INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Vocal learning in oscine songbirds is increasingly used as a model system for studying the mechanisms of sensorimotor learning. The male zebra finch is one of the best-studied songbirds, both in its singing behavior and the underlying neural mechanisms. A male zebra finch acquires its song by learning from another male, usually his father. Song learning occurs through two phases (Fig. 1A) (Arnold 1975; Immelmann 1969). During the initial sensory phase, which starts ~20 days post hatching (DPH) and lasts ~30 days for male zebra finches, he listens to a tutor singing and forms a memory or "template" of the tutor song. After 30 DPH, the sensorimotor phase begins. The bird starts practicing to sing and, using auditory feedback, gradually learns to match his own song to the template tutor song. By 60 DPH, his song has many adult-like features and by 90 DPH, the song becomes highly stereotyped or crystallized.

View larger version (20K): [in this window] [in a new window]   Fig. 1. A: time line for zebra finch song learning. The sensory learning phase starts at ~20 days post hatching (DPH) and ends at ~50 DPH. The sensorimotor learning phase starts at ~30 DPH and ends at ~90 DPH, after which songs gradually become more stereotyped and are said to be crystallized. B: highly simplified schematic diagram of the song system. Large open ellipses indicate nuclei in the motor pathway and large filled ellipses indicate those in the anterior forebrain pathway. The connection from area X to DLM is inhibitory ().

More than a dozen interconnected nuclei comprise a circuit known as the song system, which underlies song learning and production. The song system is often simplified into two main neural pathways with different functions (reviewed in Brenowitz et al. 1997). The motor pathway is essential for song production, as lesions in this pathway disrupt singing. It includes nucleus HVc (used as the proper name), nucleus robustus archistriatalis (RA), and brain stem motor and premotor nuclei controlling the syrinx and respiration (Fig. 1B) (Nottebohm et al. 1976; Vicario 1991).

The anterior forebrain pathway (AFP) provides an indirect connection between HVc and RA. It consists of a topographically organized loop connecting area X, the medial portion of the dorsolateral nucleus of the thalamus (DLM) and the lateral portion of the magnocellular nucleus of the anterior neostriatum (lMAN) (Bottjer et al. 1989; Luo et al. 2001; Nottebohm et al. 1976). The AFP in adults has many anatomical and physiological similarities to the mammalian cortico-basal ganglia-thalamocortical loop. Many neurons in area X have intrinsic properties resembling those of mammalian basal ganglia neurons (Farries and Perkel 2002). The area X neurons projecting to DLM provide a strong GABAergic input to DLM (Luo and Perkel 1999a,b). In addition, the majority of DLM neurons have intrinsic properties strikingly similar to those of mammalian thalamocortical neurons (Luo and Perkel 1999a). While lesions of this pathway dramatically interrupt song learning in juveniles, they do not alter song production in adult birds (Bottjer et al. 1984; Scharff and Nottebohm 1991; Sohrabji et al. 1990), indicating that the AFP plays a critical role in song learning. It could do so by providing feedback signals to the motor pathway to guide vocal learning in juveniles (Brainard and Doupe 2000). However, our finding of an inhibitory connection between area X and DLM in adults indicates that this loop does not simply serve an excitatory relay function. Whatever the precise function of the AFP, knowing whether it uses similar processing mechanisms during learning as in adulthood depends on knowing its functional properties in juvenile birds.

The area X  DLM projection forms early (Johnson and Bottjer 1992), but the electrophysiological properties of DLM neurons have been studied only in adult animals (Luo and Perkel 1999a). It is possible that the functional properties of DLM neurons change during development as do those of their targets in lMAN (Boettiger and Doupe 1998; Bottjer et al. 1998; Livingston and Mooney 1997) as well as properties of other synapses in the song system. In addition, the GABAergic area X  DLM connection may not even be functional in juveniles undergoing song learning. Most importantly, in other systems, GABA_A receptor-mediated postsynaptic currents (PSCs) undergo changes during development that include a reduction of the decay time constant of the PSC (Tia et al. 1996) and the emergence of progressively more negative reversal potentials (Ben-Ari et al. 1997). Such changes could render the computation performed by the area X  DLM projection during the sensorimotor learning phase quite different from that in adults. Thus we have examined the intrinsic and synaptic properties of juvenile zebra finch DLM neurons using whole cell recording in a brain slice preparation. We found that the major intrinsic properties of DLM neurons are already in place during the sensorimotor phase of song learning. In addition, the synaptic input to DLM from area X is established and functional at that time. However, voltage-clamp analysis revealed that GABAergic inhibitory PSCs (IPSCs) in juvenile neurons decay more slowly and have a more depolarized reversal potential than in adults, suggesting that processing by this synaptic connection undergoes developmental changes during the sensorimotor phase of song learning.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Juvenile (29-60 DPH) or adult (>110 DPH) male zebra finches (Taeniopygia guttata) were obtained from our breeding colony at the University of Pennsylvania or from a local breeder. All procedures used here were in accordance with a protocol approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania and were consistent with the policy of the American Physiological Society.

Slice preparation

Slice preparation was similar to that described previously (Luo and Perkel 1999a). Briefly, halothane or isoflurane was used to anesthetize a bird deeply. The bird was then decapitated and the brain was rapidly removed. The brain was blocked by first cutting nearly parasagittally ~4 mm from the midline with the anterior end of the blade rotated ~20° laterally and then by cutting midsagittally to separate the hemispheres. The separated hemispheres were glued to the stage of a vibrating microtome with the lateral surface down, and 300-µm-thick slices were then prepared. To improve slice viability, NaCl was replaced with sucrose in the artificial cerebrospinal fluid (ACSF) for slice preparation (Aghajanian and Rasmussen 1989). The ACSF for slicing contained (in mM) 238 sucrose; 1.3 KCl, 1.3 MgCl₂, 2.5 CaCl₂, 1.0 NaH₂PO₄, 26.2 NaHCO₃, and 11 glucose. Slices were incubated for 30 min. in 30°C oxygenated solution that had half of the NaCl replaced by sucrose and for 30 additional min at 22-25°C in normal recording solution, which was identical to the slicing solution except that it contained 119 mM NaCl and no sucrose and had an osmolarity of 304 mosM. To improve the quality of cell filling, in some experiments the osmolarity of the ACSF was increased to 314 by increasing the sucrose concentration to 248 mM for slice preparation or by increasing the NaCl to 124 mM for recording (see RESULTS).

Electrophysiological recording

Slices were submerged and superfused with ACSF at 1-2 ml/min at 22-25°C. Neurons in DLM were recorded using the "blind" whole cell method (Blanton et al. 1989) in current- or voltage-clamp mode. All electrodes were pulled with a P-97 micropipette puller (Sutter Instruments, Novato, CA). The intracellular solution in the recording pipette contained (in mM) 120 K-gluconate, 10 HEPES, 10 EGTA, 8 NaCl, 2 MgATP, and 0.3 Na₃GTP (pH = 7.2-7.4). Electrodes typically had a resistance of 5-12 M for current-clamp recording and 3-6 M for voltage-clamp recording. Whole cell recordings were routinely obtained and usually remained stable for 45 min and sometimes up to 3 h. To evoke synaptic activity, afferents were activated with a single pulse (100 µs; 2-100 V or 0.1-10 mA) with a bipolar stainless-steel stimulating electrode (FHC, Brunswick, ME; 2-5 M) placed within the axonal pathway anterior to DLM, 400-800 µm away from the anterior edge of the nucleus.

Current-clamp signals were amplified using an Axoclamp 2 amplifier (Axon Instruments, Foster City, CA), low-pass filtered at 2-5 kHz, and digitized at twice the filter cutoff frequency. Voltage-clamp signals were amplified using an Axopatch 1D amplifier (Axon Instruments), low-pass filtered at 6-10 kHz, and digitized at twice the filter cutoff frequency. Input resistance and series resistance were monitored throughout the recording by delivering 10-pA hyperpolarizing current pulses or 5-mV voltage pulses and measuring the cellular responses. Only the cells with stable membrane potential, input resistance and series resistance were analyzed. Except for those illustrating firing patterns, current or voltage records are presented here as averages of three to five consecutive traces.

Cell filling

A subset of neurons was filled with neurobiotin (0.5%, Vector Laboratories, Burlingame, CA), which was included in the whole cell patch recording electrodes. Slices were fixed with 4% formaldehyde, cryoprotected in 30% sucrose in 0.1 M phosphate buffer (PB), and sectioned (60 µm) with a freezing microtome. Filled cells were visualized using Cy3-conjugated streptavidin, and data were collected with confocal microscopy. Camera lucida drawing was used in some cells to illustrate their morphology.

Lesions and GAD immunostaining

Surgery and histology were similar to those described previously (Luo and Perkel 1999b). Briefly, animals were anesthetized with pentobarbital sodium (40 mg/kg) and mounted in a stereotaxic apparatus. The skin over the skull was opened, and a small craniotomy was performed over the desired targets. Ibotenic acid (0.1-0.5 µl, 10 µg/µl) was unilaterally pressure injected into area X with a glass micropipette (tip: 30-50 µm) glued to a Hamilton syringe. After the injection, the pipette was withdrawn, and the wound was closed with surgical adhesive (Nexaband, Closure Medical, Raleigh, NC). After a survival time of 3-4 days, the animal was killed with pentobarbital sodium (250 mg/kg) and was perfused transcardially with saline followed by 4% formaldehyde in 0.1 M PB. Each brain was then postfixed in 4% formaldehyde for 2 h and cryoprotected in 30% sucrose in 0.1 M PB. Parasagittal sections were prepared (30 µm thickness) using a freezing microtome.

For GAD immunostaining, sections were incubated in 1:1,500 anti-GAD antibody (AB108, Chemicon, Temecula, CA) and 10% normal goat serum (NGS) in 0.1 M PB for 48 h at 4°C. After rinsing in PB, sections were incubated with goat anti-rabbit biotinylated secondary antibody (1:200) and 10% NGS in 0.1 M PB overnight at 4°C. After rinsing in PB, sections were incubated with 1:1000 Cy3-conjugated streptavidin. After rinsing in PB, sections were mounted with Vectashield (Vector Laboratories), coverslipped, and sealed with nail polish. All data were collected with confocal microscopy (Leica TCS).

Materials

Except as noted, chemicals were obtained from Sigma (St. Louis, MO) or Fisher Scientific (Pittsburgh, PA). Research Biochemicals (Natick, MA) supplied the N-methyl-D-aspartate (NMDA) receptor antagonist DL-2-amino-5-phosphonovaleric acid (APV) and the AMPA receptor antagonist 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX). All drugs were added to the superfusion medium by dilution of a stock solution. Stock solutions of CNQX were prepared in DMSO; all other stock solutions were made in water.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Current-clamp data described here are from 19 juvenile neurons whose resting membrane potential was more negative than 50 mV and showed some overshooting action potentials. In addition, 57 cells recorded under voltage-clamp conditions met our criteria of a stable holding current and a series resistance <30 M. Of these 57 neurons, 12 neurons were from adult male zebra finches and the rest were from juvenile male zebra finches of 29-60 DPH. To facilitate examination of the development of this pathway, some data from our previous studies of current-clamp recording of adult DLM neurons were incorporated. GAD immunostaining was performed on four juvenile males (40-48 DPH), two of which had ibotenic acid injected into area X in one hemisphere (1 right and 1 left). We first report on the intrinsic properties of the DLM neurons, then the GAD immunoreactivity patterns, and finally the synaptic responses to stimulation of the DLM afferent pathway.

Intrinsic properties

Similar to their adult counterparts, juvenile DLM neurons could be classified by their intrinsic properties into two major types, closely resembling the two types found in the adult (Luo and Perkel 1999a). Type I neurons resemble mammalian thalamocortical neurons in having rebound burst firing (Llinás and Jahnsen 1982), and type II neurons resemble mammalian thalamic interneurons (Pape and McCormick 1995; Turner et al. 1997; Williams et al. 1996). In addition, we used voltage-clamp techniques to study the conductances underlying some intrinsic and synaptic properties.

TYPE I NEURONS. Every type I neuron (n = 18) exhibited either a low-threshold Ca spike (Fig. 2A1) or current (B1), which none of the type II neurons (n = 0/6) exhibited. On hyperpolarization, type I neurons exhibited a depolarizing sag ~200 ms after the onset of hyperpolarization (Fig. 2A1, ), which was very small or nonexistent in type II neurons. Current-voltage relations revealed fast outward rectification as the membrane potential change resulting from a depolarizing current pulse was much smaller than that produced by the same amount of hyperpolarizing current (Fig. 2A, 1 and 2). When recorded in voltage-clamp mode, inward currents were usually activated during large hyperpolarizing pulses with an activating time constant ranging from 300 to 1,100 ms (Fig. 2B, ). Following the end of hyperpolarizing pulses from holding potentials near the resting potential (approximately 55 mV), a low-threshold inward current (probably a t-type calcium current) was usually seen (Fig. 2B1). In many cases, rapidly decaying transient outward currents were activated at the beginning of depolarizing pulses, suggesting the existence of I_A (Fig. 2B, 1 and 2).

View larger version (25K): [in this window] [in a new window]   Fig. 2. Membrane properties of type I neurons in the medial portion of the dorsolateral nucleus of thalamus (DLM) from juvenile birds. A1: membrane potential change with current injection of various intensities in a type I DLM neuron recorded in current-clamp mode from a 44-day-old bird. These juvenile type I neurons exhibited a time-dependent hyperpolarization-activated depolarizing sag () and a low-threshold calcium action potential following the end of a hyperpolarizing pulse. Inset: magnified view or one rebound burst at the end of a pulse. A2: the current-voltage relation shows the effects of the slow hyperpolarization-activated inward rectification and a rapid outward rectification. , the membrane potential change measured early in the current injection. , the membrane potential deflection at the end of current injection. - - -, the linear regression of membrane potential changes at the beginning of hyperpolarizing current injection. B1: the currents activated in a type I DLM neuron recorded in voltage-clamp mode from a 29-day-old bird. The cell was held at the resting membrane potential of 60 mV. The time-dependent, hyperpolarization-activated current is indicated (). The transient outward current at the beginning of the depolarizing voltage pulse was probably an A current. B2: the current-voltage relations for this cell show both the outward rectification and inward rectification effect. , current values measured at the beginning of the voltage pulse; , those measured at the end the pulse.

View larger version (18K): [in this window] [in a new window]   Fig. 3. Properties of Ih in juvenile type I neurons. A: current traces from a type I neuron recorded in voltage-clamp mode show that the slow hyperpolarization-activated inward current (A1) could be blocked by 3 mM Cs+ (A2), suggesting it is the H current. The bird was 55 DPH. Voltage commands are shown below A2. B: the amplitude and activation time constant of the Ih at different membrane potentials for 12 type I juvenile neurons. The holding potential was 60 mV. , Ih amplitude; , activation time constant.

View larger version (27K): [in this window] [in a new window]   Fig. 4. Firing properties of juvenile DLM type I neurons. A: in response to depolarizing current injection at the resting membrane potential, a type I neuron from a 44-DPH bird fired action potentials with an intermittent train pattern. The amount of current injected was 50 pA in 1, 100 pA in 2, and 200 pA in 3. B: when the cell was held at a hyperpolarized potential, then given the same depolarizing current pulse as in A3, a burst of action potentials riding on the top of a Ca2+ spike were elicited. Same cell as in A. A and B have the same scales. C: a cell from another 44-DPH animal burst spontaneously during prolonged hyperpolarization, possibly through an interaction of low-threshold Ca2+ current and the inward rectification contributed by H current. D: long-lasting hyperpolarization elicited spontaneous bursting activity at the frequency of the delta oscillation. C and D are from the same cell as in Fig. 2A.

View larger version (31K): [in this window] [in a new window]   Fig. 5. Morphology of type I neurons. A1: camera lucida drawing of a 49-DPH neuron. This class of neuron was characterized by a medium-sized soma and long dendrites. A2: confocal microscopic image of the dendritic processes within the box in 1. B: a 34-DPH type I neuron showing the second type of morphology seen with a large soma and short processes that are often restricted to one side of the soma. Scale bars: A1 = 25 µm, A2 = 10 µm, B = 25 µm.

TYPE II NEURONS. The basic features of type II neurons are already established by 35 DPH. They have much larger input resistance (usually >1 G; range: 655-1,885 M) than type I cells and lack low-threshold Ca²⁺ spikes (Table 1). Six type II neurons were recorded in current-clamp mode. The firing patterns in two of these were different from those of adult neurons (see following text). Voltage-clamp recording revealed additional details regarding this difference. Resting membrane potential and input resistance were not significantly different in juveniles and adults, and no significant correlation was observed between either of these parameters and age within the juvenile category (range: 34-55 DPH; n = 13; P > 0.49).

View larger version (34K): [in this window] [in a new window]   Fig. 6. Membrane properties of type II DLM neurons from juvenile birds. A1: the membrane potential deflections of a 48-DPH type II DLM neuron are shown in response to hyperpolarizing and depolarizing current pulses (shown below). A2: the current-voltage relation for this cell. The dashed line is determined by the resting potential and the membrane potential change with a 10-pA depolarizing pulse. All the points were on or above the line, suggesting rapid inward rectification. B1: the membrane currents activated by clamping a 37-DPH cell to various potentials between 40 and 100 mV from the resting membrane potential of 60 mV. B2: the current-voltage plot for the same cell. The dashed line is determined by the rest condition and the currents immediately following a step to 55 mV. The inward rectification was also revealed by voltage-clamp recording in that most of the points were below the dashed line. Time-dependent inward currents like Ih were activated in some cells, but the amplitude was much smaller (~5 pA at 90 mV) than that observed in type I neurons. Open circles and filled circles represent current at the beginning and end of pulse, respectively.

View larger version (26K): [in this window] [in a new window]   Fig. 7. Firing pattern of type II DLM neurons from juvenile birds. A, 1-3: a type II neuron from a 47-DPH bird exhibited a regular firing pattern in response to depolarizing pulses of 30, 50, and 70 pA. A4: the instantaneous firing rates shown in A, 1-3. Adaptation of firing rates was observed for most of the type II neurons. B, 1-3: another type II neuron from a 38-DPH bird exhibiting initial bursting behavior followed by large adaptation of firing rate in response to depolarizing pulses of 30, 50, and 70 pA. B4: instantaneous firing rates shown in B, 1-3.

GAD immunostaining

To begin to address the synaptic input to DLM, we carried out GAD immunostaining in juvenile animals (n = 4). This revealed that by 40 days the juveniles had an adult-like GAD immunoreactivity pattern in the song system. Large, intensely GAD+ somata were found sparsely distributed in area X (Fig. 8A). Strong GAD immunoreactivity was observed throughout DLM (Fig. 8B1). At high power, dense GAD+ terminals were observed in DLM but no GAD+ somata were found (Fig. 8B2). In many cases, these GAD+ terminals formed baskets the size of DLM somata. In addition, ibotenic acid lesions of part of ipsilateral area X depleted the GAD immunoreactivity in part of DLM (n = 2). The lesioned area in area X (Fig. 8C1) corresponded largely to the GAD-depleted area in DLM (Fig. 8C2). For example, lesion of posterior area X substantially reduced GAD immunoreactivity in posterior and slightly dorsal DLM only, consistent with the topographic organization of the X  DLM projection in adults (Luo et al. 2001). These data suggest that by 40 DPH, the area X  DLM projection is already GABAergic and at least a coarse topography is already established.

View larger version (139K): [in this window] [in a new window]   Fig. 8. GAD immunoreactivity pattern in area X and DLM in juvenile male zebra finches. A1: low-power (×5x) image of GAD immunoreactivity in area X (- - -). , pallial lateral magnocellular nucleus of the anterior neostriatum (lMAN); A2: high-power (×40) image of GAD immunoreactivity in area X showing sparsely distributed, large and strongly GAD+ somata. B1: low-power view of GAD immunoreactivity in DLM; B2: high-power view of GAD immunoreactivity in DLM showing that only terminals were GAD+. C1: a lesion in the posterior and dorsal portion of area X was made by injection of ibotenic acid in the hemisphere contralateral to that shown in B. C2: the area within the posterior and dorsal portion of DLM ipsilateral to the lesioned area X was depleted of GAD immunoreactivity. The ibotenic acid was injected at 44 DHP, and the animal was perfused at 48 DPH. In A-C, anterior is to the right and dorsal is up. Scale bar for A1: 200 µm; A2: 25 µm; B1: 100 µm; B2: 25 µm; C1: 100 µm; and C2: 100 µm.

Synaptic physiology

To test the functionality of inputs to DLM neurons, we examined several aspects of synaptic physiology. In general, it was much more difficult to elicit synaptic responses in juvenile neurons, especially GABAergic ones, than in adult cells. In most cases, higher-intensity stimulation (>1 mA or >70 V) and substantial repositioning of the stimulating electrodes was required. A number of recordings were lost during this process.

Stimulating the afferent pathway ~500 µm anterior to DLM often evoked both inhibitory and excitatory postsynaptic responses. All of the excitatory postsynaptic potentials or currents (EPSPs or EPSCs) were blocked by the AMPA-type glutamate receptor blocker CNQX (10 µM). Although some of the excitatory responses appeared to be monosynaptic, most were polysynaptic, as the latency could be >100 ms.

In the presence of CNQX, GABAergic postsynaptic responses were evoked in 12 juvenile and 4 adult type I neurons. This population represents all cells in which a response was seen prior to addition of CNQX. In all juvenile neurons tested (n = 9), the GABA_A receptor antagonist bicuculline methiodide (BMI) blocked the inhibitory PSP or PSC (IPSP or IPSC, n = 1/1 for IPSP and n = 8/8 for IPSC) in a reversible manner (Fig. 9A), indicating that these responses were mediated by GABA_A receptors. At resting membrane potentials of 51 and 54 mV, the two IPSPs recorded in juvenile cells had peak amplitudes of 3.2 and 2.5 mV, respectively.

View larger version (29K): [in this window] [in a new window]   Fig. 9. Juvenile type I neurons receive functional GABAergic input. A: an evoked postsynaptic current (PSC) for a 60-DPH cell is shown. The PSC was evoked by stimulating the afferent pathway outside of DLM (control). The PSC was resistant to 10 µM 6-cyano-7-nitroquinoxalene-2,3-dione (CNQX) and was completely blocked by 40 µM bicuculline methiodide (BMI; CNQX + BMI). The blockade by BMI was partially reversible (wash). The cell was held at 60 mV and the reversal potential was 54 mV. B: spontaneous PSCs for a 37-DPH neuron are shown. The PSCs had a decay time constant of 8.3 ± 0.8 ms and were completely blocked by CNQX. Bottom: the details of current traces surrounded by the dashed box. Cell was held at 60 mV. C: spontaneous PSCs in a cell from a 55-DPH zebra finch. As in A, these PSCs were resistant to CNQX and were reversibly blocked by BMI. Bottom: the details of the trace defined by the dashed box. The decay time constant for the PSC of this cell in the presence of CNQX was 9.5 ± 1.1 ms. The cell was held at 80 mV.

To test for developmental changes of IPSC waveform, we examined the latency, rise time and decay time constant of the IPSC in juveniles and adults. The IPSC decay time constant for juveniles (15.2 ± 6.3 ms; range: 7.7-25.0 ms), was significantly longer than for adults (9.3 ± 1.8 ms; range: 7.4-10.9 ms; P < 0.05). Within the juvenile group (age range: 34-55 DPH), there was no significant correlation of decay time constant with age (P > 0.17; n = 12). There was no significant difference in either the latency or rise time between juveniles (n = 10) and adults (n = 5, Table 2), or significant linear correlation of these parameters with age within the juvenile group (P > 0.1; n = 12).

Evidence of functional GABAergic input to DLM neurons also came from spontaneous IPSCs. While the vast majority of the type I neurons exhibited spontaneous PSCs with frequencies ranging from 1 to 100 Hz, many of these PSCs had fast decay time constants (3-9 ms) and amplitudes <40 pA at 60 mV. Application of CNQX blocked most of these small PSCs, suggesting they were glutamatergic EPSCs (Fig. 9B). In five cells, however, some of the spontaneous PSCs were resistant to the application of CNQX. These PSCs had longer decay time constants (8-12 ms) and a wide range of amplitudes (40-500 pA). In the one cell tested, these PSCs were completely blocked by BMI and this blockade was reversible (Fig. 9C). The wide range of spontaneous IPSC amplitudes observed suggests the possibility of a spontaneously active GABAergic cell population remaining within the slice and projecting to DLM neurons.

The reversal potential of the evoked IPSCs was measured by stimulating the afferent pathway in the presence of CNQX while clamping the cell at different membrane potentials (Fig. 10A1) and then calculating the reversal potential by using linear regression (Fig. 10A2). In juvenile cells, the IPSC had a reversal potential of 55.4 ± 3.4 (mean ± SE, n = 7). This value is significantly different from 70.5 mV, the equilibrium potential for Cl calculated using the Nernst equation using the concentrations of Cl in the pipette and extracellular solutions (paired t-test; P < 0.01). It is also significantly less negative than the reversal potential of IPSPs and IPSCs in adult cells (65.9 ± 1.5, n = 8; P < 0.05; Fig. 10B). Within the juvenile group, the reversal potential showed a weak but statistically significant positive correlation with age (range 34-55 DPH; P < 0.035; r² = 0.049). Based on the reversal potential and the amplitude of IPSC, the peak conductance of the IPSCs was 12.8 ± 4.1 (SE) nS.

View larger version (16K): [in this window] [in a new window]   Fig. 10. Reversal potential of IPSCs in juvenile type I DLM neurons. A1: evoked PSCs of a cell clamped at different membrane potentials (range: 80 to 32 mV) during the application of 10 µM CNQX. A2: the PSC reversal potential for this cell was 51.5 mV. The straight line indicates the linear regression of IPSC amplitudes at different holding potentials. This cell was from a 47 DPH bird. B: the reversal potential of IPSCs for the juvenile type I cells (n = 7) was significantly (t-test; P < 0.01) more depolarized than ECl (dashed line) determined from the [Cl] in pipette and extracellular solutions. It was also significantly (t-test; P < 0.05) more depolarized than that measured for the adult cells using current-clamp recording.

In each of three cells examined, the amplitude of the GABAergic PSC exhibited all-or-none properties with regard to the stimulation intensity. In the presence of CNQX, stimulation below threshold intensity elicited no response, but slightly stronger stimulation elicited a full-amplitude IPSC, whose amplitude was not increased even by much stronger stimulation (Fig. 11).

View larger version (18K): [in this window] [in a new window]   Fig. 11. All-or-none property of the IPSCs in a juvenile type I DLM neuron. Stimulation intensities <1.4 mA failed to elicit any PSC during the application of CNQX. More intense stimulation (>1.4 mA) elicited IPSCs of similar amplitudes. This cell was from a 55-DPH bird. Top: sample traces with stimulation intensity and IPSC amplitude indicated by the same numbers in the bottom panel.

Stimulation of the afferent pathway elicited PSCs in juvenile type II neurons that were mostly blocked by the application of CNQX (n = 4/4) when the cell was held at its resting membrane potential (Fig. 12A). The latency of the EPSC was 4.4 ± 2.3 ms, longer than that of the IPSC of type I neurons (2.4 ± 1.0 ms, P < 0.05). Spontaneous EPSCs were observed in many of the type II neurons (n = 9/11), and all spontaneous EPSCs were completely blocked by CNQX (n = 2/2; Fig. 12B).

View larger version (19K): [in this window] [in a new window]   Fig. 12. Juvenile DLM type II neurons receive glutamatergic input. A: stimulation of the afferent pathway evoked an EPSC in a 40-DPH cell that was reversibly blocked by 10 µM CNQX. The cell was held at 60 mV. B: spontaneous EPSCs in a 34-DPH type II cell are shown. All spontaneous EPSCs were blocked by CNQX. The expanded trace shows 3 EPSCs of different amplitudes within the dashed box.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The major results of this study are that the intrinsic properties of DLM neurons are largely established by post-hatch day 44, near the beginning of the sensorimotor phase of song learning; the area X  DLM projection is GABAergic and functional during this phase, and in many features resembles that in adults; two physiological properties of this synaptic connection in juveniles, the IPSC decay time and reversal potential, differ from those in the adult. These results suggest that the processing mechanism of the area X  DLM projection is basically similar during the sensorimotor learning phase and adulthood but that quantitative changes in synaptic function occur during development that could affect neural information processing.

Intrinsic properties

By the onset of sensorimotor learning, juvenile DLM neurons have already established their adult-like phenotypes. They fall into two categories, each of which has features similar to those of its adult counterpart. Type I neurons have low-threshold Ca²⁺ spikes and time-dependent inward rectification activated by hyperpolarization. The current underlying the inward rectification closely resembles the I_h observed in mammals. In addition, more detailed studies of the current amplitude and activation time constant as a function of membrane potential revealed strong similarities between the I_h of DLM type I neurons and that of mammalian thalamic relay neurons (McCormick and Pape 1990), further strengthening the similarities between avian thalamic neurons and those in mammals (Strohmann et al. 1994).

The morphological properties observed in juvenile DLM neurons suggests the possibility of at least two subtypes of the type I class. It is possible that the morphological variation among type I neurons corresponds to cells with different physiological properties. Heterogeneity was observed in the firing properties of both juvenile and adult type I neurons, such as the delayed firing in some but not all type I neurons. Quantitative morphometry coupled with a more detailed characterization of intrinsic properties should help test for subtypes of type I neurons. In adults, however, we reported one main type of morphology (Luo and Perkel 1999a). Instead of reflecting a developmental change in morphology, this difference is more likely the result of the larger number of cells filled (18 vs. 6) and improvements in slice preparation and cell filling techniques (e.g., higher osmolarity of ACSF) in the present study.

This study revealed some additional detailed intrinsic properties of type II neurons. Similar to the adult type II neurons, they have large input resistances and lack a low-threshold Ca spike (Luo and Perkel 1999a). With voltage-clamp recording, we found that these neurons have rapid inward rectification and a small degree of time-dependent inward rectification. In addition to the cells with regular firing patterns (n = 4/6), two of these neurons showed bursts of action potentials at ~50 Hz at the beginning of current pulses followed by strong adaptation. This firing pattern is not observed in adult type II neurons, suggesting some developmental change in their firing properties. We had great difficulty in filling the type II neurons, which might be explained by their small somata, an inference based on their large input resistance. The single type II neuron that we filled had a soma ~10 µm in diameter and had short and twisted dendrites. More type II neurons need to be filled to examine whether these are characteristic features.

Synaptic input

Similar to their adult counterparts, juvenile type I neurons receive both glutamatergic and GABAergic input in response to stimulation anterior to DLM, while type II neurons appear to receive only glutamatergic input. Evoked and spontaneous glutamatergic synaptic activity occurs in both types of neurons. Evoked glutamatergic EPSCs with long latencies are likely polysynaptic; apparently monosynaptic glutamatergic EPSCs are also observed, however. These could come from orthodromic or antidromic activation of other DLM neurons making collateral excitatory synapses within DLM or through orthodromic activation of neurons outside of DLM, although it is unclear where the somata of such neurons would lie. One possibility is nucleus RA, which makes a weak anatomical projection to DLM (Vicario 1993; Wild 1993). Another possibility is an as-yet undescribed glutamatergic projection from the forebrain to DLM. It seems unlikely that such a projection could come from area X because all area X somata retrogradely labeled after tracer injections in DLM were GAD positive (Luo and Perkel 1999b).

Evidence for a strong GABAergic projection from area X to DLM in juveniles comes from three observations. First, immunostaining for GAD shows that DLM has dense GAD+ terminals but no GAD+ somata. Strong GAD immunoreactivity is observed in large area X neurons, the soma size of which is comparable to that of the projection neurons (Luo and Perkel 1999b). Lesions of area X substantially reduce GAD immunoreactivity in the corresponding portion of DLM. The indistinguishable GAD immunoreactivity pattern and lesion effect in adult and juvenile animals indicates that by 40 days, the GABAergic area X  DLM projection is well established. Second, type I neurons have spontaneous synaptic activity that is resistant to the glutamatergic AMPA receptor blocker CNQX but is completely blocked by the GABA_A receptor blocker BMI. The spontaneous GABAergic synaptic activity suggests that GABAergic synapses are functional in juvenile animals. Last and most important, stimulating the afferent pathway elicits large all-or-none IPSCs in type I neurons that are resistant to CNQX and completely blocked by BMI. These data indicate that DLM receives strong GABAergic input from outside of the nucleus. Because lesions of area X deplete the GAD immunoreactivity in DLM and DLM itself does not contain GAD+ somata, this GABAergic input most likely arises from area X. The all-or-none nature of the synaptic input in juveniles suggests that, as in adults, there is a low degree of convergence of area X axons onto DLM neurons. In fact, our voltage-clamp data from juveniles complement our current-clamp data from adults (Luo and Perkel 1999a), in which nonlinear summation could conceivably have masked convergent inputs.

While both juvenile and adult DLM type I neurons receive strong GABAergic input that is mediated by GABA_A receptors, developmental changes do occur in this pathway. As in some mammalian systems, the GABA_A receptor-mediated synaptic currents in juvenile DLM neurons have a significantly longer decay time constant than those in adult neurons (Tia et al. 1996; Vicini 1999). Slower decay of the GABAergic input would result in longer and thus more effective inhibitory action on the postsynaptic cell, which could more reliably generate postinhibitory rebound in thalamic neurons with a low threshold Ca²⁺ spike. This effect is confounded, however, by the more depolarized reversal potential in juveniles (see following text). In any case, a longer-duration synaptic current would enhance the efficacy of the synaptic response.

We observed that the reversal potential of the GABA_A receptor-mediated currents in juvenile DLM cells is more depolarized than that in adults. This connection could thus have a different effect on the computations performed during sensorimotor learning. In many mammalian systems, the GABA_A receptor-mediated PSC in neonatal animals is much more depolarizing than that in adults (LoTurco et al. 1995) and is also implicated in a neurotrophic effect (Ben-Ari et al. 1997; Lauder et al. 1998). The changed reversal potential is due to a higher concentration of intracellular Cl in the young cells (Owens et al. 1996; Rohrbough and Spitzer 1996). If the reversal potential is above the membrane potential threshold for spike generation, it could be excitatory rather than inhibitory to the postsynaptic cell. The reversal potential of GABA_A receptor-mediated currents in juvenile DLM neurons (55 mV) is significantly more depolarized than in adults (65 mV) and also more depolarized than the E_Cl determined from internal and external chloride concentrations (70 mV). Cells were recorded using the whole cell technique and thus the intracellular [Cl] was likely affected by the [Cl] in the recording pipette through dialysis. This suggests that the intracellular [Cl] in juveniles is higher than that in the adult and higher than that in the pipette, possibly because of active transport through a Cl transporter in the postsynaptic cell (Rohrbough and Spitzer 1996). At a reversal potential of 55 mV, the GABAergic current evoked in this pathway is still well below the firing threshold of about 40 mV and is thus inhibitory. More accurate methods, such as perforated cell-attached recording (Horn and Marty 1988) or sharp electrode recording, are needed to determine more accurately the intracellular [Cl] and IPSP reversal potential in the juvenile and cells.

Whether area X induces bursting in juvenile DLM neurons depends on the membrane potential of DLM neurons. If their resting membrane potential is hyperpolarized in vivo, firing of the GABAergic afferent pathway will generate depolarizing currents in the DLM neurons and possibly activate Ca²⁺ spikes, which in turn will generate bursts of action potentials and thus become excitatory. Overall, whereas the GABAergic input in juveniles has a reversal potential below the firing threshold and is inhibitory to the postsynaptic cells in the slice preparation, accurate measurement of intracellular [Cl] and the resting membrane potential in vivo will be needed to correctly understand the action of the GABAergic input from area X to DLM during sensorimotor learning (Owens et al. 1996; Verheugen et al. 1999).

Summary

The data from this study directly demonstrate that in juveniles in the midst of the sensorimotor learning phase, the area X  DLM projection in the male zebra finch is both anatomically and physiologically GABAergic. In addition, the intrinsic properties of the DLM neurons are largely the same as those of the adult neurons. Our data thus suggest that the basic processing mechanism of the area X  DLM projection has been established by the start of the sensorimotor phase of song learning. More detailed characterization of the intrinsic properties of DLM neurons, such as the I_h-like current, has strengthened the similarity between avian and mammalian thalamic neurons (McCormick and Pape 1990) and added to the parallels between the AFP and the mammalian basal ganglia-thalamocortical pathway at a cellular level. The computations performed by the AFP during song learning may have many similarities with those made by the mammalian basal ganglia-thalamocortical pathway in sensorimotor learning.

Some properties of the GABA_A receptor-mediated synaptic currents in juvenile DLM neurons are different from those in adult neurons. The change in reversal potential may be especially important, as it could indicate a switch in the area X  DLM projection from excitation during song learning to inhibition after the song is learned. Our measurement of the reversal potential in this study is, however, limited by our invasive method of recording. To understand better the function performed by this pathway in vivo in juveniles, less invasive methods are necessary to accurately measure the reversal potential of GABA_A receptor-mediated currents and the resting membrane potential in vivo.