Researchers often turn to mice when they need an animal model for what microglia do in Alzheimer’s and other neurodegenerative diseases. Alas, along with the mice come nagging doubts about how closely their microglia resemble the human counterpart. If the two don’t match up, using mice for translational research makes little sense. Now, researchers in the labs of Bart Eggen, University of Groningen in the Netherlands, and Suely Marie, University of São Paulo in Brazil, offer a note of caution to mouse modelers. Analyzing global gene expression in human microglia isolated from 39 postmortem human brain samples, they find that human and mice microglia indeed have a lot in common. Not everything, though—the way microglia age is distinct. Importantly for studies of age-related diseases, the changes in gene expression that distinguish cells from young and old donors showed barely any overlap between the two species. The work appeared July 3 in Nature Neuroscience.

“This is an extremely important study,” said Oleg Butovsky, Harvard Medical School, Boston. “More studies will need to be done to determine whether human microglia are really different from mouse, or if they are changed due to their different environment, but clearly this is a start to what we should be doing now, that is, studying real human microglia.”

Don’t toss those mouse models just yet, though, said Rickie Patani, University College London. “This study provides a valuable resource across the disciplines of neuroscience and immunology. The interspecies age-related divergence in microglial transcriptional signatures adds to accumulating evidence that supports the use of human postmortem and iPSC platforms to complement, but not replace, existing in vivo models of aging and neurodegeneration,” Patani wrote in an email to Alzforum.

For the growing number of investigators who are already switching from mouse to human microglia in their studies, the results provide a needed benchmark, said Shane Liddelow of Stanford University. “This resource is going to be heavily used by a lot of researchers—not only to determine the baseline expression of genes that should be looked for in any human microglia cell culture experiments, but also as a great comparative resource for people to purify microglia from postmortem brain with neurodegenerative disease, infection, or trauma.

“The differences with aging are interesting,” Liddelow added. “These genes could become key targets for further investigation of neurodegenerative disease.”

To source human microglia, first authors Thais Galatro and Inge Holtman turned to autopsy samples collected either in Groningen or São Paulo. Using fluorescence-activated cell sorting, they rapidly isolated viable CD11B+/CD45+ microglia from parietal cortex samples from 39 people who had died between the ages of 34 and 102 without a diagnosed brain disease. After deep sequencing the transcripts in each sample, the researchers identified a set of 1,297 genes that were expressed at least eight times higher in microglia compared to an unsorted cortical tissue sample. The genes, which the scientists called the human microglia core signature, fell into expected functional classes, with many related to the scavenger and innate immune activity of microglia. A proteomic analysis, western blotting, and immunostaining supported the RNA expression data. 

Common core: Biological functions, gene families, and representative genes of the human microglia transcriptome. [Galatro et al., Nature Neuroscience.]

Eggen's group documented extensive overlap of expression profiles between human and mouse microglia, both for mouse data generated in–house and taken from previously published reports. “The results should reassure researchers that human and mouse microglia are extremely similar,” said Eggen. “They share the majority of genes, they have the same function. They are the same cell type,” he said. 

Nonetheless, differences stood out. Compared with mouse, human cells expressed higher levels of genes related to immune function. Part of the difference lay in human-specific genes such as SIGLECs and MHCs; part of it lay in upregulation of shared genes, indicating a more active ongoing inflammatory response in humans than mice. 

Point of Divergence.

Venn diagram of changes in gene expression in human or mouse microglia. Only 14 genes go up in both mouse and man, and only nine go down in both, for an overlap of less than 1 percent between species. [Galatro et al., Nature Neuroscience.]

With regard to aging, Eggen and colleagues tallied 212 genes that increased and 360 that decreased their expression with age. Hierarchical clustering of the top 100 most affected genes clearly distinguished young and old profiles. But compared with profiles of aging mouse microglia (Grabert et al., 2016), the aging human microglia were starkly different—only 14 genes increased and only nine decreased in both species.

The results suggest that human microglia age differently than mice, and do so in an interesting way. Many human genes that change with age are involved in regulating the actin cytoskeleton, and most of those go down. This could lead to microglia that are less able to remodel their cytoskeleton, and hence less nimble when called upon. “We haven’t done the experiments, but it could be that older microglia are less able to scan their environment, because their protrusions are less mobile. Maybe these cells are less able to migrate to sites of damage or infiltration, because their actin cytoskeleton is not as functional anymore as it was at the beginning,” Eggen said.

Aging Actin: Microglia may slow down with age, as they downregulate genes involved in cytoskeletal dynamics. [Galatro et al., Nature Neuroscience.]

Data on the human microglial proteome is scarce. Until recently, only one study had been published, on three tissue samples derived from temporal lobe surgery (Zhang et al., 2016; see Jan 2016 news). Overall, Eggen’s new data looked similar to that study. 

More recently, Christopher Glass’s lab at Stanford published a transcriptome analysis of highly purified microglia from another human source, i.e., surgical biopsies from 19 children with epilepsy (Jun 2017 news). Fortuitously, the Glass and Eggen groups used similar methods for microglia isolation and gene expression analysis, Glass wrote to Alzforum by email. That allowed them to directly compare their data. When they did, they uncovered an extremely high correlation, with a Pearson coefficient of 0.94, between the average gene expression values of both cohorts. “This is important because it means that the data sets are likely to reflect a largely normal human microglia transcriptome, and that they can be combined to provide increased power for downstream analysis. This is clearly an important new resource for the field,” Glass wrote about the current paper. “It’s reassuring that we independently find the same thing,” Eggen agreed.

Do the differences in gene expression, either at steady state or with age, reflect intrinsic species differences, or other factors? Microglia are by nature responsive to their environment, and the increased expression of immune function, for example, could well be due to surroundings, Eggen said. His study used six-month-old mice raised in relatively clean surroundings and facing few immune challenges compared with humans who are out and about in the world for years. Events before and around the time of death, including comorbidities, may rev up microglia in a way that does not occur in samples taken from healthy mice. The study only looked at one region of the brain, the right parietal cortex, and it is possible that different regions will give different results. “There could be many factors that play into what we see,” said Eggen.

Neuropathologist Hans Lassmann of the Medical University of Vienna said the new work echoes the main contrast he sees between human and mouse microglial phenotypes. In humans, microglia appear to be what he calls “pre-activated”—they express proinflammatory markers, whose expression increases with age (Zrzavy et al., 2017). Rodents do not show the same markers of inflammation. In humans, then, disease-specific events are superimposed on this state of pre-activated microglia, Lassmann explained.

Eggen and his group are continuing to characterize human microglia, in both normal brain and disease. Currently his group is isolating microglia from people with AD and other neurodegenerative diseases, and comparing them to mouse models, he told Alzforum. Besides profiling microglia from different brain regions, they are profiling single cells. The researchers also would like to explore sex differences. The current study by chance included seven women to 32 men, and the scientists are collecting more samples to even that out.—Pat McCaffrey


  1. Overall the paper is a great resource—there really is nothing out there at the moment for microglia researchers to use as a gold standard for what the transcriptome of human microglia should be.

    The validation of selected human microglial transcriptome targets with immunostaining of human postmortem tissue is a great addition. Of course this tissue is from similar postmortem delay, and so does not remove the concern that these genes may only be expressed due to activation in a dead brain. But it does wonderfully show their localization in microglia and not other cells.

    The differences with aging are interesting. In figure 5, the heatmap shows numbers of genes upregulated in senescence and others downregulated over the same aging timeframe. These could become key targets for further investigation of neurodegenerative disease.

    This new human microglial transcriptome database was produced from microglia isolated from the right parietal cortex. Considering there is now considerable information about the heterogeneity of microglia, are the differences seen between the human and rodent cells due to species differences or regional heterogeneity differences? I.e., were the compared mouse samples prepared from the same brain region?

    The postmortem delay, which is always the problem when dealing with microglia, is reasonable and likely renders moot most arguments about cells randomly becoming activated due to slow cell preparations.

    This resource is going to be heavily used by a lot of researchers—not only to determine the baseline expression of genes that should be looked for in any human microglia cell culture experiments, but also as a great comparative resource for people to purify microglia from patient postmortem brain with neurodegenerative disease, infection, or trauma.

  2. It is reassuring that these data show good concordance with the data set from the Barres Lab in 2016. We now have an additional treasure trove for investigators interested in mechanistic learnings from neuro-glia (microglia and astrocyte) interactions and interactors in rodents, with potential to inform on human studies.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Purification of Adult Human Astrocytes Shows: They Are Unique
  2. What Makes a Microglia? Tales from the Transcriptome

Paper Citations

  1. . Purification and Characterization of Progenitor and Mature Human Astrocytes Reveals Transcriptional and Functional Differences with Mouse. Neuron. 2016 Jan 6;89(1):37-53. Epub 2015 Dec 10 PubMed.
  2. . Loss of 'homeostatic' microglia and patterns of their activation in active multiple sclerosis. Brain. 2017 Jul 1;140(7):1900-1913. PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . Transcriptomic analysis of purified human cortical microglia reveals age-associated changes. Nat Neurosci. 2017 Aug;20(8):1162-1171. Epub 2017 Jul 3 PubMed.