HVC projection neurons in juveniles exhibit more variable spiking and bursting patterns compared to adults
The song development of zebra finches involves a gradual refinement of vocalizations, transitioning from variable juvenile songs to a stable, stereotyped adult form. During the crystallization phase, juveniles begin to produce structured syllables, signaling progress towards their final song. However, the temporal features of their songs, such as the timing between syllables, remain immature and undergo further refinement in the sensorimotor phase23. To understand the neural mechanisms underlying this temporal refinement, we focused on HVC, a premotor nucleus crucial for song learning and production. In adult zebra finches, the HVC motor program for song consists of precise, time-locked bursts of activity14–16. We aimed to investigate how this program develops during the plastic song phase, a stage of late-stage song learning.
To achieve this, we conducted intracellular recordings of HVC projection neurons in freely behaving male juvenile zebra finches (74–94 days post hatch) during singing. Given HVC's role in controlling syllable and gap duration during this phase13,24, we analyzed the neural activity of HVC projection neurons in both juveniles and adults to identify potential differences that could contribute to the ongoing refinement of song timing.
We found that HVC projection neurons elicited sparse bursts of action potentials during a specific time point during song production (Fig. 1A, 1B). The distribution of bursts spanned the entire song motif in both age groups15,16 (Fig. 1A, 1B). This observation suggests that the neural dynamics underlying song production are already established during the late phase of song learning, leading to the production of stereotyped song structure. However, this finding alone cannot account for the observed differences in temporal variability between juvenile and adult songs23, indicating that other factors contribute to the ongoing refinement of song timing.
It has been previously shown that in adult zebra finches HVC projection neurons display a sparse number of bursts during song production14–16,24–26. To explore whether our recorded neurons produced ultra-sparse bursting patterns as well, we calculated the maximum number of bursts per trial for each HVC projection neuron. In juveniles we observed a larger number of bursts during motif per HVC projection neuron when compared to adults (maximum number of bursts per motif: juveniles = 0–4 bursts per motif, median = 2, 10 neurons in 4 juveniles; adults = 0–4 bursts per motif, median = 1, 54 neurons in 10 adults, p = 0.03, Wilcoxon rank sum test, Fig. 1C). Next, we measured the degree of stereotypy of bursting activity across song motifs. We identified bursts as reoccurring bursts across song motifs, if their onset time was within ± 20ms across motifs. In adults, 85.94% of all recorded bursts were consistently repeated across multiple renditions of the song motif, indicating a high degree of stereotypy. Conversely, in juveniles, only 60% of bursts were reoccurring (number of reoccurring bursts, juveniles = 12/20 bursts, adults = 55/64 bursts, p = 0.02, Fisher Exact Test). These observations are in line with previous work11 and suggest, that the functional connectivity within HVC exhibits greater variability in late-stage juveniles compared to adults.
In addition to bursting activity some neurons also exhibited single action potentials during song production which reduces the sparseness of the neural code and might induce less reliable behavioral outcomes in terms of song consistency and timing. To test whether single action potentials might contribute to a less temporally stereotyped song we explored the spiking activity of the HVC neurons (spikes within a burst and single spikes). We observed, that the number of spikes per motif tended to be higher in juvenile birds than in adults despite not significantly so (spikes per motif: juveniles = 6.5 spikes/motif, adults = 4 spikes/motif, p = 0.08, one-way ANOVA, Fig. 1D). Higher number of spikes per motif could also be attributed to single spikes alone outside of bursts. To account for the potential impact of single spikes, we separately analyzed their occurrence across song renditions. This analysis revealed that HVC neurons in juvenile zebra finches exhibited a significantly higher number of single action potentials compared to adults (single spikes per motif: juveniles = 0.25 single spikes/motif, adults = 0 single spikes/motif, p = 0.03, one-way ANOVA, Fig. 1E). The number of spikes recruited for bursting activity per song motif did not differ between juveniles and adults on a single neuron level (number of spikes within bursts: juveniles = 5.5 ± 3.54, median = 5.25, adults = 3.91 ± 3.96, median = 3.75, p = 0.23, one-way ANOVA, Fig. 1F). These results show that in late-stage development HVC projection neurons in juveniles exhibit detectable differences in spiking activity compared to adults. However, this increased spiking activity in juveniles is not reflected in the spiking patterns within bursts themselves. Instead, it can be attributed to a higher frequency of single spikes occurring outside of bursts. This increased incidence of single spikes might contribute to the greater variability observed in juvenile song production. We next explored, whether the song variability in juveniles during the late-state development can also be attributed to temporal dynamics within bursts in HVC projection neurons.
Temporal dynamics of HVC neuron bursting activity are slower in juveniles compared to adults
We explored the intrinsic dynamics of individual bursts by analyzing their individual number of spikes and temporal characteristics of these spikes during song production (Fig. 2A, 2B). First, we quantified the number of spikes occurring per burst. In both juveniles and adults, we observed a similar number of spikes occurring per burst (median spikes per burst ± std: juveniles = 3 ± 1.95, adults = 3 ± 1.94, p = 0.26, Wilcoxon rank sum test, Fig. 2C). Next, we assessed the variability of the number of spikes occurring in each reoccurring burst. HVC projection neurons in juveniles had a comparable distribution of Δ number of spikes per burst to neurons in adults (Δ number of spikes per burst: juveniles = 0.32 ± 0.61, adults = 0.37 ± 0.29, p = 0.09, Wilcoxon rank sum test, Fig. 2D), which indicates stereotyped reoccurring bursts during the late-stage development. We next investigated whether the temporal structure of bursts in juveniles and adults differed by quantifying their duration. Unlike previously reported11, we did not observe differences in the duration of bursts between juveniles and adults on a population level potentially due to our smaller sample size (duration of bursts: juveniles = 9.64 ms, adults = 6.83 ms, Wilcoxon rank sum test, p = 0.15, Fig. 2E). However, when assessing the instantaneous firing rate within bursts, we found that it was lower in juveniles than adults (firing rate: juveniles = 226.09 Hz, adults = 387.68 Hz, p < 0.001, Wilcoxon rank sum test, Fig. 2F) indicating that premotor signals are produced on a slower timescale compared to in the adult brain. To quantify whether the lower instantaneous firing rate also produced a distinct temporal pattern of spiking progression within bursts, we next compared the normalized pattern of spiking progression in juveniles and adults (Fig. 2G). Once we accounted for the overall higher instantaneous firing rate in adults by demeaning the progression pattern in juveniles and adults, the relative progression in juveniles was not distinguishable from that of the adult HVC projection neurons (demeaned firing rate: juveniles = 3.41*10− 14±25.45 Hz, adults=-1.14*10− 14±55.04 Hz, p = 0.84, Wilcoxon rank sum test, Fig. 2H).
To verify, whether the synaptically connected neurons provide stereotyped excitatory inputs preceding the bursts, we analyzed the membrane potential rise during a 15 ms window preceding all recorded bursts. Excitatory input accounting to the membrane potential rise preceding the bursts was as stereotyped in singing juveniles as in quiet juveniles where we elicited bursts using current injection (Wilcoxon rank sum test, p = 0.21) or in singing adults (Wilcoxon rank sum test, p = 0.11, Supplementary Fig. 1). This finding indicates, that the elicitation of bursts in singing juveniles occurs in a stereotyped way that is comparable to adults. The overall stereotyped characteristics of bursts (i.e., number of spikes, burst duration and stereotyped excitatory input) and the lower instantaneous firing rate indicate precise yet slower signal transmission in juvenile HVC projection neurons than in adults. We hypothesized, that these temporal dynamics might also be exhibited in a focal microcircuit within HVC. Here we leveraged the ability to quantify subthreshold activity of individual HVC projection neurons. This metric reflects the integrated input received by a focal neuron from its presynaptic network.
HVC projection neurons receive temporally distinct subthreshold inputs in juveniles compared to in adults
To quantify the temporal dynamics of inputs that HVC projection neurons are receiving, we assessed the stereotypy of the membrane potential by correlating subthreshold activity across motif renditions for each recorded neuron (Fig. 3A). In juveniles, the subthreshold activity was as stereotyped as in adults (subthreshold precision: juveniles = 0.82, adults = 0.83, p = 0.59, Wilcoxon rank sum test, Fig. 3B), suggesting a stable neural representation of song production. Since we previously reported temporal differences in bursting activity, we hypothesized, that these differences could also be reflected in the summation of excitatory and inhibitory postsynaptic potentials (PSPs) received by a focal neuron. To address these temporal dynamics, we quantified the duration of the PSPs and observed, that PSP events in juveniles were of longer duration than in adults (PSP duration: juveniles = 21.11 ± 2.94, adults = 18.09 ± 3.54, p = 0.01, Wilcoxon rank sum test, Fig. 3C), which is in line with our previously observed slower temporal dynamic within bursts in juveniles. Further we quantified the membrane potential amplitude, and found that PSPs in juveniles had a higher membrane potential amplitude than in adults (PSP amplitude: juveniles = 6.37 ± 1.2mV, adults = 5.64 ± 1.43, p < 0.01, Wilcoxon rank sum test, Fig. 3D), which could potentially be accounted to a higher resting membrane potential in juveniles27. In juveniles, a higher resting membrane potential is associated with altered neuronal excitability and could likely manifest as changes in the frequency or amplitude of postsynaptic potential events. The frequency of the PSP element occurrence was lower in juveniles than adults (PSP element frequency: juveniles = 12.85 ± 4.99, adults = 16.03 ± 4.91, p = 0.045, Wilcoxon rank sum test, Fig. 3E), suggesting a slower-paced, more sparse input from synaptically connected neurons. The slower and more sparse membrane potential dynamics during singing in juveniles might explain the temporal differences during song performance4,23.