The Alzheimer's risk allele ApoE4 is firmly linked to both amyloid plaques and neurofibrillary tangles, and yet the apolipoprotein may rattle the brain long before these pathologic lesions do. In the October 11 Nature Aging, researchers led by Sudeshna Das, Massachusetts General Hospital, Charlestown, report that among older people who have nary a whiff of AD pathology, ApoE4 may have already altered the fate of microglia. In ApoE4 carriers, these cells expressed a different suite of genes than did microglia in people who carry ApoE2, the protective ApoE allele. “The surprising discovery suggests that ApoE4 primes microglia toward a more phagocytic and inflammation phenotype,” Das told Alzforum.

  • Microglial gene-expression cluster found in ApoE4 carriers.
  • The genes are expressed regardless of Alzheimer’s pathology.
  • The cluster may prime microglia to respond to brain damage.

Others pointed out that the findings are based on transcriptomic profiling of bulk RNA from human brain, which lacks the resolution of single-cell or single-nuclei analyses that have taken the field by storm of late. “It is very difficult to come to any solid conclusion based on whole-tissue analysis,” noted Oleg Butovsky, Brigham and Women’s Hospital, Boston. Das agreed that single-cell analysis will provide valuable information. “We will need a higher-resolution picture of expression at the individual cell level, and we will also need to understand the spatial relationship between these changes and plaques and tangles in the brain,” she said.

First author Alberto Serrano-Pozo and colleagues spotted the inflammatory microglial signature by analyzing RNA taken postmortem from the forebrains of volunteers in the Religious Orders Study/Memory and Aging Project at Rush University, Chicago. The researchers analyzed the RNA for microglial and astrocyte gene-expression patterns that could explain why ApoE2 protects the brain, while ApoE4 causes mayhem. They took a spectral clustering approach, looking for suites of genes that were more highly expressed in ApoE3 versus ApoE2 carriers, and higher again in ApoE4s. They also searched the other way around, for expression that followed an ApoE2>ApoE3>ApoE4 gradient. Either pattern might explain the benefits and risks associated with the different alleles.

The scientists focused on 519 genes known to be predominantly microglial. One cluster of 172 transcripts stood out. Relative to samples from ApoE3 carriers, this was downregulated in ApoE2 samples and upregulated in ApoE4s. All participants had a CERAD rating of 0, i.e., no detectable amyloid in the brain at postmortem.

Spectral Clustering. Among ROSMAP samples, a cluster of 172 microglial genes was more highly expressed in ApoE4 carriers than in ApoE3s or ApoE2s.

Analysis of bulk RNA from people who had no tangles in the forebrain, namely with Braak scores of 0, II, or III, identified a cluster of 200 genes that was similarly downregulated in ApoE2s and upregulated in ApoE4s. Half of these genes also turned up in the 172-gene cluster identified in the CERAD 0 population. All told, the data suggests that ApoE4 drives expression of a particular cadre of microglial genes in the absence of any plaque or tangle pathology. Serrano-Pozo found a similar cluster when he analyzed bulk transcriptomic data from the Mount Sinai Brain Bank.

What does the cluster do? Genes regulating inflammation and phagocytosis, including TREM2 and TYROBP, were strongly represented, suggesting the cells may be primed to respond to pathological changes in the brain. In support of this, the authors found a similar cluster in brain samples from people who had abundant amyloid plaques, i.e., a CERAD score of 3.

Next, the scientists probed this ApoE-gene cluster in ApoE knock-in mice. Serrano-Pozo analyzed transcriptomes of ApoE2, E3, and E4 mice that had been reported by researchers in Guojun Bu’s lab at the Mayo Clinic, Jacksonville, Florida (Zhao et al., 2020). Alas, only two of the 172 genes in the ApoE cluster were linked to ApoE genotype in these mice. Das found this unsurprising, since she said it confirmed previous reports of little overlap in expression among different ApoE knock-in lines. But Yadong Huang, Gladstone Institute, San Francisco, found it interesting. “This suggests that the APOE genotype acts, in concert with the cumulative effect of other insults on microglia during brain aging, to distinguish this microglia-APOE cluster in brains of elders,” he wrote.

Indeed, Das and colleagues report that an age-associated expression cluster in ApoE4 knock-in mice overlapped with 37 of the 172 genes in the human ApoE4 cluster. Among these transcripts were some involved in phagocytosis and inflammation, including TREM2 and TYROBP.

They authors tested whether this expression would be further exacerbated by other insults, such as amyloid plaques, by examining transcriptomes of 12-month-old APP/PS1 mice, which have abundant plaques. Compared to ApoE3 x APP/PS1 mice, those with an ApoE4 knock-in more highly expressed some of the key genes of the ApoE cluster, including Trem2, Tyrobp, C1qa, Cd68, and Cd74. “Taken together, these mouse experiments support the idea that APOE regulates microglial gene expression through the TREM2–TYROBP axis and can bias the microglial transcriptomic phenotype in a scenario of chronic stimulation,” write the authors.

In the same issue of Nature Aging, Huang and colleagues published a drug-repurposing study in which the diuretic bumetanide reversed ApoE4 expression signatures in microglia in ApoE4 knock-in mice only when amyloid pathology is present (Oct 2021 news). “It is worth testing whether bumetanide can reverse the transcriptomic profile of the microglia-APOE cluster identified by Serrano-Pozo et al.,” he wrote.—Tom Fagan

Comments

  1. In this nice study analyzing RNA-seq data from two cohorts of human brain samples, Serrano-Pozo et al. identified a cluster of microglial genes that are upregulated in APOE4 and downregulated in APOE2 carriers relative to APOE3 homozygotes in normal (neuritic plaque-free) brains. Importantly, this microglia-APOE cluster is enriched in phagocytosis and proinflammatory genes, and is also detectable in brains with frequent neuritic plaques. Since the association of this microglial gene cluster with APOE4 was strongest in normal, neuritic plaque-free brains, this suggests that APOE4 itself renders microglia toward a disease-associated state during normal brain aging.

    Interestingly, comparing the human microglia-APOE cluster with available RNA-seq data from human APOE knock-in (APOE-KI) mice showed very little overlap. This suggests that the APOE genotype acts, in concert with the cumulative effect of other insults on microglia during brain aging, to distinguish this microglia-APOE cluster in brains of elders.

    It would be interesting to see whether the microglia-APOE cluster (or a similar cluster) could also be found in single-cell (or single-nucleus) RNA-Seq datasets with APOE genotype information, once they are available in the future. Additionally, since APOE can be made in multiple types of cells in the central nervous system (CNS), including astrocytes as well as stressed microglia and neurons, this study raises an important question as to whether this microglia-APOE cluster reflects the effects of microglia-produced APOE or the combined effects of different cellular sources of APOE in the CNS.

    Linking this study to our recently reported computational repurposing of bumetanide for APOE4-related Alzheimer’s disease (AD) (Taubes et al., 2021), it is interesting to note that bumetanide treatment reversed the APOE4/4-specific transcriptomic signature of AD in neurons, but not in glia cells, in aged APOE4-KI mice, while the same treatment reversed the transcriptomic signature in both neurons and microglia in APOE4-KI mice with human Ab accumulation (APOE4-KI/J20 mice). We hypothesized that this was likely due to the fact that microglial activation (microgliosis), which represents a pathological hallmark of AD, occurs in APOE4-KI/J20 mouse brains, but not in aged APOE4-KI mouse brains. It is conceivable that the transcriptomic profile of activated microglia in response to Ab accumulation make them more responsive to bumetanide’s beneficial effects on transcriptome. Thus, it is worth testing whether bumetanide can reverse the transcriptomic profile of the microglia-APOE cluster identified by Serrano-Pozo et al.

    References:

    . Experimental and real-world evidence supporting the computational repurposing of bumetanide for APOE4-related Alzheimer’s disease. Nature Aging, 1, 2021, pp. 932–47. Nat Aging.

  2. APOE’s role in immune regulation is widely acknowledged these days. Out of the three human APOE isoforms, APOE4 is the ancestral allele from which APOE3 and APOE2 evolved. APOE4 confers protection against infection and is associated with a higher level of immune response, and it was hypothesized to have evolved when infectious pathogens were a major cause of mortality. APOE4 is also reported to mount higher anti-tumor immunity and suppress cancer progression.

    In Alzheimer’s disease, we and others found that APOE4 is associated with a higher level of microglial activation that is linked to exacerbated AD pathologies and neurodegeneration. However, as the level of microglial activation always positively correlates with the degree of pathology and neurodegeneration, it’s usually difficult to separate out a direct effect of APOE on microglial activation from an indirect effect resulting from its role in regulating pathology and neurodegeneration, which subsequently induces different levels of microglial activation.

    In this interesting paper, the authors found higher upregulation of phagocytotic and proinflammatory microglial genes in normal aging APOE4 carriers who don’t bear neuritic plaque pathology in the brain, suggesting direct modulation of microglial function by APOE4 in human subjects. This stands in line with other evidence of enhanced immunity conferred by APOE4. The upregulation of TREM2 and TYROBP by APOE4 suggests that the APOE-TREM2-TYROBP axis may play an essential role in mediating APOE4’s immunomodulatory effects on microglia.

    On the other hand, whether the effect of APOE4 on microglial gene expression is inherent in APOE4 carriers under physiological conditions, i.e., present in the young population, thus serving as a predisposition to exacerbated microglial responses in neurodegenerative conditions, or whether it manifests upon the presence of microglial-activating signals, such as cell damage in the aged brain or other comorbid pathologies, still awaits further investigation.

    It’s possible that the observed phenotype results from APOE-regulated microglial response to damage-associated molecular patterns (DAMPs) in the aged brain, which most likely involve lipids released from damaged cells or myelin. As a lipid modulator, ApoE isoform-dependently influences brain cell lipid composition, and thus may regulate microglial immune responses by affecting the lipid signals being released, or by controlling microglial lipid metabolism and associated activating pathways.

  3. Serrano-Pozo and colleagues tested the hypothesis that APOE alleles differently impact glial responses independent of neuritic amyloid plaques. They performed spectral clustering analyses on the expression levels of 519 microglia-predominant and 397 astrocyte-predominant genes from the dorsolateral prefrontal cortex (DLPFC) using the bulk RNA-Seq dataset obtained from the Religious Orders Study and Memory and Aging Project (ROSMAP). They subdivided the cohort based on the degree of neuritic plaques (NPs).

    In the brains with no NPs, they identified microglia and astrocyte gene clusters that were upregulated in APOE4+ but downregulated in APOE2+ compared to APOE3/3 individuals, or vice versa. These genes were defined as the APOE allele-related genes. Interestingly, they found that APOE4 was significantly correlated with the upregulation of a cluster of phagocytic and proinflammatory microglia genes, such as TREM2, TYROBP, CD68, etc., whereas these genes were downregulated in the presence of APOE2 allele when comparing to APOE3/3 cases.

    These analyses were then performed in the dataset obtained from the Mount Sinai Brain Bank (MSBB), which includes four different brain regions. They found these APOE allele-related genes were marginally significant (p=0.052) in APOE4+ cases, but did not differ in APOE2+ carriers, when compared to APOE3/3 cases. Similar expression analyses were performed with astrocyte-predominant genes, and the authors found a dysregulation of lipid metabolism and the extracellular matrix associated with APOE genotype.

    This is an important study investigating the role of APOE in glia cells with interesting findings. However, one limitation of these analyses is that the number of APOE4+ cases is relatively small. For example, in ROSMAP with no NPs, there were 16 APOE4 cases, comparing to 113 APOE3/3 and 36 APOE2 cases; similarly, in MSBB with no NPs, there were seven to nine APOE4 cases, comparing to 49 to 58 APOE3/3 and eight to nine APOE2 cases. Although these brains did not have NPs, they could have other comorbid pathologies including cerebral amyloid angiopathy, arteriosclerosis, Lewy body pathology, TDP-43 pathology, and hippocampal sclerosis. In particular, in the ROSMAP cohort, APOE4 carriers seem to have higher frequency of cerebral amyloid angiopathy (37.9 percent in APOE2, 29.5 percent in APOE3, and 56.2 percent in APOE4) and TDP-43 pathology (42.5 percent in APOE2, 43.3 percent in APOE3, and 52.5 percent in APOE4). These co-pathologies can also contribute to the microglia response and therefore potentially affect the correlation of the microglia gene profile with APOE gene alleles. Therefore, the conclusion needs to be further validated using a well-controlled and balanced cohort.

    When testing the list of APOE allele-related genes in the bulk RNA-Seq dataset against human ApoE-targeted replacement (TR) mice generated by our group, the authors found these genes were highly overlapped with aging-induced changes in these mice, but had minimal overlap with APOE4’s effect. This is not too surprising considering the known differences between humans and mice. The ApoE-TR models are devoid of any pathologies; therefore, aging is the strongest driver impacting pathways such as those related to immune responses with the top hub genes including Trem2, Tyrobp, and Cd68.

    In general, the APOE-genotype effect on brain transcriptomics was much milder compared to the aging effect in these mice; however, the interactive effects between aging and APOE genotype do exist, with APOE4 mice showing greater upregulation of immune response genes in aged cohorts compared to APOE2 and APOE3 mice. The network, or co-expression, of the immune response-related genes in mice might not be the same as in humans; therefore, the direct comparison with selected genes from humans to mice may lead to minimal similarity.

    Lastly, the analyses in this study are based on bulk RNA-Seq datasets with selected 519 microglial and 397 astrocytic genes. Considering the complexity of gene network regulations in the brain, the study of APOE allele effects on different cell types would require higher resolution profiling by single cell/nuclei transcriptomics.

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Can an Old Diuretic Drug Disarm APOE4, Prevent Alzheimer’s?

Paper Citations

  1. . Alzheimer's Risk Factors Age, APOE Genotype, and Sex Drive Distinct Molecular Pathways. Neuron. 2020 Jun 3;106(5):727-742.e6. Epub 2020 Mar 20 PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . Effect of APOE alleles on the glial transcriptome in normal aging and Alzheimer’s disease. Nature Aging, 1, 2021, pp. 919-31. Nat Aging.