Gene Therapy (2000) 7, 377–383  2000 Macmillan Publishers Ltd All rights reserved 0969-7128/00 $15.00 www.nature.com/gt VIRAL TRANSFER TECHNOLOGY RESEARCH ARTICLE Transduction of the contralateral ear after adenovirus- mediated cochlear gene transfer

T Sto¨ver1,2, M Yagi1,3 and Y Raphael1 1Kresge Hearing Research Institute, Department of Otolaryngology, The University of Michigan Medical School, Ann Arbor, MI, USA; 2Department of Otolaryngology, Medizinische Hochschule Hannover, Hannover, Germany; and 3Department of Otolaryngology, Kansai Medical University, Osaka, Japan

Cochlear gene transfer is a promising new approach for cochleae, whereas virus inoculation into the bloodstream did inner ear therapy. Previous studies have demonstrated hair not. The cochlear aqueduct was identified as the most likely cell protection with cochlear gene transfer not only in the route of virus spread to the contralateral cochlea. These data inoculated, but also in the uninoculated ear. To characterize enhance our understanding of the kinetics of virus-mediated the kinetics of viral spread, we investigated the extent of transgene expression in the inner ear, and assist in the transgene expression in the contralateral (uninoculated) development of clinical applications for inner ear gene ther- cochlea after unilateral adenoviral cochlear gene transfer. apy. Our results showed a functional communication We used a lacZ reporter gene vector, and demonstrated between the CSF and the perilymphatic space of the inner spread of the adenovirus into the cerebrospinal fluid (CSF) ear, that is not only of importance for otological gene trans- after cochlear inoculation of 25 ␮l viral vector. Direct virus fer, but also for CNS gene transfer. Gene Therapy (2000) 7, application into the CSF resulted in transduction of both 377–383.

Keywords: adenovirus; gene transfer; cochlea; guinea pig

Introduction low. In contrast, viral vectors have been shown to be effective vehicles for gene transfer. Several viral vectors A variety of pathological processes may lead to cochlear have been used for experimental inner ear gene transfer.4 and vestibular impairments. Prevention or correction of Among the most promising are the adenovirus,5,6 the such pathological processes is, thus far, clinically restric- adeno-associated virus,7 the herpes simplex virus,8 and ted to the systemic application of substances. Cochlear the retrovirus.9 Cochlear gene transfer for over- gene transfer provides both an opportunity for localized expression of glial cell line-derived neurotrophic factor administration and the potential for molecular-based (GDNF) and brain-derived neurotrophic factor (BDNF) interventions. Such intervention may facilitate over- showed protective effects on hair cells and spiral expression of certain gene products that may influence ganglion cells.10,11 These studies also demonstrated the repair and regeneration in the inner ear epithelia. For potential for cochlear gene therapy for therapeutic instance, manipulating levels of retinoic acid in cultured applications. Kip1 developing cochleae and knocking out the p27 gene Following unilateral adeno-associated virus gene trans- 1–3 have been shown to lead to overproduction of hair cells. fer into guinea pig cochlea, Lalwani et al12,13 observed The former example deals with a secreted (diffusible) transduction of vestibular cells of the uninoculated gene product and can be done in the mature cochlea. The (contralateral) cochlea. They speculated that the viral vec- Kip1 latter example, p27 knock out, can presently be tor was transferred to the uninoculated ear via the blood- accomplished only at the germline intervention level. stream, the bone marrow, or the CSF. Contralateral trans- However, once gene therapy in somatic cells shuts down gene expression following virus-mediated gene transfer the expression of specific genes, similar results can be has not been reported elsewhere. However, a trend for accomplished in a tissue-specific manner. Such inter- the contralateral protective effects of GDNF has been ventions are particularly exciting for clinical applications observed.10 If the contralateral effect is true, protection as they would provide a treatment of inner ear experiments based on a comparison between the inocu- pathologies that currently have no cure. lated and contralateral sides could underestimate the There are several possible ways to deliver transgenes degree of protection, since the contralateral ear might into tissues. Nonviral vectors pose little or no risk of also have been treated and protected. The mechanism for immune response, but their transduction efficiency is the contralateral protection is unclear. Therefore, the first goal of this study was to characterize the contralateral effects after unilateral cochlear gene transfer. Specifically, Correspondence: Y Raphael, MSRB 3, Room 9303, 1150 W. Medical we wished to determine if contralateral transgene Center Drive, Ann Arbor, MI 48109–0648, USA expression can occur after adenovirus inoculation of the Received 29 July 1999; accepted 28 October 1999 guinea pig ear and, if so, to elucidate the route of migration of the viral particles to the contralateral ear. Transduction of the contralateral cochlea T Sto¨ver et al 378 We chose an adenoviral vector system, encoding the Transduced cochlear cells are detectable 6 h after gene reporter gene lacZ, to address these questions. Inner ear transfer inoculation with the adenoviral vector is well estab- Time-course experiments were performed with cochlear lished5,14 and it has been demonstrated to be therapeuti- inoculations of 25 ␮l adenoviral vector into the left coch- cally effective on hair cells.10 As third-generation adeno- lea. Transfected cochlear cells could be detected 6 h after virus vectors are being developed with reduced inoculation. None of the animals killed from 30 min to immunogenicity and increased duration of gene 5.5 h after cochlear gene transfer showed lacZ-positive expression, these vectors are potential candidates for cells. In animals killed 6 h and 10 h after inoculation, cochlear gene transfer in clinical trials. lacZ-positive cells were identified in the scala tympani of The second goal of this study was to determine other both ears. Histological sections obtained from these speci- important features of virus-mediated cochlear gene trans- mens identified lacZ-positive cells predominantly in con- fer in the cochlea. Without doubt, clinical application will nective tissue cells within the spiral ligament, mesothelial require detailed information on the possible side-effects cells beneath the organ of Corti, and fibrocyctes lining of viral vehicles used for cochlear gene transfer. Determi- the perilymphatic fluid spaces (Figure 1D). Furthermore, nation of the transgene distribution after inoculation with in some specimens, we found lacZ-staining in Hensen increased amounts of viral vector is of great interest as cells and pillar cells (Figure 1E) and also in the region of no information is currently available on the maximal vol- Rosenthal’s canal. The transfected cells in Rosenthal’s umes of virus solution applicable to the inner ear as a canal were predominantly Schwann cells, lying between bolus injection. Evaluation of the resulting transgene dis- spiral ganglion cells (Figure 1F). Six (and 10) h after virus tribution and identification of the cell types susceptible inoculation into the scala tympani, we also observed to the viral vector after inoculation with increased vol- staining of the middle ear mucosa in the inoculated side umes of vector have to occur before gene transfer can (data not shown). be practically used in clinical trials. Other issues to be addressed include unintended transfection of nontar- The transduction of the contralateral cochlea is geted tissues with increased virus volume and possible mediated by the cochlear aqueduct changes in vector cell specificity. To determine if the route of transduction for the contrala- We demonstrate that cochlear expression of lacZ was teral cochlear cells was via the blood or CSF mediation, detectable as soon as 6 h after virus inoculation. Contrala- 25 ␮l of adenoviral solution was injected directly into the teral transgene expression was volume dependent. It was jugular vein or into the CSF space of the skull. No lacZ- observed after an inoculation with 25 ␮l, but not after a positive cells were found on the brain surface or in coch- 5 ␮l inoculation into the cochlea. No systemic transgene lear tissue of any systemically inoculated animal (n = 4). expression was found after cochlear inoculation even In contrast, lacZ-positive cells were identified in both with the increased volume of the viral vector, whereas cochleae, mostly in the basal and the second cochlear expression in liver and spleen was observed after inocu- turns, and on the brain surface in all animals inoculated lation into the bloodstream and the CSF. Inoculation into via the CSF (n = 6). Careful inspection of the cochlear the cranial CSF resulted in detection of lacZ-positive cells tissue revealed a predominant cluster region of lacZ-posi- in the cochlear opening of the cochlear aqueduct bilater- tive cells close to the ‘hook’ region of the cochlea ally, implicating the CSF aqueduct as the route for ear to (Figure 3A and B). This region represents anatomically ear vector transfer. Finally, application of the viral vector the opening of the cochlear aqueduct that connects the into the lumbar region of the CSF space was also ident- CSF space with the perilymphatic space of the scala tym- ified as a potential route for delivering vectors into the pani. While 5 ␮l of the viral vector applied to the cochlea cochlea. were not sufficient to induce a contralateral gene transfer effect, 25 ␮l of the virus solution applied to the cochlea or the CSF space induced a response. Results To investigate further viral passage through the coch- lear aqueduct in guinea pigs, we injected 200 ␮l of the adenoviral vector into the CSF space of the lumbar region Cochlear inoculation of increased vector volume results via a lumbar puncture (n = 2). We found that one of the in transduction of the contralateral ear animals showed lacZ-positive cells in both cochleae. In Cochlear gene transfer via the cochleostomy with 5 ␮lof this animal, the brain surface was also positively stained the adenoviral vector, carrying the reporter gene lacZ, for lacZ. The spinal cord was lacZ-positive in both resulted in the transduction of cochlear cells only in the animals. inoculated left ear (n = 6). Transfected cells were localized in the scala tympani of the basal and second turn. After Systemic expression of lacZ-positive cells after cochlear inoculation with 25 ␮l(n = 6) of adenoviral sol- cochlear, blood, or CSF inoculation ution, lacZ-positive cells were observed in both the inocu- Following cochlear inoculation with 5 ␮l of vector sol- lated (Figure 1A) and the uninoculated (contralateral) ution (n = 6), we did not detect any systemic expression cochleae (Figure 1B and C). Transduced cells in the unin- of lacZ in any organ (liver, spleen, lung, kidney or brain) oculated contralateral (right) ear were localized predomi- other than the inoculated ear. After injection of 25 ␮lof nantly in the scala tympani, at the opening of the per- vector solution into the cochlea (n = 6), we did not find ilymphatic aqueduct. LacZ-positive cells were also found expression of lacZ outside the above described basal on the basal surface of the brain following a 25 ␮l coch- brain region and the contralateral ear. Middle ear stain- lear inoculation (Figure 2A). No transduction of the brain ing was observed with 5 ␮l and 25 ␮l inoculation of the surface or the contralateral cochlea was detected follow- cochlea only at the site of inoculation. Virus application ing a 5 ␮l viral inoculation (n = 6) (Table 1). directly to the CSF (25 ␮l), however, led to the systemic

Gene Therapy Transduction of the contralateral cochlea T Sto¨ver et al 379

Figure 1 (A) A stereo-micrograph of the left cochlea after cochlear inoculation of the left ear with 25 ␮L adenoviral vector. After X-gal staining, transduced (blue) cells are located in first and second turns of the scala tympani (bar: 1 mm). (B) A stereo-micrograph of the right cochlea after cochlear inoculation of the contralateral (left) ear with 25 ␮l adenoviral vector. After X-gal staining, transduced cells are located predominately in the scala tympani around the ‘hook’ region of the basal turn of the cochlea. The opening of the cochlear aqueduct into the scala tympani is indicated with a hair (arrow) that has been inserted into the duct lumen (bar: 1 mm). C: A stereo-micrograph of the same specimen as (B) (higher magnification) showing the area with the highest density of lacZ-positive cells at the opening of the cochlear aqueduct into the scala tympani. A hair (arrow) has been placed inside the cochlear aqueduct to indicate the location of its opening into the scala tympani (bar: 1.5 mm). D: A lightmicroscopic (LM) section of the organ of Corti after 25 ␮l adenovirus inoculation into the cochlea. Predominantly mesothelial cells (m) beneath the organ of Corti and the fibrocyctes (f) lining the perilymphatic fluid spaces are transduced by the adenoviral vector. The framed box indicates the area of (E) (bar: 100 ␮m). (E) A LM section from the same region as shown in (D) at higher magnification. ␤-galactosidase could be detected in Hensen cells (h) and pillar cells (p) of the organ of Corti (bar: 20 ␮m). (F) A LM section of a left cochlea (Rosenthal’s canal) after cochlear inoculation with 25 ␮l adenoviral vector. LacZ-positive cells (stained blue) are located between spiral ganglion cells and identified as Schwann cells (bar: 20 ␮m).

Gene Therapy Transduction of the contralateral cochlea T Sto¨ver et al 380 Table 1 Localization of lacZ-positive cells according to inoculation side and volume of adenoviral vector used

Localization of lacZ- Adenovirus inoculation volume and site positive cells 5 ␮l25␮l25␮l25␮l CSF cochlea cochlea blood (n = 6) (n = 6) (n = 6) (n = 4)

Inoculated cochlea yes yes no yes Non-inoculated cochlea no yes no yes Middle ear yes yes no no Brain surface no yes no yes Liver no no yes yes Spleen no no yes yes Lung no no no no Kidney no no no no

expression of lacZ in both the liver and spleen (n = 6) but not in the middle ear space or in lung or kidney tissue. Virus application (25 ␮l) directly into the bloodstream via the jugular vein resulted in even stronger expression of lacZ in the liver (Figure 2B and C) and the spleen (Figure 2D and E), whereas no expression was found in the cochlea, middle ear space, cells of the brain surface, the lung, or the kidney (n = 4). These results are summar- ized in Table 1.

Discussion Following inoculation of the left cochleae with the adeno- viral vector, contralateral transfection of cochlear cells was found. The contralateral transfection appeared to be volume dependent, since transfection of the right (contralateral) ear was restricted to experiments in which 25 ␮l of vector solution was inoculated into the left coch-

Figure 2 (A) A stereo-micrograph of the basal brain surface after cochlear inoculation with 25 ␮l adenoviral vector and X-gal staining. LacZ-posi- tive cells (blue stained) are located on the basal brain surface and on blood vessels. Stereo-optical inspection clearly showed blue-colored particles spread along the outside of the blood vessel walls (arrow) (bar: 1.5 mm). (B) A stereo-micrograph of X-gal stained liver of an animal intravenously inoculated with 25 ␮l adenoviral vector. LacZ-positive cells (blue stained) are uniformly distributed throughout the liver (bar: 0.5 mm). (C) A LM section of the liver of an animal intravenously inoculated with 25 ␮l aden- Figure 3 (A) A stereo-micrograph of a left cochlea after CSF inoculation oviral vector. LacZ-positive cells were identified as hepatocytes (arrows) with 25 ␮l adenoviral vector and X-gal staining. Note that the lacZ-posi- (bar: 20 ␮m). (D) A stereo-micrograph of X-gal stained spleen of an ani- tive cells are located predominantly (dark blue staining) around the ‘hook’ mal intravenously inoculated with 25 ␮l adenoviral vector. LacZ-positive region in the scala tymani of the cochlea. This region represents the open- cells (blue stained) are distributed throughout the organ (bar: 0.5 mm). ing of the cochlear aqueduct into the scala tympani. Blue cells are also (E) A LM section of the spleen of an animal intravenously inoculated found in the second turn of the cochlea (bar: 1 mm). (B) A stereo-micro- with 25 ␮l adenoviral vector. LacZ-positive cells appear like reticular cells graph of the right cochlea of the same animal as shown in (A). After CSF in the spleen cords (arrows) (bar: 20 ␮m). inoculation with 25 ␮l adenoviral vector and X-gal staining, lacZ-positive cells are located predominantly around the ‘hook’ region of the basal turn (bar: 1 mm).

Gene Therapy Transduction of the contralateral cochlea T Sto¨ver et al 381 lea. Contralateral transfection was not observed with a small volume (5 ␮l) adenovirus inoculation. A bilateral lack of transfected cells following a systemic intravenous inoculation ruled out the blood-borne transmission of vector for intercochlear spread of virus. Inoculation of the viral vector into the CSF led to a bilateral cochlear trans- fection, manifested mainly in the mesothelial cells in the scala tympani near the cochlear aqueduct opening. LacZ- positive cells were also found in the surface of the brain following inoculation with 25 ␮l of viral vector into the cochlea or the CSF. These results provide strong evidence that the cochlear aqueduct is the mediating route for intercochlear communication leading to contralateral vector spread. Our data conclusively demonstrate contralateral trans- fection of cochlear cells with adenovirus vectors as early as 6 h after inoculation. Transfection of contralateral cells might have occurred earlier, considering that the expression of ␤-galactosidase may require several hours to manifest. Contralateral transfection has previously been observed with a smaller viral vector, the adeno- associated virus.12,13 These authors speculated that mediation of the vector to the uninoculated ear was via the blood, bone marrow, or CSF. As the perilymphatic fluids are not compressible and perilymph leakage was not observed during our surgical procedures, fluid dis- placement inside the cochlea had to take place during the application of additional volume into the perilymphatic space. This is even more obvious, considering that the viral volume (25 ␮l) applied to the perilymphatic space exceeded the volume of the perilymphatic space (16 ␮l).15 We speculate that fluid application to the cochlea results Figure 4 (A) A schematic of the ear anatomy, emphasizing the connection in a displacement of perilymphatic fluid via the cochlear between the perilymphatic space of the cochlea (scala tympani) and the aqueduct into the CSF. Our findings support this hypoth- CSF space. (B) The same schematic as (A), with red arrows indicating esis on four different points. First, we demonstrated that the route of vector spread after inoculation of 25 ␮l viral vector into the viral vector inoculation into one cochlea leads to the scala tympani of the cochlea. Virus spread is from the perilymph of the detection of cells expressing the transgene in the contrala- scala tympani via the cochlear aqueduct to the CSF, and further via the contralateral cochlear aqueduct to the opposite cochlea. (C) The red arrows teral cochlea. Second, in these animals we found the cells indicate the route of vector spread after direct inoculation of 25 ␮L viral on the brain surface consistently transduced, showing the solution into the CSF. The viral vector is distributed within the CSF and presence of vector in the CSF space, which is continuous reaches the perilymphatic space of both cochleae via the cochlear aqueducts. with the cochlear aqueduct. Third, when vector was applied directly to the CSF, it not only transduced the brain surface, it also spread into the scala tympani of both assessing protective effects (following bilateral insults) by cochleae. Fourth, after CSF inoculation, transduced cells comparing the inoculated with the uninoculated ear were located predominantly around the opening of the would underestimate the degree of protection. The impli- cochlear aqueduct near the ‘hook’ region of the scala tym- cation of the contralateral transfection for gene therapy, pani of both cochleae. This distribution pattern was as a consequence of exceeding a certain inoculation vol- almost symmetrical and identical to the distribution ume, is spread of the viral vector or diffusion of secreted found in the uninoculated (right) ear after cochlear inocu- proteins beyond the inoculated cochlea. Therefore, lation with 25 ␮l of the vector. Finally, we ruled out localized cochlear gene transfer can be achieved only if blood-borne transduction of cochlear cells as the cause of inoculation is performed with relatively low volumes of contralateral gene transfer as no cochlear gene transfer viral vector. On the positive side, our findings imply that was observed after direct inoculation of the vector into the cochlear aqueduct might be used as a route for gene the bloodstream. These data strongly suggest that the transfer of the cochlea. Our results after inoculation of cochlear aqueduct and the CSF facilitate inter-aural the vector directly into the CSF demonstrate that, in prin- vector spread (see Figure 4 for a schematic). ciple, gene transfer in the cochlea can be accomplished These findings are of practical importance for the clini- without directly manipulating the cochlea. In the present cal application of gene therapy in the inner ear. The con- animal experiments, we have shown that even after nection between both cochleae may account for several application of the vector to the CSF via a lumbar punc- unexplained contralateral gene transfer effects after uni- ture, transfected cells were observed in both cochleae. lateral inoculation. An adenovirus vector encoding the These results might lead to a new approach for inner ear human GDNF gene was shown to protect the inoculated gene transfer or drug application, both from the CSF to (left) ears from ototoxic drugs and, in addition, to afford the inner ear and from the inner ear to the CSF. some protection to the contralateral (uninoculated) ear.10 The cochlear aqueduct as an accessible gate to the Due to the contralateral protection, it is likely that cochlea via the CSF has implications beyond the area of

Gene Therapy Transduction of the contralateral cochlea T Sto¨ver et al 382 gene transfer. Bilateral deafness after meningitis, parti- middle ear mucosa, as has been demonstrated by cularly in young children with bacterial meningitis or Mondain et al.22 viral mumps meningitis, might progress via the cochlear In conclusion, we detected contralateral cochlear trans- aqueduct to the cochlear space.16 Clinically, it has been gene expression. The contralateral spread was dependent found that an elevated CSF pressure is a predicting factor on the volume of inoculated vector solution. Virus inocu- for the occurrence of deafness after bacterial meningitis.16 lation into the CSF resulted in bilateral cochlear transduc- Since we have demonstrated the functional connection tion, whereas systemic virus inoculation did not. The between the cochleae and the CSF space via the cochlear cochlear aqueduct was identified as the most likely route aqueduct, pathological processes occurring at the CSF of virus spread to the contralateral cochlea. These data level, eg pneumococcal meningitis,17 might be considered enhance our understanding of the kinetics of virus- as an explanation for the bilateral effects on the mediated transgene expression in the inner ear, and assist cochlear tissue. in the development of clinical applications for inner ear Localization of lacZ-positive cells in the cochlear tissue gene therapy. Furthermore, our data provide evidence after inoculation of 25 ␮l of the adenoviral vector showed for a close functional communication between the CSF transduction predominantly of mesothelial cells beneath and the perilymphatic space of the inner ear, a finding the organ of Corti and fibrocyctes lining the perilym- that is of importance for otological gene transfer as well phatic fluid spaces (Figure 1D). X-gal staining was also as CNS gene therapy. found in Hensen cells, pillar cells (Figure 1E), and Schwann cells in Rosenthal’s canal (Figure 1F). Thus, Materials and methods comparison between the types of cells transduced with a 5 ␮lor25␮l vector revealed a higher quantity and larger spread of the transfection with the higher volume of vec- Adenoviral vector tor. In addition, the higher volume of vector inoculation To perform cochlear gene transfer experiments, we used resulted in transduction of two types of supporting cells a replication-deficient adenoviral vector (E1A/B and part in the organ of Corti, which did not express the reporter of E3 deleted) based on the human adenoviral backbone transgene following the lower volume of inoculation. The (serotype 5). The vector carried the reporter gene lacZ CMVlacZ E. coli ␤ transduction of supporting cells is important and exciting (Ad. ), encoding -galactosidase, driven by the cytomegalovirus promoter. The viral vector was because of their key role for repair and regeneration,18–20 constructed as has been described previously23 and presenting these cells as a main target for gene transfer obtained from the University of Michigan Vector Core. in the inner ear. The vector was stored at −80°C until use. It was diluted Systemic expression of lacZ in the liver and spleen was with Ringers’ solution (145 mm NaCl, 2.7 mm KCl, 2 mm found after CSF and intravenous inoculation of the ␮ MgSO4, 1.2 mm CaCl2,5mm Hepes) to a final concen- adenovirus (25 l), but not after cochlear inoculation with 10 ␮ tration of 10 plaque-forming units (p.f.u.)/ml immedi- the same increased volume of virus solution (25 l), indi- ately before inoculation. In experiments comparing virus cating less vector being distributed into the CSF after inoculation with 5 ␮l and 25 ␮l, the same batch of cochlear inoculation compared with direct CSF inocu- adenoviral vector was used. lation. No lacZ expression was found in lung or kidney tissue. Cochlear inoculations with lower amounts of the Animal procedures and gene transfer adenoviral vector (5 ␮l), exclusively and consistently All animal experiments were approved by the University showed that transduction was restricted to the inoculated Committee on the Use and Care of Animals at the Uni- (left) ear. Therefore, the data demonstrate that small vol- ␮ versity of Michigan and were performed using accepted umes (5 l) of vector can be safely applied for cochlear veterinary standards. Guinea pigs of either sex (Murphy gene therapy (as a bolus injection), without risking vector Breeding Laboratory, Plainfield, IN, USA), weighing 237– spread outside the inoculated cochlea. 387 g, were used. Viral administration into the cochlea In our experiments, transfected cells were also found was via a cochleostomy, as previously described.24 The in the middle ear space under some experimental animals were inoculated with either 5 ␮l(n = 6) or 25 ␮l approaches. This occurred only in animals after direct (n = 6) adenoviral solution into the scala tympani of the cochlear inoculation, but not after blood or CSF inocu- left cochlea. A movement of the round window mem- lation. No apparent difference was noted in the amount brane could be observed during inoculation of 25 ␮l per- ␮ ␮ of middle ear staining after either 5 lor25 l vector formed over approximately 3 min. This movement was inoculation. The perilymphatic flow rate has been shown used to monitor the injection velocity. Application of the to increase by about two orders of magnitude following viral vector into the bloodstream (n = 4) was performed an experimentally induced fistula.21 Therefore, an with a 25 ␮l virus solution injected directly into the right increased perilymph flow as a result of the inoculation jugular vein. Virus application into the CSF was perfor- may increase the risk of viral spread into the middle ear. med with 25 ␮l virus solution injected into the subdural Perilymph leakage may be due to an incomplete seal at space (n = 6). the injection site. Even though a leak was undetected dur- For adenovirus application to the CSF, animals were ing the procedure, the presence of transfected middle ear anesthetized (xylazine 10 mg/kg i.m. and ketamine mucosal cells suggests that a leak probably occurred. 40 mg/kg i.m.), and a 3 × 3 mm piece of skull bone, near Unwanted middle ear transfection involves the potential to the bregma, was carefully removed with a diamond risk of vector spreading via the Eustachian tube to the burr. The dura was incised and a 5 cm piece of vinyl tub- airways and to the digestive tract. However, transfection ing V/4 (Scientific Commodities Incorporation, Lake of middle ear cells also presents therapeutic opport- Havasu City, AZ, USA) was inserted about 4 mm into the unities that may be harnessed for interventions in the subdural space. To prevent CSF leakage, the cannula was

Gene Therapy Transduction of the contralateral cochlea T Sto¨ver et al 383 secured to the skull with carboxylate cement (Durelon, 2 Chen P, Segil N. P27Kip1 links cell proliferation to morphogen- ESPE, Germany). After the cement dried (approximately esis in the developing organ of Corti. Development 1999; 126: 10 min), the viral solution was injected into the CSF. The 1581–1590. Kip1 cannula was left in place for 10 min and then withdrawn. 3Lo¨wenheim H et al. Gene disruption of p27 allows cell pro- liferation in the postnatal and adult organ of corti. Proc Natl The skull opening was sealed with bone wax (Ethicon, Acad Sci USA 1999; 96: 4084–4088. Summerville, NJ, USA) and the skin sutured in two lay- 4 Raphael Y, Yagi M. Gene transfer and the inner ear. Curr Opin ers. Alternatively, a lumbar puncture using a 25-gauge Otolaryngol Head Neck Surg 1998; 6: 311–315. needle (n = 2) was performed to inject 200 ␮l of virus sol- 5 Raphael Y, Frisancho JC, Roessler BJ. Adenoviral-mediated gene ution into the CSF space. Animals were carefully moni- transfer into guinea pig cochlear cells in vivo. Neurosci Lett 1996; tored postsurgically for discomfort, head-tilt, feeding, 207: 137–141. and general activity level. Recovery was uneventful in all 6 Weiss MA, Frisancho JC, Roessler BJ, Raphael Y. Viral-mediated animals. The animals were decapitated 5 days after vec- gene transfer in the cochlea. Int J Dev Neurosci 1997; 15: 577–583. tor inoculation and temporal bones, lung, spleen, liver, 7 Lalwani AK et al. Development of in vivo gene therapy for hear- kidneys, and brain tissue were removed. ing disorders: introduction of adeno-associated virus into the cochlea of the guinea pig. Gene Therapy 1996; 3: 588–592. 8 Geschwind MD et al. Defective HSV-1 vector expressing BDNF Tissue processing in auditory glia elicits neurite outgrowth: model for treatment Collected tissue samples (temporal bones, brain, lung, of neuron loss following cochlear degeneration. Hum Gene Ther spleen, liver and kidney) were fixed with 4% paraformal- 1996; 7: 173–182. dehyde for 2 h at 4°C and then rinsed twice for 10 min 9 Kiernan AE, Fekete DM. In vivo gene transfer into the embryonic with phosphate-buffered saline (PBS). To detect inner ear using retroviral vectors. Audiol Neurootol 1997; 2: 12–24. expression of the transgenic ␤-galactosidase, the tissues 10 Yagi M et al. Hair cells are protected from aminoglycoside oto- were incubated overnight at 37°C in X-gal solution (5- toxicity by adenoviral-mediated overexpression of GDNF. Hum ␤ Gene Ther 1999; 10: 813–823. bromo-4-chloro-3-indolyl- -d-galactoside) as described 11 Staecker H, Gabaizadeh R, Federoff H, Van De Water TR. Brain- 23 previously. Following a PBS rinse, the tissues were derived neurotrophic factor gene therapy prevents spiral gang- photographed under a dissection microscope (Wild MZ- lion degeneration after hair cell loss. Otolaryngol Head Neck Surg 12). Tissues were sectioned for histological localization of 1998; 119: 7–13. lacZ-positive cells. The cochleae were decalcified in 3% 12 Lalwani AK et al. Green fluorescent protein as a reporter for EDTA with 0.25% glutaraldehyde for about 10 days. gene transfer studies in the cochlea. Hear Res 1997; 114: 139–147. Tissues were then dehydrated in an ethanol series and 13 Lalwani AK et al. Expression of adeno-associated virus inte- embedded in JB-4 media (Electron Microscopy Sciences, grated transgene within the mammalian vestibular organs. Am Washington, PA, USA). Sections obtained at the mid- J Otol 1998; 19: 390–395. ␮ 14 Komeda M, Roessler BJ, Raphael Y. The influence of interleukin- modiolar area (4 m thick) were stained with toluidin- 1 receptor antagonist transgene on spiral ganglion neurons. Hear blue (1%) or eosin-red (0.5%) and cover-slipped with Res 1999; 131: 1–10. Permount (Fisher Scientific, Springfield, NJ, USA). 15 Salt AN, Thalmann R. Cochlear fluid dynamics. In: Jahn AF and Santos-Sacchi J (eds). Physiology of the Ear. Raven Press: New Gene transfer time-course York, 1988, pp 341–357. For the time-course experiment, cochlear gene transfer 16 Woolley AL et al. Risk factors for hearing loss from meningitis via a cochleostomy24 was performed with 25 ␮l of the in children: the Children’s Hospital experience. Arch Otolaryngol adenoviral solution in the 17 guinea pigs, as described Head Neck Surg 1999; 125: 509–514. 17 Kesser BW et al. Time course of hearing loss in an animal model above. After vector inoculation, the animals were killed of pneumococcal meningitis. Otolaryngol Head Neck Surg 1999; after 30 min, 1 h, 2 h, 3 h and 3.5 h (one animal each), and 120: 628–637. after 4 h, 4.5 h, 5 h, 5.5 h, 6 h, and 10 h (two animals each). 18 Forge A. Outer hair cell loss and supporting cell expansion fol- The temporal bones and the brain tissues were collected lowing chronic gentamicin treatment. Hear Res 1985; 19: 171–182. and stained with X-gal solution as described above. 19 Raphael Y, Altschuler RA. Scar formation after drug-induced cochlear insult. Hear Res 1991; 51: 173–184. 20 Leonova EV, Raphael Y. Application of a platinum replica Acknowledgements method to the study of the cytoskeleton of isolated hair cells, supporting cells and whole mounts of the organ of Corti. Hear We thank Drs RP Bobbin and D McCullum for their help- Res 1999; 130: 137–154. ful advice, Brian Shin for technical assistance, and Nadine 21 Salt AN, Inamura N, Thalmann R, Vora AR. Evaluation of pro- Brown for critical comments on the manuscript. This cedures to reduce fluid flow in the fistulized guinea-pig cochlea. work was supported by NIH NIDCD Grant 2 P01 Acta Otolaryngol (Stockh) 1991; 111: 899–907. DC00078 (YR). TS is a scholar of the Alexander von 22 Mondain M et al. Adenovirus-mediated in vivo gene transfer in Humboldt-Foundation. guinea pig middle ear mucosa. Hum Gene Ther 1998; 9: 1217– 1221. 23 Davidson BL et al. A model system for in vivo gene transfer into References the central nervous system using an adenoviral vector. Nat Genet 1993; 3: 219–223. 1 Kelley MW et al. The developing organ of Corti contains retinoic 24 Sto¨ver T, Yagi M, Raphael Y. Cochlear gene transfer: round win- acid and forms supernumerary hair cells in response to exogen- dow versus cochleostomy inoculation. Hear Res 1999; 136: ous retinoic acid in culture. Development 1993; 119: 1041–1053. 124–130.

Gene Therapy