 ms (magenta, 16 EPSPs). F: summarized results from 6 recordings.

Immature IHCs express inward rectifier potassium channels that can be inhibited by extracellular Cs⁺ (Marcotti et al. 2003). Indeed, the frequency of synaptic events was higher in CsCl (control vs. CsCl: 1.01 ± 1.66 vs. 2.23 ± 2.44 events/s, n = 16, 4,357 EPSP or EPSCs analyzed, P = 0.025). To exclude the possibility that CsCl by some presynaptic effect changes the EPSC waveform and therefore affects the EPSP waveform, we tested the effect of CsCl on EPSCs and found no significant difference compared with control (control vs. CsCl: EPSC mean amplitude: 159 ± 91 vs. 168 ± 99 pA, P = 0.838; 10–90% rise time: 0.38 ± 0.14 vs. 0.38 ± 0.14 ms, P = 0.796; τ_decay: 1.24 ± 0.40 vs. 1.43 ± 0.57 ms, P = 0.383; n = 3, 626 EPSCs analyzed). These data suggest that changes in the EPSP waveform during CsCl application are due to its effect on the afferent dendrite.

Application of the irreversible HCN channel blocker ZD7288 exhibited greater effects than those of CsCl on the EPSP waveform, without changing the EPSP frequency (control vs. ZD7288: 0.73 ± 0.76 vs. 0.70 ± 0.71 events/s, P = 0.582) (Fig. 7, D–F). During application of 50 μM ZD7288 (in the presence of 200 μM cAMP intracellularly and, additionally, 200 μM 8-Br-cAMP extracellularly), τ_decay increased by 47 ± 24% (from 4.22 ± 0.98 to 6.19 ± 1.67 ms, n = 6, 593 EPSPs analyzed, P = 0.002) and the resting membrane potential was hyperpolarized by about 4 mV (from −63 ± 6 to −67 ± 7 mV, P < 0.001, n = 11). EPSP amplitude (8 ± 3 vs. 9 ± 4 mV, P = 0.378) and 10–90% rise time (1.40 ± 0.32 vs. 1.49 ± 0.31 ms, P = 0.082) did not significantly change. The I_h-induced hyperpolarization of the membrane potential could change the activity of additional ion channels that also might affect the EPSP waveform as it has been shown in a computational model of cochlear nucleus neurons (Rothman and Manis 2003b) or in recordings from dendrites of both hippocampal neurons (George et al. 2009) and frontal cortex pyramidal neurons (Day et al. 2005). Therefore to exclude possible effects on the EPSP waveform by hyperpolarization, membrane potentials were manually reset during ZD7288 application to their control values by constant current injection in three recordings. A significant increase of 30 ± 11% in τ_decay compared with control was still observed (Fig. 7F, τ_decay in control: 4.22 ± 0.98 ms; τ_decay in ZD7288: 5.25 ± 1.51 ms, P = 0.03). This result indicates that I_h shapes the EPSP waveform directly by contributing to the membrane resting conductance. The application not only of CsCl but also of ZD7288 increased τ_decay over the entire range of EPSP amplitudes (Fig. 7, B and E). We conclude that the activity of I_h in afferent dendrites shortens the EPSP waveform over the whole range of EPSP amplitudes.

Application of ZD7288 also caused minor changes in the spike waveform. The spike amplitude increased by 9 ± 4% (from 32 ± 5 to 35 ± 4 mV, P = 0.026; n = 3, 155 spikes analyzed) and spike half-width increased by 14 ± 8% (from 3.05 ± 0.53 to 3.45 ± 0.39 ms, P = 0.046). The 10–90% rise time did not change significantly (control vs. ZD7288: 1.24 ± 0.20 vs. 1.29 ± 0.19 ms, P = 0.757).

DISCUSSION

Properties of I_h in IHC afferent dendrites

Dendritic I_h has been shown to play an important role in setting firing rates in nerve fibers (Magee and Johnston 2005). Here we have used whole cell recordings to provide an initial characterization of I_h present in postnatal IHC afferent dendrites. The pharmacological profile of I_h, its reversal potential, its sensitivity to cAMP, and the expression of the HCN subunits 1, 2, and 4 in afferent dendrites indicate that I_h is mediated by HCN channels. In immunolabeling experiments we found HCN1 and HCN4 in postnatal afferent fibers and HCN1, HCN2, and HCN4 after hearing onset. No immunolabeling of HCN3 above background was found in either the spiral ganglion or the afferent dendrites at both ages and thus we cannot make a statement about the expression of HCN3. In adult guinea pig spiral ganglion somata, the expression of all four HCN subunits has been reported (Bakondi et al. 2009). The different result regarding HCN3 expression might be due to a difference in the species or to the different antibodies used. No specific HCN1 or HCN2 labeling has been found in mouse cochlear hair cells and mice lacking HCN1, HCN2, or both exhibited normal transduction currents (Horwitz et al. 2010).

We found each HCN subunit in a high percentage of afferent dendrites. It is therefore highly likely that heteromeric channels are formed. However, it is difficult to unequivocally deduce the composition of HCN channels in afferent dendrites from the properties of the recorded I_h. When measured in the same experimental conditions, homomeric HCN1 channels activate at more positive voltages and show faster kinetics than homomeric HCN2 and HCN4 channels. On the other hand, HCN1 shows the least sensitivity to cyclic nucleotides, followed by HCN2 to HCN4 (Moosmang et al. 2001; Santoro et al. 2000). Heteromeric HCN channels have been shown to exhibit intermediate properties compared with those of homomeric channels (Chen et al. 2001; Ulens and Tytgat 2001). Additionally, HCN channels with the same subunit composition exhibit variable characteristics when measured in different expression systems or tissues (Moosmang et al. 2001; Santoro et al. 2000; Wainger et al. 2001). Knock-out animals for HCN1, HCN2, and HCN4 do exist, although reports on their behavior do not mention any test of auditory function (Harzheim et al. 2008; Herrmann et al. 2007; Ludwig et al. 2003; Nolan et al. 2003, 2004; Stieber et al. 2003).

A range of activation voltages and cyclic nucleotide sensitivities has been described for I_h in studies on auditory neurons. For example, in bushy cells, a V_half of −94 to −84 mV and positive shift of V_half with cAMP were found (Cuttle et al. 2001; Leao K et al. 2006), whereas in octopus cells, a V_half of −65 mV and no shift with cAMP were reported (Bal and Oertel 2000). The properties we found for I_h in afferent dendrites are comparable to those found for I_h in spiral ganglion somata. In afferent dendrites, V_half was −104 mV, similar to V_half reported for spiral ganglion somata (−101 mV; Chen 1997; −78 to −122 mV; Mo and Davis 1997b). Both in spiral ganglion somata and afferent dendrites, V_half shifted positively in the presence of cAMP analogs.

In control conditions, we found 3% and, in the presence of cAMP analogs, 9% of I_h open at rest. The experimental conditions might underestimate the size of I_h in vivo; we recorded at room temperature and the amplitude of I_h has been found to be temperature sensitive in auditory neurons (Cao and Oertel 2005; Cao et al. 2007; Rodrigues and Oertel 2006). We recorded in immature cochleae, however, our immunolabeling experiments suggest a maturation of HCN channel subunit expression in hearing animals. For example, HCN2 labeling of afferent dendrites was found in 3-wk-old cochleae but not before hearing onset. HCN2 has a higher sensitivity to cAMP compared with that of HCN1 and thus properties of I_h may change with maturation. In our whole cell recordings, second messengers may be subject to wash-out and therefore the V_half measured may be more negative compared with that under in vivo conditions. Additionally, under in vivo conditions, input from lateral efferent fibers may up-regulate intradendritic cAMP levels (see last paragraph) and in vitro transmitter release from these fibers may be absent or abnormal.

Role of I_h in auditory neurons

Multiple studies have shown that auditory neurons involved in processing timing information express I_h. Examples include bushy cells (Cao et al. 2007; Leao K et al. 2006; Leao R et al. 2005; Rothman and Manis 2003a), octopus cells (Bal and Oertel 2000; Cao and Oertel 2005; Golding et al. 1995, 1999), and stellate cells (Rodrigues and Oertel 2006; Rothman and Manis 2003b) of the cochlear nucleus, medial superior olive neurons (Scott et al. 2005), nucleus laminaris neurons (Kuba et al. 2005; Yamada et al. 2005), lateral superior olive neurons (Barnes-Davies et al. 2004; Leao K et al. 2006), medial nucleus of the trapezoid body (MNTB) neurons (Banks et al. 1993; Cuttle et al. 2001; Leao K et al. 2006), the calyx of Held (Cuttle et al. 2001), and inferior colliculus neurons (Koch and Grothe 2003). Additionally, I_h was found in the somata of spiral ganglion neurons (Chen 1997; Mo and Davis 1997b). I_h can be partially active at rest, contributing to a lower membrane input resistance and thus shortening EPSPs. Shorter EPSPs reduce the time window of temporal summation and therefore improve coincidence detection (Golding et al. 1995; Koch and Grothe 2003; Yamada et al. 2005). Similarly, I_h expressed in dendrites of nonauditory CNS neurons shortened EPSPs and thereby reduced summation of synaptic activity (Day et al. 2005; Magee 2000; Magee and Johnston 2005). The nonuniform distribution of HCN channels along dendrites compartmentalized distal dendrites from the somata (Berger et al. 2003), limiting action potential (AP) back-propagation, and thereby decreasing hyperexcitability (Tsay et al. 2007; Ying et al. 2007). In motion-sensitive neurons in the superior colliculus (Endo et al. 2008), dendritic I_h was shown to keep AP timing at a minimum jitter and with short latencies. It is not surprising that similar mechanisms might be used at the IHC afferent synapse, as discussed in the following text.

Properties of EPSPs in afferent fibers and effects of I_h on EPSPs

Here we describe EPSP waveforms at the IHC afferent synapse. EPSPs have 10–90% rise times of about 1 ms and time constants of decay of about 5 ms. In comparison, the duration of EPSPs recorded in vivo in guinea pig (Palmer and Russell 1986; Siegel 1992; Siegel and Dallos 1986) was about 1–2 ms. The slower kinetics of EPSPs reported here could be due to the immature age of the recorded dendrites. AMPA mediated EPSCs at IHC afferent synapses speed up by a factor of 2, in a comparison of the immature age used in this study with 3-wk-old animals (Grant et al. 2010). Also, our recordings were performed at room temperature and AMPA receptor kinetics are temperature sensitive, with a Q₁₀ of >2 (Postlethwaite et al. 2007). The amplitudes and gating kinetics of voltage-gated ion channels, such as I_h and I_KL that are known to shorten EPSPs and APs in auditory neurons, are generally larger and faster in posthearing animals (Scott et al. 2005) and at body temperature (Cao and Oertel 2005; Cao et al. 2007; Rodrigues and Oertel 2006).

As shown for EPSCs (Glowatzki and Fuchs 2002; Grant et al. 2010), EPSP amplitudes covered an impressively wide range, between 1 and 35 mV. The median EPSP amplitudes varied widely between fibers, between 2 and 14 mV. Spikes recorded in afferent dendrites had a threshold at about −50 mV and roughly 20% of EPSPs activated spikes. Therefore for small EPSPs, postsynaptic summation of EPSPs might be necessary to reach the threshold for AP generation. In simultaneous recordings from hair cells and afferent dendrites, summation of EPSCs was observed during the early response to step depolarizations of the IHC membrane potential (Goutman and Glowatzki 2007). Our results showed that EPSPs slowed down by 47% when I_h was blocked by ZD7288. These data suggest that depending on the level of I_h active, the time window of EPSP summation may vary, allowing for regulation of firing rate in auditory nerve fibers.

Block of I_h with ZD7288 also induced a hyperpolarization of the membrane potential by about 4 mV. When the membrane potential was reset to control values during block of I_h, EPSPs still slowed down by roughly 30%. This result indicates that I_h shapes the EPSP waveform directly by contributing to the resting membrane conductance. I_h may also act indirectly on the EPSP waveform, by depolarizing the membrane potential and thereby activating other ion channels such as I_KL that might additionally change the resting membrane conductance (Rothman and Manis 2003b). However, our data set here does not prove or reject this scenario because the changes in EPSP waveform during block of I_h with and without hyperpolarization were not significantly different.

During block of I_h, and with a change in the resting membrane conductance, we expected to see not only slowing of the EPSP waveform but also an increase in the EPSP amplitude (Magee 1998). The EPSP amplitude increased slightly, although not significantly. This is most likely due to the wide range of EPSP amplitudes in individual recordings. We suspect that if a larger data set was available, the difference in amplitude might have reached statistical significance. The fact that there was a small but significant increase in the spike amplitude (representing a less variable waveform compared with the EPSPs) further supports this idea.

HCN channels in afferent dendrites as possible targets for lateral efferent transmission

IHC afferent dendrites receive efferent innervation from the lateral superior olivary complex. Multiple transmitters such as acetylcholine, dopamine, γ-aminobutyric acid, and opiate peptides have been found in lateral efferent terminals (Eybalin 1993). Lesion of lateral efferent nerves disrupted temporal coding and increased susceptibility to acute acoustic trauma (Darrow et al. 2006, 2007). Intracochlear perfusion of putative lateral efferent neurotransmitters like dopamine affected firing rates in the auditory nerve (Ruel et al. 2001). However, the cellular mechanisms underlying these modulatory effects are unclear. One possibility is that lateral efferent transmitters modulate dendritic ion channels such as I_h, for example via second messenger cascades, thereby subsequently affecting auditory nerve firing. Indeed, such modulation of I_h by a neurotransmitter has been demonstrated in auditory neurons. In rat MNTB neurons, noradrenaline and cAMP analogs increased the amplitude and shifted the activation curve of I_h (Banks et al. 1993). In chick nucleus laminaris, noradrenaline enhanced temporal precision of EPSPs by regulating I_h activity in the postsynapse (Yamada et al. 2005). In addition to cyclic nucleotides, phosphorylation by protein kinases, pH, and lipid second messengers have been shown to modulate HCN channel activity (Fogle et al. 2007; Pian et al. 2007; Robinson and Siegelbaum 2003; Zolles et al. 2006). Our finding that a cAMP-sensitive I_h in afferent dendrites modulates the EPSP waveform supports the idea that modulation of temporal coding could occur directly at the first synapse of the auditory pathway, possibly via lateral efferent inputs.

DISCLOSURES

No conflicts of interest are declared by the authors.