Myriad studies have placed microglia right in the middle of AD pathogenesis, but defining what, when, and how these cells influence the disease process in a person’s brain is difficult. Three recent gene-expression studies—one in Nature and two posted on bioRχiv—took the plunge into postmortem brain samples to try to find out how microglia and other cells were altered in Alzheimer’s. Two studies used single-nuclei RNA sequencing approaches to expose transcriptomic clusters of disease-related cells, while another sequenced RNA from sorted myeloid cells. In general, the single-nuclei approaches were limited by the small number of microglia cells, while the sorting approach was constrained by the quality of the RNA. For the most part, the studies found surprisingly little overlap between microglial gene-expression signatures in human Alzheimer’s with those previously identified in mouse models. An exception was ApoE, which ramped up across studies and species.

  • Three studies reported AD-linked gene-expression profiles from human postmortem brains.
  • Human microglia had an AD-related gene signature distinct from that seen earlier in mouse models.
  • ApoE expression was up in AD brain microglia, which expressed an “enhanced aging phenotype.”

“It is exciting to see all these studies examining cell-type-specific responses to Alzheimer’s disease pathology at the transcriptomic levels,” commented Jason Ulrich of Washington University in St. Louis. “The isolated cell-type approach and the single-nucleus approach shared some common findings that challenge whether microglial phenotypes described in mouse models of neurodegeneration exist in human disease.”

Genome-wide association studies have placed genetic risk for AD at the feet of microglia (Apr 2018 news). Gene-expression studies in AD mouse models have revealed how AD risk genes such as TREM2 and ApoE make microglia ditch a homeostatic gene-expression signature in favor of a disease-associated one as pathology worsens (Jun 2017 news; Sep 2017 news; Dec 2018 news). But do human microglia respond similarly to neurodegeneration? To take on the challenge of analyzing postmortem brain tissue, scientists have devised techniques that read disease-associated gene-expression signatures at the cell-type or single-cell level. As the rarest cell type in the brain, microglia remain the most difficult nut to crack, but that hasn’t stopped researchers from trying (Jul 2018 conference news). 

From Chaos to Clusters
On May 1 in Nature, researchers led by Li-Huei Tsai and Manolis Kellis at the Massachusetts Institute of Technology published a single-cell transcriptomic analysis of 48 postmortem brain samples from the Religious Orders Study and Memory and Aging Project (ROSMAP). Alzforum covered the portion of this work that Tsai presented at Keystone last year (Jul 2018 news). The samples came from 24 people with amyloid plaques and neurofibrillary tangles, and 24 with little to no AD pathology. Starting with a chunk of prefrontal cortex from each brain, first author Hansruedi Mathys and colleagues isolated nuclei from all of the cells and subjected them to single-nucleus RNA sequencing, aka snRNA-Seq or snuc-Seq.

The researchers made use of the 10x Genomics protocol. It separates each nucleus into a single droplet. In it, all nuclear RNA is tagged with a unique oligonucleotide barcode prior to sequencing. Compared with single-cell RNA sequencing, which requires brain tissue to be gently dissociated into a single-cell suspension, snRNA-Seq allows researchers to skip the dissociation step and directly isolate nuclei. For this reason, researchers think that snRNA-Seq will be particularly useful in analyzing individual transcriptomes from archived, frozen samples, from which single, reasonably intact cells are difficult to generate.

Postmortem SnRNA-Seq. Nuclei isolated from the brains of 48 people after death were subjected to snRNA-Seq. Transcriptomes were clustered by cell type, then subclustered into populations within each cell type. This process connected gene-expression changes to disease. [Courtesy of Mathys et al., Nature, 2019.]

In total, Mathys profiled 80,660 single-nucleus transcriptomes from the 48 people. Using cell-type-specific transcripts, they grouped the nuclei into eight broad clusters: excitatory neurons, inhibitory neurons, astrocytes, microglia, oligodendrocytes, oligodendrocyte progenitor cells (OPCs), endothelial cells, and pericytes.

They identified more than 1,000 differentially expressed genes (DEGs) in nuclei from AD versus control. While most DEGs in neurons were repressed in AD, a majority in microglia, astrocytes, and oligodendrocytes were turned up. Interestingly, while ApoE transcripts were higher in microglial nuclei from AD brain samples, they were lower in astrocytes.

When the researchers partitioned their samples based on Braak stage, they found that the cell-type specificity of DEGs waned as disease progressed. In early Braak stages, DEGs were highly cell-type specific, while in later stages, genes were more commonly upregulated across cell types. This included genes for protein folding and integrity, apoptosis, autophagy, and the general stress response. Christos Proukakis of University College London found this stage-specific analysis compelling. “This helps distinguish changes in an end-stage brain, which may simply be secondary and nonspecific,” he wrote to Alzforum.

Did each cell type respond differently to disease? The researchers addressed this by identifying modules of DEGs that correlated with specific pathological traits, essentially endophenotypes, such as Aβ level, neurofibrillary tangle burden, and cognition. They found 10 modules, M1–10, that correlated with AD phenotypes, and some of these modules comprised genes expressed predominantly by one cell type. For example, M6 and M7 contained genes whose expression in microglia correlated with AD pathology. They include the AD risk genes ApoE, TREM2, MEF2C, PICALM, and MHC Class II. M9 contained genes that were turned up in oligodendrocytes with increasing pathology. These genes were involved in myelination and oligodendrocyte differentiation, likely signifying the brain’s response to axonal damage.

The researchers broke down their eight cell-type clusters into subclusters based on their transcriptomes. The more abundant a cell type in the brain, the more transcriptomic subclusters it contained. In all, the scientists pinpointed 13 excitatory neuron (Ex), 12 inhibitory neuron (In), four astrocyte (Ast), five oligodendrocyte (Oli), three OPC, and four microglial (Mic) subclusters.

For each cell type, the researchers found subclusters that were overrepresented in AD or control samples. For example, Ex4, In0, Ast1, Oli0, and Mic1 tended to come from cells derived from AD brain samples, while Ex6, In2, Ast0, and Oli1 nuclei were more commonly from controls. Mathys and colleagues found that the AD-associated subpopulations were overrepresented in the brains of women, hinting at a transcriptional basis for the known gender disparity in AD.

Focusing in on the AD-associated microglial cluster Mic1, the scientists saw that many of the transcripts that defined the subpopulation were involved in immunity and inflammation. Of the 229 genes previously identified as upregulated in the disease-associated microglial (DAM) signature in 5xFAD mice, 28 were also part of the Mic1 signature, including ApoE, SPP1, and CD74 (Keren-Shaul et al., 2017). Of the 480 genes that Tsai had pegged previously as a late-disease microglial signature in p25 mice, 35 were also turned up in human Mic1 cells, including MHC class II genes such as HLA-DRB1 (Mathys et al., 2017). 

However, the human Mic1 signature also contained 49 genes not previously identified in mouse models, including complement component C1QB and the pattern-recognition receptor CD14. Some Mic1 genes overlapped with human microglia genes previously linked to aging (Olah et al., 2018). Overall, Mic1 expressed a suite of genes with some resemblance to the suite expressed in AD mice or in aged human microglia, but that also had its own distinctive aspects. Notably, TREM2, which had been tied to AD pathology in their gene module analysis, was not among the Mic1 signature genes.

Proukakis called the study a tour de force. Oleg Butovsky of Brigham and Women’s Hospital in Boston considers it the most comprehensive single-cell transcriptomic analysis of the human brain to date. To his mind, the findings support the idea that ApoE signaling in microglia is important in AD. However, he noted that because microglia are rare cells, the findings hinge on data from very few nuclei. Across all samples, the researchers analyzed about 2,000 microglial nuclei, 500 of which comprised the Mic1 subpopulation.

Tsai agreed that the small number of microglial nuclei was a limitation, but noted that the power to detect microglial gene-expression signatures will grow as more samples are analyzed. Compared with other brain regions, the dorsolateral prefrontal cortex, from which the scientists derived the nuclei, has relatively few microglia. “More and more people are doing single-cell RNA sequencing in human brain samples and in different brain regions,” Tsai wrote to Alzforum. “I am optimistic that we will soon accumulate enough data to properly analyze the microglia in healthy and disease conditions.”

Perfecting the Pipeline
To that end, researchers at Washington University in St. Louis have now amassed transcriptomic data from around one million single nuclei derived from 50 human brains, according to Carlos Cruchaga of WashU. The samples come from the parietal cortices of people with autosomal-dominant and sporadic AD, as well as age-matched controls, he said. While the scientists have yet to publish findings from this massive data set, on March 30, they uploaded a pilot study to bioRχiv, in which they used data from three of the brains to optimize methodological aspects of snRNA-Seq analysis.

First author Jorge Del-Aguila and colleagues sampled the parietal cortices of three women from the same family, who died between the ages of 82 and 89. All three had AD pathology upon autopsy. Only one carried the autosomal-dominant PSEN1-A79V mutation known for its wide range of age at onset. Based on scores on the clinical dementia rating (CDR) scale prior to death, the woman with autosomal-dominant AD and one sibling had severe dementia at death, while the other sibling had mild dementia.

The researchers performed snRNA-Seq on more than 26,000 nuclei from the three samples. For comparison, they sequenced RNA from bulk parietal tissue from the same samples, and found the two were highly concordant. After trying out several techniques to quantify and cluster the transcriptomes, they emerged with 13 clusters: Six came from excitatory neurons and two from inhibitory neurons, while astrocytes, microglia, oligodendrocytes, OPCs, and endothelial cells had only one transcriptome cluster each. The preprint manuscript does not present an extensive comparison of gene-expression profiles between the familial and sporadic AD samples. However, it does report that the sample from the mutation carrier had fewer excitatory neurons than the sporadic AD samples, in keeping with previous reports that excitatory neurons are most vulnerable in familial AD.

Strikingly, 90 percent of the nuclei in the astrocyte cluster derived from the PSEN1 mutation carrier. Why this is so is unknown, but Cruchaga speculated that this probably occurred during sampling of the tissue. Perhaps the researchers inadvertently selected a chunk of prefrontal cortex that had more astrocytes in one sample than the others. Cruchaga said the uneven distribution of cells in the brain makes this problem very difficult to control, but larger numbers of samples should diminish sampling inconsistencies.

Though they detected but one transcriptomic cluster of microglia, Del-Aguila still attempted to see if their gene expression profile aligned with the mouse DAM signature. Of the 500 genes linked to the DAM signature in Keren-Shaul et al., the researchers identified 326 human homologues, 92 of which were detected in the snRNA-Seq samples. They found 20 DAM markers elevated in the PSEN1 carrier, and 15 and 18 in the microglia from each of the two sporadic AD brains.

The researchers concluded that they had too few microglial nuclei to detect the DAM signature in the human cells, if it exists. Cruchaga said that as more samples are analyzed, a disease-associated signature could emerge, but he doubts that it would bear strong resemblance to mouse signatures. He added that in addition to doing snRNA-Seq on all of the nuclei from each sample, the scientists are taking a complementary approach to enrich for microglia, whereby they deplete neuronal nuclei prior to sequencing.

Single-nucleus RNA-Seq has worked well in exposing transcriptomic changes in neurons and microglia in autism spectrum disorder (ASD). In a paper published May 17 in Science, a team led by Arnold Kriegstein at the University of California, San Francisco, used snRNA-Seq to pinpoint sets of genes in upper-layer projection neurons and microglia that correlated with clinical severity of ASD (Velmeshev et al., 2019). In addition, by analyzing a cohort of patients with sporadic epilepsy, they were able to disentangle gene-expression changes related to seizures from those primarily associated with ASD.

Introducing HAM
In a preprint posted on bioRχiv April 19, researchers led by David Hansen and Brad Friedman at Genentech in South San Francisco used a different approach to enrich for certain cell types, including microglia. Rather than taking a single-nuclear approach to investigate AD-associated transcriptomes in all cell types in the brain, first author Karpagam Srinivasan and colleagues used cell sorting to isolate CD11b+ myeloid cells, which in the brain predominantly mean microglia. Sampling the human superior frontal gyrus, they then sequenced RNA from those myeloid cells in bulk from each sample. The SFG samples came from 22 people with AD pathology in Braak stage 5–6, plus 21 neurologically normal, age-matched controls at the Banner Sun Health Research Institute Brain and Body Donation Program in Arizona.

Because the researchers had to dissociate and fix the cells to sort them, the RNA quality of some of the samples was too poor to include in the analysis. Ultimately, the scientists analyzed myeloid cell gene-expression profiles from 10 people with AD and 15 controls.

Sort First, Ask Questions Later. Karpagam Srinivasan and colleagues dissociated, fixed, and sorted cells from postmortem brain samples using cell-type specific markers, including CD11b for myeloid cells. They presented RNA sequencing data from myeloid cells in their paper. [Courtesy of Srinivasan et al., BioRxiv, 2019.]

“The beauty of this method is that it does not require fresh tissue to capture the microglia signature,” commented Elizabeth Bradshaw of Columbia University in New York, who was not involved in the study. “However, the concern is that there might be a biological connection to the large number of samples that failed,” she added. Bradshaw suggested future studies could rule this out by comparing the findings from the sorted myeloid cells with single-nuclei transcriptomics from the same samples. “[This] would be a very informative comparison as the field of nuclei transcriptomics explodes,” she wrote. 

Hansen and colleagues reported that relative to controls, expression of 45 genes increased and 21 genes decreased in the AD myeloid cells. The researchers dubbed these gene sets “Myeloid AD-Up” and “Myeloid AD-Down,” respectively.

To validate their results, the researchers compared them with those from whole-tissue gene expression analysis from three cohorts. These included previously published RNA sequencing from AD and control fusiform gyrus, as well as newly analyzed FuG samples from a separate cohort (Friedman et al., 2018). They also cross-referenced their findings to bulk RNA sequencing data from the ROSMAP cohort of prefrontal cortex samples. In all three whole-tissue RNA data sets, expression of the Myeloid AD-Up genes was higher, and of the Myeloid AD-Down genes lower, in AD samples relative to controls. These expression differences widened at higher Braak stages. To the authors, the data from whole tissue not only helped validate their myeloid sorting method, but also suggested that the AD myeloid signature applied in different regions of the brain.

How does the AD myeloid gene set compare with disease gene signatures reported in mice? Friedman et al. had previously published just such a comparison using whole-tissue RNA sequencing from a larger number of human FuG samples. In it, they reported that human AD samples bore but a slight resemblance to the gene expression signatures of mouse neurodegeneration models (Feb 2018 news). 

Now, using their data from sorted myeloid cells instead, they found even fewer commonalities between the mouse and human gene signatures. In fact, ApoE was the lone upregulated gene in both mouse and human. Expression of most of the mouse disease-associated microglial genes had at best feeble upward or downward trends in the human AD myeloid data set. Similarly, genes identified as belonging to a “homeostatic signature” in mice, i.e., genes that turn down expression in response to disease or injury, were not similarly turned down in the human AD myeloid cells. Only one mouse homeostatic gene—SERPINF1—had significantly reduced expression in the human AD samples.

HAM, DAM, or Sometimes a DAM HAM
Because their cells looked nothing like DAM, Hansen and colleagues gave them a new acronym: human Alzheimer’s microglia, or HAM. Comparing the HAM profile to other published human data sets, they found that roughly half of the HAM genes overlapped with age-associated gene-expression changes reported in myeloid cells from fresh autopsy tissue (Jul 2017 news). Genes that upped expression with age, e.g., IL-15, also went up in AD relative to controls in Hansen’s cohort, whereas transcripts that sagged with age, such as CECR2, went down in Hansen’s controls. That said, the other half of HAM genes, including ApoE, did not change expression in microglia with age. The researchers summarized the HAM profile as part “enhanced aging” phenotype, and part age-independent, disease-related phenotype.

Friedman also compared the HAM profile to a single-cell RNA-Seq data set from tumefactive brain lesions, which mark a rare form of multiple sclerosis (Feb 2019 news). Surprisingly, he found that microglia from these MS lesions shared attributes with HAM and DAM profiles. Roughly half of DAM genes were elevated in the MS lesions, while about half of HAM downregulated genes and a third of HAM upregulated genes were similarly altered in the lesions. That human microglia were capable of expressing a DAM-like signature makes its absence in AD brains even more notable, the authors contend. It also supports the idea that human and mouse microglia are capable of mounting similar responses, in certain conditions.

Bradshaw also found this result noteworthy. “Importantly, they found that it is possible for human microglia to have a DAM phenotype, but it is not seen in the human microglia from Alzheimer’s disease patients,” she wrote. “This will have big implications as we work toward targeting microglia therapeutically.”

Despite the differences between DAM and HAM, the researchers also noted some commonalities. Both DAM and HAM genes overlap with genes induced by natural aging, though the specific genes differ between species. In particular, many genes involved in lipid and lysosomal biology—if not the exact same ones—were part of DAM and HAM profiles.

What explains the differences between HAM and DAM? Hansen et al. favor the idea that DAM activation in neurodegeneration mice reflect a protective response, while the HAM profile reflects defective microglia. GWAS indicate that subpar microglial responses can cause disease, Hansen noted. That, combined with the enhanced aging phenotype of HAM, suggests that age-associated microglial dysfunction fuels disease, he added.

Butovsky, who has identified gene profiles of microglia in several mouse models of disease, was skeptical of the findings because the RNA was in such bad shape. The authors reported an RNA integrity number (RIN)—a measure of RNA quality—of between 1 and 3 for their sorted myeloid samples. Butovsky said it’s best to work with RNA whose RIN is closer to 7. For his part, Cruchaga was not so quick to dismiss the findings. He noted that while high-quality RNA is crucial for microarray studies that require hybridization, lower-quality RNA can suffice for sequencing studies. Besides the RNA quality issues, Tsai noted that using myeloid markers to sort cells can be problematic because such markers might change in the diseased brain.

Hansen acknowledged the RNA quality issues, but countered that his cell-sorting method still offers important advantages over single-cell techniques. “By collecting tens of thousands of microglia from each sample, we get a richer picture of the whole microglial transcriptome. This allows for easier identification of differentially expressed genes,” he said. Friedman will compare his data to those from single-cell studies as they become available. “Hopefully, such data sets will yield additional information regarding whether the HAM profile within a tissue occurs in all microglia or in a distinct subpopulation, whether the HAM profile represents a single state or a combination of different cell states, and whether any individual cells exhibit a more DAM-like response,” Hansen wrote.

WashU’s Ulrich noted that DAM cells as described by Keren-Shaul et al. comprise a small proportion of microglia in the 5xFAD model. If the cells were as rare in humans, they could be difficult to capture using single-nucleus studies. “It would be interesting to use the cell-extraction approach described by Srinivasan et al. using putative DAM markers such as CD11c to enrich for microglia in DAM-like transcriptomic states,” he wrote. He noted that the SlideSeq approach—in which RNA from frozen tissue sections is transferred onto a surface covered with oligonucleotide barcodes—could assess the transcriptomic state of cells relative to their spatial proximity to pathology (Rodriques et al., 2019). 

Butovsky proposed a similar approach for single-nucleus studies. He believes the microglial nuclei analyzed in Mathys et al. likely derived from cells that were situated both near and far from plaques. Emerging technologies such as MERFISH and Nanostring Digital Spatial Profiling may allow analysis of microglia transcriptomes within their spatial context, he said, a feat that could bring pathology-associated gene expression changes into view (Moffitt et al., 2018; Decalf et al., 2019). 

Scientists can browse the data from each of these studies within interactive databases. The data from Mathys et al. is available to eligible users on the AMP-AD Knowledge portal:!Synapse:syn18485175. The burgeoning single-nuclei data set from WashU will be made publicly available for perusal ahead of formal publication, Cruchaga said, using this interactive browser: As of now, the site houses data from the three brains in the pilot study. Finally, the gene-expression data from Srinivasan et al., along with previously published results from RNA sequencing of bulk tissue and mouse models, is publicly available on the Genentech researchers’ database:

“The new single-cell studies provide the highest-resolution transcriptomic data sets of AD brain cells currently available, which can now be bioinformatically mined by the AD community,” commented Jerold Chun of the Sanford Burnham Prebys Medical Discovery Institute in La Jolla, California. “These are great, new resources.”—Jessica Shugart


  1. This is a tour de force by the teams of Li-Huei Tsai and Manolis Kellis. A particular strength is the separate treatment of early and advanced pathology, with analyses of early versus control, and late versus early. This helps distinguish changes in an end-stage brain, which may simply be secondary and non-specific.

    Comparison of these groups indeed shows very interesting results. Late changes are different, with upregulated genes often shared by most cell types, and consistent with a general response to stress. It would be interesting to determine if similar changes are seen at advanced stages of other neurodegenerative disorders, or other brain regions. Early changes are almost always specific to neurons, or one glial type. These changes were additionally shown to be distinct from profiles related to postmortem interval, or age at death.

    Another important message is that bulk RNA signal, when compared to single-cell data, was dominated by changes in excitatory neurons and oligodendrocytes. Microglia, emerging as key players in Alzheimer's, were very poorly represented in bulk RNA, highlighting the need for cell-type specific analysis. The importance of microglia is supported by the detection of a specific disease-associated microglial sub-population, although disease-related sub-populations of astrocytes and oligodendrocytes are now also reported.

    Interestingly, although there was clear overlap of disease-related microglia expression patterns with those reported in aged microglia, there were important differences, with the key risk factor APOE only found in the disease-related population. In another new study which analysed the microglial response in an App knock-in mouse model, which included prominent ApoE upregulation, this response was severely impaired in the absence of ApoE (Sala Frigerio et al., 2019). These studies suggest a role of APOE in the microglial response in Alzheimer’s.

    Tsai and Kellis correctly conclude that they cannot differentiate cause and effect in their findings, or beneficial from harmful changes, but they have clearly advanced our understanding of how pathology develops.


    . The Major Risk Factors for Alzheimer's Disease: Age, Sex, and Genes Modulate the Microglia Response to Aβ Plaques. Cell Rep. 2019 Apr 23;27(4):1293-1306.e6. PubMed.

    View all comments by Christos Proukakis
  2. We have DAM, BAM, PAM, and now HAM (Human Alzheimer’s Microglia/Myeloid cells signature) to help us better understand how microglia change with disease.

    Srinivasan et al. used a novel method to isolate various cell types from frozen brain tissue, where they successfully isolated RNA from about half of the CD11b+ samples. The beauty of this method is that it does not require fresh tissue to capture the microglia signature as our method using viable microglia does. This allows the authors to carefully select annotated samples from the Banner Sun Health Research Institute Brain and Body Donation Program.

    However, the concern is that there might be a biological connection to the large number of samples that failed. This could be ruled out by comparing the finding to transcriptomics from microglia-specific nuclei from the same subjects, which would be a very informative comparison as the field of nuclei transcriptomics explodes. 

    Interestingly, Srinivasan et al. did not find the mouse DAM phenotype enriched in their human microglia from Alzheimer’s patients. Rather, half of their signature appeared to resemble more of an enhanced aging phenotype rather than cell activation as in the DAM phenotype.

    Only APOE was found both in the DAM and HAM signatures. We had also found APOE upregulated in our human aged microglia profile (HuMi_Aged) of aged individuals, who all had some degree of Alzheimer’s pathology if not a clinical diagnosis.

    Importantly, Srinivasan et al. found that it is possible for human microglia to have a DAM phenotype, but it is not seen in the human microglia from Alzheimer’s disease patients. This will have big implications as we work toward targeting microglia therapeutically. The caveat to this work is the small sample size, but it is an important beginning.

    View all comments by Elizabeth Bradshaw
  3. It is exciting to see all these studies examining cell-type-specific responses to Alzheimer’s disease pathology at the transcriptomic levels. The isolated cell-type approach from Srinivasan, Friedman et al. and the single-nucleus approach from Mathys, Davila-Velderrain et al. shared some common findings that challenge whether microglial phenotypes described in mouse models of neurodegeneration exist in human disease. Although both studies identified upregulation of APOE in microglia from AD brain, other “disease-associated microglia” (DAM) markers were not as readily detectable.

    It is worth noting that DAM cells as described by Keren-Shaul et al. comprised a relatively small proportion of microglia in the 5xFAD amyloid model, which may make it difficult to capture with single-cell/nucleus analysis (Keren-Shaul et al., 2017). In this regard, it would be interesting to use the cell-extraction approach described by Srinivasan et al; using putative DAM markers such as CD11c would enrich for microglia in DAM-like transcriptomic states.

    Additionally, the SlideSeq approach recently described by Rodriques, Stickels et al. would be a powerful way to get around the difficulty in enriching for particular cell types such as microglia or astrocytes and assess the transcriptomic state of cells in regards to their spatial proximity to pathology (Rodriques et al., 2019). 


    . 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.

    . Slide-seq: A scalable technology for measuring genome-wide expression at high spatial resolution. Science. 2019 Mar 29;363(6434):1463-1467. Epub 2019 Mar 28 PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Hot DAM: Specific Microglia Engulf Plaques
  2. ApoE and Trem2 Flip a Microglial Switch in Neurodegenerative Disease
  3. Microglia Reveal Formidable Complexity, Deep Culpability in AD
  4. A Delicate Frontier: Human Microglia Focus of Attention at Keystone
  5. Microglial Transcriptome Hints at Shortcomings of AD Model
  6. Human and Mouse Microglia Look Alike, but Age Differently
  7. Single-Cell Profiling Maps Human Microglial Diversity, Flexibility

Mutations Citations

  1. PSEN1 A79V

Paper Citations

  1. . 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. . Temporal Tracking of Microglia Activation in Neurodegeneration at Single-Cell Resolution. Cell Rep. 2017 Oct 10;21(2):366-380. PubMed.
  3. . A transcriptomic atlas of aged human microglia. Nat Commun. 2018 Feb 7;9(1):539. PubMed.
  4. . Single-cell genomics identifies cell type-specific molecular changes in autism. Science. 2019 May 17;364(6441):685-689. PubMed.
  5. . Diverse Brain Myeloid Expression Profiles Reveal Distinct Microglial Activation States and Aspects of Alzheimer's Disease Not Evident in Mouse Models. Cell Rep. 2018 Jan 16;22(3):832-847. PubMed.
  6. . Slide-seq: A scalable technology for measuring genome-wide expression at high spatial resolution. Science. 2019 Mar 29;363(6434):1463-1467. Epub 2019 Mar 28 PubMed.
  7. . Molecular, spatial, and functional single-cell profiling of the hypothalamic preoptic region. Science. 2018 Nov 16;362(6416) Epub 2018 Nov 1 PubMed.
  8. . New tools for pathology: a user's review of a highly multiplexed method for in situ analysis of protein and RNA expression in tissue. J Pathol. 2019 Apr;247(5):650-661. Epub 2019 Feb 20 PubMed.

Other Citations

  1. Apr 2018 news

External Citations


Further Reading

No Available Further Reading

Primary Papers

  1. . Single-cell transcriptomic analysis of Alzheimer's disease. Nature. 2019 Jun;570(7761):332-337. Epub 2019 May 1 PubMed.
  2. . A single-nuclei RNA sequencing study of Mendelian and sporadic AD in the human brain. Alzheimers Res Ther. 2019 Aug 9;11(1):71. PubMed.
  3. . Alzheimer's Patient Microglia Exhibit Enhanced Aging and Unique Transcriptional Activation. Cell Rep. 2020 Jun 30;31(13):107843. PubMed.