|Home | About | Journals | Submit | Contact Us | Français|
In the inferior colliculus (IC), a brief period of acoustic conditioning can transiently enhance evoked discharge rate. The cellular basis of this phenomenon was assessed with whole cell current-clamp recordings in a gerbil IC brain slice preparation. The current needed to elicit a single action potential was first established for each neuron. A 5s synaptic stimulus train was delivered to the lateral lemniscus (LL), and followed immediately by the initial current pulse to assess a change in postsynaptic gain. The majority of IC neurons (66%) displayed an increase in current-evoked action potentials (positive gain). Despite the blockade of ionotropic glutamate receptors, this effect was correlated with membrane depolarization that occurred during the synaptic train. The postsynaptic mechanism for positive gain was examined by selective blockade of specific neurotransmitter receptors. Gain in action potentials was enhanced by antagonists of metabotropic glutamate, acetylcholine, GABAA and glycine receptors. In contrast, the gain was blocked or reduced by an antagonist to ionotropic serotonin receptors (5-HT3R). Blocking voltage-activated calcium channels with verapamil also reduced the effect. These results suggest that 5-HT3R activation, coupled with increased intracellular calcium, can transiently alter postsynaptic excitability in IC neurons.
The modification of neural coding properties can last for periods of milliseconds to days. An example of auditory coding plasticity over a short interval (~10 ms) is the precedence effect (Blauert 1972; Litovsky et al. 1999; Tollin and Yin 2002). Neural correlates of this phenomenon, which involve a neuron’s rapid change in sensitivity to an acoustic cue, have been identified in the IC (Burger and Pollak 2001; Litovsky and Yin 1998). There are also many studies that demonstrate an altered response to acoustic signals over longer intervals (≥100 ms). Examples that are relevant to the present study include the increased responses that are observed following modulation of phase or level which simulates acoustic motion, as well as frequency modulation (Sanes et al. 1998; Wilson and O’Neill 1998; Malone and Semple 2001). The altered discharge rate persists for 100s of milliseconds to seconds, suggesting that a form of short-term plasticity exists in the IC (Finlayson and Adam 1997; Kvale and Schreiner 2004; Sanes et al. 1998; Spitzer and Semple 1998; Thornton et al. 1999). It is not yet known whether this form of plasticity arises from intrinsic membrane properties, inputs from primary auditory nuclei, or inputs from neuromodulatory systems (Adams 1979; Coleman and Clerici 1987; Faingold 1991; Sivaramakrishnan and Oliver 2001). Therefore, the present study was designed to identify cellular mechanisms that could support these forms of short-term plasticity.
The central nucleus of the IC is a locus of synaptic integration in the central auditory system. Both monaural and binaural projections from the brainstem nuclei converge there, and their terminals release the excitatory and inhibitory neurotransmitters, glutamate, GABA and glycine (Glendenning and Baker 1988; Saint Marie et al 1999; Glendenning et al 1992; Gonzalez-Hernandez et al 1996, Zhang et al 1998). Furthermore, the balance of excitation and inhibition in the IC is influenced by modulator neurotransmitters. Both acetylcholine and serotonin are present in the IC and their effects on auditory processing have been examined (Contreras and Bachelard, 1979; Hurley et al 2002). Frequency tuning properties can be altered either by stimulation of ascending cholinergic afferents, or focal application of serotonin (Ji et al 2001, Ma and Suga 2003; Hurley and Pollak 1999, 2001; Hurley 2006; Hurley 2007; Hurley 2008). Indeed, activation of serotonin can have a variety of effects on the timing and magnitude of frequency tuning, due to the diversity of receptor subtypes present in the IC (Hall and Hurley 2007; Hurley 2007; Hurley et al. 2008). The recruitment of postsynaptic calcium currents in IC neurons may also support response enhancement. Rebound calcium-activated depolarizations can last for several hundred milliseconds in IC neurons (Sivramakrishnan and Oliver 2001), while other calcium currents have been shown to last for seconds (N’Guemo and Morad 2003).
Whereas much attention has been devoted to the cellular basis of short and long-term plasticity (e.g., paired-pulse depression or facilitation, long-term potentiation), little is known about intermediate forms of plasticity (Buonomano and Merzenich 1998; Shin et al. 2001). In this paper, we describe a novel form of short-term plasticity in IC neurons, elicited by a brief period of synaptic stimulation in a brain slice preparation. Here we identify a serotonergic mechanism involving ionotropic serotonin receptors (5-HT3) that causes a transient gain in current-evoked action potentials. We propose that transient states of increased excitability of IC neurons may underlie the short-term increases in discharge rates observed in vivo. Interestingly, an independent report (Bohorquez and Hurley, co-submitted with this manuscript) shows that a 5-HT3 receptor-dependent mechanism operates during auditory processing in the mouse inferior colliculus.
Gerbils (Meriones unguiculatus) aged postnatal day (P) 10–20 were anaesthetized with chloral hydrate (350 mg/kg), decapitated, and used to make 300–350 μm coronal brain slices through the IC. Slices were obtained at 4°C for animals aged P10–12 and at room temperature (25°C) for animals aged P13–20. A broad age range covering periods before and after hearing onset (~P12) were chosen to track developmental or use-dependent differences. All slices contained visible fibers of the lateral lemniscus (LL) ventral to the IC. The artificial cerebrospinal fluid (ACSF) contained (in mM): 125 NaCl, 4 KCl, 1.2 KH2PO4, 1.3 MgSO4, 24 NaHCO3, 15 glucose, 2.4 CaCl2, and 0.4 L-ascorbic acid (pH = 7.3 when bubbled with 95% O2/5% CO2). Low Ca2+ ACSF contained 0.24 mM CaCl2, and all other ingredients remained the same. The slices were continuously superfused in the recording chamber with oxygenated ACSF at 4 ml/min.
Whole-cell current-clamp recordings (Warner Instruments PC-501A) were obtained at room temperature (≈ 23°C), with electrodes fabricated from 1.5 mm outer diameter borosylicate glass microcapillaries (10–15MΩ). The internal pipet solution contained (in mM): 130 potassium gluconate, 0.6 EGTA, 10 Hepes, 2 MgCl2, 5 KCl, 2 ATP, 0.3 GTP, 5 phosphocreatine (pH = 7.2). All neuronal recordings were made in the position of the IC slice known to contain the heaviest SOC innervation (Beyerl 1978; Brunso-Bechtold et al. 1981; Nordeen et al. 1983) and greatest neuronal density, defined as the central nucleus by Nissl, Golgi, and cell-myelin method (Faye-Lund and Osen 1985). A GΩ seal was obtained, the membrane was ruptured, and electrode and access resistance were then compensated to about 70% (Kotak et al. 1998).
Each neuron was activated with 500 ms depolarizing current steps in 10pA increments, to determine the threshold for a single action potential, and the suprathreshold firing pattern. We restricted our analyses to neurons that displayed a repetitive action potential pattern in response to suprathreshold current pulses, either with an initial firing burst, or without (Figure 2A and B). Spike adaptation was often observed during the current pulse, similar to rat IC neurons (Sivaramakrishnan and Oliver 2001). Neurons with onset and transient discharge patterns were excluded because their intrinsic membrane properties do not permit an increase in the number of action potentials following a synaptic train stimulus.
After a 5–10 minute rest, each neuron was also characterized with 500ms, 10pA steps of hyperpolarizing current. The presence or absence of rebound-spiking and delayed rectifier currents (sag current, or Ih) was recorded for subsequent analyses.
To assess the influence of afferent activity on postsynaptic gain, a paradigm was designed which tested the excitability of a neuron after discrete periods of synaptic stimulation. A 500 ms depolarizing current pulse was first injected into each IC neuron, and the pulse amplitude was adjusted to elicit a single action potential. A train of 200 μs stimulation pulses was then delivered to the LL at 100Hz for 5s, a form of tetanic synaptic stimulation. Immediately after the synaptic stimulation, the original intracellular current pulse was again injected into the IC neuron. To follow the time course of changes in current-evoked action potentials, the current pulse was redelivered at 30s intervals, until the response returned to the pre-tetanic level (Figure 1A). No action potentials were elicited during the train of synaptic stimuli.
To facilitate comparison with earlier in vivo results (Sanes et al., 1998; Malone and Semple 2001), the synaptic stimulation was designed to approximate acoustic conditioning stimuli, which lasted 1 or more seconds. In the slice model, the stimulation rate of 100Hz approximates acoustically-driven rates of afferent fibers to the IC. A 5s duration was chosen based on preliminary findings that shorter stimulus durations were occasionally ineffective.
Although a range of current magnitudes could elicit a single action potential, we chose the minimum current necessary. By defining the current magnitude at this lower end of spike threshold, estimates of gain remained conservative. Any increase in the action potential number after the synaptic stimulation period would be less likely to fall within a range of response variability, and therefore more likely to reflect a true postsynaptic change. The stimulation paradigm was also designed to minimize response rundown during the course of the experiment. At the first sign of rundown, indicated by either depolarized resting potential or action potential broadening, data acquisition from that neuron was terminated. Thus, after each tetanic stimulation paradigm and subsequent recovery series of test pulses, the neuron was left unstimulated for 10–20 mins prior to retesting with a pharmacological agent.
Brain slices were bathed in 4 mM kynurenic acid (KYN) (Sigma) to block ionotropic glutamatergic receptors (Moore et al. 1998) and prevent over-stimulation of the neurons. Specific pharmacological agents were delivered in ACSF at the following concentrations: bicuculline (BIC), 30 μM (Sigma); strychnine (SN), 1 μM (Sigma); DNQX, 0.5 μM (Tocris); AP-5, 1 μM (Tocris); a-methyl-4-carboxyphenylglycine (MCPG), 1 μm (Tocris); SCH-50911, 20 μM (Tocris); atropine, 1 μM (Sigma); mecamylamine (Mec), 5 μM (Sigma); LY-278,584, 1 μm (Sigma); verapamil, 20 μM (Sigma); 2-methyl-5-hydroxytryptamine maleate (2-M-5HT), 30 μM (Sigma).
The responses during synaptic stimulation were measured and analyzed offline with the Igor Slice Analysis Macro (http://www.cns.nyu.edu/~sanes/slice_software/). As a neuron’s voltage response during afferent stimulation was obscured by the stimulation artifacts, these artifacts were removed (Figure 3). The artifacts at 1–3 ms after each stimulus onset were deleted, and the adjacent segments of the trace were joined. Changes in membrane potential (Vm) and the duration of these changes were measured directly from traces in Igor. Statistical comparisons were performed in JMP 5.0, using multivariate ANOVA, and Student’s t-test (alpha = 0.05). Unless otherwise noted, all values are reported as mean ± standard error of the mean (SE).
All protocols were reviewed and approved by the New York University Institutional Animal Care and Use Committee.
The stimulus paradigm was designed to assess a postsynaptic change in the response to current pulses, following short periods of synaptic activity. The initial design of the present in vitro experiments was motivated by the hypothesis that short term changes in gain of IC neurons was dependent on synaptic inhibition intrinsic to the IC Sanes et al., 1998). Therefore, all recordings were performed in the presence of KYN, which blocks ionotropic glutamatergic neurotransmission.
As illustrated in Figure 1A, a control current pulse was delivered at spike threshold, and was repeated 3 times to establish fidelity of the response (repetitions not shown). A train of stimuli was then delivered to the LL afferent pathway, followed by a series of current pulses to reassess postsynaptic gain. An elevated response was termed Positive Gain (Figure 1B). A subset of neurons displayed an elevated response only after a 30 s delay (Figure 1C). A minority of neurons tested (12%) exhibited a transient suppression of the current-evoked response (loss of action potential), which recovered within 30–250 s (n = 32), and was termed Negative Gain (Figure 1D). Approximately 20% of tested neurons displayed no change in the current-evoked response after the synaptic stimulation train (No Gain, n = 36).
In 66% of IC neurons studied (n = 164), the number of current-evoked action potentials increased immediately after the synaptic train. This value includes neurons that displayed an increased gain immediately or after a 30 s delay. Thus, a majority displayed Positive Gain (Figure 1E). This elevated response returned to baseline (i.e., one spike) within 30–250 s. For all neurons tested, the average change in number of current-evoked action potentials (Δ) was +1.8 ± 0.13 action potentials (Figure 1F), including Negative and No Gain effects.
Similar distributions of each gain effect were observed whether or not the neuron displayed an initial burst of action potentials (Figure 2A). To establish the discharge pattern of each neuron, the response to a range of 500 ms depolarizing current steps was obtained. No correlation was found between suprathreshold discharge pattern and gain type (chi-square likelihood ratio = 13.1, p > 0.05) (Figure 2B).
Neurons were also classified by their response to 10pA, 500ms hyperpolarizing current steps. While both rebound-spiking and delayed rectifier currents (sag current, or Ih) were present in the population of recorded IC neurons, no correlation was observed between these intrinsic membrane properties and gain category (chi-square, p > 0.05) (Figures 2C and D).
Recordings were made from IC neurons in a range of postnatal ages, to determine whether gain effects emerged during development. Interestingly, all types of gain were present around the time of hearing onset (P10–13), and the distribution of gain effects did not differ significantly from P14–20 neurons (pairwise t-test, p > 0.05).
Afferent stimulation artifacts were removed from each trace prior to analysis, as described in the Methods (Figure 3A). During the period of synaptic stimulation, membrane potential was typically displaced from rest. The majority of recorded neurons displayed a hyperpolarizing component (61%), whereas about one third (35%) displayed both hyperpolarizing and depolarizing components. The hyperpolarizing component appeared during at the outset of the stimulus train, and typically returned to the resting membrane potential before the end of the train.
To determine whether the magnitude and duration of hyperpolarizing and depolarizing segments were correlated with gain effect, analyses of deflections from Vrest (resting membrane potential) during the synaptic stimulation train were performed (Figure 3B). Both the maximum amplitude and duration of each component were measured and compared between 4 gain categories: No Gain, Negative Gain, Positive Gain of 1– 2 action potentials (1–2 APs), or Positive Gain of > 2 action potentials (>2 APs). The magnitude and duration of membrane depolarization were largest in the Positive Gain group (Figure 3C, grey bars). In contrast, the duration of hyperpolarization was longest for neurons that displayed Negative Gain (Figure 3C, black bars).
To determine whether the gain effect was dependent, on a depolarizing shift specific to synaptic stimulation, a tonic 5 pA current was injected during the 5s pre-test pulse period, in place of synaptic stimulation. Positive gain neurons (n = 3) were allowed to recover for >20 mins, and the depolarizing current injection alone was delivered. As illustrated in Figure 4A, this stimulus did not change the number of action potentials evoked by the subsequent current pulse for any of the cells tested. As expected, when a tonic current injection persisted beyond the synaptic stimulation period, and throughout the current test pulse, there was an increase in the evoked response (Figure 4B).
For both the Positive and Negative gain groups, the mean recovery time was 57s. The recovery of Positive Gain varied widely (80 ± 57 s), and was significantly longer than that of the Negative Gain (40 ± 29 s) (ANOVA, F = 21.4, r = 0.527, p < 0.001). No significant correlations were found between age groups and recovery time (ANOVA, p > 0.05).
We examined the role of several neurotransmitter systems with receptor-specific antagonists (Table 1). To determine which synaptic afferents were involved in the recruitment of the Positive Gain effect, we first focused on three major ionotropic systems known to be present in the LL input pathway: GABA, glycine, and glutamate. Previous studies in vivo using focal application of GABA and glycine in the IC showed that single unit responses were briefly elevated following the cessation of a pulse of inhibitory neurotransmitter (Sanes et al., 1998). This post-inhibitory enhancement lasts up to 10 s, and implicates a GABAergic or glycinergic receptor-mediated mechanism. In the IC brain slice, blockade of inhibition with BIC and SN did not prevent the positive gain after synaptic stimulation (n = 8). In fact, this manipulation enhanced the positive gain effect observed prior to inhibitory antagonist treatment in 5 of the neurons tested (Figure 5). Blockade of metabotropic GABAB receptors with the antagonist, SCH-50911 (20 μM), was also ineffective at reducing positive gain (n = 4).
Since ionotropic glutamate receptors were blocked in all experiments, they played no role in this effect. However, as an internal control, KYN was replaced by specific antagonists to AMPA and NMDA receptors (CNQX and APV). No change occurred in the gain effect under these receptor-specific manipulations (n = 3). Addition of MCPG (1 μM), a group I and II metabotropic glutamate receptor antagonist, also failed to block the positive gain (n = 5).
To test whether cholinergic pathways were involved, experiments were performed in the presence of either a nicotinic antagonist (5 μM mecamylamine) or a muscarinic antagonist (1 μM atropine) (Figure 6). Neither of these agents was able to block the synaptically-induced increase in action potentials. In fact, each of these antagonists enhanced the gain effect in 5 of 7 neurons tested.
To determine whether the LL-evoked synaptic response in IC neurons contained a serotonergic component, a 5-HT3 receptor antagonist (LY-278, 584 at 1 μM) was added to the ACSF (KYN was not present). The LL-evoked EPSP amplitude decreased following application of LY-278,584, indicating the presence of ascending serotonergic afferents to the IC (n = 3) (Figure 7A). Furthermore, in the absence of any synaptic stimulation, bath application of a 5-HT3 agonist, 2-M-5HT (30 μM) enhanced the current-evoked discharge compared to control pulses (Figure 7B).
To determine whether LL-evoked serotonergic transmission initiates the gain effect, slices were bathed in LY-278, 584 prior to delivery of a 5 s afferent stimulation train. This 5-HT3 antagonist led to a marked decrease or elimination of positive gain in 6 of 8 IC neurons tested (Figure 8). Furthermore, the drug effect was reversible after a 30 min washout interval. Unlike all other pharmacological agents tested, the 5-HT3 antagonist reduced the steady depolarization that occurred during the synaptic stimulation train (Figure 8, bottom).
The 5-HT3 receptor-coupled channel is permeable to Na+ and Ca2+ ions (Barnes and Sharp, 1999), so other mechanisms that permit Ca2+ influx could also facilitate the positive gain effect. To examine the role of Ca2+, experiments were performed in a low Ca2+ ACSF (n = 3). Synaptic activation was present in 0.24 mM Ca2+ ACSF, but there was a reduction in the positive gain in each neuron tested (Figure 9A). Since this result cannot rule out a possible reduction in presynaptic release, the involvement of Ca2+ channels during synaptic stimulation was examined directly, with the L-channel blocker, verapamil (20 μM). In 3 of 5 neurons, verapamil reduced the duration of the positive gain effect; recovery time to pre-stimulation levels was shortened (Figure 9B). In two neurons, the positive gain was completely eliminated. For the three cells in which LY-278, 584 did not eliminate the positive gain, additional co-application of verapamil was able to eliminate the gain response. Thus, verapamil attenuated some aspect of the positive gain effect each time it was applied. The complementary effects of these two drugs are summarized in Figure 9C.
The amplitude and duration of depolarization from Vrest that occurs during the synaptic train (Figure 3), was examined for neurons treated with verapamil and LY. Direct measurements of gain in number of action potentials were also compared. Only verapamil and LY showed a consistent reduction in all of these parameters (Table 1).
Previous studies demonstrated that the neural response to an acoustic cue depends on the recent history of stimulation (Malone and Semple 2001; McAlpine et al. 2000; Spitzer and Semple 1993, 1998). A period of stimulation with a specific sound level, frequency, interaural phase or level can enhance the subsequent discharge rate of an IC neuron when any of these parameters are varied. More importantly, the enhanced discharge rate can persist for several seconds. Since persistent enhancement of discharge rate could influence the representation of all ongoing auditory stimuli, such as speech, the physiological basis for this short-term plasticity will be fundamental to our understanding of IC coding.
Investigation of in vitro mechanisms that support acoustic conditioning led to the discovery of a novel form of short-term plasticity in IC neurons. After short periods of synaptic stimulation, there was an increase in the number of action potentials evoked by a current pulse, indicating a temporary change in postsynaptic excitability. This positive gain was observed in the majority of recorded neurons (66%), although negative gain did exist in a minority (12%) of neurons (Figure 1E), and a minority displayed no change in gain (22%). Subsequent analysis showed that the positive gain effect is associated with a transient depolarization elicited by the synaptic train. These positive gain effects did not depend on the presence of either an outward cation current (Ih) that rectifies large negative changes from Vrest, or rebound action potentials (Figure 2). Therefore, while certain intrinsic properties could add to the positive gain effect, they are not necessary for initiating it.
Periods of hyperpolarization and depolarization that occurred during the synaptic train suggested that a postsynaptic mechanism could underlie positive gain effects. In most cases, the magnitude of depolarization did predict a positive gain effect (Figure 3C). While the hyperpolarizing forces are at least partly attributable to GABA- and glycinergic transmission, the source of the depolarization is not ionotropic glutamatergic transmission, because KYN was present throughout the experiments. Furthermore, current injection controls (Figure 4A) indicate that depolarization during the synaptic stimulation period is not sufficient to elicit a Positive Gain. Rather, this depolarization must be recruited by a seconds-long synaptic drive, and outlast the synaptic activation period. Thus, a persistent depolarization, when coupled with the depolarizing pulse, does lead to a greater output (Figure 4B).
The lateral lemniscus is the primary ascending pathway to the IC and contains axons from many brainstem nuclei (Adams 1979; Coleman and Clerici 1987; Faingold 1991; Oliver 1991). Pharmacological manipulation of postsynaptic responses in the IC have also demonstrated the presence of a range of neurotransmitter systems, including serotonin (Faingold 1991). Given this diversity, stimulation of the LL pathway at the ventral aspect of the IC could activate several neurotransmitter systems. Therefore, the chemical basis of positive gain effects were assessed with neurotransmitter antagonists. \
All non-serotonin antagonists tested in this study were unable to block the positive gain effect. In contrast, incubation with BIC and SN led an increase in positive gain, indicating that GABA and glycinergic systems may serve to dampen the increase in gain observed following afferent stimulation (Figure 5). Similarly, blockade of muscarinic and nicotinic acetylcholine receptors facilitated the positive gain effect. This result can be explained if the site of activation for acetylcholine in the IC is primarily on GABAergic synaptic terminals. Indeed, activation of muscarinic receptors upregulates the frequency of spontaneous GABAergic postsynaptic currents in the IC (Yigit et al 2003), so it is likely that the increase in positive gain effects with atropine is due to down regulation of synaptic inhibition. Moreover, in vivo iontophoretic application of nicotinic antagonists increases discharge rates in IC neurons (Habbicht and Vater 1996). As nicotinic receptors are known to gate calcium and facilitate postsynaptic depolarization, it is possible that their site of activation is also inhibitory synaptic terminals. Therefore, we cannot exclude disynaptic effects within the IC.
The IC contains a heterogeneous neuronal population, and the variability of pharmacological effects could reflect intrinsic differences among the recorded neurons, although each tested neuron displayed repetitive firing (Figure 2). Since positive gain did occur in neurons that did not display rebound potentials, it is unlikely that our findings are due solely to Ca2+-dependent rebound depolarization or anode-break excitation (Sivaramakrishnan and Oliver, 2001). The variability of the pharmacological effects may be partially explained by variability in membrane characteristics, such as neurotransmitter receptor and Ca2+ channel density. Despite the similarity of results for animals of different ages, we must acknowledge that all recordings were obtained during a relatively early period of development, and many synaptic and biophysical properties do not yet display mature characteristics in the auditory central nervous system (Sanes, 1993; Balakrishnan et al., 2003: Scott et al., 2005; Vale and Sanes, 2000; Oswald and Reyes, 2008). Therefore, it is possible that the 5HT3R-dependent mechanism described above may display different kinetics in adults.
Serotonergic fibers in the IC originate primarily from the dorsal raphe, and follow the LL afferent pathway to both the external cortex and central nucleus, where they display a wide distribution (Klepper and Herbert 1991; Thompson et al. 1994). The existence of serotonin at synaptic sites is also supported by the morphological structure of serotonin-immunoreactive varicosities in the neuropil, some of which are apposed closely to IC cell bodies (Hurley and Thompson 2001; Hurley et al. 2002). Indeed, serotonin fibers are found throughout the rodent SOC nuclei (Thompson and Hurley 2004) and are known to modulate synaptic transmission in the LSO (Fitzgerald and Sanes 1999).
A functional influence of serotonin on action potential generation in the brain slice preparation is consistent with previous observations in vivo. Local iontophoresis of serotonin into the IC can facilitate or depress discharge rates in both spontaneous firing and acoustically driven responses to modulated sounds (Hurley and Pollak 1999). Even more interesting is the observation that local serotonin iontophoresis can increase the duration of responses to FM sweeps up to three-fold, indicating a post-stimulus enhancement in excitability which outlasts acoustic stimulation (Hurley and Pollak 1999). Serotonin iontophoresis is also more likely to enhance neural responses to FM sweeps than to pure tones, in a single neuron. This observation suggests serotonin preferentially affects acoustic signals that change over time.
Recent studies that pharmacologically-controlled serotonergic receptor activation and serotonin release in the IC have also demonstrated a role for serotonin in auditory coding (Hurley 2006; Hurley 2007; Hall and Hurley 2007; Hurley et al 2008). Thus, endogenous serotonin is present in the intact IC, and is likely to modulate acoustic processing. In step with these in vivo findings, our result that serotonin contributes to a portion of the evoked EPSP after LL stimulation is the first in vitro evidence of postsynaptic depolarization by serotonin in IC neurons. In this issue, a companion study of IC neurons in vivo reports enhancements in discharge rate following 5-HT3R activation (Bohorquez and Hurley, co-submitted with this manuscript). Indeed, our report that 5-HT3R activation exerts a depolarizing effect in vitro is consistent with this result. Our focus on gain changes near action potential threshold demonstrated that this depolarizing influence could change the input/output ratio of an IC neuron near rest. Interestingly, the in vivo study also reported a depressive effect for 5-HT3R at high firing rates. While we did not observe any depressive effects of 5-HT3R activation in vitro, this is likely because we examined postsynaptic gain at spike threshold, the lowest firing rate of a neuron. Furthermore, our pharmacological testing was restricted to neurons that could demonstrate the positive gain phenomenon, whereas the in vivo study, which demonstrated a variety of pharmacological effects, had no cell selection criteria. Therefore, it is not surprising that we observed less diverse effects than the in vivo study. An intriguing trend among these in vivo results was that, as the spike count gets higher, the effect of the 5-HT3R agonist moves to less facilitation/more depression (Bohorquez and Hurley, co-submitted with this manuscript). Thus, our finding is consistent with pharmacological effects found for lower firing rates in vivo. Despite the limitations in comparing in vitro with in vivo studies, our result of depolarizing 5HT3R activation leading to increased discharge rate supports the observation of enhancements in spike rate following pharmacological 5-HT3R activation in vivo.
The gain effect in IC neurons is remarkably similar to findings obtained in turtle spinal motorneurons. In this system, a short synaptic train elicits a transient period of excitability, also measured as an increase in current-evoked action potentials (Hounsgaard and Kiehn 1989). This temporary elevation in gain lasts for tens of seconds to minutes, and is attributed to an L-type Ca2+ plateau potential. Furthermore, this synaptically-evoked Ca2+ current is regulated by serotonin, and is independent of ionotropic glutamatergic and glycinergic transmission (Delgado-Lezama et al. 1999; Delgado-Lezama et al. 1997). The same transient states of excitability were also found in cat motorneurons, and were also regulated by serotonin (Hounsgaard et al. 1988).
The prominent reduction of excitability caused by the 5-HT3 antagonist LY-278,584 strongly suggests a role for serotonin via its sole ionotropic receptor. This receptor is known to gate a fast-activating, nonselective cation channel, and cause rapid depolarization, making it a likely contributor to the net depolarization we observe in IC neurons (Barnes and Sharp 1999; Davies et al. 1999; Dubin et al. 1999). Consistent with this observation, the 5-HT3 agonist (2M5HT) increases membrane excitability enough to influence action potential generation (Figure 7B). Indeed, the Vm depolarized 4.7 mV immediately after application of this 5-HT3 agonist. Therefore, the contribution of serotonin via this ionotropic receptor may have a significant effect on discharge rate coding in the IC. Taken together, these results indicate that direct activation of 5-HT3 receptors depolarizes the neuron enough to increase discharge rate.
5-HT3 receptor-mediated excitability has been observed in other neural systems. Iontophoresis of a 5-HT3 agonist increases mean discharge rate of amygdala neurons in vivo. In mouse neuroblastoma cells containing a native form of the 5-HT3 receptor, local application of serotonin elicits depolarizing currents of monovalent and divalent cations, and increases [Ca2+]in (Stewart et al. 2003). Also, contrary to the expectation that neuromodulator receptors are primarily extrasynaptic, electron micrograph studies of the hippocampus show postsynaptic 5-HT3 receptor labeling preferentially localizes to the postsynaptic density (Miquel et al. 2002). Such a postsynaptic location suggests this receptor plays an active role in dendritic neurotransmission. Indeed, a survey of 5-HT3 receptor antibody staining in the rat brain shows immunopositive cell bodies in the IC. (Morales et al. 1998). To date, there is no further anatomical evidence of 5-HT3 receptor distribution in the IC. However, anatomical and physiological examination of 5-HT1 and 5-HT2 subtypes indicates a growing relevance to both auditory processing and the neural basis of anxiety (Castilho et al. 1999; Peruzzi and Dut 2004).
The duration of synaptically-elicited positive gain in IC neurons cannot be fully explained with a fast ionotropic receptor, such as 5-HT3. The elevated response over tens of seconds is more likely due to a postsynaptic effect of Ca2+. An intracellular rise in Ca2+ could be elicited by 5-HT3 receptor activity or could independently result from other signals that depolarize the membrane. Our results with verapamil suggest that voltage-gated Ca2+ currents in IC neurons are involved in the enhanced excitability to postsynaptic current pulses. The membrane depolarization that is elicited by the tetanic synaptic stimulation is significantly reduced by verapamil (Figure 9B), showing a possible L-type channel contribution to this response. Sustained L-type voltage activated Ca2+ currents in IC neurons have been described in approximately 63% of IC cells (N’Gouemo and Morad 2003). Furthermore, L-type currents in these IC neurons displayed a slow voltage-dependent inactivation, causing them to last up to several minutes. Such a time course could account for the recovery periods observed in our experiments.
The role of Ca2+ in the long recovery of IC neurons in our stimulation paradigm could also reflect the presence of a plateau potential. An example of brief periods of synaptic drive causing elevated and sustained spiking activity in the entorhinal cortex shows that Ca2+ plateaus underlie enhanced responses to successive stimuli (Egorov et al. 2002). Similarly, serotonin application in ferret thalamus slice can cause minutes-long depolarizations, which were sufficient to elicit action potentials (Monckton and McCormick 2002). Even more pertinent are results which demonstrate that serotonin application facilitates long, slow L-type Ca2+ plateau potentials, which increase the spike discharge rate to subsequent injections of current in spinal motorneurons (Perrier et al. 2002). While these plateaus are triggered by the activation of metabotropic serotonin receptors, it is possible that the positive gain effects we see involve serotonin and Ca2+ in a similar way. However, activation of a plateau potential would require depolarizations much greater than the typical 5mV deflections seen in the current study. We have not ruled out the possibility of metabotropic serotonin involvement during our synaptic stimulation paradigm.
Meaningful in vivo - in vitro comparisons have been made in the visual system, wherein a slow potassium current observed in the slice is proposed as the basis for in vivo contrast adaptation, a temporary gain change in V1 neurons (Sanchez-Vives et al. 2000). In this case, a seconds-long subthreshold event directly impacts the discharge rate of a single neuron, and ultimately changes the neural coding of subsequent contrast stimuli. Similar to the positive gain phenomenon we see IC neurons, the mechanisms elucidated in the V1 slice experiments are also longer-lasting than those observed in vivo. This is an encouraging case of in vivo experiments motivating in vitro studies of cellular mechanisms underlying sensory function. Nonetheless, the IC slice model does fall short of direct comparison with the in vivo phenomenon of dynamic conditioning in several ways. Stimulation of the LL in vitro causes a broad activation of afferents, whereas acoustic stimulation presumably stimulates specific combinations of afferents. Also, the in vitro stimulation is always performed with ionotropic glutamate blocked (in KYN), making this broad afferent activation somewhat restricted. As ionotropic glutamate is known to influence the accumulation of postsynaptic calcium, the in vitro stimulation is likely to under-recruit postsynaptic calcium. Thus, the synaptic stimulation period in vitro does not have a specific acoustic equivalent. Furthermore, the in vitro model does not specifically address presynaptic mechanisms that may contribute to the in vivo phenomenon. Adaptation of afferents may influence postsynaptic gain effects during both dynamic conditioning and tetanic synaptic stimulation, however the contribution of afferents was not assessed with either stimulation paradigm.
The prevalence of positive gain effects in IC neurons suggests a possible effect on many aspects of auditory processing. Although the kinetics of the slice are slow compared to in vivo conditions, our findings may point to the cellular bases of sensitivity to stimulus order and context in IC neurons, particularly on the time scale of 100s of ms. Stimulus context has proven important for neural coding in the IC, as shown with studies of increased sensitivity to gap detection in background noise, reflecting a form of IC gain control (Wilson and Walton 2002), or the temporary enhancement of spike rate depending on the interval between two sequential tones (Finlayson, 1999). Indeed, temporal order of sound stimuli does affect our understanding of speech components (Holt and Lotto 2002). Furthermore, speech recognition appears most dependent on low frequency components rather than complex spectral components (Shannon et al. 1995). Such low frequencies occupy temporal periods that would definitely be affected by neurophysiological changes on the order of 100s of ms, such as those described herein.
Changes on an intermediate time scale (100s of ms) are likely to have an important impact on auditory perception in the typical acoustic environment (e.g., reverberating). For example, human subjects report the apparent target location of a moving sound to be displaced in the direction of motion (Mateeff and Hohnsbein 1988; Perrott and Musicant 1981, 1977). Because the IC appears to integrate synaptic input over longer time intervals than lower brainstem auditory nuclei, it is a prime candidate for studies of short-term plasticity of coding properties.
The authors wish to acknowledge the technical and intellectual contributions of Dr. Vibhakar Kotak during the course of this study. Supported by DC006864 and DC05455.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.