Hidden Chemical Tags on DNA Reveal Immune Chaos in Fragile X-Linked Brain Disease

Fragile X-associated tremor/ataxia syndrome, or FXTAS, has long been explained by a single, brutal mechanism: a stretched stretch of CGG repeats in the FMR1 gene produces toxic RNA that clogs the machinery of brain cells. But a new study of postmortem human brains suggests the story is considerably more tangled. By mapping the genome-wide pattern of DNA methylation, chemical tags that sit on cytosines and help decide which genes are quiet or active, researchers have uncovered a previously unrecognized epigenetic landscape in the FXTAS prefrontal cortex, one that converges on the immune system in an unexpected and directionally surprising way.

The research team, led by investigators at the Icahn School of Medicine at Mount Sinai and the University of California, Davis, analyzed postmortem prefrontal cortex tissue from 27 men with FXTAS and 29 male controls. All samples came from Brodmann area 9, the same region where the disorder’s molecular damage plays out in life. Female donors were deliberately excluded to avoid the confounding influence of X-chromosome inactivation, given that FMR1 itself resides on the X chromosome. Diagnosis in each FXTAS case was confirmed by the presence of the premutation allele and, in many cases, by the hallmark ubiquitin-positive intranuclear inclusions seen under the microscope after death.

Technically, the work was demanding. The team profiled methylation using two versions of the Illumina MethylationEPIC array, version one covering roughly 850,000 probes and version two roughly 900,000, and merged the datasets by retaining the 718,960 CpG probes common to both platforms. To guard against hidden confounders, from batch effects to differences in postmortem interval, they applied surrogate variable analysis, a statistical framework that estimates unmeasured sources of variation and adjusts for them while protecting disease status from being inadvertently scrubbed out of the data. As a sensitivity check, the entire analysis was repeated with methylation beta values converted to M values; effect sizes and adjusted p values from the two approaches correlated at roughly 0.98 and 0.95, respectively, indicating that the findings were not artifacts of the statistical parameterization.

Using a strict definition that required both a methylation difference greater than 0.10 on the beta scale and a false discovery rate below 0.05, the team identified 98 differentially methylated positions, 48 hypermethylated and 50 hypomethylated, mapping to 70 genes. That number may sound modest against the hundreds of thousands of probes tested, but individual CpG sites are noisy messengers, and the fact that any signal survived the thresholds at all is notable for a rare, late-onset disorder. Moreover, several genes carried multiple, directionally consistent changes, which boosted confidence in those loci.

The genes themselves read like a map of processes that neuroscientists already suspect in FXTAS. CYP2E1, a cytochrome P450 enzyme that is a major enzymatic source of reactive oxygen species in neurons and astrocytes, carried three hypomethylated probes in FXTAS. LINGO1, a negative regulator of myelination and oligodendrocyte differentiation and a known genetic risk factor for essential tremor, was hypomethylated, a finding that resonates with the prominent white matter disease of FXTAS. Hypomethylation also touched the zinc importer SLC39A4 and the cerebellar zinc-finger transcription factor ZIC1, pointing toward altered zinc homeostasis, a theme linked to mitochondrial dysfunction and antioxidant defense in neurodegeneration. Genes governing synaptic function, including ADD2 and SYT3, and chromatin regulators such as RCOR1 and PRDM16 appeared as well.

The single most striking immune-related finding was TRAF3, which contains one of the most statistically significant methylation changes in the entire dataset. TRAF3 acts as a brake on non-canonical NF-κB signaling and type I interferon responses, so hypermethylation at this locus, if it dampens expression, could loosen a critical restraint on inflammatory signaling. The directional pattern among other immune genes was equally intriguing: negative regulators such as TRAF3 and the lectin receptor CLEC1A were hypermethylated, while genes promoting myeloid activation and inflammatory effector function, including RHBDF2/iRhom2, which controls TNF-α shedding, the myeloid transcription factor MEIS1, and the phagocytosis regulator ELMO3, were hypomethylated. Notably, no large methylation changes appeared at the FMR1 locus itself, consistent with the premutation state, in which the promoter remains largely unmethylated unlike the full mutation seen in fragile X syndrome.

Pathway analyses gave biological texture to these gene-level hits, though the authors are careful about their limits. Reactome analysis highlighted themes of inflammatory resolution, including biosynthesis of maresin-like pro-resolving mediators, axonal growth inhibition through RHOA signaling, zinc influx via the SLC39 family, and CYP2E1 metabolism. None of these pathways survived correction for multiple testing, and transcription-factor motif enrichment and protein-protein interaction analyses likewise found no significant higher-order modules, although AP-1/bZIP family motifs, relevant to innate immune responses, ranked nominally at the top in both HOMER and JASPAR sequence analyses. Comparison with published brain methylation signatures of Alzheimer’s, Parkinson’s, and Huntington’s disease found no gene-level overlap, while seven of the FXTAS-associated genes did overlap with differentially expressed genes in an independent mouse model of the disorder.

The proteomic arm of the study delivered the most dramatic result. Using the Olink Inflammation panel, a proximity extension assay capable of quantifying 366 inflammation-related proteins, the team profiled prefrontal cortex from 9 FXTAS cases and 12 controls. Sixty proteins differed significantly between the groups, and 59 of them, essentially all of them, were less abundant in FXTAS. The depleted proteins included chemokines such as CXCL8 and CXCL10, interleukins including IL6, IL15, and IL18, core components of innate immune and NF-κB signaling such as TLR3, IRAK1, IRAK4, and IKBKG/NEMO, the classical complement component C1QA, and microglial receptors like SIGLEC1 and SIGLEC10. The team then showed through a series of technical checks that this near-uniform reduction was not explained by sequencing depth, normalization behavior, or internal control artifacts.

A reduced inflammatory protein signature in a neurodegenerative disease may seem paradoxical, and the authors do not interpret it as the absence of inflammation. Instead, they propose that the combined epigenetic and proteomic findings may reflect a chronically engaged but dysregulated or inadequately resolved immune state, in which the brakes and the accelerators of innate immunity are both malfunctioning. Supporting this view, an independent mass spectrometry proteomic study of FXTAS cortex from the same UC Davis brain repository had already reported reduced abundance of inflammation-associated proteins, and among the 19 proteins measured by both platforms, 13 showed concordant directionality, including IRAK1, IKBKG/NEMO, and PARP1. The picture is also unlikely to be uniform across the brain: elevated IL-12 and TNF-α have previously been reported in the FXTAS cerebellum, hinting at region-specific inflammatory profiles.

The study has important caveats that the authors openly acknowledge. The methylation data come from bulk tissue, so cell-type-specific effects, particularly glial inflammatory responses, cannot be directly resolved. The work is cross-sectional and cannot establish whether methylation changes precede or follow disease onset. The Olink cohort was small, the assay was restricted to inflammation-related proteins, and methylation changes cannot be assumed to map one-to-one onto gene expression or protein abundance, especially since many differentially methylated sites sat outside promoters. Still, the convergence is what makes this study compelling: methylation alterations in immune-regulatory genes such as TRAF3 sit alongside a proteomic signature of coordinated disruption in innate immune and NF-κB pathways, measured with entirely independent technology. As the first genome-wide methylation map of the FXTAS brain, the work reframes a disorder once seen purely as RNA toxicity, and it hands researchers a set of concrete, testable hypotheses linking CGG-repeat biology to oxidative stress, myelination, zinc homeostasis, and an immune system that appears both fired up and strangely depleted in the same tissue.

Subject of Research: Epigenetic and inflammatory molecular alterations in fragile X-associated tremor/ataxia syndrome brain tissue

Article Title: DNA methylation and proteomic profiling of postmortem brain tissue reveals epigenetic dysregulation and inflammatory signaling in fragile X-associated tremor/ataxia syndrome

Article References: Lozano, R., Lin, X., Hagerman, R., Martínez-Cerdeño, V., & Pinto, D. (2026). DNA methylation and proteomic profiling of postmortem brain tissue reveals epigenetic dysregulation and inflammatory signaling in fragile X-associated tremor/ataxia syndrome. Frontiers in Molecular Neuroscience, 19, Article 1932396. https://doi.org/10.3389/fnmol.2026.1932396

Image Credits: AI Generated

DOI: 10.3389/fnmol.2026.1932396

Keywords: FXTAS, FMR1, DNA methylation, epigenetics, NF-κB signaling, neuroinflammation, TRAF3, CYP2E1, LINGO1, proteomics, Olink, neurodegeneration