Are microglia the masterminds of amyloid plaque formation? In the August 21 Nature Communications, researchers led by Kim Green at the University of California, Irvine, suggest these cells seed amyloid deposits. In a mouse model of amyloidosis, obliterating microglia at a young age all but prevented parenchymal plaques. Instead, Aβ ended up in blood vessels, inducing a cerebral amyloid angiopathy (CAA). Those few plaques that did form in the brain wreaked havoc on surrounding neurites, however, possibly because they were not encapsulated by microglia. “At early stages of Alzheimer’s disease, microglia have a complicated involvement, both protecting and harming the brain,” Green said. He presented some of these data at a recent Keystone symposium (Jul 2019 conference news).

  • A second study reports that eliminating microglia abolishes plaque formation.
  • Excess Aβ accumulates in blood vessels instead.
  • Restoring microglia seeds new plaques, while vascular amyloid remains.

Other researchers said the findings elucidate the contribution of microglia to amyloidosis. “The authors present clear data that the presence of microglia is required to allow amyloid plaque deposition,” Christian Haass at the German Center for Neurodegenerative Diseases in Munich wrote to Alzforum (comment below). Mathias Jucker at the University of Tübingen, Germany, who did a similar study 10 years ago and found the opposite effect on plaques, focused on the vascular component. “To me, the most fascinating point is the increase in CAA … Quite possibly, vascular amyloid could be more damaging to the brain than parenchymal plaques, which would render microglial seeding of parenchymal deposits protective,” he wrote (comment below).

This study was made possible by the development of drugs to selectively kill off microglia. Five years ago, Green and colleagues, in collaboration with scientists at Plexxikon Inc., developed an inhibitor of colony-stimulating factor 1 receptor (CSF1R), an essential microglial survival factor. The inhibitor, PLX3397, almost completely wiped out these cells in wild-type mouse brain. After withdrawing the drug, microglia replenished themselves (Apr 2014 news). 

Brain or Blood? In the cortices of 5xFAD mice that lack microglia, amyloid (green) accumulates only in blood vessels, not in gray matter (left, low mag; right, high mag). [Courtesy of Spangenberg et al., Nature Communications.]

Green and colleagues used PLX3397 to abolish microglia in 10-month-old, plaque-ridden 5xFAD mice. Their plaque burden did not budge, but more neurons and synapses survived and cognition improved (Spangenberg et al., 2016). Meanwhile, Charles Glabe and colleagues at UC Irvine fed PLX3397 to young, pre-plaque 5xFAD mice. This eliminated 80 percent of microglia and prevented 90 percent of plaques, while maintaining normal performance on a fear-conditioning test (Mar 2018 news). 

However, PLX3397 inhibited not only CSF1R but also two homologous immune cell receptors. Hence Green’s group collaborated with Plexxikon to redesign it to be more specific and brain-penetrant. The new version, PLX5622, is 20-fold more selective for CSF1R than for homologous receptors, and 20 percent of it enters the brain. Indeed, it eliminated virtually all microglia from wild-type mice though, curiously, these mice maintained normal learning and memory six months after their microglia were gone.

Armed with this new pharmacological tool, Green and colleagues examined what microglial depletion would do to young 5xFAD mice. First author Elizabeth Spangenberg fed PLX5622 to 6-week-old animals until they were either 4 or 7 months old, then analyzed their brains. In the 4-month-old mice, microglia were absent from the cortex, but a few remained in the thalamus, retrosplenial cortex, and subiculum. Unexpectedly, amyloid plaques paralleled this distribution, with almost none in the cortex, but a few near remaining microglia in the other regions. “Because microglia are believed to phagocytose amyloid, we assumed that when we got rid of microglia, more plaques would form. But we saw the opposite,” Green said.

By 7 months of age, microglia hung on only in the subiculum. Still, some plaques persisted in the thalamus and retrosplenial cortex. These were more diffuse and irregularly shaped than plaques in untreated 5xFAD mice and contained little ApoE, consistent with prior reports that microglia are the main source of this protein in plaques (Jan 2019 news). The halo of dystrophic neurites that typically surrounds amyloid plaques was 50 percent larger in PLX5622 mice than in controls. This fits with the idea that microglia form a barrier around plaques that protects surrounding tissue (May 2016 news; Jun 2017). 

Meanwhile, the scientists ascertained that the treated 5xFAD mice continued to produce as much amyloid precursor protein and its cleavage products as ever. In both 4- and 7-month-old 5xFAD mice on PLX5622, this excess Aβ peptide deposited in cerebral blood vessels. CAA can cause hemorrhages, and indeed, the researchers found microbleeds in these animals in the thalamus, where vessels are particularly leaky in mouse models of CAA (Davis et al., 2004). Notably, 5xFAD usually develop no CAA, and the untreated controls in this study did not, either. “The vascular pathology was immense,” Green said. “These results almost suggest that [one of] the functions of microglia in the normal brain is to protect us against CAA.”

Seven-month-old 5xFAD mice have no measurable defect in learning; however, they do spend more time in the open arm of an elevated plus maze, a sign of impairment in a normal anxiety behavior for these nocturnal animals. The 5xFAD mice without microglia spent even more time in the open arm.

What happens to the CAA mice if microglia return? Green and colleagues stopped feeding PLX5622 to 12 5xFAD mice at 4 months of age and, a month later, examined the brains. Lo and behold, they were speckled with just as many plaques as those of age-matched 5xFAD mice never exposed to the inhibitor, although their plaques were 25 percent smaller on average. And their CAA remained high, matching that in 5-month-old 5xFAD mice on PLX5622. “The fact that plaques come back is a new finding, and very intriguing,” commented Oleg Butovsky at Brigham and Women’s Hospital, Boston. Butovsky was a reviewer on Green’s paper.

The big question raised by this work is how microglia might promote plaque formation. Opinions vary. Green believes microglia take up extracellular Aβ and induce it to aggregate inside themselves, thus seeding plaques. In support of this, the authors found Aβ aggregates lurking inside microglial lysosomes in 5xFAD and 3xTg mice. These microglia were often located far from plaques, making it unlikely the cells were simply ingesting plaque material. Green and colleagues saw the same phenomenon in samples from postmortem human brains with AD pathology. Some earlier research also reported finding amyloid fibrils forming inside microglia, and suggested a role for these cells in seeding plaques (Wisniewski et al., 1990; Frackowiak et al., 1992; Stalder et al., 1999). 

For his part, Glabe believes amyloid forms inside neurons, because plaques are typically surrounded by neuronal remnants (Mar 2013 conference news). Glabe previously suggested that crosstalk between microglia and neurons might stimulate intraneuronal plaque formation. Haass favors a third idea, that microglia secrete factors, such as ApoE or the protein complexes known as ASC specks, that facilitate extracellular Aβ fibrillization (Dec 2017 news). 

Others raised caveats with the findings. Jucker noted that eliminating microglia for four weeks from 3-month-old APPPS1 mice, which more aggressively deposit amyloid than 5xFAD mice, did not dent plaque formation (Grathwohl et al., 2009). “In an aggressive model, microglia may play less of a role in deposition,” Jucker speculated. Butovsky pointed out that eliminating a cell type from the brain changes the milieu and the crosstalk between cells. Therefore, these experiments do not prove that plaque formation is a direct effect of microglia, he said.

Researchers led by Sangram Sisodia at the University of Chicago reported in the August 21 Journal of Neuroscience that PLX5622 eliminates microglia from mice carrying the pathogenic M146V mutation in presenilin 1. The inhibitor rescued neurogenesis, even though these 9-week-old animals had no evidence of amyloid plaques, and it reduced anxiety to wild-type levels, as seen by animals spending more time in lit areas, grooming less, and rearing up more to explore their surroundings. In both PS1 and 5xFAD mice, PLX5622 lowered anxiety.

One thing all the researchers agreed on is that eliminating microglia is no therapeutic strategy for Alzheimer’s disease. Mice are raised in a sterile environment, and so fare well without immune cells in the brain. In people, lack of microglia could lead to infections such as meningitis. Instead, the goal is to figure out what microglial phenotype best wards off Alzheimer’s pathology, Butovsky said. In future work, Green will try to answer this by studying how microglia contribute to plaque formation, and how these cells affect tau pathology.—Madolyn Bowman Rogers


  1. The role of microglia in amyloid generation/removal appears to be very complex. The authors now present clear data that the presence of microglia is required to allow amyloid plaque deposition. We favor their argument that microglia may secrete factors that facilitate Aβ fibrillization. The authors nicely demonstrate that in the absence of microglia, dense core amyloid plaques have 50–70 percent less ApoE, which is in good agreement with our recent findings in TREM2 loss-of-function mice (Parhizkar et al., 2019). It may be that the absence of microglial ApoE (or other seeding factors released from microglia such as ASC specks (Venegas et al., 2017)) reduces amyloidogenesis and therefore prevents plaque deposition and compaction.

    Our longitudinal amyloid-PET-imaging experiments also revealed an increased fibrillar amyloidogenesis early during pathogenesis in APPPS1/Trem2−/− mice at 3 and 6 months, which leveled out during late disease progression at 12 months of age. This could have been due to the failure of mutant microglia to cluster around plaques and to increase ApoE expression (or other factors driving amyloid plaque compaction and growth). Nevertheless, one should be careful comparing total microglial absence with TREM2 loss-of-function mutants, because the latter selectively target a microglia population with a disease-associated signature but leave homeostatic microglia intact (Krasemann et al., 2017; Keren-Shaul et al., 2017). 


    . Loss of TREM2 function increases amyloid seeding but reduces plaque-associated ApoE. Nat Neurosci. 2019 Feb;22(2):191-204. Epub 2019 Jan 7 PubMed.

    . Microglia-derived ASC specks cross-seed amyloid-β in Alzheimer's disease. Nature. 2017 Dec 20;552(7685):355-361. PubMed.

    . The TREM2-APOE Pathway Drives the Transcriptional Phenotype of Dysfunctional Microglia in Neurodegenerative Diseases. Immunity. 2017 Sep 19;47(3):566-581.e9. PubMed.

    . A Unique Microglia Type Associated with Restricting Development of Alzheimer's Disease. Cell. 2017 Jun 15;169(7):1276-1290.e17. Epub 2017 Jun 8 PubMed.

  2. This nice study by Green’s group investigates what happens after elimination of microglia in the 5xFAD mouse model. The hypothesis they propose, i.e., that microglia take up Aβ and form fibrils, is interesting but not novel. Wisniewski and others came to a similar conclusion many years ago (e.g. Wisniewski et al., 1990; Frackowiak et al., 1992; for more references see Stalder et al., 1999). We previously reported that eliminating microglia in the APPPS1 mouse model for four weeks does not inhibit amyloid plaque formation, and I wonder whether the difference is the aggressiveness of that model (Grathwohl et al., 2009). APPPS1 mice have heavy Aβ42-driven pathology and carry a PS1 mutation, which in humans causes one of the earliest onsets of AD. Thus, in an aggressive model the microglia may play less of a role in deposition compared to a less aggressive model.

    To me, the most fascinating point is the increase in CAA. The role of microglia in cerebral β-amyloidosis is undisputed but it may be related to the recent discovery of microglia heterogeneity, because there is also mechanistic heterogeneity in amyloid fibril formation. Knocking out one pathway of amyloid deposition may cause another one to become more active. Interestingly, the equivalent amounts of total Aβ in PLX562-treated vs. non-treated animals suggest that microglial ablation shifts Aβ from the parenchyma to the vasculature, where it is then presumably deposited. Quite possibly, vascular amyloid could be more damaging to the human brain than parenchymal plaques, in which case microglial seeding of parenchymal deposits would be protective. Vascular amyloid can be prevented by early reduction of Aβ (Schelle et al., 2019). 


    . Ultrastructure of the cells forming amyloid fibers in Alzheimer disease and scrapie. Am J Med Genet Suppl. 1990;7:287-97. PubMed.

    . Ultrastructure of the microglia that phagocytose amyloid and the microglia that produce beta-amyloid fibrils. Acta Neuropathol. 1992;84(3):225-33. PubMed.

    . Association of microglia with amyloid plaques in brains of APP23 transgenic mice. Am J Pathol. 1999 Jun;154(6):1673-84. PubMed.

    . Formation and maintenance of Alzheimer's disease beta-amyloid plaques in the absence of microglia. Nat Neurosci. 2009 Nov;12(11):1361-3. PubMed.

    . Early Aβ reduction prevents progression of cerebral amyloid angiopathy. Ann Neurol. 2019 Oct;86(4):561-571. Epub 2019 Aug 19 PubMed.

  3. This latest work from Kim Green’s laboratory shows that prolonged microglia depletion in 5xFAD mice (a common mouse model for AD) greatly reduces plaque formation, thus confirming recent findings by another group (Sosna et al., 2018). Notably, this phenomenon occurs when microglia are depleted before plaques begin to form (around 3–4 months of age). However, if the depletion starts after plaque formation, no obvious change in plaque load can be observed (Spangenberg et al., 2016). Together, these data suggest that microglia are critical for plaque deposition, at least at the early stages of disease.

    These studies will probably raise a lively debate. One may intuitively interpret these findings as a demonstration that microglia are primarily detrimental during amyloid pathology. However, the current paradigm identifies the soluble Aβ (especially oligomers) as the most neurotoxic form, whereas fibrillar Aβ is considered relatively benign. According to this view, fibrillization and subsequent deposition of soluble Aβ would reduce the bioavailability and diffusion of the most toxic Aβ species. Alongside the reduced plaque burden, Green’s data show an aggravated CAA (cerebral amyloid angiopathy), increased dystrophic neurites, and higher APP levels in 5xFAD mice with depleted microglia, as compared with non-depleted 5xFAD. Additionally, microglia depletion did not alter the production of soluble Aβ fraction, and no change could be seen at the learning and memory tests.

    Taken together, while microglia may be truly important for the formation of amyloid plaques, it remains to be determined if microglia depletion may represent a promising therapeutic strategy for AD.  Overall, the full picture of the role of microglia in AD may be more complicated than the idea that microglia are simply “beneficial” or “detrimental.” This work by Green’s laboratory certainly opens new perspective on the mechanisms of Aβ aggregation and plaque deposition.


    . Early long-term administration of the CSF1R inhibitor PLX3397 ablates microglia and reduces accumulation of intraneuronal amyloid, neuritic plaque deposition and pre-fibrillar oligomers in 5XFAD mouse model of Alzheimer's disease. Mol Neurodegener. 2018 Mar 1;13(1):11. PubMed.

    . Eliminating microglia in Alzheimer's mice prevents neuronal loss without modulating amyloid-β pathology. Brain. 2016 Apr;139(Pt 4):1265-81. Epub 2016 Feb 26 PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. TREM2, Microglia Dampen Dangerous Liaisons Between Aβ and Tau
  2. Microglial Magic: Drug Wipes Them Out, New Set Appears
  3. Wiping Out Microglia Prevents Neuritic Plaques
  4. Without TREM2, Plaques Grow Fast in Mice, Have Less ApoE
  5. Barrier Function: TREM2 Helps Microglia to Compact Amyloid Plaques
  6. Hot DAM: Specific Microglia Engulf Plaques
  7. Like Star Born of Supernova, Plaque Born of Exploded Neuron?
  8. Do Microglia Spread Aβ Plaques?

Research Models Citations

  1. 5xFAD (B6SJL)
  2. 3xTg
  3. APPPS1

Paper Citations

  1. . Eliminating microglia in Alzheimer's mice prevents neuronal loss without modulating amyloid-β pathology. Brain. 2016 Apr;139(Pt 4):1265-81. Epub 2016 Feb 26 PubMed.
  2. . Early-onset and robust cerebral microvascular accumulation of amyloid beta-protein in transgenic mice expressing low levels of a vasculotropic Dutch/Iowa mutant form of amyloid beta-protein precursor. J Biol Chem. 2004 May 7;279(19):20296-306. Epub 2004 Feb 25 PubMed.
  3. . Ultrastructure of the cells forming amyloid fibers in Alzheimer disease and scrapie. Am J Med Genet Suppl. 1990;7:287-97. PubMed.
  4. . Ultrastructure of the microglia that phagocytose amyloid and the microglia that produce beta-amyloid fibrils. Acta Neuropathol. 1992;84(3):225-33. PubMed.
  5. . Association of microglia with amyloid plaques in brains of APP23 transgenic mice. Am J Pathol. 1999 Jun;154(6):1673-84. PubMed.
  6. . Formation and maintenance of Alzheimer's disease beta-amyloid plaques in the absence of microglia. Nat Neurosci. 2009 Nov;12(11):1361-3. PubMed.

Further Reading

Primary Papers

  1. . Sustained microglial depletion with CSF1R inhibitor impairs parenchymal plaque development in an Alzheimer's disease model. Nat Commun. 2019 Aug 21;10(1):3758. PubMed.
  2. . Deficits in Enrichment-Dependent Neurogenesis and Enhanced Anxiety Behaviors Mediated by Expression of Alzheimer's Disease-Linked Ps1 Variants Are Rescued by Microglial Depletion. J Neurosci. 2019 Aug 21;39(34):6766-6780. Epub 2019 Jun 19 PubMed.