Progressive neuroinflammation and deficits in motor function in a mouse model with an Epg5 pathogenic variant of Vici syndrome
Bradley T. Thornton, Alexandra G. Hardinger, Laramie Pence, Priyanka Prem Kumar, Nikolas Connolly, Scott J. Weir, Jay L. Vivian
TL;DR
Researchers created new mouse models of Vici syndrome to study how genetic mutations cause neurological decline and found that brain inflammation plays a key role.
Contribution
The study introduces novel Epg5 mutant mouse models that mimic patient-derived pathogenic variants and reveals a role for neuroinflammation in Vici syndrome.
Findings
Epg5 mutant mice show progressive neurological deficits and perinatal lethality.
Transcriptomic analysis reveals robust neuroinflammatory signatures in the central nervous system.
The findings suggest that neuroglial activation contributes to the pathogenesis of Vici syndrome.
Abstract
Vici syndrome (VS) is a rare pediatric genetic disorder characterized by profound developmental delay, seizures, immune deficits, cardiomyopathy and progressive motor dysfunction. This devastating condition is caused by pathogenic variants in the EPG5 gene, which encodes a regulator of autophagy, leading to the accumulation of toxic intracellular material and widespread cellular dysfunction. Less-severe EPG5 pathogenic variants have recently been linked to rare familial forms of Parkinson’s disease, suggesting deficits in EPG5 function drive a range of neurodegenerative disorders. Currently, there are no effective treatments for any disorders associated with pathogenic variants of EPG5. The underlying cellular mechanisms driving the progressive neurological decline in VS remain poorly understood. Previous studies using Epg5 knockout models have demonstrated severe neurological…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7- —Mercy Research Partners of Children’s Mercy Kansas City Vici Syndrome Foundation Vici syndrome families
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAutophagy in Disease and Therapy · Mitochondrial Function and Pathology · Cellular transport and secretion
Introduction
Vici syndrome (VS; OMIM 242840; also known as EPG5-related disorder) is a rare, autosomal recessive, childhood-onset multisystem disorder that primarily affects the central nervous system (CNS), cardiac function and immune response, among other organ systems^1–3^. Clinical manifestations typically emerge within the first year of life and include failure to thrive, global developmental delay, immunodeficiency with recurrent infections^4^ and cardiomyopathy. Neurological features include hypotonia, developmental delay^5^, seizures, progressive motor decline and structural brain abnormalities^1^. Affected children are typically nonverbal and unable to ambulate^5^. VS follows a relatively progressive course, with a median survival of approximately 42 months; however, a range of phenotypes is observed, suggesting a spectrum influenced by the associated pathogenic variants^6,7^. Mortality is commonly attributed to a combination of progressive cardiomyopathy, recurrent infections and respiratory compromise due to neuromuscular decline. Neurological deterioration is a nearly universal hallmark of this disease and substantially impairs quality of life^2^. Currently, no disease-modifying therapies exist and clinical management remains largely supportive and palliative. The rarity of VS, coupled with limited access to patient-derived samples^8,9^ and the absence of well-characterized disease models, has hindered both mechanistic understanding and therapeutic development.
A definitive diagnosis of VS is established through genetic testing, which reveals pathogenic variants in the Ectopic P-granules protein 5 (EPG5) gene^10,11^. EPG5 encodes a critical regulator of autophagy, facilitating the fusion of autophagosomes with lysosomes^12^. Most pathogenic variants identified in VS are nonsense or frameshift mutations^10^ resulting in loss of function, although some missense mutations have also been reported^6^. Patient-derived cells exhibit impaired autophagic flux, characterized by the accumulation of undegraded autophagic material due to defective EPG5-mediated vesicle fusion with the lysosome^13,14^. VS is now recognized as part of a broader group of rare disorders^15,16^ termed congenital disorders of autophagy. Deficits in autophagy may drive a range of neurological disorders and other diseases^17,18^, suggesting new avenues for treatment using targeted therapies that modulate autophagy. A recent study of rare familial forms of Parkinson’s disease have identified pathogenic variants in EPG5^19^, suggesting deficits in EPG5 may drive a spectrum of neurodegenerative diseases beyond VS.
Previous studies using Epg5 knockout models have demonstrated severe neurological phenotypes^20^; however, the molecular and cellular mechanisms underlying the progressive motor deficits in VS remain poorly understood. Moreover, no existing mouse models that recapitulate patient-specific EPG5 mutations have been reported. To address this gap, we generated and characterized two novel Epg5 mutant mouse strains: one harboring a truncating W860X pathogenic variant identified in a patient (case report in preparation) and another carrying a newly developed null allele. Both models exhibit early onset and progressive neurological phenotypes, with subtle but significant differences between alleles. Whole-animal deficits in motor function were evident by 6 weeks of age and worsen over time. Transcriptomic analyses of brain and spinal cord tissue from these models revealed robust neuroinflammatory signatures, implicating glial activation, particularly of microglia and astrocytes, in disease progression. Notably, the molecular profiles observed in Epg5-deficient mice share features with disease-associated microglia^21^ identified in models of neurodegenerative disorders such as Alzheimer’s disease and amyotrophic lateral sclerosis, suggesting potential mechanistic overlap. These findings enhance our understanding of VS pathogenesis and provide candidate biomarkers for disease monitoring. Furthermore, these models establish a critical platform for preclinical evaluation of therapeutic strategies for this devastating disorder.
Materials and methods
Animal use
The mice were generated and maintained on a C57BL/6J genetic background, and the mutant stocks were maintained as heterozygotes by backcrossing to vendor-derived animals (Jackson Laboratories). Homozygous mutant mice were closely monitored for animal welfare concerns related to neurological dysfunction and removed from the study via approved euthanasia methods if any effects on welfare were identified. The animals were housed in a specific pathogen-free vivarium operating as a barrier facility in single-use disposable mouse cages (Innovive) with a 12-h light/dark cycle and maintained at 22 ± 1.0 °C and 40 ± 5% humidity. Pregnant dams and preweaning pups were fed high-fat chow and water ad libitum, and postweaning aged mice were fed standard laboratory chow and water ad libitum.
Generation of mouse strains with Epg5 W860X and Epg5 KO alleles
The design and validation of the CRISPR reagents used for in vivo genome editing were performed as described previously^22^. To generate the Epg5 W860X allele, a CRISPR target site (CCCGAATCACAGCTATCTCT) was identified in exon 14 of the mouse Epg5 locus. To generate the Epg5 KO allele, a CRISPR target site (CGGGCTCTCCTGCTCGCGAC) was identified at the 5’ end of exon 2 of the mouse Epg5 locus. CRISPR reagents were synthesized as crRNAs and annealed to universal tracrRNAs (Integrated DNA Technologies). Before production, the activity of the CRISPR reagents was verified via embryo electroporation of the Cas9–crRNA–tracrRNA ribonucleoprotein (RNP) complex. The Epg5 W860X mutant mice were produced as described^22,23^ via CRISPR-mediated genome editing and homology-directed repair. Single-stranded donor DNA was designed for the W860X allele (Supplementary Table 1) and generated via DNA synthesis (Integrated DNA Technologies). For Epg5 W860X, the donor DNA incorporated a single nucleotide change, resulting in a premature stop codon in exon 14 followed by a restriction site and flanked by 88 bp and 68 bp of Epg5 homology for targeting via homology-directed repair. The CRISPR RNP complexes were coinjected with donor DNA into C57BL/6J zygotes^24^, which were subsequently transferred to pseudopregnant females^23^ for delivery. The resulting founders were confirmed for correct targeting of the donor DNA via allele-specific PCR and sequencing (Supplementary Table 1). To produce the Epg5 KO strain, the CRISPR reagents were electroporated into C57BL/6J zygotes, which were subsequently transferred to pseudopregnant females for delivery. The resulting founders were analyzed for deletions generated via nonhomologous end-joining via PCR and the nature of the resulting mutations was confirmed by next-generation sequencing of PCR amplicons. One founder harboring a 22-nucleotide deletion in exon 2 resulting in a frameshift mutation was expanded by breeding towild-type (WT) C57Bl/6J mice. Germline transmission of each desired targeted allele was confirmed in F1 generation pups via PCR and sequencing. Homozygous mutant mice were generated by intercrossing of heterozygous mice, and homozygous WT mice were obtained from the same litters as controls.
Behavior analysis
The analysis of motor coordination, endurance and fatigue resistance was performed via the rotarod test (Panlab). The mice were acclimated and housed in rooms with an adjusted light‒dark cycle to allow for testing during the active dark phase. Test subjects were given two training days with the apparatus set at 4 rpm for 1 min for three runs each training day. One week after training, the mice were tested every 2 weeks with the apparatus set at a ramp speed from 4 to 40 rpm over the course of 300 s. The mice were tested for three trials per testing day with an intertrial interval of 10–15 min. Failure to initiate was defined as a trial in which the animal was unable to remain on the rod for one full revolution at 4 rpm at the beginning of a test. The latency to fall and speed at fall data from each trial were recorded via SEDACOM 2.0 software (Panlab). Data visualization and statistical analysis were performed via Prism software (GraphPad). Statistical analysis of data from WT and mutant mice at a given time point was performed via an unpaired t-test.
Protein analysis
Spinal cord tissues from 6-month-old WT and mutant siblings of each of the Epg5 mutant strains were dissociated in RIPA buffer (Thermo Fisher) with a TissueRuptor homogenizer (Qiagen). The supernatant was collected, and a protease and phosphatase inhibitor cocktail (Thermo Fisher) was added. The protein concentration of the supernatant was quantified via a Pierce BCA assay (Thermo Fisher) per the manufacturer’s protocol. Protein samples were denatured in Laemmli sample buffer (Bio-Rad) and 20 µg of protein from each sample was used for western blot analysis. Antibodies against LC3 (Cell Signaling Technology) and α-tubulin (Cell Signaling Technology) and secondary antibodies (LI-COR Biosciences) were used to detect each protein. Fluorescent imaging was performed on the membrane via an Odyssey CLx Imager (LI-COR Biosciences). Each band was quantified via ImageStudioLite analysis software (LI-COR Biosciences).
Purification of microglia
Primary microglia were isolated from 7-month-old WT and homozygous Epg5 W860X mutant mice. Mice were euthanized via exsanguination via cardiac perfusion with cold PBS under isoflurane anesthesia, after which the whole brain was isolated. Whole brains were processed according to manufacturer’s protocol using enzymatic digestion and dissociation (Adult Brain Dissociation Kit, GentleMACS Dissociator and Debris Removal Solution; Miltenyi Biotec). Subsequent magnetic separation of purified microglia was performed using CD11b microbeads and separation columns (Miltenyi Biotec). The purified microglia were lysed immediately in Qiazol (Qiagen) for use in RNA purification.
RNA analysis
Gene expression analyses were performed on purified RNA from tissues and from purified microglia collected from 6–7-month-old WT and homozygous mutants of the Epg5 strains. The tissues or cells were lysed via Qiazol/chloroform extraction and the RNA was purified via a Qiagen RNeasy column purification kit (Qiagen) according to the manufacturer’s protocols. The extracted RNA (1 µg of total RNA for tissues, 180 ng from purified microglia) was then either given to the institutional genomics core for RNA sequencing or was converted into cDNA for quantitative PCR. The qPCRs were performed using TaqMan Gene Expression Master Mix (Thermo Fisher) and the indicated probes (Supplementary Table 2). The samples were analyzed via the ViiA 7 real-time PCR system (Applied Biosystems) according to the manufacturer’s protocols. For plotting, the expression data were normalized to the data of the WT control siblings.
RNA-seq bioinformatics analysis
Transcriptomic profiling via bulk RNA sequencing (RNA-seq) was performed on RNA from the cerebellum and spinal cord of homozygous W860X mice or their WT control siblings. RNA-seq was performed with stranded total RNA with a RiboZero Gold library preparation kit (Illumina) per the manufacturer’s protocols. The RNA sequences were processed for gene expression quantification. The standard two-group differential expression analysis of a set of RNA-seq samples was processed with Kallisto^25^. Output files were processed with tximport^26^ to collapse transcript counts to gene counts via the transcript-to-gene mapping table. The mouse reference genome GRCm39 was used in the analysis. To generate a volcano plot, the processed average differential expression data from the RNA-seq data consisting of 13,991 genes in the cerebellum and 25,527 genes in the spinal cord were input into Prism (GraphPad). For comparative differential gene expression analysis between tissues, genes that had a log_2_(fold change) of +1/−1 and a false discovery rate (adjusted P value) of less than 0.01 were used. Overlapping differentially expressed genes (DEGs) between tissues were identified via Venny (https://bioinfogp.cnb.csic.es/tools/venny/index.html). The overlapping genes were then input into Ingenuity Pathway Analysis (Qiagen) for pathway analysis and to g:Profiler^27^ for Gene Ontology (GO) analysis. Altered pathways and associated P values were visualized via Prism (GraphPad).
Histology
Tissues were collected and processed as previously described^28^. In brief, 6-month-old mutant or control mice were placed under anesthesia (isoflurane) for transcardial perfusion with 4% paraformaldehyde in PBS. CNS tissues were then removed and fixed overnight. The tissues were then washed two times in PBS before preparation for histology.
For immunohistochemistry, the tissues were dehydrated and embedded in paraffin. The 10 µm sections were rehydrated via citrate antigen retrieval buffer for use in staining to detect GFAP (HRP/DAB IHC Detection Kit, Abcam) per the manufacturer’s protocol. A colorimetric reaction was performed using DAB.
For immunofluorescence imaging, the tissues were frozen for cryosectioning as described previously^29^. In brief, 20 µm sections were cut. Direct conjugated antibodies to detect GFAP or IBA-1 were used to detect astrocytes and microglia, respectively, and costained with DAPI to detect nuclei. The sections were mounted with a coverslip with Fluormount G mounting media (Thermo Fisher). Immunofluorescence imaging was performed on a Zeiss Axio Observer 7 microscope (Zeiss). For quantification, images of distinct regions of the spinal cord were imported into the Imaris software package (Oxford Instruments). For each fluorophore, an absolute intensity threshold value was adjusted via the histogram and standardized for each image. A new surface was created to measure the total area of the image. The summed value of the combined area from each surface was divided by the total area to obtain the percentage value of positivity for each image.
Results
To develop mouse models that recapitulate a pathogenic EPG5 variant associated with VS, a novel mouse strain was generated via CRISPR-mediated zygotic mutagenesis^22–24^ on an inbred C57Bl/6J genetic background (Table 1 and Supplementary Table 1). One strain was engineered to carry the Epg5 W860X mutation, which mirrors a recently identified pathogenic variant in a child with severe VS (manuscript in preparation). For comparison, a knockout (KO) strain was also generated by introducing a 22 bp interstitial deletion near the start of exon 2, resulting in a frameshift and predicted early truncation of the EPG5 protein. Following generation and colony expansion, each strain was intercrossed to produce homozygous mutant mice.Table 1. Summary of mutant mouse strains.Mouse alleleEpg5 exonMutation^1^Human pathogenic variant^2^Human gnomAD variant IDHuman RefSNP IDKO2p.R41QfsTer8 (c.122_143del)N/AN/AN/AW860X14p.W860XW865X (c. 2595 G > A)SNV: 18-45925861 - C-Trs1160786631^1^Mouse accession number: NM_001195633.2.^2^Human accession number: NM_020964.3.
Genotypic analysis of offspring at postnatal day 14 (P14) revealed a significant underrepresentation of homozygous mutants for both the W860X and KO alleles (Table 2). Chi-square analysis indicated that this loss was due to a reduced number of surviving males (P = 0.043 for KO; P = 0.035 for W860X). A small number of deceased pups at P1 were found to be homozygous mutants, suggesting perinatal lethality associated with these alleles.Table 2. Summary of the genotypes of offspring from intercrosses of heterozygous Epg5 mutant mice.Epg5 alleleKO**W860XTotalMFTotalMFGenotypeExpected1075651249126123 WT (+/+)25%31.8%32.1%31.4%30.1%29.4%30.9%Heterozygous50%55.1%57.1%52.9%51.8%55.6%48.0%Homozygous mutant25%13.1%10.7%15.7%18.1%15.1%21.1%P value0.0140.0430.2610.0230.0350.280significanceNSNSTotal: the total number of mice analyzed from heterozygous intercrosses.M: number of male mice.F: number of female mice.*P < 0.05 according to the chi-square test.
Surviving homozygous mutants were monitored longitudinally postweaning. Mice homozygous for either the W860X or KO mutation were generally smaller than their WT littermates, although only W860X mutants consistently exhibited reduced body size throughout life (Fig. 1a,b and Supplementary Fig. 1). All the homozygous mutants displayed marked motor deficits, particularly in hindlimb coordination. Compared to age-matched controls, mutant mice exhibited abnormal hindlimb posture, characterized by a splayed stance and inability to support body weight (Fig. 1c,d). A moderate ataxic phenotype was evident, with wobbly, halting, slow gait, and a high frequency tremor. (Supplementary Movie 1). Owing to progressive motor impairment, most of the mice were removed from the study by 7 months of age for animal welfare reasons (Fig. 1e). No sex differences were observed in the severity of motor deficits.Fig. 1. Summary of whole-animal deficits in Epg5 mutant mouse models.a,b Weight comparison of mutant homozygous mutant with sibling control WT mice at 4 weeks (a) and 8 weeks (wk) of age (b). c–e In comparison to control mice (c), W860X homozygous mutant mice (d) display poorly placed hindlimbs when attempting to bear weight at 4 months of age. Hindlimb function in homozygous W860X mice is progressively more severe by 7 months of age (e). f–i An assessment of motor function of W860X (f and g) and KO (h and i) mutant mice via rotarod. In comparison to age-matched control siblings, homozygous mutant mice have poor capacity to navigate rotarod test of coordination. Mutant mice fail the test at lower speed of rotation (‘speed at fall’; f and h) and at shorter time (‘latency to fall’; g and i) on the apparatus. At later ages starting at 12 weeks, some homozygous mutant animals are unable to initiate the rotarod test, and all W860X mice fail at 14 weeks and all KO mice fail at 16 weeks. Analysis of data between WT and mutant mice at a given time point was performed via an unpaired t-test. ****P < 0.0001, ***P < 0.001, **P < 0.01 and *P < 0.05.
To quantitatively assess motor function, rotarod testing was conducted longitudinally beginning at 8 weeks of age. Compared to their control littermates, both W860X and KO homozygous mutants presented significantly impaired performance (Fig. 1f–i). By 12 weeks, some mutants were unable to remain on the rotating cylinder, by 14 weeks, all W860X homozygous mutants and by 16 weeks, all KO homozygous mutants were unable to initiate the test. Both male and female mutants were equally affected. Owing to the severity of motor impairment, mutants were unable to perform additional coordination tests, such as beam walking. These findings of deficits in motor function were similar to a previous analysis of other Epg5 mutant mouse models^20^. These studies indicate early onset and progressive deficits in survival, growth and motor function in both Epg5 mutant strains, with modest differences in severity between the two mutant alleles analyzed.
Tissues from each mutant strain were collected for transcript and autophagy analyses. Compared to those of WT controls, RNA extracted from the spinal cords of W860X homozygous mice presented significantly lower Epg5 transcript levels (Fig. 2a). Although not statistically significant, a similar trend was observed in the cerebellum, suggesting tissue-specific transcript instability, potentially due to nonsense-mediated decay. By contrast, the transcript levels in the KO mice were not reduced. Western blot analysis of the LC3 protein (microtubule-associated protein 1 light chain 3 alpha) was used to assess autophagy (Fig. 2b,c and Supplementary Fig. 2). Analysis of LC3 via western blotting distinguishes the cytoplasmic form of LC3 (LC3-I) from the autophagosome-localized LC3 (LC3-II). LC3-II is subject to autophagy-mediated degradation, thus the steady-state levels of LC3-II allow for an assessment of autophagy. The autophagosome-associated LC3-II was elevated in spinal cord lysates from both mutant strains compared to tissues from WT controls, indicating that autophagic flux was impaired with a resulting LC3-II accumulation.Fig. 2. Molecular deficits in Epg5 transcript stability and markers of autophagy from tissues derived from Epg5 mutant mice.a Assessment of Epg5 transcript stability in control and Epg5 mutant mice at 6 months of age using a RT–qPCR primer pair that amplifies the 3′ end of the Epg5 transcript. Unstable Epg5 transcript in spinal cord tissue from mice with mutations that recapitulate pathogenic variants. Epg5 transcript levels from cerebellum tissues were reduced but not significant. b,c Western blot analysis (b) and quantification (c) of spinal cord tissue of Epg5 mutant mice at 7 months of age to detect LC3, a marker of autophagy. Note the increased accumulation of autophagosome-associated LC3-II compared to cytoplasmic LC3. Analysis of gene expression levels between WT and mutant mice performed via an unpaired *t-*test. ***P < 0.001.
To characterize the neurological phenotype further, bulk whole-transcriptome sequencing was performed on RNA from cerebellar and spinal cord tissues from 6-month-old W860X homozygous mice and control littermates. These CNS tissues were selected because of their roles in motor control. The transcriptomic profiles (n = 4 of each genotype, split evenly by sex) were consistent between the sexes and were combined for this analysis. A substantial number of DEGs were identified, with more genes upregulated than downregulated (Fig. 3a,b). Notably, genes associated with glial activation, including microglial markers (Gpnmb and Itgax) and the astrocytic marker Gfap were upregulated, suggestive of a gliosis in these tissues. Several Toll-like receptor genes (Tlr2 and Tlr7), key components of innate immune signaling, were also upregulated. Among the downregulated genes, Myoc (myocilin) was nearly absent in the spinal cord of W860X mutants but unchanged in the cerebellum. This gene is specifically expressed in the glia limitans superficialis (GLS), a specialized astrocyte population localized to the thin surface layer throughout the CNS.Fig. 3. Transcriptomic analysis of CNS tissues shows a range of molecular deficits associated with Epg5 pathogenic variants.a,b Volcano plots from whole-tissue transcriptomic analysis of anterior spinal cord (a) and cerebellum (b) showing DEGs Epg5 W860X homozygous mutant mice compared to control age-matched siblings. Select genes of interest are marked. c A Venn diagram of overlapping DEGs in Epg5 W860X mutant CNS tissues. For this analysis, genes were identified that have both twofold or greater up- or downregulation and adjusted P value less than 0.01. d,e Pathway (d) and GO (e) analyses of overlapping CNS DEG gene signature in Epg5 W860X mutant mice. f A heat map of DEGs from spinal cord and cerebellum of Epg5 W860X from the transcriptomic analysis of select genes associated with a previously identified disease-associated microglia signature in other mouse neurological damage models. Genes that were absent from the transcriptomic analysis of the cerebellum (Clec7a and Ccl4) were assessed with Taqman qPCR. Note that a substantial number of these genes are upregulated in the CNS of Epg5 W860X mutant mice. g,h Gene expression analysis of purified microglia from brain of 7-month-old control mice or W860X homozygous mutant mice. Microglia were purified from whole brain via enzymatic dissociation and CD11b magnetic microbeads, then used for total RNA purification for gene expression analysis. The upregulation of disease-associated microglia (g) and other markers associated with microglial activation (h) in mutant microglia compared to control mice indicates increased expression of these genes in microglia.
Although distinct sets of DEGs were identified in the cerebellum and spinal cord of Epg5 W860X mutants, a shared gene expression signature was also observed across both tissues (Fig. 3c and Supplementary Table 3). Analysis of these overlapping DEGs highlighted key cellular pathways affected by the W860X mutation in CNS tissues (Fig. 3d,e). GO and pathway enrichment analyses confirmed that these shared DEGs were significantly associated with inflammatory and immune responses. Pathways identified in this analysis include cytokine, interferon and complement inflammatory pathways. Cellular pathways identified as neutrophil degranulation and phagosome formation may suggest components of a microglial activation. Although several pathways were identified with pathogen response, this analysis likely does not represent an inflammatory response to microbial insult given the specific pathogen-free status of the vivarium within which these animals were housed.
Previous studies have identified distinct molecular signatures associated with microglial responses to neurological injury in various mouse disease models^21^ and in human neurodegenerative disorders^30^. Among these, a disease-associated microglia (DAM) signature, also referred to as the neurodegenerative microglial phenotype, has been characterized by the upregulation of genes such as Clec7a, Cst7 and Trem2^21^. Transcriptomic analysis of CNS tissues from Epg5 W860X homozygous mutant mice revealed upregulation of many DAM-associated genes (Fig. 3f), suggesting a glial response reminiscent of other CNS injury models. To confirm that the increased expression of DAM markers observed in the bulk transcriptomic data is due to upregulated expression in the microglia, microglia were purified from dissociated whole brain from W860X homozygous mice or control WT siblings at 7 months of age using CD11b magnetic microbeads. qPCR analysis of gene expression of DAM markers Clec7a, Cst7, Itgax and Ccl4 (Fig. 3g,h) and other markers indicative of an activated microglial state in neurological disorders^31–33^ such as Lilrb4a and Irf7 (Fig. 3h) were observed to be higher in the purified mutant microglia compared to control when normalized for equal cell numbers. This important result confirms that the microglia in the Epg5 mutant mice display upregulated expression of these markers, indicative of a DAM phenotype. Collectively, these transcriptomic data define a novel and complex neuroinflammatory and glial activation signature in mice carrying the Epg5 W860X pathogenic variant that is reminiscent of other neurological disease or damage models involving neuroglia.
To further evaluate the expression of selected DEGs, RNA was isolated from the CNS tissues of 7-month-old mutant and control mice for qPCR analysis (Fig. 4). Genes were selected on the basis of the transcriptomic findings, with a focus on glial markers (astrocytic and microglial), inflammatory mediators, interferon response genes and other relevant targets. All DEGs identified in the W860X model were also dysregulated in the Epg5 KO mice, although with varying magnitudes (Fig. 4a–d and Supplementary Figs. 3 and 4). For example, Gpnmb and Gfap expression was more severely elevated in W860X mutants than in KO mice, whereas other genes presented similar expression levels across both models. Notably, the severity of gene expression deficits varied by tissue. Tlr7 expression, for example, was more severely affected in the cerebellum of W860X mutants, whereas spinal cord expression levels did not differ significantly between the two models. These molecular findings parallel the phenotypic severity observed in vivo, with similar levels of severity between the mutant alleles, with the W860X mutants exhibiting somewhat more pronounced deficits than KO mice.Fig. 4. Comparison of gene expression defects between Epg5 mutant alleles.The expression of select DEGs identified from the transcriptomic analysis of Epg5 mutant mice was examined via qPCR via TaqMan assays. a–d A comparison of gene expression deficits in the spinal cord (a and b) and cerebellum (c andd) of the Epg5 homozygous mutant mice and age-matched sibling WT control mice at 6 months of age. Compared to the WT controls, all the mutants presented significant deficits in gene expression of these selected markers at 6 months of age. Analysis of gene expression levels between WT and mutant mice via one-way ANOVA. ****P < 0.0001, ***P < 0.001, **P < 0.01 and *P < 0.05.
To assess the temporal progression of the gene expression deficits associated with the Epg5 mutations, the expression of a subset of glial-specific genes was evaluated in 2- and 6-month-old W860X and KO mutant mice compared to WT controls (Fig. 5 and Supplementary Fig. 5). The selected genes included the astrocyte marker Gfap and the microglial markers Gpnmb and Itgax. Many of these genes were already significantly upregulated at 2 months of age compared to WT mice, particularly in the spinal cord of both mutant models (Fig. 5a–d), and their expression increased further by 6 months. By contrast, the changes in gene expression in the cerebellum of the KO mice were relatively modest at 2 months but became more pronounced with age. Gene expression defects in additional cell subtypes were confirmed in this analysis. Tnc, which encodes Tenascin C, a factor specifically expressed in the Bergmann glia of the cerebellum^34^, was substantially reduced at 6 months of age in both Epg5 mutant mouse models. Aqp4, which encodes Aquaporin 4, a marker of a subtype of perivascular-associated astrocytes^35^, was significantly increased 3.3-fold in the W860X homozygous mutant cerebellum in the RNA-seq analysis. These results indicate that glial phenotypes in Epg5 W860X mutants were already apparent at early adulthood stages and progress with age, which is consistent with the age of observed deficits in motor function.Fig. 5. Comparison of gene expression defects between Epg5 mutant alleles and effects on age.a–d Age effects on gene expression deficits in Epg5 W860X (a and b) and Epg5 KO (c and d) CNS tissues. Tissues from the anterior spinal cord (a and c) and cerebellum (b and d) were isolated from the indicated mutants between 2 and 6 months of age for gene expression analysis. Relative expression is based on comparisons with WT mice at the 2-month time point. Analysis of gene expression levels between WT and mutant mice in a given cohort via one-way ANOVA. ****P < 0.0001, ***P < 0.001, **P < 0.01 and *P < 0.05.
Given the strong glial signature observed via transcriptomic and qPCR analyses, histological studies were performed to assess astrocyte and microglial distribution and phenotype in CNS tissues from 7-month-old Epg5 W860X mutants and control WT littermates. Immunohistochemistry revealed a substantial increase in GFAP-positive astrocytes across all CNS regions examined, which was consistent with elevated Gfap transcript levels (Figs. 6a,b and 7a–d). Immunofluorescence staining for GFAP and the microglial marker IBA-1 confirmed widespread astrocyte and microglial expansion in the spinal cord (Fig. 6c–n) and cerebellum (Fig. 7e–g) in Epg5 mutant mice. Microglia exhibited a hypertrophic, spider-like morphology with extended processes (Fig. 6g, arrowhead), indicative of activation. In the cerebellum, increased numbers of astrocytes and microglia were observed across multiple layers, including the molecular layer (ML), internal granule layer (IGL) and white matter (WM) (Fig. 7f,g). Despite these cellular changes, the overall cerebellar architecture remained intact.Fig. 6. Glial expansion in the spinal cord of Epg5 mutant mice.a–m Anterior spinal cord tissues, corresponding to the region around cervical vertebra 6, were collected from 7-month-old control (a, c, f, i and l) or Epg5 W860X homozygous mutant mice (b, d, g, j and m) for histological analysis. Histochemical analysis to detect GFAP in the ventral spinal cord (a and b). A substantial increase in GFAP-positive cells was observed throughout the spinal cord in the Epg5 W860X homozygous mutant mice. Scale bar, 100 µm. Immunofluorescence was used to detect astrocytes (GFAP; green), microglia (IBA-1; red) and nuclei (DAPI; blue) in the spinal cord (c–n). Expanded microglia and astrocytes were observed throughout the spinal cord in the Epg5 W860X homozygous mutant mice (c and d). Scale bar, 300 µm. High-magnification assessment of microglia and astrocytes in specific regions of the spinal cord (e–n). High-magnification image of the ventral gray matter of the spinal cord showing expanded astrocytes and microglia, with microglia demonstrating morphological changes, including thickened and more extensive processes and a large size, suggesting an activated phenotype (arrowheads) (f and g). Scale bar, 100 µm. Quantification of the increase in microglia in the ventral horn gray matter of the spinal cord as a percentage of the total image area, demonstrating a quantitative increase in microglial content (h). A region of the dorsal funiculus of the spinal cord (i and j). Scale bar, 150 µm. Substantial expansion of microglia in the dorsal-most region of the dorsal funiculus (k). Details of the ventral border of the spinal cord (l and m). The top region of each image shows microglia, which are substantially increased in this region. The bottom portion of each image shows astrocytes, including the GLS, which are substantially expanded in the border region of the spinal cord. Scale bar, 100 µm. Quantification of astrocyte content in the ventral spinal cord and ventral border (n). Analysis of cell quantification between WT and mutant mice was performed via an unpaired *t-*test. ****P < 0.0001. The diagram in e was generated with Biorender.com.Fig. 7. Glial expansion in the brain and cerebellum of Epg5 mutant mice. Whole brains were collected from 7-month-old Epg5 W860X homozygous mutant mice for histological analysis.a–d Histochemical analysis to detect GFAP, a marker of astrocytes in the brain (a and b; scale bar, 1 mm) and focus on the cerebellum (c and d) in control (a and c) or Epg5 W860X homozygous mutant mice (b and d). A substantial increase in GFAP-positive cells was observed throughout all regions of the brain, including the cerebellum. e–g Immunofluorescence was used to detect astrocytes (GFAP; green), microglia (IBA-1; red) and nuclei (DAPI; blue) in the indicated regions of the cerebellum (e) in control (f) or Epg5 W860X homozygous mutant mice (g). Expanded microglia and astrocytes are observed in all layers of the cerebellum of Epg5 W860X homozygous mutant mice. Scale bar, 200 µm. Layers of the cerebellum are indicated. PCL, Purkinje cell layer. The diagram in e was generated with Biorender.
While glial expansion in Epg5 mutant mice was broadly distributed, certain white matter regions presented particularly dense microglia and astrocyte accumulation. In the spinal cord, microglia were notably concentrated in the dorsal funiculus (Fig. 6i–k), a region containing descending corticospinal and ascending sensory tracts. In addition, both astrocytes and microglia were enriched along the surface of the spinal cord (Fig. 6c,d,l–n). While astrocytes are normally present in this region, their numbers and process extension were markedly increased in mutants (Fig. 6d,m). Microglia, which are typically sparse at the surface, were also significantly expanded (Fig. 6m). These findings demonstrate a robust and progressive glial response in the CNS of Epg5 W860X mutant mice, which is correlated with the observed motor deficits and supports a role for glial dysfunction in the pathogenesis of VS.
Discussion
The complex etiology of VS remains incompletely understood, contributing to the current lack of effective therapeutic strategies. In particular, the molecular and cellular basis for the neurological manifestations of VS, which are universally present in individuals with VS, remain poorly characterized. This study presents the first published analysis of a novel mouse model harboring a pathogenic variant in Epg5 that mirrors a mutation identified in patients with VS. Specifically, the Epg5 W860X allele recapitulates the EPG5-W865X nonsense mutation found in a clinically severe VS patient (manuscript in preparation). Homozygous W860X mutant mice exhibit profound motor deficits that are comparable in severity to those observed in both the Epg5 KO mice analyzed in this study and the previously reported Epg5-deficient models^20^.
Focusing on the molecular and cellular underpinnings of neurological dysfunction, our analysis revealed previously unreported CNS phenotypes in these models. Using established molecular markers, we observed widespread and progressive expansion of astrocytes and microglia throughout the CNS in Epg5 mutant mice. This glial expansion increases with age and correlates with the severity of motor dysfunction, with significant motor deficits and glial marker upregulation evident by 2 months of age. Notably, a molecular signature of neuroinflammation, characterized by astrocyte and microglial expansion, suggests a contributory role for glial cells in VS pathogenesis and disease progression. Previous studies using Epg5 mutant mice revealed inflammatory and immune-related phenotypes^36,37^, including gastrointestinal and peripheral immune dysfunction. In conjunction with these prior studies, the findings in this study suggest the presence of an inflammatory phenotype across multiple organ systems in Epg5 mutant mice, underscoring the potential for inflammation contributing to the multisystemic nature of the disease. Specific CNS molecular markers may thus serve as indicators of disease progression and severity.
Comparative analysis of the mutant alleles confirmed that the W860X and KO alleles presented severe loss-of-function mutations. This aligns with the clinical severity observed in the patient diagnosed with VS harboring the corresponding W865X mutation, who succumbed to the disease at 16 months of age (clinical report in preparation). The reduced Epg5 transcript levels in W860X mutant mice suggest that the mRNA is subject to nonsense-mediated decay. Owing to the lack of a high-quality antibody for mouse EPG5, protein levels associated with the mutations generated in this report could not be assessed via western blot. However, unpublished data from patient-derived cells from our laboratory indicate that the W860X pathogenic variant does not produce a stable truncated protein, further supporting its classification as a null allele.
A novel and striking observation in our study was the reduced survival of male mice homozygous for either the W860X or KO allele, an outcome not previously reported in other Epg5 mutant models. This discrepancy may reflect differences in experimental design, such as the depth of analysis of progeny from heterozygous intercrosses or potential residual activity of the allele used in earlier studies. In humans, VS is associated with early onset hypotonia and failure to thrive. The reduced perinatal survival of homozygous mutant mouse pups may mirror this clinical phenotype. Further investigations are needed to elucidate the mechanisms underlying the observed male-specific lethality. Interestingly, previous studies reported sex-dependent differences in microglial abundance and activity during both embryonic development and adulthood^38^. Future analyses of Epg5 mutant tissues at late prenatal and early postnatal stages will be critical for understanding the origins of this lethality and defining the earliest manifestations of glial dysregulation associated with the defective autophagy in this model.
One of the most striking transcriptional changes observed in the spinal cord of Epg5 mutant mice was the downregulation of myocilin (Myoc), a marker of a specialized astrocyte subtype known as the GLS^39^. GLS astrocytes are localized to the pial surface of the brain and spinal cord and are thought to contribute to barrier function and immune surveillance. Myoc expression is initiated postnatally in these cells, suggesting Myoc is a marker for developmental maturation of GLS cells. The substantial downregulation of Myoc in Epg5 mutant mice implicates a requirement for autophagy, either directly or indirectly, in the survival or maturation of GLS astrocytes. Interestingly, our histological analysis revealed an expansion of astrocytes and microglia at the surface of the spinal cord, suggesting that reactive gliosis may interfere with GLS maturation. Further studies are needed to determine whether this represents a cell-autonomous defect in GLS astrocytes or a secondary effect of microglial activation and inflammation. Understanding the role of GLS astrocytes in CNS homeostasis and their potential involvement in VS pathogenesis could provide new insights into how autophagy deficits disrupt glial development and function. The effect on glial maturation, function or survival associated with loss of EPG5 function is likely to be highly cell subtype specific. For example, expression of genes associated with astrocyte function such as Kcnj10 or Slc1a3, which encodes Kir4.1 channel and EAAT1 respectively, with functions to clear extracellular potassium and glutamate^40,41^, were not seen to be disrupted in the transcriptomic datasets. Single-cell or spatial gene expression analyses methods will be important to assess these cell subtype-specific deficits in detail.
Previous reports on Epg5 mutant mice using a null allele have demonstrated partial phenotypic overlap with the disease presentations in patients with VS^20^, though not all hallmark features were recapitulated to the same level of severity^9^. Patients with VS exhibit variable age of death^7,11^ attributed to a combination of loss of cardiac function, progressive loss of motor function and immune system deficits. The analysis presented here of the newly developed Epg5 mutant mouse models have identified perinatal lethality in approximately half of the male homozygotes, potentially modeling the perinatal failure-to-thrive phenotype observed in many patients with VS^2^, possibly due to deficits in swallowing. While cardiac and immune deficits in these new mouse models remain areas for future investigation, this study focuses on neurological manifestations. Although patients with VS typically present with more severe CNS involvement^11^, our mouse model exhibits a progressive neurodegenerative phenotype, which may be attenuated by the controlled laboratory environment. Future studies incorporating different genetic backgrounds or environmental stressors may unmask additional phenotypes in the mouse models. Importantly, while cellular and molecular deficits in the CNS of patients with VS remain poorly defined, including a gliosis^11^, our work significantly expands potential molecular and cellular aspects of the disease. The neuroinflammatory signatures and glial expansion in the mouse models suggest potential parallels in human pathology. Validation of these findings in precious VS patient CNS samples^8^ will be critical to confirm this translational relevance.
Astrocytes and microglia are essential for maintaining CNS homeostasis and play increasingly recognized roles in modulating disease progression^42,43^. Astrocyte homeostatic functions^44,45^ include synaptic maintenance, neurotransmitter recycling, integrity of the blood‒brain barrier, and metabolic support of neurons. Microglia, the resident innate immune cells of the CNS, are responsible for the phagocytosis of cellular debris, apoptotic cells, and pathogens^46^. These glial cells not only are passive support cells but are also highly dynamic^47,48^, responding to environmental cues and pathological states with complex molecular and functional changes in a coordinated fashion^49^. In disease or injury contexts, glial cells display a range of functional plasticity, which can adopt a range of activation states, including proinflammatory and neuroprotective phenotypes^50^. Importantly, these cell types can adopt complex molecular phenotypes^48,51^ that may be either neuroprotective or neurotoxic depending on the context and duration of activation. Both astrocytes and microglia can secrete cytokines and chemokines depending on the activation state^52,53^, which influences the local microenvironment and contributes to neuroinflammatory cascades. Microglia play a critical role in synaptic pruning during development and in response to injury^54^.
There is growing appreciation for the dual roles of glial cells in contributing to the pathogenesis of neurological disorders. Dysregulation of glial function has been implicated in a wide range of neurodegenerative diseases^55–57^, including Alzheimer’s disease, Parkinson’s disease and amyotrophic lateral sclerosis (ALS). In models of Alzheimer’s disease, Parkinson’s disease and ALS, microglial activation has been shown to be correlated with disease progression. For example, APOE alleles associated with increased Alzheimer’s disease risk have been linked to altered microglial responses^58^, whereas SOD1 mutations in ALS models have demonstrated that glial dysfunction can exacerbate motor neuron degeneration^59^. These findings underscore the importance of glial cells not only as responders to neuronal injury but also as active participants in disease mechanisms. A recent study reported the exciting observation of non-null alleles of EPG5 as pathogenic variants for rare early onset Parkinson’s disease^19^. Future studies will be needed to ascertain the extent of the glial deficits in the CNS of these patients, and the role in Parkinson’s disease progression.
Whole-transcriptome analysis of CNS tissues from Epg5 mutant mice revealed a signature of activated microglia resembling other mouse CNS injury or disease models^21,60^. This distinct transcriptional profile, known as DAM, also referred to as the neurodegenerative microglial phenotype, has been identified in multiple models of neurodegeneration or injury. This phenotype is characterized by the upregulation of genes such as Apoe, Trem2 and Cst7, which are not typically induced in classical inflammatory responses such as lipopolysaccharide exposure^21^. DAM microglia are thought to play complex roles in disease progression, being initially protective via microglial phagocytic activity by clearing debris, but there is evidence that they may contribute to disease progression if chronically activated^61^. Future studies involving single cell analysis of microglia in the Epg5 mutant mice will be needed to uncover the spectrum of the microglial phenotypes in these mice and the regional differences in microglial states within the CNS, and how these may change over the course of the lifespan.
The complex neuroinflammatory signature seen in Epg5 mutant mice suggest that glial dysfunction may be a key driver of disease progression in VS. The localized accumulation of microglia in specific regions of the spinal cord in Epg5 mutant mice, specifically in white matter, may point to possible cellular mechanisms for a role of microglia in neuronal function in this model. Notably, microglial localization to specific CNS regions, such as the corticospinal tract in models of ALS^62^, has been associated with region-specific vulnerability. Whether spatial differences in glial activation in Epg5 mutant mice or VS remains to be determined. The neurological phenotypes observed in Epg5 mutant mice probably arise from a combination of neuronal and glial dysfunction driven by impaired autophagy. Future studies using primary cell cultures from these mice, as well as human induced pluripotent stem cell models derived from patients with VS^63^, will be essential for dissecting the cell type-specific consequences associated with EPG5 loss. The development of new mouse and cellular models that allow for tissue-specific manipulation of EPG5 function and autophagy will be critical for determining the extent to which glial dysfunction contributes to disease progression in VS and other congenital disorders of autophagy. Ultimately these models will be critical to determine the molecular mechanisms that give rise to the complex cellular deficits seen in VS. These studies will have major implications into potential therapeutic targets.
In conclusion, this study presents a detailed analysis of novel mouse models of VS, including a strain harboring a pathogenic variant in Epg5 that mirrors a mutation identified in patients with VS. Our transcriptomic analysis of CNS tissues revealed a complex molecular signature of neuroinflammation. Widespread and progressive expansion of astrocytes and microglia was observed throughout the CNS in these mutant mice, and this glial activation correlated with age and the severity of motor dysfunction. The detailed analysis of these mouse models suggests a contributory role for glial cells in the pathogenesis and disease progression of VS. These studies validate the utility of these mouse models for study the progression of VS, and for use in therapeutic assessment for this devastating disorder, including the potential for targeting neuroglia as therapeutic strategy.
Supplementary information
Supplementary Information Supplementary Table 1. Summary of Epg5 mutant alleles. Supplementary Table 2. Thermo Fisher TaqMan gene expression assays used in this study. Supplementary Table 3. Overlapping DEGs between spinal cord and cerebellum in Epg5 mutant mice. Supplementary Movie 1. Ataxic locomotion associated with Epg5 W860X homozygous mutant mice.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Dafsari, H. S. et al. EPG 5-Related Disorder (2022). In Gene Reviews [Internet] (eds Adam, M. P. et al.) (University of Washington, Seattle, 1993-present).
- 2Bond, S., Saxena, S. & Sierra-Delgado, J. A. Microglia in ALS: insights into mechanisms and therapeutic potential. Cells 14, 421-439 (2025).
- 3Migliarini, S. et al. Microglia morphological changes in the motor cortex of h SOD 1(G 93A) transgenic ALS mice. Brain Sci.11, 807–815 (2021).