INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The tonotopic array of tuned auditory nerve fibers brings to the brain a spectral and temporal representation of sound from which biologically useful information must be extracted. Extraction of the location and meaning of sounds from the spatiotemporal pattern of activation of auditory nerve fibers begins in the cochlear nuclei, where all auditory nerve fibers terminate and distribute information to several parallel ascending pathways. Octopus cells form a pathway to the superior paraolivary nucleus and to the ventral nucleus of the lateral lemniscus (Adams 1997; Oertel 1999; Schofield 1995; Schofield and Cant 1997; Vater et al. 1997; Warr 1969). This pathway is present in all mammals and is especially prominent in humans (Adams 1997). Octopus cells convey features of sound that are critical for the recognition of natural sounds including speech. They convey the presence of acoustic transients, periodicity, and direction of frequency sweeps in their temporal firing patterns (Godfrey et al. 1975; Oertel et al. 2000; Rhode 1994, 1998; Rhode and Smith 1986; Rhode et al. 1983; Smith et al. 1993).

Octopus cells detect the coincident input of auditory nerve fibers. The dendrites of octopus cells cross the bundles of auditory nerve fibers and are thus accessible to input from many fibers (Brawer et al. 1974; Golding et al. 1995; Oertel et al. 2000; Osen 1969; Willott and Bross 1990). Summation of multiple auditory nerve inputs is required to bring an octopus cell to threshold (Golding et al. 1995). Auditory nerve fibers excite octopus cells through glutamate receptors of the AMPA subtype that have rapid kinetics (decay time constants average 350 µs) (Gardner et al. 1999, 2001). The low input resistances and short time constants of octopus cells allow the rapid synaptic currents to produce synaptic potentials whose duration is short, generally about 1 ms. With patch-clamp recordings, input resistances were measured to be between about 2 and 6 M and time constants to be about 200 µs (Bal and Oertel 2000; Golding et al. 1999). The low input resistance arises largely from two voltage-sensitive conductances that are partly activated at rest, a Cs⁺- and ZD7288-sensitive, inward rectifier (g_h) and a 4-aminopyridine (4AP) and -dendrotoxin-sensitive, low-threshold potassium conductance, g_KL (Bal and Oertel 2000, 2001; Golding et al. 1995, 1999). These conductances have similar properties but are larger than those recorded in other neurons, including other brain stem auditory neurons with especially sharp timing. The partial activation of these two, strong voltage-sensitive conductances at rest indicated that the firing of octopus cells might be sensitive to the rate at which they are depolarized. The present study shows that this is true.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Slices were made from young mice of the strains CBA and ICR. Animals between 17 and 26 days old were decapitated, and their brains were dissected in physiological saline at 31°C. The block of tissue that contained the cochlear nuclei was glued to a Teflon block at the transverse cut through the caudal inferior colliculus. Coronal slices, 150-200 µm thick, that contained the caudal posteroventral cochlear nucleus were cut with an oscillating tissue slicer (Frederick Haer, New Brunswick, ME). Slices were allowed to recover for about 1 h in a recording chamber containing about 0.3 ml, which was continuously superfused with oxygenated physiological saline, 33°C, at about 8 ml/min. Physiological saline contained (in mM) 130 NaCl, 3 KCl, 1.3 MgSO₄, 2.4 CaCl₂, 20 NaHCO₃, 3 HEPES, 10 glucose, and 1.2 KH₂PO₄, pH 7.4 (Sigma, St. Louis, MO). In some experiments, -dendrotoxin (Alomone Labs, Jerusalem, Israel) was added to the normal physiological saline.

Recordings were made with an Axopatch 200A amplifier in the "fast" current-clamp mode. A Digidata 1200 interface (Axon Instruments, Foster City, CA) controlled by pClamp software (version 6.0; Axon Instruments) was used to control current and shock stimuli and the sampling of membrane potential. Voltages were low-pass filtered at 10 kHz and sampled digitally at 25 or 50 kHz. The fire-polished borosilicate glass patch-pipettes were filled with a solution with the following composition (in mM) 140 KGluconate, 5 NaCl, 1 MgSO₄, 1 CaCl₂, 11 EGTA, and 10 HEPES, at a pH of 7.25 (Sigma). Electrodes had resistances between 5 and 10 M before they touched a cell. All traces are corrected for a junction potential of 12 mV.

In many of the present experiments, ramps as well as square pulses of current were injected through the recording electrode. Balancing the resistance and capacitance of the electrode in the voltage traces was performed off-line. Resistive components were balanced by recording the injected current pulse together with the voltage response and subtracting a voltage that was proportional to the current in the conventional way. The proportionality factor was determined by eliminating the most rapidly falling component at the end of the current pulse. The proportionality factor was constant in any one recording and reflected the resistance of the recording electrode after the seal was made (which was sometimes greater than before the seal was made, 5-22 M). In responses to depolarizing current ramps in which the electrode resistance, but not capacitance, was balanced, the voltage appeared to be displaced by a few millivolts in the hyperpolarizing direction during the current ramps with respect to the voltage responses before or after the end of the ramp (Fig. 1). Because all current was in a depolarizing direction, this apparent hyperpolarization had to be an experimental artifact of unbalanced capacitance. To balance the loss of current to the bath across the capacitance of the microelectrode as the voltage changed during the injection of current ramps, a constant voltage increment was added to the segments of the traces where current ramps were injected. The voltage increment, V_cap, corresponds in magnitude to the product of the capacitative current, I_cap, lost to the bath across the capacitance of the electrode, C_elec, and the resistance of the electrode, R_elec, during the current ramp, dI/dt

(1)

(2)

The capacitance of the electrodes could be estimated from the magnitude of the balancing voltage pulse.

(3)

The balanced capacitance was calculated in 12 recordings where it was on average 5.6 pF.

View larger version (29K): [in this window] [in a new window]   Fig. 1. Artifacts appear in traces in which the resistance has been balanced but in which the capacitance has not been balanced. A: a family of responses to ramps of equal durations whose rate of rise is varied by varying the steady-state current. B: a family of responses to ramps whose rate of rise is varied by varying the duration of a ramp to a steady level. In A and B, the resistance of the electrode was balanced by subtracting the current waveform multiplied by the resistance of the electrode. The resistance of the electrode was determined in the conventional way by "balancing the bridge" at the offset of the current. The membrane potential appears to be hyperpolarized in proportion to the rate of rise of the current ramp. These "hyperpolarizations" must be experimental artifacts because depolarizing current cannot hyperpolarize cells. In Figs. 3-6, the capacitative artifacts were balanced by adding a square voltage pulse whose amplitude was proportional to the rate of rise of the ramp.

In some experiments, inputs to neurons were activated through a glass pipette with shocks to a fiber tract within about 100 µm of the cell body. Pipettes for stimulation were similar to those used for recording, but they were filled with extracellular saline solution; voltage pulses between 10 and 100 V were 100 µs in duration.

Many octopus cells were identified morphologically. Biocytin (Sigma; 0.1%) was often included in the pipette solution. After a recording, slices were fixed in 4% paraformaldehyde and stored at 4°C, embedded in a gelatin-albumin mixture (Oertel et al. 1990), and sectioned at 60 µm in the plane of the slice. Biocytin-filled cells were visualized using the avidin-biotinylated horseradish peroxidase (HRP) complex reaction (Vectastain ABC Elite Kit; Vector Laboratories, Burlingame, CA), using nickel/cobalt-intensified 3,3'-diaminobenzidine tetra HCL (DAB) (Golding et al. 1995). Reconstructions were made with a camera lucida. With practice the octopus cell area could be visualized routinely. Because octopus cells are so distinctive electrophysiologically, not all tissues were processed histologically in later experiments.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patch-clamp recordings from 58 octopus cells, of which 42 were anatomically identified, form the basis of this study. These cells had resting potentials 60.6 ± 5.8 mV and input resistances of 3.2 ± 1.5M.

Reconstructions of four of those cells are shown in Fig. 2. The morphology was consistent with what was reported on the basis of earlier recordings in parasagittal and coronal slices (Golding et al. 1995, 1999; Oertel et al. 1990). Recordings were made from the cell bodies that lay near the posterior cut surface of slices. The dendrites spread toward the anterior through the slices. The dendrites were generally thick, 2-4 µm in diameter, and often terminated with an abrupt taper. The axons of all cells headed medially and dorsally between the deep layer of the dorsal cochlear nucleus and the restiform body in the intermediate acoutic stria. The axons were narrow near the cell bodies and widened to about 1 µm diameter. Some axons terminated in the octopus cell area and in the adjacent granule cell domains with local collaterals (Fig. 2, top right).

View larger version (18K): [in this window] [in a new window]   Fig. 2. Octopus cells labeled by internal dialysis with biocytin were reconstructed with a camera lucida. Reconstructions of the cells are shown in reconstructions of the cochlear nuclear complex in coronally cut slices. The lateral border of the brain stem is represented with a thick line. The thinner lines enclose granule cell regions. The granule cell lamina (GrL) separates the ventral from the dorsal cochlear nucleus and extends into the fusiform cell layer in the dorsal cochlear nucleus. All octopus cells lie in the octopus cell area, an area that occupies the most caudal and dorsal part of the ventral cochlear nucleus. Their axons exit the cochlear nucleus through the intermediate acoustic stria (IAS) in a fiber bundle that lies interposed between the deep layer of the dorsal cochlear nucleus and the restiform body. Protrusions in the axon at the top right (arrows) were likely to be nodes of Ranvier.

An estimate of the internodal length of octopus cell axons could be made. The branch and protrusions, spreading from 1 to 3 µm from the axon, probably indicate the location of nodes of Ranvier (Fig. 2, arrows). The distances between the branch and the more distally located protrusions was 110, 125, and 145 µm, averaging 127 µm. In peripheral nerve, it has been shown that internodal conduction time is roughly constant at 19.7 µs at 37°C over a wide range of internodal lengths (Rasminsky and Sears 1972). If the relationship between internodal length and conduction velocity is similar in central axons, the axons of octopus cells would be expected to conduct at about 6.5 m/s.

In every octopus cell tested, firing depended strongly on the rate at which they were depolarized. Rapid depolarizations caused octopus cells to fire, whereas slow depolarizations did not. This observation is illustrated in Fig. 3. Figure 3B shows responses to ramps of current that led up to an identical family of steady current pulses, 2.0-3.8 nA. The duration of the ramps differed in the three sets of traces (1, 1.2, and 1.4 ms). A comparison of the groups of traces shows that the cell fired in response to smaller currents when the current pulse rose in 1 ms than when the current pulse rose in 1.4 ms. Plots of peak voltage responses as a function of the steady-state current have a step where the current reaches threshold (Fig. 3C). With square current pulses, only 1.5 nA was required to bring this cell to threshold whereas over 3 nA were required when the current rose over a 1.2 ms ramp. The plot of the amplitude of the response as a function of the rate of depolarization indicates that the firing threshold is a consistent function of the rate of depolarization over all groups of recordings (Fig. 3D).

View larger version (35K): [in this window] [in a new window]   Fig. 3. Generation of action potentials in an octopus cell depends on the rate of rise of depolarization. A: cell was depolarized by 3 families of current pulses consisting of ramps of durations (t) that led to a steady current pulse as shown schematically. Steady-state current pulses varied identically from 2 to 3.8 nA in 0.2-nA increments in each family of traces. B: ramps produced linearly rising depolarizations that often, but not always, led to the firing of the cell. Rapidly rising ramps were more likely to cause firing than slowly rising ramps. These traces were adjusted off-line to balance the 8.6 M series resistance and 5.5 pF that were contributed by the electrode. C: the amplitude of the peak depolarization from the 62-mV resting potential as a function of the steady-state current pulse. Threshold currents for the generation of action potentials, evident as step changes in the amplitude were smallest for square current pulses and greatest for the longest ramps. D: the same responses were plotted as a function of the rate of rise of the membrane potential from rest. The rate of rise was determined from a linear fit of the depolarization in the octopus cell that preceded the action potential. The firing threshold is a consistent function of the rate of rise. In responses to each of the families of currents, action potentials were generated only when the octopus cell was depolarized at a rate 12 mV/ms or greater.

The rate of depolarization can be varied systematically either by varying the level to which the current pulse rises or by varying the duration of the ramp toward a steady-state current. Figure 3 illustrates an example in which the current was varied in small increments for three ramp durations. Figure 4 illustrates two examples for which the duration of ramps was varied in small increments to several current levels. Slower depolarizations were not only less likely to evoke action potentials than more rapid ones, but the action potentials recorded in the cell body were smaller than those evoked by rapid depolarizations. The threshold of action potentials, indicated by step changes in the amplitude of peak responses, occurred at a consistent rate of depolarization for any one cell. The lowest rate of depolarization with which an action potential could be evoked varied among cells between 5 and 15 mV/ms (mean, 10.0 ± 2.5 mV/ms; n = 12).

View larger version (28K): [in this window] [in a new window]   Fig. 4. Systematic variation in the rate of rise with changes in duration of the ramp confirms that the firing threshold is a consistent function of the rate of rise of depolarization in any 1 cell but varies between cells. A: families of current ramps were generated by varying the duration of the ramp systematically in small increments. For each cell, the upper family of traces resulted from ramps to 2 nA and the lower family of traces from ramps to 4 nA. B and C: the amplitude of the peak of the response as a function of rate of depolarization is shown for 2 different cells in B and C. Plots of the amplitude of the response as a function of rate of depolarization are shown on the right. The threshold of action potentials, reflected in the step change in the amplitude of the response was a consistent function of the rate of rise of membrane potential in each cell. The cell whose responses are shown in B required only a 5-mV/ms depolarization, the lowest of the cells measured, whereas the cell in C required 12 mV/ms and had a more graded response. Off-line adjustment was made to balance 12 M and 5.5 pF in B and 17 M and 3.4 pF in C.

The finding that the firing of octopus cells depends on the rate of depolarization by current injection raised the question whether responses to synaptic activation are similarly sensitive to the rate of depolarization. Figure 5 illustrates that the responses in three octopus cells to synaptic depolarization and depolarization by current injection follow similar patterns. Three sample responses to shocks in the vicinity of the recorded cell and three sample responses to ramps are shown for each of the cells in the traces on the left. Shocks evoked excitatory synaptic responses in octopus cells that rose from the resting potential about 0.5 ms after the beginning of the shock. As is typical of octopus cells, synaptic responses to shocks are brief, lasting only about 1 ms, and vary in their rise and amplitude as a function of shock strength (Golding et al. 1995). The results of these measurements are summarized in the plots on the right. In all three cells, the firing threshold in response to synaptic stimulation roughly matched the firing threshold in responses to injection of current ramps (Fig. 5).

View larger version (25K): [in this window] [in a new window]   Fig. 5. Comparison of firing threshold when octopus cells are depolarized synaptically with when they are depolarized with current in 3 cells, A-C. For each cell, the upper group of traces shows 3 superimposed responses to shocks to the auditory nerve of different strengths (arrow). Synaptic stimulation caused EPSPs whose rate of rise from rest (dotted line) varied as a function of shock strength. Lower group of traces show responses to families of current ramps like those shown in Fig. 4 in the same cell. Plots of the amplitude of the response as a function of the rate of depolarization (the slope of the dotted line) are shown on the right. The threshold for firing in responses to shocks and injected current matched. Stimulus artifacts were deleted in traces depicting responses to shocks.

To test whether g_KL plays a role in determining the threshold rate of depolarization, the sensitivity of firing to the rate of depolarization was tested when g_KL was blocked. In octopus cells, g_KL could be specifically blocked by -dendrotoxin (Bal and Oertel 2001). -Dendrotoxin blocked g_KL with high affinity; with 5 nM, g_KL was half-maximally blocked. At concentrations over 50 nM, more than 90% of g_KL was blocked. Figure 6 shows that -dendrotoxin caused the firing of octopus cells to become less sensitive to the rate of depolarization. Under control conditions, current ramps that rose to 3.5 nA in more than 1 ms failed to evoke action potentials (Fig. 6A). When 100 nM -dendrotoxin was added to the bath, even slow depolarizations that rose to 3.5 nA in 40 ms evoked large, long action potentials (Fig. 6B). The threshold rate of rise changed by about one order of magnitude, from 12.4 to 1.1 mV/ms (not shown).

View larger version (22K): [in this window] [in a new window]   Fig. 6. -Dendrotoxin, a blocker of the low-threshold K+ current, affects the sensitivity to rate of depolarization of an octopus cell. A: in normal saline, responses to depolarizing current ramps had to rise within <2 ms to evoke an action potential. B: in the presence of 100 nM -dendrotoxin, the same octopus cell fired large, repetitive action potentials. Those large action potentials could be generated even with slowly rising current ramps.

The conversion of small, brief action potentials to large, broad ones by -dendrotoxin shows that g_KL plays a major role in the repolarization of action potentials at the cell body and draws attention to two related features of action potentials in octopus cells, the graded nature and the timing of the peaks recorded in the cell body. The traces in Fig. 6A illustrate characteristic action potentials recorded in cell bodies of octopus cells. Action potentials were small and varied in the height of their peaks depending on how they were evoked. Rapid depolarizations evoked larger action potentials than slow depolarizations. In the presence of -dendrotoxin, the action potentials were larger and the integration time of octopus cells expanded; Fig. 6B shows that depolarization could elicit action potentials whose peaks occurred over a period that was an order of magnitude longer. These findings indicate that the gradedness and small size of action potentials in octopus cells arise from shunting of the inward current by g_KL. In shunting all but the earliest inward current, g_KL restricted the timing of the peak of action potentials to a narrow time window near the beginning of a depolarization (Fig. 6A). These results show that the -dendrotoxin-sensitive conductance contributes not only to the sensitivity to the rate of rise of depolarizations but also to the sharp timing of the occurrence of action potentials in octopus cells under physiological conditions.

The breadth of the action potentials in the presence of -dendrotoxin raised the question whether voltage-sensitive Na⁺ channels or voltage-sensitive Ca²⁺ channels predominated in their generation. Both types of channels have been demonstrated to be present in octopus cells (Golding et al. 1999). The observation that 1 µM tetrodotoxin (TTX) blocked those action potentials indicates that the action potentials are generated at least in part by voltage-sensitive Na⁺ channels (data not shown).

Low-threshold potassium conductances have been shown to prevent repetitive firing and to cause rectification in some cells but not others (Forsythe and Barnes-Davies 1993; Manis and Marx 1991; Rathouz and Trussell 1998). T stellate cells lie immediately adjacent to the octopus cell area in the caudal end of the ventral cochlear nucleus and fire repetitively when they are depolarized with current with action potentials that have a single undershoot (Fujino and Oertel 2001; Oertel et al. 1990). Figure 7 shows that T stellate cells fired action potentials even when they were depolarized slowly. The greater the injected current, the more rapidly they reached threshold after the previous action potential and the more rapidly they fired. The experiment in Fig. 7 shows that this cell's threshold lay in a narrow range of voltages near 52 mV. This T stellate cell, like the two others tested, was insensitive to -dendrotoxin. Neither the shape of the action potential nor the current-voltage relationship was significantly affected by the application of 100 nM -dendrotoxin. (Fig. 8).

View larger version (72K): [in this window] [in a new window]   Fig. 7. In T stellate cells, action potentials are generated independently of the rate at which the cell reaches a threshold voltage. Responses to rapidly rising and slowly rising current pulses are superimposed. In the top traces, current pulses were ramped quickly and slowly to a maximum of 0.2 nA; in the bottom traces, current pulses were ramped over similar times to 0.4 nA. Ramps affected the rate, but not the shape or threshold, of action potentials.

View larger version (14K): [in this window] [in a new window]   Fig. 8. T stellate cells had small or undetectable -dendrotoxin-sensitive conductances. A: the shape of action potentials in a T stellate cell is affected little by 100 nM -dendrotoxin. Note that the undershoot of the action potential has a single component, a characteristic of T stellate cells that distinguishes them from D stellate cells (Fujino and Oertel 2001). B: the current-voltage relationship over the hyperpolarizing range of the same T stellate cell was unaffected by 100 nM -dendrotoxin.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present experiments show that the rate at which cochlear nuclear octopus cells are depolarized determines whether they fire action potentials. They fire over a narrow time range when they are depolarized rapidly and fail to fire when they are depolarized slowly. The sensitivity to rate of rise results from the presence of an -dendrotoxin-sensitive, low-threshold potassium conductance, g_KL. The g_KL also determines the timing of the peaks of action potentials in that it shunts all but the earliest regenerative inward current and restricts the timing of action potentials to near the beginning of a depolarization. Neighboring T stellate cells lack an -dendrotoxin-sensitive conductance and fire action potentials in response to depolarizations independently of whether they are rapid or slow.

Cochlear nuclear octopus cells spread their dendrites perpendicularly across the array of auditory nerve fibers from which they receive their major excitatory input. On average at least 50 auditory nerve fibers terminate on the dendrites of one octopus cell with small boutons that are more uniform in size than in the adjacent multipolar cell area (Golding et al. 1995; R. E. Wickesberg and D. Oertel, unpublished results). Synaptic excitation is through AMPA receptors with large conductance and fast kinetics (Gardner et al. 1999, 2001). The voltage changes produced by synaptic currents are shaped by the biophysical properties of octopus cells. The input resistances and time constants of octopus cells are so low that the duration of synaptic potentials (Golding et al. 1995) is similar to the duration of the synaptic currents (Gardner et al. 1999). EPSPs are between 1 and 2 ms in duration (Golding et al. 1995). The low input resistance also makes the amplitude of voltage changes associated with the activation of individual auditory nerve fibers small (Golding et al. 1995). Octopus cells respond to shocks of the auditory nerve with synaptic responses that are graded as a function of shock strength reflecting the recruitment of variable proportions of large numbers of inputs (Golding et al. 1995). To evoke an action potential requires that between 1/10th and one-half of a cell's total inputs be activated synchronously with a shock (Golding et al. 1995). The more temporal spread in the arrival of excitation, the slower the rate of rise of the resulting EPSP and the larger the number of activated inputs required to bring an octopus cell to threshold (Cai et al. 1997, 2000; Levy and Kipke 1997).

The present findings from experiments in slices are consistent with what is known about the responses of octopus cells to sound. In vivo the convergence of many auditory nerve fibers on octopus cells is reflected in their broad tuning and in their sensitivity to broadband stimuli (Godfrey et al. 1975; Oertel et al. 2000; Rhode and Smith 1986; Rhode et al. 1983; Smith et al. 1993). Octopus cells fire vigorously in response to tones at moderate and high levels at frequencies less than 1,500 Hz, frequencies at which phase-locking synchronizes the firing of auditory nerve fibers (Rhode and Smith 1986). They encode the fundamental frequency in complex sounds with exceptional precision (Rhode 1994, 1998). Octopus cells respond to tones at high frequencies with only a single action potential at the onset transients of tones (Godfrey et al. 1975; Rhode and Smith 1986; Rhode et al. 1983; Smith et al. 1993). Clicks evoke action potentials whose temporal jitter is less than 100 µs (Godfrey et al. 1975; Oertel et al. 2000). Octopus cells can encode periodic stimuli cycle-for-cycle to rates up to 800 Hz, firing rates that are more than double the maximum firing rates of their auditory nerve inputs (Oertel et al. 2000; Rhode and Smith 1986).

The properties of octopus cells contrast with those of T stellate cells, which are their immediate neighbors and which are also driven by input from auditory nerve fibers. T stellate cells are driven by fewer than 1/10th the number of auditory nerve fiber inputs, between 4 and 6 (Ferragamo et al. 1998). Correspondingly, T stellate cells are narrowly tuned and respond tonically with "chopping" (Rhode et al. 1983; Smith and Rhode 1989). The firing properties of T stellate cells can be modulated by small currents, whereas those in octopus cells cannot (Fujino and Oertel 2001). The present results show that octopus cells fire only when they are depolarized quickly, whereas T stellate cells fire independently of how slowly or how quickly they are depolarized.

The sensitivity to rate of rise depends strongly on a conductance that is sensitive to -dendrotoxin. Dendrotoxins isolated from the venom of the Mamba snakes specifically block potassium channels of the Kv1 family (also called KCNA or Shaker) (Grissmer et al. 1994; Hopkins 1998; Hopkins et al. 1994; Owen et al. 1997; Robertson et al. 1996; Stühmer et al. 1989; Tytgat et al. 1995). The  subunits Kv1.1, Kv1.2 (Wang et al. 1994), and Kv1.4 (Fonseca et al. 1998) have been shown to be strongly expressed in brain stem auditory nuclei and especially strongly expressed in the octopus cell area. The -dendrotoxin-sensitive conductance is extraordinarily large in octopus cells, the maximum conductance being on average 514 nS (Bal and Oertel 2001).

Potassium conductances with low thresholds are widespread in neurons that encode timing in the brain stem auditory nuclei of vertebrates. Strong, voltage-sensitive currents activated by small depolarizations from rest are observed as a rectification in many of the neurons that encode timing in the mammalian auditory pathway including bushy cells of the mammalian ventral cochlear nucleus and its avian homologue (Isaacson and Walmsley 1995; Manis and Marx 1991; Oertel 1983; Rathouz and Trussell 1998; Reyes et al. 1994; Wu and Oertel 1984; Zhang and Trussell 1994), principal cells of the MNTB (Banks and Smith 1992; Brew and Forsythe 1995; Forsythe and Barnes-Davies 1993; Wang et al. 1998; Wu and Kelly 1991), principal cells of the MSO (Smith 1995), and neurons in the ventral nucleus of the lateral lemniscus (Wu 1999; reviewed by Oertel 1997, 1999; Trussell 1997, 1999). The low-threshold potassium current was first identified by its sensitivity to 4AP in isolated bushy cells (Manis and Marx 1991). In all the auditory neurons in which it has been studied, the rectification results from the presence of a low-threshold, 4AP- and dendrotoxin-sensitive potassium current (Bal and Oertel 2001; Banks and Smith 1992; Brew and Forsythe 1995; Forsythe and Barnes-Davies 1993; Golding et al. 1999; Isaacson and Walmsley 1995; Manis and Marx 1991; Rathouz and Trussell 1998). Unlike octopus cells, which are contacted by large numbers of excitatory inputs and whose responses to the activation of auditory nerve fibers are graded, other cells in this group receive fewer inputs (1 to about 7), many of which are suprathreshold. Only in neurons in which multiple inputs are subthreshold does the sensitivity to rate of rise increase the sensitivity of a neuron to synchronicity. Synchronicity has been shown to enhance the precision in the timing of firing in bushy cells (Joris et al. 1994a,b; Rothman and Young 1996). In the cells with few suprathreshold inputs, the sharpening of the timing of the peak of the action potential may be the major functional consequence of the presence of g_KL (Forsythe and Barnes-Davies 1993; Rathouz and Trussell 1998; Reyes et al. 1996). Like in octopus cells, g_KL seems to reduce the peak of the action potential as it sharpens its timing.