Science1 publisher2 min readPublished
Alzheimer's brain cells fold their genomes differently, Pitt and CMU researchers report
A Science paper measured genome folding and gene activity in the same single cells from postmortem prefrontal cortex, and its authors place chromatin architecture alongside amyloid and tau in the disease's molecular pathology.
The Scientist · Science desk
What happened
- A Science paper reports that the 3D organization of DNA is disrupted in several types of brain cells affected by Alzheimer's disease, altering how important genes are switched on and off.
- The work came from Carnegie Mellon's School of Computer Science, the University of Pittsburgh School of Medicine and the University of Washington, in a release Pitt issued on September 13, 2026.
- The tissue was postmortem prefrontal cortex from people with and without Alzheimer's who had joined a long-term dementia study and later donated their brains for research.
- The team also built Hicformer, a model that takes DNA sequence, broad folding patterns and DNA contact maps, and predicts gene activity across different cell types.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- constraint An end-state comparison of donated brains can rank cell types by how much their folding differs, but nothing in it dates that change against plaque or tangle formation. Causal claims wait on perturbation work.
- capability Because contacts and transcripts come from one cell, a candidate regulatory change arrives with a cell type attached. A screening team can pick the cell type in which to run its assay.
- decision Anyone designing a postmortem brain cohort now has a choice to make about tissue budget: expression data alone leaves the folding layer unmeasured, and contact assays plus spatial maps cost samples.
- precedent Model-inferred expression will start appearing in target papers as supporting evidence. The standard for accepting it should be whether someone perturbed a contact and watched the gene respond.
DNA does not sit inside a cell as a straight strand. It folds, and the folding helps determine which genes are accessible and active [5]. So a difference in folding is a difference in regulation, in principle. Measuring that difference inside one cell, rather than across a mixed piece of tissue, is the hard part.
GAGE-seq is what lets this work get at it. The technique reads gene expression and three-dimensional genome contacts within the same individual cell [7]. Those readouts were then combined with spatial transcriptomic maps, which preserve where in intact tissue the activity occurred [8]. The combination let the group tie genome organization to gene regulation and see where the Alzheimer's-related changes sat in the surrounding tissue [9].
Every sample was taken after death, so the comparison is between end states [6]. It can show that folding differs where plaques and tangles are found; it cannot order the two in time. Hansruedi Mathys, the Pitt neurobiologist who directed that arm of the study, said the results "establish higher-order chromatin alterations as a component of the molecular pathology associated with the disease" [12]. In the same sentence he put the American case count at seven million and rising [13]. Amyloid-beta plaques and tau tangles remain the best known features of the disease, and the paper's argument is that chromatin belongs on the list too [16].
Xinyue Lu, the doctoral student who co-led the work, described Hicformer as a computational test bed [11]. A model that infers expression from structure produces candidates for perturbation experiments. The release does not report one, or the number of donors, or the size of the folding differences.
For a group hunting targets, the usable output is a shortlist of cell types and genes where folding and expression differ together in Alzheimer's cortex [4]. Pitt frames that as a possible route toward new treatments [18]; whether any candidate on such a list is druggable is a separate question this data cannot answer. Jian Ma, who led and supervised the study, said "Alzheimer's disease cannot be understood one layer at a time" [14], and described the goal as moving "beyond cataloging disease-associated changes toward understanding how they fit together and which mechanisms to test next" [15].
What to watch
- Donor counts and effect sizes in the Science paper itself. Those numbers decide whether the folding differences separate individuals or only cohorts.
- A follow-up that perturbs one of the implicated DNA contacts in the implicated cell type and measures the gene. That result would move this past co-occurrence.
- Whether the pattern holds in a second cohort or a brain region other than prefrontal cortex.