INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

There are two primary nuclei in the avian auditory brain stem: nucleus magnocellularis (NM) and nucleus angularis (NA; Fig. 1), both of which receive direct input from the auditory nerve (Boord 1969; Carr and Boudreau 1991; Jhaveri and Morest 1982a,b; Parks and Rubel 1978; Ramón y Cajal 1971; Rubel and Parks 1975). NM sends bilateral projections to nucleus laminaris (NL), which is specialized for coincidence detection of interaural time delays (Boord 1969; Carr and Konishi 1990; Joseph and Hyson 1993; Overholt et al. 1992; Parks and Rubel 1975; Smith and Rubel 1979; Young and Rubel 1983). NA, on the other hand, has been classified as the initial site for processing sound intensity information (Moiseff and Konishi 1983; Takahashi et al. 1984). Its neurons send excitatory contralateral input to the dorsal nucleus of the lateral lemniscus (LLD, formerly VLVp: Sullivan and Konishi 1984; Takahashi and Konishi 1988a) and the inferior colliculus (IC). NA also projects to the superior olive (SO), which provides a GABAergic projection back to NA, NL, and NM (Fig. 1; Brückner and Hyson 1998; Carr et al. 1989; Code et al. 1989; Conlee and Parks 1986; Lachica et al. 1994; Takahashi and Konishi 1988b; von Bartheld et al. 1989; Yang et al. 1999).

View larger version (20K): [in this window] [in a new window]   Fig. 1. A: diagram of circuit summarizing nucleus angularis (NA) connections. NA receives excitatory ipsilateral inputs (glutamate) from the auditory portion of the 8th nerve. Cb, cerebellum; NM, nucleus magnocellularis; NL, nucleus laminaris; 8N, auditory portion of 8th cranial nerve. Insert: plane of section. Bar = 1 mm. B: schematic diagram of left NA with the location and orientation of the cells shown in subsequent figures. Frequency axis is indicated by the gradient arrow on the right where dark = low frequency. NA contains a single tonotopic representation, with the highest best frequencies mapped lateral and dorsal and the lowest best frequencies ventral and medial (sparrow, Konishi 1965; pigeon, Boord and Rasmussen 1963; Hotta 1971; chick, Warchol and Dallos 1990; barn owl, Köppl, 2001). Right: data are from a single penetration through chicken NA and show best frequencies of single units from 2,659 Hz (dorsal) to 247 Hz (ventral). Vertical bar = 1 mm and shows penetration depth within NA (from Warchol and Dallos 1990).

Earlier studies of the avian auditory brain stem emphasized the morphological and physiological specializations of NM and NL neurons for temporal coding (Carr and Konishi 1990; Hackett et al. 1982; Jackson and Parks 1982; Parks and Rubel 1975; Raman and Trussell 1992; Raman et al. 1994; Reyes et al. 1994, 1996; Rubel and Parks 1975; Smith and Rubel 1979; Sullivan and Konishi 1984; Trussell et al. 1993; Warchol and Dallos 1990; Young and Rubel 1983). In contrast, much less is known about NA and its putative role in sound intensity processing. We have recently described the cell morphology of NA neurons in the barn owl (Soares and Carr 2001) and the pigeon (Häusler et al. 1999). The in vivo physiology of NA cells has also been described (duck, Konishi 1973; pigeon, Boord and Rasmussen 1963; Hotta 1971; blackbird, Sachs and Sinnot 1978; Sachs et al. 1978, 1980; chick, Warchol and Dallos 1990; barn owl, Köppl et al. 2001; Sullivan 1985; Sullivan and Konishi 1984). Extracellular recordings have shown that many NA neurons exhibit firing patterns similar to those observed in the mammalian cochlear nuclei (Sachs and Sinnot 1978; Sachs et al. 1978, 1980; Sullivan 1985; Sullivan and Konishi 1984; Warchol and Dallos 1990; see Rhode and Greenberg 1992 and Young et al. 1988 for review).

Since NM is devoted to temporal processing, NA must be the source of all other ascending auditory information. It may therefore mediate more aspects of sound processing than just intensity coding. To explore this hypothesis, we have used in vitro whole cell recording to describe the intrinsic firing properties and single-cell morphology of NA. We show that NA is both physiologically and morphologically heterogeneous.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We used in vitro slice preparations of chick embryo (E16 to E19) brainstems to investigate the physiology and morphology of single NA cells. Hearing onset in chickens occurs in ovo at E11 (Saunders et al. 1973), and auditory thresholds are near their adult levels by hatching (Gray and Rubel 1985; Rubel and Parks 1988). In vitro electrophysiology and immunohistochemical studies in NM and NL have found equivalent results between young chickens (age 1-3 wk) and late embryos (Hyson et al. 1995; Kubke and Carr, 2000; Smith and Rubel 1979; Young and Rubel 1986). Experiments were performed in accordance with the guidelines approved by the Marine Biological Laboratory (Woods Hole, MA) and the Florida State University, (Tallahassee, FL) Institutional Animal Care and Use Committees (IACUC). This material was presented in abstract form (Soares et al., 2000) along with a similar physiological study (Harnett and Trussell, 2000).

Brain slice preparation

Chicken embryos were rapidly decapitated. A ~4-mm segment of the caudal skull containing the brain stem was removed with a razor blade and quickly submerged in artificial cerebral spinal fluid (ACSF). ACSF contained the following (in mM): 130 NaCl, 26 NaH₂CO₃, 3 KCl, 2 CaCl₂, 2 MgCl₂, 1.25 NaH₂PO₄, and 10 dextrose. The ACSF was constantly gassed with 95% O₂-5% CO₂ (pH 7.4). The brain stem segment was dissected out of the cranium and transferred to a vibrating blade tissue slicer (Campden) where it was mounted with cyanoacrylate glue, supported by a gel solution (30% gelatin in ACSF), and cut in ACSF. Transverse slices (300 µm) containing NA were collected and placed in a holding chamber at room temperature (25-27°C) in oxygenated ACSF (95% O₂-5% CO₂). Slices were transferred as needed to a submersion-type recording chamber and perfused (1 ml/min) with heated oxygenated ACSF (29-33°C), using either a custom made in-line heater or a Warner TC 324B (Warner Instruments, Hamden, CT). All reported physiological measurements were made from heated slices.

Whole cell recordings

Whole cell patch clamp recordings were made from visually identified NA neurons using infrared/differential interference contrast (IR/DIC) video microscopy (Stuart et al. 1993). Initial pipette resistance was 3-7 M. Cell bodies were patched 50-100 µm from the surface of the slice. Pipette saline was composed of the following (mM): 120 K gluconate, 20 KCl, 0.1 EGTA, 2 MgCl, 2 Na₂ATP, and 10 HEPES (pH 7.3). Pipettes were backfilled with saline containing Lucifer yellow (Molecular Probes, Eugene, OR) or Sulforhodamine (Molecular Probes). Electrical recordings were made using an Axoclamp-2B (Axon Instruments, Foster City, CA) in bridge mode and digitized while connected to a PC running PClamp 8.0 (Axon Instruments) acquisition software or connected to a Macintosh computer running Superscope II software (GW Instruments, Somerville, MA). Analysis of electrophysiological data was performed using AxonGraph 4.5 software (Axon Instruments). Voltage-current (V-I) relationships were constructed from a series of hyperpolarizing and depolarizing current injections. Input resistance (R_N) was calculated by using linear regression analysis of the V-I relationships at the linear range of membrane potential (approximately ±10 mV of resting potential). The slowest membrane time constant (₀) was determined by fitting either single or double exponential functions to identical current injections near resting potential so that no "sag" was included. Half-width was measured at the half-maximum amplitude above the threshold of the action potential.

Anatomy

Whole cell recordings with either Lucifer yellow lithium salt (LY)- or Sulfarhodamine 101 (SR)-filled pipettes allowed for visualization of individual NA neurons. All tracers were diluted in intracellular solution (0.5%). Tracers diffused into cells during recordings, but the loading was enhanced by passing current pulses (1,000 to 1,000 pA) for 10 min. Afterward, slices were kept in the recording chamber under oxygenated ACSF at room temperature for 30 min. Slices were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer. Whole-mount preparations of fluorescent double-labeled slices were prepared by clearing in glycerol and mounting in antifade solution (Molecular Probes).

Multivariate cluster and Matlab analyses

Qualitative neuronal classification was complemented with cluster analysis (SPSS 10 software; SPSS, Chicago, IL) and a Matlab three-dimensional plot (Fig. 5). These techniques were independent, aided in uncovering structure in data sets, and suggested a scheme for classification. Cluster analysis was performed using three physiological measurements from each cell. Data were coded so that the person performing the cluster analysis had no preconceived notions of composition or preferred number of clusters. In this study, clusters were determined by the Ward method, which analyzed the squared Euclidean distance between points. This method allowed for an objective characterization of cell groupings. The plots in Fig. 5, A and B, show aggregation schemes, which began with individual cells at the bottom and then grouped them into hierarchically nested clusters with increasing linkage distance. When linkage distance is small, highly similar cells are joined into a small, low-level cluster, whereas high linkage distances yield large clusters of more dissimilar cells, often spanning multiple lower branches of the dendrogram. Thus linkage distance provides a criterion for progressively joining together cells with increasing dissimilarity. A MATLAB plot was used to represent the data in a three-dimensional format and display clustering results observed in the dendrogram.

Morphological analysis

Slices that contained fluorescent-filled neurons were imaged using either a Zeiss LSM510 or a Biorad MRC 1024 confocal microscope. Images of individual neurons were captured in distinct optical layers. Digital photomontages were reconstructed and adjusted for brightness and contrast using Adobe PhotoShop software (Adobe Systems, San Jose, CA). All photomicrographs were negative images of reconstructions from stacked fluorescent images. National Institutes of Health image software was used to measure the major and minor axis of each cell. Although the tonotopic organization of NA is yet to be precisely described, Warchol and Dallos (1990) showed that this nucleus is organized in a tonotopic manner, with higher frequencies located lateral-dorsal and lower frequencies located medial-ventral (see Fig. 1). The tonotopic axis was used to determine the aspect ratio of each cell by dividing the isofrequency dendritic axis by the axis perpendicular to it. Thus a low-aspect ratio cell had dendrites perpendicular to the presumed isofrequency plane (vertical cells). Intermediate-aspect ratio cells had dendrites that extended in all directions (radiate and stubby cells). High-aspect ratio cells had primary dendrites within the isofrequency plane (planar cells).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Criteria for cell discrimination

Cells were classified according to their responses to current injection and their morphology. First, we determined if the cell responded with one or more action potentials to suprathreshold current injection. Based on this criterion, three classes were observed. The first class responded with only one spike. If a cell responded with a train of action potentials, voltage responses to an increasing family of current injections were determined. If a progressive decrease in amplitude occurred within the train, the cells were classified as "damped." If no decrease in amplitude was observed, cells were classified as tonic. The dendritic morphology was correlated with the physiological responses. Using these criteria, the NA cells were segregated into three types: one-spike, damped, and tonic. Neutral descriptive terms were chosen for these cells until their physiology and morphology are correlated with in vivo physiological types.

Summary of morphological and physiological classification

Although NA neurons are diverse, there is an underlying organization that supports division into functional cell types, each with a unique set of morphological and physiological characteristics. Neurons that responded to current injection with only one spike had stubby radiate dendrites. Neurons whose action potentials showed a decrease in amplitude during current injection showed a consistent planar morphology, with their major dendritic axis located within the presumed isofrequency lamina (Warchol and Dallos 1990). Neurons that responded with tonic action potentials had a radiate morphology. There were three variations of tonic responses: tonic I, with an onset spike at all levels of positive current injection and a small radiate form; tonic II, with a delayed first spike and major dendritic branches perpendicular to the tonotopic axis (vertical); and tonic III, with no onset spikes and large radiate morphology. While each cell class had characteristic physiological responses, morphological features of tonics I and III graded into each other. Therefore a cluster analysis was developed to support tonic cell subdivisions (see Membrane properties and morphology of NA neurons; Fig. 5, A and B). This allowed classification of physiological response types independent of morphological categories.

The morphology of the homologous cells in the barn owl NA was previously described (Soares and Carr 2001). The owls were raised in our animal facility and were not available in sufficient numbers to justify an in vitro study. Nevertheless, the chicken cell types resemble the owl cell types, much as the cat and gerbil cell types resemble each other in the mammalian cochlear nucleus (for review, see Romand and Avan 1997). The four common morphological types in barn owl NA were classified on the basis of their dendritic organization. Planar and stubby cells were confined to an isofrequency plane. Radiate and vertical cells, on the other hand, extended their dendrites across isofrequency planes. These four groupings were largely distinct. Our in vitro physiological types conform to the morphological types described for the barn owl. Furthermore, the observed physiological properties have allowed us to differentiate between two morphologically similar radiate neurons (see Membrane properties and morphology of NA neurons).

Membrane properties and morphology of NA neurons

A total of 110 cells were examined. The entire dendritic arbor could be clearly seen in 41 cells and therefore satisfied our criteria for morphological studies. NA neurons showed a high degree of heterogeneity with respect to their firing properties and morphology, and all cell classes were distributed throughout the nucleus.

One-spike cells

We recorded from 22 cells that produced one spike in response to depolarization. The action potential was characterized by a faster rising phase and a slower decay (Fig. 2C, Table 1). Following the spike, the membrane potential remained depolarized and quickly returned to the resting potential after the end of the current pulse (Fig. 2, A and D). Equal amounts of depolarizing and hyperpolarizing current injection had asymmetrical effects on membrane potential, with rectification in the depolarizing voltage range. The mean resting potential of one-spike neurons was 60 mV (±5.6 mV), and the mean input resistance was 104 M (±27 M; Fig. 2B), the lowest value for any cell type within NA. Seven of 22 one-spike neurons were reconstructed morphologically (Table 1). Cell body areas were about 364 µm² and tended to be round, although a wide variety of shapes were observed. Variations observed included a stubby multipolar branching pattern with two thick, short, primary dendrites emerging on either side of the body, with or without additional thinner primary dendrites. In some cases, the dendrites did not emerge in a radiate fashion but arose from only part of the cell body (Fig. 2E). For all cells in NA, we determined the aspect ratio, or relationship between dendritic span within a frequency band and dendritic span orthogonal to this plane. A symmetrical cell would have a ratio of 1. The aspect ratio was 1.3 for one-spike cells (Table 1).

View larger version (15K): [in this window] [in a new window]   Fig. 2. A: voltage responses of one-spike NA neuron to 300-ms steps of steady current injection. The resting membrane potential of this neuron was 60 mV. Note "sag" during hyperpolarization (arrow). B: plot of the relationship between current injected (pA) and maximum membrane voltage deflection (mV) for this cell. Measurements at peaks (dotted line) and at sustained responses (solid line). Slopes of I/V function obtained from the sustained response were used to determine the mean input resistance of NA cells. C: expanded view of an action potential in a one-spike neuron, evoked with a suprathreshold current step. Firing threshold was defined as the start of the fast rise in the voltage waveform. D: diagram showing the location of spikes with increased current injections. Note that large currents were used (3,000 pA) to be certain that these cells would only produce a single spike. E: photomicrographs of one-spike (stubby) neurons filled with Sulfarhodamine and Lucifer yellow. Bar = 20 µm.

Damped cells

Cells were classified as "damped" when depolarization evoked a large, long-lasting depolarizing event with one or two action potentials followed by subthreshold membrane oscillations (Fig. 3, A and D). A total of 39 damped cells had a mean resting potential of 60 mV (±7.3 mV) and a mean input resistance of 188.2 M (±85.5 M; Table 1). Their action potentials exhibited a successive decrement in their peak amplitudes during the stimulus, and the first action potential was generally narrower than the succeeding ones. For the cell shown (Fig. 3C, bottom), the width of the first spike (measured at one-half the peak height) was 1.8 ms, whereas the action potentials and oscillations that followed had widths of 4-9.8 ms at one-half height. Damping constants varied from fast to slow during depolarizing pulses (Fig. 4). Both slow and fast damped cells showed an increase in damping, or decrease in spike height, when injected with progressively larger positive currents. Negative current produced a hyperpolarization of the membrane potentials characterized by a time-dependent depolarizing "sag." On termination of the hyperpolarizing current, the membrane potential would sometimes rebound above the normal resting potential and result in an action potential (not shown). Damped cells were characterized by a unique planar morphology, with dendrites oriented parallel to the presumed isofrequency contour (Figs. 3C and 1B). The aspect ratio, or relationship between dendritic span within a frequency band and orthogonal to it, was 2.11 (Table 1). Dendrites extended about 117 µm (±37 µm) within the frequency axis versus about 62 µm (±13.5 µm) across it. Damped cells had oval-shaped cell bodies with areas averaging 302 µm² and two or more, long, branched dendrites that originated from opposite poles of the cell body.

View larger version (19K): [in this window] [in a new window]   Fig. 3. Voltage responses of a damped NA neuron to current steps. The resting membrane potential of this neuron was 60 mV. Note "sag" during hyperpolarization (arrow). A: plot showing the relationship of current injected and maximum membrane voltage deflection for the same cell. B: photomicrographs of 2 damped (planar) neurons filled with Sulfarhodamine. Bar = 20 µm. D: location within the nucleus of 3 filled damped neurons. Dorsal is up and lateral is right. Bar = 100 µm.

View larger version (18K): [in this window] [in a new window]   Fig. 4. Damping constant histogram showing variation from fast to slow damping during depolarizing pulses. Damping constants were calculated by fitting an exponential curve to the spike amplitude (ms). All constants were calculated from the first response to show damping.

Tonic cells

Forty-nine of the 110 cells recorded from NA were tonic. Tonic cells fired action potentials, usually in a sustained fashion, during the duration of a positive current step. Upon further increases in the current intensity, tonic cells fired at a rate proportional to the current magnitude, with the number of action potentials increasing until the firing rate saturated. In response to hyperpolarizing current pulses (900 to 3000 pA), membrane voltage deflections exhibited both transient and sustained components. The initial transient deflection was not evident with smaller hyperpolarizing currents (100 to 600 pA). The sustained component appeared as a "sag" in the voltage deflection (Figs. 6-8) and increased in amplitude with hyperpolarization. Within the tonic class, three physiological variations were present: I, II, and III (Fig. 5). These subtypes were differentiated according to both the timing of subsequent action potentials (see next paragraph; Fig. 5), and by morphological criteria.

View larger version (25K): [in this window] [in a new window]   Fig. 5. A: MATLAB 3-dimensional plot of the variations among the tonic cells. Note the 3 clusters. Since the 2 other cell types were easily classifiable, we used these analyses to differentiate among the tonic cell type only. Matlab parameters were chosen to optimize the nature of these cell variations. The primary axis plots the log of the timing of the first spike with respect to stimulus onset. The secondary axis is a dimensionless number which arises from calculating the variable (timing of the last spike  timing of the first spike)/number of spikes. The secondary axis provides a measure of interspike interval. The tertiary axis is the ratio of the timing of the second spike over the timing of the first. It provides a measure of response duration. Open circles, tonic II; gray circles, tonic I; black circles, tonic III. B: dendrogram of tonic cell types obtained by hierarchical cluster analysis using the parameters described above. The cluster analysis was performed using the Ward method analyzing squared Euclidean distances between points.

Twenty-three tonic cells were reconstructed morphologically (Table 1). These cells were characterized by dendrites that were oriented in all directions (radiate) or perpendicular to the tonotopic organization of NA (vertical). Primary dendrites of all tonic cells were long and robust and ramified broadly. They tended to be smooth and arose from a smooth cell body. Secondary and tertiary dendrites were often sparsely spiny. Tonic II cells were differentiated from tonics I and III because their dendrites were oriented orthogonal to the isofrequency plane, conforming to the vertical morphological type of Soares and Carr (2001). Tonics I and III could not, however, be differentiated on the basis of dendritic morphology. To determine if tonic cells were distinct types, we used three physiological measurements from each cell to develop a quantitative neuronal classification scheme (Fig. 5). These measures were derived from the timing of the first spike with respect to stimulus onset, the interspike interval, and response duration (see Fig. 5 for details). The plots in Fig. 5, A and B, show two aggregation schemes. Figure 5A plotted data from all neurons, while Fig. 5B is a dendrogram that began with individual cells at the bottom and grouped them into hierarchically nested clusters with increasing linkage distance. When linkage distance was small, highly similar cells were joined into a small, low-level cluster, whereas high linkage distances yielded large clusters of more dissimilar cells, often spanning multiple lower branches of the dendrogram. Thus linkage distance provided a criterion for progressively joining together cells with increasing dissimilarity. Based on this analysis, the tonic cell class was sorted into three groups. Tonic I (Fig. 5A, gray circles) and tonic III (Fig. 5A, black circles) shared similar response onsets, durations, and interspike intervals and formed two high-level clusters, distinct from tonic II. In this analysis, tonics I and III might be regarded as a continuum except that they had distinct action potential shapes (see Tonic I and Tonic III). Note also that although tonics I and III formed a continuum in Fig. 5A, members of each group were not intermixed.

Tonic I

Tonic I cells showed an increase in spike rate with increasing depolarizing current injections. The distinguishing feature of this type was the order in which the new action potentials appeared. At lower current injections, only one or two spikes occurred at the onset of the stimulus. With increasing current, more spikes appeared in sequential fashion (Figs. 6A, D). Subthreshold membrane oscillations were observed after spikes. Nineteen out of 49 cells were Tonic I. Tonic I cells resembled Tonic III, except that Tonic I had a brief stereotyped action potential undershoot following a spike. The current-voltage relations of Tonic I cells were linear near rest (Fig. 6B), with mean resting potentials of about 57 mV, and input resistances of about 248 M (Table 1). Four tonic I cells were intracellularly labeled (Fig. 6C). They had a radiate dendritic organization, mean cell body sizes of about 321 µm² and a frequency-axis ratio of 1.9. 

View larger version (23K): [in this window] [in a new window]   Fig. 6. A: voltage responses of a tonic I neuron to 300-ms steps of steady current injection. The resting membrane potential of this neuron was 60 mV. Note the "sag" during hyperpolarization (top arrow), the subthreshold membrane oscillations (3-arrow set), and rapid repolarization (star). B: relationship of current injected (pA) and maximum membrane voltage deflection (mV) for this cell. C: photomicrograph of a tonic I (radiate) neuron filled with Lucifer yellow. Bar = 20 µm. D: diagram showing the distribution of spikes with increased current injections.

Tonic II

When depolarized above rest, these cells generated only one or two spikes toward the end of the stimulus (Fig. 7A). With progressively larger positive current injected, an onset spike, then a series of spikes appeared (Fig. 7, A and D). The timing between the first and the second spike was used to distinguish this cell type. It first appeared at the end of the stimulus, but the interspike interval between the first and the second spike decreased with more current until a sustained tonic response was achieved. Subthreshold membrane oscillations were also present. Nine out of 49 cells showed this characteristic. The current-voltage relation was linear near rest (Fig. 7B), the mean resting potential was 62 mV, and the input resistance was about 237 M (Table 1). Tonic II cells were morphologically unique, because they had vertically oriented dendrites. Tonic II cells were not common, and only 3 were intracellularly labeled (Fig. 7C). They had a mean aspect ratios of 0.68, with a mean dendritic length perpendicular to the isofrequency slab of about 109 µm (±9.5 µm) and a within isofrequency slab length of 76 µm (±33.6 µm). Tonic II cell bodies were oval with mean areas of 335 µm² (Table 1).

View larger version (23K): [in this window] [in a new window]   Fig. 7. A: voltage responses of a tonic II neuron to 300-ms steps of steady current injection. The resting membrane potential of this neuron was 64 mV. Note "sag" during hyperpolarization (top arrow), subthreshold membrane oscillations (3-arrow set), and rapid repolarization ramp (star). B: relationship of current injected (pA) and maximum membrane voltage deflection (mV) for this cell. C: photomicrograph of a tonic II (vertical) neuron filled with Lucifer yellow. Bar = 20 µm. D: diagram showing the location of spikes with increased current injections.

Tonic III

Tonic III were distinguished principally by the scalloped shape of their action potential undershoots (Fig. 8A). At lower current injections, tonic III cells showed only one or two spikes in the middle of the stimulus. This pattern did not appear to be consistent from stimulus to stimulus. When more current was applied, more spikes appeared in a nonconsecutive pattern (Fig. 8, A and D). Tonic III cells were common, being 21 of 49 tonic cells. Their current-voltage relation was linear near resting potential (Fig. 8B), the mean resting potential was 58 mV, and input resistance was 295 M (Table 1). Sixteen tonic III cells were reconstructed; cell bodies were round or oval (Fig. 8C; Table 1), with a mean area of 354 µm². The dendritic organization was radiate, and the aspect ratio of dendrites was 1.44. 

View larger version (24K): [in this window] [in a new window]   Fig. 8. A: voltage responses of a tonic III neuron to 300-ms steps of steady current injection. The resting membrane potential of this neuron was 60 mV. Note "sag" during hyperpolarization (top arrow) and scalloped repolarization after the action potential (star). B: relationship of current injected and maximum membrane voltage deflection for this cell. C: photomicrograph of a tonic III (radiate) neuron filled with Lucifer yellow. Bar = 20 µm. D: distribution of spikes with increased current injections.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Understanding the neural mechanisms responsible for sound processing requires a detailed knowledge of the neurons and connections of auditory brain stem. The intrinsic circuitry of NA is important because of its elusive role in sound computation. Toward this goal, we have explored membrane properties and corresponding morphological features of NA neurons. Using a combination of whole cell physiology and a morphological classification previously employed (Soares and Carr 2001), we have identified five physiological types in the chick NA. The heterogeneous nature of NA suggests that it may process more than one aspect of the ascending auditory stream.

Previous studies in nucleus angularis

Previous in vivo physiology studies showed that NA exhibits a variety of physiological responses (Köppl et al. 2001; Sachs and Sinnot 1978; Sachs et al. 1978, 1980; Sullivan 1985; Sullivan and Konishi 1984; Warchol and Dallos 1990). Primary-like, chopper, and onset responses were found in the barn owl and the chicken, while primary-like, pauser, and onset responses were found in the redwing blackbird. Since correlation between in vivo and in vitro responses is difficult to achieve post hoc, we are unable to determine relationships between the in vivo and in vitro data except to note that there are similar numbers of physiological types.

Previous anatomical studies described NA neurons in the barn owl (Soares and Carr 2001) and the pigeon (Haüsler et al. 1999). The barn owl NA has four morphological cell types: planar, radiate, vertical, and stubby. Planar neurons are restricted to an isofrequency band, whereas the dendrites of radiate neurons extend across a presumed band. Vertical cells have long dendrites oriented perpendicularly to isofrequency bands. Stubby cells are the most numerous in the barn owl, and their dendrites are short enough to be confined to an isofrequency band. These cell types are largely discrete categories when considering major morphological characteristics, although some overlap exists (i.e., between large radiates and planars and between large stubbies and small radiates; Soares and Carr 2001). We applied the "owl" classification scheme to our chicken data and found similar types. Further comparative studies will be needed to ascertain that these cell types are found throughout birds.

In the chick in vitro material, neurons with stubby dendrites responded with only one spike. Planar neurons showed a damped response in the amplitude of its action potentials. Vertical neurons showed a delayed response (tonic II) and radiate neurons showed tonic responses to current injection (tonics I and III). Thus physiological response types were congruent with the morphological types first described in the barn owl (Soares and Carr 2001). Are these discrete cell types? Morphological analyses of multipolar cell types in both bird and mammal cochlear nuclei yield what could be considered peaks in a continuum, rather than discrete, nonoverlapping types (Doucet and Ryugo 1997; Soares and Carr 2001). This point becomes clear when all data are presented, rather than representative cell types selected to illustrate the different types (Fig. 5A). The example of tonic I and III cells is a good one, because one might ask whether they belong on a continuum of morphological and physiological characteristics or whether they are separate types. Although tonic I and III cells were not morphologically distinct, in this study we separated them on the basis of their action potential undershoot, which is scalloped in tonic III and rapidly repolarizing in tonic I. Both in vivo physiology and dissection of underlying conductances may support either the continuum or the unique cell type hypotheses.

Comparisons with other avian brain stem nuclei

The neurons of other brain stem auditory nuclei, NM, NL, and superior olivary nucleus (SON), all have physiological and morphological features in common with NA. NM neurons are morphologically and physiologically specialized for the encoding of temporal information (Carr 1993; Oertel 1999; Trussell 1997). They have large round cell bodies with few or no dendrites and phase lock to the auditory stimulus (Jhaveri and Morest 1982a,b; Köppl 1997; Parks 1981; Sullivan and Konishi 1984). By preserving temporal characteristics of acoustic inputs, NM neurons provide the information necessary for coincidence detection of interaural time differences (Carr and Konishi 1990; Joseph and Hyson 1993; Overholt et al. 1992; Warchol and Dallos 1990; Young and Rubel 1983). In vitro studies have shown that NM and NL neurons fire a single action potential in response to depolarizing current injection (Reyes et al. 1994, 1996; Zhang and Trussell 1994). In our study, the one-spike neurons responded similarly to depolarizing current injection, although further studies will be necessary to determine whether this neuron can encode rapid transients and phase lock to the auditory stimulus. Morphologically, one-spike neurons reported here resembled NL neurons in that both have short dendrites, although one-spike neurons showed more variation in the number of dendrites. Unlike NM and NL neurons, SON neurons fire repetitively to depolarizing current steps and show a multipolar morphology with extensive dendritic arborization (Yang et al. 1999). These characteristics parallel those of tonic cells in NA.

Comparisons with mammals

Previous in vivo extracellular recordings from NA have shown that many neurons exhibit firing patterns similar to those observed in the mammalian cochlear nucleus (pigeon: Hotta 1971; Warchol and Dallos 1990; barn owl: Sullivan 1985; Sullivan and Konishi 1984; redwing blackbird: Sachs and Sinnot 1978; Sachs et al. 1978, 1980; see Rhode and Greenberg 1992 and Young et al. 1988 for mammalian review). Morphological types in NA are also similar to those described in the mammalian cochlear nuclei. The morphological classification scheme we used for NA was based on whether or not neurons were confined to an isofrequency plane. This scheme was developed by Doucet and Ryugo (1997) for multipolar cells in the mammalian ventral cochlear nucleus (see One-spike neurons), but it applies well to the avian NA (Soares and Carr 2001). There are also in vitro physiological similarities between mammalian and avian neurons.

One-spike neurons

One-spike NA responses are similar to those exhibited by both bushy and octopus cells in mammalian ventral cochlear nucleus (Golding et al. 1995; Manis and Marx 1991; Wu and Oertel 1984). In these cell types, depolarizing current pulses usually produce a single action potential at the onset of the depolarization. The characteristic properties of both bushy and octopus cells are conferred by a depolarization-activated, dendrotoxin-sensitive, low-threshold K⁺ conductance that is activated at rest and that dominates the biophysical properties of these cells (Bal and Oertel, 2000; Brew and Forsythe 1995; Manis and Marx 1991). A similar low threshold conductance may underlie the responses of NA one-spike neurons, since it is also found in both NM and NL neurons (Rathouz and Trussell 1998; Reyes et al. 1994, 1996) and in the irregularly firing principal cells of the tangential nucleus (Gamkrelidze et al. 1998, 2000). These similarities suggest that one-spike neurons, like bushy, octopus, NM, and NL neurons, may mediate accurate transmission of temporal information (Oertel 1999; Trussell 1999).

Furthermore, in both one-spike and octopus cells, hyperpolarizing current pulses produce transient hyperpolarizations with a time-dependent depolarizing sag of the membrane potential (Golding et al. 1999; Wu and Oertel 1984). In octopus cells, this sag is due to a hyperpolarization-activated, mixed-cation conductance, gh (Bal and Oertel 2000). We do not know if the sag in the membrane potential observed here is due to I_h.

Damped neurons

There are no cells that have been described in the mammalian cochlear nucleus that correspond to damped neurons in NA. Transient neurons in the IC of the rat (Sivaramakrishnan and Oliver 2001) however, show a similar progressive damping of action potentials to membrane oscillations to current injection. This response was unaffected by 4-aminopyridine (4-AP) but altered by charybdotoxin which changed the damped firing pattern to a sustained one.

Tonic cells

During depolarization pulses, tonic I and II cells in NA produced subthreshold oscillations similar to those in the pyramidal cells in the DCN (Manis and Molitor 2001). These low-frequency oscillations were blocked by tetrodotoxin in pyramidal cells (TTX, 500 nM). Voltage-gated Na⁺ channels are therefore required to generate membrane oscillations, and Manis and Molitor (2001) suggest that they play a role in controlling spike timing in neurons when the membrane potential slowly approaches, or hovers near spike threshold. Tonic II cells also resemble DCN pyramidal cells in that both exhibit similar delayed firing patterns in vitro (Kanold and Manis 1999). In our study however, NA tonic II cells did not require prior hyperpolarization to exhibit delayed spikes in response to depolarization. In this regard, tonic II cells also resemble the striatal spiny neurons of the zebra finch basal ganglia (Farries and Perkel 1997). Tonic II, pyramidal, and spiny striatal neurons share two defining properties: fast inward rectification in response to hyperpolarizing current and delayed spike response to depolarizing current. In mammalian DCN and in the bird basal ganglia, the delayed response is mediated by an A-type potassium current that rapidly activates on depolarization and gradually inactivates.

Tonic cells also share features with both D- and T-stellate neurons in the ventral cochlear nucleus (VCN). T- and D-stellates may be physiologically differentiated in vitro by the shape of the action potential undershoot that is rapidly repolarizing in D-stellates and scalloped in T-stellates. T- and D-stellates are further differentiated by differences in inward rectification, which is more prominent and more rapid in D-stellates (Fujino and Oertel 2001). Tonics I and III are similarly distinguished. Tonic I neurons have prominent, rapid inward rectification, while tonic III neurons have a scalloped undershoot and weak rectification. Despite these similarities, tonic cells in NA and neurons in the VCN are unlikely to be homologous, because T-stellates are excitatory neurons that project to the contralateral VCN, and D-stellate neurons are inhibitory glycingeric neurons that suppress activity in T-stellate cells (Davis and Young, 2000; Ferragamo et al. 1998; Gates et al. 1996; Moore et al. 1996; Oertel et al. 1990; Saint Marie et al. 1991; Wickesberg et al. 1994; Wu and Oertel 1984). In the barn owl, all NA neurons project to the midbrain (Soares and Carr 2001).

The parallel evolution of coding strategies: convergence of birds and mammals

Birds and mammals are amniotes that share a common ancestor in the carboniferous (Carroll 1988). Many amniote synapomorphies, or shared derived homologous characters, are widely interpreted as adaptations to life on land. We propose that some features of the mammalian auditory system may be apomorphic, or derived and different from the ancestral condition, and therefore without homology to birds. The lack of homology is due to the separate development of true tympanic ears in the ancestors of both birds and mammals (Clack 1997). Well-known mammalian modifications include the emergence of multiple ossicles and moveable ears. These peripheral changes would have had different reorganizing effects on the ancestral population of brain stem auditory neurons, leading to their parallel evolution with respect to birds.

The observed convergence of morphology and physiology of cochlear neurons is a plausible outcome of parallel evolution, because neurons in both birds and mammals experience similar constraints in detecting the stimulus (i.e., sounds). Thus, although a common population of brain stem auditory neurons existed in the tetrapod ancestor, distinct evolutionary forces acted on these two groups allowing for the emergence of different ears and in turn, dissimilar organization in the brain stem. When the parsimony principle¹ is applied to our comparisons of bird and mammal cochlear nuclei, the many small differences make it unlikely that the cochlear nucleus neurons are homologous.

Although the avian and mammalian cochlear nuclei may have evolved in parallel, they share many physiological and morphological features. The previous notion that NA neurons are only responsible for sound intensity processing (Takahashi et al. 1984) may represent an incomplete story. The diverse response types in NA, and the specialization of the other brain stem auditory pathway (NM to NL), support the argument that NA neurons may also detect other features of sound, perhaps in a similar fashion to cells in the mammalian cochlear nuclei.