How the Brain Removes Its Own Connections: Part Three: Not Only Microglia

Abstract

Two parts of this series have been about microglia. A 2022 paper in Nature Aging shows that for one class of synapse, in a mouse model of tauopathy, the cell doing most of the eating is an astrocyte.

It also shows a division of labour nobody predicted: astrocytic lysosomes contained relatively more excitatory synapse material, microglial lysosomes more inhibitory.

Both processes ran through the same complement protein, and deleting it reduced both. This part is about what that means for a field that had settled on a single cell type.

1.The Model

Borislav Dejanovic, Morgan Sheng, Jesse Hanson and colleagues at Genentech and the Broad Institute worked in Tau P301S mice — animals carrying a mutant human tau gene that develop tau pathology and synapse loss with age.

This is a model of tauopathy rather than of Alzheimer's disease as a whole. It reproduces one arm of the pathology, which is a strength for mechanism and a limit for translation.

The method was integrative multi-omics applied to biochemically isolated synapse fractions — proteins present at synapses, measured directly, across ages and genotypes.

2.The Finding

Astrocytic and microglial proteins were increased in Tau P301S synapse fractions with age, and this increase was C1q-dependent.

In addition to microglia, the authors identified that astrocytes contribute substantially to synapse elimination in Tau P301S hippocampi. That is the sentence that reframes the topic.

Then the division of labour: relatively more excitatory synapse marker proteins were found in astrocytic lysosomes, whereas microglial lysosomes contained more inhibitory synapse material.

Two different cell types, preferentially consuming two different classes of connection, in the same tissue at the same time. Neither had been assigned that role.

Deleting C1q reduced astrocyte–synapse association and decreased both astrocytic and microglial synapse engulfment — establishing that the same complement protein governs both arms.

3.Why the Division of Labour Matters

Excitatory and inhibitory synapses are not interchangeable. The balance between them determines whether a circuit is stable, and shifting it in either direction produces dysfunction — including, at the extreme, seizures.

If one cell type preferentially removes excitatory connections and another preferentially removes inhibitory ones, then the net effect on a circuit depends on which cell is more active. A single measure of "synapse loss" conceals that entirely.

It also complicates any therapeutic reasoning. Blocking complement would reduce both arms, which sounds desirable until one asks what happens to a network that stops pruning either class.

4.Reading Astrocytes Into the Story

Astrocytes have appeared in this library before, as the cells that supply neurons with metabolic substrate. That they also consume synapses is a different function entirely, and it fits a pattern this field keeps producing: cells assumed to be supporting infrastructure turn out to be doing structural work on the circuit.

It also means the popular shorthand — that the brain's immune cells prune its connections — is at best incomplete. Astrocytes are not immune cells.

The complement protein C1q is the common factor. It is doing the tagging in both cases, which suggests the tag is general and the readers of that tag are several.

5.What the Study Does Not Establish

It is a mouse model carrying a human mutation that causes a rare inherited tauopathy. Most Alzheimer's disease is not that.

The multi-omics approach measures protein content in isolated fractions and in lysosomes. It shows what material is where; it does not directly show a cell removing a functioning synapse in real time.

And the direction of causation between tau pathology, complement activation and synapse loss is not settled by this design. The three occur together and C1q deletion reduces the third, which is a strong link, not a full account.

The commercial context should be stated because it is relevant to what gets studied. Several authors were full-time employees of Genentech during the study, and the senior author reports scientific advisory roles and a co-founding position across several neuroscience therapeutics companies. Complement inhibition is an active drug-development area; this is high-quality work produced inside that interest.

Editorial Comment

MindHeaven® makes no claim relating to complement, astrocytes, microglia or synapse elimination, and nothing in our formulations is proposed to influence them.

The reason this part is in the series is a general one. The story had a satisfying shape — immune cells sculpt the brain — and the shape was too simple. When a well-designed study looked with a method that did not presuppose which cell was responsible, a second cell type appeared, doing the opposite half of the job.

Part Four closes the series with the only study of the four conducted on human-derived cells, and with what that does and does not buy.