Intra-CNS AAV9-GBA1 delivery yields species and route of administration differences in safety and transgene expression Authors: Ana Claudia Amaral, Branka Grubor, Davide Gianni, Lee Koetzner, Neeta Abraham, Shawn Bourque, Danielle Brown, Yi Chen, Kay-Ellen Chicoine, Pete Clarner, PJ De Giovanni, Stefan Hamann, Melissa Kirkland, Odete R Mendes, Merilla Michael, Murali VP Nadella, Kalyani Nambiar, Jennifer Sebalusky, Weike Zeng, Shanqin Xu, Patrick Trapa, Edward D Plowey, Eric Tien, Jim Fikes, Dominic M Walsh, Warren D Hirst, Junghae Suh, Kelly E Glajch Published: 2026 ABSTRACT AbstractAdeno-associated virus (AAV) delivery of lysosomal enzymes to the brain is currently being tested in several clinical trials. However, publicly available non-human primate (NHP) studies comprehensively reporting biodistribution, transgene expression, and safety across the central nervous system (CNS) remain limited. Here, we examined AAV9-GBA1 (encoding glucocerebrosidase [GCase]) delivery to mice and NHPs. Loss-of-function GBA1 mutations cause Gaucher’s disease and are risk factors for Parkinson’s disease and dementia with Lewy bodies. In mice, early postnatal intracerebroventricular administration produced appreciable transgene expression and increased GCase activity in the brain without adverse findings. Two independent adult NHP studies evaluated AAV9-GBA1 brain delivery via intracisterna magna (ICM) and intraparenchymal (IPa) administration. ICM delivery produced considerable transgene expression in the spinal cord and dorsal root ganglia (DRG) neurons along with adverse microscopic findings, but limited brain expression and activity. In contrast, IPa delivery into the thalamus and/or putamen produced substantial transgene expression and GCase activity in injected and connected brain regions, with limited effects in distal regions, including the spinal cord. While no AAV-related DRG toxicity was detected with IPa, procedure- and AAV9-GBA1-related adverse microscopic brain findings were observed and associated with clinical outcomes. Further technological advances are needed to achieve safe and therapeutic AAV transgene expression in NHP brain.Keywords: GBA1, AAV, intraparenchymal, intracisterna magna, dorsal root ganglia toxicity, NHP Introduction The lysosome is a vital organelle involved in numerous cellular processes, including autophagy, endocytosis, and lipid homeostasis. Mutations in lysosomal proteins are associated with a wide range of diseases, including many neurological disorders. Specifically, mutations in lysosomal hydrolases and membrane transporters cause lysosomal storage disorders (LSDs), a group of approximately 70 rare, monogenic diseases typically appearing early in life and accompanied by neurological and non-neurological deficits.1 While some LSDs have approved enzyme replacement therapies addressing peripheral symptoms, these therapies fail to alleviate central nervous system (CNS) symptoms as the enzymes cannot cross the blood-brain barrier.2 In addition, mutations in lysosomal proteins increase the risk of developing several chronic neurodegenerative diseases, such as Parkinson’s disease (PD), frontotemporal dementia, and amyotrophic lateral sclerosis,3 all of which lack sufficient disease-modifying therapies. To address the challenge of CNS delivery in these lysosomal disorders, extensive research has been focused on gene therapy, predominantly utilizing adeno-associated viruses (AAVs), to express functional lysosomal genes directly in the brain.4 Rodent studies have demonstrated that AAV-mediated delivery of lysosomal enzymes either through direct brain injection (intraparenchymal [IPa]) or through intra-cerebrospinal fluid (intra-CSF including intracerebroventricular [ICV]) administration is efficacious in correcting CNS pathway deficits and neurological symptoms in LSD rodent models (summarized in Massaro et al.5). In addition, these approaches have also been tested in indications with naturally occurring non-rodent animal models (cats and dogs; summarized in Massaro et al.5). Moreover, both intra-CSF via lumbar intrathecal (IT) or intracisterna magna (ICM) and IPa delivery of AAVs have been tested in the clinic for LSDs (reviewed in Kido et al.4) with many trials still ongoing (NCT04408625, NCT04127578, NCT04411654, NCT04713475, NCT04669535, NCT04771416, NCT04771416, NCT03580083, and NCT03566043). As non-human primates (NHPs) are a critical species for clinical translation to humans, publicly available data on AAV-mediated lysosomal protein expression in the CNS via ICM or IPa delivery5 would benefit from additional studies in these models. In recent years, AAV9 has emerged as a common capsid for clinical gene therapy trials, with multiple ongoing studies utilizing AAV9 intra-CSF delivery of lysosomal enzymes for mucopolysaccharidosis type I (NCT03560083), mucopolysaccharidosis type II (NCT03566042), PD (NCT04127578), and frontotemporal dementia (NCT04408625). However, a limited number of studies have comprehensively evaluated the biodistribution, transgene expression, and safety of AAV9 ICM administration in NHPs.6,7,8,9,10 Although AAV9 NHP IPa delivery has been reported previously,11 these studies do not provide a comprehensive assessment of brain biodistribution, transgene expression, and safety outcomes following AAV9 administration. GBA1 encodes for the lysosomal hydrolase glucocerebrosidase (GCase). Homozygous loss-of-function mutations in GBA1 cause Gaucher’s disease (GD). Heterozygous and homozygous mutations are major genetic risk factors for PD, with ∼10% of PD patients harboring GBA1 mutations.12 Furthermore, heterozygous GBA1 mutations are a major risk factor for dementia with Lewy bodies (DLB).13 Given the lack of disease-modifying therapies to address the neurological symptoms of patients with GD, PD, or DLB, targeting GBA1 is an important area of focus for drug discovery. Indeed, there is an ongoing trial for GBA1-PD utilizing AAV9 delivery of GBA1 via ICM administration (NCT04127578). In this report, we used AAV9 to deliver GBA1 to the brain of mice (C57BL/6N-Gba1tm1.1Mjff/J [here referred to as D409V+/+] and C57BL/6NTac [referred to as wild type (WT)) and cynomolgus NHPs. In mice, early postnatal ICV administration of AAV9-GBA1 produced broad vector biodistribution and a dose-dependent increase in functional protein in several brain regions without significant histopathologic changes in the brain. Next, we performed two separate studies to administer AAV9-GBA1 into the adult NHP brain via ICM or IPa delivery. ICM delivery led to low viral genome biodistribution and negligible transgene expression in the brain and produced microscopic findings in the dorsal root ganglia (DRG), spinal cord, and related regions. IPa delivery into the NHP thalamus and/or putamen produced robust viral genome biodistribution and transgene expression throughout many brain regions, but this was accompanied by significant clinical signs and adverse microscopic findings in several brain regions. In summary, in contrast to the mouse data, AAV9-GBA1 delivery in NHPs (via ICM or IPa) did not achieve the desired biodistribution and transgene expression levels in the brain with acceptable safety profiles. This suggests that further technological advances are needed to obtain robust transduction and expression of GBA1 with AAV delivery to the NHP brain without significant safety findings, including exploration of engineered tissue-tropic capsids and cargo solution to control transgene expression. Results AAV9-GBA1 ICV delivery into mice restores WT GCase activity levels without significant safety findings To deliver AAV9-GBA1 to the mouse brain, we utilized ICV administration into early postnatal mice (postnatal day 0 [P0]) harboring homozygous D409V GBA1 mutations (D409V+/+). D409V+/+ mice have a significant reduction in GCase enzymatic activity compared with WT control mice14 and thus provide a good pharmacodynamic model for diseases with GCase deficiency. Mice were injected bilaterally with a total of 1 × 109, 7 × 109, 1 × 1010, or 4 × 1010 genomic copies (GCs) of AAV9-GBA1 (expressing human GBA1) or 4 × 1010 GC of AAV9-YFP (as a control) in a total volume of 4 mL and were euthanized at 2 months post-injection. The amount of vector present in the brain (vector genome biodistribution) was quantified using quantitative PCR (qPCR) with a probe targeting the WPRE element of the AAV genome and normalized to the amount of total DNA. AAV9-GBA1 treatment produced a dose-dependent increase in vector biodistribution in the cortex and striatum (Figure 1A) with AAV9-YFP control-injected animals showing comparable vector genome levels to human GBA1 (hGBA1) at the same dose level (data not shown). To assess the expression of hGBA1 mRNA, in situ hybridization (ISH) RNAscope was performed using a custom probe designed to selectively measure hGBA1. Widespread hGBA1 RNA expression was observed throughout the brain, especially in the cortex and striatum, in a dose-dependent manner (Figure 1B). Moreover, there was a dose-dependent increase in hGBA1 bulk tissue RNA levels in the cortex, as assessed by qPCR using a custom probe selective for hGBA1, (Figure S1) which was in line with the ISH RNAscope data. AAV9-GBA1 ICV delivery into mice produces robust expression of active hGCase in the brain (A) Vector genome biodistribution, as measured by genomic copies (GC)/μg DNA, was quantified in the cortex (left) and striatum (right) from D409V+/+ mice injected with 1 × 109 (pink, n = 9), 7 × 109 (green, n = 13), 1 × 1010 (purple, n = 10), and 4 × 1010 (light blue, n = 23) total GC of AAV9-GBA1. (B) Representative images of in situ hybridization staining for hGBA1 mRNA from coronal mouse brain sections. From left to right, images of WT control mice (injected with AAV9-YFP at 4 × 1010 GC) and D409V+/+ mice injected with 1 × 109, 7 × 109, 1 × 1010, and 4 × 1010 total GC doses. (C) GCase activity (nmol/mg/h) in the cortex (left) and striatum (right) from WT mice injected with AAV9-YFP 4 × 1010 GC dose (WT control, gray; n = 21) and D409V+/+ mice injected with a range of doses: AAV9-YFP 4 × 1010 GC dose (“Control,” black; n = 22) and AAV9-GBA1 1 × 109 (pink, n = 18), 7 × 109 (green, n = 23), 1 × 1010 (purple, n = 20), and 4 × 1010 (light blue, n = 33 cortex and n = 31 striatum) total GC. Scale bar represents 1 mm. Mixed-effects model with Dunnett’s multiple comparisons test. Data are presented as medians and min to max. ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. To assess the level of functional GCase protein, we utilized the 4-methylumbelliferyl-β-D-glucopyranoside (4MUG) enzyme activity assay that measures the hydrolysis of a synthetic substrate of GCase (4MUG) into its product (4-methylumbelliferyl), which is detected by fluorescence emission. This GCase activity assay is used routinely in the field as a surrogate for hydrolysis of endogenous GCase substrates.15 In D409V+/+ mice, a mixed-effects model demonstrated an overall effect of brain region (F(1, 129) = 52.35, p < 0.0001), dose (F(5, 131) = 68.44, p < 0.0001), and brain region × dose (F(5, 129) = 24.84, <0.0001) on GCase activity. Specifically, GCase activity was significantly increased in the cortex in the 7 × 109 GC (p = 0.0084), 1 × 1010 GC (p = 0.0004), and 4 × 1010 GC (p < 0.0001) AAV9-GBA1 dose groups compared with AAV9-YFP-injected (control) mice. In the striatum, GCase activity was significantly increased only in the 4 × 1010 GC AAV9-GBA1 dose group (p < 0.0001) with a trend toward an increase in the 1 × 1010 GC AAV9-GBA1 dose group (p = 0.0665) when compared with control-injected mice. All dose levels, except the 1 × 109 GC group, were numerically higher than control-injected D409V+/+ mice (Figure 1C). In contrast, GCase activity in the 1 × 109 GC dose group was not significantly different than AAV9-YFP-injected mice in the cortex (p > 0.9999) or striatum (p > 0.9999) which was in line with the more modest vector biodistribution (Figure 1A), ISH mRNA (Figure 1B), and bulk tissue mRNA expression data (Figure S1) from this group. Altogether, these data demonstrated that the 7 × 109 GC dose was sufficient to restore GCase activity to or just above WT control levels (Figure 1C), our target level, and thus was deemed to be our approximate minimal efficacious dose. In parallel to the D409V+/+ mouse studies, AAV9-GBA1 was studied in age-matched WT mice at the same dose levels (1 × 109, 7 × 109, 1 × 1010, or 4 × 1010 total GC). In WT mice, a mixed-effects model demonstrated an overall effect of brain region (F(1, 73) = 40.03, p < 0.0001), dose (F(4, 74) = 31.24, p < 0.0001), and brain region × dose (F(4,73) = 10.67, <0.0001) on GCase activity. Specifically, there was a dose-dependent increase in GCase activity in both the cortex and striatum with a significant increase at 7 × 109 GC (p = 0.0001), 1 × 1010 GC (p = 0.0075), and 4 × 1010 GC (p < 0.0001) in the cortex and at 4 × 1010 GC (p < 0.0001) in the striatum compared with control-injected WT mice (Figure S2). Overall, in WT mice with AAV9-GBA1 treatment, GCase activity was increased ∼2- to 3-fold in 7 × 109 GC group (the minimal efficacious dose from the D409V+/+ study) and up to 4-fold and 10-fold in the 1 × 1010 GC and 4 × 1010 GC groups, respectively, compared with WT control (AAV9-YFP-treated) mice (Figure S2). In addition to mRNA expression and functional protein levels, we assessed the safety outcomes following AAV9-GBA1 injection by monitoring in-life clinical signs and brain histopathology. In D409V+/+ and WT AAV9-GBA1-treated animals, no clinical signs were observed and no evidence of neuronal pathology, as assessed by H&E (data not shown), was detected. Inflammation was monitored using microglial process perimeter quantification of Iba1/Pu.1-positive cells (markers for microglia). While a two-way repeated measures (RM) ANOVA demonstrated an overall effect of brain region (F(1,44) = 44.66, p < 0.0001), animal (F(44, 44) = 4.76, p < 0.0001), and brain region × treatment (F(4,44) = 4.57, p = 0.0036), there was no overall effect of treatment (F(4,44) = 0.521, p = 0.7204) or any change in microglial process perimeter between control (AAV9-YFP treated) mice and any AAV9-GBA1 dose group (p > 0.05) (Figures S3A and S3B) in the cortex or striatum of D409V+/+ mice. The only histological finding was a higher incidence in neuronal vacuolation in the cortex in the 4 × 1010 GC AAV9-GBA1-treated group (p < 0.0001) (Figures S3C and S3D). This was not considered adverse and deemed to be due to high levels of overexpression of GBA1 illustrated by the very high vector biodistribution (Figure 1A) and mRNA expression (Figures 1B and S1) in this dose group. Similar histopathology data (microglial quantification and neuronal vacuolation) were obtained from WT AAV9-GBA1 ICV-injected mice (Figure S3), except for an increase in microglial process perimeter only in the cortex of the 7 × 109 GC AAV9-GBA1 group (p = 0.0022). Together these data demonstrate that ICV delivery of AAV9-GBA1 at and above doses sufficient to restore GCase activity back to WT levels (up through 1 × 1010 GC) in the brain of D409V+/+ mice was not associated with any widespread or adverse safety signals. AAV9-GBA1 IPa delivery into NHPs yields increased vector biodistribution and mRNA expression in the brain compared with ICM delivery We next completed two different studies assessing the vector genome biodistribution, transgene expression (mRNA and functional protein), and safety of AAV9-GBA1 delivery into cynomolgus monkeys via ICM or IPa delivery (Figure S4). Doses for the ICM study were selected based on dose scaling from the minimal efficacious dose in the rodent study (Figure 1), and IPa doses were selected from published literature.8,11 Our goal was to achieve widespread expression of GBA1 throughout the brain regions involved in PD including the striatum, substantia nigra, and cortex. For the ICM study, a single, 1-mL injection of AAV9-GBA1 was delivered into the cisterna magna CSF at three different concentrations: 1× 1013 GC/mL (ICM-low), 3 × 1013 GC/mL (ICM-medium), and 1 × 1014 GC/mL (ICM-high). For the IPa study, 0.15 or 0.125 mL of AAV9-GBA1 was delivered bilaterally into the putamen (P) and/or thalamus (T), respectively, at two different concentrations (low: 2.5 × 1012 GC/mL and high: 7.5 × 1012 GC/mL) across three different experimental groups: IPa-Low P + T, IPa-High P + T, and IPa-High P only (Figure S4). The putamen and thalamus were chosen as these regions are highly connected to critical subcortical and cortical regions implicated in PD.11,16 Each study also included a reference group that received an equal volume of vehicle (artificial cerebrospinal fluid, aCSF) delivered ICM or IPa (P + T). All NHPs were euthanized at 6 weeks post-injection, except for the IPA-High P + T group that was euthanized early due to clinical signs (Table S1). In the ICM study, low levels of vector genomes (∼101–103 GC/μg DNA) were detected throughout the brain, including cortical, basal ganglia, and midbrain regions, whereas ∼10-fold higher vector genomes (∼102–104 GC/μg DNA) were detected in the spinal cord and the peripheral nervous system ([PNS]; DRG and trigeminal ganglia [TG]; Figures 2A and S5A). In contrast, IPa administration produced robust vector distribution in multiple brain regions (∼104–107 GC/μg DNA), comparable to the levels detected in the mouse study (Figure 1A), with highest levels of vector in the injected regions (putamen and thalamus) followed by midbrain and cortical regions. However, limited amounts of vector genomes (up to ∼102 GC/μg DNA) were detected in the spinal cord, DRG, and TG (Figures 2A and S5B). In agreement with the vector biodistribution data, bulk tissue hGBA1 RNA expression, as assessed by qPCR, in the ICM study was low throughout the brain (∼102–104 ddCT hGBA1/RPP30) but up to 1,000-fold higher in the spinal cord and DRG tissue. Specifically, robust RNA expression (∼105–106 ddCT hGBA1/RPP30) was observed at multiple levels of the spinal cord (cervical, thoracic, and lumbar) in all dose groups. In addition, hGBA1 RNA expression exhibited a dose-dependent increase in multiple DRG levels (cervical, thoracic, and lumbar) up to 106 ddCT hGBA1/RPP30 (Figures 2B and S5A). In contrast, in the IPa study, robust hGBA1 RNA expression was detected in multiple brain regions (∼104–106 ddCT hGBA1/RPP30) with the highest levels in the injected regions followed by midbrain and cortical regions. Moreover, hGBA1 RNA expression was ∼100- to 10,000-fold lower in the TG and DRG (∼101–102 ddCT hGBA1/RPP30) than in the brain. In the spinal cord, hGBA1 RNA expression was higher (∼102–104 ddCT hGBA1/RPP30) compared with that in DRG and TG but still lower than that in most brain regions in the IPa study (Figures 2B and S5B). Similar to previous studies,17 significant variability was observed within groups for both vector biodistribution and RNA expression (Figure S5). Together these data demonstrate that IPa delivery yielded robust transduction in the brain with limited transduction in the spinal cord and PNS. In contrast, ICM delivery yielded low brain transduction and high spinal cord and PNS transduction. AAV9-GBA1 IPa delivery into NHP yields increased vector biodistribution and mRNA expression in the brain compared with ICM delivery Heatmaps illustrating (A) vector biodistribution (GC/μg DNA) and (B) bulk tissue RNA expression (ddCT hGBA1/RPP30) data from the ICM study (left) and IPa study (right). Data are plotted as the mean values for each tissue sample and group. Scale bar indicates the range of the mean values for each dataset (white boxes with an “X” indicate samples not available), and dotted lines are used to denote those tissues in the brain (“Brain”; above the line) vs. the spinal cord and peripheral nervous system (“Spinal cord/PNS”; below the blue line). AMY, amygdala; Ca, caudate; CaPu, caudate/putamen; Pu, putamen; FCx, frontal cortex; PFC, prefrontal cortex; Ocx, occipital cortex; PCx, parietal cortex, TCx, temporal cortex; THM, thalamus; SC, C4, spinal cord, cervical level 4 level; SC, T9, spinal cord, thoracic level 9; SC, L2, spinal cord, lumbar level 2; TG, trigeminal nerve; DRG, C2, dorsal root ganglia, cervical level 4; DRG, C3, dorsal root ganglia, cervical level 3; DRG, C4, dorsal root ganglia, cervical level 4; DRG, C6, dorsal root ganglia, cervical level 6; DRG, C7, dorsal root ganglia, cervical level 7; DRG, T9, dorsal root ganglia, thoracic level 9; DRG, L1, dorsal root ganglia, lumbar level 1; DRG, L2, dorsal root ganglia, lumbar level 2; DRG, L3, dorsal root ganglia, lumbar level 3. Injected regions in the IPa study are denoted with an “∗.” IPa, but not ICM, delivery into NHPs produced enhanced transduction and functional hGCase expression in the brain We further explored the expression and function of hGBA1 in multiple brain regions. In the ICM study, ISH RNAscope results demonstrated sparse transduction of cells in the brain in all dose groups as measured by hGBA1 RNAscope (Figure 3A), similar to the low abundance of hGBA1 RNA throughout the brain (Figure 2B). Occasional clusters of hGBA1 RNA ISH signal were observed in the brain particularly in the medium- and high-dose groups (Figure 3A) consistent with previously reported perivascular astrocyte transduction.18 In contrast, IPa administration produced hGBA1 RNA ISH signal throughout multiple injected and non-injected regions (Figure 3B) similar to the bulk tissue hGBA1 RNA expression data (Figure 2B) but without the same clustered transduction pattern as seen in the ICM study. IPa, but not ICM, delivery yields widespread transduction and target GCase activity in NHP brain Representative in situ hybridization images from sections stained for hGBA1 mRNA (hGBA1 ISH, red) from the Control (Ref), Low (ICM-low), Medium (ICM-medium), and High (ICM-high) dose groups (left to right) from the ICM study (A) and Control (Ref), Low P + T, High P + T, and High P only (left to right) from the IPa study (B). GCase activity, as quantified by the 4MUG assay, was measured in all groups across various brain regions for both the ICM study (C) and the IPa study (D) and is presented as fold change relative to the control group. Solid blue line denotes the target GCase activity based on rodent data. FCx, frontal cortex; CaPu, caudate/putamen; SN, substantia nigra; LC, locus coeruleus; THM, thalamus; AMY, amygdala; PFC, prefrontal cortex; Pu, putamen; MCx, motor cortex; Ca, caudate; GP, globus pallidus; In, insula. Brown boxes around the Pu and THM denote the injected regions in the IPa study. Scale bars represent 5 mm. Mixed-effects model with Dunnett’s multiple comparisons test. Data are presented as medians and range. ∗p < 0.05. To detect the levels of functional protein in these studies, we utilized the 4MUG enzyme activity assay as before. In the ICM study, no significant changes in GCase activity were detected in any brain region analyzed when compared with aCSF control animals (Figure 3C) as indicated by a lack of effect of brain region (F(2.40, 18.75) = 1.84, p = 0.18), dose group (F(3, 8) = 3.81, p = 0.06), or interaction between brain region and dose (F(15, 39) = 0.80, p = 0.67). In contrast, with IPa delivery, there was a trend toward a significant effect of brain region (F(1.827, 3.654) = 6.056, p = 0.07) and dose group (F(1.194, 2.388) = 10.97, p = 0.06) on GCase activity in treated groups compared with aCSF control animals with a two-way RM ANOVA. There was a strong overall trend toward an increase in GCase activity with 13 of 21 and 18 of 21 data points at or above the target activity level in the IPa-High P + T group and IPa-High P only group, respectively (Figure 3D). The lack of statistical significance is likely a consequence of the small sample size and high degree of animal-to-animal variability rather than a true lack of a biological effect. Qualitative changes in GCase activity above aCSF control levels were observed in many regions with the greatest changes detected in the injected regions (putamen and thalamus) followed by the substantia nigra. Modest increases were observed in the prefrontal cortex, motor cortex, and amygdala with more variable responses detected in the caudate. In the mouse study, our minimal efficacious dose (7 × 109 GC in GBA-mutant D409V+/+ mice) produced a ∼2- to 3-fold increase in GCase activity in the cortex and striatum when dosed into WT mice (Figure S2). This fold change increase matched the changes seen in the NHP IPa study from the Low P + T group in the prefrontal cortex (2.10 ± 0.27), motor cortex (1.90 ± 0.21), and putamen (3.07 ± 0.72) (Figure 3D). Therefore, based on our mouse studies, we achieved our target fold changes in cortical and striatal regions in the NHP IPa-Low P + T group. Together these data agree with the bulk tissue RNA and ISH RNAscope data demonstrating that IPa delivery into the thalamus and/or the putamen, but not ICM delivery, led to a robust increase in hGBA1 mRNA expression and functional hGCase protein in the NHP brain. AAV9-GBA1 ICM and IPa delivery into NHPs produced distinct safety findings in the areas of highest viral genome biodistribution and transgene expression We also assessed the effects of AAV9-GBA1 treatment on in-life observations, clinical pathology, and histopathology in NHPs. In the ICM study, one ICM-mid dose animal exhibited transient clinical symptoms, including abnormal gait and limited use of the hindlimb (no longer observed 4 h after initial observation), and one ICM-low dose animal had increased CSF total nucleated cell count that was not considered adverse. AAV9-GBA1-related histopathology findings were present at all dose levels and affected the DRG, TG, spinal cord, cauda equina, dorsal nerve roots, and peripheral nerves (individual animal data presented in Table S2). The DRG changes consisted of minimal to moderate neuronal cell body degeneration/necrosis often associated with minimal to moderate mononuclear cell infiltration and minimal to marked dorsal nerve root nerve fiber degeneration (Figures 4A, 5A, and S6). In the spinal cord, there was AAV9-GBA1-related minimal to moderate nerve fiber degeneration predominantly of the dorsal funiculi (i.e., ascending dorsal white matter tracts) (Figures 4A, 5A, and S6), which corresponded to above-described sensory neuron changes in DRGs. Spinal cord neuronal vacuolation was also observed in AAV9-GBA1-treated animals (Figures 4A, S6, and S7A) characterized by large well-demarcated vacuoles within neuronal cell bodies filled with pale eosinophilic glossy material that was hGCase immunoreactive (Figure S7B). In addition, there were a few degenerate neurons surrounded by increased number of glial cells within the lumbar spinal cord of one ICM-low-dose male and one ICM-high-dose female (Figure 4A). Spinal cord mononuclear cell infiltration was also found in all treatment groups, including aCSF control animals, and was deemed procedure related, possibly exacerbated by the AAV9-GBA1 treatment (Figures 4A and S6). Furthermore, minimal to marked nerve fiber degeneration was observed in the peripheral nerves and cauda equina and minimal mononuclear cell infiltration was seen in the TG without obvious neuronal injury (data not shown). In contrast to the findings in the spinal cord and PNS, there were no AAV9-GBA1-related microscopic findings in any brain region (Figure 4A) likely due to the low transduction levels (Figures 2B and 3C). Microscopic findings from the ICM study that were considered adverse included any severity of neuronal degeneration/necrosis in the DRG and spinal cord and nerve fiber degeneration in the spinal cord (present at all dose levels) and moderate to marked nerve fiber degeneration of the peripheral nerves (present at mid and high dose). The above-described microscopic findings were considered adverse based on inability of the CNS to regenerate or high severity of changes in the PNS that can contribute to clinical signs. AAV9-GBA1 ICM and IPa delivery into NHP produces histopathology findings in different nervous system regions Heatmap of major histopathology findings observed in the (A) ICM study and (B) IPa study in the brain (putamen, thalamus, brain stem, and cortex), spinal cord, and dorsal root ganglia (DRG) regions. Each box denotes the highest severity of finding in an individual animal for the specified tissue/brain region. Severity of findings from no findings to most severe: none (green), minimal (yellow), mild (light orange), moderate (dark orange), marked (red), and severe (black). MN infiltration, mononuclear cell infiltration. AAV9-GBA1 ICM and IPa delivery into NHP produces microscopic lesions in CNS and PNS (A) NHP ICM study: H&E staining of the cervical spinal cord dorsal funiculi and DRG from the control animal no. 1501—within normal limits (top panels, 5×: scale bar, 500 μm, and 20×: scale bar, 100 μm). H&E staining of the cervical (C6) spinal cord and DRG from animal no. 3501: moderate bilateral nerve fiber degeneration of the dorsal funiculi of the spinal cord (arrow) and neuronal cell body degeneration/necrosis (arrow) with mononuclear cell infiltration (∗) and neuronophagia (star) in the DRG (bottom panels, 5×: scale bar, 500 μm, and 20×: scale bar, 100 μm, respectively). (B) NHP IPa study: H&E staining of the brain from control animal no. 1002, block 11/slice 9—within normal limits (1×: scale bar, 3 mm, and 20×: scale bar, 200 μm). H&E staining of the brain from a low-dose P + T animal no. 2501, block 11/slice 9—within the thalamus (Th), there was neuronal eosinophilic material accumulation and neuronal degeneration/necrosis (40×: scale bar, 100 μm) accompanied by perivascular inflammation extending into the surrounding neuropil and gliosis (20×: scale bar, 200 μm). Substantia nigra (SN): within normal limits (1×: scale bar, 3 mm). In the IPa study, clinical signs in all surviving groups including controls consisted of transient tremors, vomiting and associated salivation, miscellaneous cranial/periorbital swelling, skin lesions/discoloration, hunched posture, and weakness and were considered associated with the surgical procedure and expected given the nature of the dosing procedure. These findings largely resolved in 2–3 weeks (4–5 weeks in monkeys treated at both sites). Surgical procedure-related histopathology findings included a linear track of neuropil necrosis accompanied by histiocytic infiltration, hemosiderin pigment (suggestive of previous hemorrhage), and/or nerve fiber degeneration extending from the cerebral cortex to the putamen/thalamus area. In one control animal (no. 1601), in the area of thalamus and putamen, neuropil necrosis was accompanied by locally extensive cavitation. The cause of cavitation may be the injection procedure-related injury of a large blood vessel, which could have led to hemorrhage (hemorrhage not obvious since more than 2 weeks after injection), thrombotic event with subsequent ischemic infarction, and/or too fast injection rate. The procedure-related brain microscopic findings in animal no. 1601, in particular cavitation, were considered adverse due to the extensive damage of the brain parenchyma and associated clinical signs of vomiting (day 1) and tremors occurring on multiple days (day 1, 3, and 40). Procedure-related nerve fiber degeneration along the cannula track in the brain (predominantly affecting the cortex, thalamus, and/or putamen; infrequently extending to the brain stem and/or amygdala) was minimal in severity in controls and minimal to moderate in AAV9-GBA1-treated animals (Figure 4B; Table S2). Minimal nerve fiber degeneration was also present in the cervical and thoracic spinal cord lateral funiculi with a higher incidence observed in group 2 and 4 animals when compared with controls (Table S2). This suggested AAV9-GBA1-related exacerbation of the injection procedure effect. All animals in the High P + T group were euthanized early, between days 17 and 30, due to poor and deteriorating condition characterized by uncoordinated behavior, tremors, hunched posture, decreased activity, and weakness and showing neurological signs (limited hindlimb usage or excessive repetitive movement of a hindlimb) (Table S1). Increases in white blood cell counts with associated increases in neutrophil, monocyte, and platelet counts combined with increased fibrinogen concentration were suggestive of an inflammatory process that correlated with the inflammation in the brain (Table S3). Histopathology results revealed AAV9-GBA1-related findings in the brain including minimal to moderate neuropil inflammation with neuronal degeneration, necrosis and/or loss, parenchymal rarefaction, astro- and microgliosis, and perivascular mononuclear cell infiltration (i.e., perivascular cuffing) at the injection sites (putamen and thalamus), frequently extending into adjacent white matter tracts or other neuronal structures (Figures 4B and 5B; Table S4). Accumulation of bright eosinophilic material in neuronal cytoplasm, which was hGCase immunoreactive (Figure S8), was observed in individual animals in the putamen and/or thalamus and was accompanied by more severe histopathologic findings (i.e., neuropil inflammation with neuronal necrosis). In addition, neuronal vacuoles, which were also hGCase immunoreactive (Figure S9), were observed but accompanied by less severe findings (i.e., perivascular cuffing with rare or no neuronal necrosis). The likely cause of morbidity in the High P + T group was extensive neuropil inflammation with neuronal necrosis/loss in the putamen and thalamus. For all other AAV9-GBA1 dose groups in the IPa study, abnormal neurological observations were noted including, but not limited to, decreased use of hindlimbs, decreased motor activity, absent or reduced withdrawal reflex and patellar reflex, incoordination, and weakness. These observations were test article related but judged non-adverse as they were of limited duration and severity. In the High P only group, clinical pathology changes were observed and suggestive of acute-phase response/inflammation as well as clinically suspected dehydration, similar to what was observed in the High P + T group. AAV9-GBA1-related microscopic findings in scheduled euthanasia animals were observed in the brain and spinal cord (Figure 4B; Table S4). Findings were similar to those described in early euthanized High P + T group animals (Figure 4B). While the severity of brain microscopic findings was greatest at the injection site (putamen and/or thalamus), microscopic findings were also observed in the adjacent cortex (prefrontal, motor), caudate nucleus, and/or substantia nigra with occasional changes also seen in the amygdala (data not shown) and brain stem (Figure 4B). Minimal to moderate perivascular/neuropil inflammation associated with regionally extensive minimal to marked neuronal degeneration/necrosis at the injection sites of the brain was attributed to AAV9-GBA1 treatment since these changes were not observed in the aCSF control group and was considered adverse due to associated tissue damage. Discussion In this study, we showed that early postnatal ICV administration of AAV9-GBA1 to mice led to widespread vector biodistribution and a dose-dependent increase in hGBA1 RNA and functional protein expression in the brain without adverse histopathological changes in age-matched postnatal GBA D409V+/+ and WT mice. In young adult NHPs, ICM delivery led to modest vector biodistribution with minimal hGBA1 RNA and functional protein expression in the brain and adverse microscopic findings in the DRG, TG, spinal cord, and associated nerves (dorsal spinal nerve roots and peripheral nerves). In contrast, IPa delivery into the thalamus and/or putamen produced high vector biodistribution, hGBA1 RNA, and functional protein expression in many brain regions, but it was also associated with significant clinical signs and adverse microscopic findings beyond procedure-related findings in brain regions with the highest hGBA1 expression. In mice, we found that P0 ICV AAV9-GBA1 administration led to a significant increase in hGBA1 expression and functional protein that was well tolerated, without substantial histopathology findings, with up to a 4-fold increase in GCase activity above WT control levels (up through 1 × 1010 GC/mL dose level). A modest but significant increase in microglial process perimeter, indicative of microglial activation, was observed at 1 × 1010 GC/mL in WT mice. However, this was not detected at a higher dose (4 × 1010 GC/mL), and no changes were seen at any dose level in D409V+/+ mice (Figure S3). In contrast, neuronal vacuolation was observed only in the high-dose group (4 × 1010 GC/mL) (Figure S3) in both genotypes. In NHPs, IPa administration also led to a significant increase in GCase activity in target brain regions (Figure 3D), but it was accompanied by significant, adverse histopathological findings (Figure 4B) and clinical signs requiring pre-terminal euthanasia in the IPa-High P + T group. Interestingly, vector biodistribution levels in the NHP IPa study (∼104–106 copies/μg DNA) were comparable to those achieved in mouse ICV study (∼104–106 copies/μg DNA), with slightly higher levels in the injected regions of some NHPs (∼106–107 copies/μg DNA). While this higher vector biodistribution in NHPs could account for some of the histopathology differences, not all animals had higher levels in the putamen and/or thalamus (Figure 2A) and there were also significant histopathology findings in other brain regions, with comparable vector levels to the mouse study (Figure 4B). These data suggest that the total amount of vector delivered to a brain region is not the primary driver, but perhaps species and/or route of administration (ICM vs. IPa) played a more critical role in the differing safety findings between these two studies. Age represents another potential confounding variable in our study, as mouse injections were performed during the early postnatal period, whereas NHP injections were conducted in young adults. Previous groups have demonstrated that adult mice show a more robust humoral immune response to AAV injection than neonatal mice across various routes of administration, including intra-CSF injection.19,20 In addition, early postnatal AAV delivery in mice predominantly transduces neurons, whereas adult administration results in relatively greater astrocyte transduction.21 We did not systematically characterize the specific cell types transduced in each of our experiments. However, based on morphology, substantial neuronal transduction was observed in both the mouse ICV and NHP IPa studies (data not shown). However, it is possible that increased astrocyte transduction in the adult NHP IPa cohort contributed to a heightened immune response, potentially exacerbating the toxicity observed in that study. Future experiments directly comparing early postnatal and young adult NHP AAV administration would be valuable to better define the role of age in the differential immune responses and toxicities observed here. Previous studies have used ICM administration to deliver AAV9, and other widely available capsids, to the NHP brain with varying results. Some groups have reported low levels of brain transduction with AAV9 (≤5% of cells or brain area),6,22 while others demonstrated high levels of brain vector biodistribution,7,8,10 more widespread brain transduction23,24 or viral spread,25 or significant improvements on other outcomes, such as increased expression of the target of interest.26 Our data agree with the findings of low vector genome biodistribution (Figure 2A) and low levels of transgene expression (Figures 2B and 3A) seen in some of these studies.6,22 In addition, many studies that show vector transduction via immunohistochemistry (IHC) or immunoflourescence report a similar clustered staining pattern to that seen in this study (Figure 3A) consistent with a perivascular staining pattern.23,27 Despite multiple reports of AAV ICM administration into NHPs, it is difficult to directly compare viral spread and expression between studies since the endpoints used vary between studies with some focusing only on vector biodistribution, whereas others reporting just transduction or expression values. In addition, the variable findings between our study and some others could also be due to manufacturing and analytical differences, utilization of promoters with differing activity or cell specificity, expression of different cargos (non-native fluorescent proteins vs. native cargo), or use of single-stranded AAV vs. self-complementary AAV (scAAV) configuration. With regard to the latter, it is known that scAAV produces an earlier peak in expression, which drives increased cargo expression at various times.28 In addition to conflicting information about vector spread and expression, mixed safety findings have been reported with intra-CSF (ICM or IT) administration of AAVs. In recent years, a growing number of studies have reported microscopic findings in DRG neurons (DRG toxicity) following intra-CSF AAV administration in NHPs.8,29,30,31,32,33,34 As DRG toxicity has been reported across many different capsids, promoters, and cargos (reviewed in Hordeaux et al.35), this appears to be a class effect with the currently available AAV capsids. However, it is important to note that not all studies report DRG toxicity with intra-CSF AAV administration.33 To date, these DRG findings generally appear to be asymptomatic with clinical symptoms and nerve fiber conduction impairment reported in only a small subset of NHPs.35,36 In humans, there has been only one report of clinical symptoms attributed to DRG toxicity following IT AAV administration. The relatively mild symptoms seen in an amyotrophic lateral sclerosis patient included transient tingling, “painful electric shocks,” and absent or reduced sensory nerve potentials,37 but it is challenging to definitively ascribe these symptoms to AAV-related DRG toxicity since they may have arisen due to an underlying lesion in the patient. Despite the unclear and limited data about DRG toxicity in humans, DRG histopathological abnormalities are widespread in intra-CSF NHP studies and raise concern about the use of ICM administration of currently available capsids in humans. Despite variable NHP expression and safety data, there are several ongoing clinical trials using intra-CSF AAV administration to deliver lysosomal enzymes to the brain of patients with various lysosomal disorders (reviewed in Daci and Flotte38). As lysosomal enzymes can be secreted, it has been shown that limited expression throughout the brain could still be efficacious in improving neurological symptoms through cross-correction.31,39 Most of the current trials target juvenile LSD patients, and while results from some early clinical trials appear promising (reviewed in Kido et al.4), there are no approved therapies in this space to date. In addition, some groups have used this approach to target adult indications of LSDs or other chronic, adult neurological diseases (reviewed in Daci and Flotte38). In particular, there are ongoing clinical trials using ICM administration of AAV9-GBA1 (PR001/LY3884961) in patients with GD (https://clinicaltrials.gov/study/NCT05487599) and in patients with PD with GBA1 mutations (https://clinicaltrials.gov/study/NCT04127578). Limited publicly disclosed preclinical data suggest that this approach is safe and able to achieve therapeutically relevant GCase activity levels in mice and NHPs.40 In contrast, our NHP ICM data show very limited viral spread, low levels of functional GCase protein in brain regions critical in PD and GD (Figure 3C), and more importantly, extensive adverse finding in DRG, spinal cord, and related regions (Figures 4A and 5A) despite the use of an identical serotype and comparable cargo configurations specifically designed to achieve sustained expression of hGBA1. It is possible that differences in the vector design, manufacturing process, analytical characterization, or measured endpoints could account for the disparate findings between our study and the published findings with PR001.40 It is noteworthy that we did not pre-screen for neutralizing antibodies or utilize immunosuppression in the reported results. While intra-CSF and IPa delivery methods are generally less affected by circulating neutralizing antibodies than systemic administration, a recent intra-CSF (IT) study showed that pre-existing AAV9 titers did not alter vector biodistribution in the CNS but did impact biodistribution in the periphery and triggered strong immune responses.41 Therefore, it is possible that pre-existing neutralizing antibodies may have contributed to the immune responses seen in our ICM study. Additionally, data from our group suggest that immunosuppression can greatly reduce but not eliminate the severity and/or incidence of DRG, spinal cord, and other adverse histopathology findings.42 In addition to intra-CSF delivery, IPa injection of AAV has been used previously for delivery of neurotrophic factors and gene replacement strategies in neurological diseases. Original studies focused primarily on AAV2, which demonstrated some spread beyond the injected region in the NHP brain.43,44,45 The only successful approach in the clinic was the use of bilateral intraputamen AAV2-hAADC IPa injection for Aromatic l-amino acid decarboxylse (AADC) deficiency that required only retrograde spread to a connected target region. This approach has been approved in the European Union and the United States,46,47 and other IPa-targeting strategies are also showing promise for this disease.48 More recently, AAV5 and AAV9 have shown promise in NHP IPa studies demonstrating improved viral spread and expression beyond the injected regions with intraputamen and/or intrathalamic injection.11,49 Consistent with these findings, our results demonstrate that intrathalamic and/or intraputamen AAV9 injection led to robust viral spread and functional protein expression in injected and connected brain regions (Figures 3B and 3D). However, this was accompanied by both procedure- and test article-related adverse findings not reported in these previous studies.11,49 While procedure-related differences may explain some of the discrepancies, vector cargo is likely also an important contributing factor as those studies utilized GFP, whereas we used hGBA1. Consistent with this, previous reports using AAV8 and AAVrh10 to deliver other lysosomal enzymes (β-N-acetylhexosaminidase and arylsulfatase A), have described safety findings similar to our NHP IPa study, including eosinophilic accumulation50 as well as local immune cell infiltration, vacuolization, and necrosis.51,52 Moreover, we utilized a codon-optimized hGBA1 sequence, which some studies suggest may be linked with untoward outcomes in the clinic.53 Taken together, the combination of lysosomal enzyme along with codon optimization may explain some of the adverse findings described in this study that were not seen in other reports. Procedure-related findings were observed in a subset of animals, including one control animal (no. 1601) in which adverse histopathology findings including locally extensive brain parenchyma cavitation at the injection sites was observed. Adverse effects in toxicology studies are any changes related to treatment with a test article that are considered potentially harmful to the well-being of the test species, resulting in dysfunction, or negatively impacting the ability of an animal to thrive or develop normally.54 Although animal no. 1601 was not treated with a test article, but injected with the aCSF, morphologic changes at the injection site and associated clinical signs indicated a potentially harmful effect of the IPa procedure in this animal. Other control animals exhibited only non-adverse changes, including gliosis and minimal nerve fiber degeneration (Table S4), suggesting that the adverse procedural effects (from no. 1601) may be attributable to procedural variability and could potentially be mitigated through refinements in technique. As other authors have reported absence of relevant procedure-related safety findings associated with IPa delivery of AAV, we can speculate that achieving a no observed adverse effect level is feasible and could likely be attained in this model with procedural optimization.11,49 In addition, IPa administration is technically challenging for physicians and invasive for patients, thus affecting the risk/benefit calculation of this approach. Therefore, the clinical uptake of any IPa-administered drug will likely remain limited until there are advances in safety and delivery technologies. As we exceeded our target GCase activity with NHP IPa delivery in some brain regions (putamen, thalamus, and substantia nigra) (Figure 3D), there is a possibility that a lower dose could achieve the pharmacodynamic target (GCase activity) without any safety findings. However, the therapeutic window is very limited and would likely be challenging to achieve especially given the intra-animal variability observed. While this may be feasible for focal IPa injections targeting only specific nuclei as with AADC deficiency,46 it is very challenging to achieve with neurological diseases that affect a more widespread area of the brain. Given the limited therapeutic window of IPa delivery for many indications and the expression and safety challenges (DRG toxicity and related findings) for ICM delivery, there is a major effort in the gene therapy field to identify novel AAV capsids that have improved brain transduction, allow sufficient expression, and mitigate safety risks, such as DRG toxicity. In this regard, systemically delivered blood-brain barrier-crossing capsids are yielding promising advances with several companies advancing programs into late preclinical development and at least one entering early clinical evaluation. The results from the initial human clinical trials will be critical for the field to help realize the potential of AAV-based gene therapy across a wider set of neurological diseases. Materials and methods Animals All mouse work was performed according to the Biogen Institutional Animal Care and Use Committee. All NHP work was performed according to the Charles River Laboratories Institutional Animal Care and Use Committee. AAV generation YFP or a codon-optimized version of hGBA1 was cloned in a cis-plasmid containing AAV2-derived inverted terminal repeats (ITRs) under the control of the CAGGs promoter, followed by a strong polyadenylation signal and a woodchuck hepatitis virus post-regulatory element. AAV9-CAGGs-GBA1 (“AAV9-GBA1”) and AAV9-CAGGs-YFP (“AAV9-YFP”; control virus) were prepared by triple transfection in HEK293-derivative cells. Viral particle purification for material used for mouse studies was performed by iodixanol purification (Packgene Biotech). Material for NHP studies was purified from clarified cell lysate by a step of affinity chromatography (POROS CaptureSelect AAV9, Thermo Fisher Scientific), followed by a polishing anion exchange chromatography step (POROS HQ, Thermo Fisher Scientific) and a final buffer exchange to physiologically compatible formulations (PBS or aCSF). To quantify titers, AAV samples were treated with DNase I, diluted, and combined with droplet digital PCR Supermix for probes (no UTP) and a primer/probe mix specific for hGBA1. Droplets were produced and analyzed using QX200 with the AutoDG system from BioRad. Mouse study AAV injection and tissue collection A total of 4 μL viral vector and 0.1% (volume/volume) fast green (Sigma) were delivered into WT or D409V+/+ mice by unilateral ICV injection, followed by passive diffusion at P0. The dye confirmed distribution into both ventricles of each injected P0 neonate. Mice were euthanized at 2 months after injection. One hemisphere of the brain was taken for histopathological analysis and another hemisphere was sub-dissected to collect key brain regions of interest, which were flash frozen in liquid nitrogen. Vector biodistribution Pulverized mouse cortex samples were subjected to tissue lysis, digestion, and DNA extraction procedures using DNeasy 96 Blood & Tissue Kit from QIAGEN. ATL and proteinase K buffer were added to the wells, and homogenates were incubated at 56°C with shaking at 225 rpm with ThermoMixer C overnight. 4 mL 100 mg/mL RNAse A (QIAGEN) was added to each well before proceeding to subsequent DNA extraction procedures following manufacturer’s instruction. All the centrifugation steps were performed with Sigma 4-16KS centrifuge at 6,000 × g (QIAGEN). DNA was eluted in AE buffer, and total DNA concentration was determined with NanoDrop 8000 (Thermo Fisher Scientific). Each eluted DNA sample was normalized to 20 ng/mL with nuclease-free water for qPCR analysis. Sample DNA (20 ng/reaction) and pre-validated linearized plasmid standards containing WPRE and mouse RPP30 sequence (0, 1, 5, 10, 50, 100, 1E3, 1E4, 1E5, 1E6, 1E7, or 1E8 copies/reaction) were analyzed by qPCR (10 mL/reaction) in a 384-well optical plate using custom TaqMan primer/probe sets specific for both vector DNA and mouse genomic DNA at final concentration of 890 nM for primers and 247.5 nM for probes in triplicates with TaqMan Universal PCR Master Mix, no AmpErase UNG (Applied Biosystems, NJ) on a CFX384 system (BioRad). Viral vector DNA detection was performed by detecting WPRE-FAM, and mouse gDNA quantification was done by detecting RPP30-HEX. Measurements of viral vector DNA and gDNA were performed as a duplex assay. PCR cycling conditions were 10 min at 95°C, 40 cycles of 15 s at 95°C, and 1 min at 60°C. The quantities of the viral DNA and the genomic DNA were interpolated from the standard curve prepared. Amplification efficiency (E) was calculated with the following equation: E = −1 + 10 (−1/S), where S is the slope of the standard curve. The unit reported from this qPCR assay for viral DNA quantification was genome copies per microgram DNA (GC/ug DNA). RNA analysis RNA was extracted from flash-frozen mouse cortex samples using the QIAzol (QIAGEN 79306)/chloroform extraction method and further purified by using the QIAGEN RNeasy Plus Mini Kit (QIAGEN 74134). 2 μg of RNA was reverse-transcribed to generate cDNA using SuperScript IV VILO Master Mix (Thermo Fisher 11756500) and then diluted 10-fold with nuclease-free water. 2 μL of cDNA was loaded into 10 μL qPCR reactions using TaqMan Fast Advanced Master Mix (Thermo Fisher 4444557) targeting a codon-optimized sequence of GBA to selectively measure vector-expressed GBA mRNA and mouse ACTB (Thermo Fisher 4352341E) as a housekeeping gene. The 2–ΔΔCt method was used to calculate fold change relative to the lowest dose group. GCase activity assay: Mouse and NHP Tissues samples were homogenized in 10 v/w buffer (250 mM sucrose, 10 mM Tris pH = 7.5, 1 mM EDTA, 0.25% Triton X-100 solution with protease/phosphatase inhibitors) using an Omni Ruptor, sonicated in a water bath, and incubated on ice for 30 min. Samples were centrifuged at 20,000 × g for 15 min at 4°C, and supernatant was collected and quantified with bicinchoninic acid. To measure relative GCase activity, 3 mM 4MUG (Sigma M3633), an artificial substrate of GCase, was incubated with 10 μg lysates with and without 100 mM conduritol-B-epoxide (EMD Millipore 234599), a GBA antagonist, at 37°C for 1 h. Reactions were stopped with 1 M glycine (pH 10.8), and fluorescence was detected with a BioTek synergy microplate reader at 365/445 excitation/emission wavelengths. The absolute GCase activity (nanomole/milligram/hour) was calculated using a 4-methylumbelliferone (Sigma M1381) standard curve. Histopathology In situ hybridization and immunohistochemical staining: Mouse and NHP Formalin-fixed paraffin-embedded (FFPE) tissue sections from murine brain and NHP brain and spinal cord from ICM and IPa studies were evaluated for RNA quality using RNAscope positive control probe Mfa-PPIB (cyclophilin B; Advanced Cellular Diagnostics [ACD] Cat # 424148) and negative control probe dapB (ACD Cat # 3120238). An RNAscope assay designed to detect human codon optimized GBA1 (hGBA1-codon-NoXMn, cat# 827888) transgene and mRNA expression using a red chromogenic kit (ACD Cat #322750). RNA ISH was performed on a Leica BOND RX automated platform according to manufacturer instructions. Bright-field images were collected with a 40× objective using a 3DHistech Pannoramic-250 digital slide scanner. Neuroanatomic regions of interest (ROIs) were manually annotated in whole-slide images and batch-analyzed using two custom image analysis algorithms developed in Visiopharm software (Horsholm, Denmark). The first algorithm was trained to segment and count all gray matter cells, and the second algorithm detected the cells positive for hGBA1 signal based on threshold analysis. The outputs (number of hGBA1-positive cells and total number of cells) were used to calculate percent cellular transduction (% Cell Transduction) per ROI. Resulting data were analyzed using GraphPad Prism, and hGBA1 RNA expression differences were determined using a Kruskal-Wallis test. Murine and NHP FFPE tissue sections were also utilized for IHC assays. Murine brain sections were immunostained for markers of microglial activation on a Ventana Discovery Ultra autostainer. Briefly, 5-μm-thick sections were deparaffinized and rehydrated. A duplex IHC labeling purine-rich box 1 (PU.1) and ionized calcium binding adaptor molecule 1 (Iba1) was performed on murine sections to label microglia cell body and processes, respectively. Heat-induced epitope retrieval was performed at 95°C for 64 min in CC1 buffer. Slides were then incubated for 1 h with a rabbit monoclonal anti-PU.1 antibody (Cell Signaling Cat # 2258) at 6 μg/mL in Discovery antibody diluent (Roche Cat # 760-108). Secondary detection was performed using the Ventana anti-Rb HQ/anti-HQ HRP amplification system and DAB (Multimer HRP). The anti-PU.1 antibody and bound HRP conjugates were denatured and neutralized. The sections were then incubated for 1 h with a primary rabbit polyclonal anti-Iba1 antibody (Wako Cat # 019-19741) at 0.125 μg/mL in Discovery antibody diluent. Secondary detection was performed with Discovery OmniMap anti-Rb HRP and Discovery Purple (Multimer HRP). Slides were counterstained with hematoxylin and coverslipped. NHP spinal cord (ICM study) and brain sections (IPa study) were immunostained for hGBA1 using a Ventana Discovery Ultra autostainer following the manufacturer’s instructions. Briefly, sections were deparaffinized, rehydrated, and subjected to heat-induced epitope retrieval at 95°C for 64 min in CC2 buffer. Tissue sections were then incubated for 1 h with a Biogen-generated rabbit polyclonal anti-hGBA1 primary antibody (5 μg/mL) in Discovery antibody diluent (Roche, Cat # 760-108). Detection was performed using the Ventana anti-Rb HQ/anti-HQ HRP amplification system and DAB, followed by hematoxylin counterstaining and coverslipping. NHP studies Experimental design and in-life assessment Housing setup for both NHP studies was conducted as described in the “Guide for the Care and Use of Laboratory Animals” (National Research Council, 2011). The following parameters and endpoints were evaluated in both NHP studies: in-life assessments (mortality, clinical observations, body weights, food appetence, physical examinations, neurological evaluation) and clinical pathology (hematology, coagulation, clinical chemistry, urinalysis, and CSF analysis [cell count and clinical chemistry]). Neurological examinations included general attitude and motor function, postural reactions, cranial nerve function, and spinal nerve function. ICM study: purpose-bred, naive, 2- to 3-year-old male and female cynomolgus monkeys, were group housed at Charles River Laboratories Montreal ULC Senneville Site (CR-SEN; Senneville, QC, Canada). Animals were dosed on day 1 via a single bolus to the cisterna magna (ICM injection) followed by passive diffusion with reference item (aCSF) or AAV9-GBA1 at dose levels 1.0 × 1013, 3.0 × 1013, or 1.0 × 1014 GC/dose (3 animals/group of mixed sex; study design: Figure S4A). ICM needle placement was confirmed by MRI imaging. Dose volume was 1 mL with injection target rate of 1 mL/min. The syringe was maintained in place for 30 s before the needle was removed. Clinical pathology samples were collected prestudy (week −2 or −3 and week −1), during week 4, and on day 43, except for CSF samples, which were collected prestudy, at dosing, and on day 43. Six weeks after dosing (day 43), animals were sedated with an intramuscular injection of a sedative cocktail (ketamine, dexmedetomidine, and glycopyrrolate) and then euthanized by an intravenous overdose of sodium pentobarbital, followed by exsanguination. Necropsies were performed and organ weights collected by Charles River Laboratories, Montreal personnel. IPa study: purpose-bred, naive, 19- to 45-month-old male and female cynomolgus monkeys of Vietnamese and Indonesian origins, were group housed at Charles River Laboratories Montreal ULC Laval Site (CR-LAV; Laval, QC, Canada). The animals were fasted overnight (at least 8 h) prior to dosing. On day 1, animals were dosed via IPa injection, i.e., infused bilaterally in the putamen and/or thalamus, via an MRI-directed catheter. MRI-assisted injections to the target regions were performed using ClearPoint systems with frames secured over the frontal and parietal lobes.55 After SmartFlow cannulas (NGS-NC-05) were inserted steadily through the drill holes over 30 s–1 min, the 2 sites on each hemisphere were infused sequentially (left putamen, left thalamus, right putamen, right thalamus) using convection-enhanced delivery. Animals were dosed at a volume of 0.15 mL/putamen and 0.125 mL/thalamus, with reference vehicle (aCSF) or low concentration (2.5 × 1012 GC/mL) or high concentration (7.5 × 1012 GC/mL) test article (3 animals/group of mixed sex; study design: Figure S4B). Dose administration in each structure used a convection-enhanced delivery strategy with a 2-step increase in dose rate i.e., 10 μL at 120 μL/h for the initial 5 min followed by the remaining volume at 300 μL/h. Following the completion of dosing, the cannula was left in place for 10 min before stepwise removal over the course of approximately 1 min. The surgical area was then closed, and a topical antibiotic was administered to all the surgical sites daily until complete healing. Clinical pathology samples (hematology, coagulation, clinical chemistry and urinalysis) were collected prior to dosing (week −1 or day 1) and prior to necropsy (day 43). Six weeks after dosing (day 43), animals were sedated with an intramuscular injection of a combination of ketamine hydrochloride and acepromazine and then euthanized by an intravenous overdose of sodium pentobarbital, followed by exsanguination. Necropsies were performed and organ weights collected by Charles River Laboratories Montreal ULC (Laval) personnel. Histopathology ICM study: tissues for microscopic evaluation were fixed in 10% neutral buffered formalin, trimmed, processed routinely, embedded in paraffin, and stained with hematoxylin and eosin by Charles River Laboratories Inc, Montreal, Canada. A detailed histopathological evaluation was performed on study plan tissues (Table S5) for all animals by a board-certified veterinary pathologist. Tissues were evaluated by light microscopy. IPa NHP study: tissues for microscopic evaluation were fixed in 10% neutral buffered formalin, trimmed, processed routinely, embedded in paraffin, and stained with hematoxylin and eosin by Charles River Laboratories Montreal ULC Laval (Québec) Canada; Inotiv Boulder, Boulder, CO; or Charles River Laboratories, Inc., Frederick, MD. Microscopic evaluation was conducted by a board-certified veterinary pathologist on all protocol-specified tissues from all animals (Table S6). Tissues were evaluated by light microscopy. Biodistribution—AAV DNA Tissue samples were subjected to lysis, digestion, and DNA extraction procedures. In brief, lysis buffer (Promega, A826B) and proteinase K buffer (Promega, MC500) were added to the tubes, samples were homogenized using a Geno Grinder, and DNA was extracted using a Maxwell RSC instrument according to the manufacturer’s instruction. DNA was quantified, and purity was assessed using a SpectraMax spectrophotometer. Sample DNA (0.14 μg/μL) and pre-validated linearized plasmid standards containing WPRE and mouse RPP30 sequence were analyzed in a 384-well optical plate by qPCR using custom TaqMan primer/probe sets specific for both vector DNA and mouse genomic DNA (QuantStudio 7). Samples were run in duplicate, and results were reported as normalized to total genomic DNA detected using RPP30. Only samples with a quantifiable RPP30 PCR result ≥ RPP30 lower limit of quantitation (685.58 pg/PCR) were used for reporting results of AAV-GBA copies (detected by WPRE) normalized to total gDNA (detected by RPP30). RNA analysis RNA was extracted from flash-frozen tissue samples using a QIAsymphony by homogenization into RLT lysis buffer followed by chloroform extraction. Total RNA was analyzed for purity and concentration using a NanoDrop8000. RNA was reverse transcribed to generate cDNA using SuperScript IV VILO Master Mix (Thermo Fisher 11756500) and then diluted with nuclease-free water. 4 μL of cDNA was loaded into 20 μL qPCR reactions using TaqMan Fast Advanced Master Mix (Thermo Fisher 4444557) targeting a codon-optimized sequence of human GBA1 (at the exon 1-2 junction) to selectively measure vector-expressed GBA1 mRNA and RPP30 as a control gene. The 2–ΔΔCt method was used to calculate fold change relative to the aCSF control group. Statistics Statistical analysis was performed using Graph Pad Prism software. Boxplots, with the central line representing the media, the edges representing the interquartile ranges, and the whiskers representing the minimum and maximum values, or scattered dot plots, with the median and min to max values, are used for graphical representation. For heatmaps, mean values were presented, except for Figure 4, where maximum values are plotted. Sample sizes are included in the figure legends for clarity. Mixed effects model (or two-way ANOVA for Figure S3) with Dunnett’s post hoc t tests were performed at a significance (a) of 0.05. Exact p values (two-tailed) are reported. Data and code availability The data supporting the findings described in this report are the property of Biogen Inc. and can be obtained from the corresponding author (L.K.) upon reasonable request. Acknowledgments We thank Ashley Mallat (for supporting and planning the NHP studies) and Terrence Dobrowski, Linda Engle, Vincent Lau, Rich Macniven, Kevin Mangaudis, Michael Nguy, John Pieracci, and Chelsea Worley (for AAV material generation and analytical characterization for the NHP studies). All the experiments included in this report were funded by Biogen Inc. Author contributions A.C.A. and N.A., investigation, methodology, visualization, and writing; B.G., L.K., and E.D.P., conceptualization, analysis, methodology, visualization, and writing; D.G., K.N., and K.E.G., conceptualization, formal analysis, methodology, supervision, visualization, and writing; K.N., formal analysis, methodology, supervision, and writing. P.T. and E.T., conceptualization and analysis; P.C., S.X., J. Sebalusky, D.B., and O.R.M., investigation and methodology; S.H., investigation, methodology, and visualization; S.B., Y.C., P.J.D.G., M.M., and W.Z., investigation only. M.V.P.N., analysis and methodology; J.F., D.M.W., W.D.H., and J. Suh, supervision and writing. K.-E.C. and M.K., project administration, methodology, and supervision. Declaration of interests A.C.A., B.G., D.G., L.K., N.A., S.B., Y.C., K.-E.C., P.C., P.J.D.G., S.H., M.K., M.M., M.V.P.N., K.N., J. Sebalusky, W.Z., S.X., P.T., E.D.P., E.T., J.F., D.M.W., W.D.H., J. Suh, and K.E.G. were all employees of Biogen when this work was completed. D.B. and O.R.M were employees of Charles River Laboratories when this work was completed. Footnotes Supplemental information can be found online at https://doi.org/10.1016/j.omta.2026.201779. Supplemental information Document S1. Figures S1–S9 and Tables S1, S3, S5, and S6 Data S1. Tables S2 and S4 Document S2. Article plus supplemental information References Associated Data Supplementary Materials Document S1. Figures S1–S9 and Tables S1, S3, S5, and S6 Data S1. Tables S2 and S4 Document S2. Article plus supplemental information Data Availability Statement The data supporting the findings described in this report are the property of Biogen Inc. and can be obtained from the corresponding author (L.K.) upon reasonable request.