Microgliosis goes hand-in-hand with Alzheimer’s disease, but the exact staging remains murky. A neuropathological study in the January 24 Nature Communications now places one type of microglial activation into the AD pathological cascade. Researchers led by Philip De Jager at Columbia University in New York found that the more plaques and tangles in a person’s cortex, the higher the proportion of amoeboid microglia there. A statistics method that teases out causal relationships indicated that these activated cells directly promoted tau accumulation, which, in turn, triggered cognitive decline. In addition, most of the genetic variants the authors linked to activated microglia in a genome-wide association analysis boosted the risk of AD, strengthening the causal connection. “These findings suggest we should time anti-inflammatory drugs to when there is amyloid pathology, but relatively few tangles,” De Jager said.

  • In human postmortem cortex, activated microglia correlate with AD pathology.
  • This microglial activation appears to come after plaques but before tangles.
  • Genes that heighten this activation also raise AD risk.

Other researchers praised the methodology and robustness of the finding. “This study is the largest to date directly demonstrating that microglia activation, and not abundance, is associated with AD,” Richard Bazinet at the University of Toronto wrote to Alzforum. Terrence Town at the University of Southern California, Los Angeles, was impressed by the number of different techniques used, including histological analysis, imaging, and high-throughput genomics and transcriptomics. “This paper nicely adds to the field because it shows a direct causal relationship between microglial activation and AD pathology,” he wrote.

Form Reveals Function. Microglia in healthy brain are highly ramified (left), while reactive microglia in inflamed brain are dense and globular (right). [Courtesy of Felsky et al., Nature Communications.]

Activated Microglia Linked Specifically to AD Pathology
Previous research flagged microglial changes as a key feature of Alzheimer’s disease (Davies et al., 2016). More recently, Ricardo Taipa and colleagues at the Centro Hospitalar do Porto, Portugal, found greater numbers of activated microglia in postmortem AD temporal cortices than in age-matched controls (Taipa et al., 2018). A recent review of 113 postmortem studies by Bazinet and colleagues concluded that activation of these cells, rather than their overall abundance, correlates best with AD (Hopperton et al., 2018). However, most of the individual studies in that review were small and did not include detailed pathological data or genotyping, nor correlate with cognitive assessments.

De Jager and colleagues undertook a more comprehensive analysis by leveraging data from the Religious Orders Study and the Memory and Aging Project. Both are longitudinal aging cohorts that enroll people without dementia, evaluate their cognitive and clinical health annually, and include brain donation after death. The authors selected a subset of 225 brains for analysis; 90 of these met pathological criteria for Alzheimer’s disease.

First author Daniel Felsky examined the appearance of microglia in two cortical and two subcortical regions of each. Healthy brains tend to contain slender, ramified microglia, while inflammation is associated with condensed, globular cells (see image above). The scientists found a greater proportion of such globular cells in the midfrontal and inferior temporal cortices of people with AD than in controls. In control brain, activated microglia were very rare, fewer than one in 200, whereas in Alzheimer’s cortex, the researchers counted about three times as many. This greater proportion of active microglia (PAM) in AD brain was highly significant. Moreover, people with PAM values in the highest quartile had about five times the risk of AD as those in the lowest quartile; this was comparable to the risk conferred by an ApoE4 allele in this cohort.

In contrast to these cortical regions, the PAM in two subcortical brain regions, the ventral medial caudate and the posterior putamen, were the same as in controls. “This regional specificity is an exciting finding,” Bazinet noted.

How does PAM relate to specific pathologies? The authors found the strongest correlation with amyloid plaques, the second-strongest with paired helical filaments (PHF) of tau and neurofibrillary tangles. However, high PAM did not correlate with Lewy bodies, TDP-43 deposits, or vascular pathology. The authors turned to mediation statistics, a type of regression analysis considered the gold standard for parsing the causal relationships between variables. With this, the authors identified a direct link between high PAM and PHF tau, and an indirect effect of PAM on global cognitive decline. They also found a causal link between amyloid plaques and high PAM, as well as the expected direct link between plaques and tau. Together, the data suggest that amyloid plaques and activated microglia act synergistically to promote tau pathology, which then triggers cognitive decline, the authors concluded.

This conclusion fits with data from Christian Haass at the German Center for Neurodegenerative Diseases (DZNE) in Munich. Haass found that high levels of the microglial marker sTREM2 in cerebrospinal fluid correlated with high p-tau, total tau, and cognitive decline, again suggesting that microglial activation lies upstream of tau (Dec 2016 news). Still, De Jager acknowledged that his findings do not prove causality. To get a better idea of the sequence of events, he is comparing microglial PET imaging to amyloid and tau PET in longitudinal studies.

Researchers were intrigued that activated microglia did not associate with other neuropathologies, despite inflammation being a common feature of neurodegenerative disease. De Jager suggested that neuroinflammation may occur in different brain regions in other diseases, or involve an alternate activation state of microglia. Taipa said evidence for the former exists, because he and others have found activated microglia in distinct brain regions in frontotemporal lobar degeneration (Lant et al., 2013; Taipa et al., 2017).

De Jager is investigating the possibility of different activation states. He recently analyzed the diversity of human microglia by single-cell transcriptomics, identifying 14 different subtypes (Jul 2018 news; Olah et al., 2018). Others have reported similar findings (Dec 2018 news). De Jager is profiling transcription in the activated, globular microglia to see whether they represent a single subtype, or multiple ones. The findings will provide a more precise determination of what form of microglial activation correlates with AD, he told Alzforum.

Others agreed this is an important question. Haass wondered if these globular microglia express markers of the previously identified disease-associated microglia (DAM; Jun 2017 news). For his part, Town has found that amyloid pathology prods microglia into a more inflammatory and less phagocytic state, suggesting the particular type of microglial activation is key (Feb 2015 news). “AD evolution may be driven by an imbalanced innate immune system, rather than generalized activation of innate immunity,” Town suggested.

What’s the Underlying Genetics?
For clues to what regulates microglia, De Jager and colleagues performed a GWAS using data from 2,067 participants in the ROS/MAP studies. This turned up two SNPs associated with PAM. One was rare, but the other, on chromosome 1, occurred in a third of the population. This SNP, rs2997325, associated with microglial activation in the midfrontal cortex specifically. It is unclear what the variant does. It lies near a non-coding RNA, LINC01361, but no genes. “Further study of this variant could be important in determining the mechanism of action of microglial activation in AD,” Bazinet suggested.

The authors tested the SNP’s association with microglial activation in 27 participants in the Indiana Memory and Aging Study. In that cohort, rs2997325 carriers had a higher microglial PET signal, as seen with the TSPO-binding ligand PBR28, in the left entorhinal cortex than noncarriers did, with an SUVR of around 1.6 versus 1.2. Given this, De Jager believes that much of the individual variation seen in microglial TSPO PET imaging may be a result of this SNP. Researchers may need to account for this when analyzing microglia PET to reduce noise in the data, he suggested. De Jager and colleagues adjusted for a TSPO variant that alters ligand binding.

In addition to the two genome-wide hits, the authors identified dozens of genes with weaker significance. They combined these into a polygenic risk score of microglial activation. Lo and behold, most of these genes are also AD risk genes. This suggests that having microglia that are easily activated puts people at risk of developing AD, the authors noted.

Researchers led by Bart De Strooper, Dementia Research Institute, U.K., and Mark Fiers at KU Leuven in Belgium have similar data. In a preprint on BioRxiv, they identified 11 new AD risk genes that are expressed in microglia and switched on by amyloid pathology in a mouse model. In keeping with De Jager’s model, these microglial genes are unaffected by tau pathology, suggesting they lie upstream of tau. De Strooper and others believe many genes affect AD risk by modulating the microglial response to Aβ. Numerous known AD genes, such as TREM2, CD33, and ABCA7, are primarily expressed in microglia. A recent large GWAS supports the idea that a large portion of AD risk comes from microglial gene expression (Jun 2017 news).—Madolyn Bowman Rogers

Comments

  1. This manuscript by Daniel Felsky, Philip De Jager, and colleagues provides a new perspective on the interplay between microglial activation, amyloid-β/tau pathology, and genetic risk. The authors infer that genetic risk can lead to microglial activation, which in turn results in AD. Data from our group (Sierksma et al., 2019) and Salih et al. (2018) similarly suggest that genetic risk for AD affects microglial response to amyloid-β, and that microglia activation is upstream of tau pathology, but our experimental approach is different, using a combination of gene expression analysis in response to tau or Aβ pathology and combining that with GWAS risk analysis. Both studies show a large part of risk for AD to be associated with microglia expressed genes.

    Felsky and colleagues demonstrate that a higher proportion of morphologically identified activated microglia (PAM) is associated with the presence of AD-specific pathology in two cortical regions (midfrontal and the inferior temporal cortex) as well as with cognitive decline, although the latter effect was predicted to be mediated through tau pathology. Several genomic variants were predicted to confer risk for region-specific microglial activation, and variant rs2997325T, possibly affecting expression of LINC01361, was also significantly associated with increased binding in the entorhinal cortex of a TSPO (microglia) PET ligand. Moreover, by building a polygenic risk model using the predicted GWAS variants for activated microglia in the two cortical regions, the authors could demonstrate that these variants also have predictive value for AD, although the inverse was not observed, i.e., a polygenic risk score model built from AD GWAS SNPs did not predict morphologically activated microglia. 

    Some questions remain:

    1. What causes the morphological activation of microglia? It would have been interesting to see if measures for amyloid-β or tau pathology were also predictive for the degree of microglial activation; i.e., to what extent are these three pathological observations co-occurring or can one pathological feature (e.g., amyloid-β pathology) drive the expression of the other (PAM). Transcriptomic data from AD mouse models suggest that amyloid-β pathology, and not tau pathology, triggers the microglial response, placing amyloid-β pathology as the instigator of the pathological cascade (Sierksma et al., 2019; Salih et al., 2018). These findings fit the predictions of the current paper where tangle pathology is positioned downstream of morphologically activated microglia. How activated microglia (and/or its interaction with amyloid-β accumulation) may lead to tangle formation remains to be determined.
    2. To what extent can the decline in cognitive performance be accounted for by cell loss? It has been previously demonstrated that gray matter volumes may partially mediate the effect of tau pathology on cognitive decline (Bejanin et al., 2017). A longitudinal study where structural imaging would be combined with tau- and microglia-specific PET imaging and cognitive assessment may validate their prediction that activated microglia mediate cognitive decline through tau accumulation.

    References:

    . Novel Alzheimer risk genes determine the microglia response to amyloid-β but not to TAU pathology. bioRxiv. January 16, 2019.

    . Genetic variability in response to Aβ deposition influences Alzheimer's risk. BioRxiv, October 8, 2018.

    . Tau pathology and neurodegeneration contribute to cognitive impairment in Alzheimer's disease. Brain. 2017 Dec 1;140(12):3286-3300. PubMed.

  2. Dr. De Jager’s group has elegantly revivified a 20-year-old technique in neuroimmunology of Alzheimer’s disease (AD)—histological assessment of microglial morphology. These investigators combined microglial stereology with high throughput genomics, transcriptomics, proteomics, imaging, and pathological analyses. From this tour de force of techniques, the authors were able to draw three primary conclusions: 1) in cortical regions, morphological evidence of microglial activation (but not proliferation) correlates with AD pathology as operationalized by Aβ deposits and paired helical filament tau; 2) microglial activation and Aβ load synergistically lead to tau pathology that induces cognitive decline; 3) a genomic propensity for activation of microglia causes increased AD risk.

    In past decades, we have linked innate immunity to AD progression and have identified mononuclear phagocytes as a primary therapeutic target. We accomplished this by 1) uncovering a circumstantial relationship between immunological traits and human AD and 2) establishing causality via genetic and pharmacological manipulation in rodent models. This paper nicely adds to the field because it shows a direct causal relationship between microglial activation and AD pathology in humans. Even more striking is the exclusivity to AD. Dr. De Jager’s group didn’t find a relationship between microglial activation and other common neuropathologies associated with aging and neuroinflammation. It is important to emphasize, though, that these results do not differentiate between different microglial activation states. This is significant, because we and others have suggested that AD progression is driven by an imbalanced innate immune system rather than generalized activation of innate immunity (Guillot-Sestier et al., 2013; Guillot-Sestier et al., 2015Guillot-Sestier et al., 2015). 

    References:

    . Innate Immunity in Alzheimer's Disease: A Complex Affair. CNS Neurol Disord Drug Targets. 2013 Apr 4; PubMed.

    . Il10 deficiency rebalances innate immunity to mitigate Alzheimer-like pathology. Neuron. 2015 Feb 4;85(3):534-48. Epub 2015 Jan 22 PubMed.

    . Innate Immunity Fights Alzheimer's Disease. Trends Neurosci. 2015 Nov;38(11):674-81. PubMed.

  3. This article presents important data for the discussion on the role of microglia in AD pathogenesis. Yet, it is still to be determined if microglia activation is the cause of neurodegeneration or a secondary reactive process; or if neurodegeneration is secondary to microglia senescence and associated loss of microglial protection. Despite the huge amount of research based on animal models, these still only mirror limited aspects of AD pathology in humans. Furthermore, brain aging is a dynamic process and due to the prolonged lifespan of CNS microglia (in humans compared to rodents), they are more susceptible to accumulate aging-related changes (beyond specific pathological conditions).

    This study represents one of the largest studies analyzing microglia activation in human brain in the context of Alzheimer´s disease and aging. The authors elegantly approach the question starting with detailed morphological studies of microglia in neuropathologically characterized postmortem brain tissue from two large clinical cohort studies of cognitive aging, then followed by genome-wide analyses to identify the genomic architecture of microglia activation.

    This an important study supporting direct microglia involvement in AD pathogenesis, showing that both microglia activation and amyloid-β contribute to tau pathology, and that microglia activation leads to cognitive decline indirectly via the accumulation of PHF-tau.  

    Despite the large cohort and accurate diagnostic quantitation of specific pathologies (namely other proteinopathies), and certainly highlighting the close relation of activated microglia to AD pathology (Aβ and Tau), the study could be missing other neurodegenerative dementias (particularly of young onset such FTLD) where these other proteinopathies have higher burden. For instance, morphological studies have reported higher levels of activated microglia in FTLD with some differences to AD in relation to anatomical distribution (Lant et al., 2014Taipa et al., 2017). It would be important that future studies, with similar approach, address the same question for other specified neurodegenerative dementias.

    The findings are also worth considering in the context of in vivo imaging using microglial markers. Hamelin and colleagues showed that higher initial 18F-DPA-714 binding is associated with better clinical prognosis after a two-year follow-up in AD (Hamelin et al., 2018). Interestingly, they also showed that patients with lowest initial 18F-DPA-714 binding had the greatest subsequent increase of microglial activation and unfavorable clinical outcome, while patients with highest initial 18F-DPA-714 binding had the lowest subsequent increase of microglial activation and more favorable clinical outcome, independently of the initial cortical amyloid load. With the advent of tau-targeted positron emission tomography tracers, it will be possible to extend the studies of the relationship between Aβ, tau, and activated microglia in AD and aging.

    References:

    . Distinct dynamic profiles of microglial activation are associated with progression of Alzheimer's disease. Brain. 2018 Jun 1;141(6):1855-1870. PubMed.

    . Patterns of microglial cell activation in frontotemporal lobar degeneration. Neuropathol Appl Neurobiol. 2014 Oct;40(6):686-96. PubMed.

    . Patterns of Microglial Cell Activation in Alzheimer Disease and Frontotemporal Lobar Degeneration. Neurodegener Dis. 2017;17(4-5):145-154. Epub 2017 Apr 27 PubMed.

  4. This is a tour de force, integrating many sophisticated forms of analysis. In the context of many other indices of neuroinflammation and microglial malactivation, the evidence that these cells participate in AD progression is difficult to refute.

    Regarding the ordering of events, however, I must confess one concern: The analyses seem to have bypassed the regions of the brain where tau pathology begins, namely entorhinal cortex and hippocampal formation. By limiting the regional distribution to neocortex (along with striatal comparisons), the authors may have fallen victim to the temptation to "look under the lamppost." Plaques become prevalent in the neocortex before they do in the hippocampus; tau spreads to the former region only in Braak Stages IV (barely) and V. So even if microglia were capable of reacting to both amyloid and tau pathology, it begs the question to state that they precede tau pathology in regions where it is late to arrive.

    I suppose what we need is a Braak-like staging of the pattern of PAM progression.

    View all comments by Steve Barger

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Paper Alert: Slotting TREM2 into Alzheimer’s Pathogenesis
  2. A Delicate Frontier: Human Microglia Focus of Attention at Keystone
  3. Microglia Reveal Formidable Complexity, Deep Culpability in AD
  4. Hot DAM: Specific Microglia Engulf Plaques
  5. Cytokine Takes Aβ Off the Menu for Microglia
  6. Microglial Master Regulator Tunes AD Risk Gene Expression, Age of Onset

Paper Citations

  1. . Microglia show altered morphology and reduced arborization in human brain during aging and Alzheimer's disease. Brain Pathol. 2016 Nov 14; PubMed.
  2. . Inflammatory pathology markers (activated microglia and reactive astrocytes) in early and late onset Alzheimer disease: a post mortem study. Neuropathol Appl Neurobiol. 2018 Apr;44(3):298-313. Epub 2017 Nov 27 PubMed.
  3. . Markers of microglia in post-mortem brain samples from patients with Alzheimer's disease: a systematic review. Mol Psychiatry. 2018 Feb;23(2):177-198. Epub 2017 Dec 12 PubMed.
  4. . Patterns of microglial cell activation in frontotemporal lobar degeneration. Neuropathol Appl Neurobiol. 2014 Oct;40(6):686-96. PubMed.
  5. . Patterns of Microglial Cell Activation in Alzheimer Disease and Frontotemporal Lobar Degeneration. Neurodegener Dis. 2017;17(4-5):145-154. Epub 2017 Apr 27 PubMed.
  6. . A single cell-based atlas of human microglial states reveals associations with neurological disorders and histopathological features of the aging brain. bioRxiv June 11, 2018

External Citations

  1. preprint

Further Reading

Primary Papers

  1. . Neuropathological correlates and genetic architecture of microglial activation in elderly human brain. Nat Commun. 2019 Jan 24;10(1):409. PubMed.