If you Google Alzheimer's and memory loss, you'll find a lot of tidy diagrams suggesting memories just sort of evaporate - like someone left your hippocampus out in the sun too long. Neat story. Slight problem: the brain is messier than that. In a new mouse study, researchers suggest some old memories may not be fully gone at first - they may be getting suppressed, almost like the wrong cells keep hitting mute at exactly the worst moment.

That is a much weirder story. Also, frankly, a much more brain-like one.

The question: are old memories erased, or just harder to reach?

One of the cruel patterns in Alzheimer's disease is retrograde amnesia - trouble recalling things that happened in the past. Early on, this often follows a gradient: newer memories go first, older ones hang on longer. But as disease progresses, even those remote memories start slipping away.

Scientists have spent years studying how Alzheimer's affects the making of new memories. Less clear is what happens to remote memories - the older ones that are thought to rely more on cortical areas such as the medial prefrontal cortex, rather than just the hippocampus. Think of it as memory's long-term storage unit, except the unit is alive, electrically active, and staffed by cells with the energy of a group chat at 2 a.m.

In this study, van Adrichem and colleagues used APP/PS1 mice, a common Alzheimer's model, to ask a simple but sneaky question: when remote memories fade, what exactly is going wrong in the cortical circuits that are supposed to help retrieve them?
Source paper: van Adrichem et al., eLife (2025). DOI: 10.7554/eLife.106866

Meet the suspects: engram cells and the brain's professional buzzkills

A lot of memory research now revolves around engram cells - neurons that were active when a memory formed and can later be reactivated during recall. They're often described as the physical trace of a memory. Not the whole story, but definitely major cast members.

Then there are parvalbumin interneurons, or PV interneurons. These are fast-spiking inhibitory neurons. Their job is to keep nearby excitatory neurons from getting too rowdy. Every brain needs this. Without inhibition, neural circuits would behave like a wedding DJ who believes every song should play at once.

But balance matters. Too little inhibition is bad. Too much inhibition can also wreck the signal.

The team found that as APP/PS1 mice aged, their remote memory got worse - and this decline lined up with hyperexcitability in PV interneurons in the medial prefrontal cortex. In other words, the cells responsible for damping neural activity were themselves getting unusually excitable. Which is a very on-brand brain paradox.

The twist: the memory cells were still there

Here's the interesting part. The researchers looked at Fos expression, a marker often used to identify recently active neurons, to see whether memory-encoding cells were failing to reactivate during recall. You might expect the old memory trace to have simply fallen apart.

That was not the main story.

They did not find the remote memory deficit was mirrored by big changes in engram-cell reactivation or PV cell reactivation in the medial prefrontal cortex. Instead, they found something subtler: engram cells in APP/PS1 mice were receiving stronger inhibitory input than non-engram cells.

So the problem may not be that the memory trace vanished. It may be that the neurons carrying that trace are being selectively over-inhibited - like trying to hold a conversation while someone keeps lowering your microphone.

That matters, because it shifts the framing. Some memory loss in Alzheimer's might come not only from destroyed circuits, but from dysregulated cortical microcircuits that make stored information harder to access.

Why this is a big deal, even though it's in mice

Mouse studies are not people. Mice do not forget where they parked, mostly because they are not cursed with parking garages. Still, this work points to a compelling mechanism for the gradual loss of remote memories in Alzheimer's.

It also fits with a broader idea in neuroscience: cognitive symptoms can emerge from network imbalance, not just cell death alone. The brain isn't a filing cabinet. It's a live electrical system. If inhibitory cells start over-controlling the wrong targets, memories could become less retrievable even before they are fully degraded.

That opens interesting doors. If future work shows a similar mechanism in humans, treatments might aim to restore the excitation-inhibition balance in specific cortical memory circuits, rather than treating memory loss as a one-way deletion event. That's still a big "if," to be clear. Neuroscience has a long and glorious tradition of finding something elegant in mice and then discovering humans prefer to be complicated.

What this study is answering - and what it isn't

This paper tackles a genuine gap: why remote memories decline progressively in Alzheimer's models. It suggests cortical inhibitory circuits, especially PV interneuron input onto engram cells, may play a bigger role than expected.

What it does not show is that human Alzheimer's memory loss can be reversed by flipping one neat cellular switch. The disease involves amyloid, tau, inflammation, vascular factors, synaptic dysfunction, and enough circuit-level chaos to keep neuroscientists employed until the sun burns out.

Still, this study adds a sharp piece to the puzzle. It says old memories may not simply disappear on schedule. Sometimes they may still be sitting there, while the surrounding circuit has become a little too eager to tell them to be quiet.

And honestly, if that isn't the most frustratingly plausible brain problem imaginable, I don't know what is.

References

1. van Adrichem JJV, van der Loo RJ, Ambrosini Defendi R, Smit AB, van den Oever MC, van Kesteren RE. Progressive remote memory decline coincides with parvalbumin interneuron hyperexcitability and enhanced inhibition of cortical engram cells in a mouse model of Alzheimer's disease. eLife. 2025;14:RP106866. doi: 10.7554/eLife.106866

2. Roy DS, Arons A, Mitchell TI, Pignatelli M, Ryan TJ, Tonegawa S. Memory retrieval by activating engram cells in mouse models of early Alzheimer's disease. Nature. 2016;531(7595):508-512. doi: 10.1038/nature17172

3. Tiwari SS, Mizuno K, Ghosh A, et al. Alzheimer's disease-related decline in spatial memory is associated with impaired coordination of hippocampal-prefrontal oscillations. J Neurosci. 2018;38(36):7789-7805. doi: 10.1523/JNEUROSCI.0582-18.2018

4. Palop JJ, Mucke L. Network abnormalities and interneuron dysfunction in Alzheimer disease. Nat Rev Neurosci. 2016;17(12):777-792. doi: 10.1038/nrn.2016.141

5. Hijazi S, Heistek TS, Scheltens P, Neumann U, Shimshek DR, Mansvelder HD, Smit AB, van Kesteren RE. Early restoration of parvalbumin interneuron function improves hippocampal network dynamics and memory in a mouse model of Alzheimer's disease. Mol Psychiatry. 2020;25(12):3380-3398. doi: 10.1038/s41380-019-0483-4

Disclaimer: The image accompanying this article is for illustrative purposes only and does not depict actual experimental results, data, or biological mechanisms.