www.elsevier.com/locate/brainres

Brain Research 997 (2004) 40–51

Research report

Time course of efferent fiber and spiral ganglion cell degeneration

following complete hair cell loss in the chinchilla

Sandra L. McFadden*, Dalian Ding, Haiyan Jiang, Richard J. Salvi

Center for Hearing and Deafness, University at Buffalo, 215 Parker Hall, Buffalo, NY 14214, USA

Accepted 20 October 2003

Abstract

Ethacrynic acid (EA) is known to interact with aminoglycoside antibiotics such as gentamicin (GM). In the chinchilla, co-administration

of GM and EA can produce hair cell lesions ranging from a small loss of outer hair cells (OHCs) in the base of the cochlea to complete

destruction of all hair cells, depending on dosing parameters. Although hair cell loss has been characterized, little is known about the fate of

efferent fibers or spiral ganglion neurons (SGNs) in this model. To study the time course of efferent fiber and SGN loss, chinchillas were

injected with GM (125 mg/kg IM) followed immediately by EA (40 mg/kg IV). Estimates of efferent fiber loss and density changes were

made after 3 days or 1, 2, 3, or 4 weeks of survival. Estimates of SGN loss and density changes were made after 15 days or 1, 2, 4, or 6

months of survival. Cochlear function was rapidly abolished and all cochlear hair cells were missing within 24 h after treatment. Inner hair

cells (IHCs) in the middle turn of the cochlea died earlier than cells in the apex or base, and OHCs in Rows 1 and 2 died earlier than OHCs in

Row 3. Degeneration of efferent nerve fibers began 3–7 days post-injection, versus 15–30 days for SGNs, and the loss of efferent fibers was

essentially complete within 1 month, versus 2–4 months for SGNs. The rapid time course of efferent fiber and SGN loss in the chinchilla

may make it a practical model for studying mechanisms of neural loss and survival in the mammalian inner ear.

D 2003 Elsevier B.V. All rights reserved.

Theme: Development and regeneration

Topic: Neuronal death/sensory systems

Keywords: Aminoglycoside antibiotic; Loop diuretic; Ototoxicity; Inner ear pathology; Ganglion neuron; Nerve fiber; Cochlea

1. Introduction

Research on topics related to the protection, rescue and

repair of neurons and their processes is aided by the use of

animal models in which neural degeneration occurs in a

predictable manner over a well-defined time course. One

method for producing an animal model of spiral ganglion

neuron (SGN) and nerve fiber degeneration in the cochlea

is to destroy cochlear hair cells by administering a large

dose of an aminoglycoside antibiotic (AAB) and ethacrynic

acid (EA) [14–16,21,24–26]. SGNs degenerate following

IHC destruction, with a time course that depends on factors

such as species and magnitude of IHC loss. The AAB/EA

deafening procedures that have been used with cats,

0006-8993/$ - see front matter D 2003 Elsevier B.V. All rights reserved.

doi:10.1016/j.brainres.2003.10.031

* Corresponding author. Tel.: +1-716-829-2001x13; fax: +1-716-829-

2980.

E-mail address: mcfadden@buffalo.edu (S.L. McFadden).

monkeys and guinea pigs typically result in variable and

incomplete hair cell loss, thereby confounding studies of

SGN degeneration and treatments aimed at preserving

them. Furthermore, SGN degeneration is a slow process

that can take more than a year to complete in cats and

guinea pigs [15,21,25]. The slow time course and variable

magnitude of SGN degeneration can be a major drawback

for studies of neural death and survival mechanisms. We

recently described a GM/EA deafening procedure that

reliably destroys all cochlear hair cells in chinchillas [17].

In these animals, we also observed rapid degeneration of

SGNs and nerve fibers in the cochlea, suggesting that the

chinchilla may be a practical model for studies of neural

protection, rescue and repair. The current study describes

the time course of SGN and nerve fiber degeneration in the

chinchilla cochlea following a single injection of GM (125

mg/kg IP) and EA (40 mg/kg IV) that rapidly destroyed

hair cells and abolished cochlear function. We were par-

ticularly interested in comparing the time course of efferent

S.L. McFadden et al. / Brain Research 997 (2004) 40–51 41

fiber degeneration and afferent fiber/SGN degeneration

caused by the GM/EA combination, given speculation

about a trophic role of lateral olivocochlear (LOC) efferent

fibers in the cochlea [18].

Fig. 1. Surface preparation from the middle turn of a normal cochlea,

‘‘stained’’ with silver nitrate for hair cell counting. Note the presence of

three rows of OHCs (OHC1, OHC2, OHC3) and one row of IHC, separated

by pillar cells (PC). The arrow shows where one OHC is missing from the

third row. OHC were replaced by a phalangeal scar (arrow) formed by

supporting cells. Scale bar = 20 Am.

2. Materials and methods

2.1. Subjects

Subjects were 41 healthy adult chinchillas obtained from

a commercial breeder. Four chinchillas were prepared for

electrophysiological recording by implanting chronic re-

cording electrodes at the round window niche (see below).

Thirty-six chinchillas, including the four animals with round

window electrodes, were anesthetized with ketamine (55

mg/kg IM) and acepromazine (0.5 mg/kg IM), then injected

with GM (gentamicin sulfate, Sigma Chemicals; 125 mg/kg

IM), followed immediately by EA (Sodium EdecrinR,Merck & Co., Inc., West Point, PA; 40 mg/kg into the right

jugular vein). For counts of efferent nerve fibers, animals

were sacrificed after survival times of 3 days or 1, 2, 3, or 4

weeks (N = 15, three animals at each time point). For counts

of SGNs and examination of nerve fibers in the habenula

perforata of the osseous spiral lamina (OSL), animals were

sacrificed after survival times of 15 days or 1, 2, 4, or 6

months (N = 15, three animals at each time point). Six

chinchillas were sacrificed at 6, 12 or 24 h post-injection

for verification of rapid hair cell loss [17]. Five untreated

chinchillas served as normal controls for SGN counts.

2.2. Electrophysiology

Compound action potentials (CAP), cochlear microphon-

ics (CM) and summating potentials (SP) were monitored in

four chinchillas to observe the time course of decline of

cochlear function following GM/EA injection. The chin-

chillas were prepared for electrophysiological recording by

implanting a Teflon-coated silver wire ball electrode at the

round window niche of the right ear under surgical anes-

thesia (ketamine, 55 mg/kg IM; acepromazine, 0.5 mg/kg

IM). Animals were allowed to recover for at least 1 week

prior to GM/EA injection. Stimuli were digitally generated

tones at 1, 2, 4 and 8 kHz, presented at a rate of 19 per s

from a loudspeaker located 25 cm directly in front of the

animal in a double-walled sound attenuating booth. For

CAP, tones were 7 ms duration, 1 ms rise/fall; for CM and

SP, tones were 17 ms duration, 1 ms rise/fall. Electrical

activity recorded from the round window electrode was

amplified (20,000� for CAP and CM, 5000� for SP),

filtered (100–3000 Hz for CAP; 10–10,000 for CM; 1–300

Hz for SP), digitized by an A/D converter on a signal

processing board in a computer, and averaged for 100

presentations at each stimulus level. Stimulus level was

incremented in 10 dB steps, from 0 to 70 dB SPL for CAP,

from 50 to 90 dB SPL for CM, and from 40 to 90 dB SPL

for SP. Responses were recorded prior to GM/EA injection,

and at 0.5, 1.0, 1.5, 2.0, 24, and 72 h following injection.

2.3. Morphological assessment

Animals were sacrificed with CO2 and decapitated, and

the bullae were quickly removed. As described in detail by

Azeredo et al. [1], two distinct populations of olivocochlear

efferent fibers, referred to as LOC and medial olivocochlear

(MOC) efferents, innervate the chinchilla cochlea. The vast

majority of olivocochlear efferent fibers are small, unmy-

elinated fibers that arise from neurons located within the

lateral superior olivary nucleus; these LOC efferent fibers

project almost exclusively to the ipsilateral cochlea, where

they form synapses with the afferent fibers beneath the IHCs.

A smaller number of olivocochlear efferent fibers are large,

myelinated fibers that arise from neurons located in the

medial region of the superior olive, particularly the dorso-

medial peri-olivary nucleus; theseMOC fibers project to both

cochleas (in a 1:4 ipsilateral to contralateral ratio in chinchil-

la), lose their myelin at the habenula perforata, and cross the

tunnel of Corti to form synapses directly with the OHCs.

For determination of LOC and MOC efferent fiber loss,

cochleas were prepared using acetylcholine esterase (AChE)

histochemistry as described previously [2,30]. In AChE

stained cochleas, MOC fibers can be seen crossing the

tunnel of Corti en route to the OHCs, whereas LOC fibers

travel in the inner spiral bundle (ISB) in the IHC region (see

Fig. 5A). Cochleas were first perfused with 4% paraformal-

Fig. 3. Mean cochleogram (N= 3 ears) showing the percentage of inner hair

cells (IHC) and OHCs in Rows 1, 2 and 3 (OHC1, OHC2, OHC3) missing

12 h after GM/EA treatment. Note that IHC loss (solid line) was greater in

the middle region of the cochlea than in the base or the apex. Based on the

frequency–place map of Greenwood [9], the place where IHC loss peaked

(approximately 55% distance from the apex) transduces energy around 2

kHz. Note also that OHC loss in Row 3 (heavy dashed line) lags behind

OHC loss in Rows 1 or 2 (light dashed and dotted lines, respectively).

Standard deviations of each 10% segment are as follows: IHC, 2.51, 10.59,

17.43, 7.17, 15.11, 7.05, 0, 0, 5.67, 0; OHC1, 9.06, 0.60, 0, 0, 0, 0, 0, 0, 0,

0; OHC2, 2.68, 1.21, 0, 0, 0, 0, 0, 0, 0, 0; OHC3, 5.07, 16.37, 0.89, 4.26,

7.12, 5.87, 0, 0, 0, 0.

Fig. 2. Mean input/output functions for the CAP (A), CM (B), and SP (C)

for 4 kHz stimuli, measured before and at various times after GM/EA

injection. Note the rapid decrease in response amplitude within 30 min

post-injection, and the transient partial recovery of amplitudes at 1.5 h.

Responses at other tested frequencies (1, 2 and 8 kHz) were also abolished

within 24 h after injection. Bars show S.E.M.

S.L. McFadden et al. / Brain Research 997 (2004) 40–5142

dehyde in 0.1 M phosphate buffer (pH 7.2) for 30 min to 1

h, then dissected in 70% ethanol to remove the organ of

Corti as a flat surface preparation. The organ of Corti was

pre-incubated in 0.1 M acetate buffer (pH 6.0) for 30 min,

then incubated in AChE staining solution for 1 h. Specimens

were washed three times with acetate buffer, then treated

sequentially (1 min in each solution, at room temperature)

with 1 ml of 1% ammonium sulfide in distilled water; five

washes of 0.1 M sodium nitrate; 1 ml of 0.1% silver nitrate

in distilled water; five washes of 0.1 M sodium nitrate and

three washes of 0.1 M acetate buffer. Specimens were

mounted in glycerin on glass slides and examined under a

light microscope.

AChE-positive tunnel crossing nerve fibers (MOC effer-

ents) were counted in successive 0.24 mm segments from the

apex of the cochlea to the base. Counts were made at a point

close to the outer spiral bundle toward the medial edge of the

tunnel, where individual MOC fibers are more easily distin-

guished before they converge to form the tight bundle that is

visible in the middle of the tunnel. Counts from each ear

were entered into a software program that computed fiber

density (fibers per mm) in 10% segments from apex to base

(thereby compensating for differences in overall cochlear

length among animals). Averages for each group (normal, 3

days, 1, 2, 3 and 4 weeks) were computed separately for

apical and basal halves of the cochlea for comparisons of

MOC fiber density as a function of group and cochlear

region. MOC efferent fiber loss was estimated by comparing

data from GM/EA-treated chinchillas to our laboratory

norms based on counts from five normal chinchillas. Efferent

fibers were counted in both cochleas of each animal, then an

average loss was computed for each animal. Statistical tests

are, therefore, based on the number of animals in each group

rather than the number of ears examined.

S.L. McFadden et al. / Brain Research 997 (2004) 40–51 43

It is not possible to count individual AChE-labeled fibers

in the ISB, and both cochlear afferents and medial efferent

tunnel crossing fibers (TFC) can be interspersed among ISB

fibers [12]; therefore, LOC efferent loss could only be

roughly estimated from our material, based on changes in

the width of the ISB over time. The ISB was examined in

10-Am long samples (one from the apical turn and one from

the basal turn of the cochlea), and measurements were taken

at the narrowest and widest points along this segment.

Average ISB widths were computed for each group based

on 5–10 samples per group.

For counts of SGNs in Rosenthal’s canal and examina-

tion of nerve fibers in the OSL, the cochleas were perfused

through the round window with approximately 1 ml of 4%

paraformaldehyde in 0.1 M phosphate buffered saline (pH

Fig. 4. Structure of the organ of Corti at various times after the loss of all cochlear h

the normal organ of Corti shows OHCs from rows 1, 2 and 3 (OHC1, OHC2, OH

(HC), and outer pillar cells (OPC) bordering the tunnel of Corti (asterisk)), and an in

hair cells. B–F. Cross sections at various times after GM/EA injection: (B) 2 days;

7.4), then immersed in fixative for approximately 24 h. The

fixed cochleas were decalcified by immersion in Decal

(Baxter Scientific Products) for 2 days, dehydrated, and

embedded in Epon 812 resin. Serial sections were cut in a

plane parallel to the modiolus at a thickness of 3 Am(Reichert Super Nova microtome), mounted on slides and

stained with toluidine blue [2]. For estimates of SGN

numbers, every neuron with a nucleus was counted in every

fifth section through the modiolus [26]. To sample the entire

area of Rosenthal’s canal, approximately 40–50 sections per

modiolus were examined. Counts were obtained from five

normal animals, and from 15 animals treated with GM and

EA. To estimate the density of SGNs, modiolar sections

from the apical and basal turns of the cochlea were exam-

ined. The number of SGNs in a 0.2-mm diameter visual

air cells (all sections from the lower middle turn). (A) Cross section through

C3), supporting cells (Dieter’s cells (DC) beneath the OHCs, Hensen’s cells

ner hair cell (IHC). The tectorial membrane (TM) extends over the top of the

(C) 15 days; (D) 1 month; (E) 2 months and (F) 4 months. Scale bar = 10 Am.

S.L. McFadden et al. / Brain Research 997 (2004) 40–5144

field was counted for five samples from the apical turn, and

five samples from the basal turn of each cochlea. Group

averages for the number of SGNs per square millimeter,

based on two to five animals per group, were computed

separately for the apical and basal turns. In addition, every

fifth section containing habenula perforatae was examined

to determine the condition of nerve fibers in the OSL in the

middle turn. Because most nerve fibers were absent even at

2 weeks post-injection, counts of habenular nerve fibers of

GM/EA-treated animals were not made. The nerve fiber

counts from normal control animals that are presented in this

paper were obtained from six normal chinchillas as part of a

separate study [4]. Nerve fibers passing through 10 basal, 10

middle turn, and 10 apical openings in the OSL of each

cochlea were counted and averaged for the group.

The opposite cochlea of each animal was used to confirm

hair cell loss, using succinate dehydrogenase (SDH) histo-

Fig. 5. Surface preparations from the lower middle turn of the organ of Corti sh

progressive loss of efferent fibers after GM/EA injection (B–F). In A, thinner whi

large white arrow points to the ISB which contains LOC efferent fibers. Scale ba

chemistry as described previously [2]. The round and oval

windows were opened and 0.2 M sodium succinate in 0.1 M

phosphate buffer (pH 7.4) was slowly perfused through the

round window. The cochlea was immersed in sodium

succinate solution for 1 h at 37 jC, then in 10% formalde-

hyde for 24 h. The organ of Corti was dissected out,

mounted in glycerin as a flat surface preparation, and

examined under a light microscope at 400� magnification.

No hair cells were present in any cochlea examined.

A separate group of six animals was used to examine the

time course of hair cell loss following GM/EA treatment.

Cochleas from these animals were removed at 6, 12, or 24

h after GM/EA treatment and prepared as described previ-

ously [2]. Briefly, cochleas were perfused through the round

window with 0.5% silver nitrate in phosphate buffered saline

(pH 7.4), then fixed in 10% formalin for 2 h. The organ of

Corti was dissected out and mounted on a glass slide as a

owing AChE-positive (efferent) fibers in the normal cochlea (A) and the

te arrow points to TCF that are MOC efferents crossing the tunnel of Corti;

r = 10 Am.

S.L. McFadden et al. / Brain Research 997 (2004) 40–51 45

surface preparation, then exposed to indirect sunlight for

approximately 1 h to enhance staining contrast. Specimens

were examined under a light microscope at 400� magnifi-

cation, and hair cell loss was determined in successive 0.24-

mm segments, beginning at the apex. Fig. 1 shows a

representative surface preparation from the middle turn of

a normal cochlea. Hair cells were counted as present if their

stereocilia or cell bodies were visible, or as missing if a

phalangeal scar was visible (Fig. 1, arrow). Cochleograms

showing the percent of missing hair cells as a function of

basilar membrane position, normalized to percent distance

from the apex, were prepared for animals in the 12 h group.

Counts from each ear were entered into a software program

that computed hair cell density (cells per mm) in 10%

segments of the cochlea from apex to base, and compared

these values to lab standards based on hair cell counts from

nine normal, young adult chinchillas. Estimates of frequency

corresponding to percent distance from the apex were made

using the frequency–place map of Greenwood [9].

All procedures regarding the use and care of animals in

this study were reviewed and approved by the Institutional

Animal Care and Use Committee at University at Buffalo.

Fig. 6. MOC efferent loss after GM/EA treatment. (A) Loss of MOC

efferents averaged across 10% segments of the cochlea, from apex to base.

Bars show one S.E.M. above or below the mean. (B) Density of MOC

efferents (fibers per mm) as a function of group (reflecting time re: GM/EA

treatment) and cochlear region (apical half of the cochlea versus the basal

half). Bars show one S.D. * Density in basal half significantly greater than

density in apical half (Tukey tests, P< 0.05). ** Density in apical half

significantly greater than density in basal half (Tukey tests, P< 0.05).

3. Results

3.1. Decline of cochlear function following GM/EA

injection

CAP, CM and SP amplitudes declined rapidly at all tested

frequencies following GM/EA injection, with complete and

permanent losses occurring in all four animals within 24 h.

Mean input/output functions for 4 kHz stimuli are shown in

Fig. 2. The CAP (Fig. 2A) was essentially abolished within

30 min after GM/EA injection. At the highest stimulus level

tested (70 dB SPL), the amplitude of the CAP at 30 min was

less than 6% of the pre-injection value. The CAP showed a

small, transient increase at 70 dB SPL between 1.0 and 1.5 h,

reaching 18% of the pre-injection value at 1.5 h, then

declining again, to 6% at 2 h. At 24 and 72 h, the CAP

was completely absent.

CM (Fig. 2B) and SP (Fig. 2C) amplitudes also showed

rapid initial decreases, followed by a transient recovery at

1–1.5 h, then complete disappearance by 24 h. At 0.5 h post-

injection, the CM was 26% and the SP was 12% of their

respective pre-injection values. At 1.5 h, the CM and SP had

both recovered to approximately 30% of their pre-injection

values at 70 dB SPL, and to approximately 70% of their pre-

injection values at 90 dB SPL. By 24 h, both the CM and the

SP had completely disappeared. The results were qualita-

tively similar at the other three frequencies tested.

3.2. Organ of Corti structure and hair cell loss

Consistent with our previous morphological findings

[17] and the physiological data described above, all cochlear

hair cells were rapidly destroyed after the high dose of GM

and EA. Examination of cochleas at 6, 12 and 24 h after

GM/EA treatment indicated that onset of hair cell loss

occurred between 6 and 12 h. Most hair cells were present

at 6 h post-injection, but at 12 h post-injection (Fig. 3),

approximately 50% of IHCs and 80% of OHCs (averaged

across Rows 1, 2 and 3) were missing. As shown in Fig. 3,

OHC loss in Row 3 lagged behind losses in Rows 1 and 2.

There was only slightly less OHC loss in Row 2 than in

Row 1 at 12 h; in both rows, OHC loss was nearly complete

over 80% of the cochlea, beginning around 20% distance

from the apex. In contrast, Row 3 OHC loss was complete

only in the region located 70–100% distance from the apex.

IHC loss was greatest in the middle turn, peaking in the 2

kHz region of the cochlea [9], located 55% distance from

S.L. McFadden et al. / Brain Research 997 (2004) 40–5146

the apex (Fig. 3, solid line). At 24 h (not shown), cochleas

were devoid of all hair cells.

Despite the loss of all hair cells within 24 h, supporting

cells (Deiters’ cells and pillar cells (PC)) were present for

several months after treatment (Fig. 4). By 4 months after

treatment (Fig. 4F), the organ of Corti had collapsed onto

the basilar membrane. The collapse of the supporting cells

followed a base to apex gradient in all cochleas (data not

shown).

3.3. Efferent fiber degeneration

Fig. 5 shows the pattern of efferent innervation as it

appears in AChE-stained surface preparations of the organ

of Corti. In the normal organ of Corti (Fig. 5A), fibers

passing through the upper portion of the tunnel of Corti and

fibers within the spiral bundles are clearly identified by their

Fig. 7. Appearance of SGNs in Rosenthal’s canal (all sections from the middle tu

surrounded by a myelin sheath (arrowhead) and have a prominent nucleolus (sma

irregularly shaped and surrounded by fluid filled vacuoles. (C) 1 month, (D) 2 m

dark brown label. The TCF are axons from MOC neurons

that terminate at the base of the OHCs, while most efferent

fibers within the ISB are axons from LOC efferent neurons

that terminate on afferent dendrites beneath the IHCs. Panels

B–F in Fig. 5 show the progressive loss of TCF and ISB

fibers between 3 days and 4 weeks after GM/EA treatment.

MOC efferent fiber loss is quantified in Fig. 6. Fig. 6A

shows the percentage of TCF missing as a function of

cochlear location and time after GM/EA injection. There

was a clear base-to-apex progression of loss. By 3 weeks,

efferent fibers were completely gone in the basal half of the

cochlea. Fiber loss in the apical half of the cochlea was 67%

at 3 weeks, and increased to 88% at 4 weeks. The base-to-

apex gradient in efferent fiber degeneration is illustrated

more clearly in Fig. 6B, which shows the density of MOC

efferent fibers in the apical and basal halves of the cochlea

as a function of Group (corresponding to time relative to

rn). (A) Normal SGNs before GM/EA treatment; note that type I SGNs are

ller arrow). (B) 15 days after GM/EA; although SGNs are present, most are

onths, (E) 4 months, and (F) 6 months. Scale bar=10 Am.

Fig. 8. Density of SGNs as a function of group (reflecting time re: GM/EA

treatment) and cochlear region. All treatment groups had lower SGN

density compared to the normal group (Tukey tests, P values < 0.05). Bars

indicate one SD. Asterisks indicate significant differences between apical

turn and basal turn SGN densities ( P values < 0.05). SGN densities are

average values based on five samples from each cochlear region (apex and

base) of an animal, and averaged across two– five animals per group

(Normal, n= 5; 15 days, n= 2; 1 month, n= 4; 2 months, n= 2; 4 months,

n= 2; 6 months, n= 3).

S.L. McFadden et al. / Brain Research 997 (2004) 40–51 47

GM/EA injection). A two-way ANOVA comparing MOC

efferent fiber density as a function of Group and Cochlear

Region (apical half vs. basal half) yielded a significant

Group�Region interaction, F(5,48) = 17.795, P < 0.001.

Fig. 5B makes the source of the interaction clear. In the

normal and 3 days groups, MOC fiber density was higher in

the basal half of the cochlea than in the apical half.

Beginning at 1 week, the density gradient reversed and

MOC efferent fiber density was higher in the apical half of

the cochlea than in the basal half. Tukey tests (P < 0.05)

indicated that MOC fiber density was significantly lower in

the apical half than in the basal half for the normal and 3

days groups, and significantly higher in the apical half than

in the basal half for the 1, 2 and 3 weeks groups. Tukey tests

also confirmed group differences (collapsed across region):

MOC fiber density was significantly higher for the normal

and 3 days groups than all other groups, for the 1 week

group compared to the 2 weeks group, and for the 2 weeks

group compared to the 4 weeks group.

LOC efferent fiber loss was estimated by measuring the

width of the ISB in the apical and basal halves of the

cochlea. The width of the ISB ranged from a minimum of

0.75 Am to a maximum of 2 Am in normal cochleas. After

GM/EA injection, the range from minimum to maximum

was 0.5 to 2 Am at 3 days, and 0 (no discernable ISB) to

1.75 Am at 1, 2 and 3 weeks. The ISB was missing from all

samples examined at 4 weeks post-injection. In normal

cochleas, the mean maximum width was 1.72F 0.38 Am.

The maximum width declined to 1.52F 0.34 Am at 3 days,

1.05F 0.48 Am at 1 weeks, 0.96F 0.26 Am at 2 weeks, and

0.84F 0.62 at 3 weeks, suggesting a progressive loss of

LOC efferents that roughly paralleled the loss of MOC

efferents. A basal gradient of loss was evident in 1 and 2

weeks samples, suggesting greater vulnerability of basal

LOC efferents versus apical LOC efferents. At 1 week post-

injection, the width of the ISB was 46% of the normal value

in the base, versus 76% in the apex. At 2 weeks, the width

of the ISB was 37% of the normal value in the base, versus

68% in the apex. In our 3 weeks samples, the width of the

ISB was similar in the base and apex (38–40% of normal).

Overall, the data suggest a progressive loss of LOC efferents

that is complete within 4 weeks.

3.4. Spiral ganglion cell and afferent nerve fiber

degeneration

The appearance of SGNs in a normal animal and at

various times after EA/GM treatment is shown in Fig. 7.

The mean number of SGNs (F 1 S.D.) in the normal

cochlea was estimated to be 17,006F 486, based on counts

from five untreated chinchillas. At 15 days after GM/EA

injection, SGN number was not significantly reduced (based

on counts of cells with visible nuclei), but cells were clearly

morphologically abnormal. As shown in Fig. 7B, cells and

their nuclei were irregularly shaped and in some cases,

condensed and darkly stained, and the cytoplasmic area of

the cell was markedly reduced. Large vacuoles, that

appeared to be fluid accumulated between the cell mem-

brane and its myelin sheath, surrounded most cells and in

some cases, distorted the shape of the cell. At 1 month, only

4561F1080 SGNs remained. Thus, more than 70% of

SGNs degenerated between 15 days and 1 month after

injection. At 2 months, the mean count was 832F 449

(approximately 5% of the normal number), and at 4 and 6

months, counts were 115F 55 and 75F 59, respectively

(both < 1% of normal).

SGN densities as a function of Group and Cochlear

Region (apex vs. base) are shown in Fig. 8. A two-way

ANOVA showed a significant interaction between Group

and Cochlear Region (F(5,48) = 3.446; P= 0.01). As with

MOC efferent density, the interaction came about because

SGN density was higher in the basal half of the cochlea than

in the apical half for the normal and 15 days groups, but not

for the 1, 2, 4, or 6 months groups (Fig. 8). Tukey tests were

used for pairwise comparisons between groups (P < 0.05).

This analysis indicated that SGN density was significantly

lower than normal 15 days and thereafter. SGN density

remained stable between 15 days and 1 month, then de-

clined significantly between 1 month and 2 months, and

again between 2 months and 4 months, when very few cells

remained in any cochlea.

The number of nerve fibers per habenula perforata varied

by cochlear location in normal chinchillas [4]. Values

(meanF S.D.) were 76F 13 fibers/habenula in the apical

turn, 130F 28 in the middle turn, 182F 30 in the upper

basal turn, and 94F 12 in the hook region. As shown in Fig.

Fig. 9. Appearance of nerve fibers within the habenula perforata of the OSL in a normal chinchilla (A) and in chinchillas at various times after GM/EA

treatment: (B) 15 days; (C) 1 month; (D) 2 months. All sections are from the middle turn. Scale bar=10 Am.

S.L. McFadden et al. / Brain Research 997 (2004) 40–5148

9, fiber density was dramatically decreased within 2 weeks,

and showed little change thereafter.

4. Discussion

Co-administration of 125 mg/kg GM and 40 mg/kg EA

abolished cochlear function and destroyed all hair cells

within 24 h after injection. Efferent fibers and SGNs

degenerated rapidly after the loss of cochlear hair cells,

with an earlier onset of loss of efferents in the basal half of

the cochlea compared to SGNs. Efferent fiber loss began

between 3 and 7 days post-injection, versus 15–30 days for

SGNs. At 1 week post-injection, approximately 50% of

MOC efferent fibers had degenerated, with a strong base to

apex gradient (73% loss in the basal half of the cochlea vs.

27% loss in the apical half); MOC efferent fiber loss

increased to 68% at 2 weeks, 84% at 3 weeks, and 94%

at 4 weeks. LOC efferent loss roughly paralleled MOC

efferent loss. At 1 week post-injection, the maximum width

of the ISB had decreased by approximately 40%, with a

strong base to apex gradient (54% decrease in the base vs.

24% decrease in the apex), and at 4 weeks the ISB was no

longer discernable in our samples. At 15 days, although

SGN density was reduced by approximately 36% and most

cells appeared morphologically abnormal, cell number was

not significantly reduced. Given that no significant SGN

loss was observed at 15 days, it is likely that the significant

decrease in SGN density was due to the presence of large

fluid filled vacuoles around the cells (Fig. 7). By 1 month,

70% of the SGNs had degenerated; the loss increased to

95% by 2 month, and to nearly 100% by 4–6 months. The

degeneration of hair cells and efferent fibers occurred

independently of supporting cell loss, because supporting

cells were present for 1 month or more after injection (see

Fig. 4). Thus, the time course of histopathology in the

chinchilla cochlea involves initial rapid loss of hair cells,

followed soon after by loss of efferent fibers and SGNs, then

supporting cells of the organ of Corti. Interestingly, a

significant number of efferent fibers were missing when

SGN loss began, and efferent fiber loss was essentially

complete when 30% of SGNs still remained at 1 month.

S.L. McFadden et al. / Brain Research 997 (2004) 40–51 49

Within 4–6 months after a single injection of GM and EA,

the chinchilla cochlea was devoid of all hair cells and

supporting cells, and nearly all nerve fibers and SGNs.

4.1. Hair cell dysfunction and loss

The onset of hair cell loss occurred between 6 h, when

most cells were present, and 12 h, when approximately 50%

of IHCs and 80% of OHCs were missing. Although hair

cells were physically intact at 6 h, it is likely that they were

functionally impaired within hours after GM/EA injection.

The physiological data suggest an initial depression of CAP,

SP and CM due to loss of the positive endocochlear

potential (EP), followed by cochlear dysfunction attributable

to permanent changes in the hair cells themselves. We

previously showed that a single dose of EA (40 mg/kg

IV) results in reversal of the EP, from approximately + 80

to � 10 mV, within 10 min after injection, followed by slow

recovery beginning 25 min post-injection [3]. CAP and SP

amplitudes declined rapidly as well, reaching their lowest

levels (close to zero) at 30 min, then beginning a slow,

progressive recovery 60 min after EA injection. The CM

showed the same pattern of decline and recovery, but with a

delay of approximately 30 min relative to the CAP and SP.

In the current study, CAP, CM and SP were depressed

across all frequencies and stimulus levels at the first

measurement time of 30 min, consistent with a loss of EP.

The CAP and SP showed some recovery at the 1 and 1.5

h measurement times, while the CM showed recovery at 1.5

h (Fig. 2). These physiological changes parallel those seen

with EA alone [3], and are likely to represent reversible

effects of EA on the cochlear EP. The subsequent depression

of CAP, SP and CM at 2 h may reflect a delayed action of

GM on cochlear hair cell function. Alternatively, the tran-

sient recovery and subsequent loss of the cochlear potentials

might reflect delayed permanent damage to the stria vascu-

laris. Regardless of the mechanisms, the results suggest a 2

h delay in the effects of GM on cochlear function versus an

almost immediate effect of EA. Preliminary data from our

lab [5] suggest that mitochondrial dysfunction contributes to

transduction failure and activation of apoptotic cell death

pathways within hours after GM/EA injection, prior to

visible morphological damage to the hair cells.

At 12 h post-injection, OHC loss in Row 3 showed a

typical base-to-apex gradient, whereas most OHCs in Rows

1 and 2 had already died (see Fig. 3). The reason for the

slower rate of death for OHCs in Row 3 versus Rows 1 and

2 is unclear at this time, but it is interesting that differences

in OHC susceptibility to AABs by rows has previously been

attributed to differential distribution of MOC efferents [13].

In contrast to Row 3 OHCs, IHCs in the middle turn of the

cochlea succumbed to GM/EA sooner than IHCs in the base

or apex. This peculiar pattern of IHC loss was observed in

an earlier study [17], when chinchillas received a lower dose

of GM (10 and 20 mg/kg IP) combined with EA (40 mg/kg

IV). Together, the data indicate that IHCs in the middle turn

of the chinchilla cochlea are more vulnerable to damage

from GM and EA than IHCs in the apex or the base. The

reasons underlying the higher vulnerability of middle turn

IHCs, and whether this enhanced vulnerability is specific to

chinchillas and to GM, remain to be determined. Interest-

ingly, complete section of the auditory nerve in chinchillas

results in the loss of approximately 50% of IHCs in the basal

turn of the cochlea, accompanied by loss of type I, but not

type II ganglion cells in the base [29]. This indicates that the

vulnerability of IHCs in the chinchilla cochlea varies

according to the damaging agent, and is not a characteristic

that is specific or unique to middle turn IHCs.

4.2. Implications of the time course of SGN and efferent

fiber degeneration

It is not possible to determine from this study whether the

loss of efferent fibers and SGNs was a primary effect of

GM/EA treatment, or a secondary effect of cell loss in the

organ of Corti. Previously, Webster and Webster [24] argued

that SGN degeneration in AAB/EA-treated animals was at

least partially due to direct neurotoxic effects of the drugs,

based on the observation that SGN degeneration occurred

much earlier in cochleas damaged by AAB/EA treatment

than in cochleas injured by noise. However, at least some

SGN degeneration may be secondary to IHC loss. Imamura

and Adams [10,11] used immunocytochemistry to investi-

gate the uptake and retention of GM in the guinea pig inner

ear. After a single injection of GM, SGNs showed only

modest uptake and short retention of GM, and there was a

long delay in cytochemical changes and cell death. Based on

these observations as well as the fact that SGN loss only

occurred in regions of the cochlea where there was complete

destruction of the organ of Corti, Imamura and Adams

suggested that SGN death was secondary to loss of trophic

and sensory input from cells of the organ of Corti. Studies

showing that SGN survival can be enhanced by neuro-

trophic factors [6,22,26,27] can be taken as support for

the hypothesis that IHC loss deprives SGNs of trophic

factors that normally promote their survival. The current

data do not rule out a trophic role of IHCs on SGN survival;

however, it is interesting that some SGNs die within weeks,

whereas others are able to survive for extended periods of

time following IHC loss. The characteristics of SGNs that

make them susceptible or resistant to degeneration, either

from neurotoxicity or from trophic deprivation, should be

investigated in future studies.

In view of the earlier loss of efferent fibers versus SGNs

and the trophic role ascribed to efferent fibers in the

cochlea [18], it is interesting to speculate that SGN survival

is influenced by the presence and condition of the efferent

nerve fibers. A large number of neuroactive substances

have been identified in efferent fibers, including enkepha-

lins and dopamine in the vesiculated endings of LOC

efferent fibers in the inner spiral sulcus below IHCs

[7,8]. Pujol [18] and others have suggested that LOC

S.L. McFadden et al. / Brain Research 997 (2004) 40–5150

efferent fibers play a role in protecting the auditory

dendrites and their synapses with IHCs and assisting in

their repair after excitotoxic injury. It is possible that lateral

efferent fibers also influence the survival and recovery of

SGNs and their peripheral processes following aminoglyco-

side toxicity. The current data provide no direct support for

this hypothesis; however, the time course of degeneration,

with the demise of efferent fibers occurring earlier than that

of SGNs, does not rule out the possibility of a trophic

interaction between lateral efferent fibers and the afferent

fibers they normally contact.

4.3. Comparisons with previous studies

Li et al. [16] studied the effects of electrical stimulation

on survival of SGNs in guinea pigs deafened with kanamy-

cin (400 mg/kg SC) and EA (40 mg/kg IV). In normal

guinea pigs, SGN numbers declined from base to apex. This

is consistent with the current data showing higher SGN

density in the basal turn versus the apical turn in normal

chinchillas. In cochleas of kanamycin/EA treated guinea

pigs, approximately 40–50% of SGNs remained 8 weeks

after treatment, versus only 5% in the present study. One

factor that may account for the apparent discrepancy in the

magnitude of SGN loss in the two studies is the number of

hair cells surviving the deafening procedure. Whereas all

hair cells were destroyed in the present study, approximately

23% of IHCs and OHCs survived the kanamycin/EA

treatment in the Li et al. study. If IHC survival promotes

SGN survival, then residual hair cell survival rather than

species differences may account for the higher percentage of

SGNs surviving at 2 months in drug-deafened guinea pigs

versus chinchillas.

Guinea pigs that were deafened by a single injection of

neomycin into the middle ear cavity showed no significant

loss of SGNs until 6 weeks, when a pronounced toxicity in

the apical turn was observed; at 10 weeks, significant SGN

loss was observed in all turns of the cochlea, but variability

among animals was large [28]. SGN loss was greatest in the

apex (82% at 10 weeks) and least in the lower middle turn

(47%), with intermediate loss in the basal turn (70%).

Again, however, hair cell loss was incomplete and variable.

The authors suggested that a longer survival time might

have led to more severe hair cell pathology and possibly

more uniform SGN loss as well.

The studies cited above highlight an important issue

regarding animal models of SGN loss. Deafening proce-

dures used in previous studies with guinea pigs [14,16,23]

and cats [20,25,28] have produced variable amounts of

IHC and OHC loss across animals and treatment groups,

thereby confounding the interpretation of SGN degenera-

tion and treatments aimed at promoting SGN survival. In

one study, GM injected into the middle ear cavity of

anesthetized guinea pigs produced near-total OHC loss,

but variable IHC loss across animals and cochlear location

[23]. Similarly, injection of a neomycin solution into the

middle ear cavity destroyed most OHCs, but spared large,

and variable, numbers of IHCs [28]. A cochleogram from

one of the most severely affected guinea pigs showed

nearly complete sparing of IHCs in the region from

approximately 45–80% distance from the apex but nearly

complete loss in all other cochlear regions. Even when a

large systemic dose of kanamycin is combined with EA,

hair cell losses appear to be variable, producing complete

or nearly complete loss of cochlear hair cells in some cases,

but sparing a significant number of hair cells in others

[14,16,20]. Clearly, if SGN survival is linked to hair cell

survival, as proposed by Zappia and Altschuler [28] and

others, variability in hair cell loss is an undesirable con-

founding factor in experimental models. The deafening

protocol used in the current study wiped out all hair cells

in the chinchilla cochlea without exception, allowing us to

document the time course of SGN degeneration without the

confounding influence of remaining hair cells. The reliable

absence of cochlear hair cells in the chinchilla model may

help reduce variability in future studies using various

protection and rescue strategies.

4.4. Implications for future studies

The interdependence between IHCs and SGNs presents a

challenge for the management of profound hearing loss.

Currently, cochlear implants can be used to bypass a

dysfunctional cochlea, providing that SGNs and their axons

projecting to the cochlear nucleus remain intact. In the

future, hair cell regeneration will be useful only if SGNs

are intact and their neuronal processes are able to establish

synaptic contact with the repopulated sensory epithelium.

Thus, there is a strong clinical need for establishing ways of

promoting SGN survival following hair cell death. The rapid

time course of cochlear degeneration in the GM/EA treated

chinchilla may make it a particularly useful model for future

studies of cell death mechanisms and strategies for promot-

ing cell survival.

Several additional issues would be interesting to pursue

in future studies. First, what makes some cells susceptible to

rapid loss, while others resist loss for months after an

ototoxic insult? Second, is the base-to-apex gradient in

efferent fiber loss related to differential production and/or

distribution of cochlear antioxidants, as has been suggested

for cochlear hair cells [19]? Third, is there a direct link

between the loss of efferent fibers and SGN loss? Fourth, is

the loss of efferent fibers is the cochlea associated with

degeneration of medial and/or LOC neurons in the brain-

stem? The present study lays the groundwork for answering

these and other questions.

Acknowledgements

Supported by grants from NIH/NIDCD (P01 DC03600)

to R.J.S. and S.L.M.

S.L. McFadden et al. / Brain Research 997 (2004) 40–51 51

References

[1] W.J. Azeredo, M.L. Kliment, B.J. Morley, E. Relkin, N.B. Slepecky,

A. Sterns, W.B. Warr, J.M. Weekly, C.I. Woods, Olivocochlear neu-

rons in the chinchilla: a retrograde fluorescent labelling study, Hear.

Res. 134 (1999) 57–70.

[2] D. Ding, X. Zheng, J. Wang, Quantitation of nerve fibers in habenula

perforata in chinchilla, Zhonghua Er Bi Yan Hou Ke Za Zhi 33 (1998)

30–31.

[3] D. Ding, S.L. McFadden, R.J. Salvi, Cochlear hair cell densities and

inner-ear staining techniques, in: J.F.Willott (Ed.), Handbook ofMouse

Auditory Research, CRC Press, Boca Raton, 2001, pp. 189–204.

[4] D. Ding, S.L. McFadden, J.M. Woo, R.J. Salvi, Ethacrynic acid rap-

idly and selectively abolishes blood flow in vessels supplying the

lateral wall of the cochlea, Hear. Res. 173 (2002) 1–9.

[5] D.L. Ding, H. Jiang, S.L. McFadden, R.J. Salvi, Hair cell loss from

the combination of gentamicin and ethacrynic acid is initiated by

caspase 9 and cytochrome c release, Assoc. Res. Otolaryngol. Abstr.

(2004) (submitted).

[6] P. Ernfors, M.L. Duan, W.M. ElShamy, B. Canlon, Protection of

auditory neurons from aminoglycoside toxicity by neurotrophin-3,

Nat. Med. 2 (1996) 463–467.

[7] M. Eybalin, Neurotransmitters and neuromodulators of the mamma-

lian cochlea, Physiol. Rev. 73 (1993) 309–373.

[8] M. Eybalin, G. Charachon, N. Renard, Dopaminergic lateral efferent

innervation of the guinea-pig cochlea: immunoelectron microscopy of

catecholamine-synthesizing enzymes and effect of 6-hydroxydop-

amine, Neuroscience 54 (1993) 133–142.

[9] D.D. Greenwood, A cochlear frequency-position function for several

species—29 years later, J. Acoust. Soc. Am. 87 (1990) 2592–2605.

[10] S.I. Imamura, J.C. Adams, Changes in cytochemistry of sensory and

nonsensory cells in gentamicin-treated cochleas, J. Assoc. Res. Oto-

laryngol. 4 (2003) 196–218.

[11] S.I. Imamura, J.C. Adams, Distribution of gentamicin in the guinea

pig inner ear after local or systemic application, J. Assoc. Res. Oto-

laryngol. 4 (2003) 176–195.

[12] S. Iurato, C.A. Smith, D.H. Eldredge, D. Henderson, C. Carr, Y.

Ueno, S. Cameron, R. Richter, Distribution of the crossed olivo-

cochlear bundle in the chinchilla’s cochlea, J. Comp. Neurol. 182

(1978) 57–76.

[13] A. Kohonen, Effect of some ototoxic drugs upon the pattern and

innervation of cochlear sensory cells in the guinea pig, Acta Otolar-

yngol. (Suppl. 208) (1965) 1–70.

[14] M. Komeda, B.J. Roessler, Y. Raphael, The influence of interleukin-1

receptor antagonist transgene on spiral ganglion neurons, Hear. Res.

131 (1999) 1–10.

[15] P.A. Leake, G.T. Hradek, Cochlear pathology of long term neomycin

induced deafness in cats, Hear. Res. 33 (1988) 11–33.

[16] L. Li, C.W. Parkins, D.B. Webster, Does electrical stimulation of deaf

cochleae prevent spiral ganglion degeneration? Hear. Res. 133

(1999) 27–39.

[17] S.L. McFadden, D. Ding, H. Jiang, J.M. Woo, R.J. Salvi, Chinchilla

models of selective cochlear hair cell loss, Hear. Res. 174 (2002)

230–238.

[18] R. Pujol, Lateral and medial efferents: a double neurochemical mech-

anism to protect and regulate inner and outer hair cell function in the

cochlea, Br. J. Audiol. 28 (1994) 185–191.

[19] S.H. Sha, R. Taylor, A. Forge, J. Schacht, Differential vulnerability of

basal and apical hair cells is based on intrinsic susceptibility to free

radicals, Hear. Res. 155 (2001) 1–8.

[20] R.K. Shepherd, E. Javel, Electrical stimulation of the auditory nerve.

I. Correlation of physiological responses with cochlear status, Hear.

Res. 108 (1997) 112–144.

[21] R.K. Shepherd, J. Matsushima, R.L. Martin, G.M. Clark, Cochlear

pathology following chronic electrical stimulation of the auditory

nerve: II. Deafened kittens, Hear. Res. 81 (1994) 150–166.

[22] H. Staecker, R. Gabaizadeh, H. Federoff, T.R. Van De Water, Brain-

derived neurotrophic factor gene therapy prevents spiral ganglion

degeneration after hair cell loss, Otolaryngol. Head Neck Surg. 119

(1998) 7–13.

[23] M. Suzuki, M. Yagi, J.N. Brown, A.L. Miller, J.M. Miller, Y. Raphael,

Effect of transgenic GDNF expression on gentamicin-induced coch-

lear and vestibular toxicity, Gene Ther. 7 (2000) 1046–1054.

[24] M. Webster, D.B. Webster, Spiral ganglion neuron loss following

organ of corti loss: a quantitative study, Brain Res. 212 (1981)

17–30.

[25] S.A. Xu, R.K. Shepherd, Y. Chen, G.M. Clark, Profound hearing loss

in the cat following the single co-administration of kanamycin and

ethacrynic acid, Hear. Res. 70 (1993) 205–215.

[26] M. Yagi, S. Kanzaki, K. Kawamoto, B. Shin, P.P. Shah, E. Magal, J.

Sheng, Y. Raphael, Spiral ganglion neurons are protected from degen-

eration by GDNF gene therapy, J. Assoc. Res. Otolaryngol. 1 (2000)

315–325.

[27] J. Ylikoski, U. Pirvola, J. Virkkala, P. Suvanto, X.Q. Liang, E. Magal,

R. Altschuler, J.M. Miller, M. Saarma, Guinea pig auditory neurons

are protected by glial cell line-derived growth factor from degener-

ation after noise trauma, Hear. Res. 124 (1998) 17–26.

[28] J.J. Zappia, R.A. Altschuler, Evaluation of the effect of ototopical

neomycin on spiral ganglion cell density in the guinea pig, Hear.

Res. 40 (1989) 29–37.

[29] X.Y. Zheng, D.L. Ding, S.L. McFadden, D. Henderson, Evidence that

inner hair cells are the major source of cochlear summating potentials,

Hear. Res. 113 (1997) 76–88.

[30] X.Y. Zheng, D. Henderson, B.H. Hu, D.L. Ding, S.L. McFadden, The

influence of the cochlear efferent system on chronic acoustic trauma,

Hear. Res. 107 (1997) 147–159.