Contralateral Effects and Binaural Interactions in Dorsal Cochlear Nucleus
- First Online:
- Cite this article as:
- Davis, K.A. JARO (2005) 6: 280. doi:10.1007/s10162-005-0008-5
- 247 Views
The dorsal cochlear nucleus (DCN) receives afferent input from the auditory nerve and is thus usually thought of as a monaural nucleus, but it also receives inputs from the contralateral cochlear nucleus as well as descending projections from binaural nuclei. Evidence suggests that some of these commissural and efferent projections are excitatory, whereas others are inhibitory. The goals of this study were to investigate the nature and effects of these inputs in the DCN by measuring DCN principal cell (type IV unit) responses to a variety of contralateral monaural and binaural stimuli. As expected, the results of contralateral stimulation demonstrate a mixture of excitatory and inhibitory influences, although inhibitory effects predominate. Most type IV units are weakly, if at all, inhibited by tones but are strongly inhibited by broadband noise (BBN). The inhibition evoked by BBN is also low threshold and short latency. This inhibition is abolished and excitation is revealed when strychnine, a glycine-receptor antagonist, is applied to the DCN; application of bicuculline, a GABAA-receptor antagonist, has similar effects but does not block the onset of inhibition. Manipulations of discrete fiber bundles suggest that the inhibitory, but not excitatory, inputs to DCN principal cells enter the DCN via its output pathway, and that the short latency inhibition is carried by commissural axons. Consistent with their respective monaural effects, responses to binaural tones as a function of interaural level difference are essentially the same as responses to ipsilateral tones, whereas binaural BBN responses decrease with increasing contralateral level. In comparison to monaural responses, binaural responses to virtual space stimuli show enhanced sensitivity to the elevation of a sound source in ipsilateral space but reduced sensitivity in contralateral space. These results show that the contralateral inputs to the DCN are functionally relevant in natural listening conditions, and that one role of these inputs is to enhance DCN processing of spectral sound localization cues produced by the pinna.
Keywordsdorsal cochlear nucleuscontralateral inputsbinaural interactionssound localization
The dorsal cochlear nucleus (DCN) receives afferent input from the auditory nerve (Osen 1970; Ryugo and May 1993) and is thus usually thought of as a monaural nucleus. In response to ipsilateral stimulation, DCN principal cells in cat are well known to exhibit type IV unit response properties (Young and Brownell 1976; Young 1980). Type IV units are excited by low-level best frequency (BF) tones and inhibited by high-level tones, but are excited by broadband noise at all levels. Type IV units also respond with inhibition when a spectral notch is placed at their BF (Spirou and Young 1991). By virtue of their unusual sensitivity to narrowband peaks and notches, such as those added to sounds by the directionally dependent filtering properties of the pinna (Musicant et al. 1990; Rice et al. 1992), DCN principal cells are thought to serve a role in the detection of the spatial location of sound sources (Young et al. 1992; Imig et al. 2000). Consistent with this interpretation, lesioning the output pathway of the DCN disrupts the sound orientation behaviors of cats (Sutherland et al. 1998; May 2000).
Several lines of evidence, however, suggest the potential for functionally significant binaural interactions at the level of the DCN. First, anatomic studies have shown that the DCN receives inputs from the contralateral cochlear nucleus (Adams and Warr 1976; Cant and Gaston 1982; Shore et al. 1992; Schofield and Cant 1996; Alibardi 2000; Arnott et al. 2004), as well as descending projections from binaural nuclei (superior olive: Brown et al. 1988; Ostapoff et al. 1997; inferior colliculus: Conlee and Kane 1982; Schofield 2001; auditory cortex: Weedman and Ryugo 1996). Some of these projections are excitatory, whereas others are inhibitory. Moreover, the inhibitory inputs are both glycinergic and GABAergic (Wenthold 1987; Ostapoff et al. 1997; Alibardi 2000). Second, electrical stimulation of the contralateral auditory nerve in an in vitro whole brain preparation produces inhibitory postsynaptic potentials in 80% of DCN principal cells (Babalian et al. 1999). The latencies of these inhibitory responses are suggestive of mono- and disynaptic connections from the contralateral cochlear nucleus. Finally, single-unit studies have demonstrated that DCN neurons are sensitive to acoustic stimulation of the contralateral ear (Mast 1970, 1973; Young and Brownell 1976; Evans and Zhao 1993; Joris and Smith 1998; Imig et al. 2000). Some units are excited by monaural stimulation of the contralateral ear, but others are inhibited. Limited data suggest that this inhibition is, at least in part, glycinergic (i.e., it is blocked by the glycine antagonist strychnine; Evans and Zhao 1993). In some studies, responses to binaural stimulation are markedly different from those to ipsilateral stimulation suggesting a functional relevance for these binaural interactions (Mast 1970; Young and Brownell 1976), whereas in other studies, the effects are small (Joris and Smith 1998; Imig et al. 2000).
The goals of this study were to gain further information about the nature and effects of contralateral inputs to the DCN by sampling type IV unit responses to a variety of contralateral and binaural stimuli. The results support prior observations that most, but not all, type IV units are inhibited by contralateral stimulation. This inhibition is stronger for noise than for tones, and it is also lower in threshold and shorter in latency. Application of either strychnine or bicuculline (a GABAA antagonist) to the recording site blocks this inhibition and reveals underlying excitation; however, only strychnine blocks the onset of inhibition. Manipulations of the acoustic striae suggest that the inhibitory, but not excitatory, inputs to the DCN enter the DCN via its output pathway, and that the short latency inhibition is carried by axons from the contralateral cochlear nucleus. Consistent with the relative strengths of monaural effects, binaural inhibition is stronger for noise than for tones. Compared to monaural responses, binaural responses to virtual space stimuli show enhanced sensitivity to the location of a sound source in ipsilateral space but reduced sensitivity in contralateral space. These results show that contralateral inputs to the DCN are functionally relevant in natural listening conditions, and that one role of these inputs is to enhance DCN processing of spectral sound localization cues produced by the pinna.
Experiments were performed on 11 adult cats (3–4 kg) with infection-free ears and clear tympanic membranes. The first nine of these experiments were conducted at Johns Hopkins University (JHU); the last two were conducted at the University of Rochester (UR). All of the following procedures were carried out using similar protocols approved by the Institutional Animal Care and Use Committee of JHU and the University Committee on Animal Resources at the UR.
Cats were anesthetized with ketamine (40 mg/kg, im) and xylazine (0.5 mg/kg, im) and were given atropine (0.05 mg/kg, im) to minimize respiratory secretions and dexamethasone (2 mg/kg, im) to reduce cerebral edema. Body temperature was maintained at 39 ± 0.5°C using a regulated heating blanket, and at the UR, breathing and heart rates were monitored. The cephalic vein was cannulated to allow intravenous infusions of fluids, including supplemental doses of ketamine (15 mg/kg) and xylazine (0.1 mg/kg) as needed (e.g., heart rate over 180 beats/min), and a tracheotomy was performed to facilitate quiet breathing.
A midline incision was made over the skull and the temporalis muscles reflected to visualize the top of the skull and the ear canals. A craniotomy was performed over parietal cortex, and cats were made decerebrate by aspirating the brainstem between the superior colliculus and the thalamus. Anesthesia was then discontinued. The ear canals were transected near the tympanic membrane to accept hollow ear bars for delivering closed-field acoustic stimuli. The animal's head was fixed in the recording position, 35° nose down with respect to stereotaxic horizontal coordinates, using a headpiece and two ear bars. The left DCN was visualized by removing the skull about the nuchal ridge and aspirating the overlying cerebellum. In some experiments, the cerebellum above the floor of the fourth ventricle was also aspirated to allow access to the acoustic striae. At the end of experiments, cats were euthanized with an overdose of sodium pentobarbital (100 mg/kg, iv). Some cats were perfused to allow histological confirmation of the completeness of decerebration and of the placement of pharmacological electrodes.
Acoustic stimuli were delivered bilaterally via electrostatic speakers that were coupled to hollow ear bars. At the start of each experiment, the frequency response of both systems was measured with a probe tube microphone that was inserted into the ear bars near the tympanic membrane. At JHU, the acoustic calibrations were relatively flat (∼100 dB SPL ± 5 dB) across frequency from 40 Hz to 40 kHz and similar in both ears (±2 dB). Therefore, applying equal attenuation to binaural tones of the same frequency was assumed to create a 0-dB interaural level difference (ILD). At the UR, the attenuations were adjusted separately for the two ears to compensate for any differences in the calibration curves. At both institutions, interaural cross talk was at least 30 dB (and typically >50 dB) down at all frequencies in the ear opposite to the sound source. The physiologically effective interaural cross talk was measured by means of binaural threshold differences to tones for single auditory nerve fibers and cochlear microphonics (JHU system: Gibson 1982) or for single DCN type IV units (UR system). In the latter case, acoustic crossover was easily detected when the response to high level contralateral tones took the characteristic highly nonmonotonic form elicited by low-level ipsilateral tones.
All test stimuli, including tones, broadband noise, notch noise, and virtual space stimuli, were 200 ms in duration, gated on and off with 10-ms rise/fall times, and presented once per second. The wideband stimuli were synthesized on-line in the frequency domain and converted to time domain waveforms by taking the inverse Fourier transform of the digitally created noise spectrum (Nelken and Young 1997). Generic head-related transfer functions (HRTFs; Rice et al. 1992) were used to filter broadband noise spectra to synthesize binaural virtual space (VS) stimuli at azimuths ranging from −60 to +60° in 15° steps and elevations from −30 to +45° in 7.5° steps. At the UR, the magnitude spectra of all wideband stimuli were corrected to compensate for nonflat calibration curves. Analog signals were created by playing the waveforms through a 16-bit D/A converter at a (usual) sampling rate of 100 kHz.
All experiments were carried out in a double-walled sound-attenuating booth. Single-unit activity was recorded with platinum–iridium electrodes. At JHU, electrodes were advanced into the DCN using a hydraulic microdrive. The signal from the electrode was amplified (10,000–30,000×) and filtered from 0.3 to 6 kHz. A variable-threshold Schmitt trigger was used to discriminate action potentials from background noise. At the UR, an Alpha–Omega system was used to advance the electrodes (via a motor-controlled multielectrode positioning system or EPS), to condition the signal (MCP-Plus), and to detect action potentials (template matching software, MSD). In all experiments, spike times relative to stimulus onset were stored for off-line analysis. Sound-driven activity was analyzed in terms of average discharge rates over the final 150 ms of the stimulus-on interval to reflect steady-state responses; spontaneous rates were computed over the last 400 ms of the stimulus-off interval of each 1-s stimulation period.
Recording electrodes were advanced dorsoventrally through the DCN, whereas 50-ms search tones or noise bursts were presented to the ipsilateral ear. When a single unit was isolated, its BF and threshold were determined using audiovisual feedback, and its response type was determined from responses to 200-ms BF-tone and broadband noise bursts presented across a range of sound levels (100-dB range in 1-dB steps). Units were classified as type IV, IV-T, III, II, or complex spiking using standard criteria (e.g., Shofner and Young 1985; Zhang and Oertel 1993a; Manis et al. 1994). Results are primarily reported for type IV units because DCN principal cells in cat usually exhibit these response properties (Young 1980). Once a unit was classified, rate-level functions were obtained for tones (at the ipsilateral BF) and broadband noise presented to the contralateral ear. Frequency response maps for contralateral tones were created by sweeping the frequency of tone bursts over a three-octave range centered on the unit's ipsilateral BF. These sweeps were presented at multiple sound levels, ranging from 0 to 60 dB above threshold. Each frequency–intensity combination was presented once.
The pharmacology of the contralateral inputs to the DCN was investigated by infusing the recording site with either the glycine antagonist strychnine or the GABAA antagonist bicuculline. Piggyback multibarreled electrodes (after Havey and Caspary 1980) were used to deliver strychnine hydrochloride or bicuculline methiodide (each 10 mM, pH 3.5–4.0, Sigma) into the DCN (Davis and Young 2000). Electrode negative retention currents of 20 nA and ejection currents of 50 nA were produced with microiontophoresis constant current generators. Rate-level functions for contralateral BF tones and broadband noise bursts were performed before and after pharmacological manipulations to verify unit stability and to confirm recovery from inhibitory blockade before moving on to the next unit.
The origin of contralateral effects in the DCN was studied using both pharmacological and surgical manipulations of the dorsal output tracts (dorsal and intermediate acoustic striae). Hypodermic needles (30 gauge) were used to record from, and to deliver solutions of, lidocaine hydrochloride (2%; Tech America) into the contralateral DAS/IAS and DAS. Needles were coupled to a 1-ml syringe to allow reliable injections of microliter amounts of lidocaine solution. The needles were either placed at the medial border of the contralateral DCN, where the fibers of the DAS and IAS intermingle, or moved medial toward the midline of the floor of the fourth ventricle to isolate the DAS (Fernandez and Karapas 1967; Davis 2002). Placement of the needle was guided by searching for noise-evoked background activity where the output tracts were expected to be found. The needle was placed in a position judged to the center of the tract(s) (i.e., where there was the maximum background activity), and then left there for the entire experiment. In some of these experiments, the medial border of the ipsilateral DCN was aspirated to sever fibers incoming through the ipsilateral DAS/IAS.
For binaural testing, two paradigms were used. First, the same pure tone or broadband noise stimulus was presented to both ears, but a 40-dB range of ILDs was created by varying the level of the contralateral stimulus relative to a fixed ipsilateral stimulus. The intensity of the ipsilateral stimulus was fixed at 10-dB re threshold. Second, responses were obtained to ipsilateral monaural and binaural VS stimulation. The sampling rate for these stimuli was set on a unit-by-unit basis to (BF/12) × 100 kHz. This stimulus manipulation, which effectively shifted the BF of each unit under study to 12 kHz, was performed because 12 kHz is near the center of the frequency range (5–18 kHz) where cat HRTFs show directionally dependent spectral notches (Musicant et al. 1990; Rice et al. 1992). Furthermore, it increased the yield of data at one frequency and thereby allowed the data to be subjected to statistical analyses.
The effects of contralateral monaural stimulation with pure tones and broadband noise (BBN) were studied on 54 type IV units. To examine further the nature and origin of the contralateral inputs to the DCN, different subsets of these units were tested under local blockade of inhibitory receptors (n = 8) and after manipulations of the dorsal output tracts (n = 18). Binaural interaction data were acquired on 20 units.
Effects of contralateral monaural acoustic stimulation on type IV units
Each unit's latency for contralateral tones (noise) is plotted against its latency for ipsilateral tones (noise) in Figure 5B and D, respectively. As above, symbols for units without a response to contralateral tones are plotted beyond the axes. Here, response latency is defined as the time from stimulus onset to the first time the driven rate in the PSTH exceeded 1 SD above (below) the SR. Two symbols (asterisks and squares) are plotted for each class ee unit in Figure 5D to highlight the dual nature of the response (inhibition followed by excitation) to contralateral noise exhibited by this class of units. Note that all of the data points for tones are above the equity (dashed) line, whereas most data points for noise are distributed along the line (including 19 units with contralateral noise latencies below corresponding ipsilateral values) indicating that latency differences were usually smaller for noise than for tones. Overall, the median difference in latency was 10 ms for tones but only 2 ms for noise. However, the unit classes did show some differences in response properties. For example, class ii units (unfilled circles) showed a wide range of latencies for contralateral tones, whereas class ei and ee units (filled symbols) showed a tighter cluster of values at intermediate latencies. In addition, units excited by contralateral noise (class ee; black squares) showed longer latencies for contralateral noise than did units inhibited by noise (P > 0.05, Mann–Whitney U-test).
Effects of contralateral monaural acoustic stimulation on other unit types
The effects of contralateral monaural stimulation with pure tones and BBN were studied on 3 type IV-T, 10 type III, 7 type II, and 5 complex-spiking units. The type IV-T and type III units exhibited all four classes of response with threshold and latency differences similar to those exhibited by type IV units. The type II units, on the other hand, appeared unresponsive to monaural stimulation of the contralateral ear; that is, driven rates were unchanged from zero, which is the SR of this unit type. If background activity was evoked in these units using low-level ipsilateral BF tones, then these units exhibited weak class ei properties. Furthermore, the excitation to contralateral tones was tuned about the ipsilateral BF (not shown). Finally, complex-spiking units exhibited little-or-no affect of contralateral stimulation.
Pharmacology of contralateral inhibitory inputs
Poststimulus time histograms showing the time course of the effects of these inhibitory antagonists on the response of the type IV unit are shown in Figure 6B. Under control conditions (heavy solid line), the PSTH showed a typical short-latency, sharp decrease in rate, followed by a sustained reduction throughout the stimulus duration. It also showed an unusually long, poststimulus suppression in rate that took several hundreds of milliseconds to recover back to SR. Under strychnine (solid line), the short-latency inhibitory response was blocked, and the initial response was excitatory. With time, the magnitude of this excitatory response decreased, and by the end of the stimulus, the unit was inhibited. In contrast, bicuculline (dashed line) did not block the short-latency component of the inhibition. Thereafter, its effects were similar to that of strychnine.
On the origins of contralateral effects
Figure 7B shows the rate-level response of a type IV unit to contralateral BBN after the medial border of the ipsilateral DCN was lesioned. This manipulation severs both dorsal output tracts from the cochlear nucleus. The unit was strictly excited by contralateral BBN (and tones, not shown) at all levels. Unlike typical class ee units, however, the PSTH for this unit exhibited no short-latency inhibition prior to the excitatory response (Fig. 7D). All units tested after this manipulation (n = 6) showed similar response properties suggesting that the axons carrying inhibition, but not excitation, enter the DCN via the DAS. The fact that inhibitory and excitatory inputs enter the DCN via dorsal and ventral routes, respectively, may explain the observation reported here and by Mast (1970, 1973) that neurons inhibited by contralateral tones predominate in the middle (fusiform cell) layer of the DCN, whereas excited neurons tend to be located in the deep layer of the DCN.
Effects of binaural stimulation on type IV units
The effects of ILD on the discharge rate properties of type IV units were quantified by comparing the driven rates at ILDs of ±20 dB for binaural tones (Fig. 8B) and noise (Fig. 8C). Each symbol in these plots indicates the driven rate of an individual unit at the −20-dB versus the +20-dB condition, with symbols falling along the dashed line representing units that showed no change in rate under the two conditions. Different symbols represent the different response classes. Consistent with qualitative observations for the representative unit in Figure 8A, the rate changes for tones were small for all units tested (n = 10; Fig. 8B), as indicated by the fact that the symbols are clustered about the equity line. Conversely, rate changes for BBN were large (Fig. 8C), with a median change in rate of 45 spikes/s. Furthermore, for 4 of the 19 units showing inhibition to contralateral BBN (circles), the unit's driven response changed from excitation to inhibition (symbols below the horizontal line) suggesting a strong functional relevance for these binaural interactions.
The threshold for transition from a contralateral to ipsilateral dominant response has been defined as the half-maximal ILD (Wenstrup et al. 1988; Park and Pollak 1993; Davis et al. 1999). The half-maximal ILD is defined to be that ILD at which the driven response to the binaural stimulus changes by 50% from the response evoked by the excitatory monaural stimulus. Using this definition, no units exhibited a transition in response to tones. This was true despite the fact that the level of the contralateral tone in the −20-dB ILD condition was more than 10 dB above the median monaural threshold of the population. By contrast, most units exhibited a transition for noise (17/20 units), with the median half-maximal ILD equal to −4 dB. Given that the intensity of the contralateral noise in the 0-dB ILD condition was just above the median monaural threshold of the population, this result suggests that only a modest increase in contralateral stimulus level above threshold was needed to attain the half-maximal ILD.
The effects of binaural versus monaural stimulation on the strength of inhibition were quantified by comparing the magnitude of inhibition at elevations along the 12-kHz notch contour as a function of azimuth. Curves connecting the median rates for six class oi units are shown in Figure 9C. The green plots indicate responses to monaural stimulation, whereas the red plots indicate responses to binaural stimulation. Nonparametric statistical tests indicated that azimuth had no significant effect on the strength of inhibition (P = 0.53; Friedman's test); however, inhibitory responses to binaural stimulation were significantly stronger than responses to monaural stimulation (P = 0.02; median change of 9 spikes/s). The ability of binaural stimulation to affect driven rates even at the edges of ipsilateral space tested here most likely reflects the fact that the levels of VS stimuli were usually well above (∼30 dB above) ipsilateral thresholds to noise, and thus above contralateral thresholds to noise as well.
Dorsal cochlear nucleus type IV units showed distinct changes in the extent of inhibitory response areas in their spatial receptive fields from monaural to binaural stimulus conditions. Figure 9D shows the borders of these areas for monaural (green curves) and binaural stimulation (red curves) as a function of azimuth. The curves connect the median values at each azimuth. Note that under monaural stimulus conditions, the inhibitory area is tuned along the 12-kHz notch contour (dashed line) from low ipsilateral to high contralateral elevations. In contrast, the inhibitory area widens significantly in the contralateral hemifield (positive azimuths) under binaural stimulus conditions. The effect of this change is to reduce the selectivity of type IV units for notches in the contralateral field (i.e., to lateralize the response to the ipsilateral hemifield).
Comparisons with previous studies
In the present study, DCN principal cell responses to contralateral monaural tones and noise could be grouped into four distinct classes (Figs. 2 and 3). Most type IV units were weakly (30%), if at all (52%), inhibited by tones, but were strongly inhibited by noise. The remainder of the units was excited by tones and either inhibited (11%) or excited by noise (7%). These results are largely consistent with those of previous studies. In particular, Young and Brownell(1976) found the same range of response classes and similar incidence rates in an earlier study in decerebrate cats. Mast (1970, 1973) found that approximately 20% of DCN cells in anesthetized chinchilla were inhibited by contralateral tones, whereas 10% of units in an unanesthetized preparation were excited. Finally, Joris and Smith (1998) found that 50% of a limited sample of units in anesthetized cat showed weak inhibition to tones, whereas 100% of DCN principal cells were strongly inhibited by contralateral noise. The lack of excitation in their study may be because of the effects of anesthesia, which is known to abolish contralateral excitatory effects (Mast 1973).
Consistent with their respective monaural effects, contralateral tones had only weak effects on ipsilaterally evoked activity, whereas contralateral noise had strong effects (Fig. 8). In particular, ILD curves for tones were relatively flat as a function of increasing contralateral level, whereas binaural BBN responses decreased with level. In agreement with these results, Young and Brownell (1976) reported that binaural responses to diotic tones (ILD = 0-dB condition) as a function of stimulus level were essentially the same as responses to ipsilateral tones. Binaural noise responses, on the other hand, were found to be intermediate between excitatory ipsilateral and inhibitory contralateral responses. Interestingly, tone ILD curves in chinchilla were found to decrease markedly with increasing contralateral level suggesting a strong effect of contralateral tones (Mast 1970). This difference could reflect an anesthetic effect (i.e., a potentiation of GABAergic inputs; Richter and Holtman 1982) or a species difference. Finally, Joris and Smith (1998) reported that noise ILD curves were flat as compared with those found in the lateral superior olive. However, they found decreases in firing rates with increasing contralateral level on the order of 30–40 spikes/s, which are comparable to the values shown here.
Stimulation of the contralateral ear had significant effects on type IV unit responses to natural sound fields (Fig. 9). In response to monaural VS stimuli, type IV units showed a tuned inhibitory response that followed a diagonal contour from low ipsilateral to high contralateral elevations. This trough in the elevation functions coincided with a spectral notch in the HRTF at the unit's BF. When tested with binaural VS stimuli, type IV units showed a deeper inhibitory trough in the ipsilateral hemifield and inhibition throughout much of contralateral space. Imig et al. (2000) also found that DCN principal cells in anesthetized cat showed response nulls in their azimuth functions to free-field monaural and binaural noise. Furthermore, spatial receptive fields showed that these response nulls followed the expected trajectory of spectral notches. The depth of modulation was slightly greater for binaural than monaural noise showing that contralateral inhibition had a small but significant effect on unit directionality.
A conceptual model of contralateral inputs to type IV units
The prototypic type IV unit is weakly, if at all, inhibited by contralateral tones, but is strongly inhibited by noise (Figs. 2 and 3). This inhibition is, in part, glycinergic and short latency (Fig. 6; Evans and Zhao 1993; Mast 1970; Joris and Smith 1998) and carried by commissural axons (Fig. 7A and C). Several lines of evidence suggest that Oc units in the contralateral VCN are the source of this inhibition. First, Oc units (large multipolar cells, Smith and Rhode 1989) respond more strongly to BBN than to tones (Winter and Palmer 1995; Palmer et al. 1996). Second, Oc units have very short response latencies, especially compared with those of DCN neurons (Rhode and Smith 1986a, b; Joris and Smith 1998), allowing for ample time to provide near-coincident input with ipsilateral inputs. Finally, the commissural pathway arises primarily from large multipolar cells in the VCN (Cant and Gaston 1982; Shore et al. 1992; Schofield and Cant 1996) that are glycinergic (Wenthold1987; Ostapoff et al. 1997; Alibardi 2000; Babalian et al. 2002). Recent studies have shown that at least some of these cells have Oc responses to sound (Needham and Paolini 2003; Arnott et al. 2004). In addition, it appears from Arnott et al. that the axons of Oc units travel across the brainstem near the fibers of the IAS. The results of the current study are consistent with this observation as pharmacological manipulation of the contralateral IAS, but not DAS, blocked the contralateral short-latency inhibition.
The possibility of glycinergic inputs from a more diverse population of cells, most notably from periolivary neurons in the superior olivary complex (SOC) (Ostapoff et al. 1990, 1997), cannot be ruled out. However, response latencies for neurons in these nuclei are long in comparison to those of auditory nerve fibers (e.g., Tsuchitani 1977; Liberman and Brown 1986). Thus, such inputs are likely to contribute only to the later, sustained portion of the inhibition.
The inhibition of type IV units is also, in part, GABAergic and long latency (Fig. 6). The largest extrinsic sources of GABAergic input to the DCN are from the periolivary nuclei within the SOC (Ostapoff et al. 1990, 1997). Most of the GABA-immunoreactive cells that project to the DCN are located bilaterally in the VNTB. The VNTB contains a heterogeneous group of cells, which are known to receive ascending input from the VCN (Robertson and Winter 1988; Thompson and Thompson 1991; Smith et al. 1993) as well as descending inputs from a variety of sources (Vetter et al. 1993; Mulders and Robertson 2000). The response properties of all the different cell types in the VNTB have not been studied systematically. In general, however, VNTB neurons are driven by tones and noise, through one or both ears, and have response latencies that are delayed on the order of 5–10 ms with respect to activity in auditory nerve fibers (Goldberg and Brown 1968; Guinan et al. 1972a, b; Liberman and Brown 1986).
The model assumes a source of excitation to type IV units to account for units that are excited by tones and noise (Figs. 2 and 3). This excitatory source has a long latency (Fig. 5B and D; Mast 1973), and its axons must enter the DCN via a ventral route (Fig. 7B and D). A strong candidate for this source is a circuit that includes neurons in the VNTB and T stellate cells in the VCN. The VNTB contains two distinct groups of cholinergic cells, medial olivocochlear (MOC) neurons and small cells, which provide all of the cholinergic input to the VCN (Sherriff and Henderson 1994). The efferent fibers of MOC neurons project to outer hair cells in the cochlea, but also give off collateral branches within the vestibular nerve root that contact the dendrites of stellate cells in the VCN (Benson and Brown 1990; Brown and Benson 1992). The small cells project to the magnocellular region of the VCN through the trapezoid body (Sherriff and Henderson 1994). In turn, in vitro studies suggest that T stellate cells excite DCN principal cells (Oertel et al. 1990; Zhang and Oertel 1993b, 1994).
Several lines of evidence suggest that contralateral acoustic stimulation could elicit long-latency excitation of T stellate cells (chopper units, Smith and Rhode 1989). First, MOC neurons, and VNTB neurons in general, are driven by tones or noise, through one or both ears, and have long response latencies (Goldberg and Brown 1968; Guinan et al. 1972a, b; Liberman and Brown 1986; Liberman 1988; Brown et al. 1998). Second, T stellate cells are excited by application of cholinergic agonists (Fujino and Oertel 2001) suggesting that activity of either the MOC or small cells in the VNTB would likely excite T stellate cells. Consistent with this hypothesis, Mulders et al. (2002) found that electrical stimulation of the MOC, after the olivocochlear effect was eliminated, elicited excitation in some VCN chopper units. However, other chopper units were inhibited suggesting a dual action of MOC efferents. Third, Needham and Paolini (2003) found that contralateral tones elicited sustained excitatory postsynaptic potentials in chopper units with a delay of 7–10 ms. Similarly, chopper units showed a short-latency membrane hyperpolarization after stimulation of the contralateral ear with noise that was soon exceeded by excitatory inputs (latency of ∼14 ms). Finally, Shore et al. (2003) found that 30% of VCN units in anesthetized guinea pig were inhibited by contralateral stimulation, but that 4.5% of cells were excited.
The possibility of other sources of excitatory input to DCN type IV units cannot be ruled out. For example, there is considerable heterogeneity in the neurons that contribute to the commissure (Shore et al. 1992; Schofield and Cant 1996; Alibardi 2000). Some of these neurons are immunonegative for both glycine and GABA (Alibardi 2000) and thus presumably excitatory. In addition, some projections from the periolivary nuclei are also immunonegative for both glycine and GABA (Ostapoff et al. 1997). Finally, projections from the inferior colliculus may be glutamatergic and excitatory (Saint Marie 1996).
Behavioral studies have shown that cats are capable of localizing the source of broadband sounds in the frontal field with considerable accuracy (May and Huang 1996). This accuracy is maintained when the bandwidth of stimuli is narrowed to contain energy from 5 to 18 kHz, but is lost when mid-frequency cues are removed by high-pass filtering or narrowing the bandwidth of the stimulus to mid-frequency pure tones (Huang and May 1996). Studies of the filter functions that describe the transformation of a free-field sound to the energy spectrum at the eardrum suggest that a directionally dependent spectral notch at mid-frequencies may be the cue to localizing broadband sounds (Musicant et al. 1990; Rice et al. 1992). Interestingly, notch frequency does not represent a single point in space, but rather a contour that connects low ipsilateral to high contralateral elevations. The ambiguity along this line could be resolved by information from low and high frequencies, from binaural time and level disparity cues (Middlebrooks and Green 1991), or knowledge of the notch frequency in the contralateral ear (the binaural first notch direction code or BiFiND code; Rice et al. 1992).
Electrophysiological studies in cats have shown that the projection neurons of the DCN are especially sensitive to spectral notches (Young et al. 1992; Imig et al. 2000; this study). Type IV units are inhibited by tones, but give excitatory responses to broadband noise (Young and Brownell 1976). When a spectral notch is added to noise near BF, type IV units show an inhibitory response (Spirou and Young 1991). Thus, type IV units show a tuned inhibition for a spectral notch at their BF. Furthermore, the spatial receptive fields of DCN principal cells show response nulls that follow the expected diagonal trajectory of spectral notches (Imig et al. 2000; this study). These observations suggest that the DCN serves as an early stage in the processing of HRTF-based sound localization cues. Consistent with this interpretation, lesioning the output pathway of the DCN disrupts the sound localization performance of cats, particularly their orientation to sound source elevation (Sutherland et al. 1998; May 2000).
The results of present study show that DCN principal cells are generally inhibited by contralateral stimuli, particularly by broadband noise. The effects of the contralateral inputs on the spatial receptive fields of type IV units are twofold: stronger inhibitory responses to notches in ipsilateral space and primarily inhibitory responses in contralateral space. The former effect increases DCN sensitivity to spectral notches and is thus likely to enhance upstream processing of spectral cues. For example, type O units in the inferior colliculus are known to receive DCN inputs (Davis 2002) and to show tuned excitatory responses for notches located just below BF (Davis et al. 2003). The stronger inhibition exhibited by DCN units for notches at BF under binaural stimulus conditions would likely prevent the upward spread of excitation in type O units. The latter effect largely confines the sensitivity of DCN principal cells to sounds in ipsilateral space. This has clear implications for the BiFiND hypothesis because it restricts toward the midline the region of space where notch frequency may be known from both ears. This effect could also explain, in part, preliminary observations that type O units show spatial tuning for sounds primarily in one hemifield (Davis et al. 2003). More studies are needed to determine the extent to which binaural interactions in DCN impact the processing of spectral-dependent information in higher-order neurons.
The author thanks Dr. Eric D. Young for support and encouragement, Drs. Nell Cant and Brett Schofield for helpful discussions, Oleg Lomakin for software development and for participating in some of the experiments, and Yue-Houng Hu and Oleg Lomakin for help in data analysis and figure preparation. This work was supported by National Institute of Deafness and Other Communication Disorders grants DC00979 (to E.D.Y.) and DC05161 (to K.A.D.).