Song Decrystallization in Adult Zebra Finches Does Not Require the Song Nucleus NIf

Abstract

In adult male zebra finches, transecting the vocal nerve causes previously stable (i.e., crystallized) song to slowly degrade, presumably because of the resulting distortion in auditory feedback. How and where distorted feedback interacts with song motor networks to induce this process of song decrystallization remains unknown. The song premotor nucleus HVC is a potential site where auditory feedback signals could interact with song motor commands. Although the forebrain nucleus interface of the nidopallium (NIf) appears to be the primary auditory input to HVC, NIf lesions made in adult zebra finches do not trigger song decrystallization. One possibility is that NIf lesions do not interfere with song maintenance, but do compromise the adult zebra finch's ability to express renewed vocal plasticity in response to feedback perturbations. To test this idea, we bilaterally lesioned NIf and then transected the vocal nerve in adult male zebra finches. We found that bilateral NIf lesions did not prevent nerve section–induced song decrystallization. To test the extent to which the NIf lesions disrupted auditory processing in the song system, we made in vivo extracellular recordings in HVC and a downstream anterior forebrain pathway (AFP) in NIf-lesioned birds. We found strong and selective auditory responses to the playback of the birds' own song persisted in HVC and the AFP following NIf lesions. These findings suggest that auditory inputs to the song system other than NIf, such as the caudal mesopallium, could act as a source of auditory feedback signals to the song motor network.

INTRODUCTION

Auditory feedback plays a critical role in maintaining learned vocal patterns in humans and in some adult songbirds (Brainard and Doupe 2000a; Doupe and Kuhl 1999). In adult zebra finches, disrupting feedback by deafening (Nordeen and Nordeen 1992), transecting the vocal nerve (Roy and Mooney 2007; Williams and McKibben 1992), or superimposing delayed feedback or noise onto ongoing singing (Leonardo and Konishi 1999; Zevin et al. 2004) causes previously stable (i.e., crystallized) song to slowly degrade. This form of renewed vocal plasticity, referred to as decrystallization, must ultimately reflect the effects of altered feedback on song motor networks (Brainard and Doupe 2000a). Vocal nerve section has the potential to affect auditory (i.e., by acutely distorting vocalization) and proprioceptive feedback, either of which might contribute to decrystallization. However, decrystallization triggered either by vocal nerve section or by deafening exhibits similarly slow time courses (Nordeen and Nordeen 1992; Williams and McKibben 1992) and involves the same neural pathways (Brainard and Doupe 2000b; Williams and Mehta 1999), suggesting vocal nerve section precipitates vocal plasticity by disrupting auditory feedback. The song system is a specialized network of interconnected nuclei in the songbird brain necessary for singing (Nottebohm et al. 1976) (Fig. 1A), providing the motor network through which auditory feedback must act to trigger decrystallization. Nonetheless, how feedback-related information enters the song system and alters the song motor program remains unclear.

FIG. 1.

Schematic of the song system and the experimental plan. A: schematic of the song system and the central auditory pathways in the songbird. The 2 major pathways of the song system—the song motor pathway (SMP, in red) and the anterior forebrain pathway (AFP, in blue)—mediate song production and learning. Auditory information ascends through the avian auditory thalamus (Ov: nucleus ovoidalis) to the primary forebrain auditory areas called field L, from where it flows to secondary auditory areas, including the caudal mesopallium (CM) and the caudal part of medial nidopallium (NCM). Auditory information enters the song system at the level of HVC (proper name for the nucleus) via the forebrain nucleus interfacialis (NIf) and CM. Note that different subdivisions of field L and the medial and lateral divisions of CM are not shown for the sake of simplicity. Single-headed and double-headed arrows denote unidirectional and bidirectional connections, respectively. RA, robust nucleus of arcopallium; area X, proper name for the nucleus; LMAN, lateral magnocellular nucleus of anterior nidopallium; DLM, dorsolateral thalamic nucleus; nXII, hypoglossal nucleus (12th cranial nerve nucleus); mMAN, medial magnocellular nucleus of anterior nidopallium; Uva, nucleus uvaeformis; resp. areas, brain stem and midbrain respiratory nuclei. B: schematic showing the time course over which vocal nerve section triggers song decrystallization in normal adult zebra finches (top) and the experimental plan used in this study (bottom).

The telencephalic nucleus HVC (proper name for the nucleus) has emerged as the earliest site for the integration of auditory information with explicit song motor commands. In anesthetized zebra finches, auditory presentation of the bird's own song (BOS) strongly excites HVC neurons (Doupe and Konishi 1991; McCasland and Konishi 1981; Mooney 2000). HVC also gives rise to two output pathways essential to song performance and plasticity (Nottebohm et al. 1976). The “song motor pathway” (SMP) (in red, Fig. 1A) generates precise motor signals necessary for singing and includes the HVC, nucleus robustus of the arcopallium (RA), and the brain stem (tracheosyringeal) nucleus XIIts (Hahnloser et al. 2002; McCaslandba 1987; Vu et al. 1994). The “anterior forebrain pathway” (AFP) (in blue, Fig. 1A) drives acute song variability and is necessary to slower forms of vocal plasticity (Brainard and Doupe 2000b; Kao et al. 2005; Olveczky et al. 2005; Williams and Mehta 1999) and indirectly links HVC to RA via area X, dorsolateral thalamic nucleus (DLM), and lateral magnocellular nucleus of anterior nidopallium (LMAN) (Nottebohm et al. 1976). Beyond their motor roles, neurons in both pathways respond to BOS playback and these auditory responses depend on input from the HVC (Doupe and Konishi 1991; Vicario and Yohay 1993). One idea is that these song-evoked auditory responses reflect a functional linkage by which feedback-related signals trigger vocal plasticity. One way to test this idea is to interfere with the normal flow of auditory information into HVC and study the effect of such interference on feedback-dependent song decrystallization.

HVC's major source of auditory input is widely presumed to be the telencephalic nucleus interface of the nidopallium (NIf) (Fortune and Margoliash 1995; Vates et al. 1996). Indeed, permanent (Cardin et al. 2005) or reversible (Coleman and Mooney 2004) NIf inactivation abolishes almost all auditory responses in HVC of the anesthetized finch. However, this result is difficult to reconcile with the finding that bilateral NIf lesions fail to trigger decrystallization in adult zebra finches (Cardin et al. 2005). Although NIf lesions do not interfere with routine song maintenance, they may prevent renewed vocal plasticity in response to feedback perturbations, perhaps because the expression of plasticity requires the full complement of auditory input to the song system. Alternatively, there may be additional sources of auditory input to the song system that are sufficient to support song plasticity.

Here we tested whether NIf is necessary for decrystallization triggered by vocal nerve section. We found bilateral NIf lesions did not prevent song decrystallization following vocal nerve section. We also found strong and selective auditory responses persisted in HVC and LMAN following NIf lesions and, in nerve-sectioned birds, the residual auditory activity in LMAN was selective for the distorted BOS. These findings suggest auditory responses that persist in the song system in the absence of NIf are sufficient to enable nerve section–induced vocal plasticity. Consistent with previous reports (Bauer et al. 2008), reversible inactivation of the caudal mesopallium (CM) strongly suppressed residual auditory activity in LMAN, indicating CM may provide an additional important source of auditory information to the song system capable of supporting vocal plasticity.

METHODS

Animals and experimental groups

Bilateral NIf lesions and vocal nerve sections were performed on a total of 11 adult (age, 125 ± 12 days posthatch, mean ± SD; range, 107 to 145 days) male zebra finches (Taeniopygia guttata) in accordance with a protocol approved by the Duke University Institutional Animal Care and Use Committee. Electrophysiological recording experiments in unilaterally NIf-lesioned birds were carried out in 18 adult birds (age, 132 ± 10 days posthatch, mean ± SD; range, 119 to 150 days).

Neurotoxic lesioning of NIf

Before NIf lesioning, songs of all birds were recorded in the presence of a female bird. Each bird was sedated with 0.04–0.05 ml of 1.5 mg/ml diazepam (Abbott Laboratories, North Chicago, IL) and NIf recording sites were stereotaxically located and marked on the birds' skull. Carbon-fiber microelectrodes (Carbostar-4, 0.4–0.8 MΩ; Kation Scientific) were lowered into the NIf area and multiunit activity was recorded. NIf was identified by its typical spontaneous bursting activity (Cardin and Schmidt 2004; Coleman and Mooney 2004) and the outline of NIf was carefully mapped out. The recording electrode was then withdrawn and a micropipette filled with ibotenic acid was lowered into NIf. Eight to 12 small injections of 25–35 nl of 7 mg/ml ibotenic acid (Sigma, St. Louis, MO) were then made into each NIf by pressure injection through the micropipette connected to a Nanoject pressure injector (Drummond Scientific, Broomall, PA). Each injection was made over the course of 1–2 min and the pipette was withdrawn very slowly to minimize spread of the neurotoxin up the pipette track. For unilateral NIf lesions, ibotenic acid was injected into the right NIf only. Following lesions the birds recovered alone in a solitary cage for a day, after which they were kept in a sound recording chamber for regular monitoring of their song.

Vocal (tracheosyringeal) nerve section

Before nerve section, songs of all birds were recorded in the presence of an adult female bird. Birds were then anesthetized with Equithesin (2 mg/kg, administered intramuscularly: 0.85 g chloral hydrate, 0.21 g pentobarbitol, 0.42 g MgSO₄, 2.2 ml 100% ethanol, and 8.6 ml propylene glycol, brought to a final 20-ml volume with dH₂O) and a 3- to 5-mm section of the right tracheosyringeal nerve was removed, followed by a resuturing of the skin. Postmortem visual inspection in a random sample of birds showed that the nerve did not grow back during the following weeks to months. Following surgery, birds were allowed to recover alone in a solitary cage for a day, after which they were transferred to the main colony where they shared cages with other nerve-sectioned birds.

Song recording and spectral similarity measurement

Spectral and temporal analyses of songs were carried out following methods published previously (Roy and Mooney 2007). Songs were recorded before and at regular intervals (∼7 days) after vocal nerve section in the presence of an adult female zebra finch. Song recording and editing were performed using custom software (LabVIEW, National Instruments). Song spectrograms were subsequently created using the software Sound Analysis-2 (Tchernichovski et al. 2000). For analyzing spectral and temporal changes in song following nerve section, we randomly chose single-song motifs (excluding introductory notes) from each song recording sessions in each bird. Then, to quantify degradation in spectral quality of song following bilateral NIf lesions and vocal nerve section, we used the “percentage significant similarity” measure in Sound Analysis-2. Briefly, this measurement contrasts acoustic features such as pitch, Wiener entropy, sound continuity, and frequency modulation between small segments of two given songs and returns a statistical estimate of overall similarity between the two songs in the form of a “percentage significant similarity” score. We calculated “percentage significant similarity” in the “overall” mode with the following settings: threshold, 92%; interval, 70 ms; and sections, 20 ms. Using this technique, we first compared with each other 10 pairs of song motifs recorded before NIf lesions or nerve section in each bird. This gave us a “pre-NIf lesion” or “pre nerve section” estimate of spectral variance already existing in normal songs. Subsequently, 10 song motifs recorded at different intervals following NIf lesions or nerve sections were compared with 10 pre-NIf lesion or pre nerve section motifs, to obtain an estimate of the degradation in spectral qualities induced by the respective manipulation. For analysis of spectral degradation following NIf lesions, such comparisons were made for songs recorded on multiple days after the lesions, all of which are shown in Fig. 2B for each bird. For nerve-sectioned birds, we wanted to obtain an average spectral degradation plot for all birds. However, song recordings in each bird were done on slightly different days following nerve section and therefore direct averaging of similarity scores for one particular time point across birds was not possible. Therefore to obtain a mean similarity measurement across birds, we pooled together from all birds the similarity scores obtained for the first, the second, and the last song recordings (marked post1, post2, and post3, respectively, in Fig. 3C) following nerve section and averaged them. The first, the second, and the last song recordings were made at 5 ± 1.1 (means ± SE), 11 ± 3.1, and 51 ± 9 days, respectively, following nerve section.

FIG. 2.

Bilateral NIf lesions do not lead to decrystallization of adult zebra finch song. A: spectral derivative plots of songs from one adult male zebra finch show that the song's spectral and temporal structures remain stable following bilateral NIf lesions. The yellow lines superimposed on each spectrogram denote the filtered and smoothed amplitude waveforms of the songs. The colored bars at the bottom of each spectrogram denote identity of individual syllables, as estimated by experienced human observers. B: bilateral NIf lesions did not degrade the spectral features of the bird's song. Each colored line denotes percentage spectral similarity (see methods) to pre-NIf lesion song at different intervals following NIf lesions in a single adult zebra finch. Time point zero (marked “pre”) represents the mean within-bird spectral similarity observed between multiple bouts of pre-NIf lesion songs; no significant decreases from this baseline mean spectral similarity were detected at any time following NIf lesions. C: bilateral NIf lesions had no effect on song temporal structure. We estimated the stability of the songs' temporal structure by comparing the amplitude envelopes of songs recorded at different times following NIf lesions to those recorded immediately before NIf lesions and expressed it as a temporal integrity (TI) score (see methods). The horizontal line marks the mean pre-NIf lesion stereotypy score minus 2SDs, a level below which song was operationally defined as decrystallized. Each symbol type and colored line are used to denote the TI scores for a single bird. Note that the TI scores in none of the 11 birds fell below this threshold following NIf lesions even at late times, indicating that the NIf lesions did not affect song temporal structure.

FIG. 3.

Bilateral NIf lesions do not prevent song decrystallization induced by vocal nerve section. A: photomicrographs of cresyl violet–stained sagittal sections of brain tissue from a bird with bilateral NIf lesions. Two sections, separated from each other by about 150 μm, from each hemisphere are shown. Black arrows on each section outline the boundaries of the lesions; note the densely packed, small cell profiles at the lesion site. Bottom: a single section from a different bird with an intact NIf is shown at higher magnification for comparison with the lesioned cases. In the lesioned bird shown here (green615; see Table 1), 100% lesion of NIf was achieved in both hemispheres. B: spectral derivative plots of songs before and at different times after vocal nerve section in 2 adult male zebra finches with bilateral NIf lesions. The colored bars below each plot denote identity of individual syllables, as estimated by experienced human observers. Note the increased spectral distortion in all syllables shortly after nerve section in both birds. The filtered and smoothed amplitude waveforms (yellow lines superimposed on each spectrogram) for the songs indicate that the large-scale song temporal structure as defined by the order of syllables was maintained shortly after nerve section in both birds, but in the plastic bird (left) underwent alterations between 12 and 57 days post nerve section, while remaining intact in the nonplastic bird (right). C: mean percentage spectral similarity to pre nerve section song (y-axis, means ± SE) at different intervals following nerve section (x-axis, means ± SE) for 11 bilaterally NIf-lesioned birds. Time point zero (denoted “pre”) represents the mean within-bird spectral similarity observed between multiple bouts of pre nerve section songs. Time points denoted post1, post2, and post3 represent the average intervals following nerve section at which the first, second, and the last song recordings were made. Dotted lines denote mean ± 1SE of spectral similarity between unrelated normal adult zebra finch songs recorded from our colony. D: TI scores for songs of normal birds (left, “between birds” or “within bird, within day” comparison) and for songs of individual birds with bilateral NIf lesions at early and late times following vocal nerve section. Open circles: TI scores for individual birds. Filled circles and error bars: means ± SE for a given group; the mean late TI scores are reported separately for plastic (filled square) and nonplastic (filled diamond) birds. The horizontal line marks the mean pre nerve section stereotypy score minus 2SDs, a level below which song was operationally defined as decrystallized. In plastic birds, the “late” TI scores had decreased to a level approaching TI scores obtained between different zebra finch songs. In nonplastic birds, songs remained stable, as reflected in TI scores resembling stereotypy scores in normal adult zebra finches. Compare these TI scores to those plotted in Fig. 2C, from birds subjected to bilateral NIf lesions but where the vocal nerves were left intact.

Analysis of song temporal structure

For analysis of song temporal structure, amplitude envelopes of song motifs were cross-correlated using custom MATLAB software. This analysis was based on the principle that if two song motifs are very similar to each other in their overall temporal structure (i.e., syllable sequence), cross-correlation between the amplitude waveforms of the two should return a high cross-correlation coefficient (for details see Roy and Mooney 2007). We half-wave rectified the signal, then filtered and smoothed the rectified-amplitude envelopes using the decimate function in MATLAB (with the original sampling rate of 22,050 Hz and r = 10). Next, we carried out cross-correlation between these filtered amplitude envelopes using the xcorr function in the default mode. The correlation coefficient was then normalized in the following manner: for a given waveform, the amplitudes of all the points were squared and summed and the square root of the sum was taken as the power in the signal. Then the cross-correlation coefficient was divided by the product of the power in the two waveforms being compared and the peak-normalized cross-correlation coefficient was taken as an indication of the match between the two waveforms. The values of the coefficients thus obtained approached zero for two very dissimilar waveforms and approached 1 for two very similar waveforms.

Assessment of song decrystallization

Using this method, we compared song motifs of a single bird from 2 different days, following NIf lesions or vocal nerve section, and the mean peak-normalized cross-correlation coefficient thus obtained was defined as the “temporal integrity” (TI) score, signifying the relative degree to which the song structure was stable between those 2 days. We also estimated the variability in song temporal structure on a given day by calculating the mean-normalized cross-correlation coefficient for 10 randomly chosen motifs within that day. This value was defined as the “stereotypy score,” which was essentially equivalent to a “within-bird, within-day” TI score. The stereotypy scores were calculated both before and on different days following NIf lesions and vocal nerve section. The overall mean stereotypy score for all birds prior to NIf lesions or vocal nerve section was high (0.86 ± 0.02, n = 11; mean ± SD). We set an operationally defined threshold for song decrystallization as 2SDs below this mean stereotypy score: following either NIf lesions or vocal nerve section, any TI score falling below this threshold was used as the criterion for detecting song decrystallization.

Following nerve section, we calculated for each bird a mean TI score by cross-correlating the amplitude envelopes of 10 song motifs from the first song recording session post nerve section (post1) with those of 10 song motifs from subsequent recording sessions. Comparison between the first and second recording sessions (post1 vs. post2, Fig. 3C) provided an estimate of temporal integrity at “early” times following nerve section (Fig. 3D). Likewise, comparisons between the first and the last song recordings (post1 vs. post3) gave us an estimate of temporal integrity at “late” times following nerve section (Fig. 3D). We did not compare the post nerve section song motifs directly to the pre nerve section song motifs for two reasons. First, although amplitude waveforms of songs immediately following nerve section roughly resembled those of pre nerve section songs, the distortions in spectral features caused by the manipulation induced some alterations in the amplitude waveforms and thus made the comparisons less faithful purely to the temporal structure of song. Second, very rarely syllables (often those involving a high pitch click) were dropped immediately after nerve section, potentially signifying that those syllables were being generated solely through the right syrinx. Thus comparisons of pre and early post nerve section motifs in those birds returned lower TI scores that did not reflect auditory feedback-mediated alterations in the song temporal structure. Therefore we took the earliest recording following nerve section as the starting point from which to monitor the integrity of song temporal structure.

In vivo electrophysiology and data analysis

On the morning of the day of the electrophysiological recording, birds were injected in the pectoral muscle with 20% urethane (75–105 μl total; Sigma), administered in three 25- to 35-μl doses at 30-min intervals. LMAN and HVC recording sites were stereotaxically located and marked on the birds' skull. Then birds were immobilized via the mounted post in a sound-attenuating chamber (Industrial Acoustics, Bronx, NY) on an air table (TMC, Peabody, MA); body temperature was maintained by an electric blanket at 37°C (Harvard Apparatus, Holliston, MA). Following topical application of xylocaine (2%), the scalp was retracted, a small craniotomy (width, <300 μm) was made over spots marked for LMAN and HVC, and the dura was slit open with a fine insect pin (Minuten, Carolina Biological Supply).

Extracellular multiunit recordings in LMAN and HVC were carried out using carbon-fiber microelectrodes (Carbostar-4, 0.4–0.8 MΩ; Kation Scientific). Recorded extracellular signals were amplified via an A-M systems model 1700 differential amplifier (Everett, WA), band-pass filtered between 300 Hz and 5 kHz, digitized at 10 kHz, and stored on a PC hard drive. Peristimulus time histograms (PSTHs; 25-ms bin width) were computed on-line (see Data analysis). On conclusion of recordings, 5% biocytin was iontophoretically injected (2-μA positive current, 7-s on–off cycle, for 8–10 min) through the barrels of Carbostar-4 electrodes to mark the recording site.

Song stimuli

To record auditory responses, ≥15 iterations of each auditory stimulus, delivered at intervals ranging from 6 to 10 s, were presented at about 70 dB (root-mean-squared, A-weighting) through a speaker positioned 20 cm directly in front of the bird. The auditory stimulus always presented during electrophysiological recordings included the bird's own song recorded on the day of electrophysiology (BOS). To assess selectivity of BOS responses in nerve-sectioned birds undergoing song decrystallization (i.e., plastic birds), we played back the most recent version of BOS recorded on the day of electrophysiology (new BOS) and the BOS recorded prior to nerve section (old BOS). The new and old BOS stimuli in all birds were chosen randomly, but in some plastic birds singing songs with variable durations a new BOS motif was chosen that was comparable in duration to that of the old BOS. In some cases, to assess general selectivity of auditory responses, we also played back the BOS played in reverse (rev BOS) and songs of other adult zebra finches (conspecific or CON).

Data analysis

The threshold for detecting units within HVC and LMAN multiunit activity was set visually by the user at an amplitude above background that excluded the smaller amplitude events. The responsiveness or firing rate (R_FR) of multiunit recording sites was calculated by R_FR = S_FR − B_FR, where S_FR and B_FR are the firing rates during each stimulus presentation and during a 1.5-s baseline period before each stimulus presentation, respectively. Average R_FR values were computed for ≥15 stimulus iterations. Significance was determined with paired t-test comparing stimulus-evoked responses to corresponding baseline measures. We rarely encountered sites that responded to auditory stimuli with decrease in activity, and therefore for the purpose of this study, focused only on increases in activity induced by stimulus presentation. Subsequent response strength and selectivity measurements were carried out only for sites that exhibited significant auditory responses to one or more of the presented auditory stimuli, including BOS, reverse BOS, or CON. Next, to compare response strengths across different birds, the response strengths calculated as before were converted to Z scores. The Z score (Z_FR) is given by the difference between the average firing rate during stimulus presentation and that during a 1.5-s baseline period before stimulus presentation, divided by the SD of this difference

where S̄_FR is the mean firing rate during the stimulus and B̄_FR is the mean firing rate during the baseline period and the denominator is the SD of S_FR − B_FR.

The selectivity of a given recording site for one stimulus versus another was measured with the psychophysical metric d′, which estimates the discriminability between two stimuli. The d− value comparing the responses to two stimuli, say, stimulus 1 and stimulus 2, is given by

where R̄ is the mean value of R_FR (as described earlier), and σ² is its variance. These d′ values were calculated for comparisons of responses to forward BOS versus reverse BOS (d_fwd–rev), BOS versus conspecific songs (d_BOS–CON), and in the case of nerve-sectioned birds, the new BOS versus the old BOS (d_N–O).

In unilaterally NIf lesioned birds, we used the same recording electrode to compare the multiunit responses of HVC and LMAN from the two hemispheres. The same manually set threshold was used for analyzing all multiunit sites recorded in the two hemispheres from an individual bird. For LMAN, in a single bird we averaged the response strengths (R_FR calculated in spikes/s) of all multiunit sites recorded on the intact side and then normalized the response strengths of all sites recorded on the lesioned side by the average value for the intact side. A similar response on both sides should thus yield a value close to 1. For HVC, there was an added complication. We found that spontaneous activity in HVC was strongly diminished following NIf lesions. Because response strength measurements subtract spontaneous activity from the stimulus-evoked activity, this reduction in spontaneous activity would result in higher BOS-response strengths being reported in NIf-lesioned birds. Therefore for estimating response strength in HVC, we directly measured the stimulus-evoked response strength without subtracting spontaneous activity. The raw response strength thus measured on the intact side was then used to normalize the responses on the lesioned side as described for LMAN.

Reversible inactivation of HVC and CM

In a subset of unilaterally NIf lesioned birds, HVC ipsilateral and contralateral to the lesioned NIf were reversibly inactivated by injections of lidocaine. Following identification of HVC with multiunit recordings, a micropipette filled with 2% lidocaine was lowered into HVC and small volumes of lidocaine were pressure injected using a Picospritzer (duration, 20 ms; pressure, 20 psi; three to six pulses; General Valve, Fairfield, NJ). For reversible inactivation of CM, small volumes (70 nl) of 0.25 M γ-aminobutyric acid (GABA) solution (mixed with 5% Alexafluor-594) were pressure injected (duration, 20 ms; pressure, 20 psi; one to four pulses) into CM. We estimated the volumes of drugs injected by each pulse of the Picospritzer before the injections were made. With the pressure and duration on the Picospritzer set to fixed values, we made a one-pulse injection of the GABA solution into a large drop of mineral oil. Then under the microscope we measured the diameter of the small droplet of GABA solution floating in the oil and calculated the volume of the single droplet, assuming the droplet was a perfect sphere. We restricted the volume injected by each pulse between 10 and 25 nl.

Histology

After each recording session, birds were deeply anesthetized with Equithesin and transcardially perfused with 0.9% saline for about 3 min, followed by 4% paraformaldehyde (PFA) in 25 mM sodium phosphate buffer for 30 min. Brains were removed and postfixed in 4% PFA with 30% sucrose overnight, blocked sagittally, and sectioned on a freezing microtome at 60 μm. We confirmed the extent of the NIf lesions using two strategies. In some birds sections were stained with cresyl violet and the outlines of the lesions were visualized under dark-field illumination. Using Axiovision 4 (Carl Zeiss) software we drew outlines around the portions of NIf that remained intact despite the injections, measured the area of remaining NIf, summed this area across multiple sections, and then multiplied this sum by the section thickness (60 μm), to obtain an estimate of the volume of NIf left intact in each hemisphere. Using the same method, we also estimated the volumes of NIf from 10 intact hemispheres from a separate set of seven nonlesioned control adult male zebra finches. The volumes of NIf measured from these control birds had a reasonably small variance (coefficient of variation, 11.2%) and thus the mean volume of the intact nucleus was then used to quantify the percentage of NIf volume remaining in each lesioned bird. For each hemisphere in the lesioned birds, the estimated NIf volume remaining intact was divided by the mean total NIf volume obtained in nonlesioned birds and the fraction multiplied by 100 to obtain the percentage NIf remaining intact. This value was subtracted from 100 to obtain the percentage NIf lesioned in each hemisphere.

In some of the NIf-lesioned birds, including all 18 birds that received unilateral NIf lesions, we confirmed the lesions by using fluorescent retrograde tracer dyes. About 200 nl of 5% dextran-conjugated Texas Red (Molecular Probes, Eugene, OR) was injected into HVC 2 days before electrophysiological recordings. Following recordings, brain tissues were fixed and sectioned as described earlier, and 60-μm sections were visualized under fluorescent illumination. In nonlesioned hemispheres, retrograde labeling was clearly observed in NIf, nucleus uvaeformis (Uva), and medial magnocellular nucleus of anterior nidopallium (mMAN), whereas on the lesioned sides labeling was seen only in Uva and mMAN. The lesioned NIf area usually did not show any retrogradely labeled cells, but exhibited some autofluorescence due to the damaged tissue. In lesioned birds in which small numbers of retrogradely labeled NIf cells were still visible, we measured the volume of remaining NIf compared with the volume of intact NIf using the same method as described earlier for cresyl violet staining. A list of the percentage NIf volume lesioned in each bird, measured either in cresyl violet–stained sections or in fluorescently labeled sections, is provided in Table 1.

TABLE 1.

Estimated percentage volume of NIf lesioned in each hemisphere of the 11 birds subjected to bilateral NIf lesions and the histological method used in each bird for this estimation

Bird ID	Method of Histological Assessment	Percentage NIf Lesioned		Decrystallization Following NIf Lesion	Decrystallization Following Nerve Section
		Left	Right
Blue603	Cresylviolet	80.9	81.0	No	Yes
Green615	Fluorescence	100.0	100.0	No	Yes
Yellow614	Fluorescence	52.3	100.0	No	Yes
Orange640	Cresylviolet	100.0	100.0	No	Yes
Purple622	Cresylviolet	81.1	81.0	No	No
Green617	Fluorescence	76.1	61.5	No	No
Purple621	Fluorescence	80.5	81.0	No	No
Blue623	Fluorescence	100.0	100.0	No	No
Yellow623	Cresylviolet	100.0	100.0	No	No
Purple641	Fluorescence	100.0	60.7	No	No
Silver640	Fluorescence	76.6	100.0	No	No

The behavioral outcomes of the NIf lesions and the vocal nerve sections are shown to the right. Note that the 18 birds subjected to unilateral NIf lesions followed by LMAN and HVC recordings are not included in this table. In all those 18 birds, the right NIf was 100% lesioned, as estimated from fluorescent retrograde tracer dye labeling (see Fig. 4A).

In one nerve-sectioned bird used for the CM inactivation experiments, we confirmed the locations of the GABA injection site by fluorescent imaging of the tracer dye (5% Alexafluor-594) that was mixed with the GABA solution.

RESULTS

Bilateral NIf lesions do not prevent decrystallization induced by vocal nerve section

To test whether NIf is necessary for vocal nerve section–induced song decrystallization, we first made ibotenic acid lesions bilaterally in NIf of adult male zebra finches (n = 11). In each bird, after the conclusion of all behavioral and physiological testing, the location and extent of the lesions were estimated from histological material (Fig. 3A; see methods). In each of the 11 birds, ≥50% of NIf was lesioned in each hemisphere (compared with the average volume of NIf measured in normal adult zebra finch brains). The detailed quantification of the lesions in each of the 11 birds is given in Table 1. Following lesions, we monitored the songs of these birds for several weeks (mean number of days postlesion: 24 ± 13 days, mean ± SD; range 8 to 48 days) and compared the spectral and temporal characteristics of songs recorded before and after the NIf lesions in each bird (see methods). Consistent with a previously published study (Cardin et al. 2005), we found that bilateral NIf lesions did not disrupt spectral or temporal characteristics of song structure (Fig. 2).

One week or more (24 ± 13 days, mean ± SD; range 8 to 48 days) after placement of the NIf lesions, we unilaterally sectioned the vocal (i.e., tracheosyringeal) nerve in the 11 birds with bilateral NIf lesions. Vocal nerve section in NIf-lesioned birds immediately and persistently distorted the spectral features of all birds' songs (Fig. 3B), similar to the effects previously observed in normal adult zebra finches (Roy and Mooney 2007; Williams and McKibben 1992). To quantify these effects on song spectral properties, we made within-bird comparisons of the song's spectral features before and at various times following nerve section (see methods) (Tchernichovski et al. 2000). The average spectral similarity scores (see methods) between songs produced by the same bird before and after nerve section were low compared with within-bird comparisons made prior to nerve section (Fig. 3C; pre: 90 ± 3%; post1: 32 ± 7%; post2: 34 ± 8%; post3: 30 ± 7%; df: 3, F = 57, P < 0.001, one-way repeated-measures ANOVA; P < 0.01, n = 11, pre vs. post1, post2, and post3, Fisher's post hoc test; all values are expressed as means ± SE). The within-bird spectral similarity scores comparing pre- and post nerve section songs at all times were as low as similarity scores measured between unrelated zebra finch songs recorded from our colony (mean spectral similarity of unrelated zebra finch songs: 28 ± 5%; df: 3, F = 1.2, P = 0.7, one-way ANOVA for unrelated and post1, post2, post3). Therefore in all bilaterally NIf lesioned birds, unilateral section of the vocal nerve precipitated immediate and long-lasting spectral distortion of their crystallized songs.

Although nerve section resulted in severe spectral distortion, it did not immediately affect the overall temporal pattern of the birds' songs, as reflected in the overall syllable sequence (Fig. 3B). However, at longer times (>2 wk following vocal nerve section), the song temporal pattern started to degrade in a subset of NIf-lesioned birds, similar to what has been observed in normal adult zebra finches following nerve section (Roy and Mooney 2007). To quantify changes in song temporal pattern, we calculated a temporal integrity (TI) score (see methods) for each bird by cross-correlating the amplitude envelopes of the bird's songs produced shortly after nerve section to those produced at longer intervals following nerve section. We also calculated a stereotypy score (essentially a “within-bird, within-day” TI score) by comparing amplitude envelopes of different bouts of songs recorded from a bird on a single day before nerve section. In all NIf-lesioned birds, the stereotypy scores calculated prior to vocal nerve section were high (Fig. 3D, mean “within-bird, within-day” TI score: 0.86 ± 0.02, n = 11; mean ± SD). We set an operationally defined threshold for song decrystallization as 2SDs below this mean stereotypy score.

Applying this threshold for assessing song decrystallization, we found that all 11 birds maintained TI scores above the threshold for the first 2 wk following nerve section (Fig. 3D, early). However, at longer times (>2 wk), the TI scores of a subset of birds (n = 4/11) fell below the threshold (Fig. 3D, late) and these birds were deemed to have undergone song decrystallization (“plastic” birds). The other 7 birds maintained TI scores above the threshold (“nonplastic” birds). The high TI scores in the nonplastic birds were comparable to the stereotypy scores calculated prior to nerve section (Fig. 3D; P = 0.3, n = 7, paired t-test), whereas the TI scores in the plastic birds were comparable to TI scores obtained by comparing the temporal patterns of songs of unrelated adult male zebra finches from our colony (i.e., “between-birds” TI scores) (Fig. 3D; P = 0.15, between-birds TI scores vs. late TI scores in 4 plastic birds, unpaired t-test). Thus a subset of NIf-lesioned birds underwent song decrystallization at >2 wk following vocal nerve section. The fraction of NIf-lesioned birds that underwent decrystallization in this study (4/11) was identical to the fraction of normal adult zebra finches that undergo decrystallization following nerve section (8/22), as described in a previous study (Roy and Mooney 2007). These observations indicate that NIf is not necessary for decrystallization of adult song triggered by vocal nerve section.

NIf lesions do not abolish auditory activity in LMAN

We found that vocal nerve section could induce decrystallization of adult zebra finch song even in the absence of NIf. Two possible explanations could account for these observations. First, if NIf lesions indeed disrupt auditory activity patterns in the song system, including the AFP, then this activity is unnecessary for feedback-dependent song plasticity in adult birds. Alternatively, the AFP may receive other sources of auditory input and the auditory drive from these other sources may be sufficient for feedback-mediated vocal plasticity. To begin to distinguish between these two ideas, we first examined the extent of disruption in LMAN auditory activity caused by NIf lesions. We focused on LMAN because LMAN is necessary for nerve section–induced song decrystallization (Williams and Mehta 1999) and plastic retuning of LMAN auditory activity is strongly correlated to this form of song plasticity (Roy and Mooney 2007). Thus we unilaterally lesioned NIf in a set of 15 adult male zebra finches and then carried out in vivo multiunit recordings under urethane anesthesia in the LMAN of these birds. Because NIf projects exclusively to the ipsilateral HVC, and thus indirectly to the ipsilateral AFP, unilateral lesions allowed us to make interhemispheric comparisons of LMAN auditory activity within individual birds. To assess the completeness of NIf lesions, we injected retrograde tracers into HVC prior to electrophysiological recordings. We were unable to detect any retrogradely labeled cells in the NIf area ipsilateral to the lesion in any of the 15 unilaterally NIf lesioned birds tested (Fig. 4A), suggesting that the lesions were complete.

FIG. 4.

Auditory responses in HVC and LMAN persist following unilateral NIf lesions. A: histology from unilaterally NIf-lesioned bird. Fluorescent images of sagittal sections are shown from a unilaterally NIf-lesioned bird that was injected in both HVC with a retrograde tracer (dextran-conjugated Texas Red). Left column: images from the intact hemisphere. Right column: images from the hemisphere with NIf lesion. For each hemisphere, the top 2 rows depict, respectively, a low- and a high-magnification image of the NIf area. On the intact side, retrogradely labeled NIf cells can be clearly seen (white arrow), whereas on the lesioned side, no such cells are visible, although the tissue surrounding the NIf lesion site autofluoresces. Bottom panel shows that in both hemispheres the thalamic nucleus Uva, another HVC afferent, was densely labeled with the retrograde tracer, confirming accurate placement of tracer injections in HVC in both the intact and lesioned hemispheres. B: multiunit spiking responses to 30 iterations of bird's own song (BOS) playback in HVC and LMAN in 2 hemispheres of a single bird are shown. The left column depicts responses in the NIf-intact side, whereas the right column depicts responses in the NIf-lesioned side of the brain. The top row depicts the peristimulus time histogram (PSTH) of spiking responses in LMAN and the middle row depicts PSTH of spiking responses in HVC. The bottom row shows oscillograms of the song stimulus (i.e., BOS) used to elicit these responses. Note the striking reduction in spontaneous activity in HVC on the NIf-lesioned side. C: comparable BOS-evoked and spontaneous activity in LMAN on both sides of the brain following unilateral NIf lesions. Top: mean response strength to playback of BOS in both LMAN was comparable in almost all of the 15 birds. Each filled circle and error bars denote means ± SE response strength to BOS playback on the NIf-lesioned side, as normalized to the mean response strength on the NIf-intact side for each bird. The dotted line marks mean normalized response Z score of 1, which signifies identical mean LMAN responses in the 2 hemispheres. The open square and error bars denote the mean of means ± SE normalized response strength for all 15 birds, which is not significantly different from 1 (P = 0.9). Bottom: the spontaneous activities in LMAN on NIf-intact and NIf-lesioned sides were also comparable. D: BOS-evoked and spontaneous activities in HVC are diminished following NIf lesions. Top: mean response strength to playback of BOS in HVC ipsilateral to the NIf lesion was weaker compared with the contralateral HVC. Each filled circle and error bars denote means ± SE response strength to BOS playback on the NIf-lesioned side, as normalized to the mean response strength on the NIf-intact side for each bird. For this response strength calculation, background activity was not subtracted from the stimulus-evoked activity. The dotted line marks mean normalized response of 1, which signifies identical mean HVC responses in the 2 hemispheres. The open square and error bars denote the mean of means ± SE normalized response strength for all 9 birds, which is significantly <1 (P < 0.0001). Bottom: the spontaneous activity in HVC ipsilateral to NIf lesion was strongly diminished compared with the contralateral HVC. Overall, the strength of HVC spontaneous activity on the lesioned side was significantly lower compared with the intact side (P < 0.0001).

We detected robust BOS-evoked auditory activity in LMAN ipsilateral to the side of the NIf lesion. In LMAN, almost all multiunit recording sites recorded in both hemispheres showed significant auditory responses to one or more of the presented auditory stimuli, including BOS, reverse BOS, or CON (Fig. 4B, top; intact side: 49/50 sites; lesioned side: 51/53 sites, n = 15 birds). Overall, the BOS-evoked responses in LMAN ipsilateral to the NIf lesion were comparable in strength to responses recorded in LMAN on the intact side (Fig. 4C, top; P = 0.9), although a weaker response was detected in a few birds. The levels of spontaneous action potential activity of LMAN neurons ipsi- and contralateral to the side of the lesion were also similar (Fig. 4C, bottom; P = 0.1). To test whether the BOS-evoked LMAN responses in the absence of NIf were still selective, we compared responses evoked by BOS with responses evoked by BOS played in reverse (rev BOS) and conspecific (CON) songs. Interestingly, LMAN neurons ipsilateral to the NIf lesion were highly selective for the BOS, as reflected in the high d′ values (see methods; on NIf-lesioned side: mean d_fwd–rev = 1.9 ± 0.15, n = 54; d_BOS–CON = 2.1 ± 0.3, n = 7). These d′ values on the lesioned side were indistinguishable from the d′ values obtained in the LMAN contralateral to NIf lesion (on NIf-intact side: mean d_fwd–rev = 1.9 ± 0.1, n = 49; d_BOS–CON = 2.1 ± 0.4, n = 7; P = 0.9 for rev BOS, P = 0.8 for CON, t-test intact vs. lesioned hemispheres). Therefore strong and selective auditory responses persist in LMAN when the ipsilateral NIf is completely lesioned.

It is generally believed that LMAN derives its auditory input solely, albeit indirectly, from HVC (Fig. 1A). We wondered whether the auditory activity we detected in LMAN following NIf lesions arises from HVC or from previously unidentified sources of auditory input to the AFP. Therefore we first examined a subset of these birds to test whether auditory responses in HVC persisted after NIf lesions and then tested whether HVC auditory activity was necessary for the residual auditory activity in LMAN. Using multiunit recordings in HVC, we detected significant BOS responses ipsilateral to the NIf lesion (Fig. 4B, middle), a result reported previously from our group (Bauer et al. 2008). However, the fraction of multiunit recording sites in HVC that showed significant responses to one or more of the presented auditory stimuli, including BOS, reverse BOS, or CON, were slightly smaller on the lesioned side compared with those on the intact side of the brain (intact side: 20/21 sites; lesioned side: 22/29 sites, n = 9 birds). On average the BOS-evoked responses in HVC ipsilateral to the NIf lesion were significantly weaker compared with those in the contralateral HVC (Fig. 4D, top, P < 0.0001). Although auditory responses in HVC were weaker on the side of the NIf lesion, we found that HVC auditory responses in both brain hemispheres were highly selective (d_fwd–rev = 3.0 ± 0.6 for intact side, 2.7 ± 0.3 for lesioned side; P = 0.6). We also observed that spontaneous activity in HVC ipsilateral to the NIf lesion was greatly reduced (Fig. 4D, bottom, P < 0.0001; also compare individual examples in Fig. 4B).

Although NIf lesions do not affect the strength of spontaneous and BOS-evoked activity in LMAN, they do affect both of these features in HVC (Fig. 4, B–D). Thus we wondered whether the auditory activity that persists in LMAN following NIf lesions derives entirely from HVC or also includes contributions from other previously undescribed sources of auditory input to the AFP. To test this idea, we used lidocaine to reversibly inactivate HVC ipsilateral to the NIf lesion while recording extracellularly from the ipsilateral LMAN. The BOS-evoked auditory responses in LMAN that persisted following unilateral NIf lesion were drastically reduced by lidocaine injections in ipsilateral HVC (Fig. 5A). In all five birds tested, inactivation of HVC led to significant reduction in BOS-evoked auditory responses in LMAN (Fig. 5B). This effect was reversible, with BOS-evoked responses in LMAN recovering within 10–15 min after lidocaine application had ceased. Therefore following unilateral NIf lesions, the remaining auditory responses in LMAN primarily originate in the ipsilateral HVC, suggesting that the AFP does not receive auditory input from sources other than HVC.

FIG. 5.

LMAN auditory activity spared by NIf lesion requires activity in ipsilateral, but not contralateral, HVC. A: in a unilaterally NIf-lesioned bird, lidocaine injection into ipsilateral HVC reversibly abolishes BOS-evoked auditory activity in the ipsilateral LMAN. The top row depicts PSTH of LMAN multiunit spiking responses to 30 iterations of BOS playback before (left), immediately after (middle), and during recovery from (right) the lidocaine injection. The middle row depicts individual raw traces of the electrode signal to playback of a single iteration of BOS. The bottom row depicts oscillograms of the song stimulus. B: the mean BOS-evoked responses recorded in the LMAN of 5 NIf-lesioned birds before and after lidocaine injections into the ipsilateral HVC. Each column represents the average strength (spikes/s) of multiunit BOS responses to 30 playbacks of BOS in 5 different birds, before and after lidocaine injections. C: auditory responses in LMAN ipsilateral to the NIf lesion were unaffected by lidocaine injection into the contralateral HVC. The panel arrangement is the same as that in A. D: mean BOS-evoked responses recorded in the LMAN of 4 NIf-lesioned birds before and after lidocaine injections into the contralateral HVC. Each column represents the average strength (spikes/s) of multiunit responses to 30 playbacks of BOS in 4 different birds, before and after lidocaine injections.

A remaining possibility is that HVC also contributes to auditory responses in the contralateral LMAN, perhaps via crossed projections from either thalamic or brain stem components of the song system (Striedter and Vu 1998; Vates et al. 1997; Vu et al. 1998). Such crossed projections could afford another mechanism for the auditory responses we detected in LMAN following lesions in the ipsilateral NIf. To test this idea, we injected 2% lidocaine into the contralateral HVC while recording in LMAN ipsilateral to NIf lesion. Inactivation of the contralateral HVC, however, had no effect on the BOS-evoked responses we recorded in LMAN (Fig. 5, C and D). Thus following unilateral NIf lesions, the BOS-evoked responses that persisted in LMAN derive entirely from the ipsilateral HVC. A recent study identified the caudal mesopallium (CM) as an important source of direct auditory input to HVC (Bauer et al. 2008). Taken together with these prior observations, the current results suggest that, following NIf lesions, the direct projection of CM onto HVC can drive normal auditory activity in LMAN (for confirmation of this idea, see the following text). Thus the auditory activity that persists in the AFP following NIf lesions may be sufficient for supporting vocal plasticity in nerve-sectioned birds.

In NIf-lesioned birds, residual LMAN auditory activity can shift in selectivity following vocal nerve section

Previously, we showed that LMAN neurons in nerve-sectioned birds can shift their auditory selectivity to the distorted BOS and that expression of such auditory plasticity in LMAN strongly correlates with song decrystallization (Roy and Mooney 2007). Therefore we wondered whether the selectivity of LMAN auditory responses spared by NIf lesions would also shift following vocal nerve section. To test this idea, we carried out in vivo multiunit recordings in LMAN of bilaterally NIf lesioned birds that decrystallized their songs following vocal nerve section (i.e., plastic birds). Three birds used for this purpose were obtained from the subset of four birds (4/11) that had undergone song decrystallization following NIf lesion and nerve section, as described earlier (Fig. 3D).

In these three behaviorally plastic birds, multiunit recordings in LMAN detected auditory responses selective for the new, distorted BOS (i.e., the decrystallized BOS) (Fig. 6A). All (20/20) multiunit sites recorded in these birds yielded significant auditory responses and all 20 sites displayed selectivity for the new BOS, as assessed with a d′ metric (Fig. 6B; mean d_N–O = 0.77 ± 0.1; P < 0.001). The mean response strength to the new BOS, expressed as a Z score, was significantly greater than the response strength to the old BOS (Fig. 6C; P < 0.0001). Therefore the BOS-evoked auditory responses that persist in LMAN of NIf-lesioned birds can shift in selectivity during song decrystallization triggered by vocal nerve section.

FIG. 6.

The auditory responses that persist in LMAN following NIf lesions can change in selectivity during song decrystallization. A: in a bird with bilateral NIf lesions that decrystallized its song following vocal nerve section, auditory responses in LMAN were tuned to the decrystallized new BOS. Top row: the PSTH of multiunit spiking responses to 30 iterations of playback of new BOS and old BOS. Bottom row: the oscillograms of the song stimuli. Whereas the new BOS elicited a strong response at this site, the old BOS evoked only very weak responses. B: scatterplot of d_N–O values for all 20 multiunit recording sites from the 3 plastic birds. Open symbols of a kind denote d_N–O value for recording sites from a single bird. The filled circle and error bars denote overall means ± SE d_N–O value for all 20 sites, which was significantly higher than zero (P < 0.001). C: the mean response strength to new BOS and old BOS, expressed as Z scores, for all 20 multiunit sites recorded, reveals significantly stronger responses to the new BOS. D: in one of the 3 plastic birds, multiunit LMAN responses to the playback of the new BOS were abolished by γ-aminobutyric acid (GABA) injections into ipsilateral CM. Top row: depicts PSTH of spiking responses to 30 iterations of new BOS playback before (left), immediately after (middle), and during recovery from (right) GABA injections in CM. The bottom row depicts oscillograms of the song stimulus. E: quantification of the LMAN response strength (spikes/s) before and after GABA injection in CM from the experiment shown in D. There was a significant decrease in mean LMAN response strength to new BOS immediately following GABA injection in CM (P = 0.006, pre-GABA vs. GABA). F: the GABA injection site (black circle), located by fluorescent imaging of the Alexafluor-594 dye mixed with the GABA solution, shown here schematically superimposed on a camera lucida drawing of a sagittal section from the mediolateral axis containing the CM recording site.

A previous study found that auditory responses remaining in HVC following NIf lesions could be suppressed by reversibly inactivating CM (Bauer et al. 2008). These previous findings suggest CM could be the source of the plastic auditory responses in LMAN we observed here. To confirm this idea, we tested whether auditory responses in LMAN selective for the distorted BOS were abolished by reversibly inactivating CM. Because of technical difficulties in simultaneously placing electrodes in CM and LMAN, we were able to perform this experiment in only one of the three behaviorally plastic birds. We found that injections of 70 nl of 0.25 M GABA in ipsilateral CM significantly suppressed the LMAN auditory responses to playback of new BOS, which was the preferred stimulus in this bird (Fig. 6, D and E). The effect was reversible and the response recovered over the following 10–12 min. Postexperimental histology confirmed the site of GABA injection to be in CM (Fig. 6F). This result indicates that the LMAN auditory responses that persist following NIf lesions and that shift in their selectivity following vocal nerve section may originate in the ipsilateral CM.

DISCUSSION

We have shown that the forebrain song nucleus NIf, thought to be a major source of auditory input to the song system, is unnecessary for vocal nerve section–induced song decrystallization in adult songbirds. In adult male zebra finches, bilateral NIf lesions failed to prevent song decrystallization caused by vocal nerve section, a manipulation that chronically distorts singing-related auditory feedback. We also found that NIf lesions failed to abolish normal auditory activity in LMAN, the output nucleus of the anterior forebrain pathway, which plays an essential role in audition-dependent vocal plasticity. The auditory responses that persisted in LMAN following NIf lesions depended on activity in ipsilateral HVC. Moreover, the residual auditory activity in LMAN could gain selectivity for the distorted BOS following vocal nerve section and required activity in ipsilateral CM. Taken together, these findings indicate that NIf is unnecessary for two feedback-dependent processes in the adult zebra finch: song maintenance and song decrystallization. Given that prior studies had indicated NIf to be the major auditory input to the song system (Cardin et al. 2005; Coleman and Mooney 2004; Fortune and Margoliash 1995; Vates et al. 1996), three possible explanations emerge. First, perturbing other forms of nonauditory (such as proprioceptive) feedback is sufficient to mediate song decrystallization in the absence of NIf. Second, auditory inputs to the HVC and the AFP do not serve an auditory feedback function. Third, auditory inputs other than those from NIf (i.e., CM) are sufficient to support feedback-mediated song plasticity.

One possibility that the current experiments cannot totally exclude is that, in the absence of NIf, the aberrant proprioceptive feedback resulting from nerve section could be sufficient to mediate song decrystallization. We think this is unlikely for several reasons. First, selective removal of the afferent fibers in the tracheosyringeal nerve fails to trigger decrystallization of adult zebra finch song, suggesting proprioceptive input from the syrinx alone is unnecessary for song maintenance (Bottjer and Arnold 1984). Second, anatomical studies suggest that NIf is a major conduit through which proprioceptive information might reach HVC; if this information were necessary for song maintenance, NIf lesions would be expected to trigger decrystallization, in contrast to observations made here and in prior studies (Cardin et al. 2005). Third, nerve section–induced decrystallization and other forms of decrystallizations induced by explicit manipulations of auditory signals (such as deafening) require an intact AFP and act over a similarly slow time course (Brainard and Doupe 2000b; Williams and Mehta 1999), suggesting that same central pathways—and thus by extension, similar mechanisms—are involved in these processes. These observations strongly argue that vocal nerve section induces song decrystallization primarily via auditory feedback.

The existing evidence also argues against the second possible explanation. The behavioral effects of altered feedback on vocal plasticity dictate that auditory feedback must influence song motor networks (Brainard and Doupe 2000a; Leonardo and Konishi 1999; Nordeen and Nordeen 1992; Roy and Mooney 2007; Williams and McKibben 1992; Zevin et al. 2004). The expression of feedback-dependent vocal plasticity requires singing-related activity in the AFP (Brainard and Doupe 2000b; Williams and Mehta 1999), suggesting that auditory feedback must ultimately modulate this activity to trigger vocal plasticity. Although LMAN neurons do not act as real-time sensors of auditory feedback (Leonardo 2004), they do gain selectivity for spectrally distorted songs following vocal nerve section (Roy and Mooney 2007), suggesting that their properties are shaped by auditory feedback. Further, chronic recordings in singing swamp sparrows and Bengalese finches show that HVC neurons projecting to the AFP (i.e., HVC_AFP cells) display motor-related activity, the pattern of which is highly similar to auditory activity evoked in these same cells when the bird is quietly listening to the same song played through an audio speaker (Prather et al. 2008). This precise auditory–vocal correspondence strongly hints that song motor-related activity in HVC_AFP cells is shaped by auditory experience, even though singing-related activity of HVC_AFP cells is insensitive to acute feedback perturbations (Kozhevnikov and Fee 2006; Prather et al. 2008). Moreover, these findings point to neurons presynaptic to HVC_AFP cells as the source of real-time auditory feedback signals important to vocal plasticity. Consistent with this view, real-time sensitivity to auditory feedback perturbations has been detected in neurons in primary and secondary regions of the auditory telencephalon in zebra finches (Keller and Hahnloser 2009) and putative HVC interneurons in Bengalese finches (Sakata and Brainard 2008). Furthermore, manipulating singing-related activity in the auditory thalamic nucleus ovoidalis (Ov), which indirectly provides auditory drive to HVC through NIf and CM, destabilizes song in a manner similar to peripheral auditory feedback manipulations (Lei et al. 2007). These various findings strongly support the idea that auditory inputs to HVC, and thus to the AFP, convey feedback-related information necessary for vocal plasticity.

Prior evidence pointed strongly to NIf as the primary candidate for providing auditory input to HVC (Cardin et al. 2005; Coleman and Mooney 2004). If the NIf input conveys singing-related auditory feedback to HVC and the AFP, then bilateral NIf lesions could constitute a form of central deafening, with deleterious consequences for song. Although NIf lesions in Bengalese finches (a close relative to the zebra finch) induce alterations in higher-order motif structure in song (Hosino and Okanoya 2000), our study confirms an earlier study showing that NIf lesions in adult zebra finches do not trigger song decrystallization (Cardin et al. 2005) (see Fig. 2). A second possibility is that when the bird experiences distorted feedback, NIf plays a critical role in the ensuing process of decrystallization by conveying altered feedback signals to the song system. This idea predicts that, in NIf-lesioned birds, altered patterns of neural activity arising from distorted feedback would fail to reach the song system and thus fail to induce decrystallization. However, we found that adult zebra finches with NIf lesions could still undergo song decrystallization following vocal nerve section, indicating that the consequences of distorted feedback could still influence the song system independent of NIf activity.

These observations lead to the third possibility: vocal nerve section–induced decrystallization requires that auditory feedback signals reach HVC (and the AFP), but that these signals can be conveyed by auditory inputs to HVC other than those from NIf. Consistent with this idea, we found that BOS-selective auditory responses persisted in HVC and LMAN following NIf lesions. The residual LMAN responses in the NIf-lesioned hemispheres were comparable to those in the NIf-intact hemispheres with respect to strength and selectivity. This is somewhat surprising, given a prior finding that acute inactivation of NIf profoundly silences HVC auditory activity (Coleman and Mooney 2004) and given our present finding that over longer intervals following unilateral NIf lesions, BOS-evoked responses in the ipsilateral HVC were significantly weaker than those recorded in the contralateral HVC. A detailed discussion on these differences in results has been provided elsewhere (Bauer et al. 2008), but briefly, one possible explanation is that, following NIf lesions, remaining sources of input to HVC can compensate for the loss of auditory drive to HVC.

Emerging evidence indicates that a subset of neurons in the caudal mesopallium (CM) can provide a direct source of auditory input to HVC (Bauer et al. 2008). This study confirms and extends those earlier findings by showing the auditory responses that persist in the AFP following NIf lesions depend on CM activity. Moreover, following NIf lesions, LMAN neurons could still gain selective responses for spectrally distorted songs, suggesting that the residual auditory drive that CM supplies to the AFP (i.e., via HVC) is sensitive to the effects of auditory feedback. Indeed, several emerging lines of evidence support the idea that CM is the proximal source of feedback-related information to the song motor network. First, chronic recordings made in freely behaving finches detect elevated neuronal activity in CM during singing and song playback (Bauer et al. 2008). Second, some neurons in lateral aspects of CM (i.e., CLM) respond to perturbations in auditory feedback presented during singing (Keller and Hahnloser 2009). Third, singing correlates with increased immediate early gene expression in the CM of normal, but not deafened, birds (Jarvis and Nottebohm 1997; Mello and Clayton 1994; Mello et al. 1992).

These various observations promote two important directions for future studies. First, additional chronic recording studies in freely behaving songbirds are needed to determine whether singing-related activity of HVC-projecting CM neurons is sensitive to distorted auditory feedback. Although such sensitivity is exhibited by some CLM neurons, whether any of these cells project to HVC is unknown (Keller and Hahnloser 2009). Second, because CM has appeared as the major source of auditory input to both NIf and HVC, lesions of CM could be used to test the importance of this input to song learning and maintenance. The spatially distributed nature of CM may make it challenging to completely lesion, but this approach represents an important future step in the quest to determine whether auditory activity in the song system serves a feedback role. Finally, CM has been implicated in the learned recognition of other birds' songs (Gentner and Margoliash 2003). The recent finding that CM provides direct auditory drive to HVC (Bauer et al. 2008) and the results presented here suggest a role for CM in auditory processing of not only self-generated songs, but also the songs of other birds.

GRANTS

This work was supported by National Institute on Deafness and Other Communication Disorders Grant DC-02524 to R. Mooney.