Excitatory action of an immature glycinergic/GABAergic sound localization pathway

Abstract

Most mammals determine the azimuthal direction of incoming sound using auditory cues arising from differences in interaural sound intensity. The first station in the ascending auditory pathway, which processes interaural intensity differences, is the lateral superior olive (LSO), a binaural nucleus in the auditory brainstem. LSO neurons encode interaural intensity differences by integrating excitatory input from the ipsilateral cochlea and inhibitory input from the contralateral cochlea. Both inputs converge on single neurons in a highly organized, frequency-specific manner. The correct development of the precise arrangement of these inputs and their physiological properties depends on neuronal activity. Previous studies have shown that inhibitory, glycinergic/GABAergic inputs to the LSO are transiently depolarizing, and it has been hypothesized that this depolarizing action enables developing inhibitory inputs to act as excitatory inputs. In support of this hypothesis, we recently demonstrated that depolarizing glycinergic/GABAergic inputs can increase the intracellular calcium concentration in immature LSO neurons and elicit action potentials. These results provide support for the notion that the influence of glycinergic/GABAergic synaptic activity on development of the LSO involves calcium-dependent signaling mechanisms.

Introduction

Fast and accurate sound localization is one of the basic functions of the auditory system. The ability to quickly determine the direction of incoming sound is essential for guiding many basic behaviors such as finding food sources, avoiding predators, and locating mating partners. To determine the direction of incoming sound, the auditory system has to extract both the elevation and azimuthal direction of the auditory stimulus. Many mammals, including humans, determine sound elevation by using elevation-dependent changes in the spectral cues that arise from the acoustic filter characteristics of the pinna [1]. Sound localization in the azimuthal plane is achieved by using interaural differences in the arrival time and the intensity of sound, both of which vary systematically with the direction of incoming sound.

Perhaps due to the fundamental importance of sound localization, interaural time and intensity differences are processed at the earliest level possible, i.e., as soon as auditory information from both ears converges on single neurons. Two binaural nuclei in the superior olivary complex of the brainstem are particularly specialized for extracting interaural sound differences: Neurons in the medial superior olive (MSO) extract interaural time differences by acting as coincidence detectors, which are tuned to specific interaural time delays [2]. In contrast, neurons in the lateral superior olive (LSO) encode interaural intensity differences by integrating excitatory inputs from the ipsilateral cochlea and inhibitory input from the contralateral cochlea [3], [4]. LSO neurons receive direct excitatory, glutamatergic inputs from the ipsilateral cochlear nucleus (CN) and indirect inhibitory, glycinergic input from the contralateral cochlear nucleus via a relay nucleus, the medial nucleus of the trapezoid body (MNTB) [5] (Fig. 1). Both inputs are tonotopically organized and converge on single LSO neurons in a frequency-specific manner, which enables LSO neurons to integrate binaural auditory information that arises from the same sound source.

In the past years, several laboratories have examined the developmental events and mechanisms by which the precise anatomical and physiological organization of the LSO becomes established in order to fulfill its function in the mature brain. These studies revealed that the developing LSO circuit undergoes numerous anatomical, physiological, and molecular changes. These events have been summarized in a recent review to which the reader is referred for a detailed overview [6]. In accordance with the spirit of this supplemental issue, we will briefly summarize some aspects of LSO development, and then recapitulate some recent experiments from our own laboratory in which we investigated the effects of glycinergic/GABAergic synaptic activity in immature LSO neurons.

In rats, the LSO becomes innervated around embryonic day 17 (E17) when the first collaterals from the CN axons start to invade the superior olivary complex [7]. Once afferent fibers have reached the LSO, functional synapses form almost immediately, as indicated by the fact that electrical stimulation of afferent axons from the CN or MNTB elicits postsynaptic potentials in LSO neurons [8], [9]. Thus, LSO neurons receive functional synaptic inputs at least 2 weeks prior to the onset of hearing airborne sound which, in rats and gerbils, begins at around P12–P14 with.

Anatomical tract tracing studies have indicated that these early connections to the LSO are already highly ordered and topographically organized. For example, axons from the CN terminate in topographically appropriate areas of the LSO [7], [10], and single MNTB axons branch in restricted areas in topographically correct LSO locations [11]. The presence of this topographic organization of LSO afferents from the earliest stages on, together with the fact that regenerating MNTB axons innervate topographically correct areas in the LSO independent of spike activity or glycinergic synaptic transmission [12], is consistent with the idea that the topographic organization of auditory pathways is primarily determined by chemical cues [13]. More detailed studies on a single cell level, however, have indicated that some of these early connections are further refined by selective pruning of axonal terminals and dendrite branches. In a series of publications, Sanes and coworkers demonstrated that individual MNTB axon terminals and LSO dendrites gradually become more restricted to smaller areas along the tonotopic axis in the LSO (reviewed in Ref. [6]). Furthermore, this topographic sharpening of the MNTB-LSO pathway appears to depend on synaptic activity, as refinement of both axonal and dendritic arbors are impaired if neuronal activity levels or patterns are disrupted by cochlear ablation or by blocking glycine receptors.

In addition to its role in sharpening anatomical topography, synaptic activity in the MNTB-LSO pathway is also critical for the maturation and/or maintenance of physiological properties of LSO synapses. For example, experimental disruption of spontaneous synaptic activity in the MNTB-LSO pathway results in the rapid, functional disconnection of many MNTB inputs and in a weakening of the remaining ones [14]. Interestingly, disruption of synaptic activity in the MNTB-LSO pathways not only changes the glycinergic/GABAergic synapses in this pathway, but also changes the physiological properties of glutamatergic synapses that contact LSO neurons [15].

However, despite the well-documented role of neuronal activity in spatial and physiological fine-tuning of LSO circuitry, the exact cellular mechanisms by which glycinergic/GABAergic synaptic activity exerts its effect in the developing LSO are still poorly understood. Based on the central role of intracellular calcium signaling in the regulation of dendritic and axonal growth, synaptic plasticity, and cell death [15], [16], it seems plausible that calcium-signaling mechanisms also mediate at least some of the effects of glycinergic/GABAergic activity on LSO development. Glycinergic/GABAergic MNTB-LSO synapses can undergo activity-dependent synaptic plasticity in the form of long-term depression (LTD) [17], and this LTD depends on intracellular calcium signaling, perhaps mediated by activation of neurotrophin tyrosine kinase receptors [18].

Another means by which MNTB-LSO synapses could have access to intracellular calcium signaling is through their transient depolarizing action that exists in neonatal animals. As in many other brain areas, glycinergic and GABAergic synapses in the immature LSO are depolarizing instead of hyperpolarizing [9]. In the LSO, this depolarizing action results from a high intracellular chloride concentration, which causes an efflux of negative chloride ions through glycine and GABA_A receptors [19]. In several neuronal systems, depolarizing GABA and glycine act as excitatory neurotransmitters in the sense that they can trigger postsynaptic action potentials and increase [Ca²⁺]_i [20], [21]. On the other hand, there is also compelling evidence that glycine and GABA act inhibitory even if they are depolarizing [22], [23].

The inhibitory action of depolarizing GABA has been particularly well characterized in the chick nucleus magnocellularis, the avian homologue of the mammalian CN [24], [25]. In this nucleus, the reversal potential of GABA is more positive than the spiking threshold, yet GABAergic depolarizations remain subthreshold and inhibit action potential generation. This seemingly paradoxical action is accomplished by a combination of three effects: activation of a low-threshold, voltage-gated K⁺-conductance that prevents GABAergic depolarizations from reaching spike threshold, a strong decrease in membrane input resistance that effectively shunts excitatory synaptic currents, and the accommodation of spike thresholds due to the long duration of GABAergic depolarizations. Inhibitory actions of depolarizing glycine and GABA have also been observed on the level of calcium responses. For example, in the inferior colliculus of neonatal gerbils, bath application of GABA or glycine increases resting [Ca²⁺]_i modestly, but when applied during glutamatergic stimulation, GABA or glycine effectively reduces glutamate-generated calcium responses [26]. Together, the above examples illustrate that the physiological actions of depolarizing glycine and GABA greatly depend on the cell type and physiological context during which inhibitory synapses are activated. Thus, it is very difficult to make generalizations and predictions concerning whether depolarizing glycine or GABA will have an excitatory or inhibitory action in a given cell type and age.

We therefore decided to directly examine whether activity in the glycinergic/GABAergic MNTB-LSO pathway can increase postsynaptic [Ca²⁺]_i in immature LSO neurons [27]. To this end, we applied calcium-imaging techniques to slices from rat and mouse brainstem, which had been labeled with the calcium-sensitive dye Fura-2. In slices from neonatal animals, direct stimulation of glycine and GABA_A receptors elicited robust increases in [Ca²⁺]_i in LSO neurons (Fig. 2). These responses persisted when spike-dependent activity was blocked by tetrodotoxin, but they were greatly reduced or abolished by strychnine, a glycine receptor antagonist, or bicuculline, a GABA_A-receptor antagonist. These results demonstrate that activation of glycine and GABA_A receptors in immature LSO neurons increases [Ca²⁺]_i. Because glycine and GABA are depolarizing in neonatal animals, it is likely that these calcium responses were caused by activation of voltage-gated calcium channels and an influx of extracellular calcium. To test this hypothesis, we applied glycine and GABA in the absence of calcium in the bath medium or in the presence of 5 mM nickel, a voltage-dependent calcium channel blocker at these concentrations. Under both conditions, calcium responses were abolished or greatly reduced. In addition, glycine or GABA increased [Ca²⁺]_i only during the first postnatal week, when the MNTB-LSO pathway is depolarizing. In two-week old animals, when MNTB-LSO synapses have switched to being hyperpolarizing, glycine and GABA failed to increase [Ca²⁺]_i but instead consistently decreased [Ca²⁺]_i (Fig. 3). Taken together, these results demonstrate that in neonatal animals, glycine and GABA elicit calcium responses via depolarization and activation of voltage-gated calcium channels.

Previous studies have shown that strong or prolonged activation of glycine and GABA receptors can cause the breakdown of the transmembrane chloride gradient and produce depolarizations that are generated by an efflux of bicarbonate [28]. Therefore, calcium responses obtained by prolonged bath application of glycine or GABA agonists might not necessarily reflect responses that occur after physiological stimulation of synaptic glycine and GABA receptors. To address this issue, we examined whether stimulation of the MNTB-LSO pathway elicits similar responses to those obtained by pharmacological stimulation. In these experiments, we stimulated MNTB neurons selectively by rapid, focal photolysis of caged glutamate, in order to avoid the activation of fibers en passage [29]. Uncaging of glutamate inside the boundaries of the MNTB reliably increased [Ca²⁺]_i in LSO neurons (Fig. 4). These synaptic calcium responses were partly or completely blocked by bicuculline or strychnine, indicating that they were mediated either by glycine, GABA, or both. These results provide evidence that synaptic activity in the MNTB-LSO pathway in neonatal rats elicits an increase in [Ca²⁺]_i in LSO neurons.

In addition to eliciting calcium responses, exogenous glycine and glycinergic/GABAergic synapses are also excitatory in neonatal LSO neurons in respect to their ability to elicit action potentials. Iontophoretic application of glycine to cell bodies of LSO neurons of neonatal rats [19] as well as electrical stimulation of the MNTB-LSO pathway in neonatal mice [30] produce spikes as long as the chloride reversal potentials are more positive than spiking thresholds.

Taken together, several lines of evidence support the idea that the glycinergic/GABAergic MNTB-LSO pathway in the developing auditory brainstem acts as an excitatory pathway. A major, but yet unresolved, question is what role this excitatory action plays in the maturation and functional specialization of the LSO. Interestingly, glycinergic and GABAergic excitation occurs during the period when the survival of LSO neurons critically depends on elevated levels of [Ca²⁺]_i [31] and when LSO neurons undergo considerable growth [32]. This raises the possibility that calcium responses elicited by spontaneous activity in the MNTB-LSO pathway plays a role in the survival of LSO neurons and their morphological fine-tuning. Recent data from our laboratory also indicate that functional connectivity of the MNTB-LSO pathway is extensively refined [33] as long as the MNTB-LSO pathway is excitatory, suggesting that the transient excitatory action of the MNTB-LSO pathway is important for the establishment of the precise topographic organization of this inhibitory pathway. Because excitatory actions of inhibitory neurotransmitters are a widespread phenomenon in immature neuronal circuits, understanding of the functional significance of the excitatory action of the MNTB-LSO pathway could provide valuable insight into the general mechanisms by which inhibitory circuits become functionally optimized in the developing brain.