Magnetic Targeting of AAV Gene Therapy for Inner Ear Following Systemic Delivery: Preliminary Findings and Transduction Pattern in Rat Cochlea

Abstract

Purpose

Delivery of therapeutics to the inner ear is complicated by their inaccessible location and the presence of the blood-labyrinth barrier that restricts most blood-borne compounds from entering the inner ear. This study addresses the challenge of optimal delivery in treating inner ear disease, focusing on magnetic targeting gene therapy using adeno-associated virus (AAV).

Methods

The investigation explores three AAV serotypes (AAV2 Quad Mut, AAV2 pANC80L65, and AAV9 PHP.eB) delivered systemically, tagged with a brain-derived nerve growth factor (BDNF) transgene and GFP reporter protein, and captured with superparamagnetic nanoparticles. External magnets target AAV delivery to the Left ear of both male and female Long Evans rats. After 2 weeks, we evaluated tropism and transduction in both cochleae and assessed distribution in other major organs (heart, lung, liver, kidney, spleen, and brain) using immunohistochemistry, real-time polymerase chain reaction, and enzyme-linked immunosorbent assays. Six animals were used for each experimental group.

Results

Immunofluorescence analysis demonstrated the qualitative distribution of AAVs in sensory cells and spiral ganglion neurons (SGN) in both ears. A significant increase in BDNF gene expression in the targeted left ear of rats administered AAV2 Quad Mut was observed. A single dose of magnetic targeting of AAV2 Quad Mut effectively transduced SGN and enhanced BDNF expression, leading to the restoration of ouabain-induced SGN loss and hearing loss (HL).

Conclusion

These findings indicate the potential of magnetic targeting to direct gene therapy following systemic delivery, paving the way for future applications in the treatment of HL.

Introduction

Hearing loss (HL) is the most prevalent sensory impairment in humans, affecting a large population worldwide [1]. Sensorineural hearing loss (SNHL) is the most common type of HL and is characterized by damage to the hair cells (HCs) in the cochlea, spiral ganglion, cranial 8th nerve, or the central processing auditory centers of the brain [1]. It can be caused by various factors, such as underlying diseases, drug ototoxicity, noise exposure, aging, or genetic factors [2]. The current treatment options for SNHL, including hearing aids and cochlear implants, have limitations in terms of amplification and clarity.

Gene therapy has emerged as a promising avenue for treating various diseases, including HL [2, 3]. However, the complexity of the inner ear poses significant challenges for effective gene delivery. The deep cranial location, the presence of the blood-labyrinth barrier (BLB), and the intricate architecture of the inner ear make it one of the most challenging organs to target with gene therapy [3]. Conventional techniques for accessing the inner ear, such as the intratympanic route or intracochlear approach, require specialized equipment and expertise and often result in disruption of micro-mechanics and altered homeostasis, leading to HL and vertigo [4–8]. Considering recent advancements in gene transfer to the inner ear, there remains a critical need to refine and develop minimally invasive methods for effective and widespread gene therapeutic transduction to the inner ear, thereby enhancing the accessibility and application of these therapies.

One potential approach is the systemic administration of substances via the intravenous route, which is a simple and minimally invasive method. However, this approach has significant limitations when it comes to delivering substances specifically to the inner ear. Firstly, the presence of the BLB, which separates the stria vascularis from the peripheral circulation, leads to variable penetration into the inner ear [9]. Secondly, there is a risk of undesirable systemic distribution and effects, such as systemic inflammatory response syndrome, which can have serious consequences [10]. Thirdly, high doses and volumes of the therapeutic substance may be required to achieve the desired therapeutic effect.

Recent studies have suggested that magnetic nanoparticles (NPs) can enhance the targeting of vectors and improve the efficacy of vector-mediated gene therapy [11, 12]. In our recent study, we demonstrated that local magnetic delivery of AAV2 (quad Y-F)-mediated brain-derived nerve growth factor (BDNF) coupled with superparamagnetic iron oxide nanoparticles (SPIONs) improves the transduction efficacy of the virus on hair cells (HCs) and restores auditory brainstem response wave 1 amplitude and synapse loss in noise-induced HL [13]. We also delivered magnetically targeted mesenchymal stem cells coupled with iron oxide NPs to the inner ear via systemic administration in a cochlear implant animal model and demonstrated improved spiral ganglion cell (SGN) survival and hearing recovery [14]. Collectively, the results indicated the effectiveness and efficiency of local delivery and magnetic targeting of gene therapy to the inner ear. Building on these findings, we aimed to explore the combination of systemic delivery and magnetic targeting for AAV-mediated BDNF therapy. This approach seeks to overcome, or at least mitigate, the barriers associated with systemic administration. To study the specificity of our novel method of drug delivery to the inner ear, we investigated the transduction pattern within the inner ear while also assessing the distribution of three serotypes of AAV constructs—AAV2 Quad Mut, AAV2 pANC80L65, and AAV9 PHP.eB in major organs including heart, lung, liver, kidney, spleen, and brain.

In addition, we used an ouabain-induced neuropathy model to examine the HL and protective effect of magnetic targeting of AAV2-BDNF in the inner ear. Ouabain is a cardiac glycoside that inhibits sodium–potassium ATPase by disrupting the ionic balance and increasing intracellular sodium and calcium levels. Consequently, ouabain is known to induce arrhythmias in cardiac tissues and auditory neuropathy in the cochlea [15, 16]. Ouabain is known to selectively and permanently damage SGN and Schwann cells, making it a disease model to study auditory neuropathy and sensorineural HL [17, 18]. It is typically applied locally to the round window membrane of the inner ear, allowing targeted and localized induction of auditory neuropathy [19]. We optimized the administration protocol for ouabain using an intratympanic route which provides minimal invasiveness and repeatability while still allowing dose-dependent hearing damage. With systemic administration and magnetic targeting of AAV2Quad Mut-BDNF, effective protection against ouabain-induced auditory threshold shift and loss of SGN was observed in the inner ear. Collectively, our findings hold great promise for the application of gene therapy in the inner ear and, subsequently, its potential in treating HL.

Results

Efficient In Vivo Delivery of AAV Serotypes to the Inner Ear: Co-localization of AAV Marker (GFP) with Hair Cell (Myosin VIIa) and Spiral Ganglion Neurons (Tuj1) Markers in Cochlear Cryosections

Our primary objective was to determine if systemically delivered selected AAV serotypes coupled with 20 µL SPIONs were successfully delivered to the Left inner ear following external magnetic targeting and to characterize their tropism pattern within this region. The right ear was preserved as contralateral control. Rats were systemically injected with an equal dose of 1.6 * 10¹⁰ AAV genome copies, and cochleae were harvested 2 weeks after tail vein injection and processed for cryosections. Although recombinant AAV vectors have been reported to extend their tropism to various cochlear cell subtypes, including IHC, OHC, supporting cells, stria vascularis cells, and SGN [3], we focused our study specifically on evaluating HC and SGN. We utilized immunostaining with the HC marker Myosin VIIa (red) and the neuronal marker Tuj1 (white) to assess the distribution and their co-localization with AAV marker GFP (green) in the inner ear (Fig. 1). The nuclei are stained blue with DAPI. All stained sections were imaged initially at 10× to check for overall GFP expression and then individual turns were imaged at 20×. For all 3 serotypes used, the brightest signals were observed in the Rosenthal canal, but weaker signals were also observed in OHC and IHC within the Organ of Corti. Although GFP signals were observed in all cochlear turns and along the mid-modiolus, we selected mid-cochlear turns from all immune-stained cross sections across various groups in this study for comparison due to their minimal disruption, uniformity, and consistency.

Fig. 1.

Tropism of AAV2 Quad Mut, AAV2pANC80L65, and AAV9 PHP.eB coupled with SPIONs following systemic delivery and magnetic targeting to left ear. Representative immunostaining images of the Rosenthal canal (A) and Organ of Corti (B) stained with Myosin VIIa (red), GFP (green), and Tuj1 (white). A 20× image of mid turn cochlea shows co-localization of GFP and Tuj1 for AAV2 Quad Mut, AAV2pANC80L65, and AAV9 PHP.eB serotypes used for both Left and right ears. GFP expression is absent in control rat that received no treatment. Scale bar, 100 µm. B 20× image of the Organ of Corti shows GFP expression in IHC and OHC within the Organ of Corti. AAV2 Quad Mut and AAV2pANC80L65-treated rats show stronger GFP expression compared to AAV9 PHP.eB-treated rat for both Left and contralateral right ear. GFP expression is absent in the control rat. Scale bar, 100 µm

Figure 1A shows 20× images of the Rosenthal canal, where GFP expression is observed in ears of rats injected with either AAV2 Quad Mut, AAV2pANC80L65, or AAV9 PHP.eB. GFP expression was colocalized with Tuj1, indicating transduction of SGN within the Rosenthal canal. The co-localization was observed among all three AAV serotypes, in both left (magnetically treated) and right (contralateral control) ears. There was no discrete difference of GFP expression comparing among left ears or right ears of the three AAV serotypes. GFP expression and colocalization were absent in control ear (rat that did not receive any intervention) indicating successful delivery and tropism of our selected AAV serotypes in SGN cells following tail vein injection and external magnetic targeting.

Figure 1B represents the mid turn Organ of Corti, including IHC and OHC. GFP expression and co-labeling of Myosin VIIa were more or less observed in the IHC and OHC for all AAV serotypes used. Within each serotype, the co-expression was observed consistently across all cochlear turns, with no discernible differences between them. Qualitatively, GFP expression was stronger for AAV2 Quad Mut and AAV2 pANC80L65 compared to AAV9 PHP.eB for both left (magnetically targeted) and right (contralateral control) ears. Control ears exhibit no GFP expression and co-localization within the Organ of Corti.

AAV Serotypes Transduced Inner and Outer Hair Cells: Co-expression Localization Study of AAV Marker (GFP) and Hair Cells (Myosin VIIa) in Cochlear Whole Mount Sections

Due to the limited number of hair cells in cochlear cross sections, we evaluated the transduction efficiency of AAV2 Quad Mut, AAV2 pANC80L65, and AAV9 PHP.eB in hair cells of micro-dissected whole mounts as shown in Fig. 2. Cochleae were harvested in 2 weeks following systemic administration of AAV and external magnetic targeting. The dissected three parts represent low (apex, < 8 kHz), middle (middle, 8–20 kHz), and high (base, > 20 kHz) frequencies. Qualitative analysis of the apical, mid, and basal regions revealed co-expression of GFP and Myosin VII in both IHC and OHC of the left ears treated with the three AAV serotypes. In addition, GFP expression was also detected in the IHC and OHC of the contralateral (right) ears across all AAV serotype administered groups. Within each AAV serotype, the apex showed stronger GFP signals compared to the mid and basal turns. GFP signals were absent in both OHCs and IHCs of the cochlear turns in control rats that received no treatment.

Fig. 2.

Transduction efficiencies of AAV2 Quad Mut, AAV2 pANC80L65, and AAV9 PHP.eB in HC. Rats were systemically injected with an equal dose of 1.6 * 10¹⁰ genome, and cochleae were harvested 2 weeks after AAV serotypes injection. The cochleae were fixed and dissected into three parts, representing low (apex, < 8 kHz), middle (middle, 8–20 kHz), and high (base, > 20 kHz) frequencies. Then, cochleae were immunohistochemically labeled by GFP and Myosin VIIa, followed by confocal imaging. Both IHC and OHC showed co-expression of GFP and Myosin VIIa across all AAV serotypes. AAV2 Quad Mut demonstrated the strongest expression in the apical region of the Left ear. No GFP signals were expressed in the control rat which did not receive treatment. Scale bar, 100 µm

AAV-Coupled SPIONs Enhanced BDNF Expression in the Cochlea: Immunohistochemistry and RT-qPCR Study

After evaluating the distribution of GFP-tagged AAV serotypes in the inner ear, a separate set of animals was assessed to determine the expression of BDNF in the neuronal cells of Rosenthal canal. Figure 2A illustrates the co-expression analysis of BDNF (red), GFP (green), and Tuj1 (white) in SGN of cochlear cryosections. There was noticeable co-expression of GFP and BDNF in SGN labeled with Tuj1 within the Rosenthal canal in all AAV serotype-treated ears. Qualitatively, within each serotype, the left ear showed stronger BDNF signals compared to the right (contralateral control) ear. Although the control showed some BDNF signal, the absence of GFP co-expression suggests the presence of endogenous BDNF within the control ear. Expression of BDNF in normal cochleae has been reported in both human and animal models [20, 21].

To validate the histology and perform a quantitative comparison, we further evaluated the level of BDNF mRNA in the magnet-targeted left ear and contralateral right ear after systemic delivery of AAV serotypes using RT-qPCR. The relative BDNF mRNA expressions in AAV-treated cochleae were compared to those in control rats that received neither AAV-mediated BDNF gene therapy nor magnet targeting. As shown in Fig. 2B, compared to control, the BDNF expression was increased in the left ear of all-treated groups. However, significance was achieved in the left ear of the group treated with AAV2 Quad Mut serotype (P < 0.01). Compared to contralateral control (right) ears, significant upregulation was also seen in the left ear of the group receiving AAV2 Quad Mut serotype (P < 0.001). No significant changes were observed in the left ears of rats treated with AAV2 pANC80L65 and AAV9 PHP.eB compared to contralateral control (right) ears.

Distribution of BDNF Expression in Brain, Lung, Kidney, Liver, Heart, and Spleen: RT-qPCR and ELISA

Delivery of a viral vector through the bloodstream will likely spread the virus to undesired tissues/organs [22, 23]. This issue is particularly important because, unlike the cochlea, other organs do not have a BLB. Apart from raising safety concerns, sequestration of AAV in these non-target locations also reduces the amount of vectors that can be delivered to the targets and hence has significant implications on the effective vector dose needed for systemic therapy [24]. Therefore, we set up an experiment in adult rats receiving AAV2 Quad Mut, AAV2 pANC80L65, and AAV9 PHP.eB to determine if AAV-BDNF delivered via the tail vein was taken up and expressed by any peripheral organs. Two weeks post-injection with an equal dose of 1.6 * 10¹⁰ genome, we harvested the brain, lungs, kidneys, liver, heart, and spleen for biochemical analysis (Fig. 3A). Compared to the control group, all examined tissues except the lungs showed upregulation of BDNF expression. The absence of any changes in the lungs is expected due to the first-pass effect achieved by administering sodium nitroprusside 5 min prior to injecting the AAV-BDNF-SPION complex. Notably, a statistically significant increase in BDNF expression was observed in the brain (P < 0.05) and heart (P < 0.01) tissues obtained from rats administered AAV2 Quad Mut. Moreover, AAV9 PHP.eB treatment led to a significant rise in BDNF expression in the brain (P < 0.05), kidney (P < 0.001), liver (P < 0.01), and spleen (P < 0.001). Additionally, AAV2 pANC80L65 demonstrated a significant increase in BDNF expression specifically in the kidney (P < 0.05), heart (P < 0.05), and spleen (P < 0.05).

Fig. 3.

Distribution to peripheral organs after BDNF gene therapy using AAV2 Quad Mut, AAV2 pANC80L65, and AAV9 PHP.eB. Two weeks post injection of AAV-BDNF-SPION complex, tissues were harvested and processed for gene and protein analysis. A RT-qPCR showing relative gene expression of the BDNF in the brain, lung, liver, kidney, heart, and spleen of control, AAV2 Quad Mut, AAV2 pANC80L65, and AAV9 PHP.eB administered rats. Lungs showed no changes due to first-pass metabolism. A significantly higher expression was observed in the brain and heart for AAV2 Quad Mut; the brain, kidney, liver, and spleen for AAV9 PHP.eB, and the kidney, heart, and spleen for AAV2 pANC80L65. B ELISA study demonstrates the concentration of the BDNF protein in the brain, liver, and kidney of control, AAV2 Quad Mut, AAV2 pANC80L65, and AAV9 PHP.eB administered rats. AAV2 Quad Mut showed a significant increase in the brain and liver. Bar diagrams represent the mean ± SEM of relative gene expression, with n = 6

Further to validate if the mRNA expression translated to corresponding protein expression, we performed ELISA on brain, being the closest organ to magnet after the ear, liver, and kidney, major organs of metabolism, and excretion (Fig. 3B). An increase in BDNF concentration was observed across all tested serotypes compared to control for the three tissues. However, the level of significance was only achieved in the brain (P < 0.05) and liver (P < 0.001) in animals treated with the AAV2 Quad Mut variant.

Intratympanic Administration of Ouabain Caused Significant SGN Reduction and Ouabain-Induced HL but Not Intraperitoneal Administration at Day 7

In order to determine the best administration method for ouabain in rats as the ototoxicity model, both intratympanic (IT) and intraperitoneal (IP) injection of ouabain were tested and compared. Rats were randomly assigned to each of the three groups: ouabain injected by IT, ouabain injected by IP, and control with saline only. Ouabain was injected IT at 10 mM concentration for approximately 0.5–0.8 mL until it filled up the ear canal. In comparison, ouabain was also injected IP at 5 mg/kg twice on day 1 and day 3. Auditory brainstem response (ABR) tests were performed before and 1 week after the ouabain and saline treatment. After 7 days, only the rat group with IT administration had significantly higher ABR threshold shifts at higher frequencies (16 k and 32 k) compared to the control and IP groups (Fig. 4A). We did not find any significant difference in the effects of administering ouabain IP at the concentration we tested compared to saline control. The functional HL is correlated with the SGN loss observed most significantly at the basal turn from cochlear cross-sections (Fig. 4B). SGN were labeled with Tuj1 at different cochlear regions from the Rosenthal canal of the modiolus of the cochleae for both the control and ouabain-treated groups (IT). Intact SGN morphology in the Rosenthal canal was seen in both saline and ouabain-treated groups in the apex region (Fig. 4B). Damaged SGN (empty spaces) were observed in ouabain administered groups (IT) at the mid turn and basal turn regions compared to the saline group.

Fig. 4.

Measurement of ABR threshold shift following ouabain treatment. A Mean ABR threshold shift at 7 days post ouabain injection via intraperitoneal (IP) administered ouabain or intratympanic (IT) injected ouabain relative to saline control. IT delivery of ouabain is more effective in causing cellular damage by ablating SGNs. IP administration of ouabain did not affect ABR threshold significantly. Post 7 days, statistical significance was observed in intratympanic administered animals at high SPL (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Data are presented as means ± SEM. n = 4 rats for each group. B Low-magnification cross-section through the cochleae comparing ouabain-treated (left) and control saline-treated (right) ears of rats revealed significant loss of SGNs (green) in mid and basal turns. Scale bar, 100 µm. Immunofluorescence images are representative images of n = 4 animals in each group (cochlear hair cells stained with Myosin VIIa, red, and Tuj1 green)

Magnetic Targeting of AAV2 Delivery of BDNF Rescue Mid and Basal Turn SGN

To test the impact of magnetic delivery of BDNF gene therapy on rescuing hearing in the model with damaged SGN by ouabain, functional ABR hearing test was measured before and after ouabain injection and magnetic delivery of BDNF. Three different groups of rats were treated with either saline, ouabain, or ouabain with magnetic targeting with AAV2(Quad Mut)-BDNF. Ouabain caused a significant and irreversible ABR threshold shift in higher experimental frequencies (16 kHz, 24 kHz, and 32 kHz). The threshold shift did not recover even after 1 month (Fig. 5A). Interestingly, we observed a moderate difference between the ouabain-treated group and the magnetic delivery group. For the rats that were treated with AAV2 Quad Mut-BDNF with magnetic targeting, there was a minor recovery of threshold shift for 10 dB at frequency of 32 kHz. The same recovery was observed 2 weeks post treatment. One month after ouabain administration, the threshold shift did not fully recover for higher frequencies (16 kHz, 24 kHz, and 32 kHz), but there was a moderately significant recovery of threshold shift at 24 kHz and 32 kHz 20 dB compared to ouabain group. Interestingly, no significant difference between left ear (where the magnet was placed) and contralateral control; right ear was observed. Overall, over 1 month period of time recovering from ouabain insult, systemic magnetic delivery of BDNF showed smaller threshold shifts compared to ouabain only group at higher frequencies (24 kHz and 32 kHz). This shows that loss of hearing due to ouabain damaging SGN is partially recovered and recovered at faster rate at high-frequency regions (mid to basal turn) in magnetic targeting of AAV2-BDNF group followed by systemic delivery.

Fig. 5.

Magnetic targeting of AAV2 Quad Mut-BDNF-SPION treatment promotes recovery of hearing function and preserves spiral ganglion neurons (SGNs) following ouabain-induced injury. A Mean ABR threshold shifts at 1 week, 2 weeks, and 2 months post-intratympanic (IT) ouabain injection relative to pre-treatment. IT injection of ouabain caused a significant threshold shift over the saline-injected control animals at higher frequencies (16 to 32 kHz) 1 week after ouabain injection. ABR threshold shifts recovered partially by magnetic delivery of AAV2 Quad Mut-GFP-BDNF-SPION 1–2 weeks following ouabain injection at 32 kHz and showed more significant rescue 1 month following ouabain injection at 24–32 kHz. (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Data are presented as means ± SEM. = 5 rats for each group. B Immunostaining of rat cochlea mid-basal turn cryosections with BDNF, GFP, Tuj1, and DAPI imaged at 20 × for treatment groups: Saline+Magnet, Ouabain+Magnet, and Ouabain + AAV2 Quad Mut + Magnet. Positive staining is observed for GFP and mBDNF in AAV2 Quad Mut group. Scale bar, 100 µm

To further investigate the histological effects of ouabain exposure and the magnetic delivery of BDNF in the inner ear, we performed immunostaining on cochlear cryosections to visualize SGNs, GFP, and BDNF. Figure 5B displays a representative immunofluorescent image of the Rosenthal canal in the mid-to-basal turn region. A significant loss of SGNs was observed in the Rosenthal canal of the cochleae in the ouabain-treated group compared to the saline control group. However, magnetic delivery of AAV2 Quad Mut-BDNF demonstrated partial recovery of SGNs 1 month after ouabain-induced damage (Fig. 5B). While the rescue of hearing loss, as measured by ABR thresholds, indicated comparable functional improvements in both the treated left ear and the contralateral right ear, immunostaining revealed more prominent GFP and BDNF expression in the magnetically targeted ear compared to the contralateral ear. This discrepancy may be partially explained by the distinct nature of these assessments. Functional recovery measured by ABR thresholds reflects the overall effects of BDNF on SGNs, including neuroprotection, synaptic repair, and enhanced neuronal survival [13, 25, 26]. In contrast, the immunostaining represents a localized view of BDNF expression in a single cochlear cross-section, focusing on a specific region. Taken together, our findings suggest that the local magnetic delivery of AAV2 Quad Mut-BDNF to the inner ear was effective in targeting SGNs, helping to rescue damaged neurons and contributing to partial hearing loss recovery in the mid-to-basal turn frequency range.

Discussion

The use of systemic delivery for drug or gene therapy offers major advantages, such as ease of clinical application and reduced time requirement, as it does not involve the complexities of invasive procedures targeting the inner ear or require skilled personnel. However, it is often associated with suboptimal drug delivery to the inner ear and can lead to undesired off-target effects [27]. To address these challenges, researchers have recently explored the use of nanoparticle systems in conjunction with systemic administration. Among the various mechanisms employed to enhance systemic delivery to the inner ear, magnetic targeting using nanoparticles has shown great promise. Magnetic targeting utilizes the forces generated by magnetic fields to guide SPIONs towards targeted tissue and interestingly, across the RWM, introducing new possibilities for treating inner ear diseases and HL [13, 14, 28, 29]. SPIONs are well-established as ideal candidates for delivering drug substances due to their excellent biocompatibility and superparamagnetism. The paramagnetic properties of SPIONs not only enable their use as MR contrast agents but also allow for targeting specific organs for cancer treatment and imaging [30, 31]. Previous in vivo studies conducted in rat and guinea pig models, as well as in vitro investigations using human temporal bone samples, have demonstrated the effectiveness and safety of SPIONs as drug delivery vehicles, including our own research [11, 13, 14]. In vivo, SPIONs are generally very safe, being cleared by uptake into phagocytic cells and the reticuloendothelial system [14, 32]. They degrade within 4–6 weeks and are recycled into the normal iron pool. These encouraging results suggest that with further modification, magnetic targeting in combination with systemic delivery of gene therapy could be translated to clinical applications for inner ear disease or HL. In our current study, we applied this approach to compare the tropism patterns and transduction efficiency of three recombinant adeno-associated virus (AAV) serotypes: AAV2 Quad Mut, AAV2 pANC80L65, and AAV9 PHP.eB, with the aim of enhancing the understanding of this approach’s potential utility in clinical drug or gene delivery to the inner ear.

Gene delivery to the inner ear requires the packaging of genes into vectors capable of entering cells. Among various viral vectors, AAV has received significant attention for inner ear gene therapy due to its many advantageous attributes. The small, single-stranded DNA-containing parvovirus with a non-enveloped protein capsid is non-pathogenic, capable of transducing both pre- and post-mitotic cells, including the neurosensory epithelia of the inner ear, and can achieve long-term expression of the transgene [3, 6, 7]. Since its initial description in the 1960 s as a replication-deficient member of the Parvoviridae family, numerous AAV serotypes and over 100 variants have been isolated from different animal species [4]. Although multiple AAV serotypes have been identified to improve their transduction to target tissues, some do not exhibit the expected tropism patterns. In addition, comparing the tissue tropism patterns of AAV serotypes is universally difficult because of variations in vector titers, doses, promoters, transgenes, delivery routes, points of administration, and age of animals at administration. For instance, Kim et al. [33] demonstrated efficient transduction of both inner and outer hair cells after injection of rAAV2/1-CMV-GFP (1.3 × 10^13 GC/mL), whereas Landegger et al. [34] reported strong transduction of inner hair cells but weak transduction of the outer hair cells after injection of the same construct (3.5 × 10^13 GC/mL). Another potential explanation for this difference in transduction is that the former study administered the gene therapy in utero, while the latter performed the injections in postnatal mice. In the present report, we evaluated three recombinant AAV serotypes for inner ear gene delivery, including (i) AAV2 Quad Mut, (ii) AAV2 pANC80L65, and (iii) AAV9 PHP.eB., that were previously identified to be successfully delivered to mammalian inner ear to target a wide range of cell types, including HCs, supporting cells (SCs), stria vascularis, and auditory neurons [6, 13, 34–38]. All three investigated AAV serotypes incorporate neurotrophins BDNF using the same promoter, CBA, in equivalent doses of approximately 1.6 * 10¹⁰ GC/animal.

Neurotrophins BDNF play a critical role in the treatment of HL by promoting cell survival, supporting hair cell regeneration, protecting auditory neurons, enhancing synaptic plasticity, and aiding in cochlear implant integration [9, 39]. BDNF supports the survival of SGNs in the inner ear, prevents their degeneration, and promotes the regeneration of damaged hair cells [40–42]. In this regard, Fukui et al. demonstrated significant neural preservation and regeneration in a mouse model of human hereditary deafness [42]. It acts as a neuroprotective agent, shields auditory neurons from damage, and improves the efficiency of auditory signal transmission through modulation of synaptic connections [24]. Previous studies, including our own, have highlighted the protective role of BDNF for SGN and enhancement in the function of cochlear implants [13, 24, 43]. Therefore, we selected the neurotrophins BDNF for this study due to its significant potential therapeutic applications in treating HL, particularly in the preservation and restoration of auditory function.

Our first objective was to determine if we could successfully deliver the AAV serotypes to the inner ear after systemic injection via the rat’s tail vein and magnetic targeting to the Left ear. To assess this, we initially assessed the expression of GFP, the AAV marker, in cross-sections of the cochleae collected 2 weeks post-treatment. Our results revealed successful transduction of SGN in all cochlear turns, with co-expression of GFP and Tuj1 in both left and right ears of treated animals (Fig. 1A). Similar transduction of SGN cells with AAV2/Anc80L65 was observed following AAV injection in the semicircular canal [6]. The Organ of Corti containing the hair cells also showed GFP expression; however, we further tested the hair cell transduction efficiency of AAV in cochlear whole mount (Fig. 6). Immunostaining of cochlear whole mounts revealed co-expression of GFP and Myosin VIIa, an actin-binding motor protein involved in the formation of hair cell stereocilia. This co-expression was observed in both OHC and IHC across the apex, mid, and basal turns of both left (magnetic targeted) and right (contralateral control) ears in animals receiving AAV-BDNF. Similar findings were previously reported by Kim et al., who evaluated the transduction distribution and efficiency of EGFP-expressing AAV2-derived vectors, particularly the PHP.eB serotype [44]. However, it is important to note that the interpretation of GFP expression is limited by variability in Tuj1 and Myosin VIIa staining intensity observed in some treated groups. This variability introduces a potential confounding factor in immunofluorescence analysis, as differences in staining intensity could affect the perceived GFP signal and compromise the accuracy of comparative assessments.

Fig. 6.

Transduction efficiencies of AAV2 Quad Mut, AAV2 pANC80L65, and AAV9 PHP.eB in HC. Rats were systemically injected with an equal dose of 1.6 * 1010 genome, and cochleae were harvested 2 weeks after AAV serotypes injection. The cochleae were fixed and dissected into three parts, representing low (apex, < 8 kHz), middle (middle, 8–20 kHz), and high (base, > 20 kHz) frequencies. Then, cochleae were immunohistochemically labeled by GFP and Myosin VIIa, followed by confocal imaging. Both IHC and OHC showed co-expression of GFP and Myosin VIIa across all AAV serotypes. AAV2 Quad Mut demonstrated the strongest expression in the apical region of the Left ear. No GFP signals were expressed in the control rat which did not receive treatment. Scale bar, 100 µm

Our previous study using a local delivery and magnetic targeting confirmed that SPION-associated signal loss was higher in the cochleae of SPION-treated ear (left ear) compared to the control untreated ear (right ear) when using magnetic targeting. Furthermore, SPIONs coupled with AAV-BDNF therapy with magnetic targeting showed no transfection in the contralateral control right ears [13]. The discrepancy in this study could be attributed to the systemic delivery approach. Due to the relatively small size of the animal’s head and the use of a 1 T magnet, the magnetic attraction of the particles is distributed across the entire head, affecting both ears. While systemic delivery can lead to distribution to undesired organs and requires a higher vector dose, it avoids the need for specialized equipment and expertise required for local applications such as intratympanic or intracochlear delivery. These local methods are often not readily available in primary care settings, such as rural hospitals.

Following the successful delivery using the systemic delivery and magnetic targeting approach, we were concerned about potential changes in BDNF expression within the inner ear. To investigate this, we immunoassayed cochlear cryosections for BDNF expression. Our results revealed co-expression of BDNF with GFP in cross-sections of the adult rat’s left ear, specifically in the spiral ganglion neurons (SGNs) of Rosenthal canal. Weak BDNF expression was also observed in the right (contralateral control) ear. The control ear that received no treatment showed slight BDNF expression, likely due to the presence of endogenous BDNF [20, 21]. To further validate our qualitative results, we performed quantitative analysis of BDNF expression using RT-qPCR. Our data revealed an upregulation of total BDNF expression in the left (magnetically targeted) ear, with a notable increase observed only with the AAV2 Quad Mut serotype compared to contralateral control (right) ear and control ear that did not receive any treatment. This discrepancy between the immunohistochemistry and RT-qPCR results for AAV2 pANC80L65 and AAV9 PHP.eB could be attributed to several factors. First, immunohistochemistry is a qualitative method that visualizes protein expression and localization, while RT-qPCR provides a quantitative measure of mRNA levels. It is possible that while BDNF protein was detected in higher amounts in the treated ears, the corresponding mRNA levels may not have shown a significant increase due to post-transcriptional regulatory mechanisms affecting mRNA stability or translation efficiency. Additionally, variations in sensitivity and specificity between the two techniques may have contributed to these findings. Immunohistochemistry relies on antibody specificity, and differences in antigen retrieval or antibody performance could lead to perceived differences in signal intensity. On the other hand, RT-qPCR is highly sensitive to experimental variability in RNA extraction, reverse transcription efficiency, and amplification steps, which could influence the quantification of mRNA levels. Another possible explanation is the spatial expression pattern of BDNF in the cochlea. Immunohistochemistry allows for the visualization of localized expression, whereas RT-qPCR provides an average expression level for the entire cochlea. If BDNF expression was localized to specific regions within the left ear, as observed in immunohistochemistry, these localized increases may not have been adequately captured by RT-qPCR.

These results highlight the importance of employing complementary methodologies to assess gene and protein expression. While the qualitative data from immunohistochemistry suggest a stronger BDNF protein presence in treated ears, the quantitative RT-qPCR findings underscore the need for further exploration of the molecular mechanisms influencing BDNF expression. Future studies incorporating additional quantitative methods, such as Western blot or targeted proteomics, could provide more insight into this discrepancy.

As previously mentioned, a primary shortcoming of systemic drug delivery is the subtherapeutic concentration in the inner ear and off-target distribution and effects [24]. This issue is particularly important because, unlike the cochlea, other organs do not have a blood-labyrinth barrier. To understand the distribution of AAV-BDNF following systemic delivery and magnetic targeting to a specific ear, we analyzed BDNF expression in several major organs, including the brain, lungs, kidneys, liver, heart, and spleen (Fig. 3). Indeed, our results provided new insights into the distribution of AAV2 Quad Mut, AAV2 pANC80L65, and AAV9 PHP.eB across these organs as we observed transduction of peripheral organs tested. The result is not surprising given that barriers like BLB do not exist in other organs and the proximity of organs like brain to the magnet. Moreover, BDNF expression has been reported in several non-neuronal tissues, including muscle, thymus, heart, liver, vascular smooth muscle cells, and spleen [45–48]. As these non-target effects reduce the amount of vectors that can be delivered to the targets, it is crucial to determine an effective vector dose needed for systemic therapy for future applications.

Based on our previous local magnetic delivery study and the current qualitative and quantitative analyses, we have selected the AAV2 Quad Mut serotype to assess gene therapy in ouabain-induced HL model [13].

Because local magnetic delivery of AAV2-BDNF showed prominent accumulation of BDNF in SGNs, we used a ouabain-induced disease model to examine the effectiveness of magnetic targeting of AAV2-BDNF on recovering lost neurons. Treatment using magnetic targeting of AAV2 Quad Mut-BDNF efficiently and effectively recovered the loss of SGNs in the mid and basal turns of the rat cochleae injected with ouabain. This treatment also allowed partial recovery of ABR threshold shift at higher frequencies (24 k to 32 k) 1 month after ouabain administration. These results support previous findings that BDNF expression can prevent HL caused by ototoxicity [49]. The partial recovery is found in high-frequency areas, correlating with the administration method that the basal turn has the most access to ouabain while the apex is further to reach. Future research should focus on optimizing AAV2-BDNF treatment (dosage and administration frequency) to achieve optimal recovery. The findings provide a novel, minimally invasive, targeted, and effective treatment for neuropathy and sensorineural HL.

In summary, our study demonstrates the feasibility of delivering genes to the cochleae via systemic administration combined with magnetic targeting, without the need for direct access to the inner ear. This approach offers the potential to minimize damage to critical structures within the middle and inner ear. While our results suggest partial restoration of HL in certain conditions, the overall effectiveness of this technique in fully reversing HL remains to be established. Additional preclinical studies, including experiments in larger animal models and non-human primates, are required to comprehensively evaluate the efficacy and safety of systemic AAV delivery with magnetic targeting.

Our findings provide a foundational step toward the development of non-invasive gene therapy strategies for HL, but significant challenges remain. Future research will focus on optimizing this method, including exploring detailed functional outcomes, refining dose selection, and further enhancing magnetic targeting to maximize therapeutic benefits. These efforts aim to advance the potential of systemic gene delivery to improve treatment strategies and quality of life for individuals with HL or related conditions.

Materials and Methods

Adeno-Associated Virus Vector, Serotypes, and Nanoparticles

Three recombinant AAV vectors, AAV2 Quad Mut, AAV2 pANC80L65, and AAV9 PHP.eB., were obtained from Ocular Gene Therapy (University of Florida). These recombinant vectors are packaged with rat BDNF transgene coding sequence with an internal ribosome entry site driven by chicken β actin (CBA) promoter. They also include expressing green fluorescent protein (GFP) receptor gene under the control of P2A promoter. Mag4C-Ad superparamagnetic iron oxide nanoparticles (SPIONs) (OZ, Bioscience Cat #: AKC11000); CA, USA) were used to capture the AAV vectors through electrostatic and hydrophobic interaction as per the manufacturer’s protocol. In brief, 7 µL of AAV2 Quad Mut (1.68 × 10^10 gc), 5 µL of AAV2 pANC80L65 (1.615 × 10^10 gc), and 3 µL of AAV9 PHP.eB (1.638 × 10^10 gc) were added to 20 µL of SPIONs and incubated in RT for 30 min before tail vein administration and magnetic targeting delivery. Preparation of vector capture and dilutions was performed in sterile conditions. It has already been established that the majority of systemically injected MSCs accumulate within pulmonary capillary beds (33). To limit this and make more AAV vectors available for accumulation within the cochlea, 25 mg of the vasodilator sodium nitroprusside (Sigma Cat # 228,710) diluted in 200 mL PBS was administered 5 min prior to MSC injection (33).

Animals and Tail Vein rAAV Injection

All animal care and handling, performed using accepted veterinary standards, were approved by the Canadian Council on Animal Care and the Sunnybrook Animal Care Committee Regulations. Wild-type Long Evans rats (obtained from Charles River Laboratories, Canada) were housed under standard conditions (12-h light: 12-h dark cycle) in temperature-controlled (22 °C ± 1 °C) rooms. Equal numbers of male and female rats aged 7 weeks at the onset of experimentation were used. The animals had access to normal commercial laboratory chow and tap water ad libitum throughout the experimental procedure.

Intravenous administration was achieved via tail vein (T/V) injection. Rats were warmed briefly using a heat lamp for 15 min to increase the visibility of tail veins prior to injection. Animals were randomly divided into 4 main groups depending on the type of AAV injected: (A) control animals receiving no AAV injection, (B) AAV2 Quad Mut, (C) AAV2 pANC80L65, (D) AAV9 PHP.eB. One hundred microliters of nitroferricyanide (50 µg) was injected initially into a lateral tail vein. After 5 min, 500 µL of captured AAV vector diluted in PBS was slowly injected via the same catheter, and the residual solution within the catheter was flushed with an additional 400 µL of PBS before pressure was applied at the injection site until bleeding cessation. The left ear of all AAV-treated groups (B, C, D) was placed near a 1.0 Tesla magnet (5 × 5 × 5 cm, K&J Magnetics, PA) for 30 min after T/V injection. The right ears were treated as contralateral controls of the treated animals.

Two cohorts of animals were set up, containing six rats per group. Two weeks post-tail-vein injection of virus, animals were deeply anesthetized using isoflurane followed by (a) Cohort 1: intra-cardiac perfusion with 4% PFA in PBS for cochlear collection and fixation or (b) Cohort 2: exsanguination of rats for cochleae and organ (lung, heart, liver, kidney, spleen, and brain) collection for gene expression and protein analysis. Figure 7 depicts a schematic representation of the experimental procedure.

Fig. 7.

Schematic diagram showing systemic delivery approach and evaluated parameters

Ouabain-Induced Hearing Loss Model

A separate group of animals was anesthetized using 2%–3% isoflurane for the injection procedure. The corneas were coated with an ophthalmic ointment to prevent drying. Endoscopy-guided intratympanic injection was performed with a 25G spinal needle at the posterosuperior quadrant of the ear drum. Both ears were injected with ouabain (10 mM). Approximately 0.5–0.8 mL of solution was injected in each ear. After injection, rats were placed on their side with the injected ear upward for 30 min to increase absorption through the round and oval windows. Control animals received saline. Animals also received ouabain by intraperitoneal injection (50 mg/kg) as a comparison group with the intratympanic injection group.

One day after the ouabain injection, animals underwent general anesthesia using 2–3% isoflurane for the tail vein AAV2 Quad Mut-GFP-BDNF delivery with magnetic targeting as described above. Animals were randomly divided into three main groups: (A) negative control animals receiving saline and no AAV injection, (B) positive control animals receiving ouabain and no AAV injection, and (C) animals receiving ouabain and AAV2 Quad Mut.

Cochlear Function Testing

ABRs were recorded using the Tucker Davis Technologies (TDT) System 3 RZ6 in an open field with individual ears tested separately as described previously with some modifications (13). All three groups of animals for ouabain and AAV treatment were subjected to functional ABR testing post injection at 1 week, 2 weeks, and 1 month time points. The animals were anesthetized with an intraperitoneal injection of ketamine (100 mg/kg) and xylazine (20 mg/kg) and then placed in a custom-made acoustically and electrically shielded chamber on a preheated non-electric heating pad (~ 37 °C). Animals’ eyes were kept moist using an ophthalmic ointment during anesthesia. Acoustic stimuli were delivered through the TDT acoustic system consisting of one speaker placed 5 cm from the ear canal.

All ABR stimuli and responses were measured using BioSigRZ software. During ABR measurement, 5 ms, single-channel cosine-squared gated tones were delivered 21 times/s at frequencies of 4 kHz, 8 kHz, 16 kHz, 24 kHz, and 32 kHz. For each frequency, the tone Levels ranged from 10 to 100 dB in 10 dB incremental intervals. Electrical responses were amplified 20× and averaged across 512 responses at each frequency and level combination. Responses were collected via inserting needle electrodes at the forehead (active electrode) below the pinna of the reference ear (reference electrode) and below the contralateral ear (ground electrode). Each wave was filtered in BioSIGRZ from 300 Hz to 3 kHz and NT at 50 Hz. To calculate wave 1 amplitude, the averaged ABR waveforms were imported into MATLAB (R2018a, The MathWorks, Inc., Natick, MA, USA). After the onset of the stimulus, each waveform was visually inspected for accuracy, and the ABR threshold was defined as the lowest sound level at which a reproducible waveform was identified following a visual inspection.

Cochlear Processing for Immunohistochemistry

Cochleae harvested after perfusion were fixed with 4% paraformaldehyde (PFA) for 2 h at room temperature (RT). After rinsing in phosphate buffered saline (PBS) 3 times for 5 min, the cochleae were decalcified with Acid-Free EDTA Decalcifier (Cancer Diagnostics Inc, Cat# EDT100) solution at RT for 48 h. Cochleae were then processed in one of 2 ways: (a) Whole Mounts: After Reissner’s membrane and lateral wall and tectorial membrane of cochleae were removed, the Organ of Corti was micro-dissected into three individual parts representing low (apex), middle (mid), and high (base) frequencies; or (b) Cryosections: Excess bones were trimmed from decalcified cochleae using a scalpel and forceps. Cochleae were rehydrated through a serial sucrose gradient, embedded in optimum cutting temperature medium (OCT) (Fisher Scientific Cat# 23,730,571), and serially sectioned coronally into 10 µm slices using a cryostat. All tissue sections were mounted on SuperFrost Plus microscope slides (ThermoFisher Scientific Cat#1,255,015, Pittsburgh, PA, USA). Slides were then stored at −80 °C until further use.

Immunofluorescence Staining and Imaging

Hair cell pieces or cryosections were blocked for 1 h at RT in 5% donkey serum. After blocking the non-specific sites, cochlear pieces were incubated with primary antibody (a) Anti-Myosin VIIa (Proteus bioscience, USA, Cat# 25–6790, used at 1:1000), (b) Anti-GFP (Goat polyclonal to GFP, Cat# Ab5450, used at 1:500), (c) Anti-BDNF (Alomone Cat#ANT-010, used at 1:400), and (d) Anti-Tubulin B3 (TUBB3 Cat#801,201, used at 1:1000) diluted in 1% donkey serum albumin overnight in a Humidified chamber at 4 °C. Hair cell pieces or cryosections were washed three times in PBS and incubated with species-specific secondary antibodies conjugated to Alexa488, Alexa555, and Alexa647 (1:1000) for 1 h in room temp. Samples/slides were washed three times with PBST for 5 min each and mounted in Prolong-gold with DAPI (ProLong® Gold antifade reagent with DAPI, Life technologies Cat# 501,128,966).

Images of AAV transduction cochleae were acquired using a Nikon confocal microscope. Objectives of 10×, 20×, and 40× digital zoom were used. Images were captured with the same setting and exposure time for all pictures.

RNA Isolation and RT-qPCR

At the end of the experimental period, rats were sacrificed, cochleae and small pieces of heart, lung, liver, kidney, spleen, and brain were collected on dry ice immediately and stored at −80 °C until further use.

For RNA isolation, cochleae and tissue samples were homogenized in 1 ml TRIzol reagent (Invitrogen Cat#15,596,026) using PolytronR PT 2100 homogenizer (Kinematica, Switzerland). Total RNA isolation was performed according to the method previously described with some modification (1). We briefly combined the advantages of both TRIzol and column purification by using the PureLink RNA mini kit (Invitrogen Cat# 12,183,025). Adding an extra step with chloroform phase separation followed by precipitation of the aqueous phase with one volume of ethanol 70% prior to being deposited on the column ensured that the RNA from the aqueous phase was free of protein and organic contaminants when purified through the column. Samples were then processed according to the manufacturer’s protocol.

Total RNA concentration and integrity were determined with a microgel bioanalyzer (Agilent Bioanalyzer 2100; Agilent, Mississauga, ON, Canada). RNA (1 µg) was treated with DNAse and reverse transcribed using qScript cDNA SuperMix (Quanta Bioscience Cat# CA101414-104), Gaithersburg, MD). The cDNA was amplified by Quantitative real-time PCR using the Taqman method with the ABI Prism 7500 PCR system (Applied Biosystems, Foster City, CA) according to the manufacturer’s instructions. RT-qPCR probe and primer sets for BDNF and GAPDH gene expression assays were purchased from Life Technologies. Relative gene expression was calculated using the delta delta CT method (Applied Biosystems).

Measurement of BDNF Protein in Brain, Liver, and Kidney

Tissue proteins were extracted using radioimmunoprecipitation assay (RIPA) buffer, and protein was quantified using (bicinchoninic acid assay) BCA method. Total BDNF protein in brain, heart, and kidney was measured using Total BDNF Quantikine ELISA kit (R & D Systems Cat# DBNT00).

Statistical Analysis

Results are reported as mean ± standard errors of the mean (SEM). N refers to the number of animals used. All statistical analyses were performed with GraphPad Prism Software version 9.05 (San Diego, CA, USA), and the data and figures were reported as mean ± standard error mean (SEM). Comparisons between groups were performed by two-way analysis of variance (ANOVA) followed by the Holm-Sidak multiple comparisons test. P-values < 0.05 were considered significant and represented as *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Author Contribution

ZF was involved in study design, programmed and implemented the experiment, collected data, performed analysis, interpreted results, wrote the first draft of the manuscript, and organized the manuscript edits and revisions. TNL contributed to the study of conception and design, interpreted the data, and provided detailed advice and comments on both the analysis and the previous manuscript versions. Both YS and ZZ conducted a portion of experiments and analysis and provided comments for previous manuscript versions.

Funding

This work presented in this manuscript was supported by the Koerner Hearing Regenerative Foundation.

Data Availability

The data used in this study are available from authors upon request.

Code Availability

Not applicable.

Declarations

Ethical Approval

All animal care and handling, performed using accepted veterinary standards, were approved by the Canadian Council on Animal Care and the Sunnybrook Animal Care Committee Regulations. Protocol #24616.

Competing Interests

The authors declare no competing interests.