A Tiny Enzyme Flaw May Explain How Dementia Begins

A tiny structural flaw in a brain-protective enzyme may reveal how dementia begins. Credit: Shutterstock

Scientists have identified a small flaw in the enzyme GPX4 that prevents neurons from defending themselves. This mutation, found in children with a rare form of early dementia, disrupts a tiny loop the enzyme uses to protect cell membranes.

Lab and animal studies confirmed widespread neuron loss when GPX4 fails. The results suggest a deeper connection between this process and other types of dementia.

Why Neurons Die in Dementia

Why do neurons die in dementia – and can this process be slowed down? An international research team led by Prof. Marcus Conrad, Director of the Institute of Metabolism and Cell Death at Helmholtz Munich and Chair of Translational Redox Biology at the Technical University of Munich (TUM), reports in Cell how neurons use a built-in system to protect themselves from ferroptotic cell death.

Central to this protection is the selenoenzyme glutathione peroxidase 4 (GPX4). A single mutation in the gene responsible for GPX4 disrupts a previously unknown part of the enzyme’s function. In children who inherit this mutation, the result is severe early-onset dementia. Under normal conditions, GPX4 inserts a short protein loop – described as a “fin” – into the inner side of the neuronal membrane.

This allows the enzyme to neutralize lipid peroxides, harmful molecules that would otherwise damage the cell.

How the GPX4 “Fin” Shields Neurons

“GPX4 is a bit like a surfboard,” says Conrad. “With its fin immersed into the cell membrane, it rides along the inner surface and swiftly detoxifies lipid peroxides as it goes.” In children with early-onset dementia, a point mutation alters this fin-like loop. The changed structure prevents GPX4 from anchoring into the membrane, leaving lipid peroxides to accumulate. When these molecules build up, they weaken the membrane, trigger ferroptosis, and ultimately cause neurons to rupture and die.

The project began with three children in the United States who have an extremely rare form of early childhood dementia. All three share the same mutation in the GPX4 gene, known as R152H. Using cells from one affected child, the team reprogrammed the samples into a stem-cell-like state. These stem cells were then used to create cortical neurons and three-dimensional structures resembling early brain tissue, known as brain organoids.

Evidence From Mouse Models and Protein Analysis

To see how the mutation affects a whole organism, the researchers introduced the R152H variant into a mouse model. This allowed them to alter GPX4 function specifically in different types of nerve cells.

As GPX4 became impaired, the mice slowly developed serious motor problems, experienced neuron loss in the cerebral cortex and cerebellum, and showed strong neuroinflammatory reactions. These features closely matched what was seen in the affected children and resembled patterns typical of neurodegenerative diseases.

At the same time, the team examined changes in protein levels in the experimental model. Many proteins that increase or decrease in Alzheimer’s disease showed similar changes in mice lacking functional GPX4. This overlap suggests that ferroptotic stress may contribute not only to this rare childhood disorder, but also to more widespread forms of dementia.

Rethinking the Roots of Dementia

“Our data indicate that ferroptosis can be a driving force behind neuronal death – not just a side effect,” says Dr. Svenja Lorenz, one of the first authors of the study. “Until now, dementia research has often focused on protein deposits in the brain, so-called amyloid ß plaques. We are now putting more emphasis on the damage to cell membranes that sets this degeneration in motion in the first place.”

Early tests show that blocking ferroptosis can slow the cell death caused by loss of GPX4 in both cell cultures and mice. “This is an important proof of principle, but it is not yet a therapy,” says Dr. Tobias Seibt, nephrologist at LMU University Hospital Munich and co-first author. Adam Wahida, also a first author of the study, adds: “In the long term, we can imagine genetic or molecular strategies to stabilize this protective system.

For now, however, our work clearly remains in the realm of basic research.”

Long-Term Collaboration Reveals a Key Molecular Detail

The findings come from a research network built over many years, combining genetics, structural biology, stem cell research, and neuroscience across multiple international sites.

“It has taken us almost 14 years to link a yet-unrecognized small structural element of a single enzyme to a severe human disease,” says Marcus Conrad. “Projects like this vividly demonstrate why we need long-term funding for basic research and international multidisciplinary teams if we are to truly understand complex diseases such as dementia and other neurodegenerative disease conditions.”

Reference: “A fin-loop-like structure in GPX4 underlies neuroprotection from ferroptosis” by Svenja M. Lorenz, Adam Wahida, Mark J. Bostock, Tobias Seibt, André Santos Dias Mourão, Anastasia Levkina, Dietrich Trümbach, Mohamed Soudy, David Emler, Nicola Rothammer, Marcel S. Woo, Jana K.

Sonner, Mariia Novikova, Bernhard Henkelmann, Maceler Aldrovandi, Daniel F. Kaemena, Eikan Mishima, Perrine Vermonden, Zhi Zong, Deng Cheng, Toshitaka Nakamura, Junya Ito, Sebastian Doll, Bettina Proneth, Erika Bürkle, Francesca Rizzollo, Abril Escamilla Ayala, Valeria Napolitano, Marta Kolonko-Adamska, Stefan Gaussmann, Juliane Merl-Pham, Stefanie Hauck, Anna Pertek, Tanja Orschmann, Emily van San, Tom Vanden Berghe, Daniela Hass, Adriano Maida, Joris M.

Frenz, Lohans Pedrera, Amalia Dolga, Markus Kraiger, Martin Hrabé de Angelis, Helmut Fuchs, Gregor Ebert, Jerica Lenberg, Jennifer Friedman, Carolin Scale, Patrizia Agostinis, Annemarie Zimprich, Daniela Vogt-Weisenhorn, Lillian Garrett, Sabine M. Hölter, Wolfgang Wurst, Enrico Glaab, Jan Lewerenz, Bastian Popper, Christian Sieben, Petra Steinacker, Hans Zischka, Ana J. Garcia-Saez, Anna Tietze, Sanath Kumar Ramesh, Scott Ayton, Michelle Vincendeau, Manuel A. Friese, Kristen Wigby, Michael Sattler, Matthias Mann, Irina Ingold, Ashok Kumar Jayavelu, Grzegorz M.

Popowicz and Marcus Conrad, 4 December 2025, Cell.

DOI: 10.1016/j.cell.2025.11.014

Never miss a breakthrough: Join the SciTechDaily newsletter.

Follow us on Google and Google News.