Helper-Dependent Adenovirus-Mediated Gene Transfer Into the Adult Mouse Cochlea

Copyright @ 200 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.7

Helper-Dependent Adenovirus-Mediated GeneTransfer Into the Adult Mouse Cochlea

*Gentiana I. Wenzel, *Anping Xia, *Etai Funk, *M. Bradley Evans,†Donna J. Palmer, †Philip Ng, *‡Fred A. Pereira, and *§John S. Oghalai

*Bobby R. Alford Department of OtolaryngologyYHead and Neck Surgery; ÞDepartment of Molecular andHuman Genetics; and þHuffington Center on Aging and Molecular and Cellular Biology, Baylor College of

Medicine; and §Department of Bioengineering, Rice University, Houston, Texas, U.S.A.

Background: Gene therapy may provide a way to restorecochlear function to deaf patients. The most successful tech-niques for cochlear gene therapy have been injection of early-generation adenoviral vectors into scala media in guinea pigs.However, it is important to be able to perform gene therapyresearch in mice because there is wide availability of trans-genic strains with hereditary hearing loss.Purpose: We demonstrate our technique for delivery of athird-generation adenoviral vector, helper-dependent adeno-virus (HDAd), to the adult mouse cochlea.Methods: Mice were injected with an HDAd that contained areporter gene for either A-galactosidase or green fluorescentprotein into scala media. After 4 days, the cochleae were har-vested for analyses. Auditory brainstem response monitoringof cochlear function was performed before making acochleostomy, after making a cochleostomy, and before killingthe animal.

Results: A-Galactosidase was identified in the spiral ligament,the organ of Corti, and spiral ganglion cells by light micro-scopy. Green fluorescent protein epifluorescence was assessedin whole-mount organ of Corti preparations using confocalmicroscopy. This demonstrated transduction of inner haircells, outer hair cells, and supporting cells. Paraffin-embeddedcross sections similarly revealed gene transduction within theorgan of Corti. Threshold shifts of 39.8 T 5.4 and 37.7 T 5.5 dBwere observed in mice injected with HDAd or control buffer,respectively.Conclusion: The technique of scala media HDAd injectionreliably infects the adult mouse cochlea, including cells with-in the organ of Corti, although the procedure itself adverselyaffects hearing. Key Words: AdenovirusVCochleaVGenetherapyVHearingVHelper dependentVMouseVScalamediaVScala tympaniVVector.Otol Neurotol 28:1100Y1108, 2007.

Deafness is the most common sensory disorder.Approximately 1 in 1,000 children are born with a se-rious permanent hearing impairment, and single-genedefects probably account for more than half of thesepatients (1). Cochlear gene therapy has been proposedas a means of hearing restoration for this patient popula-tion (2).

The most efficient cochlear gene delivery techniquesdemonstrated to date have been with early-generationadenoviral vectors (3). All adenoviral infections provokean acute inflammatory response to the viral capsid pro-teins. First- and second-generation adenoviral vectors,which contain some native viral DNA, also are directlycytotoxic and provoke an adaptive cellular immuneresponse as new capsid proteins are produced by thehost cell (reviewed by 4Y6). Because these effects leadto death of the host cell, early-generation adenovirus-mediated gene transfer only produces short-term effectsin most tissues of the body and is unsuitable for manygene therapy applications (7,8).

Helper-dependent adenovirus (HDAd) is considereda third-generation adenoviral vector. It is devoid ofviral coding sequences and has been shown to mediatehigh-efficiency transduction in vivo and to producea long-term, high level of transgene expression with

Address correspondence and reprint requests to John S. Oghalai,M.D., Bobby R. Alford Department of Otolaryngology-Head andNeck Surgery, Baylor College of Medicine, One Baylor Plaza,NA102, Houston, TX 77030; E-mail: [email protected]

These data were presented at the Spring Meeting of the AmericanNeurotology Society in 2006.

This study was supported by NIH grant DC006671, The AmericanHearing Research Foundation, and The Caroline Weiss Law Fund forResearch in Molecular Medicine (to J.S.O.), NIH grant R01-067324,and The American Heart Association (grant 0465102Y to P.N.).

Supplemental digital content for this article is available on thejournal’s Web site at www.otology-neurotology.com.

Otology & Neurotology28:1100Y1108 � 2007, Otology & Neurotology, Inc.

1100


minimal chronic toxicity (4). These features would be ofbenefit for cochlear gene therapy because of the need tominimize trauma in this delicate, nonregenerative organand to have long-term gene expression. Another poten-tial benefit of HDAd is that it permits the transfer ofDNA greater than 37 kb long.

In vivo studies in guinea pigs and mice have beenperformed to test cochlear gene therapy techniques(9Y21). All have been hindered by the inherent difficul-ties in accessing the cochlea and cytotoxicity of thetransduction vector. The cochlea is quite small andencased within the densest bone in the body, the oticcapsule. It is a very delicate structure easily damaged bysurgical trauma. Studies in guinea pigs have been moresuccessful likely because of the larger size of itscochlea. This animal model may be preferable to usein studies of hair cell toxicity because of the abilityto experimentally produce hair cell death (22,23).However, mice provide a much better model to studyhereditary abnormalities pertinent to human disease, andthere are multiple mouse models of genetic hearingimpairment that can be used to test gene therapy tech-niques before human experimentation (24Y29). Herein,we present the technique we have developed to intro-duce HDAd into the mouse organ of Corti in an effort totransduce the cochlear epithelium.

MATERIALS AND METHODS

Surgical ProceduresThe Baylor College of Medicine Institutional Animal Care

and Use Committee approved the study protocol. Adult(4Y8 wk old) C57/Bl6 mice weighing 20 to 30 g were studied.The mice were anesthetized using an injection of ketaminehydrochloride (100 mg/kg, i.p.) and xylazine hydrochloride(5Y10 mg/kg, i.p.). The head was held secure by a custom

FIG. 1. Schematic overview of the incision and required expo-sure for scala media injection. The opening in the bulla (B) per-mits visualization of the stapedial artery (SA) and the spiralligament (SL) as landmarks. The glass micropipette (P) isinserted through the cochleostomy in the basal turn (BT). Notethat the head of the mouse is tilted so that the apical turn (AT) canbe identified. The tympanic annulus (TA) is also shown.

FIG. 2. Surgical approach to the mouse cochlea. A, An incisionhas been made posterior and inferior to the ear canal to exposethe sternocleidomastoid muscle with the overlying greater auricu-lar nerve (GAN) and the trapezius muscle (TM). The sternoclei-domastoid muscle was divided and the ends retracted outward.B, The underlying digastric muscle was divided and retracted toexpose the tympanic bulla (TB) and the tympanic annulus (TA). C,The bulla was opened to expose the stapedial artery (SA). Thelocation of the scala media (SM) of the basal turn can be delin-eated by the pigmented spiral ligament, which courses paralleland anterior to the stapedial artery.

1101HDAD-MEDIATED GENE TRANSFER INTO MOUSE COCHLEA

Otology & Neurotology, Vol. 28, No. 8, 2007


bite block. Body temperature was maintained at 39-C usingan electric heating pad (FHC, Bowdoinham, ME, USA). Allprocedures were performed on the left ear and can be observedin video format (see supplemental online video).

An incision was made ventral to the ear canal, and thetympanic bulla was exposed (Figs. 1 and 2). The bulla wasthen opened with a pick, with care taken to preserve the integ-rity of the tympanic annulus. The cochlea could be visualizedthrough the opening in the bulla and the stapedial artery iden-tified as it courses over the basal cochlear turn. The spiralligament could be visualized through the otic capsule bonebecause of its pigmentation.

The otic capsule bone overlying the middle of the cochlearbasal turn was carefully drilled away until the surface of thespiral ligament was reached using a Skeeter microdrill with a0.6-mm-diameter diamond burr (Xomed, Jacksonville, FL,USA) (Fig. 3). A glass micropipette with a tip diameter ofapproximately 10 Km was loaded with 1 Hl of virus suspension(8 � 109 viral particles) by aspirating from a drop of the viralsolution (see supplementary movie). The tip of the pipette wasthen slowly advanced through the spiral ligament into scalamedia. The virus suspension was then infused into the scalamedia during 1 minute by connecting an air-filled syringe tothe pipette holder and applying gentle manual pressure. Thefluid level in the tip of the micropipette could be observedto slowly drop as the infusion progressed. There occasionallywas some backflow noted, and the actual amount of deliveredvirus may be less than expected. Once completed, the micro-pipette was then withdrawn and the cochleostomy coveredwith a small piece of Gelfoam (Pharmacia Upjohn Company,Kalamazoo, MI, USA). The wound was closed in 2 layers withnonabsorbable sutures.

The mice were allowed to awaken from anesthesia, and theirpain was controlled with 0.15 mg/kg buprenorphine hydro-chloride (Abbott Laboratories, Chicago, IL, USA). After4 days, the animals were killed by cervical dislocation whileunder anesthesia.

HDAd FormulationsWe used 2 different formulations of HDAd. The first was

HDAd-LacZ, which contained a porcine cytomegalovirusYLacZ expression cassette (30) and was amplified with 116cells and AdNG163 (31,32). Characterization of the vector,as described previously (30), revealed the expected genomicstructure and a level of helper virus contamination of less than0.05%. The Escherichia coli lacZ gene was used as a reportergene. Transduced cells produce the enzyme A-galactosidase,which forms a blue precipitate when combined with 5-bromo-4-chloro-3 indolyl-A-D-galactopyranoside as a substrate.This can be visualized under light microscopy. The secondformulation of vector we used was HDAd-hemagglutinin(HA)-prestinYgreen fluorescent protein (GFP), which includedboth GFP and HA-tagged prestin as transgenes (33). Genetransduction within individual cells was assessed using confo-cal microscopy for GFP fluorescence or immunolabeling forHA expression. The viral suspension of HDAd was at a con-centration of 8 � 1012 viral particles per milliliter in a buffer of10 mmol/L Tris and 10% glycerol. As a control solution, thebuffer used to suspend the virus (10 mmol/L Tris and 10%glycerol) was also tested.

Auditory Evoked Brainstem Response MeasurementsTo measure auditory evoked brainstem responses (ABRs),

sound stimuli were generated digitally, attenuated using

FIG. 3. Injection into scala media. A, A microdrill was used toremove a small area of bone over the scala media, anterior to thestapedial artery (SA). B, The cochleostomy to enter the scalamedia (SM) was visible as a minute opening. The spiral ligamentwas just inside the cochleostomy, so there was no endolymphaticleakage. C, The pipette (P) was brought in from the left and wasmostly out of the plane of focus. It was inserted through thecochleostomy using a micromanipulator and penetrated throughthe spiral ligament to enter scala media.

1102 G. I. WENZEL ET AL.



programmable attenuators, and delivered via electrostaticspeakers (RP-2, PA-5, ED-1, and EC-1; Tucker-Davis Tech-nologies, Alachua, FL, USA) (34). The speakers were cali-brated across the frequency range before each experimentusing a probe-tip microphone (microphone Type 8192,NEXUS conditioning amplifier; Bruel and Kjaer, Naerum,Demark). This was performed after connecting the micro-phone and speakers to an ear bar inserted into the animal’sear canal.

The ABR was recorded using needle electrodes. One waspositioned at the vertex of the skull, and the other was insertedinto the muscle ventral to the temporal bone through the openwound. A ground electrode was placed in the hind leg. Thesignals were amplified 10,000 times using a biological ampli-

fier (HS4/DB4; Tucker-Davis Technologies), digitized at 100kHz, and digitally band-pass filtered from 300 to 3,000 Hz.

The stimulus for eliciting the ABR was a 5-millisecond sinewave tone pip with cos2 envelope rise and fall times of0.5 millisecond. The repetition time was 50 milliseconds,and 250 trials were averaged. The peak-to-peak ABR signal

FIG. 4. Representative images of the basal cochlear turn 4 daysafter HDAd-LacZ injection. Transgene expression is visible as ablue precipitate. A, An uninjected control cochlea after undergoingthe A-gal reaction demonstrated no evidence of blue precipitate.B, Representative basal turn from an HDAd-LacZ injectedcochlea. There was strong transgene expression at the injectionpoint (*). A linear staining pattern along the organ of Corti wasvisible (arrows). In addition, strong transgene expression can benoted within the modiolus (arrowheads). The scale bar is 200 Hm.

FIG. 5. Representative whole-mount preparations of the basalorgan of Corti from adult mice 4 days after HDAd-HA-prestin-GFPinjection. These are confocal images; GFP expression is green,and actin stained with phalloidin is red. Control contralateralcochlea (noninjected) (A) and cochlea injected with control buffer(B) show normal anatomy of the organ of Corti. There were 3 rowsof outer hair cells and 1 row of inner hair cells. The outer hair cellstereociliary bundle forms a ‘‘V’’ shape. C, Transduced inner haircells (red arrows) and pillar cells (green arrow) demonstratedGFP fluorescence. A nontransduced inner hair cell is also high-lighted (white arrow). D, Green fluorescent protein expression canalso be identified in outer hair cells (red arrows) and supportingcells such as Hensen cells (green arrows). A nontransducedouter hair cell is shown (white arrow). Scale bars are 20 Km.




was measured at stimulus intensities ranging from 10 to 80 dBsound pressure level in 10-dB steps for each frequency. Thesedata were interpolated off-line to analyze frequency-specificthresholds. The threshold shift for each frequency in each indi-vidual animal and each stage in the procedure was analyzed.We measured ABR thresholds at frequencies between 4 and 80kHz. This interrogates nearly the entire length of the mousecochlear duct (35).

Auditory evoked brainstem response threshold measurementswere performed before and after opening the cochleostomy onthe day of HDAd injection. Auditory evoked brainstemresponse thresholds were also measured on the day of killing.This was done after anesthetizing the animal and reopeningthe wound to verify that the middle ear space was aerated.The mean threshold for each animal was calculated by aver-aging the thresholds at each frequency. Differences betweenmean threshold values were calculated using the paired Stu-dent’s t test. All statistical analyses were performed usingExcel (Microsoft, Redmond, WA, USA) and SPSS (SPSS,Inc., Chicago, IL, USA). Statistical significance was assessedif p G 0.05. All presented values are mean T SEM. Plotting wasperformed using SigmaPlot (SPSS, Inc.).

Histologic ProcessingCochleae injected with HDAd-LacZ were opened at the

apex and base and immersed in freshly mixed fixative contain-ing 0.4 mL 25% glutaraldehyde, 2.5 mL 100 mmol/L EGTA,0.1 mL 1 mol/L MgCl2, 2 mL 37% formaldehyde, and 45 mL0.1 mol/L sodium phosphate buffer. After sitting in the fixativefor 1 hour, the cochleae were washed in a buffer containing

FIG. 6. Paraffin-embedded cross sections of the organ of Corti 4 days after HDAd-HA-prestin-GFP injection. Immunolabeling wasperformed for HA; a transgene marker linked to prestin was imaged by itself in A and C. Myosin VIIa, a hair cellYspecific marker and4¶,6-diamidino-2-phenylindole staining, to identify cell nuclei, were also imaged (B and D). A and B, A representative section from a control,noninjected cochlea. One inner and 3 outer hair cells can be noted using myosin VIIa labeling. There was no HA labeling. C and D,Representative sections from a cochlea injected with HDAd-HA-prestin-GFP. Normal inner and outer hair organization can be noted.Hemagglutinin expression was found in multiple cell types within the organ of Corti, including supporting cells and Hensen cells (bluearrows), pillar cells (yellow arrow), and probably outer hair cells (white arrow). Because the HA-prestin protein is a transmembrane protein,HA expression was localized along cell membranes. Scale bar is 20 Km.

FIG. 7. Auditory evoked brainstem response thresholds. Withthe scala media injection technique, the cochleostomy caused amild threshold shift across the middle of the frequency range.Significant differences in threshold are marked with an asterisk(*). Large threshold shifts were found 4 days after the injectionacross all frequencies. There were no statistically significantthreshold differences between cochleae injected with HDAd andcochleae injected with control buffer. This suggests that most ofthe cochlear dysfunction occurred due to injection trauma.




0.4 mL 1 mol/L MgCl2, 2.0 ml 1% deoxycholate, 0.4 mL 10%NP-40, and 197.2 mL 0.1 mol/L sodium phosphate buffer andincubated up to 16 hours in 5-bromo-4-chloro-3 indolyl-A-D-galactopyranoside. Finally, the enzyme reaction was stoppedby rinsing the cochleae in phosphate-buffered saline (PBS)and postfixed for 2 days in 4% paraformaldehyde. Onecochlear turn was microdissected, mounted on a glass slidewith Fluoromount G (Electron Microscopy Sciences, Ft.Washington, PA, USA), and visualized with a dissectingmicroscope.

Cochleae injected with HDAd-HA-prestin-GFP were per-fused with 4% paraformaldehyde and then placed in a jar con-taining 4% paraformaldehyde for 1 hour. Cell membranes werelightly permeabilized with 0.1% Triton-X for 10 minutes. Thecochlea was then rinsed with PBS twice for 5 minutes, incu-bated in Alexa Fluor 546-phalloidin (A22283; MolecularProbes, Eugene, OR, USA) at a concentration of 1:200 for1 hour, and then rinsed with PBS. This was done to stainF-actin. Whole-mount preparations of the organ of Cortiwere then microdissected out of the cochlea and mounted.Confocal microscopy was performed to visualize GFP andphalloidin independently (LSM 510; Zeiss, Jena, Germany).

Cochlear cross sections were also prepared from animalsinjected with HDAd-HA-prestin-GFP. The cochleae were har-vested and immersed in 4% paraformaldehyde for 2 days,decalcified in 0.2 mol/L EDTA for 2 weeks, embedded inparaffin, and cut into 7-Hm sections. After deparaffinizationand washing with PBS 3 times, the preparations wereimmersed with 4% normal goat serum in PBS for 1 hourat room temperature and incubated at 4-C overnight in a mix-ture of mouse anti-HA (dilution, 1:1,200) and rabbit antimyo-sin VIIa (dilution, 1:200; Affinity Bioreagents, Golden, CO,USA). After washing with PBS the next day, the preparationswere immersed with a mixture of Alexa Fluor 594 goat anti-mouse and Alexa Fluor 488 goat antirabbit secondary anti-bodies at room temperature for 1 hour. After washing withPBS, the cultures were mounted on glass slides with theVECTASHIELD mounting medium with 4¶,6-Diamidino-2-Phenylindole DAPI (Vector Laboratories, Burlingame, CA,USA) and sealed with nail polish.

RESULTS

We studied 41 mice. The first 8 mice were used todevelop the technique of scala media injection. Elevenmice were injected with HDAd-LacZ and 18 withHDAd-HA-prestin-GFP. Four mice were injected withcontrol solutions. All cochleae were studied histologi-cally. All mice had ABR measurements before thecochleostomy, 29 had ABR measurements after thecochleostomy, and 17 had ABR measurements justbefore they were sacrificed.

Light Microscopy of Cochlea Injected WithHDAd-LacZ

Gross cochlear dissections were visualized using lightmicroscopy to assess for blue precipitate indicative ofsuccessful HDAd-LacZ transduction. Mice injected withHDAd-LacZ demonstrated blue precipitate within thecochlea (n = 11 of 11). A representative specimen isshown in Figure 4. At the site of injection, there wasparticularly strong labeling (Fig. 4B). The spiral liga-

ment and spiral ganglion cells always demonstratedblue precipitate at the injection site. Most cochleaealso demonstrated labeling within the organ of Corti(n = 8 of 11). This can be noted as the continuous linearpattern of precipitate along the organ of Corti extendingaway from the injection site. Some leakage of the blueprecipitate occurred out of the transduced, degeneratingcells that seemed to discolor adjacent nontransducedcells. Because of this problem, it was difficult to con-fidently differentiate which cell types were transducedeven under high magnification.

Confocal Microscopy of Cochleae Injected WithHDAd-HA-prestin-GFP

We studied the transduction of individual cells withinthe organ of Corti using confocal microscopy usingHDAd-HA-prestin-GFP. Confocal imaging provides avery thin plane of sectioning and permits the counter-staining of actin within hair cells and supporting cellswith phalloidin. Within hair cells, large concentrationsof actin are found within the stereocilia and the cuticularplate (36). Green fluorescent protein is a cytoplasmicprotein and is not found in the plasma membrane. Theactin within the stereocilia and cuticular plate displacethe cytoplasm so GFP is only visible within the soma ofhair cells (37). Thus, with this method, we can confi-dently identify transduced hair cells and supporting cellswithin microdissected segments of the epithelium. Typi-cally, the green GFP can be observed wrapping aroundthe red cuticular plate.

Uninjected cochleae and cochleae injected with con-trol buffer demonstrated normal epithelial architecture,with 1 row of inner hair cells and 3 rows of outer haircells (OHCs) (Figs. 5A and B). Although all hair cellstereocilia demonstrated splaying, as is typical of micro-dissected whole-mount preparations, the general organi-zation of the stereocilia remained. The stereocilia ofinner hair cells tended to be aligned in a row, whereasthe stereocilia of OHCs tended to be aligned in a BV[shape. Because the epithelium is not completely flat,different regions of the hair cells can be identified inFigure 5A. In the third row of OHCs (top row), the cuti-cular plate is predominantly visible, whereas the stereo-cilia are more clearly visible within the first 2 rows.

Cochlea injected with HDAd-HA-prestin-GFP stu-died under whole-mount demonstrated GFP expression(n = 12) in nearly all cell types. This included inner haircells, OHCs, pillar cells, and other supporting cells. Fivetransduced inner hair cells are shown in Figure 5C (redarrows). In these cells, the GFP can be observed withinthe cytoplasm surrounding the cuticular plate. An exam-ple of a transduced pillar cell, which sits between 2 innerhair cells, is also shown (green arrow). In Figure 5D, 2transduced OHCs are highlighted (red arrows), althoughseveral more infected OHCs are also visible. Many trans-duced supporting cells are also visible in this image.

The stereocilia were not as easily identifiable incochlea injected with HDAd-HA-prestin-GFP com-pared with those injected with control buffer. In some




preparations, all 3 rows of OHCs could be found,whereas in some other preparations, only 1 or 2 rowswere found. It is unknown whether this change reflectstrauma from the dissection after animal sacrifice or wasa side effect of transduction process.

Cochlear Cross SectionsBecause of the concern that the stereociliary loss

might represent hair cell loss, we also studied paraffin-embedded cochlear cross sections. In this preparation,hair cells can be accurately identified by immunolabelingfor myosin VIIa, a hair cellYspecific protein. Transgeneprotein expression was identified using immunolabelingfor HA (n = 6 of 6 cochleae) (Fig. 6). Because the HAepitope was tagged to the transmembrane protein prestin,labeling was only found within cell membranes. Hemag-glutinin-prestin expression was predominantly foundwithin the supporting cells of the organ of Corti. How-ever, supporting cell and hair cell plasma membranesare adjacent to each other in the epithelium. Thus, itwas impossible to determine conclusively from thesecross sections whether hair cell membranes expressedHA-prestin or whether the labeling around the haircells represented adjacent transduced supporting cells.Nevertheless, there was clear evidence of supportingcell transduction and inner and OHC survival.

Assessment of Cochlear FunctionThere were substantial impacts of these manipulations

on cochlear function (Fig. 7). At the beginning of theexperiment, the mice had normal cochlear function withABR thresholds averaging 26.2 T 5.3 dB. Drilling thecochleostomy caused an average threshold elevation of5.1 T 5.7 dB, which was statistically significant withinthe midfrequency range (p G 0.05) and not significant atthe extremes (p 9 0.1). After 4 days of incubation, ABRthresholds were found to be elevated even further. Thesedifferences averaged another 39.8 T 5.4 dB and werestatistically significant at all frequencies (p G 0.05).The threshold shifts in mice injected with control bufferaveraged 37.7 T 5.5 dB and were equivalent to thosefound in the mice injected with HDAd at all frequencies(p 9 0.1).

DISCUSSION

This study demonstrates that HDAd can infect adultmouse cochlear cells, including inner hair cells, OHCs,and supporting cells. The ability to reliably transducethe mouse cochlea will permit studies of gene ther-apy in transgenic models of genetic deafness, a criticalgoal on the path to developing treatments for humandeafness. The benefit of this approach is that it is arelatively straightforward technique to produce trans-duced cells suitable for further study. For example, weare using the patch clamp technique to assess forchanges in the nonlinear capacitance and electromotilityof cultured OHCs from the prestin-null transgenicmouse that have been transduced by our HDAd to

express prestin (38). In addition, because HDAd trans-duces supporting cells quite well, this technique may bevaluable in studies involving gene therapy to restorehearing in mouse models of hearing loss due to muta-tions that alter supporting cell functions (e.g., connexinmutations).

The ability to effectively transduce cells within theadult mouse organ of Corti is generally accepted as anecessary step in developing strategies for human co-chlear gene therapy (3,10). Human hair cells are physio-logically similar to animal cochlear hair cells (39,40).Most other groups have found that hair cells are nottypically transduced in mice when adenovirus is appliedto scala tympani (13,14,17). This is consistent with therecent report that hair cells in the mouse cochlea losetheir coxsackie-adenovirus receptors before adulthood(41). However, it should be noted that 2 groups havedemonstrated mouse hair cell infection using adenoviralvectors with a scala tympani inoculation technique(13,20,42). Certainly, hair cells can be transduced byadenovirus in culture (37). Other gene therapy strategiesthat may prove more valuable in the mouse includenonadenoviral vectors (3,10).

The guinea pig is also a useful and commonly usedmodel for cochlear gene therapy experiments. However,the lack of transgenics and the outbred nature of guineapigs limit the ability of this model to study potentialtreatments for genetic hearing loss. Similar to themouse, the scala media technique in guinea pigs doesseem to be the most reliable technique to transducesupporting cells within the organ of Corti (17,18,22).However, 2 groups have reported widespread hair celltransduction using adenovirus in guinea pigs with thescala tympani technique (15,16,43). One group per-formed a slow infusion of 5 � 108 particle flush unitsof Ad(E1-, E3-, pol-) into scala tympani at 1 HL/h dur-ing 8 days with an implanted osmotic pump (16). Theother group rapidly injected 5 HL of 1.6 � 108 particleflush units /mL Ad5 through the round window (43).Both demonstrated near-complete transduction of haircells, with no transduction of mesothelial cells.

A downside to this technique is that substantial hear-ing loss occurs. Because the amount of hearing loss isthe same with control injections, the injection volumemay be too large. We are continuing to refine our injec-tion technique to minimize cochlear trauma, and furtherconcentration of the virus may be helpful. However,hearing loss may not be as significant a problem inhumans, the eventual goal of cochlear gene therapy.The human scala media contains 7.7 HL of endolymphcompared with the mouse scala media, which contains0.19 HL (44). Injections of artificial endolymph of up to1 HL during an 18-minute period have been performedinto the guinea pig scala media, which has an endo-lymph volume of approximately 2 HL (45). There wereno changes in the endocochlear potential during theinjection, suggesting that the scala media membranesremained intact with this volume load. Although thehuman cochlea has a substantially larger endolymphatic




volume, it has only approximately 5 times more haircells than the mouse cochlea (15,500 vs. 3,150, respec-tively) (46Y48). We speculate that a larger viral dosemay not be needed. In fact, if the membranes that sur-round the human scala media are able to tolerate andcontain the 1-Hl injection volume, it is possible that thenumber of viral particles needed to achieve adequatetransduction may actually be less in the human than inthe mouse.

Helper-dependent adenovirus may well be the adeno-virus of choice for cochlear gene therapy trials inhumans. In other organs, it has been shown to haveno direct cytotoxicity or late immune response, com-mon problems with early generation adenoviral vectorsthat only lead to short-term transgene expression (4).There is another potential benefit to using HDAd.First-generation adenoviral vectors inhibit mechano-electric transduction, whereas second-generation adeno-viral vectors (deleted E1, E3, the viral polymerase, andthe preterminal protein) do not (49). Consistent withthis, an in vivo study demonstrated that first-generationvectors are ototoxic, whereas second-generation vectorsare not (16). The reasons for these effects are unknown.However, HDAd also would not be expected to have anydirect ototoxic effects on mechanoelectric transduction.

CONCLUSION

Scala media injection of HDAd in the adult mousecochlea is successful in transducing cells within theorgan of Corti, including hair cells, although the tech-nique itself is associated with hearing loss.

Acknowledgments: The authors thank Dr. Robert Raphaeland his laboratory members for assistance with the confocalmicroscope and Isaac Ayala for the illustration.

REFERENCES

1. Smith RJ, Bale JF Jr, White KR. Sensorineural hearing loss inchildren. Lancet 2005;365:879Y90.

2. Wareing MJ, Lalwani AK. Cochlear gene therapy: current per-spectives. Int J Pediatr Otorhinolaryngol 1999;49:S27Y30.

3. Lalwani AK, Jero J, Mhatre AN. Current issues in cochlear genetransfer. Audiol Neurootol 2002;7:146Y51.

4. Palmer DJ, Ng P. Helper-dependent adenoviral vectors for genetherapy. Hum Gene Ther 2005;16:1Y16.

5. Tomanin R, Scarpa M. Why do we need new gene therapy viralvectors? Characteristics, limitations and future perspectives ofviral vector transduction. Curr Gene Ther 2004;4:357Y72.

6. St George JA. Gene therapy progress and prospects: adenoviralvectors. Gene Ther 2003;10:1135Y41.

7. Morral N, Parks RJ, Zhou H, et al. High doses of a helper-dependent adenoviral vector yield supraphysiological levels ofalpha1-antitrypsin with negligible toxicity. Hum Gene Ther1998;9:2709Y16.

8. Morsy MA, Gu M, Motzel S, et al. An adenoviral vector deletedfor all viral coding sequences results in enhanced safety andextended expression of a leptin transgene. Proc Natl Acad SciU S A 1998;95:7866Y71.

9. Lalwani AK, Han JJ, Walsh BJ, et al. Green fluorescent protein asa reporter for gene transfer studies in the cochlea. Hear Res1997;114:139Y47.

10. Lalwani AK, Mhatre AN. Cochlear gene therapy. Ear Hear2003;24:342Y8.

11. Lalwani AK, Walsh BJ, Reilly PG, et al. Development of in vivogene therapy for hearing disorders: introduction of adeno-associated virus into the cochlea of the guinea pig. Gene Ther1996;3:588Y92.

12. Kho ST, Pettis RM, Mhatre AN, et al. Safety of adeno-associatedvirus as cochlear gene transfer vector: analysis of distant spreadbeyond injected cochleae. Mol Ther 2000;2:368Y73.

13. Kawamoto K, Oh SH, Kanzaki S, et al. The functional and struc-tural outcome of inner ear gene transfer via the vestibular andcochlear fluids in mice. Mol Ther 2001;4:575Y85.

14. Jero J, Mhatre AN, Tseng CJ, et al. Cochlear gene deliverythrough an intact round window membrane in mouse. HumGene Ther 2001;12:539Y48.

15. Luebke AE, Foster PK, Muller CD, et al. Cochlear function andtransgene expression in the guinea pig cochlea, using adenovirus-and adeno-associated virus-directed gene transfer. Hum Gene Ther2001;12:773Y81.

16. Luebke AE, Steiger JD, Hodges BL, et al. A modified adenoviruscan transfect cochlear hair cells in vivo without compromisingcochlear function. Gene Ther 2001;8:789Y94.

17. Ishimoto S, Kawamoto K, Kanzaki S, et al. Gene transfer intosupporting cells of the organ of Corti. Hear Res 2002;173:187Y97.

18. Izumikawa M, Minoda R, Kawamoto K, et al. Auditory hair cellreplacement and hearing improvement by Atoh1 gene therapy indeaf mammals. Nat Med 2005;11:271Y6.

19. Martin DM, Raphael Y. Gene-based diagnostic and treatmentmethods for tinnitus. Int Tinnitus J 2003;9:3Y10.

20. Praetorius M, Baker K, Weich CM, et al. Hearing preserva-tion after inner ear gene therapy: the effect of vector andsurgical approach. ORL J Otorhinolaryngol Relat Spec 2003;65:211Y4.

21. Sugahara K, Shimogori H, Okuda T, et al. Novel method forhomogeneous gene transfer to the inner ear. Acta Otolaryngol2004:19Y22.

22. Staecker H, Praetorius M, Baker K, et al. Vestibular hair cellregeneration and restoration of balance function induced bymath1 gene transfer. Otol Neurotol 2007;28:223Y31.

23. Kawamoto K, Ishimoto S, Minoda R, et al. Math1 gene transfergenerates new cochlear hair cells in mature guinea pigs in vivo.J Neurosci 2003;23:4395Y400.

24. Weil D, Blanchard S, Kaplan J, et al. Defective myosin VIIAgene responsible for Usher syndrome type 1B. Nature 1995;374:60Y1.

25. Gibson F, Walsh J, Mburu P, et al. A type VII myosin encoded bythe mouse deafness gene shaker-1. Nature 1995;374:62Y4.

26. Probst FJ, Chen KS, Zhao Q, et al. A physical map of the mouseshaker-2 region contains many of the genes commonly deleted inSmith-Magenis syndrome (del17p11.2p11.2). Genomics 1999;55:348Y52.

27. Oghalai JS, Holt JR, Nakagawa T, et al. Ionic currents and elec-tromotility in inner ear hair cells from humans. Journal Neurophy-siol 1998;79:2235Y9.

28. Oghalai JS, Holt JR, Nakagawa T, et al. Harvesting human haircells. Ann Otol Rhinol Laryngol 2000;109:9Y16.

29. Anagnostopoulos AV. A compendium of mouse knockouts withinner ear defects. Trends Genet 2002;18:499.

30. Palmer D, Ng P. Improved system for helper-dependent adenoviralvector production. Mol Ther 2003;8:846Y52.

31. Palmer DJ, Ng P. Physical and infectious titers of helper-dependent adenoviral vectors: a method of direct comparison tothe adenovirus reference material. Mol Ther 2004;10:792Y8.

32. Ng P, Parks RJ, Graham FL. Preparation of helper-dependentadenoviral vectors. Methods Mol Med 2002;69:371Y88.

33. Xia A, Ugrinov R, Wenzel GI, et al. Development and expressionof helper-dependent adenovirus containing prestin in HEK 293cells and the mouse cochlea. Presented at the Association forResearch in Otolaryngology Annual Meeting, New Orleans, LA,February 20, 2005.

34. Oghalai JS. Chlorpromazine inhibits cochlear function in guineapigs. Hear Res 2004;198:59Y68.




35. Muller M, von Hunerbein K, Hoidis S, et al. A physiologicalplace-frequency map of the cochlea in the CBA/J mouse. HearRes 2005;202:63Y73.

36. Slepecky N, Chamberlain SC. Distribution and polarity of actin inthe sensory hair cells of the chinchilla cochlea. Cell Tissue Res1982;224:15Y24.

37. Holt JR, Johns DC, Wang S, et al. Functional expression ofexogenous proteins in mammalian sensory hair cells infectedwith adenoviral vectors. J Neurophysiol 1999;81:1881Y8.

38. Xia A, Wenzel GI, Palmer DJ, et al. Prestin transduction in theprestin null mouse cochlea in vitro and in vivo. Presented at theAssociation for Research in Otolaryngology Annual Meeting, Bal-timore, MD, February 6, 2006.

39. Oghalai JS, Patel AA, Nakagawa T, et al. Fluorescence-imagedmicrodeformation of the outer hair cell lateral wall. J Neurosci1998;18:48Y58.

40. Oghalai JS, Zhao HB, Kutz JW, et al. Voltage- and tension-dependent lipid mobility in the outer hair cell plasma membrane.Science 2000;287:658Y61.

41. Excoffon KJ, Avenarius MR, Hansen MR, et al. The Coxsackie-virus and adenovirus receptor: a new adhesion protein in cochleardevelopment. Hear Res 2006;215:1Y9.

42. Staecker H, Li D, O’Malley BW Jr, et al. Gene expression in the

mammalian cochlea: a study of multiple vector systems. ActaOtolaryngol 2001;121:157Y63.

43. Han D, Yu Z, Fan E, et al. Morphology of auditory hair cells inguinea pig cochlea after transgene expression. Hear Res 2004;190:25Y30.

44 Igarashi M, Ohashi K, Ishii M. Morphometric comparison of endo-lymphatic and perilymphatic spaces in human temporal bones. ActaOtolaryngol 1986;101:161Y4.

45. Rask-Andersen H, DeMott JE, Bagger-Sjoback D, et al. Morpho-logical changes of the endolymphatic sac induced by microinjec-tion of artificial endolymph into the cochlea. Hear Res 1999;138:81Y90.

46. Ding D, McFadden SL, Salvi RJ. Cochlear hair cell densities andinner ear staining techniques. In: Willott JF, ed. Handbook of MouseAuditory Research. Boca Raton, FL: CRC Press, 2001:189Y204.

47. Nadol JB Jr. Comparative anatomy of the cochlea and auditorynerve in mammals. Hear Res 1988;34:253Y66.

48. Retzius G. Das gehororgan der Wirbeltiere. II. Das gehororgander Wirbeltiere, der Vogel and der Saugetiere. Stockholm, Sweden:Samson and Wallin, 1884.

49. Holt JR. Viral-mediated gene transfer to study the molecular phy-siology of the mammalian inner ear. Audiol Neurootol 2002;7:157Y60.



Documents

Helper-Dependent Adenovirus-Mediated Gene Transfer Into the Adult Mouse Cochlea