Ultrasound-Mediated Gene Therapy in Alzheimer’s Disease Validated through In Vivo PET Imaging

12 Efficient, spatially selective delivery of adeno -associated virus (AAV) therapeutics to deep 13 brain structures remains a major challenge to gene therapy for Alzheimer’s disease (AD), owing to 14 limited transport across the blood-brain barrier (BBB) and poor penetration to target neurons. Here, 15 we establish an integrated, noninvasive imaging and therapy platform that combines microbubble-16 enhanced focused ultrasound (MB -FUS) with positron emission tomography/computed 17 tomography (PET/CT) to transiently modulate the BBB, enhance region -specific AAV delivery 18 following systemic dosing, and longitudinally track transduction in vivo. Optimized MB -FUS 19 achieved targeted hippocampal delivery of systemically administered AAV9 in healthy mice, 20 resulting in a 10 -fold enhancement of neuronal transduction as compared to non-FUS controls. 21 Importantly, longitudinal PET reporter gene imaging in the 5xFAD AD model demonstrated robust 22 brain AAV transduction that remained stable for at least seven months. Finally, to assess therapeutic 23 impact, we used brain -derived neurotrophic factor (BDNF) as a test cargo. MB -FUS-facilitated 24 delivery elevated BDNF expression in targeted regions and produced short -term improvements in 25 synaptic signaling in 5xFAD mice. Collectively, these results highlight MB -FUS as a next -26 generation delivery platform to overcome barriers to AAV therapeutic delivery in Alzheimer’s 27 disease and position longitudinal PET assessment as a critical, translatable tool for monitoring and 28 optimizing gene therapy. 29

30 Alzheimer’s disease (AD), the most common form of dementia, remains a major unmet medical 31 challenge, affecting over 32 million people worldwide. 1–3 Although recent FDA approvals of 32 amyloid-targeting drugs ( e.g., aducanumab, lecanemab, donanemab) have sparked hope, their 33 modest cognitive benefit and potential side effects underscore the need for approaches that act 34 beyond amyloid. 3–5 Gene therapy offers a complementary, mechanism -diverse strategy by 35 delivering transgenes that promote neuroprotection, restore synaptic and circuit function, enhance 36 autophagy-mediated clearance of disease -related proteins, and modulate disease -associated 37 pathways.6–8 Among available platforms, a deno-associated virus (AAV) vectors are particularly 38 attractive for central nervous system (CNS) applications owing to their high transduction efficiency, 39 low immunogenicity, and capacity for long-term transgene expression.7,9,10 40 A major obstacle to systemic AAV delivery is the blood-brain barrier (BBB), a tightly regulated 41 neurovascular interface that restricts macromolecular transport from the circulation into the CNS 42 parenchyma.11,12 Several alternative delivery routes have been evaluated clinically but each fall 43 short: stereotactic injection is invasive and limited by targeting error and restricted coverage within 44 the intended structure , whereas intra cerebrospinal fluid (CSF) delivery offers limited spatial 45 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 2 of 26 control and inefficient access to deep parenchyma .11,12 Engineered BBB -crossing capsids can 46 increase CNS delivery in select settings but again lack spatial targeting. 10,13–17 Beyond baseline 47 BBB permeability, AD -associated pathology, including BBB dysfunction, reduced transporter 48 expression (e.g., LRP1), amyloid accumulation, and neuroinflammation, introduces structural and 49 functional heterogeneity that adds complexity to vector delivery. 18,19 Whether this pathological 50 heterogeneity facilitates or impedes AAV transduction remains unclear, as regional permeability 51 changes may coexist with perivascular debris accumulation that restricts parenchymal access and 52 functional effectiveness.18 Together, these constraints motivate a delivery strategy that is 53 noninvasive and scalable while enabling efficient, region-specific vector delivery, robust neuronal 54 transduction, and minimal off-target biodistribution and peripheral toxicity. 55 Microbubble-enhanced focused ultrasound (MB -FUS) offers a unique noninvasive approach 56 for targeted drug delivery to the brain. Focused ultrasound (FUS) beams can be precisely directed 57 through the intact skull to specific brain regions with sub -millimeter accuracy. When combined 58 with intravenously administered microbubbles ( MB; lipid- or albumin-shelled gas pockets, 1 -10 59 μm diameter), low -intensity FUS induces controlled oscillations that exert mechanical stress in 60 brain vessels, transiently increasing BBB permeability (4-24 hours)20–26 and facilitating interstitial 61 transport27,28. This spatiotemporal control creates a therapeutic window for region -selective 62 delivery, and several preclinical studies have demonstrated enhanced AAV transduction in targeted 63 brain regions using MB-FUS.25,26,29–39 More importantly, completed and ongoing clinical trials have 64 demonstrated that FUS-mediated BBB opening is safe and feasible for AD treatment22–24,40–49, and 65 a recent study combining MRI-guided FUS with aducanumab reported greater amyloid-β reduction 66 in targeted regions 50, underscoring the platform's translational potential for targeted gene therapy 67 in AD. 68 Despite promising preclinical and clinical studies exploring MB-FUS-mediated AAV delivery, 69 critical gaps limit translation and optimization. Variations across studies in AAV serotype s, 70 dosages, FUS parameters, and target regions, without standardized quantification methods, make 71 direct comparisons and protocol optimization challenging. Moreover, existing readouts for 72 assessing AAV delivery have significant limitations. Endpoint histology requires tissue sacrifice 73 and cannot provide whole -organ quantification , while gadolinium -enhanced MRI and CSF 74 sampling provide only indirect evidence of BBB opening, not direct measures of transduction or 75 sustained transgene expression. Positron emission tomography (PET) reporter gene imaging 76 addresses this gap by offering whole-body, quantitative, and clinically translatable visualization of 77 transgene expression with high sensitivity and temporal resolution.51–53 We previously demonstrated 78 using PET imaging of radiolabeled AAV9 and PET reporter gene imaging that MB -FUS 79 substantially enhances vector delivery and transgene expression in targeted brain regions of healthy 80 mice.51 However, whether these findings translate to AD, where BBB dysfunction and pathological 81 heterogeneity may unpredictably alter delivery efficiency and transduction, and whether improved 82 delivery yields meaningful therapeutic efficacy remain unclear . Together, these limitations 83 underscore the need for noninvasive, quantitative imaging tools that can longitudinally monitor 84 transgene expression and assess transduction efficiency in diseased brains to guide protocol 85 optimization and therapeutic translation. 86 Here, we establish an integrated noninvasive platform combining MB-FUS with PET reporter 87 gene imaging to enable region -specific AAV delivery and longitudinal monitoring of transgene 88 expression in an AD murine model. We first optimized MB -FUS parameters using mathematical 89 modeling of microbubble dynamics and in vivo validation across two frequencies (0.8 MHz and 90 1.5 MHz) to maximize hippocampal targeting precision, delivery efficiency, and neuronal 91 transduction, assessed via contrast -enhanced MRI, longitudinal fluorescent reporter imaging, and 92 immunofluorescence. Critically, we then applied quantitative PET reporter gene imaging in both 93 healthy and 5xFAD AD model mice to directly test whether parameters optimized in healthy brains 94 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 3 of 26 maintain delivery and transduction efficiency in the context of AD -related BBB dysfunction and 95 pathological heterogeneity, demonstrating robust and sustained transgene expression in targeted 96 regions over seven months. Finally, to evaluate therapeutic potential, we delivered brain -derived 97 neurotrophic factor (BDNF) via MB-FUS-enhanced AAV9 and assessed functional improvements 98 in synaptic signaling and neuroprotection in the 5xFAD hippocampus. Together, this work validates 99 MB-FUS as a scalable, clinically translatable strategy for targeted gene therapy in AD and positions 100 quantitative PET imaging as a critical tool for optimizing delivery protocols across disease contexts 101 and monitoring treatment efficacy in vivo. 102

103 Clinically translatable multi -frequency FUS systems achieve region -specific and safe AAV 104 transduction in the healthy brain. 105 To optimize MB -FUS parameters for hippocampal targeting, we first confirmed that 5 µm 106 monodisperse lipid-shelled microbubbles are more potent in increasing BBB permeability than 2 107 µm microbubbles, as assessed by contrast-enhanced MRI and dextran extravasation (Suppl. Method 108 1; Suppl. Fig. 1 ), consistent with prior reports. 54–59 Therefore, we used 5 µm microbubbles 109 throughout this work. Next, to identify a frequency that balances effective skull penetration with 110 robust microbubble-mediated BBB modulation and is compatible with clinical transcranial FUS 111 systems, we tested 0.8 MHz and 1.5 MHz in parallel (Suppl. Fig. 2). To mitigate differences in 112 Fig. 1. Multi-frequency ultrasound systems enable BBB opening for targeted AAV delivery. A) Schematic of microbubble-enhanced focused ultrasound for improved targeted AAV delivery to the hippocampus. B) Measurement of ultrasound insertion loss through the mouse skull at 0.8 MHz and 1.5 MHz from 2 to 12 months of age. C) Mathematical modeling of microbubble oscillation s inside a 1 0 μm vessel excited with 0.8 MHz and 1.5 MHz frequencies at 175kPa peak negative pressure . D) Representative power levels of acous tic emissions originating from MB oscillations during a typical 120s treatment using 0.8 MHz (upper) and 1.5 MHz (lower) frequency. Emissions are processed in the frequency domain in specific bands (H: harmonics, UH: ultra-harmonics, B: broadband). E) Representative contrast enhanced T1-weighted MR image 10 minutes after MB-FUS treatment using 0.8 MHz (upper) and 1.5 MHz (lower) frequency. F) Representative Hematoxylin-eosin (H&E) staining images of the brain treated with 0.8 MHz (upper) and 1.5 MHz (lower) frequency. G) Reverse transcription quantitative polymerase chain reaction (RT-qPCR) assessment of ssDNA of AAV9 transgene (tdTomato) in brain hemispheres at 21 hours after the treatment. P-values were determined by one-way analysis of variance (ANOVA) and were adjusted using Bonferroni correction. Plots show mean ± S.D (n = 4). *P ≤ 0.05, **P ≤ 0.01, **** P ≤ 0.0001. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 4 of 26 focal dimensions between frequencies , we used phased -array electronic steering to deliver 113 volumetric sonication for 1.5 MHz (see Methods), enabling more direct frequency comparison. Our 114 FUS system enables precise focal targeting with real -time monitoring of MB activity via passive 115 acoustic mapping (PAM) during sonication (Suppl. Fig. 2). As expected, 0.8 MHz showed a 116 reduced skull attenuation compared to 1.5 MHz (Fig. 1B), though insertion loss increased with 117 mouse age, particularly in animals younger than 3 months, likely due to ongoing skull maturation 118 (Fig. 1B). To predict how frequency influences 5 µm MB oscillation dynamics and resulting 119 bioeffects within brain vasculature, we simulated MB oscillations within brain vasculature using a 120 mathematical model based on the Marmottant model at two frequencies: 0.8 MHz and 1.5 MHz.55,60 121 The model predicted that MB excited at 0.8 MHz oscillate at relative greater amplitude compared 122 to 1.5 MHz (ΔR/R₀ of 16% vs. 3%, respectively) under the same acoustic pressures (Fig. 1C), 123 suggesting that lower -frequency ultrasound may induce stronger mechanical effects on the 124 vasculature and enhance therapeutic delivery. To test this prediction in vivo, we performed MB -125 FUS treatment targeting the right hippocampus, a brain region critical for memory and cognition 126 and one of the earliest affected in AD. FUS was applied during intravenous microbubble injection 127 (2.5 × 10⁸ MBs/kg) using a peak negative pressure of 320 kPa ( Supple. Method 2; 1 ms pulses, 5 128 Hz PRF), calibrated for skull insertion loss at both frequencies. During sonication, we monitored 129 MBs acoustic emissions using passive acoustic monitoring, confirming MB oscillations through 130 the detection of signature harmonic signals (Fig. 1D ; Suppl. Fig. 2). Moreover, we observed 131 minimal broadband emissions at both frequencies throughout the treatment, indicating that MB 132 underwent stable oscillations rather than inertial cavitation. This is critical for safety, as broadband 133 emissions are associated with microbubble collapse and potential hemorrhaging (Fig. 1D). Ten 134 minutes after MB -FUS treatment, T1 -weighted magnetic resonance imaging (MRI) following 135 intravenous injection of a gadolinium -based contrast agent confirmed increased vascular 136 permeability in the hippocampus, as evidenced by contrast extravasation (Fig. 1E) . This result 137 validates both the enhanced permeability and the spatial precision of our targeting approach at both 138 ultrasound frequencies setups. Moreover, histological analysis of brain slices showed that the BBB 139 opening in the FUS-targeted areas did not result in hemorrhage, further demonstrating the safety of 140 the procedure at this pressure (Fig. 1F). 141 After confirming targeting precision, we assessed the ability of MB -FUS to enhance delivery 142 of intravenously administered AAV9 (3 × 1013 vg/kg), selected for its established CNS tropism 143 following systemic administration, to the brain using both frequencies. MB -FUS was started 144 immediately following systemic AAV9 injection. Quantitative PCR analysis of AAV9 single -145 stranded DNA copies in the sonicated hemisphere 21 hours post -treatment revealed significant 146 increases in vector genome delivery: 25-fold at 0.8 MHz (p < 0.0001) and 15-fold at 1.5 MHz (p = 147 0.0028) compared to AAV -only controls (Fig. 1G). The greater enhancement at 0.8 MHz is 148 consistent with the 5 µm MB oscillation amplitude predicted by our mathematical model. Together, 149 these findings establish our multi -frequency FUS platform, particularly at 0.8 MHz, as a safe and 150 effective strategy for region-specific AAV delivery in the healthy brain. 151 Neuron-specific and liver-detargeted AAVs transduction achieved by MB -FUS revealed by 152 fluorescent reporter imaging 153 Having established that our MB -FUS systems achieve enhanced, targeted AAV9 delivery, we 154 next assessed transduction efficiency in healthy brains following MB -FUS treatment using a 155 fluorescent reporter gene, tdTomato (Fig. 2A). In vivo optical imaging revealed that AAV9 156 injection alone produced moderate tdTomato expression in the brain compared to non-AAV, non-157 FUS controls. However, MB-FUS treatment at both 0.8 MHz and 1.5 MHz substantially enhanced 158 transgene expression relative to AAV alone, with elevated levels sustained for up to 6 months (Fig. 159 2B-C; Suppl. Fig. 3). Notably, while both frequencies yielded comparable transduction within the 160 first month, the 0.8 MHz group demonstrated consistently higher tdTomato expression thereafter, 161 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 5 of 26 indicating a potential long-term advantage (Fig. 2C; Suppl. Fig. 3). Ex vivo brain analysis at 1 and 162 7 months after treatment confirmed these findings, with significantly enhanced expression in both 163 MB-FUS groups compared to non-FUS controls and higher transduction with 0.8 MHz at 7 months 164 (Fig. 2D-F). 165 Interestingly, MB-FUS treatment markedly reduced off -target AAV9 expression in the liver, 166 potentially addressing a major challenge in systemic AAV gene therapy, where AAV liver tropism 167 raises concerns about hepatotoxicity, immune responses, and dose-limiting toxicity (Fig. 2C; Suppl. 168 Fig. 2. In vivo fluorescent reporter gene imaging reveals enhanced and stable brain -targeted AAV9 transduction with liver detargeting following MB -FUS treatment. A) In vivo experimental protocol for AAV delivery with MB-FUS. B) Representative in vivo imaging of AAV9-tdTomato expression in the brain at 2 weeks, 1 month, and 3 months after treatment. C) Quantification of in vivo imaging of maximum tdTomato expression in the brain (left) and liver (right) over a 6 -month period following treatment. D) Representative ex vivo images of tdTomato transduction biodistribution 1 month after treatment. E) Quantification of ex vivo imaging of tdTomato expression in the brain 1 month after treatment. P-values were determined by one -way analysis of variance (ANOVA) and were adjusted using Bonferroni correction. F) Quantification of ex vivo imaging of tdTomato transduction biodistribution 1 month (left) and 7 months (right) after treatment. P-values were determined by two- way analysis of variance (ANOVA) and were adjusted using Bonferroni correction. Plots show mean ± S.D (n=5). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001. Abbreviations: tdT: tdTomato. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 6 of 26 Fig. 4).61 In contrast to the sustained transduction observed in the brain, liver expression in AAV9-169 treated groups peaked at 1 month and subsequently declined. Unexpectedly, both in vivo and ex 170 vivo optical imaging revealed that liver tdTomato expression was selectively reduced in the 0.8 171 MHz MB-FUS group compared to other treatment groups, with the most pronounced reduction 172 observed between 1 and 2 months post -treatment ( Fig. 2C -D, F ; Suppl. Fig. 4 ). Quantitative 173 analysis at 1 month confirmed that while the liver was the primary site of AAV9 accumulation in 174 non-FUS controls (liver-to-brain ratio: 25 ± 4.6; maximum intensity), 0.8 MHz MB-FUS reversed 175 this biodistribution, achieving a brain-to-liver ratio of 3.6 ± 0.8 (p = 0.013 vs. non -FUS; Fig. 2F). 176 Although further investigation is needed to elucidate the underlying mechanism, these findings 177 suggest that MB -FUS not only enables region -specific AAV9 delivery but may also reduce 178 peripheral exposure and systemic toxicity by limiting off-target hepatic accumulation. 179 Fig. 3. Immunofluorescent microscopy demonstrates that MB -FUS-enhanced delivery enables spatially targeted and neuron-specific AAV transduction in the brain. A) Representative fluorescent microscopy images showing AAV9-tdTomato transduction in the brains of healthy mice at 6 months post-treatment. From left to right: untreated control, MB-FUS only, AAV only, AAV + MB -FUS at 0.8 MHz, and AAV + MB -FUS at 1.5 MHz. B) Representative immunofluorescent staining of neurons and blood vessels at 6 months post-treatment, demonstrating cell-specific AAV transduction. Green: tdTomato; Ma genta: NeuN (neuronal nuclei); Red: Lectin (vasculature); Blue: DAPI (nuclei). C) Quantification of brain tdTomato expression across treatment groups at 6 months post - treatment. (n = 4) D) Quantification of neuron -specific transduction, expressed as the pe rcentage of NeuN⁺ cells among all tdTomato⁺ cells in the brain at 6 months post -treatment. (n = 3) E) Quantification of vessel cell -specific transduction, expressed as the percentage of lectin⁺ cells among all tdTomato⁺ cells in the brain at 6 months post - treatment. (n = 3) P -values were determined by one -way analysis of variance (ANOVA) and were adjusted using Bonferroni correction. Plots show mean ± S.D. n.s. no significance , P >0.05, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, **** P ≤ 0.0001. Abbreviations: HC: Hippocampus. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 7 of 26 To evaluate the spatial and cell -type specificity of AAV9 transduction, we performed 180 immunofluorescence analysis on brain sections harvested at 7 months post -treatment. Confocal 181 microscopy revealed robust tdTomato expression in the hippocampus of MB -FUS-treated mice at 182 both 0.8 MHz and 1.5 MHz, confirming effective transduction and precise regional targeting (Fig. 183 3A, C ). Consistent with our prior observations 51, we also detected transgene expression in the 184 contralateral (non -sonicated) hippocampus, suggesting potential vector diffusion or axonal 185 transport via interhemispheric connections following MB-FUS-mediated delivery62 (Fig. 3A). To 186 elucidate cellular specificity, we quantified colocalization of AAV transduction (tdTomato) with 187 neurons (NeuN) and vasculature (lectin), revealing a marked increase in neuronal transduction in 188 MB-FUS-treated animals at both frequencies (0.8 MHz: 37.4% ± 6.5%; 1.5 MHz: 36% ± 9.1%) 189 compared to the AAV-only group (4.2% ± 0.4%, p < 0.05), representing a 9-fold enhancement (Fig. 190 3B, D). Conversely, the majority of transduced cells in the AAV -only group were associated with 191 blood vessels (28% ± 2.3% lectin⁺ vs. 4.2% ± 0.4% NeuN⁺; Fig. 3D -E), indicating limited 192 parenchymal penetration and perivascular confinement. These findings demonstrate that MB-FUS 193 overcomes vascular barriers to facilitate AAV9 extravasation into the brain parenchyma, thereby 194 enhancing neuronal transduction efficiency. This shift from endothelial -restricted to neuron -195 preferential targeting underscores the capacity of MB-FUS to modulate vector biodistribution at the 196 cellular level and improve therapeutic specificity in CNS gene delivery. 197 PET/CT reporter gene imaging enables longitudinal in vivo monitoring of gene expression 198 and reveals enhanced AAV9 transduction following MB -FUS treatment in an Alzheimer’s 199 disease mouse model. 200 While fluorescent reporter gene imaging provides valuable information in small animal models, 201 its limited tissue penetration, sensitivity, and quantification accuracy restrict its translational 202 potential. In contrast, PET/CT imaging offers deep tissue penetration, high sensitivity, and 203 quantitative capability, making it a clinically translatable modality for evaluating AAV transduction 204 efficiency and spatial distribution. Importantly, PET enables direct comparison of MB -FUS 205 performance across healthy and diseased brains, addressing the critical question of whether AD -206 related BBB dysfunction and pathological heterogeneity alter the transduction efficiency. To 207 leverage these advantages, we developed a PET-based method to assess MB-FUS-enhanced AAV9 208 transduction in both healthy and 5xFAD mice. Specifically, we engineered AAV9 vectors encoding 209 the PET reporter gene pyruvate kinase M2 (PKM2) and employed [¹⁸F]DASA -10, a BBB -210 permeable radiotracer51,53, to noninvasively monitor and quantify PKM2 expression in vivo. 211 Following the same treatment protocol, mice received MB -FUS and AAV9-PKM2 delivery at 212 three months of age (Fig 2A). Six months post -treatment, [¹⁸F]DASA-10 PET imaging revealed 213 significantly enhanced AAV9-PKM2 expression in the sonicated brain regions of both healthy and 214 5xFAD mice (Fig 4; Suppl. Fig. 5 ). In healthy mice, standardized uptake values (SUVs) in the 215 targeted regions were 0. 64 ± 0.08 for the 0.8 MHz group and 0. 63 ± 0.27 for the 1.5 MHz group, 216 both significantly higher than the 0.20 ± 0.0 6 observed in non -FUS controls (p = 0.0003 and p = 217 0.0005, respectively) (Fig 4A -B). More importantly , 5xFAD mice showed similar trend of 218 increased SUVs of 0.89 ± 0.13 (0.8 MHz) and 0.84 ± 0.12 (1.5 MHz), compared to 0.49 ± 0.07 in 219 non-FUS controls (p = 0.0004 and p = 0.0028, respectively), indicating enhanced and sustained 220 transgene expression in diseased brain environment with MB -FUS (Fig. 4D -E). Notably, SUVs 221 were consistently higher in 5xFAD mice than in healthy mice across all groups, potentially 222 reflecting both increased baseline permeability in the diseased vasculature and endogenous PKM2 223 upregulation associated with AD pathology63. qPCR analysis of PKM2 mRNA levels in the treated 224 hemisphere confirmed the PET imaging findings. In healthy mice, MB -FUS showed higher 225 expression with 25 -fold (0.8 MHz) and 16 -fold (1.5 MHz) increases vs. non -FUS controls (p < 226 0.003 for both frequencies; Fig. 4C). In 5xFAD mice, enhancement was even more pronounced 227 with 0.8 MHz, yielding a 45-fold increase in expression (p = 0.0017; Fig. 4F). Across both mouse 228 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 8 of 26 models, 0.8 MHz yielded significantly higher expression than 1.5 MHz (p < 0.05), consistent with 229 the frequency-dependent enhancement observed in optical imaging quantification. Together, these 230

validate PET imaging as a noninvasive and clinically translatable approach for tracking 231 AAV transduction and demonstrate effective, region-specific gene delivery with MB-FUS in both 232 healthy and AD (5xFAD) mouse models. 233 Fig. 4. In vivo PET/CT imaging with [¹⁸F]DASA-10 quantifies AAV9-mediated PKM2 transduction following MB-FUS in both healthy and 5xFAD Alzheimer’s model mice. A) Representative PET/CT images (coronal, axial, sagittal) obtained immediately following [¹⁸F]DASA -10 administration, 6 months after MB -FUS-assisted AAV9-PKM2 delivery in healthy mice. Regions of MB -FUS treated region are marked with dotted green circles. B) Image-based quantification of standardized uptake values (SUVs) in regions of interest across treatment groups in healthy mice following [¹⁸F]DASA-10 injection. C) Reverse transcription quantitative polymerase chain reaction (RT-qPCR) quantification of PKM2 mRNA levels in MB -FUS-treated brain hemispheres of healthy mice 6 month following MB-FUS-mediated AAV9-PKM2 delivery. D) Representative PET/CT images (coronal, axial, sagittal) obtained immediately following [¹⁸F]DASA -10 administration, 6 months after MB -FUS-assisted AAV9 -PKM2 delivery in 5xFAD mice. Regions of MB-FUS treated region are marked with dotted green circles. E) Image-based quantification of standardized uptake values (SUVs) in regions of interest across treatment groups in 5xFAD mice following [¹⁸F]DASA-10 injection. F) RT-qPCR quantification of PKM2 mRNA levels in MB -FUS-treated brain hemispheres of 5xFAD mice 6 month following MB -FUS-mediated AAV9 -PKM2 delivery. P -values were determined by one-way analysis of variance (ANOVA) and were adjusted using Bonferroni correct ion. Plots show mean ± S.D. (n = 4-5) n.s. no significance, P >0.05, *P ≤ 0.05, **P ≤ 0.01, **** P ≤ 0.0001. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 9 of 26 MB-FUS-enhanced AAV9:BDNF delivery produces transient synaptic improvements in 234 5xFAD mice 235 To evaluate the therapeutic potential of MB-FUS-mediated gene delivery, we assessed whether 236 enhanced hippocampal delivery of AAV9 encoding brain -derived neurotrophic factor (BDNF) 237 could improve synaptic function in 5xFAD mice. BDNF was selected as a therapeutic payload due 238 to its established capacity to reduce neuronal death, enhance synaptic signaling, and improve 239 hippocampal-dependent memory in preclinical AD models 64–69, which has led to its recent 240 advancement into a first -in-human AAV-based gene therapy trial (NCT05040217). Quantitative 241 PCR confirmed robust BDNF transgene expression following MB -FUS treatment, with 50 -112-242 fold improvement vs. non-FUS controls observed at 1 and 7 months post-treatment (p < 0.05; Fig. 243 5A, E). Importantly, at 1 -month post -treatment, MB -FUS-treated animals showed a significant 244 increase in the presynaptic marker synapsin 1 (Syn1) compared to both AAV -only and non -245 treatment controls (Fig. 5B), alongside trends toward elevated postsynaptic markers synaptophysin 246 (Syp) and PSD-95 (DLG4), though not statistically significant (Fig. 5C–D). By 7 months, synaptic 247 marker levels had returned to baseline across all treatment groups (Fig. 5F –H). Together, these 248 findings provide proof-of-concept that MB-FUS-enhanced AAV9-BDNF delivery can elicit early 249 synaptic improvements in 5xFAD mice, and suggest that optimized dosing, earlier intervention, or 250 sustained expression strategies may be required to achieve long -term functional rescue in AD 251 models. 252 Fig. 5. MB-FUS-enhanced AAV9-BDNF delivery produces transient synaptic improvements in 5xFAD mice. A) Reverse transcription quantitative polymerase chain reaction (RT -qPCR) quantification of BDNF mRNA levels in the treated hemisphere of 5xFAD mice 1 month following MB-FUS-mediated AAV9-BDNF delivery. RT-qPCR quantification of the B) presynaptic marker synapsin 1 (Syn1), C-D) postsynaptic markers synaptophysin (Syp) and PSD-95 (Dlg4) one month post treatment. E) RT -qPCR quantification of BDNF mRNA levels in the hipp ocampus 7 months following MB -FUS-mediated AAV9 -BDNF delivery. RT -qPCR quantification of synaptic markers F) Syn1, G) Syp, and H) Dlg4 at 7 months post -treatment. P-values were determined by one-way analysis of variance (ANOVA) and were adjusted using Bonf erroni correction. Plots show mean ± S.D. (n = 3) n.s. no significance P >0.05, *P ≤ 0.05, **P ≤ 0.01. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 10 of 26

253 In this study, we established an integrated platform combining microbubble-enhanced focused 254 ultrasound (MB-FUS) with positron emission tomography (PET) reporter gene imaging to enable 255 noninvasive, region - and cell -specific AAV delivery and longitudinal monitoring of transgene 256 expression in AD. Through systematic optimization, we identified 0.8 MHz as the preferred 257 frequency for hippocampal targeting, achieving up to 25 -fold enhancement in vector delivery, 9 -258 fold improvement in neuronal transduction, and sustained expression over 7 months in healthy mice, 259 with reduced off -target liver accumulation. Critically, quantitative PET imaging confirmed that 260 delivery parameters optimized in healthy brains translated effectively to the AD context, 261 demonstrating durable transgene expression despite BBB dysfunction and pathological 262 heterogeneity. Moreover, proof-of-concept therapeutic testing with BDNF revealed early synaptic 263 improvements, validating the platform's capacity to deliver functional therapeutics. Together, these 264 findings establish MB-FUS with PET monitoring as a clinically translatable strategy for targeted, 265 quantifiable gene therapy in neurodegenerative disease. 266 A central finding of this work is the optimization of MB-FUS parameters to 0.8 MHz based on 267 mathematical modeling and in vivo validation. Modeling predicted that 5 µm microbubbles driven 268 at 0.8 MHz would oscillate at greater amplitude than at 1.5 MHz under equivalent acoustic pressures, 269 which we confirmed experimentally through enhanced vector delivery, sustained neuronal 270 transduction, and reduced hepatic off -target expression. Importantly, this frequency bridges a 271 translational gap between preclinical and clinical practice. While preclinical studies typically 272 employ frequencies above 1 MHz for smaller focal zones and better precision, clinical transcranial 273 FUS systems operate at 200 kHz -1 MHz to minimize skull attenuation, enhance penetration, and 274 reduce heating risk.70 Since FUS parameters critically influence bioeffects in a nonlinear manner55, 275 using a frequency compatible with both preclinical models and clinical systems is essential for 276 scalable translation. Our finding that 0.8 MHz achieves robust, neuron -specific hippocampal 277 delivery while balancing penetration and spatial precision suggests that this frequency can bridge 278 preclinical and clinical applications, allowing optimized protocols to translate directly to patients. 279 The enhanced delivery achieved by MB-FUS translated not only to increased overall transgene 280 expression but also to a shift in cellular tropism, redirecting systemically administered AAV9 from 281 predominantly perivascular and glial cells to robust neuronal transduction in adult mice. Our AAV-282 only controls showed less than 5% neuronal labeling, consistent with prior reports and providing 283 further evidence that this limitation stems from perivascular trapping by mature astrocytic endfeet 284 that restricts parenchymal penetration. 71–73 Critically, this suggests that limited neuronal 285 transduction reflects a delivery barrier that restricts extravasation and diffusion, rather than inherent 286 capsid tropism. By transiently modulating BBB permeability and enhancing interstitial transport, 287 MB-FUS overcomes perivascular entrapment, enabling deep neuronal access and converting a 288 systemically scalable vector into a regionally targeted, neuron-specific platform. 289 A critical challenge for clinical translation is determining whether parameters optimized in 290 healthy brains maintain efficacy in the pathological context of AD. To address this, we integrated 291 quantitative PET reporter gene imaging, which provides noninvasive, longitudinal assessment of 292 transgene expression and addresses a critical gap in the field by providing standardized methods to 293 monitor AAV transduction in vivo. Prior preclinical studies have relied on endpoint histology or 294 indirect readouts such as gadolinium -enhanced MRI, which provide limited quantitative 295 information and cannot assess longitudinal transgene expression or compare delivery efficiency 296 across cohorts and disease contexts. By employing PET imaging of the reporter gene PKM2 with 297 the BBB-permeable radiotracer [¹⁸F]DASA -10, we demonstrated for the first time that MB -FUS 298 parameters optimized in healthy mice translate effectively to the 5xFAD AD model, yielding robust 299 and stable transgene expression over 7 months despite BBB dysfunction and pathological 300 heterogeneity. Critically, molecular confirmation via qPCR revealed 45 -112-fold enhancement in 301 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 11 of 26 AAV transduction in the 5xFAD brain across reporter and therapeutic transgenes compared to non-302 FUS controls, suggesting that combining focal BBB modulation with systemic administration could 303 enable substantially lower vector doses while maintaining therapeutic efficacy, thereby reducing 304 systemic exposure, off-target toxicity, and manufacturing costs. Beyond technical validation, PET 305 imaging offers translational potential by enabling systematic dose optimization, facilitating cross -306 laboratory protocol comparison, and providing a clinically compatible platform for real -time 307 monitoring and individualized therapy adjustment in patients. Importantly, this approach is broadly 308 applicable beyond Alzheimer's disease to other CNS disorders requiring region -specific gene 309 delivery, including Parkinson's disease, epilepsy, and glioblastoma, positioning PET -guided MB-310 FUS as a generalizable strategy for precision gene therapy in the brain. 311 To assess therapeutic potential, we delivered AAV9 -BDNF via MB-FUS in 5xFAD mice and 312 observed sustained transgene expression over 7 months with early synaptic improvements, but these 313 effects were not maintained at later timepoints. While BDNF has shown neuroprotective efficacy 314 in prior preclinical studies 64–69, our findings in the 5xFAD model revealed only transient benefit, 315 highlighting that successful delivery does not automatically ensure sustained therapeutic efficacy. 316 These results provide proof -of-concept that the MB -FUS platform can deliver functional 317 therapeutics and elicit measurable biological responses in the AD brain, while also revealing key 318 areas for optimization and further investigation . Several factors may contribute to the transient 319 therapeutic effects observed. BDNF dose was not optimized in this study, and the relationship 320 between vector dose, sustained BDNF expression, and long-term functional outcomes in the 5xFAD 321 brain remains to be defined. Importantly, the PET reporter imaging platform established here 322 enables noninvasive, longitudinal quantification of AAV transduction across different vector doses, 323 facilitating systematic dose -titration studies to identify optimal dosing regimens that maximize 324 therapeutic benefit, with direct translatability to individualized patients dosing in clinical trials. 325 Moreover, the 5xFAD model is an aggressive, amyloid -driven system in which ongoing synapse 326 loss may outpace BDNF-mediated rescue, particularly at later disease stages. Testing in tau-based, 327 neuroinflammation-driven, or apolipoprotein E4 (APOE4) knock-in models will be essential to 328 assess platform performance across the heterogeneity of AD pathology. Additional limitations 329 include the mechanisms underlying the observed reduction in liver transduction at 0.8 MHz, which 330 remain unclear and warrant further investigation. Potential explanations include altered AAV 331 pharmacokinetics, increased brain accumulation reducing hepatic bioavailability, or frequency -332 dependent effects on systemic circulation, but rigorous biodistribution studies will be needed to 333 distinguish these possibilities. 334 Collectively, our findings demonstrate that MB -FUS with PET monitoring addresses critical 335 barriers to CNS gene therapy by enabling noninvasive, region -specific delivery with real -time 336 quantification, and provides a scalable, clinically viable platform for therapeutic development in 337 Alzheimer's disease and beyond. 338

339 In vivo experiments 340 All animal experiments were conducted with a protocol approved by the Institutional Animal 341 Care & Use Committee (IACUC) at Stanford University. Female mice aged 3 to 9 months were 342 used in this study, with treatments initiated at 3 months of age and longitudinal monitoring until 343 study endpoints at 9 months. C57BL/6 mice (Charles River Laboratories) were used to characterize 344 BBB opening induced by multi -frequency MB-FUS. B6 albino mice (Jackson Laboratory) were 345 used to assess AAV9-tdTomato transduction via in vivo and ex vivo optical imaging. To investigate 346 AAV9 transduction and therapeutic efficacy in the context of AD, 5xFAD transgenic mice and age-347 matched B6SJLF1/J wild -type controls (Jackson Laboratory) were used. These cohorts were 348 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 12 of 26 subjected to AAV9-mediated delivery of either PKM2 for PET/CT imaging or BDNF for functional 349 therapeutic evaluation, with or without MB-FUS. 350 Focused ultrasound system and transducer calibration 351 We developed custom multi -frequency FUS systems operating at 0.8 MHz and 1.5 MHz to 352 investigate the effect of ultrasound frequency on MB -FUS-induced BBB opening and to assess 353 whether a lower -frequency, which is more amenable to clinical translation, can support efficient 354 and robust AAV-mediated gene delivery in the context of AD. The 0.8 MHz system consists of a 355 single-element therapeutic transducer ( H-115, Sonic Concepts, WA; fundamental frequency: 250 356 kHz; third harmonic frequency with matching network: 829 kHz ) driven at 829 kHz with an 357 arbitrary waveform generator (model 33500B, Agilent) and a 40 -dB power amplifier (UltraX10, 358 E&I) and a passive cavitation detector (5 MHz single element, unfocused, 3.175 -mm aperture) 359 oriented toward the acoustic focus of the therapeutic transducer with a 45° angle to reduce 360 sensitivity to direct transmit beam reflection . The 1.5 MHz system , as previously described 51, 361 features a custom 128-element 2D therapeutic array (center frequency: 1.5 MHz) integrated with 362 an L12-5 linear imaging transducer (38 mm aperture, Philips/ATL) positioned within the array’s 363 central opening. A Vantage 256 ultrasound system (Verasonics) was used to do both image 364 guidance, FUS therapy and acoustic emission recording during treatment. To account for smaller 365 focal dimensions at 1.5 MHz, we expanded the treated region by raster scanning the 1.5 MHz focus 366 over a 5 × 5 grid with 0.5 mm spacing (total sonicated volume ~2.5 × 2.5 × 2.7 mm³ ≈ 17 mm³), 367 allowing easier comparison between frequencies. Real-time analysis of acoustic emissions via 368 passive cavitation detection (PCD) was employed to characterize MB oscillations throughout 369 sonication for both frequencies. With the 1.5 MHz setup, the imaging transducer was also used to 370 generate passive acoustic maps to localize acoustic emissions.74 This feedback mechanism allowed 371 us to monitor cavitation dynamics and ensure sonications remained within predefined safety 372 thresholds. The mechanical index (MI) during imaging was maintained at or below 0.35 to ensure 373 safety. 374 The therapeutic transducers were calibrated both in free-field and transcranial conditions with 375 a calibrated needle hydrophone (HNP0400, Onda, Sunnyvale, CA, USA). For accurate comparison 376 between the trans -skull pressure between 0.8 and 1.5 MHz, transcranial focal pressure was used 377 throughout the study . The transcranial focal pressure was obtained by locating the focus using 378 hydrophone and 3D positioning system after placing a mouse skull harvested across age from 2-12 379 month in between the transducer and hydrophone. The vertical position of skull was carefully 380 adjusted using pulse/echo so that the geometric focus of transducer was at approximately 4 mm 381 inside the skull structure; lateral position of skull was also adjusted to a similar location used to 382 target mouse hippocampus. At each frequency and conditions, a 30 cycle pulse was used, and the 383 maximum peak-to-peak pressures were used to obtain calibration curve. 384 Mathematical modeling of microbubble dynamics 385 To investigate the frequency -dependent behavior of microbubble (MB) dynamics in brain 386 vasculature, we used a mathematical model that simulates ultrasound-driven MB oscillations, fluid 387 transport across the vessel wall, and interstitial flow. 55 MBs were modeled using a modified 388 Rayleigh–Plesset equation that incorporates the non-linear effect of bubble oscillation due to its 389 shell.60,75–77 To capture interactions with the vessel wall and surrounding fluid, the model was 390 implemented in a finite element model comprising a lumen, vascular wall, and interstitial space , 391 with the microbubble centrally positioned within the vessel . MB oscillations were coupled to the 392 surrounding fluid via a pressure boundary condition, and vessel fluid dynamics was governed by 393 the incompressible Navier –Stokes equations. The simulations were performed using the 394 commercial finite element software, COMSOL (version 6.1, Burlington, MA, USA) , where 395 necessary equations were added using the Mathematics module. 396 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 13 of 26 Microbubble preparation and characterization 397 The MBs used in this study were lipid -shelled MBs produced in -house 398 (distearoylphosphatidylcholine (DSPC):1,2distearoyl -sn-glycero-3-phosphoethanolamine-N-[ma 399 leimide(polyethylene glycol)-2000 (DSPE -PEG2000); 90:10 molecular ratio). To independently 400 assess the impact of ultrasound frequency on the bioeffect and to ensure robust and consistent BBB 401 opening, we utilized monodisperse MBs with a 5 μm diameter. These MBs were isolated through 402 size-selective centrifugation of freshly activated polydisperse solution . For each 20 g mouse, a 403 bolus injection of 50 μL containing 5 × 10⁶ MBs (equivalent to 2.5 × 10⁸ MBs/kg) was administered 404 via the tail vein. To make sure the MB sizes and number of MBs selected were consistent using this 405 method, the selected MB were counted and characterized by AccuSizer 770A Optical Particle Sizer 406 (Santa Barbara, CA, USA) in triplicates before every MB-FUS treatment. 407 Focused ultrasound treatment procedure 408 For the MB-FUS experiments, the animal was placed in the supine position with its head held 409 by a stereotaxic frame designed in-house and attached to a 3D stage for fine positioning. To achieve 410 precise FUS targeting, the hippocampal region of interest (ROI) in the right hemisphere was 411 identified using ultrasound guidance for the 1.5 MHz setup and a needle-based alignment approach 412 for the 0.8 MHz system. Target localization was guided by the skull contour and cross -referenced 413 with anatomical landmarks from the Allen Mouse Brain Atlas and Allen Reference Atlas. To assess 414 AAVs delivery, AAVs vectors were injected intravenously (i.v.) 1 minute before the MB -FUS 415 treatment. The following exposure settings were employed: 1.5 MHz/0.8MHz therapy, 1 msec 416 bursts, 5 Hz repetition rate, 320 kPa focal pressure; 5.2 MHz imaging, 1 cycle, 250 kPa, for a total 417 treatment time of 2 minutes with concurrent i.v. tail vein administration of MB (2.5 × 10⁸ MBs/kg). 418 Finally, the brains were harvested for further processing after performing transcardial perfusion 419 with ice-cold PBS. 420 Magnetic resonance imaging 421 To confirm the BBB opening, immediately after the sonication, the animals were injected with 422 gadolinium contrast agent, Gd-HPDO3A (Prohance®, 0.5 μmol/g mouse body weight), and MRI 423 was performed using a Bruker 11.7 Tesla small animal scanner (Bruker BioSpin MRI, Ettlingen, 424 Germany) equipped with a cross coil configuration with a mouse body resonator for transmit and a 425 mouse surface coil for receive. Images were acquired using ParaVision 360 (Bruker BioSpin MRI). 426 Permeability of the BBB was determined with a T1 weighted (T1w) sequence (2D RARE sequence, 427 RARE factor = 2, repetition time (TR) 250 ms, echo time (TE) 6.7 ms, 1 mm slice thickness, 1 mm 428 interslice distance, 13 images, field of view (FOV) = 2 x 2 cm, matrix = 384 x 384, number of 429 acquisitions (NA) = 6). Hemorrhage was assess using T2 weight MRI sequence (2D FLASH 430 sequence, flip angle (FA) = 15°, repetition time (TR) 250 ms, echo time (TE) 15 ms, 1 mm slice 431 thickness, 1 mm interslice distance, 13 images, field of view (FOV) = 2 x 2 cm2, matrix = 192 x 432 192, number of acquisitions (NA) = 4). 433 Immunohistochemistry staining 434 Hematoxylin-eosin (H&E) staining was performed to examine tissue damage and safety. 20 µm 435 thick frozen sections (Leica 3050 S Cryostat) were dehydrated beforehand and stained using a Leica 436 Autostainer (ST5010). The sections were imaged with a 20x objective using a brightfield 437 microscope (Eclipse Ti2, Nikon). 438 AAV production 439 AAV9 vectors were produced by the UNC Vector Core (University of North Carolina). Briefly, 440 AAVs were harvested 5 days after triple transfection in HEK293 cells by PEG precipitation of 3 - 441 and 5-days media and osmotic lysis of cell pellets. Crude AAVs were then purified by extraction 442 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 14 of 26 from iodixanol density gradients and buffer exchanged into Dulbecco’s phosphate -buffered saline 443 (DPBS). Viral titers were determined by qPCR on a woodchuck hepatitis virus post-transcriptional 444 regulatory element (WPRE) present in all packaged AAV genomes. 445 DNA extraction 446 To assess AAV9 vector biodistribution, mice were euthanized 21 hours post -treatment, and 447 brains were extracted and bisected into sonicated and non-sonicated hemispheres on ice. The brain 448 sample was immediately stabilized in Allprotect Tissue Reagent (Qiagen) until processing. DNA 449 was isolated from tissue homogenates using the DNeasy Blood & Tissue Kit (Qiagen) according to 450 the manufacturer's protocol. DNA concentration and purity were determined by NanoDrop 451 spectrophotometry (Thermo Fisher Scientific), and samples were diluted to a uniform concentration 452 for qPCR analysis. 453 RNA extraction and cDNA synthesis 454 To assess transgene expression, brain tissue was collected and immediately stabilized in RNA-455 protect tissue reagent (Qiagen) on ice. Total RNA was extracted from brain hemispheres using the 456 RNeasy Midi Kit (Qiagen) following the manufacturer's instructions. During this process, genomic 457 DNA was removed by DNase. RNA integrity and concentration were assessed by NanoDrop 458 spectrophotometry, and samples with an A260/A280 ratio between 1.8 and 2.0 were used for 459 downstream analysis. Complementary DNA (cDNA) was synthesized from 500 ng -1 µg of total 460 RNA using SuperScript IV VILO Master mix (ThermoFisher Scientific) according to the 461 manufacturer's protocol. 462 Quantitative PCR (qPCR) 463 qPCR was performed using TaqMan -based assays with primers and probe designed to detect 464 specific genome copies ( tdTomato, PKM2, hBDNF, Syn1, Syp, and Dlg4) and normalized to the 465 housekeeping gene , β-actin. Relative expression was calculated using the ΔΔCt method and 466 expressed as fold-change relative to control groups. All reactions were performed in triplicate on 467 a CFX96 (Bio-Rad) system using the following cycling conditions: 95°C for 10 min, followed by 468 40 cycles of 95°C for 15 s and 60°C for 1 min. 469 Fluorescent reporter gene optical imaging 470 In vivo and ex vivo fluorescent imaging was performed using LAGO system (Spectral Instruments 471 Imaging) with an exposure of 2s. 472 In vivo and ex vivo fluorescent imaging was performed using the LAGO imaging system (Spectral 473 Instruments Imaging, Tucson, AZ) with excitation at 535 nm and emission at 589 nm (optimized 474 for tdTomato fluorescence). Images were acquired with a 2-second exposure time, and fluorescence 475 intensity was quantified using Aura, the manufacturer's software. Regions of interest (ROIs) were 476 drawn over the targeted hippocampal region, and fluorescence signal was quantified as maximum 477 and mean radiance (photons/second/cm²/steradian) or total emission (photons/second). 478 PET reporter gene imaging 479 Radiotracer synthesis was performed at the Stanford Cyclotron and Radiochemistry Facility 480 (CRF). To noninvasively assess PKM2 expression, we employed [¹⁸F]DASA -10, a second -481 generation PET radiotracer with improved physicochemical and pharmacokinetic properties 482 compared to [¹⁸F]DASA-23.78 [¹⁸F]DASA-10 was synthesized from nucleophilic displacement of 483 the nitro group within 1-((2,3-dihydrobenzo[b][1,4]dioxin-6-yl)sulfonyl)-4-((2-fluoro-6-484 nitrophenyl)sulfonyl)piperazine, as previously described. 78 The radiotracer ([¹⁸F]DASA -10; 485 specific activity: 74.4 GBq/μmol) was administered via tail vein injection (~5.55 MBq per mouse), 486 followed by dynamic PET/CT imaging initiated at the time of injection and continued for 30 487 minutes. 488 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 15 of 26 Raw list mode data were reconstructed using 3D ordered -subset expectation maximization 489 using maximum aposteriori (3D -OSEM/MAP) image reconstruction and converted to Standard 490 Uptake Value (SUV). For dynamic image analysis of [18F]DASA-10, the 30-min list mode data was 491 segmented into 20 static time frames (15 x 8, 60 x 8, 300 x 4; seconds x frames) and reconstructed 492 as stated above. Quantitative PET image analysis was performed with Inveon Research Workplace 493 (IRW) software after the co -registration of PET and CT images. Images were quantified by 494 manually drawing ROI in the FUS treated area. PET images were quantified using standardized 495 uptake values (SUV) to account for significant body weight variation across cohorts, as 496 comparisons were made between healthy and 5xFAD mice ranging from 4 to 10 months of age . 497 Images from 20-30 min post-injection were used for quantification to optimize signal-to-noise ratio. 498 SUV max values represent the maximum voxel intensity within each ROI within this time inteval, 499 not the temporal peak across the imaging window. 500 Immunofluorescence staining and microscopy 501 For protein expression analysis, after the animals were euthanized at the time point based on 502 their treatment protocols, the brains were harvested and were fixed with 4% paraformaldehyde 503 overnight at 4°C. The next day, 20 µm sections were cut using a microtome (Leica 3050 S Cryostat). 504 To assess the biological effect induced by MB -FUS AAVs delivery , immunofluorescence 505 staining was performed on the brain tissue. Tissues were prepared for staining by fixing in 4% 506 paraformaldehyde at room temperature for 10 min (For sections requiring staining of intracellular 507 markers (e.g., NeuN), they were permeabilized with 0.1% Triton X -100 in PBS for 5 minutes, 508 subsequently). After washed with PBS, the sections were blocked for 1 hour at room temperature 509 (2% Bovine Serum Albumin, 5% goat serum in PBS). It was then incubated with primary antibody 510 diluted in 1% Bovine Serum Albumin (1:100) for 12 hours at 4°C. Next, the sections were incubated 511 with secondary antibody diluted in 1% Bovine Serum Albumin (1:5 00) for 1 hour at room 512 temperature. To stain the cell nucleus, samples were incubated with DAPI diluted in PBS (1:1000, 513 62248, Invitrogen) for 10 minutes after washing. Finally, the sections were rinsed with PBS to 514 remove excess antibody, mounted with mounting medium (Prolong Glass Antifade Mountant, Lot# 515 2018752, Invitrogen), and covered with coverslips. Samples were cured with a mounting medium 516 for 24 hours in dark at room temperature before imaging. 517 The sections were imaged with a 20x objective using a fluorescence confocal microscope (Leica 518 DMi8 Inverted Microscope). The quantification of the fluorescence images was performed using 519 ImageJ. 520 Statistical analysis 521 All statistical analyses were performed using GraphPad Prism. P values P < 0.05 was considered 522 statistically significant. (n.s. no significance, * P ≤ 0.05, ** P ≤ 0.01, ****P ≤ 0.0001). In the case 523 of multiple comparisons, the p-values were adjusted using Bonferroni correction. 524 Data availability: All data supporting the findings of this study are available within the article and 525 its supplementary files. Any additional requests for information can be directed to, and will be 526 fulfilled by, the corresponding authors. Source data are provided in this paper. 527 Acknowledgments: We thank Dr. Frezghi Habte, Dr. Edwin Chang, and Laura Jean Pisani at the 528 Stanford Center for Innovation in In Vivo Imaging (SCi3) for technical support with PET/CT, 529 optical imaging, and MRI studies . [¹⁸F]DASA-10 production was carried out at the Stanford 530 Cyclotron and Radiochemistry Facility (CRF) . This study was supported by NIH grants R01 531 CA112356, R01 EB028646 and the Focused Ultrasound Foundation. Y.G. was partially supported 532 by the Focused Ultrasound Foundation Lockhart Postdoctoral Fellowship. 533 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 16 of 26 Author contributions: Y.G. and K.W.F. designed research; Y.G, J.F., and N.Z. performed FUS 534 treatment; Y.G, J.A., N.Z., J.W.S., J.W., and B.L.J. performed PET/CT imaging , R.M. and C.B. 535 developed and produced the PET radiotracer used in this study, Y.G., S.K.T and J.A. performed the 536 bioassay and analyzed the data, B.L.J. and M.N.R. performed animal care, Y.G. and K.W.F. wrote 537 the manuscript. All authors reviewed the manuscript and approved the final version. 538 Competing interests: The authors declare that they have no competing interests. 539

540 1. Self, W. K. & Holtzman, D. M. Emerging diagnostics and therapeutics for Alzheimer disease. Nat Med 541 29, 2187–2199 (2023). 542 2. Dementia. https://www.who.int/news-room/fact-sheets/detail/dementia. 543 3. Aye, S., Johansson, G., Hock, C., Lannfelt, L., Sims, J. R., Blennow, K., Frederiksen, K. S., Graff, C., 544 Molinuevo, J. L., Scheltens, P., Palmqvist, S., Schöll, M., Wimo, A., Kivipelto, M., Handels, R., Frölich, 545 L., Zilka, N., Tolar, M., Johannsen, P., Jönsson, L. & Winblad, B. Point of view: Challenges in 546 implementation of new immunotherapies for Alzheimer’s disease. The Journal of Prevention of 547 Alzheimer’s Disease 12, 100022 (2025). 548 4. Patwardhan, A. G. & Belemkar, S. An update on Alzheimer’s disease: Immunotherapeutic agents, 549 stem cell therapy and gene editing. Life Sciences 282, 119790 (2021). 550 5. Breijyeh, Z. & Karaman, R. Comprehensive Review on Alzheimer’s Disease: Causes and Treatment. 551 Molecules 25, 5789 (2020). 552 6. Chen, W., Hu, Y. & Ju, D. Gene therapy for neurodegenerative disorders: advances, insights and 553 prospects. Acta Pharm Sin B 10, 1347–1359 (2020). 554 7. Sun, J. & Roy, S. Gene-based therapies for neurodegenerative diseases. Nat Neurosci 24, 297–311 555 (2021). 556 8. Morroni, F., Caccamo, A., Moreira, P. I., Avila, J., Galimberti, D., Pappolla, M. A., Plascencia-Villa, G., 557 Sorensen, A. A., Zhu, X. & Perry, G. Advances and Challenges in Gene Therapy for Alzheimer’s 558 Disease. Journal of Alzheimer’s Disease 101, S417–S431 (2024). 559 9. Wang, S. & Xiao, L. Progress in AAV-Mediated In Vivo Gene Therapy and Its Applications in Central 560 Nervous System Diseases. Int J Mol Sci 26, 2213 (2025). 561 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 17 of 26 10. Wang, J.-H., Gessler, D. J., Zhan, W., Gallagher, T. L. & Gao, G. Adeno-associated virus as a delivery 562 vector for gene therapy of human diseases. Sig Transduct Target Ther 9, 78 (2024). 563 11. Ye, D., Chukwu, C., Yang, Y., Hu, Z. & Chen, H. Adeno-associated virus vector delivery to the brain: 564 Technology advancements and clinical applications. Advanced Drug Delivery Reviews 211, 115363 565 (2024). 566 12. Kang, L., Jin, S., Wang, J., Lv, Z., Xin, C., Tan, C., Zhao, M., Wang, L. & Liu, J. AAV vectors applied to the 567 treatment of CNS disorders: Clinical status and challenges. Journal of Controlled Release 355, 458–568 473 (2023). 569 13. Yao, Y., Wang, J., Liu, Y., Qu, Y., Wang, K., Zhang, Y., Chang, Y., Yang, Z., Wan, J., Liu, J., Nakashima, H., 570 Lawler, S. E., Chiocca, E. A., Cho, C.-F. & Bei, F. Variants of the adeno-associated virus serotype 9 with 571 enhanced penetration of the blood–brain barrier in rodents and primates. Nat. Biomed. Eng 6, 1257–572 1271 (2022). 573 14. Deverman, B. E., Pravdo, P. L., Simpson, B. P., Kumar, S. R., Chan, K. Y., Banerjee, A., Wu, W.-L., Yang, 574 B., Huber, N., Pasca, S. P. & Gradinaru, V. Cre-dependent selection yields AAV variants for 575 widespread gene transfer to the adult brain. Nat Biotechnol 34, 204–209 (2016). 576 15. Nonnenmacher, M., Wang, W., Child, M. A., Ren, X.-Q., Huang, C., Ren, A. Z., Tocci, J., Chen, Q., 577 Bittner, K., Tyson, K., Pande, N., Chung, C. H.-Y., Paul, S. M. & Hou, J. Rapid evolution of blood-brain-578 barrier-penetrating AAV capsids by RNA-driven biopanning. Molecular Therapy Methods & Clinical 579 Development 20, 366–378 (2021). 580 16. Chuapoco, M. R., Flytzanis, N. C., Goeden, N., Christopher Octeau, J., Roxas, K. M., Chan, K. Y., 581 Scherrer, J., Winchester, J., Blackburn, R. J., Campos, L. J., Man, K. N. M., Sun, J., Chen, X., Lefevre, A., 582 Singh, V. P., Arokiaraj, C. M., Shay, T. F., Vendemiatti, J., Jang, M. J., Mich, J. K., Bishaw, Y., Gore, B. B., 583 Omstead, V., Taskin, N., Weed, N., Levi, B. P., Ting, J. T., Miller, C. T., Deverman, B. E., Pickel, J., Tian, 584 L., Fox, A. S. & Gradinaru, V. Adeno-associated viral vectors for functional intravenous gene transfer 585 throughout the non-human primate brain. Nat. Nanotechnol. 18, 1241–1251 (2023). 586 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 18 of 26 17. Nisanov, A. M., Rivera de Jesús, J. A. & Schaffer, D. V. Advances in AAV capsid engineering: 587 Integrating rational design, directed evolution and machine learning. Molecular Therapy 33, 1937–588 1945 (2025). 589 18. Sweeney, M. D., Sagare, A. P. & Zlokovic, B. V. Blood–brain barrier breakdown in Alzheimer disease 590 and other neurodegenerative disorders. Nat Rev Neurol 14, 133–150 (2018). 591 19. Cai, Z., Qiao, P.-F., Wan, C.-Q., Cai, M., Zhou, N.-K. & Li, Q. Role of Blood-Brain Barrier in Alzheimer’s 592 Disease. J Alzheimers Dis 63, 1223–1234 (2018). 593 20. Perolina, E., Meissner, S., Raos, B., Harland, B., Thakur, S. & Svirskis, D. Translating ultrasound-594 mediated drug delivery technologies for CNS applications. Advanced Drug Delivery Reviews 208, 595 115274 (2024). 596 21. Kong, C. & Chang, W. S. Preclinical Research on Focused Ultrasound-Mediated Blood-Brain Barrier 597 Opening for Neurological Disorders: A Review. Neurol Int 15, 285–300 (2023). 598 22. Lipsman, N., Meng, Y., Bethune, A. J., Huang, Y., Lam, B., Masellis, M., Herrmann, N., Heyn, C., 599 Aubert, I., Boutet, A., Smith, G. S., Hynynen, K. & Black, S. E. Blood–brain barrier opening in 600 Alzheimer’s disease using MR-guided focused ultrasound. Nat Commun 9, 2336 (2018). 601 23. Meng, Y., Goubran, M., Rabin, J. S., McSweeney, M., Ottoy, J., Pople, C. B., Huang, Y., Storace, A., 602 Ozzoude, M., Bethune, A., Lam, B., Swardfager, W., Heyn, C., Abrahao, A., Davidson, B., Hamani, C., 603 Aubert, I., Zetterberg, H., Ashton, N. J., Karikari, T. K., Blennow, K., Black, S. E., Hynynen, K. & 604 Lipsman, N. Blood–brain barrier opening of the default mode network in Alzheimer’s disease with 605 magnetic resonance-guided focused ultrasound. Brain 146, 865–872 (2023). 606 24. Meng, Y., MacIntosh, B. J., Shirzadi, Z., Kiss, A., Bethune, A., Heyn, C., Mithani, K., Hamani, C., Black, 607 S. E., Hynynen, K. & Lipsman, N. Resting state functional connectivity changes after MR-guided 608 focused ultrasound mediated blood-brain barrier opening in patients with Alzheimer’s disease. 609 NeuroImage 200, 275–280 (2019). 610 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 19 of 26 25. Kofoed, R. H., Dibia, C. L., Noseworthy, K., Xhima, K., Vacaresse, N., Hynynen, K. & Aubert, I. Efficacy 611 of gene delivery to the brain using AAV and ultrasound depends on serotypes and brain areas. 612 Journal of Controlled Release 351, 667–680 (2022). 613 26. Thévenot, E., Jordão, J. F., O’Reilly, M. A., Markham, K., Weng, Y.-Q., Foust, K. D., Kaspar, B. K., 614 Hynynen, K. & Aubert, I. Targeted delivery of self-complementary adeno-associated virus serotype 9 615 to the brain, using magnetic resonance imaging-guided focused ultrasound. Hum Gene Ther 23, 616 1144–1155 (2012). 617 27. Arvanitis, C. D., Askoxylakis, V., Guo, Y., Datta, M., Kloepper, J., Ferraro, G. B., Bernabeu, M. O., 618 Fukumura, D., McDannold, N. & Jain, R. K. Mechanisms of enhanced drug delivery in brain 619 metastases with focused ultrasound-induced blood-tumor barrier disruption. Proc Natl Acad Sci U S A 620 115, E8717–E8726 (2018). 621 28. Curley, C. T., Mead, B. P., Negron, K., Kim, N., Garrison, W. J., Miller, G. W., Kingsmore, K. M., Thim, E. 622 A., Song, J., Munson, J. M., Klibanov, A. L., Suk, J. S., Hanes, J. & Price, R. J. Augmentation of brain 623 tumor interstitial flow via focused ultrasound promotes brain-penetrating nanoparticle dispersion 624 and transfection. Science Advances 6, eaay1344 (2020). 625 29. Alonso, A., Reinz, E., Leuchs, B., Kleinschmidt, J., Fatar, M., Geers, B., Lentacker, I., Hennerici, M. G., 626 Smedt, S. C. de & Meairs, S. Focal Delivery of AAV2/1-transgenes Into the Rat Brain by Localized 627 Ultrasound-induced BBB Opening. Molecular Therapy Nucleic Acids 2, (2013). 628 30. Hsu, P.-H., Wei, K.-C., Huang, C.-Y., Wen, C.-J., Yen, T.-C., Liu, C.-L., Lin, Y.-T., Chen, J.-C., Shen, C.-R. & 629 Liu, H.-L. Noninvasive and Targeted Gene Delivery into the Brain Using Microbubble-Facilitated 630 Focused Ultrasound. PLOS ONE 8, e57682 (2013). 631 31. Kofoed, R. H., Heinen, S., Silburt, J., Dubey, S., Dibia, C. L., Maes, M., Simpson, E. M., Hynynen, K. & 632 Aubert, I. Transgene distribution and immune response after ultrasound delivery of rAAV9 and PHP.B 633 to the brain in a mouse model of amyloidosis. Molecular Therapy Methods & Clinical Development 634 23, 390–405 (2021). 635 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 20 of 26 32. Wang, S., Kugelman, T., Buch, A., Herman, M., Han, Y., Karakatsani, M. E., Hussaini, S. A., Duff, K. & 636 Konofagou, E. E. Non-invasive, Focused Ultrasound-Facilitated Gene Delivery for Optogenetics. Sci 637 Rep 7, 39955 (2017). 638 33. Weber-Adrian, D., Thévenot, E., O’Reilly, M. A., Oakden, W., Akens, M. K., Ellens, N., Markham-639 Coultes, K., Burgess, A., Finkelstein, J., Yee, A. J. M., Whyne, C. M., Foust, K. D., Kaspar, B. K., Stanisz, 640 G. J., Chopra, R., Hynynen, K. & Aubert, I. Gene delivery to the spinal cord using MRI-guided focused 641 ultrasound. Gene Ther 22, 568–577 (2015). 642 34. Wang, S., Olumolade, O. O., Sun, T., Samiotaki, G. & Konofagou, E. E. Noninvasive, neuron-specific 643 gene therapy can be facilitated by focused ultrasound and recombinant adeno-associated virus. Gene 644 Ther 22, 104–110 (2015). 645 35. Weber-Adrian, D., Kofoed, R. H., Chan, J. W. Y., Silburt, J., Noroozian, Z., Kügler, S., Hynynen, K. & 646 Aubert, I. Strategy to enhance transgene expression in proximity of amyloid plaques in a mouse 647 model of Alzheimer’s disease. Theranostics 9, 8127–8137 (2019). 648 36. Xhima, K., Nabbouh, F., Hynynen, K., Aubert, I. & Tandon, A. Noninvasive delivery of an α-synuclein 649 gene silencing vector with magnetic resonance–guided focused ultrasound. Movement Disorders 33, 650 1567–1579 (2018). 651 37. Foiret, J., Zhang, H., Fite, B. Z., Ilovitsh, T., Mahakian, L. M., Tam, S., Beitnere, U., Pyles, B., Segal, D. J. 652 & Ferrara, K. W. Blood-brain barrier disruption for the delivery of non-infectious viral vectors and 653 proteins, preliminary study. in 2017 IEEE International Ultrasonics Symposium (IUS) 1–4 (2017). 654 doi:10.1109/ULTSYM.2017.8092207. 655 38. Stavarache, M. A., Petersen, N., Jurgens, E. M., Milstein, E. R., Rosenfeld, Z. B., Ballon, D. J. & Kaplitt, 656 M. G. Safe and stable noninvasive focal gene delivery to the mammalian brain following focused 657 ultrasound. Journal of Neurosurgery 130, 989–998 (2018). 658 39. Weber-Adrian, D., Kofoed, R. H., Silburt, J., Noroozian, Z., Shah, K., Burgess, A., Rideout, S., Kügler, S., 659 Hynynen, K. & Aubert, I. Systemic AAV6-synapsin-GFP administration results in lower liver 660 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 21 of 26 biodistribution, compared to AAV1&2 and AAV9, with neuronal expression following ultrasound-661 mediated brain delivery. Sci Rep 11, 1934 (2021). 662 40. Epelbaum, S., Burgos, N., Canney, M., Matthews, D., Houot, M., Santin, M. D., Desseaux, C., 663 Bouchoux, G., Stroer, S., Martin, C., Habert, M.-O., Levy, M., Bah, A., Martin, K., Delatour, B., Riche, 664 M., Dubois, B., Belin, L. & Carpentier, A. Pilot study of repeated blood-brain barrier disruption in 665 patients with mild Alzheimer’s disease with an implantable ultrasound device. Alz Res Therapy 14, 40 666 (2022). 667 41. Mehta, R. I., Carpenter, J. S., Mehta, R. I., Haut, M. W., Ranjan, M., Najib, U., Lockman, P., Wang, P., 668 D’haese, P.-F. & Rezai, A. R. Blood-Brain Barrier Opening with MRI-guided Focused Ultrasound 669 Elicits Meningeal Venous Permeability in Humans with Early Alzheimer Disease. Radiology 670 298, 654–662 (2021). 671 42. Rezai, A. R., Ranjan, M., Haut, M. W., Carpenter, J., D’Haese, P.-F., Mehta, R. I., Najib, U., Wang, P., 672 Claassen, D. O., Chazen, J. L., Krishna, V., Deib, G., Zibly, Z., Hodder, S. L., Wilhelmsen, K. C., 673 Finomore, V., Konrad, P. E. & Kaplitt, M. Focused ultrasound–mediated blood-brain barrier opening 674 in Alzheimer’s disease: long-term safety, imaging, and cognitive outcomes. 675 https://doi.org/10.3171/2022.9.JNS221565 (2022) doi:10.3171/2022.9.JNS221565. 676 43. Mehta, R. I., Carpenter, J. S., Mehta, R. I., Haut, M. W., Wang, P., Ranjan, M., Najib, U., D’Haese, P.-F. 677 & Rezai, A. R. Ultrasound-mediated blood–brain barrier opening uncovers an intracerebral 678 perivenous fluid network in persons with Alzheimer’s disease. Fluids Barriers CNS 20, 46 (2023). 679 44. Bae, S., Liu, K., Pouliopoulos, A. N., Ji, R., Jiménez-Gambín, S., Yousefian, O., Kline-Schoder, A. R., 680 Batts, A. J., Tsitsos, F. N., Kokossis, D., Mintz, A., Honig, L. S. & Konofagou, E. E. Transcranial blood-681 brain barrier opening in Alzheimer’s disease patients using a portable focused ultrasound system 682 with real-time 2-D cavitation mapping. Theranostics 14, 4519–4535 (2024). 683 45. Park, S. H., Baik, K., Jeon, S., Chang, W. S., Ye, B. S. & Chang, J. W. Extensive frontal focused 684 ultrasound mediated blood-brain barrier opening for the treatment of Alzheimer’s disease: a proof-685 of-concept study. Transl Neurodegener 10, 44 (2021). 686 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 22 of 26 46. Rezai, A. R., Ranjan, M., Haut, M. W., Carpenter, J., D’Haese, P.-F., Mehta, R. I., Najib, U., Wang, P., 687 Claassen, D. O., Chazen, J. L., Krishna, V., Deib, G., Zibly, Z., Hodder, S. L., Wilhelmsen, K. C., 688 Finomore, V., Konrad, P. E., Kaplitt, M., & Alzheimer’s Disease Neuroimaging Initiative. Focused 689 ultrasound-mediated blood-brain barrier opening in Alzheimer’s disease: long-term safety, imaging, 690 and cognitive outcomes. J Neurosurg 139, 275–283 (2023). 691 47. D’Haese, P.-F., Ranjan, M., Song, A., Haut, M. W., Carpenter, J., Dieb, G., Najib, U., Wang, P., Mehta, 692 R. I., Chazen, J. L., Hodder, S., Claassen, D., Kaplitt, M. & Rezai, A. R. β-Amyloid Plaque Reduction in 693 the Hippocampus After Focused Ultrasound-Induced Blood-Brain Barrier Opening in Alzheimer’s 694 Disease. Front Hum Neurosci 14, 593672 (2020). 695 48. Epelbaum, S., Burgos, N., Canney, M., Matthews, D., Houot, M., Santin, M. D., Desseaux, C., 696 Bouchoux, G., Stroer, S., Martin, C., Habert, M.-O., Levy, M., Bah, A., Martin, K., Delatour, B., Riche, 697 M., Dubois, B., Belin, L. & Carpentier, A. Pilot study of repeated blood-brain barrier disruption in 698 patients with mild Alzheimer’s disease with an implantable ultrasound device. Alzheimers Res Ther 699 14, 40 (2022). 700 49. Jeong, H., Song, I.-U., Chung, Y.-A., Park, J.-S., Na, S.-H., Im, J. J., Bikson, M., Lee, W. & Yoo, S.-S. 701 Short-Term Efficacy of Transcranial Focused Ultrasound to the Hippocampus in Alzheimer’s Disease: 702 A Preliminary Study. J Pers Med 12, 250 (2022). 703 50. Rezai, A. R., D, ’Haese Pierre-Francois, Finomore, V., Carpenter, J., Ranjan, M., Wilhelmsen, K., 704 Mehta, R. I., Wang, P., Najib, U., Vieira, L. T. C., Arsiwala, T., Tarabishy, A., Tirumalai, P., Claassen, D. 705 O., Hodder, S. & Haut, M. W. Ultrasound Blood–Brain Barrier Opening and Aducanumab in 706 Alzheimer’s Disease. New England Journal of Medicine 390, 55–62 (2024). 707 51. Ajenjo, J., Seo, J. W., Foiret, J., Wu, B., Raie, M. N., Wang, J., Fite, B. Z., Zhang, N., Malek, R., Beinat, 708 C., Malik, N., Anders, D. A. & Ferrara, K. W. PET imaging of focused-ultrasound enhanced delivery of 709 AAVs into the murine brain. Theranostics 13, 5151–5169 (2023). 710 52. Seo, J. W., Ingham, E. S., Mahakian, L., Tumbale, S., Wu, B., Aghevlian, S., Shams, S., Baikoghli, M., 711 Jain, P., Ding, X., Goeden, N., Dobreva, T., Flytzanis, N. C., Chavez, M., Singhal, K., Leib, R., James, M. 712 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 23 of 26 L., Segal, D. J., Cheng, R. H., Silva, E. A., Gradinaru, V. & Ferrara, K. W. Positron emission tomography 713 imaging of novel AAV capsids maps rapid brain accumulation. Nat Commun 11, 2102 (2020). 714 53. Seo, J. W., Ajenjo, J., Wu, B., Robinson, E., Raie, M. N., Wang, J., Tumbale, S. K., Buccino, P., Anders, 715 D. A., Shen, B., Habte, F. G., Beinat, C., James, M. L., Reyes, S. T., Ravindra Kumar, S., Miles, T. F., Lee, 716 J. T., Gradinaru, V. & Ferrara, K. W. Multimodal imaging of capsid and cargo reveals differential brain 717 targeting and liver detargeting of systemically-administered AAVs. Biomaterials 288, 121701 (2022). 718 54. Choi, J. J., Feshitan, J. A., Baseri, B., Wang, S., Tung, Y.-S., Borden, M. A. & Konofagou, E. E. 719 Microbubble-Size Dependence of Focused Ultrasound-Induced Blood–Brain Barrier Opening in Mice 720 In Vivo. IEEE Transactions on Biomedical Engineering 57, 145–154 (2010). 721 55. Guo, Y., Lee, H., Kim, C., Park, C., Yamamichi, A., Chuntova, P., Gallus, M., Bernabeu, M. O., Okada, H., 722 Jo, H. & Arvanitis, C. Ultrasound frequency-controlled microbubble dynamics in brain vessels regulate 723 the enrichment of inflammatory pathways in the blood-brain barrier. Nat Commun 15, 8021 (2024). 724 56. Song, K.-H., Fan, A. C., Hinkle, J. J., Newman, J., Borden, M. A. & Harvey, B. K. Microbubble gas 725 volume: A unifying dose parameter in blood-brain barrier opening by focused ultrasound. 726 Theranostics 7, 144–152 (2017). 727 57. Samiotaki, G., Vlachos, F., Tung, Y.-S. & Konofagou, E. E. A quantitative pressure and microbubble-728 size dependence study of focused ultrasound-induced blood-brain barrier opening reversibility in 729 vivo using MRI. Magnetic Resonance in Medicine 67, 769–777 (2012). 730 58. Vlachos, F., Tung, Y.-S. & Konofagou, E. Permeability dependence study of the focused ultrasound-731 induced blood–brain barrier opening at distinct pressures and microbubble diameters using DCE-732 MRI. Magn. Reson. Med. 66, 821–830 (2011). 733 59. Wang, S., Samiotaki, G., Olumolade, O., Feshitan, J. A. & Konofagou, E. E. Microbubble Type and 734 Distribution Dependence of Focused Ultrasound-Induced Blood–Brain Barrier Opening. Ultrasound in 735 Medicine & Biology 40, 130–137 (2014). 736 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 24 of 26 60. Marmottant, P., van der Meer, S., Emmer, M., Versluis, M., de Jong, N., Hilgenfeldt, S. & Lohse, D. A 737 model for large amplitude oscillations of coated bubbles accounting for buckling and rupture. The 738 Journal of the Acoustical Society of America 118, 3499–3505 (2005). 739 61. Ling, Q., Herstine, J. A., Bradbury, A. & Gray, S. J. AAV-based in vivo gene therapy for neurological 740 disorders. Nat Rev Drug Discov 22, 789–806 (2023). 741 62. Castle, M. J., Gershenson, Z. T., Giles, A. R., Holzbaur, E. L. F. & Wolfe, J. H. Adeno-Associated Virus 742 Serotypes 1, 8, and 9 Share Conserved Mechanisms for Anterograde and Retrograde Axonal 743 Transport. Hum Gene Ther 25, 705–720 (2014). 744 63. Wang, Y., Wang, Y., Zhang, W., Xu, Y., Narayan, N. S., Cheng, H., Yang, H., Huang, Y., Zhang, C. & 745 Wang, C. Development of a Novel PET Radioligand Targeting PKM2 for Brain Imaging and Alzheimer’s 746 Disease Characterization. J. Med. Chem. 68, 25456–25468 (2025). 747 64. Nagahara, A. H., Mateling, M., Kovacs, I., Wang, L., Eggert, S., Rockenstein, E., Koo, E. H., Masliah, E. 748 & Tuszynski, M. H. Early BDNF Treatment Ameliorates Cell Loss in the Entorhinal Cortex of APP 749 Transgenic Mice. J Neurosci 33, 15596–15602 (2013). 750 65. Nagahara, A. H., Merrill, D. A., Coppola, G., Tsukada, S., Schroeder, B. E., Shaked, G. M., Wang, L., 751 Blesch, A., Kim, A., Conner, J. M., Rockenstein, E., Chao, M. V., Koo, E. H., Geschwind, D., Masliah, E., 752 Chiba, A. A. & Tuszynski, M. H. Neuroprotective effects of brain-derived neurotrophic factor in 753 rodent and primate models of Alzheimer’s disease. Nat Med 15, 331–337 (2009). 754 66. de Pins, B., Cifuentes-Díaz, C., Farah, A. T., López-Molina, L., Montalban, E., Sancho-Balsells, A., 755 López, A., Ginés, S., Delgado-García, J. M., Alberch, J., Gruart, A., Girault, J.-A. & Giralt, A. Conditional 756 BDNF Delivery from Astrocytes Rescues Memory Deficits, Spine Density, and Synaptic Properties in 757 the 5xFAD Mouse Model of Alzheimer Disease. J Neurosci 39, 2441–2458 (2019). 758 67. Kopec, B. M., Zhao, L., Rosa-Molinar, E. & Siahaan, T. J. Non-invasive Brain Delivery and Efficacy of 759 BDNF in APP/PS1 Transgenic Mice as a Model of Alzheimer’s Disease. Med Res Arch 8, 2043 (2020). 760 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 25 of 26 68. Ibrahim, A. M., Chauhan, L., Bhardwaj, A., Sharma, A., Fayaz, F., Kumar, B., Alhashmi, M., AlHajri, N., 761 Alam, M. S. & Pottoo, F. H. Brain-Derived Neurotropic Factor in Neurodegenerative Disorders. 762 Biomedicines 10, 1143 (2022). 763 69. Cirulli, F., Berry, A., Chiarotti, F. & Alleva, E. Intrahippocampal administration of BDNF in adult rats 764 affects short-term behavioral plasticity in the Morris water maze and performance in the elevated 765 plus-maze. Hippocampus 14, 802–807 (2004). 766 70. Durham, P. G., Butnariu, A., Alghorazi, R., Pinton, G., Krishna, V. & Dayton, P. A. Current clinical 767 investigations of focused ultrasound blood-brain barrier disruption: A review. Neurotherapeutics 21, 768 e00352 (2024). 769 71. Foust, K. D., Nurre, E., Montgomery, C. L., Hernandez, A., Chan, C. M. & Kaspar, B. K. Intravascular 770 AAV9 preferentially targets neonatal neurons and adult astrocytes. Nat Biotechnol 27, 59–65 (2009). 771 72. Foust, K. D., Salazar, D. L., Likhite, S., Ferraiuolo, L., Ditsworth, D., Ilieva, H., Meyer, K., Schmelzer, L., 772 Braun, L., Cleveland, D. W. & Kaspar, B. K. Therapeutic AAV9-mediated Suppression of Mutant SOD1 773 Slows Disease Progression and Extends Survival in Models of Inherited ALS. Molecular Therapy 21, 774 2148–2159 (2013). 775 73. Choudhury, S. R., Harris, A. F., Cabral, D. J., Keeler, A. M., Sapp, E., Ferreira, J. S., Gray-Edwards, H. L., 776 Johnson, J. A., Johnson, A. K., Su, Q., Stoica, L., DiFiglia, M., Aronin, N., Martin, D. R., Gao, G. & Sena-777 Esteves, M. Widespread Central Nervous System Gene Transfer and Silencing After Systemic Delivery 778 of Novel AAV-AS Vector. Molecular Therapy 24, 726–735 (2016). 779 74. Arvanitis, C. D., Crake, C., McDannold, N. & Clement, G. T. Passive Acoustic Mapping with the Angular 780 Spectrum Method. IEEE Transactions on Medical Imaging 36, 983–993 (2017). 781 75. Khodabakhshi, Z., Hosseinkhah, N. & Ghadiri, H. Pulsating Microbubble in a Micro-vessel and 782 Mechanical Effect on Vessel Wall: A Simulation Study. Journal of Biomedical Physics and Engineering 783 0, (2020). 784 76. Hosseinkhah, N., Chen, H., Matula, T. J., Burns, P. N. & Hynynen, K. Mechanisms of microbubble–785 vessel interactions and induced stresses: A numerical study. J Acoust Soc Am 134, 1875–1885 (2013). 786 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint Page 26 of 26 77. van der Meer, S., Dollet, B., Chin, C. T., Bouakaz, A., Voormolen, M., de Jong, N., Versluis, M. & Lohse, 787 D. Microbubble spectroscopy of ultrasound contrast agents. The Journal of the Acoustical Society of 788 America 120, 3327–3327 (2006). 789 78. Kendirli, M. T., Malek, R., Silveira, M. B., Acosta, C., Zhang, S., Azevedo, C., Nagy, S. C., Habte, F., 790 James, M. L., Recht, L. D. & Beinat, C. Development of [18F]DASA-10 for enhanced imaging of 791 pyruvate kinase M2. Nucl Med Biol 124–125, 108382 (2023). 792 793 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.02.703398doi: bioRxiv preprint