Auditory sensitivity regulation via rapid changes in expression of surface AMPA receptors

Abstract

We report a robust regulation of surface AMPA receptors in mouse auditory neurons, both with application of glutamate receptor agonists in cultured neurons and in response to acoustic stimulation in vivo. The reversible reduction of surface AMPA receptors following acoustic stimulation correlated with changes in acoustic sensitivity. Thus we show that AMPA receptor cycling is important for optimizing synaptic transfer at one of the most exacting synapses in the body.

Peripheral auditory neurons encode acoustic frequencies out to 6 kHz over nearly three orders of magnitude of graded stimulus intensity^1,2. This temporal precision and dynamic range is astonishing when one considers that this process is mediated by the hair cell releasing packets of neurotransmitter, and that the entire input to the auditory nerve fiber is encoded in a single ribbon synapse³. The achievement of these stimulus coding capabilities in a single synapse requires a number of specializations, not the least of which must be some formidable mechanisms for regulating synaptic strength over time.

AMPA receptors (mainly GluR2–4) mediate glutamatergic transmission between the hair cell and the auditory neurons⁴. Recent studies in cultured hippocampal neurons indicate that synaptic strength may be dynamically regulated by AMPA receptor cycling at the postsynaptic membrane in response to the activation of glutamatergic receptors^5–10.

We first determined whether cultured auditory neurons could regulate surface AMPA receptors in a manner similar to the regulation of hippocampal neurons. Auditory neurons were identified by labeling with an antibody to neurofilament (a neuronal marker) and by their appearance in differential interference contrast microscopy (Supplementary Methods online). At 15–18 h after plating, cell cultures from the spiral ganglia of postnatal day 6–8 mice mainly comprised neurons with short projections. We distinguished surface from total AMPA receptors with an antibody directed to an extracellular epitope of the GluR2 subunit in the absence and presence of membrane permeabilization (Fig. 1). In the absence of permeabilization, the GluR2 label was primarily confined to the cell surfaces; following permeabilization, the GluR2 label was present throughout the neuronal cell body and neurites. The labeling of total GluR2 was approximately 1.3-fold that of surface receptor labeling. To estimate the membrane permeability as a result of fixation alone, label for neurofilament (an intracellular protein) was quantified with and without permeabilization of the cultured auditory neurons. The amount of neurofilament label seen before permeabilization was approximately 11% of that seen after. Thus we estimate surface GluR2 to be about two-thirds of the total.

Figure 1.

Glutamate receptor agonists and antagonists altered surface and total GluR2 receptors in cultured mouse auditory neurons. (a) Differential interference contrast light microscopy, neurofilament label, and GluR2 label (total and surface) in auditory neurons. (b) Changes in surface and total GluR2 in response to application of 20 μM AMPA, NMDA or glutamate for 10 min. All glutamate agonists reversibly decreased surface GluR2 when compared with application of artificial perilymph (AP) (P < 0.05 for all times less than 40 min). Total GluR2 was reduced by NMDA and glutamate (P < 0.05), but not by AMPA application. Each data point represents the analysis of five images of neuronal clusters from each of three different cultures. The number of neurons in each image ranged from 3–40. Means ± s.e.m. are plotted. (c) Both NMDA and AMPA receptors mediated the removal of surface GluR2. Neither APV (50 μM) nor DNQX (20 μM) alone produced a significant (P < 0.05) reduction of surface GluR2. DNQX blocked AMPA-induced, but not NMDA-induced, removal of surface GluR2, whereas APV inhibited NMDA-induced, but not AMPA-induced, surface GluR2 removal. Plotted are means ± s.e.m. * P < 0.05, compared with control surface GluR2 as analyzed with a nonpaired Student’s t test.

A 10-min exposure to 20 μM glutamate, NMDA or AMPA reduced the number of surface GluR2 receptors by 55–60% (Fig. 1), while cell morphology remained intact. Surface GluR2 began to recover at 10–20 min, and returned to prestimulation levels by 60 min. The effects were concentration dependent (Supplementary Fig. 1 online), with NMDA being the most potent and AMPA being the least potent. NMDA-induced surface GluR2 removal was blocked by the NMDA-receptor antagonist d-(–)-2-amino-5-phosphonovaleric acid, APV (50 μM), but not by the AMPA-receptor antagonist 6,7-dinitro-quinoxaline-2,3-dione, DNQX (20 μM) (Fig. 1). Correspondingly, AMPA-induced removal of surface AMPA receptors was inhibited by DNQX, but not by APV.

Excluding calcium from the extracellular medium nearly eliminated the decrease in surface AMPA receptors that was induced by AMPA, NMDA or glutamate, suggesting that an influx of calcium is required for surface GluR2 removal (Supplementary Fig. 2 online). We investigated the role of voltage-gated calcium channels by adding nifedipine (10 μM), an L-type channel blocker, or ω-conotoxin GVIA (1 μM), an N-type channel blocker, to the bathing medium. Neither blocker altered surface AMPA receptor quantities by itself, but each could partially (40–50%) reduce agonist-induced removal of surface GluR2 receptors.

In cultured hippocampal neurons, surface AMPA receptors, once internalized, can be recycled or degraded¹¹. Auditory neurons behaved similarly; NMDA decreased total GluR2 receptor by ~25%, while reducing surface GluR2 by 60%, indicating that about 40% of internalized GluR2 was degraded. In contrast, AMPA did not change the total GluR2. Differential effects of AMPA and NMDA on internalized AMPA receptor degradation are thought to be due to their differential activation of NSF (N-ethylmaleimide sensitive fusion) protein^5,11,12, which inhibits the diversion of AMPA receptors to lysosomes and promotes receptor recycling back to the cell membrane.

To investigate whether our findings in cultured auditory neurons are relevant to transmission of acoustic information in vivo, we monitored sound-evoked changes in cochlear surface AMPA receptors and auditory threshold sensitivity in CBA/CaJ mice (6–8 weeks old; Supplementary Methods) with procedures that were approved by the Massachusetts Eye and Ear Infirmary Animal Care Committee. Our rationale was that acoustic stimulation should increase the release of neurotransmitter from the hair cell to activate glutamate receptors on auditory neurons. If the consequent removal of surface AMPA receptors has a role in synaptic strength in the mouse cochlea, we should detect a reduction in auditory sensitivity.

A 10-min broad-band acoustic-noise exposure, chosen to activate most of the cochlear sensory epithelium, caused threshold (sensitivity) shifts in the neural-based auditory brainstem responses (ABRs) of ~15 dB at 8 kHz, ~20 dB at 20 kHz and ~30 dB at 45.2 kHz, measured in 2–6 min after noise. To measure changes in surface AMPA receptors in the mouse cochlea after noise exposure, we biotinylated all cochlear surface proteins and then extracted the biotinylated proteins from inner ear homogenates for western blot analysis. Surface AMPA receptors were detected and quantified by densitometry from bands in the western blot in protein extracted before, and at varying times after, noise exposure (Fig. 2). At 2 min following noise exposure, surface AMPA receptors decreased to 49% of that in cochleae without noise exposure, and recovered to 64% at 20 min, and to 91% 1 h after noise.

Figure 2.

Noise exposure reversibly decreased both cochlear surface AMPA receptor and auditory sensitivity in vivo. (a) Western blots of brain and cochlear extracts. The left two columns illustrate the total GluR2 in brain and cochlea. Surface GluR2 (columns 3–6) was determined after biotinylation of cochlear surface. (b) Following noise exposure, the time course of recovery of auditory sensitivity (reversal of threshold shift) correlated with that of the quantity of surface AMPA receptor. In ten mice, threshold shifts, measured after a 10-min broad-band noise exposure (1–40 kHz, 116 dB sound pressure level) were averaged over a broad range of cochlear frequencies (8, 20 and 45 kHz; threshold shifts for each of these frequencies are presented in Fig. 3c). For western blot analysis of surface GluR2, each data point represents five samples and each sample comprised cochlear tissue from three mice. Plotted are means ± s.e.m.

Most of the proteins detected in these biotinylated extracts were indeed surface proteins. We measured the amount of β-actin as an indicator of intracellular proteins that might be present in each sample. Only trace β-actin bands were found in the biotinylated protein sample, whereas large amounts of β-actin were detected in extracts of brain and whole cochlea samples that were not subjected to biotinylation (Fig. 2).

The time course of the decrease in surface GluR2 in vivo correlated with the decrease in auditory sensitivity following noise exposure (Fig. 2). ABRs recovered gradually with time post-exposure and returned to near baseline levels by 60–70 min. Over the same period of postexposure monitoring, the cochlear distortion product otoacoustic emissions, generated by the outer hair cells, remained stable. These findings are thus consistent with a site of action at the hair cell–auditory neuron synapse. The time course of the reduction and recovery of surface AMPA receptors in the cochlea following noise exposure was correlated to the change and recovery of ABR thresholds to sound stimuli (Fig. 3), consistent with the idea that the reduction in number of surface AMPA receptors at the synapse makes the auditory neuron less sensitive to acoustic stimuli.

Figure 3.

Pharmacology of surface GluR2 removal. (a,b) Myr-Dyn (200 μM for 90 min), APV + DNQX (3 mM each for 60 min) and APV (3 mM) alone all blocked noise-induced surface GluR2 removal. Each data point represents analysis of three samples and each sample comprised two cochleae (one cochlea taken per mouse). Data are expressed as mean ± s.e.m. and analyzed using a nonpaired Student’s t test. * P < 0.05, compared with cochlear surface GluR2 without noise. (c) APV reduced noise-induced ABR threshold shifts. The time course of recovery of threshold shifts for three different cochlear regions are shown in uninfused cochleae (noise only), cochleae infused with an artificial perilymph solution (AP + noise) and in those infused with APV (APV + noise). APV reduced the amount of threshold shift following acoustic overstimulation and reduced the time for recovery. Data with noise only were summarized from ten mice. Data for AP or APV perfusion were taken from 4–6 mice each. Means ± s.e.m. are plotted. * P < 0.05, comparing APV + noise with noise only.

To determine whether the noise-induced changes in surface AMPA receptors occurred via activation of glutamate receptors, we infused the cochlea with a combination of DNQX (3 mM) and APV (3 mM), exposed the animal to noise, and then collected the cochleae immediately after noise for western blot analysis. Cochlear infusion with these agents completely blocked both the neural response to acoustic stimulation and the noise-induced removal of surface GluR2 receptors (Fig. 3). Indeed, a slight increase in surface GluR2 was observed in the presence of these antagonists compared with that observed in silence, consistent with a block of spontaneous discharge rate, which is present in silence and is mediated by glutamatergic transmitter released from the hair cells.

Because AMPA receptor antagonists block neuronal responses to acoustic stimulation, it was not possible to examine the effects of AMPA antagonists alone on recovery of auditory sensitivity following exposure to high-level noise. However, we could examine the effects of APV on this process, as APV produces no acute effects on auditory thresholds. In the presence of APV, noise induced a smaller decrease (~10–15 dB less) in auditory sensitivity, followed by a more rapid recovery. We also noted a reduction in noise-induced removal of surface GluR2 (Fig. 3).

The number of surface AMPA receptors present at any moment is determined by the rate of internalization of the receptor, presumably by endocytosis, and the rate of externalization of the receptor, presumably by exocytosis. We analyzed the effects of myristoylated dynamin inhibitory peptide (myr-Dyn), a membrane-permeable agent known to block endocytosis of the AMPA receptor¹³, on the removal of surface GluR2 in response to noise. myr-Dyn blocked the noise-induced reduction in surface receptor, suggesting that endocytosis is involved in this process (Fig. 3). Another indicator of the specificity of this process to the AMPA receptor is that we did not observe a decrease in GABA_B receptor, also present exclusively on auditory neurons, on the cochlear surface following noise stimulation (Supplementary Fig. 3 online).

The correlation of a decrease in auditory sensitivity with a decrease in surface AMPA receptor suggests that the efficiency of afferent transmission in the cochlea is regulated by changing the number of surface AMPA receptors. A decrease in the number of postsynaptic glutamate receptors should decrease the probability of a response to a given quantity of transmitter released. Because the auditory nerve fiber responds probabilistically to the temporal and intensity characteristics of the acoustic stimulus^1,2, even subtle effects on synaptic efficiency might be expected to alter the response characteristics of the auditory nerve fiber.

Rapid regulation of synaptic efficiency in auditory transmission may have several important roles. First, one method for optimizing the response range at this synapse may be to regulate the number of postsynaptic receptors responding to transmitter released from the hair cell. Second, mammalian auditory neurons encode the temporal characteristics of an acoustic stimulus as a modulation of afferent discharge around a baseline (or spontaneous) discharge rate². Maintenance of a stable spontaneous rate in the face of vagaries in presynaptic influences may be important in this regard. Finally, a reduction in the number of glutamate receptors in the face of continuing high-intensity acoustic input might be important for reducing excitotoxicity.

Although NMDA receptors are present on afferent fibers¹⁴, their role in auditory function has not previously been elucidated. They are not thought to be directly involved in fast neurotransmission because selective NMDA antagonists have little acute effect on cochlear potentials¹⁵. Our results suggest that NMDA receptors are involved in modulating synaptic efficiency by regulating the number of surface AMPA receptors in auditory neurons. Consistent with this is our finding that NMDA antagonists alter the recovery of cochlear potentials following the presentation of high-level acoustic stimulation.

By characterizing the phenomenon in auditory neurons and demonstrating its responsibility for regulating auditory sensitivity in vivo, we establish that AMPA receptor cycling is important for optimizing synaptic transfer at one of the most exacting synapses in the body. This report is also one of the first to demonstrate a specific functional role in vivo for rapid cycling of glutamate (AMPA) receptors. Perhaps the demands of regulating synaptic strength in the auditory periphery make the phenomenon robust and simpler to characterize in vivo than at other synapses, though it seems likely that AMPA cycling has a similar role throughout the nervous system.