Plasticity of stereotyped birdsong driven by chronic manipulation of cortical-basal ganglia activity

Abstract

Cortical-basal ganglia (CBG) circuits are critical for motor learning and performance, and are a major site of pathology. In songbirds, a CBG circuit regulates moment-by-moment variability in song and also enables song plasticity. Studies have shown that variable burst firing in LMAN, the output nucleus of this CBG circuit, actively drives acute song variability, but whether and how LMAN drives long-lasting changes in song remains unclear. Here, we ask whether chronic pharmacological augmentation of LMAN bursting is sufficient to drive plasticity in birds singing stereotyped songs. We show that altered LMAN activity drives cumulative changes in acoustic structure, timing, and sequencing over multiple days, and induces repetitions and silent pauses reminiscent of human stuttering. Changes persisted when LMAN was subsequently inactivated, indicating plasticity in song motor regions. Following cessation of pharmacological treatment, acoustic features and song sequence gradually recovered to their baseline values over a period of days to weeks. Together, our findings show that augmented bursting in CBG circuitry drives plasticity in well-learned motor skills, and may inform treatments for basal ganglia movement disorders.

Keywords: basal ganglia; motor performance; motor variability; movement disorders; sequence variability; songbird; stuttering; vocal learning; vocal plasticity; zebra finch.

Figure 1.. Testing a causal role of LMAN burst firing in song plasticity.

A: Schematic showing the song motor pathway (white) and the BG-thalamo-cortical circuit (red) in songbirds. LMAN activity modulates song variability, indicated by different colored arrows. See text for abbreviations. B: Table summarizing known features of singing-related LMAN activity at different developmental stages. Frequent burst firing and more variable spike timing in LMAN are associated with higher levels of song variability and song plasticity. C: Experimental paradigm: To test whether augmented burst firing in LMAN can drive long-lasting changes to song, 10mM bicuculline methiodide (BMI) was chronically infused into LMAN, and the effects on song were examined. D: Extracellular recording of singing-related LMAN activity. Shown are the song-aligned spike rasters for a single unit during undirected song when PBS (top) or 10mM BMI (bottom) was infused and the corresponding distribution of inter-spike intervals (right). Shading highlights ISIs ≤ 5 ms. A higher proportion of spikes occur in bursts with BMI.

Figure 2.. Example illustrating accumulation of song changes induced by chronic BMI treatment in LMAN.

A: Experimental timeline: Bilateral infusions of PBS (pre and post) or 10 mM bicuculline methiodide (BMI; range 8–13 d) were performed 2–5 times per day. B: Spectrograms of two representative bouts of an adult male zebra finch (o54w8) during PBS infusions 1 d prior to BMI treatment. C-E: Spectrograms of three song bouts per day during chronic BMI treatment. All songs were recorded when the male was alone (‘undirected’ song). Syllables are labelled with letters; question marks denote unidentifiable syllables; ‘stop’ indicates end of a song bout. Atypical syllables and sequences are indicated in red. Variability in syllable structure, song tempo, and song sequence increased across repeated days of BMI infusion. See also Figure S1.

Figure 3.. Variability in syllable structure increases across multiple days of BMI infusions in LMAN and persists following cessation of BMI treatment.

A: Example spectrograms of syllable ‘a’ on 1 d pre-BMI treatment (left), d 11 of BMI treatment (middle), and 1 d post-BMI treatment (right) illustrating increased variability in syllable structure and syllable duration during and after BMI treatment. Scale bars in E. B-D: Summary of changes in spectral and temporal properties of motif syllables across experimental days. Variability in pitch (B), Wiener entropy (C), and syllable duration (D) increased with BMI in LMAN, and remained higher on the first day after cessation of BMI treatment compared to control recordings pre-BMI. “Mean PBS pre” data point is the mean value across three days of PBS infusion prior to BMI. BMI was infused for up to 2 weeks (range: 8–13 d). For each bird, data are plotted for the first day of BMI infusion (BMI first), half-way through BMI infusions (BMI middle), on the last day of BMI infusion (BMI last), and one day after cessation of BMI treatment (PBS post +1). Different symbols denote different birds; error bars denote SEM across syllables for each bird (n=2–7 syllables per bird). Gray lines connect data from the same bird across days; black symbols indicate the mean across birds (n=8; n=7 on PBS post +1, as data is missing for bird k77r57). Related-samples Wilcoxon’s pairwise comparisons: #p<0.10, *p<0.05, **p<0.01. E: Example spectrograms of vocalizations from three birds that did not resemble syllables produced prior to BMI infusions (‘unidentifiable syllables’). F: The percentage of unidentifiable syllables increased across days of BMI infusion, and remained higher 1 d post-BMI. Conventions as in B–D. G: Distributions of the durations of motif syllables 1 d pre-BMI treatment (blue; 12,725 syllables) and on the last day of BMI treatment (red, 7,285 syllables), and distribution of the durations of unidentifiable syllables during BMI treatment (BMI first, mid and last combined; dashed black; 1,034 syllables) across all birds. Multiple narrow peaks in the distribution prior to BMI infusion (blue) correspond to distinct, stereotyped syllables characteristic of adult song. In contrast, unidentifiable syllables induced by BMI treatment were highly variable in duration (dashed black, wide, unimodal distribution). The distribution of unidentifiable syllables differed significantly from the distributions of motif syllables recorded during PBS and BMI infusions (pairwise K-S tests, ***p<0.0001). For comparison, the light gray bar indicates the range of proto-syllable durations and the light gray circle indicates the mean duration of proto-syllables in juvenile zebra finches. See also Figure S2.

Figure 4.. Long-lasting disruption of song sequence induced by chronic BMI treatment in LMAN.

For each recording, variability in syllable sequencing was quantified by calculating the transition entropy (TE) for each syllable (see STAR Methods) and averaging across syllables (ATE). Normalized ATE increased across days of BMI infusion in 7/8 birds (BMI last versus Mean PBS pre, p=0.016, Wilcoxon’s), and remained higher in 6/7 birds after switching to vehicle (PBS post+1 vs Mean PBS pre, p=0.031, Wilcoxon’s). See also Figure S3.

Figure 5.. Stuttering-like phenotype in BMI-treated birds.

A-C: BMI in LMAN drove an abnormal increase in syllable repetitions. A: Example spectrograms from one bird (k71o7) who repeated syllables ‘b’ and ‘c’ on d 7 of BMI treatment. Note that repetitions of ‘c’ were typically followed by the next canonical motif syllable. B: The frequency of syllable repetitions increased across days of BMI infusions (BMI last versus Mean PBS pre, p=0.016). Syllable repetitions remained higher in 4/7 birds 1 d after cessation of BMI infusions, and across all birds, there was a trend towards increased repetitions after cessation of BMI (PBS post+1 vs. Mean PBS pre, p=0.063). Conventions as in Figure 3. C: Birds (7/8) typically continued with their normal song after syllable repetitions. D-F: Long gap durations induced by chronic BMI treatment. D: Example spectrograms of the canonical motif aligned at the onset of syllable ‘a’ (dashed line) for bird k77r57 1 d pre-BMI and on d 8 of BMI treatment illustrate longer gaps (shaded bars) between syllables (e.g., gap between ‘b’ and ‘c’) and increased variability in gap durations. E: Histograms of the summed gap durations when birds sang a complete motif for the birds in A and D. Gap durations were greater on the last day of BMI treatment compared to PBS pre (pairwise two-sample K-S tests, p<0.0001). F: Histograms of normalized summed gap durations (see Methods) before (PBS pre-1) and on the last day of BMI treatment across all 8 birds. The BMI distribution is significantly different from the control distribution (pairwise two sample K-S test, p<0.0001), and skewed to the right (skew = 1.15 vs. 0.53, respectively).

Figure 6.. BMI-driven changes in song are not rescued by LMAN inactivation, but song does recover post-BMI treatment.

A: Complete experimental timeline for 6 birds prior to and following BMI treatment. B: Example spectrograms of bird y3y18 prior to BMI infusion and 1 d after cessation of BMI infusion. After 11 days of BMI treatment, the bird sang aberrant syllables and sequences (red) when vehicle was infused, including skipping and repeating syllables (PBS post+1). Note that the ‘f’ to ‘c’ transition became infrequent. C: Variability in syllable morphology and song sequence persisted in this bird when LMAN was inactivated with 1mM muscimol 1 d after BMI infusions. D: Song recovered gradually over a period of weeks following cessation of BMI infusions, including the lost ‘f’ to ‘c’ transition. Scale bars are for all spectrograms in B-D. E-H: Muscimol infusions did not completely rescue performance. Song variability remained high in at least one feature in all 6 birds. E: Percentage of unidentifiable syllables. F: Variability of Wiener entropy. G: Percentage of syllable repetitions. H: Normalized sequence entropy. Conventions as in Figure 3. Bars indicate the mean across birds (n=6). For comparison, data from the last recording (PBS post last; range 13–27d after cessation of BMI treatment) show that songs eventually recovered after cessation of chronic BMI treatment. The last day song was recorded is indicated for each bird relative to cessation of BMI treatment (H). See also Figure S4.

Figure 7.. Gradual recovery of song following cessation of BMI treatment.

A-D: Time course of changes in syllable structure and sequence following cessation of BMI treatment. Variability in pitch (A), Wiener entropy (B), syllable duration (C) and syllable sequence (D) gradually recovered towards baseline values after cessation of BMI treatment. Wilcoxon’s pairwise comparisons between control data (Mean PBS pre) and data recorded on the last day of BMI infusions or during PBS infusions post-BMI treatment: #p<0.10; *p<0.05. Data are shown for 7 birds; 1 bird was only recorded for up to 4 days post-BMI. See also Figure S5.