The present study investigated the efficiency of 2 methods of gene transfection in neonatal mice. The investigation of gene therapy in mice is justified by the importance of mouse models in the research of genetic hearing loss and the early onset of this loss in many cases. After examining the challenges of high mortality after transfection surgery and the difficulty of operating on the small cochleae of neonatal mice, this study compared the transfection efficiency of 2 gene transfection approaches: RWM puncture; and trans-RWM after digestion. Using appropriate procedures and a new generation of AAV vector (with mutations of surface-exposed tyrosine residuals), we were able to obtain satisfactory transfection of the marker gene to the IHCs and OHCs, the most important cell types in the cochlea. However, transfection to type I SGNs was surprisingly low compared with that previously reported in adult guinea pigs 
. Moreover, we found the trans-RWM approach to be slightly less effective, and the outcomes were more varied.
Cochlear gene transfection using AAV vectors is relatively easy and safe. In a previous study, we found the trans-RWM approach to be effective and minimally invasive in guinea pigs 
. However, successful cochlear gene transfection in neonatal mice had the additional problem of a high postoperative mortality rate, primarily as a result of infanticide by the parents. The reasons for this are not clear, as no comprehensive study of this behavior is available. However, as it varies across strains, infanticide is likely dependent on some genotypic features of the animals 
. When males remain in the colony, the colony is more aggressive, and weaker pups are more likely to be eaten by the parents. In our preliminary study, we tested the effect of 2 factors and their combination: removing the father and sham surgery. The purpose of the sham surgery was to equalize the overall health status of all pups in the colony. Although this was not intended to be a systematic study, the preliminary study demonstrated that a combination of the 2 procedures significantly improved the survival rate of the neonates. Therefore, father isolation and sham surgery were used in the colonies involved in the gene transfection experiments.
Several procedures were considered beneficial and were adopted in this study. First, the mice were deeply anesthetized because of the separation of sternocleidomastoideus 
. Our aesthetic dose was adequate to maintain a deep sleep for approximately 40–60 min. Second, we did not use gas inhalation for anesthesia because the muzzle required for gas may have interfered with the operation 
. A further difficulty in the surgery was the exposure of the RWM, because it was largely covered by soft tissue. The narrow operating field made it difficult to separate the soft tissue mechanically without damaging the underlying RWM. Digestion using collagenase appears to be an effective way to overcome this problem. After digestion, the soft tissue became spongy and could be dissected with a paper tip.
Unlike microinjection through the bony wall of the cochlea into the scala media, the local transfection used in the present study did not cause cochlear damage 
. We found no threshold differences between the operated and control ears during the ABR test in either the trans-RWM or RWM-puncture groups. Application of collagenase solution facilitated dissection of the soft tissue that covered the RWM, and it partially damaged the RWM, which temporarily increased membrane permeability to the viral vector. The high regenerative potency of epithelial cells and connective tissues ensured that the RWM damage induced by collagenase digestion soon healed, as demonstrated in our previous report 
. This autonomic healing explains why the hearing function remained unchanged. Moreover, the ABR test results, which revealed no threshold difference between operated and contralateral ears, indicated that the injury caused by RWM puncture had healed by 2 weeks post-surgery. We used a micromanipulator to hold and advance the glass needle so that the RWM perforation was extremely small, as determined by the needle tip diameter, which was approximately 10 µm. We determined that this hole was well sealed after the needle was removed because any observed perilymph leakage was not significant. RWM treatment using 90 mg/mL collagenase caused over-digestion and significant hearing loss.
A comparison of the trans-RWM and RWM-puncture methods in terms of hearing function maintenance, transfection efficiency, and surgical difficulty revealed that both procedures preserved hearing function equally well, although the trans-RWM method did not disturb the perilymph with the administration of a large volume of exogenous agent. Our findings suggest that transfection via RWM puncture is more effective than that via the trans-RWM method, based on equal amounts of viral solution. However, more viral solution reached the perilymph in the RWM puncture group than in the trans-RWM diffusion group. The difficulty of the surgery was comparable in both approaches. The collagenase treatment and application of the virus increased the surgical time in the trans-RWM procedure. However, the vertical insertion of the needle tip across the RWM, which is hidden approximately 1 mm below the RW niche, required great caution and skill.
Our results indicated that cochlear transfection in neonatal mice is characterized by an even longitudinal transfection in the hair cells and low transfection efficiency in the SGNs. The mechanisms underlying the even transfection are not clear. In guinea pigs, transfection rates are reportedly higher in the basal turn, which is closer to the site of virus application 
. One explanation for this is that the vector was not fully diffused to the apex of the cochlea. However, evidence for quick longitude transportation of material in the perilymph has been reported, so the diffusion distance does not explain the transfection gradient 
. AAV entry into the cell is mediated by special receptors; differences in receptor distribution may establish a gradient of cellular tropism to AAVs along the cochlea 
. The third possible reason is a difference in basilar membrane permeability to AAVs across the cochlea. However, no evidence for AAV tropism or basilar membrane permeability along the cochlea has been reported in either guinea pigs or mice.
The reasons for the low SGN transfection rate in neonatal mice observed in the present study are also not clear. 2 pathways have been proposed for AAV transportation to the organ of Corti: (1) diffusion across the basilar membrane, which lacks tight junctions 
; and (2) transport along nerve fibers via the habenula perforata after AAV transfection of the SGNs. The 2nd approach was proposed to explain the higher transduction rates in IHCs than in OHCs following AAV transfection, suggesting that IHCs receive richer innervations from thick type I fibers. However, no clear evidence exists to support this postulation. The low SGN transfection rate observed in the present study and in others 
essentially refutes the possibility of the latter pathway.
Some untreated ears in the RWM-puncture group exhibited sparse IHC transfection. This was likely the result of the connection between the cochlear fluid and cerebrospinal fluid (CSF) via the cochlear aqueduct 
, and the fact that the 1 µL of viral solution injected in the present study was greater than the perilymph volume, which is reportedly 0.62 µL in adult mice 
. We injected a large volume of viral solution to maximize gene transfection. However, the original perilymph was completely replaced by the viral solution, and some may have leaked into the CSF. Leakage of the virus into the CSF suggests that the cochlea, at least in neonatal mice, is not entirely isolated and local gene transfection into the cochlea may have unpredictable effects in the central nervous system. Thus, the quantity of viral solution should be carefully planed to balance transfection and the risk of an unplanned leakage into the CSF. An advantage of the trans-RWM method is that leakage into the CSF is not a potential problem.
In summary, a high survival rate was obtained when father isolation and sham surgery were combined. The present study demonstrated satisfactory gene transfection to hair cells in neonatal mice at P7 via 2 methods, without causing significant hearing loss. The RWM puncture did not have an adverse effect on hearing preservation, and the ABR test results of this group were similar to those of the trans-RWM group. This was most likely the result of autonomic healing in the RWM of young mice. The outcome of our study paves the way for cochlear gene therapy in neonatal mice using AAV vectors.