INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The ability to see clearly during self motion depends on the conjoint operation of the vestibuloocular and optokinetic reflexes (VOR and OKR, respectively). The VOR, evoked by motion of the head, produces compensatory eye movements that rotate the eyes in the opposite direction of head movement. Full-field image motion across the retina evokes the OKR, which produces eye movements in the direction of the visual motion. Differences in vestibular and visual sensory transduction machinery cause the two reflexes to operate over complementary stimulus ranges, such that the VOR responds best to rapid or high-frequency motion, while slow, low-frequency stimuli primarily activate the OKR (Baarsma and Collewijn 1974). Together, the VOR and OKR operate linearly over a wide range of frequencies and amplitudes, ensuring excellent image stability during self motion.

Visual and vestibular sensory signals converge onto neurons in the medial vestibular nucleus (MVN), many of which project to ocular motoneurons that drive compensatory eye movements. The intrinsic firing properties of MVN neurons are well suited to the combined demands of the VOR and OKR: MVN neurons respond to their inputs in a remarkably linear fashion over a wide range of input amplitudes and frequencies, showing little adaptation in response to sustained inputs (du Lac and Lisberger 1995a,b). As such, the MVN neuronal spike generator (which transforms input current into firing rate) can be thought of as a broadband, linear filter capable of signaling information about visual or vestibular stimuli with proportionate changes in firing rate.

The sensory systems responsible for initiating the VOR and OKR mature at different stages of development. In rodents, head motion signals from the semicircular canal afferents are capable of being transmitted to the central vestibular nuclei on the first postnatal day of life, and the vestibular periphery appears to be mature by the end of the second postnatal week (Curthoys 1978, 1979a, 1982; Rusch et al. 1998). In contrast, the eyes do not open until about the 13th day of postnatal life, and central neurons do not respond to visual signals until many days later (Lannou et al. 1979, 1980; Reber-Pelle 1984). Whether the maturation of MVN neuronal firing properties occurs in accordance with vestibular or visual development is not clear. Spontaneous firing rates increase during the first few postnatal weeks in mouse MVN neurons (Dutia and Johnston 1998; Dutia et al. 1995; Johnston and Dutia 1996). However, nothing is known about when MVN neurons develop their mature filtering properties, including the ability to modulate firing rate linearly in response to a broad range of stimuli.

This study examines the development of firing properties in rat MVN neurons during the first month of postnatal life. Spike generation was studied by recording firing responses to intracellular current injected into neurons recorded with whole cell patch electrodes in brain stem slices. The results indicate that the filtering properties of the MVN neuronal spike generator change dramatically during the first few weeks of postnatal development, attaining maturity around the onset of visual experience.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Slice preparation

Brain stem slices were prepared from Long Evans rats from the 1st through the 30th day of postnatal life. Rats were deeply anesthetized with pentobarbital sodium and decapitated prior to dissection of the brain stem from the surrounding skull and nerves. The dissected brain stem was glued onto a chuck with cyanoacrylate, and transverse slices (300-400 µm thick) were cut with a Campden vibrating microtome in ice-cold artificial cerebrospinal fluid (ACSF). Slices were incubated at room temperature in ACSF aerated with 95% O₂-5% CO₂ for at least 1 h before being transferred to the recording chamber.

Electrophysiology

Slices were placed in a recording chamber superfused with aerated ACSF warmed to 32-33°C. Slices were illuminated from below and visualized with a dissecting microscope. The MVN could be identified by its characteristic position adjacent to the fourth ventricles and dorsolateral to the nucleus prepositus hypoglossi. ACSF contained (in mM) 124 NaCl, 2.5 KCl, 1.3 MgSO₄, 26.0 NaHCO₃, 2.5 CaCl₂, 1.0 NaH₂PO₄, and 11 dextrose (pH 7.4; 300-310 mOsm).

Neurons were recorded intracellularly with whole cell patch pipettes pulled from borosilicate glass on a Flaming/Brown electrode puller. Electrodes were filled with an internal solution intended to match the internal composition of the cell. This solution contained (in mM): 122.5 K⁺-gluconate, 17.5 KCl, 8 NaCl, 10 HEPES, 0.1 EGTA, 2 ATP, and 0.3 GTP (pH 7.2; 280-285 mOsm). Electrode resistances ranged from 3 to 14 M (typically 5-8 M).

Electrophysiological recordings were made in bridge mode with an Axoclamp 2B amplifier (Axon Instruments). Weak positive pressure was ejected through the electrode tip to prevent clogging of the electrode during advancement through the slice. The electrode was advanced in ~5-µm steps until a neuron was encountered. On formation of a tight seal (typically 2-4 G), break-in was achieved with a small amount of negative pressure, and the amplifier bridge was balanced. Voltage offsets were corrected for after removal of the electrode from the cell. Junction potentials were not measured. Voltage signals were amplified 50 times, filtered (3 kHz), sampled at 20 kHz, and collected on a PowerMac computer (Apple) using software written in the laboratory with IgorPro (Wavemetrics).

Data analysis

Analyses were restricted to neurons that had action potential amplitudes of at least 50 mV, input resistances of >100 M, and consistent responses to small steps of current throughout the experiment. Spontaneous firing rate was analyzed during at least 5 s of firing. Spike generation was analyzed by measuring firing rate responses to at least three repetitions of each of a variety of amplitudes of intracellularly injected steps of current. To determine the maximum firing rate that could be sustained during current injection, each neuron was challenged with 1-s steps of current that increased in amplitude by 25-100 pA. At some level of current (which varied across neurons and developmental ages), MVN neurons could no longer sustain firing throughout the step (e.g., Fig. 1B). The maximum sustainable firing rate was defined as the mean firing rate evoked in response to three repetitions of the largest input step from which firing could be sustained throughout the stimulus. Control experiments in which step amplitudes were delivered in random order demonstrated that none of the results presented here were influenced by the order of stimulus presentation.

View larger version (45K): [in this window] [in a new window]   Fig. 1. Firing rate responses evoked by intracellular current injection in a postnatal day 3 (P3) and a P20 medial vestibular nucleus (MVN) neuron. In each panel, membrane potential is plotted as a function of time in response to intracellular depolarization with current steps. Responses to 2 different amplitudes of input current (75 and 275 pA) are plotted in the top and bottom panels, respectively, for each neuron. A and B: responses of a P3 neuron. C and D: responses of a P20 neuron.

Spike generation gain was defined as the slope of the mean firing rate-current relationship. The correlation coefficient for this relationship (R²) defined spike generation linearity. The dynamics of spike generation were quantified by an adaptation index (AI), defined as AI = 1  (FR_LATE/FR_EARLY), where FR_LATE is the mean firing rate during the final 200 ms of the step and FR_EARLY is the mean rate from peak firing rate to 100 ms after step onset. AIs were calculated from firing rate responses with peak rates between 30 and 50 spikes/s. Analyses of the time course of adaptation during firing rate responses to 10 s of current were performed on a subset of neurons by fitting a single exponential function to instantaneous firing rate values versus time. Firing rate values evoked during the initial 50 ms following step onset were not well fit by an exponential and were excluded from the fits.

Analyses of action potentials were performed on averages of at least 10 spikes, aligned at their peak membrane potential. As has been discussed by Johnston et al. (1994), in MVN neurons the membrane potential depolarizes gradually preceding each action potential such that action potentials do not have a distinct threshold. In the current study, action potential threshold was determined as the intersection between linear fits to the gradual depolarizing phase preceding the action potential (from 10 ms preceding the action potential peak) and the rapidly rising phase of the action potential (to the peak of the derivative of membrane potential). Action potential height was defined as the difference in membrane potential between threshold and the peak. Action potential width was defined as the time from spike threshold to the threshold crossing during repolarization. Afterhyperpolarization (AHP) amplitude was defined as the difference between action potential threshold and the most negative membrane potential attained during the AHP.

Input resistance was calculated from the change in membrane potential evoked by hyperpolarizing current steps (500 ms) when the neuron was held hyperpolarized with DC current (typically at about 60 mV). To minimize contribution from subthreshold, voltage-dependent conductances, stimuli consisted of small-amplitude (10-50 pA) steps of current. Input resistance values were obtained from averages of responses to 6-10 steps of current. All values are reported as means ± SE unless otherwise noted.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

MVN neurons recorded in brain stem slices fired spontaneous action potentials at all ages tested, from the day of birth [postnatal day 0 (P0)] through the 30th postnatal day of age (P30). During the first few weeks of development, spontaneous firing rates increased in MVN neurons, as has been described previously in mice (Dutia et al. 1995; Johnston and Dutia 1996). Neurons from animals younger than P12 never fired faster than 10 spikes/s (mean 5.2 ± 0.06, mean ± SE, n = 42). In contrast, spontaneous firing rates in older neurons ranged up to 25 spikes/s (mean 9.8 ± 0.09, n = 53).

The filtering properties of MVN neurons (their ability to respond to a broad range of inputs with sustained, linear changes in firing rate) matured over the first 2-3 wk of postnatal life. To study spike generation (the transformation from inputs into firing rate), MVN neurons were challenged with steps of current delivered intracellularly. At all ages, depolarization with intracellular current injection evoked increases in firing rate. However, the pattern and range of evoked responses changed markedly during development. Figure 1 shows the responses of a P3 and a P20 MVN neuron to depolarizing current steps of two different amplitudes. At both ages, the neurons fired spontaneously and responded to a small step of current with a sustained increase in firing rate (Fig. 1, A and C). In the older neuron, larger current steps elicited further increases in firing rates (Fig. 1D). However, the younger neuron was unable to fire reliably during the same amplitude input (Fig. 1B). Instead, action potential amplitude dropped dramatically following the onset the step, and the membrane oscillated around a depolarized level of 25 to 30 mV. This behavior is likely to reflect cumulative sodium channel inactivation consequent to inadequate membrane repolarization following each action potential (Erisir et al. 1999).

Input and firing rate range

Both the range of input currents that MVN neurons could fire in response to, and the range of firing rates that could be sustained during steady depolarization, increased substantially during the first 2 wk of postnatal life. To determine input and firing rate ranges, each neuron was injected with depolarizing current steps that increased in amplitude by 25-100 pA until the neuron could no longer sustain action potentials throughout the 1-s stimulus (e.g., Fig. 1B). A lower bound on the maximum sustainable firing rate was obtained by calculating the mean firing rate during the final 100 ms of the largest current step that could elicit sustained firing. The corresponding input current and firing rate values for each neuron are plotted as a function of age in Fig. 2, A and B, respectively. The filled symbols in Fig. 2 represent the average values of data grouped in a 3-day interval. The dynamic range of inputs to which MVN neurons could respond with modulations in firing rate increased substantially during the first 2 wk of postnatal life. During the first few postnatal days, neurons could not sustain firing in response to input currents >500 pA (Fig. 2A). In contrast, by the third postnatal week, input currents of more than 2500 pA could elicit modulations in firing rate responses. Concomitantly, the range of sustainable firing rates also increased. During the first 4 days of postnatal life, neurons could not sustain firing rates over 50 spikes/s. Thereafter, firing range increased with age until, by P17, most neurons (28/34) could sustain firing rates >100 spikes/s.

View larger version (21K): [in this window] [in a new window]   Fig. 2. Dynamic range and firing rate range of MVN neurons both increase during development. Neurons were challenged with current steps of increasing amplitudes. A: the largest current step that evoked increases in firing rate that were sustained for the duration of the 1-s stimulus is plotted as function of age. B: the mean sustained firing rate evoked by the corresponding current step is plotted as a function of age. Open symbols indicate data from individual neurons. Filled symbols indicate averages of data grouped in 3-day intervals centered on each symbol; lines between filled symbols connect the points.

Linearity

Figure 3 shows typical examples of mean firing rate evoked by intracellular depolarization plotted as a function of input current amplitude in neurons recorded from P4 and P25 rats. Each plot includes data from two neurons at each age recorded from the same slice and covers the entire range of inputs to which the neurons could sustain firing rate responses. At all ages, responses to repetitions of the same stimulus were highly reproducible, such that standard deviations were smaller than the symbols in each plot in Fig. 3. Curves through the data represent the best linear fit (- - -) and the best second-order polynomial fit () to the current-firing rate relationship for each neuron. In the older neurons, firing rate increased linearly with input amplitude over the sustainable range of firing rates (Fig. 3B). However, in the younger neurons, the relationship between input current and firing rate exhibited a nonlinear saturation at the high range of input currents (Fig. 3A).

View larger version (20K): [in this window] [in a new window]   Fig. 3. Relationship between input current and evoked firing rate in immature and mature MVN neurons. A and B plot mean firing rate evoked during a 1-s current step is as a function of current amplitude for each of 2 neurons recorded in single slices obtained from a P4 and a P25 rat, respectively. Data from each neuron at a given age are plotted as triangles and filled circles; error bars representing standard deviations of 3 repetitions of each input step were smaller than the symbols. Solid lines indicate the best 2nd-order polynomial fit, and dotted lines indicate the best linear fit, to each current-firing rate relationship. Gain values (slope of the linear fits) were 94 and 197 (spikes/s)/nA for the P4 neurons and 139 and 213 (spikes/s)/nA for the P25 neurons. The corresponding linear correlation coefficients (R2 values) were 0.959 and 0.984 for the P4 neurons and 0.999 and 0.997 for the P25 neurons, respectively.

Linear correlation coefficients (R² values) of firing rate versus input current from data taken over the sustainable range of firing rates for each neuron are plotted as a function of age in Fig. 4. In neurons younger than P12, R² values ranged from 0.93 to 1.0 (mean 0.982 ± 0.003, n = 42), whereas in older neurons, R² values tended to cluster above 0.99 (mean 0.993 ± 0.002, n = 53). As exemplified by the excellent polynomial fits to the data plotted in Fig. 3, the lower R² values do not arise from noise in the spike generation process, but instead reflect systematic deviations from linearity such that the higher the input current, the smaller the change in firing rate evoked by a fixed change in input current. A measure of this nonlinearity was obtained by calculating the ratio of tangents to polynomial fits to each neuron's current-firing rate relationship evaluated at zero input current and at the maximum input current. A ratio of 1 indicates a perfectly linear spike generator, while ratios less than or greater than 1 indicate a tendency toward sublinearity or supralinearity, respectively. In neurons less than P12, the mean ratio was 0.54 ± 0.01 (n = 42), whereas in older neurons, the mean ratio was 0.94 ± 0.01 (n = 53), confirming that young MVN neurons exhibit a more pronounced tendency toward nonlinear saturation than do their older counterparts.

View larger version (23K): [in this window] [in a new window]   Fig. 4. Spike generation linearity changes during postnatal development. Correlation coefficients (R2 values) derived from linear fits to the current-firing rate relationship for each neuron are plotted as a function of age.

Gain

It is evident from Fig. 3 that spike generation gain (linear slope of the current-firing rate relationship) can vary widely across neurons recorded at the same age and in the same slice preparation. Figure 5A plots spike generation gain as a function of developmental age. Gain values tended to be highest in the first few postnatal days but varied widely at all ages and were poorly correlated with age (R² = 0.12, n = 95). In neurons with sublinear spike generators, such as those plotted in Fig. 3A, gain values derived from linear fits to the entire current-firing rate curve underestimate the responses to small inputs. To exclude this bias, an instantaneous gain was measured for each neuron by fitting the current-firing rate relationship with a second-order polynomial and evaluating the tangent at zero input current. Although instantaneous gain values were somewhat better correlated with age (R² = 0.21, n = 95) than were linear gain values, they also varied considerably across neurons at any given ages (data not shown).

View larger version (20K): [in this window] [in a new window]   Fig. 5. Spike generation gain does not change substantially during development. A: spike generation gain, the slope of the best linear fit to the current-firing rate relationship for each neuron, is plotted as a function of developmental age. Gain values tend to be highest during the 1st few postnatal days but are not well correlated with age. B: input resistance vs. age. Input resistance was calculated from the steady-state change in membrane potential evoked by small hyperpolarizing current steps applied on a DC hyperpolarization sufficient to bring the membrane potential to a stable resting value. C: spike generation gain divided by input resistance for each neuron is plotted as a function of age. Open symbols indicate data from individual neurons. Filled symbols indicate averages of data grouped in 3-day intervals centered on each symbol; lines between filled symbols connect the points.

The observed variability of spike generation gain could reflect heterogeneity in the number of leak channels and/or morphology in the population of MVN neurons recorded. Input resistance varied significantly across ages (Fig. 5B), as expected from a heterogeneous sample of neurons. The slight tendency for younger neurons to have larger input resistances suggests that membrane surface area increases during postnatal development. To determine whether the distribution of gain values shown in Fig. 5A result from variations in input resistance, spike generation gain divided by input resistance is plotted versus age in Fig. 5C. The resulting normalized gain values varied considerably both within and across ages. These data suggest that the ionic mechanisms that dictate spike generation do not change systematically during the first month of postnatal life.

Dynamics

In contrast with gain, the dynamics of spike generation change substantially during the first 2 wk of postnatal development. Figure 6A shows examples of instantaneous firing rate versus time in response to current injection in a P3 and a P25 MVN neuron. In both neurons, firing rate increased to about 40 spikes/s following the onset of the current step, then declined (adapted) during the maintained depolarization. However, the magnitude of adaptation was greater in the younger neuron. The extent of firing rate adaptation was quantified by an AI, which is a measure of the decline in firing rate during a step of input current relative to the firing rate at the onset of the step (see METHODS). Figure 6B shows that the AI tended to decline with age during the first two postnatal weeks. In neurons younger than P10, the AI ranged from 0.11 to 0.56 (mean 0.38 ± 0.003, n = 35). In contrast, the AI range dropped to 0.06-0.38 in P15-30 neurons (mean 0.17 ± 0.002, n = 43).

View larger version (21K): [in this window] [in a new window]   Fig. 6. Spike generation dynamics change during development. A: instantaneous firing rate in response to 3 repetitions of a current step (shown below) is plotted as a function of time for 2 neurons, age P3 and P25. Spontaneous firing rates (5 spikes/s in the P3 neuron, 9 spikes/s in the P25 neuron) have been subtracted from the results to facilitate the comparison. Although the peak firing rates evoked by the current steps were similar in the 2 neurons, firing rates adapted more in the younger than in the older neuron. B: an index of the strength of adaptation (AI; see METHODS) is plotted as a function of age. An AI value of 0 indicates complete lack of adaptation; greater values indicate stronger adaptation. Neurons younger than P10 adapted significantly more than did neurons older than P20. Open symbols indicate data from individual neurons. Filled symbols indicate averages of data grouped in 3-day intervals centered on each symbol; lines between filled symbols connect the points.

To determine whether the time course of adaptation differed in mature and immature MVN neurons, a subset of neurons were challenged with 10-s current steps. Firing rate responses immediately following step onset were not well fit by single or double exponentials, as has been described for hypoglossal neurons (Sawczuk et al. 1995). However, the time course of the step response beginning 50 ms following step onset was well fit by a single exponential in all of the neurons tested. Time constants of adaptation ranged from 0.95 to 6.67 s across all ages (mean 2.4 ± 0.4 s; n = 34) and were not correlated with age (R² = 0.001, n = 34).

Ionic mechanisms underlying developmental changes in spike generation

Developmental changes in the ionic mechanisms that underlie spike generation in MVN neurons can be inferred from differences in the time course of the membrane potential during spontaneous firing in neurons recorded at different ages (Fig. 7). The width of the action potential decreased markedly during development, from a mean of 3.04 ± 0.03 ms in neurons younger than P10 (n = 33) to a steady-state value of 0.98 ± 0.01 ms in P17-P30 neurons (n = 34). Variability in action potential width decreased until about P17 (Fig. 7B), which may reflect differences in the rates of maturation in a heterogeneous population of neurons. Decreases in both rise time (Fig. 7C) and fall time (Fig. 7D) contributed to developmental changes in action potential width, as has been reported in mice (Dutia and Johnston 1998). Such decreases in action potential width during development have been observed in a variety of cell types and are presumed to reflect concomittant changes in delayed rectifier potassium channels that are responsible for rapid repolarization of the action potential (Vincent et al. 2000).

View larger version (29K): [in this window] [in a new window]   Fig. 7. Developmental changes in action potential properties. A: membrane potential is plotted vs. time in these examples of action potentials averaged from each of 3 different neurons, ages P3, P13, and P23. Traces have been offset for visual clarity. B: action potential width decreases during the 1st 17 days of postnatal development. The changes in width are mirrored by developmental decreases in action potential rise time (C) and fall time (D). E: afterhyperpolarization (AHP) peak amplitude is plotted vs. age. F: the time to the peak of the AHP is plotted vs. age, showing a pronounced developmental decline. Open symbols indicate data from individual neurons. Filled symbols indicate averages of data grouped in 3-day intervals centered on each symbol; lines between filled symbols connect the points.

The time course of the AHP also changed during the first few weeks of postnatal development. In young neurons, the membrane potential dropped slowly following each action potential, and AHPs attained their peak amplitude tens of milliseconds after spike repolarization (Fig. 7, A and F). As neurons matured, the AHP became faster, such that by P15, the peak of the AHP occurred within 2-23 ms of the peak of the action potential (Fig. 7, A and F). The amplitude of the AHP varied considerably both within and across age groups but did not change appreciably as a function of age (Fig. 7E). In MVN neurons, the AHP arises from the combined action of calcium-activated K currents, TEA-sensitive potassium currents, and a calcium-independent current (Johnston et al. 1994; Serafin et al. 1991). The slow time course of the AHP in neurons recorded from young animals could reflect developmental changes in calcium buffering or in channel density, kinetics, and/or voltage sensitivity.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

As early as the first day of postnatal life, MVN neurons fire spontaneously and are capable of modulating their firing rate in response to input currents. However, the filtering properties of the MVN spike generator change significantly over the subsequent 2 wk of development. Immature MVN neurons adapt strongly in response to sustained depolarization, can respond to a limited range of input amplitudes, are unable to fire faster than about 60 spikes/s, and tend to show a nonlinear saturation of firing rate responses to increasing inputs. In contrast, by P17, MVN neurons show little adaptation, can fire up to 200 spikes/s in response to a broad range of inputs, and are predominantly linear. Concomittant with these developmental changes in spike generation is a decrease in the width of the action potential and increase in the AHP, suggesting that changes in the expression or properties of potassium channels may contribute to the maturation of the MVN neuronal spike generator.

Functional consequences

The changes in spike generation that occur postnatally have functional implications for the ability of immature MVN neurons to transmit afferent information about vestibular and visual motion signals. The limited dynamic range and tendency toward nonlinearity observed during the first postnatal week implies that immature MVN neurons can respond with proportionate changes in firing rate to a more restricted range of stimuli than their mature counterparts. However, the relatively low response gains of immature vestibular afferents (Curthoys 1979a, 1982) suggests that the head movement signals impinging on immature MVN neurons may fall within their linear operating range. The pronounced spike frequency adaptation observed at early postnatal stages implies that young MVN neurons are best suited to transmit transient or high-frequency information, such as that conveyed by the vestibular system, but are less capable of faithfully transmitting information about sustained or low-frequency movements, such as slow visual motion signals.

Development of vestibular and visual function

The maturation of the spike generator in rodent MVN neurons occurs in the context of a developing peripheral vestibular system. In rats, the semicircular canals expand rapidly after birth and achieve their adult size by the sixth postnatal day (Curthoys 1981). Utricular hair cells acquire their mature electrophysiological phenotypes by the eighth postnatal day in mice (Rusch et al. 1998). Concomittantly, vestibular afferents increase their sensitivity to both galvanic stimulation (Desmadryl 1991) and head rotation (Curthoys 1979a, 1982). MVN neurons are capable of responding to head rotation on the first day of postnatal life (Lannou et al. 1979), indicating that peripheral vestibular signals are being transmitted to the central vestibular system. The gain of MVN neuronal responses to head movement increases during the first 6 days of life (Lannou et al. 1979). This developmental change cannot be explained by increases in spike generation gain, which tends to decrease during the first few postnatal days (Fig. 5), but rather is consistent with increases in vestibular afferent response gains (Curthoys 1979a, 1982).

The functional significance of vestibular neuronal firing early in postnatal life is unclear, since relatively little is known about the development of vestibulo-motor behaviors in rodents. As early as P4, eye movements can be evoked by electrical stimulation of the vestibular periphery (Curthoys 1979b). Given that the VOR is not needed to stabilize images on the retina until after the eyes open at about the end of the second postnatal week, it seems likely that this early processing of head movement information by MVN neurons is used for vestibulo-spinal reflexes and/or to establish appropriate physiological or anatomical properties of the circuitry for the VOR.

In addition to their role in signaling head movement information, MVN neurons respond to moving visual stimuli (Allum et al. 1976; Cazin et al. 1980; Keller and Precht 1979; Waespe and Henn 1977) and are thought to play a role in mediating the optokinetic response (Lisberger et al. 1981). Spike generation in MVN neurons attains a mature state within a few days of eye opening, which typically occurs between P12 and P13 in the Long Evans rats used for this study. Given that the earliest responses to visual stimuli in MVN neurons do not occur until P22 (Lannou et al. 1980), it would appear that the firing properties of MVN neurons are fully mature by the time they are required for the proper function of the optokinetic response. In many sensory systems, peripheral activity influences the development of central neurons (Cline 1991; Kaas et al. 1983; Katz and Shatz 1996). The gradual development of spike generation in MVN neurons and its relatively mature state prior to the time of eye opening precludes a predominant role for visual experience. Whether vestibular nerve activity is required for the proper maturation of MVN neurons remains an open question.

Heterogeneity in spike generation properties

The medial vestibular nucleus comprises a heterogeneous population of neurons that differ in their physiological response properties (Lisberger and Miles 1980; Scudder and Fuchs 1992; Stahl and Simpson 1995), anatomical connections (De Zeeuw and Berrebi 1995; McCrea et al. 1987), and intrinsic ionic conductances (du Lac and Lisberger 1995a; Johnston et al. 1994; Serafin et al. 1991). The wide range of spike generation properties observed in both mature and immature neurons is likely to reflect this heterogeneity of cell types. A rostrocaudal gradient in the development of action potential shape and spontaneous firing rate has been reported previously in the mouse MVN (Dutia et al. 1995) and may account for some of the variability in immature action potential parameters described in this study.

Ionic mechanisms

Developmental changes in potassium channel expression or properties are likely to play a significant role in the maturation of action potential parameters, firing rate range, and spike frequency adaptation observed in MVN neurons. Decreases in action potential repolarization with age occur in many types of neurons and are thought to arise from changes in the expression of delayed rectifier potassium channels (Vincent et al. 2000), while changes in action potential rise times might reflect increases in sodium channel density or kinetics (O'Dowd et al. 1988). The high firing rates that can be sustained by mature MVN neurons are likely to depend on the presence of the Kv3 family of potassium channels, which underlie the ability to fire rapidly in auditory brain stem neurons and fast spiking cortical interneurons (Erisir et al. 1999; Massengill et al. 1997; Wang et al. 1998). Kv3 gene products are developmentally regulated during the first few postnatal weeks (Rudy et al. 1999), and Kv3 channel subunits are expressed at high concentrations in the MVN (C. Sekirnjak, B. Rudy, and S. du Lac, unpublished observations).

Previous studies of the development of vestibular nucleus neurons have implicated changes in potassium channels in maturation of firing patterns in vestibular neurons (Peusner et al. 1998). Immature neurons in the chick tangential vestibular nucleus fire a single spike during maintained depolarization, while their mature counterparts are able to fire repetitively (Gamkrelidze et al. 1998; Peusner et al. 1998). Changes in the expression of a dendrotoxin-sensitive K⁺ channel that occur prior to hatching can account for the developmental switch in firing modes. Rodent MVN neurons fire repetitively in response to sustained depolarization at all postnatal ages, but it is possible that changes in slowly inactivating K⁺ currents could contribute to the maturation of action potential repolarization or firing patterns.

The ionic currents that govern spike frequency adaptation in either mature or immature MVN neurons remain to be identified. Given that calcium channel blockade has little effect on adaptation in MVN neurons (du Lac 1996), it is unlikely that changes in calcium-dependent potassium channels are responsible for the maturation of spike generation dynamics. A sodium-dependent potassium current has been implicated in spike frequency adaptation in cortical neurons (Sanchez-Vives et al. 2000). If such a current were to exist in MVN neurons, then developmental changes in sodium buffering capacities or the current itself could underlie the changes in dynamics observed in MVN neurons.