In Alzheimer’s disease, microglia reveal their two faces: On the one hand, they serve as protective scavengers of toxic Aβ protein deposits. On the other, amyloid incites them to phagocytose healthy synapses and contribute to neurodegeneration. Now, a study uncovers an unexpected connection between the ALS/FTD-associated protein TDP-43, microglial phagocytic activity, and synapse loss that could have ramifications for multiple neurodegenerative diseases. When researchers in the lab of Lawrence Rajendran, University of Zurich, deleted TDP-43 in microglia, they saw the cells’ production of lysosomes and their phagocytic activity shoot up. On the up side, that led to more aggressive clearance of Aβ and lower plaque load in mouse models of AD. On the down side, the TDP-less microglia avidly gobbled up synapses, even in animals without amyloid. The authors contend that clinical data support their idea that people with TDP-43 pathology clear amyloid more effectively: They appear to have a lower incidence of AD and less amyloid deposition. Published online June 29 in Neuron, the work proposes a physiological role for TDP-43 in regulating microglial phagocytosis, and suggests that dysfunctional microglia could be directly responsible for neurodegeneration in ALS/FTD and other TDP-43 proteinopathies.

“This is a really interesting paper, which provides more support for the role of microglia in neurodegenerative diseases generally,” said Cynthia Lemere, Brigham and Women’s Hospital, Boston. The results mirror her own lab’s recent work on microglia and complement-mediated synapse loss in AD models, where less-active microglia oversee less clearance of amyloid but also less synapse loss. Those mice have increased amyloid load, but better cognitive function (see Jun 2017 news). “Loss of TDP-43 moves everything exactly in the opposite direction—more active microglia, more clearance of Aβ, and at the same time, more synapse loss,” Lemere said. “These results say we have to be careful when we think about stimulating phagocytosis as a therapeutic approach—we really need to think through what that might be doing.”

Marco Colonna of Washington University in St. Louis called the paper “provocative” for its focus on loss of microglial TDP-43 function in neurodegeneration. “TDP-43 is supposed to be neurotoxic because it accumulates as misfolded protein in cytosolic granules in neurons. This work shows that loss of TDP-43 function also impacts disease, and not only in neurons,” he told Alzforum.

First author Rosa Paolicelli stumbled onto TDP-43 when she screened a panel of candidate proteins, all associated with late-onset AD or neurodegeneration, for their ability to regulate microglial clearance of Aβ. Paolicelli used siRNA to knock down candidates in the BV2 microglia cell line in culture, then measured the cells’ ability to take up and degrade Aβ. Of 18 genes tested, she found one hit: After knockdown of TDP-43, the BV2 cells became voracious consumers of Aβ. They cleared Aβ from conditioned media from either a wild-type mouse cell line or from HeLa cells expressing mutated human APP.

The result came as a surprise, Rajendran said. TDP-43 is an RNA-binding protein, transcriptional repressor, and splicing factor. Normally a nuclear protein, TDP-43 forms cytoplasmic inclusions in glia and neurons in ALS and FTD, and in other neurodegenerative diseases. Loss of TDP-43 function in neurons causes cell death; before Paolicelli’s experiment, nothing was known of its role in microglia.

How did loss of TDP-43 increase Aβ uptake? Paolicelli found those cells nonspecifically jacked up phagocytosis, rapidly taking up Aβ and other cargoes, including labeled dextran and transferrin. After uptake, the cells degraded Aβ, thanks to increased lysosomal function and biogenesis in the BV2 cells. Primary microglia behaved in the same way. 

Synaptic Feast.

Microglia (green) lacking TDP-43 (cKO) are more phagocytic and contain higher levels of engulfed synaptic markers (PSD95, red) than intact microglia (WT). [Courtesy of Neuron, Paolicelli et al.]

To assess how TDP-43 affects amyloid clearance in vivo, the researchers created mice in which TDP-43 was inducibly knocked out only in microglia. In the mice, the researchers downregulated the TDP-43 transcript and protein by administering tamoxifen, then injected Aβ42 oligomers into the cortex. In the knockouts, microglia accumulated in the amyloid core and its vicinity in greater numbers, and engulfed more Aβ, than wild-type littermates. The researchers got similar results when they crossed the conditional TDP-43 knockout with the ArcAβ/APParc mouse model of AD. After tamoxifen treatment, the mice showed a 60 percent reduction in Aβ peptide levels in brain homogenates, and a similar reduction in plaque load, with no effect on Aβ production. Such active removal of amyloid seemed promising, Rajendran said, but his team’s enthusiasm waned when they noticed that the phagocytic microglia also set upon synapses, resulting in a significant drop in the levels of synaptic markers PSD95, synapsin, synaptophysin, and vGlut1 in cortical areas. 

In mouse models of AD, microglial pruning of synapses depends on complement and on the presence of amyloid. Pruning in the TDP-43 knockout mice, however, did not require amyloid deposition. Even in non-AD, wild-type mice, microglial TDP-43 knockout led to a loss of synaptic markers and decreased cortical dendritic spine density. More microglia contained traces of engulfed synapses than in wild-type mice (see image above). “This is important because it shows we don't need another pathology, or that microglia activation is a secondary effect of Aβ load,” said Rajendran. “Removal of TDP-43 is sufficient to activate an aberrant phagocytosis phenotype, and as a result the microglia gulp up Aβ and also synapses.” The scientists have not tested whether the synaptic clearing was still dependent on complement protein C3, he said.

Are these findings relevant to human disease? If microglia lacking TDP-43 are turbocharged to remove Aβ, the authors reasoned, people with TDP-43 pathology might show a lower incidence of AD. They analyzed a cohort of 698 people with ALS, of whom 98 percent would be expected to have TDP-43 inclusions. They found that, among the 168 patients older than 75, 12 had AD, an incidence of 7.1 percent, compared to an expected incidence of 17 percent in this age group. This supports the idea that TDP43 pathology might be associated with delayed onset or less AD, Rajendran believes, presumably due to less amyloid deposition. At the same time, patients with ALS did experience a subtle decrease in cognitive function with age, which the authors speculate may be due to synapse loss.

The researchers also measured amyloid load in a brain autopsy cohort, comparing 40 controls with 62 subjects with AD and 60 with TDP-43 pathology (35 cases of ALS and 25 of FTLD with TDP-43). They saw a similar plaque burden across several brain regions in younger controls (65-74 years old) and the TDP-43 patients. In older cases (75 years old or more), they detected increased Aβ in normal healthy brains and even higher levels in AD cases. In contrast, Aβ levels in TDP-43 patients were lower than either AD or control subjects. The authors interpret this to indicate increased Aβ clearance in the TDP-43 patients. However, this conclusion was questioned by several researchers who told Alzforum they would like to see more details of the demographic differences between the groups, especially age and sex. ALS and FTLD-TDP occur at younger ages than AD, and a difference in age distribution in the groups might lead to apparent differences in amyloid load, for example.

The part of the study dealing with the relationship between amyloid and TDP-43 pathology in people raised skepticism from researchers not involved in the work. In their comment (see below), Carol Brayne, Suvi Hokkanen, and Sally Hunter, University of Cambridge, England, U.K., sound a cautionary note that the limited data shown may not represent a larger universe of TDP-43 pathology patients. In their work and the work of others, they find no relationship between TDP-43 and amyloid pathology in population-representative samples (Keage et al., 2014). 

Rik Ossenkoppele, VU University Medical Center, Amsterdam, is first author on a meta-analysis of PET studies looking at amyloid pathology across multiple neurodegenerative diseases (May 2015 news). His analysis did show a lower prevalence of amyloid pathology in FTD patients than in healthy controls, he wrote to Alzforum in an email. “Still, I would not consider this direct evidence for enhanced clearance of amyloid pathology in people with TDP-43 diseases, as FTD is caused by a variety of pathogenic proteins (i.e., 3R/4R tau, TDP type A/B/C/D/U, FUS), and it’s unclear from our clinical data whether it is TDP or the other pathologies that are associated with a reduction of amyloid pathology,” he wrote. Ossenkoppele told Alzforum that he has additional unpublished data on a disease group with homogenous TDP-43 pathology, where the prevalence of amyloid pathology is not lower than in the general population.

What about a role for TDP-43 microglial pathology in ALS and other TDP-43 proteinopathies? The authors report finding microglial TDP-43 cytoplasmic inclusions, albeit rare, in four out of four patients with motor neuron disease. Glial inclusions have been reported previously, Brayne and colleagues wrote, but are frequently disregarded, and “the current study usefully highlights this important aspect.” The authors also report an increase in CD68 staining in the vicinity of TDP-43 pathology in human brain tissues. Whether this indicates a critical role for TDP-43 in regulating microglia, or is merely an expected accompaniment to neurodegeneration independent of TDP-53, remains to be clarified.—Pat McCaffrey


  1. This study examines the relationships between TDP-43 inclusions, microglia, and dementia in ALS, FTLD, and AD. Work presented was from mouse or cell culture with human comparative studies. The main findings were that depletion of TDP-43 in microglia is associated with both enhanced loss of dendritic spines and synapses and enhanced Aβ clearance via microglial phagocytosis and lysosomal degradation, but not with enhanced Aβ production. Synapse and dendritic spine loss occurred independently from presence/absence of amyloid and was associated only with loss of microglial TDP-43. The human studies suggest that TDP-43 pathology is associated with lower deposition of amyloid pathology. Among ALS and FTLD-TDP patients, AD prevalence was lower compared to age-matched controls, and ALS with TDP-43 is reported to show more markers of synapse loss.

    We see the following reasons to be cautious. The human comparative studies are from autopsy cohorts and are not population-representative. In the Cambridge City over 75s Cohort study (CC75C) (Keage et al., 2014), TDP-43 pathology is common and can occur in those with and those without dementia. There were significant associations between TDP-43 and neuronal loss and dementia but no significant associations between TDP-43 inclusions and amyloid deposition in this population- representative sample. Work investigating the association between TDP-43, hippocampal sclerosis, and amyloid deposition in CC75C and also the Cognitive Ageing and Function Study (CFAS) found no significant association between amyloid deposition and TDP-43 in terms of HScl neuronal loss in the older population (Hokkanen et al., 2017, submitted). TDP-43 has been previously associated with neurofibrillary tangles, but not with amyloid pathology (review Wilson et al., 2011). TDP-43 in AD may aggravate the clinical symptoms and progression of AD and in human postmortem studies, and the presence of TDP-43 is reported to be higher in AD cases. This suggests that while the interaction between TDP-43 and microglia is interesting, within the older population where most AD-type dementia occurs, the relevance of this mechanism to dementia is questionable and requires further investigation.

    The associations found in the autopsy cohorts listed in Paolicelli et al. have not been controlled for sex, and the age ranges investigated suggest that this study is weighted toward the younger old. The selection of cases and controls may have biased the findings, making translation to the human population difficult. Synaptic loss through microglia activation is an interesting suggested mechanism for clinical dementia, but needs further confirmation and replication in unbiased settings. More detailed population- and community-based evaluations of TDP-43, amyloid, and synapses are needed.

    As the authors cited, most studies on hypothesized toxic gain of function of TDP-43 aggregates and TDP-43 loss of function have been associated with neuronal inclusions, even though TDP-43 pathology is observed in both neurons and glia. Glial TDP-43 is frequently disregarded and the Paolicelli et al. study usefully highlights this important aspect.

    It is important to note that results from studies looking at neuronal TDP-43 may be conflicting. Amyloid plaque formation has been reported to be reduced also in mice models overexpressing neuronal TDP-43 (Davis et al., 2017), as well as in mice with depletion of neuronal TDP-43 (LaClair et al., 2016). This study requires replication and investigation in other laboratory models.

    AD is a complex clinicopathologically defined syndrome with many different players that cannot be reduced to simple models in the lab. It is not really possible to separate the contributions of distinct glial versus neuronal TDP-43 pathologies nor separate gain-of-function vs. loss-of function mechanisms in living systems such as mouse models, as this does not represent the actual situation in the living brain.

    In summary, this work usefully highlights the role of microglia in neurodegenerative disease, but we cannot say on the basis of this evidence whether this mechanism is relevant in all neurodegenerative disease. Population studies of brain aging suggest that mechanisms related to TDP-43-associated neurodegeneration and dementia in the older population may be different from those seen in ALS and FTLD. Detailed population-based investigations of all factors associated with dementia are required to translate laboratory based findings to actual neurodegenerative disease in the human population.

    Sally Hunter also contributed to this comment.


    . TDP-43 expression influences amyloidβ plaque deposition and tau aggregation. Neurobiol Dis. 2017 Jul;103:154-162. Epub 2017 Apr 20 PubMed.

    . TDP-43 pathology in the population: prevalence and associations with dementia and age. J Alzheimers Dis. 2014;42(2):641-50. PubMed.

    . Depletion of TDP-43 decreases fibril and plaque β-amyloid and exacerbates neurodegeneration in an Alzheimer's mouse model. Acta Neuropathol. 2016 Dec;132(6):859-873. Epub 2016 Oct 26 PubMed.

    . TDP-43 in aging and Alzheimer's disease - a review. Int J Clin Exp Pathol. 2011;4(2):147-55. PubMed.

  2. The recent discoveries that microglia produce ASC specks, thereby contributing to Aβ plaque formation, while their capacity to remove plaques and synapses is enhanced by loss of TDP-43 highlight, once again, the paradoxical nature of the evidence regarding the role of microglia in Alzheimer's disease.

    Microglia appear to do anything and the opposite. Thus, they have been reported to limit, potentiate, or have no effect on plaque formation, to aberrantly prune synapses via the complement system, to foster aberrant tau aggregation, and to be dystrophic and severely damaged by tau, which would incapacitate them to carry out any of the aforementioned protective or detrimental actions.

    I doubt that a unitary explanation exists for this conflicting evidence. Rather, the problem may be that the experimental manipulations and models—e.g., pharmacological or genetic ablation of microglia, transgenic overexpression or deletion of genes identified as risk factors in GWAS, and massive overexpression of mutated APP and tau proteins—are overly aggressive and stir microglia to extreme phenotypes with little relevance to human aged brains. More experimental finesse is necessary.

    I also believe that key questions remain overlooked: what the functions of microglia and their striking motility are in the adult, healthy brain, and whether these functions, whose molecular underpinnings are still poorly understood, are affected by age and genetic risk factors. The answers to these questions may help us understand the consequences of microglia malfunction in neurodegeneration.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Sans Complement: Amyloid Grows, Synapses and Memory Stay
  2. Meta-Analyses Deliver Most Definitive Data Yet on Amyloid Prevalence

Research Models Citations

  1. ArcAβ

Paper Citations

  1. . TDP-43 pathology in the population: prevalence and associations with dementia and age. J Alzheimers Dis. 2014;42(2):641-50. PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . TDP-43 Depletion in Microglia Promotes Amyloid Clearance but Also Induces Synapse Loss. Neuron. 2017 Jul 19;95(2):297-308.e6. Epub 2017 Jun 29 PubMed.