INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Parallel organization of synaptic signaling pathways is a prominent feature of mammalian auditory thalamus where acoustic relay is carried out by segregated lemniscal and nonlemniscal relay neurons (Imig and Morel 1983; Winer 1992). The former is located ventrally (MGv) in the medial geniculate body (MGB). They receive ascending input from the central nucleus of the inferior colliculus (IC) and project to the primary auditory cortex. The nonlemniscal neurons are situated in the dorsal (MGd), medial (MGm) and shell nuclei of the MGB. They receive input from the IC as well as other brain regions (Winer 1992) and provide a substantial and direct projection to the lateral amygdala nuclei (Doron and LeDoux 2000). Nonlemniscal pathway is critically involved in auditory associative learning and memory (Komura et al. 2001; LeDoux 2000). Previous in vitro studies have shown that MGd neurons tend to discharge in bursts in responding to sensory afferent stimulation, whereas MGv neurons fire mostly single or double spikes (Hu 1995). Such a pathway-specific allocation of burst neurons observed in vitro is not a trivial epiphenomenon. Burst or high-frequency firing not only enhances sensory event detection but also can modulate synaptic plasticity in their target neurons (Huang et al. 2000; Mooney and Hu 2002; Sherman 2001). In this study, we examined whether burst firing can be evoked via physiological acoustic stimuli in vivo and whether bursting neurons are preferentially associated with nonlemniscal nuclei.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Preparations

The methodology used in this paper has been previously described in detail (He 2001) and approved by the Animal Subjects Ethics Committee of the Hong Kong Polytechnic University. Briefly, guinea pigs (400-627 g) with clean external ears and normal auditory thresholds were anesthetized with ketamine/xylazine (initially 40 and 10 mg/kg im, supplemented at 10 and 2.5 mg · kg¹ · h¹ im). After mounted to a stereotaxic frame with the right ear freed from the ear bar, a craniotomy was performed to allow vertical access to the left MGB.

Acoustic stimuli, generated by a MALab system (Kaiser Instruments, Irvine, CA), were delivered via a dynamic earphone (Bayer DT-48) mounted in a probe in a double-walled sound-proof room (NAP, Clayton, Australia). The sound pressure level (SPL) was calibrated over a frequency range of 100 Hz to 35 kHz via a condenser microphone (Brüel and Kjær 1/4-in). Repeated noise bursts (100-ms duration) with 1 s or longer interstimulus intervals and a 5-ms rise/fall time were used. Repeated pure tones were used to characterize the best frequencies of the recorded neurons.

Recording

Platinum or tungsten microelectrodes with impedance of 9-12 M (Frederick Haer & Co.) were advanced by a stepping-motor micro-drive according to atlas (Rapisarda and Bacchelli 1977). A single electrode was used for each experiment so that the depth coordinates could be kept consistent for different penetrations and used for physiological map reconstruction based on a single lesion at the last penetration (He 2001). Single units were isolated via high-impedance recording electrodes and further confirmed on-line by means of their waveforms displayed on a digital oscilloscope. Only those units that exhibited stable waveforms throughout a trial session (typically <7 min) were recorded using an amplitude and time window-discriminator (He 2001)and displayed as raster or rate meter graphs for on-line monitoring and off-line analysis.

Histology

The animals were deeply anesthetized and transcardially perfused with conventional fixative. The brains were removed and stored overnight in 0.1 M phosphate buffer containing 30% sucrose. Transversal sections (40 µm thick) were cut on a freezing microtome and Nissl stained. To compensate tissue shrinkage, the Nissl images were symmetrically enlarged by 1013% during reconstruction of the physiology maps (He 2001). Nonparametric and Fisher exact tests were used to examine statistical significance. Data are expressed as means ± SE.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Lemniscal and nonlemniscal MGB neurons exhibited varying degrees of spontaneous activity, some of which appeared as rhythmic high-frequency bursts resembling spindle oscillations (Steriade and Llinas 1988). The following criteria were then adopted in the selection of responsive neurons. 1) The average spike counts (20 trials) within the first 100 ms of a peristimulus histogram must be three times higher than that in the rest of the histogram (900 ms). 2) An evoked burst response must have multiple spikes (2-6) occurring within a 15-ms time period, and a bursting neuron must show burst responses in 50% of the trials. 3) A single spike neuron must show no more than two spikes during a 15-ms poststimulus window in 75% trials. A lower ratio was adopted for bursting cells because they tended to have a higher rate of response failures (see following text) than single-spike neurons. Furthermore, we found that <2% of neurons in our study showed frequent two-spike responses and none of which could be classified as burst neuron because they did not appear in >50% trials together with high-frequency bursts. Hence, neurons with frequent two-spike responses were all in the single or nonbursting cell category. Based on the preceding criteria, our database included 63 bursting neurons and 60 single-spike cells.

Figure 1 illustrates typical bursting versus single-spike neurons recorded from the MGv and several subregions of the nonlemniscal shell (He 2001). Data pooled from two animals showed that ~77% or 33 of 43 nonlemniscal neurons were bursting cells (Fig. 1D, a-d); in contrast, only 17% or 5 of 29 lemniscal neurons were found to discharge high-frequency bursts (P < 0.001). An inverse distribution was, however, observed for single spike cells (Fig. 1D, h and i), which were observed in 83% MGv cells and 23% nonlemniscal neurons (P < 0.001). Bursting neurons often co-localized with off-responding cells (He 2001) and exhibited longer on-response latencies (see following text).

View larger version (49K): [in this window] [in a new window]   Fig. 1. Distribution of burst and nonburst neurons in the medial geniculate body (MGB). The recording sites (A-C) and pattern of spike activities (D) from 9 representative neurons from a single animal are shown. The penetrations coordinates are rostrocaudal (RC) and mediolateral (ML). A frontal map indicating ON-OFF responding neurons is superimposed on the Nissl section at the same level. Note the "sudden" response transition from burst to single spike between dorsal and ventral MG (MGd and MGv; f and g). R: response. L: long latency (>50 ms). W: weak response. N: no response. Numbers indicate the best frequencies. The scale bar is in mm.

Our study also uncovered a significant population (~35%) of lemniscal and nonlemniscal MGB cells that cannot be confidently classified as either single or burst cell type based on above-mentioned criteria. However, the poststimulus time histograms (PSTHs) constructed from "nonbursting" (i.e., single, double, and multiple spike) cells clearly exhibited a different temporal pattern of spike activities from bursting neurons. The latter, as a population, showed a more sustained pattern of excitation (Figs. 1 and 2A) and a more diverse and delayed first spike latencies that lagged nonbursting response by ~7 ms (Fig. 2B). Furthermore, the failure of bursting responses during individual trials often took place in an all-or-none fashion (n = 6; Fig. 3A).

View larger version (56K): [in this window] [in a new window]   Fig. 2. Poststimulus time histograms (PSTHs) and1st-spike latency distributions of bursting and nonbursting neurons. A: PSTHs of nonbursting and bursting () responses (binwidth = 5 ms) with time 0 corresponding to the onset of stimulation. For the sake of easy comparison, only data obtained from the 1st 50 ms of individual trials are shown. B: 1st-spike latency distribution of bursting () and nonbursting responses. Each bar represents the mean 1st-spike latency of a group of cells (binwidth = 2ms).

View larger version (25K): [in this window] [in a new window]   Fig. 3. Temporal features of evoked bursts. A: raster graphs derived from 20 stimulus trials from 4 different bursting neurons. Each burst consists of 2-6 successive spikes with varying latencies. Some neurons exhibit all-or-none failures (top right and bottom left). B, left: frequency distribution of intraburst mean interspike intervals (ISIs). Right: the sequence of lengths of the within-burst ISIs from a subpopulation of bursts made up of 2-4 spikes. The intervals are normalized by taking the 1st intraburst ISI as 100%. A bi-exponential function was used for curve fitting.

Quantitative analysis of the intra-burst structure in a subpopulation of neurons revealed that the evoked burst response had a mean interspike interval of 3-4 ms (Fig. 3B; n = 22). Interspike interval increased progressively within a burst (Fig. 3B). The underlying trend of this increase can be clearly seen by fitting the data with an arbitrary bi-exponential curve. Interestingly, spontaneous and evoked bursts recorded from the same neuron often exhibited similar features (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study examined the firing pattern of lemniscal and nonlemniscal MGB neurons responding to acoustic stimulation. Our data show that many neurons located within the nonlemniscal MGB discharged high-frequency bursts. The burst is, however, not a common feature of lemniscal neurons where the acoustic responses are dominated by single spikes or spike doublets of significantly shorter latency.

To our knowledge, the phenomenon of differential distribution of bursting neurons in vivo has not been documented before. Previous studies, however, show that lemniscal and nonlemniscal auditory neurons have distinctive physiological features (Bordi and LeDoux 1994; Calford 1983; Edeline et al. 1999). In contrast to MGv cells, the synaptic responses evoked in many nonlemniscal neurons, particularly those located in the caudodorsal part of the MGB, are characterized by delayed spike firing, widely fluctuating latency, a broad frequency tuning, and an absence of tonotopic organization (Bordi and LeDoux 1994; Calford 1983).

The differential anatomical distribution of single-spike and bursting responses cannot be readily ascribed to the antagonistic effect of ketamine anesthesia on N-methyl-D-aspartate (NMDA) receptors. Previous study has shown that the fast or early excitatory postsynaptic potentials (EPSPs) that mediate single-spike response in MGv neurons are largely mediated by non-NMDA receptors whereas burst firing in MGd requires NMDA receptor activation (Hu et al. 1994). In this context, we may have underestimated the proportion of bursting neurons in the nonlemniscal auditory thalamus. The persistence of bursting responses in the MGd may result from an incomplete blockade of NMDA receptor synaptic potentials (Deschenes and Hu 1990) and/or a relatively low sensitivity of NMDA receptors comprising NR2D subunits to uncompetitive NMDA receptor antagonists e.g., MK-801, (Bresink et al. 1996). Interestingly, NR2D subunits are abundantly expressed in caudal and dorsal MGB (Buller et al. 1994).

The extracellular recording approach adopted here provides only limited information regarding the cellular mechanisms underlying single-spike and burst firing. A commonly cited mechanism of thalamic neuronal bursting relies on voltage-dependent activation of a low-threshold Ca²⁺ spike (LTS) or the T-type Ca²⁺ current (Hu 1995; Steriade and Llinas 1988). Such LTS bursts exhibit long and variable latencies, all-or-none failure of occurrence, and successive increase in intraburst spike intervals (Hu 1995; Steriade and Llinas 1988), features that are consistent with the bursting responses reported here. Furthermore, MGd cells have a strong tendency to discharge bursts because of a relatively hyperpolarized membrane potential and a large NMDA receptor-mediated synaptic input (Hu et al. 1994). On the other hand, a more positive membrane potential, smaller EPSPs and a depolarization response to neuromodulators (e.g., acetylcholine) dramatically reduces the bursting responses in the MGv (Hu et al. 1994; Mooney et al. 1995).

Lemniscal thalamocortical transmission, with its faithful preservation of stimulus characteristics in the temporal and spatial domains, depends critically on a high membrane excitability, a tonic mode of neuronal discharge, and a tonotopical/retinotopic organization (Imig and Morel 1983; McCormick and Bal 1994; Steriade and Llinas 1988). The relatively slow responding mode of bursts, due to a long refractory period, seems unfit for rapid and precise signaling although bursts may be involved in other aspects of sensory processing. For instance, burst firing in the dorsal lateral geniculate nucleus is associated with improved detectability of sensory events (Sherman 2001), whereas bursting or high-frequency firing of nonlemniscal MGB efferents to the amygdala induces long term synaptic potentiation (Huang et al. 2000; Mooney and Hu 2002).