The differing biophysical properties of neurons the axons of which form the different pathways from the ventral cochlear nucleus (VCN) determine what acoustic information they can convey. T stellate cells, excitatory neurons the axons of which project locally and to the inferior colliculus, and D stellate cells, inhibitory neurons the axons of which project to the ipsi- and contralateral cochlear nuclei, fire tonically when they are depolarized, and, unlike other cell types in the VCN, their firing rates are sensitive to small changes in resting currents. In both types of neurons, the hyperpolarization-activated current (Ih) reversed at –40 mV, was activated at voltages negative to −60 mV, and half-activated at approximately −88 mV; maximum hyperpolarization-activated conductances (gh max) were 19.1 ± 2.3 nS in T and 30.3 ± 2.6 nS in D stellate cells (means ± SE). Activation and deactivation were slower in T than in D stellate cells. In both types of stellate cells, 50 μM 4(N-ethyl-N-phenylamino)1,2-dimethyl-6-(methylamino) pyridinium chloride (ZD7288) and 2 mM Cs+ blocked a 6- to 10-fold greater conductance than the voltage-dependent gh determined from Boltzmann analyses at –62 mV. The voltage-insensitive, ZD7288-sensitive conductance was proportional to gh max and ginput. 8-Br-cAMP shifted the voltage dependence of Ih in the depolarizing direction, increased the rate of activation, and slowed its deactivation in both T and D stellate cells. Reduction in temperature did not change the voltage dependence but reduced the maximal gh with a Q10 of 1.3 and slowed the kinetics with a Q10 of 3.3.
Hyperpolarization-activated cation currents (Ih) perform a wide range of functions in excitable cells. In many neurons, Ih contributes to setting the resting potential (Bal and Oertel 2000; Banks et al. 1993; Doan and Kunze 1999; Pape 1996) and to controlling the rate of spontaneous and evoked firing (Maccaferri and McBain 1996; McCormick and Pape 1990b; Spain et al. 1987). In hippocampal cells, Ih contributes to synaptic integration in dendrites (Magee 1998; Williams and Stuart 2000). These functions can be modulated through cAMP-mediated signaling pathways (Cathala and Paupardin-Tritsch 1999; DiFrancesco and Tortora 1991; Ingram and Williams 1996; Ludwig et al. 1998; McCormick and Pape 1990a; Pedarzani and Storm 1995; Saitow and Konishi 2000; Tokimasa and Akasu 1990; van Ginneken and Giles 1991; Vargas and Lucero 1999).
The shapes, projections, and biophysical characteristics of the four groups of principal cells of the VCN, bushy, octopus, T stellate, and D stellate cells, are distinct. Octopus cells occupy the most caudal and dorsal VCN in a sharply delineated area (Golding et al. 1995). The remainder of the VCN contains intermingled bushy and stellate cells with bushy cells being more common anteriorly and stellate cells more common posteriorly. The low input resistances that result from the activation of gh and a low-voltage activated K+ conductance, gKL, make responses to synaptic currents in octopus and bushy cells fast and precisely timed, and responses to current pulses transient (Bal and Oertel 2000; Cuttle et al. 2001; Manis and Marx 1991; Oertel et al. 2000). The absence (or near absence) of gKL in stellate cells allows them to fire tonically when they are depolarized; their rates of firing are sensitive to small changes in resting currents (Ferragamo and Oertel 2002; Fujino and Oertel 2001; Oertel et al. 1990). The responses of VCN cells to sound have been shown in vivo to be sensitive to neuromodulation (Kossl and Vater 1989).
Two types of stellate cells have been distinguished and named on the basis of several differences. In slices from mice, they were named for their projection patterns. Both types project locally in the ventral and dorsal cochlear nuclei but the main axon projects through the Trapezoid body in T stellate cells but Dorsalward through the intermediate acoustic stria in D stellate cells (Oertel et al. 1990). Other authors have distinguished T and D stellate cells on the basis of other differences: “type 1” and “type 2” vary in the degree of somatic inputs in electron micrographs (Cant 1981), responses to tones are “chopper” or more transient “onset-chopper” (Blackburn and Sachs 1989; Smith and Rhode 1989), the dendrites of “planar” stellate cells are aligned with auditory nerve fibers, whereas those of “radial” stellate cells are not (Doucet and Ryugo 1997). T stellate cells are glutamatergic and form a major excitatory pathway to the contralateral inferior colliculus (Adams and Warr 1976), whereas D stellate cells are glycinergic and inhibitory (Ferragamo et al. 1998; Needham and Paolini 2003). Many, but probably not all, D stellate cells project to the contralateral cochlear nuclei (Arnott et al. 2004; Cant and Gaston 1982; Needham and Paolini 2003). T stellate cells are narrowly tuned and fire tonically in responses to tones, whereas D stellate cells are more broadly tuned and fire transiently at the onset of tones (Arnott et al. 2004; Blackburn and Sachs 1989; Jiang et al. 1996; Nelken and Young 1994; Palmer and Winter 1996; Palmer et al. 1996; Smith and Rhode 1989; Smith et al. 2005).
The present experiments show that gh regulates the firing of stellate cells through its influence on the resting potential and input resistance. The sensitivity of gh to cAMP levels indicates that firing in both types of stellate cells can be regulated neuronally. We were surprised to discover in the course of our analysis that ZD7288 and extracellular Cs+, drugs that are commonly used to block gh, block more than the voltage-sensitive gh.
Preparation of slices
All experiments were done in accordance with the guidelines of the Research Animal Resources Center at the University of Wisconsin-Madison. Parasagittal slices were prepared from the cochlear nuclei of ICR mice between 16 and 18 days old. During the surgery, the auditory nerve was cut where it exits the internal auditory meatus. A block of brain tissue containing the cochlear nucleus was trimmed and glued onto a Teflon-coated specimen disc. The slices (200 μm thick) were prepared with a VT1000S oscillating tissue slicer (Leica, Nussloch, Germany) using a sapphire blade (Delaware Diamonds Knives, Wilmington, DE). The tissue was kept submerged in physiological saline of the following composition (mM): 127 NaCl, 3 KCl, 1.2 KH2PO4, 2.4 CaCl2, 1.3 MgSO4, 3 HEPES, 20 NaHCO3, and 10 glucose, pH 7.4 when saturated with 95% O2-5% CO2 at 26°C. The osmolality, measured with a 3D3 Osmometer (Advanced Instruments, Norwood, MA), was 302 mOsm/kg. Slices were transferred to a recording chamber (∼0.6 ml) and superfused continually at 5–6 ml/min. The temperature of the inflowing perfusate was monitored with a small thermistor, IT-23, and a Thermalert thermometer TH-5 (Physitemp, Clifton, NJ) and kept at 33 ± 0.5°C in most of the experiments by a custom-built, feedback-controlled heater. The recording chamber was placed on the stage of an Axioskop 2FS microscope (Zeiss, Oberkochen, Germany) and individual neurons were visualized using differential interference contrast optics (×63/0.9 w water-immersion lens).
Patch pipettes were made from borosilicate glass capillaries (1B120F-4, World Precision Instruments) and had resistances of between 3.8 and 5 MΩ when filled with a solution containing (in mM) 110 potassium gluconate, 9 HEPES, 9 EGTA, 4.5 MgCl2, 14 phosphocreatinine, 0.3 GTP, and 4 Na2ATP, pH 7.3 (KOH), 293 mOsm/kg. In normal extracellular saline, this solution yielded a junction potential of −12 mV that was added to all voltages. Whole cell current- and voltage-clamp recordings were performed with an Axopatch 200B amplifier controlled by a PC computer through a Digidata 1200 computer interface and pClamp 8.0 software (Axon Instruments, Foster City, CA). The access resistance (Ra) was monitored throughout all experiments and ranged from 9 to 17 MΩ. Ra was partially compensated on-line between 60 and 75% with a 20-μs lag time. For analysis of the voltage dependence of activation, the voltage drop caused by the uncompensated Ra was subtracted off-line from the applied voltage command to yield a precise value of transmembrane voltage (Rothman and Manis 2003). Voltage responses to current injection and current responses to voltage steps were recorded at 40 and 5 kHz and filtered at 10 and 1 kHz, respectively.
The input resistance (Rinput) was determined in a subset of cells from the slope of the relationship between the mean voltage responses in three consecutive recordings to the last 50 ms of 300-ms current pulses from −60 to +60 pA in 10-pA steps. To achieve the greatest possible accuracy in these measurements, the voltage drop across the access resistance was subtracted (bridge was balanced) off-line in the following way. The 0.2 ms after the end of the current pulse that was contaminated by electrode artifacts was eliminated and reconstructed by fitting the time course of decay after the offset with a single exponential and extrapolating back to the offset of the current pulse. Activation and deactivation time constants of Ih were determined by fitting the current evoked during an activating or deactivating pulse to double exponential functions of the form: where Ih(t) is the current at time t, ISS is the steady state current, Af and As are the initial amplitudes of the fast (τf) and slow (τs) exponential components, respectively. To avoid contamination by the capacitative artifacts, the first 12 to 15 ms of the recorded current were not included in the fit. The voltage sensitivity of gh was measured from tail currents. The amplitude of individual tail currents, measured 12 ms after the end of the conditioning step (I), was normalized as a function of maximal and minimal tail currents as I(V) = (I − Imin)/(Imax − Imin) and plotted as a function of the voltage of the preceding step (V) and reflects the relative conductance (g). Each of the measurements from individual cells was fitted by a Boltzmann function of the form: g(V) = 1/1 + e[(V − V0.5)/k] from which the value of half-maximal activation voltage (V0.5) and the slope factor (k) were derived. The maximal conductance (gh max) was determined from tail currents using the Ohm's law g = (Imax − Imin)/(V − Vrev), where Imax was the tail current after a fully activating voltage step, Imin was the tail current after a step to −52 mV, V was the voltage at which the tail currents were measured (−77 mV), and Vrev the reversal potential of Ih. The Q10 for the change in gh max and kinetics of activation was determined from the relationship: . Data are expressed as means ± SE. Statistical significance was determined using Student's t-test; data were considered significant when the probability of the null hypothesis was <0.05. Measurements were initially analyzed with Clampfit 9.0 (Axon Instruments) and all the plots, fits and statistics were performed with Origin 7.5 (Microcal, Northampton, MA) software.
All measurements of Ih under voltage-clamp were made after blocking other currents pharmacologically. Glycinergic and glutamatergic synaptic currents were blocked with 0.5 μM strychnine and 40 μM 6,7-dinitroquinoxaline-2,3(1H,4H)-dione (DNQX). Voltage-sensitive Na+ and K+ currents were blocked with 1 μM tetrodotoxin (TTX) and 2 mM 4-aminopyridine (4-AP). All drugs were added to the saline solution and applied in the bath. In some other recordings, Ih was blocked with 50 μM 4(N-ethyl-N-phenylamino)1,2-dimethyl-6-(methylamino) pyridinium chloride (ZD7288) or with 2 mM CsCl. Most drugs and reagents were purchased from Sigma (St. Louis, MO), ZD7288 was obtained from Tocris (Bristol, UK) and TTX from Alomone Labs (Jerusalem, Israel).
The following results are based on whole cell patch-clamp recordings from 120 T and 15 D stellate cells in the VCN in which the access resistance, input resistance, and resting potential remained stable throughout the recording. Our sample of recordings from T stellate cells is larger than that of D stellate cells because there are more of them in the VCN. The electrophysiological distinctions between T and D stellate cells are illustrated in Fig. 1. While both T and D stellate cells responded to depolarizing current pulses with sustained firing, the repolarization after action potentials in responses to depolarizing current and the rectification of responses to hyperpolarizing current were subtly different (Ferragamo et al. 1998; Fujino and Oertel 2001; Oertel et al. 1990). In both groups of cells, repolarizing currents associated with the action potential caused a fast and pronounced voltage undershoot (Fig. 1, *). Unlike T stellate cells, repolarization in D stellate cells exhibited an additional slower and more variable undershoot (Fig. 1, bottom, ▴). In response to injection of hyperpolarizing current, voltages fell to a hyperpolarizing peak and then sagged back toward rest. Experiments described below indicate that the sag results from the activation of Ih. The sag was slower and smaller in T stellate than in D stellate cells (Fujino and Oertel 2001). Rebound action potentials after the end of a hyperpolarizing current step were common in both types of cells.
The resting potentials and input resistances of T and D stellate neurons in the absence of any ion channel blockers were not significantly different. The resting potentials were –64.4 ± 0.8 mV (n = 22) in T stellate and –64.8 ± 0.4 mV (n = 13) in D stellate cells. In a subset of cells, responses were measured to small (10 pA) steps of current in triplicate as described in methods. In this sample of cells, the input resistances around rest (Rinput) had a mean of 74 ± 11 MΩ (n = 12) in T stellate and 60 ± 5 MΩ (n = 11) in D stellate cells.
Isolation and estimation of Ih
To examine Ih, recordings were made in voltage clamp. It is unlikely that stellate cells are isopotential under voltage clamp because they have long, tapering dendrites and long segments of axons in slices (Oertel et al. 1990). In clamping these cells at depolarized potentials, escaped action potentials were often observed in the absence of Na+ channels blockers. Outward currents exceeded 12 nA at +10 mV but were unlikely to be clamped well because plots of the instantaneous component of the tail current versus the voltage were not linear and in some cells repeated voltage steps evoked inconsistent current responses. Responses to hyperpolarizing voltages were, however, generally stable under voltage clamp. One interpretation of these observations is that the majority of gh is located electrically near the recording site and that depolarization-activated gK and gNa are often located more distantly on axons or other fine processes.
After recording the responses to current pulses in current-clamp mode in the absence and presence of synaptic blockers, 1 μM TTX was added to the physiological saline to block Na+ currents. Figure 2 shows voltage-clamp recordings of Ih in a T and a D stellate cell when the voltage was stepped from a holding potential of –62 mV to voltages between –57 and –117 mV. Shifts in voltage elicited an instantaneous current (Iinst) followed by a slowly developing Ih (Fig. 2A). The capacitative current at the onset of a voltage step lasted 6–10 ms. To avoid distortion of Iinst by the capacitative artifact, Iinst was measured as the mean value of the current over 1 ms centered 12 ms after the onset of steps to between –57 to –77 mV. For larger hyperpolarizing steps, the first 12 ms were excluded, and Iinst was measured by fitting the activation of the current with a double-exponential function and extrapolating back to the onset of the step. To isolate Ih, recordings were made in the presence of 2 mM 4-AP. Figure 2B shows that 2 mM 4-AP reduced Iinst, perhaps as a result of blocking a small, low-voltage activated K+ current (IKL) (Rothman and Manis 2003). In some cells, the plot of Iinst versus the applied voltage deviated from linearity in the absence of 4-AP, causing measurements of the input conductance (ginput) (slope) to depend on the range of voltages over which the slope was fit. The ginput at –62 mV was therefore estimated from linear fits of Iinst-V plots between −57 and −77 mV (Fig. 2C). Adding 2 mM 4-AP decreased the slope and made Iinst a more linear function of voltage. In the T and D stellate neurons shown in Fig. 2C, the ginput decreased from 12.9 to 10.4 nS and from 17.7 to 15.8 nS, respectively. On average, in the presence of 2 mM of 4-AP, the ginput was 12.0 ± 1.3 nS (n = 16) in T stellate cells and 12.4 ± 1.2 nS (n = 6) in D stellate cells. Adding 4-AP reduced ginput by 19.4 ± 4.6% (n = 6) in T stellate and by 17.8 ± 3.8% (n = 5) in D stellate cells. To determine whether 4-AP affects Ih, Ih was first separated from other currents by subtracting Iinst from the current evoked by a voltage step from −62 to −92 mV (not shown). Ih was reduced by 18.4% (n = 8) on average in T stellate and by 14.4% (n = 3) in D stellate cells by 2 mM 4-AP.
Reversal potential of Ih
The reversal potentials were measured from the intersection in Iinst-V plots after conditioning pulses to three different voltages. Figure 3 illustrates how the measurement of the reversal potential for Ih was made in a D stellate cell. Ih was activated by hyperpolarizing, conditioning pulses to three different voltages (Fig. 3A). Pulses were longer for the smaller hyperpolarizations to allow currents to approach the steady state. Plots of Iinst 12 ms after the end of the conditioning pulses were linear as a function of the voltage with the slopes reflecting the magnitude of the previous, steady-state conductance (Fig. 3B). At the reversal potential, no current flows through gh so that Ih is equal (0 nA) under all conditions. The reversal potential was measured as the point at which the regression lines, fitted to the Iinst-V relationships, intersect. Reversal potentials in T stellate cells were −40 ± 2 mV (n = 6), those in D stellate cells were −41 ± 4 mV (n = 4) (Fig. 3C). The Iinst-V relationships were linear and intersected at one point, indicating that gh was reasonably well space-clamped.
Voltage sensitivity and maximum amplitude of gh
Tail currents were used to measure the voltage sensitivity and maximum amplitude of gh. The steady-state activation of gh is reflected in the amplitude of tail currents on return to −77 mV after steps to voltages between −52 to −132 mV (Fig. 4). The tail currents measured at the test voltage (–77 mV) result from the activation or deactivation of Ih after the end of the conditioning pulse. Over the depolarizing voltage range where little or no gh was activated by the conditioning pulse, the step to –77 mV caused activation of gh; when the variable voltage pulse was strongly hyperpolarizing, the step to –77 mV caused gh to be deactivated (Fig. 4A). The relative amplitudes of tail currents between the two extremes are a measure of the voltage sensitivity of gh. The relationship of the normalized conductance as a function of voltage is shown for seven T stellate and eight D stellate cells in Fig. 4B. Half-maximal activation of gh occurred at −88.9 ± 0.7 mV in T stellate and −86.8 ± 1.2 mV in D stellate cells and were not significantly different. The slope factors (k) that reflect the steepness of the relation between voltage and fractional activation of Ih, were 7.8 ± 0.6 and 6.6 ± 0.2 mV in T and D stellate cells, respectively, and were not significantly different. The Boltzmann relationships indicate that only a small fraction of gh max is activated at the resting potential, on average 4.1 and 4.5% of the maximal gh in T and D stellate neurons respectively (Fig. 4B). To test whether measurements of the voltage sensitivity might have been distorted by the failure of conditioning pulses to allow the activation of currents to reach the steady state, gh-V curves generated from tail currents after 3-s pulses and from responses to pulses that varied from 3 to 6.2 s in a T stellate cell were compared. Differences were minimal (Fig. 4C). The slope factor differed significantly in T stellate cells when measurements were made with 2- and 3-s pulses, however (data not shown).
Pharmacological block by ZD7288 and Cs+
In other cells, ZD7288 (BoSmith et al. 1993; Gasparini and DiFrancesco 1997; Harris and Constanti 1995) and Cs+ (Harris and Constanti 1995; Magee 1998; McCormick and Pape 1990b) have been shown to block Ih. We examined the sensitivity of the currents activated from a holding potential of −62 mV to 50 μM ZD7288. In both T and D stellate cells, ZD7288 blocked Ih almost completely, and it significantly reduced Iinst (Fig. 5, A and B). In responses to strongly hyperpolarizing voltages, some unblocking of ZD7288 occurred with time as has also been observed in other types of neurons (Harris and Constanti 1995; Shin et al. 2001). The Iinst-V plot shows that ginput at –62 mV in the T stellate cell was reduced from 10.4 to 4.4 nS and in the D stellate cell from 10.1 to 6.1 nS (Fig. 5C). On average, 50 μM ZD7288 reduced the ginput by 49.4 ± 3.1% (n = 11) in T and by 46.2 ± 3.1% (n = 3) in D stellate neurons. The sensitivity to Cs+ was determined in a similar way. Figure 5 illustrates that 2 mM Cs+ blocked a substantial proportion of Ih in a T stellate cell. Strong hyperpolarizing pulses revealed that a small, hyperpolarization-activated current remained in the presence of Cs+, however, indicating that the block by 2 mM Cs+ was not complete. The ginput at –62 mV in this cell was reduced from 10.0 to 8.2 nS. On average, 2 mM Cs+ reduced ginput in T stellate cells by 13.2 ± 3.5% (n = 7). The block by 2 mM Cs+ in T stellate cells was significantly smaller than that by 50 μM ZD7288.
Voltage-insensitive, ZD7288-sensitive conductance
A goal of the present experiments was to understand how gh contributes to regulating the excitability of stellate cells and how it contributes to the setting of the resting potential. The large differences, even within an individual cell, of estimates of gh near rest made from the Boltzmann function, and sensitivity of the input conductances near rest to ZD7288 and Cs+ prompted a quantitative analysis of conductances near rest.
Measurements were made of gh max and ginput as described in the preceding text. Maximal gh was 19.1 ± 2.3 nS (n = 16) in T and 30.3 ± 2.6 nS (n = 6) in D stellate cells (Fig. 6A), and the two were significantly different. All recordings that form the basis of this analysis of conductances at –62 mV were made in the presence of 2 mM 4-AP. Correcting for the fraction of gh that was blocked by 4-AP, gh max were 23.4 nS for T and 35.4 nS for D stellate cells.
There was considerable variability in the ginput and gh max among recorded neurons. This might be expected if recordings were made from stellate cells the processes of which were cut to varying degrees in the making of slices. We therefore plotted gh max as a function of ginput and found that they covaried and that the T and D stellate cells with the largest ginput also had the largest gh max (Fig. 6A).
Figure 6B shows a comparison of the estimates of gh active at –62 mV derived from the Boltzmann functions and those derived from pharmacological sensitivity to 50 μM ZD7288 and 2 mM Cs+. In every T and D stellate cell for which measurements were made, the measurement of the voltage-sensitive gh was smaller than the conductances blocked by ZD7288 and Cs+ at −62 mV. Distortions in the Boltzmann function are unlikely to have been large enough to account for these differences. One explanation for these results is that ZD7288 and Cs+ are not specific for gh and that each blocks other, unidentified conductances. Such an explanation cannot be excluded but there is a more interesting possibility.
It has recently been demonstrated that in an expression system, the expression of gh is associated with a voltage-insensitive conductance, an instantaneous current (Macri and Accili 2004; Proenza et al. 2002). If the expression of a voltage-sensitive gh is associated with a voltage-insensitive conductance in T and D stellate cells, then the magnitude of the voltage-insensitive, ZD7288-sensitive conductance would be expected to be related to the voltage-sensitive conductance. Figure 6C shows that it was the case in T stellate cells, and perhaps also in D stellate cells, that the magnitude of the voltage-insensitive ZD7288-sensitive instantaneous conductance is proportional to the gh max in stellate cells. Whether the voltage-insensitive Cs+-sensitive conductance follows the same pattern was unclear (because the spread of gh max in our sample of recorded neurons was not wide enough). However, Fig. 6A implies that the voltage-insensitive, ZD7288-sensitive conductance is also proportional to ginput and that its expression may simply be related to the surface area of the cell.
Kinetics of Ih activation and deactivation
The activation and deactivation kinetics of Ih in T and D stellate cells are compared in Fig. 7. Ih activated more rapidly in D than in T stellate cells (Fig. 7A). The time course of the activation of currents was well described by the sum of two exponentials, the fast and slow time constants, τf and τs, which were voltage-dependent and in the tens and hundreds of milliseconds, respectively. Not only were each of the time constants significantly faster, but also the relative contribution of the fast time constant was greater in D than in T stellate cells. For instance, in response to a voltage step to −97 mV, τf was 137 ± 16 ms in T stellate cells (n = 8) and 75 ± 9 ms in D stellate cells (n = 7); in the same cells τs was 966 ± 89 ms and 567 ± 45 ms in T and D stellate cells. The time constants became shorter with increasing hyperpolarization (Fig. 7B). The deactivation of Ih was examined by fitting double-exponential functions to the decay of the tail currents. After pulses to −122 mV that were long enough for the current to reach steady state, steps to between −62 and −82 mV caused Ih to deactivate. Deactivation kinetics were voltage-dependent and were significantly faster in D stellate than T stellate neurons. When the voltage was returned to near rest (−67 mV), the fast components (τf) were 148 ± 17 ms and 56 ± 7 and slow components (τs) were 866 ± 93 ms and 356 ± 57 in T (n = 6) and D (n = 4) stellate neurons, respectively, and the fast components were more dominant in D than T stellate neurons.
Ih contribution to stellate cells' excitability
To gain an understanding of how ZD7288- and Cs+-sensitive currents affect the excitability of stellate cells, we compared the T and D stellate cells' behavior in current clamp in the absence and presence of these blockers. 50 μM ZD7288 and 2 mM Cs+ hyperpolarized T and D stellate cells, increased their input resistance, abolished the sag in the voltage response, and prevented action potentials at the end of hyperpolarizing current steps (Fig. 8). Figure 8A shows that application of 50 μM ZD7288 to a T stellate cell increased the input resistance from 45 to 124 MΩ and induced a hyperpolarization of 10 mV; injecting 40 pA of positive current brought the resting potential at the cell body to its original level. Figure 8B shows that 2 mM Cs+ increased the input resistance from 70 to 102 MΩ in a D stellate cell; the 3-mV hyperpolarization of the resting potential could be compensated by 10 pA. Οn average, 50 μM ZD7288 increased the input resistance in T stellate cells from 69 ± 5 to 130 ± 12 MΩ and hyperpolarized the resting potential by 8 ± 2 mV (n = 5). In D stellate cells, ZD7288 increased the input resistance from 96 ± 10 to 167 ± 12 MΩ and hyperpolarized the resting potential by 4 ± 1 mV (n = 3). In T stellate cells, 2 mM Cs+ increased the input resistance from 86 ± 5 to 131 ± 17 MΩ and hyperpolarized the resting potential by 3.2 ± 0.7 mV (n = 3). Both ZD7288 and Cs+ hyperpolarize the resting potential, indicate that they block a net depolarizing current; ZD7288 hyperpolarizes the resting potential significantly more than Cs+.
The shapes of action potentials were altered by 50 μM ZD7288 or 2 mM Cs+ (Fig. 8, A and B). Less current was required to bring cells to threshold in the presence of both blockers. Action potentials were broader and the first, fast undershoot following the action potential was less deep when Ih was blocked. In the presence of 50 μM ZD7288, large depolarizing currents generated a few action potentials at the onset of the pulse the peaks of which became progressively lower and eventually cells ceased firing (Fig. 8A). In the presence of Cs+, failure to generate action potentials was less pronounced than in ZD7288 but was evident when cells fired at high frequencies.
In the presence of blockers of Ih or Ihg ZD7288 or Cs+, two repolarizing components were clearly visible after action potentials, much like those illustrated for a D stellate cell in Fig. 1. Presumably the increased input resistance in the presence of ion channel blockers made voltage changes by small currents more prominent. Under control conditions, the voltage always rises monotonically after an action potential in T stellate cells, whereas double undershoots are observed in D stellate cells, characteristics that have served to distinguish these two types of cells (Fig. 1) (Fujino and Oertel 2001; Oertel et al. 1990). These results show that differences in the repolarization of action potentials in T and D stellate cells arise from differences in the relative magnitudes of the currents rather than from their presence or absence.
Modulation of Ih
One of the properties of Ih that makes it biologically interesting is that it can be modulated by changes in intracellular levels of cAMP (Robinson and Siegelbaum 2003). The observation that the voltage range of activation of Ih in octopus cells, cells that lie immediately adjacent to stellate cells, was not affected by cyclic nucleotides (Bal and Oertel 2000) raised the question whether Ih in stellate cells is modulated by cAMP. As in the previous study, modulation of Ih in stellate neurons was examined by perfusing a membrane-permeable analogue of cAMP, 8-Br-cAMP, in the bath. Both the voltage sensitivity and the kinetics of Ih in stellate cells were affected by 8-Br-cAMP.
In every T and D stellate cell tested (n = 8), the voltage-activation curve shifted in the depolarizing direction on addition of 8-Br-cAMP. Figure 9 shows an example of the action of 500 μM 8-Br-cAMP on a T stellate cell and 100 μM 8-Br-cAMP on a D stellate cell. The negative shift in the holding current at −62 mV reveals the activation of a depolarizing current. At 100 μM, 8-Br-cAMP caused a shift in the half-activation voltage (V0.5) of Ih of 4.2 ± 1.0 mV (n = 3) in T stellate cells and 6.7 ± 0.4 mV (n = 2) in D stellate cells. The effect of 8-Br-cAMP was concentration-dependent; in T stellate cells 500 μM 8-Br-cAMP shifted V0.5 by 11.7 ± 1.4 mV (n = 4).
In the presence of 8-Br-cAMP, activation of Ih was accelerated and deactivation was slowed. Changes in the kinetics of Ih that were observed in all T and D stellate cells tested are illustrated in a D stellate cell in the presence of 500 μM 8-Br-cAMP in Fig. 10. Activation by a pulse to –97 mV proceeded with time constants τs = 1050 ms and τf = 148 ms in control conditions and τs = 784 and τf = 116 ms in the presence of 8-Br-cAMP with the fast component being more prominent (Fig. 10B). Deactivation time constants slowed and the slower component increased in prominence in the presence of 8-Br-cAMP (Fig. 10C). In this cell, the voltage range of activation of Ih shifted by 11.2 mV in the presence of 8-Br-cAMP (Fig. 10D).
Influence of temperature on Ih
Changes in temperature affected the kinetics and the magnitude of Ih in stellate neurons (Fig. 11). Lowering the temperature from 33 to 27οC reduced the amplitude of the evoked inward current as well as the rates of activation and deactivation. The changes in amplitude of Ih in this cell, as in all others tested, were stable over the 12 min over which it was monitored. The reduction in temperature did not alter the voltage dependence of activation; in T stellate cells V0.5 was −91 ± 1.6 at 33οC and −90 ± 2.0 at 27οC (n = 5). In T stellate cells gh max was reduced with a Q10 1.3 ± 0.05 and the fast activation time constant was slowed with a Q10 3.3 ± 0.3 when the temperature was reduced from 33 to 27οC (n = 6).
Hyperpolarization-activated conductances are common in neurons even though the voltage range of activation spans membrane potentials that are largely beyond the physiological range. The present experiments show that gh, the voltage sensitivity of which is regulated through cAMP, affects the excitability of stellate neurons by its influence on the resting potential and on the input conductance. In making measurements in T and D stellate cells that were sufficiently detailed to illustrate how consistent certain features of gh are among neurons and also how other features of gh seem to be tailored to suit different types of neurons in the VCN, we found unexpectedly that these cells also have a voltage-insensitive, ZD7288-sensitive, and Cs+-sensitive conductances that are 5–10 times larger than gh at –62 mV and the magnitude of which is proportional to these cells' maximum gh.
Characteristics of Ih in neurons from the VCN
Ih reverses at approximately –40 mV in both T and D stellate cells. In other cells, too, reversal potentials of this mixed cation current generally lie between –30 and –44 mV (Bal and Oertel 2000; Banks et al. 1993; Chen 1997; Maccaferri and McBain 1996; McCormick and Pape 1990b; Mo and Davis 1997; Saitow and Konishi 2000; Spain et al. 1987). This reversal potential reflects a permeability ratio PNa/PK of ∼0.3 (Wollmuth and Hille 1992). The gh max of stellate cells, ∼20 nS in T and ∼30 nS in D stellate cells, was considerably smaller than that in octopus cells (∼150 nS) and the half-activation voltages around −88 mV were more hyperpolarized than that measured in octopus cells (approximately −65 mV) (Bal and Oertel 2000). The differing maximal conductances and voltage ranges of activation result in large differences in the resting gh. On average only ∼4% gh max, is activated in T and D stellate cells at rest, whereas 62 nS gh, 41% of gh max, is activated in octopus cells (Bal and Oertel 2000). The kinetics of Ih is more rapid in D stellate than in T stellate cells, and both are slower than in octopus cells. In each of the cells, τfast dominates the time course of the activation of Ih. Currents evoked by steps from –62 mV to the voltage where gh was about half-maximally activated had a τfast of 50 ms (−67 mV) in octopus cells, 140 ms (−97 mV) in T stellate, and 75 ms (−97 mV) in D stellate cells. For the same voltage steps, τslow was 190 ms in octopus, 960 ms in T stellate, and 560 ms in D stellate cells. Bushy cells express gh with characteristics generally similar to those described here for stellate neurons; measurements in bushy cells are not directly comparable because they were made from younger animals and at lower temperatures (Cuttle et al. 2001); the expression and possibly the voltage dependence of Ih are developmentally regulated in bushy cells as in other cell types (Cuttle et al. 2001; Tanaka et al. 2003).
A characteristic feature of gh is that its biophysical properties are regulated by cAMP. In T and D stellate cells, 8-Br-cAMP shifted its voltage dependence in the depolarizing direction. Increases in intracellular cAMP shift the voltage range of activation in the depolarizing direction with little or no change in gh max in many cell types including bushy cells (Banks et al. 1993; Cathala and Paupardin-Tritsch 1999; Chen et al. 2001; Cuttle et al. 2001; DiFrancesco and Tortora 1991; Ingram and Williams 1996; Ludwig et al. 1998; McCormick and Pape 1990b; Saitow and Konishi 2000; Santoro and Tibbs 1999; Santoro et al. 1998; Tokimasa and Akasu 1990; Vargas and Lucero 1999). The maximal shift is related to the type of channels that form Ih and also to the amount of basal modulation by the resting levels of cAMP (Chen et al. 2001). In octopus cells, gh seems not to be modulated in similar experiments (Bal and Oertel 2000), suggesting that in these cells, Ih channels are either fully modulated or that they are formed largely from subunits whose sensitivity to cAMP is small. In octopus cells, the magnitude of Ih also changes with temperature and then adapts over tens of minutes (Cao and Oertel 2005). We did not observe such adaptation in T or D stellate cells.
Possible molecular identity of Ih in neurons from the VCN
Ih is mediated through hyperpolarization-activated cyclic nucleotide-gated (HCN) channels. Four subunits, HCN1-HCN4, form these channels (Ludwig et al. 1998; Santoro and Tibbs 1999; Santoro et al. 1997, 1998). Homomeric channels expressed in heterologous expression systems have differing voltage sensitivity, kinetics, and sensitivity to cyclic nucleotides. Channels of HCN1 subunits are fastest, HCN2 channels activate and deactivate at intermediate rates, and HCN4 channels are the slowest (Moosmang et al. 2001). HCN1 are less sensitive to cyclic nucleotide modulation than the others (Santoro et al. 1998; Ulens and Tytgat 2001), whereas HCN2 is more sensitive to cAMP (Chen et al. 2001). HCN subunits also combine to form heteromeric channels (Altomare et al. 2003; Chen et al. 2001; Ulens and Tytgat 2001).
The protein of HCN1 and HCN2 subunits has been detected in the octopus cell area and in adjacent regions where T and D stellate cells are located (Koch et al. 2004). The observed properties indicate that stellate neurons seem preferentially to express the relatively slow and modulatable HCN2 subunits. On the other hand, the fast kinetics, depolarized voltage range of activation, and apparent insensitivity to cAMP indicate that HCN1 is the most prevalent subunit expressed in octopus cells. Measurements of the levels of mRNA expression in the VCN also indicate that HCN1 and HCN2 are the most commonly expressed subunits (Santoro et al. 2000).
Assessment of three methods for isolating Ih
In our experiments, Ih was isolated by widely used electrophysiological and pharmacological techniques, its voltage sensitivity, ZD7288 sensitivity, and Cs+ sensitivity. It was surprising that the ZD7288-sensitive and Cs+-sensitive conductances were so much greater than the estimate of gh from the Boltzmann relationship (Fig. 6B).
It has long been known that Ih is sensitive to extracellular Cs+ (Maccaferri et al. 1993; McCormick and Pape 1990b; Spain et al. 1987). In the present experiments, 2 mM Cs+ blocked a conductance that was about sixfold greater than the voltage-sensitive gh but about half as great as that blocked by 50 μM ZD7288 at –62 mV in T stellate cells. In the MNTB, Cs+ also blocked a larger conductance at rest than expected from the Boltzmann analysis (Banks et al. 1993). The block may not have been complete or entirely specific. The block of Ih by Cs+ is known to be more complete at hyperpolarized than at depolarized voltages; for example, at –62 mV, only 60% of heterologously expressed HCN2 channels were blocked by 5 mM Cs+ (DiFrancesco 1982; Moroni et al. 2000). There is also the possibility that Cs+ is not entirely specific for inwardly rectifying mixed cation channels; it could, for example, affect inwardly rectifying K+ channels (Lesage et al. 1995; Mermelstein et al. 1998). In stellate cells, extracellular Cs+ at 2 mM seems to block primarily gh because its application causes the resting potentials to hyperpolarize. The present results are consistent with the conclusion that Cs+ blocks the voltage-sensitive gh as well as a voltage-insensitive conductance but that the block is incomplete.
ZD7288 has been widely used to test the action of Ih because it is considered to block this current nearly completely and with considerable specificity (BoSmith et al. 1993). This drug has been used to assay Ih in octopus cells of the VCN (Bal and Oertel 2000), substantia nigra neurons (Harris and Constanti 1995), CA1 hippocampal pyramidal cells (Gasparini and DiFrancesco 1997; Macaferri and McBain 1996), thalamic neurons (Luthi et al. 1998), and cerebellar basket cells (Saitow and Konishi 2000). In hippocampal CA1 neurons, the half-maximal blocking of Ih required 10.5 μM, and at 50 μM, the concentration used in the present experiments, the block was nearly saturated (Gasparini and DiFrancesco 1997). At 50 μM, ZD7288 has been reported to block glutamate receptors of both AMPA and N-methyl-d-aspartate (NMDA) subtypes (Chen 2004) and a low-voltage-activated Ca2+ conductance (Felix et al. 2003), but these conductances were blocked by DNQX and by voltage, respectively, in our experiments. In hippocampal pyramidal cells and in thalamic relay neurons, neither Na+- nor Ca2+-dependent action potentials were affected by ZD7288 (Gasparini and DiFrancesco 1997; Luthi et al. 1998; Maccaferri and McBain 1996). Furthermore, the removal of Ca2+ does not affect the resting potential or the shape of action potentials in T stellate cells (Wickesberg and Oertel 1989), indicating that voltage-sensitive Ca2+ conductances are small. It thus seems unlikely that the block of these conductances accounts for the unexpectedly large-conductance block by ZD7288 near rest.
An interesting possibility is that ZD7288 blocks a voltage-insensitive current that is mediated through HCN channels. Accili and his colleagues have suggested that the expression of HCN2 channels is associated with the expression of an “instantaneous current” that is voltage insensitive over the tested voltage range and the magnitude of which is directly proportional to the amount of heterologously expressed HCN protein and that increased on coexpression with the regulatory subunit minK-related peptide 1 (Macri and Accili 2004; Proenza et al. 2002). These authors concluded that HCN channels mediate a voltage-insensitive, instantaneous current through leaky channels. In contrast to the present results, the instantaneous current was insensitive to extracellular Cs+, but its sensitivity to ZD7288 was not reported. In hippocampal pyramidal cells ZD7288, at 10–20 μM, blocked ∼18% of the resting conductance (Gasparini and DiFrancesco 1997). Other authors, however, did not observe an instantaneous, ZD7288-sensitive conductance in thalamic relay neurons (Luthi et al. 1998) and with coexpression of HCN with the MiRP1 protein in cardiac sinoatrial cells (Qu et al. 2004).
Boltzmann analyses of conductances activated and deactivated by voltage are also subject to errors. First, if the conditioning pulses are too short, activation can be underestimated especially for voltage pulses that activate the conductance slowly. Our tests suggest this was not a significant problem when the pulses were ≥3 s. Second, the voltage range over which gh-V curves were measured should be wide enough for activation and deactivation to be saturated. The range of activation used, –50 and –125 mV, was generally wide enough to include saturation (Fig. 4). Third, with imperfect space-clamping, the slope of the gh-V function can appear less steep than it is. One indication that gh was reasonably well space-clamped is that plots of the chord conductances were linear and intersected at one point (Fig. 3). It is unlikelly that the small errors in the gh-V curves in our measurements can underly the large difference in resting activation of gh and resting ZD7288 and Cs+ sensitivity, indicating that 50 μM ZD7288 and 2 mM Cs+ block a voltage-insensitive conductance in addition to gh.
Factors that determine the excitability of T and D stellate cells
The electrical excitability of T and D stellate cells is sensitive to small changes in resting potential and resting conductance. The resting potentials of the T and D stellate cells recorded here in parasagittal slices were on average 5 mV more negative and the input resistances lower by more than a factor of two than those recorded from coronal slices (Fujino and Oertel 2001). These results suggest that processes of stellate cells have a greater tendency to be cut in coronal than parasagittal slices, a conclusion consistent with the morphology of these cells (Oertel et al. 1990). These results also suggest that the resting potential at the cell body is hyperpolarized by the presence of these processes.
Several conductances have been identified that contribute to setting the resting potential of stellate cells. Of the resting conductances, ∼15 nS in both types of stellate cells, ∼20% is blocked by 2 mM 4-AP. 4-AP blocks low-voltage-activated K+ conductances (Bal and Oertel 2001; Banks et al. 1993; Brew and Forsythe 1995; Manis and Marx 1991; Rathouz and Trussell 1998; Rothman and Manis 2003; Wu 1999) and a small proportion of gh (Bal and Oertel 2000; present experiments). The present experiments show that the voltage-sensitive gh accounts for ∼10% of the input conductance and that the voltage-insensitive, ZD7288-sensitive conductance contributes roughly another 20%. A muscarine-sensitive, voltage-insensitive conductance of ∼1 nS affects T stellate, but not D stellate, cells (Fujino and Oertel 2001).
The present experiments also reveal the importance of the input conductance in the generation of action potentials. Increasing the input resistance with ZD7288 or Cs+ weakens or prevents firing. In the presence of these blockers, action potentials become broader and their peaks become lower with repeated firing, presumably because slower depolarization results in greater inactivation of Na+ currents. By increasing the input conductance at rest, ZD7288- and Cs+-sensitive conductances promote rapid firing in stellate cells.
The work was supported by National Institute of Deafness and Other Communications Disorders Grant DC-00176.
M. J. McGinley, X. Cao, and P. Poon read the manuscript critically more than once, debated its intricacies, and made thoughtful suggestions for which we are indeed grateful. We are lucky to have had help from I. Siggelkow, S. Shatadal, and R. Kochhar.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 by the American Physiological Society