Alzheimer's model mice engineered to report on mitochondrial recycling develop large deposits of acidic and neutral mitochondria inside neuronal processes, structures the authors name mitochondrial plaques, according to a paper by Xiuli Dan, Deborah Croteau, Paul Robbins, Vilhelm Bohr and colleagues published online in Nature Neuroscience on 29 July 2026. The team reports that the plaques arise from an abnormal build-up of mitochondria, that lysosomes are recruited late, and that the resulting clearance (mitophagy) is incomplete because lysosomal function is impaired. The abstract states the plaques frequently co-develop with amyloid to form mixed plaques, but can also emerge on their own at early stages of disease. The authors describe them as a new pathological entity in Alzheimer's, distinct from the two long-recognised hallmarks, extracellular amyloid plaques and neurofibrillary tangles.

The core work is animal work. The main model was a transgenic mouse carrying APP and PSEN1 mutations crossed with a pH sensitive mitochondrial reporter, mt-Keima, which glows differently inside acidic compartments; the authors report the structures in cortex and hippocampus of these mice but not in wild-type reporter mice. In the earlier preprint version of the study, posted to bioRxiv on 20 February 2025, measurements came from 844 cortical and 229 hippocampal structures in six Alzheimer's model mice aged 50 to 60 weeks, and three-dimensional reconstruction put the average spatial overlap between the mitochondrial structures and amyloid plaques in mouse brain at 7.3 per cent. The published paper reports that the structures were also found in a second, independent mouse model, 5xFAD, and in post-mortem human Alzheimer's brain. The human comparison in the preprint version was small: hippocampal tissue from four people with Alzheimer's and four controls, with the mitochondrial aggregates largely overlapping lysosomal markers in the Alzheimer's cases and absent from controls.

The finding is being read against the clinical record of anti-amyloid antibodies, which lower amyloid plaque burden but produce modest cognitive effects. If a second, largely separate lesion sits inside neurons, the authors argue, removing amyloid would leave it in place. That inference is not tested in the paper: the study measures pathology in mice and in autopsy tissue, not treatment response in patients, and it does not show that the structures cause cognitive decline.

The authors themselves stop short of a timeline. Robbins told Newsweek it is too early to say how much earlier the mitochondrial structures appear than amyloid, and that settling the sequence will require long-running animal studies and human tissue spanning different disease stages. Laura Bojarskaite, a neuroscientist at the University of Oslo who was not involved in the work, said the results need independent replication, confirmation in living people and evidence that they track disease course, and that studies like this should be read as promising rather than practice changing. The team says its next steps are to look for biomarkers of the structures and to screen for compounds that prevent them accumulating.