A Novel Mouse Model for Cerebral Inflammatory Demyelination in X-linked Adrenoleukodystrophy: Insights into Pathogenesis and Potential Therapeutic Targets

Abstract

Objective:

X-linked adrenoleukodystrophy (ALD) is caused by mutations in ABCD1, a peroxisomal gene. More than half of males with an ABCD1 mutation develop inflammatory cerebral demyelination (cALD), but underlying mechanisms remain unknown and therapies are limited. We sought to develop and characterize a mouse model of cALD to facilitate study of disease mechanisms and therapy development.

Methods:

We used immunoassays and immunohistochemistry to assess novel (IL-18) and established molecular markers in CSF and post-mortem brain tissue from cALD patients. We generated a cALD phenotype in Abcd1-knockout mice using a two-hit method that combines cuprizone and experimental autoimmune encephalomyelitis models. We then used MRI and immunohistochemistry to assess the fidelity of cALD molecular markers in the mice.

Results:

Human and mouse cALD lesions shared histologic features of myelin phagocytosis, myelin loss, abundant microglial activation, T and B-cell infiltration, and astrogliosis. Compared to wild-type controls, Abcd1-knockout mice displayed more cerebral demyelination, blood-brain barrier disruption, and perivascular immune cell infiltration. This enhanced inflammatory response was associated with higher levels of fibrin deposition, oxidative stress, demyelination, and axonal injury. IL-18 immunoreactivity co-localized with perivascular monocytes/macrophages in both human and mouse brain tissue. In cALD patients, CSF IL-18 levels correlated with MRI lesion severity.

Interpretation:

Our results suggest loss of Abcd1 function in mice predisposes to more severe blood-brain barrier disruption, cerebral inflammation driven by the infiltration of peripheral immune cells, demyelination, and axonal damage, replicating human cALD features. This novel mouse model could shed light on cALD mechanisms and accelerate cALD therapy development.

Graphical Abstract

Much like humans, environmental stressors combined with loss of Abcd1 gene function are sufficient to produce the characteristic features of the cerebral demyelinating phenotype of adrenoleukodystrophy in mice. This offers a preclinical model for cerebral ALD.

Introduction

X-linked adrenoleukodystrophy (ALD) is caused by ABCD1 mutations, impairing peroxisomal very long-chain fatty acid (VLCFA) transport and degradation. Affected males have elevated serum VLCFAs and develop adrenal insufficiency and progressive spastic paraplegia by their fifth decade. The cerebral demyelinating form (cALD) occurs unpredictably, with the highest incidence in the first decade, affecting approximately one-third of boys^1,2. Although the risk of cALD rises again in the 3^rd and 4^th decades, approximately one-third of males with an ABCD1 mutation never manifest cALD³.

cALD is a devastating, rapidly progressive form of inflammatory demyelination⁴. The molecular mechanisms underlying cALD pathogenesis are not well understood, although fibrin deposition, oxidative stress, and NLRP3 activation have been previously implicated^5,6. Current treatments include hematopoietic stem cell transplantation using donor or gene-corrected cells, but this is limited to early-stage cALD cases and can only halt progression, not reverse damage^7,8. Therapy development for cALD has been hindered by a limited understanding of cALD’s pathophysiology and the absence of a preclinical model.

Similar to males with ALD, Abcd1-knockout mice exhibit increased levels of VLCFAs in brain and immune tissue; they also develop signs and symptoms of myelopathy with advanced age⁹. In both humans and mice, Abcd1 deficiency alone is insufficient to induce cALD lesions. In contrast to humans, spontaneous cerebral demyelination has never been reported in Abcd1-knockout mice¹⁰.

Researchers have tried various methods to induce cerebral lesions in mice that replicate human pathology. Pujol et al. (2004) developed transgenic mice with a double knockout of Abcd1 and Abcd2 genes. Although these mice develop symptoms of motor axonopathy several months earlier than Abcd1-knockout mice, they did not develop the inflammatory cerebral demyelination¹¹. Kettwig et al. (2021) found that Abcd1-knockout mice induced with MOG-peptide experimental autoimmune encephalomyelitis (EAE) exhibited greater clinical disability than similarly treated wild-type mice; however, the study did not report histologic findings¹². Recently, Martinovic et al. observed that Abcd1-knockout mice fed a cuprizone (CPZ) diet, which induces oxidative stress and transient demyelination, exhibited an accelerated reduction of mature oligodendrocytes and increased axonal damage after 3 weeks of CPZ diet, without evidence of motor disability. However, after 5 weeks on the CPZ diet, both Abcd1-knockout and wild-type mice exhibited similar repair of axonal damage¹³.

We hypothesized that CPZ diet, which enhances oxidative stress, combined with EAE, which induces CNS autoimmunity, in Abcd1-knockout mice would induce a sufficiently robust model for preclinical study of cALD. To assess the fidelity of our murine model to human cALD, we compared the histological and molecular findings in the cALD mouse model to human cALD tissue samples as a gold standard, and to similarly treated wild-type mice as a genotype control.

Methods

Post-mortem human tissue

Formalin-fixed brain tissue from three healthy pediatric controls and two pediatric patients with cerebral adrenoleukodystrophy (<21 years old) were provided by the NIH NeuroBioBank at the University of Maryland, Baltimore, MD. Gross tissue examination was used to identify suspected demyelinating cALD lesions; these were validated using histological criteria^14,15. For each cALD patient, we procured three distinct sites with active demyelination. We used representative samples from age and gender-matched healthy individual as a control. All specimens were archived by consent at autopsy.

Cytokine assay and analysis

Cerebrospinal fluid (CSF) samples were collected from 20 cALD patients prior to hematopoietic stem cell transplantation at the University of Minnesota. Control CSF was collected from eight pediatric patients without clinical or cytologic evidence of inflammation undergoing lumbar puncture for clinical purposes. All CSF samples were obtained following provision of informed consent in accordance with the Declaration of Helsinki and were approved by the institutional review boards at the University of Minnesota and Stanford University. Interleukin-18 (IL-18) and interleukin-1β (IL-1β) levels were measured using ProcartaPlex Simplex bead sets (eBioscience) and a Bio-Plex instrument (Bio-Rad, Hercules, CA, USA). IL-18 and IL-1β concentrations were calculated for cALD and control groups, and significance was determined using a two-tailed Mann-Whitney U test. The correlation between IL-18 levels and Loes MRI severity score was evaluated using Spearman’s coefficient in Prism version 10.2.

Experimental animals

Abcd1 heterozygote knockout female mice (Jax lab, B6.129-Abcd1tm1Kan/J, Stock No. 003716) were crossed with wild-type C57BL/6 male mice to generate both female and male knockout mice, though only male mice were used in studies. In these mice, exon 1 of the Abcd1 gene is disrupted, which eliminates the translation code¹⁶. Pure wild-type mice (Jax lab, C57BL/6) were utilized as controls. Mice were housed in groups of three to five with a 12-hour light/dark cycle at a controlled temperature with food and water provided ad libitum. The Administrative Panel on Laboratory Animal Care (APLAC) at Stanford University approved all animal protocols (#32955; #34033) described herein.

Cuprizone intoxication and EAE induction

Eight-week-old Abcd1-knockout male (Abcd1^y/-) and wild-type male mice were randomly assigned to one of four arms: no treatment (arm 1), cuprizone (CPZ) treatment alone (arm 2), EAE treatment alone (arm 3), and combination of CPZ/EAE treatment (arm 4). Tables S1 and S2 detail the distribution of mice across each experimental arm and assay.

At eight weeks of age, Abcd1^y/- and wild-type male mice in the CPZ/EAE treatment group were fed a 0.2% CPZ diet for two weeks, followed by EAE-MOG_35-55 immunization. Mice in the CPZ-only group initiated the 0.2% CPZ diet at ten weeks of age to align with the CPZ/EAE group’s age for tissue collection and histological analysis, thereby minimizing the influence of age on our data. This two-week duration of CPZ treatment was chosen for its significant immune cell infiltration and loss of mature oligodendrocytes in the brain compared to 4-6 week periods¹⁷ ( Fig S1).

At ten weeks of age, EAE and CPZ/EAE mice underwent EAE induction as previously described¹⁸. Briefly, mice were immunized with an emulsion of 200 μg of Complete Freund’s Adjuvant (CFA, Difco), 100 μg of myelin oligodendrocyte glycoprotein peptide35–55 (MOG_35–55, Genemed Synthesis, Inc), and 200 ng of Bordetella pertussis toxin (lot# 181236A1, List Biological Laboratories)^18,19. Pertussis was injected twice: on the day of MOG injection and again on the second day post MOG injection. The MOG_35–55 emulsion was subcutaneously injected into both flanks, near the axillary lymph nodes.

EAE motor disability scoring and tissue collection

For mice in EAE alone and CPZ/EAE treatment groups, EAE motor disability symptoms were scored following MOG immunization as described previously¹⁸. Mice treated with CPZ alone do not exhibit motor disability. We collected tissue samples when the mice reached disability scores of 2 or 3, typically between days 15-22 post-MOG immunization (at 12-13 weeks of age). If the mice displayed no clinical sign of EAE or had a score below 2 by day 22 (age 13 weeks), we collected the samples on that day. Therefore, the motor disability scores are presented until day 15 to include all mice in the clinical scoring analysis. No mice were excluded from the analysis.

Mouse in vivo MRI acquisition and processing

We obtained in vivo MRI images utilizing an actively shielded Bruker horizontal bore scanner (Bruker Corp, Billerica MA) with 11.7 T field strength and International Electric Co. (IECO) 750/400 gradient drivers at the Stanford Center for Innovation in In vivo Imaging (SCi3). For each mouse, we obtained T1 and T2-weighted images. Gadolinium (Gadavist, 100mg/kg) was administered via intraperitoneal injection 30 minutes prior to imaging²⁰. OsiriX software was used to quantify signal intensity and myelin loss in the medial corpus callosum (MCC) in the rostral diencephalon.

Immunofluorescence and immunohistochemistry

The collection of tissue, luxol fast blue (LFB) staining, and immunofluorescence were performed based on the procedure detailed in our previous study¹⁷. The Table S3 contains the details of the antibodies including their concentrations and manufacturing information.

Quantification procedures

All experiments were assessed by at least two trained readers blinded to mouse identity and treatment arm. CD68⁺ and Olig2⁺ cell counts were conducted in 0.1 mm² region of MCC and normalized to DAPI using ImageJ. The intensity of myelin (MBP staining) in MCC was measured using ImageJ. Final counts/scores were averaged from 2-3 coronal tissue sections per mouse. Analysis was performed between the Bregma -2.355 and -1.255 mm in mice.

Two blinded readers identified and classified each perivascular cuff (PVC) in 2 to 3 brain sections per mouse. We assigned each PVC a score based on lesion severity. We used DAPI staining to analyze the degree of perivascular immune cell infiltration and its severity was graded on the scale of I to IV as follows: Grade I: leukocyte foci with an aggregation of inflammatory cells; Grade II: accumulation of immune cells around the vessels without parenchymal penetration; Grade III: penetration of immune cells into the parenchyma, and Grade IV: prominent immune cell penetration and formation of a large lesion (Fig S2A). Scores for all lesions in 2-3 coronal sections were summed, scoring one point for each grade 1 PVC, two points for each grade 2 PVC, etc. Average lesion severity score was calculated per coronal section to provide a single PVC severity score for each mouse.

This same analysis was applied to determine CD68⁺ monocyte/macrophage infiltration (Fig S2B). Perivascular infiltration of T and B cells was graded similarly, with one exception that Grade I represented fewer cells attached to endothelial cells in the vessels (Fig S2C).

The grading of astrogliosis was determined by GFAP staining, which highlighted the cell body and arborization of astrocytes (Fig S2D). Grade I represented minimal or no astrogliosis, while grade IV represented severe lesions.

Severity of demyelination in MCC was based on the qualitative assessment of intensity of LFB staining (Fig S3A).

We used ImageJ to quantify total area of staining for Fibrin(ogen), Gp91-phox, and SMI-32 immunoreactive areas as previously described^21,22.

Statistical analyses

Data were analyzed with GraphPad Prism 10. Significant differences between two groups were determined using unpaired t-tests with Welch’s correction. For multiple group comparisons between wild-type and Abcd1-knockout groups, we used a two-way ANOVA with Sidak post-hoc analysis. One-way ANOVA was used for multiple comparisons between the groups within each genotype. The normality of distribution was determined using the Shapiro-Wilk test. A p-value of <0.05 was considered significant. Sample size estimation using G*Power 3.1 recommended 5 mice per group for 90% statistical power, based on effect size estimated from previous MBP staining data¹⁷.

Results

cALD donors had advanced cerebral disease at time of diagnosis

Figure 1 highlights key radiologic and histologic features in two males affected by cALD. cALD donor 1 was 7 years old at the time of autopsy. Brain MRI, obtained 10 months prior to death, demonstrates extensive diffuse contiguous symmetric T2 hyperintense lesions involving frontotemporal, central, and anterosuperior occipitoparietal white matter. These lesions exhibited gadolinium contrast enhancement and received a Loes MRI severity score of 16 (Fig 1A). His clinical course was complicated by primary adrenal insufficiency, seizures, g-tube dependence and loss of consistent bowel and bladder though he continued to ambulate independently.

FIGURE 1. In boys with cALD, brain lesions exhibit important radiologic and histologic hallmarks.

(A, E). Axial MR imaging from two affected boys with cALD highlight areas of demyelination (T2 FLAIR intensities) and vascular breakdown (contrast enhancement). Red boxes indicate representative areas sampled for histopathological images. (B-K) LFB + H&E histology of postmortem tissue from cALD patients with varying severity of demyelination in lesions. Histology highlights regions of extensive, active demyelination (C, G) and early gliosis (D, H) compared to healthy, unaffected control (I-K). B, F, I: scale bar: 400 μm. C, D, G, H, J, K: scale bar: 100 μm (L) Immunohistochemistry of representative demyelinated brain lesion shows infiltration of microglia/ macrophages (CD163), and to a lesser extent T-cells (CD3) within the lesion. B cells (CD20) are present in lower abundance. Top row scale bar: 400μm; black boxes indicate corresponding areas shown at higher magnification. Remaining images: scale bar: 50 μm.

cALD donor 2 was 10 years old at the time of autopsy. Prior to his diagnosis, he had experienced progressive behavioral changes and vision processing disorder. Brain MRI, obtained 28 months prior to death, demonstrated extensive T2 hyperintense, posterior predominant, confluent lesions including involvement of splenium with contrast enhancement, and received a Loes score of 13 (Fig 1E). His clinical course was complicated by adrenal insufficiency.

Both donors had confirmed mutations in ABCD1 and elevated plasma VLCFA levels. Given their extensive brain lesions at diagnosis, neither donor was eligible for transplant.

Perivascular immune cell infiltration in human cALD

cALD brain lesions radiate outward creating distinct concentric histopathological regions with central area of gliosis (Fig 1D, H), surrounded by concentric areas of progressive demyelination (Fig 1C, G). We confirmed that the human brain tissue sections had perilesional, active demyelination, and gliosis regions based on previously defined criteria^14,15. Histology and LFB staining identified active demyelination areas and extensive inflammatory infiltration of monocyte/macrophage and scattered perivascular T cells (Fig 1L). Sections also contained regions consistent with perilesional white matter characterized by pale LFB staining and scattered macrophage/microglia immunoreactivity (Fig 1B, D, F, H).

We noted perivascular infiltration consisting predominantly of cells of monocyte/macrophage lineage (Fig 1L). Furthermore, there was a consistent presence of CD3⁺ T-cells in the perivascular region (Fig 1L). In agreement with prior literature²³, perivascular CD20⁺ B-cell infiltration was inconsistent and present to a lesser degree compared to T-cells, and monocyte/macrophage (Fig 1L).

CSF and brain tissue from boys with cALD show evidence of NLRP3 inflammasome activation

CSF IL-18 was significantly elevated in cALD (median [interquartile range]: 12.89 [8.85-23.45]) compared to controls (5.68 [5.56-6.28], p = 0.0001; Fig 2A). CSF IL-1β was not significantly different between cALD patients and controls (Fig 2B). CSF IL-1β levels were lower than IL-18, possibly due to IL-1β’s short half-life or degradation by matrix metalloproteinases, which are known to be increased in CSF of ALD patients^24,25. In cALD, increased CSF IL-18 positively correlated to higher Loes MRI severity scores, with higher scores indicative of more severe cALD brain lesions (Spearman coefficient (r) = 0.82 with p < 0.001 (Fig 2C)). This provides initial evidence that IL-18 may correlate with neuroinflammation in cALD. Table S4 details patient ages, Loes scores, IL-18 and IL-1β concentrations in CSF.

FIGURE 2. Interleukin-18, an NLRP3 inflammasome effector, is increased in spinal fluid and brain tissue of boys with cALD.

(A) IL-18, but not IL-1ß, is significantly elevated in the CSF of cALD boys (n=20) compared to unaffected, age-matched controls (n=8) with p= 0.0001. Significance was determined using a two-tailed Mann-Whitney U test (B) In boys with cALD, increased IL-18 levels in the CSF correlate with higher Loes MRI severity scores. Spearman co-efficient r = 0.818, p<0.001. Solid line: mean; dashed lines: 95% confidence interval. (C) Within areas of early brain demyelination in cALD (green arrowheads), the distribution of IL-18⁺ cells align with microglia (Iba1) rather than astrocytes (GFAP). Top Row: Scale bar represents 400 μm; black boxes indicate corresponding areas shown at higher magnification. Remaining images: scale bar 100 μm. (D) Immunofluorescence highlights co-localization of IL-18 with perivascular microglia/macrophages (Iba1⁺) and, to a lesser extent, intraparenchymal astrocytes (GFAP). Insets highlight co-localization at 60X magnification. Scale bar: 100μm.

Within cALD lesions, IL-18 immunoreactive cells resembled the morphology and distribution of microglia/macrophages, and astrocytes (Fig 2D, E). In perivascular regions, IL-18 co-localized with Iba1⁺ monocyte/macrophages (Fig 2E).

Combined CPZ/EAE induces a robust neurologic phenotype in Abcd1^y/- mice

We hypothesized that the combination of Abcd1 deficiency with both CPZ and EAE, as opposed to EAE or CPZ alone, would yield clinical, radiologic, and histologic features similar to human cALD (experimental design in Fig 3A).

FIGURE 3. In Abcd1y/- mice, exposure to a combination of CPZ and EAE induced greater motor disability compared to similarly treated wild-type controls or treatment with EAE or CPZ alone.

(A) A graphical overview summarizes the treatment regimens and timelines for the four experimental groups applied to both Abcd1^y/- and wild-type mice. Mice aged 8 weeks were given a 0.2% CPZ diet for 2 weeks, followed by EAE induction through MOG_35-55 injection. For naive and CPZ-treated mice (Arms 1 and 2), tissue was collected at 12 weeks of age. For EAE and CPZ/EAE groups (Arms 3 and 4), tissue was collected when mice reached a disability score of 2-3 between 15 and 22 days post-MOG_35-55 injection, when the mice were 12-13 weeks old. (B) Clinical motor disability scores in the CPZ/EAE group (arm 4) (n=12-13 mice/genotype), EAE induction alone (arm 3) (n=6-7 mice/genotype) and CPZ alone treatment (arm 2) (n=7-12 mice/genotype). (C) Motor disability scores in Abcd1^y/- mice with the CPZ/EAE induction were higher than those in wild-type mice with the CPZ/EAE induction at days 12 and 13 post-MOG injection (n=12-13 mice/genotype). (D) Abcd1^y/- mice induced with CPZ/EAE had significantly higher disability than mice treated with EAE alone at days 11-13 post MOG_35-55 injection. In graphs B, C, and D, each dot represents the mean motor disability score ± SEM. Significant differences between the two groups were determined using an unpaired t-test with Welch’s correction. (E) The graph illustrates the area under the curve of motor disability, presented as mean ± SEM. This measure was elevated in Abcd1^y/- mice compared to wild-type mice following CPZ/EAE treatment. However, a two-way ANOVA followed by Sidak’s post-hoc test did not indicate a significant difference between groups. CPZ: cuprizone, EAE: experimental autoimmune encephalomyelitis, KO: Abcd1^y/-, WT; wild-type, ROI: region of interest.

Abcd1^y/- mice with CPZ/EAE treatment (cALD mice) exhibited a more severe motor disability compared to wild-type mice on days 12 and 13 post-MOG immunization (p = 0.04, and p = 0.03, respectively) (Fig 3B, C). cALD mice also exhibited a more severe motor disability compared to EAE alone on days 11-13 post-MOG immunization (p ≤ 0.03) (Fig 3B, D). Mice provided CPZ diet alone did not exhibit motor disability, regardless of genotype (Fig 3B, D). We used unpaired t-tests with Welch’s correction to compare motor disability scores between two groups (Fig 3C, D).

Despite a trend toward higher motor disability in cALD mice, two-way ANOVA showed no significant difference in the area under the curve of mototr disability(Fig 3E).

In EAE alone, 7 out of 7 wild-type and 3 out of 6 Abcd1^y/- mice developed motor disability. However, in CPZ/EAE, 11 out of 13 wild-type mice and 12 out of 13 Abcd1^y/- mice developed motor disability. Both genotypes in CPZ/EAE group showed onset of motor disability around day 10 post-MOG immunization.

The use of either EAE or CPZ alone also yielded similar motor disability levels across genotypes (Figs 3B and S3A). Wild-type mice in CPZ/EAE or EAE alone had statistically similar levels of motor disability (Figs 3B and S3B).

Motor disability was highest in cALD mice (Fig 3B–E), though statistical significance varied depending on comparison groups. The combination of CPZ/EAE in Abcd1^y/- mice induced a more rapid onset and higher severity of motor disability than EAE (Figs 3B, D and S3C).

CPZ/EAE in Abcd1^y/- mice induces inflammatory cerebral demyelination

We examined whether the cALD mouse model developed inflammatory cerebral demyelination, a hallmark of human cALD pathology. We focused on characterizing demyelination in the MCC, given that human cALD brain lesions typically originate in the MCC at the splenium or genu^26,27. cALD mice showed significantly higher T2 hyperintensity in the MCC when compared to both naïve Abcd1^y/- mice and wild-type CPZ/EAE mice (p < 0.001 and p = 0.007, respectively); (Fig 4A, B).

FIGURE 4. Abcd1y/- mice induced with CPZ/EAE exhibit radiological and histological features of demyelination, blood-brain barrier disruption, and axonal damage.

(A) Representative coronal MRI images demonstrate a high T2 signal in the MCC (red box) of Abcd1^y/- mice treated with CPZ/EAE, resembling the characteristic pattern seen in human cALD patients (far right). (B) Quantitative analysis of T2-weighted MRI (T2) of the MCC shows a higher T2 signal in the MCC (highlighted by red box) of Abcd1^y/- mice treated with CPZ/EAE (n=4-5 mice/group) compared to controls, similar to early lesions in human cALD. (C) Representative histological sections of MBP staining in the MCC depict differing levels of demyelination across experimental arms. The red box denotes the representative area for quantitative analysis. Scale bars: 50 μm. (D) Quantitative analysis of MBP intensity in the MCC shows higher levels of demyelination in Abcd1^y/- mice treated with CPZ/EAE compared to Abcd1^y/- naïve mice with p value <0.001 (n=4-6 mice/group) (E) Representative images of post-gadolinium T1-weighted MRI (T1) show focal areas of gadolinium enhancement (yellow arrows) in Abcd1^y/- mice treated with CPZ/EAE, similar to characteristic findings in human cALD (n=4-5 mice/group). (F, G) Representative images and quantitative measures of immunostaining for fibrin(ogen), a marker of BBB disruption, show higher levels in Abcd1^y/- mice treated with CPZ/EAE compared to wild-type mice. (H, I) Similarly, representative and quantitative analysis of immunostaining for gp91-phox, a marker of oxidative stress, are highest in Abcd1^y/- mice treated with CPZ/EAE (n=3-7 mice/group). Scale bars, 200 μm. (J, K) Representative images and quantitative measures of immunostaining for SMI-32, a marker of axonal damage, show higher levels in Abcd1^y/- mice treated with CPZ/EAE compared to wild-type mice (n=4-6 mice/group) Scale bars; 100 μm. In the graphs, each circle symbol represents an individual mouse. Gray circle symbols in the graphs denote the corresponding images. Data are presented as mean ± SD. Two-way ANOVA and post-hoc Sidak test was used to determine significance between groups. MCC: medial corpus callosum; MBP: myelin basic protein; CPZ: cuprizone, EAE: experimental autoimmune encephalomyelitis, KO: Abcd1^y/-, WT; Wild-type.

cALD mice showed significant demyelination in the MCC compared to Abcd1^y/- naïve mice (p < 0.001), indicated by MBP staining (Fig 4C, D). There was nominally greater demyelination in cALD mice compared to wild-type CPZ/EAE mice, but it was not statistically significant. LFB staining further confirmed myelin loss in Abcd1^y/- mice compared to wild-type mice in the CPZ/EAE treatment group ( Fig S4A, B). Due to the extensive demyelination, we quantified oligodendrocytes in the MCC and observed a significant decrease in Olig2⁺ cells within MCC in mice with CPZ/EAE treatment compared to naïve mice (p = 0.02) (Fig S5A, B).

CPZ/EAE in Abcd1^y/- mice induces BBB disruption, oxidative stress, and axonal damage

To assess BBB integrity, we used T1 weighted MRI with contrast at 5 weeks after initiating CPZ/EAE treatment. We observed contrast enhancement at the MCC lesion site, resembling human cALD (Fig 4E). Previous studies showed reduced claudin-5, a tight junction protein, in cerebral endothelial cells of cALD patients⁶. Qualitatively, Abcd1^y/- mice showed variable expression of Claudin-5 in CPZ/EAE treatment compare to naïve mice ( Fig S6A–D).

BBB disruption was confirmed by quantifying brain parenchymal fibrin/fibrinogen deposition. In the CPZ/EAE group, fibrin(ogen) deposition was detected in white and gray matter, with highest intensity in the hippocampal region. cALD mice showed significantly increased fibrin(ogen) deposition compared to wild-type CPZ/EAE mice (p = 0.003) (Fig 4F, G). Fibrin(ogen) activates NADPH oxidase and oxidative stress in microglia and macrophages and inhibition of fibrin-induced inflammation protects from axonal damage in EAE^21,22,28. We observed upregulated gp91-phox levels, a marker of oxidative stress, in the cALD mice compared to wild-type mice induced with CPZ/EAE (p = 0.002) (Fig 4H, I).

We assessed axonal damage using SMI-32 antibody, which targets the non-phosphorylated epitope of neurofilament heavy proteins. cALD mice showed significantly higher axonal damage compared to wild-type CPZ/EAE mice (p = 0.009) (Fig 4J, K). Table S2 provides mean ± SD for each group in MRI and immunohistology tests.

cALD mice exhibit perivascular infiltration of monocytes/macrophages and lymphocytes

Perivascular cuffs (PVCs), consisting of immune cell infiltration in perivascular areas, occur following viral or immune-mediated inflammatory processes^29,30. Fig 5A illustrates PVC spatial distribution in representative mice from all four experimental arms.

FIGURE 5. Increased perivascular infiltration of immune cells in Abcd1y/- mice induced with CPZ/EAE.

(A) The schematics illustrate the spatial distribution of perivascular cuffs in a representative mouse, as determined by DAPI staining, across experimental arms. Grading criteria based on infiltrating immune cell penetration and lesion size: Grade I: leukocyte foci with aggregation of inflammatory cells; Grade II: accumulation of immune cells around vessels without parenchymal penetration; Grade III: penetration of immune cells into the parenchyma; Grade IV: prominent immune cell penetration and formation of a large lesion. (B-C) Representative histological images depict perivascular infiltration of (B) CD68+ monocytes/macrophages in the cortex, (C) CD3⁺ T cells and CD20⁺ B cells at the lateral callosal-striatal border in the Abcd1^y/- mice induced with CPZ/EAE. Scale bars: 50 μm. (D-G) Quantitative assessment reveals severity of (D) total immune cell infiltration, (E) monocyte/macrophage infiltration, (F) T cell infiltration, and (G) B cell infiltration in the perivascular region in Abcd1^y/- and wild-type mice with CPZ alone, EAE alone, and combined CPZ/EAE (n=4-7 mice/group). In the graphs, each circle symbol represents an individual mouse. Gray circle symbols in the graphs denote the corresponding images. Data are presented as mean ± SD. Significance between the groups was determined by two-way ANOVA with Sidak’s post-hoc test. CPZ: cuprizone, EAE: experimental autoimmune encephalomyelitis.

Abcd1^y/- and wild-type mice in CPZ or EAE alone groups had few and low grade PVCs (Fig 5A). The cALD mouse exhibited significantly higher PVC severity compared to wild-type CPZ/EAE mice (p < 0.001) (Fig 5D). These findings replicate the perivascular infiltration seen in cALD human pathology (Figs 1L and 2D).

Using flow cytometry, we analyzed the peripheral immune cells (CD45^hi) relative to total immune cells (CD45⁺) within the brain as an indicator of peripheral immune cell infiltration (Fig S1). We observed higher levels of CD45^hi within the brain at 2 weeks of CPZ diet compared to 4 and 6 weeks. At 2 weeks of CPZ diet, approximately 16.4 ± 6.2% (mean ± SD) of total brain immune cells were of peripheral origin (CD45^hi) (Fig S1A). The majority of these infiltrating cells were CD11b⁺ myeloid cells, comprising 10.3 ± 3.5% (mean ± SD) of total immune cells (Fig S1B). A smaller fraction of infiltrating immune cells were T cells (CD3⁺), accounting for 1.3 ± 0.46% (mean ± SD) of total immune cells in brain (Fig S1C).

We further analyzed the cellular composition of perivascular infiltration by evaluating monocyte/macrophage, T-cells, and B-cells. cALD mice exhibited significantly more severe monocyte/macrophage perivascular infiltration compared to wild-type mice in the CPZ/EAE group (p < 0.001) (Fig 5B, E). T-cell (Fig 5C, F) and B-cell (Fig 5C, G) perivascular infiltration was also more severe in the cALD mouse model compared to wild-type CPZ/EAE mice (p < 0.001). BBB damage and infiltration of immune cells were present in both white and gray matter, including cortex (Fig 5B) and lateral callosal-striatal border (Fig 5C). Although wild-type CPZ/EAE mice had fewer PVCs overall, we were still able to identify select areas with T-cell and B-cell infiltration ( Fig S7).

cALD mice exhibit microgliosis and astrogliosis

CD68⁺ activated macrophages/microglia were increased in the MCC of both Abcd1^y/- and wild-type mice after CPZ treatment alone. Combined CPZ/EAE treatment significantly increased CD68⁺ cells compared to EAE alone in both genotypes (p ≤ 0.001), without significant differences between genotypes (Fig 6A, C).

FIGURE 6. Comparable microgliosis and astrogliosis in the medial corpus callosum of Abcd1y/- and wild-type mice induced with CPZ/EAE.

Representative histological sections show (A) CD68 and (B) GFAP staining across the four experimental arms (n=3-7 mice/group). The white boxes highlight the MCC. Quantitative analysis of the MCC shows elevated (C) CD68⁺ cell count and (D) GFAP grading in the CPZ and CPZ/EAE experimental arms (n=3-7 mice/group). In the graphs, each circle symbol represents an individual mouse. Gray circle symbols in the graphs denote the corresponding images. Data are presented as mean ± SD. Statistics calculated via two-way ANOVA and Sidak post-hoc tests. Scale bar: 200μm. CPZ: Cuprizone, EAE: experimental autoimmune encephalomyelitis, MCC: medial corpus callosum, KO: Abcd1^y/-, WT; wild-type.

GFAP staining showed comparable astrogliosis between naïve Abcd1^y/- and wild-type mice. In the CPZ/EAE group, Abcd1^y/- mice showed higher, though not statistically significant, astrogliosis compared to wild-type (Fig 6B, D). Higher magnification images of CD68 and GFAP staining in MCC are provided in Fig S8A, B.

Demyelination coincides with perivascular infiltration and myelin phagocytosis in cALD mice

In human cALD, affected white matter is characterized by perivascular inflammatory cells and demyelination. Our cALD mouse model revealed similar demyelination in regions with accumulation of CD68⁺ cells around the vessels (Fig 7A, B). Perivascular monocyte/macrophage and parenchymal macrophage/microglia show evidence of intracellular myelin, indicating myelin phagocytosis (Fig 7B, C).

FIGURE 7. Perivascular monocyte/macrophage infiltration, associated myelin phagocytosis, and IL-18 expression in cALD mouse model.

The co-staining of CD68 (monocyte/macrophage lineage) and PLP (myelin) indicates monocyte/macrophage infiltration and myelin phagocytosis. The images represent (A) grade II and (B) grade IV of perivascular monocyte/macrophage infiltration. (C) The images display myelin phagocytosis by macrophages/microglia. The upper right image depicts a higher magnification of myelin phagocytosis at the site in the white box of the image. Arrows indicate colocalization of CD68 and PLP. (D) IL-18 expression in perivascular monocytes/macrophages. Arrows indicate colocalization. The upper right image depicts a higher magnification of IL-18 and CD68 colocalization. The scale bar represents 50 μm

cALD mice exhibit IL-18 expression in perivascular monocytes and macrophages

Our findings of human postmortem cALD brain tissue indicates increased IL-18 immunoreactivity in perivascular monocytes/macrophages. The cALD mouse model also demonstrates IL-18 immunoreactivity in perivascular monocytes/macrophages (Fig 7D). IL-18 colocalized with CD68 in the lateral corpus callosum of Abcd1^y/- and wild-type mice in CPZ/EAE model, but is higher in Abcd1^y/- mice (Fig S9).

Among wild-type and Abcd1-knockout mice receiving CPZ-only, we observed IL-18 expression predominantly within the hippocampus and, in contrast with patterns in CPZ/EAE, IL-18 staining was observed primarily in astrocytes, and rarely in CD68⁺ microglia/macrophages, (Fig S10).

Cerebral demyelination persisted for at least 10 weeks in cALD mice

In cALD mice, demyelination in the MCC persisted for up to 10 weeks after initiating CPZ/EAE induction, as confirmed by T2-weighted MRI compared to naïve mice (p = 0.01) (Fig S11A). Sustained microgliosis in the MCC was evident 10 weeks after CPZ/EAE initiation. However, our study may have been underpowered to detect a significant difference in gliosis between naïve and CPZ/EAE conditions in Abcd1^y/- mice (p = 0.09) (Fig S11B, C).

DISCUSSION

More than half of males with a mutation in ABCD1 will develop inflammatory cerebral demyelination, which can be fatal if untreated. The absence of spontaneous cerebral demyelination in Abcd1-knockout mice has hampered investigations into both the mechanism and treatment of cALD for decades. Here, we introduce a two-hit method combining an oxidative stressor (CPZ) and an immune trigger (MOG-EAE) that induces a robust cerebral demyelinating phenotype in male Abcd1-knockout mice recapitulating key radiologic, histologic, and molecular features of cALD, including BBB disruption, fibrin deposition, oxidative stress, IL-18 expression, demyelination, and axonal damage (Table 1). Collectively, these findings suggest that our mouse model offers new avenues for investigating both disease mechanisms and therapy development for cALD.

Table 1:

Histological features of brain lesions in Abcd1^y/- mice in four groups: no induction, CPZ, EAE, and combined CPZ/EAE induction. Mice subjected to the CPZ-EAE combination displayed histological features that resemble those seen in brain lesions of humans with ABCD1 mutations.

Classical features of brain lesions in humans with ABCD1 mutations	Comparative features observed in mice with Abcd1y/- genotype
	No induction	EAE alone	CPZ alone	CPZ + EAE
Elevated VLCFAs in brain and immune cells	+	+	+	+
Cerebral demyelination of corpus callosum	-	-	+	+
Microglial/macrophage activation & myelin phagocytosis in cerebral white matter	-	-	+	+
Astrogliosis	-	-	+	+
Oxidative stress	-	+	+	+
Elevated IL-18	-	-	+	+
Blood-brain barrier disruption	-	-	-	+
Perivascular cuffing	-	-	-	+
Vascular infiltration of monocyte/macrophage into brain parenchyma	-	-	-	+
Vascular infiltration of T-cell into brain parenchyma	-	-	-	+
Vascular infiltration of B-cell into brain parenchyma	-	-	-	+

EAE: Experimental Autoimmune Encephalitis induced with MOG_35–55 and pertussis; CPZ: Cuprizone diet.

Since both oxidative stress and inflammation have previously been proposed as potential triggers in cALD^31–33, we combined CPZ and MOG-EAE in Abcd1^y/-knockout mice to induce a cALD phenotype. This approach complements recent efforts combining CPZ and EAE in wild-type mice to study inflammatory demyelination³⁴.

Traditional EAE protocols, using CNS-derived myelin peptides, primarily affect the spinal cord with minimal cerebral involvement³⁵. Kettwig et al. reported significant motor disabilities in Abcd1^y/- mice versus wild-type mice in the EAE model, immunizing mice at 20-35 weeks with MOG_35-55. Our study used younger mice (10 weeks old) and collected tissue upon reaching a score of 2-3, 15-22 days post-immunization, rather than observing for 40 days. Their protocol used less MOG_35-55 (20 μg) but more mycobacterium (1.25 mg) per mouse. These variations in age, observation period, and mycobacterium dose may contribute to differing disability results. While Kettwig et al. did not report on CNS histology, our data in Abcd1^y/- mice induced with EAE alone suggest these mice exhibit minimal cerebral pathology, as summarized in Table 1.

Dietary CPZ induces demyelination, via oligodendrocyte dysfunction, and innate immune activation, via microglia and astrocytes, but lacks infiltration of adaptive immune cells^36,37. Remyelination typically begins around week 5 of a 6-week CPZ diet¹⁷ . In contrast, our model demonstrates persistent cerebral demyelination for at least 8 weeks after returning to standard feed.

In Martinovic et al.’s study, there were no significant differences in microgliosis, astrogliosis, demyelination, or oligodendrocyte number in the MCC of Abcd1-knockout and wild-type mice after 3 and 5 weeks on 0.3% CPZ diet¹³. Our findings in the CPZ/EAE model are consistent with this, although we were able to identify the difference in BBB disruption and demyelination through MRI, immunohistochemistry, and LFB staining. Additionally, we observed greater inflammation, likely due to more infiltration of immune cells within the brain. Martinovic et al also demonstrated a significant reduction in mature oligodendrocytes and an increase in axonal damage in the MCC of Abcd1-knockout mice compared to wild-type mice within 3 weeks of starting a CPZ diet. However, these differences moderated after 5 weeks on a CPZ diet and following a return to a normal diet. Although we did not examine the impact of Abcd1 dysfunction on oligodendrocyte maturation, our findings indicated that Abcd1-knockout mice exhibited more axonal damage compared to wild-type mice following treatment with CPZ/EAE. This finding of more severe axonal damage in cALD mouse is similar to the high levels of axonal damage reported in human cALD patients³⁸.

We designed our experiments to investigate the brain histology in the early stage of the disease when the motor disability reached the EAE scores of 2 and 3 on days 15-22 post MOG_35-55 immunization. Among mice evaluated up to 10 weeks after initiation of CPZ/EAE treatment, we observed persistent evidence of inflammatory demyelination (Fig S11). This should be sufficient duration of disease activity to allow for preclinical screening of potential abortive therapies for cALD. Further studies are required to observe lesion evolution at later stages.

Both cALD and MS lesions in humans preferentially affect the medial corpus callosum³⁹, possibly due to high myelin density, low vascularity, or metabolic vulnerabilities^6,40,41. Lesions in both Abcd1^y/- and wild-type CPZ/EAE treated mice show similar predilection, suggesting possible overlapping mechanisms in cALD and MS mouse models³⁶.

The mechanisms linking ABCD1-deficiency and VLCFA accumulation to severe inflammatory demyelination in cALD are unclear. Inflammasomes, such as the NLRP3 inflammasome, mediate innate immune responses in various neurodegenerative and metabolic diseases⁴². NLRP3 inflammasome activation leads to IL-1β and IL-18 production and pyroptosis. The NLRP3 inflammasome can be activated by metabolic disruptions, including oxidative stress, mitochondrial and lysosomal dysfunction, and altered lipid metabolism^43,44. We hypothesized that NLRP3 inflammasome activation might drive the neuroinflammatory cascade in cALD.

Elevated NLRP3 inflammasome effectors in human cALD and our mouse model support our hypothesis. Furthermore, CSF IL-18 positively correlates with cALD severity (Loes scores). Future studies are needed to investigate the NLPR3 inflammasome as a therapeutic target for cALD. Our mouse model may offer a novel preclinical tool such investigations.

Our analysis of immune cell infiltration in both human and murine cALD focuses on perivascular immune cell migration. However, immune cells can also infiltrate the brain through non-vascular routes⁴⁵. Previous work in MS and the CPZ/EAE model show immune cell infiltration across the ventricles and choroid plexus^46-48. To our knowledge, these avenues have not been studied in cALD. Future studies of cell migration across non-vascular routes in human cALD and murine models like ours may offer additional insights and therapeutic targets.

BBB disruption and perivascular immune cell infiltration, key features of cALD, are well-represented in our mouse model and can promote neuroinflammation^6,40. BBB disruption allows fibrinogen extravasation, forming proinflammatory fibrin that activates innate immune responses^22,49 Fibrin can trigger oxidative stress and NLRP3/IL-18 activation in macrophages/microglia impeding remyelination and neurogenesis ^22,50–52. In cALD mice, perivascular monocytes/macrophages highly express IL-18, which can promote a toxic astrocyte phenotype that propagates inflammation and exacerbates neurodegeneration⁵³. Our model could be used to screen therapies targeting BBB disruption, fibrin deposition, and/or NLRP3 inflammasome.

The cALD model offers an opportunity to study lesion arrest and brain remyelination. In human cALD, active brain lesions are characterized by contrast enhancement on MRI, indicating BBB disruption and predicting lesion expansion^6,15,39. This contrast enhancement and lesion growth can arrest in some patients, most often after hematopoietic stem cell transplantation. Unfortunately, the white matter injury and demyelination are irreversible⁵⁴. The cALD mouse model recapitulates active demyelinating lesions with the presence of contrast enhancement on MRI. Unlike mice exposed to CPZ alone that undergo spontaneous remyelination, our cALD mice showed persistent T2 hyperintensities and contrast enhancement at 5 and 10 weeks after starting a 2-week CPZ diet, suggesting sustained lesion activity akin to human cALD.

Our study and the proposed cALD model have important limitations. First, the triggers for cALD lesion formation in humans are likely diverse, involving pathways not fully captured by our induction protocol or mouse biology. Second, our work concentrates on the acute phase of lesion formation, showing parallels to human cALD lesions, but future studies on chronic phases are needed to assess demyelination persistence and tissue injury evolution. Additionally, the most promising therapeutic targets (e.g., fibrin, oxidative stress, NLRP3, IL-18) are yet to be determined. While we have established a causal role for Abcd1 deficiency in lesion evolution, the downstream mediators and therapeutic targets have yet to be elucidated.

CONCLUSION

By combining Abcd1 deficiency with an oxidative stressor (CPZ diet) and immune trigger (EAE), we have developed a novel mouse model of cALD that recapitulates key features of human cALD, including robust levels of cerebral BBB disruption, perivascular infiltration of innate and adaptive immune cells, demyelination, elevated markers of oxidative stress, and axonal damage. Our cALD mouse model demonstrates robust fibrin deposition and increased IL-18 within perivascular monocytes/macrophages, which aligns with similar pathology in human cALD. Finally, we use histological findings as well as non-invasive MRI markers to demonstrate sustained inflammation and demyelination more than two months after CPZ/EAE induction. As the first preclinical model for cALD, we expect this to enable long-sought investigations into disease mechanisms and accelerate development of candidate therapies for lesion prevention, cessation, and remyelination.

Summary of Social Media:

1. If you and/or a co-author has a Twitter handle that you would like to be tagged, please enter it here. (format: @AUTHORSHANDLE)

@DrDrIsha

2. What is the current knowledge on the topic?

X-linked adrenoleukodystrophy (ALD) is a progressive neurometabolic disorder caused by mutations in ABCD1. Up to 60% of males with ALD develop cerebral ALD (cALD), a fatal and neurologically devastating phenotype characterized by inflammatory cerebral demyelination. The lack of a robust preclinical model for cALD has hampered the study of disease mechanisms and therapeutics.

3. What question/problem did this study address?

The lack of a robust preclinical model for cALD has hampered the study of disease mechanisms and therapeutics.

1. We sought to develop a robust mouse model of cALD to overcome the limitations of existing Abcd1-null mice, which fail to spontaneously develop inflammatory cerebral demyelination.

2. We used human cALD brain tissue and CSF to benchmark the model against both new and established cALD markers allowing an assessment of the model’s robustness and utility for future research.

4. What does this study add to our knowledge?

1. We introduce a novel mouse model of cALD that recapitulates human disease features, providing a valuable tool for studying disease mechanisms and developing new therapies.

2. We demonstrate that loss of Abcd1 function predisposes to more severe cerebral inflammation and demyelination, advancing our understanding of cALD pathogenesis.

3. We identify potential therapeutic targets for cALD, including IL-18, which is associated with NLRP3 inflammasome activation.

5. How might this potentially impact on the practice of neurology?

1. We describe a robust new model for cALD that will facilitate the study of cALD’s molecular mechanisms and accelerate therapy development.

2. We identify IL-18, and the NLRP3 pathway, as a potential therapeutic target for cALD.

3. These insights may lead to the development of novel therapies aimed at: a) Preventing, b) Halting, or c) Reversing neuroinflammatory demyelination in patients with cALD or similar neuroinflammatory diseases.

Abbreviations

ABCD1

ATP binding cassette subfamily D member 1

cALD

Cerebral Adrenoleukodystrophy

BBB

Blood-brain barrier

CPZ

Cuprizone

EAE

Experimental Autoimmune Encephalomyelitis

MCC

Medial Corpus Callosum

NLRP3

NOD-, LRR- and Pyrin Domain-Containing Protein 3

PVC

Perivascular Cuff

VLCFAs

Very Long Chain Fatty Acids

X-ALD

X-linked Adrenoleukodystrophy

Data availability

Data presented in the manuscript is available in the article and Supplemental materials. Additional data is available from the corresponding author, KPV, upon reasonable request.