Abstract
The acoustic environment an animal experiences early in life shapes the structure and function of its auditory system. This process of experience-dependent development is thought to be primarily orchestrated by potentiation and depression of synapses, but plasticity of intrinsic voltage dynamics may also contribute. Here, we show that in juvenile male and female zebra finches, neurons in a cortical-level auditory area, the caudal mesopallium (CM), can rapidly change their firing dynamics. This plasticity was only observed in birds that were reared in a complex acoustic and social environment, which also caused increased expression of the low-threshold potassium channel Kv1.1 in the plasma membrane and endoplasmic reticulum (ER). Intrinsic plasticity depended on activity, was reversed by blocking low-threshold potassium currents, and was prevented by blocking intracellular calcium signaling. Taken together, these results suggest that Kv1.1 is rapidly mobilized to the plasma membrane by activity-dependent elevation of intracellular calcium. This produces a shift in the excitability and temporal integration of CM neurons that may be permissive for auditory learning in complex acoustic environments during a crucial period for the development of vocal perception and production.
SIGNIFICANCE STATEMENT Neurons can change not only the strength of their connections to other neurons, but also how they integrate synaptic currents to produce patterns of action potentials. In contrast to synaptic plasticity, the mechanisms and functional roles of intrinisic plasticity remain poorly understood. We found that neurons in the zebra finch auditory cortex can rapidly shift their spiking dynamics within a few minutes in response to intracellular stimulation. This plasticity involves increased conductance of a low-threshold potassium current associated with the Kv1.1 channel, but it only occurs in birds reared in a rich acoustic environment. Thus, auditory experience regulates a mechanism of neural plasticity that allows neurons to rapidly adapt their firing dynamics to stimulation.
Introduction
The voltage-gated ion channels an individual neuron contains in its plasma membrane determine how it integrates synaptic inputs over time to produce patterns of action potentials, affecting its functional response properties (Llinás, 1988; Storm, 1988). The diverse intrinsic cellular dynamics found within many brain areas contribute to sensory coding (Svirskis et al., 2002; Rothman and Manis, 2003; Padmanabhan and Urban, 2010), memory consolidation (Titley et al., 2017; Lisman et al., 2018), and maintaining network stability (Cudmore et al., 2010). Although the complement of channels expressed by individual neurons may primarily depend on the genetic programs that define cell types (Bishop et al., 2015), intrinsic dynamics often remain plastic, changing in response to activity and experience (Daoudal and Debanne, 2003; Zhang and Linden, 2003; McKay et al., 2009; Debanne et al., 2019; Santin and Schulz, 2019). Alongside synaptic plasticity, intrinsic plasticity may contribute substantially to experience-dependent functional changes during learning and development (Titley et al., 2017; Lisman et al., 2018; Daou and Margoliash, 2021), but in contrast to synaptic plasticity, the underlying mechanisms remain poorly understood.
The zebra finch (Taeniopygia guttata) is a social songbird that communicates with learned and innate vocalizations (Elie and Theunissen, 2016). As in many other species and sensory systems, development of the zebra finch auditory system is shaped by the acoustic environment experienced early in life (Amin et al., 2007; Woolley, 2012; Moore and Woolley, 2019). The intrinsic dynamics of neurons in the caudal mesopallium (CM), a cortical-level auditory area implicated in discriminating and learning species-specific vocalizations (Gentner and Margoliash, 2003; Meliza and Margoliash, 2012), are sensitive to acoustic experience during the developmental window when young males begin to memorize a song to copy (Chen and Meliza, 2020). Just after zebra finches fledge from the nest (18–20 d after hatch), almost all of the putative excitatory neurons in CM exhibit tonic dynamics, spiking continuously when depolarized. In typical rearing conditions, where nests are in a breeding colony with many other families, by 30–35 d after hatch about half of these cells are phasic, spiking only at the onset of stimulation. If the birds are instead reared in a quiet environment, phasic neurons do not appear until much later in life, suggesting that intrinsic plasticity is an adaptive mechanism that permits normal auditory development in challenging acoustic conditions.
Phasic dynamics in CM depend on a low-threshold outward current (Chen and Meliza, 2018), and the experience-dependent emergence of phasic dynamics between P30 and P35 is correlated with expression of Kv1.1 (Chen and Meliza, 2020), a low-threshold voltage-gated potassium channel that facilitates phasic firing and is regulated by activity in a number of other systems (Brew and Forsythe, 1995; Bal and Oertel, 2001; Lu et al., 2004; Watanabe et al., 2014; Dehorter et al., 2015; Akter et al., 2018; Morgan et al., 2019). However, expression alone is not sufficient to cause a shift in intrinsic dynamics, because Kv1.1 must be exported to the plasma membrane, a process that may require activity-dependent elevation of intracellular calcium levels (Manganas et al., 2001; Rangaraju et al., 2010; Vacher and Trimmer, 2012; Galea et al., 2014). Indeed, at light microscopy resolution, much of the Kv1.1 staining in CM neurons is confined within the cell body (Chen and Meliza, 2020), where it might comprise a pool of low-threshold potassium channels that could be trafficked to the plasma membrane in response to activity.
In the present study, we tested this model using several approaches. We examined how acoustic experience influences the expression and localization of Kv1.1 using confocal and electron microcopy in CM neurons from birds between 30 and 35 d of age, confirming the existence of an experience-dependent pool of Kv1.1 in the endoplasmic reticulum (ER) that could be exported to the membrane. We made whole-cell recordings from putative excitatory neurons during this window to determine whether Kv1.1 expression is sufficient to predict whether an individual cell is tonic or phasic. Finally, we tested whether repeatedly stimulating neurons to spike would cause them to become more phasic, using pharmacology to determine whether this change in intrinsic dynamics depends on low-threshold potassium currents.
Materials and Methods
Animals
All procedures were performed according to National Institutes of Health guidelines and protocols approved by the University of Virginia Institutional Animal Care and Use committee. Experiments used tissue from juvenile zebra finches (T. guttata; 52 birds) between 30 and 35 d after hatch, the developmental period when phasic excitability peaks in colony-reared (CR) birds (Chen and Meliza, 2020). Finches were bred in our colony from 25 different pairs. All birds received finch seed (Abba Products) and water ad libitum and were kept on a 16/8 h light/dark schedule in temperature-controlled and humidity-controlled rooms (22–24°C).
Experimental rearing conditions
All of the animals in the study were bred in individual cages that were initially placed in a room housing dozens of male and female finches of varying ages. Colony-reared (CR) chicks remained with their parents in the colony room until they were used in an experiment. Pair-reared (PR) chicks were moved with their parents into an acoustic isolation box (Eckel Industries) within 7 d after the first egg hatched. Both PR and CR chicks heard the songs of their father and the calls of their whole family, but PR birds did not experience the acoustic environment of the colony room. Based on nestbox recordings reported in Chen and Meliza (2020), the sound amplitude (calculated using RMS in 1-min intervals) is higher in the colony room than (mean ± SD =
Immunofluorescence
We compared Kv1.1 expression in CR and PR animals using a similar procedure to the one described previously (Chen and Meliza, 2020). Animals were administered a lethal intramuscular injection of Euthasol and perfused transcardially with a 10 U/ml solution of sodium heparin in PBS (in mm: 10 Na2HPO4, 154 NaCl, pH 7.4) followed by 4% paraformaldehyde (in PBS). Brains were immediately removed from the skull, postfixed overnight in 4% paraformaldehyde at 4°C, and then transferred to PBS. Brains were blocked along the same plane used for making acute slices, and sections were cut at 60 μm on a vibratome and stored in PBS with 0.05% sodium azide until ready for staining. Sections were rinsed four times in PBS and then incubated in citrate buffer (0.01 m, pH 8.5) at 80°C for 25 min. The sections were cooled to room temperature, washed twice in PBS, permeabilized in PBS-T (0.1% Tween 20 in PBS), and blocked in 5% goat serum and 2% glycine in PBS-T. The tissue was stained for Kv1.1 using a monoclonal mouse antibody (1:500 in blocking solution; UC Davis/NIH Neuromab clone K20/78; RRID:AB_10673165) for 60–72 h at 4°C. Sections were washed four times for 15 min in PBS-T and then incubated with a fluorescent secondary antibody (1:2000 in PBS-T; goat anti-mouse IgG1 conjugated to Alexa Fluor 488; Invitrogen; RRID:AB_2534069). The tissue was washed twice with PBS, counterstained for Nissl (Neurotrace 640/660, 1:1000 in PBS; ThermoFisher, catalog #N21483; RRID:AB_2572212), washed four times with PBS, and then mounted and coverslipped in Prolong Gold with DAPI (ThermoFisher, catalog #P36934; RRID:SCR_015961).
Stained tissue was imaged on a Zeiss LSM 800 confocal microscope using a 40× objective (water immersion, NA 1.2) in stacks with 0.44 μm between optical sections. At least three stacks were imaged in each section, and two sections were imaged per animal for a total of six sections. To quantify expression, we counted the number of Kv1.1-positive neurons in each stack manually in Imaris (version 9.2.1) after using the automatic background subtraction, with the default smoothing radius of 39.9 μm. The total number of neurons in each stack was counted from the DAPI channel using the automatic spot detection function in Imaris, followed by a manual check. Samples were processed and analyzed in two batches, and each batch contained samples from both conditions. Identical laser power and gain settings were used for the Kv1.1 channel in all the samples from each batch, but the DAPI settings were adjusted freely to optimize signal quality. Experimenters were blind to animal identity and rearing condition throughout staining, image collection, and analysis.
Slices used for acute recordings were stained for Kv1.1 and biocytin using a similar procedure with modifications for thick sections. Permeabilization, blocking, and staining solutions were prepared with a stronger detergent (0.3% Triton X-100 instead of 0.1% Tween 20). The blocking step was 2.5 h; the primary antibody incubation was 7 d; and the secondary staining incubation lasted for 20 h and included streptavidin-conjugated Alexa Fluor 635 (1:1000, Invitrogen) to label biocytin. The biocytin-filled neuron was imaged with the same microscope, objective, and spacing between optical sections described above.
Cells were only included for analysis if the recorded neuron could be unambiguously identified in the slice (sometimes biocytin expelled from the pipette during patching was taken up by neighboring cells) and if Kv1.1 staining was good enough to clearly see positive neurons at the depth of the cell within the section. Included neurons were classified as Kv1.1-negative or Kv1.1-positive by an observer (C.D.M.) who was blind to the electrophysiological characteristics of the cell. Cells were deemed to be positive only if they had clearly defined, punctate Kv1.1 staining that was elevated higher than the background, co-extensive with the biocytin signal in the soma, and showing a region of lower staining corresponding to the nucleus (Fig. 1A). Almost all (
Early auditory experience modulates expression and localization of Kv1.1. A, Examples of α-Kv1.1 staining in CM at P30–P35 in CR and PR zebra finches. Images are single confocal slices (0.44) with identical laser, gain, and contrast settings. Scale bar: 20 μm. B, Proportion of Kv1.1-positive neurons in each rearing condition from
The biocytin-filled neurons were also analyzed using an automated pipeline to quantify expression by the number of puncta in the soma. The 3D stacks were first cropped in Imaris to include only the volume immediately surrounding the soma. One cell was discarded because of a blood vessel within this volume that stained strongly in the Kv1.1 channel. The cropped image was exported for analysis in Cell Profiler (version 4.2.4). Illumination was corrected separately in the biocytin and Kv1.1 channels using the regular illumination function with a polynomial smoothing function. An Otsu global method strategy with three-class thresholding to identify objects from six to 14 pixels in diameter in the Kv1.1 channel. The Otsu global method strategy with two-class thresholding was used to identify the soma from the biocytin channel. Finally, the ClassifyObjects method was used to count Kv1.1 puncta located with the cell soma. Density was calculated as the total number of colocated puncta divided by the sum of the area in each optical section times the section thickness (0.44 μm).
Immuno-electron microscopy (EM)
Animals were deeply anesthetized with an overdose of euthasol and transcardially perfused with Tyrode’s solution followed by 100 ml of fixative solution containing 4% paraformaldehyde and 0.1–0.3% glutaraldehyde in PBS. Brains were immediately removed from the skull, postfixed overnight in 4% paraformaldehyde at 4°C, blocked for modified coronal sections, and cut at 60 μm on a vibratome. Sections were rinsed in 1% sodium borohydride and stored in 0.05% sodium azide in PBS at 4°C.
Kv1.1 was stained as for immunofluorescence, with the following changes: sections were incubated in citrate buffer (0.01 m, pH 8.5) at 80°C for 2× 15 min. Sections were cooled to room temperature, washed three times in PBS, and preincubated in a solution with 1% bovine serum albumin (BSA) and 0.1% Triton X-100 in PBS for only 30 min. Sections were stained with a solution of primary antibody (1:500), 1% BSA, and 0.05% sodium azide for 72 h at 4°C on a shaker. The secondary antibody incubation was with horse anti-mouse IgG conjugated to biotin (1:100 in PBS; Vector Labs, catalog #BA-2000-1.5) for 2 h at room temperature. Sections were treated with avidin-biotinylated peroxidase (VECTASTAIN ABC-HRP staining kit; Vector Labs, catalog #PK-4000) for 2 h and developed in a solution of hydrogen peroxide (H2O2) and 0.05% diaminobenzidine (DAB) for 2–7 min.
Sections were embedded for electron microscopy by incubation with 1% osmium tetroxide (OsO4) in 0.1 m sodium phosphate buffer (PB; pH 7.4) for 1 h then with filtered 4% uranyl acetate in 70% alcohol for 1 h, followed by dehydration in acetone and treatment with a 1:1 acetone/resin mixture overnight. The following day, sections were transferred to full resin and left overnight. Sections were then flat-embedded between two aclar sheets and cured in a 60°C oven overnight. The region of interest from each section was placed in a BEEM capsule (EMS). The capsules were filled with resin and cured at 60°C for 24–48 h, or until polymerized. The region of interest (now capsule-embedded) was traced with a camera lucida and trimmed down to a 1 × 2 mm trapezoid. Ultrathin sections of 50–80 nm in thickness were collected on 400 mesh copper grids (Ted Pella) using an ultramicrotome (Ultracut UCT7; Leica). The ultrathin sections were examined on a JEOL1010 electron microscope equipped with a 16-megapixel CCD camera (SIA). Images for quantitative and immuno-labeled terminal analysis were taken at 8000–15,000× magnification, yielding a resolution of 576.06–1134.92 pixels/μm. Kv1.1 localization was analyzed in 25 neurons from each bird (
Brain slice preparation and electrophysiology
Whole-cell recordings were obtained from acute slices of the caudal mesopallium. Birds were administered a lethal intramuscular injection of Euthasol (pentobarbital sodium and phenytoin sodium; 200 mg/kg; Hospira) and perfused transcardially with ice-cold cutting buffer (in mm: 92 NaCl, 2.5 KCl, 1.2 NaH2PO4, 30 NaHCO3, 20 HEPES, 25 glucose, 5 sodium ascorbate, 2 thiourea, three sodium pyruvate, 10 MgSO4·7H2O, 0.5 CaCl2·2H2O, pH 7.3–7.4, 310–330 mmol/kg). The brain was blocked using a custom 3D-printed brain holder (http://3dprint.nih.gov/discover/3dpx-003953) on a dorsorostral to ventrocaudal plane pitched 65° forward of the transverse plane, so that slices through the auditory pallium would be approximately orthogonal to the ventral mesopallial lamina (LMV; Chen and Meliza, 2018). Sections 300 μm thick were cut in room-temperature cutting buffer on a VF-200 or VF-500 Compresstome (Precisionary Instruments), and then transferred to 32°C cutting buffer for 10–15 min for initial recovery. Slices were then transferred to room-temperature holding buffer (in mm: 92 NaCl, 2.5 KCl, 1.2 NaH2PO4, 30 NaHCO3, 20 HEPES, 25 glucose, 5 sodium ascorbate, 2 thiourea, three sodium pyruvate, 2 MgSO4·7H2O, 2 CaCl2·2H2O, pH 7.3–7.4, 300–310 mmol/kg) to recover for at least 1 h before use in recordings. All solutions were bubbled continuously with 95% O2·5% CO2 mixture starting at least 10 min before use.
Slice recordings were conducted in a RC-26G recording chamber (Warner Instruments) perfused with standard recording artificial CSF (ACSF; in mm: 124 NaCl, 2.5 KCl, 1.2 NaH2PO4, 24 NaHCO3, 5 HEPES, 12.5 glucose, 2 MgSO4·7H2O, 2 CaCl2·2H2O; pH 7.3–7.4, 300–310 mmol/kg) at a rate of 1–2 ml/min at 32°C. Whole-cell patch-clamp recordings were obtained under 40× infrared (900 nm) DIC optics. CM was located relative to the lamina mesopallialis ventralis (LMV) and the internal occipital capsule (CIO), which both comprise dense myelinated fibers visible as dark bands under bright-field or IR illumination. Recording pipettes were pulled from filamented borosilicate glass pipettes (1.5 mm outer diameter, 1.10 mm inner diameter; Sutter Instruments) using a P-1000 Micropipette Puller (Sutter Instruments) and were filled with internal solution (in mm: 135 K-gluconate, 10 HEPES, 8 NaCl, 0.1 EGTA, 4 MgATP, 0.3 NaGTP, 10 Na-phosphocreatine, pH 7.2–7.3, 300–310 mmol/kg). In some recordings, 0.2% biocytin (Sigma) was included in the internal solution.
Electrodes had a resistance of 4–8 MΩ in the bath. Voltages were amplified with a Multiclamp 700B amplifier (Molecular Devices) in current-clamp mode, low-pass filtered at 10 kHz, and digitized at 50 kHz with a Digidata 1440A. Pipette capacitance was neutralized, and 8–12 MΩ of series resistance was subtracted by bridge balance. Recorded voltage was corrected offline for a liquid junction potential of –11.6 mV (as measured at 32°C). Current injections and data collection were controlled by pCLAMP (version 10.4; Molecular Devices). Neurons were excluded from analysis if the resting membrane potential was above –55 mV or if action potentials failed to cross –10 mV.
After a brief test to ascertain how much current was needed to evoke spikes, each neuron was recorded in epochs of 20 sweeps. Each sweep was 6 s in duration: 100 ms at 0 pA, a 2000-ms depolarizing step, 1000 ms at 0 pA, 500 ms at –20 pA, 500 ms at –40 pA, 500 ms at –20 pA, and 1400 ms at 0 pA. The amplitude of the depolarizing pulse was systematically varied within each epoch over a range that was set so that at least 10 sweeps in each epoch evoked action potentials without eliciting depolarization block. The range of current steps was adjusted throughout the recording as needed. For the colocalization experiments, the recordings were terminated after one or two epochs by slowly withdrawing the electrode from the cell until it detached and formed an outside-out patch, as evidenced by a sharp increase in resistance. The slice was allowed to rest in the recording chamber for 5–10 min and then fixed in 4% formaldehyde for 15 h before staining. In plasticity experiments, the standard condition was to record epochs one after another as long as the cell appeared to remain healthy. Additional exclusion criteria were applied post hoc (see below). In the reduced stimulation condition, after recording a single epoch at the beginning of the experiment, the cell was voltage-clamped at resting potential for at least 400 s, followed by another recording epoch. During the voltage-clamp interval, performing a seal test (–2-mV pulses at 100 Hz) tended to help avoid resealing.
Pharmacology
To test whether blocking potassium currents would reverse intrinsic plasticity, recordings were performed in standard ACSF with an added cocktail of drugs to block fast synaptic transmission: 10 μm (
±)−3-(2-carboxyp-iperazin-4-yl)propyl-1-phosphonic acid (CPP; Alomone Labs), 20 μm 6-cyano-7-nitroquinoxaline-2,3-dione disodium salt hydrate (CNQX; Alomone Labs), and 10 μm (+)-bicuculline (Sigma). After the neuron became phasic (4–11 epochs), 2 mm 4-aminopyridine (4-AP; Sigma), or 100 nm α-dendrotoxin (α-DTX; Alomone Labs) was added to the perfusate. In the dendrotoxin experiments, solutions were prepared with 0.2% BSA to minimize nonspecific binding of the drug to the perfusion tubing and chamber. BSA appeared to cause series resistance to be higher and less stable, so the exclusion criterion for series resistance deviation (see below) was relaxed to 50% from baseline. In experiments to reduce intracellular Ca
Signal processing and analysis
Recordings were analyzed using custom Python code (https://github.com/melizalab/cm_plasticity). If the average series resistance deviated in any epoch by >30% from baseline, the resting membrane potential increased by >10 mV, or the cell spiked spontaneously more than five times, that epoch and all the ones that followed were excluded. Spontaneous firing was used as an exclusion criterion because it indicated poor health, it interfered with analysis, and it introduced uncontrolled variation in activity. Input resistance was not used as an exclusion criterion because it was expected to change with an increase or decrease in low-threshold potassium conductance. After excluding bad epochs, individual sweeps were excluded if the resting membrane potential was transiently unstable, defined as a fluctuation >10 times the median absolute deviance. Action potentials were detected and the spike trains were analyzed following previously described procedures (Chen and Meliza, 2018, 2020). The response duration in each sweep was measured as the time elapsed between the first and the last spikes of the depolarizing pulse. In responses with only one spike, this difference was undefined, so we imputed a value equal to the width of the spike plus the duration of the afterhyperpolarization, with the rationale that this value would be approximately what we would measure if the cell had spiked two times in rapid succession. The rheobase was defined as the average of the largest current that produced no spikes and the smallest current that produced one or more spikes. The slope of the frequency-current (f-I) curve was calculated for each epoch from the above-threshold sweeps as the average of the change in firing rate divided by the average of the increase in current. Current-voltage (I-V) curves were measured from all sweeps using the steady-state voltages produced by both depolarizing and hyperpolarizing pulses. For the depolarizing currents, the voltage values were calculated from the last 500 ms of the pulse and were only included if there were no action potentials during this interval.
Experimental design and statistical analysis
Intrinsic spiking dynamics were quantified using two measures: the average duration of spiking in response to depolarizing step currents, and the average slope of the f-I curve. In contrast to our previous study, we did not log-transform the durations, because the residuals of statistical models using untransformed durations (i.e., arithmetic difference) were closer to a normal distribution than the residuals in models with log-transformed durations (i.e., geometric difference), as assessed by visual examination of a Q-Q plot and the Shapiro–Wilk test. Duration and slope were either compared with the expression of Kv1.1 measured with immunohistochemistry or tracked over time to determine under what conditions intrinsic dynamics were stable or plastic. At least seven animals were used for slice experiments in each condition tested; this sample size was chosen based on previously published studies from our lab examining intrinsic dynamics in these neurons (Chen and Meliza, 2018, 2020). The experimenter was aware of the rearing conditions of the animals and the experimental conditions during recording. The experimental design was not preregistered.
Statistical inference was performed for confocal and electrophysiology data using mixed-effects models with bird and family identity as random effects to account for repeated measures. When the dependent measure was spike duration, individual sweeps were treated as the unit of observation, and a random intercept (and slope, if applicable) was added for each neuron. Although the results of the four experiments measuring intrinsic plasticity are presented separately, it is important to note that the effects were estimated in a single unified model with all four conditions. Fixed-effect contrasts were planned to compare the last good epoch with the first epoch, and the experimental conditions (PR birds, BAPTA-AM, and reduced stimulation) with the standard condition (CR birds, standard ACSF, continuous stimulation). The reversal experiment was analyzed separately, again with a single model for both experimental conditions (4-AP and α-DTX). This analysis could not be combined with the plasticity analysis because it only included neurons with an average response duration that was initially >0.5 s and that decreased by at least 0.4 s over the course of recording. The last epoch recorded after applying 4-AP/α-DTX was compared with the epoch before it was applied. The proportion of cells expressing Kv1.1 in CR and PR birds was estimated using a mixed-effects model with a binomial error distribution. Image stacks were treated as the unit of observation, with section, bird, and staining batch as random effects, and rearing condition as the sole fixed effect. All of the models were fit in R using the lme4 package (version 1.1–29), and the effects, post hoc comparisons, confidence intervals, and p values were calculated using the emmeans package (version 1.8.1). This package uses the Tukey method to correct p values for multiple comparisons, and it reports t statistics using Satterthwaite approximations for the effective degrees of freedom, which are denoted in the text as subscripts and reflect both the sample size and within-group variances; thus, not all effects tested in a model have the same degrees of freedom. Post hoc tests were only performed if there was a significant difference identified by ANOVA.
Nonparametric statistical tests (Wilcoxon rank-sum test) were used for the immuno-EM data because of the small number of cells and animals.
Results
Kv1.1 expression and localization depend on experience
To test whether there is an experience-dependent pool of ready-to-use Kv1.1 that could support rapid changes of intrinsic dynamics, we used immunohistochemistry to quantify expression and localization of Kv1.1 in CM neurons. We compared tissue from colony-reared (CR) and pair-reared (PR) birds between P30 and P35 (days after hatch). Both CR and PR birds are exposed to songs from their father and calls from their immediate family, but CR birds hear these vocalizations against a noisy, complex background produced by many other individuals, while PR birds hear them in relative quiet. In confocal images, Kv1.1 stains weakly throughout the neuropil and intensely in cell bodies outside the nucleus. CR birds had higher levels of Kv1.1 expression than PR birds, with much stronger staining in the neuropil and a much larger proportion of neurons expressing detectable levels of Kv1.1 (Fig. 1A,B). This result is consistent with our earlier finding that Kv1.1 expression is higher in CR birds compared with female-reared birds, who do not hear any song (Chen and Meliza, 2020), and it confirms that Kv1.1 expression reflects the properties of the acoustic environment rather than exposure to a tutor song.
To determine what subcellular compartments contained Kv1.1, we used electron microscopy to compare immunolabeled tissue from CR and PR birds. Kv1.1 protein was associated with the plasma membrane, the ER, and the cytosolic side of the nuclear membrane (Fig. 1C). A similar pattern is seen in the mammalian neocortex (Goldberg et al., 2008). Staining was observed in dendrites and spines near the postsynaptic density; these occurrences were not scored but may also represent Kv1.1 that could contribute to membrane voltage dynamics. Staining was also present around mitochondria and in regions of the cytoplasm not obviously occupied by any organelles. As in the confocal images, there was more Kv1.1 in the CR birds than in the PR birds (Fig. 1G). CR birds had more dense staining along the plasma membrane (Fig. 1D) and along the nuclear membrane (Fig. 1E), and there were more clusters in the ER (Fig. 1F) and in all the compartments (Fig. 1G). The cytosolic area was statistically indistinguishable between CR and PR birds (Wilcoxon test:
Kv1.1 is not sufficient for phasic excitability
The existence of an intracellular pool of Kv1.1 implies that expression of Kv1.1 is necessary but not sufficient for phasic excitability. That is, a neuron might express high levels of Kv1.1, but if most was inside the cell instead of on the membrane, the spiking dynamics would still be tonic. To test this prediction, we performed whole-cell recordings from neurons in acute slices of CM from colony-reared chicks. Cells were stimulated with 2-s depolarizing current steps at a range of amplitudes, and as we have seen previously, some neurons responded throughout the depolarization (tonic spiking) while others responded only near the onset (phasic spiking). Other neurons fell somewhere between these two extremes, producing tonic responses on some trials and phasic responses on others. We quantified each neuron’s tendency toward phasic or tonic spiking along this continuum by averaging the duration of spiking across all the suprathreshold amplitudes. Longer durations indicate a tendency for tonic spiking (Fig. 2A,B), and shorter durations indicate a tendency for phasic spiking (Fig. 2C). The recording pipette was filled with biocytin, and after the recording was completed, the slices were stained for biocytin and Kv1.1 to determine what, if any, relationship there was between intrinsic dynamics and Kv1.1 expression.
Kv1.1 is not sufficient for phasic excitability. A, Example of a Kv1.1-negative tonic neuron. Image is a single confocal slice (0.44 μm) showing fluorescent staining for Kv1.1 (cyan) and biocytin (magenta). Scale bar: 40 μm. Bottom shows exemplar voltage traces in response to step current injections. Scale bars: 20 mV, 50 pA, 500 ms. B, Example of a Kv1.1-positive tonic neuron. C, Example of a Kv1.1-positive phasic neuron. D, Average response duration for Kv1.1-negative (
As illustrated in Figure 2A–C, biocytin-filled neurons had quasi-pyramidal somata with dendrites radiating in all directions. All but one of the neurons with good staining in the dendrites (
Rapid, experience-dependent plasticity of intrinsic dynamics
CM neurons that contain a pool of Kv1.1 in the ER and other intracellular locations should be able to rapidly shift toward more phasic spiking by exporting Kv1.1 to the plasma membrane. To test this hypothesis, we tracked evoked spiking patterns for 100–2500 s in whole-cell recordings from CR chicks (Fig. 3A). Many neurons exhibited a shift during the recording from tonic to phasic spiking (Fig. 3B,C), along with an increase in the rheobase and a decrease in the slope of the relationship between firing rate and injected current (f-I curve; Fig. 3D,F). Phasic spiking was typically accompanied by the emergence of outward rectification at around –60 mV (Fig. 3E), which is characteristic of a low-threshold voltage-gated potassium current. Resting membrane voltage and input resistance tended to remain stable or to decrease, which is what would be expected if the low-threshold potassium current is partly activated at rest.
Rapid emergence of phasic excitability in colony-reared (CR) birds. A, Neurons from colony-reared chicks were stimulated with epochs of depolarizing step currents 2 s in duration with varying amplitude. B, Tonic responses of an exemplar neuron immediately after establishing the whole-cell recording. Top is a raster plot showing spike times. Below are voltage and current traces for three selected trials. Scale bars: 40 mV, 100 pA, 500 ms. C, Phasic response of the same exemplar neuron 841 s after the start of the recording. Same format as in A. D, Firing rate versus injected current amplitude (f-I curve) for the first and last epochs. E, Steady-state voltage versus injected current amplitude for the first and last epochs of the recording. The line is plotted only when the cell did not spike during the measurement window. Data for the negative current values was taken from hyperpolarizing pulses used in every sweep to track input and series resistance. Note the emergence of strong outward rectification in the latter epoch. F, Firing duration, slope of frequency versus current, rheobase (Ith), resting potential (Vm), and resting input resistance (Rm) for each epoch in the recording. Vertical lines show standard error of mean for duration, Vm, and Rm, which were averaged across all the sweeps in the epoch. Pauses in the recording occurred when the range of stimulating currents was adjusted to ensure an adequate number of traces included spikes (see Materials and Methods). G, Change in duration from first to last epoch versus total recording time (top) and total number of spikes (bottom). Each symbol corresponds to the last epoch recorded in an individual neuron. H, Comparison of average firing duration in the first and last epochs for
Large changes in intrinsic dynamics were observed only after ∼400 s of recording, and persisted throughout recordings that lasted for over 30 min (Fig. 3G). Not every neuron was plastic. When intrinsic dynamics changed, it was almost always in the direction of more phasic responses. Nearly half of the neurons (43%,
If this shift in intrinsic dynamics is caused by export of Kv1.1 to the plasma membrane, then it may depend on the amount of readily available intracellular Kv1.1. Pair-reared birds have lower levels of Kv1.1 expression in the ER and overall (Fig. 1), so we expected that the intrinsic dynamics in neurons from PR birds would be much less plastic compared with what we observed in CR birds. This was indeed the case. As illustrated in Figure 4, although there was a tendency for neurons in PR birds to become less excitable, their firing patterns remained stable. Decreased excitability was sometimes accompanied by a shift in the I-V curve at hyperpolarized voltages (Fig. 4E), but the low-threshold outward current that appeared in plastic CR neurons (Fig. 3E) was not observed. Overall, in PR birds there was no significant change in firing duration (Fig. 4H), even in recordings lasting close to 30 min (Fig. 4G). There was a modest but significant decrease in f-I slope (Fig. 4I), suggesting that other mechanisms may contribute to the decrease in excitability.
Tonic dynamics are stable in pair-reared (PR) birds. A, Experimental design: neurons from PR chicks were stimulated with epochs of depolarizing step currents. B–F, Responses of an exemplar neuron to depolarizing step currents immediately after establishing the whole-cell recording and 1001 s later. Same format as Figure 3B–F. G, Change in duration from first to last epoch versus total recording time (top) and total number of spikes (bottom). H, Comparison of average firing duration in the first and last epochs for
This result rules out the possibility that changes in firing patterns are a consequence of degrading recordings, which can lead to reduced excitability as cells become unhealthy and electrochemical gradients run down. It is unlikely that the neurons in PR birds are inherently healthier than in CR birds. There was no statistically significant difference between CR and PR birds in initial resting membrane potential (LMM:
Intrinsic plasticity requires low-threshold potassium currents
To further confirm whether neurons in CR birds were becoming more phasic because of Kv1.1, we tested whether the plasticity could be reversed by blocking low-threshold potassium currents. We previously showed that 4-AP, a relatively nonselective blocker of voltage-gated potassium channels, and α-DTX, which is selective for channels in the Kv1 family, both cause phasic cells to become more tonic (Chen and Meliza, 2018). Here, when we applied α-DTX or 4-AP to tonic neurons that had become phasic over the course of recording (Fig. 5A), the cells reverted to tonic firing. As seen for an example neuron treated with α-DTX (Fig. 5B), the re-emergence of tonic firing was accompanied by reduction in outward rectification (Fig. 5C) and in the rheobase (Fig. 5D). Both drugs caused a significant reversal of the plasticity (Fig. 5E). In addition to restoring tonic firing, 4-AP also caused a broadening of action potentials consistent with its ability to block other voltage-gated potassium channels involved in repolarization at the concentration used here (2 mm). The magnitude of the reversal was slightly greater with 4-AP (LMM:
Blocking low-threshold potassium currents reverses the shift to phasic dynamics. A, Neurons from CR birds were stimulated with epochs of depolarizing step currents. After 700–1700 s of recording, either 4-AP (2 mm) or α-DTX (100 nm) was applied in the bath. B, Responses of an exemplar neuron during the first epoch, the last epoch before drug application (predrug), and the epoch after drug application (α-DTX). Format is the same as in Figure 3B,C. Scale bars: 40 mV, 500 ms. Current steps: 130, 150, 170 pA. C, Steady-state voltage versus injected current for the epochs shown in B. D, Firing rate versus injected current amplitude for the epochs in B. E, Comparison of average firing duration in the first, predrug, and postdrug epochs for neurons treated with 4-AP (brown;
Intrinsic plasticity depends on activity and intracellular calcium
Neuronal activity drives an increase in Kv1 currents in CA1 cells of the mammalian hippocampus (Morgan et al., 2019) and in the avian nucleus magnocellularis (Adachi et al., 2019), so we hypothesized that intrinsic plasticity in CM was also triggered by activity. It was not possible to completely eliminate neural activity, because we had to evoke spiking to measure neuronal firing patterns. Instead, we used a minimal stimulation protocol with only a single epoch recorded at the beginning and end of the experiment, and the neuron was voltage-clamped at its resting potential in the interim (Fig. 6A). As predicted, neurons were much less plastic when they were stimulated less (Fig. 6B). Only one of the cells (6%,
Intrinsic plasticity depends on activity and intracellular calcium A, In the minimal stimulation condition, neurons from CR chicks were tested with a single epoch of depolarizing step currents, clamped at the resting potential for at least 400 s, and then tested with another epoch of current steps. B, Comparison of average firing duration (left) and f-I slope (right) in the first and last epochs for the minimal stimulation condition (
One possibility whereby activity could trigger export of Kv1.1 to the plasma membrane is through an increase in intracellular calcium concentrations. To test this hypothesis, intracellular calcium dynamics were clamped using the fast calcium chelator BAPTA, administered by preincubating the slices with the AM ester of BAPTA (Fig. 6C). We used this approach because it allowed us to keep our internal pipette solution the same for all experiments. With this treatment, there was no significant change in duration or f-I slope (Fig. 6D).
A significant change in firing duration was only observed in CR birds under normal recording conditions. To compare plasticity across all the conditions tested (and avoid the fallacy that a difference in significance implies a significant difference), the data were analyzed in unified linear mixed-effects models, one for firing patterns (Fig. 7A) and one for excitability (Fig. 7B). On average, spiking duration decreased by over 500 ms in CR neurons under normal conditions, significantly more than in PR neurons, in the minimal stimulation condition, and with BAPTA treatment. The PR, minimal stimulation, and BAPTA conditions were not statistically distinguishable from each other. For excitability (slope of the f-I relationship), there was no significant difference among the conditions. This result shows that there was both a nonspecific decrease in excitability across all our experiments and a specific change in firing patterns related to experience, activity, and intracellular calcium.
Intrinsic plasticity depends on experience, activity, and intracellular calcium, and is correlated with changes in passive membrane properties. A, Average change in response duration for the CR, PR, minimal stimulation, and BAPTA conditions. Hollow circles show means; thick and thin whiskers show 50% and 90% credible intervals for the mean. The change in duration in the CR condition is greater than for the PR condition (
In further support of a specific mechanism, the changes in duration we observed were correlated with changes in intrinsic properties that would be expected from an increase in the conductance of a Kv1.1-like potassium current with an activation voltage close to the resting membrane potential (Fig. 7C): decreasing excitability (f-I slope), decreasing resting membrane potential (Vm), and increasing rheobase. Although the change in input resistance (Rm) was not statistically significant, the trend was also in a direction expected from increased resting conductance (and therefore lower Rm). Changes in duration were not accompanied by what is typically seen during rundown (increasing Vm) or closing of the opening from the recording pipette into the cell (increasing Rm).
Discussion
These results demonstrate that neurons in the caudal mesopallium exhibit rapid, activity-dependent plasticity of intrinsic dynamics when zebra finches are near the beginning of the sensitive period for song memorization (P30-P35; P stands for days post-hatch). This plasticity requires early exposure to an acoustically complex environment, which causes elevated expression of the low-threshold potassium channel Kv1.1 in the endoplasmic reticulum and plasma membrane (Fig. 1). Plasticity is primarily observed in tonic-spiking neurons: suprathreshold stimulation for more than a few minutes causes a shift to phasic spiking (Fig. 3) that is reversed by blocking low-threshold potassium currents (Fig. 5) and prevented by blocking intracellular calcium signaling (Fig. 6). In constrast, phasic neurons almost never become tonic. Spiking patterns are not plastic in neurons from pair-reared birds (Fig. 4), which have much lower levels of Kv1.1 expression. Taken together, these findings support a model in which spiking triggers calcium-dependent trafficking of Kv1.1 from intracellular compartments to the plasma membrane. This rapid form of plasticity could facilitate early auditory development by allowing neurons in this higher-order brain region to nimbly adjust their excitability and temporal integration properties in response to the challenging “cocktail-party” acoustic background of the colony.
The present study sheds light on the mechanisms underlying the diverse intrinsic dynamics in CM (Chen and Meliza, 2018) and their dependence on age and experience (Chen and Meliza, 2020). Phasic spiking appears to require both a slow process of Kv1.1 synthesis and a rapid process of Kv1.1 mobilization, as illustrated in Figure 8. The slow process begins around the time zebra finches fledge from the nest and reflects the environment the chicks experience over the next two weeks. Compared with pair-reared chicks, colony-reared chicks experience a louder and more complex acoustic environment, which may result in increased activity-dependent Kv1.1 expression, as seen in the avian auditory hindbrain (Lu et al., 2004; Akter et al., 2018). Nonauditory and neuromodulatory factors could also contribute, as CR birds would be able to engage in more visual and vocal interactions with other animals in the colony. By the time CR birds reach P30–P35, almost all of the excitatory neurons express Kv1.1, but only some of these exhibit phasic spiking at the start of recording, and there is no correlation between intracellular Kv1.1 expression and spiking dynamics (Fig. 2). However, many of the tonic neurons are plastic, with the potential to become phasic when they are stimulated, suggesting that the intracellular Kv1.1 is a ready-to-use pool that can be rapidly mobilized to the plasma membrane to alter intrinsic dynamics.
Model of slow and fast intrinsic plasticity. When zebra finches fledge (P18–P20), CM neurons have low levels of Kv1.1 expression and only produce tonic responses. Exposure to a complex environment stimulates Kv1.1 expression in CR birds, but neurons remain tonic so long as the Kv1.1 remains within the neuron. Acute stimulation that elevates intracellular calcium causes Kv1.1 to be exported to the plasma membrane, rapidly shifting the dynamics to phasic spiking. Kv1.1 expression remains low in PR animals, so their neurons are not able to become phasic. An unknown process may allow phasic neurons to internalize Kv1.1 and shift back to tonic spiking.
Although changes in intrinsic dynamics can be the consequence of nonspecific rundown, all of the evidence points to a specific mechanism involving a low-threshold potassium current that is likely mediated at least in part by Kv1.1. First, we did not observe plasticity in PR birds, which have much lower levels of Kv1.1 expression. Second, plasticity was reversed by blocking low-threshold potassium channels with α-dendrotoxin or with 4-AP. Third, plasticity required a minimum level of stimulation, typically occurring only after at least 120 spikes. Finally, changes in duration were not associated with changes in passive membrane properties that typically accompany degraded recordings. Further evidence could be obtained with more direct measures of Kv1.1 trafficking and low-threshold potassium currents (e.g., with live microscopy or voltage-clamp recordings), but these experiments would likely require using dissociated cell culture to obtain better optical and electrical access.
The time course of the plasticity and the abundance of Kv1.1 in the ER suggest that Kv1.1 is being mobilized by trafficking to the plasma membrane. Kv1.1 is retained in the ER by binding of the external pore region of the channel to matrix metalloprotease 23 (Rangaraju et al., 2010; Vacher and Trimmer, 2012), which can be cleaved by the calcium-dependent endoprotease furin (Galea et al., 2014). As proposed by Adachi et al. (2019), activity-dependent elevation of intracellular calcium could thereby trigger the release of Kv1.1 from the ER to the plasma membrane. This model is consistent with our observation that pretreating neurons with the calcium chelator BAPTA-AM blocks plasticity (Fig. 6). Intracellular calcium could also regulate Kv1.1 by modifying the phosphorylation state of channels already localized to the membrane, although this explanation is less likely, because calcium-dependent kinases typically downregulate Kv1 conductance (Hoffman and Johnston, 1998) or have only long-lasting effects on expression (Winklhofer et al., 2003).
These results strongly implicate Kv1.1 but do not rule out contributions from other channels. Changes in firing duration were reversed completely by α-DTX, a specific blocker of Kv1 channels (Guan et al., 2007), but the magnitude of the reversal was slightly greater with 4-AP, which blocks a much broader range of voltage-gated potassium channels. This implies that changes in duration may involve 4-AP-sensitive channels outside the Kv1 family. Changes in excitability, measured as a decrease in the f-I slope, were not reversed by α-DTX or 4-AP. We also often observed a shift in the current–voltage relationship at hyperpolarized potentials (Figs. 3E, 4E), which could be caused by increased Ih currents (Fan et al., 2005; Ross et al., 2019). Thus, multiple mechanisms may contribute to the maintenance of cellular and network homeostasis.
Activity-dependent intrinsic plasticity has been observed widely in the brain, involving a diverse range of ion channels and mechanisms for modulating their conductances (Zhang and Linden, 2003; Titley et al., 2017; Debanne et al., 2019; Daou and Margoliash, 2021). Plasticity of the low-threshold (D-type) potassium current mediated by Kv1.1 and other members of the Kv1 family has been reported in the cortex (Dehorter et al., 2015; Gainey et al., 2018) and avian nucleus magnocellularis (Lu et al., 2004; Adachi et al., 2019), and also in the hippocampus, where it can occur on the fast time scales we see in CM (Morgan et al., 2019). Whereas hippocampal neurons require intense stimulation to trigger intrinsic plasticity, CM neurons are much more sensitive, requiring as few as 100 spikes over 400 s. Given that we almost never observed phasic cells becoming tonic, this raises the question of why only about half of the neurons in CM are phasic at this age in CR birds, as we would expect in vivo spontaneous and auditory-evoked activity to have driven all the plastic neurons toward phasic spiking before we cut slices. It seems likely that there is another mechanism of plasticity that can induce phasic-spiking neurons to become more tonic. This putative mechanism could involve sustained inhibitory or metabotropic inputs (Cudmore et al., 2010; Campanac et al., 2013) that were not active in slices. Maturation of these inputs during the closing of the critical period for sensory learning (London, 2019) or more general development of the auditory cortex could explain why phasic spiking essentially disappears in CM by P65 (Chen and Meliza, 2020). It is also possible that the tonic, plastic neurons we recorded were comparatively silent in vivo up until the time when we cut slices. This could occur if these neurons were not yet strongly integrated into the developing circuitry of CM and the broader auditory system, and indeed intrinsic plasticity may be an essential mechanism for maintaining network homeostasis as connectivity develops. Given the critical importance of this age to the development of vocal production and perception in these animals, it is perhaps unsurprising that neurons in the auditory cortex can so rapidly and sensitively change their intrinsic dynamics.
Footnotes
This work was supported by the National Institutes of Health Grant 1R01-DC018621, the National Science Foundation Grant IOS-1942480, and the University of Virginia Brain Institute. Mark Beenhakker provided helpful feedback on the manuscript.
The authors declare no competing financial interests.
- Correspondence should be addressed to C. Daniel Meliza at cdm8j{at}virginia.edu
