AAV hamartin gene therapy in a stochastic, cerebral mouse model of tuberous sclerosis type 1

Abstract

Tuberous sclerosis complex (TSC) is a dominantly inherited disease in which most individuals are born with one defective allele encoding for either hamartin (TSC1) or tuberin (TSC2), with a somatic loss of the other allele leading to abnormal neurodevelopment and upregulation of cell growth in susceptible tissues. Ninety percent of affected individuals have brain involvement, including epilepsy, cognitive impairment, autism, and/or sleep disorders. In the stochastic, cerebral mouse model of Tsc1, loss of function of hamartin is induced in the CNS by injection of an adeno-associated virus (AAV) vector encoding Cre recombinase into the cerebral ventricles of homozygous Tsc1^flox/flox mice at birth. In the brain, Tsc1 loss leads to increased proliferation of subventricular zone cells, disrupted neuronal migration and cortical cytoarchitecture, dysmyelination, and microglia-mediated inflammation, ultimately resulting in early mortality. Systemic administration of an AAV9 vector encoding human hamartin at postnatal day 21 significantly ameliorated these abnormalities at 3 and 6 weeks post-injection and markedly extended survival in this TSC1 mouse model. This work reveals the ability of hamartin replacement therapy to reverse some of the brain abnormalities caused by its loss in different cell types and provides support for the potential use of gene replacement therapy in the treatment of TSC1 patients.

Graphical abstract

Systemic AAV9-mediated hamartin delivery reversed key neuropathological defects and extended survival in Tsc1-deficient mice, supporting its therapeutic potential in tuberous sclerosis complex type 1.

Introduction

Tuberous sclerosis complex (TSC) is a hereditary disease affecting multiple organs, including skin, eye, heart, lung, and kidney with over 90% of affected individuals having neurologic manifestations—epilepsy, autism, cognitive disability, hydrocephalus, and mental health issues.^1,2,3 These neurologic/psychiatric symptoms are associated with neuropathological features, including cortical tubers, subependymal nodules, and subependymal giant cell astrocytomas, as well as alterations in morphology, number and positioning of many cell types, including neurons, astrocytes and oligodendrocytes.⁴ Affected individuals typically inherit one defective allele of TSC1 encoding hamartin or TSC2 encoding tuberin with these proteins acting as a complex with guanosine triphosphatase activation of Ras homolog enriched in the brain to downregulate activity of the mechanistic target of rapamycin complex 1 (mTORC1).⁵ If, at some time during development or later in life, a somatic mutation knocks out the complementary normal allele in a susceptible cell type, mTORC1 is activated leading to increased protein translation associated with enlargement and/or proliferation of affected cells, as well as changes in morphology and signaling. In the brain, the latter includes altered dendritic spine morphology, multiple axons per neuron, and hyperactive synapses,^4,6 as well as immature phenotypes of astrocytes with poor ability to control extracellular glutamate⁷ and oligodendrocytes with impaired myelin production.^3,8

We have focused on the evaluation of gene replacement therapy in a mouse model of Tsc1 in which hamartin function is lost in different cell types in the brain at birth in the homozygous Tsc1^flox/flox model⁹—termed a stochastic, cerebral model.¹⁰ This is achieved by intraventricular injection of adeno-associated virus vector serotype 1 (AAV1) encoding Cre recombinase at birth, in contrast to other mouse models where it is knocked out in all cells of a specific type in the brain of mouse embryos.^9,11,12,13 The stochastic cerebral model is more similar to knocking out hamartin expression in a variety of cell types in the brain in utero with electroporation of a Cre expression construct.¹⁴

In our 2019 study,¹⁰ we showed that AAV9-mediated hamartin replacement could dramatically improve survival and pathology in a Tsc1-deficient cerebral mouse model. A single intravenous (i.v.) dose of AAV9-hamartin extended median survival from 32 days to over 429 days. The treatment corrected several key brain abnormalities, including ventricular enlargement, which could lead to hydrocephalus, subependymal overgrowths, enlarged neuronal size, and elevated mTORC1 signaling. Mice also showed improved motor performance on the rotarod and steady weight gain. Importantly, there were no signs of toxicity or damage to other organs, and the therapeutic effects were long-lasting after just one dose.

Building on that work, we used a clinically optimized version of the vector (AAV9-CB6-hamartin, provided by BridgeBio) to further test and refine this approach. In this study, we added molecular analyses, magnetic resonance imaging (MRI) to assess ventricular volume and dysmyelinated lesions, and experiments to define the minimum effective dose to extend survival. We evaluated proliferation of cells in the subventricular zone (SVZ), neuronal morphology, and cortical architecture at P42, 3 weeks after injection of AAV9-hamartin vector. Myelination and microglial activation were assessed at multiple time points (P21, P42, and P63) to track long-term effects. These experiments showed significant improvements across all these parameters and confirmed a strong survival benefit, further supporting the promise of AAV9-hamartin as a therapeutic approach for TSC1.

Results

AAV9-hamartin treatment shows CNS expression and improves survival in a dose-dependent manner in Tsc1-deficient mice

To determine the minimum effective dose of the AAV9-CB6-hamartin vector (hereafter termed AAV9-hamartin, Figure S1A), we utilized an in vivo model of conditional Tsc1 deletion. Homozygous Tsc1^flox/flox mouse pups were injected intracerebroventricularly (i.c.v.) at postnatal day 0 (P0) with AAV1-CBA-Cre recombinase (2 × 10¹⁰ vector genome [vg] total, 1 μL per lateral ventricle) (Figure 1A), resulting in biallelic Tsc1 deletion in neurons, astrocytes, oligodendrocytes, subependymal cells, and other ventricular zone cell populations. Recombination spread from the injection site into adjacent subcortical and cortical regions. At P21, mice were randomized into five treatment groups and received retro-orbital (RO) injections of either phosphate-buffered saline (PBS) (group 1, n = 20) or escalating doses of AAV9-hamartin: group 2 (1 × 10¹² vg/kg, n = 18), group 3 (5 × 10¹² vg/kg, n = 21), group 4 (1 × 10¹³ vg/kg, n = 22), and group 5 (5 × 10¹³ vg/kg, n = 20) (Figure 1A). Survival was monitored until P120. PBS-treated mice (group 1) showed a median survival of 46 days. AAV9-hamartin conferred a dose-dependent rescue: median survival 52 days (group 2), 63 days (group 3), and 73 days (group 4). In group 5, most mice survived past the 120-day endpoint, indicating a robust survival benefit (Figure 1D; ∗∗p = 0.0012 for group 2 vs. group 1, ∗∗∗∗p < 0.0001 for groups 3–5 vs. group 1).

Figure 1.

Minimum effective dose determination and expression of Cre recombinase and human hamartin in Tsc1-deficient mice

(A) Schematic representation of the experimental setup. Tsc1-floxed mouse pups received i.c.v. injections of AAV1-Cre (2 × 10¹⁰ vg) into both cerebral lateral ventricles at P0 to induce Tsc1 deletion. At P21, mice were randomly assigned to five groups for RO injections at the medial canthus with PBS (group 1, n = 18) or escalating doses of AAV9-hamartin (group 2: 1 × 10¹² vg/kg, n = 18; group 3: 5 × 10¹² vg/kg, n = 21; group 4: 1 × 10¹³ vg/kg, n = 22; group 5: 5 × 10¹³ vg/kg, n = 20). Survival was monitored up to P120, at which point animals were euthanized. (B) RT-qPCR analysis of Cre recombinase transcript levels in cortex, hippocampus, and cerebellum from WT mice (not shown), mice treated with AAV1-Cre or AAV1-Cre + AAV9-hamartin (5 × 10¹³ vg/kg, n = 3 per group). Ct values are shown (inverted y axis); lower Ct indicates higher transcript abundance. Cre expression was detected in all regions, with no significant differences between cohorts across brain regions (p = 0.1, p = 0.2, p > 0.09). Data are presented as mean ± SD. Statistical analysis were made using two-tailed unpaired Mann-Whitney U test. DL, detection level. (C) RT-qPCR detection of hamartin transcript in the same brain regions. Signal was detected exclusively in AAV9-hamartin-treated mice, confirming transgene expression (∗∗p = 0.009, ∗∗p = 0.004, ∗∗∗p = 0.0004). Data are presented as mean ± SD. Statistical comparisons were made using unpaired t test with Welch’s correction. ND, not detected. (D) Kaplan-Meier survival analysis revealed dose-dependent increases in survival. Median survival times were 46, 52, 63, 73, and >120 days, for groups 1 through 5, respectively. Statistical significance was determined using the log rank Mantel-Cox test (∗∗p = 0.0012, ∗∗∗∗p < 0.0001).

To assess AAV transduction and transgene expression, we performed RT-qPCR for Cre and hamartin transcripts across the cortex, hippocampus, and cerebellum in wild-type (WT) mice (not shown), and mice injected with either AAV1-Cre alone or AAV1-Cre + AAV9-hamartin (5 × 10¹³ vg/kg).

Cre expression was detected in all three regions of both Cre-injected groups (Figure 1B). Cre Ct values were comparable between groups (mean Ct: cortex, 15.7 vs. 18.7; hippocampus, 16.5 vs. 18.5; and cerebellum, 20.8 vs. 20.7, respectively; p = 0.1 p = 0.16, p = 0.9). Overall, these results indicate balanced transduction and expression across groups in different brain regions.

Hamartin expression, detected exclusively in the AAV9-hamartin group, showed increased transcript levels in all regions (Figure 1C), with Ct values of 30.3 in the cortex, 30.0 in the hippocampus, and 31.6 in the cerebellum, consistent with widespread AAV9-mediated transgene delivery (∗∗p = 0.009, ∗∗p = 0.004, ∗∗∗p = 0.0004).

To validate RNA input, GAPDH Ct values were measured across all brain regions and groups (Figure S1B). For Cre expression, mean GAPDH Ct values were consistent: cortex (17.0 vs. 19.9), hippocampus (16.7 vs. 19.0), and cerebellum (17.0 vs. 17.3), with no significant intergroup differences (p = 0.1, p = 0.1, p = 0.11). Similarly, for hamartin expression, mean GAPDH Ct values remained stable across samples: cortex (21.0 vs. 23.7), hippocampus (20.9 vs. 23.5), and cerebellum (21.4 vs. 20.9), with no significant intergroup differences (p = 0.11, p = 0.1, p = 0.42). These results confirm uniform RNA input across groups in different brain regions.

In summary, consistent Cre and hamartin expression confirmed efficient and balanced vector delivery and transduction across brain regions. AAV9-hamartin administration achieved dose-dependent survival benefit in Tsc1-deficient mice, with widespread transgene expression across the CNS.

Widespread neuronal Cre expression induces cortical and hippocampal disorganization

To assess the spatial reach and cell-type specificity of Cre recombination in our Tsc1-floxed model carrying a Cre-inducible lacZ reporter, we performed X-gal staining on brains collected at P42 following neonatal AAV1-Cre injection. Strong β-galactosidase (β-gal) activity was seen across the forebrain regions including the cortex, hippocampus, and SVZ, confirming widespread recombination near and beyond the ventricular injection site (Figure 2A).

Figure 2.

Widespread neuronal Cre expression induces cortical and hippocampal disorganization

(A) Representative β-gal-stained coronal brain sections from AAV1-Cre-injected mice at P42 showed Cre recombinase activity in blue. Insets highlight specific ROI, including the hippocampus (left inset), cortex (left and right insets), and lateral ventricles (right inset). Scale bars, 1,000 μm and 100 μm (insets). (B) Immunofluorescent analysis of Cre recombinase (magenta) in the cortex colocalized with cell-type-specific markers: NeuN (neurons, green), GFAP (astrocytes, green), Olig2 (oligodendrocytes, green), and Iba1 (microglia, green). Arrowheads indicate colocalization. DAPI (blue) was used for nuclear staining. Scale bars, 100 μm. (C) Representative sections stained for NeuN (magenta) from WT, AAV1-Cre, and AAV9-hamartin-treated Tsc1-floxed mice at P42. Images were acquired across four anatomically matched regions: (1) cortex + hippocampus, (2) somatosensory cortex, (3) hippocampal CA1, and (4) layer 5 of the primary somatosensory cortex (S1L5). AAV1-Cre-injected mice and AA9-hamartin-treated mice displayed abnormal neuronal migration, uneven dispersion and cortical cytoarchitecture disruption (white and yellow arrowheads in 1), neuronal clusters and aggregates (white arrowheads in 2), attenuated CA1 sector of hippocampal formation (white and yellow arrowheads in 2 and 3). Dashed lines delineate cortical and hippocampal boundaries in 1, 2, and 3; S1L5 in 4. Brain images are annotated based on the Allen Mouse Brain Atlas. CA1 (cornu ammonis area 1), S1L5 (primary somatosensory cortex layer V). Scale bars, 100 μm.

We also used immunofluorescence to identify the types of cells expressing Cre. Most Cre-positive cells co-labeled with NeuN, indicating that they were neurons. Occasional Cre expression was observed in GFAP⁺ astrocytes, Olig2⁺ oligodendrocytes, and Iba1⁺-lineage cells, suggesting that, while neurons are the main target, there is some transduction of glial cells as well (Figure 2B). This pattern aligns with previous work showing that neonatal i.c.v. AAV1 delivery at P0 results in broad CNS expression, with strong neuronal tropism.¹⁵

Given the established role of TSC1 in cortical development and lamination,¹⁶ we next assessed the impact of conditional Tsc1 loss on neuronal architecture using NeuN and MAP2 staining at P42. In WT mice, NeuN immunostaining revealed well-defined cortical layers and normal hippocampal architecture, including the cornu ammonis area 1 (CA1) and layer V of the primary somatosensory cortex (S1L5) regions (Figure 2C). In contrast, AAV1-Cre-injected Tsc1-floxed mice exhibited marked neuronal disorganization, which suggested migration abnormalities. Neurons were unevenly dispersed with some clusters and aggregates suggestive of cortical dysplasia. CA1 sector of hippocampal formation appeared attenuated. This is consistent with prior findings in mouse models of Tsc1/Tsc2 loss, where aberrant positions of neurons and cytoarchitectural abnormalities were attributed to altered radial glial differentiation and migration defects during cortical development.¹⁷

Moreover, treatment with AAV9-hamartin at the highest dose (5 × 10¹³ vg/kg) did not markedly improve cortical disorganization and neuronal aberrant migration compared with WT (Figure 2C).

To support these observations, MAP2 staining (Figure S2A) was patchy with disorganized dendritic branching and irregular stratification. In WT brains, MAP2+ dendrites were organized in parallel arrays, consistent with intact cortical scaffolding. These disruptions reflect known roles of TSC1 in regulating cytoskeletal dynamics and neuron-glia interactions during development.¹⁸

Collectively, these data demonstrated that early postnatal Tsc1 deletion lead to pronounced defects in cortical organization, neuronal migration, and hippocampal integrity, and that systemic hamartin replacement did not notably rescue organization.

AAV9-hamartin treatment reduces increased cell proliferation in the dorsolateral SVZ of Tsc1-deficient mice

To assess the effect of Tsc1 loss and subsequent hamartin replacement on neural progenitor cell proliferation, we examined Ki67 expression in the dorsolateral SVZ¹⁵ at P42. In WT mice, Ki67+ cells were sparsely distributed along the ventricular wall (Figure 3A(i)), consistent with low physiological levels of neurogenic proliferation in this region.

Figure 3.

Ki67 immunostaining shows that cell proliferation in the dorsolateral SVZ of Tsc1-deficient mice is reduced by AAV9-hamartin treatment

(A) Representative sections showing Ki67 (red) and DAPI (blue) staining in the SVZ of the lateral ventricle in WT, AAV1-Cre, and AAV1-Cre + AAV9-hamartin-treated mice at P42. Arrowheads indicate Ki67-positive cells within the dorsolateral SVZ. AAV1-Cre-injected animals (ii) exhibited a marked increase in Ki67+ cells compared with WT (i), indicating elevated cell proliferation. This phenotype was attenuated in AAV1-Cre + AAV9-hamartin-treated mice (iii), suggesting rescue. Scale bars, 100 μm. (B) Quantification of Ki67+ cells in the dorsolateral SVZ (n = 3 mice per group). AAV1-Cre-injected mice showed significant increase in SVZ proliferation compared with WT, which was normalized by hamartin treatment. Brain image was adapted from the Allen Mouse Brain Atlas. Data are presented as mean ± SD. Statistical significance was determined using one-way ANOVA with Bonferroni’s post hoc test (∗p = 0.0171, ∗p = 0.0339).

In contrast, AAV1-Cre-injected Tsc1-floxed mice exhibited a marked increase in Ki67+ cells within the dorsolateral SVZ, with proliferating cells extending deeper into the parenchyma (Figure 3A(ii), arrowheads), suggesting aberrant progenitor morphology. Quantification revealed a significant increase in Ki67+ cells in the AAV-Cre-injected group (average 112 cells) compared with WT (average 66 cells; ∗p = 0.0171; Figure 3B).

Notably, treatment with AAV9-hamartin at 5 × 10¹³ vg/kg significantly reduced this proliferative response (average 72 cells; ∗p = 0.0339 vs. AAV1-Cre), although levels remained modestly higher than WT, indicating incomplete normalization of the SVZ proliferation phenotype (Figure 3A(iii) and 3B). Nested analysis (Figure S2B) quantifies the number of Ki67-positive cells per ventricle, with four ventricles analyzed per mouse and three mice per group. These findings suggest that hamartin replacement effectively dampened aberrant SVZ proliferation induced by Tsc1 loss. Some of the observed changes may reflect downstream consequences of disrupted ependymal cell maturation, which can impair cerebrospinal fluid (CSF) circulation and contribute to ventricular enlargement. Moreover, given the broad tropism of AAV1-Cre, unintended transduction of choroid plexus cell types could alter CSF production and further influence ventricular volume.¹⁹

AAV9-hamartin treatment normalizes ventricular volume and partially reduces T2-hyperintense lesion burden in Tsc1-deficient mice

To evaluate structural brain abnormalities and treatment response, we performed high-resolution MRI in living mice at P42. Tsc1-floxed mice injected with AAV1-Cre exhibited a marked reduction in total ventricular volume compared with WT controls, consistent with previous reports linking Tsc1 loss to abnormal neuroprogenitor proliferation and impaired ependymal maturation.²⁰ The mean ventricular volume in AAV1-Cre mice was significantly decreased (mean log₁₀: 0.74) compared with WT (mean log₁₀: 1; ∗p = 0.0348), while AAV9-hamartin treatment restored ventricular size to near-WT levels (mean log₁₀: 0.95; ∗p = 0.0411) (Figures 4A and 4B; Table S1).

Figure 4.

MRI analysis reveals AAV9-hamartin treatment restores ventricular volume and partially reduces T2-hyperintense lesion burden in Tsc1-deficient mice

(A) Representative T2-weighted MRI scans showing enlarged lateral ventricles in AAV1-Cre-injected mice (middle panel) compared with WT controls (left), with normalization of ventricular volume following AAV9-hamartin treatment (right; red arrows) at P42. (B) Quantification of lateral ventricular volume (mm³) showed a significant reduction in AAV1-Cre-injected mice (n = 5) compared with WT (n = 3), and restoration to near-WT levels in AAV1-Cre + AAV9-hamartin-treated mice (n = 6). Data are log₁₀ transformed, shown as mean ± SD. Statistical significance was determined using one-way ANOVA with Dunnett’s post hoc test (∗p = 0.0348, ∗p = 0.0411). (C) Representative MRI scans highlighting increased T2-hyperintense lesions representing dysmyelination in AAV1-Cre-injected and AAV1-Cre + AAV9-hamartin-treated mice (red arrows, middle and right) compared with WT (left) at P42. (D) Quantification of mean total lesions size per mouse revealed a significant increase in AAV1-Cre-injected mice (n = 5) compared with WT (n = 3), with a trend toward reduction in AAV9-hamartin-treated animals (n = 6). Data are log₁₀ transformed, shown as mean ± SD. Statistical significance was determined using one-way ANOVA with Dunnett’s post hoc test (∗p = 0.0156, p = 0.1192).

We also assessed the size of T2-hyperintense lesions, which is indicative of dysmyelination within the brain parenchyma. AAV1-Cre-injected mice displayed a substantial increase in total T2 lesion size (mean log₁₀: 1.47) relative to WT (mean log₁₀: 0.65; ∗p = 0.0156). In mice treated with AAV9-hamartin, lesion size was reduced to mean log₁₀: 1.04. However, this did not reach statistical significance compared with AAV1-Cre alone (p = 0.1192) (Figures 4C and 4D; Table S1). To contextualize the statistical findings, the corresponding raw values were as follows: mean ventricular volume was 10.3 mm³ in WT mice, 5.9 mm³ in AAV1-Cre mice, and 9.0 mm³ in AAV1-Cre + AAV9-hamartin-treated mice. For T2-hyperintense lesion area, mean values were 4.6 mm² in WT mice, 40.6 mm² in AAV1-Cre-injected mice, and 14.4 mm² in the hamartin-treated group.

Collectively, these MRI findings suggested that systemic delivery of AAV9-hamartin could reverse ventricular compression at P42 and partially reduce lesion burden following early Tsc1 loss, supporting its therapeutic potential.

AAV9-hamartin treatment restores myelination in Tsc1-deficient mice over time

We performed immunohistochemistry for myelin basic protein (MBP) in WT, AAV1-Cre-, and AAV1-Cre + AAV9-hamartin-treated mice at P21, P42, and P63 to assess whether hamartin gene replacement could reverse myelination deficits caused by Tsc1 loss (Figure 5). At P21, AAV1-Cre mice exhibited significantly reduced MBP-positive area in the dorsal forebrain (including cortex and corpus callosum) (17.9%, mean log₁₀: 1.23) and hippocampus (7.0%, mean log₁₀: 0.84) compared with WT (26.3%, mean log₁₀: 1.41 and 12.5%, mean log_10: 1.09, respectively; ∗p = 0.03, ∗p = 0.02), indicating early myelination defects (Figures 5A-i, 5B, and 5C). This deficit persisted at P42 (dorsal forebrain: 19.7%, mean log₁₀: 1.28; hippocampus: 6.1%, mean log₁₀: 0.77, vs. WT: 25.6%, mean log₁₀: 1.4 and 11%, mean log₁₀: 1.02, respectively; ∗p = 0.03, ∗p = 0.02) but was partially restored in AAV9-hamartin-treated animals (dorsal forebrain: 26.2%, mean log₁₀: 1.38; hippocampus: 7.8%, mean log₁₀: 0.89; p = 0.19, p = 0.14) (Figures 5A-ii, 5B, and 5C). By P63, myelin coverage in treated mice reached levels comparable with WT in both regions (dorsal forebrain: 42.1%, mean log₁₀: 1.62; hippocampus: 24.2%, mean log₁₀: 1.28, vs. 43.2%, mean log₁₀: 1.63 and vs. 19.6%, mean log₁₀: 1.38), with no significant differences (p > 0.99, p = 0.99) (Figures 5Aiii, 5B, and 5C).

Figure 5.

AAV9-hamartin treatment rescues myelination deficits in Tsc1-deficient mice over time

(A) Representative brain sections showing immunostaining for MBP (red) at P21 (i), P42 (ii), and P63 (iii). Images are shown for WT, AAV1-Cre-injected, and AAV1-Cre + AAV9-hamartin treatment groups. The thresholded MBP channel used for quantification is shown in the far-right column. Dashed lines outline MBP level in the cortex. Scale bars, 1,000 μm. (B) Quantification of percent MBP-positive area in the dorsal forebrain (cortex + corpus callosum) across treatment groups and time points (n = 3 per group). At P21 and P42, AAV1-Cre-injected mice exhibited significantly reduced MBP signal compared with WT controls (∗p = 0.03 for both time points). At P42, AAV9-hamartin treatment resulted in partial recovery that did not reach statistical significance (p = 0.19). By P63, MBP levels in treated mice were comparable with WT (p > 0.99). Data are log₁₀ transformed, shown as mean ± SD. Statistical significance was determined using two-way ANOVA with Sidak’s post hoc test. D, death. (C) Quantification of MBP-positive area in the hippocampus across treatment groups and time points (n = 3 per group). AAV1-Cre-injected mice showed significantly reduced MBP signal compared with WT at P21 and P42 (∗p = 0.02 for both). AAV9-hamartin treatment showed a trend toward recovery at P42 but did not reach significance (p = 0.14). By P63, MBP levels in treated mice remained statistically comparable with WT (p = 0.99). Data are shown as mean ± SD. Statistical significance was determined using two-way ANOVA with Sidak’s post hoc test. D, death.

When comparing within-group changes over time, WT mice exhibited a significant increase in MBP between P42 and P63 in both the dorsal forebrain (25.6%, mean log₁₀: 1.4 to 43.2%, mean log₁₀: 1.63; ∗∗p = 0.001) and hippocampus (11.0%, mean log₁₀: 1.02 to 19.6%, mean log₁₀: 1.28; ∗∗∗p = 0.0001), consistent with normal developmental myelination. AAV1-Cre mice could not be evaluated at P63 due to early mortality (D). AAV1-Cre + AAV9-hamartin-treated animals showed a similar pattern of progressive myelin recovery, with MBP-positive area increasing at P42 from 26.2%, mean log₁₀: 1.39 to 42.1%, mean log₁₀: 1.62 in the dorsal forebrain at P63 (∗∗p = 0.001) and from 7.8%, mean log₁₀: 0.89, at P42, to 24.2%, mean log₁₀: 1.38 in the hippocampus at P63 (∗∗∗p < 0.0001), reaching levels indistinguishable from WT by the final time point.

These findings demonstrated that AAV9-hamartin treatment not only reversed early myelin loss but promoted sustained, developmentally appropriate myelin recovery over time.

AAV9-hamartin treatment modulates microglial activation and potentially induces a pro-regenerative phenotype following Tsc1 deletion

To probe the inflammatory dynamics associated with myelin repair, we co-stained brain cells for Iba1, a microglial marker, and CD68, a transmembrane glycoprotein expressed by monocytes and macrophages, associated with phagocytic activity and prominently detected in microglia within white matter lesions, where it suggests active clearance of myelin debris.²¹ In the hemi-section of mouse brain treated with AAV1-Cre + AAV9-hamartin (5 × 10¹³ vg/kg), we observed substantial colocalization of Iba1 (red) and CD68 (green), consistent with the presence of activated microglia. This colocalization was most pronounced in the cortical region and white matter-enriched regions including the corpus callosum and internal capsule, suggesting regional engagement of microglia in phagocytic clearance and tissue remodeling (Figures 6A and S3). We subsequently quantified CD68 expression in the hemisphere of WT, AAV1-Cre, and AAV1-Cre + AAV9-hamartin-treated mice at P21, P42, and P63 (Figure 6B). While resting microglia in WT brain express CD68, it was upregulated upon activation. AAV1-Cre-injected mice exhibited a robust increase in CD68 integrated density at P21 (4.5 × 10⁴ arbitrary units [a.u.], mean log₁₀: 5.21) and P42 (6.1 × 10⁵ a.u., mean log₁₀: 5.51) compared with WT (5.4 × 10³ a.u., mean log₁₀: 3.71, and 1.5 × 10⁵ a.u., mean log₁₀: 4.66, respectively; ∗∗p = 0.0092), indicating early and sustained microglial activation. AAV1-Cre + AAV9-hamartin-treated animals showed a reduction in CD68 expression at P42 (1.2 × 10⁵ a.u., mean log₁₀: 4.87; p = 0.42), suggesting partial suppression of microglial reactivity. By P63, WT animals exhibited a physiological increase in CD68 compared with P42 (8.3 × 10⁵ a.u., mean log₁₀: 5.91; ∗∗p = 0.001), while AAV9-hamartin-treated mice displayed a further rise compared with P42 (4.5 × 10⁶ a.u., mean log₁₀: 6.53; ∗∗p = 0.001), consistent with a prolonged remodeling phase in response to therapy. At P63, although the CD68 increase in treated mice was higher compared with WT, it was not significant (p = 0.67). To characterize the nature of this late-phase microglial activation, we performed immunofluorescent staining for microglial identity and polarization markers in AAV9-hamartin-treated brains at P63 (Figure 6C). In coronal sections from WT and treated animals, widespread Iba1⁺/CD68⁺ microglia were observed in the cortex, corpus callosum, and internal capsule (Figures 6A and S3).²² Importantly, CD68⁺ cells colocalized with IL-10, a canonical anti-inflammatory cytokine,²³ as well as CD206, a mannose receptor enriched in tissue remodeling macrophages,^{24,25,26,27,28} and ARG1, an enzyme whose catalytic activity produces polyamines that support extracellular matrix repair²⁹ (Figure 6C). The presence of these markers across the cortex, corpus callosum, and internal capsule, as well as their relative absence in WT controls, suggests that AAV9-hamartin promoted a non-inflammatory, pro-repair microglial phenotype by P63.²⁹ Together, these findings indicate that AAV9-hamartin treatment not only limited early inflammatory activation, but induced a reparative microglial state that may have supported tissue remodeling and long-term myelin regeneration.

Figure 6.

AAV9-hamartin treatment potentially induces a shift toward an activated, reparative microglial phenotype

(A) Representative brain section showing immunostaining for Iba1 (microglial marker, red) and CD68 (activation marker, green) in the brain hemi-section of AAV1-Cre-injected and AAV9-hamartin (5 × 10¹⁰ vg/kg)-treated mice at P63. The merged panel demonstrates regional overlap of CD68 with Iba1-positive microglia (white arrowheads), indicating activated microglial populations in cortex, corpus callosum, and internal capsule. Brain image was adapted from the Allen Mouse Brain Atlas. Scale bar, 1,000 μm. C, cortex; CC, corpus callosum; IC, internal capsule. (B) Quantification of CD68 integrated density in the hemisphere of WT, AAV1-Cre-injected, and AAV1-Cre + AAV9-hamartin-treated mice (n = 3 per group) at P21, P42, and P63. AAV1-Cre-injected mice displayed significantly increased CD68 expression at P21 and P42 compared with WT (∗∗p = 0.0092). AAV9-hamartin treatment partially attenuated CD68 expression at P42 (p = 0.42). At P63, both WT and AAV9-hamartin-treated mice showed increased CD68 signal compared with P42 (∗∗p = 0.001), but the difference between treated and WT animals was not statistically significant (p = 0.67). Data are log₁₀ transformed, shown as mean ± SD. Statistical significance was determined using two-way ANOVA with Sidak’s post hoc test. (C) Immunostaining in the cortex, corpus callosum, and internal capsule of WT and AAV9-hamartin-treated mice at P63 showed CD68 colocalization with different microglial markers only in the treated mice. Left panel (cortex): CD68 (green) colocalized with CD206 (red). Right panel (internal capsule): CD68 (green) colocalized with IL-10 (red). Bottom panel: CD68 (green) colocalized with ARG1 (red). Merged panels and insets highlight areas of signal overlap. Scale bars, 100 μm and 10 μm (insets). D, death.

Discussion

In individuals with TSC, disease typically results from a germline heterozygous mutation in either TSC1 or TSC2. Focal brain lesions such as cortical tubers or subependymal nodules are believed to arise through a somatic second-hit event, often involving loss of heterozygosity in a restricted subset of cells.^30,31 These second-hit mutations can occur during fetal development, most likely in radial glial progenitors during mid-gestation, leading to localized biallelic inactivation and the formation of dysplastic regions.^1,32

To model this in mice, we used an AAV1-Cre approach to induce conditional Tsc1 loss by injecting this vector-producing Cre recombinase into the lateral ventricles at P0 of Tsc1^flox/flox mice. This stage of murine development corresponds to approximately 23–32 weeks of human gestation, a period of active gliogenesis and cortical organization, and is thus considered developmentally analogous to mid-to-late second trimester in humans.^33,34 At this time point, AAV1 preferentially transduces periventricular radial glial cells and other neuroglial precursors in the ventricular zone,¹⁵ cell populations that have also been implicated as the origin of human cortical tuber pathology.

While this mouse model does not replicate the focal and mosaic nature of human lesions, the broader recombination pattern induced by ventricular AAV-Cre results in a larger population and different types of Tsc1-deficient cells, providing a robust and reproducible system to examine how early biallelic Tsc1 loss affects cortical development, myelination, and glial responses. Additionally, this strategy enables postnatal temporal control of gene deletion, circumventing the embryonic lethality associated with early, pan-neural knockout mouse models.^16,35

Although many of the symptoms of TSC can be remedied by continuous administration of rapalog drugs that inhibit mTORC1 activity, these drugs are not totally effective at remedying some of the symptoms, including epilepsy, cognitive problems, and lymphangioleiomyomatosis.^36,37,38 Furthermore, these drugs can potentially compromise brain development in infants and children and are immune suppressive.^39,40 Clearly new therapeutic approaches are needed. Gene replacement therapy using AAV vectors could be used in combination with drugs and neurosurgical therapies, and AAV9 gene replacement therapy has the advantage that it has proven effective in human diseases.^41,42 AAV9-hamartin can replace this protein in deficient cells and potentially provide “extra” protective copies to heterozygous cells. This study demonstrates that several of the brain abnormalities caused by loss of hamartin in different cell types in the mouse brain during early postnatal development can be normalized by systemic delivery of an AAV9-hamartin vector which crosses the blood-brain barrier. Furthermore, at high doses of this vector under a strong constitutive promoter there was marked extension of lifespan.

Most mouse models of TSC1/2 with brain involvement show early death thought to be due to hydrocephalus or epilepsy.¹⁴ In the stochastic, cerebral model, the timing of death can be controlled by the dose of AAV1-Cre-injected i.c.v. At a dose of 2 × 10¹⁰ vg at birth, mice lived on average 46 days with the cause of death not determined. A subset of animals developed a hunched posture accompanied by decreased grooming and hypoactivity prior to death, consistent with hydrocephalus, but others did not exhibit any signs of illness. Lifespan was extended in about 80% of these mice to at least 120 days when AAV9-CB6-hamartin was given i.v. at a dose of 5 × 10¹³ vg/kg on P21, with the surviving animals appearing healthy and active. This dose of therapeutic vector is in the range found to be effective in other mouse models of neurologic disease with AAV9.^43,44 It typically takes an AAV vector about 7 days to express its transgene after entry into a cell,⁴⁵ so it is not clear whether earlier injection might have been more effective. Postnatal age 21 days in mice is equivalent to 6 months gestational age in humans⁴⁶ with prenatal gene therapy currently not an option in humans. Many of the symptoms of TSC, such as epilepsy and mental deficits arise from neurodevelopmental abnormalities in the brain, which could probably not be fully normalized after birth. Still, from a neurologic perspective, gene replacement therapy in infancy or childhood might prevent exacerbation of symptoms, for example, by increased myelination, and reduced seizure frequency.

In the AAV1-Cre-injected Tsc1-floxed mice, we observed neuronal abnormalities in regions near the ventricles, including the cortex and hippocampus. Neuronal staining showed disorganized cortical layers and disrupted hippocampal structure, with neurons appearing misaligned, unevenly spaced, and dysplastic. Dendritic architecture was also affected, with fragmented, shortened processes that were less radially oriented. These features suggest that loss of Tsc1 interferes with proper neuronal positioning and dendritic development. After AAV9-hamartin treatment, abnormalities were marginally improved with neuronal layers appearing more defined. The persistence of some defects caused by early Tsc1 loss suggests that earlier intervention might be needed for more complete correction of neuronal development and migration. The unchanged neuronal density in the cortex following AAV9-hamartin treatment suggests that hamartin overexpression is well tolerated by cells, likely because its activity is regulated through binding to normal levels of endogenous tuberin.

Neuropathological features of tuberous sclerosis include subependymal nodules (SENs), which form in infancy and can progress to subependymal giant cell astrocytoma (SEGA), with 10% of TSC patients susceptible to hydrocephalus without drug or neurosurgical intervention.³ Emerging evidence suggests that SEGAs may not always develop from SENs but could arise from a distinct progenitor population with a unique transcriptional profile.⁴⁷ By staining for the proliferation marker Ki67, around the lateral ventricles, we noted an increase in dividing cells after i.c.v. AAV-Cre injection, which was normalized after i.v. injection with AAV9-hamartin. The latter correlated with a significant decrease in ventricular volume after AAV-Cre injection, as determined by MRI of living animals. This was normalized after AAV9-hamartin injection. We hypothesize that overgrowth of the ventricular layer would reduce ventricular volume until it reached an extent where it would interfere with flow of cerebral spinal fluid, potentially leading to hydrocephalus and death.

In vivo MR imaging also revealed abnormal T2-hyperintense regions in the brains of AAV-Cre-treated mice, which suggested abnormal myelination with a trend toward normalization after AAV9-hamartin injection at day 42. In 95% of humans with TSC, MRI revealed a range of white matter abnormalities, including nonspecific conglomerate foci, similar to those seen in the MRI of our mice.^48,49,50 Most white matter abnormalities in humans are related to gliosis and hypomyelination associated with cortical tubers⁵⁰ and are typically hyperintense on T2-weighted sequences after maturation of myelin.⁵¹ Pathologically, resected tubers from TSC patients have shown varying levels of MBP in gray and white matter,⁵² and immunohistochemical studies have demonstrated upregulation of pS6 and p4E-BP1 proteins in white matter regions.⁵³ Myelination defects are commonly observed in TSC brains, both focally in tubers and diffusely throughout the brain.^54,55 In mouse models of TSC1 and TSC2, an increase in pS6 and p4E-BP1 have been demonstrated within white matter tracts, along with a decrease in myelin-associated proteins, such as MBP, again consistent with a hyperactive mTOR pathway and disrupted myelination.⁵⁶ In a study involving the Tsc1-floxed alleles and a synapsin I promotor-driven Cre allele to eliminate Tsc1 in differentiating neurons, reduced myelination, as evidenced by decreased MBP expression, was observed.⁵⁷

Our study shows that Tsc1 loss during early mouse development leads to dysmyelination and microglial activation. At early time points (P21 and P42) in AAV1-Cre-injected mice, MBP staining was significantly reduced in the dorsal forebrain and hippocampus, suggesting a failure in early myelin formation. These results align with previous studies showing that mTORC1 hyperactivation disrupts oligodendrocyte differentiation and impairs myelination.⁵⁸ Notably, hamartin treatment improved MBP expression by P42 and fully restored it by P63, indicating that gene replacement can rescue myelination, although this process appears to require time to take effect.

In parallel, we observed an increase in CD68 expression in AAV1-Cre-injected brains, particularly at P21 and P42, pointing to heightened microglial activation. While this likely reflects a reactive response to myelin loss and other aspects of disrupted brain development, it also suggests that neuroinflammation may be contributing to the delayed myelination. CD68 is a well-established marker of activated, phagocytic microglia, particularly in white matter.²¹ Interestingly, CD68 levels remained elevated in hamartin-treated animals even at P63, which may reflect a shift in microglial function, being less about inflammation and more about supporting repair.

Microglia are not simply responders to injury; they are active participants in CNS development and remodeling. Arising from yolk sac progenitors,⁵⁹ they help shape brain circuits and myelination through debris clearance and cytokine signaling. Depending on the environment, microglia can take on pro- or anti-inflammatory roles. Markers like CD206, IL-10, and ARG1, which we detected in AAV9-hamartin-treated but not in WT mice, are associated with a reparative phenotype.²⁹ Prior work has also shown that mTORC1 signaling can influence microglial polarization and function, and that inhibition of this pathway, whether by rapamycin or genetic deletion, promotes an anti-inflammatory state and supports recovery in models of CNS injury.^60,61 Altogether, these findings suggest that AAV9-hamartin not only promotes remyelination but may also shape the microglial response in a way that favors tissue repair. Importantly, abnormal and reduced myelination has been correlated with epileptic activity.⁶² Since myelination is active in mouse brains up to P60⁶³ and that the gene therapy-treated mice live much longer, it is consistent that myelination was restored to a greater extent over time in our Tsc1 mouse model. This is highly relevant to human TSC, since restored myelination can reduce seizures. In humans, myelination peaks during the first year of life but continues into the 20s and 30s, thus leaving a window to restore myelination in TSC patients and possibly reduce epilepsy.

In summary, these findings in a stochastic, cerebral mouse model of TSC1 support the potential of gene replacement therapy for TSC1. Several aspects stand out. First, the extent of recovery of lifespan, reduction in proliferation of ventricular cells, and remyelination in response to AAV9-hamartin therapy suggest that even replacing hamartin in a subset of the cells that have lost it can have dramatic effects in the brain. These studies support further investigation into the potentially therapeutic effect of AAV gene replacement in alleviating symptoms in TSC1 patients, including epilepsy⁸ and possibly associated neuropsychiatric disorders.⁵²

Materials and methods

AAV design and production

AAV1-CBA-Cre (AAV1-Cre) and AAV9-CB6-TSC1 (AAV9-hamartin) were designed and provided by BridgeBio. AAV1-Cre contains the Cre recombinase gene under the control of the chicken β-actin (CBA) promoter with a bovine growth hormone polyadenylation signal. AAV9-hamartin encodes the human TSC1 gene (hamartin) under the CB6 promoter with a rabbit β-globin polyadenylation signal.

Both vectors were produced via triple transfection of a HEK293 suspension cell line cultured in Expi293 Expression Medium (Thermo Fisher Scientific, Waltham, MA) at 37°C and 5% CO₂, and 130 rpm within Optimum Growth flasks (Thomson, Oceanside, CA). Transfection was performed using PEImax (Polysciences, Warrington, PA). Seventy-two hours after transfection, lysis buffer containing 20% Tween 20 (VWR, Radnor, PA), 40 mM magnesium chloride (Sigma-Aldrich, St. Louis, MO), and 1 M Tris (Sigma-Aldrich) at pH 8.0 was added to a final concentration of 1% Tween 20, 2 mM magnesium chloride, and 50 mM Tris. Salt-active nuclease (SAN) (Arcticzymes, Tromsø, Norway) was added to a final concentration of 25 U/mL and sodium chloride (Promega, Madison, WI) was added to a final concentration of 0.5 M for digestion of nucleic acids. Finally, the lysate was 0.2 μm filtered prior to affinity chromatography.

AAV purification by affinity chromatography

Clarified lysate was purified across POROS CaptureSelect AAV9 or AAVX Affinity column (Thermo Fisher, Rockford, IL) and eluted via a low pH elution. The elution was collected, pooled, neutralized, and stored at 2°C–8°C overnight prior to anion-exchange chromatography. The affinity eluate was adjusted to pH 9 and loaded across a POROS 50 HQ Strong Anion Exchange Column (Thermo Fisher). Post-loading, the column was subjected to a sodium acetate gradient to elicit elution of the full capsid particles. The eluate was subsequently concentrated using a 100 kDa Amicon ultra centrifugal spin filter (Sigma-Aldrich) prior to sterile 0.2 μm filtration.

Determination of genome titer

Viral genome titer was determined by droplet digital polymerase chain reaction (ddPCR). Samples were treated with SAN for 30 min at 37°C, followed by Proteinase K digestion at 55°C for 30 min, and finally an incubation at 95°C for 15 min to inactive the enzymes. Samples were diluted and combined with Bio-Rad ddPCR Supermix (Bio-Rad, Hercules, CA) for Probes (no dUTP) and a polyadenylation signal-specific primer-probe set. After droplet generation, samples were immediately subjected to PCR: 95°C for 10 min, 45 cycles of 95°C for 30 s and 60°C for 1 min, 98°C for 10 min, and a 30-min hold at 4^oC. Droplet detection was conducted on a Bio-Rad QX200 droplet reader using FAM detection and data were analyzed using the accompanying QuantSoft Analysis Pro software.

Animals and injections

Experimental research protocols were approved by the Institutional Animal Care and Use Committee for the Massachusetts General Hospital (MGH) following the guidelines of the National Institutes of Health for the Care and Use of Laboratory Animals. Experiments were performed on Tsc1^flox/flox mice, which also carried the Cre-inducible ROSA26 lacZ marker allele, as described in Meikle and co-workers.^57,64 In response to Cre recombinase, the Tsc1^flox/flox allele is converted to a null allele, and the lacZ allele expresses β-gal. These mice have a normal and healthy lifespan.

For vector injections, in the neonatal period (P0 to P1), pups were cryo-anesthetized and injected with 1 μL of viral vector AAV1-CBA-Cre into each cerebral lateral ventricle (2 × 10¹⁰ vg) with a glass micropipette (70–100 mm in diameter at the tip) using a Narishige IM300 microinjector at a rate of 2.4 psi/s (Narshige International, East Meadow, NY). Mice were then placed on a warming pad and returned to their mothers after regaining normal color and full activity typical of newborn mice.

At age 3 weeks (P21), mice were anesthetized with isoflurane (Baxter Healthcare, Deerfield, IL) inhalation (3.5% isoflurane in an induction chamber and then maintained anesthetized with 2%–3% isoflurane and oxygen [1–2 L/min] for the duration of the injection). AAV vectors were injected RO into the vasculature in a volume of 60 μL AAV9-CB6-hamartin in four doses 1 × 10¹², 5 × 10¹², 1 × 10¹³, and 5 × 10¹³ vg/kg using a 0.3-mL insulin syringe over less than 2 min⁶⁵ or buffer.

Animals were bred in-house and were housed under controlled temperatures with a 12-h light/dark cycle. The mice were given standard animal feed, and water was provided ad libitum.

RNA extraction and reverse transcription

Total RNA was extracted from the cortex, hippocampus, and cerebellum of three mice per experimental group (WT non-injected, AAV1-Cre only injected, and AAV1-Cre and AAV9-hamartin injected) using the QIAGEN miRNeasy Micro Kit (catalog no. 217084) according to the manufacturer’s protocol. RNA concentration and purity were assessed using a NanoDrop spectrophotometer (ND-1000).

To generate cDNA, 400 ng of total RNA from each sample was reverse transcribed using the SuperScript IV VILO Master Mix with ezDNase Enzyme (Thermo Fisher Scientific, catalog no. 11766050), following the manufacturer’s protocol. This single-step method integrates DNase treatment to remove any residual genomic DNA contamination during reverse transcription. Each reaction was carried out in a 20 μL volume containing 4 μL of SuperScript IV VILO Master Mix with ezDNase Enzyme, 400 of RNA, and RNase-free water to volume. Reactions were incubated as follows: 25°C for 10 min (ezDNase step), 50°C for 10 min (reverse transcription), and 85°C for 5 min to inactivate the enzyme.

RT-qPCR

RT-qPCR was conducted using both SYBR Green and TaqMan chemistries to assess gene expression levels. For SYBR Green assays, PowerUp SYBR Green Master Mix (Applied Biosystems) was utilized on a QuantStudio 3 Real-Time PCR System (Applied Biosystems). Each 20 μL reaction contained 10 μL of SYBR Green Master Mix, 2 μL of forward primer, 2 μL of reverse primer (final concentration of 400 nM each), and 0.5 μL of cDNA (10 ng). The thermal cycling protocol was 50°C for 2 min and 95°C for 2 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. SYBR Green primers for GAPDH were obtained from Origene (http://www.origene.com/), while primers for Cre recombinase were custom-designed using SnapGene software.

For TaqMan assays, qPCR was performed using TaqMan Fast Advanced Master Mix (Applied Biosystems) and pre-designed TaqMan Gene Expression Assays (Thermo Fisher Scientific). Each 20 μL reaction included 10 μL of TaqMan Master Mix, 1 μL of TaqMan Gene Expression Assay, and 0.3 μL of cDNA (6 ng). Thermal cycling conditions were 50°C for 2 min, 95°C for 20 s, followed by 40 cycles of 95°C for 1 s and 60°C for 20 s. TaqMan primer and probe assays included pre-designed GAPDH (Thermo Fisher Scientific) and hamartin primer and probe sequence provided by BridgeBio.

A complete list of primers and probes used is provided in Table S2. All reactions were performed in triplicate.

MRI

Quantification of demyelinated areas on MRI was assessed by a board-certified neuroradiologist with more than 20 years of experience blinded to the identity of the mice, using Horos v.4.0 (https://horosproject.org/). T2-hyperintense areas in the entire brain parenchyma were first segmented using the pencil region of interest (ROI) tool and then summed for each mouse.

MRI of mouse brains were imaged on a Bruker (Billerica, MA) 7.0 Tesla MRI with a 2.2 cm diameter volume coil using a 3D TurboRARE sequence (repetition time (TR) 1,800 ms; echo time (TE). 8.6 ms; RARE factor, 16) with an isotropic voxel size of 130 μm. The mouse brains were segmented using a supervised threshold method using Amira software (Thermo Fisher Scientific). CSF was then manually segmented, in areas where partial volume effect was not pervasive, to find the ROI statistics of the CSF. After which, the volume of the ventricles was segmented by using a threshold that was higher than 3 standard deviations (SDs) of the mean of the CSF ROI in the segmented brain.

Immunohistochemistry and image acquisition

Mice were euthanized at days 21, 42, and 63 using ketamine/xylazine (100:10; Akorn, Lake Forest, IL), followed by transcardiac perfusion with PBS. Brains were post-fixed overnight at 4°C, cryoprotected in 30% sucrose, and embedded in Tissue-Tek O.C.T. Compound (Sakura Finetek, Torrance, CA). Coronal sections (12 μm thick) were prepared and directly mounted on glass slides. Sections were blocked with 1% BSA and 5% goat serum in 1× PBS containing 0.1% Triton X-100 for 1 h at room temperature (RT) and incubated overnight at 4°C with primary antibodies, including anti-Ki67, anti-NeuN, anti-MBP, anti-Olig2, anti-CD68, anti-GFAP, anti-Iba1, anti-CD68, anti-CD206, anti-IL-10, anti-ARG1, and anti-Cre recombinase. Details of antibody dilutions and sources are provided in Table S3.

Following three washes with 1× PBS, sections were incubated with Alexa Fluor 555 or Alexa Fluor 647-conjugated secondary antibodies (1:500; Invitrogen) for 1 h at RT. Sections were washed three times with 1× PBS, and autofluorescence around the lateral ventricles (with Ki67 staining) was quenched using TrueBlack (catalog no. 23007, Biotium, Fremont, CA). Slides were mounted using ProLong Diamond Antifade Mountant with DAPI (P36962, Invitrogen).

Imaging was conducted using a Keyence BZ-X800 microscope (KEYENCE Corporation of America, Itasca, IL) and a NIKON CSU-W1 spinning disk confocal microscope.

Image processing, analysis, and quantification

Cell proliferation within the dorsolateral SVZ of the lateral ventricles was assessed by counting Ki67-positive cells per field of view. For each mouse, two non-consecutive coronal sections (totaling four lateral ventricle regions per mouse) were analyzed at 10× magnification. The analysis was performed in biological triplicates to ensure reproducibility.

A ROI encompassing the dorsolateral SVZ was defined for each section. Quantification was conducted using the "Find Maxima" algorithm integrated into the Fiji (ImageJ) software,⁴¹ which enables automated detection and enumeration of local maxima corresponding to Ki67-positive nuclei.

The results are presented as the average number of Ki67-positive cells per mouse (Figure 3B). Additionally, Figure S2B provides a nested analysis of the average number of Ki67-positive cells per ventricle.

For MBP and CD68, quantification was performed on three biological replicates per group at three developmental time points (postnatal days 21, 42, and 63), using comparable ROI and imaging parameters across samples. Full-brain stitched images were obtained at 4× magnification, capturing the entire coronal plane at the hippocampal level (bregma −1.5 mm to −2.5 mm) to allow for comprehensive regional analysis. The analysis was performed in biological triplicates to ensure reproducibility. All images were analyzed using ImageJ (Fiji, v.2.16.0/1.54p).

For quantification of MBP, images were first converted to 8-bit grayscale using Fiji (ImageJ). A consistent thresholding method was applied across all samples to minimize inter-sample variability. ROI were manually defined in anatomically matched areas of the dorsal forebrain (including cortex and corpus callosum) and hippocampus. The MBP-positive area fraction was calculated as the percentage of the ROI occupied by MBP signal using the “Analyze → Measure” function in Fiji.

CD68 expression, used as a marker of microglial activation, was quantified by calculating the total integrated density (sum of pixel intensity values) within one anatomically defined brain hemisphere. Thresholding was standardized across all images to ensure consistency in signal detection.

β-gal staining

β-gal staining was performed on brain sections using the Senescence β-Galactosidase Staining Kit (no. 9860, Cell Signaling Technology, Danvers, MA). Brain tissues were fixed in 4% paraformaldehyde for 24 h at 4°C, cryoprotected in 30% sucrose in PBS, and sectioned into 12 μm coronal slices. Sections were mounted onto glass slides, washed with 1× PBS, and fixed in 1× fixative solution (diluted from 10× stock) for 10–15 min at RT. Fixed sections were rinsed twice in PBS to remove residual fixative.

The β-gal staining solution was freshly prepared by combining 930 μL of 1× staining solution, 10 μL of 100× solution A, 10 μL of 100× solution B, and 50 μL of 20 mg/mL X-gal stock solution (prepared in DMSO). Staining solution was applied to the sections, and slides were incubated at 37°C in a CO₂-free environment overnight.

After incubation, sections were examined under a light microscope for blue staining, indicative of β-gal. For long-term storage, staining solution was removed, and sections were overlaid with 70% glycerol and stored at 4°C.

Statistical analysis

All analyses were conducted using GraphPad Prism v.9.0. Statistical significance was defined as p < 0.05. Data are presented as mean ± SD.

One-way ANOVA with Bonferroni’s multiple comparisons test was used to evaluate Ki67-positive cell counts. MRI-derived measurements of ventricular volume and T2-hyperintense lesion area were analyzed using one-way ANOVA followed by Dunnett’s post hoc test. Myelin (MBP) and microglial (CD68) quantifications across treatment groups and time points (P21, P42, P63) were assessed using two-way ANOVA with Sidak’s multiple comparisons test. RT-qPCR measurements of Cre and hamartin mRNA expression across treatment groups and brain regions (cortex, hippocampus, cerebellum) were evaluated using two-tailed unpaired Mann-Whitney U test and unpaired t test with Welch’s correction. Survival curves were analyzed using the log rank (Mantel-Cox) test.

Log₁₀ transformation was applied to MRI measurements, % MBP-positive area, and CD68 integrated density values to stabilize variance and meet assumptions of normality. Sample sizes (n) are indicated in the corresponding figure legends.

Data availability

The data used to support the findings of this study are included within the article and the supplemental information.

Acknowledgments

We thank Ms. Suzanne McDavitt for skilled editorial assistance, and BridgeBio, Inc for vector manufacturing. We thank Dr. Kristi Viles, Dr. David Scott, and Dr. Clayton Beard from BridgeBio for their valuable insights⁶⁴ and Dr. Kevin Costa Leandro for the generation of Cre primers. Graphical abstract and Figure 1A were created with BioRender.com. Brain images and anatomical references were adapted from the Allen Mouse Brain Atlas, courtesy of the Allen Institute for Brain Science. K.B. is funded by the National Institutes of Health (NIH) K22 CA2802019-01 and Department of Defense (DOD) award HT9425-24-1-0119. This work was supported by a Sponsored Research Agreement (SRA) from BridgeBio Pharma Inc. and Department of Defense (DOD) award HT9425-23-1-0338.

Author contributions

X.O.B. and E.A.H. conceived and designed the experiments. J.W.C. and G.R.W. performed MRI and quantified ventricular volume and T2-hyperintense lesion size. M.M. and R.S. provided guidance on microscopic quantification and analysis. K.B. advised on data analysis and statistics. T.T., E.R., J.S.H., and M.G. assisted with experiments. A.S.-R. and A.L.G. provided expertise on brain neuropathology and cytoarchitecture. E.A.H. and S.P. performed the experiments. E.A.H. analyzed the data and prepared the figures. E.A.H. and X.O.B. wrote the manuscript. All authors edited and commented on the paper.

Declaration of interests

The authors declare no competing interests.