Blockade of c-Jun N-terminal kinase pathway attenuates gentamicin-induced cochlear and vestibular hair cell death

Hearing Research 163 (2002) 71^81 www.elsevier.com/locate/heares

Blockade of c-Jun N-terminal kinase pathway attenuates gentamicin-induced cochlear and vestibular hair cell death Jukka Ylikoski a

a;b;

*, Liang Xing-Qun

a;b

, Jussi Virkkala b , Ulla Pirvola

a;b

Institute of Biotechnology, University of Helsinki, P.O. Box 56 (Viikinkaari 9), 00014 Helsinki, Finland b Department of ORL, University of Helsinki, 00290 Helsinki, Finland Received 16 July 2001; accepted 23 August 2001

Abstract The ototoxic action of aminoglycoside antibiotics leading to the loss of hair cells of the inner ear is well documented. However, the molecular mechanisms are poorly defined. We have previously shown that in neomycin-exposed organotypic cultures of the cochlea, the c-Jun N-terminal kinase (JNK) pathway ^ associated with stress, injury and apoptosis ^ is activated in hair cells and leads to their death. We have also shown that hair cell death can be attenuated by CEP-1347, an inhibitor of JNK signalling [Pirvola et al., J. Neurosci. 20 (2000) 43^50]. In the present study, we demonstrate that gentamicin-induced ototoxicity leads to JNK activation and apoptosis in the inner ear hair cells in vivo. We also show that systemic administration of CEP-1347 attenuates gentamicin-induced decrease of auditory sensitivity and cochlear hair cell damage. In addition, CEP-1347 treatment reduces the extent of hair cell loss in the ampullary cristae after gentamicin intoxication. Particularly, the inner hair cells of the cochlea and type I hair cells of the vestibular organs are protected. We have previously shown that also acoustic overstimulation leads to apoptosis of cochlear hair cells and that CEP-1347 can attenuate noise-induced sensory cell loss. These results suggest that activation of the JNK cascade may be a common molecular outcome of cellular stress in the inner ear sensory epithelia, and that attenuation of the lesion can be provided by inhibiting JNK activation. ß 2002 Elsevier Science B.V. All rights reserved. Key words: Cellular stress; c-Jun N-terminal kinase signalling; Cellular death; Inner ear; Hair cell; Ototoxicity

1. Introduction The ototoxic potential of aminoglycoside antibiotics is well known. Therefore, their clinical use has been

* Corresponding author. Tel.: +358 (9) 47173325; Fax: +358 (9) 19159366. E-mail address: jukka.ylikoski@helsinki.¢ (J. Ylikoski). Abbreviations: ABR, auditory brainstem response; DAPI, 4P,6diamidino-2-phenylindole; DIC, di¡erential interference contrast; GA, glutaraldehyde; GM, gentamicin; GSTp, glutathione S-transferase Pi; HC, hair cell; IHC, inner hair cells; JNK, c-Jun N-terminal kinase; OHC, outer hair cells; PBS, phosphate-bu¡ered saline; PFA, paraformaldehyde; ROS, reactive oxygen species; TUNEL, terminal deoxynucleotidyl transferase-mediated biotinylated UTP nick end labeling

limited to the treatment of severe infections caused by Gram-negative microorganisms. The ototoxic capacity of aminoglycoside antibiotics has been widely used in experimental research to create inner ear lesions, despite the fact that the molecular mechanisms of ototoxicity have remained obscure. This has further been complicated by the recent demonstration that megalin, a higha¤nity receptor for aminoglycosides, is expressed in the secretory, but not in the sensory epithelia of the inner ear (Ylikoski et al., 1997). Further understanding of the ototoxic mechanisms of aminoglycosides has been provided by a series of articles during the last decade indicating that the ototoxic e¡ects of gentamicin (GM) require an `activated' form of the drug (Huang and Schacht, 1990). This `activated' GM leads to the formation of a redox-active iron^GM complex (Priuska and Schacht, 1995) and the generation of reactive oxygen species (ROS) (Clerici et al., 1996; Hirose et al., 1997 ; Sha and Schacht, 1999). This hypothesis is sup-

0378-5955 / 02 / $ ^ see front matter ß 2002 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 8 - 5 9 5 5 ( 0 1 ) 0 0 3 8 0 - X

HEARES 3778 3-1-02 Cyaan Magenta Geel Zwart

72

J. Ylikoski et al. / Hearing Research 163 (2002) 71^81

ported by the ability of certain antioxidants or free radical scavengers and iron chelators to protect from aminoglycoside-induced ototoxicity (Hulka et al., 1993 ; Garetz et al., 1994; Song and Schacht, 1996; Song et al., 1997). Another step towards understanding the mechanisms of aminoglycoside ototoxicity came from a study showing that concurrent administration of GM and N-methyl-D-aspartate antagonists markedly attenuates both hearing loss and destruction of cochlear hair cells (HCs) in guinea pigs (Basile et al., 1996). From that study it was concluded that aminoglycoside-induced ototoxicity is mediated, in part, through an excitotoxic process. During recent years, an increasing body of evidence suggests that HC death after aminoglycoside ototoxicity can occur through apoptosis. This possibility was mentioned 15 years ago in GM-intoxicated cochleas (Forge, 1985). Apoptosis and necrosis are two forms of cell death that are de¢ned based on morphological and biochemical criteria. In apoptosis, chromatin condensation, cellular shrinkage and early preservation of plasma membrane integrity contrast with the cytoplasmic disintegration and disorganized clumping of chromatin in necrosis (Kerr et al., 1972; Wyllie et al., 1980). Apoptosis is a gene-directed self-destruction program that mainly results from posttranslational activation of a set of proteins, which are involved in intracellular signalling cascades (Ra¡, 1992; Weil et al., 1996). In contrast, necrosis is thought to result from more passive mechanisms triggered by extrinsic insults (e.g. trauma, toxins, microbes). The apoptosis versus necrosis classi¢cation has been useful in categorizing cell death in numerous settings, but the relationship between these modes of cell death is not always clear. For instance, following excitotoxic or anoxic^ischemic injury, biochemical features of apoptosis and morphological evidence of necrosis have been observed even in the same individual neurons of the adult brain (PorteraCaillau et al., 1997). Revealing the intracellular signalling pathways that are activated in stressed HCs might o¡er a possibility to attenuate stress-induced HC death. We have recently demonstrated that the c-Jun N-terminal kinase (JNK) pathway (Derijard et al., 1994; Kyriakis et al., 1994) is activated in stressed cochlear HCs in vitro and that CEP-1347, an indolocarbazole that inhibits JNK signalling (Maroney et al., 1998), protects auditory HCs from neomycin damage in vitro and from noise trauma in vivo (Pirvola et al., 2000). In the present study, we have studied whether aminoglycoside antibiotics cause activation of the JNK pathway and induction of apoptosis in the inner ear HCs in vivo. Further, we have investigated whether CEP-1347 can protect cochlear and vestibular HCs from aminoglycoside-induced death in vivo.

2. Materials and methods 2.1. Animals, tissues, lesioning and delivery of test compounds Adult Dunkin^Hartley female guinea pigs (weight 300^400 g) were used. They were given free access to water and a regular guinea pig diet. Two experimental groups were formed, six guinea pigs in each group: b

b

Group I, GM only. The animals were injected s.c. with GM (Gensumycin, Hoechst Marion Roussel) (120 mg/kg body weight), once daily for 14 days. One animal of this group died on day 13 due to intoxication and its inner ears could not be investigated. Two animals of this group were decapitated 1 day after the last GM injection (day 15). They showed signs of severe intoxication and their inner ears were used for apoptosis studies (see below). The remaining three guinea pigs were decapitated under deep anesthesia 30 days after the initiation of GM injections. Five ears were used for cytocochleograms and one ear was embedded in para¤n. Group II, GM plus CEP-1347. The animals were injected s.c. with GM (120 mg/kg body weight) once daily for 14 days. In addition, these animals received a daily injection of CEP-1347 (Cephalon, 1 mg/kg, s.c.) dissolved in 5% Solutol (BASF) in phosphate-bu¡ered saline (PBS, pH 7.4). The 1 mg/kg dose of CEP-1347 was prepared daily from a 5 or 10 mg/ml stock in 25% Solutol that was stored protected from light at +4³C. CEP-1347 treatment was started 1 day before GM injections and continued for 28 subsequent days. These animals were decapitated under deep anesthesia 30 days after the initiation of GM injections. Of the six CEP-1347-treated guinea pigs, one was used for apoptosis study (decapitated on day 15) and the remaining ¢ve animals survived the 30 day study period. Of the inner ears of these ¢ve guinea pigs, two cochleas were destroyed during the preparation of cytocochleograms and could not be used for morphological studies. Six ears of the remaining four animals were used for cytocochleograms, and the other ears were embedded in para¤n. In addition, ¢ve guinea pigs served as normal, untreated controls.

2.2. Evaluation of auditory function Thresholds of auditory brainstem responses (ABRs) were determined from each ear 2 or 3 days before the injections started (baseline values) and at the end of the

HEARES 3778 3-1-02 Cyaan Magenta Geel Zwart

J. Ylikoski et al. / Hearing Research 163 (2002) 71^81

73

exposure period (30 days after the initiation of GM treatment). Guinea pigs were anesthetized with a mixture of ketamine (40 mg/kg, s.c.) and xylazine (10 mg/ kg, s.c.) and placed on a heated pad inside a soundisolated booth. ABRs were measured with System II hardware and BioSig software (Tucker Davis Technology). Stimuli with alternating polarity and 0.5 ms cos2 rise/fall and 9 ms plateau were presented at a 10 Hz rate with high-frequency transducer (Intelligent Hearing) connected to a speculum, which was placed in the meatus of the external ear canal. Stimuli were calibrated against a BruelpKjaer (BpK) 4133 microphone connected to a BpK 2606 sound level meter. Responses between vertex and mastoid subcutaneous electrodes were ampli¢ed with a digital ampli¢er (DB4/HS4, Tucker Davis Technology). Threshold was determined for frequencies 2.0, 4.0, 8.0, 16.0 and 32.0 kHz from a set of responses at varying intensities with 5 dB intervals and 1000 sweeps near threshold. 2.3. Processing the inner ears for histological analyses The animals were decapitated under deep anesthesia and perilymphatically perfused with 4% paraformaldehyde (PFA) in PBS (pH 7.4) or 2.5% glutaraldehyde (GA) in 0.1 M phosphate bu¡er (pH 7.4). Immersion in these ¢xatives continued overnight at +4³C. GA¢xed specimens were post¢xed with 1% osmium tetroxide, embedded in Epon and processed for cytocochleograms (Ylikoski, 1975). The dissected, Epon-embedded vestibular organs were cut for semithin (1 Wm) sections and stained with 1% toluidine blue. 2.4. HC counts The cochleas were prepared for Epon-embedded halfcoil surface preparations as earlier described (Ylikoski, 1975). An Olympus Provis microscope equipped with a 40U oil objective and di¡erential interference contrast (DIC) optics was used for cell counting. HCs were characterized as missing if no stereocilia, cuticular plate, cell membrane or nucleus in the appropriate location were observed. The quantity of cellular damage in the vestibular organs was evaluated from semithin plastic sections from the middle part of the horizontal or superior ampullary cristae by counting HCs of at least 10 corresponding transverse sections, 25 Wm apart. For each section, the length of the sensory epithelium was measured by an ocular micrometer. HCs were identi¢ed based on their morphology and location of their nucleus near the luminal part of the epithelium. A HC was counted as present if its nucleus was clearly visible. In further morphometric evaluations, type I HCs were identi¢ed on the basis of their prominent nerve calyx.

Fig. 1. Following 15 days of GM treatment, sensory epithelium of crista ampullaris of the guinea pig shows JNK activation as revealed by using phospho-JNK antibodies. (A) A normal (nontreated) crista does not show JNK induction. (B) JNK induction is seen in the crista of a guinea pig treated with GM. (C) High magni¢cation shows that HC nuclei stain strongly, cytoplasm weakly, whereas supporting cells show no staining. Scale bar: A,B, 200 Wm; C, 20 Wm.

Results are graphed as the mean þ S.E.M. Di¡erences were assessed using paired Student's t-test. P values of 6 0.05 were considered signi¢cant. 2.5. Immunohistochemistry PFA-¢xed inner ears were decalci¢ed in 0.5 M EDTA (pH 8.0) at +4³C and embedded in para¤n. The blocks

HEARES 3778 3-1-02 Cyaan Magenta Geel Zwart

74

J. Ylikoski et al. / Hearing Research 163 (2002) 71^81

Fig. 2. GM-induced HC apoptosis in guinea pigs in vivo. A para¤n section doublestained with the DAPI nuclear stain (A) and the TUNEL-method (B) shows fragmented nuclei of OHCs of the guinea pig cochlea 15 days after the initiation of GM treatment. Small arrows mark OHCs, large arrow the IHC. SM, scala media; ST, scala tympani; is, inner sulcus. (C) A semithin plastic section of the normal vestibular sensory epithelium. (D) A semithin plastic section of an ampullary crista treated with GM for 15 days shows several vestibular HCs that are in the process of apoptosis as revealed by their fragmented nuclei (arrows). Most of the nuclei of apoptotic HCs are located the level of supporting cell nuclei. Scale bar: A,B, 80 Wm; C,D, 20 Wm. C

were cut midmodiolarly to 5 Wm thick sections and mounted on 3-aminopropyl ethoxysilane-coated (Sigma) slides. For immunostaining, para¤n-embedded sections were depara¤nized and incubated with polyclonal phospho-JNK antibodies (Thr183/Tyr185, New England Biolabs) (1:1000 dilution). For detection, tyramide signal ampli¢cation (NEN), ABC Elite kit (Vector) and 3,3P-diaminobenzidine were used as previously described (Pirvola et al., 2000). Sections were lightly counterstained with 1% methyl green. 2.6. Assessment of apoptosis Para¤n-embedded sections were stained by the terminal deoxynucleotidyl transferase-mediated biotinylated UTP nick end labeling (TUNEL) method using the £uorescein kit by Boehringer-Mannheim as previously described (Pirvola et al., 2000). 4P,6-Diamidino-2phenylindole (DAPI) was used to visualize the nuclei. In addition, fragmented nuclei were veri¢ed from Eponembedded semithin sections. All microscopic (Olympus Provis) images under bright-¢eld, £uorescence and DIC optics were digitized using a Photometrics SenSys CCD video camera. Figures were processed using Image-Pro Plus 3.0, Adobe Photoshop 4.0, and Micrografx Designer 6.0. 2.7. Organotypic cochlear cultures Cochlear explants including the basal and middle turns were dissected from postnatal day 2 BALB/c mice and kept on Nuclepore ¢lters (pore size 0.1 Wm, Pleasanton) placed on a metal grid as previously described (Pirvola et al., 2000). The explants were maintained in F12 medium (Life Technologies) containing 15% fetal calf serum (Life Technologies) at 37³C with 5% CO2 . At the time of initiation of incubations, CEP1347 (Cephalon) was added to cochlear cultures at concentrations of 100 nM, 500 nM, 1000 nM and 5000 nM. After a 2 h stabilization period, neomycin sulfate (Sigma) was added at a concentration of 100 WM. After a 48 h exposure, cochlear explants were ¢xed with 4% PFA/0.5% GA in PBS (pH 7.4) for 1 h at room

temperature. They were stained with rhodamine-phalloidin (Molecular Probes) and DAPI, and surface preparations were prepared as previously described (Pirvola et al., 2000). Phalloidin is a speci¢c marker for cellular F-actin and DAPI shows cell nuclei under UV illumi-

HEARES 3778 3-1-02 Cyaan Magenta Geel Zwart

J. Ylikoski et al. / Hearing Research 163 (2002) 71^81

75

Fig. 3. Cytocochleograms of GM-treated (control 5^13) and GM+CEP-1347-treated (CEP 8^11) cochleas. Cytocochleograms (n = 5) of GMonly cochleas show a severe loss of OHCs in the basal turns. In two of these guinea pig cochleas, the entire organ of Corti, including the IHCs is destroyed in the basal two-third region. Cytocochleograms of GM+CEP-1347-treated cochleas (n = 6) also show a severe OHC loss in the basal halves, but the loss is signi¢cantly smaller than in the GM-only group. The IHC loss is substantially smaller than in the GM-only group. F OHCs, 8 IHCs.

nation. Because inner hair cells (IHCs) could not always be revealed in the surface preparations, only outer hair cells (OHCs) were included in the analysis performed under an Olympus Provis microscope using epi£uorescence. HCs were characterized as missing if no cuticular plate or regular nuclei were observed. Analyses of OHCs were done using a 40U objective lens and

an ocular grid. Because of the resistance of HCs of the middle (and apical) turn to neomycin, only the basal turn was included in the analysis. Three separate experiments, each including at least three explants at each concentration, were analyzed. Student's t-test was used for statistical analysis. P 6 0.05 was considered statistically signi¢cant.

HEARES 3778 3-1-02 Cyaan Magenta Geel Zwart

76

J. Ylikoski et al. / Hearing Research 163 (2002) 71^81

3. Results 3.1. JNK activation and apoptosis in HCs in vivo Two guinea pigs were decapitated after a 14 day period of GM injections (day 15). Also inner ears of normal (non-treated) guinea pigs were prepared for histology. By staining para¤n sections with phospho-JNK antibodies, induction of JNK activation was found in GM-treated, but not in normal vestibular HCs (Fig. 1A^C ). In GM-exposed vestibular organs, weak staining was seen in the cytoplasm and strong staining in the nuclei of HCs (Fig. 1B,C). In the traumatized organ of Corti of adult animals, these antibodies caused background problems, perhaps due to the unavoidable decalci¢cation process, and we could not reliably document JNK phosphorylation there. However, we have previously shown by immunohistochemistry that an ototoxic insult leads to JNK activation in neonatal cochlear HCs in vitro (Pirvola et al., 2000). To detect apoptosis, para¤n sections were doublestained with the nuclear dye DAPI (Fig. 2A) and the TUNEL method (Fig. 2B). As analyzed after a 14 day period of GM injections (day 15), both cochlear (Fig. 2A,B) and vestibular (data not shown) HCs showed apoptotic pro¢les. TUNEL staining was not seen in the inner ear sensory epithelia of normal guinea pigs (data not shown). To verify apoptosis, contralateral ears (n = 2) were embedded in plastic and cut to semithin sections, and cellular morphology was investigated. Cells were interpreted as apoptotic if they showed shrunken and fragmented nuclei. In contrast to GMexposed cochleas and vestibular organs, the sensory epithelia of normal guinea pigs did not show nuclear fragmentation (Fig. 2C). In GM-exposed vestibular organs, apoptotic HC nuclei seemed to move downwards towards the level of supporting cell nuclei (Fig. 2D). Thus, disappearing HCs seemed to be processed and eliminated through apoptosis within the sensory epithelium. However, some HCs with normal-looking nuclei

Fig. 4. Histograms (average þ S.E.M.) of cochlear HC loss after GM treatment (n = 5 cochleas) and after GM+CEP-1347 treatment (n = 6 cochleas) in vivo. 4480 OHCs and 572 IHCs are lost in GMonly cochleas. After GM+CEP-1347 treatment, these values are 2952 and 73, respectively.

Fig. 5. The average ( þ S.E.M.) auditory brainstem (ABR) thresholds in GM-treated ears (n = 6) and GM+CEP-1347-treated ears (n = 8) of the guinea pig. The average threshold elevations are greater in the GM-only group (range from about 51 to 86 dB SPL) than in the GM+CEP-1347 group (range from about 6 to 50 dB) at all test frequencies tested (at 2.0, 4.0 and 16.0 kHz **P 6 0.01; at 8.0 and 32.0 kHz *P 6 0.05).

showed bulging of their luminal cytoplasm (data not shown). These HCs may become extruded into the endolymphatic space. The mode of death of these cells remains to be elucidated. 3.2. GM-induced cochlear HC damage in vivo and its attenuation by CEP-1347 3.2.1. Morphometric data of the cochlea As analyzed at day 30 post-exposure (14 day period of GM injections followed by a post-exposure period of 16 days), the cochleas treated with GM only (n = 5) showed a severe loss of OHCs in their basal part. As seen in the individual cytocochleograms (Fig. 3), the entire organ of Corti, including the IHCs, was destroyed in the basal region of two cochleas. The other three cochleas showed smaller areas (extending over

Fig. 6. Dose dependence of CEP-1347 for OHC survival in neomycin-challenged (100 WM, 48 h) cochlear cultures prepared from postnatal day 2 mice. The protective e¡ect increased from 70% with a CEP-1347 concentration of 100 nM to 95% with a concentration of 1000 nM. A concentration of 5000 nM of CEP-1347 had no protective e¡ect. For each concentration, the average þ S.E.M. is based on three separate experiments, each including at least three cochlear explants.

HEARES 3778 3-1-02 Cyaan Magenta Geel Zwart

J. Ylikoski et al. / Hearing Research 163 (2002) 71^81

77

Fig. 7. Vestibular HC damage after GM treatment and GM+CEP-1347 treatment in vivo. (A,B) Ampullary cristae of GM-only-treated guinea pigs show loss of most of the type I HCs. (C,D) A smaller proportion of type I HCs are lost in ampullary cristae of GM+CEP-1374-treated animals. Scale bar: A,C, 160 Wm; B,D, 40 Wm.

0.1^0.3 mm) in the basal turn where the entire organ of Corti had disappeared. The cochleas (n = 6) of GM+CEP-1347-treated guinea pigs (GM exposure as above plus CEP-1347 treatment covering the lesioning and post-exposure periods) showed a loss of OHCs in the their basal halves, but this loss was less severe than that in the GM-only cochleas (Fig. 3). Total destruction of the organ of Corti was not seen following CEP-1347 treatment. Especially the loss of IHCs (Figs. 3 and 4) and pillar cells (data not shown) was smaller in CEP-1347-treated cochleas than in the GM-only group. HC counts show that total OHC loss ranged from 3300 to 5500, the average being 4480 (66%) (total number of OHCs per cochlea averaged 6800) in GM-onlytreated ears (Fig. 4). Two cochleas had an IHC loss of 1010 and 1210, whereas this loss ranged from 250 to 570 in the other three cochleas. The average IHC loss was 572 (29%) (total number of IHCs per cochlea averaged 2000).

In the CEP-1347-treated group, total OHC loss ranged from 1670 to 3700, the average being 2952 (44%). The average IHC loss was only 73 (range 40^ 90) (3.5%) (Fig. 4). The di¡erence in the numbers of survived OHCs and IHCs between the GM-only group and GM+CEP-1347 group is statistically signi¢cant (P 6 0.05 for OHCs and P 6 0.01 for IHCs). 3.2.2. Functional data of the cochlea Baseline ABRs were comparable in all guinea pigs prior to the ototoxic lesion. As analyzed at day 30 post-exposure, substantial threshold shifts could be seen both in the GM-only group (n = 6 ears) and in the GM+CEP-1347 group (n = 8 ears) (Fig. 5). In the GM-only group, elevation of ABR thresholds ranged from 51 to 86 dB SPL. The ears of the GM+CEP1347 group showed smaller threshold shifts, the average threshold shift being in the range from 6 dB (at 2 kHz) to about 50 dB (at 8^32 kHz). There was a statistically signi¢cant di¡erence in the average threshold elevations

HEARES 3778 3-1-02 Cyaan Magenta Geel Zwart

78

J. Ylikoski et al. / Hearing Research 163 (2002) 71^81

between the two groups at all frequencies tested (at 2.0, 4.0 and 16.0 kHz P 6 0.01; at 8.0 and 32.0 kHz P 6 0.05). 3.3. Dose-dependent e¡ects of CEP-1347 on HC survival in neomycin-challenged cochlear cultures Cochlear explants were prepared from postnatal day 2 mice and coincubated with 100 WM of neomycin and CEP-1347 at concentrations of 100 nM, 500 nM, 1000 nM and 5000 nM. After a 48 h incubation period, the specimens were ¢xed and stained with phalloidin, and the preserved OHCs of the basal turn were counted. As shown in Fig. 6, the maximal protective e¡ect of CEP1347 on OHC survival was seen at the concentration of 1000 nM. Together, these in vivo and in vitro data suggest that CEP-1347, an inhibitor of JNK signalling, attenuates cochlear HC loss and hearing loss following GM intoxication. Next, we studied the e¤ciency of CEP-1347 on ototoxically damaged vestibular HCs in vivo. 3.4. GM-induced vestibular HC damage in vivo and its attenuation by CEP-1347 GM exposure induced moderate damage of vestibular HCs in all three ampullary cristae, whereas the utricular and saccular maculae showed a mild sensory cell loss. Therefore, quantitative analyses were performed on the ampullary sensory epithelium. In ampullary cristae of GM-treated guinea pigs, most type I HCs were lost (Fig. 7A,B). Consistent with this ¢nding, it has been shown that type I HCs are more sensitive to aminoglycoside ototoxicity than type II HCs, and HCs in general are more sensitive than supporting cells (Lindeman, 1969). In addition, it has been shown that central regions of the cristae are most sensitive, with little difference in sensitivity between the canals (Lindeman, 1969). In ampullary cristae of GM+CEP-1347-treated guinea pigs, the numbers of preserved HCs were greater than those in the GM-only group (Fig. 7C,D). Particularly type I HCs showed good preservation. Based on quantitative analyses, in GM-treated cristae (n = 6), the average densities of all preserved HCs (54.5/ mm) and particularly the densities of type I HCs (8.3/ mm) were signi¢cantly lower than in GM +CEP-1347treated cristae (n = 6) (77.1/mm and 24.2/mm, respectively) (Fig. 8). 4. Discussion Earlier studies have suggested that HCs die by apoptosis following aminoglycoside ototoxicity both in the cochlea (Forge, 1985; Nagakawa et al., 1998; Vago et

Fig. 8. HC densities in traumatized ampullary cristae of the guinea pig in vivo. In GM-treated (genta) cristae, the average ( þ S.E.M.) densities of all preserved HCs (54.5 cells/mm) are lower than in GM+CEP-1347-treated cristae (77.1 cells/mm). Densities of type 1 HCs are signi¢cantly lower in GM-treated than in GM+CEP-1347treated cristae (8.3 cells/mm and 24.2 cells/mm, respectively).

al., 1998) and in the vestibular organs (Li et al., 1995; Nagakawa et al., 1997 ; Forge and Li, 2000). It has been speculated that oxidative stress is involved in HC death (Schacht, 1993). We have been focusing on an intracellular signalling cascade, the JNK cascade, which has been shown to couple cellular stress to apoptosis (reviewed by Ip and Davis, 1998 ; Mielke and Herdegen, 2000). The JNK pathway is activated in response to various environmental stressors and this activation leads to cell death, as shown in various cell types (Ip and Davis, 1998 ; Mielke and Herdegen, 2000). Although JNK signalling may also be involved in cellular di¡erentiation, regeneration and repair, there is increasing evidence that one of its major roles is to act as a mediator of apoptosis (Ip and Davis, 1998; Mielke and Herdegen, 2000). In the current study, we show in vivo that the ototoxic action of aminoglycosides on the inner ear HCs of guinea pigs leads to JNK activation and apoptosis of vestibular HCs. The GM dosage used in the present study induced an extensive loss of cochlear HCs and a moderate loss of HCs in the vestibular cristae. These losses were attenuated by treatment with CEP-1347, a non-protein compound that is a speci¢c inhibitor of JNK signalling (Maroney et al., 1998). These ¢ndings are in accordance with our earlier in vitro observations showing that the JNK cascade is activated in neomycinexposed cochlear HCs. In addition, we have previously shown that CEP-1347 protects cultured auditory HCs

HEARES 3778 3-1-02 Cyaan Magenta Geel Zwart

J. Ylikoski et al. / Hearing Research 163 (2002) 71^81

from neomycin damage and that it also attenuates noise-induced HC loss and hearing loss in vivo (Pirvola et al., 2000). In this study, we demonstrate that, in organotypic cultures, CEP-1347 protects OHCs in a dose-dependent manner, so that OHC survival increases from 70% at a concentration of 100 nM to 95% at a concentration of 1000 nM. CEP-1347 has no protective e¡ect at a concentration of 5000 nM. Together, JNK activation appears to be a response by which HCs react to di¡erent inner ear speci¢c stresses in vivo and in vitro. Our data show that both cochlear and vestibular HCs degenerate through apoptosis following ototoxic stress. Cochlear HC apoptosis has been reported also after acoustic overstimulation (Hu et al., 2000; Pirvola et al., 2000) and with ageing (Usami et al., 1997). Thus apoptosis may be a predominant mode of death of HCs in response to noxious stimuli and aging. This may be linked to the defence mechanism by which inner ear sensory organs try to preserve tissue integrity and permeability barriers and avoid additional damage, which would result from potassium-rich endolymph entering into the sensory epithelia (Li et al., 1995). It has been suggested that dying HCs are eliminated from the sensory epithelia through two routes: they can be extruded into the endolymph or translocated basally and phagocytosed within the epithelium (Li et al., 1995). Preservation of tissue integrity and phagocytosis of dying cells without in£ammation are characteristic features of apoptotic cell death (Kerr et al., 1972; Wyllie et al., 1980). In accordance, in the specimens of the present study, we frequently observed apoptotic HCs, which had been translocated to the basal parts of the sensory epithelia towards the level of supporting cell nuclei. Aminoglycosides might stress HCs by changing the microhomeostasis of the inner ear. It is not known how a cell senses stress. The response of an individual cell to exogenous stress is suggested to occur at three levels : cell membrane, cytoplasm and nucleus (Ronai, 1999). It is possible that cytoplasmic mechanisms predominate in stressed HCs. Then, ROS-regulated molecules that alter the cytoplasmic redox potential might be upstream regulators of JNK activation (reviewed by Adler et al., 1999). These molecules might activate cascades of stress-inducible kinases and lead to transcriptional activation of nuclear transcription factors, such as c-Jun. The cellular outcome is likely to depend on the type and dose of stress. The morphometric part of the present study shows that our dosing schedule of GM induced severe HC death both in the cochleas and in the vestibular organs. How large a proportion of HC death was associated with apoptosis (vs. necrosis) is not known. A recent in vitro study showed that apoptotic HC death predominated in GM-exposed utricular sensory epithelia (Forge and Li, 2000). In vivo, however,

79

quantitative evaluation of apoptosis in a chronic experiment, such as the present study, is di¤cult because HC death is progressive and apoptotic cell appearance is limited to only a few hours before they are phagocytosed (Wyllie et al., 1980 ; Kerr et al., 1987 ; Brusch et al., 1990). Cytocochleograms of the present study indicate that inhibition of JNK activation by CEP-1347 attenuates HC loss both in the cochlea and in the vestibular organs. We selected the dosing paradigm on the basis of an earlier study in which GM, 120 mg/kg, was used for 19 days (Song et al., 1997). Based on HC counts and functional data, cochlear and vestibular lesions were only moderate in that study. Surprisingly, in our experiments, 120 mg/kg of GM for 14 days led to a severe functional lesion and widespread loss of both cochlear and vestibular HCs. In the cochleas of our experiments, the main protection was seen on the IHCs of which only 71% were preserved in GM-only-exposed cochleas, whereas an average IHC preservation of 97% was achieved in GM+CEP-1347-treated cochleas. Although OHC damage was signi¢cantly less in CEP-1347-treated than in GM-only-exposed cochleas, protection of OHCs was not as prominent as in the case of IHCs. IHCs are known to be more resistant to ototoxic damage than OHCs (Ylikoski, 1975). Both IHCs and OHCs have been shown to undergo apoptotic death after aminoglycoside exposure in vivo (Nagakawa et al., 1998). It is possible that by using lower dosages of GM, a better protection of OHCs by CEP-1347 could have been achieved. The ABR results of the present study were in correlation with anatomical results. Thus, GM-exposed cochleas showed an average threshold shift ranging from about 60 to 80 dB SPL, which can be explained by an almost 80% loss of OHCs and substantial loss of IHCs as well. CEP-1347-treated cochleas showed an average threshold shift ranging from about 5 to 50 dB, which corresponds to a reduction of auditory sensitivity when the IHCs are still functional, but the major part of OHCs are destroyed (Ylikoski, 1975). The best protection by CEP-1347 was seen at low frequencies, which corresponds to the preservation of signi¢cantly more OHCs (about 1500) in upper turns of the CEP-1347-treated cochleas as compared to GM-exposed cochleas. A central role for the generation of ROS and for changes in the cochlea's antioxidant defense system has been implicated in ototoxic drug-induced HC damage (reviewed by Kopke et al., 1999). Sequential reduction of oxygen along the univalent pathway generates superoxide anion, hydrogen peroxide, hydroxyl radical and water. A large body of in vitro and in vivo evidence indicates that these partially reduced oxygen metabolites are important mediators of aminoglycoside ototox-

HEARES 3778 3-1-02 Cyaan Magenta Geel Zwart

80

J. Ylikoski et al. / Hearing Research 163 (2002) 71^81

icity (Sha and Schacht, 1999; Forge and Schacht, 2000). In other tissues, ROS have been shown to activate JNKs (reviewed by Finkel, 1998), and JNK activation has been shown to be inhibited by pretreatment of cells with antioxidants (Lo et al., 1996). Further, the JNK pathway has been shown to mediate apoptotic cell death in response to free radicals generated by UVand Q-radiation (Derijard et al., 1994), and activation of the JNK pathway precedes the apoptotic process (Derijard et al., 1994; Kyriakis et al., 1994). In addition, glutathione S-transferase Pi (GSTp) has been identi¢ed as an endogenous inhibitor of JNK. GSTp was found to associate with JNKs under basal conditions and dissociate following exposure to UV light or hydrogen peroxide with subsequent JNK activation (Adler et al., 1999). Thus, the protective e¡ects of antioxidants and free radical scavengers on inner ear HC damage can be considered indirect evidence of involvement of the JNK pathway in HC trauma. This mechanism might also explain the protective e¡ects of neurotrophins on HC damage, reported both in the cochlea (Ruan et al., 1999) and in the vestibular organs (Lopez et al., 1999), because neurotrophins are known to have antioxidative properties (Dugan et al., 1997). Furthermore, heat stress, another stressful condition, was recently reported to induce 100^200-fold upregulation of mRNA of the heat shock protein Hsp72 in the inner ear, with simultaneous protection of mice from acoustic injury (Yoshida et al., 1999). Thus, upregulation of Hsp72 might increase the tolerance of cochlear cells to stress. This might occur through suppression of the JNK pathway by Hsp72, since in other cells, Hsp72 interferes with JNK activation and prevents apoptotic signalling in response to stress such as heat shock and ethanol (Meriin et al., 1998). Together, we propose a model in which aminoglycoside exposure leads to stress in the inner ear HCs by ^ possibly generating ROS and ^ activating the JNK pathway and apoptosis. Together with our earlier results (Pirvola et al., 2000), we suggest that JNK activation could be one of the major intracellular cascades by which HCs respond to changes in the homeostasis of their environment following exposure to ototoxic drugs or noise ^ a similar response might also occur in conjunction with presbyacusis. Recent evidence indicates that this might be a mechanism by which many di¡erent cell types respond to stressful environmental stimuli. The inner ear sensory compartments might be useful model systems ^ as organs which can be accurately monitored both functionally and morphologically ^ for studies on insult-induced cellular death. In conclusion, the present results show that the JNK pathway is activated in conjunction with aminoglycoside antibiotic-induced HC death in vivo and that CEP-1347, an inhibitor of JNK signalling, attenuates cochlear and vestibular HC damage.

Acknowledgements We are indebted to Dr. Anne Camoratto for discussions and comments on the manuscript and to Maria von Numers for technical assistance. This work was supported by a grant from the Sigrid Juse¨lius Foundation. References Adler, V., Yin, Z., Tew, K.D., Ronai, Z., 1999. Role of redox potential and reactive oxygen species in stress signalling. Oncogene 18, 6104^6111. Basile, A.S., Huang, J.-M., Xie, C., Webstre, D., Berlin, C., Skolnick, P., 1996. N-Methyl-D-aspartate antagonists limit aminoglycoside antibiotic-induced hearing loss. Nature Med. 2, 1338^1343. Brusch, W., Kleine, L., Tenniswood, M., 1990. The biochemistry of cell death by apoptosis. Biochem. Cell. Biol. 88, 1071^1074. Clerici, W.J., Hensley, K., DiMartino, D.L., Butter¢eld, D.A., 1996. Direct detection of ototoxicant-induced reactive oxygen species generation in cochlear explants. Hear. Res. 98, 116^124. Derijard, B., Hibi, M., Wu, I.H., Barrett, T., Su, B., Deng, T., Karin, M., Davis, R.J., 1994. JNK1: a protein kinase stimulated by UV light and Ha-Ras that binds and phosphorylates the c-Jun activation domain. Cell 76, 1025^1037. Dugan, L.L., Creedon, D.J., Johnson, E.M.J., Holtzman, D.M., 1997. Rapid suppression of free radical formation by NGF involves the mitogen-activated protein kinase pathway. Proc. Natl. Acad. Sci. USA 94, 4086^4091. Finkel, T., 1998. Oxygen radicals and signaling. Curr. Opin. Cell Biol. 10, 248^253. Forge, A., 1985. Outer hair cell loss and supporting cell expansion following chronic gentamicin treatment. Hear. Res. 19, 171^182. Forge, A., Li, L., 2000. Apoptotic death of hair cells in mammalian vestibular sensory epithelia. Hear. Res. 139, 97^111. Forge, A., Schacht, J., 2000. Aminoglycoside antibiotics. Audiol. Neuro-Otol. 5, 3^22. Garetz, S.L., Altschuler, R.A., Schacht, J., 1994. Attenuation of gentamicin ototoxicity by glutathione in the guinea pig in vivo. Hear. Res. 77, 81^87. Hirose, K., Hockenberry, D.N., Rubel, E.W., 1997. Reactive oxygen species in chick hair cells after gentamicin exposure in vitro. Hear. Res. 104, 1^14. Hu, B.H., Guo, W., Wang, P.Y., Henderson, D., Jiang, S.C., 2000. Intense noise-induced apoptosis in hair cells of guinea pig cochlea. Acta Otolaryngol. 120, 9^24. Huang, M.Y., Schacht, J., 1990. Formation of a cytotoxic metabolite from gentamicin by liver. Biochem. Pharmacol. 40, R11^R14. Hulka, G.F., Prazma, J., Brownlee, R.E., Pulver, S., Pillsbury, H.C., 1993. Use of poly-L-aspartic acid to inhibit aminoglycoside cochlear ototoxicity. Am. J. Otol. 14, 352^356. Ip, Y.T., Davis, R.J., 1998. Signal transduction by the c-Jun N-terminal kinase (JNK) ^ from in£ammation to development. Curr. Opin. Cell Biol. 10, 205^219. Kerr, J.F.R., Wyllie, A.H., Currie, A.R., 1972. Apoptosis: A basic biological phenomenon with wide-ranging implications in tissue kinetics. Br. J. Cancer 26, 239^257. Kerr, J.R., Searle, J., Harmon, B.V., Bishop, C.J., 1987. Apoptosis. In: Potten, C.S. (Ed.), Perspectives on Mammalian Cell Death. Oxford University Press, Oxford, pp. 93^128. Kopke, R., Allen, K.A., Henderson, D., Ho¡er, M., Frenz, D., VandeWater, T., 1999. A radical demise; Toxins and trauma share common pathways in hair cell death. Ann. NY Acad. Sci. 884, 171^191.

HEARES 3778 3-1-02 Cyaan Magenta Geel Zwart

J. Ylikoski et al. / Hearing Research 163 (2002) 71^81 Kyriakis, J.M., Banerjee, P., Nikolakaki, E., Dai, T., Rubie, E.A., Ahmad, M., Avruch, J., Woodgett, R.J., 1994. The stress-activated protein kinase subfamily of c-Jun kinases. Nature 369, 156^160. Li, L., Nevill, G., Forge, A., 1995. Two modes of hair cell loss from the vestibular sensory epithelia of the guinea pig inner ear. J. Comp. Neurol. 355, 405^417. Lindeman, H.H., 1969. Regional di¡erences in sensitivity of the vestibular sensory epithelia to ototoxic antibiotics. Acta Otolaryngol. 67, 177^189. Lo, Y.Y.C., Wong, J.M.S., Cruz, T.F., 1996. Reactive oxygen species mediate cytokine activation of c-Jun NH2-terminal kinases. J. Biol. Chem. 271, 15703^15707. Lopez, I., Honrubia, V., Lee, S.-C., Chung, W.-H., Li, G., Beykirch, K., Micevych, P., 1999. The protective e¡ect of brain-derived neurotrophic factor after gentamicin ototoxicity. Am. J. Otol. 20, 317^324. Maroney, A.C., Glicksman, M.A., Basma, A.N., Walton, K.M., Knight, E., Murphy, C.A., Bartlett, B.A., Finn, J.P., Angeles, T., Matsuda, Y., Ne¡, N.T., Dionne, G.A., 1998. CEP-1347/ KT7515 rescues motoneurons from apoptosis and inhibits a stress activated protein kinase pathway. J. Neurosci. 18, 104^111. Meriin, A.B., Gabai, V.L., Yaglom, J., Shifrin, V.L., Sherman, M.Y., 1998. Proteasome inhibitors activate stress kinases and induce Hsp72. J. Biol. Chem. 273, 6373^6379. Mielke, K., Herdegen, T., 2000. JNK and p38 stresskinases-degenerative e¡ectors of signal-transduction-cascades in the nervous system. Prog. Neurobiol. 61, 45^60. Nagakawa, Y., Yamane, H., Shibata, S., Nakai, Y., 1997. Gentamicin ototoxicity induced apoptosis of the vestibular hair cells of guinea pigs. Eur. Arch. Otorhinolaryngol. 254, 9^14. Nagakawa, Y., Yamane, H., Shibata, S., Sunami, K., Nakai, Y., 1998. Apoptosis of guinea pig cochlear hair cells following chronic aminoglycoside treatment. Eur. Arch. Otorhinolaryngol. 255, 127^131. Pirvola, U., Xing-Qun, L., Virkkala, J., Saarma, M., Camoratto, A.M., Walton, K.M., Ylikoski, J., 2000. Rescue of auditory hair cells and neurons by CEP-1347/KT515, an inhibitor of c-Jun Nterminal kinase activation. J. Neurosci. 20, 43^50. Portera-Caillau, C., Price, D.L., Martin, L.J., 1997. Excitotoxic neuronal death in the immature brain is an apoptosis-necrosis morphological continuum. J. Comp. Neurol. 378, 70^87. Priuska, E.M., Schacht, J., 1995. Formation of free radicals by gen-

81

tamicin and iron and evidence for an iron-gentamicin complex. Biochem. Pharmacol. 50, 1749^1752. Ra¡, M.C., 1992. Social control of cell survival and cell death. Nature 356, 97^400. Ronai, Z., 1999. Deciphering the mammalian stress response ^ a stressful task. Oncogene 18, 6084^6086. Ruan, R.S., Leong, S.K., Mark, I., Yeoh, K.H., 1999. E¡ects of BDNF and NT-3 on hair cell survival in guinea pig cochlea damaged by kanamycin treatment. NeuroReport 10, 2067^2071. Schacht, J., 1993. Biochemical basis of aminoglycoside ototoxicity. Otolaryngol. Clin. North. Am. 26, 845^856. Sha, S.-H., Schacht, J., 1999. Formation of free radicals by aminoglycoside antibiotics. Hear. Res. 128, 12^118. Song, B.-B., Schacht, J., 1996. Variable e¤cacy of radical scavengers and iron chelators to attenuate gentamicin ototoxicity in guinea pig in vivo. Hear. Res. 94, 87^93. Song, B.-B., Anderson, D.J., Schacht, J., 1997. Protection from gentamicin ototoxicity by iron chelators in guinea pig in vivo. J. Pharmacol. Exp. Ther. 282, 369^377. Usami, S.-i., Takumi, Y., Fujita, S., Shinkawa, H., Hosokawa, M., 1997. Cell death in the inner ear associated with aging is apoptotic? Brain Res. 747, 147^150. Vago, P., Humbert, G., Lenoir, M., 1998. Amikacin intoxication induces apoptosis and cell proliferation in rat organ of Corti. NeuroReport 9, 431^436. Weil, M., Jacobson, M.D., Coles, H.S.R., Davies, T.J., Gardner, R.L., Ra¡, K.D., Ra¡, M.C., 1996. Constitutive expression of the machinery for programmed cell death. J. Cell Biol. 133, 1053^1059. Wyllie, A.H., Kerr, J.F.R., Currie, A.R., 1980. Cell death: The signi¢cance of apoptosis. Int. Rev. Cytol. 68, 251^306. Ylikoski, J., 1975. Correlative studies on the cochlear pathology and hearing loss in guinea pigs after intoxication with ototoxic antibiotics. Acta Otolaryngol. 326 (Suppl.), 7^62. Ylikoski, J., Pirvola, U., Suoqiang, Z., Eriksson, U., Solin, M.-L., Miettinen, A., 1997. Aminoglycoside ototoxicity: High a¤nity receptors are expressed in secretory epithelia. Audit. Neurosci. 3, 415^424. Yoshida, N., Kristiansen, A., Liberman, M.C., 1999. Heat stress and protection from permanent acoustic injury in mice. J. Neurosci. 19, 10116^10124.

HEARES 3778 3-1-02 Cyaan Magenta Geel Zwart