Distinct forms of synaptic plasticity during ascending vs. descending control of medial olivocochlear efferent neurons

Activity in each brain region is shaped by the convergence of ascending and descending axonal pathways, and the balance and characteristics of these determine neural output. The medial olivocochlear (MOC) efferent system is part of a reflex arc that critically controls auditory sensitivity. Multiple central pathways contact MOC neurons, raising the question of how a reflex arc could be engaged by diverse inputs. We examined functional properties of synapses onto brainstem MOC neurons from ascending (ventral cochlear nucleus, VCN), and descending (inferior colliculus, IC) sources in mice using an optogenetic approach. We found that these pathways exhibited opposing forms of short-term plasticity, with VCN input showing depression and IC input showing marked facilitation. By using a conductance clamp approach, we found that combinations of facilitating and depressing inputs enabled firing of MOC neurons over a surprisingly wide dynamic range, suggesting an essential role for descending signaling to a brainstem nucleus.


Abstract:
7 Activity in each brain region is shaped by the convergence of ascending and descending axonal 8 pathways, and the balance and characteristics of these determine neural output. The medial 9 olivocochlear (MOC) efferent system is part of a reflex arc that critically controls auditory  Introduction: 20 The cochlea is the peripheral organ of hearing. As such, it communicates with the central 21 nervous system by its centrally-projecting afferent fibers. However, the cochlea also receives 22 input from a population of cochlear efferent fibers that originate in the brainstem. The medial 23 olivocochlear (MOC) system provides many of these efferent fibers, and may serve to protect the 24 cochlea from acoustic trauma (Rajan, 1988; Kujawa and Liberman, 1997;Darrow et al., 2007) 25 and to dynamically enhance the detection of salient sound in diverse sensory environments 26 (Winslow and Sachs, 1987a; Kawase and Liberman, 1993) by controlling cochlear gain in a 27 frequency and intensity specific manner. MOC efferent fibers arise from cholinergic neurons 28 whose somata primarily reside in the ventral nucleus of the trapezoid body (VNTB) of the 29 superior olivary complex (SOC) (Warr, 1975), and project to outer hair cells in the cochlea 30 (Guinan et al., 1983(Guinan et al., , 1984Wilson et al., 1991), and this peripheral control by efferents has been 31 extensively studied (Guinan, 2010(Guinan, , 2018. MOC fibers respond to sound and form a negative 32 feedback system, and is thus described as a reflex providing frequency-specific feedback to the 33 cochlea (Liberman and Brown, 1986;Winslow and Sachs, 1987b;Brown, 2016). This feedback 34 is mediated by acetylcholine released from terminals of MOC fibers, thereby inhibiting outer hair 35 cell motility and decreasing cochlear sensitivity (Wiederhold and Kiang, 1970). arc or as mediators of descending control by higher brain regions. For example, excitatory 41 synaptic inputs that modulate and control MOC neuron function are made both by ascending 42 mice. We found that the majority of MOC neurons had a resting membrane potential of -80.4 ± 112 0.8 mV (N = 56), and were silent at rest (only 3/59 neurons were spontaneously active), 113 consistent with the low frequency of spontaneous activity observed in vivo (Fex, 1962;Cody and 114 Johnstone, 1982; Robertson, 1984; Robertson and Gummer, 1985). The membrane capacitance 115 (Cm) and resistance (Rm) were 36.5 ± 1.6 pF and 123 ± 9 MΩ (N = 59), respectively. In response  Table 1. For 162 comparison between double and single exponential fits, ꚍ fast and ꚍ slow were converted to a 163 weighted decay time constant, Afast and Aslow are the absolute amplitudes of each component. There was no significant 166 difference between ꚍ from single exponential fits and ꚍ w. Current voltage relations were 167 constructed by plotting the peak amplitude of EPSCs evoked at holding potentials between -82.8  neurons that receive auditory nerve input (Oertel et al., 2011). As a population, they may encode 181 sound intensity and frequency spectrum. T-stellate cells are a major ascending pathway of the 182 auditory system which project widely to many targets, and are the only VCN cell which projects 183 to the IC.  (Adams, 1979;Thompson, 1998). In recordings from tdTomato positive VCN neurons (N = 13) in AAVrg-pmSyn1-EBFP-Cre 203 infected ChAT-Cre/tdTomato mice, all neurons exhibited responses to current injections that 204 were characteristic of T-stellate cells (see example in Figure 4C). Action potentials fired 205 tonically with a sustained rate in response to depolarizing current injections ( Figure 4D).

222
These results definitively show that at least a subset of IC projecting T-stellate cells provide 223 glutamatergic excitatory input to MOC neurons.     Table 1, and example Figure 6D). The average ꚍ fast and ꚍ slow of mEPSCs were 0.17 ± 0.01 ms 279 and 1.72 ± 0.43 ms, respectively (N = 3 neurons, 1873 mEPSCs). The inter-event interval (IEI) 280 between mEPSCs ranged from 6.8 ms to 2.6 seconds, and each event was counted and sorted 281 into 20 ms bins ( Figure 6E). The distribution of binned mEPSC inter-event intervals was best 282 described with a single exponential equation (ꚍ = 0.20 seconds), reflecting that the miniature 283 events were stochastic in nature (Fatt and Katz, 1952). The average and median mEPSC 284 amplitudes were 57.5 ± 0.9 pA and 52.5 pA, respectively, and ranged from 27.2 pA to 146.9 pA 285 ( Figure 6F). In comparison with light-evoked EPSCs from IC and VCN (Table 1), these results 286 confirmed that the majority of mEPSCs were due to fast-gating AMPARs. Together, supportive 287 of data from light-evoked EPSCs and glutamate-puff evoked currents, these results suggest that  Ascending and descending inputs to medial olivocochlear neurons show distinct, opposite forms 292 of short-term plasticity 293 The AMPARs mediating transmission from VCN and IC were biophysically similar (Table 1). 294 However, input-specific repetitive activation of VCN or IC inputs revealed strikingly opposing 295 forms of short-term plasticity (Figure 7). During 20-pulse tetanus stimuli (20 or 50 Hz), light-296 evoked VCN-originating EPSCs depressed whereas IC-originating EPSCs facilitated ( Figure 7A To analyze recovery from facilitation or depression, a test EPSC was evoked after a 20-pulse 311 tetanus at time intervals increasing from 100 ms to 25.6 seconds ( Figure 7A and E). This was 312 repeated five to twenty times for each test pulse with a 30-second or greater gap between sweeps 313 and the results were averaged. We fit recovery data with single exponential functions, and found 314 that depression observed by VCN input recovered with a time-course (ꚍ 20 Hz = 3.5 ± 0.7 sec, ꚍ 50 315 Hz = 3.1 ± 0.4 sec) comparable to the recovery from IC input facilitation (ꚍ 20 Hz = 4.5 ± 1.4 sec, ꚍ 316 50 Hz = 4.4 ± 1.7 sec, Figure 7E). While classical short-term facilitation lasts for only hundreds of 317 milliseconds after tetanus stimuli (Zucker and Regehr, 2002), IC input facilitation onto MOC 318 neurons lasted for tens of seconds. This longer-lasting facilitation resembles synaptic 319 augmentation, which has a longer lifespan (seconds) than classical short-term facilitation 320 (milliseconds) and a recovery time-course that is insensitive to the duration or frequency of 321 repetitive activation (Magleby, 1987;Zucker and Regehr, 2002). Thus, while ascending and 322 descending inputs to MOC neurons employ similar postsynaptic receptors, they differ 323 dramatically in short-term plasticity.

325
The onset and dynamic range of MOC neuron output is controlled by integrating facilitating and 326 depressing inputs 327 We showed above that the intrinsic properties of MOC neurons permit them to fire over a wide 328 range. Moreover it has been previously shown that MOC neurons respond dynamically to a wide 329 variety of binaural sound intensities and frequencies (Liberman and Brown, 1986;Brown, 1989;     response to EPSG waveforms significantly increased with presynaptic firing rate ( Figure 8E). FS 383 was 0.095 ± 0.014 for 40 Hz, and 0.31 ± 0.08 for 180 Hz (p = 1.6 × 10 -4 , paired samples 384 Student's t-test). Some MOC neurons failed to reach action potential threshold in response to 385 depressing waveforms at ~40 Hz (6/6 failures with 10 inputs, and 3/6 with 20 inputs), and ~180 386 Hz (3/6 failures with 10 inputs) (e.g., last two rows of Figure 8D). When action potentials were 387 elicited and the number of simulated inputs were equivalent, depressing waveforms at ~180 Hz 388 always drove MOC neurons to threshold earlier than those at ~40 Hz ( Figure 8F). At ~40 Hz, 389 with 20 to 40 simulated inputs, depressing waveforms often elicited an onset response ( Figure   390 8Ci) that occurred earlier than facilitating waveforms at the same rate ( Figure 8D and F). When 391 the presynaptic firing rate was increased to ~180 Hz, facilitating waveforms with 10 to 20 392 simulated inputs generally elicited an onset response sooner than with depressing inputs. As our       Neurons throughout the brain receive mixtures of synaptic inputs that vary not only in their 555 origin or information content, but their short-term plasticity. A prominent example is that of 556 cerebellar Purkinje neurons, whose parallel fiber inputs facilitate while climbing fiber inputs 557 depress (Sakurai, 1987;Hansel and Linden, 2000). The physiological functions served by this 558 diversity likely vary with brain region. In MOC neurons, we found that synaptic responses 559 having properties of the ascending or descending inputs alone were not capable of encoding 560 firing over a wide range and with short latency. However by combining these different types of 561 input and varying input number and firing rate, sustained MOC output could vary over 20-fold. 562 We suggest that this central synaptic mechanism could aid in grading the level of efferent 563 dampening of cochlear function according to sound level.   Leica, or 7000smz-2, Campden) in warm aCSF. Throughout sectioning, brain slices were 636 collected and stored in aCSF at 31°C. When sectioning was completed slices were incubated an 637 additional 30 minutes at 31°C, followed by storage at room temperature, ~23°C.  Figure 1D). unit was set to rectifying mode and Erev was set to +10 mV, to simulate MOC neuron CP-714 AMPARs ( Fig. 3F and 5E).

716
Conductance waveforms were created using Igor Pro 8 and modeled using physiological data. was based on a fit to averaged EPSCs from IC and VCN, with ꚍrise = 0.27 ms, and ꚍdecay = 1.9 ms.

726
Timing and frequency of EPSGs were convolved to action potential timing from T-stellate cells 727 in response to repeated 500-ms current injections, and each repetition (trial) was considered an 728 input ( Fig. 8A and 9A). T-stellate cells were identified by virally mediated retrograde labeling 729 (AAVrg-pmSyn1-EBFP-Cre) (Fig. 4) from the contralateral IC in Ai9(RCL-tdT) mice. For each 730 individual input, short term plasticity was simulated in a frequency invariant manner by 731 weighting Gmax of unitary EPSGs according to exponential fits of physiological data (Fig. 7D).