Evaluation of safety and early efficacy of AAV gene therapy in mouse models of vanishing white matter disease

Abstract

Leukoencephalopathy with vanishing white matter (VWM) is a progressive incurable white matter disease that most commonly occurs in childhood and presents with ataxia, spasticity, neurological degeneration, seizures, and premature death. A distinctive feature is episodes of rapid neurological deterioration provoked by stressors such as infection, seizures, or trauma. VWM is caused by autosomal recessive mutations in one of five genes that encode the eukaryotic initiation factor 2B complex, which is necessary for protein translation and regulation of the integrated stress response. The majority of mutations are in EIF2B5. Astrocytic dysfunction is central to pathophysiology, thereby constituting a potential therapeutic target. Herein we characterize two VWM murine models and investigate astrocyte-targeted adeno-associated virus serotype 9 (AAV9)-mediated EIF2B5 gene supplementation therapy as a therapeutic option for VWM. Our results demonstrate significant rescue in body weight, motor function, gait normalization, life extension, and finally, evidence that gene supplementation attenuates demyelination. Last, the greatest rescue results from a vector using a modified glial fibrillary acidic protein (GFAP) promoter-AAV9-gfaABC(1)D-EIF2B5-thereby supporting that astrocytic targeting is critical for disease correction. In conclusion, we demonstrate safety and early efficacy through treatment with a translatable astrocyte-targeted gene supplementation therapy for a disease that has no cure.

Keywords: AAV9; EIF2B5; MRI; VWM; astrocytes; gene supplementation; gene therapy; integrated stress response; leukodystrophy; myelin; vanishing white matter disease.

Graphical abstract

Figure 1

Comparison of eGFP expression in mouse tissue after AAV9-mediated gene delivery (A) Brain biodistribution study schematic depicting three AAV9 vectors encoding eGFP under the astrocyte-targeting promoters (GFAP, gfaABC(1)D) and ubiquitous CAG promoter. Each construct was packaged into AAV9 and delivered individually via a single ICV injection into the left ventricle of WT postnatal day 0–1 (P0-P1) mouse pups. Injected mice were then euthanized at 4 weeks (n = 10) and 8 weeks (n = 7–10) to collect tissue for analysis of eGFP distribution. (B) Visualization of eGFP expression patterns (green) and distribution throughout the entire brain 4 weeks following injection with one of the vectors. 1x magnification; combination eGFP and brightfield filters (n = 2 brains per construct) using a fluorescent microscope. (C–E) Fluorescent images displaying eGFP biodistribution in sagittal mouse brain sections 4 weeks after ICV injection of AAV9 vectors using either the GFAP (C), gfaABC(1)D (D), or CAG (E) promoters, scale bar, 1 mm. (C1–E1) 40X images of eGFP-positive cells from each construct, scale bar, 50 μm. (F) eGFP biodistribution quantified as a percentage (%) of eGFP-positive brain area relative to total brain area (DAPI signal) 4 and 8 weeks post-AAV9 infusion (n = 8–10) and controls (n = 6). (G) Extended depth of focus (EDF) confocal image of co-expression patterns (white) between GFAP (purple) and eGFP (green) in a GFAP+ brain area within the hippocampus, 4 weeks after AAV9-GFAP-eGFP ICV injection. Neurons labeled by NeuN (red) and all cells by DAPI, scale bar, 25 μm. (H) Quantification of colocalization of S100β and eGFP 4 weeks (left) and 8 weeks (right) as a percentage relative to an uninjected WT control. Each data point represents an average of three repetitions, with 19 fields of view collected from seven brain regions in each repetition. (I) Schematic showing the regions of the brain that correspond to brain areas 1, 2, and 3. (J) eGFP expression was quantified via RT-qPCR in multiple tissue types, from 8-week-old mice injected with each of the three eGFP vectors at P0-P1. Expression is calculated relative to the housekeeping gene β-Actin. Tissues examined include three different regions of the brain (BA1/2/3 = Brain Area 1/2/3), the cervical spinal cord (CSC), thoracic spinal cord (TSC), and lumbar spinal cord (LSC), along with two peripheral tissues, heart, and liver. Significance was calculated using ordinary one-way ANOVA (F and H) and two-way ANOVA (J) for all graphs in this figure. Significance indicators are ∗∗∗∗p ≤ 0.0001, ∗∗∗p ≤ 0.001, ∗∗p ≤ 0.01, ∗p ≤ 0.05.

Figure 2

R191H model characterization through behavior, myelin quantification, neuroinflammation, and ISR activation (A) Body weights from 4 weeks of age from male (left) and female (right), R191H (red), and WT (black) mice are displayed monthly (every 4 weeks), showing significant differences in body weight. (B) Rotations per minute (RPMs) at which 8-month-old R191H mice (red) fall from the accelerating rotarod indicates significantly poorer motor performance as compared with age-matched, WT mice (white). (C and D) Luxol Fast Blue myelin staining and quantification of the brain (C) and cervical spinal cord (D) of WT (left) mice and 8-month-old R191H (right). In both regions, R191H mice have significantly less Luxol-positive area relative to WT mice, indicating loss of myelin. (E) Iba1 immunofluorescence staining in the cervical spinal cord indicates significant microglia activation in WT mice (left) as opposed to 8-month-old R191H mice (right). (F) RT-qPCR results show elevation of multiple ISR transcripts relative to Gapdh in BA3 and cervical spinal cord of R191H compared with WT 8-month-old mice. Data are mean ± standard deviation. Significance was calculated using Mann-Whitney t tests followed by post hoc Holm-Šídák tests (A and B), or unpaired t tests (C–F). Significance indicators are ∗∗∗∗p ≤ 0.0001, ∗∗∗p ≤ 0.001, ∗∗p ≤ 0.01, ∗p ≤ 0.05.

Figure 3

In-life characterization of the Eif2b5^I98M toy murine model using survival, body weight, rotarod, and gait mapping (A) I98M (males and females) survival graphed over time (in days) demonstrates a significantly reduced median probability of survival of 154.5 days. (B) Body weights starting from 4 weeks of age from male (solid line) and female (dashed line), I98M (red), and WT (black) mice are displayed monthly (every 4 weeks), showing significant differences in weight. (C) Latency (in seconds) at which 4.5-month-old I98M mice (red) fall from the accelerating rotarod indicates significantly poorer motor performance as compared with age-matched, WT mice (white). (D–F) Quantification of gait was obtained from the Noldus CatWalk XT system using 4.5-month-old mice (6–13 mice, 1–4 runs per animal). Five parameters were quantified: (D) cadence (steps per second), (E) forelimb and hindlimb print area (cm²), and (F) forelimb and hindlimb swing (time of no contact in seconds). I98M mice showed significant gait alterations in all parameters quantified except cadence. (G) Foot maps captured from the CatWalk XT of WT (left) and affected I98M (right) 4.5-month-old mice display significant alterations in gait. Left front (LF—yellow), right front (RF—turquoise), left hind (LH—green), and right hind (RH—magenta) all indicate which limb corresponds to which print. Significance was calculated using the Log rank (Mantel-Cox) test (A), multiple paired t tests (B), and Mann-Whitney t tests (C–F) for all graphs in this figure. Significance indicators for graphs are ∗∗∗∗p ≤ 0.0001, ∗∗∗p ≤ 0.001, ∗∗p ≤ 0.01, ∗p ≤ 0.05, ns = not significant.

Figure 4

Histological and molecular characterization of pathology in the I98M model (A–D) Comparison of sagittal brain sections from 5-month-old WT control and I98M mice, scale bar, 1 mm. (A and B) Visualization of demyelination (FluoroMyelin stain, green) in a I98M mouse brain (neurons visualized with NeuroTrace, red) overlayed (A) or as individual FluoroMyelin signal (B). (E) Higher magnification image of the brain highlighting the structural components of the cerebral cortex and corpus callosum in both I98M and control (FluoroMyelin/NeuroTrace signals), scale bar, 250 μm. (C) IF image of MBP (green) revealing increased protein levels observed in I98M mice. (D) IF image illustrating elevated GFAP immunoreactivity (purple) indicating astrogliosis in I98M mice. (F) Higher magnification of the brain encompassing astrocytic reactivity (GFAP, purple), and MBP (green) in the cerebral cortex, scale bar, 250 μm. (G) Comparative quantification of myelin, MBP, and GFAP protein expression by IF in I98M vs. WT mice, n = 5. The graph depicts percentage of protein expression relative to control, with the x axis representing various protein markers, and the y axis denotes percent change in expression compared with WT. (H) RT-qPCR results display fold change in gene expression relative to Gapdh of multiple ISR genes in BA3 and cervical spinal cord of 5-month-old I98M mice (n = 5), demonstrating significant upregulation of the ISR in the I98M model. One-way ANOVA (G) and two-way ANOVA (H) were used to determine significance. Indicators for all graphs in this figure are ∗∗∗∗p ≤ 0.0001, ∗∗∗p ≤ 0.001, ∗∗p ≤ 0.01, ∗p ≤ 0.05.

Figure 5

AAV9-GFAP-EIF2B5, AAV9-gfaABC(1)D-EIF2B5, and AAV9-CAG-EIF2B5 gene therapy safety and efficacy in R191H mice at 7 months of age (A) Body weights starting at 4 weeks of age from male (left) and female (right), sham-injected R191H (red), WT (black), and treated AAV9-GFAP-EIF2B5 (pink), AAV9-CAG-EIF2B5 (green), and AAV9-gfaABC(1)D-EIF2B5 (blue) R191H mice are displayed as monthly (every 4 weeks) measurements. AAV9-gfaABC(1)D-EIF2B5 and AAV9-CAG-EIF2B5 showed significant rescue in body weight (p ≤ 0.05–0.001). (B) RPMs at which 7-month-old, gene therapy-treated R191H mice fall from the accelerating rotarod indicates significant rescue in motor performance in the AAV9-gfaABC(1)D-EIF2B5 (blue; p ≤ 0.01) cohort when compared with sham-injected R191H mice (red). AAV9-CAG-EIF2B5 (green) and AAV9-GFAP-EIF2B5 (pink) cohorts show no improvement on rotarod. (C) Representative T2-weighted magnetic resonance imaging (MRI) in WT (left), R191H sham-injected (middle), and AAV9-gfaABC(1)D-EIF2B5 treated R191H (right) mice with the corpus callosum outlined in red. (D) RT-qPCR results display EIF2B5 transgene expression relative to the Gapdh housekeeping gene using the ΔCT method times 1000. Significance was calculated using mix-effects model (A) or one-way ANOVA (B and D) followed by Dunnett’s post hoc test (A and B). Significant indicators are ∗∗∗∗p ≤ 0.0001, ∗∗∗ or %%% p ≤ 0.001, ∗∗ or ## or %% p ≤ 0.01, ∗ or % or #p ≤ 0.05.

Figure 6

AAV9-GFAP-EIF2B5, AAV9-gfaABC(1)D-EIF2B5, and AAV9-CAG-EIF2B5 gene therapy safety and efficacy in I98M mice (A) Survival graphed over time (in days) of WT (black dashed; p ≤ 0.0001), untreated I98M (red; ns), sham-injected (gray), AAV9-GFAP-EIF2B5 (pink; p ≤ 0.05), AAV9-gfaABC(1)D-EIF2B5 (blue; p ≤ 0.0001), and AAV9-CAG-EIF2B5 (green; p ≤ 0.001) demonstrates extension of life expectancy in treatment cohorts as compared with sham treated I98M mice. (B) Body weights starting at 4 weeks of age from both male (left) and female (right), untreated WT (black), I98M (red), sham-injected (gray), and treated AAV9-GFAP-EIF2B5 (pink), AAV9-gfaABC(1)D-EIF2B5 (blue), and AAV9-CAG-EIF2B5 (green), I98M mice are displayed as monthly (every 4 weeks) measurements. AAV9-gfaABC(1)D-EIF2B5 showed significant rescue in body weight (p ≤ 0.01–0.0001) of both sexes of mice, while AAV9-CAG-EIF2B5 showed significant rescue in only female body weight (p ≤ 0.05–0.001) as compared with I98M sham treated mice. (C) Latency (seconds) at which 5-month-old, gene therapy-treated I98M mice fall from the accelerating rotarod shows rescue in motor performance in the AAV9-gfaABC(1)D-EIF2B5 (blue) and AAV9-CAG-EIF2B5 (green) cohort compared with sham I98M mice (red). (D–F) Quantification of gait was obtained from the Noldus CatWalk XT system using WT (white), untreated I98M (red), sham-injected (gray), and treated AAV9-GFAP-EIF2B5 (pink), AAV9-gfaABC(1)D-EIF2B5 (blue), and AAV9-CAG-EIF2B5 (green) I98M mice at 4.5 months old (6–13 mice, one to four runs per animal). Five parameters were quantified: (D) Cadence (steps per second), (E)forelimb and hindlimb print area (cm²) and (F) forelimb and hindlimb swing (time of no contact [s]). (G–L) Representative T2-weighted MRI in WT (G), I98M untreated (H), sham-injected (I), AAV9-GFAP-EIF2B5 (J), AAV9-gfaABC(1)D-EIF2B5 (K), and AAV9-CAG-EIF2B5 (L) treated I98M mice with the corpus callosum outlined in red. Significance was calculated using Log rank (Mantel-Cox) test (A), multiple paired t tests (B), one-way ANOVA (C), and Mann-Whitney t tests (D–F) for all graphs in this figure. Indicators are ∗∗∗∗ or %%%% p ≤ 0.0001, ∗∗∗ or ### p ≤ 0.001, ∗∗ or %% or ## p ≤ 0.01, and ∗ or # p ≤ 0.05.

Figure 7

Endogenous ISR activation that rescues homeostasis as compared with the unregulated ISR activation in R191H and I98M models (A) Diagram depicting the molecular pathway that is activated during the ISR. Genes that are commonly upregulated during ISR activation are displayed as upward-facing green arrows. ISR activation leads to the phosphorylation of eIF2α thereby inhibiting eIF2B, as well as triggering an upregulation cascade of the following: PERK, GCN2, eIF4EBP1, SLC7A5, ATF5, TRIB3, and CHOP. GADD34 is also upregulated; however, this triggers a negative feedback loop where GADD34 will dephosphorylate eIF2α, thereby rescuing homeostasis. (B) Diagram displaying the deregulated ISR as it relates to eIF2B5 mutations, in both R191H and I98M models. ISR genes found to be upregulated by RT-qPCR for BA3 are displayed as dark blue (R191H), and dark yellow (I98M) upward-facing arrows. Cervical spinal cord is also represented in light blue (R191H) and yellow (I98M). Significance is depicted by the size of the arrows as well as asterisks. Significance indicators: ∗∗∗∗p ≤ 0.0001, ∗∗∗p ≤ 0.001, ∗∗p ≤ 0.01, ∗p ≤ 0.05.