How the Brain Removes Its Own Connections: Part Two: The Eat-Me Signal

Abstract

Part One established that microglia remove synapses as a normal part of how circuits are built. That raises an obvious question the field could not answer for years: how does a microglial cell know which synapse to remove?

A 2020 paper in The EMBO Journal identified the missing piece. The answer is a lipid, normally kept on the inside of a cell membrane, that gets flipped to the outside — and it works as a message meaning eat me.

This is the most mechanistically satisfying result in the series, and it is entirely in mice and cultured cells.

1.The Gap in the Story

By 2020 the microglial side of the process was reasonably well described. The microglial receptor TREM2 and the soluble complement proteins C1q and C3 were recognised as key players — the machinery that recognises and engulfs.

What was missing was the neuronal side. As Natalie Scott-Hewitt, Francesca Perrucci and colleagues put it, the neuronal molecular components that specify synapses to be eliminated were still undefined.

Without that, the mechanism was half a story. Complement can tag a synapse, but something has to determine which synapses get tagged, or the process is indiscriminate.

2.Phosphatidylserine

Phosphatidylserine is a phospholipid, and in a healthy cell it is held on the inner leaflet of the membrane — facing inward. Its appearance on the outer surface is one of biology's oldest signals: it is how a dying cell announces itself to phagocytes.

The finding here is that this same signal is used locally, at synapses, in a healthy developing brain. Exposed phosphatidylserine represents a neuronal eat-me signal involved in microglial-mediated pruning.

The elegance of that is worth stating. A signal that normally means "this whole cell is dying" is deployed at a sub-cellular scale to mean "this one connection is surplus" — without the neuron dying at all.

3.How They Showed It

The evidence comes in three layers, and the layering is what makes it convincing.

In co-cultures of hippocampal neurons and microglia, synapse elimination could be partially prevented in two independent ways: by blocking access to exposed phosphatidylserine using Annexin V, a protein that binds it, or by removing TREM2 from the microglia. Blocking either end of the interaction reduced pruning.

In living animals, phosphatidylserine exposure was observed at both hippocampal and retinogeniculate synapses, and microglia were seen engulfing phosphatidylserine-labelled material — during the established developmental periods when microglial pruning is known to occur. The timing matched.

And in mice lacking C1q, which are known to fail at properly refining retinogeniculate connections, presynaptic phosphatidylserine exposure was elevated while engulfment by microglia was reduced. The tags accumulated because nothing was collecting them.

That last experiment is the strongest of the three, because it connects a pre-existing genetic phenotype to the newly proposed mechanism rather than merely observing a correlation.

4.What It Explains

The authors note that developmentally regulated phosphatidylserine exposure appears common among developing brain structures — hippocampus and visual system both — which suggests a general rule rather than a local quirk.

It also supplies a plausible account of specificity. If exposure is regulated locally, at individual synapses, then activity could determine which connections expose the signal and which do not. The long-standing principle that connections which fail to fire together get eliminated acquires a physical implementation.

The paper does not demonstrate that activity controls the flip. It establishes the tag, and identifies what reads it.

5.The Distance to a Person

Hippocampal co-cultures, retinogeniculate connections in developing mice, and a C1q knockout line. Nothing here is human, and the developmental windows involved are early.

The relevance to an adult brain is an inference rather than a finding. Pruning continues into adulthood — Part One covers that — but whether the same phosphatidylserine mechanism operates in a fifty-year-old hippocampus has not been shown.

One further caution belongs here because of what phosphatidylserine also is. It is sold as a supplement, derived from soy or sunflower lecithin, marketed for memory. Nothing in this paper bears on that. Oral phosphatidylserine is a dietary lipid entering general metabolism; the mechanism described here is about where a lipid already present in a neuronal membrane is positioned. These are not the same subject, and the shared name is the only connection.

Editorial Comment

MindHeaven® does not use phosphatidylserine and makes no claim about it, about synaptic pruning, or about microglia.

We flag the naming coincidence explicitly because it is exactly the kind of overlap that produces a plausible-sounding sentence with nothing behind it. A supplement sharing a name with a molecule in a mechanism is not participating in that mechanism.

Part Three complicates the picture established so far. A 2022 study in a mouse model of tauopathy found that the cell doing most of the eating, for one class of synapse, is not a microglial cell at all.