| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Otolaryngology, Head and Neck Surgery, University of California, San Francisco, California
Submitted 30 August 2004; accepted in final form 13 January 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
300 pps). Acutely deafened and implanted adult cats served as controls. Spiral ganglion cell density in all long-deafened animals was markedly reduced (mean <5.8% of normal). Both long-term deafness and chronic electrical stimulation altered temporal resolution of neurons in the central nucleus (ICC) but not in the external nucleus. Specifically, LDU animals exhibited significantly poorer temporal resolution of ICC neurons (lower Fmax, longer response latencies) as compared with control animals. In contrast, chronic stimulation in LDS animals led to a significant increase in temporal resolution. Changes in temporal resolution after long-term deafness and chronic stimulation occurred broadly across the entire ICC and were not correlated with its tonotopic gradient. These results indicate that chronic electrical stimulation can reverse the degradation in temporal resolution in the auditory midbrain after long-term deafness and suggest the importance of factors other than peripheral pathology on plastic changes in the temporal processing capabilities of the central auditory system. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Nadol et al. 1989; Nishiyama et al. 2000; Otte et al. 1978; Snyder et al. 1995; Vollmer et al. 1999, 2000). The age at onset of deafness is an important factor determining the extent of these pathological changes, suggesting a critical period for normal development of the auditory system (human studies: Eggermont and Bock 1986; Ruben and Rapin 1980; animal studies: Silverman and Clopton 1977; Webster 1983, 1988).
Animal studies have shown that sensorineural hearing loss before the onset of hearing or during early postnatal periods results in more profound anatomical degeneration (e.g., Moore 1990; Nishiyama et al. 2000; Nordeen et al. 1983) and functional degradation or reorganization as compared with changes observed after auditory deprivation later in life (e.g., Hardie and Shepherd 1999; Hardie et al. 1998; Moore 1994; Raggio and Schreiner 1999; Shepherd et al. 1997; Silverman and Clopton 1977; Trune 1982).
Clinical studies also suggest that the acquisition of speech and language in humans has a critical period within the first few years of life (Eggermont and Bock 1986; Ruben and Rapin 1980). The deficits in speech and language acquisition are generally more pronounced with earlier and more extensive auditory deprivation. The greatest deficits are observed when deafness occurs around birth (Ruben 1986).
Several studies indicate that early chronic electrical stimulation of the auditory system during maturation can ameliorate or prevent some of these negative effects of auditory deprivation. Histological studies have shown that chronic electrical stimulation of the cochlea prevents or delays the degeneration of spiral ganglion cells (SGC) (e.g., Hartshorn et al. 1991; Leake et al. 1991, 1992, 1999) and ameliorates degenerative changes in the cochlear nucleus (Lustig et al. 1994; Matsushima et al. 1991) that occur after deafness. Electrophysiological studies have demonstrated that early introduction of chronic electrical stimulation in neonatally deafened cats either maintains or increases temporal resolution of neurons in the ICC (Snyder et al. 1995; Vollmer et al. 1999) depending on the stimulus frequency.
Moreover, clinical studies indicate that deaf children implanted at a younger age demonstrate better speech recognition than those implanted at an older age (Hassanzadeh et al. 2002; Miyamoto et al. 2003; Moog and Geers 1999; Osberger 1995). Congenitally and prelingually deaf cochlear implant (CI) users who are implanted as adults generally demonstrate particularly poor speech discrimination (Busby et al. 1991). Other studies suggest that the immature deaf auditory system may be more plastic or adaptable to the electrical information provided by a CI than the mature, long-deafened auditory system (e.g., Blamey et al. 1996; Dawson et al. 1992; Osberger et al. 1998; Tyler et al. 1997; Waltzman 1997). These studies highlight the important role of age at implantation and duration of deafness for speech recognition performance in CI users. However, even prelingually deafened individuals implanted as adults demonstrate gradual improvements in speech recognition over time, indicating that auditory experience and implant use are important factors for enhanced performance (Busby et al. 1991). These results suggest that reorganization of auditory processing capacities even after prolonged periods of deafness is presumably a reflection of plastic mechanisms in the central auditory system.
The goal of the present study was to examine temporal processing in the inferior colliculus (IC) in an adult animal model of congenital deafness. Specifically, we assessed the temporal response properties [maximum following frequencies (Fmax) and first spike latencies] of single IC neurons to address the following questions: first, how is temporal resolution affected by long-term auditory deprivation? We hypothesized that the severe cochlear pathology induced by prolonged periods of deafness would impair temporal processing in the IC. The present study extends the very limited data set reported in a previous investigation on temporal response properties of IC neurons in a single long-deafened animal (Shepherd et al. 1999) and examines neuronal responses in neonatally deafened, unstimulated animals that were studied after long durations (>2.5 yr) of deafness (LDU). Second, what is the effect of chronic electrical stimulation on temporal processing in the neonatally deafened adult auditory system? The current study reports for the first time temporal resolution data of IC neurons obtained from long-term (>3.5 yr) neonatally deafened animals that received chronic electrical stimulation as adults (LDS). Earlier studies using chronic stimulation with identical signals indicated a significant increase in temporal resolution of ICC neurons in the developing auditory system (Vollmer et al. 1999). Thus these data provide a valuable basis for a comparison of functional alteration after chronic electrical stimulation between the mature and developing auditory system.
In addition, SGC density was determined for the individual animals to evaluate the effects of long-term deafness on peripheral cochlear pathology and the consequences of severe spiral ganglion cell loss on temporal resolution in the central auditory system.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Deafening and implantation
Before all surgical procedures, animals were sedated with an intramuscular injection of ketamine (22–33 mg/kg) or initial anesthesia was induced with inhaled isoflurane. An intravenous catheter was inserted into the cephalic vein for fluid or drug administration. General anesthesia was induced with pentobarbital sodium (7–10 mg/kg iv) and maintained at a surgical areflexic level with supplementary intravenous infusion of pentobarbital sodium (2–6 mg · kg–1 · h–1) in Ringer solution. Vital functions (heart rate, respiration, CO2 or O2 saturation, body temperature) were monitored continuously and maintained at physiological levels.
Two experimental groups of neonatally deafened animals (n = 11) were studied after prolonged periods of neonatal deafness (>2.5 yr): six unstimulated cats received a unilateral cochlear implant as adults and were studied acutely (LDU group); the other five cats were also implanted as adults and received several weeks to months of chronic electrical stimulation prior to study (LDS group; Table 1). All long-deafened animals were deafened as newborns by systemic administration of neomycin sulfate (40–70 mg/kg im/SID) beginning 24 h after birth and continuing for the first 14–25 days after birth. Neomycin injections were terminated when profound hearing loss (>108 dB) was confirmed by the absence of auditory brain stem responses to clicks (0.2 ms/ph, 20 pps) and frequency following responses to tonal stimuli (500 Hz). None of the animals demonstrated any residual hearing. All animals in both long-deafened groups were maintained for periods ranging from 2.5 to 7.2 yr prior to study. Two of the LDU animals (K16 and K24) were implanted immediately before study, the others were implanted 1 wk before the electrophysiological experiment to allow thresholds to stabilize. LDS animals were implanted, and electrical stimulation of the auditory nerve was initiated at ages ranging from 3.5 to 7.0 yr with an average of 5.3 yr.
|
). Control cats usually were deafened 1.5–3.5 wk before study by intravenous administration of kanamycin and ethacrynic acid (Xu et al. 1993). About half of the control animals were implanted immediately before the experiment, the others were usually implanted 
1–2 wk before study.
Comparison of temporal resolution data from cats studied directly after implantation and those studied 1 wk after implantation did not show any statistically significant differences in either the LDU or the control group. The immediate and delayed implantation data were therefore pooled for subsequent analysis.
All electrode arrays were inserted through the round window into the left scala tympani. Chronic implantations were carried out under sterile conditions. Implants consisted of four platinum-iridium ball-shaped electrode contacts (
300 mm diam) in a molded silicone rubber carrier that was custom designed for the cat cochlea (Rebscher et al. 1988). Electrode contacts were designated 1–4 from apical to basal cochlear locations and were arranged as two bipolar offset-radial pairs (apical pair: 1,2; basal pair: 3,4). The separation between the electrodes comprising a pair was 1 mm, and the two pairs were separated by 3 mm. A percutaneous connector allowed direct electrical connection to the electrodes.
Chronic stimulation
Chronic electrical stimulation was applied for 4 h/d, 5 days/wk for a mean duration of 21 ± 11 (SD) wk, with maximum signal intensity adjusted to 2 dB above electrically evoked auditory brain stem responses (EABR) threshold for each individual subject. All LDS animals were stimulated with the apical electrode pair 1,2. Due to lead failure, however, subjects CD393 and CH539 were stimulated with electrode pairs 2,3 and 1,4, respectively, during the final weeks of stimulation. Because the level of chronic stimulation generally exceeds the dynamic range of IC neurons both for widely and narrowly spaced electrode pairs and leads to an activation of virtually the entire IC (unpublished data), it seems unlikely that the wider spacing of the stimulating electrodes in CH539 affected temporal resolution in this animal.
Electrical stimulation was delivered either by an analogue speech processor (SP) that transduced ambient environmental sounds or by computer-generated amplitude modulated pulse trains. For stimulation with the analogue speech processor, the frequency spectrum of the analogue stimulation was band-pass filtered from 250 Hz to 3 kHz with a roll-off at the shoulder frequencies of 6 dB/octave. The computer-generated signal was a continuous train of electrical pulses (200 µs/phase, charge-balanced biphasic square-wave pulses) delivered at a 300 pps carrier rate and sinusoidally amplitude modulated (SAM) at a frequency of 30 Hz with a modulation depth of 100% (300/30 SAM). The choice of these stimuli was based on previous studies showing that chronic electrical stimulation of the developing auditory system using these signals resulted in a significant increase in temporal resolution of IC neurons (Snyder et al. 1995; Vollmer et al. 1999).
One animal (CH611) received stimulation with 80 pps delivered by a backpack stimulator for a period of 2 wk. This animal did not tolerate stimulation outside its home cage, and the 80 pps stimulation was used until an analogue SP became available as a backpack stimulator for use in the home cage.
All but one cat received additional stimulation during behavioral training sessions (Table 1; 5 day/wk). The total duration of suprathreshold stimulation during the behavioral sessions was very brief (a total of usually <30 s) and, therefore very limited compared with the duration of chronic passive stimulation (4 h/day at suprathreshold level).
Acute electrophysiological experiments
The anesthetic procedures were virtually identical to those already described for implantation. The animal's head was stabilized in a head holder, and the IC contralateral to the cochlear implant was exposed. Neuronal responses were recorded differentially using two tungsten microelectrodes matched in impedance (0.8–1.5 M
). The active electrode was advanced through the IC from dorsolateral to ventromedial parallel to the tonotopic gradient (Brown et al. 1997) with low frequencies represented more superficially and high frequencies at progressively deeper locations within the ICC. Biphasic square–wave pulses (0.2 ms/ph, 3 pps) were used as a search stimulus. Neural activity was recorded differentially, band-pass filtered, amplified, and monitored on an oscilloscope (Tektronix 565) and an audio monitor.
At intervals of 100 µm along each penetration, the minimum response threshold levels for three cycles of a 100-Hz sinusoidal signal and for pulses (0.2 ms/ph, 3–10 pps) were determined audiovisually for either single- or multiple-neuron responses. Thresholds were plotted as a function of IC depth to obtain a spatial tuning curve (STC; Fig. 1). As reported previously, STCs were typically W-shaped (Vollmer et al. 1999). The highest threshold region between the two locations of minimum thresholds was defined as the border between the two nuclei (see Fig. 1, - - -) and allowed neurons to be assigned to either the external nucleus of the IC (ICX) or the ICC. Due to the difference in charge per phase of the 100-Hz sinusoidal (5 ms/ph) and pulsatile (0.2 ms/ph) signals, STCs obtained with sinusoidal stimuli generally had lower thresholds, markedly sharper tuning, and greater dynamic ranges than those obtained with pulses. Therefore in most penetrations the border between ICX and ICC was determined by the STCs for sinusoidal stimuli.
|
To assess the temporal processing capabilities of isolated neurons, peristimulus-time histograms (PSTHs) were plotted, and the maximum stimulus frequency (Fmax) to which each neuron phase locked (P < 0.01, Raleigh-test) (Mardia 1972) was determined. Figure 2A illustrates examples of PSTHs constructed for one single neuron responding to intracochlear pulse trains of increasing frequencies (10–90 pps). Vector strength (VS) and significance (P) of phase-locking are noted to the right of each histogram. The response to the first pulse in each pulse train, i.e., the onset response, was excluded from the analysis. In this example, the highest frequency to which the neuron responded in a phase-locked manner (P < 0.01, Fmax) was 70 pps. In addition, the entrainment, i.e., the average number of driven spikes per stimulus pulse, normalized for the number of stimulus trains, is plotted for each stimulus repetition rate in Fig. 2B.
|
Based on the STCs, the locations of single neurons were assigned to either ICX or ICC. Fmax and first spike latencies were analyzed separately for the two nuclei (Vollmer et al. 1999). Neurons from penetrations in which STCs were incomplete and did not allow a clear definition of the border between the two nuclei for either pulsatile or sinusoidal stimulation were assigned as ICX or ICC neurons if their recording locations fell into the nonoverlapping depth range of either nucleus calculated for the given animal group. For example, in control animals the recording depths for ICX neurons ranged from 300 to 1,880 µm, and for ICC neurons the range was 1,355–5,500 µm. Therefore neurons from incomplete STCs were included as ICX neurons if their recording depth was <1,354 µm and included as ICC neurons if their recording location was >1,881 µm. Neurons that did not meet these criteria were excluded from the analysis (see RESULTS) (Vollmer et al. 1999).
Cochlear histology
After completion of the electrophysiological experiment, the cochleae of all long-deafened animals were preserved for histology. The methods for the preparation of cochlear specimens were identical to those described by Leake et al. (1999) and will be described here only briefly. The cochleae were perfused, decalcified and embedded in LX resin. Semithin sections (1–2 µm) were stained with toluidin-blue to assess the presence or condition of the organ of corti and the survival of SGC as a function of cochlear location. SGC density in Rosenthal's canal was determined using a point-counting method (Leake and Hradek 1988; Leake et al. 1999). Earlier studies in normal hearing animals using this method have provided normative data for the cat spiral ganglion (Leake and Hradek 1988). These data served as a control reference in the present study and allowed the SGC density of the long-deafened cats to be expressed as percent of normal.
Statistical comparisons
The level of significance (
) is specified as P < 0.01 in the present study. The Student's t-test was used for comparisons of the SGC survival (Fig. 3). Because the physiological data were not normally distributed, the Mann-Whitney U test was used for comparisons of Fmax and latencies among the three different groups and between IC nuclei. Sets of three independent statistical tests were required for comparisons between groups within the ICX and the ICC. Because multiple statistical comparisons increase the likelihood of erroneously obtaining significant differences (type I error), a Bonferroni correction was used to adjust the level of significance ( adj = /3 = 0.01/3 = 0.003) for the statistical comparisons. For comparisons between ICX and ICC for the individual groups, pairwise single comparisons without correction were used.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Figure 3A illustrates the organ of corti of a normal hearing cat. In contrast, Fig. 3B shows an example of the severe degeneration of the organ of corti observed in the long-term deafened animals included in this study. Histological examination revealed that there were no surviving inner or outer hair cells in any of the LDU or LDS animals. Further, cochleae from long-deafened animals (Fig. 3D) demonstrated a severe loss of SGC and myelinated nerve fibers when compared with cochleae from control animals (Fig. 3C).
Figure 4A shows the morphometric SGC data for the six LDU and the five LDS animals. The volume ratios (densities) of the SGC somata are averaged for the 10% sectors of the spiral ganglion from the cochlear base to the apex (normalized for basilar membrane length of each individual cochlea). All SGC densities reported in this study refer to the implanted left cochlea in each animal. The data in each group are expressed as percent of normal SGC density as described previously (Leake and Hradek 1988). Severe degeneration was observed throughout the cochlea, and the mean SGC density was <14% of normal for each of the sectors in both long-deafened groups. Figure 4B displays the SGC density averaged across all cochlear sectors. The mean SGC density in the LDU group was reduced to 7.9% of normal, whereas in the LDS group the mean SGC density was reduced to only 3.1% of normal. This difference was statistically significant (Student's t-test; P < 0.01).
|
Table 1 summarizes the density of SGC (as percent of normal) averaged across all cochlea sectors for the individual cats. The data indicate that degeneration of the spiral ganglion was very severe in all 11 animals.
Maximum following frequencies (Fmax)
The responses of 659 single neurons to pulse trains of increasing frequencies were recorded in control, LDU and LDS animals. Fmax was estimated for 261 neurons in control animals, for 164 neurons in LDU animals, and for 215 neurons in LDS animals. As assessed by the STC for each recording penetration (e.g., Fig. 1), 15% of all neurons (n = 97) were located in the ICX, and 78% of all neurons (n = 499) were located in the ICC.
Approximately 7% of all neurons (n = 44) could not be assigned to either ICX or ICC and were therefore excluded from the analyses. Because STC widths tend to be very broad and dynamic ranges very small in long-deafened animals (Vollmer et al. 2000), a clear W-shape of the STCs often could not be identified, and boundaries between ICX and ICC were difficult to determine. Thus the number of neurons that could not be assigned to either ICX or ICC was somewhat larger in both LDU (9.8%, n = 16 neurons) and LDS animals (9.8%, n = 21 neurons) compared with unassigned neurons in the control group (2.7%, n = 7 neurons).
Quantitative distributions of Fmax.
Figure 5 illustrates the distributions of Fmax values and the median Fmax for all IC neurons that were assigned to either ICX (
) or ICC (
) in the three experimental groups. Across the three groups of cats, >99% of all ICC neurons had Fmax <300 pps. Generally, Fmax values for IC neurons in each group covered a fairly broad range of frequencies, but the distributions of Fmax for ICC neurons always extended to higher frequencies than that of ICX neurons. For ICC neurons, however, the distributions differed markedly for the three groups. The distribution of ICC neurons in LDU cats peaks at the lowest Fmax (60–80 pps, median Fmax = 70 pps; Fig. 5B), the distribution of the control group peaks at an intermediate Fmax (80–100 pps, median = 95 pps; Fig. 5A), and the distribution of the LDS cats peaks at the highest Fmax (120–140 pps, median = 135 pps; Fig. 5C).
|
|
Topographic distribution of Fmax.
Previous studies have suggested that ICX neurons have a weak tonotopic organization with characteristic frequencies (CFs) tending to decrease with increasing recording depth (Aitkin et al. 1975). In contrast, the ICC has a clear tonotopic gradient from low to high CFs with increasing recording depth (Brown et al. 1997; Merzenich and Reid 1974; Snyder et al. 1990). To determine if there is a correlation between the frequency-following capability of IC neurons and the tonotopic gradients in the nuclei, it was necessary to first exclude the possibility of a confounding relationship between temporal resolution and the locations of minimum ICC thresholds (best location; BL) for each stimulating electrode pair. Analyses showed that temporal resolution did not vary systematically relative to BL. Instead, the distributions of both Fmax and latencies as a function of depth relative to BL were relatively flat for all three groups of animals. Thus inter-animal or -electrode differences in BLs did not influence the analysis of temporal resolution versus recording depth when normalized to the ICX/ICC border.
We then analyzed the spatial distributions of Fmax along IC depth for all neurons assigned to either ICX or ICC in the three groups (Fig. 7). The recording depths are not normalized to the border between ICX and ICC in these plots. Instead, the actual recording depths noted during the experiment are shown. Neurons from STCs that allowed a clear identification as either ICX or ICC neurons and neurons that were assigned to either nucleus based on criteria described in METHODS are shown separately. As mentioned in the preceding text, the relatively flat STCs and the severely reduced dynamic ranges in the long-deafened animals (Vollmer et al. 2000) made it particularly difficult to determine the boundaries between ICX and ICC. As a result, the number of identified ICX and ICC neurons is markedly reduced in these animals. However, Fig. 7 documents that the assigned neurons fall within the range of identified neurons for both ICX and ICC in each group of animals. Although we cannot exclude the possibility of a misassignment of individual neurons, Fig. 7 shows that misassignments would be limited to a few neurons located close to the border between ICX and ICC and would not influence the overall outcome of our topographic analysis.
|
If temporal resolution of IC neurons and the tonotopy in the IC are correlated, Fmax should decrease with recording depth (decreasing CF) in the ICX and should increase with recording depth (increasing CF) in the ICC. Consequently, the linear regression lines for Fmax in the ICX should have negative slopes with increasing IC depth, and those in the ICC should have positive slopes.
In fact, with the exception of the ICC in control animals, all the slopes of the regression lines for Fmax in the ICX and ICC are positive (Fig. 7). However, all correlation coefficients for both nuclei data are relatively small (ICX: all R < 0.45, ICC: all R < 0.16), and with the exception of the ICX data in the LDS groups, none of the correlations are statistically significant. If the exceptional regression analysis was repeated with the depths of the recording locations normalized to the border in a more limited set of ICX neurons (raw data from Fig. 8C), the correlation between Fmax and ICX depth in LDS animals was no longer significant. Overall the findings indicate that there was no systematic relationship between Fmax and the tonotopic (CF) gradient of either ICX or ICC in any of the groups. In the three groups of cats, neurons in the ICC exhibited a relatively broad range of Fmax at any given recording depth with the exceptions of the sparcely sampled superficial and deepest recording locations.
|
Plotted in this manner, there is no apparent relationship between Fmax and IC depth (i.e., CF gradient) for ICX neurons. In the ICC of control animals (Fig. 8A) neurons show a broad maximum in Fmax around the center of the ICC, whereas temporal resolution decreases at the more superficial (dorsolateral) and deep (ventromedial) recording locations. In the ICC of LDU animals (Fig. 8B), the median Fmax values exhibit a broad decline throughout the nucleus, and at all depth sectors the values are equal to or smaller than the median Fmax for controls. In contrast, ICC neurons from LDS animals (Fig. 8C) show a broad increase in Fmax across the entire ICC compared with control animals. These results indicate that the regions of increased Fmax after high-frequency stimulation are not correlated with the tonotopic gradient in the ICC.
First spike response latencies
Onset latencies for single neurons were recorded as a second measure of temporal resolution. Of the 645 neurons for which latency values were obtained, 270 were recorded in control animals, 159 in LDU animals, and 216 in LDS animals. A total of 100 neurons (15.5% of all neurons) were located in the ICX, 501 neurons (77.7%) were located in the ICC, and 44 neurons (6.8%) could not be assigned to either ICX or ICC and were excluded from the analyses.
Quantitative distributions of first spike latencies.
In Fig. 9 the distributions of first-spike latencies in the three groups are shown separately for the ICX () and the ICC (). Values and arrows indicate the median latencies for each nucleus. Generally, latencies of ICC neurons are clearly distributed toward shorter latencies compared with ICX neurons.
|
The median first spike latency data for ICX and ICC neurons and statistical comparisons (Mann-Whitney U test) are summarized in Fig. 10. In control and LDS animals, neuronal latencies in the ICX are significantly longer than those in the ICC (P < 0.01). The ICX neurons from LDU animals show a tendency toward longer median latencies [9.5 ± 1.7 (Q) ms] when compared with either control or LDS groups [8.9 ± 1.6 and 8.9 ± 1.0 (Q) ms, respectively]. However, these differences are not statistically significant.
|
Topographic distribution of latencies.
The spatial distributions of onset latencies along IC depth are shown in Fig. 11. The data are identified as ICX (
and 
) and ICC ( and ) neurons. Neurons from STCs with a well-defined border that allowed a definite identification as either ICX or ICC neurons, and those neurons that were assigned to either nucleus by the criteria described in METHODS are identified separately. Because both Fmax and latencies were determined for the majority of neurons, the ranges of recording locations in the two nuclei and the regions of overlap between ICX and ICC neurons in each group are virtually identical to those reported for the spatial distributions of Fmax (see Fig. 7).
|
The slopes of the linear regression lines in both ICX and ICC of all three groups are negative. However, all correlation coefficients are very small (all R < 0.33), and with the exception of those for the ICX and ICC in LDS animals the correlations are not statistically significant. When the regression analysis was repeated for neurons from STCs with a well-defined border and with the depth normalized to the border (raw data from Fig. 12C), the correlation between Fmax and both ICX and ICC depths in LDS animals is no longer significant.
|
Median latencies from control and LDS animals (Fig. 11, A and C, respectively), show relatively flat distributions across the IC with a slight tendency for the shortest latencies to be located around the center of the ICC. Median latencies in the LDU animals decrease progressively with increasing ICC depth (Fig. 12B). However, the number of ICC neurons per depth range is very limited in these animals, and no neurons were recorded at the deepest ICC locations (>3 mm re border) where control animals usually show a tendency for an increase in latencies.
With the exception of the median latencies at depths of 2.5–3 mm (n = 3), the median latencies in the LDU group are longer throughout the entire IC as compared with both control and LDS animals.
The results indicate that first spike latencies in control and LDS animals are not systematically correlated with the tonotopic gradient of either nucleus, and similar to the changes in Fmax, changes in latencies occurred relatively uniformly across the entire ICC.
Correlation between Fmax and response latencies
Langner and colleagues (1987) reported in normal hearing cats a significant correlation between the onset latencies of IC neurons and their best modulation frequencies (BMF; modulation frequency that evokes the strongest neuronal response) in response to amplitude modulated (AM) tones.
Despite different stimulus conditions (acoustic vs. electric) and different criteria used for the determination of temporal resolution (BMF vs. Fmax), similar estimates of temporal resolution of IC neurons have been reported previously for both acoustic, AM-evoked responses and electrically evoked responses to unmodulated pulse trains (Snyder et al. 1995; Vollmer et al. 1999). Among the similarities are the temporal patterns of IC responses (PSTHs), the latency distribution, and the range and distribution of the frequency following capabilities of IC neurons (Fmax and BMF).
Figure 13 shows the relationship between and the covariation of onset latencies and Fmax for ICX and ICC neurons in the three groups of animals. To produce the curves in Fig. 13, the equation used by Langner and colleagues was corrected for the instantaneous onset of electrical pulses and the lack of cochlear delays in electrical stimulation [onset latency = (5.1 ± 0.9) ms + (1.2 ± 0.2) 1/CF + 0.16 ± 0.03) 1/Fmax] (cf. Snyder et al. 1995). The two curves in each panel encompass the lowest and the highest first spike latencies predicted by the modified equation. The majority of both ICX and ICC neurons from each experimental group have latencies within the predicted range of latencies for given Fmax marked by the two curves. The closest agreement between latencies predicted by the modified equation and the observed data are found for neurons from LDS animals. Generally, the correlations in Fig. 13 show a decrease in latencies with increasing Fmax, suggesting an inverse correlation between Fmax and onset latencies in response to electrical stimulation of the auditory nerve.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
The SGC density in both unstimulated and chronically stimulated long-deafened cats was severely diminished as compared with normal. LDU animals had an average SGC density of 8% of normal, and LDS animals an average of 3% of normal. This difference in SGC density between the two groups of long-deafened animals was statistically significant. Previous studies have shown that SGC degeneration continues progressively for many years after aminoglycoside-induced hair cell loss (Hardie and Shepherd 1999; Leake and Hradek 1988; Leake and Rebscher 2004; Shepherd and Javel 1997; Xu et al. 1993). Thus the overall lower SGC survival in the LDS animals may be explained on the basis of the older age at study (68.8 mo) in these animals as compared with the LDU animals (duration of deafness = age at study: 46.5 mo).
Effects of long-term deafness and chronic electrical stimulation on temporal resolution
Central nucleus of the IC (ICC). Long-term auditory deprivation and chronic electrical stimulation greatly affected temporal resolution in the ICC of long-deafened animals.
Long-term auditory deprivation.
Long-term deafness per se (LDU cats) resulted in a significant decrease in temporal resolution of ICC neurons (i.e., lower Fmax and longer latencies) as compared with control subjects. A previous study (Snyder et al. 1995) of neonatally deafened, unstimulated adult animals also reported a lower mean Fmax value (86 pps) as compared with normal control cats (93 pps), but this difference in Fmax did not achieve statistical significance. The mean age at study of these animals was 41.7 mo (unpublished data) and was, therefore only slightly lower than that of the LDU animals included in the present study. Because Snyder and colleagues did not distinguish between responses from ICX and ICC neurons, it seems likely that inclusion of ICX neurons with low temporal resolution in the analysis masked the differences in Fmax between the unstimulated and control animals.
Shepherd and colleagues (1999) also reported a reduction in the temporal resolution of ICC neurons (lower Fmax and longer latencies) in neonatally bilaterally deafened, unstimulated animals when compared with control animals. It should be noted, however, that the two animals studied had substantially shorter durations of deafness (12 and 13 mo) compared with the LDU animals in the present study (mean: 46.5 mo), suggesting that the negative effects of auditory deprivation on temporal resolution of ICC neurons may occur earlier than the prolonged durations of deafness reported in the present study.
Shepherd and colleagues (1999) also reported an even greater increase in latency for ICC neurons in a single long-deafened (deafened as juvenile), unstimulated animal. However, given the limited data available, no conclusion can be drawn about the exact time course of functional changes caused by auditory deprivation.
Overall, however, previous work (Shepherd et al. 1999; Snyder et al. 1995) and the present study agree that auditory deprivation clearly reduces temporal resolution of neurons in the ICC.
Chronic electrical stimulation.
In a previous study, we reported that chronic electrical stimulation delivered in the developing auditory system caused significant functional changes in the IC. Specifically, in neonatally deafened animals that were implanted 6–8 wk after birth and chronically stimulated for several months with high-frequency pulsatile signals, neurons in the ICC had significantly higher temporal resolution than neurons recorded from control animals (Vollmer et al. 1999). The present study extends these findings by examining stimulation effects in the mature, long deafened auditory system and demonstrates that the introduction of chronic high-frequency stimulation even after long periods of complete auditory deprivation results in a significant increase in temporal resolution of ICC neurons. In fact, the increase in temporal resolution observed in LDS animals is virtually identical to that previously reported for the neonatally deafened, early stimulated animals (Vollmer et al. 1999).
Moreover, similar to the previous findings in early-stimulated animals, the increase in Fmax after chronic high-frequency stimulation of LDS animals was not only significantly higher than in LDU animals but also higher than in control animals. These findings are remarkable because the prolonged periods of neonatal deafness in the LDS animals resulted in severe degeneration of the spiral ganglion and auditory nerve fibers. Initially we wondered if the severe peripheral pathology observed in long-term deafness would reduce or eliminate any benefit for temporal resolution of chronic electrical stimulation of the cochlea. However, the data reported here clearly indicate that the long-deafened, mature auditory system is highly capable of plastic changes and substantial functional recovery including the potential to reverse at least one aspect of functional degradation (temporal resolution) observed after prolonged periods of auditory deprivation.
In fact, the LDS animal with the shortest period of chronic stimulation (7 wk; K56) demonstrated an increase in median Fmax of ICC neurons that exceeded that of control animals and reached the second highest Fmax value (142 pps) in the group of LDS animals. These data suggest that despite severe peripheral pathology, even short periods of chronic stimulation are sufficient to induce marked changes in temporal resolution in the long-deafened auditory system. However, the present study was not designed to systematically investigate the exact time course of functional changes after chronic stimulation.
Consistent with increased temporal resolution of ICC neurons in neonatally deafened animals that were exposed to chronic electrical stimulation either during development or as adults, clinical studies also have shown evidence of improved temporal processing with device use even in prelingually deafened adult cochlear implant subjects (Busby et al. 1991). Presumably, peripheral pathology is not the limiting factor for functional plasticity in the long-deafened central auditory system.
Topographic distribution of temporal resolution versus recording location. The normalization of the recording depth to the border between ICX and ICC allows a more exact investigation and interpretation of the relationship between different neuronal response properties and their recording location in the IC (i.e., inferred relative CF). Because of the large variability in the location of the border between ICX and ICC, these relationships or correlations can otherwise easily be masked or distorted.
Temporal resolution was not systematically correlated with the tonotopic gradient in the ICC in any of the investigated groups. That is, increasing depth in the ICC does not correspond to increasingly higher temporal resolution. Instead, temporal resolution of ICC neurons was relatively evenly distributed across each of the nuclei and, in the control and LDS animals, showed only a broad maximum around the center of the nucleus with a decrease in temporal resolution toward both more dorsolateral (superficial) and more ventromedial (deep) regions. Long-term deafness resulted in a broad decrease in temporal resolution across the entire ICC, and chronic electrical stimulation resulted in an increase of temporal resolution in neurons throughout the ICC. There was no evidence of a preferential or selective effect on neurons in any particular CF region of the ICC, regardless of whether the animals had a history of complete auditory deprivation or chronic electrical stimulation.
These results support findings reported in earlier studies from our laboratory of responses to electrical stimulation in control and neonatally deafened, early stimulated animals (Vollmer et al. 1999). Further, they are also consistent with earlier studies using acoustic stimulation by Langner and colleagues (1987), who described a relatively weak correlation between onset latency and CF in normal hearing cats. Shirane and Harrison (1991) described a small but significant relationship between latency and electrode depth in the IC of two normal control chinchillas. However, the lack of differentiation between ICX and ICC neurons may be responsible for this apparent correlation.
In contrast, Shepherd and colleagues (1999) reported an orderly decrease in latency of ICX neurons from 25 to 10 ms with increasing recording depth in control, bi- and unilaterally deaf animals but not in their long-deafened animal in which neurons could not be recorded in IC depths <2,500 µm. We also saw a weak tendency for ICX latencies to decrease with increasing depth in our three groups of animals. However, the significance of this finding is unclear because the tonotopic gradient of the ICX is from high to low frequencies. Therefore if temporal resolution is related to CF, we would expect latency to increase with lower CFs (i.e., increasing depth), not to decrease.
Mechanisms.
The specific mechanisms underlying the described changes in temporal resolution of ICC neurons after long-term deafness and after chronic electrical stimulation of the long-deafened auditory system are not known. Among possible explanations for the decreased frequency following ability of ICC neurons after long-term auditory deprivation are changes in the balance of excitatory and inhibitory influences on IC neurons (Raggio and Schreiner 1999; Schreiner and Raggio 1996), a weakening of individual excitatory synapses (Kotak and Sanes 1997), a decrease in excitatory neurotransmitter release (Vale and Sanes 2002), and a decrease in synaptic density in the IC of neonatally deafened cats (Hardie et al. 1998). Moreover, the loss of myelin observed after long-term deafness may lead to an increase in membrane capacitance (Koles and Rasminsky 1972; Tasaki 1955) that could reduce the efficiency of a neuron in responding to electrical stimuli and increase the likelihood of conduction block. Also associated with the loss of myelin and partial neural degeneration are prolonged refractory periods and an increased vulnerability of the propagating spike (Cragg and Thomas 1964; Felts et al. 1997; Koles and Rasminsky 1972; McDonald and Sears 1970; Shepherd and Javel 1997; Smith and McDonald 1999; Tasaki 1955). These changes result not only in an increase in response latency and jitter but in a general reduction in conduction and synaptic efficacy of afferent connections along the ascending central auditory system. This reduced efficacy of pathway and synaptic transmission may contribute to lower temporal resolution particularly in neurons in the ICC that receives mainly input from afferent auditory neurons. In contrast, neurons in the ICX receive major inputs from the somatosensory system (e.g., Aitkin 1986) and, therefore may be less affected by auditory deprivation than ICC neurons.
The present study shows that chronic electrical activation of the cochlea in LDS animals not only reverses the described functional degradation and restores the temporal resolving capacity to a "normal" level, but also significantly increases the Fmax of ICC neurons above the level of control animals. These findings suggest that neural remodeling or auditory plasticity can occur despite extensive structural and functional degeneration of the auditory nerve and the central auditory system. The specific underlying functional and/or structural modifications in afferent projections are unknown. Changes may occur prior to and/or at the level of the IC (e.g., Eysel et al. 1981; Kaas 1996; Keller et al. 1990) and may include local Hebbian-type synaptic processes (Cruishank and Weinberger 1996; Diamond et al. 1993), changes in the synaptic organization or strength of existing afferent connections (e.g., modification in synaptic size or relocation of synapses to more effective sites on the target neuron), sprouting of new afferents that results in increased synaptic density, and/or alterations in membrane properties. Each of these mechanisms could lead to an increase in synaptic efficacy, higher synchrony in the neuronal excitation pattern and, thus to an increase in the temporal resolution of IC neurons. Moreover, together with previously reported results from neonatally deafened, early stimulated animals, the present findings suggest that higher-frequency electrical stimulation may be more effective in modulating the inhibitory and excitatory mechanisms in the ICC than lower-frequency stimulation or normal acoustic stimulation (Vollmer et al. 1999).
Based on these hypotheses, the broad increase in temporal resolution observed across the entire ICC after chronic stimulation (LDS animals) could be related to a relatively broad spread of current into the modiolus. In long-deafened animals, the severe reduction of peripheral auditory nerve processes and SGC may provide a particularly low impedance pathway for the current to spread into the modiolus (Frijns 1995) and to excite auditory nerve fibers over a very broad range of CFs. Also, the broader STC widths and reduced dynamic ranges observed in these animals would lead to broader current spread at a given suprathreshold level compared with a subject with narrower STC widths and larger dynamic ranges. Given the assumption that high-frequency stimulation is especially effective in modulating the efficacy of afferent stimulation prior to and/or at the level of the ICC (Vollmer et al. 1999), the particularly broad current spread of suprathreshold electrical pulsatile stimuli in long-deafened animals (Vollmer et al. 2000) may explain the broad increase in temporal resolution of single neurons across most of the ICC.
External nucleus of the IC (ICX)
The present study provides the first report characterizing driven activity and temporal resolution in the ICX of long-deafened cats. There were no significant differences in the temporal resolution (median Fmax and first spike latencies) of ICX neurons among the three groups of animals. These observations are consistent with previous data from neonatally deafened animals that received chronic electrical stimulation immediately after deafening at an early age (Vollmer et al. 1999). Those animals also did not show any differences in the Fmax of ICX neurons. Moreover, ICX data from all groups in the present study are very similar to previously reported data from ICX neurons in those early-stimulated animals. Together these findings suggest that the temporal resolution of neurons in the ICX is not affected by the stimulation history or auditory experience (e.g., the temporal parameters of stimulation) (Vollmer et al. 1999), by the duration of deafness, or by the age at onset of stimulation.
One explanation for the absence of plastic changes in the temporal response properties of ICX neurons after long-duration deafness and chronic electrical stimulation is the strong somatosensory input to the ICX (e.g., Aitkin 1986). This input may maintain or dominate the frequency following capabilities of ICX neurons and make them less sensitive to auditory deprivation or changes in auditory inputs. Another possible explanation is that the ICX, in contrast to the ICC, receives major descending projections from both the primary and nonprimary auditory cortices (e.g., Andersen et al. 1980; Coleman and Clerici 1987
); n, number of animals. B: mean SGC densities averaged over all cochlear sectors for the 2 experimental groups summarized by the bars and numbers. *, statistical significance (Student's t-test; P < 0.01). Error bars indicate SD.
and values indicate median latencies calculated individually for the ICX (
