PMC

a) Illustration of the manipulation of timing of paired burst stimulation. Negative lags indicate when HVC is leading LMAN, positive lags indicate when LMAN is leading HVC. b) The sign and magnitude of plasticity in both LMAN and HVC inputs depend on the timing between burst stimulation. c–e) Representative examples at −100 ms (n=5), 0 ms (n=7), and 100 ms (n=3) lags of the changes in the two pathways. All values are mean ± s.e.m.

a) Specific syllables (denoted by letters) can be targeted with negative reinforcement (white noise, WN) contingent on the fundamental frequency (expanded view of targeted syllable in (b)), which drives changes in the pitch of targeted syllables (c). Adult birds reliably shifted the fundamental frequency of targeted syllables when PBS was infused through microdialysis cannulae, while infusion of the mGlur2/3 blocker LY341495 into RA prevented learned changes in pitch (0.11 ± 0.07% of control, t(9)=5.14, p=0.0006, n=10 experiments in 5 birds). (c,d). Placement of microdialysis probes and drug spread were confirmed by injection of Muscimol-BODIPY. Area circled in (e) delineates the borders of RA (scale bar = 500um). All values are mean ± s.e.m.

a) Burst stimulation of HVC alone did not cause changes in the strength of inputs from either HVC or LMAN HVC stimulation only: LMAN: 99 ± 0.1% of control, t(2)= −0.19, p=0.87, n=3 cells from 2 birds; HVC: 97 ± 2% of control, t(5)= −2.06, p = 0.09, n=6 cells from 5 birds, paired t-test. b) Burst stimulation of LMAN alone did not cause changes in the strength of inputs from either HVC or LMAN (LMAN: 97 ± 2% of control, t(7)= −1.24, p=0.25, n=8 cells from 6 birds; HVC: 94 ± 5% of control, t(4)= −1.37, p = 0.24, n=5 cells from 3 birds, paired t-tests). c) Pairing HVC bursts with single stimulation of LMAN failed to produce lasting changes in synaptic strength (HVC: 98±3% of control, t(3)= −0.72, p=0.53, LMAN: 97±4% of control, t(3)= −0.83, p=0.47, n=4 cells from 4 birds, paired t-tests). d) Paired, low frequency burst stimulation failed to produce lasting changes in synaptic strength in either input (HVC: 99±10% of control, t(3)= −0.09, p=0.93, LMAN 107±8% of control, t(3)=0.87, p=0.45, n=4 cells from 2 birds, paired t-tests). All values are mean ± s.e.m.

a) Diagram of the major nuclei involved in song generation. The motor region RA receives inputs primarily from two sources- HVC (used as a proper name) and LMAN, the output of a cortical-basal ganglia loop. RA in turn directly drives brainstem motor neurons which control the vocal musculature. The inputs from HVC and LMAN are anatomically distinct (b). RA has well defined borders (c), and RA neurons are easily visualized under transillumination (d). e) The two pathways innervating RA PNs are pharmacologically distinct: HVC is weakly blocked by APV (84 ± 6% of control, t(4)= −2.70, p = 0.054, n=5 cells from 4 birds, paired t-test), while LMAN gives rise to APV-sensitive NMDA receptor mediated currents^, (6% ± 2% of control, t(7)= −53.95,p < 0.0001, n=8 cells from 6 birds, paired t-test) Scale bar in (c), 100 um, in (d) 20 um. All values are mean ± s.e.m.

a) Stimulation of either HVC (black) or LMAN (red) at high frequencies (800 Hz and 500 Hz respectively) drove a significantly larger summed current in comparison to single stimulations, suggesting increased glutamate release after burst stimulation, even at these high frequencies; (HVC: 107 ± 12% increase, t(12)=4.96, p=0.0003, n=13, LMAN: 188 ±38% increase, p=0.0000009, n=13 cells from 9 birds, paired t-tests). b) Schematic of protocol used for plasticity induction- HVC and LMAN inputs were paired 8 times at 0.5 Hz, with HVC driven at 800 Hz and LMAN driven at 500 Hz. After the induction protocol, LMAN inputs showed a modest increase in synaptic strength (127 ± 10% of control, t(6)=2.76, p=0.035, n=7 cells/slices from 6 birds, paired t-test), while HVC inputs showed a large decrease in synaptic strength (54% ± 16% of control, t(6)=2.86, p=0.033, n=7 cells/slices from 6 birds, paired t-test). A representative experiment (c,d) and summary data (e) are illustrated. f) The induction protocol increased the paired-pulse ratio of HVC inputs (23% ± 10% increase, t(6)= −2.56, p = 0.043, 7 cells/slices from 6 birds, paired t-test), but did not affect the paired pulse ratio of LMAN inputs (0.8% ± 4%, t(6)= 0.09, p = 0.93, n=7 cells/slices from 6 birds, paired t-test). All values are mean ± s.e.m.

Depression of HVC synapses in response to the 0 ms lag burst induction protocol (diagrammed in a) did not require activation of NMDA receptors (56 ± 10% of control, t(3)= −4.28, p=0.023, n=4 cells from 3 birds, paired t-test) (b), but lasting plasticity was blocked by chelating of intracellular calcium (HVC: 98 ± 23% of control, t(3)= −0.10, p=0.92, LMAN:107 ± 10% of control, t(3)=0.68, p=0.55, n=4 cells from 3 birds, paired t-test) (c), blockade of calcium release from intracellular stores (HVC: 98 ± 7% of control, t(3)= 0.30, p=0.78, LMAN:101 ± 3% of control, t(3)=0.20, p=0.85, n=4 cells from 2 birds, paired t-test) (d) and blockade of Group II mGluR receptors (HVC: 96 ± 5% of control, t(3)= −0.73, p=0.52, LMAN: 103 ± 10% of control, t(3)=0.24, p=0.82, n=4 cells from 4 birds, paired t-test) (e). When a different lag between stimulations was adopted, the same pattern held (f–j). When HVC stimulation led LMAN stimulation by 100ms (f), the increase in HVC synaptic strength did not depend on NMDA receptor activation (153±16% of control, t(3)=3.28, p=0.047, n=4 cells from 3 birds, paired t-test) (g). Buffering intracellular calcium rises (HVC: 105 ± 11% of control, t(4)=0.43, p=0.69; LMAN 103 ± 12% of control after BAPTA, t(4)=0.29, p=0.78, n=5 cells from 3 birds, paired t-tests) (h) or release from intracellular stores (HVC: 104 ± 7% of control, t(2)=0.30, p=0.79, LMAN 109 ± 9% of control after thapsigargin, t(2)=1.15, p=0.36, n=3 cells from 2 birds, paired t-tests) (i) prevented changes in either pathway. j) Activation of group II mGluRs was again required for changes in both pathways (HVC: 87 ± 6%, t(5)=2.54, p = 0.06, LMAN 108 ± 13%, t(5)=0.63, p=0.056, n=6 cells from 4 birds, paired t-tests.) All values are mean ± s.e.m.