JARO 04: 91–105 (2002)

DOI: 10.1007/s10162-002-3016-8

JARO
Journal of the Association for Research in Otolaryngology

Hair Cell Death in the Avian Basilar Papilla:

Characterization of the in vitro Model and
Caspase Activation
ALAN G. CHENG, LISA L. CUNNINGHAM, AND EDWIN W RUBEL
Virginia Merrill Bloedel Hearing Research Center, Department of Otolaryngology-Head and Neck Surgery,
University of Washington, Seattle, WA 98195, USA
Received: 11 April 2002; Accepted: 13 July 2002; Online publication: 7 November 2002

ABSTRACT controls. Our data indicate that auditory hair cells

degenerate as a result of gentamicin exposure in a
Caspases are a family of proteases that have been caspase-dependent manner. Specifically, the up-
implicated as key mediators of cell death. Although stream caspases, caspase-8 and caspase-9, and the
nonspecific inhibition of caspase activation has been downstream caspase-3 are activated in aminoglyco-
reported to prevent mammalian sensory hair cell side-damaged hair cells.
death, the exact roles of individual caspases during Keywords: hair cell death, gentamicin, aminogly-
hair cell death are unclear. In other systems, the ac- coside, caspase, basilar papilla
tivation of initiator caspases, such as caspase-8 and
caspase-9, can lead to the activation of the effector
caspase-3. We have begun to systematically charac-
terize hair cell death in an in vitro system by exam-
ining the activation of these specific caspases in INTRODUCTION
degenerating hair cells after acutely damaging the
whole avian basilar papilla with gentamicin. Basilar Sensory hair cells are susceptible to damage under
papillae (BP) displayed a dose-dependent hair cell numerous conditions such as excessive noise, aging,
loss after a 24-h treatment with gentamicin at con- and exposure to certain classes of therapeutic drugs
centrations of 0.1, 0.5, and 2.0 mM. When treated including the aminoglycosides. The mechanisms of
with 0.5 mM gentamicin for 6, 12, or 24 h, hair cells hair cell death as a result of these diverse stresses ap-
first began to degenerate in the basal third of the BP pear to be similar and share many of the characteris-
and damage progressed apically. Supplementation of tics of programmed cell death (apoptosis; Forge 1985;
z-VAD-fmk, a general caspase inhibitor, provided Li et al. 1995; Lang and Liu 1997; Usami et al. 1997;
short-term protection against gentamicin-induced Zheng et al. 1998; Lenoir et al. 1999; Forge and Li
hair cell death. Treatment with gentamicin for 6 or 12 2000; Hu et al. 2000; Pirvola et al. 2000; Matsui et al.
h promoted the expression of active caspase-3 and 2002). Hair cells dying in an apoptotic-like manner
active caspase-9 in many hair cells along the BP as participate in an active, organized cellular degrada-
shown by immunohistochemistry. At these time- tion and disposal with preserved tissue integrity and
points, specific fluorescent-labeled peptide substrates minimal inflammatory response (Li et al. 1995; Lang
detected more active caspase-3, caspase-8, and cas- and Liu 1997; Forge and Li 2000). Such preservation
pase-9 in gentamicin-treated hair cells relative to of the architecture of the sensory epithelium (SE) is
crucial in order to retain organ function, particularly
Correspondence to: Edwin W Rubel Æ Virginia Merrill Bloedel Hearing in tissues where hair cells are continuously produced
Research Center Æ Box 357923 Æ University of Washington. Tele-
phone: (206) 543-8360; fax: (206) 221-5685; email: rubel@ throughout life (Weisleder and Rubel 1993; Kil et al.
u.washington.edu 1997; Williams and Holder 2000).

91
92 CHENG ET AL.: Active Caspase in Hair Cells

Apoptotic cells are characterized by their stereo- Compared with other organ systems, the roles of
typical morphologic changes, including chromatin caspases in the overall cascade of cellular events
condensation and margination, cytoplasmic conden- leading to inner ear cell death have received relatively
sation, and formation of apoptotic bodies (Kerr et al. little attention. Liu et al. (1998) used a nonspecific
1972; Clarke and Clarke 1995). On a molecular level, caspase inhibitor to protect rat auditory hair cells
DNA fragmentation, as a result of activated endo- from cisplatin-induced death in vitro. More recently,
nucleases and degraded DNA repair enzymes, is a aminoglycoside-induced hair cell death has been
hallmark of apoptosis (Gavrieli et al. 1992). After shown to be caspase-mediated in the vestibular organs
exposure to aminoglycosides, dying hair cells show from mammals (Forge and Li 2000; Cunningham et
these morphologic changes in vitro (Li et al. 1995; al. 2002) and birds (Matsui et al. 2002). Individual
Lang and Liu 1997; Forge and Li 2000; Matsui et al. caspases have not been described in the context of
2002) and in vivo (Forge 1985; Lenoir et al. 1999). hair cell degeneration. It will prove essential for pre-
Apoptotic nuclear changes following aminoglycoside venting cell death of the inner ear sensory elements to
treatment have been observed in the vestibular and understand cell death and cell survival cascades as
auditory organs in mammals and birds (Forge 1985; completely as possible. This report characterizes in
Li et al. 1995; Lang and Liu 1997; Lenoir et al. 1999; vitro model of gentamicin-induced hair cell death in
Torchinsky et al. 1999; Forge and Li 2000; Matsui the avian hearing organ, the basilar papilla. Using this
et al. 2002). model, we have examined the effects of caspase inhi-
Cellular stresses, such as ultraviolet light, toxins, bition on hair cell survival as well as the activation of
and neurotrophin withdrawal, can be translated into specific caspases in gentamicin-treated hair cells.
cell death signals and thus can activate the apoptotic
machinery (for reviews, see Earnshaw et al. 1999; Slee
et al. 1999a). If these cell death signals overwhelm METHODS
survival factors (e.g., bcl-2), they activate a family of
Animals
cell death mediators termed caspases. Caspases com-
prise a unique family of cysteine-dependent, aspar- White Leghorn chicken (Gallus domesticus) eggs
tate-specific proteases that normally exist as inactive were obtained from a local distributor (H&N Inter-
zymogens (pro-caspases; for reviews, see Earnshaw et national, Redmond, WA). These eggs were hatched in
al. 1999 and Nicholson 1999). Several genes involved incubators and hatchlings were housed in brooders
in apoptosis, including that of caspase-3 (ced-3), were with food and water in the University of Washington
first described in the nematode C. elegans (Yuan et al. animal care facility. All experimental protocols were
1993). Caspases are widely conserved through evolu- reviewed and approved by the University of Wash-
tion and are found in multicellular organisms from ington Institutional Animal Care Committee and
worms to mammals (Earnshaw et al. 1999). Caspases conform to NIH guidelines.
recognize specific tetrapeptide sequences and cleave
protein elements including their own prodomains.
Organ culture techniques
Cleavage of the prodomain activates the caspase. Ac-
tivated caspases dismantle components of the cyto- Five- to 10-day-old hatchlings were rapidly decapitated
skeleton (Mashima et al. 1995, 1997; Janicke et al. and the basilar papillae (BP) were exposed, removed,
1998) and nuclear scaffolds (Lazebnik et al. 1995; and cultured as previously described (Oesterle et al.
Casiano et al. 1996). Fourteen caspase members have 1993). Under sterile conditions each BP was cultured
thus far been identified, with those mediating cell as a whole, free-floating organ with an intact teg-
death falling into two general categories (Earnshaw et mentum vasculosum. One BP was placed in 700 lL
al. 1999; Nicholson 1999): initiators (caspase-8 and culture medium in individual wells of a 48-well tissue
caspase-9) and effectors (caspase-3, caspase-6, and culture plate. Culture medium was composed of
caspase-7). At least 70 individual substrates of casp- 66.7% Basal Medium Eagle (Sigma, St. Louis, MO),
ases have thus far been identified (Nicholson 1999). 33.3% Earle’s balanced salt solution (EBSS) (Gibco/
Other caspase-mediated events include disabling BRL, Gaithersburg, MD), and 5% fetal bovine serum
DNA repair enzymes (Casiano et al. 1996) and acti- (Gibco/BRL). Cultured whole organs were incubated
vating DNA endonuclease (Enari et al. 1998; Tang at 37
C in a 5% CO2 atmosphere in a Forma Scientific
and Kidd 1998). In addition to cellular digestion, (Marietta, OH) incubator.
caspases target and inactivate bcl-XL, a member of the
survival-promoting bcl-2 family (Clem et al. 1998).
Experimental paradigms
Perhaps most importantly, caspases cleave and acti-
vate other caspases, creating a positive feedback loop Whole organ BPs were cultured for 18–24 h before
(Earnshaw et al. 1999; Nicholson 1999). any pharmacologic manipulations. Gentamicin (Sig-
CHENG ET AL.: Active Caspase in Hair Cells 93

ma) was stored as a 20 mM stock solution and diluted Caspase substrates

in culture medium immediately before application. A
The fluorogenic substrates (Intergen, Purchase, NY)
general caspase inhibitor, z-VAD-fmk (Enzyme Sys-
consist of short peptide consensus sequences conju-
tems, Livermore, CA), and its inactive structural an-
gated to a fluorescent probe. They specifically recog-
alog z-FA-fmk (BD Biosciences Pharmingen, San
nize and irreversibly bind the active sites of individual
Diego, CA), were both stored as 40 mM solutions in
activated caspases (Bedner et al. 2000). These sub-
DMSO (Sigma) and prepared shortly before use. A
strates were prepared as 150x stock solutions in sterile
subset of the cultures was preincubated with z-VAD-
PBS and stored at )20
C. Fam-VAD-fmk substrate
fmk or z-FA-fmk for 2 h before gentamicin was added
(10 lM) was used to detect general caspase activity,
(Forge and Li 2000) to ensure adequate cellular
fam-DEVD-fmk substrate (5 lM) was used for caspase-
penetration of the agents. The viability of hair cells in
3 activity, fam-LEHD-fmk substrate (5 lM) was used to
cultured BPs treated with z-VAD-fmk alone showed no
detect caspase-9 activity, and fam-LETD-fmk substrate
obvious difference from that of untreated, cultured
(5 lM) was used to detect caspase-8 activity. These
BPs.
substrates were diluted shortly before being directly
Tissues were exposed to 0.1, 0.5, or 2.0 mM
applied to the culture medium for 1h at the end of the
gentamicin for 24 h for dose–response measure-
culture periods. After the 1h incubation, tissues were
ments. The 0.5 mM gentamicin dose was chosen for
washed with buffer (Intergen) and fixed with buffered
the remainder of the experiments including those
10% formalin (Intergen) at 4
C overnight.
examining the time course of hair cell death (6, 12,
and 24 h). Tissues were fixed immediately at the end
of each culture period by immersion in buffered
Data analysis
4% paraformaldehyde in 0.1 M Na+/K+ phosphate
buffer (pH 7.4) for 30–60 min at room temperature Hair cell quantification in whole mount BPs. Tissue anal-
(RT) and rinsed in 0.1 M phosphate-buffered saline ysis consisted of three steps: First, tissues were exam-
(PBS). ined under a Zeiss Axiophot epifluorescent
microscope with a 10x objective. Several digital im-
ages were taken using Object-Image (NIH) to cover
Immunohistochemistry
the entire sensory epithelium (SE). A manual z-series
Oregon green- and rhodamine-conjugated phalloidin of images was collected to account for the convex
were obtained from Molecular Probes (Eugene, OR). surface of the SE. Adobe Photoshop (Adobe Systems,
Primary antibodies directed against cleaved (acti- San Jose, CA) was then used to montage all of the
vated) caspase-3 (dilution 1:200) and caspase-9 digital images captured from a single BP. In the sec-
(dilution 1:100) were obtained from R&D Systems ond step of the analysis, the SE was manually outlined
(Minneapolis, MN) and Cell Signaling (Beverly, MA), and bisected longitudinally from the apex to base.
respectively. Dr. Anthony Frankfurter (University of This step allowed the computer to create 10 boxes at
Virginia) generously donated the monoclonal Tuj1 equal intervals perpendicular to the longitudinal axis
antibody (dilution 1:1000). of the BP. The computer computes the orientation of
At the end of the culture periods, organs were the longitudinal axis at any point by averaging the
fixed and the tegmentum vasculosum and tectorial curvature of the surrounding 10% length of the co-
membrane were dissected free. Hair cells were either chlea. Each box had a width exactly equal to 2% of
stained with anti-Tuj1 or phalloidin. Oregon green- the axial length of the BP and was divided into su-
or rhodamine-phalloidin diluted 1:100 in 0.05% perior and inferior segments. Figure 1 shows a mon-
Triton X-100/PBS (Sigma) was applied to tissues at tage image of an untreated BP cultured for 2 days
RT for 2 h. When anti-Tuj1 was used as a hair cell with the 10 boxes and a midline drawn over it.
marker (Stone et al. 1996), tissues were first blocked Finally, a single experimenter counted the number
with 10% normal goat serum (Sigma) in 0.05% Triton of hair cells within the segments of each box. A hair
X-100/PBS for 30 min before being bathed in anti- cell was included in the counts if it was brightly la-
Tuj1 at 4
C overnight. The same histological protocol beled with phalloidin and at least partially located
was followed for the use of other primary antibodies within a box. Because gentamicin-damaged hair cells
at the dilutions indicated above. Further immunola- might lack stereocilia bundles, controls using a spe-
beling was done using BODIPY, Alexa 488, or Alexa cific hair cell marker, Tuj1, were performed to ensure
594 conjugated to IgG (Molecular Probes) as sec- that loss of phalloidin staining accurately reflected
ondary antibodies, diluted at 1:200 in block solution hair cell loss (see below). Boxes containing overtly
at RT for 2 h. Organs were mounted onto slides using mechanically damaged SE or poor imagery were dis-
Vectashield mounting medium (Vector Laboratory, carded. This was less than 1% of the sampled areas.
Burlingame, CA) and coverslipped. Hair cell density was calculated as hair cells per 100
94 CHENG ET AL.: Active Caspase in Hair Cells

statistical significance of the difference between

treatment groups.

RESULTS
Dose- and time-dependent hair cell death
In comparison with culture-free organs, there was a
significant decline (p < 0.001) in overall hair cell
density of untreated, cultured BPs after 2 days in vitro.
In normal (culture-free) BPs the mean (±SEM) hair
cell densities were 695 (±36), 855 (±14), and 386
FIG. 1. Method of quantification. Whole-organ basilar papillae (BP) (±34) hair cells per 10,000 lm2 for the apical, middle,
from posthatch 5- to 10-day-old chicks were incubated in culture and basal thirds, respectively. The corresponding
media supplemented with 5 fetal bovine serum and then fixed and densities after 18–24 h in culture were 449 (±76), 458
labeled with phalloidin. A montage of the sensory epithelium (SE)
(±57), and 164 (±17), respectively. This represents an
was constructed under low power. The SE was then outlined, and a
line was drawn from apex to base along the middle of the SE. The overall hair density of about 55%, similar to the data
computer then constructed 10 boxes at equal intervals perpendicular reported by Frenz et al. (1998). This drop in hair cell
to the longitudinal axis of the BP. Because the computer estimates density can be attributed to spreading of the sensory
the longitudinal axis by averaging the curvature of the surrounding epithelium in vitro and hair cell death as a result of
10 length along the cochlea, boxes in greatly curved areas (e.g.,
explantation and being placed in the culture system.
boxes 6 and 10) might not appear completely perpendicular to the
line drawn manually. The width of each box equals 2 of the axial After the initial 18–24 h of culture, hair cell densities
length of the BP. Hair cell density was calculated in each box along stabilized as BPs cultured for 2 and 3 days showed no
the length of the BP to assess hair cell survival (see Materials and statistical difference in their hair cell densities (data
Methods). Scale bar = 100 lm. not shown).
Hair cell survival decreased as the concentration of
gentamicin increased in the culture media. Figure 2
lm · 100 lm (10,000 lm2). Analysis of variance shows the relationship between gentamicin concen-
(ANOVA) was followed by appropriate individual tration and hair cell survival. At each of the three
comparisons used for statistical calculations; p < 0.05 doses of gentamicin tested, hair cell density was the
was considered statistically significant. lowest in the basal region and increased in a gradient
To ensure the reproducibility of this method, 10 fashion toward the apex. In addition, a decline in the
random BPs (3 controls, 3 exposed to gentamicin for overall hair cell density was noted throughout the BP
12 h, 2 exposed to gentamicin for 24 h, and 2 ex- as the concentration of gentamicin increased from
posed to gentamicin and z-VAD-fmk for 24 h) were 0.1 (n = 4) to 0.5 (n = 14) to 2.0 mM (n = 4). A sig-
chosen by a second investigator. The original exper- nificant interaction between the concentration of
imenter, now blinded to the treatment groups, gentamicin and hair cell densities along the BP was
repeated the hair cell counts. The correlation coeffi- observed (p < 0.001; two-way ANOVA). Tissues treated
cient (r) relating the two hair cell counts was 0.94. with any dose of gentamicin had significantly lower
A separate control was done to ensure that loss of (p < 0.001) hair cell densities than the controls
phalloidin staining was indicative of hair cell loss. (n = 10).
Four BPs (2 controls and 2 exposed to 0.5 mM gen- The influence of aminoglycoside exposure time at
tamicin for 6 h) were double-labeled with anti-Tuj1 a single gentamicin concentration (0.5 mM) is dem-
and phalloidin. The number of cells stained with anti- onstrated in Figure 3. This figure shows the hair cell
Tuj1, but not phalloidin, was then estimated from densities, grouped according to their location along
these tissues. Of the total number of hair cells directly the BP, of gentamicin-treated BPs normalized to their
examined through double-labeling (3000), 0.26% parallel cultured controls. In comparison to parallel
were Tuj1-positive and phalloidin-negative, whereas control cultures, drug-treated tissues displayed pro-
1.66% were Tuj1-negative and phalloidin-positive. gressively lower hair cell density as the duration of
Quantification of cells labeled for active caspase-3 and gentamicin exposure increased. No significant change
active caspase-9. Whole mount BPs were assessed using in hair cell density in the basal region was observed
a confocal microscope with a 40x objective. The total between the 12- and 24-h treatment paradigms.
numbers of double-labeled (Tuj1+ and active caspase- Meanwhile the basal–apical gradient of hair cell den-
3+ or active caspase-9+) and single-labeled (active sities was retained at all time-points. There was a sig-
caspase-3+ or active caspase-9+ only) hair cells were nificant interaction between time of exposure to
counted in each BP. Student’s t-test was used to assess 0.5 mM gentamicin and hair cell densities along the
CHENG ET AL.: Active Caspase in Hair Cells 95

FIG. 2. Relationship between gentamicin dose and hair cell survival density along the BP relative to controls. In addition, at the doses
along the length of the basilar papilla (BP). Organotypic cultures used, the apical–basal gradient of hair cell density was blunted with
were incubated in antibiotic-free media for 1 day before exposure to gentamicin treatment. There was a significant interaction between
0 (control), 0.1, 0.5, or 2.0 mM gentamicin for 24 h. Treatment with the concentration of gentamicin and hair cell position along the BP
any of the tested doses significantly lowered (p < 0.001) hair cell (p < 0.001; two-way ANOVA). Error bars = SEM.

of exposure to gentamicin (n = 11) in comparison to

untreated BPs (n = 7). After 12 h of gentamicin ex-
posure, a significant drop in hair cell density was ob-
served in both the middle and basal thirds of the BP
(n = 15, p < 0.001). At the end of a 24-h treatment
period, gentamicin-exposed BPs had significantly
lower (n = 14, p < 0.001) hair cell densities (<35% of
controls) along the entire length of the BP. Figure 4
shows representative photomicrographs of the un-
treated, control BPs (Fig. 4A–C) and those treated
with 0.5 mM gentamicin for 12 hours (Fig. 4D–F).

FIG. 3. Hair cell survival decreased in a time-dependent manner

Caspase inhibition
after 0.5 mM gentamicin treatment. Hair cell densities along the z-VAD-fmk, a broad-spectrum caspase inhibitor, works
basilar papilla (BP) were divided into apical (3 apical boxes in Fig.
1), middle (3 middle boxes), and basal thirds (4 basal boxes). Hair
by irreversibly binding the active sites on caspases,
cell densities from gentamicin-treated BPs were then normalized to thus quenching the caspase activity. When z-VAD-fmk
those of parallel untreated, cultured BPs according to their location (100 lM) was applied along with gentamicin to BPs,
on the BP (apical, middle, or basal). The density of apical hair cells hair cell integrity was promoted at both 12 (n = 4)
from BPs treated with gentamicin for 6 h is not statistically different and 24 h (n = 6) of gentamicin exposure. This in-
from that of controls. Basal hair cell loss was readily apparent as
early as 6 h after gentamicin exposure. Both the middle and basal
crease in hair cell survival was significant at both 12
thirds demonstrated decreased hair cell survival after aminoglyco- and 24 h (p < 0.001). Hair cell densities of BPs treated
side treatment for 12h. At this time-point, the hair cell density in the with gentamicin and z-VAD-fmk for 12 h is not sta-
apical region remained comparable to that of controls. By 24 h of tistically different from that of untreated, cultured
gentamicin exposure, over 60 of hair cells throughout the BP had control organs (data not shown). Table 1 shows hair
degenerated. Using two-way ANOVA, we found a significant main
effect of exposure time (p < 0.001) and a significant interaction be-
cell densities of drug-treated BPs normalized to their
tween exposure time to gentamicin and hair cell position along the respective controls according to the location along
BP (p < 0.001). Error bars = SEM. the BP (apical, middle, basal). An additional negative
control experiment using 100 lM z-FA-fmk, an inac-
BP (p < 0.001; two-way ANOVA). Notable hair cell loss tive analog of z-VAD-fmk, and 0.5 mM gentamicin
was first detected in the basal third of the BP after 6 h showed no increased hair cell survival with respect to
96 CHENG ET AL.: Active Caspase in Hair Cells

FIG. 4. Representative examples of cultured whole-organ basilar treatment with 0.5 mM gentamicin for 12 h after 24 h in control
papillae (BP) stained with fluorescent-conjugated phalloidin. Thirty- media. There is noticeable hair cell loss in the middle and basal
six hours in culture. A–C. Untreated control BPs showing apical, thirds of the BP. The hair cell density in the apical region is similar to
middle, and basal segments. Note a dense population of hair cells that of control BPs. Scale bar = 20 lm.
with some loss in the basal region. D–F. Gentamicin-treated BPs:

TABLE 1

Effects of nonspecific caspase inhibition on hair cell survivala

Treatment duration (h) Apex Middle Base

Gentamicinb alone 12 0.85 ± 0.05 0.54 ± 0.07 0.17 ± 0.03

Gentamicinb + z-VAD-fmkc 12 1.28 ± 0.10 0.96 ± 0.09 1.15 ± 0.25
Gentamicinb alone 24 0.32 ± 0.08 0.37 ± 0.10 0.31 ± 0.08
Gentamicinb + z-VAD-fmkc 24 0.59 ± 0.09 0.48 ± 0.11 0.42 ± 0.07
a
Hair cell density normalized to control cultures ± SEM.
b
Gentamicin diluted to 0.5 mM in all treatments.
c
z-VAD-fmk (100 lM), a general caspase inhibitor, was applied for 2 h and then along with gentamicin for the specified duration.

gentamicin treatment alone (n = 4; data not shown). densities of BPs cultured for 2 and 3 days are not
Despite concurrent treatment with caspase inhibitors statistically different.
and gentamicin, there was some decline in hair cell
density from 12 to 24 h (p < 0.001).
Immunolabeling for active caspases
While it is possible for z-VAD-fmk to have pre-
vented hair cell death resulting from the culture en- To begin studying hair cells in the process of de-
vironment, we expect this effect to be minimal for 2 generation, we used immunochemical detection of
reasons: (1) little general caspase activation was de- activated caspases in control basilar papillae and
tected in the untreated, cultured BPs with the fam- those treated for 6 or 12 h with gentamicin (0.5 mM).
VAD-fmk fluorogenic substrates and (2) hair cell Cultures were also double-labeled with an antibody to
CHENG ET AL.: Active Caspase in Hair Cells 97

FIG. 5. Confocal images showing expression of active caspase-3 in F. Gentamicin-treated BPs; after exposure to 0.5 mM gentamicin for
hair cells from control and gentamicin-treated basilar papillae (BP). 12 h, many hair cells remaining in the sensory epithelium showed
BPs were labeled for B-tubulin with anti-Tuj1 (green), a specific immunostaining for active caspase-3 (Tuj1+/AC3+). Its expression
marker for chick inner ear hair cells and neurons, and anti-active was noted throughout the apical, middle, and basal regions with no
caspase-3 (red). A–C. Untreated cultured BPs. Few hair cells ex- apparent gradient (arrows). In the basal third, a large number of Tuj1-
pressed active caspase-3 in all three regions of untreated BPs. Some negative hair cells expressed active caspase-3 as well (Tuj1-/AC3+;
hair cells with active caspase-3 near the luminal surface of the arrowheads). Scale bar = 20 lm.
sensory epithelium have lost their Tuj1 antigenicity (arrowheads). D–

class 3 B-tubulin (Tuj1) in order to label hair cell all of the Tuj1-/AC3+ cells were distributed in the
cytoplasm (Stone et al. 1996). basal third in both control and gentamicin-damaged
Active caspase-3. Untreated control BPs contained BPs. Figure 6 summarizes the cell counts for both
few hair cells with active caspase-3 (Tuj1+/AC3+), control cultures and BPs cultured in the presence of
scattered in the apical, middle, and basal thirds of the gentamicin. Note that many more cells showed active
BP (Fig. 5A–C). In comparison to controls (6 h, n = 6; caspase-3 labeling after 12 h than after 6 h of genta-
12 h, n = 9), gentamicin-treated BPs showed a robust micin exposure (note scale difference on ordinates of
increase in Tuj1+/AC3+ cells in all three segments Fig. 6A and B). After 6 h of gentamicin treatment, the
(apical, middle, basal) at both 6 (n = 5, data not average numbers of both Tuj1+/AC3+ (p < 0.001) and
shown) and 12 h of gentamicin exposure (n = 9, Fig. Tuj1-/AC3+ (p < 0.05) cells were increased sig-
5D–F). nificantly compared with controls. By 12 h, the counts
Occasional cells in the control tissues were labeled of Tuj1+/AC3+ and Tuj1-/AC3+ cells remained low
for active caspase-3 but not Tuj1 (Tuj1-/AC3+) (Fig. in the untreated organs. Again, gentamicin sig-
5B-C). These cells (Tuj1-/AC3+) were interpreted as nificantly increased both the Tuj1+/AC3+ (p < 0.001)
hair cells because they had sizes and shapes similar to and Tuj1-/AC3+ (p < 0.01) cells at this time-point.
adjacent Tuj1-positive hair cells and were located Active caspase-8 and caspase-9. Immunolabeling for
within 10 lm of the luminal surface of the SE. Sup- active caspase-9 was performed on BPs treated with
porting cells were easily distinguishable from hair gentamicin for 6 or 12 h and their respective controls.
cells by their deeper location within the SE. Unlike Representative micrographs from control (Fig. 7A–C)
the Tuj1+/AC3+ cells at these two time-points, almost and 12-h gentamicin-treated BPs (Fig. 7D–F) are
98 CHENG ET AL.: Active Caspase in Hair Cells

comparing the counts of AC9+ to AC3+ hair cells, we

have observed significantly more Tuj1-/AC3+ cells
than Tuj1-/AC9+ cells after gentamicin treatment for
6 (p < 0.01) or 12 h (p < 0.01).
Three antibodies against cleaved caspase-8 (BD
Pharmingen; Cell Signaling; SmithKline Beecham,
King of Prussia, PA) were tested in control or genta-
micin-treated tissues. At several concentrations these
antibodies produced either high background or no
apparent cellular staining. Despite reliable labeling in
positive controls using mouse tissues, they were un-
able to stain positive controls using chick cochleae
treated with staurosporine, a known cell-death-in-
ducing agent. These antibodies are likely inappro-
priate for immunohistochemistry in chicken tissues.

Detection of caspase activity with fluorogenic

substrates
In addition to immunohistochemical identification,
caspase activity was assessed by observing substrate
activation in live cultures. Hence we examined non-
specific caspase binding as well as binding to sub-
strates thought to be relatively specific for the
activated caspase-3, activated caspase-8, and activated
caspase-9 (Thornberry et al. 1997; Garcia-Calvo et al.
1998).
The fam-VAD-fmk fluorogenic substrate was used
FIG. 6. Cell counts of active caspase-3 expression in control and to detect general caspase activity (data not shown).
gentamicin-treated basilar papillae (BP). Cells double-labeled for
After 12 h there was noticeably more general caspase
Tuj1 and active caspase-3 (Tuj1+/AC3+) and Tuj1-negative cells
expressing active caspase-3 (Tuj1-/AC3+) near the luminal surface of activity in the gentamicin-treated (0.5 mM, n = 4)
the sensory epithelium were counted in each gentamicin-treated and than control tissues (n = 4). This difference was seen
parallel control BP. A. A 6-h treatment with gentamicin significantly in the apical, middle, and basal thirds of all samples
promoted active caspase-3 expression in both Tuj1-positive (p < tested. Thus, we proceeded to examine the activation
0.001) and Tuj1-negative (p < 0.05) hair cells in comparison to
of specific caspases.
controls. B. By 12 h of gentamicin exposure, these differences have
further widened. Significantly more Tuj1-positive (p < 0.001) and Caspase-3 activity was assessed in damaged and
Tuj1-negative (p < 0.01) cells expressed active caspase-3 compared control BPs using the fluorescent substrate fam-
with controls, (n’s: control 6 h, n = 6; control 12 h, n = 9; gent 6 h, DEVD-fmk. Figure 9 shows representative images se-
n = 5; gent 12 h, n = 9). Error bars = SEM. lected from control (Fig. 9A–C, n = 8) as well as
gentamicin-treated BPs (Fig. 9D–F, 0.5 mM gentami-
cin, n = 8). Very few hair cells (phalloidin-labeled)
shown. Fewer hair cells were labeled for active cas- with caspase-3 activity were present in the control BPs.
pase-9 (Tuj1+/AC9+) in control tissues in compari- This contrasts greatly with BPs damaged with genta-
son to those treated with gentamicin for 6 (n = 5) or micin for 12 h. Apical, middle, and basal thirds from
12 h (n = 7). Tuj1-/AC9+ cells were rarely observed in gentamicin-treated BPs typically contained substantial
control or gentamicin-damaged BPs at either time- caspase-3 activity with this assay as well. The middle
point. Cell counts are shown in Figure 8. In both the and basal thirds showed greater labeling than the
6-h (n = 6) and 12-h (n = 5) control groups, the total apex.
number of hair cells expressing active caspase-9 av- Using fam-LETD-fmk and fam-LEHD-fmk fluoro-
eraged fewer than 30 per BP. Exposure to gentamicin genic substrates, we also found more caspase-8 (Fig.
for 6 h significantly increased the number of the 10D–F, n = 8) and caspase-9 (Fig. 11D–F, n = 12) ac-
Tuj1+/AC9+ (p < 0.001) but not that of Tuj1-/AC9+ tivities in gentamicin-damaged (0.5 mM for 12 h)
cells. At 12 h, gentamicin-treated BPs again contained than respective control tissues (Fig. 10A–C, n = 8; Fig.
significantly more Tuj1+/AC9+ cells than control BPs 10A–C, n = 12). Substantial activation of caspase-8
(p < 0.001), with the number of Tuj1-/AC9+ re- and caspase-9 was evident in the apical, middle, and
maining low and similar to that of the controls. When basal segments of the gentamicin-damaged BPs. The
CHENG ET AL.: Active Caspase in Hair Cells 99

FIG. 7. Activation of caspase-9 in untreated and gentamicin-treated Control BPs; without antibiotic exposure, very few caspase-9-posi-
auditory hair cells. Whole organ basilar papillae (BP) were treated tive cells (arrowheads) were observed. D–F. Gentamicin-treated BPs;
with 0.5 mM gentamicin for 12 h and tested alongside untreated hair cells from gentamicin-treated BPs show robust expression of
controls. These tissues were then fixed and processed for immuno- active caspase-9 in the apical, middle, and basal regions (Tuj1+/
histochemistry for Tuj1 (green) and active caspase-9 (red). A–C. AC9+; arrows). Scale bar = 20 lm.

majority of the hair cells with active caspase-8 or active caspases are key mediators of aminoglycoside-in-
caspase-9 were distributed near the junction of the duced hair cell death. It is important to point out that
basal and middle thirds of the BP, similar to the caspase inhibition did not completely block genta-
pattern seen with caspase-3 activation. micin-induced hair cell death, and that, despite cas-
pase inhibition, hair cell survival decreased when the
exposure to gentamicin was extended from 12 to 24
DISCUSSION h. This suggests that caspase-independent mecha-
nisms may also mediate hair cell death; a similar
Despite extensive literature on the loss of sensory hair phenomenon has been previously reported in the
cells as a result of aminoglycoside exposure, the cel- inner ear (Cheng et al. 1999) and other systems
lular cascades determining hair cell death or survival (Miller et al. 1997; Samejima et al. 1998; Stefanis et al.
remain unclear. Several lines of independent re- 1999). Although more sustained (3–6 days) protec-
search have recently described the involvement of tion has been reported with caspase inhibition in
caspases in hair cell degeneration (Liu et al. 1998; vestibular hair cells, this protection was also incom-
Forge and Li 2000; Matsui et al. 2002). We were able plete (Forge and Li 2000; Matsui et al. 2002).
to extend these findings to a mature avian auditory Investigators have suggested that differences in
receptor epithelium using fam-VAD-fmk substrate to susceptibility of apical versus basal hair cells in
show enhanced caspase (general) activity in genta- mammals may depend on intrinsic antioxidant levels
micin-treated hair cells. Using z-VAD-fmk, a broad- (Sha et al. 2001). Although our in vitro preparation of
spectrum caspase inhibitor, we were able to protect chick BP shows this difference in susceptibility, our
hair cells against gentamicin-induced death. These subsequent investigations of caspase activation did
results further support the hypothesis that activated not expressly address this issue. It is noteworthy,
100 CHENG ET AL.: Active Caspase in Hair Cells

negative cell death regulator (caspase-2), or have

unknown functions (caspase-10 and caspase-14). Up-
stream initiator caspases are activated in response to
specific cell death triggers. The specific upstream
caspase that is activated corresponds to the subcellu-
lar site of damage. Activation of certain cell surface
death receptors will result in activation of caspase-8,
whereas caspase-9 is preferentially activated by cellu-
lar stresses targeting the mitochondria (Earnshaw et
al. 1999; Hengartner 2000; Krammer 2000). Once
activated, these upstream caspases can cleave and
activate downstream effector caspases, thus forming a
caspase cascade. We have begun to look for the acti-
vation of specific caspases in degenerating hair cells
of the chick BP.
Significantly more immunologic expression is
demonstrated for both active caspase-3 and active
caspase-9 in gentamicin-treated hair cells than con-
trols cultured for the same duration. Using caspase
substrates to detect active caspase-3 and active cas-
pase-9, we correlated the increased immunoreactivity
with their actual activities in situ in response to gen-
tamicin treatment. Thus, the two methods used to
detect activation of specific caspases were in agree-
ment. In comparison to immunodetection, the sub-
strate activation assay appears to be more sensitive but
less specific. Like specific caspase inhibitors, the
substrates can potentially bind other individual casp-
ases with much lower affinity (Thornberry et al. 1997;
FIG. 8. Quantification of hair cells expressing active caspase-9 in
undamaged and gentamicin-damaged basilar papillae (BP). Orga- Garcia-Calvo et al. 1998; reviewed in Stennicke and
notypic cultures were exposed to no or 0.5 mM gentamicin for 6 or 12 Salvesen 1999). In addition, immunohistochemistry
h (n’s: control 6 h, n = 6; control 12 h, n = 5; gent 6 h, n = 5; gent 12 examines tissues at a single time-point, whereas cas-
h, n = 7). A. Significantly more Tuj1-positive (Tuj1+) hair cells ex- pase substrate in our experiment was exposed to ac-
pressed active caspase-9 (AC9+) after 6 h of gentamicin treatment in
tive caspase within live tissues for 1 h.
comparison to controls (p < 0.001). B. The number of Tuj1+/AC9+
cells remained significantly higher than controls in BPs treated with Our data strongly implicate the activation of cas-
gentamicin for 12 h (p < 0.001). Unlike active caspase-3 expression, pase-9 as one of the key events mediating gentamicin-
active caspase-9 was rarely seen in association with Tuj1-negative induced hair cell degeneration. Cellular stresses that
hair cells under any treatment paradigms tested. Error bars = SEM. result in depletion of cellular metabolic energy such
as reactive oxygen species (ROS) are associated with
however, that similar patterns of caspase-3, caspase-8, activation of caspase-9 via the binding of apoptotic
and caspase-9 activation were seen in the basal, mid- protease activating factor (Apaf-1) and cytochrome c
dle, and apical regions of the BP at an aminoglyco- released by the mitochondria (Earnshaw et al. 1999;
side level that causes hair cell loss in all three regions. Hengartner 2000). Generation of ROS and calcium
These findings suggest that hair cells with different dysregulation are known to occur in auditory hair
levels of sensitivity have similar cell death mediators. cells damaged by gentamicin in vivo and in vitro (re-
Only a subset of the 14 known mammalian casp- viewed in Rybak 1996; Hirose et al. 1997, 1999; Sha
ases functions as cell death mediators. These fall into and Schacht 1999; reviewed in Forge and Schacht
two general categories: initiators and effectors (see 2000). Three additional observations support the
reviews: Earnshaw et al. 1999; Nicholson 1999; Hen- concept that mitochondria play a key role in medi-
gartner 2000). Initiator caspases include caspase-8 ating aminoglycoside ototoxicity. First, patients with
and caspase-9, while effector caspases include cas- mutations of the mitochondrial ribosomal RNA are
pase-3, caspase-6, and caspase-7. Other caspase predisposed to aminoglycoside-induced hearing loss
members function to process cytokines (caspase-1, (Prezant et al. 1993; Estivill et al. 1997; Casano et al.
caspase-4, caspase-5, caspase-11, and caspase-13), 1999). Second, concurrent addition of chloramphe-
mediate damage directed at the endoplasmic reticu- nicol, a specific mitochondrial poison, has been
lum (caspase-12), act as a tissue-specific positive or shown to increase the extent of gentamicin-induced
CHENG ET AL.: Active Caspase in Hair Cells 101

FIG. 9. Confocal images of active caspase-3 detected with a fluo- cells with active caspase-3 were demonstrated along the control BPs
rescent-conjugated substrate. Basilar papillae (BP) were incubated in (arrowheads). D–F. Gentamicin-treated BPs; there was robust ex-
media supplemented with no or 0.5 mM gentamicin for 12 h. During pression of active caspase-3 in hair cells from the apical, middle, and
the last hour of this culture period, fam-DEVD-fmk was added to basal segments. Arrows point to examples of double-labeled cells.
specifically detect the presence of active caspase-3 (red) in these Despite considerable hair cell loss at this time-point, many of the
tissues, which were then fixed and stained with phalloidin (green) to remaining hair cells expressed active caspase-3. Scale bar = 20 lm.
label hair cell stereocilia. A–C. Control BPs; very few scattered hair

hair cell loss in vivo (Hyde and Rubel 1995). Third, application of a specific inhibitor of caspase-3 pre-
we have observed a caspase-independent transloca- vented cisplatin-induced hair cell death and reduced
tion of cytochrome c from the mitochondria to the TUNEL labeling. Our data indicate that gentamicin-
cytoplasm during gentamicin-induced hair cell de- treated hair cells strongly express active caspase-3.
generation in vitro (Cheng et al. 2002). Since nu- Interestingly, we also observed significantly more
merous pro-apoptotic (e.g., bax, bid) and anti- Tuj1-/AC3+ cells than Tuj1-/AC9+ cells after genta-
apoptotic (e.g., bcl-2, bcl-XL) signals converge at this micin treatment for 6 or 12 h. A reasonable expla-
organelle, the mitochondrion is thought to act as a nation for this observation is that the loss of Tuj1
central checkpoint for the apoptotic machinery in antigenicity is a late event during hair cell degener-
several systems (Earnshaw et al. 1999; Slee et al. ation. Therefore, we can speculate that the activation
1999a; Hengartner 2000). Our observation of signif- of caspase-3 occurs later than caspase-9 activation,
icant activation of caspase-9 with gentamicin expo- coinciding approximately with the time when anti-
sure is in agreement with these lines of evidence. genicity of Tuj1 is lost. While our experiments do not
Thus far we have examined only one effector cas- provide direct evidence that the initiator caspase-9 is
pase, caspase-3. Features of apoptosis such as nuclear activated prior to caspase-3 activation, this proposed
condensation and DNA fragmentation are mediated sequence of caspase activation corresponds to that
by the activity of caspase-3 (Woo et al. 1998; D’Mello described in the literature in other systems (Slee et al.
et al. 2000), and several studies have reported such 1999a,b; Hengartner 2000).
changes in gentamicin-treated hair cells (Li et al. Although we were unable to find any immunologic
1995; Lang and Liu 1997; Zheng et al. 1999; Forge evidence of active caspase-8, its potential role in me-
and Li 2000). Liu et al. (1998) showed that in vitro diating hair cell death cannot be ruled out. The
102 CHENG ET AL.: Active Caspase in Hair Cells
CHENG ET AL.: Active Caspase in Hair Cells 103

b Whether or not the mitochondrial or the death

FIG. 10. Detection of active caspase-9 in control and gentamicin- receptor pathway plays a more important role during
treated basilar papillae (BP) using fluorescent-labeled substrates. aminoglycoside-induced hair cell degeneration re-
Confocal images. A subset of the BP cultures was treated with 0.5 mains to be determined. Future experiments using
mM gentamicin for 12 h before fam-LEHD-fmk, a peptide that spe- specific inhibitors of caspase-8 or caspase-9 will fur-
cifically binds to active caspase-9 (red), was added. All tissues were
then fixed and stained with phalloidin (green). A–C. Control BPs; ther clarify the roles of these pathways during hair
healthy sheets of hair cells with little active caspase-9 labeling (ar- cell death and the relationships between the initiator
rowheads) were noted along the BP. D–F. Gentamicin-treated BPs; and effector caspases.
gentamicin treatment decreased the hair cell density and increased
the expression of active caspase-9 in hair cells throughout the BP.
Arrows indicate examples of double-labeled cells. The majority of Technical considerations
active caspase-9 expression was observed near the junction of the
middle and basal regions. Scale bar = 20 lm. Studies on hair cell damage in nonavian systems have
shown that loss of the apical stereocilia can precede
hair cell death (Sobkowicz et al. 1992; Zheng et al.
1999; Gale et al. 2002). We considered this compli-
b cation and carried out experiments to compare loss
FIG. 11. Active caspase-8 expression in untreated and gentamicin- of phalloidin with loss of Tuj1 antigenicity. In the
treated basilar papillae (BP). Confocal images. After exposure to no
or 0.5 mM gentamicin for 11 h, BP cultures received fam-LETD-fmk
gentamicin-treated chick BPs, we show phalloidin to
to specifically detect the activity of active caspase-8 (red). Fluores- be an accurate hair cell marker of surviving hair cells.
cent-conjugated phalloidin was used to label hair cells (green). A–C. The population of Tuj1-positive, phalloidin-negative
Control BPs; apical, middle, basal regions of untreated BPs con- cells was estimated to be less than 0.5% of all the
tained many hair cells but few labeled with active caspase-8 (ar- remaining hair cells. This interpretation, of course,
rowheads). D–F. Gentamicin-treated BPs; gentamicin treatment
promoted the activation of caspase-8 in many hair cells throughout
depends on the assumption that the disappearance of
the BP. Arrows point to examples of double-labeled cells. There are Tuj1 antigenicity is a late degenerative event.
occasional phalloidin-negative cells expressing active caspase-8 The observed basal–apical gradient of hair cell loss
noted near the luminal surface of the sensory epithelium. Similar to in chick BP is consistent with previous in vivo (Cruz et
the expression of active caspase-3 and active caspase-9, many of the al. 1987; Tucci and Rubel 1990; Janas et al. 1995,
hair cells showing active caspase-8 are concentrated in the middle
and basal thirds of the BP. Scale bar = 20 lm.
Torchinsky et al. 1999) and in vitro reports (Frenz et
al. 1998). With respect to both the temporal pattern
and gentamicin concentration, hair cells degenerated
finding that caspase-1 inhibitor protected against in a basal–apical fashion. Hair cell death in the apical
cisplatin-induced hair cell death (Liu et al. 1998) third is seldom seen with in vivo application of am-
supports a role for the death receptor pathway during inoglycosides. Our findings of hair cell death in these
hair cell degeneration. Although caspase-1 is pri- regions are in agreement with those reported in the
marily known to mediate inflammatory responses, literature for in vitro preparations as well (Hirose et
caspase-1-deficient cells have been reported to be al. 1997; Frenz et al. 1998).
more resistant to cell death stimulated by the death
receptor pathway (Earnshaw et al. 1999). Pro-caspase-
8 is closely associated with membrane-bound recep- ACKNOWLEDGMENTS
tors carrying so-called death domains, a model well
characterized in the immune system (Slee et al. We thank Dr. Sarah Woolley for assistance with data anal-
1999a; Hengartner 2000; Krammer 2000). Upon ac- ysis, Laurie Johnson for help with preparation of the man-
tivation of these receptors by TNFA or Fas (both uscript, Glen MacDonald, Dr. Jialin Shang, and Mae del
paracrines secreted by inflammatory cells), pro-cas- Puerto for their excellent technical support. This work was
supported by NIDCD grants DC-00018 and DC-00054 and
pase-8 molecules are attracted to and cluster at the
the American Academy of Otolaryngology–Head and Neck
receptor level to allow for autocleavage and thus ac-
Surgery Resident Research Grant.
tivation. Macrophages and other as-yet-unspecified
inflammatory cells normally reside within the audi-
tory SE (Warchol 1997; Bhave et al. 1998). Within REFERENCES
hours of aminoglycoside exposure, their numbers
increase, especially in areas of hair cell lesions (War- BEDNER E, SMOLEWSKI P, AMSTAD P, DARZYNKIEWICZ Z. Activation of
chol 1997; Bhave et al. 1998). Although their precise caspases measured in situ by binding of fluorochrome-labeled
inhibitors of caspases (FLICA): correlation with DNA frag-
function(s) within the sensory organ remains un-
mentation. Exp. Cell Res. 259:308–313, 2000.
clear, these inflammatory cells may potentially pro- BHAVE SA, OESTERLE EC, COLTRERA MD. Macrophage and microglia-
mote hair cell degeneration by releasing TNFA, Fas, like cells in the avian inner ear. J. Comp. Neurol. 398:241–256,
or other death receptor ligands. 1998.
104 CHENG ET AL.: Active Caspase in Hair Cells

CASANO RA, JOHNSON DF, BYKHOVSKAYA Y, TORRICELLI F, BIGOZZI M, HIROSE K, WESTRUM LE, STONE JS, ZIRPEL L, RUBEL EW. Dynamic
FISCHEL-GHODSIAN N. Inherited susceptibility to aminoglycoside studies of ototoxicity in mature avian auditory epithelium. Ann.
ototoxicity: genetic heterogeneity and clinical implications. Am. N. Y. Acad. Sci. 884:389–409, 1999.
J. Otolaryngol 20:151–156, 1999. HU BH, GUO W, WANG PY, HENDERSON D, JIANG SC. Intense noise-
CASIANO CA, MARTIN SJ, GREEN DR, TAN EM. Selective cleavage of induced apoptosis in hair cells of guinea pig cochleae. Acta
nuclear autoantigens during CD95 Fas/APO-1-mediated T cell Otolaryngol. 120:19–24, 2000.
apoptosis. J. Exp. Med. 184:765–770, 1996. HYDE GE, RUBEL EW. Mitochondrial role in hair cell survival after
CHENG AG, CUNNINGHAM LL, RUBEL EW. Translocation of cyto- injury. Otolaryngol. Head Neck Surg. 113:530–540, 1995.
chrome c in hair cells of gentamicin-treated avian basilar pa- JANAS JD, COTANCHE DA, RUBEL EW. Avian cochlear hair cell re-
pilla. Abstr. Assoc. Res. Otolaryngol. 25:35, 2002. generation: stereological analyses of damage and recovery from
CHENG AG, HUANG T, STRACHER A, KIM A, LIU W, MALGRANGE B, a single high dose of gentamicin. Hear. Res. 92:17–29, 1995.
LEFEBVRE PP, SCHULMAN A, VAN DE WATER TR. Calpain inhibi- JANICKE RU, NG P, SPRENGART ML, PORTER AG. Caspase-3 is required
tors protect auditory sensory cells from hypoxia and neurotro- for alpha-fodrin cleavage but dispensable for cleavage of other
phin-withdrawal induced apoptosis. Brain Res. 850:234–243, death substrates in apoptosis. J. Biol. Chem. 273:15540–15545,
1999. 1998.
CLARKE PG, CLARKE S. Historic apoptosis. Nature 378:230, 1995. KERR JF, WYLLIE AH, CURRIE AR. Apoptosis: a basic biological phe-
CLEM RJ, CHENG EH, KARP CL, KIRSCH DG, UENO K, TAKASHI A, nomenon with wide-ranging implications in tissue kinetics. Br. J.
KATAN MB, GRIFFIN DE, EARNSHAW WC, VELIUONA MA, HARDWICK Cancer 26:239–257, 1972.
JM. Modulation of cell death by bcl-XL through caspase inter- KIL J, WARCHOL ME, CORWIN JT. Cell death, cell proliferation, and
action. Proc. Natl. Acad. Sci. USA 95:554–559, 1998. estimates of hair cell life spans in the vestibular organs of chicks.
CRUZ RM, LAMBERT PR, RUBEL EW. Light microscopic evidence of hair Hear. Res. 114:117–126, 1997.
cell regeneration after gentamicin toxicity in chick cochlea. KRAMMER PH. CD95 is deadly mission in the immune system. Nature
Arch. Otolaryngol. Head Neck Surg. 113:1058–1062, 1987. 407:789–795, 2000.
CUNNINGHAM LL, CHENG AG, RUBEL EW. Caspase activation in hair LANG H, LIU C. Apoptosis and hair cell degeneration in the ves-
cells of the mouse utricles exposed to neomycin. J. Neurosci. tibular sensory epithelia of the guinea pig following a genta-
2002 in press. micin insult. Hear. Res. 111:177–184, 1997.
D’MELLO SR, KUAN C-Y, FLAVELL RA, RAKIC P. Caspase-3 is required LAZEBNIK YA, TAKAHASHI A, MOIR RD, GOLDMAN RD, POIRIER GG,
for apoptosis-associated DNA fragmentation but not for cell KAUFMANN SH, EARNSHAW WC. Studies of the lamin proteinase
death in neurons deprived of potassium. J. Neurosci. Res. reveal multiple parallel biochemical pathways during apoptotic
59:24–31, 2000. execution. Proc. Natl. Acad Sci. USA 92:9042–9046, 1995.
EARNSHAW WC, MARTINS LM, KAUFMANN SH. Mammalian caspases: LENOIR M, DAUDE N, HUMBERT G, RENARD N, GALLEGO M, PUJOL R,
structure, activation, substrates, and functions during apoptosis. EYBALIN M, VAGO P. Morphological and molecular changes in
Annu. Rev. Biochem. 68:383–424, 1999. the inner hair cell region of the rat cochlea after amikacin
ENARI M, SAKAHIRA H, YOKOYAMA H, OKAWA K, IWAMATSU A, NAGATA S. treatment. J. Neurocytol. 28:925–937, 1999.
A caspase-activated DNase that degrades DNA during apoptosis, LI L, NEVILL G, FORGE A. Two modes of hair cell loss from the
and its inhibitor ICAD [see comments]. Nature 391:43–50, vestibular sensory epithelia of the guinea pig inner ear. J. Comp.
1998. Neurol. 355:405–417, 1995.
ESTIVILL X, GOVEA N, BARCELÓ A, PERELLÓ E, BADENAS C, ROMERO E, LIU W, STAECKER H, STUPAK H, MALGRANGE B, LEFEBVRE P, VAN DE
MORAL L, SCOZZARI R, D’URBANO L, ZEVIANI M, TORRONI A. WATER TR. Caspase inhibitors prevent cisplatin-induced apop-
Familial progressive sensorineural deafness is mainly due to the tosis of auditory sensory cells. Neuroreport 9:2609–2614, 1998.
mtDNA A1555G mutation and is enhanced by the treatment of MASHIMA T, NAITO M, FUJITA N, NOGUCHI K, TSURUO T. Identification
aminoglycosides. Am. J. Hum. Genet. 62:27–35, 1997. of actin as a substrate of ICE and an ICE-like protease and
FORGE A. Outer hair cell loss and supporting cell expansion fol- involvement of an ICE-like protease but not ICE in VP-16-in-
lowing chronic gentamicin treatment. Hear. Res. 19:171–182, duced U937 apoptosis. Biochem. Biophys. Res. Commun.
1985. 217:1185–1192, 1995.
FORGE A, LI L. Apoptotic death of hair cells in mammalian vestib- MASHIMA T, NAITO M, NOGUCHI K, MILLER DK, NICHOLSON DW,
ular sensory epithelia. Hear. Res. 139:97–115, 2000. TSURUO T. Actin cleavage by CPP-32/apopain during the de-
FORGE A, SCHACHT J. Aminoglycoside antibiotics. Audiol. Neurootol. velopment of apoptosis. Oncogene 14:1007–1012, 1997.
5:3–22, 2000. MATSUI JI, OGILVIE JM, WARCHOL ME. Inhibition of caspases pre-
FRENZ DA, YOO H, LIU W. Basilar papilla explants: a model to study vents ototoxic and ongoing hair cell death. J. Neurosci.
hair cell regeneration–repair and protection. Acta Otolaryngol. 22:1218–1227, 2002.
118:651–659, 1998. MILLER TM, MOULDER KL, KNUDSON CM, CREEDON DJ, DESHMUKH M,
GALE JE, MEYERS JR, PERIASAMY A, CORWIN JT. Survival of bundleless KORSMEYER SJ, JOHNSON JR EM. Bax deletion further orders the
hair cells and subsequent bundle replacement in the bullfrog’s cell death pathway in cerebellar granule cells and suggests a
saccule. J. Neurobiol. 50:81–92, 2002. caspase-independent pathway to cell death. J. Cell. Biol.
GARCIA–CALVO M, PETERSON EP, LEITING B, RUEL R, NICHOLSON DW, 139:205–217, 1997.
THORNBERRY NA. Inhibition of human caspases by peptide-based NICHOLSON DW. Caspase structure, proteolytic substrates, and
and macromolecular inhibitors. J. Biol. Chem. 273:32608– function during apoptotic cell death. Cell Death Differ. 6:1028–
32613, 1998. 1042, 1999.
GAVRIELI Y, SHERMAN Y, BEN-SASSON SA. Identification of pro- OESTERLE EC, TSUE TT, REH TA, RUBEL EW. Hair-cell regeneration
grammed cell death in situ via specific labeling of nuclear DNA in organ cultures of the postnatal chicken inner ear. Hear. Res.
fragmentation. J. Cell. Biol. 119:493–501, 1992. 70:85–108, 1993.
HENGARTNER MO. The biochemistry of apoptosis. Nature 407:770– PIRVOLA U, XING-QUN L, VIRKKALA J, SAARMA M, MURAKATA C,
776, 2000. CAMORATTO AM, WALTON KM, YLIKOSKI J. Rescue of hearing,
HIROSE K, HOCKENBERY DM, RUBEL EW. Reactive oxygen species in auditory hair cells, and neurons by CEP-1347/KT7515, an in-
chick hair cells after gentamicin exposure in vitro. Hear. Res. hibitor of c-Jun N-terminal kinase activation. J. Neurosci. 20:43–
104:1–14, 1997. 50, 2000.
CHENG ET AL.: Active Caspase in Hair Cells 105

PREZANT TR, AGAPIAN JV, BOHLMAN MC, BU X, OZTAS S, QIU WQ, THORNBERRY NA, RANO TA, RASPER DM, TIMKEY T, GARCIO–CALVO M,
ARNOS KS, CORTOPASSI GA, JABER L, ROTTER JI. Mitochondrial HOUTZAGER VM, NORDSTROM PA, ROY S, VAILLANCOURT JP,
ribosomal RNA mutation associated with both antibiotic-induced CHAPMAN KT, NICHOLSON DW. A combinatorial approach de-
and non-syndromic deafness. Nat. Genet. 4:289–294, 1993. fines specificities of members of the caspase family and gran-
RYBAK LP. Ototoxicity. Curr. Opin. Otolaryngol. Head Neck Surg. zyme B. J. Biol. Chem. 272:17907–17911, 1997.
4:302–307, 1996. TORCHINSKY C, MESSANA EP, ARSURA M, COTANCHE DA. Regulation of
SAMEJIMA K, TONE S, KOTTKE TJ, ENARI M, SAKAHIRA H, COOKE CA, p27Kip1 during gentamicin mediated hair cell death. J. Neu-
DURRIEU F, MARTINS LM, NAGATA S, KAUFMANN SH, EARNSHAW rocytol. 28:913–924, 1999.
WC. Transition from caspase-dependent to caspase-indepen- TUCCI DL, RUBEL EW. Physiologic status of regenerated hair cells in
dent mechanisms at the onset of apoptotic execution. J. Cell the avian inner ear following aminoglycoside ototoxicity. Oto-
Biol. 143:225–239, 1998. laryngol. Head Neck Surg. 103:443–450, 1990.
SHA SH, SCHACHT J. Stimulation of free radical formation by ami- USAMI S, TAKUMI Y, FUJITA S, SHINKAWA H, HOSOKAWA M. Cell death
noglycoside antibiotics. Hear. Res. 128:112–118, 1999. in the inner ear associated with aging is apoptosis? Brain Res.
SHA SH, TAYLOR R, FORGE A, SCHACHT J. Differential vulnerability of 747:147–150, 1997.
basal and apical hair cells is based on intrinsic susceptibility to WARCHOL ME. Macrophage activity in organ cultures of the avian
free radicals. Hear. Res. 155(1–2):1–8, 2001. cochlea: demonstration of a resident population and recruit-
SLEE EA, ADRAIN C, MARTIN SJ. Serial killers: ordering caspase acti- ment to sites of hair cell lesions. J. Neurobiol. 33:724–734,
vation events in apoptosis. Cell Death Differ. 6:1067–1074, 1999a. 1997.
SLEE EA, HARTE MT, KLUCK RM, WOLF BB, CASIANO CA, NEWMEYER WEISLEDER P, RUBEL EW. Hair cell regeneration after streptomycin
DD, WANG H, REED JC, NICHOLSON DW, ALNEMRI ES, GREEN DR, toxicity in the avian vestibular epithelium. J. Comp. Neurol.
MARTIN SJ. Ordering the cytochrome c-initiated caspase cascade: 331:97–110, 1993.
hierarchical activation of caspases-2, -3, -6, -7, -8, and -10 in a WILLIAMS JA, HOLDER N. Cell turnover in neuromasts of zebrafish
caspase-9-dependent manner. J. Biol. Chem. 144:281–292, larvae. Hear. Res. 143:171–181, 2000.
1999b. WOO M, HAKEM R, SOENGAS MS, DUNCAN GS, SHAHINIAN A, KAGI D,
SOBKOWICZ HM, AUGUST BK, SLAPNICK SM. Epithelial repair follow- HAKEM A, MCCURRACH M, KHOO W, KAUFMAN SA, SENALDI G,
ing mechanical injury of the developing organ of corti in cul- HOWARD T, LOWE SW, MAK TW. Essential contribution of caspase
ture: an electron microscopic and autoradiographic study. Exp. 3/CPP32 to apoptosis and its associated nuclear changes. Genes
Neurol. 115:44–49, 1992. Dev. 12:806–819, 1998.
STEFANIS L, PARK DS, FRIEDMAN WJ, GREEN LA. Caspase-dependent YUAN J, SHAHAM S, LEDOUX S, ELLIS HM, HORVITZ HR. The C. elegans
and -independent death of camptothecin-treated embryonic cell death gene ced-3 encodes a protein similar to mamma-
cortical neurons. J. Neurosci. 19:6235–6247, 1999. lian interleukin-1 beta-converting enzyme. Cell 75:641–652,
STENNICKE HR, SALVESEN GS. Caspases: preparation and character- 1993.
ization. Methods 17:313–319, 1999. ZHENG JL, KELLER G, GAO W-Q. Immunocytochemical and mor-
STONE JS, LEANO SG, BAKER LP, RUBEL EW. Hair cell differentiation phological evidence for intracellular self-repair as an important
in chick cochlear epithelium after aminoglycoside toxicity: in contributor to mammalian hair cell recovery. J. Neurosci.
vivo and in vitro observations. J. Neurosci. 16:6157–6174, 1996. 19:2161–2170, 1999.
TANG D, KIDD VJ. Cleavage of Dff-45/ICAD by multiple caspases is ZHENG Y, IKEDA K, NAKAMURA M, TAKASAKA T. Endonuclease cleavage
essential for its function during apoptosis. J. Biol. Chem. of DNA in the aged cochlea of Mongolian gerbil. Hear. Res.
273:28549–28552, 1998. 126:11–18, 1998.