Focused Ultrasounds as an Adeno-Associated Virus Gene Therapy-Empowering Tool in Juvenile Mice via Intracerebroventricular Administration

Abstract

Systemic delivery of adeno-associated virus (AAV) vectors targeting the central nervous system has the potential to solve many neurodevelopmental disorders, yet it is made difficult by the filtering effect of the blood-brain barrier and systemic complications. To overcome this limitation, we attempted to inject a Venus-expressing, oligodendrocyte-selective AAV9 viral vector in the ventricles together with lipid microbubbles and subjected them to focused ultrasound (FUS); the resulting mechanical stimulation on the brain ventricles is able to open small, temporary gaps from which vector particles can leak and spread. Our findings indicate that FUS can increase viral vector diffusion across both the anteroposterior and left-right axes without influencing cell tropism; significant effects were found with 60 and 90 s exposure time, but no effects were observed with longer intervals. Taken together, these results highlight a new strategy for the safe and effective delivery of viral vectors and offer new perspectives for the development and application of gene therapies for central nervous system diseases.

INTRODUCTION

Neurodevelopmental disorders (NDDs) are a broad category of pathologies affecting the brain in its early development, seriously compromising cognition, social behavior, mobility, and survivability of affected patients. It is estimated that 53 million children and young people worldwide are struggling with NDDs.¹

Among NDDs, hypomyelinating leukodystrophies are rare yet severe pathologies affecting myelin sheath formation and often manifest in childhood and infancy, including motor and cognitive function development delay, hypotonia, spasticity, dystonia, and ataxia.² Since the gene variations leading to more than 20 known hypomyelinating leukodystrophies have been identified,³ the use of gene therapy as a tool to either suppress or compensate for pathological mutations would represent an important step in the development of therapeutic options for patients worldwide.

One major obstacle to this approach is the blood–brain barrier (BBB), a broad network of capillaries, endothelial cells, and astrocyte feet held together by tight junctions that selectively filter molecular passage from blood flow to the brain. Injecting microbubbles, a commercially available type of contrast agent used in ultrasonography, into the circulatory system and subjecting them to focused ultrasounds (FUSs) has been shown to open small, temporary gaps, presumably because of stimulation of BBB tight junction mechanoreceptors^4,5; this approach is promising, but it still faces the issues of nonspecific biodistribution, immunogenic response, and hepatic accumulation of transgenic payload,⁶ which complicates the use of gene therapy.

Intracerebroventricular (ICV) administration of transgene vectors can overcome many obstacles presented by the bloodstream route: from the brain ventricles, cerebrospinal fluid (CSF) flows through the subarachnoid space and then the spinal cord and the whole brain⁷; this privileged route allows the administration of lower doses of vector, thus reducing peripheral uptake and immune response. ICV injection of an adeno-associated virus 9 (AAV9) viral vector showed higher biodistribution than equivalent approaches such as intrathecal and intranasal injection (which are seen as less invasive than ICV), but even so, the vector managed to penetrate mostly ventricle-adjacent areas such as cortex and hippocampus.⁸ Since ventricles rely on the same types of tight junctions holding BBB together, it is possible to use a combination of FUSs and lipidic microbubbles to produce controlled leaks of CSF directly in the brain. This approach has been tested with nonviral vectors^9,10; however, it suffers from inefficient tropism and limited payload size.

In the following study, we sought to fill this gap in experimental knowledge by examining how FUS applied to lipid microbubbles affects brain diffusion of an ICV-injected AAV vector and the specific conditions to achieve the most desirable outcome.

MATERIALS AND METHODS

Animals

Jcl:B6C3F1 mice were purchased from CLEA Japan, Inc. They were mated over 10 generations while housed in an EBAC-S breeding apparatus (CL-5351, CLEA) with ad libitum food and water, controlled temperature (22°C ± 1°C) and humidity (60% ± 5%), and a 12/12-h light/dark cycle. From the resulting offspring, a total of 26 pups aged P10 were assigned randomly to experimental groups. All experiments have been performed in accordance with the guidelines for animal care regulated by the animal committee of the National Institute of Neuroscience, National Center of Neurology and Psychiatry, Tokyo, Japan.

Self-complementary AAV vector production and titration

pscw.CNP.Venus.miRneg vector plasmid was produced as previously described¹¹ and structured as shown in Fig. 1A: under a human cyclic-nucleotide 3′-phosphodiesterase (CNP) promoter, Venus fluorescent protein cDNA and a scramble miRNA (miRneg) were cloned into a pscw self-complementary AAV scaffold. The resulting plasmid was then transfected into AAV-293 cells, together with the adenovirus helper plasmid pHelper (Agilent Technologies) and pAAV2/9n (a gift from James M. Wilson; Addgene plasmid #112865; http://n2t.net/addgene:112865; RRID:Addgene_112865) using polyethyleneimine as described elsewhere.¹² Cells were harvested 72 h after transfection, and AAVs were purified using an AAVpro Purification Kit (Takara Bio Inc.). Vector genome titers were determined by quantitative PCR using an AAVpro Titration Kit (Takara Bio Inc.).

Figure 1.

Schematics of plasmid design and administration. (A) The scAAV backbone vector contains human CNP promoter, the coding sequence of fluorescent protein Venus, and the scramble miRNA sequence. The human CNP promoter was a 1.8-kilobase 5′ flanking sequence of the human CNP gene. The engineered miRNA cassette was placed in the 3′ UTR of the Venus sequence and consisted of a 5′ random sequence with no complementarity for any known vertebrate gene (miRneg), a 19-nucleotide loop sequence derived from miR155, and nucleotides 1–8 and 11–21 of miRneg’s complementary sequence. (B) After removing fur from the scalp and hypothermia anesthesia, P10 mice injection point is found as such: picture the rectangle formed by right eye, right ear, bregma (blue dot between the eyes) and lambda (blue dot between the ears), then mark its center (green dot where diagonals cross); find the midpoint between rectangle center and bregma (green dot on the bregma-ear diagonal), then trace an imaginary line from it perpendicular to the bregma–lambda axis; inject on the midpoint of said diagonal (red X mark), 3 mm deep from scalp. CNP, cyclic-nucleotide 3′-phosphodiesterase; scAAV, self-complementary AAV; UTR, untranslated region.

ICV injection and FUS treatment

Twenty-six P10 mice had fur on their scalp removed with depilatory cream; after anesthesia by hypothermia, they were injected ICV with 4 µL of a 1:1 mix (2 µL each) of viral vector (titred 2 × 10¹² v.g./mL) and TBS-Prime-3 microbubbles (NepaGene, 2.5 × 10¹⁰ bubbles/mL). Because of their delicate skull, injection coordinates could not be determined using a stereotactic apparatus, and the procedure illustrated in Fig. 1B was instead followed: once the jaw was placed on a flat support and while holding the body perpendicular to the head, the midpoint between the eyes and the one between the ears on the top of the head (cursory corresponding to bregma and lambda) were marked; the center of the imaginary square formed by these two points, right eye and right ear, was determined, and the midpoint of the line connecting this new point and putative bregma was marked as well; originating from that point, a line perpendicular to the putative bregma–lambda axis was traced and injection was performed at the midpoint of the line, 3 mm deep from the scalp, on the right hemisphere. Seven out of the 26 animals did not receive any further treatment and were left as controls, while the others received FUSs from a Sonitron GTS plain wave probe (NepaGene; 1 W/cm², on/off duty: 3 ms on, 3 ms off) placed over the injection point. Probe irradiation surface was 28 mm², which was significantly smaller than P10 mice skulls (∼100 mm²). Different exposure times were used for the different mouse groups: 60 s (N = 7), 90 s (N = 6), and 120 s (N = 6). After the procedure, pups were returned to their home cages and perfused at P25.

Antibodies

The following primary antibodies were used: glial fibrillary acidic protein (GFAP) (clone 6F2, Dako/Agilent Technologies), neuronal nuclei (NeuN) protein (clone A60, Chemicon/MilliporeSigma), ionized calcium binding adaptor molecule 1 (Iba1) (catalog #016–20001, WAKO), and glutathione S-transferase π (Gst-π) (MBL312, Medical and Biological Laboratories).

Immunostaining and imaging

The mice were anesthetized with isoflurane inhalant liquid and transcardially perfused with fresh 4% paraformaldehyde. The brains were further fixed in 4% paraformaldehyde overnight at 4°C, cryoprotected in 30% sucrose, embedded in Tissue-Tek optimal cutting temperature compound (Sakura Finetek), and coronally cryosectioned at 10-μm thickness for immunostaining and 20-µm thickness for the remaining experiments. The sections were blocked with 5% goat serum in phosphate-buffered saline with 0.05% Triton X-100 for 2 h. Primary antibodies were diluted in PBS with 3% goat serum and 0.05% Triton X-100, whereas secondary antibodies were diluted in 0.05% Triton X-100. The following dilutions of the primary antibodies were incubated overnight at 4°C: anti-GFAP, 1:400; anti-NeuN, 1:400, anti-Iba1 1:300 and anti-Gst-π, 1:300. Secondary antibody incubation was performed for 1 h at room temperature (22–25°C) with 1:400 dilutions of Alexa Fluor 647–goat anti-mouse, and Alexa Fluor 594–goat anti-rabbit (Thermo Fisher Scientific). Cell nuclei were visualized using 4′,6′-diamidino-2-phenylindole (DAPI; MilliporeSigma). The slides were mounted with ProLong Diamond mounting agent (Thermo Fisher Scientific) and examined under a Zeiss LSM 710 confocal microscope (Carl Zeiss AG). Venus-positive cells, marker-positive cells, and double-positive cells on images from 10-μm thick slices were manually counted by a group-blind operator (N = 3 animals per group, 4 images per animal). Cell count and total Venus-positive area on images from 20-µm-thick slices were calculated using ImageJ v1.54d software (National Institutes of Health); to avoid false positives, a 5 px² minimal threshold was set for cell detection. Images were taken at anteroposterior bregma coordinate +1.5, +0.5, −0.5, −1.5, −2.5, and −3.5 mm, focusing on ventricular dorsal region (Supplementary Fig. S1).

Statistics

All values are expressed as the mean ± SEM. One-way analysis of variance (ANOVA) with Bonferroni’s post hoc test was used to determine statistical significance, defined as p < 0.05, in comparisons within groups. Two-way ANOVA was used for comparisons between different groups; bregma coordinates (row factor) and treatment (column factor) were used as categorical variables. All statistical analyses were performed using the GraphPad Prism software.

RESULTS

FUSs do not alter viral vector specificity

The plasmid used in this study has been proven to be highly selective for oligodendrocytes in its previous application thanks to the implementation of a human CNP promoter.¹¹ However, the use of the AAV9 capsid instead of AAV1/2 and the application of FUSs introduce new variables that need to be accounted for. For this reason, Gst-π, Iba1, NeuN, and GFAP were selected as markers for oligodendrocytes, microglia, neurons, and astrocytes, respectively, and their overlap with Venus-derived signals was detected by immunostaining (Fig. 2A).

Figure 2.

Vector tropism in response to ultrasound stimulation. (A) 20× magnification images from periventricular zone of ultrasound treated animals after perfusion at P25 and immunostaining with cell type-specific antibodies (right column: secondary antibody fluorescence; left column: Venus fluorescence; central column: merge). Scale bar: 100 µm. Brightness and contrast in middle column images have been increased by 40% for better visualization. (B) Quantification of the percentages of each cell type in Venus-positive cells (n = 3 mice, 4 sections per mouse). Statistical significance was determined using one-way ANOVA with Bonferroni’s post hoc test. ANOVA, analysis of variance; CTR, control (injected with vector and microbubbles but no focused ultrasounds); FUS, focused ultrasound.

Near total overlap between Venus and the Gst-π signal was detected in both ultrasound-treated and untreated animals, while nearly no colocalization of Venus with the remaining three markers was observed (Fig. 2B).

These findings confirm that FUSs do not influence the expression profile of the vector, and thus, they can be used in combination with viral particles without increasing the chance of off-target expression.

Effects of FUSs on viral vector diffusion

Images were taken from brain slices of both control (CTR) and 60 s ultrasound-treated animals (FUS 60 s), and both the percentage of Venus⁺ cells over DAPI⁺ cells and the total area of Venus-expressing cells were measured. All animals were injected in the right ventricle, and coronal sections were taken every millimeter from bregma coordinates of 1.5 mm to −3.5 mm.

In both groups, Venus expression is the strongest at bregma −0.5 mm, close to injection site, and in controls it sharply decreases in more posterior regions (Fig. 3A−C); in the ultrasound-treated group, however, Venus expression increases posteriorly to the injection site, and its total area at bregma −1.5 mm is comparable to that measured at the bregma −0.5 mm injection site (Fig. 3D−F). It is also worth mentioning that in the left hemisphere, both the relative number of Venus⁺ cells and total Venus⁺ surface were significantly higher than in the most anterior (bregma 1.5 mm) and posterior (bregma −3.5 mm) regions, implying an effect of ultrasound on both the anteroposterior and left-right axes.

Figure 3.

Quantification of focused ultrasound effect on viral vector diffusion across the bregma–lambda axis. (A) Percentage of transfected, Venus⁺ cells over total, DAPI⁺ cells and (B) total Venus⁺ area in CTR animals at fixed bregma coordinates (L: left hemisphere, uninjected; R: right hemisphere, injected. N = 7 animals). (C) Coronal sections of left and right hemispheres of a CTR animal at 4x magnification (top row: DAPI; bottom row: Venus), bregma coordinate −1.5 mm. Scale bar: 100 µm. (D) Percentage of Venus⁺ cells over DAPI⁺ cells and (E) total Venus⁺ area in 60 s treated ultrasound (US60) animals at fixed bregma coordinates. (F) Coronal sections of left and right hemispheres of a US60 animal at 4× magnification, bregma coordinate −1.5 mm (N = 7 animals). Data are reported as mean ± SEM. Statistical significance was determined using one-way ANOVA with Bonferroni’s post hoc test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. DAPI, 4′,6′-diamidino-2-phenylindole; SEM, standard error of mean.

When comparing data between controls and ultrasound-treated animals, a significant effect of the treatment was observed in both the relative number of Venus⁺ cells (p = 0.0088, Fig. 4A) and total Venus⁺ surface (p = 0.005, Fig. 4C) when looking at the right ipsilateral hemisphere. While no such significance can be seen in the left and contralateral hemispheres (Fig. 4B, D), bregma coordinates also had a significant effect not only for the right (p < 0.0001) but also for the left (p = 0.0002) hemispheres on both variables. These encouraging results prompted us to test new conditions as a means to increase the efficacy of the treatment.

Figure 4.

Effect of focused ultrasounds in relation to control animals. (A) Percentage of Venus⁺ cells over DAPI⁺ cells in the right hemisphere and (B) left hemisphere at fixed bregma coordinates. (C) Total Venus⁺ area in the right hemisphere and (D) left hemisphere at fixed bregma coordinates. Blue: CTR; light red: US60. Data are reported as mean ± SEM. Statistical significance was determined using two-way ANOVA. Bregma coordinates (row factor) and treatment (column factor) were used as categorical variables.

Different exposition times to FUSs alter viral vector diffusion

To assess the correlation between ultrasound exposure time and vector diffusion, 90 s and 120 s ultrasound-treated mice were analyzed (FUS 90 s and FUS 120 s). Like in the 60 s group, 90 s group total Venus⁺ area at bregma −1.5 mm is comparable to that measured at injection site; however, this time also relative number of Venus⁺ cells at bregma −1.5 mm is comparable to injection site values.

The same effect observed in 60 s group’s left hemispheres is replicated in the 90 s group, and this time the difference in Venus⁺ cell and total Venus⁺ surface between bregma −0.5 mm and both the anterior coordinates examined (bregma 1.5 mm and 0.5 mm), as well as bregma −3.5 mm, is statistically significant (Fig. 5A−C).

Figure 5.

Effect of different ultrasound exposure times on viral vector diffusion across the bregma–lambda axis. (A) Percentage of transfected, Venus⁺ cells over total, DAPI⁺ cells and (B) total Venus⁺ area in 90 s treated (US90) animals at fixed bregma coordinates (N = 6 animals). (C) Coronal sections of left and right hemispheres of a US90 animal at 4× magnification (top row: DAPI; bottom row: Venus), bregma coordinate −1.5 mm. Scale bar: 100 µm. (D) Percentage of Venus⁺ cells over DAPI⁺ cells and (E) total Venus⁺ area in 120 s treated ultrasound (US120) animals at fixed bregma coordinates. (F) Coronal sections of left and right hemispheres of a US60 animal at 4× magnification, bregma coordinate −1.5 mm (N = 6 animals). Data are reported as mean ± SEM. Statistical significance was determined using one-way ANOVA with Bonferroni’s post hoc test. *p < 0.05; **p < 0.01; ***p < 0.001.

Exposition to 120 s ultrasounds resulted in overall uniform values, with no significant difference between coordinated (Fig. 5D−F).

Compared to controls, 90 s treatment data reflect those of 60 s treatment on the ipsilateral hemisphere (Fig. 6A, B; Venus⁺ cell p = 0.0081, total Venus⁺ surface p = 0.0095); however, on the contralateral hemisphere, the treatment effect was statistically significant for both variables (Venus⁺ cell p = 0.0002, total Venus⁺ surface p = 0.0004) especially at bregma −0.5 mm (Venus⁺ cell p = 0.0073, total Venus⁺ surface p = 0.0166) and bregma −1.5 mm (Venus⁺ cell p = 0.047, total Venus⁺ surface p = 0.0234; Fig. 6C, D). Again, the bregma coordinates also had a significant effect on both hemispheres (p < 0.0001).

Figure 6.

Effect of different ultrasound exposure times in relation to control animals. (A, E) Percentage of Venus⁺ cells over DAPI⁺ cells in the right hemisphere and (C, G) left hemisphere at fixed bregma coordinates. (B, F) Total Venus⁺ area in the right hemisphere and (D, H) left hemisphere at fixed bregma coordinates. Blue: CTR; red: US90; deep red: US120. Data are reported as mean ± SEM. Statistical significance was determined using two-way ANOVA. *p < 0.05; **p < 0.01. Bregma coordinates (row factor) and treatment (column factor) were used as categorical variables.

No difference was found between the control group and the 120 s treatment group, regardless of hemisphere (Fig. 6E−H), and the coordinate effect was weaker (right Venus⁺ cell, p = 0.0003; right total Venus⁺ surface, p = 0.0006; left Venus⁺ cell, p = 0.0223; left Venus⁺ surface, p = 0.034).

When comparing all groups together, the group effect was statistically significant in the right (Venus⁺ cell p = 0.0399, total Venus⁺ surface p = 0.0216) and left (Venus⁺ cell p = 0.0072, total Venus⁺ surface p = 0.0093) hemispheres (Supplementary Fig. S2), and the coordinate effect was significant regardless of the hemisphere (p < 0.0001). This confirmed the presence of an overall treatment effect. Mean fluorescence intensity data substantially reflect cell number and cell area data (Supplementary Fig. S3).

In summary, both 60 s and 90 s ultrasound exposition can increase vectorial diffusion, and 90 s treatment can also influence the contralateral hemisphere.

DISCUSSION

We confirmed that applying FUS to lipidic microbubbles can increase diffusion in the brain of a viral vector injected intracerebroventricularly, and under optimal exposition time, the effect of FUS can be seen on both the left-right and anteroposterior axes. Our work not only represents the first documented use of FUS in facilitating AAVs passage across brain ventricle walls but also shows the extent to which this approach can benefit current viral vector-based gene therapy studies, as well as lay the groundwork for future applications by individuating ideal conditions of use.

Our capsid of choice, AAV9, has been reported to be able to bypass ventricular walls when injected intrathecally.¹³ However, our data show, consistently with previous findings,¹¹ that AAV9 diffusion without external influences is mostly limited to the region of injection, whereas AAV9 in combination with microbubbles and FUSs can penetrate as posterior as 3 mm from the delivery point. We also found an important effect of FUS on the contralateral hemisphere, which could be explained by considering the small brain size of our subjects and the relative proximity of the two ventricles at the injection bregma coordinate. It is possible that either the flow of CSF was strong enough to carry viral particles from one periventricular zone to the other, or that ultrasounds were able to stimulate microbubbles on the contralateral side; because the patterns of Venus⁺ cells were coherent between the two hemispheres, we believe the latter to be true. The enhanced diffusion in the contralateral hemisphere is further evidence of the beneficial role of microbubble-coupled FUS on viral vector diffusion; this is an important step toward achieving whole-brain transfection.

Depending on brain area, AAV9 vectors may display variable tropism; including the human CNP promoter in the plasmid allowed us to achieve near total selectivity for oligodendrocytes. Such cell selectivity may be important to avoid off-target gene expression. In our experimental setting, FUS did not affect the specificity of the CNP promoter in oligodendrocytes.

In the BBB, the use of FUS exposes vector particles to the surrounding astrocytes and can lead to unintended transfection if tropism is not tightly regulated^14,15; similarly, the subventricular zone is enriched with neural progenitor cells (NPCs) and neural stem cells, which can become undesired targets when the ventricular walls are loosened by mechanical stimulation. Because inverted terminal repeats in AAVs plasmids are extremely harmful to these cellular populations,¹⁶ FUS may reduce their numbers as a collateral effect. However, our initial findings show that the number of cells positive for GFAP, which is a marker both for astrocytes and NPC, in the transfected regions isn’t significantly different between controls and FUS-treated animals, regardless of brain hemisphere (Supplementary Fig. S4). Being the first study to use viral vectors and FUS after ICV injection, our study is also the first one to address the issue of unintended target overexposure in the ventricular context, which we hope can pave the road for further studies in the field of viral vector-mediated gene therapy.

The microbubbles used in this study had a phospholipid shell filled with perfluoropropane (C₃F₈), which can be produced and purchased cheaply. They were also successfully implemented in ultrasound-mediated transfection after ICV injection using non-viral vectors,¹⁰ making them a valid starting point for our experiments. Nevertheless, it is reasonable to assume that modifications to the shell structure may increase the microbubble’s ability to induce sonoporation and/or decrease the energy requirements. Further studies on this subject are required to assess ideal microbubble parameters. We performed our injections on postnatal day 10 mice because injecting earlier than P10 results in more non-specific distribution, while injecting later than P10 leads to reduced distribution of AAV. This is also ideal for FUS treatment, since the mouse skull is not yet calcified and thus does not affect ultrasound transmission. However, recently, a group working on BBB permeabilization was able to establish a direct relationship between mouse body mass and ultrasound transmission loss when crossing the skull.¹⁷ A similar relationship could be applied to ICV injections; therefore, FUS conditions such as exposure time should be optimized for the age of mice, or even other animals, to be used.

Contrary to our original assumptions, FUS exposition for 120 s has no noticeable effect on transfection; we speculate this may be caused by mechanical overstimulation: it is possible that prolonged FUS stimulation in the cerebroventricular microenvironment may have caused some microbubbles to fracture, either disrupting the stable cavitation motions needed for ventricular opening or affecting vector capsid in a way that disrupts its function; further studies on the subject are needed to reach a definitive conclusion.

Finally, a different administration site could improve treatment efficacy even further: we performed ICV injection because ventricles’ extension in relation to brain volume makes them easier to target, but that procedure is invasive and causes undesired damage to the above cortex; intracisternal and intralumbar injection have been successfully tried for AAV9 viral vectors,¹⁸ and they are significantly less invasive for the recipient subject. Further studies on FUS-assisted viral vector delivery should investigate these aspects.

CONCLUSIONS

We successfully managed to increase AAV vector diffusion in the brain of juvenile mice using lipid microbubbles and FUSs after ICV injection and optimized ultrasound exposition time in an attempt to achieve the broadest vector diffusion. We believe that our findings, while still preliminary, can be a step toward the development of safer, more effective gene therapies for human neurodevelopmental and neurodegenerative diseases, as well as a stepping stone for further research on viral vectors applied to neuropathology.

AUTHORS’ CONTRIBUTIONS

A.Z. performed the experiments, data collection and analysis, and article draft writing; H.L. offered reagents, consulting, and supervising the experimental procedures; K.I. provided project planning, supervision, and article reviewing and editing.

AUTHOR DISCLOSURE

The authors declare that they have no conflicts of interest.

FUNDING INFORMATION

This study was supported in part by grants from the Japan Society for the Promotion of Science KAKENHI under grant number JP21H02886 (K.I.) and AMED under grant numbers 21ek0109522 and 24ek0109705 (K.I.).