Functional and structural properties of ion channels at the nerve terminal depends on compact myelin

Abstract

Key points

• In the present study, we document the role of compact myelin in regulating the structural and functional properties of ion channels at the nerve terminals, using electrophysiology, dynamic Na⁺ imaging and immunohistochemistry.

• The subcellular segregation of Na⁺ channel expression and intracellular Na⁺ dynamics at the heminode and terminal was lost in the dysmyelinated axon from Long–Evans shaker rats, which lack compact myelin.

• In Long–Evans shaker rats, loss of the Na_vβ4 subunit specifically at the heminode reduced resurgent and persistent Na⁺ currents, whereas K⁺ channel expression and currents were increased.

• The results of the present study suggest that there is a specific role for compact myelin in dictating protein expression and function at the axon heminode and in regulating excitability of the nerve terminal.

Abstract

Axon myelination increases the conduction velocity and precision of action potential propagation. Although the negative effects of demyelination are generally attributed to conduction failure, accumulating evidence suggests that myelination also regulates the structural properties and molecular composition of the axonal membrane. In the present study, we investigated how myelination affects ion channel expression and function, particularly at the last axon heminode before the nerve terminal, which regulates the presynaptic excitability of the nerve terminal. We compared the structure and physiology of normal axons and those of the Long–Evans shaker (LES) rat, which lacks compact myelin. The normal segregation of Na⁺ channel expression and dynamics at the heminode and terminal was lost in the LES rat. Specifically, Na_V‐α subunits were dispersed and Na_Vβ4 subunit was absent, whereas the density of K⁺ channels was increased at the heminode. Correspondingly, resurgent and persistent Na⁺ currents were reduced and K⁺ current was increased. Taken together, these data suggest a specific role for compact myelin in the orchestration of ion channel expression and function at the axon heminode and in regulating excitability of the nerve terminal.

Key points

• In the present study, we document the role of compact myelin in regulating the structural and functional properties of ion channels at the nerve terminals, using electrophysiology, dynamic Na⁺ imaging and immunohistochemistry.

• The subcellular segregation of Na⁺ channel expression and intracellular Na⁺ dynamics at the heminode and terminal was lost in the dysmyelinated axon from Long–Evans shaker rats, which lack compact myelin.

• In Long–Evans shaker rats, loss of the Na_vβ4 subunit specifically at the heminode reduced resurgent and persistent Na⁺ currents, whereas K⁺ channel expression and currents were increased.

• The results of the present study suggest that there is a specific role for compact myelin in dictating protein expression and function at the axon heminode and in regulating excitability of the nerve terminal.

Abbreviations

aCSF

artificial cerebrospinal fluid

AIS

the axon initial segment

AP

action potential

HVA

high‐voltage activated

LE

Long–Evans rats

LES

Long–Evans shaker

MBP

myelin basic protein

MNTB

medial nucleus of the trapezoid body

Na_v

voltage‐activated Na⁺ channel

PBS

phosphate‐buffered saline

ROI

region of interest

TEA

tetraethylammonium chloride

UTHSCSA

University of Texas Health Science Center, San Antonio

WT

wild‐type

Introduction

A high density of voltage‐activated Na⁺ channels (Na_v) clustered at discrete subcellular regions along the axon is necessary for the reliability and temporal fidelity of action potential (AP) firing. In addition to its role as an electrical insulator, the myelin sheath contributes to ion channel clustering. A recent study described alterations in the structure of the node of Ranvier and the relocation of Na_v clusters at the axon initial segment (AIS) during demyelination (Hamada and Kole, 2015). These structural changes in axons following demyelination affect neuronal excitability, AP propagation, and neurotransmission.

The final destination of an AP is the nerve terminal where neurotransmitter release occurs. One of the critical factors for presynaptic excitability is the type, density and location of Na_v channels (Engel and Jonas, 2005, Leão et al. 2005, Kawaguchi and Sakaba, 2015). Specifically, transient, persistent and resurgent Na⁺ currents, originating from the axon heminode, are required for the reliability of APs firing at high frequency and proper high fidelity signalling (Leão et al. 2005, Huang et al. 2008, Kim et al. 2010). In previous studies, we found that dysmyelination of auditory afferent axons reduces the reliability of presynaptic firing and temporal fidelity of synaptic transmission during high‐frequency firing in the calyx of Held synapse (Kim et al. 2013 a; Kim et al. 2013 b). These observations raise the possibility that loss of compact myelin may alter the pattern of ion channel expression at the last heminode and in the nerve terminal. The role of myelin on the structural and functional properties of the last axon heminode is an essential part of a complete understanding of the consequence of myelin loss on presynaptic excitability and neurotransmission. Currently, the extent to which channel proteins underlying accurate signalling and high‐fidelity performance are disrupted in dysmyelinated nerve terminal is largely unknown. Here, we tested the idea that myelin is an active participant in setting the structural and functional properties at the last heminode, where propagated APs are finely tuned before invading the terminal.

To examine the impact of lack of compact myelin on the axon heminode and terminal, we took advantage of the calyx of Held in the auditory brainstem that allows direct access to the heminode and terminal for electrophysiology and live cell imaging. Na_v channels form clusters at nodes of Ranvier and the last heminode adjacent to the calyx terminal, but not within the presynaptic terminal (Leão et al. 2005). This distribution of Na_v channels may contribute to low membrane excitability of the terminal and to the short duration of presynaptic APs (Alle and Geiger, 2006, Kawaguchi and Sakaba, 2015). Previously, we found electrophysiological abnormalities in presynaptic firing and synaptic transmission at the calyx of Held in the Long Evans shaker (LES) rat (Kim et al. 2013 a,b). The LES rat has a genetic deletion of myelin basic protein (MBP) and fails to condense the myelin sheath (Kwiecien et al. 1998, O´Conner et al. 1999). Axons from young LES rats before postnatal 4 weeks are loosely covered by 2–3 thin layers of myelin, but then undergo progressive demyelination until, by three months of age, most axons are demyelinated (Smith et al. 2013). Using the LES rat, we investigated how myelin loss impacts the structural and physiological properties of ion channels at the heminode and the calyx of Held terminal, using presynaptic Na⁺ imaging, Na⁺ and K⁺ current recording, and immunostaining. We found that lack of compact myelin specifically impacts expression of Na_v and K_v channels at the axon heminodes and consequently alters subcellular Na⁺ dynamics at heminodes and terminals. The loss of Na_vβ4 subunit at the last heminode underlies the subcellular specific disruption of Na_v. These results suggest that myelination is essential for proper location of Na_vβ4 subunit and the clustering of Na⁺ and K⁺ channels at axon heminodes and terminals, to maintain presynaptic excitability at nerve terminals in the auditory nervous system.

Methods

Ethical approval

Long–Evans rats (LE; wild‐type) and Long‐Evans Shaker rats (LES; from Dr Kwiecien, McMaster University, Ontario, Canada) of both sexes between the ages of postnatal day 13 (P13) and P17 were used for electrophysiological recordings and Na⁺ imaging, and at P20–P25 or P50 for immunohistochemistry. All procedures followed the approved animal care protocols by the University of Texas Health Science Center, San Antonio (UTHSCSA) Institutional Animal Care and Use Committee. All authors understood the ethical principles that The Journal of Physiology operates under and the work complied with the animal ethics checklist reported by Grundy (2015).

Animals were maintained as a heterozygous stock. A fraction of offspring from the heterozygous animals exhibited the LES phenotype, a distinguishable tremor, starting ∼2 weeks following birth (Delaney et al. 1995; O'Connor et al. 1999; Kim et al. 2013). The LES phenotype is autosomal recessive, present only in animals that are homozygous for the MBP mutation, whereas heterozygous and wild‐type (WT) animals show no signs of locomotion disturbance (Kwiecien et al. 1998; O'Connor et al. 1999; Kwiecien 2010; Smith et al. 2013, Kim et al. 2013 a,b). In the present study, we used WT LE (mbp^+/+) and LES (homozygous mutant, mbp^–/–) rats for all experiments.

Slice preparation

Transverse brainstem slices were prepared from rat pups of P13–P17. All procedures followed approved UTHSCSA animal care protocols. After rapid decapitation following isoflurane anaesthesia, the brainstem was quickly removed from the skull and immersed in ice‐cold low‐Ca²⁺ artificial cerebrospinal fluid (aCSF) containing (in mm): 125 NaCl, 2.5 KCl, 3 MgCl₂, 0.1 CaCl₂, 25 glucose, 25 NaHCO₃, 1.25 NaH₂PO₄, 0.4 ascorbic acid, 3 myo‐inositol and 2 Na‐pyruvate, pH 7.3–7.4 when bubbled with carbogen (95% O₂, 5% CO₂; osmolarity of 310–320 mosmol l^–1). The brainstem was dissected and secured in a chamber with glue, and 200 μm (for electrophysiology and Na⁺ imaging) or 100 μm (for immunostaining) sections were cut with a vibratome (VT1200S; Leica, Bannockburn, IL, USA). Slices were then transferred to an incubation chamber containing normal aCSF bubbled with carbogen and maintained at 35°C for 30 min and thereafter at room temperature. Normal aCSF was the same as the slicing aCSF, except that it contained 1 mm MgCl₂ and 2 mm CaCl₂.

Electrophysiology

Slices were perfused at 2 ml min⁻¹ and visualized using an infrared differential interference contrast microscope (AxoExaminer; Zeiss, Oberkochen Germany) with a 60× water‐immersion objective and a CMOS camera (ORCA‐Flash2.8; Hamamatsu, Hamamatsu City, Japan). Whole‐cell patch clamp recordings were performed in normal aCSF at room temperature (22–24°C) using an EPC‐10 amplifier controlled by Patch master software (HEKA, Elektronik, Lambrecht/Pfalz, Germany). For presynaptic current clamp recordings, the pipette solution contained (in mm): 125 K‐gluconate, 20 KCl, 5 Na₂‐phosphocreatine, 10 Hepes, 4 Mg‐ATP, 0.2 EGTA and 0.3 GTP, pH adjusted to 7.3 with KOH. Presynaptic Na⁺ current recordings used a pipette solution containing (in mm): 130 CsCl, 10 tetraethylammonium chloride (TEA)‐Cl, 5 Na₂‐phosphocreatine, 10 Hepes, 4 Mg‐ATP, 10 EGTA and 0.3 GTP, pH adjusted to 7.3 with KOH, and the external solution contained 2 mm 4‐aminopyridine, 10 mm TEA‐Cl and 0.2 mm CdCl₂ to block presynaptic K⁺ and Ca²⁺ channels. Recordings were not corrected for the predicted liquid junction potential of 4 mV (CsCl‐based internal solution) or 11 mV (K‐gluconate‐based internal solution). Patch electrodes had resistances of 3–4 MΩ. For voltage clamp experiments, series resistance was <15 MΩ and was compensated 80%. Current clamp recordings were continued only if the initial uncompensated series resistance was <20 MΩ (Kim et al. 2013). The presynaptic axons of the calyx of Held terminal were stimulated with a bipolar platinum‐iridium electrode (Frederick Haer, Bowdoinham, ME, USA) placed near the midline spanning the afferent fibre tract of the medial nucleus of the trapezoid body (MNTB). An Iso‐Flex stimulator driven by a Master 10 pulse generator (AMPI, Jerusalem, Israel) delivered 100 μs pulses at 1.2 times the threshold (<15 V constant voltage). Signals were filtered at 2.9 kHz and acquired at a sampling rate of 10–50 μs.

Presynaptic Na⁺ imaging

The calyx of Held terminals were whole‐cell patch clamped using a pipette containing K‐gluconate based solution (see above) supplemented with 100 μm CoroNa Green (a Na⁺ indicator) and 50 μm Alexa 568 (Invitrogen, Carlsbad, CA, USA). We used an LED illumination system (Calibri; Zeiss) to excite indicators at 488 and 555 nm. Na⁺ imaging was performed with simultaneous whole‐cell current clamp recordings at the calyx terminals. Presynaptic APs were generated at 180 Hz for 1.5 or 3 s. The use of Alexa 568 permitted us to image cellular structures along the axon without bleaching the Na⁺‐sensitive dye. Na⁺ transients were recorded from the axon heminodes and presynaptic terminals. Fluorescence signals were acquired at 10–15 Hz and analysed using Zen Pro software (Zeiss). For analysis of CoroNa signals, regions of interest (ROI) within the axon heminode or the nerve terminal were drawn manually. Background fluorescence was measured from an area outside the cell, and all data are presented after background subtraction. All values are expressed as ΔF/F, where ΔF is the change in fluorescence during or after stimulation and F is the basal fluorescence of the ROI measured before stimulation. Measurements of intracellular Na⁺ concentration were calibrated by perfusing the slices with aCSF containing different concentrations of Na⁺ (ranging from 5 to 125 mm) in the presence of the Na⁺ ionophore monesin (50 μm) and ouabain (100 μm) to block the Na⁺ pump (Meier et al. 2006; Huang and Trussel, 2014). When NaCl was reduced, it was replaced by equimolar NMDG‐Cl to maintain osmolality.

Immunohistochemistry

Rat brainstem slices (100 μm) were fixed with 4% (w/v) paraformaldehyde in phosphate‐buffered saline (PBS) for 30 min. Free‐floating sections were blocked in 4% goat serum and 0.3% Triton X‐100 in PBS for 1 h. For immunohistochemistry, slices were incubated with the primary antibody overnight at 4°C. The primary antibodies used were: mouse anti‐sodium channel (Pan; dilution 1:200; Sigma, St Louis, MO, USA), mouse anti‐K_v3.1 (dilution 1:50; Neuromab, Davis, CA, USA), mouse anti‐Kv1.2 (dilution 1:200; Neuromab), rabbit anti‐Na_v1.6 (dilution 1:100; Alomone Labs, Jerusalem, Israel), mouse anti‐Na_v β4 (dilution 1:50; Neuromab), rabbit anti‐calretinin (dilution 1:200; Invitrogen), guinea pig anti‐Caspr (dilution 1:1000; from Dr Manzoor Bhat, UTHSCSA) and guinea pig polyclonal anti‐vesicular glutamate transporter 1 (vGluT1; dilution 1:1000; Millipore, Billerica, MA, USA). Antibody labelling was reported by incubation with appropriate Alexa dye‐conjugated secondary antibodies (dilution 1:500; Invitrogen) for 2 h at room temperature. After five rinses with PBS, coverslips were mounted onto slides using mounting medium (Fluoroshield; Sigma). Stained slices were imaged with laser lines at 488, 568 and 647 nm using an 60× oil‐immersion objective on a confocal laser scanning microscope (IX81 Fluoview 1000; Olympus, Tokyo, Japan).

Statistical analysis

Data are reported as the mean ± SEM. Each result established from immunostaining data was based on an analysis of at least 20 cells in 10 slices from three animals. Experimental data were analysed and presented using Igor Pro (Wavemetrics, Lake Oswego, OR, USA). Statistical significance was determined using Student's t test (Excel; Microsoft Corp., Redmond, WA, USA). P < 0.05 was considered statistically significant.

Results

AP‐evoked Na⁺ transients at the normal myelinated heminode and calyx nerve terminal

To investigate AP‐evoked intracellular Na⁺ transients in the axon heminode and nerve terminal, we recorded intracellular Na⁺ dynamics from the calyx of Held terminal and axon using the Na⁺‐sensitive dye CoroNa green during whole‐cell patch clamp recordings. Brief injections of current generated AP trains (180 Hz, 1.5 or 3 s in duration) and increased CoroNa green fluorescence at the axon heminode and at the presynaptic terminal (Fig. 1 A and B).

Figure 1. Na⁺ transients during action potentials in the calyx of Held nerve terminal .

A, differential interference contrast and pseudocolour fluorescence images of the calyx of Held axon and terminal in the MNTB of the WT rat (P15) loaded with the Na⁺ indicator, CoroNa‐green (100 μm), during whole‐cell recording. Ovals indicate subcellular regions of interest [axon heminode (A, black) and terminal (T, grey)] from which fluorescence measurements were obtained. B, intracellular Na⁺ transients, measured as relative changes in CoroNa‐green fluorescence (ΔF/F), elicited by AP trains (180 Hz, 1.5 s and 3 s) recorded at the axon heminode (A, black trace) and presynaptic terminal (T1 and T2, grey traces). Note the rapid rise in Na⁺ at the axon heminode (A) and the slower and smaller rise at the presynaptic terminal (T1 and T2). Inset, representative trace of AP train at 180 Hz (1.5 s) and a scaled trace of the presynaptic terminal (grey) to the peak of signal from the axon heminode (black) to compare the decay time course of Na⁺ transient at the heminode (black, τ = 12 s) and terminal (grey, τ = 30 s). C, summary of the intensity (ΔF/F) and decay time constant (τ) of Na⁺ transients elicited by AP trains for (3 s, 180 Hz) at heminodes and terminals in the individual myelinated axon. Broken lines join paired data points obtained from the same axon. Data were analysed using a paired t test: ^*** P < 0.0001 and ^* P < 0.05, respectively. D, differential interference contrast and fluorescence images of the calyx of Held terminal, loaded with a Na⁺ indicator, CoroNa‐green (100 μm), using whole‐cell recordings. The black circle indicates the measuring area at presynaptic terminal and the white circle indicates the background area. Na⁺ calibration curve relating the change in fluorescence intensity as function of the applied Na⁺ concentrations ([Na⁺]): Fitting curve (red) was obtained using the Hill equation. [Colour figure can be viewed at wileyonlinelibrary.com]

Despite the smaller surface to volume ratio of the axon compared to the nerve terminal, the increase in CoroNa green fluorescence following an AP train (180 Hz, 3 s) was significantly higher at the axon heminode (ΔF⁄F = 15 ± 1.2%) than at the presynaptic terminal in the same axons (ΔF⁄F = 10 ± 1.1%) (n = 15, paired t test, P < 0.0001) (Fig. 1 B and C). Based on a calibration curve determined in the presence of monensin (50 μm; a Na⁺ ionophore) and ouabain (100 μm; to block the Na⁺ pump), the changes in CoroNa fluorescence correspond to intracellular Na⁺ concentration increases of ∼50 mm and 30 mm at the heminode and terminal, respectively (Fig. 1 D). No increases in fluorescence were observed when Na⁺ channels were blocked with TTX (1 μm; data not shown). In addition, the rise time of the fluorescence increase was faster at the heminode (t _10%–90% = 2.6 ± 0.17 s) than at the terminal (t _10%–90% = 3.7 ± 0.24 s), as was the exponential decay time constant (τ  = 6.4 ± 1.2 s in the heminode vs. 22.3 ± 4.5 s in the terminal, n = 5, P = 0.02, paired t test) (Fig. 1 B and C). The earlier, steeper and faster Na⁺ transient recorded at the heminode, together with the fact that the rise time of the transient recorded in the terminal is longer than the AP train, suggests that, in the normal myelinated axon terminal, most of the Na⁺ currents enters at the heminode and then diffuse to the presynaptic terminal.

Lack of condensed myelination alters AP‐evoked Na⁺ transients at axon heminodes and presynaptic terminals

To examine the effect of the loss of condensed myelination on Na⁺ dynamics, we repeated the same experiments in the LES rat, a myelin‐deficient model (Fig. 2 A). By contrast to the normal axon, Na⁺ transients following APs in LES axons were significantly larger in the presynaptic terminal than at the heminode (Fig. 2 B and C). AP trains (180 Hz, 3 s) elevated intracellular Na⁺ (ΔF⁄F) by 10 ± 1.5% at the heminode and 13 ± 1.7% at the presynaptic terminal of the same axon (n = 11, paired t test, P = 0.04). The decay kinetics of Na⁺ transients along dysmyelinated axons were more variable than for the WT. Unlike that observed in the WT, there were no significant differences in the decay time constant of Na⁺ transients at axon heminodes and at presynaptic terminals, which were 38.6 ± 5.8 s and 44.4 ± 6.4 s, respectively (n = 9, P = 0.41, paired t test) (Fig. 2 C). Thus, in the LES, Na⁺ appears to enter simultaneously at both the heminode and the terminal, and the evidence for intercellular diffusion of Na⁺ described above for the normal axon terminal was not apparent in the dysmyelinated axon.

Figure 2. Lack of myelin alters the kinetics of Na⁺ transients at axon heminodes and terminals from LES rats .

A, differential interference contrast and fluorescence images of the LES calyx axon and terminal (P16), which was loaded with CoroNa‐green (100 μm) during whole‐cell recordings. Ovals indicate heminode and terminal regions of interest in the dysmyelinated calyx axon from which fluorescence measurements were obtained. B, representative trace of Na⁺ transient (ΔF/F) in response to AP trains (180 Hz, 3 s) at the axon heminode (black) and presynaptic terminal (grey) in the LES rat. C, summary of changes in the intensity (ΔF/F) and decay time constant (τ) of Na⁺ transients at heminode and terminal in the same dysmyelinated axon from LES rats. Broken lines link data from the axon heminode and terminal of the same identified cell. Paired t test: ^* P < 0.05 and non‐significant (NS), respectively. D, comparison of Na⁺ transient (ΔF/F) in response to AP trains (180 Hz, 3 s) at axon heminodes and terminals from WT and LES rats. E, comparison of decay time constant (τ) of Na⁺ transients elicited by AP trains at axon heminodes and terminals from WT and LES rats. ^*** P < 0.0001, ^** P < 0.001 and ^* P < 0.05, respectively. [Colour figure can be viewed at wileyonlinelibrary.com]

The differences in Na⁺ dynamics between WT and LES axons are summarized in Fig. 2(D) and (E). At axon heminodes, the increase in [Na⁺]_i following APs was significantly higher in the WT than the LES (ΔF⁄F = 15 ± 1.2% and 10 ± 1.5%, n = 15 and 11, P = 0.03, unpaired t test). At presynaptic terminals, however, there was no significant difference in AP‐triggered increase in [Na⁺]_i between myelinated and dysmyelinated axon terminals (ΔF⁄F = 10 ± 1.1% and 13 ± 1.7%, n = 15 and 11, P = 0.23, unpaired t test) (Fig. 2 D). Compared to myelinated axons, the decay of the Na⁺ transient was much slower at both axon heminodes and terminals in LES rats (n = 8) (Fig. 2 E). At heminodes, the decay time constant of the Na⁺ transients was 6.4 ± 1.2 s (n = 11) vs. 38.6 ± 5.8 s (n = 9) for WT and LES, respectively (P < 0.001), whereas, at the terminal, time constants were 22.3 ± 4.5 s (n = 7) and 44.4 ± 6.4 s (n = 9), respectively (P = 0.018). The level of myelination thus controls the intensity and kinetics of Na⁺ dynamics at axon heminodes and presynaptic terminals.

Loss of compact myelination disrupts Na⁺ channel clusters specifically at the last heminode next to the calyx terminal

Because Na⁺ influx via Na⁺ channels is one of the main sources for Na⁺ transients at the heminode and terminal, we examined the expression of Na⁺ channels by immunofluorescence microscopy of the AIS, nodes of Ranvier and the last heminode before the terminal (Fig. 3). The size of the Na_v cluster in the AIS was similar in WT and LES axons for both the cochlear nucleus globular bushy cells (the projections of which form the calyx of held nerve terminal) and MNTB neurons (data not shown). Surprisingly, a lack of compact myelin did not change the immunohistochemical staining pattern of Na_v cluster at nodes of Ranvier, nor was there any difference in the distance between nodes in the LES rat. The integrity of Na_V channel pattering at nodes of Ranvier was confirmed in LES rats at age P50 (data not shown). The major alteration observed at nodes of Ranvier concerned the pattern of expression of Caspr, a prominent component of the paranodal axon glial junctions. Normal calyx axons, which were labelled with calretinin (a Ca²⁺‐binding protein expressed in nerve fibres), presented strong and well‐defined labelling for Caspr at both edges of the Na_v channel cluster, whereas Caspr labelling was weak and disperse in nodes of Ranvier from the LES (Fig. 3 A and B). Despite the disruption of paranodal structure in LES axons, the length of the Na_v cluster at nodes of Ranvier and the distance between nodes was not significantly different in WT and LES axons (Fig. 3 B). The lengths of nodes were 2.2 ± 0.08 μm (n = 54) in WT and 2.0 ± 0.09 μm (n = 55) in LES rats (unpaired t test, P = 0.27) and the internodal distances were 56 ± 3.8 μm in WT and 49± 2.7 μm in LES (n = 14 and 8, P = 0.248, unpaired t test) (Fig. 3 D and G).

Figure 3. Myelin loss specifically disrupts Na⁺ channels at the axon terminal heminode .

A, expression of Na_vPan (green), Caspr (red) and calretinin (CR, blue) immunoreactivity along the calyx of Held axon and terminal in the MNTB in the WT rat. The dashed lines delimit nodal and heminodal Nav cluster, as well as the calyx of Held. Blue horizontal arrows indicate the distance between the heminode and the next node, and the internodal distance, respectively. B, expression of Caspr (red) and Na_vPan (green) at axon nodes in the MNTB from WT and LES rats. Arrows indicate Na_vPan at nodes. C, expression of Na_vPan (green) and vGluT1 (red) at axon heminodes and calyx terminals. Arrows indicate the end of Na_vPan signals for measuring the length of Na⁺ channel cluster. D and E, summary of the length of Na⁺ channel cluster at nodes of Ranvier and axon heminodes. F and G, summary of the distance between the axon heminode and the next node, and between two nodes (internodal distance). ^*** P < 0.0001.

The last heminode separates the myelinated axon from the non‐mylinated nerve terminal. By contrast to the relatively normal clustering of Na_v channels at the AIS and nodes of Ranvier, the expression of Na_v channels was markedly altered in the last heminode and the calyx nerve terminal of the LES axon. In WT axons, Na_v channels clustered in the heminode, which was located 14.1 ± 7.1 μm away from the neck of the calyx terminal (n = 28) (Fig. 3 A and C). In LES rats, the heminodal Na_v cluster was much longer than in WT (19.7 ± 7.1 μm, n = 19, in the LES vs. 6.7 ± 2.2 μm, n = 17, in WT; P < 0.0001 unpaired t test) (Fig. 3 E). As a result of the dispersed Na_v cluster at the heminode, the distance between the heminode and the next node was reduced in LES compared to WT (28 ± 4.5 μm vs. 44± 2.8 μm, n = 23 and 14, P < 0.001, unpaired t test) (Fig. 3 F). In addition, unlike for WT axons, in the LES axon, Na_v channels invaded the calyx nerve terminal. This result indicates that compact myelination is essential for the expression pattern of Na_v channels at the heminode adjacent to the presynaptic terminal.

These experiments were repeated using specific antibodies for Na_v1.2 and Na_v1.6 channel isoforms to test the possibility of differential or compensatory expression of two different Na_v isoforms (Boiko et al. 2001, Vega et al. 2008). Expression of Na_v1.6 followed the pattern described above both in WT and LES axons (Fig. 4 A), whereas Na_v1.2 was not detected (data not shown), suggesting that the main subtype of Na_v channels at axon heminodes and nodes is Na_v1.6.

Figure 4. Dispersed Na_v1.6 channel expression extends into the presynaptic terminal .

A, heminodal staining at calyx of Held presynaptic terminal with Na_v1.6 (green) and vGluT (red) in WT and LES rats (left). Nodal and paranodal staining with Na_v1.6 (green, yellow arrows) and Caspr (red) in WT and LES rats (right). B, representative traces of the outside‐out patch recordings from the calyx terminals in WT and LES rats. Voltage‐activated Na⁺ and K⁺ currents were evoked by step depolarization (from −100 mV to 50 mV; in the presence of TEA 10 mm and 4‐AP 2 mm) in the outside‐out patches. Inset, expanded scale of Na⁺ currents in the boxed area from the LES recording. C, summary of the proportion of the presynaptic terminals displaying detectable I _Na in WT and LES rats in the outside‐out patch recordings.

The immunostaining data described above demonstrate that the Na_v channel expression pattern extends into the presynaptic terminals in the LES axon, whereas, in WT axons, Na_v channels are restricted to a cluster located 10–20 μm away from the terminal. We thus tested whether functional Na⁺ currents can be detected at presynaptic terminals using outside–out patch recording from the calyx terminal (Fig. 4 B). In WT rats, we rarely detected Na⁺ currents from outside‐out patch (one in 10 patches). By contrast, four of 10 patches from LES calyx terminals expressed measurable Na⁺ currents (−147 ± 16.8 pA at 0 mV, n = 4). Together with the immunostaining, our Na⁺ current recordings suggest that the lack of compact myelination disrupted the Na_v cluster at the axon heminode and allowed Na_v channels to invade the presynaptic terminal.

Dysmyelination reduces resurgent and persistent Na⁺ currents at the calyx of Held terminals

The detection of Na⁺ currents using outside‐out patches from LES terminals indicates the changes in the distribution of Na_v channels, rather than an increase in the total whole‐cell Na⁺ currents at the LES heminode and tsierminal. Consistent with this idea, the Na⁺ imaging experiments and dispersed expression of Na_v channels described above suggest smaller Na⁺ currents in the LES axon heminode. We tested whether alteration in Na_v channel distribution affects Na⁺ currents under voltage clamp recordings in the presence of K⁺ and Ca²⁺ channel blockers (10 mm TEA, 2 mm 4‐AP and 0.2 mm CdCl₂). A step‐depolarization from the holding potential of −80 mV to +30 mV induced large and rapidly inactivating transient Na⁺ currents with amplitudes of −4.2 ± 0.48 nA (n = 4) for WT and −4.1 ± 0.61 nA for the LES (n = 5, P > 0.05). Upon repolarization to potentials ranging from +20 to −70 mV, a voltage‐ and time‐dependent resurgent Na⁺ current (I _NaR) was present (Fig. 5 A) (Raman and Bean, 1997, Kim et al. 2010, Lewis and Raman, 2014). The peak of resurgent Na⁺ current, measured at −40 mV, was significantly decreased in LES calyx terminals (WT: –310 ± 78.1 pA vs. LES: –61 ± 16.5 pA, n = 4 and 5, P < 0.001) (Fig. 5 B and C). We next examined persistent Na⁺ current (I _NaP) using voltage ramps of 100 mV s⁻¹ (from −90 mV to +50 mV) in the calyx of Held in WT and LES rats. During the voltage ramps, the fast transient current was fully inactivated to isolate I _NaP, which was completely blocked by 1 μm TTX (data not shown). By subtracting a trace with 1 μm TTX from a control trace, the current–voltage relationship for I _NaP was determined, revealing peak values of −376 ± 19.1 pA in WT calyces and −170 ± 26.8 pA in LES calyces (n = 4 and 6, P < 0.05) (Fig. 5 D). Thus, I _NaR and I _NaP are significantly reduced in the axon terminals from LES rats.

Figure 5. Reduced resurgent and persistent Na⁺ currents at axon terminal in the LES rat .

A and B, representative traces of resurgent Na⁺ currents evoked by step repolarization (from +20 mV to −70 mV) after a brief depolarization (+30 mV, 5 ms) to generate transient Na⁺ currents (I_NaT) at the calyx of Held terminals in WT (A) and LES rats (B). C, current–voltage (I–V) relationship of resurgent Na⁺ currents form WT (black) and LES rats (red). D, persistent Na⁺ currents were recorded by ramp depolarization from −90 mV to +40 mV (100 mV/s) at the calyx of Held terminals in WT (black) and LES rats (red). Representative traces of persistent Na⁺ currents after TTX subtraction. E, in current clamp recordings, TTX (1 μm) hyperpolarized the calyx terminal in the MNTB from WT but not LES rats. F, summary of changes in calyx membrane resting potential (ΔV _m) by the application of TTX (1 μm) in the MNTB from WT and LES rats. G, Na_vβ4 (green) expression at the axon heminode with calretinin (CR, red) in WT and LES rats. Circles indicate the heminodal region and arrows indicate the Na_vβ4 cluster. Unpaired t test, ^* P < 0.05. [Colour figure can be viewed at wileyonlinelibrary.com]

To determine the function consequences of I _NaP on the resting potential, we first set the resting potential to −70 mV by applying small holding currents (−18.7 ± 3.4 pA vs. −9.7 ± 1.8 pA for WT and LES calyx terminals, respectively, n = 9 and 9, P = 0.03). Then, membrane potential was recorded during the application of 1 μm TTX to inhibit I _NaP. TTX hyperpolarized WT calyx terminals (ΔV _m = −7.8 ± 3.5 mV, n = 5, P = 0.008, paired t test), whereas, in LES, TTX application had no significant effect on the resting membrane potential (ΔV _m = −0.2 ± 2.3 mV; n = 5, P = 0.814) (Fig. 5 E and F).

Previous studies demonstrated that the Na_vβ4 subunit underlies the generation of I _NaR and I _NaP (Grieco et al. 2005, Aman et al. 2009, Bant and Raman, 2010, Kim et al. 2010). To test whether a lack of compact myelin selectively impacts on Na_vβ4 expression, we examined Na_vβ4 expression at axon heminodes and terminals from WT and LES rats. We found that Na_vβ4 clusters were located at the last axon heminodes but not at terminals in the WT, whereas LES heminodes did not exhibit detectable Na_vβ4 (n = 17 cells in four slices) (Fig. 5 G). It thus appears that reduction of Na_vβ4 expression in LES heminodes explains the observed reduction in I _NaR.

Alterations in the distribution of K_v channels in the dysmyelinated axon terminal

In addition to Na_v channels, axons depend on K_v channels to shape the AP waveform. We therefore examined the expression of low‐voltage activated K_v1.2 channels and high‐voltage activated (HVA) K_v3.1 channels because these two channel types play an important role in reliable AP firing at the calyx of Held terminal (Ishikawa et al. 2003, Johnston et al. 2010). Along the normally myelinated axon, clusters of K_v3.1 and K_v1.2 were located, respectively, at nodes and heminodes, and at juxtaparanodes (Fig. 6 A). In LES, K_v1.2 clusters located at juxtaparanodes were completely disrupted, whereas clusters of K_v3.1 could still be identified at nodes (Fig. 6 B). At the heminode, however, K_v3.1 expression was disperse and extended into the calyx nerve terminals (Fig. 6 C). To quantify K_v3.1 expression at presynaptic terminals, we calculated the ratio of immunofluorescence of K_v3.1 to that of vGluT1 (a presynaptic marker). This measure of K_v3.1 expression at the calyx terminal was higher in LES (0.93 ± 0.12, n = 8) than WT (0.6 ± 0.09, I = 7, P = 0.048, unpaired t test) (Fig. 6 D), indicating that K_v3.1 expression increases at the calyx terminal from the LES.

Figure 6. Lack of compact myelin disrupts the cluster of K_v1.2 at juxtaparanodes and K_v3.1 at heminodes .

A, the calyx of Held axon and terminal were immunolabelled for K_v3.1 (green), K_v1.2 (red) and calretinin (CR, blue) in the WT rat brainstem. Dashed arrows indicate heminodes, nodes and juxtaparanodes, as well as the calyx terminal. Note, K_v3.1 was expressed at the node and the heminode and K_v1.2 was expressed at the juxtaparanode. B, expression of K_v3.1 (green), Caspr (blue) and K_v1.2 (red) at nodes, paranodes and juxtaparanodes along the axon in WT and LES rats. Arrows indicate K_v3.1 at nodes. C, at the axon heminode and terminal, expression of K_v3.1 (green) and vGluT1 (red) in WT and LES rats. Yellow arrows indicate the extent of K_v3.1 expression, which invades the presynaptic terminal in the LES. D, summary of the intensity ratio of K_v3.1 signal by vGluT1 at presynaptic terminals in WT and LES rats obtained in (C). E, representative trace of TEA‐sensitive HVA‐K⁺ (at +30 mV; sensitive to TEA 1 mm) at holding −70 mV recorded in the calyx terminal from WT and LES rats. Traces were obtained after TEA subtraction. The current–voltage (I–V) relationships of TEA‐sensitive HVA‐K⁺) currents in WT and LES rats obtained after the TEA subtraction. Unpaired t test, ^* P < 0.05. [Colour figure can be viewed at wileyonlinelibrary.com]

We examined how altered expression of K_v3.1 channels affects functional HVA K⁺ currents at the calyx terminals in LES rats. K_v3.1 channels are sensitive to a low concentration of TEA (1 mm) (Coetzee et al. 1999). TEA‐sensitive HVA K⁺ currents were significantly larger at presynaptic terminals in LES (542 ± 77 pA at 30 mV, n = 6) than in WT rats (335 ± 55 pA at 30 mV, n = 6, unpaired t test, P = 0.0012) (Fig. 6 E). This observed increase in HVA K⁺ current is consistent with and provides a mechanistic explanation for the larger after‐hyperpolarization after the action potential in calyx terminals from LES rats described previously (Kim et al. 2010, 2013 a).

We examined the expression and co‐localization patterns for Nav1.6 and K_v3.1 at nodes of Ranvier, the heminode and the nerve terminal for normal and LES axons in the MNTB. In the WT axon, Na_v1.6 and K_v3.1 clusters are co‐localized to the axon node and heminode. At the last heminode, Na_v1.6 clusters were located 10–20 μm from the terminal, with only sparse Na_v1.6 channels at the presynaptic terminal, whereas K_v3.1 expression extended to the terminal. In the LES rat, expression of both Na_v1.6 and K_v3.1 was dispersed and channels invaded the terminal (Fig. 7 A and C). By contrast, the overall structure of distal nodes of Ranvier in LES rats appeared to be less disrupted than the nerve terminal. K_v1.2 channels, which are normally clustered at the juxtaparanode, were dispersed throughout the node in the LES axon, although the co‐localization of Na_v1.6 and K_v3.1 clusters at nodes from LES rats appeared to be similar to WT nodes (Fig. 7 B). This appears to be sufficient for low frequency firing of the presynaptic terminal because LES calyx terminals can follow action potential trains up to 100 Hz without conduction failure (Kim et al. 2013 a). However, the observation that the reliability and firing at high frequencies (>300 Hz) (as occurs physiologically) is severely compromised in the LES calyx nerve terminal (Kim et al. 2013 a) suggests that the precise distribution of ion channels is essential for its function.

Figure 7. Lack of condensed myelin affects the distribution of Na_v and K_v clusters at the heminode and calyx of Held terminal in the MNTB .

A, expression of K_v3.1 (green) and Na_v1.6 (red) at the axon heminode and terminal marked with vGluT1 (blue) in WT and LES rats. Kv3.1 (yellow arrow) and Nav1.6 clusters (white arrow) were well co‐localized at the axon heminode in WT rats, whereas their clusters completely disrupted at heminode and dispersed to the terminal in LES rats (yellow arrows). B, nodal and juxtaparanodal staining with K_v3.1 (green), Na_v1.6 (red) and K_v1.2 (blue) in WT and LES rats. Kv3.1 and Nav1.6 clusters are well preserved at the node, although juxtaparanodal structures stained with Kv1.2 are disrupted in LES axons. C, summary diagram of effects of myelin loss on the distribution of voltage‐gate activated ion channels (Na_v1.6, Na_vβ, K_v1 and K_v3) at the last heminode. [Colour figure can be viewed at wileyonlinelibrary.com]

Discussion

Alterations in the structure of the node of Ranvier during demyelination have been described, although little is known about the role of myelination in regulating the structural and functional properties of the nerve terminal. In the present study, we provide evidence indicating that proper myelination plays a critical role in fine‐tuning presynaptic excitability and Na⁺ dynamics in the CNS nerve terminal. Specifically, compact myelination is required for the structural and functional properties of Na_v and K_v channels located in the last axon heminode. Loss of compact myelin disrupts Na_v channel clustering, and also changes Na⁺ dynamics and presynaptic excitability at the nerve terminal. We propose that the loss of Na_vβ4 (leading to the loss of resurgent Na⁺ current) and specific disruption of Na_v and K_v clusters at the axon heminode and terminal underlies the loss of action potential reliability in axons and nerve terminals in the LES model of dys‐/demyelinating diseases.

Axon myelination regulates AP propagation to the calyx terminal and presynaptic excitability

Changes in ion channel expression in the last axon heminode critically affect presynaptic excitability at the nerve terminal. Previous studies indicate that small changes to the conductance or spatial distribution of voltage‐dependent Na⁺ and K⁺ channels can determine the success or failure of AP invasion from a myelinated axon into a non‐myelinated segment, such as the heminode (Waxman and Wood, 1984). The auditory nervous system depends on a high reliability and temporal fidelity of impulse propagation to process auditory information. The calyx of Held terminal, located in the auditory brainstem, can fire APs at rates approaching 1 kHz, and its postsynaptic partner, the MNTB neuron, can keep up with transmission at frequencies up to 600 Hz in vitro without failure (Wu and Kelly 1993; Taschenberger and von Gersdorff 2000; Kopp‐Scheinpflug et al. 2008; Kim et al. 2013 a). For this fast and accurate spiking performance, the projections from cochlear nucleus cells require highly myelinated axons, as well as Na_v kinetics with fast recovery from inactivation and the presence of resurgent Na⁺ currents at the heminode next to the terminal.

LES rats present slower conduction to the calyx of Held with a higher failure rate at high frequency than WT (Kim et al. 2013 a). Conduction velocity is determined by several parameters, including axon diameter and internode length (internodal distance), as well as myelin thickness (Moore et al. 1978, Ford et al. 2015). In LES calyces, axon diameter increases from P18 (2.3 ± 0.19 μm, n = 10) to P38 (3.0 ± 0.26 μm, n = 14), whereas WT calyces showed no change in axon diameter (1.8 ± 0.1 μm at P18, n = 10 vs. 1.8 ± 0.1 μm at P38; data not shown). Therefore, changes to axon diameter probably do not account for slower conduction and increased AP failure in the LES rat. There was no significant difference with respect to internodal length in LES and WT axons, and the distance between the last node and the heminode was significantly reduced in LES calyces. Therefore, changes to internode distances also cannot account for the LES phenotype. In principle, a larger axon diameter and smaller internodal length would rather increase conduction velocity (Ford et al. 2015). Therefore, a reduced conduction velocity in the LES may be a direct consequence of dys‐/demyelination because modelling studies predict that the degree of myelination is one of the most important determinants of conduction velocity (Moore et al. 1978).

With respect to the frequency‐dependent AP failures observed in the LES rat, our recordings made in the nerve terminal cannot determine where these failures occur along the axon. In a previous study, we showed that the mature calyx of Held synapse relies on resurgent Na⁺ currents for reliable high‐frequency firing, and that AP failures during train in the immature calyx terminal at P5–7 could be rescued by dialysing the terminal with Na_Vβ4 peptide (Kim et al. 2010). Together with the lack of expression of Na_Vβ4 in the LES heminode, these data suggest that resurgent Na⁺ current in the heminode is essential for proper excitability of the nerve terminal. However, in the aforementioned study, we were unable to track the diffusion of the peptide back along the axon. Possibly, resurgent Na⁺ current is important for proper excitability along the entire axon and the site of rescue by the β4 peptide may have been distal to the heminode.

Another key element for determining presynaptic excitability is the expression pattern and kinetics of K_v channels, which were also affected by myelin loss. A stronger and more dispersed expression of K_v3.1 at the heminode and presynaptic terminal was accompanied by larger HVA‐K⁺ currents and a larger after‐hyperpolarization, whereas persistent and resurgent Na⁺ current was reduced. Together, these alterations are expected to reduce membrane excitability (Waxman and Wood, 1984, Kawaguchi and Sakaba, 2015).

In WT rats, calyces at the age studied can follow AP trains of up to several hundred Hz without failure. The LES rat displays AP failures at stimulations frequencies above 300 Hz but is still able to fire APs at up to 100 Hz (Kim et al. 2013 a). Although we cannot measure Na⁺ and K⁺ currents along the axon, our immunohistochemistry indicates apparently normal clustering and co‐localization of Na_v1.6 and K_v3.1 at nodes (Fig. 7), which must be sufficient to maintain its limited excitability despite disruption of K_v1.2 at juxtaparanodes (Fig. 6). A recent study suggests that clustering of channel proteins in nodes increases conduction velocity, independent of the presence of myelin (Freeman et al. 2015). Another factor that probably helps maintain AP propagation in the LES is that the distance between the last node of Ranvier and the heminode was significant decreased in LES calyces (Fig. 3 F). The boundary between myelinated and non‐myelinated segments of an axon terminal is a potential site for AP failure and depends critically on the length of the non‐myelinated segment (Waxman and Brill, 1978). In the LES, the reduced intermodal length closest to the nerve terminal may facilitate conduction at the nerve terminal, allowing it to function without failure in response to 100 Hz stimulation (Kim et al. 2013).

Axon myelination regulates Na⁺ dynamics at the nerve terminal

Where is the main entry of presynaptic Na⁺ during AP firing? In WT, Na⁺ transients measured at the heminode are larger and faster and occur earlier than at the presynaptic terminal (Fig. 1). Immunostaining of Na_v channels showed strong labelling in the heminode with little or no expression in the calyx terminal (Fig. 3). Together, these data indicate that, in the normal myelinated axon, most Na⁺ enters at the last heminode via Na_v channels and then diffuses passively to the calyx terminal. This hypothesis is similar to the retrograde diffusion of Na⁺ from the AIS to the soma in cortical pyramidal neurons (Fleidervish et al. 2010). What determines Na⁺ dynamics at the nerve terminal? The Na⁺/K⁺ ATPase is the main mechanisms regulating presynaptic Na⁺ gradients and the resting potential because blocking of Na⁺/K⁺ ATPase with ouabain (100 μm) gradually increased the basal Na⁺ level at presynaptic terminals (data not shown). However, in myelinated axons, the rapid decay of the Na⁺ transient after a train of APs mainly depends on fast bilateral diffusion along the axon (Fleidervish et al. 2010). Axon dysmyelination impacts the intensity and decay time of the Na⁺ transient, as well as Na_v expression at the heminode and the presynaptic terminal (Figs 1, 2, 3). In the LES, a reduced and dispersed expression pattern of Na_v channels probably explains the smaller and slower Na⁺ transient measured in the heminode. Conversely, the increased expression of Na_v channels in the nerve terminal probably underlies the increased size of the Na⁺ transient measured in the terminal of LES calyces. Moreover, a more spatially uniform Na⁺ entry, as a result of a more spatially uniform Na_v channel distribution, reduces the ability of Na⁺ to clear by diffusion and is expected to increase the time for Na⁺ transients to decay. However, we cannot rule out the possibility that demyelization also affects Na⁺ extrusion mechanisms mediated by Na⁺/K⁺ ATPase and NCX (Kim et al. 2005, 2007; Lee and Kim, 2015) in the LES rat.

Compact myelination is essential for clustering of Na_v and K_v at the axon heminode and terminal, rather than at distal nodes

In the present study, we revealed the role of myelin in the segregation of Na_v channel expression at the nerve terminal to maintain a high density of Na_v channels present at the heminode and a sparse density of Na_v channels in the presynaptic terminal. This conserved structure around the nerve terminal was severely disrupted by a loss of compact myelination in the LES rat, whereas alterations at distal axon were mostly confined to an altered expression pattern for caspr and K_v1.2 at paranodes and juxtaparanodes (Figs 3 and 6).

Previous studies using cuprizone to induced demyelination (Dupree et al. 2004, Crawford et al. 2009, Hamada and Kole, 2015) or mice deficient for ceramide galactosyltransferase (Dupree et al. 2004) demonstrated disruption of the nodal Na_v cluster. Unlike these previous studies, we observed that Na_v1.6 and K_v3.1 at distal nodes remained well clustered in the LES rat and appeared quite similar to the WT, although paranodal and juxtaparanodal structures were disrupted. Nodal Na⁺ clusters require axon–oligodendrocyte interactions and axonal cytoskeletal mechanisms at nodes (Rasband et al. 1999, Susuki et al. 2013). The afferent fibres to the MNTB display Na_v clusters at nodes of Ranvier in the first postnatal week, before MBP‐mediated formation of compact myelination, whereas the Na_v channel is detectable at the last heminode only after the onset of myelination (after P7–8; data not shown). This suggests that MBP‐mediated compact myelination is required for Na_v cluster formation specifically at the last heminode but not at nodes of Ranvier (Freeman et al. 2015). Instead, the interaction between axon and oligodendrocytes or the formation of axonglial junctions, prior to MBP expression, might play an important role in forming a nodal Na_v cluster (Rasband et al. 1999, Susuki et al. 2013, Freeman et al. 2015).

In some demyelination models (e.g. experimental autoimmune encephalomyelitis and the cuprizone models), nodal clusters of Na_v channels deteriorated during demyelination. In the LES model, we observed normal Na⁺ clustering at nodes. The discrepancy may depend on the underlying mechanisms of clustering and the cause of demyelination. If axon–oligodendrocyte interactions are the initial determinant for the location of Na⁺ channels at nodes, oligodendrocyte or axon degeneration could alter nodal structures in demyelinated models. In the LES model, we observed no strong effect on axon–oligodendrocyte interaction, and no axon or oligodendrocyte degeneration within the first month of age, although MBP deletion strongly inhibited the formation of compact myelination around axons. Our data suggest that the presence of compact myelination maintains the heminodal Na_v cluster near the terminal. In the present study, nodal Na_v channels maintained a normal cluster pattern along dysmyelinated LES axons in young adult animals (up to P50; data not shown). It remains to be determined whether nodal Na_v clusters are maintained in ageing in the LES rat.

In summary, the results of the present study demonstrate that compact myelin is critical to maintain a proper location and function of Na_v and K_v at the axon heminode and terminal.

Additional information

Competing interests

The authors declare that they have no competing interests.

Author contributions

JHK designed the research. EB, SEK, SYL, and JHK collected data. EB, SEK, SYL, CK and JHK analysed and interpreted data. EB, CK and JHK wrote the manuscript. JHK provided financial support. All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported by grants from the NIDCD (R01DC013157) to JHK and by grants from CNPq (478663/2012‐9 and 306293/2012‐9) to CK.