Picture an aggressive case of food coma. Microglia that noshed on tau-laden neurons soon regurgitated the aggregates, lost their appetites, and sank into a nefarious slumber, Maria Grazia Spillantini and Aviva Tolkovsky of the University of Cambridge, U.K., reported October 22 in Science Advances. Only when microglia consumed neurons riddled with tau tangles—not tau fibrils alone—did the cells undergo this transformation. The scientists found evidence of similar phenomena in the brains of people who had died with tauopathies, and of mice that overexpressed tau. They propose that dysfunctional microglia are part of a vicious cycle that exacerbates tauopathies.

  • After ingesting neurons containing tau aggregates, microglia spat out tau seeds.
  • They also stopped phagocytosis.
  • The microglia took on a senescent phenotype, including expression of metalloprotease.

Previously, Spillantini had found that, when overloaded with tau aggregates, neurons flipped lipids in their cell membranes, exposing phosphatidyl serine to the extracellular space. This inversion is a sign of impending apoptosis. It baited microglia to devour the neurons, tau aggregates and all, in a phagocytic binge (Aug 2018 news). 

How did these microglia fare, postprandially? First author Jack Brelstaff and colleagues tackled this question in the current study. They started by co-culturing dorsal root ganglia neurons from 5-month-old P301S-tau mice with microglia from wild-type neonatal mice. As they had reported previously, microglia readily gobbled up the tau-ridden neurons. Curiously, the researchers also detected tau aggregates floating in the culture medium, but only when microglia were present. This hinted that tau was being released from the microglia, not from the P301S neurons. Indeed, after Brelstaff separated microglia from their neuronal prey, the glia continued to spew out tau aggregates for at least four more days.

Importantly, the tau the microglia released was an insoluble, fibrillar form that seeded aggregation of normal, soluble tau in biosensor cell lines.

Had bits of the tau-laden neurons made the microglia sick? It seems so, because the cells lost their appetites. When offered more P301S neurons, or latex beads, a commonly used phagocytotic bait, the cells did not imbibe. Oddly, neither tau aggregates nor phosphatidyl serine liposomes alone pushed microglia into this hypophagic state, suggesting that taking bites out of a tau-bearing neuron was required to sate-and-sedate the microglia. The researchers found evidence of a similar phenomenon in mouse brain: microglia in cortical slices from P301S mice consumed fewer latex beads than did microglia in slices from wild-type animals (see image below).

What factors within the neurons cause microglia to spew tau seeds and shut down phagocytosis? This is the subject of ongoing work, Spillantini and Tolkovsky said.

None for Me, Thanks. When offered a helping of latex beads (red), microglia (green) in tissue slices from C57BL6 mice (left) chow down (yellow, co-localization). Microglia from P301S mice (right) take a pass. [Courtesy of Brelstaff et al., Science Advances, 2021.]

Because the microglia shut down phagocytosis, the researchers wondered if the cells might be senescent. Indeed, these microglia tripled production of β-galactosidase, a biomarker of senescence, compared to microglia that mingled with wild-type neurons. Interestingly, this senescence marker also appeared in naïve microglia exposed to the medium from P301S neuron/microglia co-cultures, but not medium from P301S neurons alone. This suggested that a soluble factor released by microglia that had internalized P301S neurons was able to instigate a senescent phenotype in other microglia in a paracrine fashion.

Senescent glial cells crank out damaging cytokines and proteases, and their accumulation has been implicated in neuronal death in tauopathy models (Streit et al., 2009Sep 2018 news). To see if these tau-spewing microglia also release other toxic cargo, the researchers surveyed the cell medium. They found that microglia co-cultured with P301S neurons secreted a unique profile of proteins. Eleven were distinct from those cranked out by microglia exposed to lipopolysaccharide or to wild-type neurons. These included two matrix metalloproteinases, MMP3 and MMP9; three chemokines, CXCL2, CXCL1, and CCL5; and IGFBP3, a growth factor previously pegged as part of the senescence-associated secretory phenotype.

MMP3 stood out as most elevated, so the researchers focused on its regulation. They ultimately found that MMP3 expression depended on NfκB. Notably, MMP3 could be induced in microglia-neuron co-cultures even if the authors blocked phagocytosis, suggesting that a soluble factor in the co-cultures triggered the senescent phenotype.

Did this microglial phenotype emerge in the tauopathy-ravaged brain? Using MMP3 as a proxy, the researchers examined wild-type mice and 2-month-old P301S mice that had not yet developed tau pathology. They detected both the inactive, zymogen form of MMP3, as well as the active, mature form, in both. However, in 5-month-old P301S, which had rampant tau aggregates, levels of active protease had tripled.

The scientists also found that people with frontotemporal dementia caused by tau mutations, Pick’s disease, or progressive supranuclear palsy had up to fourfold more MMP3 in their brains at postmortem. Active MMP3 was also elevated in one person with FTD caused by the C9ORF72 mutation, which causes accumulation of TDP-43 rather than tau. 

Slide Into Slumber. This model proposes that microglia become hypophagic and senescent after feasting on neurons with tau aggregates in them. The microglia then “vomit” tau seeds, which spread pathology, and cytokines, which lull other microglia into senescence. [Courtesy of Brelstaff et al., Science Advances, 2021.]

In all, the findings suggest that when microglia ingest tau aggregates, they become unwitting sowers of tau seeds in the brain. At the same time, the cells shirk phagocytosis and drive senescence in adjacent microglia, feeding a vicious cycle that could exacerbate tauopathy (see image above).

“The study will shed new light on our understanding of tau propagation in glia and non-synaptic mechanisms,” commented Tsuneya Ikezu of the Mayo Clinic in Jacksonville, Florida. Ikezu’s group previously proposed that after engulfing tangle-containing neurons, microglia help disseminate tau by packaging it into extracellular vesicles (Delpech et al., 2019). Ikezu added that his lab also observed secretion of free-form tau aggregates by microglia.

Diego Gómez-Nicola of the University of Southampton, U.K., noted that it remains unclear exactly how the neurons change the microglia, and how those changes are related to each other. “Is senescence triggered first, leading to defective phagocytosis, or is indigestion caused by over phagocytosis of tau leading to senescence,” Gómez-Nicola asked?

The new findings validate previous studies implicating tau in induction of microglial senescence (Bussian et al., 2021). Aβ, too, has been tied to microglial senescence as well as to the concept of “frustrated phagocytosis.” Gómez-Nicola said the new tau findings jibe with evidence from his group that Aβ pathology both invokes and is exacerbated by microglial senescence (Jun 2021 news). 

The study highlights that sleepy microglia are capable of causing trouble, noted Kiran Bhaskar and Gary Rosenberg of University of New Mexico, Albuquerque. “Despite adopting two functionally deficient phenotypes—hypophagy and senescence—the microglia still activate NFκB and secrete tau seeds,” they wrote. They recently reported that tau tangles switch on Nfκb in microglia (Jiang et al., 2021).—Jessica Shugart


  1. This interesting paper is novel in discovering the change of microglia after phagocytosis of pathogenic tau into a senescent phenotype, and in demonstrating their role in seeding tau aggregates after phagocytosis of tau-containing neurons. The study will shed new light for our understanding of tau propagation in glia and non-synaptic mechanism.

    We have previously published a hypothesis proposing a role of microglia for disseminating tau-containing extracellular vesicles after phagocytosis of tau-containing neurons (Delpech et al., 2019, Figure 2). We also observed secretion of tau aggregates as free form after microglial phagocytosis.

    In terms of microglial senescence, it is difficult to distinguish inflammatory activation and senescence, since activated microglia are both phagocytic and inflammatory and are known to secrete inflammatory molecules including MMPs. More relevant markers, such as expression of p16/CDKN2A or DNA damage, are more convincing for the evaluation of cellular senescence.


    . Neuroimmune Crosstalk through Extracellular Vesicles in Health and Disease. Trends Neurosci. 2019 May;42(5):361-372. Epub 2019 Mar 26 PubMed.

  2. This article nicely links the phagocytosis of pathological tau material with the engagement of microglia in a senescence-like phenotype, which is informative to understand the long-term consequences of microglia being exposed to tau.

    A number of experiments are performed in vitro. One should be cautious about extrapolating in vitro mechanisms to the in vivo condition. It is well established that cultured microglia differ quite significantly from the microglia that we can find in the living brain, in terms of phenotype, motility, as well as phagocytic capacity.

    The in vitro paradigm is also challenging when understanding, and defining, what a “live” neuron is, versus what a “dead” neuron is, versus all the shades of gray in between. The data presented in this article indicates neurons are expressing PtdSer on their membranes; that should be seen as an indication of a compromised neuron (let’s say, in critical care), and not a healthy neuron. I believe this is an important consideration when studying whether microglia can eat alive, healthy neurons; so far the literature indicates that that is a rather rare phenomenon.

    The data indicating the seeding ability of the tau secreted by microglia is interesting, and serves a strong independent validation of the data reported in Asai et al. (2015). Checking for presence of exosomes containing tau would be a useful follow-up.

    The data generated in slice cultures suggests that microglia have an impaired phagocytic capacity, which, in turn, coexists with a senescent phenotype. These results validate findings reported in Bussian et al. (2018). Collectively, they suggest tau is strongly linked to an induction of senescence in microglia, and add to the evidence from our group indicating that Aβ pathology also is associated with microglial senescence (Hu et al., 2021). 

    The link of tau-dependent microglial senescence and reduced phagocytosis still lacks a direct mechanistic experiment differentiating association from causation: Is senescence triggered first and lead to defective phagocytosis, or is the indigestion caused by ove- phagocytosis of tau, leading to senescence? How widespread is this phenotype within the microglial population? Does it depend on the specific tau species/model?

    In sum, clearly the study of microglial senescence in Alzheimer’s pathology is offering new insights on how these cells interact with the pathological components of the disease, and generate insight into the long-term consequences of chronic microglial activation.


    . Depletion of microglia and inhibition of exosome synthesis halt tau propagation. Nat Neurosci. 2015 Nov;18(11):1584-93. Epub 2015 Oct 5 PubMed.

    . Clearance of senescent glial cells prevents tau-dependent pathology and cognitive decline. Nature. 2018 Oct;562(7728):578-582. Epub 2018 Sep 19 PubMed.

    . Replicative senescence dictates the emergence of disease-associated microglia and contributes to Aβ pathology. Cell Rep. 2021 Jun 8;35(10):109228. PubMed.

  3. This is a very interesting paper. The model system is quite novel, the experiments are rigorous. The senescence onset and hypophagocytic phenotype of 5M P301L DRGn engulfed microglia is compelling. That these microglia are still capable of activating NFκB and actively secreting tau seeds, despite undergoing two functionally deficient phenotypes (senescence and hypophagocytic), is very perplexing. We have recently (Jiang et al., 2021) observed that neurofibrillary tangles derived from human AD brains are capable of increasing the expression of various members of the NFκB family, including NFKB1A, in human primary microglia, and of inducing NFκB activation via an MyD88-dependent pathway. This nicely complements the observations made by here.

    Related to the cytokine array analyses, it would be interesting to use genome-wide RNA-Seq in post-DRGn microglia (besides assaying for selected proteins) to determine if they show other functional/canonical pathway(s) related to innate immunity etc., and if any of them belong to DAM subtypes.

    Furthermore, there are some interesting trends in the datasets, which may need further investigation or clarification. For example, the LDH assay did not show evidence for cell death of tau aggregate-containing DRGns, suggesting that tau seeds released were not due to death of tau aggregate-containing DRGns. However, annexin V treatment prevented the death of these neurons. That means it is possible that some amount of tau is still released due to cell death. The  importance of MMP-3 in the senescent microglia could explain their role in the chronic inflammation seen in dementia, where MMP-3 appears in the cerebrospinal fluid and brain tissue.

    Overall, this study nicely extends our understanding of the phenotype of microglia upon engulfment of tau containing neurons.

    — Gary Rosenberg is a co-author of this comment.


    . Proteopathic tau primes and activates interleukin-1β via myeloid-cell-specific MyD88- and NLRP3-ASC-inflammasome pathway. Cell Rep. 2021 Sep 21;36(12):109720. PubMed.

  4. Evidence from the past years showed that microglia participate in tau spread (Asai et al., 2015), and that senescent glia cells are implicated in tau pathology (Bussian et al., 2018). This new study expands our knowledge on this topic and adds some interesting new functional aspects.

    The authors show that microglia that phagocytose tau aggregate-bearing neurons release seed-competent insoluble tau. Interestingly, co-culture with tau-positive neurons drives microglia into a senescent-like state characterized by a senescence-associated secretory profile (SASP) with a lower phagocytosis rate. Using tau-positive neurons similar to those in their previous study (Brelstaff et al., 2018) is an elegant way of investigating the impact of tau pathologies on microglia and how the microglia response can feed back to the neurons.

    The important results presented here raise interesting questions. For one, it will be important to identify the driving force that leads to the senescence-like phenotype, and to investigate if this is independent of the phagocytosis of neurons.

    In addition, the data suggest that the reduced phagocytic activity of microglia is independent of the presence of tau. But it remains unclear if the hypophagic phenotype develops only upon ingestion of the neurons or if it is mediated via a neuronal-secreted factor in a paracrine fashion. Investigating this in the future will be important to find potential new treatment targets. The work by Jack Brelstaff and colleagues provides the basis for analyzing this on a translational trajectory.


    . Depletion of microglia and inhibition of exosome synthesis halt tau propagation. Nat Neurosci. 2015 Nov;18(11):1584-93. Epub 2015 Oct 5 PubMed.

    . Clearance of senescent glial cells prevents tau-dependent pathology and cognitive decline. Nature. 2018 Oct;562(7728):578-582. Epub 2018 Sep 19 PubMed.

    . Living Neurons with Tau Filaments Aberrantly Expose Phosphatidylserine and Are Phagocytosed by Microglia. Cell Rep. 2018 Aug 21;24(8):1939-1948.e4. PubMed.

  5. Interestingly, in 1996 Frederick Maxfield’s group reported somewhat similar results for microglia and fibrillar Aβ.

    From their abstract: They examined the uptake, degradation, and release of small aggregates of fibrillar Aβ (fAβ) or soluble Aβ (sAβ) by microglia. They found that although some degradation of fAβ was observed over three days, no further degradation was observed over the next nine days. Instead, there was a slow release of intact Aβ. However, the poor degradation was not due to inhibition of lysosomal function, since the rate of α2-macroglobulin degradation was unaffected by the presence of fAβ in the late endosomes or lysosomes.

    Finally, the authors showed that while microglia internalize a large fraction of fAβ and sAβ, both are released without much degradation (Chung et al., 1999).


    . Uptake, degradation, and release of fibrillar and soluble forms of Alzheimer's amyloid beta-peptide by microglial cells. J Biol Chem. 1999 Nov 5;274(45):32301-8. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Tangles Turn Neuronal Membranes Inside Out, Give Microglia License to Eat Their Fill
  2. Are Tauopathies Caused by Neuronal and Glial Senescence?
  3. DAMned to Death? Microglia May Proliferate to Senescence

Paper Citations

  1. . Dystrophic (senescent) rather than activated microglial cells are associated with tau pathology and likely precede neurodegeneration in Alzheimer's disease. Acta Neuropathol. 2009 Oct;118(4):475-85. PubMed.
  2. . Neuroimmune Crosstalk through Extracellular Vesicles in Health and Disease. Trends Neurosci. 2019 May;42(5):361-372. Epub 2019 Mar 26 PubMed.
  3. . Clearance of senescent glial cells prevents tau-dependent pathology and cognitive decline. Nature. 2018 Oct;562(7728):578-582. Epub 2018 Sep 19 PubMed.
  4. . Proteopathic tau primes and activates interleukin-1β via myeloid-cell-specific MyD88- and NLRP3-ASC-inflammasome pathway. Cell Rep. 2021 Sep 21;36(12):109720. PubMed.

Other Citations

  1. frustrated phagocytosis

Further Reading

No Available Further Reading

Primary Papers

  1. . Microglia become hypofunctional and release metalloproteases and tau seeds when phagocytosing live neurons with P301S tau aggregates. Sci Adv. 2021 Oct 22;7(43):eabg4980. Epub 2021 Oct 20 PubMed.