GFAP mutations selectively reshape lipids in Alexander disease cell model

Alexander disease-associated mutations in the astrocyte protein GFAP caused selective and extensive changes to cellular lipid composition in an experimental model, with the R239C variant producing the strongest response. Researchers engineered human U-87 MG astrocytoma cells to express fluorescently tagged wild-type GFAP or four disease-associated variants and used targeted mass spectrometry to measure 488 lipid species. The most pronounced changes involved membrane lipids, energy-storage compounds, sphingolipids and ether-linked phospholipids, while the extent and character of the response varied markedly among mutations. The findings identify disturbed lipid homeostasis as a possible component of the cellular pathology associated with mutant GFAP, although the experiments were conducted in a cultured cell line rather than people or animals with the disease.

Alexander disease is a rare neurodegenerative disorder classified as a leukodystrophy and caused mainly by new, or de novo, mutations in the gene encoding glial fibrillary acidic protein, or GFAP. GFAP is a type III intermediate filament protein expressed predominantly in astrocytes, whose functions include cell migration, proliferation, organelle positioning, maintenance of the blood-brain barrier and support of glutamatergic neurotransmission. Disease-associated mutations act predominantly through a toxic gain of function, disrupting filament assembly and promoting the formation of Rosenthal fibers, which are cytoplasmic aggregates rich in GFAP and other proteins and show signs of lipoxidative damage. The variants examined in the study—R79C, R239C, R239G and E373K—occur in different subdomains of the GFAP rod domain. R79C, R239C and E373K have been associated with early-onset disease, whereas R239G is associated with a later-onset form.

Before examining lipids, the researchers verified the expression and assembly behavior of the engineered GFAP proteins. Non-reducing immunoblotting detected high-molecular-weight oligomeric forms of wild-type and mutant GFAP, consistent with the altered aggregation previously reported for several of these variants. Confocal microscopy showed that wild-type GFAP formed an extended, regular filament network running from the nuclear periphery toward the cell membrane. The four mutants produced different arrangements. R79C generated a heterogeneous network containing bundles and scattered accumulations, while E373K caused pronounced perinuclear aggregates and almost complete loss of normal filaments. R239C produced a mixture of thick bundles, aggregates, filaments and areas of diffuse fluorescence that was suggestive of incompletely polymerized protein. R239G was less uniformly affected: approximately 70% of GFP-positive cells retained a wild-type-like network, while the remaining 30% displayed aggregates and bundle-like accumulations resembling the R239C phenotype. These observations confirmed that the mutations induced distinct filament phenotypes, whereas the wild-type construct retained the expected extended network.

The targeted lipidomic screen detected and quantified 488 species from a panel of 686, spanning major classes that included phospholipids, sphingolipids, glycerolipids, sterols, acylcarnitines and ubiquinone. Comparison with non-transfected cells showed that expressing wild-type GFP-GFAP itself was already associated with substantial remodeling: 131 lipid species changed significantly. Wild-type expression reduced acylcarnitines, glycerolipids and cholesteryl esters while increasing ether-linked phospholipids and most sphingolipids, with monohexosylceramide as the exception. Changes among canonical phospholipids were species-specific rather than uniformly directional. Because U-87 MG cells have undetectable endogenous GFAP, non-transfected cells were used as the control for this comparison rather than cells expressing GFP alone. The authors noted that the cell population was itself heterogeneous, but principal component analysis and hierarchical clustering consistently separated the groups across 12 biological replicates per condition, indicating a reproducible association between wild-type GFAP expression and lipid remodeling.

Each disease-associated mutation generated a distinct lipidomic signature when compared with cells expressing wild-type GFAP. R239C produced the largest and most consistent set of changes, followed by R79C and E373K, while R239G caused more moderate changes. Ceramides, for example, decreased globally in R239C but increased in R79C; neither R239G nor E373K showed a significant overall ceramide change. Oxidized phosphatidylethanolamines fell in R239C but rose in the other variants, and triglyceride changes were distinctive to R239C. Despite these differences, all four mutants increased ether-linked phospholipids and cholesteryl esters. Monohexosylceramide and dihexosylceramide fell in both R239C and R79C, but the broader hexosylceramide response was not uniform: trihexosylceramide increased in R239C. E373K shared the fewest features with the other mutants and reversed several glycerophospholipid changes seen with R239C, indicating that mutations linked to different clinical onsets cannot be treated as interchangeable at the molecular level.

Detailed analysis of R239C, the most disruptive variant, identified 221 significantly altered lipid species relative to wild-type GFAP, comprising 132 increases and 89 decreases. Ceramide metabolism and triglyceride metabolism were the most affected pathways, with 64–100% of measured species in the former and 62–70% in the latter showing changes. Ether-linked phosphatidylcholines, lysophosphatidylglycerol, trihexosylceramides, triglycerides and ether-linked triglycerides increased, while monohexosylceramides, dihexosylceramides and sulfatide decreased. Ten of 13 affected sphingomyelins increased, whereas 19 of 29 affected ceramides decreased and 10 increased. Most modulated phospholipids showed species-specific rather than uniform class-wide responses. For example, seven phosphatidylcholines increased and 13 decreased, while four phosphatidylethanolamines increased and two decreased. Key changes were reproduced in an independent experiment, although the data do not establish which biosynthetic, degradative or remodeling processes produced them.

The lipid shifts were accompanied by evidence of increased intracellular storage. Nile Red staining and live-cell confocal microscopy showed that the relative volume occupied by lipid droplets was significantly greater in R239C-expressing cells than in cells expressing wild-type GFAP. The quantification was based on 20 cells across two independent experiments per condition and compared Nile Red-positive volume with GFP-defined cell volume, excluding the nucleus. The result was consistent with the mass-spectrometry findings that 57 of 61 modulated triglycerides and four cholesteryl esters increased, together with rises in the acylcarnitines AC(16:1) and AC(18:1). The authors suggested that lipid-droplet accumulation might represent an adaptive response to stressed or impaired fatty-acid oxidation, but they did not directly test that mechanism. Their broader interpretation was that the combination of altered membrane organization, antioxidant-related lipid changes and disturbed bioenergetic lipid pools may contribute to astrocyte dysfunction, without establishing a causal sequence.

Calculated structural indices indicated that the R239C mutation did not change every lipid class in the same way. Its average chain length decreased across several glycerophospholipid classes, sphingolipid subclasses and cholesteryl esters, but increased in acylcarnitines, triglycerides and ether-linked triglycerides. The double-bond index fell in ether phosphatidylcholines, phosphatidylethanolamines, phosphatidylserines and cholesteryl esters, while rising in acylcarnitines, lysophospholipids, several sphingolipids and triglycerides. A fluidity index derived from chain length and unsaturation increased in numerous classes but decreased in ether phosphatidylcholines, phosphatidylserines and cholesteryl esters. Lipid diversity rose in acylcarnitines, dihydroceramides and glycosphingolipids but fell in lysophosphatidylcholines, ether phosphatidylcholines, phosphatidylethanolamines, sphingomyelins, triglycerides and cholesteryl esters, pointing to selective rather than uniform remodeling.

An independent fluorescent-probe assay supported the calculated prediction that R239C cells had greater overall membrane fluidity, measuring about 1.5 times the value obtained in wild-type GFAP cells across four experiments. The phosphatidylcholine-to-cholesterol and ether phosphatidylcholine-to-cholesterol ratios also increased, whereas free cholesterol itself did not differ significantly between the groups. This combination suggests that altered membrane fluidity was driven by changes in lipid classes beyond free cholesterol. Such shifts could affect membrane organization, lipid-raft properties, oxygen permeability and stress responses, but the study did not connect the measured biophysical changes to specific astrocyte functions or neuronal injury. A possible route is the disruption of the GFAP filament network, which could disturb interactions with vimentin, an intermediate filament protein involved in lipid-droplet regulation and organelle homeostasis, although this mechanism was not directly investigated.

Overall, the study demonstrates that Alexander disease-associated GFAP mutations can reshape the lipidome of astrocyte-like cells in mutation-specific and class-selective ways. The shared increase in ether-linked phospholipids and cholesteryl esters suggests a common response to GFAP dysfunction, while pronounced differences among R239C, R79C, R239G and E373K mirror the clinical diversity of the disorder. The results do not show that lipid changes cause Alexander disease in people, nor do they establish benefit from targeting lipids. Instead, they identify membrane architecture, bioenergetic metabolism, glycosphingolipid metabolism and cellular storage as areas for further study. The principal limitations are the use of an immortalized astrocytoma line, standard culture conditions that do not reproduce the central nervous system’s lipid environment or oxygen levels, and the absence of direct experiments on the mechanisms underlying the observed remodeling. Follow-up work in more physiological models would be needed to determine whether the signatures found here contribute to astrocyte injury in the intact disease.

Subject of Research: Lipid remodeling caused by Alexander disease-associated GFAP mutations in a human astrocytoma cell model

Article Title: GFAP mutants cause imbalance of selective lipid classes in a cellular model of Alexander disease

Article References: Fernàndez-Bernal, A., Goya-Iglesias, N., Company-Marin, I., Mota-Martorell, N., Pekna, M., Pekny, M., Sol, J., Obis, È., Pamplona, R., Pajares, M. A., Pérez-Sala, D., & Jové, M. (2026). GFAP mutants cause imbalance of selective lipid classes in a cellular model of Alexander disease. iScience, 29(11), Article 117903. https://doi.org/10.1016/j.isci.2026.117903

Image Credits: AI Generated

DOI: 10.1016/j.isci.2026.117903

Keywords: Alexander disease, GFAP, astrocytes, lipidomics, R239C mutation, membrane fluidity, sphingolipids, ether lipids, neurodegeneration, cellular model, mutants, cause