The Role of Basal Forebrain Neurons in Adrenomyeloneuropathy in Mouse and Human

Abstract

Objective:

X-linked adrenoleukodystrophy is caused by mutations in the peroxisomal half-transporter ABCD1. The most common manifestation is adrenomyeloneuropathy, a hereditary spastic paraplegia of adulthood. The study set out to understand the role of neuronal ABCD1 in mice and humans with adrenomyeloneuropathy.

Methods:

Neuronal expression of ABCD1 during development was assessed in mice and humans. ABCD1 deficient mice and human brain tissues were examined for corresponding pathology. Next, we silenced ABCD1 in cholinergic Sh-sy5y neurons to investigate its impact upon neuronal function. Finally, we tested adeno-associated virus vector mediated ABCD1 delivery to brain in mice with adrenomyeloneuropathy.

Results:

ABCD1 is highly expressed in neurons located in the periaqueductal gray matter, basal forebrain and hypothalamus. In ABCD1-deficient mice (Abcd1-/y), these structures exhibited mild accumulations of α-synuclein. Similarly, healthy human controls had high expression of ABCD1 in deep gray nuclei, while X-ALD patients displayed increased levels of phosphorylated tau, gliosis and complement activation in those same regions, albeit not to the degree seen in neurodegenerative tauopathies. Silencing ABCD1 in Sh-sy5y neurons impaired expression of functional proteins and decreased acetylcholine levels, similar to observations in plasma of Abcd1-/y mice. Notably, hind limb clasping in Abcd1-/y mice was corrected through transduction of ABCD1 in basal forebrain neurons following intracerebroventricular gene delivery.

Interpretation:

Our study suggests that the basal forebrain-cortical cholinergic pathway may contribute to dysfunction in adrenomyeloneuropathy. Rescuing peroxisomal transport activity in basal forebrain neurons and supporting glial cells might represent a viable therapeutic strategy.

Summary for Social Media If Published

Adrenomyeloneuropathy is a neurodegenerative disease caused by ABCD1 deficiency. While glial cells are traditionally thought to be the key player in this disease due to their abundant expression of ABCD1, the role of neuronal ABCD1 remains elusive. The present study set out to understand the role of neuronal ABCD1 in mice and humans with adrenomyeloneuropathy. We demonstrate that ABCD1 is highly expressed in neurons specifically located in brain regions like periaqueductal gray matter, basal forebrain and hypothalamus. ABCD1 deficiency in cholinergic neurons causes significant neuronal dysfunction while correction of basal forebrain cholinergic neurons by AAV-mediated ABCD1 gene delivery leads to improvement of hind limb clasping behavior. Thus, we postulate that the basal-cortical cholinergic pathway contributes to dysfunction in adrenomyeloneuropathy. Rescuing peroxisomal transport activity in basal forebrain neurons and supporting glial cells might represent a viable therapeutic strategy.

Introduction

X-linked adrenoleukodystrophy (X-ALD) is a progressive neurological disorder caused by mutations in the ABCD1 gene^1–3. ABCD1 encodes a peroxisomal ATP-binding cassette transporter responsible for the transport of CoA-activated very long-chain fatty acids (VLCFA) into the peroxisome for degradation. The most common phenotype of X-ALD is adrenomyeloneuropathy (AMN), a debilitating myeloneuropathy that affects nearly all adult patients. The myelopathy in AMN manifests as slowly progressive paraparesis with sphincter disturbances and sensory loss for which no treatment is available⁴. By age 60, more than 80% of surviving X-ALD men show evidence of AMN⁵. Mice deficient in ABCD1 (Abcd1-/y) have a phenotype similar to AMN, with motor and sensory symptoms emerging around a year of age ^6,7.

Prior studies have largely focused on the role of ABCD1 in glial cells due to the relative abundance of ABCD1 expression in these cell types⁸. ABCD1 deficiency affects microglia and endothelial cells in both human tissue and mouse models ^9–12 and sensitizes oligodendrocytes and astrocytes, leading to inflammation and oxidative stress ^7,16. Correction of oligodendrocytes and astrocytes by adeno-associated virus in Abcd1−/− mice demonstrates significant functional improvement¹³. Yet the importance of neurons underlying the axonopathy may be greater than previously recognized ⁸. Höftberger and co-authors previously reported high levels of ABCD1 expression in certain neurons in the hypothalamus, basal nucleus of Meynert, and periaqueductal gray ¹⁴. However, there is currently no in-depth study available on the pathological changes in these regions.

In the present work, we performed a systematic analysis of neuronal ABCD1 expression across the mouse brain and characterized the corresponding pathological changes in mouse and cellular models with ABCD1 deficiency, as well as postmortem brains of X-ALD patients. Additionally, we explored the behavioral impact of adeno-associated virus (AAV) mediated gene therapy in these models and determined the cell types responsible for therapeutic benefit.

Materials and Methods

Animals.

Congenic C57BL/6J Abcd1−/− and wild type C57BL/6J mice were ordered from the Jackson laboratory. C57BL/6J Abcd1−/− were back crossed onto a pure C57BL/6J background over 6 generations. They were then bred from homozygous founders and genotyped periodically. C57BL/6J CHAT^BAC-eGFP were purchased from Jackson laboratory. All mice were kept in the animal housing facility of the Massachusetts General Hospital (MGH) Center for Comparative Medicine, had ad libitum access to water and standard rodent food, and were kept on a 12-hour light and dark cycle. A protocol for early euthanasia/humane endpoints was executed if one of the following criteria was met: loss of body weight more than 15% or a wound that could not be improved with medication. All animal experiments were approved by the Institutional Animal Care and Use Committee at MGH.

Human brain specimen.

Human tissue studies were performed on post-mortem brain tissues from 4 adult X-ALD patients and 6 unaffected control cases obtained from NIH NeuroBioBank. Use of this material was approved by the Institutional Review Board of MGH (Table S1).

Neuronal cell culture and ABCD1 silencing.

IMR32 and Sh-sy5y neuronal cell lines were obtained from ATCC and cultured according to the manufacturer’s instructions. Briefly, cells were cultured in DMEM/F12 medium plus 10% fetal bovine serum (Atlanta Biologicals) supplemented with Amphotericin B, Penicillin and Streptomycin (Thermo Fisher) at 37 °C for maintenance. For neuronal differentiation, IMR32 and Sh-sy5y cells were grown in DMEM/F12 plus 2% FBS and differentiated with 2mM sodium butyrate or 1–10μM retinoic acid respectively for 6 days. For ABCD1 silencing, 2X10⁵ Sh-sy5y cells were seeded in 6 well plates and differentiated for 3 days. ABCD1 was silenced via targeted siRNA (DharmaFECT^®, GE healthcare) for 3 days with non-targeting siRNA treatment as control.

Recombinant AAV9 vectors.

Human ABCD1 cDNA was kindly provided by Johannes Berger, University of Vienna, and inserted into pAAV-CBA plasmid using flanking HindIII and XhoI restriction sites. This rAAV plasmid contains the strong CMV enhancer/chicken β-actin hybrid promoter, a woodchuck post-transcriptional regulatory element, and SV40 and BGH poly A sequences flanked by inverted terminal repeats. This encodes a single stranded AAV genome. rAAV production has been previously described^15–17. To optimize detection of ABCD1, a C-terminal HA-tagged human ABCD1 cDNA was separately cloned into pAAV-CBA vector, named pAAV-CBA-ABCD1-HA and packaged into rAAV9 in 293T cells.

Vector administration.

Intravenous (IV) delivery:

C57BL/6J Abcd1-/y mice were placed in a restrainer (Braintree Scientific, Inc.; Braintree, MA). Next, the tail was warmed in 40 °C water for 30 seconds, before disinfection with 70% isopropyl alcohol pads. Using a 100–300 μl volume of viral vector in PBS containing 1–3X10¹² genome copies rAAV9 vector (packaging a GFP or hABCD1 expression cassette) was slowly injected into a lateral tail vein, followed by gently clamping the injection site to stop bleeding.

Stereotactic intracerebroventricular (ICV) injection:

C57BL/6J Abcd1-/y mice or C57BL/6J CHAT^BAC-eGFP mice were anesthetized using isoflurane during the whole surgical procedure and placed in a rodent stereotactic frame. rAAV9 vectors were infused into the left lateral ventricle (coordinates from Bregma in mm: AP-0.2, ML+1.0, DV-3.0) using a Harvard 22 syringe pump (Harvard Apparatus) to drive a gas-tight Hamilton syringe attached to a 33-gauge steel needle (Hamilton Co.). Briefly, 1X10¹¹ genome copies rAAV9 in 10μl (carrying GFP or hABCD1) were injected at a rate of 5μl/min, after which the needle was left in place for 5 min to prevent backflow before withdrawal. The same volume of PBS was injected by the same approach in the control mice.

Intrathecal (IT) delivery:

C57BL/6J Abcd1-/y mice were anesthetized using isoflurane. After the skin over the lumbar region was shaved and cleaned, a 3~4 cm mid-sagittal incision was made through the skin exposing the muscle and spine. A catheter was inserted into the L4-L5 spine region and attached to an Alzet^® mini osmotic pump (Alzet, Cupertino, CA) containing the rAAV9 vectors in 200μl volume. The osmotic pump was implanted under the skin for 24h and removed the next day. For bolus injection, the catheter was attached to a gas-tight Hamilton syringe with a 33-gauge steel needle. 1X10¹¹ genome copies rAAV9 vectors (carrying GFP or hABCD1) in 10μl volume were slowly injected at a rate of 5 μl /min.

Tissue and plasma preparation.

Mice were anesthetized by isoflurane and blood was collected for plasma preparation. Mice were sacrificed by transcardial perfusion of PBS. After removal, brain tissues were snap-frozen and stored at −80°C until use. For some mice, dissected tissues were fixed by 4% paraformaldehyde and equilibrated in 30% sucrose prior to sectioning.

Western blotting.

Tissue and cell lysates were prepared by using RIPA buffer (Sigma-Aldrich, St.Louts, MO) with 1% Halt protease and phosphatase Inhibitor Cocktail (Roche, Indianapolis, IN). Protein samples were separated on NuPAGE 4–12% Bis-tris gels (Invitrogen, Carlsbad, CA) and transferred on PVDF membranes. Membranes were blocked with 5% non-fat milk in PBS containing 0.05% Tween 20 and probed with antibodies listed in supplementary Table S2 followed by corresponding HRP-conjugated secondary antibodies. Membranes were developed with SuperSignal West Pico Chemiluminescent Substrate (Thermo).

Immunofluorescence staining and confocal microscopy imaging.

Tissue sections (10–14 μm) were cut at −25 °C using a cryostat microtome (Leica), then immunostained following the routine immunofluorescence staining protocol. In brief, slides were permeabilized in blocking buffer containing 0.3% Triton X and 2% goat serum for 1 hour and then stained with antibodies separately or in combination at 4 °C overnight. All first antibodies used were listed in supplementary Table S2. Alexa Fluor 488 and 555 goat anti-mouse, Alexa Fluor 488 goat anti-rabbit antibody (Invitrogen, Eugene, OR) and cy3 goat anti-rabbit (Jackson lab) were used as secondary antibodies. Sections were mounted in mounting medium with DAPI (Vector Lab) and analyzed using fluorescence microscopy (Zeiss LSM 800 Airyscan).

Immunohistochemistry.

Tissues fixed in 4% paraformaldehyde were equilibrated in PBS followed by paraffin embedding and sectioning in the MGH histology core. Slides were stained with antibodies against ABCD1, A-beta42, α-synuclein or p-tau (Table S2) according to routine immunohistochemistry staining protocol. Appropriate positive and negative controls were included in each IHC run.

Quantitative real time reverse transcription-PCR.

Total RNA from cells or tissues was isolated using Qiagen RNeasy Mini Kit (Qiagen). First-strand cDNA synthesis used 100 ng random primer (Life Technologies), 1.0 μg total RNA, 10 mM dNTP, and 200 units of reverse transcriptase (Life Technologies) per 20 μL reactions. PCRs were performed in duplicates in a 25μl final volume by using SYBR Green master mix from Applied Biosystems (Life Technologies). The data was analyzed by calculating the delta Ct value between the tested gene and an internal control.

Acetylcholine measurement.

Acetylcholine levels in cell culture medium and mouse plasma were determined by QuickDetect^™ ACh ELISA Kit (Biovision) according to the manufacturer’s instructions.

Mouse rotarod testing motor coordination and balance.

An accelerating Rotarod apparatus (Economex, Columbus Instruments) was used to assess murine motor coordination and balance. Mice were placed in separate lanes on the rotating cylinder with parameters set as Motor = 0, Accelerator = 0.3. The latency time to fall off the apparatus was recorded in seconds. We used a 3-day training/testing scheme with three consecutive trials on each day. Between each trial, mice were allowed to rest for at least 10 min. Mean values of the three trials for each mouse were taken for statistical analysis.

Mouse hind limb extension reflex test.

Each mouse was gently lifted by the tail for 10 seconds and recorded by video. The hind limb extension reflex was analyzed by an observer blinded to genotype and intervention and scored as previously published¹⁸, 0: paralysis of both hind limbs; 0.5: paralysis of one hind limb; 1: clasping of hind limbs; 1.5: alternating clasping and flexion of hind limbs; 2: flexion of hind limbs; 2.5: alternating flexion and extension of hind limbs at an angle < 90°; 3: extension of hind limbs at an angle < 90°; 3.5: alternating extension of hind limbs at an angle < 90° and ≥90°; 4: extension of hind limbs at an angle ≥ 90°. The performance score of each animal was recorded for statistical analysis.

Mouse beam walking test.

For this test, the beam was 0.6 cm wide and 120 cm in length and suspended about 60 cm above foam pads. The average score (total time crossing the beam averaged in 3 trials) was calculated for each animal.

Statistical analysis.

Data were expressed as means ± SEM and analyzed for statistical significance by ANOVA followed by Bonferroni test among experimental groups using Graphpad prism 9. A p < 0.05 was considered statistically significant.

Results

Regional differences and developmental expression of ABCD1.

To assess ABCD1 protein expression in neurons, we dissected wild type C57BL/6J mouse CNS tissues at two different ages, postnatal day 2 and postnatal day 17, and performed immunofluorescence co-staining with antibodies against ABCD1, NeuN (neuronal marker) and tyrosine hydroxylase. Surprisingly, ABCD1 protein expression was abundant in neurons across different brain areas at postnatal day 2 but declined markedly by postnatal day 17 in most brain areas such as cerebral cortex, substantia nigra and thalamus (Figure S1). Similarly, in the ventral horn of the spinal cord, where motor neurons reside, ABCD1 expression was high at postnatal day 2 but had declined markedly by postnatal day 17 (Figure S2). In contrast, certain neurons in the hypothalamus, basal forebrain (in particular the diagonal band nucleus, NDB) and periaqueductal gray retained high ABCD1 expression at postnatal day 17 (Figure 1A), as previously reported in human tissue ¹⁴. To further examine age-dependent ABCD1 expression, we dissected mouse tissues of different ages: postnatal day 6, 1 month and 22 months, and analyzed protein expression in 3 different areas of the brain: cerebral cortex, cerebellum and brain stem, as well as spinal cord and dorsal root ganglia. Consistent with immunofluorescence data presented above, there was a dramatic decrease of ABCD1 protein in cerebral cortex, brain stem, and spinal cord by 1 month of age, but no further decrease by 22 months (Figure 1B, C).

Figure 1. Regional differences in murine neuronal ABCD1 and the impact of development.

A. ABCD1 immunostaining (green) in neurons of wild type C57BL/6 mouse periaqueductal gray (PAG), hypothalamus and basal forebrain at postnatal day 2 (P2) and postnatal day 17 (P17). Tissue is co-stained with the neuronal marker NeuN (red) B. Representative western blot images showing ABCD1 expression in different CNS regions at P6, 1 month and 22 months of age. White arrow indicates neuron staining. C. ABCD1 protein quantification in different CNS regions at P6, 1 month and 22 months of age. Results were expressed as mean ± SEM.

Neuronal subtypes highly expressing ABCD1.

To determine the subtype of neurons with higher ABCD1 expression, we first co-stained ABCD1 with tyrosine hydroxylase, a marker for dopamine, norepinephrine, and epinephrine-containing neurons¹⁹. As shown in Figure 2A, the majority of the ABCD1-expressing neurons in basal forebrain, hypothalamus and periaqueductal gray did not co-express with tyrosine hydroxylase at postnatal day 16. However, there was a cluster of tyrosine hydroxylase positive neurons in pontine central gray highly expressing ABCD1. To assess expression of ABCD1 in cholinergic neurons, we used a C57BL/6 CHAT^BAC-eGFP mouse model that expresses GFP in all choline acetyltransferase (ChAT)-expressing neurons. Analysis in C57BL/6 CHAT^BAC-eGFP mouse brain confirmed higher expression of ABCD1 at postnatal day 5 but a decline with age in most regions (Figure S3). In the basal forebrain and in particular the diagonal band nucleus, which are major sources of cholinergic innervation²⁰, high ABCD1 expression colocalizes with ChAT (Figure 2B and Figure S4).

Figure 2. Murine neuronal subtypes expressing high levels of ABCD1.

A. Co-staining of tyrosine hydroxylase (TH) (red) and ABCD1 (green) (indicated by white arrow) in brain regions with higher ABCD1 neuronal level at P16 (hypothalamus, basal forebrain, periaqueductal gray (PAG), and certain hindbrain regions near the fourth ventricle). B. Co-staining of ChAT and ABCD1 in basal forebrain, hypothalamus and periaqueductal gray (PAG) at P5 and P16 in CHAT^BAC-eGFP mice.

Neuropathology of deep gray nuclei in the X-ALD mouse model.

Pathological changes in the Abcd1-/y mouse brain are mild, with no clear activation of glial cells (data not shown) compared to significant microglial activation in spinal cord¹⁰. However, analysis of basal forebrain and hypothalamus showed slightly more α-synuclein accumulation in some Abcd1-/y mice (Figure 3A), suggesting mild neuronal pathology. Further, Abcd1-/y mice also displayed mild reductions of acetylcholine in plasma (Figure 3B).

Figure 3. Neuropathology of deep gray nuclei in the X-ALD mouse model.

A. Representative immunohistochemistry staining of α-synuclein in mouse basal forebrain and hypothalamus at 20 months of age. B. Slightly reduced acetylcholine level in Abcd1-/y mouse plasma at 13–15 months of age (n=14 for WT and n=15 for Abcd1-/y). Results were expressed as mean ± SEM.

Regional differences of neuronal ABCD1 expression in postmortem human brain.

To determine expression of ABCD1 in control human brain tissues, we examined basal ganglia (nucleus accumbens, and globus pallidus), hypothalamus as well as motor cortex (Brodmann area 4). Immunohistochemistry in these tissues demonstrated moderate ABCD1 expression in neurons within nucleus accumbens, globus pallidus and hypothalamus regions. In contrast, ABCD1 protein was barely detectable in neurons within the motor cortex (Figure 4). These results corroborated prior studies that reported high ABCD1 expression in neurons of the basal nucleus and hypothalamus ¹⁴.

Figure 4. Regional differences of neuronal ABCD1 expression in unaffected human brain tissue.

ABCD1 immunohistochemistry in different human brain regions nucleus accumbens (NA), medial globus pallidus (GPi), hypothalamus and motor cortex (Brodmann area 4). Enlarged images showing ABCD1 expression in neurons.

Pathology of deep gray nuclei in X-ALD patients.

The high ABCD1 expression in neurons of basal forebrain and hypothalamus prompted us to assess the molecular neuropathology in these specific regions within the X-ALD brain. Analysis from 4 X-ALD samples, not reported to be affected by brain demyelination, and 6 control samples showed increased mRNA expression glial activation markers, GFAP, C1QA and C3, as well as partial reduction of neuronal plasticity marker SYP gene in the nucleus accumbens, medial globus pallidus, lateral globus pallidus and hypothalamus tissues from X-ALD patients (Figure 5A-D). Protein expression analysis confirmed significantly increased C3 expression in X-ALD basal ganglia and significantly reduced synapse protein PSD95 expression in X-ALD medial globus pallidus and hypothalamus (Figure 5E-F). These data suggest increased neuroinflammation and complement activation, concurrent with impaired synaptic plasticity and integrity in ALD patient brains. Immunohistochemistry staining with A-beta 42 revealed mild staining in basal ganglia neurons as previously reported²¹; however, neither staining with A-beta 42 nor staining with α-synuclein in ALD patient brains differed significantly from that in age-matched controls (Figure S5). In contrast, staining with p-tau demonstrated increased p-tau signal in the basal ganglia of several X-ALD samples but rarely in controls (Figure 5G). Extensive p-tau staining was found in the Alzheimer’s brain and served as a positive control (Figure S6). As comparison, we also performed analysis in the motor cortex. While increased GFAP expression persisted, pointing to astrocyte activation, no clear indication of complement activation such as increased C1qa or C3 expression was observed. Despite the lack of change in p-tau expression in X-ALD motor cortex, we found a significant reduction in synaptic proteins (SYP and PSD95) (Figure 5H-J).

Figure 5. Pathological analysis of deep gray nuclei in X-ALD patients.

mRNA analysis of several neuronal, glial as well as complement markers in (A) nucleus accumbens (NA), (B) Medial globus pallidus (GPi), (C) Lateral Globus pallidus (GPe) and (D) hypothalamus (HT) of X-ALD patients (n=4) compared to unaffected controls at similar ages (n=6). E. Representative western blot images showing neuronal and complement markers in NA, GPi, GPe and HT of X-ALD patients (n=4) compared to unaffected controls at similar ages (n=6). F. Protein quantification of neuronal and complement markers in in NA, GPi, GPe and HT of X-ALD patients (n=4) compared to unaffected controls. G. Representative immunohistochemistry staining showing positive p-tau signal (Enlarged square) in the NA, GPi, GPe and HT of X-ALD patients (n=4) compared to unaffected controls (n=6). H. mRNA analysis of several neuronal, glial and complement markers in motor cortex BA4 (Brodmann area 4) of X-ALD patients (n=5) compared to unaffected controls (n=4). I. Representative western blot images showing neuronal and complement markers in motor cortex of X-ALD patients (n=5) versus unaffected control (n=4). J. Protein quantification of neuronal and complement markers in motor cortex. Results were expressed as mean ± SEM. *p<0.05, **p<0.01 as compared with control.

Loss of ABCD1 impacts cholinergic neurons.

To determine the potential role of ABCD1 in neurons, we performed in vitro culture and differentiation of IMR32 and Sh-sy5y neurons. We found Sh-sy5y cells to have higher ABCD1 expression compared to IMR32 cells, and retinoic acid treatment between 1–10μM successfully induced cell differentiation into cholinergic neurons, confirmed by high expression of ChAT (Figure S7). Surprisingly, ABCD1 silencing in Sh-sy5y cells caused a significant reduction of cholinergic markers, CHAT, VACHT (vesicular acetylcholine transporter) as well as TH, DAT (dopamine transporter) and PSD75. At a protein level we verified the reduction of ChAT and TUJ1, a neuron specific class-III β-tubulin. Interestingly, ABCD1 silencing also caused significant reduction in acetylcholine secretion in the culture media, suggesting an impact upon cholinergic neuron function (Figure 6A-E). Surprisingly, no significant change in VLCFA levels were detected after ABCD1 silencing (Figure 6F). Instead, we found significantly increased oxidative stress marker GPX1, as well as marked decreases in SIRT1 and EZH2 levels, suggesting molecular perturbances caused by ABCD1 silencing and activation of further downstream signaling pathways implicated in regulating neuronal function^{22, 23} (Figure 6C, G). To further understand whether these stress response and chromatin silencing factors are relevant in human X-ALD, we examined postmortem brain tissue of ALD patients. Analysis in human deep gray nuclei revealed reductions of SIRT1 protein across all structures, in line with our in vitro observation. Reductions of EZH2 protein were found in nucleus accumbens and hypothalamus but not globus pallidus (Figure 7).

Figure 6. Loss of ABCD1 impacts cholinergic neurons in differentiated Sh-sy5y cell model.

A. Morphology of Sh-sy5y cells after ABCD1 silencing as compared to non-targeting control (NT). B. Impact of ABCD1 silencing on the mRNA expression of ABCD family members as well as various neuronal markers (n=6 for each). C. Representative western blot images showing impact of ABCD1 silencing on protein markers. D. Protein quantification of neuronal markers in ABCD1 silenced Shsy5y cells versus a NT siRNA control group. (n=6 for each). E. Impact of ABCD1 silencing on acetylcholine (ACh) production (n=6 for each). F. ABCD1 silencing in differentiated SHsy5y cells had no significant impact on very long chain fatty acid ratios (C26:0/C22:0 and C24:0/C22:0). G. Protein quantification of signaling markers in ABCD1 silenced Shsy5y cells versus NT siRNA control (n=6 for each). Results were expressed as means ± SEM. *p<0.05, **p<0.01 as compared with NT control.

Figure 7. Reduced SIRT1 and EZH2 protein expression in deep gray nuclei of X-ALD brain.

A. Representative western blot images showing SIRT1, EZH2, and CREB proteins in nucleus accumbens (NA), Medial globus pallidus (GPi), Lateral globus pallidus (GPe) and hypothalamus (HT) of ALD patients (n=4) compared to age-matched unaffected controls (n=6). Protein (B) and mRNA (C) quantification of SIRT1 in deep gray nuclei. Protein quantification of EZH2 (D) and CREB (E) in deep gray nuclei. Results were expressed as means ± SEM. *p<0.05, ***p<0.001 as compared with control.

Intracerebroventricular delivery of rAAV9-CBA-GFP led to extensive brain transduction.

By comparing the distribution of rAAV9-CBA-GFP across brain and spinal cord after different routes of delivery, we demonstrated that intracerebroventricular (ICV) delivery achieved efficient brain transduction (Figure 8A) compared to intravenous (IV) delivery and intrathecal (IT) delivery, which mainly targets the spinal cord (Figure 8B), Given these data, ICV delivery was chosen as the route of AAV-mediated ABCD1 gene delivery to the brain.

Figure 8. rAAV9-CBA-GFP mediates transgene expression across the mouse brain and spinal cord following different routes of delivery.

Representative images show rAAV9-CBA-GFP distribution across various regions of the brain (A) and spinal cord (B) after different routes of delivery (intrathecal osmotic pump (IT pump) at 1×10¹¹ gc/mouse (n=3), intrathecal bolus (IT bolus) at 1×10¹¹ genome copies (gc)/mouse (n=4), intravenous (IV) at 1×10¹² gc/mouse (n=3), and intracerebroventricular (ICV) at 1×10¹¹ gc/mouse (n=3) in Abcd1-/y mice.

Intracerebroventricular delivery of rAAV9-CBA-hABCD1-HA transduced neurons in deep gray nuclei.

To determine neuronal transduction of rAAV9-CBA-hABCD1-HA after ICV delivery, we harvested mouse brain tissue 15 days after delivery and performed HA-tag co-staining with neuronal markers NeuN and tyrosine hydroxylase. ICV delivery of rAAV9-CBA-hABCD1-HA led to widespread hABCD1-HA expression across different regions of brain (Figure S8). In particular, neurons in cerebral cortex and hippocampus that were in close proximity to the lateral ventricle demonstrated robust transduction. In addition to neurons, astrocytes were also significantly transduced, as previously reported^{16, 17} and shown in Figure S6. Surprisingly, neurons in basal forebrain were also greatly transduced, with sporadic transduction of neurons in hypothalamus and substantia nigra (Figure 9A). Further, ICV delivery of rAAV9-CBA-hABCD1-HA into CHAT^BAC-eGFP mice showed high transduction of ChAT positive cholinergic neurons in the basal forebrain, in particular the diagonal band nucleus region, which expresses endogenous ABCD1 at high levels (Figure 9B). While delivery did not occur exclusively to neurons, these data suggest that ICV delivery of AAV9 corrects cholinergic neurons in the basal forebrain.

Figure 9. ICV delivery of rAAV9-CBA-hABCD1-HA transduces neurons in deep gray matter nuclei.

A. Representative confocal images showing the expression of hABCD1-HA in neurons (indicated by white arrows and in high magnification) of different brain regions following ICV delivery at 1×10¹¹ gc/mouse in Abcd1-/y mice (n=3). B, Representative confocal images showing the colocalization of hABCD1-HA in CHAT-GFP positive neurons (indicated by white arrows and in high magnification) of different brain regions, particularly diagonal band nucleus (NDB) in basal forebrain following ICV delivery at 1×10¹¹ gc/mouse in CHAT^BAC-eGFP mice(n=3).

Intracerebroventricular delivery of rAAV9-CBA-hABCD1 improved motor performance and histopathology of mice with AMN.

For each delivery route (IV, IT, ICV) we treated a cohort of male Abcd1-/y mice (5–13 months) with rAAV9-CBA-hABCD1 and compared the behavioral correction in a set of measures. While IV and IT delivery did not show significant correction of hind limb reflex extension (Figure 10A-C), ICV injection of 1X10¹¹ gc at both young and old age led to significant improvement. Scores of hind limb reflex extension in untreated Abcd1-/y mice dropped to 1.75 at 18 months of age compared to age-matched wide type (around 3.5), whereas male Abcd1-/y mice treated via ICV at an old age retained scores around 3.25 (p<0.05) (Figure 10D). Abcd1-/y mice treated via ICV at a young age had an average score around 3.5 at 15 and 16 months (Figure 10E), and a replication study confirmed the improvement of hind limb reflex extension scores by ICV injection (Figure 10F). Compared with the other two behavioral motor testing methods, rotarod and beam crossing, hind limb reflex extension more clearly distinguished the behavioral changes between wide type and Abcd1-/y mice at advanced age (Figure 10G, H). Furthermore, the beneficial effect of gene addition was most clearly evident in hind limb extension. These data suggested that efficient transduction of brain by AAV9-hABCD1 could be important for correction of motor behavior in AMN. Furthermore, accumulation of α-synuclein in the basal forebrain and hypothalamus of untreated Abcd1-/y (Figure 3A) was rarely seen in mice after ICV treatment (Figure 10I), suggesting functional correction at a cellular and molecular level.

Figure 10. ICV delivery of rAAV9-CBA-hABCD1 improves hind limb reflex extension score and histopathology in Abcd1-/y mice.

A. Representative images showing hind limb clasping behavior in old Abcd1-/y mice. Changes of hind limb reflex extension score in Abcd1-/y mice following (B) intravenous (IV) delivery at 1×10¹² gc/mouse, Intrathecal (IT) delivery at 1×10¹¹ gc/mouse (C) and intracerebroventricular (ICV) delivery at 1×10¹¹ gc/mouse in both old (D) and young mice (E, F) mice. Mouse motor behavioral assessment by beam walking (G) and Rotarod (H) following ICV delivery of rAAV9-CBA-hABCD1. (I) Representative immunohistochemistry staining of α-synuclein in mouse basal forebrain and hypothalamus following rAAV9-CBA-hABCD1 ICV delivery with PBS as control. Results were expressed as means ± SEM. *p<0.05, ** p<0.01 as compared with wide type (WT) control and #p<0.05, ## p<0.01 as compared to untreated group.

Discussion

X-ALD is an X-linked progressive neurological condition caused by mutations in ABCD1 that affects the entire central nervous system, precipitating inflammatory demyelination in the brain and axonal degeneration in the spinal cord^24–26. While glial cells have long been considered the major contributor to disease progression, this study investigates the role of neurons, a subject of great relevance to the axonopathy of AMN. In postmortem human brains, we detected high ABCD1 expression in certain neurons of the basal forebrain, but not motor cortex. In mice, we found that ABCD1 was expressed at high levels across the CNS in the early postnatal period but by postnatal day 17, only a small group of neurons in the basal forebrain retained high expression.

The basal forebrain contains the basal nucleus of Meynert, a large collection of neurons that serves as the primary source of cholinergic input into the cortical mantle and functionally modulates the neocortex ²⁷. Adjacent to the basal nucleus of Meynert is the diagonal band nucleus. This is the second highest source of cholinergic input to the cortex. We showed that the neurons localizing to the diagonal band express ChAT and retain high ABCD1 expression postnatally. Thus, basal-cortical connectivity may play a key role in regulating motor activity and contribute to motor dysfunction in AMN²⁰.

To decipher the role of ABCD1 in neurons, we performed in vitro differentiation of Sh-sy5y neurons into cholinergic neurons and examined the effect of ABCD1 silencing in these cells. ABCD1 inhibition had a significant impact on neuronal function by reducing expression of postsynaptic protein PSD95, ChAT and class-III β-tubulin (TUJ1). These changes seem to be linked to abnormalities in metabolic signaling pathways and alter expression levels of molecules such as SIRT1 and EZH2, which have been implicated in regulating neuronal function^{22, 23}. Importantly, acetylcholine production was significantly lower in ABCD1 silenced neurons, suggesting that the absence of ABCD1 directly impacts neurotransmitter levels.

AMN is classically thought of as a disease of the spinal cord, affecting the long tracts of the upper motor neurons ²⁸. Hence, we were surprised to find low expression of ABCD1 in motor cortex. We also initially expected that lumbar intrathecal delivery of AAV9-ABCD1 would correct the motor abnormalities of the AMN mouse model^21,26. However, our data suggest that ICV delivery of gene therapy was in fact more efficacious in correcting abnormalities in hind limb clasping. We suspect that the superior effect of ICV delivery was due to transfection of basal forebrain neurons, independent of delivery to long tracts of the upper motor neurons. Similar to others¹⁸, we observed an earlier more profound manifestation of hind limb clasping starting around 15 months compared to rotarod performance abnormalities that are seen later (18–22months), possibly pointing to earlier changes in tone modulation in ABCD1 deficient mice.

The mechanism by which brain abnormalities contribute to the spastic paraparesis of AMN is unknown. Efferents of the basal ganglia influence the primary motor area of the cerebral cortex³⁰. Several movement disorders impact abnormal signaling in the basal ganglia-thalamocortical circuitry. Motor problems in Parkinson and dystonia relate to the basal ganglia cholinergic system^31–33. Prior studies identified the medial globus pallidus as the origin of major basal ganglia projections to multiple regions in the primary motor area^34–36. These findings suggest that basal ganglia outputs influence the generation and modulation of movement ³⁷. As the primary motor cortex projects directly to spinal cord ³⁸, the pallidal projections to cortical motor areas may provide the basal ganglia with a direct route to modulate the motor output of the spinal cord. While the exact mechanism of the murine hind limb extension reflex is not fully understood, here too the basal ganglia are thought to play a role³⁹. Beyond the direct impact upon motor performance in Parkinson’s disease and also seen after focal injury ^{40, 41}, the basal ganglia may have an influence in regulating cortical motor output in human and thus an important role in AMN as reflected by the high endogenous ABCD1 protein expression ⁴².

In our study, the deep gray nuclei in ALD patient brain specimens showed increased glial activation and upregulation of complement. Astrocytes are known to disrupt the vasculature in diseases of the basal ganglia but may also mediate plasticity of neuronal function^{43, 44}. Consistent upregulation of GFAP, an astrocyte activation marker, in the ALD basal ganglia suggests that astrocytes contribute to the neurodegeneration in X-ALD. On the other hand, significantly increased p-tau (ser202/thr205) within the basal ganglia (nucleus accumbens and globus pallidus) points to neuronal dysfunction as well.

As a neuronal microtubule-associated protein, tau plays a key role in regulating microtubule dynamics, axonal transport and neurite outgrowth by site-specific phosphorylation⁴⁵. Phosphorylation at residues ser202/thr205 drives tau aggregation, which is an essential pathological change in many neurological diseases ^{46, 47}. Interestingly, no distinct p-tau increase was observed in the motor cortex, which connects to basal ganglion output circuits, implying a lack of direct neuronal damage. Yet, the marked decrease in presynaptic SYP and postsynaptic PSD95 protein indicates substantial loss of synapses. As no significant upregulation of complement expression was detected, synapse loss cannot be attributed to complement activation alone. Considering that in vitro ABCD1 silencing inhibits synaptic protein expression and in vivo tau pathology directly impacts microtubule dynamics, we postulate that basal ganglia dysfunction impairs the synaptic circuit connecting to the motor cortex, resulting in motor dysfunction in AMN.

In conclusion, ABCD1 appears to be an essential protein for certain groups of neurons in the brain, particularly within the basal forebrain. The pathological changes in mice and humans with AMN highlight the potential importance of the basal forebrain. Among the different routes of AAV9-ABCD1 delivery, ICV delivery has the ability to transduce a greater region of the brain, and in particular the rostral region. Our data suggest that AAV9-ABCD1 ICV delivery corrects both neurons and astrocytes (Figure S9) in the basal forebrain. The correction of these two key cell types may underlie the observed clinical improvement of hind limb clasping in mice. Clinical Improvement following ICV delivery of AAV9-ABCD1 emphasizes the importance of the basal forebrain-cortical circuitry in hind limb motor control. Further, it suggests that therapeutic targeting of the brain may be an important component in addressing AMN pathology.

Supplementary Material

Fig S1

Figure S1. ABCD1 immunostaining (green) in neurons of wild type C57BL/6 mouse brain cortex, substantia nigra and thalamus at postnatal day 2 (P2) and postnatal day 17 (P17). Tissue is co-stained with the neuronal markers for neuronal nuclear protein (NeuN, red) and tyrosine hydroxylase (TH, red).

Fig S2

Figure S2. ABCD1 immunostaining (green) in neurons of wild type C57BL/6 mouse cerebellum and spinal cord and co-stained with the neuronal marker NeuN (red) at postnatal day 2 (P2) and postnatal day 17 (P17).

Fig S3

Figure S3. Co-staining of ChAT (GFP reporter) and ABCD1 in cortex, thalamus and cerebellum at P5 and P16 in CHAT^BAC-eGFP mice.

Fig S4

Figure S4. Confocal images with separate channels showing the neurons highly expressing ABCD1 (indicated by white arrows), colocalized with CHAT-GFP in BF.

Fig S5a

Figure S5. Representative immunohistochemistry staining of A-beta 42 (A) and α-synuclein (B) in nucleus accumbens (NA), medial globus pallidus (GPi), and hypothalamus (HT) of X-ALD patients (n=4) compared to unaffected controls (n=6).

Fig S6

Figure S6. Immunohistochemistry staining of p-tau in Alzheimer’s brain serves as positive control.

Fig S7

Figure S7. Protein (A) and mRNA marker analysis (B) in differentiated Sh-sy5y cell line. Sh-sy5y cells were seeded in either poly-L-polyline (PLL) or matrigel coated plates and then differentiated with 2mM sodium butyrate (SB) or retinoic acid (RA) (1μM and 10μM) respectively for 6 days. Cells were harvested for protein and mRNA analysis of markers for cholinergic neurons (CHAT and VACHT) and dopaminergic neurons (TH and DAT).

Fig S8

Figure S8. Confocal imaging of whole brain showing the distribution of rAAV9-CBA-hABCD1-HA transgene expression 15 days post ICV delivery at 1×10¹¹ gc/mouse. Brain sections were stained with anti-HA antibody and tile scanning was performed.

Fig S9

Figure S9. Representative confocal images showing the expression of hABCD1-HA in astrocytes of different brain regions following ICV delivery of rAAV9-CBA-hABCD1-HA at 1×10¹¹ gc/mouse in Abcd1-/y mice (n=3).

Data Availability:

The data that support the findings of this study are available from the corresponding author, upon reasonable request.